linux/drivers/dma-buf/dma-buf.c

1747 lines
47 KiB
C
Raw Permalink Normal View History

// SPDX-License-Identifier: GPL-2.0-only
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
/*
* Framework for buffer objects that can be shared across devices/subsystems.
*
* Copyright(C) 2011 Linaro Limited. All rights reserved.
* Author: Sumit Semwal <sumit.semwal@ti.com>
*
* Many thanks to linaro-mm-sig list, and specially
* Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and
* Daniel Vetter <daniel@ffwll.ch> for their support in creation and
* refining of this idea.
*/
#include <linux/fs.h>
#include <linux/slab.h>
#include <linux/dma-buf.h>
dma-buf: Rename struct fence to dma_fence I plan to usurp the short name of struct fence for a core kernel struct, and so I need to rename the specialised fence/timeline for DMA operations to make room. A consensus was reached in https://lists.freedesktop.org/archives/dri-devel/2016-July/113083.html that making clear this fence applies to DMA operations was a good thing. Since then the patch has grown a bit as usage increases, so hopefully it remains a good thing! (v2...: rebase, rerun spatch) v3: Compile on msm, spotted a manual fixup that I broke. v4: Try again for msm, sorry Daniel coccinelle script: @@ @@ - struct fence + struct dma_fence @@ @@ - struct fence_ops + struct dma_fence_ops @@ @@ - struct fence_cb + struct dma_fence_cb @@ @@ - struct fence_array + struct dma_fence_array @@ @@ - enum fence_flag_bits + enum dma_fence_flag_bits @@ @@ ( - fence_init + dma_fence_init | - fence_release + dma_fence_release | - fence_free + dma_fence_free | - fence_get + dma_fence_get | - fence_get_rcu + dma_fence_get_rcu | - fence_put + dma_fence_put | - fence_signal + dma_fence_signal | - fence_signal_locked + dma_fence_signal_locked | - fence_default_wait + dma_fence_default_wait | - fence_add_callback + dma_fence_add_callback | - fence_remove_callback + dma_fence_remove_callback | - fence_enable_sw_signaling + dma_fence_enable_sw_signaling | - fence_is_signaled_locked + dma_fence_is_signaled_locked | - fence_is_signaled + dma_fence_is_signaled | - fence_is_later + dma_fence_is_later | - fence_later + dma_fence_later | - fence_wait_timeout + dma_fence_wait_timeout | - fence_wait_any_timeout + dma_fence_wait_any_timeout | - fence_wait + dma_fence_wait | - fence_context_alloc + dma_fence_context_alloc | - fence_array_create + dma_fence_array_create | - to_fence_array + to_dma_fence_array | - fence_is_array + dma_fence_is_array | - trace_fence_emit + trace_dma_fence_emit | - FENCE_TRACE + DMA_FENCE_TRACE | - FENCE_WARN + DMA_FENCE_WARN | - FENCE_ERR + DMA_FENCE_ERR ) ( ... ) Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Reviewed-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk> Acked-by: Sumit Semwal <sumit.semwal@linaro.org> Acked-by: Christian König <christian.koenig@amd.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: http://patchwork.freedesktop.org/patch/msgid/20161025120045.28839-1-chris@chris-wilson.co.uk
2016-10-25 20:00:45 +08:00
#include <linux/dma-fence.h>
#include <linux/dma-fence-unwrap.h>
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
#include <linux/anon_inodes.h>
#include <linux/export.h>
#include <linux/debugfs.h>
#include <linux/module.h>
#include <linux/seq_file.h>
dma-buf: Add an API for exporting sync files (v14) Modern userspace APIs like Vulkan are built on an explicit synchronization model. This doesn't always play nicely with the implicit synchronization used in the kernel and assumed by X11 and Wayland. The client -> compositor half of the synchronization isn't too bad, at least on intel, because we can control whether or not i915 synchronizes on the buffer and whether or not it's considered written. The harder part is the compositor -> client synchronization when we get the buffer back from the compositor. We're required to be able to provide the client with a VkSemaphore and VkFence representing the point in time where the window system (compositor and/or display) finished using the buffer. With current APIs, it's very hard to do this in such a way that we don't get confused by the Vulkan driver's access of the buffer. In particular, once we tell the kernel that we're rendering to the buffer again, any CPU waits on the buffer or GPU dependencies will wait on some of the client rendering and not just the compositor. This new IOCTL solves this problem by allowing us to get a snapshot of the implicit synchronization state of a given dma-buf in the form of a sync file. It's effectively the same as a poll() or I915_GEM_WAIT only, instead of CPU waiting directly, it encapsulates the wait operation, at the current moment in time, in a sync_file so we can check/wait on it later. As long as the Vulkan driver does the sync_file export from the dma-buf before we re-introduce it for rendering, it will only contain fences from the compositor or display. This allows to accurately turn it into a VkFence or VkSemaphore without any over-synchronization. By making this an ioctl on the dma-buf itself, it allows this new functionality to be used in an entirely driver-agnostic way without having access to a DRM fd. This makes it ideal for use in driver-generic code in Mesa or in a client such as a compositor where the DRM fd may be hard to reach. v2 (Jason Ekstrand): - Use a wrapper dma_fence_array of all fences including the new one when importing an exclusive fence. v3 (Jason Ekstrand): - Lock around setting shared fences as well as exclusive - Mark SIGNAL_SYNC_FILE as a read-write ioctl. - Initialize ret to 0 in dma_buf_wait_sync_file v4 (Jason Ekstrand): - Use the new dma_resv_get_singleton helper v5 (Jason Ekstrand): - Rename the IOCTLs to import/export rather than wait/signal - Drop the WRITE flag and always get/set the exclusive fence v6 (Jason Ekstrand): - Drop the sync_file import as it was all-around sketchy and not nearly as useful as import. - Re-introduce READ/WRITE flag support for export - Rework the commit message v7 (Jason Ekstrand): - Require at least one sync flag - Fix a refcounting bug: dma_resv_get_excl() doesn't take a reference - Use _rcu helpers since we're accessing the dma_resv read-only v8 (Jason Ekstrand): - Return -ENOMEM if the sync_file_create fails - Predicate support on IS_ENABLED(CONFIG_SYNC_FILE) v9 (Jason Ekstrand): - Add documentation for the new ioctl v10 (Jason Ekstrand): - Go back to dma_buf_sync_file as the ioctl struct name v11 (Daniel Vetter): - Go back to dma_buf_export_sync_file as the ioctl struct name - Better kerneldoc describing what the read/write flags do v12 (Christian König): - Document why we chose to make it an ioctl on dma-buf v13 (Jason Ekstrand): - Rebase on Christian König's fence rework v14 (Daniel Vetter & Christian König): - Use dma_rev_usage_rw to get the properly inverted usage to pass to dma_resv_get_singleton() - Clean up the sync_file and fd if copy_to_user() fails Signed-off-by: Jason Ekstrand <jason@jlekstrand.net> Signed-off-by: Jason Ekstrand <jason.ekstrand@intel.com> Signed-off-by: Jason Ekstrand <jason.ekstrand@collabora.com> Acked-by: Simon Ser <contact@emersion.fr> Reviewed-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: Sumit Semwal <sumit.semwal@linaro.org> Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Signed-off-by: Simon Ser <contact@emersion.fr> Link: https://patchwork.freedesktop.org/patch/msgid/20220608152142.14495-2-jason@jlekstrand.net
2022-06-08 23:21:41 +08:00
#include <linux/sync_file.h>
#include <linux/poll.h>
#include <linux/dma-resv.h>
#include <linux/mm.h>
#include <linux/mount.h>
#include <linux/pseudo_fs.h>
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
dma-buf: Add ioctls to allow userspace to flush The userspace might need some sort of cache coherency management e.g. when CPU and GPU domains are being accessed through dma-buf at the same time. To circumvent this problem there are begin/end coherency markers, that forward directly to existing dma-buf device drivers vfunc hooks. Userspace can make use of those markers through the DMA_BUF_IOCTL_SYNC ioctl. The sequence would be used like following: - mmap dma-buf fd - for each drawing/upload cycle in CPU 1. SYNC_START ioctl, 2. read/write to mmap area 3. SYNC_END ioctl. This can be repeated as often as you want (with the new data being consumed by the GPU or say scanout device) - munmap once you don't need the buffer any more v2 (Tiago): Fix header file type names (u64 -> __u64) v3 (Tiago): Add documentation. Use enum dma_buf_sync_flags to the begin/end dma-buf functions. Check for overflows in start/length. v4 (Tiago): use 2d regions for sync. v5 (Tiago): forget about 2d regions (v4); use _IOW in DMA_BUF_IOCTL_SYNC and remove range information from struct dma_buf_sync. v6 (Tiago): use __u64 structured padded flags instead enum. Adjust documentation about the recommendation on using sync ioctls. v7 (Tiago): Alex' nit on flags definition and being even more wording in the doc about sync usage. v9 (Tiago): remove useless is_dma_buf_file check. Fix sync.flags conditionals and its mask order check. Add <linux/types.h> include in dma-buf.h. Cc: Ville Syrjälä <ville.syrjala@linux.intel.com> Cc: David Herrmann <dh.herrmann@gmail.com> Cc: Sumit Semwal <sumit.semwal@linaro.org> Reviewed-by: Stéphane Marchesin <marcheu@chromium.org> Signed-off-by: Daniel Vetter <daniel.vetter@intel.com> Signed-off-by: Tiago Vignatti <tiago.vignatti@intel.com> Reviewed-by: David Herrmann <dh.herrmann@gmail.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: http://patchwork.freedesktop.org/patch/msgid/1455228291-29640-1-git-send-email-tiago.vignatti@intel.com
2016-02-12 06:04:51 +08:00
#include <uapi/linux/dma-buf.h>
#include <uapi/linux/magic.h>
dma-buf: Add ioctls to allow userspace to flush The userspace might need some sort of cache coherency management e.g. when CPU and GPU domains are being accessed through dma-buf at the same time. To circumvent this problem there are begin/end coherency markers, that forward directly to existing dma-buf device drivers vfunc hooks. Userspace can make use of those markers through the DMA_BUF_IOCTL_SYNC ioctl. The sequence would be used like following: - mmap dma-buf fd - for each drawing/upload cycle in CPU 1. SYNC_START ioctl, 2. read/write to mmap area 3. SYNC_END ioctl. This can be repeated as often as you want (with the new data being consumed by the GPU or say scanout device) - munmap once you don't need the buffer any more v2 (Tiago): Fix header file type names (u64 -> __u64) v3 (Tiago): Add documentation. Use enum dma_buf_sync_flags to the begin/end dma-buf functions. Check for overflows in start/length. v4 (Tiago): use 2d regions for sync. v5 (Tiago): forget about 2d regions (v4); use _IOW in DMA_BUF_IOCTL_SYNC and remove range information from struct dma_buf_sync. v6 (Tiago): use __u64 structured padded flags instead enum. Adjust documentation about the recommendation on using sync ioctls. v7 (Tiago): Alex' nit on flags definition and being even more wording in the doc about sync usage. v9 (Tiago): remove useless is_dma_buf_file check. Fix sync.flags conditionals and its mask order check. Add <linux/types.h> include in dma-buf.h. Cc: Ville Syrjälä <ville.syrjala@linux.intel.com> Cc: David Herrmann <dh.herrmann@gmail.com> Cc: Sumit Semwal <sumit.semwal@linaro.org> Reviewed-by: Stéphane Marchesin <marcheu@chromium.org> Signed-off-by: Daniel Vetter <daniel.vetter@intel.com> Signed-off-by: Tiago Vignatti <tiago.vignatti@intel.com> Reviewed-by: David Herrmann <dh.herrmann@gmail.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: http://patchwork.freedesktop.org/patch/msgid/1455228291-29640-1-git-send-email-tiago.vignatti@intel.com
2016-02-12 06:04:51 +08:00
dmabuf: Add the capability to expose DMA-BUF stats in sysfs Overview ======== The patch adds DMA-BUF statistics to /sys/kernel/dmabuf/buffers. It allows statistics to be enabled for each DMA-BUF in sysfs by enabling the config CONFIG_DMABUF_SYSFS_STATS. The following stats will be exposed by the interface: /sys/kernel/dmabuf/buffers/<inode_number>/exporter_name /sys/kernel/dmabuf/buffers/<inode_number>/size /sys/kernel/dmabuf/buffers/<inode_number>/attachments/<attach_uid>/device /sys/kernel/dmabuf/buffers/<inode_number>/attachments/<attach_uid>/map_counter The inode_number is unique for each DMA-BUF and was added earlier [1] in order to allow userspace to track DMA-BUF usage across different processes. Use Cases ========= The interface provides a way to gather DMA-BUF per-buffer statistics from production devices. These statistics will be used to derive DMA-BUF per-exporter stats and per-device usage stats for Android Bug reports. The corresponding userspace changes can be found at [2]. Telemetry tools will also capture this information(along with other memory metrics) periodically as well as on important events like a foreground app kill (which might have been triggered by Low Memory Killer). It will also contribute to provide a snapshot of the system memory usage on other events such as OOM kills and Application Not Responding events. Background ========== Currently, there are two existing interfaces that provide information about DMA-BUFs. 1) /sys/kernel/debug/dma_buf/bufinfo debugfs is however unsuitable to be mounted in production systems and cannot be considered as an alternative to the sysfs interface being proposed. 2) proc/<pid>/fdinfo/<fd> The proc/<pid>/fdinfo/<fd> files expose information about DMA-BUF fds. However, the existing procfs interfaces can only provide information about the buffers for which processes hold fds or have the buffers mmapped into their address space. Since the procfs interfaces alone cannot provide a full picture of all DMA-BUFs in the system, there is the need for an alternate interface to provide this information on production systems. The patch contains the following major improvements over v1: 1) Each attachment is represented by its own directory to allow creating a symlink to the importing device and to also provide room for future expansion. 2) The number of distinct mappings of each attachment is exposed in a separate file. 3) The per-buffer statistics are now in /sys/kernel/dmabuf/buffers inorder to make the interface expandable in future. All of the improvements above are based on suggestions/feedback from Daniel Vetter and Christian König. A shell script that can be run on a classic Linux environment to read out the DMA-BUF statistics can be found at [3](suggested by John Stultz). [1]: https://lore.kernel.org/patchwork/patch/1088791/ [2]: https://android-review.googlesource.com/q/topic:%22dmabuf-sysfs%22+(status:open%20OR%20status:merged) [3]: https://android-review.googlesource.com/c/platform/system/memory/libmeminfo/+/1549734 Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Hridya Valsaraju <hridya@google.com> Reported-by: kernel test robot <lkp@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/msgid/20210603214758.2955251-1-hridya@google.com
2021-06-04 05:47:51 +08:00
#include "dma-buf-sysfs-stats.h"
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
static inline int is_dma_buf_file(struct file *);
#if IS_ENABLED(CONFIG_DEBUG_FS)
static DEFINE_MUTEX(debugfs_list_mutex);
static LIST_HEAD(debugfs_list);
static void __dma_buf_debugfs_list_add(struct dma_buf *dmabuf)
{
mutex_lock(&debugfs_list_mutex);
list_add(&dmabuf->list_node, &debugfs_list);
mutex_unlock(&debugfs_list_mutex);
}
static void __dma_buf_debugfs_list_del(struct dma_buf *dmabuf)
{
if (!dmabuf)
return;
mutex_lock(&debugfs_list_mutex);
list_del(&dmabuf->list_node);
mutex_unlock(&debugfs_list_mutex);
}
#else
static void __dma_buf_debugfs_list_add(struct dma_buf *dmabuf)
{
}
static void __dma_buf_debugfs_list_del(struct file *file)
{
}
#endif
static char *dmabuffs_dname(struct dentry *dentry, char *buffer, int buflen)
{
struct dma_buf *dmabuf;
char name[DMA_BUF_NAME_LEN];
ssize_t ret = 0;
dmabuf = dentry->d_fsdata;
dmabuf: use spinlock to access dmabuf->name There exists a sleep-while-atomic bug while accessing the dmabuf->name under mutex in the dmabuffs_dname(). This is caused from the SELinux permissions checks on a process where it tries to validate the inherited files from fork() by traversing them through iterate_fd() (which traverse files under spin_lock) and call match_file(security/selinux/hooks.c) where the permission checks happen. This audit information is logged using dump_common_audit_data() where it calls d_path() to get the file path name. If the file check happen on the dmabuf's fd, then it ends up in ->dmabuffs_dname() and use mutex to access dmabuf->name. The flow will be like below: flush_unauthorized_files() iterate_fd() spin_lock() --> Start of the atomic section. match_file() file_has_perm() avc_has_perm() avc_audit() slow_avc_audit() common_lsm_audit() dump_common_audit_data() audit_log_d_path() d_path() dmabuffs_dname() mutex_lock()--> Sleep while atomic. Call trace captured (on 4.19 kernels) is below: ___might_sleep+0x204/0x208 __might_sleep+0x50/0x88 __mutex_lock_common+0x5c/0x1068 __mutex_lock_common+0x5c/0x1068 mutex_lock_nested+0x40/0x50 dmabuffs_dname+0xa0/0x170 d_path+0x84/0x290 audit_log_d_path+0x74/0x130 common_lsm_audit+0x334/0x6e8 slow_avc_audit+0xb8/0xf8 avc_has_perm+0x154/0x218 file_has_perm+0x70/0x180 match_file+0x60/0x78 iterate_fd+0x128/0x168 selinux_bprm_committing_creds+0x178/0x248 security_bprm_committing_creds+0x30/0x48 install_exec_creds+0x1c/0x68 load_elf_binary+0x3a4/0x14e0 search_binary_handler+0xb0/0x1e0 So, use spinlock to access dmabuf->name to avoid sleep-while-atomic. Cc: <stable@vger.kernel.org> [5.3+] Signed-off-by: Charan Teja Kalla <charante@codeaurora.org> Reviewed-by: Michael J. Ruhl <michael.j.ruhl@intel.com> Acked-by: Christian König <christian.koenig@amd.com> [sumits: added comment to spinlock_t definition to avoid warning] Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Link: https://patchwork.freedesktop.org/patch/msgid/a83e7f0d-4e54-9848-4b58-e1acdbe06735@codeaurora.org
2020-06-19 19:57:19 +08:00
spin_lock(&dmabuf->name_lock);
if (dmabuf->name)
ret = strscpy(name, dmabuf->name, sizeof(name));
dmabuf: use spinlock to access dmabuf->name There exists a sleep-while-atomic bug while accessing the dmabuf->name under mutex in the dmabuffs_dname(). This is caused from the SELinux permissions checks on a process where it tries to validate the inherited files from fork() by traversing them through iterate_fd() (which traverse files under spin_lock) and call match_file(security/selinux/hooks.c) where the permission checks happen. This audit information is logged using dump_common_audit_data() where it calls d_path() to get the file path name. If the file check happen on the dmabuf's fd, then it ends up in ->dmabuffs_dname() and use mutex to access dmabuf->name. The flow will be like below: flush_unauthorized_files() iterate_fd() spin_lock() --> Start of the atomic section. match_file() file_has_perm() avc_has_perm() avc_audit() slow_avc_audit() common_lsm_audit() dump_common_audit_data() audit_log_d_path() d_path() dmabuffs_dname() mutex_lock()--> Sleep while atomic. Call trace captured (on 4.19 kernels) is below: ___might_sleep+0x204/0x208 __might_sleep+0x50/0x88 __mutex_lock_common+0x5c/0x1068 __mutex_lock_common+0x5c/0x1068 mutex_lock_nested+0x40/0x50 dmabuffs_dname+0xa0/0x170 d_path+0x84/0x290 audit_log_d_path+0x74/0x130 common_lsm_audit+0x334/0x6e8 slow_avc_audit+0xb8/0xf8 avc_has_perm+0x154/0x218 file_has_perm+0x70/0x180 match_file+0x60/0x78 iterate_fd+0x128/0x168 selinux_bprm_committing_creds+0x178/0x248 security_bprm_committing_creds+0x30/0x48 install_exec_creds+0x1c/0x68 load_elf_binary+0x3a4/0x14e0 search_binary_handler+0xb0/0x1e0 So, use spinlock to access dmabuf->name to avoid sleep-while-atomic. Cc: <stable@vger.kernel.org> [5.3+] Signed-off-by: Charan Teja Kalla <charante@codeaurora.org> Reviewed-by: Michael J. Ruhl <michael.j.ruhl@intel.com> Acked-by: Christian König <christian.koenig@amd.com> [sumits: added comment to spinlock_t definition to avoid warning] Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Link: https://patchwork.freedesktop.org/patch/msgid/a83e7f0d-4e54-9848-4b58-e1acdbe06735@codeaurora.org
2020-06-19 19:57:19 +08:00
spin_unlock(&dmabuf->name_lock);
return dynamic_dname(buffer, buflen, "/%s:%s",
dentry->d_name.name, ret > 0 ? name : "");
}
dma-buf: Move dma_buf_release() from fops to dentry_ops Charan Teja reported a 'use-after-free' in dmabuffs_dname [1], which happens if the dma_buf_release() is called while the userspace is accessing the dma_buf pseudo fs's dmabuffs_dname() in another process, and dma_buf_release() releases the dmabuf object when the last reference to the struct file goes away. I discussed with Arnd Bergmann, and he suggested that rather than tying the dma_buf_release() to the file_operations' release(), we can tie it to the dentry_operations' d_release(), which will be called when the last ref to the dentry is removed. The path exercised by __fput() calls f_op->release() first, and then calls dput, which eventually calls d_op->d_release(). In the 'normal' case, when no userspace access is happening via dma_buf pseudo fs, there should be exactly one fd, file, dentry and inode, so closing the fd will kill of everything right away. In the presented case, the dentry's d_release() will be called only when the dentry's last ref is released. Therefore, lets move dma_buf_release() from fops->release() to d_ops->d_release() Many thanks to Arnd for his FS insights :) [1]: https://lore.kernel.org/patchwork/patch/1238278/ Fixes: bb2bb9030425 ("dma-buf: add DMA_BUF_SET_NAME ioctls") Reported-by: syzbot+3643a18836bce555bff6@syzkaller.appspotmail.com Cc: <stable@vger.kernel.org> [5.3+] Cc: Arnd Bergmann <arnd@arndb.de> Reported-by: Charan Teja Reddy <charante@codeaurora.org> Reviewed-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Tested-by: Charan Teja Reddy <charante@codeaurora.org> Link: https://patchwork.freedesktop.org/patch/msgid/20200611114418.19852-1-sumit.semwal@linaro.org
2020-06-11 19:44:18 +08:00
static void dma_buf_release(struct dentry *dentry)
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
{
struct dma_buf *dmabuf;
dma-buf: Move dma_buf_release() from fops to dentry_ops Charan Teja reported a 'use-after-free' in dmabuffs_dname [1], which happens if the dma_buf_release() is called while the userspace is accessing the dma_buf pseudo fs's dmabuffs_dname() in another process, and dma_buf_release() releases the dmabuf object when the last reference to the struct file goes away. I discussed with Arnd Bergmann, and he suggested that rather than tying the dma_buf_release() to the file_operations' release(), we can tie it to the dentry_operations' d_release(), which will be called when the last ref to the dentry is removed. The path exercised by __fput() calls f_op->release() first, and then calls dput, which eventually calls d_op->d_release(). In the 'normal' case, when no userspace access is happening via dma_buf pseudo fs, there should be exactly one fd, file, dentry and inode, so closing the fd will kill of everything right away. In the presented case, the dentry's d_release() will be called only when the dentry's last ref is released. Therefore, lets move dma_buf_release() from fops->release() to d_ops->d_release() Many thanks to Arnd for his FS insights :) [1]: https://lore.kernel.org/patchwork/patch/1238278/ Fixes: bb2bb9030425 ("dma-buf: add DMA_BUF_SET_NAME ioctls") Reported-by: syzbot+3643a18836bce555bff6@syzkaller.appspotmail.com Cc: <stable@vger.kernel.org> [5.3+] Cc: Arnd Bergmann <arnd@arndb.de> Reported-by: Charan Teja Reddy <charante@codeaurora.org> Reviewed-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Tested-by: Charan Teja Reddy <charante@codeaurora.org> Link: https://patchwork.freedesktop.org/patch/msgid/20200611114418.19852-1-sumit.semwal@linaro.org
2020-06-11 19:44:18 +08:00
dmabuf = dentry->d_fsdata;
if (unlikely(!dmabuf))
return;
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
BUG_ON(dmabuf->vmapping_counter);
/*
* If you hit this BUG() it could mean:
* * There's a file reference imbalance in dma_buf_poll / dma_buf_poll_cb or somewhere else
* * dmabuf->cb_in/out.active are non-0 despite no pending fence callback
*/
BUG_ON(dmabuf->cb_in.active || dmabuf->cb_out.active);
dma_buf_stats_teardown(dmabuf);
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
dmabuf->ops->release(dmabuf);
if (dmabuf->resv == (struct dma_resv *)&dmabuf[1])
dma_resv_fini(dmabuf->resv);
WARN_ON(!list_empty(&dmabuf->attachments));
module_put(dmabuf->owner);
kfree(dmabuf->name);
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
kfree(dmabuf);
dma-buf: Move dma_buf_release() from fops to dentry_ops Charan Teja reported a 'use-after-free' in dmabuffs_dname [1], which happens if the dma_buf_release() is called while the userspace is accessing the dma_buf pseudo fs's dmabuffs_dname() in another process, and dma_buf_release() releases the dmabuf object when the last reference to the struct file goes away. I discussed with Arnd Bergmann, and he suggested that rather than tying the dma_buf_release() to the file_operations' release(), we can tie it to the dentry_operations' d_release(), which will be called when the last ref to the dentry is removed. The path exercised by __fput() calls f_op->release() first, and then calls dput, which eventually calls d_op->d_release(). In the 'normal' case, when no userspace access is happening via dma_buf pseudo fs, there should be exactly one fd, file, dentry and inode, so closing the fd will kill of everything right away. In the presented case, the dentry's d_release() will be called only when the dentry's last ref is released. Therefore, lets move dma_buf_release() from fops->release() to d_ops->d_release() Many thanks to Arnd for his FS insights :) [1]: https://lore.kernel.org/patchwork/patch/1238278/ Fixes: bb2bb9030425 ("dma-buf: add DMA_BUF_SET_NAME ioctls") Reported-by: syzbot+3643a18836bce555bff6@syzkaller.appspotmail.com Cc: <stable@vger.kernel.org> [5.3+] Cc: Arnd Bergmann <arnd@arndb.de> Reported-by: Charan Teja Reddy <charante@codeaurora.org> Reviewed-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Tested-by: Charan Teja Reddy <charante@codeaurora.org> Link: https://patchwork.freedesktop.org/patch/msgid/20200611114418.19852-1-sumit.semwal@linaro.org
2020-06-11 19:44:18 +08:00
}
dmabuf: fix use-after-free of dmabuf's file->f_inode It is observed 'use-after-free' on the dmabuf's file->f_inode with the race between closing the dmabuf file and reading the dmabuf's debug info. Consider the below scenario where P1 is closing the dma_buf file and P2 is reading the dma_buf's debug info in the system: P1 P2 dma_buf_debug_show() dma_buf_put() __fput() file->f_op->release() dput() .... dentry_unlink_inode() iput(dentry->d_inode) (where the inode is freed) mutex_lock(&db_list.lock) read 'dma_buf->file->f_inode' (the same inode is freed by P1) mutex_unlock(&db_list.lock) dentry->d_op->d_release()--> dma_buf_release() ..... mutex_lock(&db_list.lock) removes the dmabuf from the list mutex_unlock(&db_list.lock) In the above scenario, when dma_buf_put() is called on a dma_buf, it first frees the dma_buf's file->f_inode(=dentry->d_inode) and then removes this dma_buf from the system db_list. In between P2 traversing the db_list tries to access this dma_buf's file->f_inode that was freed by P1 which is a use-after-free case. Since, __fput() calls f_op->release first and then later calls the d_op->d_release, move the dma_buf's db_list removal from d_release() to f_op->release(). This ensures that dma_buf's file->f_inode is not accessed after it is released. Cc: <stable@vger.kernel.org> # 5.4.x- Fixes: 4ab59c3c638c ("dma-buf: Move dma_buf_release() from fops to dentry_ops") Acked-by: Christian König <christian.koenig@amd.com> Signed-off-by: Charan Teja Reddy <charante@codeaurora.org> Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de> Link: https://patchwork.freedesktop.org/patch/msgid/1609857399-31549-1-git-send-email-charante@codeaurora.org
2021-01-05 22:36:39 +08:00
static int dma_buf_file_release(struct inode *inode, struct file *file)
{
if (!is_dma_buf_file(file))
return -EINVAL;
__dma_buf_debugfs_list_del(file->private_data);
dmabuf: fix use-after-free of dmabuf's file->f_inode It is observed 'use-after-free' on the dmabuf's file->f_inode with the race between closing the dmabuf file and reading the dmabuf's debug info. Consider the below scenario where P1 is closing the dma_buf file and P2 is reading the dma_buf's debug info in the system: P1 P2 dma_buf_debug_show() dma_buf_put() __fput() file->f_op->release() dput() .... dentry_unlink_inode() iput(dentry->d_inode) (where the inode is freed) mutex_lock(&db_list.lock) read 'dma_buf->file->f_inode' (the same inode is freed by P1) mutex_unlock(&db_list.lock) dentry->d_op->d_release()--> dma_buf_release() ..... mutex_lock(&db_list.lock) removes the dmabuf from the list mutex_unlock(&db_list.lock) In the above scenario, when dma_buf_put() is called on a dma_buf, it first frees the dma_buf's file->f_inode(=dentry->d_inode) and then removes this dma_buf from the system db_list. In between P2 traversing the db_list tries to access this dma_buf's file->f_inode that was freed by P1 which is a use-after-free case. Since, __fput() calls f_op->release first and then later calls the d_op->d_release, move the dma_buf's db_list removal from d_release() to f_op->release(). This ensures that dma_buf's file->f_inode is not accessed after it is released. Cc: <stable@vger.kernel.org> # 5.4.x- Fixes: 4ab59c3c638c ("dma-buf: Move dma_buf_release() from fops to dentry_ops") Acked-by: Christian König <christian.koenig@amd.com> Signed-off-by: Charan Teja Reddy <charante@codeaurora.org> Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de> Link: https://patchwork.freedesktop.org/patch/msgid/1609857399-31549-1-git-send-email-charante@codeaurora.org
2021-01-05 22:36:39 +08:00
return 0;
}
dma-buf: Move dma_buf_release() from fops to dentry_ops Charan Teja reported a 'use-after-free' in dmabuffs_dname [1], which happens if the dma_buf_release() is called while the userspace is accessing the dma_buf pseudo fs's dmabuffs_dname() in another process, and dma_buf_release() releases the dmabuf object when the last reference to the struct file goes away. I discussed with Arnd Bergmann, and he suggested that rather than tying the dma_buf_release() to the file_operations' release(), we can tie it to the dentry_operations' d_release(), which will be called when the last ref to the dentry is removed. The path exercised by __fput() calls f_op->release() first, and then calls dput, which eventually calls d_op->d_release(). In the 'normal' case, when no userspace access is happening via dma_buf pseudo fs, there should be exactly one fd, file, dentry and inode, so closing the fd will kill of everything right away. In the presented case, the dentry's d_release() will be called only when the dentry's last ref is released. Therefore, lets move dma_buf_release() from fops->release() to d_ops->d_release() Many thanks to Arnd for his FS insights :) [1]: https://lore.kernel.org/patchwork/patch/1238278/ Fixes: bb2bb9030425 ("dma-buf: add DMA_BUF_SET_NAME ioctls") Reported-by: syzbot+3643a18836bce555bff6@syzkaller.appspotmail.com Cc: <stable@vger.kernel.org> [5.3+] Cc: Arnd Bergmann <arnd@arndb.de> Reported-by: Charan Teja Reddy <charante@codeaurora.org> Reviewed-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Tested-by: Charan Teja Reddy <charante@codeaurora.org> Link: https://patchwork.freedesktop.org/patch/msgid/20200611114418.19852-1-sumit.semwal@linaro.org
2020-06-11 19:44:18 +08:00
static const struct dentry_operations dma_buf_dentry_ops = {
.d_dname = dmabuffs_dname,
.d_release = dma_buf_release,
};
static struct vfsmount *dma_buf_mnt;
static int dma_buf_fs_init_context(struct fs_context *fc)
{
struct pseudo_fs_context *ctx;
ctx = init_pseudo(fc, DMA_BUF_MAGIC);
if (!ctx)
return -ENOMEM;
ctx->dops = &dma_buf_dentry_ops;
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
return 0;
}
dma-buf: Move dma_buf_release() from fops to dentry_ops Charan Teja reported a 'use-after-free' in dmabuffs_dname [1], which happens if the dma_buf_release() is called while the userspace is accessing the dma_buf pseudo fs's dmabuffs_dname() in another process, and dma_buf_release() releases the dmabuf object when the last reference to the struct file goes away. I discussed with Arnd Bergmann, and he suggested that rather than tying the dma_buf_release() to the file_operations' release(), we can tie it to the dentry_operations' d_release(), which will be called when the last ref to the dentry is removed. The path exercised by __fput() calls f_op->release() first, and then calls dput, which eventually calls d_op->d_release(). In the 'normal' case, when no userspace access is happening via dma_buf pseudo fs, there should be exactly one fd, file, dentry and inode, so closing the fd will kill of everything right away. In the presented case, the dentry's d_release() will be called only when the dentry's last ref is released. Therefore, lets move dma_buf_release() from fops->release() to d_ops->d_release() Many thanks to Arnd for his FS insights :) [1]: https://lore.kernel.org/patchwork/patch/1238278/ Fixes: bb2bb9030425 ("dma-buf: add DMA_BUF_SET_NAME ioctls") Reported-by: syzbot+3643a18836bce555bff6@syzkaller.appspotmail.com Cc: <stable@vger.kernel.org> [5.3+] Cc: Arnd Bergmann <arnd@arndb.de> Reported-by: Charan Teja Reddy <charante@codeaurora.org> Reviewed-by: Arnd Bergmann <arnd@arndb.de> Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Tested-by: Charan Teja Reddy <charante@codeaurora.org> Link: https://patchwork.freedesktop.org/patch/msgid/20200611114418.19852-1-sumit.semwal@linaro.org
2020-06-11 19:44:18 +08:00
static struct file_system_type dma_buf_fs_type = {
.name = "dmabuf",
.init_fs_context = dma_buf_fs_init_context,
.kill_sb = kill_anon_super,
};
dma-buf: mmap support Compared to Rob Clark's RFC I've ditched the prepare/finish hooks and corresponding ioctls on the dma_buf file. The major reason for that is that many people seem to be under the impression that this is also for synchronization with outstanding asynchronous processsing. I'm pretty massively opposed to this because: - It boils down reinventing a new rather general-purpose userspace synchronization interface. If we look at things like futexes, this is hard to get right. - Furthermore a lot of kernel code has to interact with this synchronization primitive. This smells a look like the dri1 hw_lock, a horror show I prefer not to reinvent. - Even more fun is that multiple different subsystems would interact here, so we have plenty of opportunities to create funny deadlock scenarios. I think synchronization is a wholesale different problem from data sharing and should be tackled as an orthogonal problem. Now we could demand that prepare/finish may only ensure cache coherency (as Rob intended), but that runs up into the next problem: We not only need mmap support to facilitate sw-only processing nodes in a pipeline (without jumping through hoops by importing the dma_buf into some sw-access only importer), which allows for a nicer ION->dma-buf upgrade path for existing Android userspace. We also need mmap support for existing importing subsystems to support existing userspace libraries. And a loot of these subsystems are expected to export coherent userspace mappings. So prepare/finish can only ever be optional and the exporter /needs/ to support coherent mappings. Given that mmap access is always somewhat fallback-y in nature I've decided to drop this optimization, instead of just making it optional. If we demonstrate a clear need for this, supported by benchmark results, we can always add it in again later as an optional extension. Other differences compared to Rob's RFC is the above mentioned support for mapping a dma-buf through facilities provided by the importer. Which results in mmap support no longer being optional. Note that this dma-buf mmap patch does _not_ support every possible insanity an existing subsystem could pull of with mmap: Because it does not allow to intercept pagefaults and shoot down ptes importing subsystems can't add some magic of their own at these points (e.g. to automatically synchronize with outstanding rendering or set up some special resources). I've done a cursory read through a few mmap implementions of various subsytems and I'm hopeful that we can avoid this (and the complexity it'd bring with it). Additonally I've extended the documentation a bit to explain the hows and whys of this mmap extension. In case we ever want to add support for explicitly cache maneged userspace mmap with a prepare/finish ioctl pair, we could specify that userspace needs to mmap a different part of the dma_buf, e.g. the range starting at dma_buf->size up to dma_buf->size*2. This works because the size of a dma_buf is invariant over it's lifetime. The exporter would obviously need to fall back to coherent mappings for both ranges if a legacy clients maps the coherent range and the architecture cannot suppor conflicting caching policies. Also, this would obviously be optional and userspace needs to be able to fall back to coherent mappings. v2: - Spelling fixes from Rob Clark. - Compile fix for !DMA_BUF from Rob Clark. - Extend commit message to explain how explicitly cache managed mmap support could be added later. - Extend the documentation with implementations notes for exporters that need to manually fake coherency. v3: - dma_buf pointer initialization goof-up noticed by Rebecca Schultz Zavin. Cc: Rob Clark <rob.clark@linaro.org> Cc: Rebecca Schultz Zavin <rebecca@android.com> Acked-by: Rob Clark <rob.clark@linaro.org> Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org>
2012-04-24 17:08:52 +08:00
static int dma_buf_mmap_internal(struct file *file, struct vm_area_struct *vma)
{
struct dma_buf *dmabuf;
if (!is_dma_buf_file(file))
return -EINVAL;
dmabuf = file->private_data;
/* check if buffer supports mmap */
if (!dmabuf->ops->mmap)
return -EINVAL;
dma-buf: mmap support Compared to Rob Clark's RFC I've ditched the prepare/finish hooks and corresponding ioctls on the dma_buf file. The major reason for that is that many people seem to be under the impression that this is also for synchronization with outstanding asynchronous processsing. I'm pretty massively opposed to this because: - It boils down reinventing a new rather general-purpose userspace synchronization interface. If we look at things like futexes, this is hard to get right. - Furthermore a lot of kernel code has to interact with this synchronization primitive. This smells a look like the dri1 hw_lock, a horror show I prefer not to reinvent. - Even more fun is that multiple different subsystems would interact here, so we have plenty of opportunities to create funny deadlock scenarios. I think synchronization is a wholesale different problem from data sharing and should be tackled as an orthogonal problem. Now we could demand that prepare/finish may only ensure cache coherency (as Rob intended), but that runs up into the next problem: We not only need mmap support to facilitate sw-only processing nodes in a pipeline (without jumping through hoops by importing the dma_buf into some sw-access only importer), which allows for a nicer ION->dma-buf upgrade path for existing Android userspace. We also need mmap support for existing importing subsystems to support existing userspace libraries. And a loot of these subsystems are expected to export coherent userspace mappings. So prepare/finish can only ever be optional and the exporter /needs/ to support coherent mappings. Given that mmap access is always somewhat fallback-y in nature I've decided to drop this optimization, instead of just making it optional. If we demonstrate a clear need for this, supported by benchmark results, we can always add it in again later as an optional extension. Other differences compared to Rob's RFC is the above mentioned support for mapping a dma-buf through facilities provided by the importer. Which results in mmap support no longer being optional. Note that this dma-buf mmap patch does _not_ support every possible insanity an existing subsystem could pull of with mmap: Because it does not allow to intercept pagefaults and shoot down ptes importing subsystems can't add some magic of their own at these points (e.g. to automatically synchronize with outstanding rendering or set up some special resources). I've done a cursory read through a few mmap implementions of various subsytems and I'm hopeful that we can avoid this (and the complexity it'd bring with it). Additonally I've extended the documentation a bit to explain the hows and whys of this mmap extension. In case we ever want to add support for explicitly cache maneged userspace mmap with a prepare/finish ioctl pair, we could specify that userspace needs to mmap a different part of the dma_buf, e.g. the range starting at dma_buf->size up to dma_buf->size*2. This works because the size of a dma_buf is invariant over it's lifetime. The exporter would obviously need to fall back to coherent mappings for both ranges if a legacy clients maps the coherent range and the architecture cannot suppor conflicting caching policies. Also, this would obviously be optional and userspace needs to be able to fall back to coherent mappings. v2: - Spelling fixes from Rob Clark. - Compile fix for !DMA_BUF from Rob Clark. - Extend commit message to explain how explicitly cache managed mmap support could be added later. - Extend the documentation with implementations notes for exporters that need to manually fake coherency. v3: - dma_buf pointer initialization goof-up noticed by Rebecca Schultz Zavin. Cc: Rob Clark <rob.clark@linaro.org> Cc: Rebecca Schultz Zavin <rebecca@android.com> Acked-by: Rob Clark <rob.clark@linaro.org> Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org>
2012-04-24 17:08:52 +08:00
/* check for overflowing the buffer's size */
if (vma->vm_pgoff + vma_pages(vma) >
dma-buf: mmap support Compared to Rob Clark's RFC I've ditched the prepare/finish hooks and corresponding ioctls on the dma_buf file. The major reason for that is that many people seem to be under the impression that this is also for synchronization with outstanding asynchronous processsing. I'm pretty massively opposed to this because: - It boils down reinventing a new rather general-purpose userspace synchronization interface. If we look at things like futexes, this is hard to get right. - Furthermore a lot of kernel code has to interact with this synchronization primitive. This smells a look like the dri1 hw_lock, a horror show I prefer not to reinvent. - Even more fun is that multiple different subsystems would interact here, so we have plenty of opportunities to create funny deadlock scenarios. I think synchronization is a wholesale different problem from data sharing and should be tackled as an orthogonal problem. Now we could demand that prepare/finish may only ensure cache coherency (as Rob intended), but that runs up into the next problem: We not only need mmap support to facilitate sw-only processing nodes in a pipeline (without jumping through hoops by importing the dma_buf into some sw-access only importer), which allows for a nicer ION->dma-buf upgrade path for existing Android userspace. We also need mmap support for existing importing subsystems to support existing userspace libraries. And a loot of these subsystems are expected to export coherent userspace mappings. So prepare/finish can only ever be optional and the exporter /needs/ to support coherent mappings. Given that mmap access is always somewhat fallback-y in nature I've decided to drop this optimization, instead of just making it optional. If we demonstrate a clear need for this, supported by benchmark results, we can always add it in again later as an optional extension. Other differences compared to Rob's RFC is the above mentioned support for mapping a dma-buf through facilities provided by the importer. Which results in mmap support no longer being optional. Note that this dma-buf mmap patch does _not_ support every possible insanity an existing subsystem could pull of with mmap: Because it does not allow to intercept pagefaults and shoot down ptes importing subsystems can't add some magic of their own at these points (e.g. to automatically synchronize with outstanding rendering or set up some special resources). I've done a cursory read through a few mmap implementions of various subsytems and I'm hopeful that we can avoid this (and the complexity it'd bring with it). Additonally I've extended the documentation a bit to explain the hows and whys of this mmap extension. In case we ever want to add support for explicitly cache maneged userspace mmap with a prepare/finish ioctl pair, we could specify that userspace needs to mmap a different part of the dma_buf, e.g. the range starting at dma_buf->size up to dma_buf->size*2. This works because the size of a dma_buf is invariant over it's lifetime. The exporter would obviously need to fall back to coherent mappings for both ranges if a legacy clients maps the coherent range and the architecture cannot suppor conflicting caching policies. Also, this would obviously be optional and userspace needs to be able to fall back to coherent mappings. v2: - Spelling fixes from Rob Clark. - Compile fix for !DMA_BUF from Rob Clark. - Extend commit message to explain how explicitly cache managed mmap support could be added later. - Extend the documentation with implementations notes for exporters that need to manually fake coherency. v3: - dma_buf pointer initialization goof-up noticed by Rebecca Schultz Zavin. Cc: Rob Clark <rob.clark@linaro.org> Cc: Rebecca Schultz Zavin <rebecca@android.com> Acked-by: Rob Clark <rob.clark@linaro.org> Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org>
2012-04-24 17:08:52 +08:00
dmabuf->size >> PAGE_SHIFT)
return -EINVAL;
return dmabuf->ops->mmap(dmabuf, vma);
dma-buf: mmap support Compared to Rob Clark's RFC I've ditched the prepare/finish hooks and corresponding ioctls on the dma_buf file. The major reason for that is that many people seem to be under the impression that this is also for synchronization with outstanding asynchronous processsing. I'm pretty massively opposed to this because: - It boils down reinventing a new rather general-purpose userspace synchronization interface. If we look at things like futexes, this is hard to get right. - Furthermore a lot of kernel code has to interact with this synchronization primitive. This smells a look like the dri1 hw_lock, a horror show I prefer not to reinvent. - Even more fun is that multiple different subsystems would interact here, so we have plenty of opportunities to create funny deadlock scenarios. I think synchronization is a wholesale different problem from data sharing and should be tackled as an orthogonal problem. Now we could demand that prepare/finish may only ensure cache coherency (as Rob intended), but that runs up into the next problem: We not only need mmap support to facilitate sw-only processing nodes in a pipeline (without jumping through hoops by importing the dma_buf into some sw-access only importer), which allows for a nicer ION->dma-buf upgrade path for existing Android userspace. We also need mmap support for existing importing subsystems to support existing userspace libraries. And a loot of these subsystems are expected to export coherent userspace mappings. So prepare/finish can only ever be optional and the exporter /needs/ to support coherent mappings. Given that mmap access is always somewhat fallback-y in nature I've decided to drop this optimization, instead of just making it optional. If we demonstrate a clear need for this, supported by benchmark results, we can always add it in again later as an optional extension. Other differences compared to Rob's RFC is the above mentioned support for mapping a dma-buf through facilities provided by the importer. Which results in mmap support no longer being optional. Note that this dma-buf mmap patch does _not_ support every possible insanity an existing subsystem could pull of with mmap: Because it does not allow to intercept pagefaults and shoot down ptes importing subsystems can't add some magic of their own at these points (e.g. to automatically synchronize with outstanding rendering or set up some special resources). I've done a cursory read through a few mmap implementions of various subsytems and I'm hopeful that we can avoid this (and the complexity it'd bring with it). Additonally I've extended the documentation a bit to explain the hows and whys of this mmap extension. In case we ever want to add support for explicitly cache maneged userspace mmap with a prepare/finish ioctl pair, we could specify that userspace needs to mmap a different part of the dma_buf, e.g. the range starting at dma_buf->size up to dma_buf->size*2. This works because the size of a dma_buf is invariant over it's lifetime. The exporter would obviously need to fall back to coherent mappings for both ranges if a legacy clients maps the coherent range and the architecture cannot suppor conflicting caching policies. Also, this would obviously be optional and userspace needs to be able to fall back to coherent mappings. v2: - Spelling fixes from Rob Clark. - Compile fix for !DMA_BUF from Rob Clark. - Extend commit message to explain how explicitly cache managed mmap support could be added later. - Extend the documentation with implementations notes for exporters that need to manually fake coherency. v3: - dma_buf pointer initialization goof-up noticed by Rebecca Schultz Zavin. Cc: Rob Clark <rob.clark@linaro.org> Cc: Rebecca Schultz Zavin <rebecca@android.com> Acked-by: Rob Clark <rob.clark@linaro.org> Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org>
2012-04-24 17:08:52 +08:00
}
static loff_t dma_buf_llseek(struct file *file, loff_t offset, int whence)
{
struct dma_buf *dmabuf;
loff_t base;
if (!is_dma_buf_file(file))
return -EBADF;
dmabuf = file->private_data;
/* only support discovering the end of the buffer,
but also allow SEEK_SET to maintain the idiomatic
SEEK_END(0), SEEK_CUR(0) pattern */
if (whence == SEEK_END)
base = dmabuf->size;
else if (whence == SEEK_SET)
base = 0;
else
return -EINVAL;
if (offset != 0)
return -EINVAL;
return base + offset;
}
/**
* DOC: implicit fence polling
*
* To support cross-device and cross-driver synchronization of buffer access
* implicit fences (represented internally in the kernel with &struct dma_fence)
* can be attached to a &dma_buf. The glue for that and a few related things are
* provided in the &dma_resv structure.
*
* Userspace can query the state of these implicitly tracked fences using poll()
* and related system calls:
*
* - Checking for EPOLLIN, i.e. read access, can be use to query the state of the
* most recent write or exclusive fence.
*
* - Checking for EPOLLOUT, i.e. write access, can be used to query the state of
* all attached fences, shared and exclusive ones.
*
* Note that this only signals the completion of the respective fences, i.e. the
* DMA transfers are complete. Cache flushing and any other necessary
* preparations before CPU access can begin still need to happen.
dma-buf: Add an API for exporting sync files (v14) Modern userspace APIs like Vulkan are built on an explicit synchronization model. This doesn't always play nicely with the implicit synchronization used in the kernel and assumed by X11 and Wayland. The client -> compositor half of the synchronization isn't too bad, at least on intel, because we can control whether or not i915 synchronizes on the buffer and whether or not it's considered written. The harder part is the compositor -> client synchronization when we get the buffer back from the compositor. We're required to be able to provide the client with a VkSemaphore and VkFence representing the point in time where the window system (compositor and/or display) finished using the buffer. With current APIs, it's very hard to do this in such a way that we don't get confused by the Vulkan driver's access of the buffer. In particular, once we tell the kernel that we're rendering to the buffer again, any CPU waits on the buffer or GPU dependencies will wait on some of the client rendering and not just the compositor. This new IOCTL solves this problem by allowing us to get a snapshot of the implicit synchronization state of a given dma-buf in the form of a sync file. It's effectively the same as a poll() or I915_GEM_WAIT only, instead of CPU waiting directly, it encapsulates the wait operation, at the current moment in time, in a sync_file so we can check/wait on it later. As long as the Vulkan driver does the sync_file export from the dma-buf before we re-introduce it for rendering, it will only contain fences from the compositor or display. This allows to accurately turn it into a VkFence or VkSemaphore without any over-synchronization. By making this an ioctl on the dma-buf itself, it allows this new functionality to be used in an entirely driver-agnostic way without having access to a DRM fd. This makes it ideal for use in driver-generic code in Mesa or in a client such as a compositor where the DRM fd may be hard to reach. v2 (Jason Ekstrand): - Use a wrapper dma_fence_array of all fences including the new one when importing an exclusive fence. v3 (Jason Ekstrand): - Lock around setting shared fences as well as exclusive - Mark SIGNAL_SYNC_FILE as a read-write ioctl. - Initialize ret to 0 in dma_buf_wait_sync_file v4 (Jason Ekstrand): - Use the new dma_resv_get_singleton helper v5 (Jason Ekstrand): - Rename the IOCTLs to import/export rather than wait/signal - Drop the WRITE flag and always get/set the exclusive fence v6 (Jason Ekstrand): - Drop the sync_file import as it was all-around sketchy and not nearly as useful as import. - Re-introduce READ/WRITE flag support for export - Rework the commit message v7 (Jason Ekstrand): - Require at least one sync flag - Fix a refcounting bug: dma_resv_get_excl() doesn't take a reference - Use _rcu helpers since we're accessing the dma_resv read-only v8 (Jason Ekstrand): - Return -ENOMEM if the sync_file_create fails - Predicate support on IS_ENABLED(CONFIG_SYNC_FILE) v9 (Jason Ekstrand): - Add documentation for the new ioctl v10 (Jason Ekstrand): - Go back to dma_buf_sync_file as the ioctl struct name v11 (Daniel Vetter): - Go back to dma_buf_export_sync_file as the ioctl struct name - Better kerneldoc describing what the read/write flags do v12 (Christian König): - Document why we chose to make it an ioctl on dma-buf v13 (Jason Ekstrand): - Rebase on Christian König's fence rework v14 (Daniel Vetter & Christian König): - Use dma_rev_usage_rw to get the properly inverted usage to pass to dma_resv_get_singleton() - Clean up the sync_file and fd if copy_to_user() fails Signed-off-by: Jason Ekstrand <jason@jlekstrand.net> Signed-off-by: Jason Ekstrand <jason.ekstrand@intel.com> Signed-off-by: Jason Ekstrand <jason.ekstrand@collabora.com> Acked-by: Simon Ser <contact@emersion.fr> Reviewed-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: Sumit Semwal <sumit.semwal@linaro.org> Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Signed-off-by: Simon Ser <contact@emersion.fr> Link: https://patchwork.freedesktop.org/patch/msgid/20220608152142.14495-2-jason@jlekstrand.net
2022-06-08 23:21:41 +08:00
*
* As an alternative to poll(), the set of fences on DMA buffer can be
* exported as a &sync_file using &dma_buf_sync_file_export.
*/
dma-buf: Rename struct fence to dma_fence I plan to usurp the short name of struct fence for a core kernel struct, and so I need to rename the specialised fence/timeline for DMA operations to make room. A consensus was reached in https://lists.freedesktop.org/archives/dri-devel/2016-July/113083.html that making clear this fence applies to DMA operations was a good thing. Since then the patch has grown a bit as usage increases, so hopefully it remains a good thing! (v2...: rebase, rerun spatch) v3: Compile on msm, spotted a manual fixup that I broke. v4: Try again for msm, sorry Daniel coccinelle script: @@ @@ - struct fence + struct dma_fence @@ @@ - struct fence_ops + struct dma_fence_ops @@ @@ - struct fence_cb + struct dma_fence_cb @@ @@ - struct fence_array + struct dma_fence_array @@ @@ - enum fence_flag_bits + enum dma_fence_flag_bits @@ @@ ( - fence_init + dma_fence_init | - fence_release + dma_fence_release | - fence_free + dma_fence_free | - fence_get + dma_fence_get | - fence_get_rcu + dma_fence_get_rcu | - fence_put + dma_fence_put | - fence_signal + dma_fence_signal | - fence_signal_locked + dma_fence_signal_locked | - fence_default_wait + dma_fence_default_wait | - fence_add_callback + dma_fence_add_callback | - fence_remove_callback + dma_fence_remove_callback | - fence_enable_sw_signaling + dma_fence_enable_sw_signaling | - fence_is_signaled_locked + dma_fence_is_signaled_locked | - fence_is_signaled + dma_fence_is_signaled | - fence_is_later + dma_fence_is_later | - fence_later + dma_fence_later | - fence_wait_timeout + dma_fence_wait_timeout | - fence_wait_any_timeout + dma_fence_wait_any_timeout | - fence_wait + dma_fence_wait | - fence_context_alloc + dma_fence_context_alloc | - fence_array_create + dma_fence_array_create | - to_fence_array + to_dma_fence_array | - fence_is_array + dma_fence_is_array | - trace_fence_emit + trace_dma_fence_emit | - FENCE_TRACE + DMA_FENCE_TRACE | - FENCE_WARN + DMA_FENCE_WARN | - FENCE_ERR + DMA_FENCE_ERR ) ( ... ) Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Reviewed-by: Gustavo Padovan <gustavo.padovan@collabora.co.uk> Acked-by: Sumit Semwal <sumit.semwal@linaro.org> Acked-by: Christian König <christian.koenig@amd.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: http://patchwork.freedesktop.org/patch/msgid/20161025120045.28839-1-chris@chris-wilson.co.uk
2016-10-25 20:00:45 +08:00
static void dma_buf_poll_cb(struct dma_fence *fence, struct dma_fence_cb *cb)
{
struct dma_buf_poll_cb_t *dcb = (struct dma_buf_poll_cb_t *)cb;
struct dma_buf *dmabuf = container_of(dcb->poll, struct dma_buf, poll);
unsigned long flags;
spin_lock_irqsave(&dcb->poll->lock, flags);
wake_up_locked_poll(dcb->poll, dcb->active);
dcb->active = 0;
spin_unlock_irqrestore(&dcb->poll->lock, flags);
dma_fence_put(fence);
/* Paired with get_file in dma_buf_poll */
fput(dmabuf->file);
}
static bool dma_buf_poll_add_cb(struct dma_resv *resv, bool write,
struct dma_buf_poll_cb_t *dcb)
{
struct dma_resv_iter cursor;
struct dma_fence *fence;
int r;
dma_resv_for_each_fence(&cursor, resv, dma_resv_usage_rw(write),
fence) {
dma_fence_get(fence);
r = dma_fence_add_callback(fence, &dcb->cb, dma_buf_poll_cb);
if (!r)
return true;
dma_fence_put(fence);
}
return false;
}
static __poll_t dma_buf_poll(struct file *file, poll_table *poll)
{
struct dma_buf *dmabuf;
struct dma_resv *resv;
__poll_t events;
dmabuf = file->private_data;
if (!dmabuf || !dmabuf->resv)
return EPOLLERR;
resv = dmabuf->resv;
poll_wait(file, &dmabuf->poll, poll);
events = poll_requested_events(poll) & (EPOLLIN | EPOLLOUT);
if (!events)
return 0;
dma_resv_lock(resv, NULL);
if (events & EPOLLOUT) {
struct dma_buf_poll_cb_t *dcb = &dmabuf->cb_out;
/* Check that callback isn't busy */
spin_lock_irq(&dmabuf->poll.lock);
if (dcb->active)
events &= ~EPOLLOUT;
else
dcb->active = EPOLLOUT;
spin_unlock_irq(&dmabuf->poll.lock);
if (events & EPOLLOUT) {
/* Paired with fput in dma_buf_poll_cb */
get_file(dmabuf->file);
if (!dma_buf_poll_add_cb(resv, true, dcb))
/* No callback queued, wake up any other waiters */
dma_buf_poll_cb(NULL, &dcb->cb);
else
events &= ~EPOLLOUT;
}
}
if (events & EPOLLIN) {
struct dma_buf_poll_cb_t *dcb = &dmabuf->cb_in;
/* Check that callback isn't busy */
spin_lock_irq(&dmabuf->poll.lock);
if (dcb->active)
events &= ~EPOLLIN;
else
dcb->active = EPOLLIN;
spin_unlock_irq(&dmabuf->poll.lock);
if (events & EPOLLIN) {
/* Paired with fput in dma_buf_poll_cb */
get_file(dmabuf->file);
if (!dma_buf_poll_add_cb(resv, false, dcb))
/* No callback queued, wake up any other waiters */
dma_buf_poll_cb(NULL, &dcb->cb);
else
events &= ~EPOLLIN;
}
}
dma_resv_unlock(resv);
return events;
}
/**
* dma_buf_set_name - Set a name to a specific dma_buf to track the usage.
* It could support changing the name of the dma-buf if the same
* piece of memory is used for multiple purpose between different devices.
*
* @dmabuf: [in] dmabuf buffer that will be renamed.
* @buf: [in] A piece of userspace memory that contains the name of
* the dma-buf.
*
* Returns 0 on success. If the dma-buf buffer is already attached to
* devices, return -EBUSY.
*
*/
static long dma_buf_set_name(struct dma_buf *dmabuf, const char __user *buf)
{
char *name = strndup_user(buf, DMA_BUF_NAME_LEN);
if (IS_ERR(name))
return PTR_ERR(name);
dmabuf: use spinlock to access dmabuf->name There exists a sleep-while-atomic bug while accessing the dmabuf->name under mutex in the dmabuffs_dname(). This is caused from the SELinux permissions checks on a process where it tries to validate the inherited files from fork() by traversing them through iterate_fd() (which traverse files under spin_lock) and call match_file(security/selinux/hooks.c) where the permission checks happen. This audit information is logged using dump_common_audit_data() where it calls d_path() to get the file path name. If the file check happen on the dmabuf's fd, then it ends up in ->dmabuffs_dname() and use mutex to access dmabuf->name. The flow will be like below: flush_unauthorized_files() iterate_fd() spin_lock() --> Start of the atomic section. match_file() file_has_perm() avc_has_perm() avc_audit() slow_avc_audit() common_lsm_audit() dump_common_audit_data() audit_log_d_path() d_path() dmabuffs_dname() mutex_lock()--> Sleep while atomic. Call trace captured (on 4.19 kernels) is below: ___might_sleep+0x204/0x208 __might_sleep+0x50/0x88 __mutex_lock_common+0x5c/0x1068 __mutex_lock_common+0x5c/0x1068 mutex_lock_nested+0x40/0x50 dmabuffs_dname+0xa0/0x170 d_path+0x84/0x290 audit_log_d_path+0x74/0x130 common_lsm_audit+0x334/0x6e8 slow_avc_audit+0xb8/0xf8 avc_has_perm+0x154/0x218 file_has_perm+0x70/0x180 match_file+0x60/0x78 iterate_fd+0x128/0x168 selinux_bprm_committing_creds+0x178/0x248 security_bprm_committing_creds+0x30/0x48 install_exec_creds+0x1c/0x68 load_elf_binary+0x3a4/0x14e0 search_binary_handler+0xb0/0x1e0 So, use spinlock to access dmabuf->name to avoid sleep-while-atomic. Cc: <stable@vger.kernel.org> [5.3+] Signed-off-by: Charan Teja Kalla <charante@codeaurora.org> Reviewed-by: Michael J. Ruhl <michael.j.ruhl@intel.com> Acked-by: Christian König <christian.koenig@amd.com> [sumits: added comment to spinlock_t definition to avoid warning] Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Link: https://patchwork.freedesktop.org/patch/msgid/a83e7f0d-4e54-9848-4b58-e1acdbe06735@codeaurora.org
2020-06-19 19:57:19 +08:00
spin_lock(&dmabuf->name_lock);
kfree(dmabuf->name);
dmabuf->name = name;
dmabuf: use spinlock to access dmabuf->name There exists a sleep-while-atomic bug while accessing the dmabuf->name under mutex in the dmabuffs_dname(). This is caused from the SELinux permissions checks on a process where it tries to validate the inherited files from fork() by traversing them through iterate_fd() (which traverse files under spin_lock) and call match_file(security/selinux/hooks.c) where the permission checks happen. This audit information is logged using dump_common_audit_data() where it calls d_path() to get the file path name. If the file check happen on the dmabuf's fd, then it ends up in ->dmabuffs_dname() and use mutex to access dmabuf->name. The flow will be like below: flush_unauthorized_files() iterate_fd() spin_lock() --> Start of the atomic section. match_file() file_has_perm() avc_has_perm() avc_audit() slow_avc_audit() common_lsm_audit() dump_common_audit_data() audit_log_d_path() d_path() dmabuffs_dname() mutex_lock()--> Sleep while atomic. Call trace captured (on 4.19 kernels) is below: ___might_sleep+0x204/0x208 __might_sleep+0x50/0x88 __mutex_lock_common+0x5c/0x1068 __mutex_lock_common+0x5c/0x1068 mutex_lock_nested+0x40/0x50 dmabuffs_dname+0xa0/0x170 d_path+0x84/0x290 audit_log_d_path+0x74/0x130 common_lsm_audit+0x334/0x6e8 slow_avc_audit+0xb8/0xf8 avc_has_perm+0x154/0x218 file_has_perm+0x70/0x180 match_file+0x60/0x78 iterate_fd+0x128/0x168 selinux_bprm_committing_creds+0x178/0x248 security_bprm_committing_creds+0x30/0x48 install_exec_creds+0x1c/0x68 load_elf_binary+0x3a4/0x14e0 search_binary_handler+0xb0/0x1e0 So, use spinlock to access dmabuf->name to avoid sleep-while-atomic. Cc: <stable@vger.kernel.org> [5.3+] Signed-off-by: Charan Teja Kalla <charante@codeaurora.org> Reviewed-by: Michael J. Ruhl <michael.j.ruhl@intel.com> Acked-by: Christian König <christian.koenig@amd.com> [sumits: added comment to spinlock_t definition to avoid warning] Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Link: https://patchwork.freedesktop.org/patch/msgid/a83e7f0d-4e54-9848-4b58-e1acdbe06735@codeaurora.org
2020-06-19 19:57:19 +08:00
spin_unlock(&dmabuf->name_lock);
return 0;
}
dma-buf: Add an API for exporting sync files (v14) Modern userspace APIs like Vulkan are built on an explicit synchronization model. This doesn't always play nicely with the implicit synchronization used in the kernel and assumed by X11 and Wayland. The client -> compositor half of the synchronization isn't too bad, at least on intel, because we can control whether or not i915 synchronizes on the buffer and whether or not it's considered written. The harder part is the compositor -> client synchronization when we get the buffer back from the compositor. We're required to be able to provide the client with a VkSemaphore and VkFence representing the point in time where the window system (compositor and/or display) finished using the buffer. With current APIs, it's very hard to do this in such a way that we don't get confused by the Vulkan driver's access of the buffer. In particular, once we tell the kernel that we're rendering to the buffer again, any CPU waits on the buffer or GPU dependencies will wait on some of the client rendering and not just the compositor. This new IOCTL solves this problem by allowing us to get a snapshot of the implicit synchronization state of a given dma-buf in the form of a sync file. It's effectively the same as a poll() or I915_GEM_WAIT only, instead of CPU waiting directly, it encapsulates the wait operation, at the current moment in time, in a sync_file so we can check/wait on it later. As long as the Vulkan driver does the sync_file export from the dma-buf before we re-introduce it for rendering, it will only contain fences from the compositor or display. This allows to accurately turn it into a VkFence or VkSemaphore without any over-synchronization. By making this an ioctl on the dma-buf itself, it allows this new functionality to be used in an entirely driver-agnostic way without having access to a DRM fd. This makes it ideal for use in driver-generic code in Mesa or in a client such as a compositor where the DRM fd may be hard to reach. v2 (Jason Ekstrand): - Use a wrapper dma_fence_array of all fences including the new one when importing an exclusive fence. v3 (Jason Ekstrand): - Lock around setting shared fences as well as exclusive - Mark SIGNAL_SYNC_FILE as a read-write ioctl. - Initialize ret to 0 in dma_buf_wait_sync_file v4 (Jason Ekstrand): - Use the new dma_resv_get_singleton helper v5 (Jason Ekstrand): - Rename the IOCTLs to import/export rather than wait/signal - Drop the WRITE flag and always get/set the exclusive fence v6 (Jason Ekstrand): - Drop the sync_file import as it was all-around sketchy and not nearly as useful as import. - Re-introduce READ/WRITE flag support for export - Rework the commit message v7 (Jason Ekstrand): - Require at least one sync flag - Fix a refcounting bug: dma_resv_get_excl() doesn't take a reference - Use _rcu helpers since we're accessing the dma_resv read-only v8 (Jason Ekstrand): - Return -ENOMEM if the sync_file_create fails - Predicate support on IS_ENABLED(CONFIG_SYNC_FILE) v9 (Jason Ekstrand): - Add documentation for the new ioctl v10 (Jason Ekstrand): - Go back to dma_buf_sync_file as the ioctl struct name v11 (Daniel Vetter): - Go back to dma_buf_export_sync_file as the ioctl struct name - Better kerneldoc describing what the read/write flags do v12 (Christian König): - Document why we chose to make it an ioctl on dma-buf v13 (Jason Ekstrand): - Rebase on Christian König's fence rework v14 (Daniel Vetter & Christian König): - Use dma_rev_usage_rw to get the properly inverted usage to pass to dma_resv_get_singleton() - Clean up the sync_file and fd if copy_to_user() fails Signed-off-by: Jason Ekstrand <jason@jlekstrand.net> Signed-off-by: Jason Ekstrand <jason.ekstrand@intel.com> Signed-off-by: Jason Ekstrand <jason.ekstrand@collabora.com> Acked-by: Simon Ser <contact@emersion.fr> Reviewed-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: Sumit Semwal <sumit.semwal@linaro.org> Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Signed-off-by: Simon Ser <contact@emersion.fr> Link: https://patchwork.freedesktop.org/patch/msgid/20220608152142.14495-2-jason@jlekstrand.net
2022-06-08 23:21:41 +08:00
#if IS_ENABLED(CONFIG_SYNC_FILE)
static long dma_buf_export_sync_file(struct dma_buf *dmabuf,
void __user *user_data)
{
struct dma_buf_export_sync_file arg;
enum dma_resv_usage usage;
struct dma_fence *fence = NULL;
struct sync_file *sync_file;
int fd, ret;
if (copy_from_user(&arg, user_data, sizeof(arg)))
return -EFAULT;
if (arg.flags & ~DMA_BUF_SYNC_RW)
return -EINVAL;
if ((arg.flags & DMA_BUF_SYNC_RW) == 0)
return -EINVAL;
fd = get_unused_fd_flags(O_CLOEXEC);
if (fd < 0)
return fd;
usage = dma_resv_usage_rw(arg.flags & DMA_BUF_SYNC_WRITE);
ret = dma_resv_get_singleton(dmabuf->resv, usage, &fence);
if (ret)
goto err_put_fd;
if (!fence)
fence = dma_fence_get_stub();
sync_file = sync_file_create(fence);
dma_fence_put(fence);
if (!sync_file) {
ret = -ENOMEM;
goto err_put_fd;
}
arg.fd = fd;
if (copy_to_user(user_data, &arg, sizeof(arg))) {
ret = -EFAULT;
goto err_put_file;
}
fd_install(fd, sync_file->file);
return 0;
err_put_file:
fput(sync_file->file);
err_put_fd:
put_unused_fd(fd);
return ret;
}
dma-buf: Add an API for importing sync files (v10) This patch is analogous to the previous sync file export patch in that it allows you to import a sync_file into a dma-buf. Unlike the previous patch, however, this does add genuinely new functionality to dma-buf. Without this, the only way to attach a sync_file to a dma-buf is to submit a batch to your driver of choice which waits on the sync_file and claims to write to the dma-buf. Even if said batch is a no-op, a submit is typically way more overhead than just attaching a fence. A submit may also imply extra synchronization with other work because it happens on a hardware queue. In the Vulkan world, this is useful for dealing with the out-fence from vkQueuePresent. Current Linux window-systems (X11, Wayland, etc.) all rely on dma-buf implicit sync. Since Vulkan is an explicit sync API, we get a set of fences (VkSemaphores) in vkQueuePresent and have to stash those as an exclusive (write) fence on the dma-buf. We handle it in Mesa today with the above mentioned dummy submit trick. This ioctl would allow us to set it directly without the dummy submit. This may also open up possibilities for GPU drivers to move away from implicit sync for their kernel driver uAPI and instead provide sync files and rely on dma-buf import/export for communicating with other implicit sync clients. We make the explicit choice here to only allow setting RW fences which translates to an exclusive fence on the dma_resv. There's no use for read-only fences for communicating with other implicit sync userspace and any such attempts are likely to be racy at best. When we got to insert the RW fence, the actual fence we set as the new exclusive fence is a combination of the sync_file provided by the user and all the other fences on the dma_resv. This ensures that the newly added exclusive fence will never signal before the old one would have and ensures that we don't break any dma_resv contracts. We require userspace to specify RW in the flags for symmetry with the export ioctl and in case we ever want to support read fences in the future. There is one downside here that's worth documenting: If two clients writing to the same dma-buf using this API race with each other, their actions on the dma-buf may happen in parallel or in an undefined order. Both with and without this API, the pattern is the same: Collect all the fences on dma-buf, submit work which depends on said fences, and then set a new exclusive (write) fence on the dma-buf which depends on said work. The difference is that, when it's all handled by the GPU driver's submit ioctl, the three operations happen atomically under the dma_resv lock. If two userspace submits race, one will happen before the other. You aren't guaranteed which but you are guaranteed that they're strictly ordered. If userspace manages the fences itself, then these three operations happen separately and the two render operations may happen genuinely in parallel or get interleaved. However, this is a case of userspace racing with itself. As long as we ensure userspace can't back the kernel into a corner, it should be fine. v2 (Jason Ekstrand): - Use a wrapper dma_fence_array of all fences including the new one when importing an exclusive fence. v3 (Jason Ekstrand): - Lock around setting shared fences as well as exclusive - Mark SIGNAL_SYNC_FILE as a read-write ioctl. - Initialize ret to 0 in dma_buf_wait_sync_file v4 (Jason Ekstrand): - Use the new dma_resv_get_singleton helper v5 (Jason Ekstrand): - Rename the IOCTLs to import/export rather than wait/signal - Drop the WRITE flag and always get/set the exclusive fence v6 (Jason Ekstrand): - Split import and export into separate patches - New commit message v7 (Daniel Vetter): - Fix the uapi header to use the right struct in the ioctl - Use a separate dma_buf_import_sync_file struct - Add kerneldoc for dma_buf_import_sync_file v8 (Jason Ekstrand): - Rebase on Christian König's fence rework v9 (Daniel Vetter): - Fix -EINVAL checks for the flags parameter - Add documentation about read/write fences - Add documentation about the expected usage of import/export and specifically call out the possible userspace race. v10 (Simon Ser): - Fix a typo in the docs Signed-off-by: Jason Ekstrand <jason@jlekstrand.net> Signed-off-by: Jason Ekstrand <jason.ekstrand@intel.com> Signed-off-by: Jason Ekstrand <jason.ekstrand@collabora.com> Reviewed-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: Sumit Semwal <sumit.semwal@linaro.org> Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Signed-off-by: Simon Ser <contact@emersion.fr> Link: https://patchwork.freedesktop.org/patch/msgid/20220608152142.14495-3-jason@jlekstrand.net
2022-06-08 23:21:42 +08:00
static long dma_buf_import_sync_file(struct dma_buf *dmabuf,
const void __user *user_data)
{
struct dma_buf_import_sync_file arg;
struct dma_fence *fence, *f;
dma-buf: Add an API for importing sync files (v10) This patch is analogous to the previous sync file export patch in that it allows you to import a sync_file into a dma-buf. Unlike the previous patch, however, this does add genuinely new functionality to dma-buf. Without this, the only way to attach a sync_file to a dma-buf is to submit a batch to your driver of choice which waits on the sync_file and claims to write to the dma-buf. Even if said batch is a no-op, a submit is typically way more overhead than just attaching a fence. A submit may also imply extra synchronization with other work because it happens on a hardware queue. In the Vulkan world, this is useful for dealing with the out-fence from vkQueuePresent. Current Linux window-systems (X11, Wayland, etc.) all rely on dma-buf implicit sync. Since Vulkan is an explicit sync API, we get a set of fences (VkSemaphores) in vkQueuePresent and have to stash those as an exclusive (write) fence on the dma-buf. We handle it in Mesa today with the above mentioned dummy submit trick. This ioctl would allow us to set it directly without the dummy submit. This may also open up possibilities for GPU drivers to move away from implicit sync for their kernel driver uAPI and instead provide sync files and rely on dma-buf import/export for communicating with other implicit sync clients. We make the explicit choice here to only allow setting RW fences which translates to an exclusive fence on the dma_resv. There's no use for read-only fences for communicating with other implicit sync userspace and any such attempts are likely to be racy at best. When we got to insert the RW fence, the actual fence we set as the new exclusive fence is a combination of the sync_file provided by the user and all the other fences on the dma_resv. This ensures that the newly added exclusive fence will never signal before the old one would have and ensures that we don't break any dma_resv contracts. We require userspace to specify RW in the flags for symmetry with the export ioctl and in case we ever want to support read fences in the future. There is one downside here that's worth documenting: If two clients writing to the same dma-buf using this API race with each other, their actions on the dma-buf may happen in parallel or in an undefined order. Both with and without this API, the pattern is the same: Collect all the fences on dma-buf, submit work which depends on said fences, and then set a new exclusive (write) fence on the dma-buf which depends on said work. The difference is that, when it's all handled by the GPU driver's submit ioctl, the three operations happen atomically under the dma_resv lock. If two userspace submits race, one will happen before the other. You aren't guaranteed which but you are guaranteed that they're strictly ordered. If userspace manages the fences itself, then these three operations happen separately and the two render operations may happen genuinely in parallel or get interleaved. However, this is a case of userspace racing with itself. As long as we ensure userspace can't back the kernel into a corner, it should be fine. v2 (Jason Ekstrand): - Use a wrapper dma_fence_array of all fences including the new one when importing an exclusive fence. v3 (Jason Ekstrand): - Lock around setting shared fences as well as exclusive - Mark SIGNAL_SYNC_FILE as a read-write ioctl. - Initialize ret to 0 in dma_buf_wait_sync_file v4 (Jason Ekstrand): - Use the new dma_resv_get_singleton helper v5 (Jason Ekstrand): - Rename the IOCTLs to import/export rather than wait/signal - Drop the WRITE flag and always get/set the exclusive fence v6 (Jason Ekstrand): - Split import and export into separate patches - New commit message v7 (Daniel Vetter): - Fix the uapi header to use the right struct in the ioctl - Use a separate dma_buf_import_sync_file struct - Add kerneldoc for dma_buf_import_sync_file v8 (Jason Ekstrand): - Rebase on Christian König's fence rework v9 (Daniel Vetter): - Fix -EINVAL checks for the flags parameter - Add documentation about read/write fences - Add documentation about the expected usage of import/export and specifically call out the possible userspace race. v10 (Simon Ser): - Fix a typo in the docs Signed-off-by: Jason Ekstrand <jason@jlekstrand.net> Signed-off-by: Jason Ekstrand <jason.ekstrand@intel.com> Signed-off-by: Jason Ekstrand <jason.ekstrand@collabora.com> Reviewed-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: Sumit Semwal <sumit.semwal@linaro.org> Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Signed-off-by: Simon Ser <contact@emersion.fr> Link: https://patchwork.freedesktop.org/patch/msgid/20220608152142.14495-3-jason@jlekstrand.net
2022-06-08 23:21:42 +08:00
enum dma_resv_usage usage;
struct dma_fence_unwrap iter;
unsigned int num_fences;
dma-buf: Add an API for importing sync files (v10) This patch is analogous to the previous sync file export patch in that it allows you to import a sync_file into a dma-buf. Unlike the previous patch, however, this does add genuinely new functionality to dma-buf. Without this, the only way to attach a sync_file to a dma-buf is to submit a batch to your driver of choice which waits on the sync_file and claims to write to the dma-buf. Even if said batch is a no-op, a submit is typically way more overhead than just attaching a fence. A submit may also imply extra synchronization with other work because it happens on a hardware queue. In the Vulkan world, this is useful for dealing with the out-fence from vkQueuePresent. Current Linux window-systems (X11, Wayland, etc.) all rely on dma-buf implicit sync. Since Vulkan is an explicit sync API, we get a set of fences (VkSemaphores) in vkQueuePresent and have to stash those as an exclusive (write) fence on the dma-buf. We handle it in Mesa today with the above mentioned dummy submit trick. This ioctl would allow us to set it directly without the dummy submit. This may also open up possibilities for GPU drivers to move away from implicit sync for their kernel driver uAPI and instead provide sync files and rely on dma-buf import/export for communicating with other implicit sync clients. We make the explicit choice here to only allow setting RW fences which translates to an exclusive fence on the dma_resv. There's no use for read-only fences for communicating with other implicit sync userspace and any such attempts are likely to be racy at best. When we got to insert the RW fence, the actual fence we set as the new exclusive fence is a combination of the sync_file provided by the user and all the other fences on the dma_resv. This ensures that the newly added exclusive fence will never signal before the old one would have and ensures that we don't break any dma_resv contracts. We require userspace to specify RW in the flags for symmetry with the export ioctl and in case we ever want to support read fences in the future. There is one downside here that's worth documenting: If two clients writing to the same dma-buf using this API race with each other, their actions on the dma-buf may happen in parallel or in an undefined order. Both with and without this API, the pattern is the same: Collect all the fences on dma-buf, submit work which depends on said fences, and then set a new exclusive (write) fence on the dma-buf which depends on said work. The difference is that, when it's all handled by the GPU driver's submit ioctl, the three operations happen atomically under the dma_resv lock. If two userspace submits race, one will happen before the other. You aren't guaranteed which but you are guaranteed that they're strictly ordered. If userspace manages the fences itself, then these three operations happen separately and the two render operations may happen genuinely in parallel or get interleaved. However, this is a case of userspace racing with itself. As long as we ensure userspace can't back the kernel into a corner, it should be fine. v2 (Jason Ekstrand): - Use a wrapper dma_fence_array of all fences including the new one when importing an exclusive fence. v3 (Jason Ekstrand): - Lock around setting shared fences as well as exclusive - Mark SIGNAL_SYNC_FILE as a read-write ioctl. - Initialize ret to 0 in dma_buf_wait_sync_file v4 (Jason Ekstrand): - Use the new dma_resv_get_singleton helper v5 (Jason Ekstrand): - Rename the IOCTLs to import/export rather than wait/signal - Drop the WRITE flag and always get/set the exclusive fence v6 (Jason Ekstrand): - Split import and export into separate patches - New commit message v7 (Daniel Vetter): - Fix the uapi header to use the right struct in the ioctl - Use a separate dma_buf_import_sync_file struct - Add kerneldoc for dma_buf_import_sync_file v8 (Jason Ekstrand): - Rebase on Christian König's fence rework v9 (Daniel Vetter): - Fix -EINVAL checks for the flags parameter - Add documentation about read/write fences - Add documentation about the expected usage of import/export and specifically call out the possible userspace race. v10 (Simon Ser): - Fix a typo in the docs Signed-off-by: Jason Ekstrand <jason@jlekstrand.net> Signed-off-by: Jason Ekstrand <jason.ekstrand@intel.com> Signed-off-by: Jason Ekstrand <jason.ekstrand@collabora.com> Reviewed-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: Sumit Semwal <sumit.semwal@linaro.org> Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Signed-off-by: Simon Ser <contact@emersion.fr> Link: https://patchwork.freedesktop.org/patch/msgid/20220608152142.14495-3-jason@jlekstrand.net
2022-06-08 23:21:42 +08:00
int ret = 0;
if (copy_from_user(&arg, user_data, sizeof(arg)))
return -EFAULT;
if (arg.flags & ~DMA_BUF_SYNC_RW)
return -EINVAL;
if ((arg.flags & DMA_BUF_SYNC_RW) == 0)
return -EINVAL;
fence = sync_file_get_fence(arg.fd);
if (!fence)
return -EINVAL;
usage = (arg.flags & DMA_BUF_SYNC_WRITE) ? DMA_RESV_USAGE_WRITE :
DMA_RESV_USAGE_READ;
num_fences = 0;
dma_fence_unwrap_for_each(f, &iter, fence)
++num_fences;
if (num_fences > 0) {
dma_resv_lock(dmabuf->resv, NULL);
dma-buf: Add an API for importing sync files (v10) This patch is analogous to the previous sync file export patch in that it allows you to import a sync_file into a dma-buf. Unlike the previous patch, however, this does add genuinely new functionality to dma-buf. Without this, the only way to attach a sync_file to a dma-buf is to submit a batch to your driver of choice which waits on the sync_file and claims to write to the dma-buf. Even if said batch is a no-op, a submit is typically way more overhead than just attaching a fence. A submit may also imply extra synchronization with other work because it happens on a hardware queue. In the Vulkan world, this is useful for dealing with the out-fence from vkQueuePresent. Current Linux window-systems (X11, Wayland, etc.) all rely on dma-buf implicit sync. Since Vulkan is an explicit sync API, we get a set of fences (VkSemaphores) in vkQueuePresent and have to stash those as an exclusive (write) fence on the dma-buf. We handle it in Mesa today with the above mentioned dummy submit trick. This ioctl would allow us to set it directly without the dummy submit. This may also open up possibilities for GPU drivers to move away from implicit sync for their kernel driver uAPI and instead provide sync files and rely on dma-buf import/export for communicating with other implicit sync clients. We make the explicit choice here to only allow setting RW fences which translates to an exclusive fence on the dma_resv. There's no use for read-only fences for communicating with other implicit sync userspace and any such attempts are likely to be racy at best. When we got to insert the RW fence, the actual fence we set as the new exclusive fence is a combination of the sync_file provided by the user and all the other fences on the dma_resv. This ensures that the newly added exclusive fence will never signal before the old one would have and ensures that we don't break any dma_resv contracts. We require userspace to specify RW in the flags for symmetry with the export ioctl and in case we ever want to support read fences in the future. There is one downside here that's worth documenting: If two clients writing to the same dma-buf using this API race with each other, their actions on the dma-buf may happen in parallel or in an undefined order. Both with and without this API, the pattern is the same: Collect all the fences on dma-buf, submit work which depends on said fences, and then set a new exclusive (write) fence on the dma-buf which depends on said work. The difference is that, when it's all handled by the GPU driver's submit ioctl, the three operations happen atomically under the dma_resv lock. If two userspace submits race, one will happen before the other. You aren't guaranteed which but you are guaranteed that they're strictly ordered. If userspace manages the fences itself, then these three operations happen separately and the two render operations may happen genuinely in parallel or get interleaved. However, this is a case of userspace racing with itself. As long as we ensure userspace can't back the kernel into a corner, it should be fine. v2 (Jason Ekstrand): - Use a wrapper dma_fence_array of all fences including the new one when importing an exclusive fence. v3 (Jason Ekstrand): - Lock around setting shared fences as well as exclusive - Mark SIGNAL_SYNC_FILE as a read-write ioctl. - Initialize ret to 0 in dma_buf_wait_sync_file v4 (Jason Ekstrand): - Use the new dma_resv_get_singleton helper v5 (Jason Ekstrand): - Rename the IOCTLs to import/export rather than wait/signal - Drop the WRITE flag and always get/set the exclusive fence v6 (Jason Ekstrand): - Split import and export into separate patches - New commit message v7 (Daniel Vetter): - Fix the uapi header to use the right struct in the ioctl - Use a separate dma_buf_import_sync_file struct - Add kerneldoc for dma_buf_import_sync_file v8 (Jason Ekstrand): - Rebase on Christian König's fence rework v9 (Daniel Vetter): - Fix -EINVAL checks for the flags parameter - Add documentation about read/write fences - Add documentation about the expected usage of import/export and specifically call out the possible userspace race. v10 (Simon Ser): - Fix a typo in the docs Signed-off-by: Jason Ekstrand <jason@jlekstrand.net> Signed-off-by: Jason Ekstrand <jason.ekstrand@intel.com> Signed-off-by: Jason Ekstrand <jason.ekstrand@collabora.com> Reviewed-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: Sumit Semwal <sumit.semwal@linaro.org> Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Signed-off-by: Simon Ser <contact@emersion.fr> Link: https://patchwork.freedesktop.org/patch/msgid/20220608152142.14495-3-jason@jlekstrand.net
2022-06-08 23:21:42 +08:00
ret = dma_resv_reserve_fences(dmabuf->resv, num_fences);
if (!ret) {
dma_fence_unwrap_for_each(f, &iter, fence)
dma_resv_add_fence(dmabuf->resv, f, usage);
}
dma-buf: Add an API for importing sync files (v10) This patch is analogous to the previous sync file export patch in that it allows you to import a sync_file into a dma-buf. Unlike the previous patch, however, this does add genuinely new functionality to dma-buf. Without this, the only way to attach a sync_file to a dma-buf is to submit a batch to your driver of choice which waits on the sync_file and claims to write to the dma-buf. Even if said batch is a no-op, a submit is typically way more overhead than just attaching a fence. A submit may also imply extra synchronization with other work because it happens on a hardware queue. In the Vulkan world, this is useful for dealing with the out-fence from vkQueuePresent. Current Linux window-systems (X11, Wayland, etc.) all rely on dma-buf implicit sync. Since Vulkan is an explicit sync API, we get a set of fences (VkSemaphores) in vkQueuePresent and have to stash those as an exclusive (write) fence on the dma-buf. We handle it in Mesa today with the above mentioned dummy submit trick. This ioctl would allow us to set it directly without the dummy submit. This may also open up possibilities for GPU drivers to move away from implicit sync for their kernel driver uAPI and instead provide sync files and rely on dma-buf import/export for communicating with other implicit sync clients. We make the explicit choice here to only allow setting RW fences which translates to an exclusive fence on the dma_resv. There's no use for read-only fences for communicating with other implicit sync userspace and any such attempts are likely to be racy at best. When we got to insert the RW fence, the actual fence we set as the new exclusive fence is a combination of the sync_file provided by the user and all the other fences on the dma_resv. This ensures that the newly added exclusive fence will never signal before the old one would have and ensures that we don't break any dma_resv contracts. We require userspace to specify RW in the flags for symmetry with the export ioctl and in case we ever want to support read fences in the future. There is one downside here that's worth documenting: If two clients writing to the same dma-buf using this API race with each other, their actions on the dma-buf may happen in parallel or in an undefined order. Both with and without this API, the pattern is the same: Collect all the fences on dma-buf, submit work which depends on said fences, and then set a new exclusive (write) fence on the dma-buf which depends on said work. The difference is that, when it's all handled by the GPU driver's submit ioctl, the three operations happen atomically under the dma_resv lock. If two userspace submits race, one will happen before the other. You aren't guaranteed which but you are guaranteed that they're strictly ordered. If userspace manages the fences itself, then these three operations happen separately and the two render operations may happen genuinely in parallel or get interleaved. However, this is a case of userspace racing with itself. As long as we ensure userspace can't back the kernel into a corner, it should be fine. v2 (Jason Ekstrand): - Use a wrapper dma_fence_array of all fences including the new one when importing an exclusive fence. v3 (Jason Ekstrand): - Lock around setting shared fences as well as exclusive - Mark SIGNAL_SYNC_FILE as a read-write ioctl. - Initialize ret to 0 in dma_buf_wait_sync_file v4 (Jason Ekstrand): - Use the new dma_resv_get_singleton helper v5 (Jason Ekstrand): - Rename the IOCTLs to import/export rather than wait/signal - Drop the WRITE flag and always get/set the exclusive fence v6 (Jason Ekstrand): - Split import and export into separate patches - New commit message v7 (Daniel Vetter): - Fix the uapi header to use the right struct in the ioctl - Use a separate dma_buf_import_sync_file struct - Add kerneldoc for dma_buf_import_sync_file v8 (Jason Ekstrand): - Rebase on Christian König's fence rework v9 (Daniel Vetter): - Fix -EINVAL checks for the flags parameter - Add documentation about read/write fences - Add documentation about the expected usage of import/export and specifically call out the possible userspace race. v10 (Simon Ser): - Fix a typo in the docs Signed-off-by: Jason Ekstrand <jason@jlekstrand.net> Signed-off-by: Jason Ekstrand <jason.ekstrand@intel.com> Signed-off-by: Jason Ekstrand <jason.ekstrand@collabora.com> Reviewed-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: Sumit Semwal <sumit.semwal@linaro.org> Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Signed-off-by: Simon Ser <contact@emersion.fr> Link: https://patchwork.freedesktop.org/patch/msgid/20220608152142.14495-3-jason@jlekstrand.net
2022-06-08 23:21:42 +08:00
dma_resv_unlock(dmabuf->resv);
}
dma-buf: Add an API for importing sync files (v10) This patch is analogous to the previous sync file export patch in that it allows you to import a sync_file into a dma-buf. Unlike the previous patch, however, this does add genuinely new functionality to dma-buf. Without this, the only way to attach a sync_file to a dma-buf is to submit a batch to your driver of choice which waits on the sync_file and claims to write to the dma-buf. Even if said batch is a no-op, a submit is typically way more overhead than just attaching a fence. A submit may also imply extra synchronization with other work because it happens on a hardware queue. In the Vulkan world, this is useful for dealing with the out-fence from vkQueuePresent. Current Linux window-systems (X11, Wayland, etc.) all rely on dma-buf implicit sync. Since Vulkan is an explicit sync API, we get a set of fences (VkSemaphores) in vkQueuePresent and have to stash those as an exclusive (write) fence on the dma-buf. We handle it in Mesa today with the above mentioned dummy submit trick. This ioctl would allow us to set it directly without the dummy submit. This may also open up possibilities for GPU drivers to move away from implicit sync for their kernel driver uAPI and instead provide sync files and rely on dma-buf import/export for communicating with other implicit sync clients. We make the explicit choice here to only allow setting RW fences which translates to an exclusive fence on the dma_resv. There's no use for read-only fences for communicating with other implicit sync userspace and any such attempts are likely to be racy at best. When we got to insert the RW fence, the actual fence we set as the new exclusive fence is a combination of the sync_file provided by the user and all the other fences on the dma_resv. This ensures that the newly added exclusive fence will never signal before the old one would have and ensures that we don't break any dma_resv contracts. We require userspace to specify RW in the flags for symmetry with the export ioctl and in case we ever want to support read fences in the future. There is one downside here that's worth documenting: If two clients writing to the same dma-buf using this API race with each other, their actions on the dma-buf may happen in parallel or in an undefined order. Both with and without this API, the pattern is the same: Collect all the fences on dma-buf, submit work which depends on said fences, and then set a new exclusive (write) fence on the dma-buf which depends on said work. The difference is that, when it's all handled by the GPU driver's submit ioctl, the three operations happen atomically under the dma_resv lock. If two userspace submits race, one will happen before the other. You aren't guaranteed which but you are guaranteed that they're strictly ordered. If userspace manages the fences itself, then these three operations happen separately and the two render operations may happen genuinely in parallel or get interleaved. However, this is a case of userspace racing with itself. As long as we ensure userspace can't back the kernel into a corner, it should be fine. v2 (Jason Ekstrand): - Use a wrapper dma_fence_array of all fences including the new one when importing an exclusive fence. v3 (Jason Ekstrand): - Lock around setting shared fences as well as exclusive - Mark SIGNAL_SYNC_FILE as a read-write ioctl. - Initialize ret to 0 in dma_buf_wait_sync_file v4 (Jason Ekstrand): - Use the new dma_resv_get_singleton helper v5 (Jason Ekstrand): - Rename the IOCTLs to import/export rather than wait/signal - Drop the WRITE flag and always get/set the exclusive fence v6 (Jason Ekstrand): - Split import and export into separate patches - New commit message v7 (Daniel Vetter): - Fix the uapi header to use the right struct in the ioctl - Use a separate dma_buf_import_sync_file struct - Add kerneldoc for dma_buf_import_sync_file v8 (Jason Ekstrand): - Rebase on Christian König's fence rework v9 (Daniel Vetter): - Fix -EINVAL checks for the flags parameter - Add documentation about read/write fences - Add documentation about the expected usage of import/export and specifically call out the possible userspace race. v10 (Simon Ser): - Fix a typo in the docs Signed-off-by: Jason Ekstrand <jason@jlekstrand.net> Signed-off-by: Jason Ekstrand <jason.ekstrand@intel.com> Signed-off-by: Jason Ekstrand <jason.ekstrand@collabora.com> Reviewed-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: Sumit Semwal <sumit.semwal@linaro.org> Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Signed-off-by: Simon Ser <contact@emersion.fr> Link: https://patchwork.freedesktop.org/patch/msgid/20220608152142.14495-3-jason@jlekstrand.net
2022-06-08 23:21:42 +08:00
dma_fence_put(fence);
return ret;
}
dma-buf: Add an API for exporting sync files (v14) Modern userspace APIs like Vulkan are built on an explicit synchronization model. This doesn't always play nicely with the implicit synchronization used in the kernel and assumed by X11 and Wayland. The client -> compositor half of the synchronization isn't too bad, at least on intel, because we can control whether or not i915 synchronizes on the buffer and whether or not it's considered written. The harder part is the compositor -> client synchronization when we get the buffer back from the compositor. We're required to be able to provide the client with a VkSemaphore and VkFence representing the point in time where the window system (compositor and/or display) finished using the buffer. With current APIs, it's very hard to do this in such a way that we don't get confused by the Vulkan driver's access of the buffer. In particular, once we tell the kernel that we're rendering to the buffer again, any CPU waits on the buffer or GPU dependencies will wait on some of the client rendering and not just the compositor. This new IOCTL solves this problem by allowing us to get a snapshot of the implicit synchronization state of a given dma-buf in the form of a sync file. It's effectively the same as a poll() or I915_GEM_WAIT only, instead of CPU waiting directly, it encapsulates the wait operation, at the current moment in time, in a sync_file so we can check/wait on it later. As long as the Vulkan driver does the sync_file export from the dma-buf before we re-introduce it for rendering, it will only contain fences from the compositor or display. This allows to accurately turn it into a VkFence or VkSemaphore without any over-synchronization. By making this an ioctl on the dma-buf itself, it allows this new functionality to be used in an entirely driver-agnostic way without having access to a DRM fd. This makes it ideal for use in driver-generic code in Mesa or in a client such as a compositor where the DRM fd may be hard to reach. v2 (Jason Ekstrand): - Use a wrapper dma_fence_array of all fences including the new one when importing an exclusive fence. v3 (Jason Ekstrand): - Lock around setting shared fences as well as exclusive - Mark SIGNAL_SYNC_FILE as a read-write ioctl. - Initialize ret to 0 in dma_buf_wait_sync_file v4 (Jason Ekstrand): - Use the new dma_resv_get_singleton helper v5 (Jason Ekstrand): - Rename the IOCTLs to import/export rather than wait/signal - Drop the WRITE flag and always get/set the exclusive fence v6 (Jason Ekstrand): - Drop the sync_file import as it was all-around sketchy and not nearly as useful as import. - Re-introduce READ/WRITE flag support for export - Rework the commit message v7 (Jason Ekstrand): - Require at least one sync flag - Fix a refcounting bug: dma_resv_get_excl() doesn't take a reference - Use _rcu helpers since we're accessing the dma_resv read-only v8 (Jason Ekstrand): - Return -ENOMEM if the sync_file_create fails - Predicate support on IS_ENABLED(CONFIG_SYNC_FILE) v9 (Jason Ekstrand): - Add documentation for the new ioctl v10 (Jason Ekstrand): - Go back to dma_buf_sync_file as the ioctl struct name v11 (Daniel Vetter): - Go back to dma_buf_export_sync_file as the ioctl struct name - Better kerneldoc describing what the read/write flags do v12 (Christian König): - Document why we chose to make it an ioctl on dma-buf v13 (Jason Ekstrand): - Rebase on Christian König's fence rework v14 (Daniel Vetter & Christian König): - Use dma_rev_usage_rw to get the properly inverted usage to pass to dma_resv_get_singleton() - Clean up the sync_file and fd if copy_to_user() fails Signed-off-by: Jason Ekstrand <jason@jlekstrand.net> Signed-off-by: Jason Ekstrand <jason.ekstrand@intel.com> Signed-off-by: Jason Ekstrand <jason.ekstrand@collabora.com> Acked-by: Simon Ser <contact@emersion.fr> Reviewed-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: Sumit Semwal <sumit.semwal@linaro.org> Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Signed-off-by: Simon Ser <contact@emersion.fr> Link: https://patchwork.freedesktop.org/patch/msgid/20220608152142.14495-2-jason@jlekstrand.net
2022-06-08 23:21:41 +08:00
#endif
dma-buf: Add ioctls to allow userspace to flush The userspace might need some sort of cache coherency management e.g. when CPU and GPU domains are being accessed through dma-buf at the same time. To circumvent this problem there are begin/end coherency markers, that forward directly to existing dma-buf device drivers vfunc hooks. Userspace can make use of those markers through the DMA_BUF_IOCTL_SYNC ioctl. The sequence would be used like following: - mmap dma-buf fd - for each drawing/upload cycle in CPU 1. SYNC_START ioctl, 2. read/write to mmap area 3. SYNC_END ioctl. This can be repeated as often as you want (with the new data being consumed by the GPU or say scanout device) - munmap once you don't need the buffer any more v2 (Tiago): Fix header file type names (u64 -> __u64) v3 (Tiago): Add documentation. Use enum dma_buf_sync_flags to the begin/end dma-buf functions. Check for overflows in start/length. v4 (Tiago): use 2d regions for sync. v5 (Tiago): forget about 2d regions (v4); use _IOW in DMA_BUF_IOCTL_SYNC and remove range information from struct dma_buf_sync. v6 (Tiago): use __u64 structured padded flags instead enum. Adjust documentation about the recommendation on using sync ioctls. v7 (Tiago): Alex' nit on flags definition and being even more wording in the doc about sync usage. v9 (Tiago): remove useless is_dma_buf_file check. Fix sync.flags conditionals and its mask order check. Add <linux/types.h> include in dma-buf.h. Cc: Ville Syrjälä <ville.syrjala@linux.intel.com> Cc: David Herrmann <dh.herrmann@gmail.com> Cc: Sumit Semwal <sumit.semwal@linaro.org> Reviewed-by: Stéphane Marchesin <marcheu@chromium.org> Signed-off-by: Daniel Vetter <daniel.vetter@intel.com> Signed-off-by: Tiago Vignatti <tiago.vignatti@intel.com> Reviewed-by: David Herrmann <dh.herrmann@gmail.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: http://patchwork.freedesktop.org/patch/msgid/1455228291-29640-1-git-send-email-tiago.vignatti@intel.com
2016-02-12 06:04:51 +08:00
static long dma_buf_ioctl(struct file *file,
unsigned int cmd, unsigned long arg)
{
struct dma_buf *dmabuf;
struct dma_buf_sync sync;
enum dma_data_direction direction;
dma-buf, drm, ion: Propagate error code from dma_buf_start_cpu_access() Drivers, especially i915.ko, can fail during the initial migration of a dma-buf for CPU access. However, the error code from the driver was not being propagated back to ioctl and so userspace was blissfully ignorant of the failure. Rendering corruption ensues. Whilst fixing the ioctl to return the error code from dma_buf_start_cpu_access(), also do the same for dma_buf_end_cpu_access(). For most drivers, dma_buf_end_cpu_access() cannot fail. i915.ko however, as most drivers would, wants to avoid being uninterruptible (as would be required to guarrantee no failure when flushing the buffer to the device). As userspace already has to handle errors from the SYNC_IOCTL, take advantage of this to be able to restart the syscall across signals. This fixes a coherency issue for i915.ko as well as reducing the uninterruptible hold upon its BKL, the struct_mutex. Fixes commit c11e391da2a8fe973c3c2398452000bed505851e Author: Daniel Vetter <daniel.vetter@ffwll.ch> Date: Thu Feb 11 20:04:51 2016 -0200 dma-buf: Add ioctls to allow userspace to flush Testcase: igt/gem_concurrent_blit/*dmabuf*interruptible Testcase: igt/prime_mmap_coherency/ioctl-errors Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: Tiago Vignatti <tiago.vignatti@intel.com> Cc: Stéphane Marchesin <marcheu@chromium.org> Cc: David Herrmann <dh.herrmann@gmail.com> Cc: Sumit Semwal <sumit.semwal@linaro.org> Cc: Daniel Vetter <daniel.vetter@intel.com> CC: linux-media@vger.kernel.org Cc: dri-devel@lists.freedesktop.org Cc: linaro-mm-sig@lists.linaro.org Cc: intel-gfx@lists.freedesktop.org Cc: devel@driverdev.osuosl.org Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: http://patchwork.freedesktop.org/patch/msgid/1458331359-2634-1-git-send-email-chris@chris-wilson.co.uk
2016-03-19 04:02:39 +08:00
int ret;
dma-buf: Add ioctls to allow userspace to flush The userspace might need some sort of cache coherency management e.g. when CPU and GPU domains are being accessed through dma-buf at the same time. To circumvent this problem there are begin/end coherency markers, that forward directly to existing dma-buf device drivers vfunc hooks. Userspace can make use of those markers through the DMA_BUF_IOCTL_SYNC ioctl. The sequence would be used like following: - mmap dma-buf fd - for each drawing/upload cycle in CPU 1. SYNC_START ioctl, 2. read/write to mmap area 3. SYNC_END ioctl. This can be repeated as often as you want (with the new data being consumed by the GPU or say scanout device) - munmap once you don't need the buffer any more v2 (Tiago): Fix header file type names (u64 -> __u64) v3 (Tiago): Add documentation. Use enum dma_buf_sync_flags to the begin/end dma-buf functions. Check for overflows in start/length. v4 (Tiago): use 2d regions for sync. v5 (Tiago): forget about 2d regions (v4); use _IOW in DMA_BUF_IOCTL_SYNC and remove range information from struct dma_buf_sync. v6 (Tiago): use __u64 structured padded flags instead enum. Adjust documentation about the recommendation on using sync ioctls. v7 (Tiago): Alex' nit on flags definition and being even more wording in the doc about sync usage. v9 (Tiago): remove useless is_dma_buf_file check. Fix sync.flags conditionals and its mask order check. Add <linux/types.h> include in dma-buf.h. Cc: Ville Syrjälä <ville.syrjala@linux.intel.com> Cc: David Herrmann <dh.herrmann@gmail.com> Cc: Sumit Semwal <sumit.semwal@linaro.org> Reviewed-by: Stéphane Marchesin <marcheu@chromium.org> Signed-off-by: Daniel Vetter <daniel.vetter@intel.com> Signed-off-by: Tiago Vignatti <tiago.vignatti@intel.com> Reviewed-by: David Herrmann <dh.herrmann@gmail.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: http://patchwork.freedesktop.org/patch/msgid/1455228291-29640-1-git-send-email-tiago.vignatti@intel.com
2016-02-12 06:04:51 +08:00
dmabuf = file->private_data;
switch (cmd) {
case DMA_BUF_IOCTL_SYNC:
if (copy_from_user(&sync, (void __user *) arg, sizeof(sync)))
return -EFAULT;
if (sync.flags & ~DMA_BUF_SYNC_VALID_FLAGS_MASK)
return -EINVAL;
switch (sync.flags & DMA_BUF_SYNC_RW) {
case DMA_BUF_SYNC_READ:
direction = DMA_FROM_DEVICE;
break;
case DMA_BUF_SYNC_WRITE:
direction = DMA_TO_DEVICE;
break;
case DMA_BUF_SYNC_RW:
direction = DMA_BIDIRECTIONAL;
break;
default:
return -EINVAL;
}
if (sync.flags & DMA_BUF_SYNC_END)
dma-buf, drm, ion: Propagate error code from dma_buf_start_cpu_access() Drivers, especially i915.ko, can fail during the initial migration of a dma-buf for CPU access. However, the error code from the driver was not being propagated back to ioctl and so userspace was blissfully ignorant of the failure. Rendering corruption ensues. Whilst fixing the ioctl to return the error code from dma_buf_start_cpu_access(), also do the same for dma_buf_end_cpu_access(). For most drivers, dma_buf_end_cpu_access() cannot fail. i915.ko however, as most drivers would, wants to avoid being uninterruptible (as would be required to guarrantee no failure when flushing the buffer to the device). As userspace already has to handle errors from the SYNC_IOCTL, take advantage of this to be able to restart the syscall across signals. This fixes a coherency issue for i915.ko as well as reducing the uninterruptible hold upon its BKL, the struct_mutex. Fixes commit c11e391da2a8fe973c3c2398452000bed505851e Author: Daniel Vetter <daniel.vetter@ffwll.ch> Date: Thu Feb 11 20:04:51 2016 -0200 dma-buf: Add ioctls to allow userspace to flush Testcase: igt/gem_concurrent_blit/*dmabuf*interruptible Testcase: igt/prime_mmap_coherency/ioctl-errors Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: Tiago Vignatti <tiago.vignatti@intel.com> Cc: Stéphane Marchesin <marcheu@chromium.org> Cc: David Herrmann <dh.herrmann@gmail.com> Cc: Sumit Semwal <sumit.semwal@linaro.org> Cc: Daniel Vetter <daniel.vetter@intel.com> CC: linux-media@vger.kernel.org Cc: dri-devel@lists.freedesktop.org Cc: linaro-mm-sig@lists.linaro.org Cc: intel-gfx@lists.freedesktop.org Cc: devel@driverdev.osuosl.org Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: http://patchwork.freedesktop.org/patch/msgid/1458331359-2634-1-git-send-email-chris@chris-wilson.co.uk
2016-03-19 04:02:39 +08:00
ret = dma_buf_end_cpu_access(dmabuf, direction);
dma-buf: Add ioctls to allow userspace to flush The userspace might need some sort of cache coherency management e.g. when CPU and GPU domains are being accessed through dma-buf at the same time. To circumvent this problem there are begin/end coherency markers, that forward directly to existing dma-buf device drivers vfunc hooks. Userspace can make use of those markers through the DMA_BUF_IOCTL_SYNC ioctl. The sequence would be used like following: - mmap dma-buf fd - for each drawing/upload cycle in CPU 1. SYNC_START ioctl, 2. read/write to mmap area 3. SYNC_END ioctl. This can be repeated as often as you want (with the new data being consumed by the GPU or say scanout device) - munmap once you don't need the buffer any more v2 (Tiago): Fix header file type names (u64 -> __u64) v3 (Tiago): Add documentation. Use enum dma_buf_sync_flags to the begin/end dma-buf functions. Check for overflows in start/length. v4 (Tiago): use 2d regions for sync. v5 (Tiago): forget about 2d regions (v4); use _IOW in DMA_BUF_IOCTL_SYNC and remove range information from struct dma_buf_sync. v6 (Tiago): use __u64 structured padded flags instead enum. Adjust documentation about the recommendation on using sync ioctls. v7 (Tiago): Alex' nit on flags definition and being even more wording in the doc about sync usage. v9 (Tiago): remove useless is_dma_buf_file check. Fix sync.flags conditionals and its mask order check. Add <linux/types.h> include in dma-buf.h. Cc: Ville Syrjälä <ville.syrjala@linux.intel.com> Cc: David Herrmann <dh.herrmann@gmail.com> Cc: Sumit Semwal <sumit.semwal@linaro.org> Reviewed-by: Stéphane Marchesin <marcheu@chromium.org> Signed-off-by: Daniel Vetter <daniel.vetter@intel.com> Signed-off-by: Tiago Vignatti <tiago.vignatti@intel.com> Reviewed-by: David Herrmann <dh.herrmann@gmail.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: http://patchwork.freedesktop.org/patch/msgid/1455228291-29640-1-git-send-email-tiago.vignatti@intel.com
2016-02-12 06:04:51 +08:00
else
dma-buf, drm, ion: Propagate error code from dma_buf_start_cpu_access() Drivers, especially i915.ko, can fail during the initial migration of a dma-buf for CPU access. However, the error code from the driver was not being propagated back to ioctl and so userspace was blissfully ignorant of the failure. Rendering corruption ensues. Whilst fixing the ioctl to return the error code from dma_buf_start_cpu_access(), also do the same for dma_buf_end_cpu_access(). For most drivers, dma_buf_end_cpu_access() cannot fail. i915.ko however, as most drivers would, wants to avoid being uninterruptible (as would be required to guarrantee no failure when flushing the buffer to the device). As userspace already has to handle errors from the SYNC_IOCTL, take advantage of this to be able to restart the syscall across signals. This fixes a coherency issue for i915.ko as well as reducing the uninterruptible hold upon its BKL, the struct_mutex. Fixes commit c11e391da2a8fe973c3c2398452000bed505851e Author: Daniel Vetter <daniel.vetter@ffwll.ch> Date: Thu Feb 11 20:04:51 2016 -0200 dma-buf: Add ioctls to allow userspace to flush Testcase: igt/gem_concurrent_blit/*dmabuf*interruptible Testcase: igt/prime_mmap_coherency/ioctl-errors Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: Tiago Vignatti <tiago.vignatti@intel.com> Cc: Stéphane Marchesin <marcheu@chromium.org> Cc: David Herrmann <dh.herrmann@gmail.com> Cc: Sumit Semwal <sumit.semwal@linaro.org> Cc: Daniel Vetter <daniel.vetter@intel.com> CC: linux-media@vger.kernel.org Cc: dri-devel@lists.freedesktop.org Cc: linaro-mm-sig@lists.linaro.org Cc: intel-gfx@lists.freedesktop.org Cc: devel@driverdev.osuosl.org Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: http://patchwork.freedesktop.org/patch/msgid/1458331359-2634-1-git-send-email-chris@chris-wilson.co.uk
2016-03-19 04:02:39 +08:00
ret = dma_buf_begin_cpu_access(dmabuf, direction);
dma-buf: Add ioctls to allow userspace to flush The userspace might need some sort of cache coherency management e.g. when CPU and GPU domains are being accessed through dma-buf at the same time. To circumvent this problem there are begin/end coherency markers, that forward directly to existing dma-buf device drivers vfunc hooks. Userspace can make use of those markers through the DMA_BUF_IOCTL_SYNC ioctl. The sequence would be used like following: - mmap dma-buf fd - for each drawing/upload cycle in CPU 1. SYNC_START ioctl, 2. read/write to mmap area 3. SYNC_END ioctl. This can be repeated as often as you want (with the new data being consumed by the GPU or say scanout device) - munmap once you don't need the buffer any more v2 (Tiago): Fix header file type names (u64 -> __u64) v3 (Tiago): Add documentation. Use enum dma_buf_sync_flags to the begin/end dma-buf functions. Check for overflows in start/length. v4 (Tiago): use 2d regions for sync. v5 (Tiago): forget about 2d regions (v4); use _IOW in DMA_BUF_IOCTL_SYNC and remove range information from struct dma_buf_sync. v6 (Tiago): use __u64 structured padded flags instead enum. Adjust documentation about the recommendation on using sync ioctls. v7 (Tiago): Alex' nit on flags definition and being even more wording in the doc about sync usage. v9 (Tiago): remove useless is_dma_buf_file check. Fix sync.flags conditionals and its mask order check. Add <linux/types.h> include in dma-buf.h. Cc: Ville Syrjälä <ville.syrjala@linux.intel.com> Cc: David Herrmann <dh.herrmann@gmail.com> Cc: Sumit Semwal <sumit.semwal@linaro.org> Reviewed-by: Stéphane Marchesin <marcheu@chromium.org> Signed-off-by: Daniel Vetter <daniel.vetter@intel.com> Signed-off-by: Tiago Vignatti <tiago.vignatti@intel.com> Reviewed-by: David Herrmann <dh.herrmann@gmail.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: http://patchwork.freedesktop.org/patch/msgid/1455228291-29640-1-git-send-email-tiago.vignatti@intel.com
2016-02-12 06:04:51 +08:00
dma-buf, drm, ion: Propagate error code from dma_buf_start_cpu_access() Drivers, especially i915.ko, can fail during the initial migration of a dma-buf for CPU access. However, the error code from the driver was not being propagated back to ioctl and so userspace was blissfully ignorant of the failure. Rendering corruption ensues. Whilst fixing the ioctl to return the error code from dma_buf_start_cpu_access(), also do the same for dma_buf_end_cpu_access(). For most drivers, dma_buf_end_cpu_access() cannot fail. i915.ko however, as most drivers would, wants to avoid being uninterruptible (as would be required to guarrantee no failure when flushing the buffer to the device). As userspace already has to handle errors from the SYNC_IOCTL, take advantage of this to be able to restart the syscall across signals. This fixes a coherency issue for i915.ko as well as reducing the uninterruptible hold upon its BKL, the struct_mutex. Fixes commit c11e391da2a8fe973c3c2398452000bed505851e Author: Daniel Vetter <daniel.vetter@ffwll.ch> Date: Thu Feb 11 20:04:51 2016 -0200 dma-buf: Add ioctls to allow userspace to flush Testcase: igt/gem_concurrent_blit/*dmabuf*interruptible Testcase: igt/prime_mmap_coherency/ioctl-errors Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: Tiago Vignatti <tiago.vignatti@intel.com> Cc: Stéphane Marchesin <marcheu@chromium.org> Cc: David Herrmann <dh.herrmann@gmail.com> Cc: Sumit Semwal <sumit.semwal@linaro.org> Cc: Daniel Vetter <daniel.vetter@intel.com> CC: linux-media@vger.kernel.org Cc: dri-devel@lists.freedesktop.org Cc: linaro-mm-sig@lists.linaro.org Cc: intel-gfx@lists.freedesktop.org Cc: devel@driverdev.osuosl.org Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: http://patchwork.freedesktop.org/patch/msgid/1458331359-2634-1-git-send-email-chris@chris-wilson.co.uk
2016-03-19 04:02:39 +08:00
return ret;
case DMA_BUF_SET_NAME_A:
case DMA_BUF_SET_NAME_B:
return dma_buf_set_name(dmabuf, (const char __user *)arg);
dma-buf: Add an API for exporting sync files (v14) Modern userspace APIs like Vulkan are built on an explicit synchronization model. This doesn't always play nicely with the implicit synchronization used in the kernel and assumed by X11 and Wayland. The client -> compositor half of the synchronization isn't too bad, at least on intel, because we can control whether or not i915 synchronizes on the buffer and whether or not it's considered written. The harder part is the compositor -> client synchronization when we get the buffer back from the compositor. We're required to be able to provide the client with a VkSemaphore and VkFence representing the point in time where the window system (compositor and/or display) finished using the buffer. With current APIs, it's very hard to do this in such a way that we don't get confused by the Vulkan driver's access of the buffer. In particular, once we tell the kernel that we're rendering to the buffer again, any CPU waits on the buffer or GPU dependencies will wait on some of the client rendering and not just the compositor. This new IOCTL solves this problem by allowing us to get a snapshot of the implicit synchronization state of a given dma-buf in the form of a sync file. It's effectively the same as a poll() or I915_GEM_WAIT only, instead of CPU waiting directly, it encapsulates the wait operation, at the current moment in time, in a sync_file so we can check/wait on it later. As long as the Vulkan driver does the sync_file export from the dma-buf before we re-introduce it for rendering, it will only contain fences from the compositor or display. This allows to accurately turn it into a VkFence or VkSemaphore without any over-synchronization. By making this an ioctl on the dma-buf itself, it allows this new functionality to be used in an entirely driver-agnostic way without having access to a DRM fd. This makes it ideal for use in driver-generic code in Mesa or in a client such as a compositor where the DRM fd may be hard to reach. v2 (Jason Ekstrand): - Use a wrapper dma_fence_array of all fences including the new one when importing an exclusive fence. v3 (Jason Ekstrand): - Lock around setting shared fences as well as exclusive - Mark SIGNAL_SYNC_FILE as a read-write ioctl. - Initialize ret to 0 in dma_buf_wait_sync_file v4 (Jason Ekstrand): - Use the new dma_resv_get_singleton helper v5 (Jason Ekstrand): - Rename the IOCTLs to import/export rather than wait/signal - Drop the WRITE flag and always get/set the exclusive fence v6 (Jason Ekstrand): - Drop the sync_file import as it was all-around sketchy and not nearly as useful as import. - Re-introduce READ/WRITE flag support for export - Rework the commit message v7 (Jason Ekstrand): - Require at least one sync flag - Fix a refcounting bug: dma_resv_get_excl() doesn't take a reference - Use _rcu helpers since we're accessing the dma_resv read-only v8 (Jason Ekstrand): - Return -ENOMEM if the sync_file_create fails - Predicate support on IS_ENABLED(CONFIG_SYNC_FILE) v9 (Jason Ekstrand): - Add documentation for the new ioctl v10 (Jason Ekstrand): - Go back to dma_buf_sync_file as the ioctl struct name v11 (Daniel Vetter): - Go back to dma_buf_export_sync_file as the ioctl struct name - Better kerneldoc describing what the read/write flags do v12 (Christian König): - Document why we chose to make it an ioctl on dma-buf v13 (Jason Ekstrand): - Rebase on Christian König's fence rework v14 (Daniel Vetter & Christian König): - Use dma_rev_usage_rw to get the properly inverted usage to pass to dma_resv_get_singleton() - Clean up the sync_file and fd if copy_to_user() fails Signed-off-by: Jason Ekstrand <jason@jlekstrand.net> Signed-off-by: Jason Ekstrand <jason.ekstrand@intel.com> Signed-off-by: Jason Ekstrand <jason.ekstrand@collabora.com> Acked-by: Simon Ser <contact@emersion.fr> Reviewed-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: Sumit Semwal <sumit.semwal@linaro.org> Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Signed-off-by: Simon Ser <contact@emersion.fr> Link: https://patchwork.freedesktop.org/patch/msgid/20220608152142.14495-2-jason@jlekstrand.net
2022-06-08 23:21:41 +08:00
#if IS_ENABLED(CONFIG_SYNC_FILE)
case DMA_BUF_IOCTL_EXPORT_SYNC_FILE:
return dma_buf_export_sync_file(dmabuf, (void __user *)arg);
dma-buf: Add an API for importing sync files (v10) This patch is analogous to the previous sync file export patch in that it allows you to import a sync_file into a dma-buf. Unlike the previous patch, however, this does add genuinely new functionality to dma-buf. Without this, the only way to attach a sync_file to a dma-buf is to submit a batch to your driver of choice which waits on the sync_file and claims to write to the dma-buf. Even if said batch is a no-op, a submit is typically way more overhead than just attaching a fence. A submit may also imply extra synchronization with other work because it happens on a hardware queue. In the Vulkan world, this is useful for dealing with the out-fence from vkQueuePresent. Current Linux window-systems (X11, Wayland, etc.) all rely on dma-buf implicit sync. Since Vulkan is an explicit sync API, we get a set of fences (VkSemaphores) in vkQueuePresent and have to stash those as an exclusive (write) fence on the dma-buf. We handle it in Mesa today with the above mentioned dummy submit trick. This ioctl would allow us to set it directly without the dummy submit. This may also open up possibilities for GPU drivers to move away from implicit sync for their kernel driver uAPI and instead provide sync files and rely on dma-buf import/export for communicating with other implicit sync clients. We make the explicit choice here to only allow setting RW fences which translates to an exclusive fence on the dma_resv. There's no use for read-only fences for communicating with other implicit sync userspace and any such attempts are likely to be racy at best. When we got to insert the RW fence, the actual fence we set as the new exclusive fence is a combination of the sync_file provided by the user and all the other fences on the dma_resv. This ensures that the newly added exclusive fence will never signal before the old one would have and ensures that we don't break any dma_resv contracts. We require userspace to specify RW in the flags for symmetry with the export ioctl and in case we ever want to support read fences in the future. There is one downside here that's worth documenting: If two clients writing to the same dma-buf using this API race with each other, their actions on the dma-buf may happen in parallel or in an undefined order. Both with and without this API, the pattern is the same: Collect all the fences on dma-buf, submit work which depends on said fences, and then set a new exclusive (write) fence on the dma-buf which depends on said work. The difference is that, when it's all handled by the GPU driver's submit ioctl, the three operations happen atomically under the dma_resv lock. If two userspace submits race, one will happen before the other. You aren't guaranteed which but you are guaranteed that they're strictly ordered. If userspace manages the fences itself, then these three operations happen separately and the two render operations may happen genuinely in parallel or get interleaved. However, this is a case of userspace racing with itself. As long as we ensure userspace can't back the kernel into a corner, it should be fine. v2 (Jason Ekstrand): - Use a wrapper dma_fence_array of all fences including the new one when importing an exclusive fence. v3 (Jason Ekstrand): - Lock around setting shared fences as well as exclusive - Mark SIGNAL_SYNC_FILE as a read-write ioctl. - Initialize ret to 0 in dma_buf_wait_sync_file v4 (Jason Ekstrand): - Use the new dma_resv_get_singleton helper v5 (Jason Ekstrand): - Rename the IOCTLs to import/export rather than wait/signal - Drop the WRITE flag and always get/set the exclusive fence v6 (Jason Ekstrand): - Split import and export into separate patches - New commit message v7 (Daniel Vetter): - Fix the uapi header to use the right struct in the ioctl - Use a separate dma_buf_import_sync_file struct - Add kerneldoc for dma_buf_import_sync_file v8 (Jason Ekstrand): - Rebase on Christian König's fence rework v9 (Daniel Vetter): - Fix -EINVAL checks for the flags parameter - Add documentation about read/write fences - Add documentation about the expected usage of import/export and specifically call out the possible userspace race. v10 (Simon Ser): - Fix a typo in the docs Signed-off-by: Jason Ekstrand <jason@jlekstrand.net> Signed-off-by: Jason Ekstrand <jason.ekstrand@intel.com> Signed-off-by: Jason Ekstrand <jason.ekstrand@collabora.com> Reviewed-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: Sumit Semwal <sumit.semwal@linaro.org> Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Signed-off-by: Simon Ser <contact@emersion.fr> Link: https://patchwork.freedesktop.org/patch/msgid/20220608152142.14495-3-jason@jlekstrand.net
2022-06-08 23:21:42 +08:00
case DMA_BUF_IOCTL_IMPORT_SYNC_FILE:
return dma_buf_import_sync_file(dmabuf, (const void __user *)arg);
dma-buf: Add an API for exporting sync files (v14) Modern userspace APIs like Vulkan are built on an explicit synchronization model. This doesn't always play nicely with the implicit synchronization used in the kernel and assumed by X11 and Wayland. The client -> compositor half of the synchronization isn't too bad, at least on intel, because we can control whether or not i915 synchronizes on the buffer and whether or not it's considered written. The harder part is the compositor -> client synchronization when we get the buffer back from the compositor. We're required to be able to provide the client with a VkSemaphore and VkFence representing the point in time where the window system (compositor and/or display) finished using the buffer. With current APIs, it's very hard to do this in such a way that we don't get confused by the Vulkan driver's access of the buffer. In particular, once we tell the kernel that we're rendering to the buffer again, any CPU waits on the buffer or GPU dependencies will wait on some of the client rendering and not just the compositor. This new IOCTL solves this problem by allowing us to get a snapshot of the implicit synchronization state of a given dma-buf in the form of a sync file. It's effectively the same as a poll() or I915_GEM_WAIT only, instead of CPU waiting directly, it encapsulates the wait operation, at the current moment in time, in a sync_file so we can check/wait on it later. As long as the Vulkan driver does the sync_file export from the dma-buf before we re-introduce it for rendering, it will only contain fences from the compositor or display. This allows to accurately turn it into a VkFence or VkSemaphore without any over-synchronization. By making this an ioctl on the dma-buf itself, it allows this new functionality to be used in an entirely driver-agnostic way without having access to a DRM fd. This makes it ideal for use in driver-generic code in Mesa or in a client such as a compositor where the DRM fd may be hard to reach. v2 (Jason Ekstrand): - Use a wrapper dma_fence_array of all fences including the new one when importing an exclusive fence. v3 (Jason Ekstrand): - Lock around setting shared fences as well as exclusive - Mark SIGNAL_SYNC_FILE as a read-write ioctl. - Initialize ret to 0 in dma_buf_wait_sync_file v4 (Jason Ekstrand): - Use the new dma_resv_get_singleton helper v5 (Jason Ekstrand): - Rename the IOCTLs to import/export rather than wait/signal - Drop the WRITE flag and always get/set the exclusive fence v6 (Jason Ekstrand): - Drop the sync_file import as it was all-around sketchy and not nearly as useful as import. - Re-introduce READ/WRITE flag support for export - Rework the commit message v7 (Jason Ekstrand): - Require at least one sync flag - Fix a refcounting bug: dma_resv_get_excl() doesn't take a reference - Use _rcu helpers since we're accessing the dma_resv read-only v8 (Jason Ekstrand): - Return -ENOMEM if the sync_file_create fails - Predicate support on IS_ENABLED(CONFIG_SYNC_FILE) v9 (Jason Ekstrand): - Add documentation for the new ioctl v10 (Jason Ekstrand): - Go back to dma_buf_sync_file as the ioctl struct name v11 (Daniel Vetter): - Go back to dma_buf_export_sync_file as the ioctl struct name - Better kerneldoc describing what the read/write flags do v12 (Christian König): - Document why we chose to make it an ioctl on dma-buf v13 (Jason Ekstrand): - Rebase on Christian König's fence rework v14 (Daniel Vetter & Christian König): - Use dma_rev_usage_rw to get the properly inverted usage to pass to dma_resv_get_singleton() - Clean up the sync_file and fd if copy_to_user() fails Signed-off-by: Jason Ekstrand <jason@jlekstrand.net> Signed-off-by: Jason Ekstrand <jason.ekstrand@intel.com> Signed-off-by: Jason Ekstrand <jason.ekstrand@collabora.com> Acked-by: Simon Ser <contact@emersion.fr> Reviewed-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Cc: Sumit Semwal <sumit.semwal@linaro.org> Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com> Signed-off-by: Simon Ser <contact@emersion.fr> Link: https://patchwork.freedesktop.org/patch/msgid/20220608152142.14495-2-jason@jlekstrand.net
2022-06-08 23:21:41 +08:00
#endif
dma-buf: Add ioctls to allow userspace to flush The userspace might need some sort of cache coherency management e.g. when CPU and GPU domains are being accessed through dma-buf at the same time. To circumvent this problem there are begin/end coherency markers, that forward directly to existing dma-buf device drivers vfunc hooks. Userspace can make use of those markers through the DMA_BUF_IOCTL_SYNC ioctl. The sequence would be used like following: - mmap dma-buf fd - for each drawing/upload cycle in CPU 1. SYNC_START ioctl, 2. read/write to mmap area 3. SYNC_END ioctl. This can be repeated as often as you want (with the new data being consumed by the GPU or say scanout device) - munmap once you don't need the buffer any more v2 (Tiago): Fix header file type names (u64 -> __u64) v3 (Tiago): Add documentation. Use enum dma_buf_sync_flags to the begin/end dma-buf functions. Check for overflows in start/length. v4 (Tiago): use 2d regions for sync. v5 (Tiago): forget about 2d regions (v4); use _IOW in DMA_BUF_IOCTL_SYNC and remove range information from struct dma_buf_sync. v6 (Tiago): use __u64 structured padded flags instead enum. Adjust documentation about the recommendation on using sync ioctls. v7 (Tiago): Alex' nit on flags definition and being even more wording in the doc about sync usage. v9 (Tiago): remove useless is_dma_buf_file check. Fix sync.flags conditionals and its mask order check. Add <linux/types.h> include in dma-buf.h. Cc: Ville Syrjälä <ville.syrjala@linux.intel.com> Cc: David Herrmann <dh.herrmann@gmail.com> Cc: Sumit Semwal <sumit.semwal@linaro.org> Reviewed-by: Stéphane Marchesin <marcheu@chromium.org> Signed-off-by: Daniel Vetter <daniel.vetter@intel.com> Signed-off-by: Tiago Vignatti <tiago.vignatti@intel.com> Reviewed-by: David Herrmann <dh.herrmann@gmail.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: http://patchwork.freedesktop.org/patch/msgid/1455228291-29640-1-git-send-email-tiago.vignatti@intel.com
2016-02-12 06:04:51 +08:00
default:
return -ENOTTY;
}
}
static void dma_buf_show_fdinfo(struct seq_file *m, struct file *file)
{
struct dma_buf *dmabuf = file->private_data;
seq_printf(m, "size:\t%zu\n", dmabuf->size);
/* Don't count the temporary reference taken inside procfs seq_show */
seq_printf(m, "count:\t%ld\n", file_count(dmabuf->file) - 1);
seq_printf(m, "exp_name:\t%s\n", dmabuf->exp_name);
dmabuf: use spinlock to access dmabuf->name There exists a sleep-while-atomic bug while accessing the dmabuf->name under mutex in the dmabuffs_dname(). This is caused from the SELinux permissions checks on a process where it tries to validate the inherited files from fork() by traversing them through iterate_fd() (which traverse files under spin_lock) and call match_file(security/selinux/hooks.c) where the permission checks happen. This audit information is logged using dump_common_audit_data() where it calls d_path() to get the file path name. If the file check happen on the dmabuf's fd, then it ends up in ->dmabuffs_dname() and use mutex to access dmabuf->name. The flow will be like below: flush_unauthorized_files() iterate_fd() spin_lock() --> Start of the atomic section. match_file() file_has_perm() avc_has_perm() avc_audit() slow_avc_audit() common_lsm_audit() dump_common_audit_data() audit_log_d_path() d_path() dmabuffs_dname() mutex_lock()--> Sleep while atomic. Call trace captured (on 4.19 kernels) is below: ___might_sleep+0x204/0x208 __might_sleep+0x50/0x88 __mutex_lock_common+0x5c/0x1068 __mutex_lock_common+0x5c/0x1068 mutex_lock_nested+0x40/0x50 dmabuffs_dname+0xa0/0x170 d_path+0x84/0x290 audit_log_d_path+0x74/0x130 common_lsm_audit+0x334/0x6e8 slow_avc_audit+0xb8/0xf8 avc_has_perm+0x154/0x218 file_has_perm+0x70/0x180 match_file+0x60/0x78 iterate_fd+0x128/0x168 selinux_bprm_committing_creds+0x178/0x248 security_bprm_committing_creds+0x30/0x48 install_exec_creds+0x1c/0x68 load_elf_binary+0x3a4/0x14e0 search_binary_handler+0xb0/0x1e0 So, use spinlock to access dmabuf->name to avoid sleep-while-atomic. Cc: <stable@vger.kernel.org> [5.3+] Signed-off-by: Charan Teja Kalla <charante@codeaurora.org> Reviewed-by: Michael J. Ruhl <michael.j.ruhl@intel.com> Acked-by: Christian König <christian.koenig@amd.com> [sumits: added comment to spinlock_t definition to avoid warning] Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Link: https://patchwork.freedesktop.org/patch/msgid/a83e7f0d-4e54-9848-4b58-e1acdbe06735@codeaurora.org
2020-06-19 19:57:19 +08:00
spin_lock(&dmabuf->name_lock);
if (dmabuf->name)
seq_printf(m, "name:\t%s\n", dmabuf->name);
dmabuf: use spinlock to access dmabuf->name There exists a sleep-while-atomic bug while accessing the dmabuf->name under mutex in the dmabuffs_dname(). This is caused from the SELinux permissions checks on a process where it tries to validate the inherited files from fork() by traversing them through iterate_fd() (which traverse files under spin_lock) and call match_file(security/selinux/hooks.c) where the permission checks happen. This audit information is logged using dump_common_audit_data() where it calls d_path() to get the file path name. If the file check happen on the dmabuf's fd, then it ends up in ->dmabuffs_dname() and use mutex to access dmabuf->name. The flow will be like below: flush_unauthorized_files() iterate_fd() spin_lock() --> Start of the atomic section. match_file() file_has_perm() avc_has_perm() avc_audit() slow_avc_audit() common_lsm_audit() dump_common_audit_data() audit_log_d_path() d_path() dmabuffs_dname() mutex_lock()--> Sleep while atomic. Call trace captured (on 4.19 kernels) is below: ___might_sleep+0x204/0x208 __might_sleep+0x50/0x88 __mutex_lock_common+0x5c/0x1068 __mutex_lock_common+0x5c/0x1068 mutex_lock_nested+0x40/0x50 dmabuffs_dname+0xa0/0x170 d_path+0x84/0x290 audit_log_d_path+0x74/0x130 common_lsm_audit+0x334/0x6e8 slow_avc_audit+0xb8/0xf8 avc_has_perm+0x154/0x218 file_has_perm+0x70/0x180 match_file+0x60/0x78 iterate_fd+0x128/0x168 selinux_bprm_committing_creds+0x178/0x248 security_bprm_committing_creds+0x30/0x48 install_exec_creds+0x1c/0x68 load_elf_binary+0x3a4/0x14e0 search_binary_handler+0xb0/0x1e0 So, use spinlock to access dmabuf->name to avoid sleep-while-atomic. Cc: <stable@vger.kernel.org> [5.3+] Signed-off-by: Charan Teja Kalla <charante@codeaurora.org> Reviewed-by: Michael J. Ruhl <michael.j.ruhl@intel.com> Acked-by: Christian König <christian.koenig@amd.com> [sumits: added comment to spinlock_t definition to avoid warning] Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Link: https://patchwork.freedesktop.org/patch/msgid/a83e7f0d-4e54-9848-4b58-e1acdbe06735@codeaurora.org
2020-06-19 19:57:19 +08:00
spin_unlock(&dmabuf->name_lock);
}
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
static const struct file_operations dma_buf_fops = {
dmabuf: fix use-after-free of dmabuf's file->f_inode It is observed 'use-after-free' on the dmabuf's file->f_inode with the race between closing the dmabuf file and reading the dmabuf's debug info. Consider the below scenario where P1 is closing the dma_buf file and P2 is reading the dma_buf's debug info in the system: P1 P2 dma_buf_debug_show() dma_buf_put() __fput() file->f_op->release() dput() .... dentry_unlink_inode() iput(dentry->d_inode) (where the inode is freed) mutex_lock(&db_list.lock) read 'dma_buf->file->f_inode' (the same inode is freed by P1) mutex_unlock(&db_list.lock) dentry->d_op->d_release()--> dma_buf_release() ..... mutex_lock(&db_list.lock) removes the dmabuf from the list mutex_unlock(&db_list.lock) In the above scenario, when dma_buf_put() is called on a dma_buf, it first frees the dma_buf's file->f_inode(=dentry->d_inode) and then removes this dma_buf from the system db_list. In between P2 traversing the db_list tries to access this dma_buf's file->f_inode that was freed by P1 which is a use-after-free case. Since, __fput() calls f_op->release first and then later calls the d_op->d_release, move the dma_buf's db_list removal from d_release() to f_op->release(). This ensures that dma_buf's file->f_inode is not accessed after it is released. Cc: <stable@vger.kernel.org> # 5.4.x- Fixes: 4ab59c3c638c ("dma-buf: Move dma_buf_release() from fops to dentry_ops") Acked-by: Christian König <christian.koenig@amd.com> Signed-off-by: Charan Teja Reddy <charante@codeaurora.org> Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de> Link: https://patchwork.freedesktop.org/patch/msgid/1609857399-31549-1-git-send-email-charante@codeaurora.org
2021-01-05 22:36:39 +08:00
.release = dma_buf_file_release,
dma-buf: mmap support Compared to Rob Clark's RFC I've ditched the prepare/finish hooks and corresponding ioctls on the dma_buf file. The major reason for that is that many people seem to be under the impression that this is also for synchronization with outstanding asynchronous processsing. I'm pretty massively opposed to this because: - It boils down reinventing a new rather general-purpose userspace synchronization interface. If we look at things like futexes, this is hard to get right. - Furthermore a lot of kernel code has to interact with this synchronization primitive. This smells a look like the dri1 hw_lock, a horror show I prefer not to reinvent. - Even more fun is that multiple different subsystems would interact here, so we have plenty of opportunities to create funny deadlock scenarios. I think synchronization is a wholesale different problem from data sharing and should be tackled as an orthogonal problem. Now we could demand that prepare/finish may only ensure cache coherency (as Rob intended), but that runs up into the next problem: We not only need mmap support to facilitate sw-only processing nodes in a pipeline (without jumping through hoops by importing the dma_buf into some sw-access only importer), which allows for a nicer ION->dma-buf upgrade path for existing Android userspace. We also need mmap support for existing importing subsystems to support existing userspace libraries. And a loot of these subsystems are expected to export coherent userspace mappings. So prepare/finish can only ever be optional and the exporter /needs/ to support coherent mappings. Given that mmap access is always somewhat fallback-y in nature I've decided to drop this optimization, instead of just making it optional. If we demonstrate a clear need for this, supported by benchmark results, we can always add it in again later as an optional extension. Other differences compared to Rob's RFC is the above mentioned support for mapping a dma-buf through facilities provided by the importer. Which results in mmap support no longer being optional. Note that this dma-buf mmap patch does _not_ support every possible insanity an existing subsystem could pull of with mmap: Because it does not allow to intercept pagefaults and shoot down ptes importing subsystems can't add some magic of their own at these points (e.g. to automatically synchronize with outstanding rendering or set up some special resources). I've done a cursory read through a few mmap implementions of various subsytems and I'm hopeful that we can avoid this (and the complexity it'd bring with it). Additonally I've extended the documentation a bit to explain the hows and whys of this mmap extension. In case we ever want to add support for explicitly cache maneged userspace mmap with a prepare/finish ioctl pair, we could specify that userspace needs to mmap a different part of the dma_buf, e.g. the range starting at dma_buf->size up to dma_buf->size*2. This works because the size of a dma_buf is invariant over it's lifetime. The exporter would obviously need to fall back to coherent mappings for both ranges if a legacy clients maps the coherent range and the architecture cannot suppor conflicting caching policies. Also, this would obviously be optional and userspace needs to be able to fall back to coherent mappings. v2: - Spelling fixes from Rob Clark. - Compile fix for !DMA_BUF from Rob Clark. - Extend commit message to explain how explicitly cache managed mmap support could be added later. - Extend the documentation with implementations notes for exporters that need to manually fake coherency. v3: - dma_buf pointer initialization goof-up noticed by Rebecca Schultz Zavin. Cc: Rob Clark <rob.clark@linaro.org> Cc: Rebecca Schultz Zavin <rebecca@android.com> Acked-by: Rob Clark <rob.clark@linaro.org> Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org>
2012-04-24 17:08:52 +08:00
.mmap = dma_buf_mmap_internal,
.llseek = dma_buf_llseek,
.poll = dma_buf_poll,
dma-buf: Add ioctls to allow userspace to flush The userspace might need some sort of cache coherency management e.g. when CPU and GPU domains are being accessed through dma-buf at the same time. To circumvent this problem there are begin/end coherency markers, that forward directly to existing dma-buf device drivers vfunc hooks. Userspace can make use of those markers through the DMA_BUF_IOCTL_SYNC ioctl. The sequence would be used like following: - mmap dma-buf fd - for each drawing/upload cycle in CPU 1. SYNC_START ioctl, 2. read/write to mmap area 3. SYNC_END ioctl. This can be repeated as often as you want (with the new data being consumed by the GPU or say scanout device) - munmap once you don't need the buffer any more v2 (Tiago): Fix header file type names (u64 -> __u64) v3 (Tiago): Add documentation. Use enum dma_buf_sync_flags to the begin/end dma-buf functions. Check for overflows in start/length. v4 (Tiago): use 2d regions for sync. v5 (Tiago): forget about 2d regions (v4); use _IOW in DMA_BUF_IOCTL_SYNC and remove range information from struct dma_buf_sync. v6 (Tiago): use __u64 structured padded flags instead enum. Adjust documentation about the recommendation on using sync ioctls. v7 (Tiago): Alex' nit on flags definition and being even more wording in the doc about sync usage. v9 (Tiago): remove useless is_dma_buf_file check. Fix sync.flags conditionals and its mask order check. Add <linux/types.h> include in dma-buf.h. Cc: Ville Syrjälä <ville.syrjala@linux.intel.com> Cc: David Herrmann <dh.herrmann@gmail.com> Cc: Sumit Semwal <sumit.semwal@linaro.org> Reviewed-by: Stéphane Marchesin <marcheu@chromium.org> Signed-off-by: Daniel Vetter <daniel.vetter@intel.com> Signed-off-by: Tiago Vignatti <tiago.vignatti@intel.com> Reviewed-by: David Herrmann <dh.herrmann@gmail.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: http://patchwork.freedesktop.org/patch/msgid/1455228291-29640-1-git-send-email-tiago.vignatti@intel.com
2016-02-12 06:04:51 +08:00
.unlocked_ioctl = dma_buf_ioctl,
.compat_ioctl = compat_ptr_ioctl,
.show_fdinfo = dma_buf_show_fdinfo,
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
};
/*
* is_dma_buf_file - Check if struct file* is associated with dma_buf
*/
static inline int is_dma_buf_file(struct file *file)
{
return file->f_op == &dma_buf_fops;
}
static struct file *dma_buf_getfile(size_t size, int flags)
{
dma-buf: ensure unique directory name for dmabuf stats The dmabuf file uses get_next_ino()(through dma_buf_getfile() -> alloc_anon_inode()) to get an inode number and uses the same as a directory name under /sys/kernel/dmabuf/buffers/<ino>. This directory is used to collect the dmabuf stats and it is created through dma_buf_stats_setup(). At current, failure to create this directory entry can make the dma_buf_export() to fail. Now, as the get_next_ino() can definitely give a repetitive inode no causing the directory entry creation to fail with -EEXIST. This is a problem on the systems where dmabuf stats functionality is enabled on the production builds can make the dma_buf_export(), though the dmabuf memory is allocated successfully, to fail just because it couldn't create stats entry. This issue we are able to see on the snapdragon system within 13 days where there already exists a directory with inode no "122602" so dma_buf_stats_setup() failed with -EEXIST as it is trying to create the same directory entry. To make the dentry name as unique, use the dmabuf fs specific inode which is based on the simple atomic variable increment. There is tmpfs subsystem too which relies on its own inode generation rather than relying on the get_next_ino() for the same reason of avoiding the duplicate inodes[1]. [1] https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/patch/?id=e809d5f0b5c912fe981dce738f3283b2010665f0 Signed-off-by: Charan Teja Kalla <quic_charante@quicinc.com> Cc: <stable@vger.kernel.org> # 5.15.x+ Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Reviewed-by: Christian König <christian.koenig@amd.com> Link: https://patchwork.freedesktop.org/patch/msgid/1652441296-1986-1-git-send-email-quic_charante@quicinc.com Signed-off-by: Christian König <christian.koenig@amd.com>
2022-05-13 19:28:16 +08:00
static atomic64_t dmabuf_inode = ATOMIC64_INIT(0);
struct inode *inode = alloc_anon_inode(dma_buf_mnt->mnt_sb);
struct file *file;
if (IS_ERR(inode))
return ERR_CAST(inode);
inode->i_size = size;
inode_set_bytes(inode, size);
dma-buf: ensure unique directory name for dmabuf stats The dmabuf file uses get_next_ino()(through dma_buf_getfile() -> alloc_anon_inode()) to get an inode number and uses the same as a directory name under /sys/kernel/dmabuf/buffers/<ino>. This directory is used to collect the dmabuf stats and it is created through dma_buf_stats_setup(). At current, failure to create this directory entry can make the dma_buf_export() to fail. Now, as the get_next_ino() can definitely give a repetitive inode no causing the directory entry creation to fail with -EEXIST. This is a problem on the systems where dmabuf stats functionality is enabled on the production builds can make the dma_buf_export(), though the dmabuf memory is allocated successfully, to fail just because it couldn't create stats entry. This issue we are able to see on the snapdragon system within 13 days where there already exists a directory with inode no "122602" so dma_buf_stats_setup() failed with -EEXIST as it is trying to create the same directory entry. To make the dentry name as unique, use the dmabuf fs specific inode which is based on the simple atomic variable increment. There is tmpfs subsystem too which relies on its own inode generation rather than relying on the get_next_ino() for the same reason of avoiding the duplicate inodes[1]. [1] https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/patch/?id=e809d5f0b5c912fe981dce738f3283b2010665f0 Signed-off-by: Charan Teja Kalla <quic_charante@quicinc.com> Cc: <stable@vger.kernel.org> # 5.15.x+ Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Reviewed-by: Christian König <christian.koenig@amd.com> Link: https://patchwork.freedesktop.org/patch/msgid/1652441296-1986-1-git-send-email-quic_charante@quicinc.com Signed-off-by: Christian König <christian.koenig@amd.com>
2022-05-13 19:28:16 +08:00
/*
* The ->i_ino acquired from get_next_ino() is not unique thus
* not suitable for using it as dentry name by dmabuf stats.
* Override ->i_ino with the unique and dmabuffs specific
* value.
*/
inode->i_ino = atomic64_add_return(1, &dmabuf_inode);
flags &= O_ACCMODE | O_NONBLOCK;
file = alloc_file_pseudo(inode, dma_buf_mnt, "dmabuf",
flags, &dma_buf_fops);
if (IS_ERR(file))
goto err_alloc_file;
return file;
err_alloc_file:
iput(inode);
return file;
}
/**
* DOC: dma buf device access
*
* For device DMA access to a shared DMA buffer the usual sequence of operations
* is fairly simple:
*
* 1. The exporter defines his exporter instance using
* DEFINE_DMA_BUF_EXPORT_INFO() and calls dma_buf_export() to wrap a private
* buffer object into a &dma_buf. It then exports that &dma_buf to userspace
* as a file descriptor by calling dma_buf_fd().
*
* 2. Userspace passes this file-descriptors to all drivers it wants this buffer
* to share with: First the file descriptor is converted to a &dma_buf using
* dma_buf_get(). Then the buffer is attached to the device using
* dma_buf_attach().
*
* Up to this stage the exporter is still free to migrate or reallocate the
* backing storage.
*
* 3. Once the buffer is attached to all devices userspace can initiate DMA
* access to the shared buffer. In the kernel this is done by calling
* dma_buf_map_attachment() and dma_buf_unmap_attachment().
*
* 4. Once a driver is done with a shared buffer it needs to call
* dma_buf_detach() (after cleaning up any mappings) and then release the
* reference acquired with dma_buf_get() by calling dma_buf_put().
*
* For the detailed semantics exporters are expected to implement see
* &dma_buf_ops.
*/
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
/**
* dma_buf_export - Creates a new dma_buf, and associates an anon file
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
* with this buffer, so it can be exported.
* Also connect the allocator specific data and ops to the buffer.
* Additionally, provide a name string for exporter; useful in debugging.
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
*
* @exp_info: [in] holds all the export related information provided
* by the exporter. see &struct dma_buf_export_info
* for further details.
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
*
* Returns, on success, a newly created struct dma_buf object, which wraps the
* supplied private data and operations for struct dma_buf_ops. On either
* missing ops, or error in allocating struct dma_buf, will return negative
* error.
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
*
* For most cases the easiest way to create @exp_info is through the
* %DEFINE_DMA_BUF_EXPORT_INFO macro.
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
*/
struct dma_buf *dma_buf_export(const struct dma_buf_export_info *exp_info)
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
{
struct dma_buf *dmabuf;
struct dma_resv *resv = exp_info->resv;
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
struct file *file;
size_t alloc_size = sizeof(struct dma_buf);
int ret;
if (WARN_ON(!exp_info->priv || !exp_info->ops
|| !exp_info->ops->map_dma_buf
|| !exp_info->ops->unmap_dma_buf
|| !exp_info->ops->release))
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
return ERR_PTR(-EINVAL);
dma-buf: change DMA-buf locking convention v3 This patch is a stripped down version of the locking changes necessary to support dynamic DMA-buf handling. It adds a dynamic flag for both importers as well as exporters so that drivers can choose if they want the reservation object locked or unlocked during mapping of attachments. For compatibility between drivers we cache the DMA-buf mapping during attaching an importer as soon as exporter/importer disagree on the dynamic handling. Issues and solutions we considered: - We can't change all existing drivers, and existing improters have strong opinions about which locks they're holding while calling dma_buf_attachment_map/unmap. Exporters also have strong opinions about which locks they can acquire in their ->map/unmap callbacks, levaing no room for change. The solution to avoid this was to move the actual map/unmap out from this call, into the attach/detach callbacks, and cache the mapping. This works because drivers don't call attach/detach from deep within their code callchains (like deep in memory management code called from cs/execbuf ioctl), but directly from the fd2handle implementation. - The caching has some troubles on some soc drivers, which set other modes than DMA_BIDIRECTIONAL. We can't have 2 incompatible mappings, and we can't re-create the mapping at _map time due to the above locking fun. We very carefuly step around that by only caching at attach time if the dynamic mode between importer/expoert mismatches. - There's been quite some discussion on dma-buf mappings which need active cache management, which would all break down when caching, plus we don't have explicit flush operations on the attachment side. The solution to this was to shrug and keep the current discrepancy between what the dma-buf docs claim and what implementations do, with the hope that the begin/end_cpu_access hooks are good enough and that all necessary flushing to keep device mappings consistent will be done there. v2: cleanup set_name merge, improve kerneldoc v3: update commit message, kerneldoc and cleanup _debug_show() Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/336788/
2018-07-03 22:42:26 +08:00
if (WARN_ON(exp_info->ops->cache_sgt_mapping &&
(exp_info->ops->pin || exp_info->ops->unpin)))
dma-buf: change DMA-buf locking convention v3 This patch is a stripped down version of the locking changes necessary to support dynamic DMA-buf handling. It adds a dynamic flag for both importers as well as exporters so that drivers can choose if they want the reservation object locked or unlocked during mapping of attachments. For compatibility between drivers we cache the DMA-buf mapping during attaching an importer as soon as exporter/importer disagree on the dynamic handling. Issues and solutions we considered: - We can't change all existing drivers, and existing improters have strong opinions about which locks they're holding while calling dma_buf_attachment_map/unmap. Exporters also have strong opinions about which locks they can acquire in their ->map/unmap callbacks, levaing no room for change. The solution to avoid this was to move the actual map/unmap out from this call, into the attach/detach callbacks, and cache the mapping. This works because drivers don't call attach/detach from deep within their code callchains (like deep in memory management code called from cs/execbuf ioctl), but directly from the fd2handle implementation. - The caching has some troubles on some soc drivers, which set other modes than DMA_BIDIRECTIONAL. We can't have 2 incompatible mappings, and we can't re-create the mapping at _map time due to the above locking fun. We very carefuly step around that by only caching at attach time if the dynamic mode between importer/expoert mismatches. - There's been quite some discussion on dma-buf mappings which need active cache management, which would all break down when caching, plus we don't have explicit flush operations on the attachment side. The solution to this was to shrug and keep the current discrepancy between what the dma-buf docs claim and what implementations do, with the hope that the begin/end_cpu_access hooks are good enough and that all necessary flushing to keep device mappings consistent will be done there. v2: cleanup set_name merge, improve kerneldoc v3: update commit message, kerneldoc and cleanup _debug_show() Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/336788/
2018-07-03 22:42:26 +08:00
return ERR_PTR(-EINVAL);
if (WARN_ON(!exp_info->ops->pin != !exp_info->ops->unpin))
dma-buf: change DMA-buf locking convention v3 This patch is a stripped down version of the locking changes necessary to support dynamic DMA-buf handling. It adds a dynamic flag for both importers as well as exporters so that drivers can choose if they want the reservation object locked or unlocked during mapping of attachments. For compatibility between drivers we cache the DMA-buf mapping during attaching an importer as soon as exporter/importer disagree on the dynamic handling. Issues and solutions we considered: - We can't change all existing drivers, and existing improters have strong opinions about which locks they're holding while calling dma_buf_attachment_map/unmap. Exporters also have strong opinions about which locks they can acquire in their ->map/unmap callbacks, levaing no room for change. The solution to avoid this was to move the actual map/unmap out from this call, into the attach/detach callbacks, and cache the mapping. This works because drivers don't call attach/detach from deep within their code callchains (like deep in memory management code called from cs/execbuf ioctl), but directly from the fd2handle implementation. - The caching has some troubles on some soc drivers, which set other modes than DMA_BIDIRECTIONAL. We can't have 2 incompatible mappings, and we can't re-create the mapping at _map time due to the above locking fun. We very carefuly step around that by only caching at attach time if the dynamic mode between importer/expoert mismatches. - There's been quite some discussion on dma-buf mappings which need active cache management, which would all break down when caching, plus we don't have explicit flush operations on the attachment side. The solution to this was to shrug and keep the current discrepancy between what the dma-buf docs claim and what implementations do, with the hope that the begin/end_cpu_access hooks are good enough and that all necessary flushing to keep device mappings consistent will be done there. v2: cleanup set_name merge, improve kerneldoc v3: update commit message, kerneldoc and cleanup _debug_show() Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/336788/
2018-07-03 22:42:26 +08:00
return ERR_PTR(-EINVAL);
if (!try_module_get(exp_info->owner))
return ERR_PTR(-ENOENT);
file = dma_buf_getfile(exp_info->size, exp_info->flags);
if (IS_ERR(file)) {
ret = PTR_ERR(file);
goto err_module;
}
if (!exp_info->resv)
alloc_size += sizeof(struct dma_resv);
else
/* prevent &dma_buf[1] == dma_buf->resv */
alloc_size += 1;
dmabuf = kzalloc(alloc_size, GFP_KERNEL);
if (!dmabuf) {
ret = -ENOMEM;
goto err_file;
}
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
dmabuf->priv = exp_info->priv;
dmabuf->ops = exp_info->ops;
dmabuf->size = exp_info->size;
dmabuf->exp_name = exp_info->exp_name;
dmabuf->owner = exp_info->owner;
dmabuf: use spinlock to access dmabuf->name There exists a sleep-while-atomic bug while accessing the dmabuf->name under mutex in the dmabuffs_dname(). This is caused from the SELinux permissions checks on a process where it tries to validate the inherited files from fork() by traversing them through iterate_fd() (which traverse files under spin_lock) and call match_file(security/selinux/hooks.c) where the permission checks happen. This audit information is logged using dump_common_audit_data() where it calls d_path() to get the file path name. If the file check happen on the dmabuf's fd, then it ends up in ->dmabuffs_dname() and use mutex to access dmabuf->name. The flow will be like below: flush_unauthorized_files() iterate_fd() spin_lock() --> Start of the atomic section. match_file() file_has_perm() avc_has_perm() avc_audit() slow_avc_audit() common_lsm_audit() dump_common_audit_data() audit_log_d_path() d_path() dmabuffs_dname() mutex_lock()--> Sleep while atomic. Call trace captured (on 4.19 kernels) is below: ___might_sleep+0x204/0x208 __might_sleep+0x50/0x88 __mutex_lock_common+0x5c/0x1068 __mutex_lock_common+0x5c/0x1068 mutex_lock_nested+0x40/0x50 dmabuffs_dname+0xa0/0x170 d_path+0x84/0x290 audit_log_d_path+0x74/0x130 common_lsm_audit+0x334/0x6e8 slow_avc_audit+0xb8/0xf8 avc_has_perm+0x154/0x218 file_has_perm+0x70/0x180 match_file+0x60/0x78 iterate_fd+0x128/0x168 selinux_bprm_committing_creds+0x178/0x248 security_bprm_committing_creds+0x30/0x48 install_exec_creds+0x1c/0x68 load_elf_binary+0x3a4/0x14e0 search_binary_handler+0xb0/0x1e0 So, use spinlock to access dmabuf->name to avoid sleep-while-atomic. Cc: <stable@vger.kernel.org> [5.3+] Signed-off-by: Charan Teja Kalla <charante@codeaurora.org> Reviewed-by: Michael J. Ruhl <michael.j.ruhl@intel.com> Acked-by: Christian König <christian.koenig@amd.com> [sumits: added comment to spinlock_t definition to avoid warning] Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Link: https://patchwork.freedesktop.org/patch/msgid/a83e7f0d-4e54-9848-4b58-e1acdbe06735@codeaurora.org
2020-06-19 19:57:19 +08:00
spin_lock_init(&dmabuf->name_lock);
init_waitqueue_head(&dmabuf->poll);
dmabuf->cb_in.poll = dmabuf->cb_out.poll = &dmabuf->poll;
dmabuf->cb_in.active = dmabuf->cb_out.active = 0;
INIT_LIST_HEAD(&dmabuf->attachments);
if (!resv) {
dmabuf->resv = (struct dma_resv *)&dmabuf[1];
dma_resv_init(dmabuf->resv);
} else {
dmabuf->resv = resv;
}
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
ret = dma_buf_stats_setup(dmabuf, file);
if (ret)
goto err_dmabuf;
file->private_data = dmabuf;
file->f_path.dentry->d_fsdata = dmabuf;
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
dmabuf->file = file;
__dma_buf_debugfs_list_add(dmabuf);
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
return dmabuf;
err_dmabuf:
if (!resv)
dma_resv_fini(dmabuf->resv);
kfree(dmabuf);
err_file:
fput(file);
err_module:
module_put(exp_info->owner);
return ERR_PTR(ret);
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
}
EXPORT_SYMBOL_NS_GPL(dma_buf_export, DMA_BUF);
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
/**
* dma_buf_fd - returns a file descriptor for the given struct dma_buf
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
* @dmabuf: [in] pointer to dma_buf for which fd is required.
* @flags: [in] flags to give to fd
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
*
* On success, returns an associated 'fd'. Else, returns error.
*/
int dma_buf_fd(struct dma_buf *dmabuf, int flags)
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
{
int fd;
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
if (!dmabuf || !dmabuf->file)
return -EINVAL;
fd = get_unused_fd_flags(flags);
if (fd < 0)
return fd;
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
fd_install(fd, dmabuf->file);
return fd;
}
EXPORT_SYMBOL_NS_GPL(dma_buf_fd, DMA_BUF);
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
/**
* dma_buf_get - returns the struct dma_buf related to an fd
* @fd: [in] fd associated with the struct dma_buf to be returned
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
*
* On success, returns the struct dma_buf associated with an fd; uses
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
* file's refcounting done by fget to increase refcount. returns ERR_PTR
* otherwise.
*/
struct dma_buf *dma_buf_get(int fd)
{
struct file *file;
file = fget(fd);
if (!file)
return ERR_PTR(-EBADF);
if (!is_dma_buf_file(file)) {
fput(file);
return ERR_PTR(-EINVAL);
}
return file->private_data;
}
EXPORT_SYMBOL_NS_GPL(dma_buf_get, DMA_BUF);
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
/**
* dma_buf_put - decreases refcount of the buffer
* @dmabuf: [in] buffer to reduce refcount of
*
* Uses file's refcounting done implicitly by fput().
*
* If, as a result of this call, the refcount becomes 0, the 'release' file
* operation related to this fd is called. It calls &dma_buf_ops.release vfunc
* in turn, and frees the memory allocated for dmabuf when exported.
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
*/
void dma_buf_put(struct dma_buf *dmabuf)
{
if (WARN_ON(!dmabuf || !dmabuf->file))
return;
fput(dmabuf->file);
}
EXPORT_SYMBOL_NS_GPL(dma_buf_put, DMA_BUF);
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
dma-buf: Add debug option We have too many people abusing the struct page they can get at but really shouldn't in importers. Aside from that the backing page might simply not exist (for dynamic p2p mappings) looking at it and using it e.g. for mmap can also wreak the page handling of the exporter completely. Importers really must go through the proper interface like dma_buf_mmap for everything. I'm semi-tempted to enforce this for dynamic importers since those really have no excuse at all to break the rules. Unfortuantely we can't store the right pointers somewhere safe to make sure we oops on something recognizable, so best is to just wrangle them a bit by flipping all the bits. At least on x86 kernel addresses have all their high bits sets and the struct page array is fairly low in the kernel mapping, so flipping all the bits gives us a very high pointer in userspace and hence excellent chances for an invalid dereference. v2: Add a note to the @map_dma_buf hook that exporters shouldn't do fancy caching tricks, which would blow up with this address scrambling trick here (Chris) Enable by default when CONFIG_DMA_API_DEBUG is enabled. v3: Only one copy of the mangle/unmangle code (Christian) v4: #ifdef, not #if (0day) v5: sg_table can also be an ERR_PTR (Chris, Christian) Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Daniel Vetter <daniel.vetter@intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Sumit Semwal <sumit.semwal@linaro.org> Cc: "Christian König" <christian.koenig@amd.com> Cc: David Stevens <stevensd@chromium.org> Cc: linux-media@vger.kernel.org Cc: linaro-mm-sig@lists.linaro.org Link: https://patchwork.freedesktop.org/patch/msgid/20210115164739.3958206-1-daniel.vetter@ffwll.ch
2021-01-16 00:47:39 +08:00
static void mangle_sg_table(struct sg_table *sg_table)
{
#ifdef CONFIG_DMABUF_DEBUG
int i;
struct scatterlist *sg;
/* To catch abuse of the underlying struct page by importers mix
* up the bits, but take care to preserve the low SG_ bits to
* not corrupt the sgt. The mixing is undone in __unmap_dma_buf
* before passing the sgt back to the exporter. */
for_each_sgtable_sg(sg_table, sg, i)
sg->page_link ^= ~0xffUL;
#endif
}
static struct sg_table * __map_dma_buf(struct dma_buf_attachment *attach,
enum dma_data_direction direction)
{
struct sg_table *sg_table;
signed long ret;
dma-buf: Add debug option We have too many people abusing the struct page they can get at but really shouldn't in importers. Aside from that the backing page might simply not exist (for dynamic p2p mappings) looking at it and using it e.g. for mmap can also wreak the page handling of the exporter completely. Importers really must go through the proper interface like dma_buf_mmap for everything. I'm semi-tempted to enforce this for dynamic importers since those really have no excuse at all to break the rules. Unfortuantely we can't store the right pointers somewhere safe to make sure we oops on something recognizable, so best is to just wrangle them a bit by flipping all the bits. At least on x86 kernel addresses have all their high bits sets and the struct page array is fairly low in the kernel mapping, so flipping all the bits gives us a very high pointer in userspace and hence excellent chances for an invalid dereference. v2: Add a note to the @map_dma_buf hook that exporters shouldn't do fancy caching tricks, which would blow up with this address scrambling trick here (Chris) Enable by default when CONFIG_DMA_API_DEBUG is enabled. v3: Only one copy of the mangle/unmangle code (Christian) v4: #ifdef, not #if (0day) v5: sg_table can also be an ERR_PTR (Chris, Christian) Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Daniel Vetter <daniel.vetter@intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Sumit Semwal <sumit.semwal@linaro.org> Cc: "Christian König" <christian.koenig@amd.com> Cc: David Stevens <stevensd@chromium.org> Cc: linux-media@vger.kernel.org Cc: linaro-mm-sig@lists.linaro.org Link: https://patchwork.freedesktop.org/patch/msgid/20210115164739.3958206-1-daniel.vetter@ffwll.ch
2021-01-16 00:47:39 +08:00
sg_table = attach->dmabuf->ops->map_dma_buf(attach, direction);
if (IS_ERR_OR_NULL(sg_table))
return sg_table;
if (!dma_buf_attachment_is_dynamic(attach)) {
ret = dma_resv_wait_timeout(attach->dmabuf->resv,
DMA_RESV_USAGE_KERNEL, true,
MAX_SCHEDULE_TIMEOUT);
if (ret < 0) {
attach->dmabuf->ops->unmap_dma_buf(attach, sg_table,
direction);
return ERR_PTR(ret);
}
}
dma-buf: Add debug option We have too many people abusing the struct page they can get at but really shouldn't in importers. Aside from that the backing page might simply not exist (for dynamic p2p mappings) looking at it and using it e.g. for mmap can also wreak the page handling of the exporter completely. Importers really must go through the proper interface like dma_buf_mmap for everything. I'm semi-tempted to enforce this for dynamic importers since those really have no excuse at all to break the rules. Unfortuantely we can't store the right pointers somewhere safe to make sure we oops on something recognizable, so best is to just wrangle them a bit by flipping all the bits. At least on x86 kernel addresses have all their high bits sets and the struct page array is fairly low in the kernel mapping, so flipping all the bits gives us a very high pointer in userspace and hence excellent chances for an invalid dereference. v2: Add a note to the @map_dma_buf hook that exporters shouldn't do fancy caching tricks, which would blow up with this address scrambling trick here (Chris) Enable by default when CONFIG_DMA_API_DEBUG is enabled. v3: Only one copy of the mangle/unmangle code (Christian) v4: #ifdef, not #if (0day) v5: sg_table can also be an ERR_PTR (Chris, Christian) Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Daniel Vetter <daniel.vetter@intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Sumit Semwal <sumit.semwal@linaro.org> Cc: "Christian König" <christian.koenig@amd.com> Cc: David Stevens <stevensd@chromium.org> Cc: linux-media@vger.kernel.org Cc: linaro-mm-sig@lists.linaro.org Link: https://patchwork.freedesktop.org/patch/msgid/20210115164739.3958206-1-daniel.vetter@ffwll.ch
2021-01-16 00:47:39 +08:00
mangle_sg_table(sg_table);
dma-buf: Add debug option We have too many people abusing the struct page they can get at but really shouldn't in importers. Aside from that the backing page might simply not exist (for dynamic p2p mappings) looking at it and using it e.g. for mmap can also wreak the page handling of the exporter completely. Importers really must go through the proper interface like dma_buf_mmap for everything. I'm semi-tempted to enforce this for dynamic importers since those really have no excuse at all to break the rules. Unfortuantely we can't store the right pointers somewhere safe to make sure we oops on something recognizable, so best is to just wrangle them a bit by flipping all the bits. At least on x86 kernel addresses have all their high bits sets and the struct page array is fairly low in the kernel mapping, so flipping all the bits gives us a very high pointer in userspace and hence excellent chances for an invalid dereference. v2: Add a note to the @map_dma_buf hook that exporters shouldn't do fancy caching tricks, which would blow up with this address scrambling trick here (Chris) Enable by default when CONFIG_DMA_API_DEBUG is enabled. v3: Only one copy of the mangle/unmangle code (Christian) v4: #ifdef, not #if (0day) v5: sg_table can also be an ERR_PTR (Chris, Christian) Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Daniel Vetter <daniel.vetter@intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Sumit Semwal <sumit.semwal@linaro.org> Cc: "Christian König" <christian.koenig@amd.com> Cc: David Stevens <stevensd@chromium.org> Cc: linux-media@vger.kernel.org Cc: linaro-mm-sig@lists.linaro.org Link: https://patchwork.freedesktop.org/patch/msgid/20210115164739.3958206-1-daniel.vetter@ffwll.ch
2021-01-16 00:47:39 +08:00
return sg_table;
}
/**
* DOC: locking convention
*
* In order to avoid deadlock situations between dma-buf exports and importers,
* all dma-buf API users must follow the common dma-buf locking convention.
*
* Convention for importers
*
* 1. Importers must hold the dma-buf reservation lock when calling these
* functions:
*
* - dma_buf_pin()
* - dma_buf_unpin()
* - dma_buf_map_attachment()
* - dma_buf_unmap_attachment()
* - dma_buf_vmap()
* - dma_buf_vunmap()
*
* 2. Importers must not hold the dma-buf reservation lock when calling these
* functions:
*
* - dma_buf_attach()
* - dma_buf_dynamic_attach()
* - dma_buf_detach()
* - dma_buf_export()
* - dma_buf_fd()
* - dma_buf_get()
* - dma_buf_put()
* - dma_buf_mmap()
* - dma_buf_begin_cpu_access()
* - dma_buf_end_cpu_access()
* - dma_buf_map_attachment_unlocked()
* - dma_buf_unmap_attachment_unlocked()
* - dma_buf_vmap_unlocked()
* - dma_buf_vunmap_unlocked()
*
* Convention for exporters
*
* 1. These &dma_buf_ops callbacks are invoked with unlocked dma-buf
* reservation and exporter can take the lock:
*
* - &dma_buf_ops.attach()
* - &dma_buf_ops.detach()
* - &dma_buf_ops.release()
* - &dma_buf_ops.begin_cpu_access()
* - &dma_buf_ops.end_cpu_access()
* - &dma_buf_ops.mmap()
*
* 2. These &dma_buf_ops callbacks are invoked with locked dma-buf
* reservation and exporter can't take the lock:
*
* - &dma_buf_ops.pin()
* - &dma_buf_ops.unpin()
* - &dma_buf_ops.map_dma_buf()
* - &dma_buf_ops.unmap_dma_buf()
* - &dma_buf_ops.vmap()
* - &dma_buf_ops.vunmap()
*
* 3. Exporters must hold the dma-buf reservation lock when calling these
* functions:
*
* - dma_buf_move_notify()
*/
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
/**
* dma_buf_dynamic_attach - Add the device to dma_buf's attachments list
dma-buf: change DMA-buf locking convention v3 This patch is a stripped down version of the locking changes necessary to support dynamic DMA-buf handling. It adds a dynamic flag for both importers as well as exporters so that drivers can choose if they want the reservation object locked or unlocked during mapping of attachments. For compatibility between drivers we cache the DMA-buf mapping during attaching an importer as soon as exporter/importer disagree on the dynamic handling. Issues and solutions we considered: - We can't change all existing drivers, and existing improters have strong opinions about which locks they're holding while calling dma_buf_attachment_map/unmap. Exporters also have strong opinions about which locks they can acquire in their ->map/unmap callbacks, levaing no room for change. The solution to avoid this was to move the actual map/unmap out from this call, into the attach/detach callbacks, and cache the mapping. This works because drivers don't call attach/detach from deep within their code callchains (like deep in memory management code called from cs/execbuf ioctl), but directly from the fd2handle implementation. - The caching has some troubles on some soc drivers, which set other modes than DMA_BIDIRECTIONAL. We can't have 2 incompatible mappings, and we can't re-create the mapping at _map time due to the above locking fun. We very carefuly step around that by only caching at attach time if the dynamic mode between importer/expoert mismatches. - There's been quite some discussion on dma-buf mappings which need active cache management, which would all break down when caching, plus we don't have explicit flush operations on the attachment side. The solution to this was to shrug and keep the current discrepancy between what the dma-buf docs claim and what implementations do, with the hope that the begin/end_cpu_access hooks are good enough and that all necessary flushing to keep device mappings consistent will be done there. v2: cleanup set_name merge, improve kerneldoc v3: update commit message, kerneldoc and cleanup _debug_show() Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/336788/
2018-07-03 22:42:26 +08:00
* @dmabuf: [in] buffer to attach device to.
* @dev: [in] device to be attached.
* @importer_ops: [in] importer operations for the attachment
* @importer_priv: [in] importer private pointer for the attachment
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
*
* Returns struct dma_buf_attachment pointer for this attachment. Attachments
* must be cleaned up by calling dma_buf_detach().
*
* Optionally this calls &dma_buf_ops.attach to allow device-specific attach
* functionality.
*
* Returns:
*
* A pointer to newly created &dma_buf_attachment on success, or a negative
* error code wrapped into a pointer on failure.
*
* Note that this can fail if the backing storage of @dmabuf is in a place not
* accessible to @dev, and cannot be moved to a more suitable place. This is
* indicated with the error code -EBUSY.
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
*/
dma-buf: change DMA-buf locking convention v3 This patch is a stripped down version of the locking changes necessary to support dynamic DMA-buf handling. It adds a dynamic flag for both importers as well as exporters so that drivers can choose if they want the reservation object locked or unlocked during mapping of attachments. For compatibility between drivers we cache the DMA-buf mapping during attaching an importer as soon as exporter/importer disagree on the dynamic handling. Issues and solutions we considered: - We can't change all existing drivers, and existing improters have strong opinions about which locks they're holding while calling dma_buf_attachment_map/unmap. Exporters also have strong opinions about which locks they can acquire in their ->map/unmap callbacks, levaing no room for change. The solution to avoid this was to move the actual map/unmap out from this call, into the attach/detach callbacks, and cache the mapping. This works because drivers don't call attach/detach from deep within their code callchains (like deep in memory management code called from cs/execbuf ioctl), but directly from the fd2handle implementation. - The caching has some troubles on some soc drivers, which set other modes than DMA_BIDIRECTIONAL. We can't have 2 incompatible mappings, and we can't re-create the mapping at _map time due to the above locking fun. We very carefuly step around that by only caching at attach time if the dynamic mode between importer/expoert mismatches. - There's been quite some discussion on dma-buf mappings which need active cache management, which would all break down when caching, plus we don't have explicit flush operations on the attachment side. The solution to this was to shrug and keep the current discrepancy between what the dma-buf docs claim and what implementations do, with the hope that the begin/end_cpu_access hooks are good enough and that all necessary flushing to keep device mappings consistent will be done there. v2: cleanup set_name merge, improve kerneldoc v3: update commit message, kerneldoc and cleanup _debug_show() Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/336788/
2018-07-03 22:42:26 +08:00
struct dma_buf_attachment *
dma_buf_dynamic_attach(struct dma_buf *dmabuf, struct device *dev,
dma-buf: add dynamic DMA-buf handling v15 On the exporter side we add optional explicit pinning callbacks. Which are called when the importer doesn't implement dynamic handling, move notification or need the DMA-buf locked in place for its use case. On the importer side we add an optional move_notify callback. This callback is used by the exporter to inform the importers that their mappings should be destroyed as soon as possible. This allows the exporter to provide the mappings without the need to pin the backing store. v2: don't try to invalidate mappings when the callback is NULL, lock the reservation obj while using the attachments, add helper to set the callback v3: move flag for invalidation support into the DMA-buf, use new attach_info structure to set the callback v4: use importer_priv field instead of mangling exporter priv. v5: drop invalidation_supported flag v6: squash together with pin/unpin changes v7: pin/unpin takes an attachment now v8: nuke dma_buf_attachment_(map|unmap)_locked, everything is now handled backward compatible v9: always cache when export/importer don't agree on dynamic handling v10: minimal style cleanup v11: drop automatically re-entry avoidance v12: rename callback to move_notify v13: add might_lock in appropriate places v14: rebase on separated locking change v15: add EXPERIMENTAL flag, some more code comments Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/353993/?series=73646&rev=1
2018-07-03 22:42:26 +08:00
const struct dma_buf_attach_ops *importer_ops,
void *importer_priv)
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
{
struct dma_buf_attachment *attach;
int ret;
if (WARN_ON(!dmabuf || !dev))
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
return ERR_PTR(-EINVAL);
if (WARN_ON(importer_ops && !importer_ops->move_notify))
return ERR_PTR(-EINVAL);
attach = kzalloc(sizeof(*attach), GFP_KERNEL);
if (!attach)
return ERR_PTR(-ENOMEM);
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
attach->dev = dev;
attach->dmabuf = dmabuf;
if (importer_ops)
attach->peer2peer = importer_ops->allow_peer2peer;
dma-buf: add dynamic DMA-buf handling v15 On the exporter side we add optional explicit pinning callbacks. Which are called when the importer doesn't implement dynamic handling, move notification or need the DMA-buf locked in place for its use case. On the importer side we add an optional move_notify callback. This callback is used by the exporter to inform the importers that their mappings should be destroyed as soon as possible. This allows the exporter to provide the mappings without the need to pin the backing store. v2: don't try to invalidate mappings when the callback is NULL, lock the reservation obj while using the attachments, add helper to set the callback v3: move flag for invalidation support into the DMA-buf, use new attach_info structure to set the callback v4: use importer_priv field instead of mangling exporter priv. v5: drop invalidation_supported flag v6: squash together with pin/unpin changes v7: pin/unpin takes an attachment now v8: nuke dma_buf_attachment_(map|unmap)_locked, everything is now handled backward compatible v9: always cache when export/importer don't agree on dynamic handling v10: minimal style cleanup v11: drop automatically re-entry avoidance v12: rename callback to move_notify v13: add might_lock in appropriate places v14: rebase on separated locking change v15: add EXPERIMENTAL flag, some more code comments Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/353993/?series=73646&rev=1
2018-07-03 22:42:26 +08:00
attach->importer_ops = importer_ops;
attach->importer_priv = importer_priv;
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
if (dmabuf->ops->attach) {
ret = dmabuf->ops->attach(dmabuf, attach);
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
if (ret)
goto err_attach;
}
dma-buf: change DMA-buf locking convention v3 This patch is a stripped down version of the locking changes necessary to support dynamic DMA-buf handling. It adds a dynamic flag for both importers as well as exporters so that drivers can choose if they want the reservation object locked or unlocked during mapping of attachments. For compatibility between drivers we cache the DMA-buf mapping during attaching an importer as soon as exporter/importer disagree on the dynamic handling. Issues and solutions we considered: - We can't change all existing drivers, and existing improters have strong opinions about which locks they're holding while calling dma_buf_attachment_map/unmap. Exporters also have strong opinions about which locks they can acquire in their ->map/unmap callbacks, levaing no room for change. The solution to avoid this was to move the actual map/unmap out from this call, into the attach/detach callbacks, and cache the mapping. This works because drivers don't call attach/detach from deep within their code callchains (like deep in memory management code called from cs/execbuf ioctl), but directly from the fd2handle implementation. - The caching has some troubles on some soc drivers, which set other modes than DMA_BIDIRECTIONAL. We can't have 2 incompatible mappings, and we can't re-create the mapping at _map time due to the above locking fun. We very carefuly step around that by only caching at attach time if the dynamic mode between importer/expoert mismatches. - There's been quite some discussion on dma-buf mappings which need active cache management, which would all break down when caching, plus we don't have explicit flush operations on the attachment side. The solution to this was to shrug and keep the current discrepancy between what the dma-buf docs claim and what implementations do, with the hope that the begin/end_cpu_access hooks are good enough and that all necessary flushing to keep device mappings consistent will be done there. v2: cleanup set_name merge, improve kerneldoc v3: update commit message, kerneldoc and cleanup _debug_show() Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/336788/
2018-07-03 22:42:26 +08:00
dma_resv_lock(dmabuf->resv, NULL);
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
list_add(&attach->node, &dmabuf->attachments);
dma-buf: change DMA-buf locking convention v3 This patch is a stripped down version of the locking changes necessary to support dynamic DMA-buf handling. It adds a dynamic flag for both importers as well as exporters so that drivers can choose if they want the reservation object locked or unlocked during mapping of attachments. For compatibility between drivers we cache the DMA-buf mapping during attaching an importer as soon as exporter/importer disagree on the dynamic handling. Issues and solutions we considered: - We can't change all existing drivers, and existing improters have strong opinions about which locks they're holding while calling dma_buf_attachment_map/unmap. Exporters also have strong opinions about which locks they can acquire in their ->map/unmap callbacks, levaing no room for change. The solution to avoid this was to move the actual map/unmap out from this call, into the attach/detach callbacks, and cache the mapping. This works because drivers don't call attach/detach from deep within their code callchains (like deep in memory management code called from cs/execbuf ioctl), but directly from the fd2handle implementation. - The caching has some troubles on some soc drivers, which set other modes than DMA_BIDIRECTIONAL. We can't have 2 incompatible mappings, and we can't re-create the mapping at _map time due to the above locking fun. We very carefuly step around that by only caching at attach time if the dynamic mode between importer/expoert mismatches. - There's been quite some discussion on dma-buf mappings which need active cache management, which would all break down when caching, plus we don't have explicit flush operations on the attachment side. The solution to this was to shrug and keep the current discrepancy between what the dma-buf docs claim and what implementations do, with the hope that the begin/end_cpu_access hooks are good enough and that all necessary flushing to keep device mappings consistent will be done there. v2: cleanup set_name merge, improve kerneldoc v3: update commit message, kerneldoc and cleanup _debug_show() Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/336788/
2018-07-03 22:42:26 +08:00
dma_resv_unlock(dmabuf->resv);
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
dma-buf: change DMA-buf locking convention v3 This patch is a stripped down version of the locking changes necessary to support dynamic DMA-buf handling. It adds a dynamic flag for both importers as well as exporters so that drivers can choose if they want the reservation object locked or unlocked during mapping of attachments. For compatibility between drivers we cache the DMA-buf mapping during attaching an importer as soon as exporter/importer disagree on the dynamic handling. Issues and solutions we considered: - We can't change all existing drivers, and existing improters have strong opinions about which locks they're holding while calling dma_buf_attachment_map/unmap. Exporters also have strong opinions about which locks they can acquire in their ->map/unmap callbacks, levaing no room for change. The solution to avoid this was to move the actual map/unmap out from this call, into the attach/detach callbacks, and cache the mapping. This works because drivers don't call attach/detach from deep within their code callchains (like deep in memory management code called from cs/execbuf ioctl), but directly from the fd2handle implementation. - The caching has some troubles on some soc drivers, which set other modes than DMA_BIDIRECTIONAL. We can't have 2 incompatible mappings, and we can't re-create the mapping at _map time due to the above locking fun. We very carefuly step around that by only caching at attach time if the dynamic mode between importer/expoert mismatches. - There's been quite some discussion on dma-buf mappings which need active cache management, which would all break down when caching, plus we don't have explicit flush operations on the attachment side. The solution to this was to shrug and keep the current discrepancy between what the dma-buf docs claim and what implementations do, with the hope that the begin/end_cpu_access hooks are good enough and that all necessary flushing to keep device mappings consistent will be done there. v2: cleanup set_name merge, improve kerneldoc v3: update commit message, kerneldoc and cleanup _debug_show() Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/336788/
2018-07-03 22:42:26 +08:00
/* When either the importer or the exporter can't handle dynamic
* mappings we cache the mapping here to avoid issues with the
* reservation object lock.
*/
if (dma_buf_attachment_is_dynamic(attach) !=
dma_buf_is_dynamic(dmabuf)) {
struct sg_table *sgt;
dma_resv_lock(attach->dmabuf->resv, NULL);
dma-buf: add dynamic DMA-buf handling v15 On the exporter side we add optional explicit pinning callbacks. Which are called when the importer doesn't implement dynamic handling, move notification or need the DMA-buf locked in place for its use case. On the importer side we add an optional move_notify callback. This callback is used by the exporter to inform the importers that their mappings should be destroyed as soon as possible. This allows the exporter to provide the mappings without the need to pin the backing store. v2: don't try to invalidate mappings when the callback is NULL, lock the reservation obj while using the attachments, add helper to set the callback v3: move flag for invalidation support into the DMA-buf, use new attach_info structure to set the callback v4: use importer_priv field instead of mangling exporter priv. v5: drop invalidation_supported flag v6: squash together with pin/unpin changes v7: pin/unpin takes an attachment now v8: nuke dma_buf_attachment_(map|unmap)_locked, everything is now handled backward compatible v9: always cache when export/importer don't agree on dynamic handling v10: minimal style cleanup v11: drop automatically re-entry avoidance v12: rename callback to move_notify v13: add might_lock in appropriate places v14: rebase on separated locking change v15: add EXPERIMENTAL flag, some more code comments Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/353993/?series=73646&rev=1
2018-07-03 22:42:26 +08:00
if (dma_buf_is_dynamic(attach->dmabuf)) {
ret = dmabuf->ops->pin(attach);
dma-buf: add dynamic DMA-buf handling v15 On the exporter side we add optional explicit pinning callbacks. Which are called when the importer doesn't implement dynamic handling, move notification or need the DMA-buf locked in place for its use case. On the importer side we add an optional move_notify callback. This callback is used by the exporter to inform the importers that their mappings should be destroyed as soon as possible. This allows the exporter to provide the mappings without the need to pin the backing store. v2: don't try to invalidate mappings when the callback is NULL, lock the reservation obj while using the attachments, add helper to set the callback v3: move flag for invalidation support into the DMA-buf, use new attach_info structure to set the callback v4: use importer_priv field instead of mangling exporter priv. v5: drop invalidation_supported flag v6: squash together with pin/unpin changes v7: pin/unpin takes an attachment now v8: nuke dma_buf_attachment_(map|unmap)_locked, everything is now handled backward compatible v9: always cache when export/importer don't agree on dynamic handling v10: minimal style cleanup v11: drop automatically re-entry avoidance v12: rename callback to move_notify v13: add might_lock in appropriate places v14: rebase on separated locking change v15: add EXPERIMENTAL flag, some more code comments Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/353993/?series=73646&rev=1
2018-07-03 22:42:26 +08:00
if (ret)
goto err_unlock;
}
dma-buf: change DMA-buf locking convention v3 This patch is a stripped down version of the locking changes necessary to support dynamic DMA-buf handling. It adds a dynamic flag for both importers as well as exporters so that drivers can choose if they want the reservation object locked or unlocked during mapping of attachments. For compatibility between drivers we cache the DMA-buf mapping during attaching an importer as soon as exporter/importer disagree on the dynamic handling. Issues and solutions we considered: - We can't change all existing drivers, and existing improters have strong opinions about which locks they're holding while calling dma_buf_attachment_map/unmap. Exporters also have strong opinions about which locks they can acquire in their ->map/unmap callbacks, levaing no room for change. The solution to avoid this was to move the actual map/unmap out from this call, into the attach/detach callbacks, and cache the mapping. This works because drivers don't call attach/detach from deep within their code callchains (like deep in memory management code called from cs/execbuf ioctl), but directly from the fd2handle implementation. - The caching has some troubles on some soc drivers, which set other modes than DMA_BIDIRECTIONAL. We can't have 2 incompatible mappings, and we can't re-create the mapping at _map time due to the above locking fun. We very carefuly step around that by only caching at attach time if the dynamic mode between importer/expoert mismatches. - There's been quite some discussion on dma-buf mappings which need active cache management, which would all break down when caching, plus we don't have explicit flush operations on the attachment side. The solution to this was to shrug and keep the current discrepancy between what the dma-buf docs claim and what implementations do, with the hope that the begin/end_cpu_access hooks are good enough and that all necessary flushing to keep device mappings consistent will be done there. v2: cleanup set_name merge, improve kerneldoc v3: update commit message, kerneldoc and cleanup _debug_show() Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/336788/
2018-07-03 22:42:26 +08:00
dma-buf: Add debug option We have too many people abusing the struct page they can get at but really shouldn't in importers. Aside from that the backing page might simply not exist (for dynamic p2p mappings) looking at it and using it e.g. for mmap can also wreak the page handling of the exporter completely. Importers really must go through the proper interface like dma_buf_mmap for everything. I'm semi-tempted to enforce this for dynamic importers since those really have no excuse at all to break the rules. Unfortuantely we can't store the right pointers somewhere safe to make sure we oops on something recognizable, so best is to just wrangle them a bit by flipping all the bits. At least on x86 kernel addresses have all their high bits sets and the struct page array is fairly low in the kernel mapping, so flipping all the bits gives us a very high pointer in userspace and hence excellent chances for an invalid dereference. v2: Add a note to the @map_dma_buf hook that exporters shouldn't do fancy caching tricks, which would blow up with this address scrambling trick here (Chris) Enable by default when CONFIG_DMA_API_DEBUG is enabled. v3: Only one copy of the mangle/unmangle code (Christian) v4: #ifdef, not #if (0day) v5: sg_table can also be an ERR_PTR (Chris, Christian) Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Daniel Vetter <daniel.vetter@intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Sumit Semwal <sumit.semwal@linaro.org> Cc: "Christian König" <christian.koenig@amd.com> Cc: David Stevens <stevensd@chromium.org> Cc: linux-media@vger.kernel.org Cc: linaro-mm-sig@lists.linaro.org Link: https://patchwork.freedesktop.org/patch/msgid/20210115164739.3958206-1-daniel.vetter@ffwll.ch
2021-01-16 00:47:39 +08:00
sgt = __map_dma_buf(attach, DMA_BIDIRECTIONAL);
dma-buf: change DMA-buf locking convention v3 This patch is a stripped down version of the locking changes necessary to support dynamic DMA-buf handling. It adds a dynamic flag for both importers as well as exporters so that drivers can choose if they want the reservation object locked or unlocked during mapping of attachments. For compatibility between drivers we cache the DMA-buf mapping during attaching an importer as soon as exporter/importer disagree on the dynamic handling. Issues and solutions we considered: - We can't change all existing drivers, and existing improters have strong opinions about which locks they're holding while calling dma_buf_attachment_map/unmap. Exporters also have strong opinions about which locks they can acquire in their ->map/unmap callbacks, levaing no room for change. The solution to avoid this was to move the actual map/unmap out from this call, into the attach/detach callbacks, and cache the mapping. This works because drivers don't call attach/detach from deep within their code callchains (like deep in memory management code called from cs/execbuf ioctl), but directly from the fd2handle implementation. - The caching has some troubles on some soc drivers, which set other modes than DMA_BIDIRECTIONAL. We can't have 2 incompatible mappings, and we can't re-create the mapping at _map time due to the above locking fun. We very carefuly step around that by only caching at attach time if the dynamic mode between importer/expoert mismatches. - There's been quite some discussion on dma-buf mappings which need active cache management, which would all break down when caching, plus we don't have explicit flush operations on the attachment side. The solution to this was to shrug and keep the current discrepancy between what the dma-buf docs claim and what implementations do, with the hope that the begin/end_cpu_access hooks are good enough and that all necessary flushing to keep device mappings consistent will be done there. v2: cleanup set_name merge, improve kerneldoc v3: update commit message, kerneldoc and cleanup _debug_show() Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/336788/
2018-07-03 22:42:26 +08:00
if (!sgt)
sgt = ERR_PTR(-ENOMEM);
if (IS_ERR(sgt)) {
ret = PTR_ERR(sgt);
dma-buf: add dynamic DMA-buf handling v15 On the exporter side we add optional explicit pinning callbacks. Which are called when the importer doesn't implement dynamic handling, move notification or need the DMA-buf locked in place for its use case. On the importer side we add an optional move_notify callback. This callback is used by the exporter to inform the importers that their mappings should be destroyed as soon as possible. This allows the exporter to provide the mappings without the need to pin the backing store. v2: don't try to invalidate mappings when the callback is NULL, lock the reservation obj while using the attachments, add helper to set the callback v3: move flag for invalidation support into the DMA-buf, use new attach_info structure to set the callback v4: use importer_priv field instead of mangling exporter priv. v5: drop invalidation_supported flag v6: squash together with pin/unpin changes v7: pin/unpin takes an attachment now v8: nuke dma_buf_attachment_(map|unmap)_locked, everything is now handled backward compatible v9: always cache when export/importer don't agree on dynamic handling v10: minimal style cleanup v11: drop automatically re-entry avoidance v12: rename callback to move_notify v13: add might_lock in appropriate places v14: rebase on separated locking change v15: add EXPERIMENTAL flag, some more code comments Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/353993/?series=73646&rev=1
2018-07-03 22:42:26 +08:00
goto err_unpin;
dma-buf: change DMA-buf locking convention v3 This patch is a stripped down version of the locking changes necessary to support dynamic DMA-buf handling. It adds a dynamic flag for both importers as well as exporters so that drivers can choose if they want the reservation object locked or unlocked during mapping of attachments. For compatibility between drivers we cache the DMA-buf mapping during attaching an importer as soon as exporter/importer disagree on the dynamic handling. Issues and solutions we considered: - We can't change all existing drivers, and existing improters have strong opinions about which locks they're holding while calling dma_buf_attachment_map/unmap. Exporters also have strong opinions about which locks they can acquire in their ->map/unmap callbacks, levaing no room for change. The solution to avoid this was to move the actual map/unmap out from this call, into the attach/detach callbacks, and cache the mapping. This works because drivers don't call attach/detach from deep within their code callchains (like deep in memory management code called from cs/execbuf ioctl), but directly from the fd2handle implementation. - The caching has some troubles on some soc drivers, which set other modes than DMA_BIDIRECTIONAL. We can't have 2 incompatible mappings, and we can't re-create the mapping at _map time due to the above locking fun. We very carefuly step around that by only caching at attach time if the dynamic mode between importer/expoert mismatches. - There's been quite some discussion on dma-buf mappings which need active cache management, which would all break down when caching, plus we don't have explicit flush operations on the attachment side. The solution to this was to shrug and keep the current discrepancy between what the dma-buf docs claim and what implementations do, with the hope that the begin/end_cpu_access hooks are good enough and that all necessary flushing to keep device mappings consistent will be done there. v2: cleanup set_name merge, improve kerneldoc v3: update commit message, kerneldoc and cleanup _debug_show() Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/336788/
2018-07-03 22:42:26 +08:00
}
dma_resv_unlock(attach->dmabuf->resv);
dma-buf: change DMA-buf locking convention v3 This patch is a stripped down version of the locking changes necessary to support dynamic DMA-buf handling. It adds a dynamic flag for both importers as well as exporters so that drivers can choose if they want the reservation object locked or unlocked during mapping of attachments. For compatibility between drivers we cache the DMA-buf mapping during attaching an importer as soon as exporter/importer disagree on the dynamic handling. Issues and solutions we considered: - We can't change all existing drivers, and existing improters have strong opinions about which locks they're holding while calling dma_buf_attachment_map/unmap. Exporters also have strong opinions about which locks they can acquire in their ->map/unmap callbacks, levaing no room for change. The solution to avoid this was to move the actual map/unmap out from this call, into the attach/detach callbacks, and cache the mapping. This works because drivers don't call attach/detach from deep within their code callchains (like deep in memory management code called from cs/execbuf ioctl), but directly from the fd2handle implementation. - The caching has some troubles on some soc drivers, which set other modes than DMA_BIDIRECTIONAL. We can't have 2 incompatible mappings, and we can't re-create the mapping at _map time due to the above locking fun. We very carefuly step around that by only caching at attach time if the dynamic mode between importer/expoert mismatches. - There's been quite some discussion on dma-buf mappings which need active cache management, which would all break down when caching, plus we don't have explicit flush operations on the attachment side. The solution to this was to shrug and keep the current discrepancy between what the dma-buf docs claim and what implementations do, with the hope that the begin/end_cpu_access hooks are good enough and that all necessary flushing to keep device mappings consistent will be done there. v2: cleanup set_name merge, improve kerneldoc v3: update commit message, kerneldoc and cleanup _debug_show() Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/336788/
2018-07-03 22:42:26 +08:00
attach->sgt = sgt;
attach->dir = DMA_BIDIRECTIONAL;
}
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
return attach;
err_attach:
kfree(attach);
return ERR_PTR(ret);
dma-buf: change DMA-buf locking convention v3 This patch is a stripped down version of the locking changes necessary to support dynamic DMA-buf handling. It adds a dynamic flag for both importers as well as exporters so that drivers can choose if they want the reservation object locked or unlocked during mapping of attachments. For compatibility between drivers we cache the DMA-buf mapping during attaching an importer as soon as exporter/importer disagree on the dynamic handling. Issues and solutions we considered: - We can't change all existing drivers, and existing improters have strong opinions about which locks they're holding while calling dma_buf_attachment_map/unmap. Exporters also have strong opinions about which locks they can acquire in their ->map/unmap callbacks, levaing no room for change. The solution to avoid this was to move the actual map/unmap out from this call, into the attach/detach callbacks, and cache the mapping. This works because drivers don't call attach/detach from deep within their code callchains (like deep in memory management code called from cs/execbuf ioctl), but directly from the fd2handle implementation. - The caching has some troubles on some soc drivers, which set other modes than DMA_BIDIRECTIONAL. We can't have 2 incompatible mappings, and we can't re-create the mapping at _map time due to the above locking fun. We very carefuly step around that by only caching at attach time if the dynamic mode between importer/expoert mismatches. - There's been quite some discussion on dma-buf mappings which need active cache management, which would all break down when caching, plus we don't have explicit flush operations on the attachment side. The solution to this was to shrug and keep the current discrepancy between what the dma-buf docs claim and what implementations do, with the hope that the begin/end_cpu_access hooks are good enough and that all necessary flushing to keep device mappings consistent will be done there. v2: cleanup set_name merge, improve kerneldoc v3: update commit message, kerneldoc and cleanup _debug_show() Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/336788/
2018-07-03 22:42:26 +08:00
dma-buf: add dynamic DMA-buf handling v15 On the exporter side we add optional explicit pinning callbacks. Which are called when the importer doesn't implement dynamic handling, move notification or need the DMA-buf locked in place for its use case. On the importer side we add an optional move_notify callback. This callback is used by the exporter to inform the importers that their mappings should be destroyed as soon as possible. This allows the exporter to provide the mappings without the need to pin the backing store. v2: don't try to invalidate mappings when the callback is NULL, lock the reservation obj while using the attachments, add helper to set the callback v3: move flag for invalidation support into the DMA-buf, use new attach_info structure to set the callback v4: use importer_priv field instead of mangling exporter priv. v5: drop invalidation_supported flag v6: squash together with pin/unpin changes v7: pin/unpin takes an attachment now v8: nuke dma_buf_attachment_(map|unmap)_locked, everything is now handled backward compatible v9: always cache when export/importer don't agree on dynamic handling v10: minimal style cleanup v11: drop automatically re-entry avoidance v12: rename callback to move_notify v13: add might_lock in appropriate places v14: rebase on separated locking change v15: add EXPERIMENTAL flag, some more code comments Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/353993/?series=73646&rev=1
2018-07-03 22:42:26 +08:00
err_unpin:
if (dma_buf_is_dynamic(attach->dmabuf))
dmabuf->ops->unpin(attach);
dma-buf: add dynamic DMA-buf handling v15 On the exporter side we add optional explicit pinning callbacks. Which are called when the importer doesn't implement dynamic handling, move notification or need the DMA-buf locked in place for its use case. On the importer side we add an optional move_notify callback. This callback is used by the exporter to inform the importers that their mappings should be destroyed as soon as possible. This allows the exporter to provide the mappings without the need to pin the backing store. v2: don't try to invalidate mappings when the callback is NULL, lock the reservation obj while using the attachments, add helper to set the callback v3: move flag for invalidation support into the DMA-buf, use new attach_info structure to set the callback v4: use importer_priv field instead of mangling exporter priv. v5: drop invalidation_supported flag v6: squash together with pin/unpin changes v7: pin/unpin takes an attachment now v8: nuke dma_buf_attachment_(map|unmap)_locked, everything is now handled backward compatible v9: always cache when export/importer don't agree on dynamic handling v10: minimal style cleanup v11: drop automatically re-entry avoidance v12: rename callback to move_notify v13: add might_lock in appropriate places v14: rebase on separated locking change v15: add EXPERIMENTAL flag, some more code comments Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/353993/?series=73646&rev=1
2018-07-03 22:42:26 +08:00
dma-buf: change DMA-buf locking convention v3 This patch is a stripped down version of the locking changes necessary to support dynamic DMA-buf handling. It adds a dynamic flag for both importers as well as exporters so that drivers can choose if they want the reservation object locked or unlocked during mapping of attachments. For compatibility between drivers we cache the DMA-buf mapping during attaching an importer as soon as exporter/importer disagree on the dynamic handling. Issues and solutions we considered: - We can't change all existing drivers, and existing improters have strong opinions about which locks they're holding while calling dma_buf_attachment_map/unmap. Exporters also have strong opinions about which locks they can acquire in their ->map/unmap callbacks, levaing no room for change. The solution to avoid this was to move the actual map/unmap out from this call, into the attach/detach callbacks, and cache the mapping. This works because drivers don't call attach/detach from deep within their code callchains (like deep in memory management code called from cs/execbuf ioctl), but directly from the fd2handle implementation. - The caching has some troubles on some soc drivers, which set other modes than DMA_BIDIRECTIONAL. We can't have 2 incompatible mappings, and we can't re-create the mapping at _map time due to the above locking fun. We very carefuly step around that by only caching at attach time if the dynamic mode between importer/expoert mismatches. - There's been quite some discussion on dma-buf mappings which need active cache management, which would all break down when caching, plus we don't have explicit flush operations on the attachment side. The solution to this was to shrug and keep the current discrepancy between what the dma-buf docs claim and what implementations do, with the hope that the begin/end_cpu_access hooks are good enough and that all necessary flushing to keep device mappings consistent will be done there. v2: cleanup set_name merge, improve kerneldoc v3: update commit message, kerneldoc and cleanup _debug_show() Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/336788/
2018-07-03 22:42:26 +08:00
err_unlock:
dma_resv_unlock(attach->dmabuf->resv);
dma-buf: change DMA-buf locking convention v3 This patch is a stripped down version of the locking changes necessary to support dynamic DMA-buf handling. It adds a dynamic flag for both importers as well as exporters so that drivers can choose if they want the reservation object locked or unlocked during mapping of attachments. For compatibility between drivers we cache the DMA-buf mapping during attaching an importer as soon as exporter/importer disagree on the dynamic handling. Issues and solutions we considered: - We can't change all existing drivers, and existing improters have strong opinions about which locks they're holding while calling dma_buf_attachment_map/unmap. Exporters also have strong opinions about which locks they can acquire in their ->map/unmap callbacks, levaing no room for change. The solution to avoid this was to move the actual map/unmap out from this call, into the attach/detach callbacks, and cache the mapping. This works because drivers don't call attach/detach from deep within their code callchains (like deep in memory management code called from cs/execbuf ioctl), but directly from the fd2handle implementation. - The caching has some troubles on some soc drivers, which set other modes than DMA_BIDIRECTIONAL. We can't have 2 incompatible mappings, and we can't re-create the mapping at _map time due to the above locking fun. We very carefuly step around that by only caching at attach time if the dynamic mode between importer/expoert mismatches. - There's been quite some discussion on dma-buf mappings which need active cache management, which would all break down when caching, plus we don't have explicit flush operations on the attachment side. The solution to this was to shrug and keep the current discrepancy between what the dma-buf docs claim and what implementations do, with the hope that the begin/end_cpu_access hooks are good enough and that all necessary flushing to keep device mappings consistent will be done there. v2: cleanup set_name merge, improve kerneldoc v3: update commit message, kerneldoc and cleanup _debug_show() Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/336788/
2018-07-03 22:42:26 +08:00
dma_buf_detach(dmabuf, attach);
return ERR_PTR(ret);
}
EXPORT_SYMBOL_NS_GPL(dma_buf_dynamic_attach, DMA_BUF);
dma-buf: change DMA-buf locking convention v3 This patch is a stripped down version of the locking changes necessary to support dynamic DMA-buf handling. It adds a dynamic flag for both importers as well as exporters so that drivers can choose if they want the reservation object locked or unlocked during mapping of attachments. For compatibility between drivers we cache the DMA-buf mapping during attaching an importer as soon as exporter/importer disagree on the dynamic handling. Issues and solutions we considered: - We can't change all existing drivers, and existing improters have strong opinions about which locks they're holding while calling dma_buf_attachment_map/unmap. Exporters also have strong opinions about which locks they can acquire in their ->map/unmap callbacks, levaing no room for change. The solution to avoid this was to move the actual map/unmap out from this call, into the attach/detach callbacks, and cache the mapping. This works because drivers don't call attach/detach from deep within their code callchains (like deep in memory management code called from cs/execbuf ioctl), but directly from the fd2handle implementation. - The caching has some troubles on some soc drivers, which set other modes than DMA_BIDIRECTIONAL. We can't have 2 incompatible mappings, and we can't re-create the mapping at _map time due to the above locking fun. We very carefuly step around that by only caching at attach time if the dynamic mode between importer/expoert mismatches. - There's been quite some discussion on dma-buf mappings which need active cache management, which would all break down when caching, plus we don't have explicit flush operations on the attachment side. The solution to this was to shrug and keep the current discrepancy between what the dma-buf docs claim and what implementations do, with the hope that the begin/end_cpu_access hooks are good enough and that all necessary flushing to keep device mappings consistent will be done there. v2: cleanup set_name merge, improve kerneldoc v3: update commit message, kerneldoc and cleanup _debug_show() Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/336788/
2018-07-03 22:42:26 +08:00
/**
* dma_buf_attach - Wrapper for dma_buf_dynamic_attach
* @dmabuf: [in] buffer to attach device to.
* @dev: [in] device to be attached.
*
* Wrapper to call dma_buf_dynamic_attach() for drivers which still use a static
* mapping.
*/
struct dma_buf_attachment *dma_buf_attach(struct dma_buf *dmabuf,
struct device *dev)
{
dma-buf: add dynamic DMA-buf handling v15 On the exporter side we add optional explicit pinning callbacks. Which are called when the importer doesn't implement dynamic handling, move notification or need the DMA-buf locked in place for its use case. On the importer side we add an optional move_notify callback. This callback is used by the exporter to inform the importers that their mappings should be destroyed as soon as possible. This allows the exporter to provide the mappings without the need to pin the backing store. v2: don't try to invalidate mappings when the callback is NULL, lock the reservation obj while using the attachments, add helper to set the callback v3: move flag for invalidation support into the DMA-buf, use new attach_info structure to set the callback v4: use importer_priv field instead of mangling exporter priv. v5: drop invalidation_supported flag v6: squash together with pin/unpin changes v7: pin/unpin takes an attachment now v8: nuke dma_buf_attachment_(map|unmap)_locked, everything is now handled backward compatible v9: always cache when export/importer don't agree on dynamic handling v10: minimal style cleanup v11: drop automatically re-entry avoidance v12: rename callback to move_notify v13: add might_lock in appropriate places v14: rebase on separated locking change v15: add EXPERIMENTAL flag, some more code comments Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/353993/?series=73646&rev=1
2018-07-03 22:42:26 +08:00
return dma_buf_dynamic_attach(dmabuf, dev, NULL, NULL);
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
}
EXPORT_SYMBOL_NS_GPL(dma_buf_attach, DMA_BUF);
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
dma-buf: Add debug option We have too many people abusing the struct page they can get at but really shouldn't in importers. Aside from that the backing page might simply not exist (for dynamic p2p mappings) looking at it and using it e.g. for mmap can also wreak the page handling of the exporter completely. Importers really must go through the proper interface like dma_buf_mmap for everything. I'm semi-tempted to enforce this for dynamic importers since those really have no excuse at all to break the rules. Unfortuantely we can't store the right pointers somewhere safe to make sure we oops on something recognizable, so best is to just wrangle them a bit by flipping all the bits. At least on x86 kernel addresses have all their high bits sets and the struct page array is fairly low in the kernel mapping, so flipping all the bits gives us a very high pointer in userspace and hence excellent chances for an invalid dereference. v2: Add a note to the @map_dma_buf hook that exporters shouldn't do fancy caching tricks, which would blow up with this address scrambling trick here (Chris) Enable by default when CONFIG_DMA_API_DEBUG is enabled. v3: Only one copy of the mangle/unmangle code (Christian) v4: #ifdef, not #if (0day) v5: sg_table can also be an ERR_PTR (Chris, Christian) Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Daniel Vetter <daniel.vetter@intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Sumit Semwal <sumit.semwal@linaro.org> Cc: "Christian König" <christian.koenig@amd.com> Cc: David Stevens <stevensd@chromium.org> Cc: linux-media@vger.kernel.org Cc: linaro-mm-sig@lists.linaro.org Link: https://patchwork.freedesktop.org/patch/msgid/20210115164739.3958206-1-daniel.vetter@ffwll.ch
2021-01-16 00:47:39 +08:00
static void __unmap_dma_buf(struct dma_buf_attachment *attach,
struct sg_table *sg_table,
enum dma_data_direction direction)
{
/* uses XOR, hence this unmangles */
mangle_sg_table(sg_table);
attach->dmabuf->ops->unmap_dma_buf(attach, sg_table, direction);
}
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
/**
* dma_buf_detach - Remove the given attachment from dmabuf's attachments list
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
* @dmabuf: [in] buffer to detach from.
* @attach: [in] attachment to be detached; is free'd after this call.
*
* Clean up a device attachment obtained by calling dma_buf_attach().
*
* Optionally this calls &dma_buf_ops.detach for device-specific detach.
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
*/
void dma_buf_detach(struct dma_buf *dmabuf, struct dma_buf_attachment *attach)
{
if (WARN_ON(!dmabuf || !attach || dmabuf != attach->dmabuf))
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
return;
dma_resv_lock(dmabuf->resv, NULL);
dma-buf: change DMA-buf locking convention v3 This patch is a stripped down version of the locking changes necessary to support dynamic DMA-buf handling. It adds a dynamic flag for both importers as well as exporters so that drivers can choose if they want the reservation object locked or unlocked during mapping of attachments. For compatibility between drivers we cache the DMA-buf mapping during attaching an importer as soon as exporter/importer disagree on the dynamic handling. Issues and solutions we considered: - We can't change all existing drivers, and existing improters have strong opinions about which locks they're holding while calling dma_buf_attachment_map/unmap. Exporters also have strong opinions about which locks they can acquire in their ->map/unmap callbacks, levaing no room for change. The solution to avoid this was to move the actual map/unmap out from this call, into the attach/detach callbacks, and cache the mapping. This works because drivers don't call attach/detach from deep within their code callchains (like deep in memory management code called from cs/execbuf ioctl), but directly from the fd2handle implementation. - The caching has some troubles on some soc drivers, which set other modes than DMA_BIDIRECTIONAL. We can't have 2 incompatible mappings, and we can't re-create the mapping at _map time due to the above locking fun. We very carefuly step around that by only caching at attach time if the dynamic mode between importer/expoert mismatches. - There's been quite some discussion on dma-buf mappings which need active cache management, which would all break down when caching, plus we don't have explicit flush operations on the attachment side. The solution to this was to shrug and keep the current discrepancy between what the dma-buf docs claim and what implementations do, with the hope that the begin/end_cpu_access hooks are good enough and that all necessary flushing to keep device mappings consistent will be done there. v2: cleanup set_name merge, improve kerneldoc v3: update commit message, kerneldoc and cleanup _debug_show() Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/336788/
2018-07-03 22:42:26 +08:00
if (attach->sgt) {
dma-buf: Add debug option We have too many people abusing the struct page they can get at but really shouldn't in importers. Aside from that the backing page might simply not exist (for dynamic p2p mappings) looking at it and using it e.g. for mmap can also wreak the page handling of the exporter completely. Importers really must go through the proper interface like dma_buf_mmap for everything. I'm semi-tempted to enforce this for dynamic importers since those really have no excuse at all to break the rules. Unfortuantely we can't store the right pointers somewhere safe to make sure we oops on something recognizable, so best is to just wrangle them a bit by flipping all the bits. At least on x86 kernel addresses have all their high bits sets and the struct page array is fairly low in the kernel mapping, so flipping all the bits gives us a very high pointer in userspace and hence excellent chances for an invalid dereference. v2: Add a note to the @map_dma_buf hook that exporters shouldn't do fancy caching tricks, which would blow up with this address scrambling trick here (Chris) Enable by default when CONFIG_DMA_API_DEBUG is enabled. v3: Only one copy of the mangle/unmangle code (Christian) v4: #ifdef, not #if (0day) v5: sg_table can also be an ERR_PTR (Chris, Christian) Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Daniel Vetter <daniel.vetter@intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Sumit Semwal <sumit.semwal@linaro.org> Cc: "Christian König" <christian.koenig@amd.com> Cc: David Stevens <stevensd@chromium.org> Cc: linux-media@vger.kernel.org Cc: linaro-mm-sig@lists.linaro.org Link: https://patchwork.freedesktop.org/patch/msgid/20210115164739.3958206-1-daniel.vetter@ffwll.ch
2021-01-16 00:47:39 +08:00
__unmap_dma_buf(attach, attach->sgt, attach->dir);
if (dma_buf_is_dynamic(attach->dmabuf))
dmabuf->ops->unpin(attach);
dma-buf: change DMA-buf locking convention v3 This patch is a stripped down version of the locking changes necessary to support dynamic DMA-buf handling. It adds a dynamic flag for both importers as well as exporters so that drivers can choose if they want the reservation object locked or unlocked during mapping of attachments. For compatibility between drivers we cache the DMA-buf mapping during attaching an importer as soon as exporter/importer disagree on the dynamic handling. Issues and solutions we considered: - We can't change all existing drivers, and existing improters have strong opinions about which locks they're holding while calling dma_buf_attachment_map/unmap. Exporters also have strong opinions about which locks they can acquire in their ->map/unmap callbacks, levaing no room for change. The solution to avoid this was to move the actual map/unmap out from this call, into the attach/detach callbacks, and cache the mapping. This works because drivers don't call attach/detach from deep within their code callchains (like deep in memory management code called from cs/execbuf ioctl), but directly from the fd2handle implementation. - The caching has some troubles on some soc drivers, which set other modes than DMA_BIDIRECTIONAL. We can't have 2 incompatible mappings, and we can't re-create the mapping at _map time due to the above locking fun. We very carefuly step around that by only caching at attach time if the dynamic mode between importer/expoert mismatches. - There's been quite some discussion on dma-buf mappings which need active cache management, which would all break down when caching, plus we don't have explicit flush operations on the attachment side. The solution to this was to shrug and keep the current discrepancy between what the dma-buf docs claim and what implementations do, with the hope that the begin/end_cpu_access hooks are good enough and that all necessary flushing to keep device mappings consistent will be done there. v2: cleanup set_name merge, improve kerneldoc v3: update commit message, kerneldoc and cleanup _debug_show() Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/336788/
2018-07-03 22:42:26 +08:00
}
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
list_del(&attach->node);
dma-buf: change DMA-buf locking convention v3 This patch is a stripped down version of the locking changes necessary to support dynamic DMA-buf handling. It adds a dynamic flag for both importers as well as exporters so that drivers can choose if they want the reservation object locked or unlocked during mapping of attachments. For compatibility between drivers we cache the DMA-buf mapping during attaching an importer as soon as exporter/importer disagree on the dynamic handling. Issues and solutions we considered: - We can't change all existing drivers, and existing improters have strong opinions about which locks they're holding while calling dma_buf_attachment_map/unmap. Exporters also have strong opinions about which locks they can acquire in their ->map/unmap callbacks, levaing no room for change. The solution to avoid this was to move the actual map/unmap out from this call, into the attach/detach callbacks, and cache the mapping. This works because drivers don't call attach/detach from deep within their code callchains (like deep in memory management code called from cs/execbuf ioctl), but directly from the fd2handle implementation. - The caching has some troubles on some soc drivers, which set other modes than DMA_BIDIRECTIONAL. We can't have 2 incompatible mappings, and we can't re-create the mapping at _map time due to the above locking fun. We very carefuly step around that by only caching at attach time if the dynamic mode between importer/expoert mismatches. - There's been quite some discussion on dma-buf mappings which need active cache management, which would all break down when caching, plus we don't have explicit flush operations on the attachment side. The solution to this was to shrug and keep the current discrepancy between what the dma-buf docs claim and what implementations do, with the hope that the begin/end_cpu_access hooks are good enough and that all necessary flushing to keep device mappings consistent will be done there. v2: cleanup set_name merge, improve kerneldoc v3: update commit message, kerneldoc and cleanup _debug_show() Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/336788/
2018-07-03 22:42:26 +08:00
dma_resv_unlock(dmabuf->resv);
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
if (dmabuf->ops->detach)
dmabuf->ops->detach(dmabuf, attach);
kfree(attach);
}
EXPORT_SYMBOL_NS_GPL(dma_buf_detach, DMA_BUF);
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
dma-buf: add dynamic DMA-buf handling v15 On the exporter side we add optional explicit pinning callbacks. Which are called when the importer doesn't implement dynamic handling, move notification or need the DMA-buf locked in place for its use case. On the importer side we add an optional move_notify callback. This callback is used by the exporter to inform the importers that their mappings should be destroyed as soon as possible. This allows the exporter to provide the mappings without the need to pin the backing store. v2: don't try to invalidate mappings when the callback is NULL, lock the reservation obj while using the attachments, add helper to set the callback v3: move flag for invalidation support into the DMA-buf, use new attach_info structure to set the callback v4: use importer_priv field instead of mangling exporter priv. v5: drop invalidation_supported flag v6: squash together with pin/unpin changes v7: pin/unpin takes an attachment now v8: nuke dma_buf_attachment_(map|unmap)_locked, everything is now handled backward compatible v9: always cache when export/importer don't agree on dynamic handling v10: minimal style cleanup v11: drop automatically re-entry avoidance v12: rename callback to move_notify v13: add might_lock in appropriate places v14: rebase on separated locking change v15: add EXPERIMENTAL flag, some more code comments Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/353993/?series=73646&rev=1
2018-07-03 22:42:26 +08:00
/**
* dma_buf_pin - Lock down the DMA-buf
* @attach: [in] attachment which should be pinned
*
* Only dynamic importers (who set up @attach with dma_buf_dynamic_attach()) may
* call this, and only for limited use cases like scanout and not for temporary
* pin operations. It is not permitted to allow userspace to pin arbitrary
* amounts of buffers through this interface.
*
* Buffers must be unpinned by calling dma_buf_unpin().
*
dma-buf: add dynamic DMA-buf handling v15 On the exporter side we add optional explicit pinning callbacks. Which are called when the importer doesn't implement dynamic handling, move notification or need the DMA-buf locked in place for its use case. On the importer side we add an optional move_notify callback. This callback is used by the exporter to inform the importers that their mappings should be destroyed as soon as possible. This allows the exporter to provide the mappings without the need to pin the backing store. v2: don't try to invalidate mappings when the callback is NULL, lock the reservation obj while using the attachments, add helper to set the callback v3: move flag for invalidation support into the DMA-buf, use new attach_info structure to set the callback v4: use importer_priv field instead of mangling exporter priv. v5: drop invalidation_supported flag v6: squash together with pin/unpin changes v7: pin/unpin takes an attachment now v8: nuke dma_buf_attachment_(map|unmap)_locked, everything is now handled backward compatible v9: always cache when export/importer don't agree on dynamic handling v10: minimal style cleanup v11: drop automatically re-entry avoidance v12: rename callback to move_notify v13: add might_lock in appropriate places v14: rebase on separated locking change v15: add EXPERIMENTAL flag, some more code comments Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/353993/?series=73646&rev=1
2018-07-03 22:42:26 +08:00
* Returns:
* 0 on success, negative error code on failure.
*/
int dma_buf_pin(struct dma_buf_attachment *attach)
{
struct dma_buf *dmabuf = attach->dmabuf;
int ret = 0;
WARN_ON(!dma_buf_attachment_is_dynamic(attach));
dma-buf: add dynamic DMA-buf handling v15 On the exporter side we add optional explicit pinning callbacks. Which are called when the importer doesn't implement dynamic handling, move notification or need the DMA-buf locked in place for its use case. On the importer side we add an optional move_notify callback. This callback is used by the exporter to inform the importers that their mappings should be destroyed as soon as possible. This allows the exporter to provide the mappings without the need to pin the backing store. v2: don't try to invalidate mappings when the callback is NULL, lock the reservation obj while using the attachments, add helper to set the callback v3: move flag for invalidation support into the DMA-buf, use new attach_info structure to set the callback v4: use importer_priv field instead of mangling exporter priv. v5: drop invalidation_supported flag v6: squash together with pin/unpin changes v7: pin/unpin takes an attachment now v8: nuke dma_buf_attachment_(map|unmap)_locked, everything is now handled backward compatible v9: always cache when export/importer don't agree on dynamic handling v10: minimal style cleanup v11: drop automatically re-entry avoidance v12: rename callback to move_notify v13: add might_lock in appropriate places v14: rebase on separated locking change v15: add EXPERIMENTAL flag, some more code comments Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/353993/?series=73646&rev=1
2018-07-03 22:42:26 +08:00
dma_resv_assert_held(dmabuf->resv);
if (dmabuf->ops->pin)
ret = dmabuf->ops->pin(attach);
return ret;
}
EXPORT_SYMBOL_NS_GPL(dma_buf_pin, DMA_BUF);
dma-buf: add dynamic DMA-buf handling v15 On the exporter side we add optional explicit pinning callbacks. Which are called when the importer doesn't implement dynamic handling, move notification or need the DMA-buf locked in place for its use case. On the importer side we add an optional move_notify callback. This callback is used by the exporter to inform the importers that their mappings should be destroyed as soon as possible. This allows the exporter to provide the mappings without the need to pin the backing store. v2: don't try to invalidate mappings when the callback is NULL, lock the reservation obj while using the attachments, add helper to set the callback v3: move flag for invalidation support into the DMA-buf, use new attach_info structure to set the callback v4: use importer_priv field instead of mangling exporter priv. v5: drop invalidation_supported flag v6: squash together with pin/unpin changes v7: pin/unpin takes an attachment now v8: nuke dma_buf_attachment_(map|unmap)_locked, everything is now handled backward compatible v9: always cache when export/importer don't agree on dynamic handling v10: minimal style cleanup v11: drop automatically re-entry avoidance v12: rename callback to move_notify v13: add might_lock in appropriate places v14: rebase on separated locking change v15: add EXPERIMENTAL flag, some more code comments Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/353993/?series=73646&rev=1
2018-07-03 22:42:26 +08:00
/**
* dma_buf_unpin - Unpin a DMA-buf
dma-buf: add dynamic DMA-buf handling v15 On the exporter side we add optional explicit pinning callbacks. Which are called when the importer doesn't implement dynamic handling, move notification or need the DMA-buf locked in place for its use case. On the importer side we add an optional move_notify callback. This callback is used by the exporter to inform the importers that their mappings should be destroyed as soon as possible. This allows the exporter to provide the mappings without the need to pin the backing store. v2: don't try to invalidate mappings when the callback is NULL, lock the reservation obj while using the attachments, add helper to set the callback v3: move flag for invalidation support into the DMA-buf, use new attach_info structure to set the callback v4: use importer_priv field instead of mangling exporter priv. v5: drop invalidation_supported flag v6: squash together with pin/unpin changes v7: pin/unpin takes an attachment now v8: nuke dma_buf_attachment_(map|unmap)_locked, everything is now handled backward compatible v9: always cache when export/importer don't agree on dynamic handling v10: minimal style cleanup v11: drop automatically re-entry avoidance v12: rename callback to move_notify v13: add might_lock in appropriate places v14: rebase on separated locking change v15: add EXPERIMENTAL flag, some more code comments Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/353993/?series=73646&rev=1
2018-07-03 22:42:26 +08:00
* @attach: [in] attachment which should be unpinned
*
* This unpins a buffer pinned by dma_buf_pin() and allows the exporter to move
* any mapping of @attach again and inform the importer through
* &dma_buf_attach_ops.move_notify.
dma-buf: add dynamic DMA-buf handling v15 On the exporter side we add optional explicit pinning callbacks. Which are called when the importer doesn't implement dynamic handling, move notification or need the DMA-buf locked in place for its use case. On the importer side we add an optional move_notify callback. This callback is used by the exporter to inform the importers that their mappings should be destroyed as soon as possible. This allows the exporter to provide the mappings without the need to pin the backing store. v2: don't try to invalidate mappings when the callback is NULL, lock the reservation obj while using the attachments, add helper to set the callback v3: move flag for invalidation support into the DMA-buf, use new attach_info structure to set the callback v4: use importer_priv field instead of mangling exporter priv. v5: drop invalidation_supported flag v6: squash together with pin/unpin changes v7: pin/unpin takes an attachment now v8: nuke dma_buf_attachment_(map|unmap)_locked, everything is now handled backward compatible v9: always cache when export/importer don't agree on dynamic handling v10: minimal style cleanup v11: drop automatically re-entry avoidance v12: rename callback to move_notify v13: add might_lock in appropriate places v14: rebase on separated locking change v15: add EXPERIMENTAL flag, some more code comments Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/353993/?series=73646&rev=1
2018-07-03 22:42:26 +08:00
*/
void dma_buf_unpin(struct dma_buf_attachment *attach)
{
struct dma_buf *dmabuf = attach->dmabuf;
WARN_ON(!dma_buf_attachment_is_dynamic(attach));
dma-buf: add dynamic DMA-buf handling v15 On the exporter side we add optional explicit pinning callbacks. Which are called when the importer doesn't implement dynamic handling, move notification or need the DMA-buf locked in place for its use case. On the importer side we add an optional move_notify callback. This callback is used by the exporter to inform the importers that their mappings should be destroyed as soon as possible. This allows the exporter to provide the mappings without the need to pin the backing store. v2: don't try to invalidate mappings when the callback is NULL, lock the reservation obj while using the attachments, add helper to set the callback v3: move flag for invalidation support into the DMA-buf, use new attach_info structure to set the callback v4: use importer_priv field instead of mangling exporter priv. v5: drop invalidation_supported flag v6: squash together with pin/unpin changes v7: pin/unpin takes an attachment now v8: nuke dma_buf_attachment_(map|unmap)_locked, everything is now handled backward compatible v9: always cache when export/importer don't agree on dynamic handling v10: minimal style cleanup v11: drop automatically re-entry avoidance v12: rename callback to move_notify v13: add might_lock in appropriate places v14: rebase on separated locking change v15: add EXPERIMENTAL flag, some more code comments Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/353993/?series=73646&rev=1
2018-07-03 22:42:26 +08:00
dma_resv_assert_held(dmabuf->resv);
if (dmabuf->ops->unpin)
dmabuf->ops->unpin(attach);
}
EXPORT_SYMBOL_NS_GPL(dma_buf_unpin, DMA_BUF);
dma-buf: add dynamic DMA-buf handling v15 On the exporter side we add optional explicit pinning callbacks. Which are called when the importer doesn't implement dynamic handling, move notification or need the DMA-buf locked in place for its use case. On the importer side we add an optional move_notify callback. This callback is used by the exporter to inform the importers that their mappings should be destroyed as soon as possible. This allows the exporter to provide the mappings without the need to pin the backing store. v2: don't try to invalidate mappings when the callback is NULL, lock the reservation obj while using the attachments, add helper to set the callback v3: move flag for invalidation support into the DMA-buf, use new attach_info structure to set the callback v4: use importer_priv field instead of mangling exporter priv. v5: drop invalidation_supported flag v6: squash together with pin/unpin changes v7: pin/unpin takes an attachment now v8: nuke dma_buf_attachment_(map|unmap)_locked, everything is now handled backward compatible v9: always cache when export/importer don't agree on dynamic handling v10: minimal style cleanup v11: drop automatically re-entry avoidance v12: rename callback to move_notify v13: add might_lock in appropriate places v14: rebase on separated locking change v15: add EXPERIMENTAL flag, some more code comments Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/353993/?series=73646&rev=1
2018-07-03 22:42:26 +08:00
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
/**
* dma_buf_map_attachment - Returns the scatterlist table of the attachment;
* mapped into _device_ address space. Is a wrapper for map_dma_buf() of the
* dma_buf_ops.
* @attach: [in] attachment whose scatterlist is to be returned
* @direction: [in] direction of DMA transfer
*
* Returns sg_table containing the scatterlist to be returned; returns ERR_PTR
* on error. May return -EINTR if it is interrupted by a signal.
*
* On success, the DMA addresses and lengths in the returned scatterlist are
* PAGE_SIZE aligned.
*
* A mapping must be unmapped by using dma_buf_unmap_attachment(). Note that
* the underlying backing storage is pinned for as long as a mapping exists,
* therefore users/importers should not hold onto a mapping for undue amounts of
* time.
*
* Important: Dynamic importers must wait for the exclusive fence of the struct
* dma_resv attached to the DMA-BUF first.
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
*/
struct sg_table *dma_buf_map_attachment(struct dma_buf_attachment *attach,
enum dma_data_direction direction)
{
struct sg_table *sg_table;
dma-buf: add dynamic DMA-buf handling v15 On the exporter side we add optional explicit pinning callbacks. Which are called when the importer doesn't implement dynamic handling, move notification or need the DMA-buf locked in place for its use case. On the importer side we add an optional move_notify callback. This callback is used by the exporter to inform the importers that their mappings should be destroyed as soon as possible. This allows the exporter to provide the mappings without the need to pin the backing store. v2: don't try to invalidate mappings when the callback is NULL, lock the reservation obj while using the attachments, add helper to set the callback v3: move flag for invalidation support into the DMA-buf, use new attach_info structure to set the callback v4: use importer_priv field instead of mangling exporter priv. v5: drop invalidation_supported flag v6: squash together with pin/unpin changes v7: pin/unpin takes an attachment now v8: nuke dma_buf_attachment_(map|unmap)_locked, everything is now handled backward compatible v9: always cache when export/importer don't agree on dynamic handling v10: minimal style cleanup v11: drop automatically re-entry avoidance v12: rename callback to move_notify v13: add might_lock in appropriate places v14: rebase on separated locking change v15: add EXPERIMENTAL flag, some more code comments Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/353993/?series=73646&rev=1
2018-07-03 22:42:26 +08:00
int r;
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
might_sleep();
if (WARN_ON(!attach || !attach->dmabuf))
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
return ERR_PTR(-EINVAL);
dma_resv_assert_held(attach->dmabuf->resv);
dma-buf: change DMA-buf locking convention v3 This patch is a stripped down version of the locking changes necessary to support dynamic DMA-buf handling. It adds a dynamic flag for both importers as well as exporters so that drivers can choose if they want the reservation object locked or unlocked during mapping of attachments. For compatibility between drivers we cache the DMA-buf mapping during attaching an importer as soon as exporter/importer disagree on the dynamic handling. Issues and solutions we considered: - We can't change all existing drivers, and existing improters have strong opinions about which locks they're holding while calling dma_buf_attachment_map/unmap. Exporters also have strong opinions about which locks they can acquire in their ->map/unmap callbacks, levaing no room for change. The solution to avoid this was to move the actual map/unmap out from this call, into the attach/detach callbacks, and cache the mapping. This works because drivers don't call attach/detach from deep within their code callchains (like deep in memory management code called from cs/execbuf ioctl), but directly from the fd2handle implementation. - The caching has some troubles on some soc drivers, which set other modes than DMA_BIDIRECTIONAL. We can't have 2 incompatible mappings, and we can't re-create the mapping at _map time due to the above locking fun. We very carefuly step around that by only caching at attach time if the dynamic mode between importer/expoert mismatches. - There's been quite some discussion on dma-buf mappings which need active cache management, which would all break down when caching, plus we don't have explicit flush operations on the attachment side. The solution to this was to shrug and keep the current discrepancy between what the dma-buf docs claim and what implementations do, with the hope that the begin/end_cpu_access hooks are good enough and that all necessary flushing to keep device mappings consistent will be done there. v2: cleanup set_name merge, improve kerneldoc v3: update commit message, kerneldoc and cleanup _debug_show() Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/336788/
2018-07-03 22:42:26 +08:00
if (attach->sgt) {
/*
* Two mappings with different directions for the same
* attachment are not allowed.
*/
if (attach->dir != direction &&
attach->dir != DMA_BIDIRECTIONAL)
return ERR_PTR(-EBUSY);
return attach->sgt;
}
dma-buf: add dynamic DMA-buf handling v15 On the exporter side we add optional explicit pinning callbacks. Which are called when the importer doesn't implement dynamic handling, move notification or need the DMA-buf locked in place for its use case. On the importer side we add an optional move_notify callback. This callback is used by the exporter to inform the importers that their mappings should be destroyed as soon as possible. This allows the exporter to provide the mappings without the need to pin the backing store. v2: don't try to invalidate mappings when the callback is NULL, lock the reservation obj while using the attachments, add helper to set the callback v3: move flag for invalidation support into the DMA-buf, use new attach_info structure to set the callback v4: use importer_priv field instead of mangling exporter priv. v5: drop invalidation_supported flag v6: squash together with pin/unpin changes v7: pin/unpin takes an attachment now v8: nuke dma_buf_attachment_(map|unmap)_locked, everything is now handled backward compatible v9: always cache when export/importer don't agree on dynamic handling v10: minimal style cleanup v11: drop automatically re-entry avoidance v12: rename callback to move_notify v13: add might_lock in appropriate places v14: rebase on separated locking change v15: add EXPERIMENTAL flag, some more code comments Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/353993/?series=73646&rev=1
2018-07-03 22:42:26 +08:00
if (dma_buf_is_dynamic(attach->dmabuf)) {
if (!IS_ENABLED(CONFIG_DMABUF_MOVE_NOTIFY)) {
r = attach->dmabuf->ops->pin(attach);
dma-buf: add dynamic DMA-buf handling v15 On the exporter side we add optional explicit pinning callbacks. Which are called when the importer doesn't implement dynamic handling, move notification or need the DMA-buf locked in place for its use case. On the importer side we add an optional move_notify callback. This callback is used by the exporter to inform the importers that their mappings should be destroyed as soon as possible. This allows the exporter to provide the mappings without the need to pin the backing store. v2: don't try to invalidate mappings when the callback is NULL, lock the reservation obj while using the attachments, add helper to set the callback v3: move flag for invalidation support into the DMA-buf, use new attach_info structure to set the callback v4: use importer_priv field instead of mangling exporter priv. v5: drop invalidation_supported flag v6: squash together with pin/unpin changes v7: pin/unpin takes an attachment now v8: nuke dma_buf_attachment_(map|unmap)_locked, everything is now handled backward compatible v9: always cache when export/importer don't agree on dynamic handling v10: minimal style cleanup v11: drop automatically re-entry avoidance v12: rename callback to move_notify v13: add might_lock in appropriate places v14: rebase on separated locking change v15: add EXPERIMENTAL flag, some more code comments Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/353993/?series=73646&rev=1
2018-07-03 22:42:26 +08:00
if (r)
return ERR_PTR(r);
}
}
dma-buf: change DMA-buf locking convention v3 This patch is a stripped down version of the locking changes necessary to support dynamic DMA-buf handling. It adds a dynamic flag for both importers as well as exporters so that drivers can choose if they want the reservation object locked or unlocked during mapping of attachments. For compatibility between drivers we cache the DMA-buf mapping during attaching an importer as soon as exporter/importer disagree on the dynamic handling. Issues and solutions we considered: - We can't change all existing drivers, and existing improters have strong opinions about which locks they're holding while calling dma_buf_attachment_map/unmap. Exporters also have strong opinions about which locks they can acquire in their ->map/unmap callbacks, levaing no room for change. The solution to avoid this was to move the actual map/unmap out from this call, into the attach/detach callbacks, and cache the mapping. This works because drivers don't call attach/detach from deep within their code callchains (like deep in memory management code called from cs/execbuf ioctl), but directly from the fd2handle implementation. - The caching has some troubles on some soc drivers, which set other modes than DMA_BIDIRECTIONAL. We can't have 2 incompatible mappings, and we can't re-create the mapping at _map time due to the above locking fun. We very carefuly step around that by only caching at attach time if the dynamic mode between importer/expoert mismatches. - There's been quite some discussion on dma-buf mappings which need active cache management, which would all break down when caching, plus we don't have explicit flush operations on the attachment side. The solution to this was to shrug and keep the current discrepancy between what the dma-buf docs claim and what implementations do, with the hope that the begin/end_cpu_access hooks are good enough and that all necessary flushing to keep device mappings consistent will be done there. v2: cleanup set_name merge, improve kerneldoc v3: update commit message, kerneldoc and cleanup _debug_show() Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/336788/
2018-07-03 22:42:26 +08:00
dma-buf: Add debug option We have too many people abusing the struct page they can get at but really shouldn't in importers. Aside from that the backing page might simply not exist (for dynamic p2p mappings) looking at it and using it e.g. for mmap can also wreak the page handling of the exporter completely. Importers really must go through the proper interface like dma_buf_mmap for everything. I'm semi-tempted to enforce this for dynamic importers since those really have no excuse at all to break the rules. Unfortuantely we can't store the right pointers somewhere safe to make sure we oops on something recognizable, so best is to just wrangle them a bit by flipping all the bits. At least on x86 kernel addresses have all their high bits sets and the struct page array is fairly low in the kernel mapping, so flipping all the bits gives us a very high pointer in userspace and hence excellent chances for an invalid dereference. v2: Add a note to the @map_dma_buf hook that exporters shouldn't do fancy caching tricks, which would blow up with this address scrambling trick here (Chris) Enable by default when CONFIG_DMA_API_DEBUG is enabled. v3: Only one copy of the mangle/unmangle code (Christian) v4: #ifdef, not #if (0day) v5: sg_table can also be an ERR_PTR (Chris, Christian) Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Daniel Vetter <daniel.vetter@intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Sumit Semwal <sumit.semwal@linaro.org> Cc: "Christian König" <christian.koenig@amd.com> Cc: David Stevens <stevensd@chromium.org> Cc: linux-media@vger.kernel.org Cc: linaro-mm-sig@lists.linaro.org Link: https://patchwork.freedesktop.org/patch/msgid/20210115164739.3958206-1-daniel.vetter@ffwll.ch
2021-01-16 00:47:39 +08:00
sg_table = __map_dma_buf(attach, direction);
if (!sg_table)
sg_table = ERR_PTR(-ENOMEM);
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
dma-buf: add dynamic DMA-buf handling v15 On the exporter side we add optional explicit pinning callbacks. Which are called when the importer doesn't implement dynamic handling, move notification or need the DMA-buf locked in place for its use case. On the importer side we add an optional move_notify callback. This callback is used by the exporter to inform the importers that their mappings should be destroyed as soon as possible. This allows the exporter to provide the mappings without the need to pin the backing store. v2: don't try to invalidate mappings when the callback is NULL, lock the reservation obj while using the attachments, add helper to set the callback v3: move flag for invalidation support into the DMA-buf, use new attach_info structure to set the callback v4: use importer_priv field instead of mangling exporter priv. v5: drop invalidation_supported flag v6: squash together with pin/unpin changes v7: pin/unpin takes an attachment now v8: nuke dma_buf_attachment_(map|unmap)_locked, everything is now handled backward compatible v9: always cache when export/importer don't agree on dynamic handling v10: minimal style cleanup v11: drop automatically re-entry avoidance v12: rename callback to move_notify v13: add might_lock in appropriate places v14: rebase on separated locking change v15: add EXPERIMENTAL flag, some more code comments Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/353993/?series=73646&rev=1
2018-07-03 22:42:26 +08:00
if (IS_ERR(sg_table) && dma_buf_is_dynamic(attach->dmabuf) &&
!IS_ENABLED(CONFIG_DMABUF_MOVE_NOTIFY))
attach->dmabuf->ops->unpin(attach);
dma-buf: add dynamic DMA-buf handling v15 On the exporter side we add optional explicit pinning callbacks. Which are called when the importer doesn't implement dynamic handling, move notification or need the DMA-buf locked in place for its use case. On the importer side we add an optional move_notify callback. This callback is used by the exporter to inform the importers that their mappings should be destroyed as soon as possible. This allows the exporter to provide the mappings without the need to pin the backing store. v2: don't try to invalidate mappings when the callback is NULL, lock the reservation obj while using the attachments, add helper to set the callback v3: move flag for invalidation support into the DMA-buf, use new attach_info structure to set the callback v4: use importer_priv field instead of mangling exporter priv. v5: drop invalidation_supported flag v6: squash together with pin/unpin changes v7: pin/unpin takes an attachment now v8: nuke dma_buf_attachment_(map|unmap)_locked, everything is now handled backward compatible v9: always cache when export/importer don't agree on dynamic handling v10: minimal style cleanup v11: drop automatically re-entry avoidance v12: rename callback to move_notify v13: add might_lock in appropriate places v14: rebase on separated locking change v15: add EXPERIMENTAL flag, some more code comments Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/353993/?series=73646&rev=1
2018-07-03 22:42:26 +08:00
if (!IS_ERR(sg_table) && attach->dmabuf->ops->cache_sgt_mapping) {
attach->sgt = sg_table;
attach->dir = direction;
}
#ifdef CONFIG_DMA_API_DEBUG
if (!IS_ERR(sg_table)) {
struct scatterlist *sg;
u64 addr;
int len;
int i;
for_each_sgtable_dma_sg(sg_table, sg, i) {
addr = sg_dma_address(sg);
len = sg_dma_len(sg);
if (!PAGE_ALIGNED(addr) || !PAGE_ALIGNED(len)) {
pr_debug("%s: addr %llx or len %x is not page aligned!\n",
__func__, addr, len);
}
}
}
#endif /* CONFIG_DMA_API_DEBUG */
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
return sg_table;
}
EXPORT_SYMBOL_NS_GPL(dma_buf_map_attachment, DMA_BUF);
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
/**
* dma_buf_map_attachment_unlocked - Returns the scatterlist table of the attachment;
* mapped into _device_ address space. Is a wrapper for map_dma_buf() of the
* dma_buf_ops.
* @attach: [in] attachment whose scatterlist is to be returned
* @direction: [in] direction of DMA transfer
*
* Unlocked variant of dma_buf_map_attachment().
*/
struct sg_table *
dma_buf_map_attachment_unlocked(struct dma_buf_attachment *attach,
enum dma_data_direction direction)
{
struct sg_table *sg_table;
might_sleep();
if (WARN_ON(!attach || !attach->dmabuf))
return ERR_PTR(-EINVAL);
dma_resv_lock(attach->dmabuf->resv, NULL);
sg_table = dma_buf_map_attachment(attach, direction);
dma_resv_unlock(attach->dmabuf->resv);
return sg_table;
}
EXPORT_SYMBOL_NS_GPL(dma_buf_map_attachment_unlocked, DMA_BUF);
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
/**
* dma_buf_unmap_attachment - unmaps and decreases usecount of the buffer;might
* deallocate the scatterlist associated. Is a wrapper for unmap_dma_buf() of
* dma_buf_ops.
* @attach: [in] attachment to unmap buffer from
* @sg_table: [in] scatterlist info of the buffer to unmap
* @direction: [in] direction of DMA transfer
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
*
* This unmaps a DMA mapping for @attached obtained by dma_buf_map_attachment().
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
*/
void dma_buf_unmap_attachment(struct dma_buf_attachment *attach,
struct sg_table *sg_table,
enum dma_data_direction direction)
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
{
might_sleep();
if (WARN_ON(!attach || !attach->dmabuf || !sg_table))
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
return;
dma_resv_assert_held(attach->dmabuf->resv);
dma-buf: change DMA-buf locking convention v3 This patch is a stripped down version of the locking changes necessary to support dynamic DMA-buf handling. It adds a dynamic flag for both importers as well as exporters so that drivers can choose if they want the reservation object locked or unlocked during mapping of attachments. For compatibility between drivers we cache the DMA-buf mapping during attaching an importer as soon as exporter/importer disagree on the dynamic handling. Issues and solutions we considered: - We can't change all existing drivers, and existing improters have strong opinions about which locks they're holding while calling dma_buf_attachment_map/unmap. Exporters also have strong opinions about which locks they can acquire in their ->map/unmap callbacks, levaing no room for change. The solution to avoid this was to move the actual map/unmap out from this call, into the attach/detach callbacks, and cache the mapping. This works because drivers don't call attach/detach from deep within their code callchains (like deep in memory management code called from cs/execbuf ioctl), but directly from the fd2handle implementation. - The caching has some troubles on some soc drivers, which set other modes than DMA_BIDIRECTIONAL. We can't have 2 incompatible mappings, and we can't re-create the mapping at _map time due to the above locking fun. We very carefuly step around that by only caching at attach time if the dynamic mode between importer/expoert mismatches. - There's been quite some discussion on dma-buf mappings which need active cache management, which would all break down when caching, plus we don't have explicit flush operations on the attachment side. The solution to this was to shrug and keep the current discrepancy between what the dma-buf docs claim and what implementations do, with the hope that the begin/end_cpu_access hooks are good enough and that all necessary flushing to keep device mappings consistent will be done there. v2: cleanup set_name merge, improve kerneldoc v3: update commit message, kerneldoc and cleanup _debug_show() Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/336788/
2018-07-03 22:42:26 +08:00
if (attach->sgt == sg_table)
return;
dma-buf: Add debug option We have too many people abusing the struct page they can get at but really shouldn't in importers. Aside from that the backing page might simply not exist (for dynamic p2p mappings) looking at it and using it e.g. for mmap can also wreak the page handling of the exporter completely. Importers really must go through the proper interface like dma_buf_mmap for everything. I'm semi-tempted to enforce this for dynamic importers since those really have no excuse at all to break the rules. Unfortuantely we can't store the right pointers somewhere safe to make sure we oops on something recognizable, so best is to just wrangle them a bit by flipping all the bits. At least on x86 kernel addresses have all their high bits sets and the struct page array is fairly low in the kernel mapping, so flipping all the bits gives us a very high pointer in userspace and hence excellent chances for an invalid dereference. v2: Add a note to the @map_dma_buf hook that exporters shouldn't do fancy caching tricks, which would blow up with this address scrambling trick here (Chris) Enable by default when CONFIG_DMA_API_DEBUG is enabled. v3: Only one copy of the mangle/unmangle code (Christian) v4: #ifdef, not #if (0day) v5: sg_table can also be an ERR_PTR (Chris, Christian) Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> Reviewed-by: Christian König <christian.koenig@amd.com> Signed-off-by: Daniel Vetter <daniel.vetter@intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Sumit Semwal <sumit.semwal@linaro.org> Cc: "Christian König" <christian.koenig@amd.com> Cc: David Stevens <stevensd@chromium.org> Cc: linux-media@vger.kernel.org Cc: linaro-mm-sig@lists.linaro.org Link: https://patchwork.freedesktop.org/patch/msgid/20210115164739.3958206-1-daniel.vetter@ffwll.ch
2021-01-16 00:47:39 +08:00
__unmap_dma_buf(attach, sg_table, direction);
dma-buf: add dynamic DMA-buf handling v15 On the exporter side we add optional explicit pinning callbacks. Which are called when the importer doesn't implement dynamic handling, move notification or need the DMA-buf locked in place for its use case. On the importer side we add an optional move_notify callback. This callback is used by the exporter to inform the importers that their mappings should be destroyed as soon as possible. This allows the exporter to provide the mappings without the need to pin the backing store. v2: don't try to invalidate mappings when the callback is NULL, lock the reservation obj while using the attachments, add helper to set the callback v3: move flag for invalidation support into the DMA-buf, use new attach_info structure to set the callback v4: use importer_priv field instead of mangling exporter priv. v5: drop invalidation_supported flag v6: squash together with pin/unpin changes v7: pin/unpin takes an attachment now v8: nuke dma_buf_attachment_(map|unmap)_locked, everything is now handled backward compatible v9: always cache when export/importer don't agree on dynamic handling v10: minimal style cleanup v11: drop automatically re-entry avoidance v12: rename callback to move_notify v13: add might_lock in appropriate places v14: rebase on separated locking change v15: add EXPERIMENTAL flag, some more code comments Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/353993/?series=73646&rev=1
2018-07-03 22:42:26 +08:00
if (dma_buf_is_dynamic(attach->dmabuf) &&
!IS_ENABLED(CONFIG_DMABUF_MOVE_NOTIFY))
dma-buf: add dynamic DMA-buf handling v15 On the exporter side we add optional explicit pinning callbacks. Which are called when the importer doesn't implement dynamic handling, move notification or need the DMA-buf locked in place for its use case. On the importer side we add an optional move_notify callback. This callback is used by the exporter to inform the importers that their mappings should be destroyed as soon as possible. This allows the exporter to provide the mappings without the need to pin the backing store. v2: don't try to invalidate mappings when the callback is NULL, lock the reservation obj while using the attachments, add helper to set the callback v3: move flag for invalidation support into the DMA-buf, use new attach_info structure to set the callback v4: use importer_priv field instead of mangling exporter priv. v5: drop invalidation_supported flag v6: squash together with pin/unpin changes v7: pin/unpin takes an attachment now v8: nuke dma_buf_attachment_(map|unmap)_locked, everything is now handled backward compatible v9: always cache when export/importer don't agree on dynamic handling v10: minimal style cleanup v11: drop automatically re-entry avoidance v12: rename callback to move_notify v13: add might_lock in appropriate places v14: rebase on separated locking change v15: add EXPERIMENTAL flag, some more code comments Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/353993/?series=73646&rev=1
2018-07-03 22:42:26 +08:00
dma_buf_unpin(attach);
dma-buf: Introduce dma buffer sharing mechanism This is the first step in defining a dma buffer sharing mechanism. A new buffer object dma_buf is added, with operations and API to allow easy sharing of this buffer object across devices. The framework allows: - creation of a buffer object, its association with a file pointer, and associated allocator-defined operations on that buffer. This operation is called the 'export' operation. - different devices to 'attach' themselves to this exported buffer object, to facilitate backing storage negotiation, using dma_buf_attach() API. - the exported buffer object to be shared with the other entity by asking for its 'file-descriptor (fd)', and sharing the fd across. - a received fd to get the buffer object back, where it can be accessed using the associated exporter-defined operations. - the exporter and user to share the scatterlist associated with this buffer object using map_dma_buf and unmap_dma_buf operations. Atleast one 'attach()' call is required to be made prior to calling the map_dma_buf() operation. Couple of building blocks in map_dma_buf() are added to ease introduction of sync'ing across exporter and users, and late allocation by the exporter. For this first version, this framework will work with certain conditions: - *ONLY* exporter will be allowed to mmap to userspace (outside of this framework - mmap is not a buffer object operation), - currently, *ONLY* users that do not need CPU access to the buffer are allowed. More details are there in the documentation patch. This is based on design suggestions from many people at the mini-summits[1], most notably from Arnd Bergmann <arnd@arndb.de>, Rob Clark <rob@ti.com> and Daniel Vetter <daniel@ffwll.ch>. The implementation is inspired from proof-of-concept patch-set from Tomasz Stanislawski <t.stanislaws@samsung.com>, who demonstrated buffer sharing between two v4l2 devices. [2] [1]: https://wiki.linaro.org/OfficeofCTO/MemoryManagement [2]: http://lwn.net/Articles/454389 Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org> Signed-off-by: Sumit Semwal <sumit.semwal@ti.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Reviewed-by: Dave Airlie <airlied@redhat.com> Reviewed-and-Tested-by: Rob Clark <rob.clark@linaro.org> Signed-off-by: Dave Airlie <airlied@redhat.com>
2011-12-26 17:23:15 +08:00
}
EXPORT_SYMBOL_NS_GPL(dma_buf_unmap_attachment, DMA_BUF);
/**
* dma_buf_unmap_attachment_unlocked - unmaps and decreases usecount of the buffer;might
* deallocate the scatterlist associated. Is a wrapper for unmap_dma_buf() of
* dma_buf_ops.
* @attach: [in] attachment to unmap buffer from
* @sg_table: [in] scatterlist info of the buffer to unmap
* @direction: [in] direction of DMA transfer
*
* Unlocked variant of dma_buf_unmap_attachment().
*/
void dma_buf_unmap_attachment_unlocked(struct dma_buf_attachment *attach,
struct sg_table *sg_table,
enum dma_data_direction direction)
{
might_sleep();
if (WARN_ON(!attach || !attach->dmabuf || !sg_table))
return;
dma_resv_lock(attach->dmabuf->resv, NULL);
dma_buf_unmap_attachment(attach, sg_table, direction);
dma_resv_unlock(attach->dmabuf->resv);
}
EXPORT_SYMBOL_NS_GPL(dma_buf_unmap_attachment_unlocked, DMA_BUF);
dma-buf: add dynamic DMA-buf handling v15 On the exporter side we add optional explicit pinning callbacks. Which are called when the importer doesn't implement dynamic handling, move notification or need the DMA-buf locked in place for its use case. On the importer side we add an optional move_notify callback. This callback is used by the exporter to inform the importers that their mappings should be destroyed as soon as possible. This allows the exporter to provide the mappings without the need to pin the backing store. v2: don't try to invalidate mappings when the callback is NULL, lock the reservation obj while using the attachments, add helper to set the callback v3: move flag for invalidation support into the DMA-buf, use new attach_info structure to set the callback v4: use importer_priv field instead of mangling exporter priv. v5: drop invalidation_supported flag v6: squash together with pin/unpin changes v7: pin/unpin takes an attachment now v8: nuke dma_buf_attachment_(map|unmap)_locked, everything is now handled backward compatible v9: always cache when export/importer don't agree on dynamic handling v10: minimal style cleanup v11: drop automatically re-entry avoidance v12: rename callback to move_notify v13: add might_lock in appropriate places v14: rebase on separated locking change v15: add EXPERIMENTAL flag, some more code comments Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/353993/?series=73646&rev=1
2018-07-03 22:42:26 +08:00
/**
* dma_buf_move_notify - notify attachments that DMA-buf is moving
*
* @dmabuf: [in] buffer which is moving
*
* Informs all attachments that they need to destroy and recreate all their
dma-buf: add dynamic DMA-buf handling v15 On the exporter side we add optional explicit pinning callbacks. Which are called when the importer doesn't implement dynamic handling, move notification or need the DMA-buf locked in place for its use case. On the importer side we add an optional move_notify callback. This callback is used by the exporter to inform the importers that their mappings should be destroyed as soon as possible. This allows the exporter to provide the mappings without the need to pin the backing store. v2: don't try to invalidate mappings when the callback is NULL, lock the reservation obj while using the attachments, add helper to set the callback v3: move flag for invalidation support into the DMA-buf, use new attach_info structure to set the callback v4: use importer_priv field instead of mangling exporter priv. v5: drop invalidation_supported flag v6: squash together with pin/unpin changes v7: pin/unpin takes an attachment now v8: nuke dma_buf_attachment_(map|unmap)_locked, everything is now handled backward compatible v9: always cache when export/importer don't agree on dynamic handling v10: minimal style cleanup v11: drop automatically re-entry avoidance v12: rename callback to move_notify v13: add might_lock in appropriate places v14: rebase on separated locking change v15: add EXPERIMENTAL flag, some more code comments Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/353993/?series=73646&rev=1
2018-07-03 22:42:26 +08:00
* mappings.
*/
void dma_buf_move_notify(struct dma_buf *dmabuf)
{
struct dma_buf_attachment *attach;
dma_resv_assert_held(dmabuf->resv);
list_for_each_entry(attach, &dmabuf->attachments, node)
if (attach->importer_ops)
dma-buf: add dynamic DMA-buf handling v15 On the exporter side we add optional explicit pinning callbacks. Which are called when the importer doesn't implement dynamic handling, move notification or need the DMA-buf locked in place for its use case. On the importer side we add an optional move_notify callback. This callback is used by the exporter to inform the importers that their mappings should be destroyed as soon as possible. This allows the exporter to provide the mappings without the need to pin the backing store. v2: don't try to invalidate mappings when the callback is NULL, lock the reservation obj while using the attachments, add helper to set the callback v3: move flag for invalidation support into the DMA-buf, use new attach_info structure to set the callback v4: use importer_priv field instead of mangling exporter priv. v5: drop invalidation_supported flag v6: squash together with pin/unpin changes v7: pin/unpin takes an attachment now v8: nuke dma_buf_attachment_(map|unmap)_locked, everything is now handled backward compatible v9: always cache when export/importer don't agree on dynamic handling v10: minimal style cleanup v11: drop automatically re-entry avoidance v12: rename callback to move_notify v13: add might_lock in appropriate places v14: rebase on separated locking change v15: add EXPERIMENTAL flag, some more code comments Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/353993/?series=73646&rev=1
2018-07-03 22:42:26 +08:00
attach->importer_ops->move_notify(attach);
}
EXPORT_SYMBOL_NS_GPL(dma_buf_move_notify, DMA_BUF);
dma-buf: add dynamic DMA-buf handling v15 On the exporter side we add optional explicit pinning callbacks. Which are called when the importer doesn't implement dynamic handling, move notification or need the DMA-buf locked in place for its use case. On the importer side we add an optional move_notify callback. This callback is used by the exporter to inform the importers that their mappings should be destroyed as soon as possible. This allows the exporter to provide the mappings without the need to pin the backing store. v2: don't try to invalidate mappings when the callback is NULL, lock the reservation obj while using the attachments, add helper to set the callback v3: move flag for invalidation support into the DMA-buf, use new attach_info structure to set the callback v4: use importer_priv field instead of mangling exporter priv. v5: drop invalidation_supported flag v6: squash together with pin/unpin changes v7: pin/unpin takes an attachment now v8: nuke dma_buf_attachment_(map|unmap)_locked, everything is now handled backward compatible v9: always cache when export/importer don't agree on dynamic handling v10: minimal style cleanup v11: drop automatically re-entry avoidance v12: rename callback to move_notify v13: add might_lock in appropriate places v14: rebase on separated locking change v15: add EXPERIMENTAL flag, some more code comments Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/353993/?series=73646&rev=1
2018-07-03 22:42:26 +08:00
/**
* DOC: cpu access
*
* There are multiple reasons for supporting CPU access to a dma buffer object:
*
* - Fallback operations in the kernel, for example when a device is connected
* over USB and the kernel needs to shuffle the data around first before
* sending it away. Cache coherency is handled by bracketing any transactions
* with calls to dma_buf_begin_cpu_access() and dma_buf_end_cpu_access()
* access.
*
* Since for most kernel internal dma-buf accesses need the entire buffer, a
* vmap interface is introduced. Note that on very old 32-bit architectures
* vmalloc space might be limited and result in vmap calls failing.
*
* Interfaces::
*
dma-buf-map: Rename to iosys-map Rename struct dma_buf_map to struct iosys_map and corresponding APIs. Over time dma-buf-map grew up to more functionality than the one used by dma-buf: in fact it's just a shim layer to abstract system memory, that can be accessed via regular load and store, from IO memory that needs to be acessed via arch helpers. The idea is to extend this API so it can fulfill other needs, internal to a single driver. Example: in the i915 driver it's desired to share the implementation for integrated graphics, which uses mostly system memory, with discrete graphics, which may need to access IO memory. The conversion was mostly done with the following semantic patch: @r1@ @@ - struct dma_buf_map + struct iosys_map @r2@ @@ ( - DMA_BUF_MAP_INIT_VADDR + IOSYS_MAP_INIT_VADDR | - dma_buf_map_set_vaddr + iosys_map_set_vaddr | - dma_buf_map_set_vaddr_iomem + iosys_map_set_vaddr_iomem | - dma_buf_map_is_equal + iosys_map_is_equal | - dma_buf_map_is_null + iosys_map_is_null | - dma_buf_map_is_set + iosys_map_is_set | - dma_buf_map_clear + iosys_map_clear | - dma_buf_map_memcpy_to + iosys_map_memcpy_to | - dma_buf_map_incr + iosys_map_incr ) @@ @@ - #include <linux/dma-buf-map.h> + #include <linux/iosys-map.h> Then some files had their includes adjusted and some comments were update to remove mentions to dma-buf-map. Since this is not specific to dma-buf anymore, move the documentation to the "Bus-Independent Device Accesses" section. v2: - Squash patches v3: - Fix wrong removal of dma-buf.h from MAINTAINERS - Move documentation from dma-buf.rst to device-io.rst v4: - Change documentation title and level Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com> Acked-by: Christian König <christian.koenig@amd.com> Acked-by: Sumit Semwal <sumit.semwal@linaro.org> Acked-by: Thomas Zimmermann <tzimmermann@suse.de> Link: https://patchwork.freedesktop.org/patch/msgid/20220204170541.829227-1-lucas.demarchi@intel.com
2022-02-05 01:05:41 +08:00
* void \*dma_buf_vmap(struct dma_buf \*dmabuf, struct iosys_map \*map)
* void dma_buf_vunmap(struct dma_buf \*dmabuf, struct iosys_map \*map)
*
* The vmap call can fail if there is no vmap support in the exporter, or if
* it runs out of vmalloc space. Note that the dma-buf layer keeps a reference
* count for all vmap access and calls down into the exporter's vmap function
* only when no vmapping exists, and only unmaps it once. Protection against
* concurrent vmap/vunmap calls is provided by taking the &dma_buf.lock mutex.
*
* - For full compatibility on the importer side with existing userspace
* interfaces, which might already support mmap'ing buffers. This is needed in
* many processing pipelines (e.g. feeding a software rendered image into a
* hardware pipeline, thumbnail creation, snapshots, ...). Also, Android's ION
* framework already supported this and for DMA buffer file descriptors to
* replace ION buffers mmap support was needed.
*
* There is no special interfaces, userspace simply calls mmap on the dma-buf
* fd. But like for CPU access there's a need to bracket the actual access,
* which is handled by the ioctl (DMA_BUF_IOCTL_SYNC). Note that
* DMA_BUF_IOCTL_SYNC can fail with -EAGAIN or -EINTR, in which case it must
* be restarted.
*
* Some systems might need some sort of cache coherency management e.g. when
* CPU and GPU domains are being accessed through dma-buf at the same time.
* To circumvent this problem there are begin/end coherency markers, that
* forward directly to existing dma-buf device drivers vfunc hooks. Userspace
* can make use of those markers through the DMA_BUF_IOCTL_SYNC ioctl. The
* sequence would be used like following:
*
* - mmap dma-buf fd
* - for each drawing/upload cycle in CPU 1. SYNC_START ioctl, 2. read/write
* to mmap area 3. SYNC_END ioctl. This can be repeated as often as you
* want (with the new data being consumed by say the GPU or the scanout
* device)
* - munmap once you don't need the buffer any more
*
* For correctness and optimal performance, it is always required to use
* SYNC_START and SYNC_END before and after, respectively, when accessing the
* mapped address. Userspace cannot rely on coherent access, even when there
* are systems where it just works without calling these ioctls.
*
* - And as a CPU fallback in userspace processing pipelines.
*
* Similar to the motivation for kernel cpu access it is again important that
* the userspace code of a given importing subsystem can use the same
* interfaces with a imported dma-buf buffer object as with a native buffer
* object. This is especially important for drm where the userspace part of
* contemporary OpenGL, X, and other drivers is huge, and reworking them to
* use a different way to mmap a buffer rather invasive.
*
* The assumption in the current dma-buf interfaces is that redirecting the
* initial mmap is all that's needed. A survey of some of the existing
* subsystems shows that no driver seems to do any nefarious thing like
* syncing up with outstanding asynchronous processing on the device or
* allocating special resources at fault time. So hopefully this is good
* enough, since adding interfaces to intercept pagefaults and allow pte
* shootdowns would increase the complexity quite a bit.
*
* Interface::
*
* int dma_buf_mmap(struct dma_buf \*, struct vm_area_struct \*,
* unsigned long);
*
* If the importing subsystem simply provides a special-purpose mmap call to
* set up a mapping in userspace, calling do_mmap with &dma_buf.file will
* equally achieve that for a dma-buf object.
*/
static int __dma_buf_begin_cpu_access(struct dma_buf *dmabuf,
enum dma_data_direction direction)
{
bool write = (direction == DMA_BIDIRECTIONAL ||
direction == DMA_TO_DEVICE);
struct dma_resv *resv = dmabuf->resv;
long ret;
/* Wait on any implicit rendering fences */
ret = dma_resv_wait_timeout(resv, dma_resv_usage_rw(write),
true, MAX_SCHEDULE_TIMEOUT);
if (ret < 0)
return ret;
return 0;
}
/**
* dma_buf_begin_cpu_access - Must be called before accessing a dma_buf from the
* cpu in the kernel context. Calls begin_cpu_access to allow exporter-specific
* preparations. Coherency is only guaranteed in the specified range for the
* specified access direction.
* @dmabuf: [in] buffer to prepare cpu access for.
* @direction: [in] direction of access.
*
* After the cpu access is complete the caller should call
* dma_buf_end_cpu_access(). Only when cpu access is bracketed by both calls is
* it guaranteed to be coherent with other DMA access.
*
* This function will also wait for any DMA transactions tracked through
* implicit synchronization in &dma_buf.resv. For DMA transactions with explicit
* synchronization this function will only ensure cache coherency, callers must
* ensure synchronization with such DMA transactions on their own.
*
* Can return negative error values, returns 0 on success.
*/
int dma_buf_begin_cpu_access(struct dma_buf *dmabuf,
enum dma_data_direction direction)
{
int ret = 0;
if (WARN_ON(!dmabuf))
return -EINVAL;
might_lock(&dmabuf->resv->lock.base);
if (dmabuf->ops->begin_cpu_access)
ret = dmabuf->ops->begin_cpu_access(dmabuf, direction);
/* Ensure that all fences are waited upon - but we first allow
* the native handler the chance to do so more efficiently if it
* chooses. A double invocation here will be reasonably cheap no-op.
*/
if (ret == 0)
ret = __dma_buf_begin_cpu_access(dmabuf, direction);
return ret;
}
EXPORT_SYMBOL_NS_GPL(dma_buf_begin_cpu_access, DMA_BUF);
/**
* dma_buf_end_cpu_access - Must be called after accessing a dma_buf from the
* cpu in the kernel context. Calls end_cpu_access to allow exporter-specific
* actions. Coherency is only guaranteed in the specified range for the
* specified access direction.
* @dmabuf: [in] buffer to complete cpu access for.
* @direction: [in] direction of access.
*
* This terminates CPU access started with dma_buf_begin_cpu_access().
*
* Can return negative error values, returns 0 on success.
*/
dma-buf, drm, ion: Propagate error code from dma_buf_start_cpu_access() Drivers, especially i915.ko, can fail during the initial migration of a dma-buf for CPU access. However, the error code from the driver was not being propagated back to ioctl and so userspace was blissfully ignorant of the failure. Rendering corruption ensues. Whilst fixing the ioctl to return the error code from dma_buf_start_cpu_access(), also do the same for dma_buf_end_cpu_access(). For most drivers, dma_buf_end_cpu_access() cannot fail. i915.ko however, as most drivers would, wants to avoid being uninterruptible (as would be required to guarrantee no failure when flushing the buffer to the device). As userspace already has to handle errors from the SYNC_IOCTL, take advantage of this to be able to restart the syscall across signals. This fixes a coherency issue for i915.ko as well as reducing the uninterruptible hold upon its BKL, the struct_mutex. Fixes commit c11e391da2a8fe973c3c2398452000bed505851e Author: Daniel Vetter <daniel.vetter@ffwll.ch> Date: Thu Feb 11 20:04:51 2016 -0200 dma-buf: Add ioctls to allow userspace to flush Testcase: igt/gem_concurrent_blit/*dmabuf*interruptible Testcase: igt/prime_mmap_coherency/ioctl-errors Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: Tiago Vignatti <tiago.vignatti@intel.com> Cc: Stéphane Marchesin <marcheu@chromium.org> Cc: David Herrmann <dh.herrmann@gmail.com> Cc: Sumit Semwal <sumit.semwal@linaro.org> Cc: Daniel Vetter <daniel.vetter@intel.com> CC: linux-media@vger.kernel.org Cc: dri-devel@lists.freedesktop.org Cc: linaro-mm-sig@lists.linaro.org Cc: intel-gfx@lists.freedesktop.org Cc: devel@driverdev.osuosl.org Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: http://patchwork.freedesktop.org/patch/msgid/1458331359-2634-1-git-send-email-chris@chris-wilson.co.uk
2016-03-19 04:02:39 +08:00
int dma_buf_end_cpu_access(struct dma_buf *dmabuf,
enum dma_data_direction direction)
{
dma-buf, drm, ion: Propagate error code from dma_buf_start_cpu_access() Drivers, especially i915.ko, can fail during the initial migration of a dma-buf for CPU access. However, the error code from the driver was not being propagated back to ioctl and so userspace was blissfully ignorant of the failure. Rendering corruption ensues. Whilst fixing the ioctl to return the error code from dma_buf_start_cpu_access(), also do the same for dma_buf_end_cpu_access(). For most drivers, dma_buf_end_cpu_access() cannot fail. i915.ko however, as most drivers would, wants to avoid being uninterruptible (as would be required to guarrantee no failure when flushing the buffer to the device). As userspace already has to handle errors from the SYNC_IOCTL, take advantage of this to be able to restart the syscall across signals. This fixes a coherency issue for i915.ko as well as reducing the uninterruptible hold upon its BKL, the struct_mutex. Fixes commit c11e391da2a8fe973c3c2398452000bed505851e Author: Daniel Vetter <daniel.vetter@ffwll.ch> Date: Thu Feb 11 20:04:51 2016 -0200 dma-buf: Add ioctls to allow userspace to flush Testcase: igt/gem_concurrent_blit/*dmabuf*interruptible Testcase: igt/prime_mmap_coherency/ioctl-errors Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: Tiago Vignatti <tiago.vignatti@intel.com> Cc: Stéphane Marchesin <marcheu@chromium.org> Cc: David Herrmann <dh.herrmann@gmail.com> Cc: Sumit Semwal <sumit.semwal@linaro.org> Cc: Daniel Vetter <daniel.vetter@intel.com> CC: linux-media@vger.kernel.org Cc: dri-devel@lists.freedesktop.org Cc: linaro-mm-sig@lists.linaro.org Cc: intel-gfx@lists.freedesktop.org Cc: devel@driverdev.osuosl.org Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: http://patchwork.freedesktop.org/patch/msgid/1458331359-2634-1-git-send-email-chris@chris-wilson.co.uk
2016-03-19 04:02:39 +08:00
int ret = 0;
WARN_ON(!dmabuf);
might_lock(&dmabuf->resv->lock.base);
if (dmabuf->ops->end_cpu_access)
dma-buf, drm, ion: Propagate error code from dma_buf_start_cpu_access() Drivers, especially i915.ko, can fail during the initial migration of a dma-buf for CPU access. However, the error code from the driver was not being propagated back to ioctl and so userspace was blissfully ignorant of the failure. Rendering corruption ensues. Whilst fixing the ioctl to return the error code from dma_buf_start_cpu_access(), also do the same for dma_buf_end_cpu_access(). For most drivers, dma_buf_end_cpu_access() cannot fail. i915.ko however, as most drivers would, wants to avoid being uninterruptible (as would be required to guarrantee no failure when flushing the buffer to the device). As userspace already has to handle errors from the SYNC_IOCTL, take advantage of this to be able to restart the syscall across signals. This fixes a coherency issue for i915.ko as well as reducing the uninterruptible hold upon its BKL, the struct_mutex. Fixes commit c11e391da2a8fe973c3c2398452000bed505851e Author: Daniel Vetter <daniel.vetter@ffwll.ch> Date: Thu Feb 11 20:04:51 2016 -0200 dma-buf: Add ioctls to allow userspace to flush Testcase: igt/gem_concurrent_blit/*dmabuf*interruptible Testcase: igt/prime_mmap_coherency/ioctl-errors Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk> Cc: Tiago Vignatti <tiago.vignatti@intel.com> Cc: Stéphane Marchesin <marcheu@chromium.org> Cc: David Herrmann <dh.herrmann@gmail.com> Cc: Sumit Semwal <sumit.semwal@linaro.org> Cc: Daniel Vetter <daniel.vetter@intel.com> CC: linux-media@vger.kernel.org Cc: dri-devel@lists.freedesktop.org Cc: linaro-mm-sig@lists.linaro.org Cc: intel-gfx@lists.freedesktop.org Cc: devel@driverdev.osuosl.org Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: http://patchwork.freedesktop.org/patch/msgid/1458331359-2634-1-git-send-email-chris@chris-wilson.co.uk
2016-03-19 04:02:39 +08:00
ret = dmabuf->ops->end_cpu_access(dmabuf, direction);
return ret;
}
EXPORT_SYMBOL_NS_GPL(dma_buf_end_cpu_access, DMA_BUF);
dma-buf: mmap support Compared to Rob Clark's RFC I've ditched the prepare/finish hooks and corresponding ioctls on the dma_buf file. The major reason for that is that many people seem to be under the impression that this is also for synchronization with outstanding asynchronous processsing. I'm pretty massively opposed to this because: - It boils down reinventing a new rather general-purpose userspace synchronization interface. If we look at things like futexes, this is hard to get right. - Furthermore a lot of kernel code has to interact with this synchronization primitive. This smells a look like the dri1 hw_lock, a horror show I prefer not to reinvent. - Even more fun is that multiple different subsystems would interact here, so we have plenty of opportunities to create funny deadlock scenarios. I think synchronization is a wholesale different problem from data sharing and should be tackled as an orthogonal problem. Now we could demand that prepare/finish may only ensure cache coherency (as Rob intended), but that runs up into the next problem: We not only need mmap support to facilitate sw-only processing nodes in a pipeline (without jumping through hoops by importing the dma_buf into some sw-access only importer), which allows for a nicer ION->dma-buf upgrade path for existing Android userspace. We also need mmap support for existing importing subsystems to support existing userspace libraries. And a loot of these subsystems are expected to export coherent userspace mappings. So prepare/finish can only ever be optional and the exporter /needs/ to support coherent mappings. Given that mmap access is always somewhat fallback-y in nature I've decided to drop this optimization, instead of just making it optional. If we demonstrate a clear need for this, supported by benchmark results, we can always add it in again later as an optional extension. Other differences compared to Rob's RFC is the above mentioned support for mapping a dma-buf through facilities provided by the importer. Which results in mmap support no longer being optional. Note that this dma-buf mmap patch does _not_ support every possible insanity an existing subsystem could pull of with mmap: Because it does not allow to intercept pagefaults and shoot down ptes importing subsystems can't add some magic of their own at these points (e.g. to automatically synchronize with outstanding rendering or set up some special resources). I've done a cursory read through a few mmap implementions of various subsytems and I'm hopeful that we can avoid this (and the complexity it'd bring with it). Additonally I've extended the documentation a bit to explain the hows and whys of this mmap extension. In case we ever want to add support for explicitly cache maneged userspace mmap with a prepare/finish ioctl pair, we could specify that userspace needs to mmap a different part of the dma_buf, e.g. the range starting at dma_buf->size up to dma_buf->size*2. This works because the size of a dma_buf is invariant over it's lifetime. The exporter would obviously need to fall back to coherent mappings for both ranges if a legacy clients maps the coherent range and the architecture cannot suppor conflicting caching policies. Also, this would obviously be optional and userspace needs to be able to fall back to coherent mappings. v2: - Spelling fixes from Rob Clark. - Compile fix for !DMA_BUF from Rob Clark. - Extend commit message to explain how explicitly cache managed mmap support could be added later. - Extend the documentation with implementations notes for exporters that need to manually fake coherency. v3: - dma_buf pointer initialization goof-up noticed by Rebecca Schultz Zavin. Cc: Rob Clark <rob.clark@linaro.org> Cc: Rebecca Schultz Zavin <rebecca@android.com> Acked-by: Rob Clark <rob.clark@linaro.org> Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org>
2012-04-24 17:08:52 +08:00
/**
* dma_buf_mmap - Setup up a userspace mmap with the given vma
* @dmabuf: [in] buffer that should back the vma
dma-buf: mmap support Compared to Rob Clark's RFC I've ditched the prepare/finish hooks and corresponding ioctls on the dma_buf file. The major reason for that is that many people seem to be under the impression that this is also for synchronization with outstanding asynchronous processsing. I'm pretty massively opposed to this because: - It boils down reinventing a new rather general-purpose userspace synchronization interface. If we look at things like futexes, this is hard to get right. - Furthermore a lot of kernel code has to interact with this synchronization primitive. This smells a look like the dri1 hw_lock, a horror show I prefer not to reinvent. - Even more fun is that multiple different subsystems would interact here, so we have plenty of opportunities to create funny deadlock scenarios. I think synchronization is a wholesale different problem from data sharing and should be tackled as an orthogonal problem. Now we could demand that prepare/finish may only ensure cache coherency (as Rob intended), but that runs up into the next problem: We not only need mmap support to facilitate sw-only processing nodes in a pipeline (without jumping through hoops by importing the dma_buf into some sw-access only importer), which allows for a nicer ION->dma-buf upgrade path for existing Android userspace. We also need mmap support for existing importing subsystems to support existing userspace libraries. And a loot of these subsystems are expected to export coherent userspace mappings. So prepare/finish can only ever be optional and the exporter /needs/ to support coherent mappings. Given that mmap access is always somewhat fallback-y in nature I've decided to drop this optimization, instead of just making it optional. If we demonstrate a clear need for this, supported by benchmark results, we can always add it in again later as an optional extension. Other differences compared to Rob's RFC is the above mentioned support for mapping a dma-buf through facilities provided by the importer. Which results in mmap support no longer being optional. Note that this dma-buf mmap patch does _not_ support every possible insanity an existing subsystem could pull of with mmap: Because it does not allow to intercept pagefaults and shoot down ptes importing subsystems can't add some magic of their own at these points (e.g. to automatically synchronize with outstanding rendering or set up some special resources). I've done a cursory read through a few mmap implementions of various subsytems and I'm hopeful that we can avoid this (and the complexity it'd bring with it). Additonally I've extended the documentation a bit to explain the hows and whys of this mmap extension. In case we ever want to add support for explicitly cache maneged userspace mmap with a prepare/finish ioctl pair, we could specify that userspace needs to mmap a different part of the dma_buf, e.g. the range starting at dma_buf->size up to dma_buf->size*2. This works because the size of a dma_buf is invariant over it's lifetime. The exporter would obviously need to fall back to coherent mappings for both ranges if a legacy clients maps the coherent range and the architecture cannot suppor conflicting caching policies. Also, this would obviously be optional and userspace needs to be able to fall back to coherent mappings. v2: - Spelling fixes from Rob Clark. - Compile fix for !DMA_BUF from Rob Clark. - Extend commit message to explain how explicitly cache managed mmap support could be added later. - Extend the documentation with implementations notes for exporters that need to manually fake coherency. v3: - dma_buf pointer initialization goof-up noticed by Rebecca Schultz Zavin. Cc: Rob Clark <rob.clark@linaro.org> Cc: Rebecca Schultz Zavin <rebecca@android.com> Acked-by: Rob Clark <rob.clark@linaro.org> Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org>
2012-04-24 17:08:52 +08:00
* @vma: [in] vma for the mmap
* @pgoff: [in] offset in pages where this mmap should start within the
* dma-buf buffer.
dma-buf: mmap support Compared to Rob Clark's RFC I've ditched the prepare/finish hooks and corresponding ioctls on the dma_buf file. The major reason for that is that many people seem to be under the impression that this is also for synchronization with outstanding asynchronous processsing. I'm pretty massively opposed to this because: - It boils down reinventing a new rather general-purpose userspace synchronization interface. If we look at things like futexes, this is hard to get right. - Furthermore a lot of kernel code has to interact with this synchronization primitive. This smells a look like the dri1 hw_lock, a horror show I prefer not to reinvent. - Even more fun is that multiple different subsystems would interact here, so we have plenty of opportunities to create funny deadlock scenarios. I think synchronization is a wholesale different problem from data sharing and should be tackled as an orthogonal problem. Now we could demand that prepare/finish may only ensure cache coherency (as Rob intended), but that runs up into the next problem: We not only need mmap support to facilitate sw-only processing nodes in a pipeline (without jumping through hoops by importing the dma_buf into some sw-access only importer), which allows for a nicer ION->dma-buf upgrade path for existing Android userspace. We also need mmap support for existing importing subsystems to support existing userspace libraries. And a loot of these subsystems are expected to export coherent userspace mappings. So prepare/finish can only ever be optional and the exporter /needs/ to support coherent mappings. Given that mmap access is always somewhat fallback-y in nature I've decided to drop this optimization, instead of just making it optional. If we demonstrate a clear need for this, supported by benchmark results, we can always add it in again later as an optional extension. Other differences compared to Rob's RFC is the above mentioned support for mapping a dma-buf through facilities provided by the importer. Which results in mmap support no longer being optional. Note that this dma-buf mmap patch does _not_ support every possible insanity an existing subsystem could pull of with mmap: Because it does not allow to intercept pagefaults and shoot down ptes importing subsystems can't add some magic of their own at these points (e.g. to automatically synchronize with outstanding rendering or set up some special resources). I've done a cursory read through a few mmap implementions of various subsytems and I'm hopeful that we can avoid this (and the complexity it'd bring with it). Additonally I've extended the documentation a bit to explain the hows and whys of this mmap extension. In case we ever want to add support for explicitly cache maneged userspace mmap with a prepare/finish ioctl pair, we could specify that userspace needs to mmap a different part of the dma_buf, e.g. the range starting at dma_buf->size up to dma_buf->size*2. This works because the size of a dma_buf is invariant over it's lifetime. The exporter would obviously need to fall back to coherent mappings for both ranges if a legacy clients maps the coherent range and the architecture cannot suppor conflicting caching policies. Also, this would obviously be optional and userspace needs to be able to fall back to coherent mappings. v2: - Spelling fixes from Rob Clark. - Compile fix for !DMA_BUF from Rob Clark. - Extend commit message to explain how explicitly cache managed mmap support could be added later. - Extend the documentation with implementations notes for exporters that need to manually fake coherency. v3: - dma_buf pointer initialization goof-up noticed by Rebecca Schultz Zavin. Cc: Rob Clark <rob.clark@linaro.org> Cc: Rebecca Schultz Zavin <rebecca@android.com> Acked-by: Rob Clark <rob.clark@linaro.org> Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org>
2012-04-24 17:08:52 +08:00
*
* This function adjusts the passed in vma so that it points at the file of the
* dma_buf operation. It also adjusts the starting pgoff and does bounds
dma-buf: mmap support Compared to Rob Clark's RFC I've ditched the prepare/finish hooks and corresponding ioctls on the dma_buf file. The major reason for that is that many people seem to be under the impression that this is also for synchronization with outstanding asynchronous processsing. I'm pretty massively opposed to this because: - It boils down reinventing a new rather general-purpose userspace synchronization interface. If we look at things like futexes, this is hard to get right. - Furthermore a lot of kernel code has to interact with this synchronization primitive. This smells a look like the dri1 hw_lock, a horror show I prefer not to reinvent. - Even more fun is that multiple different subsystems would interact here, so we have plenty of opportunities to create funny deadlock scenarios. I think synchronization is a wholesale different problem from data sharing and should be tackled as an orthogonal problem. Now we could demand that prepare/finish may only ensure cache coherency (as Rob intended), but that runs up into the next problem: We not only need mmap support to facilitate sw-only processing nodes in a pipeline (without jumping through hoops by importing the dma_buf into some sw-access only importer), which allows for a nicer ION->dma-buf upgrade path for existing Android userspace. We also need mmap support for existing importing subsystems to support existing userspace libraries. And a loot of these subsystems are expected to export coherent userspace mappings. So prepare/finish can only ever be optional and the exporter /needs/ to support coherent mappings. Given that mmap access is always somewhat fallback-y in nature I've decided to drop this optimization, instead of just making it optional. If we demonstrate a clear need for this, supported by benchmark results, we can always add it in again later as an optional extension. Other differences compared to Rob's RFC is the above mentioned support for mapping a dma-buf through facilities provided by the importer. Which results in mmap support no longer being optional. Note that this dma-buf mmap patch does _not_ support every possible insanity an existing subsystem could pull of with mmap: Because it does not allow to intercept pagefaults and shoot down ptes importing subsystems can't add some magic of their own at these points (e.g. to automatically synchronize with outstanding rendering or set up some special resources). I've done a cursory read through a few mmap implementions of various subsytems and I'm hopeful that we can avoid this (and the complexity it'd bring with it). Additonally I've extended the documentation a bit to explain the hows and whys of this mmap extension. In case we ever want to add support for explicitly cache maneged userspace mmap with a prepare/finish ioctl pair, we could specify that userspace needs to mmap a different part of the dma_buf, e.g. the range starting at dma_buf->size up to dma_buf->size*2. This works because the size of a dma_buf is invariant over it's lifetime. The exporter would obviously need to fall back to coherent mappings for both ranges if a legacy clients maps the coherent range and the architecture cannot suppor conflicting caching policies. Also, this would obviously be optional and userspace needs to be able to fall back to coherent mappings. v2: - Spelling fixes from Rob Clark. - Compile fix for !DMA_BUF from Rob Clark. - Extend commit message to explain how explicitly cache managed mmap support could be added later. - Extend the documentation with implementations notes for exporters that need to manually fake coherency. v3: - dma_buf pointer initialization goof-up noticed by Rebecca Schultz Zavin. Cc: Rob Clark <rob.clark@linaro.org> Cc: Rebecca Schultz Zavin <rebecca@android.com> Acked-by: Rob Clark <rob.clark@linaro.org> Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org>
2012-04-24 17:08:52 +08:00
* checking on the size of the vma. Then it calls the exporters mmap function to
* set up the mapping.
*
* Can return negative error values, returns 0 on success.
*/
int dma_buf_mmap(struct dma_buf *dmabuf, struct vm_area_struct *vma,
unsigned long pgoff)
{
if (WARN_ON(!dmabuf || !vma))
return -EINVAL;
/* check if buffer supports mmap */
if (!dmabuf->ops->mmap)
return -EINVAL;
dma-buf: mmap support Compared to Rob Clark's RFC I've ditched the prepare/finish hooks and corresponding ioctls on the dma_buf file. The major reason for that is that many people seem to be under the impression that this is also for synchronization with outstanding asynchronous processsing. I'm pretty massively opposed to this because: - It boils down reinventing a new rather general-purpose userspace synchronization interface. If we look at things like futexes, this is hard to get right. - Furthermore a lot of kernel code has to interact with this synchronization primitive. This smells a look like the dri1 hw_lock, a horror show I prefer not to reinvent. - Even more fun is that multiple different subsystems would interact here, so we have plenty of opportunities to create funny deadlock scenarios. I think synchronization is a wholesale different problem from data sharing and should be tackled as an orthogonal problem. Now we could demand that prepare/finish may only ensure cache coherency (as Rob intended), but that runs up into the next problem: We not only need mmap support to facilitate sw-only processing nodes in a pipeline (without jumping through hoops by importing the dma_buf into some sw-access only importer), which allows for a nicer ION->dma-buf upgrade path for existing Android userspace. We also need mmap support for existing importing subsystems to support existing userspace libraries. And a loot of these subsystems are expected to export coherent userspace mappings. So prepare/finish can only ever be optional and the exporter /needs/ to support coherent mappings. Given that mmap access is always somewhat fallback-y in nature I've decided to drop this optimization, instead of just making it optional. If we demonstrate a clear need for this, supported by benchmark results, we can always add it in again later as an optional extension. Other differences compared to Rob's RFC is the above mentioned support for mapping a dma-buf through facilities provided by the importer. Which results in mmap support no longer being optional. Note that this dma-buf mmap patch does _not_ support every possible insanity an existing subsystem could pull of with mmap: Because it does not allow to intercept pagefaults and shoot down ptes importing subsystems can't add some magic of their own at these points (e.g. to automatically synchronize with outstanding rendering or set up some special resources). I've done a cursory read through a few mmap implementions of various subsytems and I'm hopeful that we can avoid this (and the complexity it'd bring with it). Additonally I've extended the documentation a bit to explain the hows and whys of this mmap extension. In case we ever want to add support for explicitly cache maneged userspace mmap with a prepare/finish ioctl pair, we could specify that userspace needs to mmap a different part of the dma_buf, e.g. the range starting at dma_buf->size up to dma_buf->size*2. This works because the size of a dma_buf is invariant over it's lifetime. The exporter would obviously need to fall back to coherent mappings for both ranges if a legacy clients maps the coherent range and the architecture cannot suppor conflicting caching policies. Also, this would obviously be optional and userspace needs to be able to fall back to coherent mappings. v2: - Spelling fixes from Rob Clark. - Compile fix for !DMA_BUF from Rob Clark. - Extend commit message to explain how explicitly cache managed mmap support could be added later. - Extend the documentation with implementations notes for exporters that need to manually fake coherency. v3: - dma_buf pointer initialization goof-up noticed by Rebecca Schultz Zavin. Cc: Rob Clark <rob.clark@linaro.org> Cc: Rebecca Schultz Zavin <rebecca@android.com> Acked-by: Rob Clark <rob.clark@linaro.org> Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org>
2012-04-24 17:08:52 +08:00
/* check for offset overflow */
if (pgoff + vma_pages(vma) < pgoff)
dma-buf: mmap support Compared to Rob Clark's RFC I've ditched the prepare/finish hooks and corresponding ioctls on the dma_buf file. The major reason for that is that many people seem to be under the impression that this is also for synchronization with outstanding asynchronous processsing. I'm pretty massively opposed to this because: - It boils down reinventing a new rather general-purpose userspace synchronization interface. If we look at things like futexes, this is hard to get right. - Furthermore a lot of kernel code has to interact with this synchronization primitive. This smells a look like the dri1 hw_lock, a horror show I prefer not to reinvent. - Even more fun is that multiple different subsystems would interact here, so we have plenty of opportunities to create funny deadlock scenarios. I think synchronization is a wholesale different problem from data sharing and should be tackled as an orthogonal problem. Now we could demand that prepare/finish may only ensure cache coherency (as Rob intended), but that runs up into the next problem: We not only need mmap support to facilitate sw-only processing nodes in a pipeline (without jumping through hoops by importing the dma_buf into some sw-access only importer), which allows for a nicer ION->dma-buf upgrade path for existing Android userspace. We also need mmap support for existing importing subsystems to support existing userspace libraries. And a loot of these subsystems are expected to export coherent userspace mappings. So prepare/finish can only ever be optional and the exporter /needs/ to support coherent mappings. Given that mmap access is always somewhat fallback-y in nature I've decided to drop this optimization, instead of just making it optional. If we demonstrate a clear need for this, supported by benchmark results, we can always add it in again later as an optional extension. Other differences compared to Rob's RFC is the above mentioned support for mapping a dma-buf through facilities provided by the importer. Which results in mmap support no longer being optional. Note that this dma-buf mmap patch does _not_ support every possible insanity an existing subsystem could pull of with mmap: Because it does not allow to intercept pagefaults and shoot down ptes importing subsystems can't add some magic of their own at these points (e.g. to automatically synchronize with outstanding rendering or set up some special resources). I've done a cursory read through a few mmap implementions of various subsytems and I'm hopeful that we can avoid this (and the complexity it'd bring with it). Additonally I've extended the documentation a bit to explain the hows and whys of this mmap extension. In case we ever want to add support for explicitly cache maneged userspace mmap with a prepare/finish ioctl pair, we could specify that userspace needs to mmap a different part of the dma_buf, e.g. the range starting at dma_buf->size up to dma_buf->size*2. This works because the size of a dma_buf is invariant over it's lifetime. The exporter would obviously need to fall back to coherent mappings for both ranges if a legacy clients maps the coherent range and the architecture cannot suppor conflicting caching policies. Also, this would obviously be optional and userspace needs to be able to fall back to coherent mappings. v2: - Spelling fixes from Rob Clark. - Compile fix for !DMA_BUF from Rob Clark. - Extend commit message to explain how explicitly cache managed mmap support could be added later. - Extend the documentation with implementations notes for exporters that need to manually fake coherency. v3: - dma_buf pointer initialization goof-up noticed by Rebecca Schultz Zavin. Cc: Rob Clark <rob.clark@linaro.org> Cc: Rebecca Schultz Zavin <rebecca@android.com> Acked-by: Rob Clark <rob.clark@linaro.org> Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org>
2012-04-24 17:08:52 +08:00
return -EOVERFLOW;
/* check for overflowing the buffer's size */
if (pgoff + vma_pages(vma) >
dma-buf: mmap support Compared to Rob Clark's RFC I've ditched the prepare/finish hooks and corresponding ioctls on the dma_buf file. The major reason for that is that many people seem to be under the impression that this is also for synchronization with outstanding asynchronous processsing. I'm pretty massively opposed to this because: - It boils down reinventing a new rather general-purpose userspace synchronization interface. If we look at things like futexes, this is hard to get right. - Furthermore a lot of kernel code has to interact with this synchronization primitive. This smells a look like the dri1 hw_lock, a horror show I prefer not to reinvent. - Even more fun is that multiple different subsystems would interact here, so we have plenty of opportunities to create funny deadlock scenarios. I think synchronization is a wholesale different problem from data sharing and should be tackled as an orthogonal problem. Now we could demand that prepare/finish may only ensure cache coherency (as Rob intended), but that runs up into the next problem: We not only need mmap support to facilitate sw-only processing nodes in a pipeline (without jumping through hoops by importing the dma_buf into some sw-access only importer), which allows for a nicer ION->dma-buf upgrade path for existing Android userspace. We also need mmap support for existing importing subsystems to support existing userspace libraries. And a loot of these subsystems are expected to export coherent userspace mappings. So prepare/finish can only ever be optional and the exporter /needs/ to support coherent mappings. Given that mmap access is always somewhat fallback-y in nature I've decided to drop this optimization, instead of just making it optional. If we demonstrate a clear need for this, supported by benchmark results, we can always add it in again later as an optional extension. Other differences compared to Rob's RFC is the above mentioned support for mapping a dma-buf through facilities provided by the importer. Which results in mmap support no longer being optional. Note that this dma-buf mmap patch does _not_ support every possible insanity an existing subsystem could pull of with mmap: Because it does not allow to intercept pagefaults and shoot down ptes importing subsystems can't add some magic of their own at these points (e.g. to automatically synchronize with outstanding rendering or set up some special resources). I've done a cursory read through a few mmap implementions of various subsytems and I'm hopeful that we can avoid this (and the complexity it'd bring with it). Additonally I've extended the documentation a bit to explain the hows and whys of this mmap extension. In case we ever want to add support for explicitly cache maneged userspace mmap with a prepare/finish ioctl pair, we could specify that userspace needs to mmap a different part of the dma_buf, e.g. the range starting at dma_buf->size up to dma_buf->size*2. This works because the size of a dma_buf is invariant over it's lifetime. The exporter would obviously need to fall back to coherent mappings for both ranges if a legacy clients maps the coherent range and the architecture cannot suppor conflicting caching policies. Also, this would obviously be optional and userspace needs to be able to fall back to coherent mappings. v2: - Spelling fixes from Rob Clark. - Compile fix for !DMA_BUF from Rob Clark. - Extend commit message to explain how explicitly cache managed mmap support could be added later. - Extend the documentation with implementations notes for exporters that need to manually fake coherency. v3: - dma_buf pointer initialization goof-up noticed by Rebecca Schultz Zavin. Cc: Rob Clark <rob.clark@linaro.org> Cc: Rebecca Schultz Zavin <rebecca@android.com> Acked-by: Rob Clark <rob.clark@linaro.org> Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org>
2012-04-24 17:08:52 +08:00
dmabuf->size >> PAGE_SHIFT)
return -EINVAL;
/* readjust the vma */
vma_set_file(vma, dmabuf->file);
dma-buf: mmap support Compared to Rob Clark's RFC I've ditched the prepare/finish hooks and corresponding ioctls on the dma_buf file. The major reason for that is that many people seem to be under the impression that this is also for synchronization with outstanding asynchronous processsing. I'm pretty massively opposed to this because: - It boils down reinventing a new rather general-purpose userspace synchronization interface. If we look at things like futexes, this is hard to get right. - Furthermore a lot of kernel code has to interact with this synchronization primitive. This smells a look like the dri1 hw_lock, a horror show I prefer not to reinvent. - Even more fun is that multiple different subsystems would interact here, so we have plenty of opportunities to create funny deadlock scenarios. I think synchronization is a wholesale different problem from data sharing and should be tackled as an orthogonal problem. Now we could demand that prepare/finish may only ensure cache coherency (as Rob intended), but that runs up into the next problem: We not only need mmap support to facilitate sw-only processing nodes in a pipeline (without jumping through hoops by importing the dma_buf into some sw-access only importer), which allows for a nicer ION->dma-buf upgrade path for existing Android userspace. We also need mmap support for existing importing subsystems to support existing userspace libraries. And a loot of these subsystems are expected to export coherent userspace mappings. So prepare/finish can only ever be optional and the exporter /needs/ to support coherent mappings. Given that mmap access is always somewhat fallback-y in nature I've decided to drop this optimization, instead of just making it optional. If we demonstrate a clear need for this, supported by benchmark results, we can always add it in again later as an optional extension. Other differences compared to Rob's RFC is the above mentioned support for mapping a dma-buf through facilities provided by the importer. Which results in mmap support no longer being optional. Note that this dma-buf mmap patch does _not_ support every possible insanity an existing subsystem could pull of with mmap: Because it does not allow to intercept pagefaults and shoot down ptes importing subsystems can't add some magic of their own at these points (e.g. to automatically synchronize with outstanding rendering or set up some special resources). I've done a cursory read through a few mmap implementions of various subsytems and I'm hopeful that we can avoid this (and the complexity it'd bring with it). Additonally I've extended the documentation a bit to explain the hows and whys of this mmap extension. In case we ever want to add support for explicitly cache maneged userspace mmap with a prepare/finish ioctl pair, we could specify that userspace needs to mmap a different part of the dma_buf, e.g. the range starting at dma_buf->size up to dma_buf->size*2. This works because the size of a dma_buf is invariant over it's lifetime. The exporter would obviously need to fall back to coherent mappings for both ranges if a legacy clients maps the coherent range and the architecture cannot suppor conflicting caching policies. Also, this would obviously be optional and userspace needs to be able to fall back to coherent mappings. v2: - Spelling fixes from Rob Clark. - Compile fix for !DMA_BUF from Rob Clark. - Extend commit message to explain how explicitly cache managed mmap support could be added later. - Extend the documentation with implementations notes for exporters that need to manually fake coherency. v3: - dma_buf pointer initialization goof-up noticed by Rebecca Schultz Zavin. Cc: Rob Clark <rob.clark@linaro.org> Cc: Rebecca Schultz Zavin <rebecca@android.com> Acked-by: Rob Clark <rob.clark@linaro.org> Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org>
2012-04-24 17:08:52 +08:00
vma->vm_pgoff = pgoff;
return dmabuf->ops->mmap(dmabuf, vma);
dma-buf: mmap support Compared to Rob Clark's RFC I've ditched the prepare/finish hooks and corresponding ioctls on the dma_buf file. The major reason for that is that many people seem to be under the impression that this is also for synchronization with outstanding asynchronous processsing. I'm pretty massively opposed to this because: - It boils down reinventing a new rather general-purpose userspace synchronization interface. If we look at things like futexes, this is hard to get right. - Furthermore a lot of kernel code has to interact with this synchronization primitive. This smells a look like the dri1 hw_lock, a horror show I prefer not to reinvent. - Even more fun is that multiple different subsystems would interact here, so we have plenty of opportunities to create funny deadlock scenarios. I think synchronization is a wholesale different problem from data sharing and should be tackled as an orthogonal problem. Now we could demand that prepare/finish may only ensure cache coherency (as Rob intended), but that runs up into the next problem: We not only need mmap support to facilitate sw-only processing nodes in a pipeline (without jumping through hoops by importing the dma_buf into some sw-access only importer), which allows for a nicer ION->dma-buf upgrade path for existing Android userspace. We also need mmap support for existing importing subsystems to support existing userspace libraries. And a loot of these subsystems are expected to export coherent userspace mappings. So prepare/finish can only ever be optional and the exporter /needs/ to support coherent mappings. Given that mmap access is always somewhat fallback-y in nature I've decided to drop this optimization, instead of just making it optional. If we demonstrate a clear need for this, supported by benchmark results, we can always add it in again later as an optional extension. Other differences compared to Rob's RFC is the above mentioned support for mapping a dma-buf through facilities provided by the importer. Which results in mmap support no longer being optional. Note that this dma-buf mmap patch does _not_ support every possible insanity an existing subsystem could pull of with mmap: Because it does not allow to intercept pagefaults and shoot down ptes importing subsystems can't add some magic of their own at these points (e.g. to automatically synchronize with outstanding rendering or set up some special resources). I've done a cursory read through a few mmap implementions of various subsytems and I'm hopeful that we can avoid this (and the complexity it'd bring with it). Additonally I've extended the documentation a bit to explain the hows and whys of this mmap extension. In case we ever want to add support for explicitly cache maneged userspace mmap with a prepare/finish ioctl pair, we could specify that userspace needs to mmap a different part of the dma_buf, e.g. the range starting at dma_buf->size up to dma_buf->size*2. This works because the size of a dma_buf is invariant over it's lifetime. The exporter would obviously need to fall back to coherent mappings for both ranges if a legacy clients maps the coherent range and the architecture cannot suppor conflicting caching policies. Also, this would obviously be optional and userspace needs to be able to fall back to coherent mappings. v2: - Spelling fixes from Rob Clark. - Compile fix for !DMA_BUF from Rob Clark. - Extend commit message to explain how explicitly cache managed mmap support could be added later. - Extend the documentation with implementations notes for exporters that need to manually fake coherency. v3: - dma_buf pointer initialization goof-up noticed by Rebecca Schultz Zavin. Cc: Rob Clark <rob.clark@linaro.org> Cc: Rebecca Schultz Zavin <rebecca@android.com> Acked-by: Rob Clark <rob.clark@linaro.org> Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Signed-off-by: Sumit Semwal <sumit.semwal@linaro.org>
2012-04-24 17:08:52 +08:00
}
EXPORT_SYMBOL_NS_GPL(dma_buf_mmap, DMA_BUF);
/**
* dma_buf_vmap - Create virtual mapping for the buffer object into kernel
* address space. Same restrictions as for vmap and friends apply.
* @dmabuf: [in] buffer to vmap
* @map: [out] returns the vmap pointer
*
* This call may fail due to lack of virtual mapping address space.
* These calls are optional in drivers. The intended use for them
* is for mapping objects linear in kernel space for high use objects.
*
* To ensure coherency users must call dma_buf_begin_cpu_access() and
* dma_buf_end_cpu_access() around any cpu access performed through this
* mapping.
*
* Returns 0 on success, or a negative errno code otherwise.
*/
dma-buf-map: Rename to iosys-map Rename struct dma_buf_map to struct iosys_map and corresponding APIs. Over time dma-buf-map grew up to more functionality than the one used by dma-buf: in fact it's just a shim layer to abstract system memory, that can be accessed via regular load and store, from IO memory that needs to be acessed via arch helpers. The idea is to extend this API so it can fulfill other needs, internal to a single driver. Example: in the i915 driver it's desired to share the implementation for integrated graphics, which uses mostly system memory, with discrete graphics, which may need to access IO memory. The conversion was mostly done with the following semantic patch: @r1@ @@ - struct dma_buf_map + struct iosys_map @r2@ @@ ( - DMA_BUF_MAP_INIT_VADDR + IOSYS_MAP_INIT_VADDR | - dma_buf_map_set_vaddr + iosys_map_set_vaddr | - dma_buf_map_set_vaddr_iomem + iosys_map_set_vaddr_iomem | - dma_buf_map_is_equal + iosys_map_is_equal | - dma_buf_map_is_null + iosys_map_is_null | - dma_buf_map_is_set + iosys_map_is_set | - dma_buf_map_clear + iosys_map_clear | - dma_buf_map_memcpy_to + iosys_map_memcpy_to | - dma_buf_map_incr + iosys_map_incr ) @@ @@ - #include <linux/dma-buf-map.h> + #include <linux/iosys-map.h> Then some files had their includes adjusted and some comments were update to remove mentions to dma-buf-map. Since this is not specific to dma-buf anymore, move the documentation to the "Bus-Independent Device Accesses" section. v2: - Squash patches v3: - Fix wrong removal of dma-buf.h from MAINTAINERS - Move documentation from dma-buf.rst to device-io.rst v4: - Change documentation title and level Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com> Acked-by: Christian König <christian.koenig@amd.com> Acked-by: Sumit Semwal <sumit.semwal@linaro.org> Acked-by: Thomas Zimmermann <tzimmermann@suse.de> Link: https://patchwork.freedesktop.org/patch/msgid/20220204170541.829227-1-lucas.demarchi@intel.com
2022-02-05 01:05:41 +08:00
int dma_buf_vmap(struct dma_buf *dmabuf, struct iosys_map *map)
{
dma-buf-map: Rename to iosys-map Rename struct dma_buf_map to struct iosys_map and corresponding APIs. Over time dma-buf-map grew up to more functionality than the one used by dma-buf: in fact it's just a shim layer to abstract system memory, that can be accessed via regular load and store, from IO memory that needs to be acessed via arch helpers. The idea is to extend this API so it can fulfill other needs, internal to a single driver. Example: in the i915 driver it's desired to share the implementation for integrated graphics, which uses mostly system memory, with discrete graphics, which may need to access IO memory. The conversion was mostly done with the following semantic patch: @r1@ @@ - struct dma_buf_map + struct iosys_map @r2@ @@ ( - DMA_BUF_MAP_INIT_VADDR + IOSYS_MAP_INIT_VADDR | - dma_buf_map_set_vaddr + iosys_map_set_vaddr | - dma_buf_map_set_vaddr_iomem + iosys_map_set_vaddr_iomem | - dma_buf_map_is_equal + iosys_map_is_equal | - dma_buf_map_is_null + iosys_map_is_null | - dma_buf_map_is_set + iosys_map_is_set | - dma_buf_map_clear + iosys_map_clear | - dma_buf_map_memcpy_to + iosys_map_memcpy_to | - dma_buf_map_incr + iosys_map_incr ) @@ @@ - #include <linux/dma-buf-map.h> + #include <linux/iosys-map.h> Then some files had their includes adjusted and some comments were update to remove mentions to dma-buf-map. Since this is not specific to dma-buf anymore, move the documentation to the "Bus-Independent Device Accesses" section. v2: - Squash patches v3: - Fix wrong removal of dma-buf.h from MAINTAINERS - Move documentation from dma-buf.rst to device-io.rst v4: - Change documentation title and level Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com> Acked-by: Christian König <christian.koenig@amd.com> Acked-by: Sumit Semwal <sumit.semwal@linaro.org> Acked-by: Thomas Zimmermann <tzimmermann@suse.de> Link: https://patchwork.freedesktop.org/patch/msgid/20220204170541.829227-1-lucas.demarchi@intel.com
2022-02-05 01:05:41 +08:00
struct iosys_map ptr;
int ret;
dma-buf-map: Rename to iosys-map Rename struct dma_buf_map to struct iosys_map and corresponding APIs. Over time dma-buf-map grew up to more functionality than the one used by dma-buf: in fact it's just a shim layer to abstract system memory, that can be accessed via regular load and store, from IO memory that needs to be acessed via arch helpers. The idea is to extend this API so it can fulfill other needs, internal to a single driver. Example: in the i915 driver it's desired to share the implementation for integrated graphics, which uses mostly system memory, with discrete graphics, which may need to access IO memory. The conversion was mostly done with the following semantic patch: @r1@ @@ - struct dma_buf_map + struct iosys_map @r2@ @@ ( - DMA_BUF_MAP_INIT_VADDR + IOSYS_MAP_INIT_VADDR | - dma_buf_map_set_vaddr + iosys_map_set_vaddr | - dma_buf_map_set_vaddr_iomem + iosys_map_set_vaddr_iomem | - dma_buf_map_is_equal + iosys_map_is_equal | - dma_buf_map_is_null + iosys_map_is_null | - dma_buf_map_is_set + iosys_map_is_set | - dma_buf_map_clear + iosys_map_clear | - dma_buf_map_memcpy_to + iosys_map_memcpy_to | - dma_buf_map_incr + iosys_map_incr ) @@ @@ - #include <linux/dma-buf-map.h> + #include <linux/iosys-map.h> Then some files had their includes adjusted and some comments were update to remove mentions to dma-buf-map. Since this is not specific to dma-buf anymore, move the documentation to the "Bus-Independent Device Accesses" section. v2: - Squash patches v3: - Fix wrong removal of dma-buf.h from MAINTAINERS - Move documentation from dma-buf.rst to device-io.rst v4: - Change documentation title and level Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com> Acked-by: Christian König <christian.koenig@amd.com> Acked-by: Sumit Semwal <sumit.semwal@linaro.org> Acked-by: Thomas Zimmermann <tzimmermann@suse.de> Link: https://patchwork.freedesktop.org/patch/msgid/20220204170541.829227-1-lucas.demarchi@intel.com
2022-02-05 01:05:41 +08:00
iosys_map_clear(map);
if (WARN_ON(!dmabuf))
return -EINVAL;
dma_resv_assert_held(dmabuf->resv);
if (!dmabuf->ops->vmap)
return -EINVAL;
if (dmabuf->vmapping_counter) {
dmabuf->vmapping_counter++;
dma-buf-map: Rename to iosys-map Rename struct dma_buf_map to struct iosys_map and corresponding APIs. Over time dma-buf-map grew up to more functionality than the one used by dma-buf: in fact it's just a shim layer to abstract system memory, that can be accessed via regular load and store, from IO memory that needs to be acessed via arch helpers. The idea is to extend this API so it can fulfill other needs, internal to a single driver. Example: in the i915 driver it's desired to share the implementation for integrated graphics, which uses mostly system memory, with discrete graphics, which may need to access IO memory. The conversion was mostly done with the following semantic patch: @r1@ @@ - struct dma_buf_map + struct iosys_map @r2@ @@ ( - DMA_BUF_MAP_INIT_VADDR + IOSYS_MAP_INIT_VADDR | - dma_buf_map_set_vaddr + iosys_map_set_vaddr | - dma_buf_map_set_vaddr_iomem + iosys_map_set_vaddr_iomem | - dma_buf_map_is_equal + iosys_map_is_equal | - dma_buf_map_is_null + iosys_map_is_null | - dma_buf_map_is_set + iosys_map_is_set | - dma_buf_map_clear + iosys_map_clear | - dma_buf_map_memcpy_to + iosys_map_memcpy_to | - dma_buf_map_incr + iosys_map_incr ) @@ @@ - #include <linux/dma-buf-map.h> + #include <linux/iosys-map.h> Then some files had their includes adjusted and some comments were update to remove mentions to dma-buf-map. Since this is not specific to dma-buf anymore, move the documentation to the "Bus-Independent Device Accesses" section. v2: - Squash patches v3: - Fix wrong removal of dma-buf.h from MAINTAINERS - Move documentation from dma-buf.rst to device-io.rst v4: - Change documentation title and level Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com> Acked-by: Christian König <christian.koenig@amd.com> Acked-by: Sumit Semwal <sumit.semwal@linaro.org> Acked-by: Thomas Zimmermann <tzimmermann@suse.de> Link: https://patchwork.freedesktop.org/patch/msgid/20220204170541.829227-1-lucas.demarchi@intel.com
2022-02-05 01:05:41 +08:00
BUG_ON(iosys_map_is_null(&dmabuf->vmap_ptr));
*map = dmabuf->vmap_ptr;
return 0;
}
dma-buf-map: Rename to iosys-map Rename struct dma_buf_map to struct iosys_map and corresponding APIs. Over time dma-buf-map grew up to more functionality than the one used by dma-buf: in fact it's just a shim layer to abstract system memory, that can be accessed via regular load and store, from IO memory that needs to be acessed via arch helpers. The idea is to extend this API so it can fulfill other needs, internal to a single driver. Example: in the i915 driver it's desired to share the implementation for integrated graphics, which uses mostly system memory, with discrete graphics, which may need to access IO memory. The conversion was mostly done with the following semantic patch: @r1@ @@ - struct dma_buf_map + struct iosys_map @r2@ @@ ( - DMA_BUF_MAP_INIT_VADDR + IOSYS_MAP_INIT_VADDR | - dma_buf_map_set_vaddr + iosys_map_set_vaddr | - dma_buf_map_set_vaddr_iomem + iosys_map_set_vaddr_iomem | - dma_buf_map_is_equal + iosys_map_is_equal | - dma_buf_map_is_null + iosys_map_is_null | - dma_buf_map_is_set + iosys_map_is_set | - dma_buf_map_clear + iosys_map_clear | - dma_buf_map_memcpy_to + iosys_map_memcpy_to | - dma_buf_map_incr + iosys_map_incr ) @@ @@ - #include <linux/dma-buf-map.h> + #include <linux/iosys-map.h> Then some files had their includes adjusted and some comments were update to remove mentions to dma-buf-map. Since this is not specific to dma-buf anymore, move the documentation to the "Bus-Independent Device Accesses" section. v2: - Squash patches v3: - Fix wrong removal of dma-buf.h from MAINTAINERS - Move documentation from dma-buf.rst to device-io.rst v4: - Change documentation title and level Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com> Acked-by: Christian König <christian.koenig@amd.com> Acked-by: Sumit Semwal <sumit.semwal@linaro.org> Acked-by: Thomas Zimmermann <tzimmermann@suse.de> Link: https://patchwork.freedesktop.org/patch/msgid/20220204170541.829227-1-lucas.demarchi@intel.com
2022-02-05 01:05:41 +08:00
BUG_ON(iosys_map_is_set(&dmabuf->vmap_ptr));
ret = dmabuf->ops->vmap(dmabuf, &ptr);
if (WARN_ON_ONCE(ret))
return ret;
dmabuf->vmap_ptr = ptr;
dmabuf->vmapping_counter = 1;
*map = dmabuf->vmap_ptr;
return 0;
}
EXPORT_SYMBOL_NS_GPL(dma_buf_vmap, DMA_BUF);
/**
* dma_buf_vmap_unlocked - Create virtual mapping for the buffer object into kernel
* address space. Same restrictions as for vmap and friends apply.
* @dmabuf: [in] buffer to vmap
* @map: [out] returns the vmap pointer
*
* Unlocked version of dma_buf_vmap()
*
* Returns 0 on success, or a negative errno code otherwise.
*/
int dma_buf_vmap_unlocked(struct dma_buf *dmabuf, struct iosys_map *map)
{
int ret;
iosys_map_clear(map);
if (WARN_ON(!dmabuf))
return -EINVAL;
dma_resv_lock(dmabuf->resv, NULL);
ret = dma_buf_vmap(dmabuf, map);
dma_resv_unlock(dmabuf->resv);
return ret;
}
EXPORT_SYMBOL_NS_GPL(dma_buf_vmap_unlocked, DMA_BUF);
/**
* dma_buf_vunmap - Unmap a vmap obtained by dma_buf_vmap.
* @dmabuf: [in] buffer to vunmap
* @map: [in] vmap pointer to vunmap
*/
dma-buf-map: Rename to iosys-map Rename struct dma_buf_map to struct iosys_map and corresponding APIs. Over time dma-buf-map grew up to more functionality than the one used by dma-buf: in fact it's just a shim layer to abstract system memory, that can be accessed via regular load and store, from IO memory that needs to be acessed via arch helpers. The idea is to extend this API so it can fulfill other needs, internal to a single driver. Example: in the i915 driver it's desired to share the implementation for integrated graphics, which uses mostly system memory, with discrete graphics, which may need to access IO memory. The conversion was mostly done with the following semantic patch: @r1@ @@ - struct dma_buf_map + struct iosys_map @r2@ @@ ( - DMA_BUF_MAP_INIT_VADDR + IOSYS_MAP_INIT_VADDR | - dma_buf_map_set_vaddr + iosys_map_set_vaddr | - dma_buf_map_set_vaddr_iomem + iosys_map_set_vaddr_iomem | - dma_buf_map_is_equal + iosys_map_is_equal | - dma_buf_map_is_null + iosys_map_is_null | - dma_buf_map_is_set + iosys_map_is_set | - dma_buf_map_clear + iosys_map_clear | - dma_buf_map_memcpy_to + iosys_map_memcpy_to | - dma_buf_map_incr + iosys_map_incr ) @@ @@ - #include <linux/dma-buf-map.h> + #include <linux/iosys-map.h> Then some files had their includes adjusted and some comments were update to remove mentions to dma-buf-map. Since this is not specific to dma-buf anymore, move the documentation to the "Bus-Independent Device Accesses" section. v2: - Squash patches v3: - Fix wrong removal of dma-buf.h from MAINTAINERS - Move documentation from dma-buf.rst to device-io.rst v4: - Change documentation title and level Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com> Acked-by: Christian König <christian.koenig@amd.com> Acked-by: Sumit Semwal <sumit.semwal@linaro.org> Acked-by: Thomas Zimmermann <tzimmermann@suse.de> Link: https://patchwork.freedesktop.org/patch/msgid/20220204170541.829227-1-lucas.demarchi@intel.com
2022-02-05 01:05:41 +08:00
void dma_buf_vunmap(struct dma_buf *dmabuf, struct iosys_map *map)
{
if (WARN_ON(!dmabuf))
return;
dma_resv_assert_held(dmabuf->resv);
dma-buf-map: Rename to iosys-map Rename struct dma_buf_map to struct iosys_map and corresponding APIs. Over time dma-buf-map grew up to more functionality than the one used by dma-buf: in fact it's just a shim layer to abstract system memory, that can be accessed via regular load and store, from IO memory that needs to be acessed via arch helpers. The idea is to extend this API so it can fulfill other needs, internal to a single driver. Example: in the i915 driver it's desired to share the implementation for integrated graphics, which uses mostly system memory, with discrete graphics, which may need to access IO memory. The conversion was mostly done with the following semantic patch: @r1@ @@ - struct dma_buf_map + struct iosys_map @r2@ @@ ( - DMA_BUF_MAP_INIT_VADDR + IOSYS_MAP_INIT_VADDR | - dma_buf_map_set_vaddr + iosys_map_set_vaddr | - dma_buf_map_set_vaddr_iomem + iosys_map_set_vaddr_iomem | - dma_buf_map_is_equal + iosys_map_is_equal | - dma_buf_map_is_null + iosys_map_is_null | - dma_buf_map_is_set + iosys_map_is_set | - dma_buf_map_clear + iosys_map_clear | - dma_buf_map_memcpy_to + iosys_map_memcpy_to | - dma_buf_map_incr + iosys_map_incr ) @@ @@ - #include <linux/dma-buf-map.h> + #include <linux/iosys-map.h> Then some files had their includes adjusted and some comments were update to remove mentions to dma-buf-map. Since this is not specific to dma-buf anymore, move the documentation to the "Bus-Independent Device Accesses" section. v2: - Squash patches v3: - Fix wrong removal of dma-buf.h from MAINTAINERS - Move documentation from dma-buf.rst to device-io.rst v4: - Change documentation title and level Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com> Acked-by: Christian König <christian.koenig@amd.com> Acked-by: Sumit Semwal <sumit.semwal@linaro.org> Acked-by: Thomas Zimmermann <tzimmermann@suse.de> Link: https://patchwork.freedesktop.org/patch/msgid/20220204170541.829227-1-lucas.demarchi@intel.com
2022-02-05 01:05:41 +08:00
BUG_ON(iosys_map_is_null(&dmabuf->vmap_ptr));
BUG_ON(dmabuf->vmapping_counter == 0);
dma-buf-map: Rename to iosys-map Rename struct dma_buf_map to struct iosys_map and corresponding APIs. Over time dma-buf-map grew up to more functionality than the one used by dma-buf: in fact it's just a shim layer to abstract system memory, that can be accessed via regular load and store, from IO memory that needs to be acessed via arch helpers. The idea is to extend this API so it can fulfill other needs, internal to a single driver. Example: in the i915 driver it's desired to share the implementation for integrated graphics, which uses mostly system memory, with discrete graphics, which may need to access IO memory. The conversion was mostly done with the following semantic patch: @r1@ @@ - struct dma_buf_map + struct iosys_map @r2@ @@ ( - DMA_BUF_MAP_INIT_VADDR + IOSYS_MAP_INIT_VADDR | - dma_buf_map_set_vaddr + iosys_map_set_vaddr | - dma_buf_map_set_vaddr_iomem + iosys_map_set_vaddr_iomem | - dma_buf_map_is_equal + iosys_map_is_equal | - dma_buf_map_is_null + iosys_map_is_null | - dma_buf_map_is_set + iosys_map_is_set | - dma_buf_map_clear + iosys_map_clear | - dma_buf_map_memcpy_to + iosys_map_memcpy_to | - dma_buf_map_incr + iosys_map_incr ) @@ @@ - #include <linux/dma-buf-map.h> + #include <linux/iosys-map.h> Then some files had their includes adjusted and some comments were update to remove mentions to dma-buf-map. Since this is not specific to dma-buf anymore, move the documentation to the "Bus-Independent Device Accesses" section. v2: - Squash patches v3: - Fix wrong removal of dma-buf.h from MAINTAINERS - Move documentation from dma-buf.rst to device-io.rst v4: - Change documentation title and level Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com> Acked-by: Christian König <christian.koenig@amd.com> Acked-by: Sumit Semwal <sumit.semwal@linaro.org> Acked-by: Thomas Zimmermann <tzimmermann@suse.de> Link: https://patchwork.freedesktop.org/patch/msgid/20220204170541.829227-1-lucas.demarchi@intel.com
2022-02-05 01:05:41 +08:00
BUG_ON(!iosys_map_is_equal(&dmabuf->vmap_ptr, map));
if (--dmabuf->vmapping_counter == 0) {
if (dmabuf->ops->vunmap)
dmabuf->ops->vunmap(dmabuf, map);
dma-buf-map: Rename to iosys-map Rename struct dma_buf_map to struct iosys_map and corresponding APIs. Over time dma-buf-map grew up to more functionality than the one used by dma-buf: in fact it's just a shim layer to abstract system memory, that can be accessed via regular load and store, from IO memory that needs to be acessed via arch helpers. The idea is to extend this API so it can fulfill other needs, internal to a single driver. Example: in the i915 driver it's desired to share the implementation for integrated graphics, which uses mostly system memory, with discrete graphics, which may need to access IO memory. The conversion was mostly done with the following semantic patch: @r1@ @@ - struct dma_buf_map + struct iosys_map @r2@ @@ ( - DMA_BUF_MAP_INIT_VADDR + IOSYS_MAP_INIT_VADDR | - dma_buf_map_set_vaddr + iosys_map_set_vaddr | - dma_buf_map_set_vaddr_iomem + iosys_map_set_vaddr_iomem | - dma_buf_map_is_equal + iosys_map_is_equal | - dma_buf_map_is_null + iosys_map_is_null | - dma_buf_map_is_set + iosys_map_is_set | - dma_buf_map_clear + iosys_map_clear | - dma_buf_map_memcpy_to + iosys_map_memcpy_to | - dma_buf_map_incr + iosys_map_incr ) @@ @@ - #include <linux/dma-buf-map.h> + #include <linux/iosys-map.h> Then some files had their includes adjusted and some comments were update to remove mentions to dma-buf-map. Since this is not specific to dma-buf anymore, move the documentation to the "Bus-Independent Device Accesses" section. v2: - Squash patches v3: - Fix wrong removal of dma-buf.h from MAINTAINERS - Move documentation from dma-buf.rst to device-io.rst v4: - Change documentation title and level Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com> Acked-by: Christian König <christian.koenig@amd.com> Acked-by: Sumit Semwal <sumit.semwal@linaro.org> Acked-by: Thomas Zimmermann <tzimmermann@suse.de> Link: https://patchwork.freedesktop.org/patch/msgid/20220204170541.829227-1-lucas.demarchi@intel.com
2022-02-05 01:05:41 +08:00
iosys_map_clear(&dmabuf->vmap_ptr);
}
}
EXPORT_SYMBOL_NS_GPL(dma_buf_vunmap, DMA_BUF);
/**
* dma_buf_vunmap_unlocked - Unmap a vmap obtained by dma_buf_vmap.
* @dmabuf: [in] buffer to vunmap
* @map: [in] vmap pointer to vunmap
*/
void dma_buf_vunmap_unlocked(struct dma_buf *dmabuf, struct iosys_map *map)
{
if (WARN_ON(!dmabuf))
return;
dma_resv_lock(dmabuf->resv, NULL);
dma_buf_vunmap(dmabuf, map);
dma_resv_unlock(dmabuf->resv);
}
EXPORT_SYMBOL_NS_GPL(dma_buf_vunmap_unlocked, DMA_BUF);
#ifdef CONFIG_DEBUG_FS
static int dma_buf_debug_show(struct seq_file *s, void *unused)
{
struct dma_buf *buf_obj;
struct dma_buf_attachment *attach_obj;
int count = 0, attach_count;
size_t size = 0;
int ret;
ret = mutex_lock_interruptible(&debugfs_list_mutex);
if (ret)
return ret;
seq_puts(s, "\nDma-buf Objects:\n");
seq_printf(s, "%-8s\t%-8s\t%-8s\t%-8s\texp_name\t%-8s\tname\n",
"size", "flags", "mode", "count", "ino");
list_for_each_entry(buf_obj, &debugfs_list, list_node) {
dma-buf: change DMA-buf locking convention v3 This patch is a stripped down version of the locking changes necessary to support dynamic DMA-buf handling. It adds a dynamic flag for both importers as well as exporters so that drivers can choose if they want the reservation object locked or unlocked during mapping of attachments. For compatibility between drivers we cache the DMA-buf mapping during attaching an importer as soon as exporter/importer disagree on the dynamic handling. Issues and solutions we considered: - We can't change all existing drivers, and existing improters have strong opinions about which locks they're holding while calling dma_buf_attachment_map/unmap. Exporters also have strong opinions about which locks they can acquire in their ->map/unmap callbacks, levaing no room for change. The solution to avoid this was to move the actual map/unmap out from this call, into the attach/detach callbacks, and cache the mapping. This works because drivers don't call attach/detach from deep within their code callchains (like deep in memory management code called from cs/execbuf ioctl), but directly from the fd2handle implementation. - The caching has some troubles on some soc drivers, which set other modes than DMA_BIDIRECTIONAL. We can't have 2 incompatible mappings, and we can't re-create the mapping at _map time due to the above locking fun. We very carefuly step around that by only caching at attach time if the dynamic mode between importer/expoert mismatches. - There's been quite some discussion on dma-buf mappings which need active cache management, which would all break down when caching, plus we don't have explicit flush operations on the attachment side. The solution to this was to shrug and keep the current discrepancy between what the dma-buf docs claim and what implementations do, with the hope that the begin/end_cpu_access hooks are good enough and that all necessary flushing to keep device mappings consistent will be done there. v2: cleanup set_name merge, improve kerneldoc v3: update commit message, kerneldoc and cleanup _debug_show() Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/336788/
2018-07-03 22:42:26 +08:00
ret = dma_resv_lock_interruptible(buf_obj->resv, NULL);
if (ret)
goto error_unlock;
spin_lock(&buf_obj->name_lock);
seq_printf(s, "%08zu\t%08x\t%08x\t%08ld\t%s\t%08lu\t%s\n",
buf_obj->size,
buf_obj->file->f_flags, buf_obj->file->f_mode,
file_count(buf_obj->file),
buf_obj->exp_name,
file_inode(buf_obj->file)->i_ino,
buf_obj->name ?: "<none>");
spin_unlock(&buf_obj->name_lock);
dma_resv_describe(buf_obj->resv, s);
seq_puts(s, "\tAttached Devices:\n");
attach_count = 0;
list_for_each_entry(attach_obj, &buf_obj->attachments, node) {
seq_printf(s, "\t%s\n", dev_name(attach_obj->dev));
attach_count++;
}
dma-buf: change DMA-buf locking convention v3 This patch is a stripped down version of the locking changes necessary to support dynamic DMA-buf handling. It adds a dynamic flag for both importers as well as exporters so that drivers can choose if they want the reservation object locked or unlocked during mapping of attachments. For compatibility between drivers we cache the DMA-buf mapping during attaching an importer as soon as exporter/importer disagree on the dynamic handling. Issues and solutions we considered: - We can't change all existing drivers, and existing improters have strong opinions about which locks they're holding while calling dma_buf_attachment_map/unmap. Exporters also have strong opinions about which locks they can acquire in their ->map/unmap callbacks, levaing no room for change. The solution to avoid this was to move the actual map/unmap out from this call, into the attach/detach callbacks, and cache the mapping. This works because drivers don't call attach/detach from deep within their code callchains (like deep in memory management code called from cs/execbuf ioctl), but directly from the fd2handle implementation. - The caching has some troubles on some soc drivers, which set other modes than DMA_BIDIRECTIONAL. We can't have 2 incompatible mappings, and we can't re-create the mapping at _map time due to the above locking fun. We very carefuly step around that by only caching at attach time if the dynamic mode between importer/expoert mismatches. - There's been quite some discussion on dma-buf mappings which need active cache management, which would all break down when caching, plus we don't have explicit flush operations on the attachment side. The solution to this was to shrug and keep the current discrepancy between what the dma-buf docs claim and what implementations do, with the hope that the begin/end_cpu_access hooks are good enough and that all necessary flushing to keep device mappings consistent will be done there. v2: cleanup set_name merge, improve kerneldoc v3: update commit message, kerneldoc and cleanup _debug_show() Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/336788/
2018-07-03 22:42:26 +08:00
dma_resv_unlock(buf_obj->resv);
seq_printf(s, "Total %d devices attached\n\n",
attach_count);
count++;
size += buf_obj->size;
}
seq_printf(s, "\nTotal %d objects, %zu bytes\n", count, size);
mutex_unlock(&debugfs_list_mutex);
return 0;
dma-buf: change DMA-buf locking convention v3 This patch is a stripped down version of the locking changes necessary to support dynamic DMA-buf handling. It adds a dynamic flag for both importers as well as exporters so that drivers can choose if they want the reservation object locked or unlocked during mapping of attachments. For compatibility between drivers we cache the DMA-buf mapping during attaching an importer as soon as exporter/importer disagree on the dynamic handling. Issues and solutions we considered: - We can't change all existing drivers, and existing improters have strong opinions about which locks they're holding while calling dma_buf_attachment_map/unmap. Exporters also have strong opinions about which locks they can acquire in their ->map/unmap callbacks, levaing no room for change. The solution to avoid this was to move the actual map/unmap out from this call, into the attach/detach callbacks, and cache the mapping. This works because drivers don't call attach/detach from deep within their code callchains (like deep in memory management code called from cs/execbuf ioctl), but directly from the fd2handle implementation. - The caching has some troubles on some soc drivers, which set other modes than DMA_BIDIRECTIONAL. We can't have 2 incompatible mappings, and we can't re-create the mapping at _map time due to the above locking fun. We very carefuly step around that by only caching at attach time if the dynamic mode between importer/expoert mismatches. - There's been quite some discussion on dma-buf mappings which need active cache management, which would all break down when caching, plus we don't have explicit flush operations on the attachment side. The solution to this was to shrug and keep the current discrepancy between what the dma-buf docs claim and what implementations do, with the hope that the begin/end_cpu_access hooks are good enough and that all necessary flushing to keep device mappings consistent will be done there. v2: cleanup set_name merge, improve kerneldoc v3: update commit message, kerneldoc and cleanup _debug_show() Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/336788/
2018-07-03 22:42:26 +08:00
error_unlock:
mutex_unlock(&debugfs_list_mutex);
dma-buf: change DMA-buf locking convention v3 This patch is a stripped down version of the locking changes necessary to support dynamic DMA-buf handling. It adds a dynamic flag for both importers as well as exporters so that drivers can choose if they want the reservation object locked or unlocked during mapping of attachments. For compatibility between drivers we cache the DMA-buf mapping during attaching an importer as soon as exporter/importer disagree on the dynamic handling. Issues and solutions we considered: - We can't change all existing drivers, and existing improters have strong opinions about which locks they're holding while calling dma_buf_attachment_map/unmap. Exporters also have strong opinions about which locks they can acquire in their ->map/unmap callbacks, levaing no room for change. The solution to avoid this was to move the actual map/unmap out from this call, into the attach/detach callbacks, and cache the mapping. This works because drivers don't call attach/detach from deep within their code callchains (like deep in memory management code called from cs/execbuf ioctl), but directly from the fd2handle implementation. - The caching has some troubles on some soc drivers, which set other modes than DMA_BIDIRECTIONAL. We can't have 2 incompatible mappings, and we can't re-create the mapping at _map time due to the above locking fun. We very carefuly step around that by only caching at attach time if the dynamic mode between importer/expoert mismatches. - There's been quite some discussion on dma-buf mappings which need active cache management, which would all break down when caching, plus we don't have explicit flush operations on the attachment side. The solution to this was to shrug and keep the current discrepancy between what the dma-buf docs claim and what implementations do, with the hope that the begin/end_cpu_access hooks are good enough and that all necessary flushing to keep device mappings consistent will be done there. v2: cleanup set_name merge, improve kerneldoc v3: update commit message, kerneldoc and cleanup _debug_show() Signed-off-by: Christian König <christian.koenig@amd.com> Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/336788/
2018-07-03 22:42:26 +08:00
return ret;
}
DEFINE_SHOW_ATTRIBUTE(dma_buf_debug);
static struct dentry *dma_buf_debugfs_dir;
static int dma_buf_init_debugfs(void)
{
struct dentry *d;
int err = 0;
d = debugfs_create_dir("dma_buf", NULL);
if (IS_ERR(d))
return PTR_ERR(d);
dma_buf_debugfs_dir = d;
d = debugfs_create_file("bufinfo", S_IRUGO, dma_buf_debugfs_dir,
NULL, &dma_buf_debug_fops);
if (IS_ERR(d)) {
pr_debug("dma_buf: debugfs: failed to create node bufinfo\n");
debugfs_remove_recursive(dma_buf_debugfs_dir);
dma_buf_debugfs_dir = NULL;
err = PTR_ERR(d);
}
return err;
}
static void dma_buf_uninit_debugfs(void)
{
debugfs_remove_recursive(dma_buf_debugfs_dir);
}
#else
static inline int dma_buf_init_debugfs(void)
{
return 0;
}
static inline void dma_buf_uninit_debugfs(void)
{
}
#endif
static int __init dma_buf_init(void)
{
dmabuf: Add the capability to expose DMA-BUF stats in sysfs Overview ======== The patch adds DMA-BUF statistics to /sys/kernel/dmabuf/buffers. It allows statistics to be enabled for each DMA-BUF in sysfs by enabling the config CONFIG_DMABUF_SYSFS_STATS. The following stats will be exposed by the interface: /sys/kernel/dmabuf/buffers/<inode_number>/exporter_name /sys/kernel/dmabuf/buffers/<inode_number>/size /sys/kernel/dmabuf/buffers/<inode_number>/attachments/<attach_uid>/device /sys/kernel/dmabuf/buffers/<inode_number>/attachments/<attach_uid>/map_counter The inode_number is unique for each DMA-BUF and was added earlier [1] in order to allow userspace to track DMA-BUF usage across different processes. Use Cases ========= The interface provides a way to gather DMA-BUF per-buffer statistics from production devices. These statistics will be used to derive DMA-BUF per-exporter stats and per-device usage stats for Android Bug reports. The corresponding userspace changes can be found at [2]. Telemetry tools will also capture this information(along with other memory metrics) periodically as well as on important events like a foreground app kill (which might have been triggered by Low Memory Killer). It will also contribute to provide a snapshot of the system memory usage on other events such as OOM kills and Application Not Responding events. Background ========== Currently, there are two existing interfaces that provide information about DMA-BUFs. 1) /sys/kernel/debug/dma_buf/bufinfo debugfs is however unsuitable to be mounted in production systems and cannot be considered as an alternative to the sysfs interface being proposed. 2) proc/<pid>/fdinfo/<fd> The proc/<pid>/fdinfo/<fd> files expose information about DMA-BUF fds. However, the existing procfs interfaces can only provide information about the buffers for which processes hold fds or have the buffers mmapped into their address space. Since the procfs interfaces alone cannot provide a full picture of all DMA-BUFs in the system, there is the need for an alternate interface to provide this information on production systems. The patch contains the following major improvements over v1: 1) Each attachment is represented by its own directory to allow creating a symlink to the importing device and to also provide room for future expansion. 2) The number of distinct mappings of each attachment is exposed in a separate file. 3) The per-buffer statistics are now in /sys/kernel/dmabuf/buffers inorder to make the interface expandable in future. All of the improvements above are based on suggestions/feedback from Daniel Vetter and Christian König. A shell script that can be run on a classic Linux environment to read out the DMA-BUF statistics can be found at [3](suggested by John Stultz). [1]: https://lore.kernel.org/patchwork/patch/1088791/ [2]: https://android-review.googlesource.com/q/topic:%22dmabuf-sysfs%22+(status:open%20OR%20status:merged) [3]: https://android-review.googlesource.com/c/platform/system/memory/libmeminfo/+/1549734 Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Hridya Valsaraju <hridya@google.com> Reported-by: kernel test robot <lkp@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/msgid/20210603214758.2955251-1-hridya@google.com
2021-06-04 05:47:51 +08:00
int ret;
ret = dma_buf_init_sysfs_statistics();
if (ret)
return ret;
dma_buf_mnt = kern_mount(&dma_buf_fs_type);
if (IS_ERR(dma_buf_mnt))
return PTR_ERR(dma_buf_mnt);
dma_buf_init_debugfs();
return 0;
}
subsys_initcall(dma_buf_init);
static void __exit dma_buf_deinit(void)
{
dma_buf_uninit_debugfs();
kern_unmount(dma_buf_mnt);
dmabuf: Add the capability to expose DMA-BUF stats in sysfs Overview ======== The patch adds DMA-BUF statistics to /sys/kernel/dmabuf/buffers. It allows statistics to be enabled for each DMA-BUF in sysfs by enabling the config CONFIG_DMABUF_SYSFS_STATS. The following stats will be exposed by the interface: /sys/kernel/dmabuf/buffers/<inode_number>/exporter_name /sys/kernel/dmabuf/buffers/<inode_number>/size /sys/kernel/dmabuf/buffers/<inode_number>/attachments/<attach_uid>/device /sys/kernel/dmabuf/buffers/<inode_number>/attachments/<attach_uid>/map_counter The inode_number is unique for each DMA-BUF and was added earlier [1] in order to allow userspace to track DMA-BUF usage across different processes. Use Cases ========= The interface provides a way to gather DMA-BUF per-buffer statistics from production devices. These statistics will be used to derive DMA-BUF per-exporter stats and per-device usage stats for Android Bug reports. The corresponding userspace changes can be found at [2]. Telemetry tools will also capture this information(along with other memory metrics) periodically as well as on important events like a foreground app kill (which might have been triggered by Low Memory Killer). It will also contribute to provide a snapshot of the system memory usage on other events such as OOM kills and Application Not Responding events. Background ========== Currently, there are two existing interfaces that provide information about DMA-BUFs. 1) /sys/kernel/debug/dma_buf/bufinfo debugfs is however unsuitable to be mounted in production systems and cannot be considered as an alternative to the sysfs interface being proposed. 2) proc/<pid>/fdinfo/<fd> The proc/<pid>/fdinfo/<fd> files expose information about DMA-BUF fds. However, the existing procfs interfaces can only provide information about the buffers for which processes hold fds or have the buffers mmapped into their address space. Since the procfs interfaces alone cannot provide a full picture of all DMA-BUFs in the system, there is the need for an alternate interface to provide this information on production systems. The patch contains the following major improvements over v1: 1) Each attachment is represented by its own directory to allow creating a symlink to the importing device and to also provide room for future expansion. 2) The number of distinct mappings of each attachment is exposed in a separate file. 3) The per-buffer statistics are now in /sys/kernel/dmabuf/buffers inorder to make the interface expandable in future. All of the improvements above are based on suggestions/feedback from Daniel Vetter and Christian König. A shell script that can be run on a classic Linux environment to read out the DMA-BUF statistics can be found at [3](suggested by John Stultz). [1]: https://lore.kernel.org/patchwork/patch/1088791/ [2]: https://android-review.googlesource.com/q/topic:%22dmabuf-sysfs%22+(status:open%20OR%20status:merged) [3]: https://android-review.googlesource.com/c/platform/system/memory/libmeminfo/+/1549734 Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org> Signed-off-by: Hridya Valsaraju <hridya@google.com> Reported-by: kernel test robot <lkp@intel.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch> Link: https://patchwork.freedesktop.org/patch/msgid/20210603214758.2955251-1-hridya@google.com
2021-06-04 05:47:51 +08:00
dma_buf_uninit_sysfs_statistics();
}
__exitcall(dma_buf_deinit);