linux/mm/hmm.c

600 lines
17 KiB
C
Raw Normal View History

treewide: Replace GPLv2 boilerplate/reference with SPDX - rule 157 Based on 3 normalized pattern(s): this program is free software you can redistribute it and or modify it under the terms of the gnu general public license as published by the free software foundation either version 2 of the license or at your option any later version this program is distributed in the hope that it will be useful but without any warranty without even the implied warranty of merchantability or fitness for a particular purpose see the gnu general public license for more details this program is free software you can redistribute it and or modify it under the terms of the gnu general public license as published by the free software foundation either version 2 of the license or at your option any later version [author] [kishon] [vijay] [abraham] [i] [kishon]@[ti] [com] this program is distributed in the hope that it will be useful but without any warranty without even the implied warranty of merchantability or fitness for a particular purpose see the gnu general public license for more details this program is free software you can redistribute it and or modify it under the terms of the gnu general public license as published by the free software foundation either version 2 of the license or at your option any later version [author] [graeme] [gregory] [gg]@[slimlogic] [co] [uk] [author] [kishon] [vijay] [abraham] [i] [kishon]@[ti] [com] [based] [on] [twl6030]_[usb] [c] [author] [hema] [hk] [hemahk]@[ti] [com] this program is distributed in the hope that it will be useful but without any warranty without even the implied warranty of merchantability or fitness for a particular purpose see the gnu general public license for more details extracted by the scancode license scanner the SPDX license identifier GPL-2.0-or-later has been chosen to replace the boilerplate/reference in 1105 file(s). Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Reviewed-by: Allison Randal <allison@lohutok.net> Reviewed-by: Richard Fontana <rfontana@redhat.com> Reviewed-by: Kate Stewart <kstewart@linuxfoundation.org> Cc: linux-spdx@vger.kernel.org Link: https://lkml.kernel.org/r/20190527070033.202006027@linutronix.de Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2019-05-27 14:55:06 +08:00
// SPDX-License-Identifier: GPL-2.0-or-later
/*
* Copyright 2013 Red Hat Inc.
*
* Authors: Jérôme Glisse <jglisse@redhat.com>
*/
/*
* Refer to include/linux/hmm.h for information about heterogeneous memory
* management or HMM for short.
*/
#include <linux/pagewalk.h>
#include <linux/hmm.h>
#include <linux/init.h>
#include <linux/rmap.h>
#include <linux/swap.h>
#include <linux/slab.h>
#include <linux/sched.h>
mm/hmm/devmem: device memory hotplug using ZONE_DEVICE This introduce a simple struct and associated helpers for device driver to use when hotpluging un-addressable device memory as ZONE_DEVICE. It will find a unuse physical address range and trigger memory hotplug for it which allocates and initialize struct page for the device memory. Device driver should use this helper during device initialization to hotplug the device memory. It should only need to remove the memory once the device is going offline (shutdown or hotremove). There should not be any userspace API to hotplug memory expect maybe for host device driver to allow to add more memory to a guest device driver. Device's memory is manage by the device driver and HMM only provides helpers to that effect. Link: http://lkml.kernel.org/r/20170817000548.32038-12-jglisse@redhat.com Signed-off-by: Jérôme Glisse <jglisse@redhat.com> Signed-off-by: Evgeny Baskakov <ebaskakov@nvidia.com> Signed-off-by: John Hubbard <jhubbard@nvidia.com> Signed-off-by: Mark Hairgrove <mhairgrove@nvidia.com> Signed-off-by: Sherry Cheung <SCheung@nvidia.com> Signed-off-by: Subhash Gutti <sgutti@nvidia.com> Signed-off-by: Balbir Singh <bsingharora@gmail.com> Cc: Aneesh Kumar <aneesh.kumar@linux.vnet.ibm.com> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Dan Williams <dan.j.williams@intel.com> Cc: David Nellans <dnellans@nvidia.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Michal Hocko <mhocko@kernel.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Ross Zwisler <ross.zwisler@linux.intel.com> Cc: Vladimir Davydov <vdavydov.dev@gmail.com> Cc: Bob Liu <liubo95@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-09-09 07:11:58 +08:00
#include <linux/mmzone.h>
#include <linux/pagemap.h>
#include <linux/swapops.h>
#include <linux/hugetlb.h>
mm/hmm/devmem: device memory hotplug using ZONE_DEVICE This introduce a simple struct and associated helpers for device driver to use when hotpluging un-addressable device memory as ZONE_DEVICE. It will find a unuse physical address range and trigger memory hotplug for it which allocates and initialize struct page for the device memory. Device driver should use this helper during device initialization to hotplug the device memory. It should only need to remove the memory once the device is going offline (shutdown or hotremove). There should not be any userspace API to hotplug memory expect maybe for host device driver to allow to add more memory to a guest device driver. Device's memory is manage by the device driver and HMM only provides helpers to that effect. Link: http://lkml.kernel.org/r/20170817000548.32038-12-jglisse@redhat.com Signed-off-by: Jérôme Glisse <jglisse@redhat.com> Signed-off-by: Evgeny Baskakov <ebaskakov@nvidia.com> Signed-off-by: John Hubbard <jhubbard@nvidia.com> Signed-off-by: Mark Hairgrove <mhairgrove@nvidia.com> Signed-off-by: Sherry Cheung <SCheung@nvidia.com> Signed-off-by: Subhash Gutti <sgutti@nvidia.com> Signed-off-by: Balbir Singh <bsingharora@gmail.com> Cc: Aneesh Kumar <aneesh.kumar@linux.vnet.ibm.com> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Dan Williams <dan.j.williams@intel.com> Cc: David Nellans <dnellans@nvidia.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Michal Hocko <mhocko@kernel.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Ross Zwisler <ross.zwisler@linux.intel.com> Cc: Vladimir Davydov <vdavydov.dev@gmail.com> Cc: Bob Liu <liubo95@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-09-09 07:11:58 +08:00
#include <linux/memremap.h>
#include <linux/sched/mm.h>
#include <linux/jump_label.h>
#include <linux/dma-mapping.h>
mm/hmm/mirror: mirror process address space on device with HMM helpers This is a heterogeneous memory management (HMM) process address space mirroring. In a nutshell this provide an API to mirror process address space on a device. This boils down to keeping CPU and device page table synchronize (we assume that both device and CPU are cache coherent like PCIe device can be). This patch provide a simple API for device driver to achieve address space mirroring thus avoiding each device driver to grow its own CPU page table walker and its own CPU page table synchronization mechanism. This is useful for NVidia GPU >= Pascal, Mellanox IB >= mlx5 and more hardware in the future. [jglisse@redhat.com: fix hmm for "mmu_notifier kill invalidate_page callback"] Link: http://lkml.kernel.org/r/20170830231955.GD9445@redhat.com Link: http://lkml.kernel.org/r/20170817000548.32038-4-jglisse@redhat.com Signed-off-by: Jérôme Glisse <jglisse@redhat.com> Signed-off-by: Evgeny Baskakov <ebaskakov@nvidia.com> Signed-off-by: John Hubbard <jhubbard@nvidia.com> Signed-off-by: Mark Hairgrove <mhairgrove@nvidia.com> Signed-off-by: Sherry Cheung <SCheung@nvidia.com> Signed-off-by: Subhash Gutti <sgutti@nvidia.com> Cc: Aneesh Kumar <aneesh.kumar@linux.vnet.ibm.com> Cc: Balbir Singh <bsingharora@gmail.com> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Dan Williams <dan.j.williams@intel.com> Cc: David Nellans <dnellans@nvidia.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Michal Hocko <mhocko@kernel.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Ross Zwisler <ross.zwisler@linux.intel.com> Cc: Vladimir Davydov <vdavydov.dev@gmail.com> Cc: Bob Liu <liubo95@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-09-09 07:11:27 +08:00
#include <linux/mmu_notifier.h>
mm/hmm/devmem: device memory hotplug using ZONE_DEVICE This introduce a simple struct and associated helpers for device driver to use when hotpluging un-addressable device memory as ZONE_DEVICE. It will find a unuse physical address range and trigger memory hotplug for it which allocates and initialize struct page for the device memory. Device driver should use this helper during device initialization to hotplug the device memory. It should only need to remove the memory once the device is going offline (shutdown or hotremove). There should not be any userspace API to hotplug memory expect maybe for host device driver to allow to add more memory to a guest device driver. Device's memory is manage by the device driver and HMM only provides helpers to that effect. Link: http://lkml.kernel.org/r/20170817000548.32038-12-jglisse@redhat.com Signed-off-by: Jérôme Glisse <jglisse@redhat.com> Signed-off-by: Evgeny Baskakov <ebaskakov@nvidia.com> Signed-off-by: John Hubbard <jhubbard@nvidia.com> Signed-off-by: Mark Hairgrove <mhairgrove@nvidia.com> Signed-off-by: Sherry Cheung <SCheung@nvidia.com> Signed-off-by: Subhash Gutti <sgutti@nvidia.com> Signed-off-by: Balbir Singh <bsingharora@gmail.com> Cc: Aneesh Kumar <aneesh.kumar@linux.vnet.ibm.com> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Dan Williams <dan.j.williams@intel.com> Cc: David Nellans <dnellans@nvidia.com> Cc: Johannes Weiner <hannes@cmpxchg.org> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Michal Hocko <mhocko@kernel.org> Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com> Cc: Ross Zwisler <ross.zwisler@linux.intel.com> Cc: Vladimir Davydov <vdavydov.dev@gmail.com> Cc: Bob Liu <liubo95@huawei.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2017-09-09 07:11:58 +08:00
#include <linux/memory_hotplug.h>
mm: device exclusive memory access Some devices require exclusive write access to shared virtual memory (SVM) ranges to perform atomic operations on that memory. This requires CPU page tables to be updated to deny access whilst atomic operations are occurring. In order to do this introduce a new swap entry type (SWP_DEVICE_EXCLUSIVE). When a SVM range needs to be marked for exclusive access by a device all page table mappings for the particular range are replaced with device exclusive swap entries. This causes any CPU access to the page to result in a fault. Faults are resovled by replacing the faulting entry with the original mapping. This results in MMU notifiers being called which a driver uses to update access permissions such as revoking atomic access. After notifiers have been called the device will no longer have exclusive access to the region. Walking of the page tables to find the target pages is handled by get_user_pages() rather than a direct page table walk. A direct page table walk similar to what migrate_vma_collect()/unmap() does could also have been utilised. However this resulted in more code similar in functionality to what get_user_pages() provides as page faulting is required to make the PTEs present and to break COW. [dan.carpenter@oracle.com: fix signedness bug in make_device_exclusive_range()] Link: https://lkml.kernel.org/r/YNIz5NVnZ5GiZ3u1@mwanda Link: https://lkml.kernel.org/r/20210616105937.23201-8-apopple@nvidia.com Signed-off-by: Alistair Popple <apopple@nvidia.com> Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Cc: Ben Skeggs <bskeggs@redhat.com> Cc: Hugh Dickins <hughd@google.com> Cc: Jason Gunthorpe <jgg@nvidia.com> Cc: John Hubbard <jhubbard@nvidia.com> Cc: "Matthew Wilcox (Oracle)" <willy@infradead.org> Cc: Peter Xu <peterx@redhat.com> Cc: Ralph Campbell <rcampbell@nvidia.com> Cc: Shakeel Butt <shakeelb@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-07-01 09:54:25 +08:00
#include "internal.h"
struct hmm_vma_walk {
struct hmm_range *range;
unsigned long last;
};
enum {
HMM_NEED_FAULT = 1 << 0,
HMM_NEED_WRITE_FAULT = 1 << 1,
HMM_NEED_ALL_BITS = HMM_NEED_FAULT | HMM_NEED_WRITE_FAULT,
};
static int hmm_pfns_fill(unsigned long addr, unsigned long end,
mm/hmm: remove the customizable pfn format from hmm_range_fault Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Tested-by: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-02 02:20:48 +08:00
struct hmm_range *range, unsigned long cpu_flags)
{
mm/hmm: remove the customizable pfn format from hmm_range_fault Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Tested-by: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-02 02:20:48 +08:00
unsigned long i = (addr - range->start) >> PAGE_SHIFT;
for (; addr < end; addr += PAGE_SIZE, i++)
mm/hmm: remove the customizable pfn format from hmm_range_fault Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Tested-by: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-02 02:20:48 +08:00
range->hmm_pfns[i] = cpu_flags;
return 0;
}
/*
* hmm_vma_fault() - fault in a range lacking valid pmd or pte(s)
* @addr: range virtual start address (inclusive)
* @end: range virtual end address (exclusive)
* @required_fault: HMM_NEED_* flags
* @walk: mm_walk structure
* Return: -EBUSY after page fault, or page fault error
*
* This function will be called whenever pmd_none() or pte_none() returns true,
* or whenever there is no page directory covering the virtual address range.
*/
static int hmm_vma_fault(unsigned long addr, unsigned long end,
unsigned int required_fault, struct mm_walk *walk)
{
struct hmm_vma_walk *hmm_vma_walk = walk->private;
struct vm_area_struct *vma = walk->vma;
unsigned int fault_flags = FAULT_FLAG_REMOTE;
WARN_ON_ONCE(!required_fault);
hmm_vma_walk->last = addr;
if (required_fault & HMM_NEED_WRITE_FAULT) {
if (!(vma->vm_flags & VM_WRITE))
return -EPERM;
fault_flags |= FAULT_FLAG_WRITE;
}
for (; addr < end; addr += PAGE_SIZE)
mm: do page fault accounting in handle_mm_fault Patch series "mm: Page fault accounting cleanups", v5. This is v5 of the pf accounting cleanup series. It originates from Gerald Schaefer's report on an issue a week ago regarding to incorrect page fault accountings for retried page fault after commit 4064b9827063 ("mm: allow VM_FAULT_RETRY for multiple times"): https://lore.kernel.org/lkml/20200610174811.44b94525@thinkpad/ What this series did: - Correct page fault accounting: we do accounting for a page fault (no matter whether it's from #PF handling, or gup, or anything else) only with the one that completed the fault. For example, page fault retries should not be counted in page fault counters. Same to the perf events. - Unify definition of PERF_COUNT_SW_PAGE_FAULTS: currently this perf event is used in an adhoc way across different archs. Case (1): for many archs it's done at the entry of a page fault handler, so that it will also cover e.g. errornous faults. Case (2): for some other archs, it is only accounted when the page fault is resolved successfully. Case (3): there're still quite some archs that have not enabled this perf event. Since this series will touch merely all the archs, we unify this perf event to always follow case (1), which is the one that makes most sense. And since we moved the accounting into handle_mm_fault, the other two MAJ/MIN perf events are well taken care of naturally. - Unify definition of "major faults": the definition of "major fault" is slightly changed when used in accounting (not VM_FAULT_MAJOR). More information in patch 1. - Always account the page fault onto the one that triggered the page fault. This does not matter much for #PF handlings, but mostly for gup. More information on this in patch 25. Patchset layout: Patch 1: Introduced the accounting in handle_mm_fault(), not enabled. Patch 2-23: Enable the new accounting for arch #PF handlers one by one. Patch 24: Enable the new accounting for the rest outliers (gup, iommu, etc.) Patch 25: Cleanup GUP task_struct pointer since it's not needed any more This patch (of 25): This is a preparation patch to move page fault accountings into the general code in handle_mm_fault(). This includes both the per task flt_maj/flt_min counters, and the major/minor page fault perf events. To do this, the pt_regs pointer is passed into handle_mm_fault(). PERF_COUNT_SW_PAGE_FAULTS should still be kept in per-arch page fault handlers. So far, all the pt_regs pointer that passed into handle_mm_fault() is NULL, which means this patch should have no intented functional change. Suggested-by: Linus Torvalds <torvalds@linux-foundation.org> Signed-off-by: Peter Xu <peterx@redhat.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Cc: Albert Ou <aou@eecs.berkeley.edu> Cc: Alexander Gordeev <agordeev@linux.ibm.com> Cc: Andy Lutomirski <luto@kernel.org> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Borislav Petkov <bp@alien8.de> Cc: Brian Cain <bcain@codeaurora.org> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Christian Borntraeger <borntraeger@de.ibm.com> Cc: Chris Zankel <chris@zankel.net> Cc: Dave Hansen <dave.hansen@linux.intel.com> Cc: David S. Miller <davem@davemloft.net> Cc: Geert Uytterhoeven <geert@linux-m68k.org> Cc: Gerald Schaefer <gerald.schaefer@de.ibm.com> Cc: Greentime Hu <green.hu@gmail.com> Cc: Guo Ren <guoren@kernel.org> Cc: Heiko Carstens <heiko.carstens@de.ibm.com> Cc: Helge Deller <deller@gmx.de> Cc: H. Peter Anvin <hpa@zytor.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru> Cc: James E.J. Bottomley <James.Bottomley@HansenPartnership.com> Cc: John Hubbard <jhubbard@nvidia.com> Cc: Jonas Bonn <jonas@southpole.se> Cc: Ley Foon Tan <ley.foon.tan@intel.com> Cc: "Luck, Tony" <tony.luck@intel.com> Cc: Matt Turner <mattst88@gmail.com> Cc: Max Filippov <jcmvbkbc@gmail.com> Cc: Michael Ellerman <mpe@ellerman.id.au> Cc: Michal Simek <monstr@monstr.eu> Cc: Nick Hu <nickhu@andestech.com> Cc: Palmer Dabbelt <palmer@dabbelt.com> Cc: Paul Mackerras <paulus@samba.org> Cc: Paul Walmsley <paul.walmsley@sifive.com> Cc: Pekka Enberg <penberg@kernel.org> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Richard Henderson <rth@twiddle.net> Cc: Rich Felker <dalias@libc.org> Cc: Russell King <linux@armlinux.org.uk> Cc: Stafford Horne <shorne@gmail.com> Cc: Stefan Kristiansson <stefan.kristiansson@saunalahti.fi> Cc: Thomas Bogendoerfer <tsbogend@alpha.franken.de> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Vasily Gorbik <gor@linux.ibm.com> Cc: Vincent Chen <deanbo422@gmail.com> Cc: Vineet Gupta <vgupta@synopsys.com> Cc: Will Deacon <will@kernel.org> Cc: Yoshinori Sato <ysato@users.sourceforge.jp> Link: http://lkml.kernel.org/r/20200707225021.200906-1-peterx@redhat.com Link: http://lkml.kernel.org/r/20200707225021.200906-2-peterx@redhat.com Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-08-12 09:37:44 +08:00
if (handle_mm_fault(vma, addr, fault_flags, NULL) &
VM_FAULT_ERROR)
return -EFAULT;
return -EBUSY;
}
static unsigned int hmm_pte_need_fault(const struct hmm_vma_walk *hmm_vma_walk,
mm/hmm: remove the customizable pfn format from hmm_range_fault Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Tested-by: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-02 02:20:48 +08:00
unsigned long pfn_req_flags,
unsigned long cpu_flags)
{
struct hmm_range *range = hmm_vma_walk->range;
/*
* So we not only consider the individual per page request we also
* consider the default flags requested for the range. The API can
* be used 2 ways. The first one where the HMM user coalesces
* multiple page faults into one request and sets flags per pfn for
* those faults. The second one where the HMM user wants to pre-
* fault a range with specific flags. For the latter one it is a
* waste to have the user pre-fill the pfn arrays with a default
* flags value.
*/
mm/hmm: remove the customizable pfn format from hmm_range_fault Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Tested-by: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-02 02:20:48 +08:00
pfn_req_flags &= range->pfn_flags_mask;
pfn_req_flags |= range->default_flags;
/* We aren't ask to do anything ... */
mm/hmm: remove the customizable pfn format from hmm_range_fault Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Tested-by: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-02 02:20:48 +08:00
if (!(pfn_req_flags & HMM_PFN_REQ_FAULT))
return 0;
/* Need to write fault ? */
mm/hmm: remove the customizable pfn format from hmm_range_fault Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Tested-by: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-02 02:20:48 +08:00
if ((pfn_req_flags & HMM_PFN_REQ_WRITE) &&
!(cpu_flags & HMM_PFN_WRITE))
return HMM_NEED_FAULT | HMM_NEED_WRITE_FAULT;
/* If CPU page table is not valid then we need to fault */
mm/hmm: remove the customizable pfn format from hmm_range_fault Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Tested-by: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-02 02:20:48 +08:00
if (!(cpu_flags & HMM_PFN_VALID))
return HMM_NEED_FAULT;
return 0;
}
static unsigned int
hmm_range_need_fault(const struct hmm_vma_walk *hmm_vma_walk,
mm/hmm: remove the customizable pfn format from hmm_range_fault Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Tested-by: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-02 02:20:48 +08:00
const unsigned long hmm_pfns[], unsigned long npages,
unsigned long cpu_flags)
{
struct hmm_range *range = hmm_vma_walk->range;
unsigned int required_fault = 0;
unsigned long i;
/*
* If the default flags do not request to fault pages, and the mask does
* not allow for individual pages to be faulted, then
* hmm_pte_need_fault() will always return 0.
*/
if (!((range->default_flags | range->pfn_flags_mask) &
mm/hmm: remove the customizable pfn format from hmm_range_fault Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Tested-by: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-02 02:20:48 +08:00
HMM_PFN_REQ_FAULT))
return 0;
for (i = 0; i < npages; ++i) {
mm/hmm: remove the customizable pfn format from hmm_range_fault Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Tested-by: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-02 02:20:48 +08:00
required_fault |= hmm_pte_need_fault(hmm_vma_walk, hmm_pfns[i],
cpu_flags);
if (required_fault == HMM_NEED_ALL_BITS)
return required_fault;
}
return required_fault;
}
static int hmm_vma_walk_hole(unsigned long addr, unsigned long end,
mm: pagewalk: add 'depth' parameter to pte_hole The pte_hole() callback is called at multiple levels of the page tables. Code dumping the kernel page tables needs to know what at what depth the missing entry is. Add this is an extra parameter to pte_hole(). When the depth isn't know (e.g. processing a vma) then -1 is passed. The depth that is reported is the actual level where the entry is missing (ignoring any folding that is in place), i.e. any levels where PTRS_PER_P?D is set to 1 are ignored. Note that depth starts at 0 for a PGD so that PUD/PMD/PTE retain their natural numbers as levels 2/3/4. Link: http://lkml.kernel.org/r/20191218162402.45610-16-steven.price@arm.com Signed-off-by: Steven Price <steven.price@arm.com> Tested-by: Zong Li <zong.li@sifive.com> Cc: Albert Ou <aou@eecs.berkeley.edu> Cc: Alexandre Ghiti <alex@ghiti.fr> Cc: Andy Lutomirski <luto@kernel.org> Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Borislav Petkov <bp@alien8.de> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Christian Borntraeger <borntraeger@de.ibm.com> Cc: Dave Hansen <dave.hansen@linux.intel.com> Cc: David S. Miller <davem@davemloft.net> Cc: Heiko Carstens <heiko.carstens@de.ibm.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: James Hogan <jhogan@kernel.org> Cc: James Morse <james.morse@arm.com> Cc: Jerome Glisse <jglisse@redhat.com> Cc: "Liang, Kan" <kan.liang@linux.intel.com> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Michael Ellerman <mpe@ellerman.id.au> Cc: Paul Burton <paul.burton@mips.com> Cc: Paul Mackerras <paulus@samba.org> Cc: Paul Walmsley <paul.walmsley@sifive.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Russell King <linux@armlinux.org.uk> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Vasily Gorbik <gor@linux.ibm.com> Cc: Vineet Gupta <vgupta@synopsys.com> Cc: Will Deacon <will@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-02-04 09:36:03 +08:00
__always_unused int depth, struct mm_walk *walk)
{
struct hmm_vma_walk *hmm_vma_walk = walk->private;
struct hmm_range *range = hmm_vma_walk->range;
unsigned int required_fault;
unsigned long i, npages;
mm/hmm: remove the customizable pfn format from hmm_range_fault Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Tested-by: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-02 02:20:48 +08:00
unsigned long *hmm_pfns;
i = (addr - range->start) >> PAGE_SHIFT;
npages = (end - addr) >> PAGE_SHIFT;
mm/hmm: remove the customizable pfn format from hmm_range_fault Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Tested-by: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-02 02:20:48 +08:00
hmm_pfns = &range->hmm_pfns[i];
required_fault =
hmm_range_need_fault(hmm_vma_walk, hmm_pfns, npages, 0);
if (!walk->vma) {
if (required_fault)
return -EFAULT;
return hmm_pfns_fill(addr, end, range, HMM_PFN_ERROR);
}
if (required_fault)
return hmm_vma_fault(addr, end, required_fault, walk);
mm/hmm: remove the customizable pfn format from hmm_range_fault Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Tested-by: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-02 02:20:48 +08:00
return hmm_pfns_fill(addr, end, range, 0);
}
static inline unsigned long hmm_pfn_flags_order(unsigned long order)
{
return order << HMM_PFN_ORDER_SHIFT;
}
mm/hmm: remove the customizable pfn format from hmm_range_fault Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Tested-by: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-02 02:20:48 +08:00
static inline unsigned long pmd_to_hmm_pfn_flags(struct hmm_range *range,
pmd_t pmd)
{
if (pmd_protnone(pmd))
return 0;
return (pmd_write(pmd) ? (HMM_PFN_VALID | HMM_PFN_WRITE) :
HMM_PFN_VALID) |
hmm_pfn_flags_order(PMD_SHIFT - PAGE_SHIFT);
}
#ifdef CONFIG_TRANSPARENT_HUGEPAGE
static int hmm_vma_handle_pmd(struct mm_walk *walk, unsigned long addr,
mm/hmm: remove the customizable pfn format from hmm_range_fault Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Tested-by: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-02 02:20:48 +08:00
unsigned long end, unsigned long hmm_pfns[],
pmd_t pmd)
{
struct hmm_vma_walk *hmm_vma_walk = walk->private;
struct hmm_range *range = hmm_vma_walk->range;
unsigned long pfn, npages, i;
unsigned int required_fault;
mm/hmm: remove the customizable pfn format from hmm_range_fault Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Tested-by: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-02 02:20:48 +08:00
unsigned long cpu_flags;
npages = (end - addr) >> PAGE_SHIFT;
cpu_flags = pmd_to_hmm_pfn_flags(range, pmd);
required_fault =
mm/hmm: remove the customizable pfn format from hmm_range_fault Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Tested-by: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-02 02:20:48 +08:00
hmm_range_need_fault(hmm_vma_walk, hmm_pfns, npages, cpu_flags);
if (required_fault)
return hmm_vma_fault(addr, end, required_fault, walk);
pfn = pmd_pfn(pmd) + ((addr & ~PMD_MASK) >> PAGE_SHIFT);
for (i = 0; addr < end; addr += PAGE_SIZE, i++, pfn++)
mm/hmm: remove the customizable pfn format from hmm_range_fault Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Tested-by: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-02 02:20:48 +08:00
hmm_pfns[i] = pfn | cpu_flags;
return 0;
}
#else /* CONFIG_TRANSPARENT_HUGEPAGE */
/* stub to allow the code below to compile */
int hmm_vma_handle_pmd(struct mm_walk *walk, unsigned long addr,
mm/hmm: remove the customizable pfn format from hmm_range_fault Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Tested-by: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-02 02:20:48 +08:00
unsigned long end, unsigned long hmm_pfns[], pmd_t pmd);
#endif /* CONFIG_TRANSPARENT_HUGEPAGE */
static inline bool hmm_is_device_private_entry(struct hmm_range *range,
swp_entry_t entry)
{
return is_device_private_entry(entry) &&
mm: remove special swap entry functions Patch series "Add support for SVM atomics in Nouveau", v11. Introduction ============ Some devices have features such as atomic PTE bits that can be used to implement atomic access to system memory. To support atomic operations to a shared virtual memory page such a device needs access to that page which is exclusive of the CPU. This series introduces a mechanism to temporarily unmap pages granting exclusive access to a device. These changes are required to support OpenCL atomic operations in Nouveau to shared virtual memory (SVM) regions allocated with the CL_MEM_SVM_ATOMICS clSVMAlloc flag. A more complete description of the OpenCL SVM feature is available at https://www.khronos.org/registry/OpenCL/specs/3.0-unified/html/ OpenCL_API.html#_shared_virtual_memory . Implementation ============== Exclusive device access is implemented by adding a new swap entry type (SWAP_DEVICE_EXCLUSIVE) which is similar to a migration entry. The main difference is that on fault the original entry is immediately restored by the fault handler instead of waiting. Restoring the entry triggers calls to MMU notifers which allows a device driver to revoke the atomic access permission from the GPU prior to the CPU finalising the entry. Patches ======= Patches 1 & 2 refactor existing migration and device private entry functions. Patches 3 & 4 rework try_to_unmap_one() by splitting out unrelated functionality into separate functions - try_to_migrate_one() and try_to_munlock_one(). Patch 5 renames some existing code but does not introduce functionality. Patch 6 is a small clean-up to swap entry handling in copy_pte_range(). Patch 7 contains the bulk of the implementation for device exclusive memory. Patch 8 contains some additions to the HMM selftests to ensure everything works as expected. Patch 9 is a cleanup for the Nouveau SVM implementation. Patch 10 contains the implementation of atomic access for the Nouveau driver. Testing ======= This has been tested with upstream Mesa 21.1.0 and a simple OpenCL program which checks that GPU atomic accesses to system memory are atomic. Without this series the test fails as there is no way of write-protecting the page mapping which results in the device clobbering CPU writes. For reference the test is available at https://ozlabs.org/~apopple/opencl_svm_atomics/ Further testing has been performed by adding support for testing exclusive access to the hmm-tests kselftests. This patch (of 10): Remove multiple similar inline functions for dealing with different types of special swap entries. Both migration and device private swap entries use the swap offset to store a pfn. Instead of multiple inline functions to obtain a struct page for each swap entry type use a common function pfn_swap_entry_to_page(). Also open-code the various entry_to_pfn() functions as this results is shorter code that is easier to understand. Link: https://lkml.kernel.org/r/20210616105937.23201-1-apopple@nvidia.com Link: https://lkml.kernel.org/r/20210616105937.23201-2-apopple@nvidia.com Signed-off-by: Alistair Popple <apopple@nvidia.com> Reviewed-by: Ralph Campbell <rcampbell@nvidia.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Cc: "Matthew Wilcox (Oracle)" <willy@infradead.org> Cc: Hugh Dickins <hughd@google.com> Cc: Peter Xu <peterx@redhat.com> Cc: Shakeel Butt <shakeelb@google.com> Cc: Ben Skeggs <bskeggs@redhat.com> Cc: Jason Gunthorpe <jgg@nvidia.com> Cc: John Hubbard <jhubbard@nvidia.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-07-01 09:54:06 +08:00
pfn_swap_entry_to_page(entry)->pgmap->owner ==
range->dev_private_owner;
}
mm/hmm: remove the customizable pfn format from hmm_range_fault Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Tested-by: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-02 02:20:48 +08:00
static inline unsigned long pte_to_hmm_pfn_flags(struct hmm_range *range,
pte_t pte)
{
if (pte_none(pte) || !pte_present(pte) || pte_protnone(pte))
return 0;
mm/hmm: remove the customizable pfn format from hmm_range_fault Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Tested-by: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-02 02:20:48 +08:00
return pte_write(pte) ? (HMM_PFN_VALID | HMM_PFN_WRITE) : HMM_PFN_VALID;
}
static int hmm_vma_handle_pte(struct mm_walk *walk, unsigned long addr,
unsigned long end, pmd_t *pmdp, pte_t *ptep,
mm/hmm: remove the customizable pfn format from hmm_range_fault Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Tested-by: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-02 02:20:48 +08:00
unsigned long *hmm_pfn)
{
struct hmm_vma_walk *hmm_vma_walk = walk->private;
struct hmm_range *range = hmm_vma_walk->range;
unsigned int required_fault;
mm/hmm: remove the customizable pfn format from hmm_range_fault Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Tested-by: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-02 02:20:48 +08:00
unsigned long cpu_flags;
pte_t pte = *ptep;
mm/hmm: remove the customizable pfn format from hmm_range_fault Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Tested-by: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-02 02:20:48 +08:00
uint64_t pfn_req_flags = *hmm_pfn;
if (pte_none(pte)) {
mm/hmm: remove the customizable pfn format from hmm_range_fault Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Tested-by: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-02 02:20:48 +08:00
required_fault =
hmm_pte_need_fault(hmm_vma_walk, pfn_req_flags, 0);
if (required_fault)
goto fault;
mm/hmm: remove the customizable pfn format from hmm_range_fault Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Tested-by: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-02 02:20:48 +08:00
*hmm_pfn = 0;
return 0;
}
if (!pte_present(pte)) {
swp_entry_t entry = pte_to_swp_entry(pte);
/*
* Never fault in device private pages, but just report
* the PFN even if not present.
*/
if (hmm_is_device_private_entry(range, entry)) {
mm/hmm: remove the customizable pfn format from hmm_range_fault Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Tested-by: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-02 02:20:48 +08:00
cpu_flags = HMM_PFN_VALID;
if (is_writable_device_private_entry(entry))
mm/hmm: remove the customizable pfn format from hmm_range_fault Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Tested-by: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-02 02:20:48 +08:00
cpu_flags |= HMM_PFN_WRITE;
mm: remove special swap entry functions Patch series "Add support for SVM atomics in Nouveau", v11. Introduction ============ Some devices have features such as atomic PTE bits that can be used to implement atomic access to system memory. To support atomic operations to a shared virtual memory page such a device needs access to that page which is exclusive of the CPU. This series introduces a mechanism to temporarily unmap pages granting exclusive access to a device. These changes are required to support OpenCL atomic operations in Nouveau to shared virtual memory (SVM) regions allocated with the CL_MEM_SVM_ATOMICS clSVMAlloc flag. A more complete description of the OpenCL SVM feature is available at https://www.khronos.org/registry/OpenCL/specs/3.0-unified/html/ OpenCL_API.html#_shared_virtual_memory . Implementation ============== Exclusive device access is implemented by adding a new swap entry type (SWAP_DEVICE_EXCLUSIVE) which is similar to a migration entry. The main difference is that on fault the original entry is immediately restored by the fault handler instead of waiting. Restoring the entry triggers calls to MMU notifers which allows a device driver to revoke the atomic access permission from the GPU prior to the CPU finalising the entry. Patches ======= Patches 1 & 2 refactor existing migration and device private entry functions. Patches 3 & 4 rework try_to_unmap_one() by splitting out unrelated functionality into separate functions - try_to_migrate_one() and try_to_munlock_one(). Patch 5 renames some existing code but does not introduce functionality. Patch 6 is a small clean-up to swap entry handling in copy_pte_range(). Patch 7 contains the bulk of the implementation for device exclusive memory. Patch 8 contains some additions to the HMM selftests to ensure everything works as expected. Patch 9 is a cleanup for the Nouveau SVM implementation. Patch 10 contains the implementation of atomic access for the Nouveau driver. Testing ======= This has been tested with upstream Mesa 21.1.0 and a simple OpenCL program which checks that GPU atomic accesses to system memory are atomic. Without this series the test fails as there is no way of write-protecting the page mapping which results in the device clobbering CPU writes. For reference the test is available at https://ozlabs.org/~apopple/opencl_svm_atomics/ Further testing has been performed by adding support for testing exclusive access to the hmm-tests kselftests. This patch (of 10): Remove multiple similar inline functions for dealing with different types of special swap entries. Both migration and device private swap entries use the swap offset to store a pfn. Instead of multiple inline functions to obtain a struct page for each swap entry type use a common function pfn_swap_entry_to_page(). Also open-code the various entry_to_pfn() functions as this results is shorter code that is easier to understand. Link: https://lkml.kernel.org/r/20210616105937.23201-1-apopple@nvidia.com Link: https://lkml.kernel.org/r/20210616105937.23201-2-apopple@nvidia.com Signed-off-by: Alistair Popple <apopple@nvidia.com> Reviewed-by: Ralph Campbell <rcampbell@nvidia.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Cc: "Matthew Wilcox (Oracle)" <willy@infradead.org> Cc: Hugh Dickins <hughd@google.com> Cc: Peter Xu <peterx@redhat.com> Cc: Shakeel Butt <shakeelb@google.com> Cc: Ben Skeggs <bskeggs@redhat.com> Cc: Jason Gunthorpe <jgg@nvidia.com> Cc: John Hubbard <jhubbard@nvidia.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-07-01 09:54:06 +08:00
*hmm_pfn = swp_offset(entry) | cpu_flags;
return 0;
}
mm/hmm: remove the customizable pfn format from hmm_range_fault Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Tested-by: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-02 02:20:48 +08:00
required_fault =
hmm_pte_need_fault(hmm_vma_walk, pfn_req_flags, 0);
if (!required_fault) {
mm/hmm: remove the customizable pfn format from hmm_range_fault Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Tested-by: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-02 02:20:48 +08:00
*hmm_pfn = 0;
return 0;
}
if (!non_swap_entry(entry))
goto fault;
mm: device exclusive memory access Some devices require exclusive write access to shared virtual memory (SVM) ranges to perform atomic operations on that memory. This requires CPU page tables to be updated to deny access whilst atomic operations are occurring. In order to do this introduce a new swap entry type (SWP_DEVICE_EXCLUSIVE). When a SVM range needs to be marked for exclusive access by a device all page table mappings for the particular range are replaced with device exclusive swap entries. This causes any CPU access to the page to result in a fault. Faults are resovled by replacing the faulting entry with the original mapping. This results in MMU notifiers being called which a driver uses to update access permissions such as revoking atomic access. After notifiers have been called the device will no longer have exclusive access to the region. Walking of the page tables to find the target pages is handled by get_user_pages() rather than a direct page table walk. A direct page table walk similar to what migrate_vma_collect()/unmap() does could also have been utilised. However this resulted in more code similar in functionality to what get_user_pages() provides as page faulting is required to make the PTEs present and to break COW. [dan.carpenter@oracle.com: fix signedness bug in make_device_exclusive_range()] Link: https://lkml.kernel.org/r/YNIz5NVnZ5GiZ3u1@mwanda Link: https://lkml.kernel.org/r/20210616105937.23201-8-apopple@nvidia.com Signed-off-by: Alistair Popple <apopple@nvidia.com> Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com> Reviewed-by: Christoph Hellwig <hch@lst.de> Cc: Ben Skeggs <bskeggs@redhat.com> Cc: Hugh Dickins <hughd@google.com> Cc: Jason Gunthorpe <jgg@nvidia.com> Cc: John Hubbard <jhubbard@nvidia.com> Cc: "Matthew Wilcox (Oracle)" <willy@infradead.org> Cc: Peter Xu <peterx@redhat.com> Cc: Ralph Campbell <rcampbell@nvidia.com> Cc: Shakeel Butt <shakeelb@google.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2021-07-01 09:54:25 +08:00
if (is_device_exclusive_entry(entry))
goto fault;
if (is_migration_entry(entry)) {
pte_unmap(ptep);
hmm_vma_walk->last = addr;
migration_entry_wait(walk->mm, pmdp, addr);
return -EBUSY;
}
/* Report error for everything else */
pte_unmap(ptep);
return -EFAULT;
}
cpu_flags = pte_to_hmm_pfn_flags(range, pte);
mm/hmm: remove the customizable pfn format from hmm_range_fault Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Tested-by: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-02 02:20:48 +08:00
required_fault =
hmm_pte_need_fault(hmm_vma_walk, pfn_req_flags, cpu_flags);
if (required_fault)
goto fault;
/*
* Since each architecture defines a struct page for the zero page, just
* fall through and treat it like a normal page.
*/
if (pte_special(pte) && !is_zero_pfn(pte_pfn(pte))) {
mm/hmm: remove the customizable pfn format from hmm_range_fault Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Tested-by: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-02 02:20:48 +08:00
if (hmm_pte_need_fault(hmm_vma_walk, pfn_req_flags, 0)) {
pte_unmap(ptep);
return -EFAULT;
}
mm/hmm: remove the customizable pfn format from hmm_range_fault Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Tested-by: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-02 02:20:48 +08:00
*hmm_pfn = HMM_PFN_ERROR;
return 0;
}
mm/hmm: remove the customizable pfn format from hmm_range_fault Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Tested-by: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-02 02:20:48 +08:00
*hmm_pfn = pte_pfn(pte) | cpu_flags;
return 0;
fault:
pte_unmap(ptep);
/* Fault any virtual address we were asked to fault */
return hmm_vma_fault(addr, end, required_fault, walk);
}
static int hmm_vma_walk_pmd(pmd_t *pmdp,
unsigned long start,
unsigned long end,
struct mm_walk *walk)
{
struct hmm_vma_walk *hmm_vma_walk = walk->private;
struct hmm_range *range = hmm_vma_walk->range;
mm/hmm: remove the customizable pfn format from hmm_range_fault Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Tested-by: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-02 02:20:48 +08:00
unsigned long *hmm_pfns =
&range->hmm_pfns[(start - range->start) >> PAGE_SHIFT];
unsigned long npages = (end - start) >> PAGE_SHIFT;
unsigned long addr = start;
pte_t *ptep;
pmd_t pmd;
again:
pmd = READ_ONCE(*pmdp);
if (pmd_none(pmd))
mm: pagewalk: add 'depth' parameter to pte_hole The pte_hole() callback is called at multiple levels of the page tables. Code dumping the kernel page tables needs to know what at what depth the missing entry is. Add this is an extra parameter to pte_hole(). When the depth isn't know (e.g. processing a vma) then -1 is passed. The depth that is reported is the actual level where the entry is missing (ignoring any folding that is in place), i.e. any levels where PTRS_PER_P?D is set to 1 are ignored. Note that depth starts at 0 for a PGD so that PUD/PMD/PTE retain their natural numbers as levels 2/3/4. Link: http://lkml.kernel.org/r/20191218162402.45610-16-steven.price@arm.com Signed-off-by: Steven Price <steven.price@arm.com> Tested-by: Zong Li <zong.li@sifive.com> Cc: Albert Ou <aou@eecs.berkeley.edu> Cc: Alexandre Ghiti <alex@ghiti.fr> Cc: Andy Lutomirski <luto@kernel.org> Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Borislav Petkov <bp@alien8.de> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Christian Borntraeger <borntraeger@de.ibm.com> Cc: Dave Hansen <dave.hansen@linux.intel.com> Cc: David S. Miller <davem@davemloft.net> Cc: Heiko Carstens <heiko.carstens@de.ibm.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: James Hogan <jhogan@kernel.org> Cc: James Morse <james.morse@arm.com> Cc: Jerome Glisse <jglisse@redhat.com> Cc: "Liang, Kan" <kan.liang@linux.intel.com> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Michael Ellerman <mpe@ellerman.id.au> Cc: Paul Burton <paul.burton@mips.com> Cc: Paul Mackerras <paulus@samba.org> Cc: Paul Walmsley <paul.walmsley@sifive.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Russell King <linux@armlinux.org.uk> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Vasily Gorbik <gor@linux.ibm.com> Cc: Vineet Gupta <vgupta@synopsys.com> Cc: Will Deacon <will@kernel.org> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-02-04 09:36:03 +08:00
return hmm_vma_walk_hole(start, end, -1, walk);
if (thp_migration_supported() && is_pmd_migration_entry(pmd)) {
mm/hmm: remove the customizable pfn format from hmm_range_fault Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Tested-by: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-02 02:20:48 +08:00
if (hmm_range_need_fault(hmm_vma_walk, hmm_pfns, npages, 0)) {
hmm_vma_walk->last = addr;
pmd_migration_entry_wait(walk->mm, pmdp);
return -EBUSY;
}
mm/hmm: remove the customizable pfn format from hmm_range_fault Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Tested-by: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-02 02:20:48 +08:00
return hmm_pfns_fill(start, end, range, 0);
}
if (!pmd_present(pmd)) {
mm/hmm: remove the customizable pfn format from hmm_range_fault Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Tested-by: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-02 02:20:48 +08:00
if (hmm_range_need_fault(hmm_vma_walk, hmm_pfns, npages, 0))
return -EFAULT;
return hmm_pfns_fill(start, end, range, HMM_PFN_ERROR);
}
if (pmd_devmap(pmd) || pmd_trans_huge(pmd)) {
/*
* No need to take pmd_lock here, even if some other thread
* is splitting the huge pmd we will get that event through
* mmu_notifier callback.
*
* So just read pmd value and check again it's a transparent
* huge or device mapping one and compute corresponding pfn
* values.
*/
pmd = pmd_read_atomic(pmdp);
barrier();
if (!pmd_devmap(pmd) && !pmd_trans_huge(pmd))
goto again;
mm/hmm: remove the customizable pfn format from hmm_range_fault Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Tested-by: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-02 02:20:48 +08:00
return hmm_vma_handle_pmd(walk, addr, end, hmm_pfns, pmd);
}
/*
* We have handled all the valid cases above ie either none, migration,
* huge or transparent huge. At this point either it is a valid pmd
* entry pointing to pte directory or it is a bad pmd that will not
* recover.
*/
if (pmd_bad(pmd)) {
mm/hmm: remove the customizable pfn format from hmm_range_fault Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Tested-by: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-02 02:20:48 +08:00
if (hmm_range_need_fault(hmm_vma_walk, hmm_pfns, npages, 0))
return -EFAULT;
return hmm_pfns_fill(start, end, range, HMM_PFN_ERROR);
}
ptep = pte_offset_map(pmdp, addr);
mm/hmm: remove the customizable pfn format from hmm_range_fault Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Tested-by: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-02 02:20:48 +08:00
for (; addr < end; addr += PAGE_SIZE, ptep++, hmm_pfns++) {
int r;
mm/hmm: remove the customizable pfn format from hmm_range_fault Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Tested-by: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-02 02:20:48 +08:00
r = hmm_vma_handle_pte(walk, addr, end, pmdp, ptep, hmm_pfns);
if (r) {
/* hmm_vma_handle_pte() did pte_unmap() */
return r;
}
}
pte_unmap(ptep - 1);
return 0;
}
#if defined(CONFIG_ARCH_HAS_PTE_DEVMAP) && \
defined(CONFIG_HAVE_ARCH_TRANSPARENT_HUGEPAGE_PUD)
mm/hmm: remove the customizable pfn format from hmm_range_fault Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Tested-by: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-02 02:20:48 +08:00
static inline unsigned long pud_to_hmm_pfn_flags(struct hmm_range *range,
pud_t pud)
{
if (!pud_present(pud))
return 0;
return (pud_write(pud) ? (HMM_PFN_VALID | HMM_PFN_WRITE) :
HMM_PFN_VALID) |
hmm_pfn_flags_order(PUD_SHIFT - PAGE_SHIFT);
}
static int hmm_vma_walk_pud(pud_t *pudp, unsigned long start, unsigned long end,
struct mm_walk *walk)
{
struct hmm_vma_walk *hmm_vma_walk = walk->private;
struct hmm_range *range = hmm_vma_walk->range;
mm: pagewalk: add p4d_entry() and pgd_entry() pgd_entry() and pud_entry() were removed by commit 0b1fbfe50006c410 ("mm/pagewalk: remove pgd_entry() and pud_entry()") because there were no users. We're about to add users so reintroduce them, along with p4d_entry() as we now have 5 levels of tables. Note that commit a00cc7d9dd93d66a ("mm, x86: add support for PUD-sized transparent hugepages") already re-added pud_entry() but with different semantics to the other callbacks. This commit reverts the semantics back to match the other callbacks. To support hmm.c which now uses the new semantics of pud_entry() a new member ('action') of struct mm_walk is added which allows the callbacks to either descend (ACTION_SUBTREE, the default), skip (ACTION_CONTINUE) or repeat the callback (ACTION_AGAIN). hmm.c is then updated to call pud_trans_huge_lock() itself and make use of the splitting/retry logic of the core code. After this change pud_entry() is called for all entries, not just transparent huge pages. [arnd@arndb.de: fix unused variable warning] Link: http://lkml.kernel.org/r/20200107204607.1533842-1-arnd@arndb.de Link: http://lkml.kernel.org/r/20191218162402.45610-12-steven.price@arm.com Signed-off-by: Steven Price <steven.price@arm.com> Signed-off-by: Arnd Bergmann <arnd@arndb.de> Cc: Albert Ou <aou@eecs.berkeley.edu> Cc: Alexandre Ghiti <alex@ghiti.fr> Cc: Andy Lutomirski <luto@kernel.org> Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Borislav Petkov <bp@alien8.de> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Christian Borntraeger <borntraeger@de.ibm.com> Cc: Dave Hansen <dave.hansen@linux.intel.com> Cc: David S. Miller <davem@davemloft.net> Cc: Heiko Carstens <heiko.carstens@de.ibm.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: James Hogan <jhogan@kernel.org> Cc: James Morse <james.morse@arm.com> Cc: Jerome Glisse <jglisse@redhat.com> Cc: "Liang, Kan" <kan.liang@linux.intel.com> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Michael Ellerman <mpe@ellerman.id.au> Cc: Paul Burton <paul.burton@mips.com> Cc: Paul Mackerras <paulus@samba.org> Cc: Paul Walmsley <paul.walmsley@sifive.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Russell King <linux@armlinux.org.uk> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Vasily Gorbik <gor@linux.ibm.com> Cc: Vineet Gupta <vgupta@synopsys.com> Cc: Will Deacon <will@kernel.org> Cc: Zong Li <zong.li@sifive.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-02-04 09:35:45 +08:00
unsigned long addr = start;
pud_t pud;
mm: pagewalk: add p4d_entry() and pgd_entry() pgd_entry() and pud_entry() were removed by commit 0b1fbfe50006c410 ("mm/pagewalk: remove pgd_entry() and pud_entry()") because there were no users. We're about to add users so reintroduce them, along with p4d_entry() as we now have 5 levels of tables. Note that commit a00cc7d9dd93d66a ("mm, x86: add support for PUD-sized transparent hugepages") already re-added pud_entry() but with different semantics to the other callbacks. This commit reverts the semantics back to match the other callbacks. To support hmm.c which now uses the new semantics of pud_entry() a new member ('action') of struct mm_walk is added which allows the callbacks to either descend (ACTION_SUBTREE, the default), skip (ACTION_CONTINUE) or repeat the callback (ACTION_AGAIN). hmm.c is then updated to call pud_trans_huge_lock() itself and make use of the splitting/retry logic of the core code. After this change pud_entry() is called for all entries, not just transparent huge pages. [arnd@arndb.de: fix unused variable warning] Link: http://lkml.kernel.org/r/20200107204607.1533842-1-arnd@arndb.de Link: http://lkml.kernel.org/r/20191218162402.45610-12-steven.price@arm.com Signed-off-by: Steven Price <steven.price@arm.com> Signed-off-by: Arnd Bergmann <arnd@arndb.de> Cc: Albert Ou <aou@eecs.berkeley.edu> Cc: Alexandre Ghiti <alex@ghiti.fr> Cc: Andy Lutomirski <luto@kernel.org> Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Borislav Petkov <bp@alien8.de> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Christian Borntraeger <borntraeger@de.ibm.com> Cc: Dave Hansen <dave.hansen@linux.intel.com> Cc: David S. Miller <davem@davemloft.net> Cc: Heiko Carstens <heiko.carstens@de.ibm.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: James Hogan <jhogan@kernel.org> Cc: James Morse <james.morse@arm.com> Cc: Jerome Glisse <jglisse@redhat.com> Cc: "Liang, Kan" <kan.liang@linux.intel.com> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Michael Ellerman <mpe@ellerman.id.au> Cc: Paul Burton <paul.burton@mips.com> Cc: Paul Mackerras <paulus@samba.org> Cc: Paul Walmsley <paul.walmsley@sifive.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Russell King <linux@armlinux.org.uk> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Vasily Gorbik <gor@linux.ibm.com> Cc: Vineet Gupta <vgupta@synopsys.com> Cc: Will Deacon <will@kernel.org> Cc: Zong Li <zong.li@sifive.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-02-04 09:35:45 +08:00
int ret = 0;
spinlock_t *ptl = pud_trans_huge_lock(pudp, walk->vma);
if (!ptl)
return 0;
/* Normally we don't want to split the huge page */
walk->action = ACTION_CONTINUE;
pud = READ_ONCE(*pudp);
mm: pagewalk: add p4d_entry() and pgd_entry() pgd_entry() and pud_entry() were removed by commit 0b1fbfe50006c410 ("mm/pagewalk: remove pgd_entry() and pud_entry()") because there were no users. We're about to add users so reintroduce them, along with p4d_entry() as we now have 5 levels of tables. Note that commit a00cc7d9dd93d66a ("mm, x86: add support for PUD-sized transparent hugepages") already re-added pud_entry() but with different semantics to the other callbacks. This commit reverts the semantics back to match the other callbacks. To support hmm.c which now uses the new semantics of pud_entry() a new member ('action') of struct mm_walk is added which allows the callbacks to either descend (ACTION_SUBTREE, the default), skip (ACTION_CONTINUE) or repeat the callback (ACTION_AGAIN). hmm.c is then updated to call pud_trans_huge_lock() itself and make use of the splitting/retry logic of the core code. After this change pud_entry() is called for all entries, not just transparent huge pages. [arnd@arndb.de: fix unused variable warning] Link: http://lkml.kernel.org/r/20200107204607.1533842-1-arnd@arndb.de Link: http://lkml.kernel.org/r/20191218162402.45610-12-steven.price@arm.com Signed-off-by: Steven Price <steven.price@arm.com> Signed-off-by: Arnd Bergmann <arnd@arndb.de> Cc: Albert Ou <aou@eecs.berkeley.edu> Cc: Alexandre Ghiti <alex@ghiti.fr> Cc: Andy Lutomirski <luto@kernel.org> Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Borislav Petkov <bp@alien8.de> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Christian Borntraeger <borntraeger@de.ibm.com> Cc: Dave Hansen <dave.hansen@linux.intel.com> Cc: David S. Miller <davem@davemloft.net> Cc: Heiko Carstens <heiko.carstens@de.ibm.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: James Hogan <jhogan@kernel.org> Cc: James Morse <james.morse@arm.com> Cc: Jerome Glisse <jglisse@redhat.com> Cc: "Liang, Kan" <kan.liang@linux.intel.com> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Michael Ellerman <mpe@ellerman.id.au> Cc: Paul Burton <paul.burton@mips.com> Cc: Paul Mackerras <paulus@samba.org> Cc: Paul Walmsley <paul.walmsley@sifive.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Russell King <linux@armlinux.org.uk> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Vasily Gorbik <gor@linux.ibm.com> Cc: Vineet Gupta <vgupta@synopsys.com> Cc: Will Deacon <will@kernel.org> Cc: Zong Li <zong.li@sifive.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-02-04 09:35:45 +08:00
if (pud_none(pud)) {
spin_unlock(ptl);
return hmm_vma_walk_hole(start, end, -1, walk);
mm: pagewalk: add p4d_entry() and pgd_entry() pgd_entry() and pud_entry() were removed by commit 0b1fbfe50006c410 ("mm/pagewalk: remove pgd_entry() and pud_entry()") because there were no users. We're about to add users so reintroduce them, along with p4d_entry() as we now have 5 levels of tables. Note that commit a00cc7d9dd93d66a ("mm, x86: add support for PUD-sized transparent hugepages") already re-added pud_entry() but with different semantics to the other callbacks. This commit reverts the semantics back to match the other callbacks. To support hmm.c which now uses the new semantics of pud_entry() a new member ('action') of struct mm_walk is added which allows the callbacks to either descend (ACTION_SUBTREE, the default), skip (ACTION_CONTINUE) or repeat the callback (ACTION_AGAIN). hmm.c is then updated to call pud_trans_huge_lock() itself and make use of the splitting/retry logic of the core code. After this change pud_entry() is called for all entries, not just transparent huge pages. [arnd@arndb.de: fix unused variable warning] Link: http://lkml.kernel.org/r/20200107204607.1533842-1-arnd@arndb.de Link: http://lkml.kernel.org/r/20191218162402.45610-12-steven.price@arm.com Signed-off-by: Steven Price <steven.price@arm.com> Signed-off-by: Arnd Bergmann <arnd@arndb.de> Cc: Albert Ou <aou@eecs.berkeley.edu> Cc: Alexandre Ghiti <alex@ghiti.fr> Cc: Andy Lutomirski <luto@kernel.org> Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Borislav Petkov <bp@alien8.de> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Christian Borntraeger <borntraeger@de.ibm.com> Cc: Dave Hansen <dave.hansen@linux.intel.com> Cc: David S. Miller <davem@davemloft.net> Cc: Heiko Carstens <heiko.carstens@de.ibm.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: James Hogan <jhogan@kernel.org> Cc: James Morse <james.morse@arm.com> Cc: Jerome Glisse <jglisse@redhat.com> Cc: "Liang, Kan" <kan.liang@linux.intel.com> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Michael Ellerman <mpe@ellerman.id.au> Cc: Paul Burton <paul.burton@mips.com> Cc: Paul Mackerras <paulus@samba.org> Cc: Paul Walmsley <paul.walmsley@sifive.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Russell King <linux@armlinux.org.uk> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Vasily Gorbik <gor@linux.ibm.com> Cc: Vineet Gupta <vgupta@synopsys.com> Cc: Will Deacon <will@kernel.org> Cc: Zong Li <zong.li@sifive.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-02-04 09:35:45 +08:00
}
if (pud_huge(pud) && pud_devmap(pud)) {
unsigned long i, npages, pfn;
unsigned int required_fault;
mm/hmm: remove the customizable pfn format from hmm_range_fault Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Tested-by: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-02 02:20:48 +08:00
unsigned long *hmm_pfns;
unsigned long cpu_flags;
mm: pagewalk: add p4d_entry() and pgd_entry() pgd_entry() and pud_entry() were removed by commit 0b1fbfe50006c410 ("mm/pagewalk: remove pgd_entry() and pud_entry()") because there were no users. We're about to add users so reintroduce them, along with p4d_entry() as we now have 5 levels of tables. Note that commit a00cc7d9dd93d66a ("mm, x86: add support for PUD-sized transparent hugepages") already re-added pud_entry() but with different semantics to the other callbacks. This commit reverts the semantics back to match the other callbacks. To support hmm.c which now uses the new semantics of pud_entry() a new member ('action') of struct mm_walk is added which allows the callbacks to either descend (ACTION_SUBTREE, the default), skip (ACTION_CONTINUE) or repeat the callback (ACTION_AGAIN). hmm.c is then updated to call pud_trans_huge_lock() itself and make use of the splitting/retry logic of the core code. After this change pud_entry() is called for all entries, not just transparent huge pages. [arnd@arndb.de: fix unused variable warning] Link: http://lkml.kernel.org/r/20200107204607.1533842-1-arnd@arndb.de Link: http://lkml.kernel.org/r/20191218162402.45610-12-steven.price@arm.com Signed-off-by: Steven Price <steven.price@arm.com> Signed-off-by: Arnd Bergmann <arnd@arndb.de> Cc: Albert Ou <aou@eecs.berkeley.edu> Cc: Alexandre Ghiti <alex@ghiti.fr> Cc: Andy Lutomirski <luto@kernel.org> Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Borislav Petkov <bp@alien8.de> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Christian Borntraeger <borntraeger@de.ibm.com> Cc: Dave Hansen <dave.hansen@linux.intel.com> Cc: David S. Miller <davem@davemloft.net> Cc: Heiko Carstens <heiko.carstens@de.ibm.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: James Hogan <jhogan@kernel.org> Cc: James Morse <james.morse@arm.com> Cc: Jerome Glisse <jglisse@redhat.com> Cc: "Liang, Kan" <kan.liang@linux.intel.com> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Michael Ellerman <mpe@ellerman.id.au> Cc: Paul Burton <paul.burton@mips.com> Cc: Paul Mackerras <paulus@samba.org> Cc: Paul Walmsley <paul.walmsley@sifive.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Russell King <linux@armlinux.org.uk> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Vasily Gorbik <gor@linux.ibm.com> Cc: Vineet Gupta <vgupta@synopsys.com> Cc: Will Deacon <will@kernel.org> Cc: Zong Li <zong.li@sifive.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-02-04 09:35:45 +08:00
if (!pud_present(pud)) {
spin_unlock(ptl);
return hmm_vma_walk_hole(start, end, -1, walk);
mm: pagewalk: add p4d_entry() and pgd_entry() pgd_entry() and pud_entry() were removed by commit 0b1fbfe50006c410 ("mm/pagewalk: remove pgd_entry() and pud_entry()") because there were no users. We're about to add users so reintroduce them, along with p4d_entry() as we now have 5 levels of tables. Note that commit a00cc7d9dd93d66a ("mm, x86: add support for PUD-sized transparent hugepages") already re-added pud_entry() but with different semantics to the other callbacks. This commit reverts the semantics back to match the other callbacks. To support hmm.c which now uses the new semantics of pud_entry() a new member ('action') of struct mm_walk is added which allows the callbacks to either descend (ACTION_SUBTREE, the default), skip (ACTION_CONTINUE) or repeat the callback (ACTION_AGAIN). hmm.c is then updated to call pud_trans_huge_lock() itself and make use of the splitting/retry logic of the core code. After this change pud_entry() is called for all entries, not just transparent huge pages. [arnd@arndb.de: fix unused variable warning] Link: http://lkml.kernel.org/r/20200107204607.1533842-1-arnd@arndb.de Link: http://lkml.kernel.org/r/20191218162402.45610-12-steven.price@arm.com Signed-off-by: Steven Price <steven.price@arm.com> Signed-off-by: Arnd Bergmann <arnd@arndb.de> Cc: Albert Ou <aou@eecs.berkeley.edu> Cc: Alexandre Ghiti <alex@ghiti.fr> Cc: Andy Lutomirski <luto@kernel.org> Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Borislav Petkov <bp@alien8.de> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Christian Borntraeger <borntraeger@de.ibm.com> Cc: Dave Hansen <dave.hansen@linux.intel.com> Cc: David S. Miller <davem@davemloft.net> Cc: Heiko Carstens <heiko.carstens@de.ibm.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: James Hogan <jhogan@kernel.org> Cc: James Morse <james.morse@arm.com> Cc: Jerome Glisse <jglisse@redhat.com> Cc: "Liang, Kan" <kan.liang@linux.intel.com> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Michael Ellerman <mpe@ellerman.id.au> Cc: Paul Burton <paul.burton@mips.com> Cc: Paul Mackerras <paulus@samba.org> Cc: Paul Walmsley <paul.walmsley@sifive.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Russell King <linux@armlinux.org.uk> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Vasily Gorbik <gor@linux.ibm.com> Cc: Vineet Gupta <vgupta@synopsys.com> Cc: Will Deacon <will@kernel.org> Cc: Zong Li <zong.li@sifive.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-02-04 09:35:45 +08:00
}
i = (addr - range->start) >> PAGE_SHIFT;
npages = (end - addr) >> PAGE_SHIFT;
mm/hmm: remove the customizable pfn format from hmm_range_fault Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Tested-by: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-02 02:20:48 +08:00
hmm_pfns = &range->hmm_pfns[i];
cpu_flags = pud_to_hmm_pfn_flags(range, pud);
mm/hmm: remove the customizable pfn format from hmm_range_fault Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Tested-by: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-02 02:20:48 +08:00
required_fault = hmm_range_need_fault(hmm_vma_walk, hmm_pfns,
npages, cpu_flags);
if (required_fault) {
spin_unlock(ptl);
return hmm_vma_fault(addr, end, required_fault, walk);
mm: pagewalk: add p4d_entry() and pgd_entry() pgd_entry() and pud_entry() were removed by commit 0b1fbfe50006c410 ("mm/pagewalk: remove pgd_entry() and pud_entry()") because there were no users. We're about to add users so reintroduce them, along with p4d_entry() as we now have 5 levels of tables. Note that commit a00cc7d9dd93d66a ("mm, x86: add support for PUD-sized transparent hugepages") already re-added pud_entry() but with different semantics to the other callbacks. This commit reverts the semantics back to match the other callbacks. To support hmm.c which now uses the new semantics of pud_entry() a new member ('action') of struct mm_walk is added which allows the callbacks to either descend (ACTION_SUBTREE, the default), skip (ACTION_CONTINUE) or repeat the callback (ACTION_AGAIN). hmm.c is then updated to call pud_trans_huge_lock() itself and make use of the splitting/retry logic of the core code. After this change pud_entry() is called for all entries, not just transparent huge pages. [arnd@arndb.de: fix unused variable warning] Link: http://lkml.kernel.org/r/20200107204607.1533842-1-arnd@arndb.de Link: http://lkml.kernel.org/r/20191218162402.45610-12-steven.price@arm.com Signed-off-by: Steven Price <steven.price@arm.com> Signed-off-by: Arnd Bergmann <arnd@arndb.de> Cc: Albert Ou <aou@eecs.berkeley.edu> Cc: Alexandre Ghiti <alex@ghiti.fr> Cc: Andy Lutomirski <luto@kernel.org> Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Borislav Petkov <bp@alien8.de> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Christian Borntraeger <borntraeger@de.ibm.com> Cc: Dave Hansen <dave.hansen@linux.intel.com> Cc: David S. Miller <davem@davemloft.net> Cc: Heiko Carstens <heiko.carstens@de.ibm.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: James Hogan <jhogan@kernel.org> Cc: James Morse <james.morse@arm.com> Cc: Jerome Glisse <jglisse@redhat.com> Cc: "Liang, Kan" <kan.liang@linux.intel.com> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Michael Ellerman <mpe@ellerman.id.au> Cc: Paul Burton <paul.burton@mips.com> Cc: Paul Mackerras <paulus@samba.org> Cc: Paul Walmsley <paul.walmsley@sifive.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Russell King <linux@armlinux.org.uk> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Vasily Gorbik <gor@linux.ibm.com> Cc: Vineet Gupta <vgupta@synopsys.com> Cc: Will Deacon <will@kernel.org> Cc: Zong Li <zong.li@sifive.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-02-04 09:35:45 +08:00
}
pfn = pud_pfn(pud) + ((addr & ~PUD_MASK) >> PAGE_SHIFT);
for (i = 0; i < npages; ++i, ++pfn)
mm/hmm: remove the customizable pfn format from hmm_range_fault Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Tested-by: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-02 02:20:48 +08:00
hmm_pfns[i] = pfn | cpu_flags;
mm: pagewalk: add p4d_entry() and pgd_entry() pgd_entry() and pud_entry() were removed by commit 0b1fbfe50006c410 ("mm/pagewalk: remove pgd_entry() and pud_entry()") because there were no users. We're about to add users so reintroduce them, along with p4d_entry() as we now have 5 levels of tables. Note that commit a00cc7d9dd93d66a ("mm, x86: add support for PUD-sized transparent hugepages") already re-added pud_entry() but with different semantics to the other callbacks. This commit reverts the semantics back to match the other callbacks. To support hmm.c which now uses the new semantics of pud_entry() a new member ('action') of struct mm_walk is added which allows the callbacks to either descend (ACTION_SUBTREE, the default), skip (ACTION_CONTINUE) or repeat the callback (ACTION_AGAIN). hmm.c is then updated to call pud_trans_huge_lock() itself and make use of the splitting/retry logic of the core code. After this change pud_entry() is called for all entries, not just transparent huge pages. [arnd@arndb.de: fix unused variable warning] Link: http://lkml.kernel.org/r/20200107204607.1533842-1-arnd@arndb.de Link: http://lkml.kernel.org/r/20191218162402.45610-12-steven.price@arm.com Signed-off-by: Steven Price <steven.price@arm.com> Signed-off-by: Arnd Bergmann <arnd@arndb.de> Cc: Albert Ou <aou@eecs.berkeley.edu> Cc: Alexandre Ghiti <alex@ghiti.fr> Cc: Andy Lutomirski <luto@kernel.org> Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Borislav Petkov <bp@alien8.de> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Christian Borntraeger <borntraeger@de.ibm.com> Cc: Dave Hansen <dave.hansen@linux.intel.com> Cc: David S. Miller <davem@davemloft.net> Cc: Heiko Carstens <heiko.carstens@de.ibm.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: James Hogan <jhogan@kernel.org> Cc: James Morse <james.morse@arm.com> Cc: Jerome Glisse <jglisse@redhat.com> Cc: "Liang, Kan" <kan.liang@linux.intel.com> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Michael Ellerman <mpe@ellerman.id.au> Cc: Paul Burton <paul.burton@mips.com> Cc: Paul Mackerras <paulus@samba.org> Cc: Paul Walmsley <paul.walmsley@sifive.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Russell King <linux@armlinux.org.uk> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Vasily Gorbik <gor@linux.ibm.com> Cc: Vineet Gupta <vgupta@synopsys.com> Cc: Will Deacon <will@kernel.org> Cc: Zong Li <zong.li@sifive.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-02-04 09:35:45 +08:00
goto out_unlock;
}
mm: pagewalk: add p4d_entry() and pgd_entry() pgd_entry() and pud_entry() were removed by commit 0b1fbfe50006c410 ("mm/pagewalk: remove pgd_entry() and pud_entry()") because there were no users. We're about to add users so reintroduce them, along with p4d_entry() as we now have 5 levels of tables. Note that commit a00cc7d9dd93d66a ("mm, x86: add support for PUD-sized transparent hugepages") already re-added pud_entry() but with different semantics to the other callbacks. This commit reverts the semantics back to match the other callbacks. To support hmm.c which now uses the new semantics of pud_entry() a new member ('action') of struct mm_walk is added which allows the callbacks to either descend (ACTION_SUBTREE, the default), skip (ACTION_CONTINUE) or repeat the callback (ACTION_AGAIN). hmm.c is then updated to call pud_trans_huge_lock() itself and make use of the splitting/retry logic of the core code. After this change pud_entry() is called for all entries, not just transparent huge pages. [arnd@arndb.de: fix unused variable warning] Link: http://lkml.kernel.org/r/20200107204607.1533842-1-arnd@arndb.de Link: http://lkml.kernel.org/r/20191218162402.45610-12-steven.price@arm.com Signed-off-by: Steven Price <steven.price@arm.com> Signed-off-by: Arnd Bergmann <arnd@arndb.de> Cc: Albert Ou <aou@eecs.berkeley.edu> Cc: Alexandre Ghiti <alex@ghiti.fr> Cc: Andy Lutomirski <luto@kernel.org> Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Borislav Petkov <bp@alien8.de> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Christian Borntraeger <borntraeger@de.ibm.com> Cc: Dave Hansen <dave.hansen@linux.intel.com> Cc: David S. Miller <davem@davemloft.net> Cc: Heiko Carstens <heiko.carstens@de.ibm.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: James Hogan <jhogan@kernel.org> Cc: James Morse <james.morse@arm.com> Cc: Jerome Glisse <jglisse@redhat.com> Cc: "Liang, Kan" <kan.liang@linux.intel.com> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Michael Ellerman <mpe@ellerman.id.au> Cc: Paul Burton <paul.burton@mips.com> Cc: Paul Mackerras <paulus@samba.org> Cc: Paul Walmsley <paul.walmsley@sifive.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Russell King <linux@armlinux.org.uk> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Vasily Gorbik <gor@linux.ibm.com> Cc: Vineet Gupta <vgupta@synopsys.com> Cc: Will Deacon <will@kernel.org> Cc: Zong Li <zong.li@sifive.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-02-04 09:35:45 +08:00
/* Ask for the PUD to be split */
walk->action = ACTION_SUBTREE;
mm: pagewalk: add p4d_entry() and pgd_entry() pgd_entry() and pud_entry() were removed by commit 0b1fbfe50006c410 ("mm/pagewalk: remove pgd_entry() and pud_entry()") because there were no users. We're about to add users so reintroduce them, along with p4d_entry() as we now have 5 levels of tables. Note that commit a00cc7d9dd93d66a ("mm, x86: add support for PUD-sized transparent hugepages") already re-added pud_entry() but with different semantics to the other callbacks. This commit reverts the semantics back to match the other callbacks. To support hmm.c which now uses the new semantics of pud_entry() a new member ('action') of struct mm_walk is added which allows the callbacks to either descend (ACTION_SUBTREE, the default), skip (ACTION_CONTINUE) or repeat the callback (ACTION_AGAIN). hmm.c is then updated to call pud_trans_huge_lock() itself and make use of the splitting/retry logic of the core code. After this change pud_entry() is called for all entries, not just transparent huge pages. [arnd@arndb.de: fix unused variable warning] Link: http://lkml.kernel.org/r/20200107204607.1533842-1-arnd@arndb.de Link: http://lkml.kernel.org/r/20191218162402.45610-12-steven.price@arm.com Signed-off-by: Steven Price <steven.price@arm.com> Signed-off-by: Arnd Bergmann <arnd@arndb.de> Cc: Albert Ou <aou@eecs.berkeley.edu> Cc: Alexandre Ghiti <alex@ghiti.fr> Cc: Andy Lutomirski <luto@kernel.org> Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org> Cc: Arnd Bergmann <arnd@arndb.de> Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Borislav Petkov <bp@alien8.de> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Christian Borntraeger <borntraeger@de.ibm.com> Cc: Dave Hansen <dave.hansen@linux.intel.com> Cc: David S. Miller <davem@davemloft.net> Cc: Heiko Carstens <heiko.carstens@de.ibm.com> Cc: "H. Peter Anvin" <hpa@zytor.com> Cc: Ingo Molnar <mingo@redhat.com> Cc: James Hogan <jhogan@kernel.org> Cc: James Morse <james.morse@arm.com> Cc: Jerome Glisse <jglisse@redhat.com> Cc: "Liang, Kan" <kan.liang@linux.intel.com> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Michael Ellerman <mpe@ellerman.id.au> Cc: Paul Burton <paul.burton@mips.com> Cc: Paul Mackerras <paulus@samba.org> Cc: Paul Walmsley <paul.walmsley@sifive.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Ralf Baechle <ralf@linux-mips.org> Cc: Russell King <linux@armlinux.org.uk> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Vasily Gorbik <gor@linux.ibm.com> Cc: Vineet Gupta <vgupta@synopsys.com> Cc: Will Deacon <will@kernel.org> Cc: Zong Li <zong.li@sifive.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2020-02-04 09:35:45 +08:00
out_unlock:
spin_unlock(ptl);
return ret;
}
#else
#define hmm_vma_walk_pud NULL
#endif
#ifdef CONFIG_HUGETLB_PAGE
static int hmm_vma_walk_hugetlb_entry(pte_t *pte, unsigned long hmask,
unsigned long start, unsigned long end,
struct mm_walk *walk)
{
unsigned long addr = start, i, pfn;
struct hmm_vma_walk *hmm_vma_walk = walk->private;
struct hmm_range *range = hmm_vma_walk->range;
struct vm_area_struct *vma = walk->vma;
unsigned int required_fault;
mm/hmm: remove the customizable pfn format from hmm_range_fault Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Tested-by: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-02 02:20:48 +08:00
unsigned long pfn_req_flags;
unsigned long cpu_flags;
spinlock_t *ptl;
pte_t entry;
ptl = huge_pte_lock(hstate_vma(vma), walk->mm, pte);
entry = huge_ptep_get(pte);
i = (start - range->start) >> PAGE_SHIFT;
mm/hmm: remove the customizable pfn format from hmm_range_fault Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Tested-by: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-02 02:20:48 +08:00
pfn_req_flags = range->hmm_pfns[i];
cpu_flags = pte_to_hmm_pfn_flags(range, entry) |
hmm_pfn_flags_order(huge_page_order(hstate_vma(vma)));
mm/hmm: remove the customizable pfn format from hmm_range_fault Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Tested-by: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-02 02:20:48 +08:00
required_fault =
hmm_pte_need_fault(hmm_vma_walk, pfn_req_flags, cpu_flags);
if (required_fault) {
spin_unlock(ptl);
return hmm_vma_fault(addr, end, required_fault, walk);
}
pfn = pte_pfn(entry) + ((start & ~hmask) >> PAGE_SHIFT);
for (; addr < end; addr += PAGE_SIZE, i++, pfn++)
mm/hmm: remove the customizable pfn format from hmm_range_fault Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Tested-by: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-02 02:20:48 +08:00
range->hmm_pfns[i] = pfn | cpu_flags;
spin_unlock(ptl);
return 0;
}
#else
#define hmm_vma_walk_hugetlb_entry NULL
#endif /* CONFIG_HUGETLB_PAGE */
static int hmm_vma_walk_test(unsigned long start, unsigned long end,
struct mm_walk *walk)
{
struct hmm_vma_walk *hmm_vma_walk = walk->private;
struct hmm_range *range = hmm_vma_walk->range;
struct vm_area_struct *vma = walk->vma;
if (!(vma->vm_flags & (VM_IO | VM_PFNMAP | VM_MIXEDMAP)) &&
vma->vm_flags & VM_READ)
return 0;
/*
* vma ranges that don't have struct page backing them or map I/O
* devices directly cannot be handled by hmm_range_fault().
*
* If the vma does not allow read access, then assume that it does not
* allow write access either. HMM does not support architectures that
* allow write without read.
*
* If a fault is requested for an unsupported range then it is a hard
* failure.
*/
if (hmm_range_need_fault(hmm_vma_walk,
mm/hmm: remove the customizable pfn format from hmm_range_fault Presumably the intent here was that hmm_range_fault() could put the data into some HW specific format and thus avoid some work. However, nothing actually does that, and it isn't clear how anything actually could do that as hmm_range_fault() provides CPU addresses which must be DMA mapped. Perhaps there is some special HW that does not need DMA mapping, but we don't have any examples of this, and the theoretical performance win of avoiding an extra scan over the pfns array doesn't seem worth the complexity. Plus pfns needs to be scanned anyhow to sort out any DEVICE_PRIVATE pages. This version replaces the uint64_t with an usigned long containing a pfn and fixed flags. On input flags is filled with the HMM_PFN_REQ_* values, on successful output it is filled with HMM_PFN_* values, describing the state of the pages. amdgpu is simple to convert, it doesn't use snapshot and doesn't use per-page flags. nouveau uses only 16 hmm_pte entries at most (ie fits in a few cache lines), and it sweeps over its pfns array a couple of times anyhow. It also has a nasty call chain before it reaches the dma map and hardware suggesting performance isn't important: nouveau_svm_fault(): args.i.m.method = NVIF_VMM_V0_PFNMAP nouveau_range_fault() nvif_object_ioctl() client->driver->ioctl() struct nvif_driver nvif_driver_nvkm: .ioctl = nvkm_client_ioctl nvkm_ioctl() nvkm_ioctl_path() nvkm_ioctl_v0[type].func(..) nvkm_ioctl_mthd() nvkm_object_mthd() struct nvkm_object_func nvkm_uvmm: .mthd = nvkm_uvmm_mthd nvkm_uvmm_mthd() nvkm_uvmm_mthd_pfnmap() nvkm_vmm_pfn_map() nvkm_vmm_ptes_get_map() func == gp100_vmm_pgt_pfn struct nvkm_vmm_desc_func gp100_vmm_desc_spt: .pfn = gp100_vmm_pgt_pfn nvkm_vmm_iter() REF_PTES == func == gp100_vmm_pgt_pfn() dma_map_page() Link: https://lore.kernel.org/r/5-v2-b4e84f444c7d+24f57-hmm_no_flags_jgg@mellanox.com Acked-by: Felix Kuehling <Felix.Kuehling@amd.com> Tested-by: Ralph Campbell <rcampbell@nvidia.com> Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Christoph Hellwig <hch@lst.de> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-02 02:20:48 +08:00
range->hmm_pfns +
((start - range->start) >> PAGE_SHIFT),
(end - start) >> PAGE_SHIFT, 0))
return -EFAULT;
hmm_pfns_fill(start, end, range, HMM_PFN_ERROR);
/* Skip this vma and continue processing the next vma. */
return 1;
}
static const struct mm_walk_ops hmm_walk_ops = {
.pud_entry = hmm_vma_walk_pud,
.pmd_entry = hmm_vma_walk_pmd,
.pte_hole = hmm_vma_walk_hole,
.hugetlb_entry = hmm_vma_walk_hugetlb_entry,
.test_walk = hmm_vma_walk_test,
};
/**
* hmm_range_fault - try to fault some address in a virtual address range
* @range: argument structure
*
* Returns 0 on success or one of the following error codes:
*
* -EINVAL: Invalid arguments or mm or virtual address is in an invalid vma
* (e.g., device file vma).
* -ENOMEM: Out of memory.
* -EPERM: Invalid permission (e.g., asking for write and range is read
* only).
* -EBUSY: The range has been invalidated and the caller needs to wait for
* the invalidation to finish.
* -EFAULT: A page was requested to be valid and could not be made valid
* ie it has no backing VMA or it is illegal to access
*
* This is similar to get_user_pages(), except that it can read the page tables
* without mutating them (ie causing faults).
*/
int hmm_range_fault(struct hmm_range *range)
{
struct hmm_vma_walk hmm_vma_walk = {
.range = range,
.last = range->start,
};
struct mm_struct *mm = range->notifier->mm;
int ret;
mmap_assert_locked(mm);
do {
/* If range is no longer valid force retry. */
if (mmu_interval_check_retry(range->notifier,
range->notifier_seq))
return -EBUSY;
ret = walk_page_range(mm, hmm_vma_walk.last, range->end,
&hmm_walk_ops, &hmm_vma_walk);
/*
* When -EBUSY is returned the loop restarts with
* hmm_vma_walk.last set to an address that has not been stored
* in pfns. All entries < last in the pfn array are set to their
* output, and all >= are still at their input values.
*/
} while (ret == -EBUSY);
return ret;
}
EXPORT_SYMBOL(hmm_range_fault);