korg/linux
0
0
Fork 0
mirror of https://mirrors.bfsu.edu.cn/git/linux.git synced 2024-11-15 00:04:15 +08:00
Code Issues Packages Projects Releases Wiki Activity
50660c5197
linux/drivers/infiniband/core/sa_query.c

2299 lines
61 KiB
C
Raw Normal View History

Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
/*
* Copyright (c) 2004 Topspin Communications. All rights reserved.
IB: Remove garbage non-ASCII characters from comments A few files had 0xa0 characters in comments. Remove them so that the files are clean ASCII text. Signed-off-by: Roland Dreier <rolandd@cisco.com>
2007-07-10 07:17:32 +08:00
* Copyright (c) 2005 Voltaire, Inc. All rights reserved.
IB/sa: Require SA registration Require users to register with SA module, to prevent the sa_query module text from going away while an SA query callback is still running. Update all in-tree users for the new interface. Signed-off-by: Michael S. Tsirkin <mst@mellanox.co.il> Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2006-08-22 07:40:12 +08:00
* Copyright (c) 2006 Intel Corporation. All rights reserved.
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
*
* This software is available to you under a choice of one of two
* licenses. You may choose to be licensed under the terms of the GNU
* General Public License (GPL) Version 2, available from the file
* COPYING in the main directory of this source tree, or the
* OpenIB.org BSD license below:
*
* Redistribution and use in source and binary forms, with or
* without modification, are permitted provided that the following
* conditions are met:
*
* - Redistributions of source code must retain the above
* copyright notice, this list of conditions and the following
* disclaimer.
*
* - Redistributions in binary form must reproduce the above
* copyright notice, this list of conditions and the following
* disclaimer in the documentation and/or other materials
* provided with the distribution.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
* EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
* MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
* NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
* BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
* ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
* CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
* SOFTWARE.
*/
#include <linux/init.h>
#include <linux/err.h>
#include <linux/random.h>
#include <linux/spinlock.h>
#include <linux/slab.h>
#include <linux/dma-mapping.h>
#include <linux/kref.h>
ib core: Convert query_idr to XArray Signed-off-by: Matthew Wilcox <willy@infradead.org> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2019-02-21 08:20:46 +08:00
#include <linux/xarray.h>
[PATCH] fix missing includes I recently picked up my older work to remove unnecessary #includes of sched.h, starting from a patch by Dave Jones to not include sched.h from module.h. This reduces the number of indirect includes of sched.h by ~300. Another ~400 pointless direct includes can be removed after this disentangling (patch to follow later). However, quite a few indirect includes need to be fixed up for this. In order to feed the patches through -mm with as little disturbance as possible, I've split out the fixes I accumulated up to now (complete for i386 and x86_64, more archs to follow later) and post them before the real patch. This way this large part of the patch is kept simple with only adding #includes, and all hunks are independent of each other. So if any hunk rejects or gets in the way of other patches, just drop it. My scripts will pick it up again in the next round. Signed-off-by: Tim Schmielau <tim@physik3.uni-rostock.de> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-10-31 07:03:48 +08:00
#include <linux/workqueue.h>
IB/core: Ethernet L2 attributes in verbs/cm structures This patch add the support for Ethernet L2 attributes in the verbs/cm/cma structures. When dealing with L2 Ethernet, we should use smac, dmac, vlan ID and priority in a similar manner that the IB L2 (and the L4 PKEY) attributes are used. Thus, those attributes were added to the following structures: * ib_ah_attr - added dmac * ib_qp_attr - added smac and vlan_id, (sl remains vlan priority) * ib_wc - added smac, vlan_id * ib_sa_path_rec - added smac, dmac, vlan_id * cm_av - added smac and vlan_id For the path record structure, extra care was taken to avoid the new fields when packing it into wire format, so we don't break the IB CM and SA wire protocol. On the active side, the CM fills. its internal structures from the path provided by the ULP. We add there taking the ETH L2 attributes and placing them into the CM Address Handle (struct cm_av). On the passive side, the CM fills its internal structures from the WC associated with the REQ message. We add there taking the ETH L2 attributes from the WC. When the HW driver provides the required ETH L2 attributes in the WC, they set the IB_WC_WITH_SMAC and IB_WC_WITH_VLAN flags. The IB core code checks for the presence of these flags, and in their absence does address resolution from the ib_init_ah_from_wc() helper function. ib_modify_qp_is_ok is also updated to consider the link layer. Some parameters are mandatory for Ethernet link layer, while they are irrelevant for IB. Vendor drivers are modified to support the new function signature. Signed-off-by: Matan Barak <matanb@mellanox.com> Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com> Signed-off-by: Roland Dreier <roland@purestorage.com>
2013-12-13 00:03:11 +08:00
#include <uapi/linux/if_ether.h>
[PATCH] IB: move include files to include/rdma Move the InfiniBand headers from drivers/infiniband/include to include/rdma. This allows InfiniBand-using code to live elsewhere, and lets us remove the ugly EXTRA_CFLAGS include path from the InfiniBand Makefiles. Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-08-26 04:40:04 +08:00
#include <rdma/ib_pack.h>
IB/sa: Add ib_init_ah_from_path() Add a call to initialize address handle attributes given a path record. This is used by the CM, and would be useful for users of UD QPs. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2006-06-18 11:37:39 +08:00
#include <rdma/ib_cache.h>
IB/sa: Route SA pathrecord query through netlink This patch routes a SA pathrecord query to netlink first and processes the response appropriately. If a failure is returned, the request will be sent through IB. The decision whether to route the request to netlink first is determined by the presence of a listener for the local service netlink multicast group. If the user-space local service netlink multicast group listener is not present, the request will be sent through IB, just like what is currently being done. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: John Fleck <john.fleck@intel.com> Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-14 20:52:09 +08:00
#include <rdma/rdma_netlink.h>
#include <net/netlink.h>
#include <uapi/rdma/ib_user_sa.h>
#include <rdma/ib_marshall.h>
IB/core: Validate route when we init ah In order to make sure API users don't try to use SGIDs which don't conform to the routing table, validate the route before searching the RoCE GID table. Signed-off-by: Matan Barak <matanb@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-12-23 20:56:53 +08:00
#include <rdma/ib_addr.h>
IB/hfi1: Determine 9B/16B L2 header type based on Address handle When address handle attributes are initialized, the LIDs are transformed to be in the 32 bit LID space. When constructing the header, hfi1 driver will look at the LID to determine the packet header to be created. Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Don Hiatt <don.hiatt@intel.com> Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-08-05 04:54:16 +08:00
#include <rdma/opa_addr.h>
RDMA/cma: Multiple path records support with netlink channel Support receiving inbound and outbound IB path records (along with GMP PathRecord) from user-space service through the RDMA netlink channel. The LIDs in these 3 PRs can be used in this way: 1. GMP PR: used as the standard local/remote LIDs; 2. DLID of outbound PR: Used as the "dlid" field for outbound traffic; 3. DLID of inbound PR: Used as the "dlid" field for outbound traffic in responder side. This is aimed to support adaptive routing. With current IB routing solution when a packet goes out it's assigned with a fixed DLID per target, meaning a fixed router will be used. The LIDs in inbound/outbound path records can be used to identify group of routers that allow communication with another subnet's entity. With them packets from an inter-subnet connection may travel through any router in the set to reach the target. As confirmed with Jason, when sending a netlink request, kernel uses LS_RESOLVE_PATH_USE_ALL so that the service knows kernel supports multiple PRs. Signed-off-by: Mark Zhang <markzhang@nvidia.com> Reviewed-by: Mark Bloch <mbloch@nvidia.com> Link: https://lore.kernel.org/r/2fa2b6c93c4c16c8915bac3cfc4f27be1d60519d.1662631201.git.leonro@nvidia.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2022-09-08 18:09:01 +08:00
#include <rdma/rdma_cm.h>
IB/sa: Track multicast join/leave requests The IB SA tracks multicast join/leave requests on a per port basis and does not do any reference counting: if two users of the same port join the same group, and one leaves that group, then the SA will remove the port from the group even though there is one user who wants to stay a member left. Therefore, in order to support multiple users of the same multicast group from the same port, we need to perform reference counting locally. To do this, add an multicast submodule to ib_sa to perform reference counting of multicast join/leave operations. Modify ib_ipoib (the only in-kernel user of multicast) to use the new interface. Signed-off-by: Roland Dreier <rolandd@cisco.com>
2007-02-16 09:00:17 +08:00
#include "sa.h"
IB/core: Validate route when we init ah In order to make sure API users don't try to use SGIDs which don't conform to the routing table, validate the route before searching the RoCE GID table. Signed-off-by: Matan Barak <matanb@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-12-23 20:56:53 +08:00
#include "core_priv.h"
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
IB/sa: Route SA pathrecord query through netlink This patch routes a SA pathrecord query to netlink first and processes the response appropriately. If a failure is returned, the request will be sent through IB. The decision whether to route the request to netlink first is determined by the presence of a listener for the local service netlink multicast group. If the user-space local service netlink multicast group listener is not present, the request will be sent through IB, just like what is currently being done. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: John Fleck <john.fleck@intel.com> Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-14 20:52:09 +08:00
#define IB_SA_LOCAL_SVC_TIMEOUT_MIN 100
#define IB_SA_LOCAL_SVC_TIMEOUT_DEFAULT 2000
#define IB_SA_LOCAL_SVC_TIMEOUT_MAX 200000
IB/SA: Modify SA to implicitly cache Class Port info SA will query and cache class port info as part of its initialization. SA will also invalidate and refresh the cache based on specific events. Callers such as IPoIB and CM can query the SA to get the classportinfo information. Apart from making the caller code much simpler, this change puts the onus on the SA to query and maintain classportinfo much like how it maitains the address handle to the SM. Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Don Hiatt <don.hiatt@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-03-21 07:38:08 +08:00
#define IB_SA_CPI_MAX_RETRY_CNT 3
#define IB_SA_CPI_RETRY_WAIT 1000 /*msecs */
IB/sa: Route SA pathrecord query through netlink This patch routes a SA pathrecord query to netlink first and processes the response appropriately. If a failure is returned, the request will be sent through IB. The decision whether to route the request to netlink first is determined by the presence of a listener for the local service netlink multicast group. If the user-space local service netlink multicast group listener is not present, the request will be sent through IB, just like what is currently being done. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: John Fleck <john.fleck@intel.com> Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-14 20:52:09 +08:00
static int sa_local_svc_timeout_ms = IB_SA_LOCAL_SVC_TIMEOUT_DEFAULT;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
struct ib_sa_sm_ah {
struct ib_ah *ah;
struct kref ref;
IB/sa: Make sure SA queries use default P_Key MADs sent to the SA should use the the default P_Key (0x7fff/0xffff). There's no requirement that the default P_Key is stored at index 0 in the local P_Key table, so add code to the sa_query module to look up the index of the default P_Key when creating an address handle for the SA (which is done any time the P_Key table might change), and use this index for all SA queries. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2007-06-19 02:03:58 +08:00
u16 pkey_index;
IB/sa: Set src_path_bits correctly in ib_init_ah_from_path() src_path_bits needs to mask off the base LID value. Signed-off-by: Sean Hefty <sean.hefty@intel.com>
2007-04-06 01:51:10 +08:00
u8 src_path_mask;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
};
IB/SA: Add support to query opa classport info. For OPA devices, SA will query the OPA classport info instead of the IB defined classport info. opa classport info exposes additional information and capabilities that are specific to OPA devices. Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-29 07:20:26 +08:00
enum rdma_class_port_info_type {
RDMA_CLASS_PORT_INFO_IB,
RDMA_CLASS_PORT_INFO_OPA
};
struct rdma_class_port_info {
enum rdma_class_port_info_type type;
union {
struct ib_class_port_info ib;
struct opa_class_port_info opa;
};
};
IB/sa: Add cached attribute containing SM information to SA port Added a new SA port attribute containing SM ClassPortInfo fields, (ClassPortInfo fields: Table 126 IB Spec 1.3.). This is useful for checking SM support for specific features. The attribute is cached to avoid resending queries, caching is done when a successful ClassPortInfo reply is received on the port. Invalidation of the attribute is done on SM change events, SM re-registration events, and SM LID change events. The fields in ClassPortInfo should not change during SM runtime without an event. Signed-off-by: Alex Vesker <valex@mellanox.com> Reviewed by: Hal Rosenstock <hal@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-07-06 21:36:34 +08:00
struct ib_sa_classport_cache {
bool valid;
IB/SA: Modify SA to implicitly cache Class Port info SA will query and cache class port info as part of its initialization. SA will also invalidate and refresh the cache based on specific events. Callers such as IPoIB and CM can query the SA to get the classportinfo information. Apart from making the caller code much simpler, this change puts the onus on the SA to query and maintain classportinfo much like how it maitains the address handle to the SM. Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Don Hiatt <don.hiatt@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-03-21 07:38:08 +08:00
int retry_cnt;
IB/SA: Add support to query opa classport info. For OPA devices, SA will query the OPA classport info instead of the IB defined classport info. opa classport info exposes additional information and capabilities that are specific to OPA devices. Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-29 07:20:26 +08:00
struct rdma_class_port_info data;
IB/sa: Add cached attribute containing SM information to SA port Added a new SA port attribute containing SM ClassPortInfo fields, (ClassPortInfo fields: Table 126 IB Spec 1.3.). This is useful for checking SM support for specific features. The attribute is cached to avoid resending queries, caching is done when a successful ClassPortInfo reply is received on the port. Invalidation of the attribute is done on SM change events, SM re-registration events, and SM LID change events. The fields in ClassPortInfo should not change during SM runtime without an event. Signed-off-by: Alex Vesker <valex@mellanox.com> Reviewed by: Hal Rosenstock <hal@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-07-06 21:36:34 +08:00
};
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
struct ib_sa_port {
struct ib_mad_agent *agent;
struct ib_sa_sm_ah *sm_ah;
struct work_struct update_task;
IB/sa: Add cached attribute containing SM information to SA port Added a new SA port attribute containing SM ClassPortInfo fields, (ClassPortInfo fields: Table 126 IB Spec 1.3.). This is useful for checking SM support for specific features. The attribute is cached to avoid resending queries, caching is done when a successful ClassPortInfo reply is received on the port. Invalidation of the attribute is done on SM change events, SM re-registration events, and SM LID change events. The fields in ClassPortInfo should not change during SM runtime without an event. Signed-off-by: Alex Vesker <valex@mellanox.com> Reviewed by: Hal Rosenstock <hal@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-07-06 21:36:34 +08:00
struct ib_sa_classport_cache classport_info;
IB/SA: Modify SA to implicitly cache Class Port info SA will query and cache class port info as part of its initialization. SA will also invalidate and refresh the cache based on specific events. Callers such as IPoIB and CM can query the SA to get the classportinfo information. Apart from making the caller code much simpler, this change puts the onus on the SA to query and maintain classportinfo much like how it maitains the address handle to the SM. Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Don Hiatt <don.hiatt@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-03-21 07:38:08 +08:00
struct delayed_work ib_cpi_work;
IB/sa: Add cached attribute containing SM information to SA port Added a new SA port attribute containing SM ClassPortInfo fields, (ClassPortInfo fields: Table 126 IB Spec 1.3.). This is useful for checking SM support for specific features. The attribute is cached to avoid resending queries, caching is done when a successful ClassPortInfo reply is received on the port. Invalidation of the attribute is done on SM change events, SM re-registration events, and SM LID change events. The fields in ClassPortInfo should not change during SM runtime without an event. Signed-off-by: Alex Vesker <valex@mellanox.com> Reviewed by: Hal Rosenstock <hal@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-07-06 21:36:34 +08:00
spinlock_t classport_lock; /* protects class port info set */
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
spinlock_t ah_lock;
RDMA: Support more than 255 rdma ports Current code uses many different types when dealing with a port of a RDMA device: u8, unsigned int and u32. Switch to u32 to clean up the logic. This allows us to make (at least) the core view consistent and use the same type. Unfortunately not all places can be converted. Many uverbs functions expect port to be u8 so keep those places in order not to break UAPIs. HW/Spec defined values must also not be changed. With the switch to u32 we now can support devices with more than 255 ports. U32_MAX is reserved to make control logic a bit easier to deal with. As a device with U32_MAX ports probably isn't going to happen any time soon this seems like a non issue. When a device with more than 255 ports is created uverbs will report the RDMA device as having 255 ports as this is the max currently supported. The verbs interface is not changed yet because the IBTA spec limits the port size in too many places to be u8 and all applications that relies in verbs won't be able to cope with this change. At this stage, we are extending the interfaces that are using vendor channel solely Once the limitation is lifted mlx5 in switchdev mode will be able to have thousands of SFs created by the device. As the only instance of an RDMA device that reports more than 255 ports will be a representor device and it exposes itself as a RAW Ethernet only device CM/MAD/IPoIB and other ULPs aren't effected by this change and their sysfs/interfaces that are exposes to userspace can remain unchanged. While here cleanup some alignment issues and remove unneeded sanity checks (mainly in rdmavt), Link: https://lore.kernel.org/r/20210301070420.439400-1-leon@kernel.org Signed-off-by: Mark Bloch <mbloch@nvidia.com> Signed-off-by: Leon Romanovsky <leonro@nvidia.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
2021-03-01 15:04:20 +08:00
u32 port_num;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
};
struct ib_sa_device {
int start_port, end_port;
struct ib_event_handler event_handler;
RDMA: Replace zero-length array with flexible-array member The current codebase makes use of the zero-length array language extension to the C90 standard, but the preferred mechanism to declare variable-length types such as these ones is a flexible array member[1][2], introduced in C99: struct foo { int stuff; struct boo array[]; }; By making use of the mechanism above, we will get a compiler warning in case the flexible array does not occur last in the structure, which will help us prevent some kind of undefined behavior bugs from being inadvertently introduced[3] to the codebase from now on. Also, notice that, dynamic memory allocations won't be affected by this change: "Flexible array members have incomplete type, and so the sizeof operator may not be applied. As a quirk of the original implementation of zero-length arrays, sizeof evaluates to zero."[1] This issue was found with the help of Coccinelle. [1] https://gcc.gnu.org/onlinedocs/gcc/Zero-Length.html [2] https://github.com/KSPP/linux/issues/21 [3] commit 76497732932f ("cxgb3/l2t: Fix undefined behaviour") Link: https://lore.kernel.org/r/20200213010425.GA13068@embeddedor.com Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com> # added a few more
2020-02-13 09:04:25 +08:00
struct ib_sa_port port[];
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
};
struct ib_sa_query {
RDMA/cma: Multiple path records support with netlink channel Support receiving inbound and outbound IB path records (along with GMP PathRecord) from user-space service through the RDMA netlink channel. The LIDs in these 3 PRs can be used in this way: 1. GMP PR: used as the standard local/remote LIDs; 2. DLID of outbound PR: Used as the "dlid" field for outbound traffic; 3. DLID of inbound PR: Used as the "dlid" field for outbound traffic in responder side. This is aimed to support adaptive routing. With current IB routing solution when a packet goes out it's assigned with a fixed DLID per target, meaning a fixed router will be used. The LIDs in inbound/outbound path records can be used to identify group of routers that allow communication with another subnet's entity. With them packets from an inter-subnet connection may travel through any router in the set to reach the target. As confirmed with Jason, when sending a netlink request, kernel uses LS_RESOLVE_PATH_USE_ALL so that the service knows kernel supports multiple PRs. Signed-off-by: Mark Zhang <markzhang@nvidia.com> Reviewed-by: Mark Bloch <mbloch@nvidia.com> Link: https://lore.kernel.org/r/2fa2b6c93c4c16c8915bac3cfc4f27be1d60519d.1662631201.git.leonro@nvidia.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2022-09-08 18:09:01 +08:00
void (*callback)(struct ib_sa_query *sa_query, int status,
RDMA/cma: Refactor the inbound/outbound path records process flow Refactors based on comments [1] of the multiple path records support patchset: - Return failure if not able to set inbound/outbound PRs; - Simplify the flow when receiving the PRs from netlink channel: When a good PR response is received, unpack it and call the path_query callback directly. This saves two memory allocations; - Define RDMA_PRIMARY_PATH_MAX_REC_NUM in a proper place. [1] https://lore.kernel.org/linux-rdma/Yyxp9E9pJtUids2o@nvidia.com/ Signed-off-by: Mark Zhang <markzhang@nvidia.com> Reviewed-by: Bart Van Assche <bvanassche@acm.org> #srp Link: https://lore.kernel.org/r/7610025d57342b8b6da0f19516c9612f9c3fdc37.1672819376.git.leonro@nvidia.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2023-01-04 16:03:41 +08:00
struct ib_sa_mad *mad);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
void (*release)(struct ib_sa_query *);
IB/sa: Require SA registration Require users to register with SA module, to prevent the sa_query module text from going away while an SA query callback is still running. Update all in-tree users for the new interface. Signed-off-by: Michael S. Tsirkin <mst@mellanox.co.il> Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2006-08-22 07:40:12 +08:00
struct ib_sa_client *client;
[IB] Fix MAD layer DMA mappings to avoid touching data buffer once mapped The MAD layer was violating the DMA API by touching data buffers used for sends after the DMA mapping was done. This causes problems on non-cache-coherent architectures, because the device doing DMA won't see updates to the payload buffers that exist only in the CPU cache. Fix this by having all MAD consumers use ib_create_send_mad() to allocate their send buffers, and moving the DMA mapping into the MAD layer so it can be done just before calling send (and after any modifications of the send buffer by the MAD layer). Tested on a non-cache-coherent PowerPC 440SPe system. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-10-26 01:51:39 +08:00
struct ib_sa_port *port;
struct ib_mad_send_buf *mad_buf;
struct ib_sa_sm_ah *sm_ah;
int id;
IB/sa: Route SA pathrecord query through netlink This patch routes a SA pathrecord query to netlink first and processes the response appropriately. If a failure is returned, the request will be sent through IB. The decision whether to route the request to netlink first is determined by the presence of a listener for the local service netlink multicast group. If the user-space local service netlink multicast group listener is not present, the request will be sent through IB, just like what is currently being done. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: John Fleck <john.fleck@intel.com> Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-14 20:52:09 +08:00
u32 flags;
struct list_head list; /* Local svc request list */
u32 seq; /* Local svc request sequence number */
unsigned long timeout; /* Local svc timeout */
u8 path_use; /* How will the pathrecord be used */
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
};
IB/sa: Route SA pathrecord query through netlink This patch routes a SA pathrecord query to netlink first and processes the response appropriately. If a failure is returned, the request will be sent through IB. The decision whether to route the request to netlink first is determined by the presence of a listener for the local service netlink multicast group. If the user-space local service netlink multicast group listener is not present, the request will be sent through IB, just like what is currently being done. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: John Fleck <john.fleck@intel.com> Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-14 20:52:09 +08:00
#define IB_SA_ENABLE_LOCAL_SERVICE 0x00000001
#define IB_SA_CANCEL 0x00000002
IB/SA: Add support to query opa classport info. For OPA devices, SA will query the OPA classport info instead of the IB defined classport info. opa classport info exposes additional information and capabilities that are specific to OPA devices. Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-29 07:20:26 +08:00
#define IB_SA_QUERY_OPA 0x00000004
IB/sa: Route SA pathrecord query through netlink This patch routes a SA pathrecord query to netlink first and processes the response appropriately. If a failure is returned, the request will be sent through IB. The decision whether to route the request to netlink first is determined by the presence of a listener for the local service netlink multicast group. If the user-space local service netlink multicast group listener is not present, the request will be sent through IB, just like what is currently being done. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: John Fleck <john.fleck@intel.com> Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-14 20:52:09 +08:00
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
struct ib_sa_path_query {
RDMA/cma: Multiple path records support with netlink channel Support receiving inbound and outbound IB path records (along with GMP PathRecord) from user-space service through the RDMA netlink channel. The LIDs in these 3 PRs can be used in this way: 1. GMP PR: used as the standard local/remote LIDs; 2. DLID of outbound PR: Used as the "dlid" field for outbound traffic; 3. DLID of inbound PR: Used as the "dlid" field for outbound traffic in responder side. This is aimed to support adaptive routing. With current IB routing solution when a packet goes out it's assigned with a fixed DLID per target, meaning a fixed router will be used. The LIDs in inbound/outbound path records can be used to identify group of routers that allow communication with another subnet's entity. With them packets from an inter-subnet connection may travel through any router in the set to reach the target. As confirmed with Jason, when sending a netlink request, kernel uses LS_RESOLVE_PATH_USE_ALL so that the service knows kernel supports multiple PRs. Signed-off-by: Mark Zhang <markzhang@nvidia.com> Reviewed-by: Mark Bloch <mbloch@nvidia.com> Link: https://lore.kernel.org/r/2fa2b6c93c4c16c8915bac3cfc4f27be1d60519d.1662631201.git.leonro@nvidia.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2022-09-08 18:09:01 +08:00
void (*callback)(int status, struct sa_path_rec *rec,
RDMA/cma: Refactor the inbound/outbound path records process flow Refactors based on comments [1] of the multiple path records support patchset: - Return failure if not able to set inbound/outbound PRs; - Simplify the flow when receiving the PRs from netlink channel: When a good PR response is received, unpack it and call the path_query callback directly. This saves two memory allocations; - Define RDMA_PRIMARY_PATH_MAX_REC_NUM in a proper place. [1] https://lore.kernel.org/linux-rdma/Yyxp9E9pJtUids2o@nvidia.com/ Signed-off-by: Mark Zhang <markzhang@nvidia.com> Reviewed-by: Bart Van Assche <bvanassche@acm.org> #srp Link: https://lore.kernel.org/r/7610025d57342b8b6da0f19516c9612f9c3fdc37.1672819376.git.leonro@nvidia.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2023-01-04 16:03:41 +08:00
unsigned int num_paths, void *context);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
void *context;
struct ib_sa_query sa_query;
IB/SA: Add support to query OPA path records When the bit 26 of capmask2 field in OPA classport info query is set, SA will query for OPA path records instead of querying for IB path records. Note that OPA path records can only be queried by kernel ULPs. Userspace clients continue to query IB path records. Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-28 07:06:02 +08:00
struct sa_path_rec *conv_pr;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
};
IB/sa: Add GuidInfoRecord query support This query is needed for SRIOV alias GUID support. The query is implemented per the IB Spec definition in section 15.2.5.18 (GuidInfoRecord). Signed-off-by: Erez Shitrit <erezsh@mellanox.co.il> Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il> Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com> Signed-off-by: Roland Dreier <roland@purestorage.com>
2012-06-19 16:21:38 +08:00
struct ib_sa_guidinfo_query {
void (*callback)(int, struct ib_sa_guidinfo_rec *, void *);
void *context;
struct ib_sa_query sa_query;
};
IB/SA Agent: Add support for SA agent get ClassPortInfo New SA query function to return the ClassPortInfo struct from the SA. If the SM supports FullMemberSendOnly mode for MCG's, it sets a capability bit in the capability_mask2 field of the response. Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-05-26 03:02:05 +08:00
struct ib_sa_classport_info_query {
IB/SA: Modify SA to implicitly cache Class Port info SA will query and cache class port info as part of its initialization. SA will also invalidate and refresh the cache based on specific events. Callers such as IPoIB and CM can query the SA to get the classportinfo information. Apart from making the caller code much simpler, this change puts the onus on the SA to query and maintain classportinfo much like how it maitains the address handle to the SM. Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Don Hiatt <don.hiatt@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-03-21 07:38:08 +08:00
void (*callback)(void *);
IB/SA Agent: Add support for SA agent get ClassPortInfo New SA query function to return the ClassPortInfo struct from the SA. If the SM supports FullMemberSendOnly mode for MCG's, it sets a capability bit in the capability_mask2 field of the response. Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-05-26 03:02:05 +08:00
void *context;
struct ib_sa_query sa_query;
};
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
struct ib_sa_mcmember_query {
void (*callback)(int, struct ib_sa_mcmember_rec *, void *);
void *context;
struct ib_sa_query sa_query;
};
IB/sa: Route SA pathrecord query through netlink This patch routes a SA pathrecord query to netlink first and processes the response appropriately. If a failure is returned, the request will be sent through IB. The decision whether to route the request to netlink first is determined by the presence of a listener for the local service netlink multicast group. If the user-space local service netlink multicast group listener is not present, the request will be sent through IB, just like what is currently being done. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: John Fleck <john.fleck@intel.com> Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-14 20:52:09 +08:00
static LIST_HEAD(ib_nl_request_list);
static DEFINE_SPINLOCK(ib_nl_request_lock);
static atomic_t ib_nl_sa_request_seq;
static struct workqueue_struct *ib_nl_wq;
static struct delayed_work ib_nl_timed_work;
static const struct nla_policy ib_nl_policy[LS_NLA_TYPE_MAX] = {
[LS_NLA_TYPE_PATH_RECORD] = {.type = NLA_BINARY,
.len = sizeof(struct ib_path_rec_data)},
[LS_NLA_TYPE_TIMEOUT] = {.type = NLA_U32},
[LS_NLA_TYPE_SERVICE_ID] = {.type = NLA_U64},
[LS_NLA_TYPE_DGID] = {.type = NLA_BINARY,
.len = sizeof(struct rdma_nla_ls_gid)},
[LS_NLA_TYPE_SGID] = {.type = NLA_BINARY,
.len = sizeof(struct rdma_nla_ls_gid)},
[LS_NLA_TYPE_TCLASS] = {.type = NLA_U8},
[LS_NLA_TYPE_PKEY] = {.type = NLA_U16},
[LS_NLA_TYPE_QOS_CLASS] = {.type = NLA_U16},
};
RDMA: Allow ib_client's to fail when add() is called When a client is added it isn't allowed to fail, but all the client's have various failure paths within their add routines. This creates the very fringe condition where the client was added, failed during add and didn't set the client_data. The core code will then still call other client_data centric ops like remove(), rename(), get_nl_info(), and get_net_dev_by_params() with NULL client_data - which is confusing and unexpected. If the add() callback fails, then do not call any more client ops for the device, even remove. Remove all the now redundant checks for NULL client_data in ops callbacks. Update all the add() callbacks to return error codes appropriately. EOPNOTSUPP is used for cases where the ULP does not support the ib_device - eg because it only works with IB. Link: https://lore.kernel.org/r/20200421172440.387069-1-leon@kernel.org Signed-off-by: Leon Romanovsky <leonro@mellanox.com> Acked-by: Ursula Braun <ubraun@linux.ibm.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-04-22 01:24:40 +08:00
static int ib_sa_add_one(struct ib_device *device);
IB/core: lock client data with lists_rwsem An ib_client callback that is called with the lists_rwsem locked only for read is protected from changes to the IB client lists, but not from ib_unregister_device() freeing its client data. This is because ib_unregister_device() will remove the device from the device list with lists_rwsem locked for write, but perform the rest of the cleanup, including the call to remove() without that lock. Mark client data that is undergoing de-registration with a new going_down flag in the client data context. Lock the client data list with lists_rwsem for write in addition to using the spinlock, so that functions calling the callback would be able to lock only lists_rwsem for read and let callbacks sleep. Since ib_unregister_client() now marks the client data context, no need for remove() to search the context again, so pass the client data directly to remove() callbacks. Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Signed-off-by: Haggai Eran <haggaie@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-07-30 22:50:14 +08:00
static void ib_sa_remove_one(struct ib_device *device, void *client_data);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
static struct ib_client sa_client = {
.name = "sa",
.add = ib_sa_add_one,
.remove = ib_sa_remove_one
};
ib core: Convert query_idr to XArray Signed-off-by: Matthew Wilcox <willy@infradead.org> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2019-02-21 08:20:46 +08:00
static DEFINE_XARRAY_FLAGS(queries, XA_FLAGS_ALLOC | XA_FLAGS_LOCK_IRQ);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
IB: Use DEFINE_SPINLOCK() for static spinlocks Rather than just defining static spinlock_t variables and then initializing them later in init functions, simply define them with DEFINE_SPINLOCK() and remove the calls to spin_lock_init(). This cleans up the source a tad and also shrinks the compiled code; eg on x86-64: add/remove: 0/0 grow/shrink: 0/3 up/down: 0/-40 (-40) function old new delta ib_uverbs_init 336 326 -10 ib_mad_init_module 147 137 -10 ib_sa_init 123 103 -20 Signed-off-by: Roland Dreier <rolandd@cisco.com>
2009-09-06 11:24:23 +08:00
static DEFINE_SPINLOCK(tid_lock);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
static u32 tid;
#define PATH_REC_FIELD(field) \
IB/SA: Rename ib_sa_path_rec to sa_path_rec Rename ib_sa_path_rec to a more generic sa_path_rec. This is part of extending ib_sa to also support OPA path records in addition to the IB defined path records. Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-28 07:05:58 +08:00
.struct_offset_bytes = offsetof(struct sa_path_rec, field), \
RDMA/core: Use sizeof_field() helper Make use of the sizeof_field() helper instead of an open-coded version. Link: https://lore.kernel.org/r/20200527144152.GA22605@embeddedor Signed-off-by: Gustavo A. R. Silva <gustavoars@kernel.org> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-27 22:41:52 +08:00
.struct_size_bytes = sizeof_field(struct sa_path_rec, field), \
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
.field_name = "sa_path_rec:" #field
static const struct ib_field path_rec_table[] = {
RDMA/SA: Fix kernel panic in CMA request handler flow Commit 9fdca4da4d8c (IB/SA: Split struct sa_path_rec based on IB and ROCE specific fields) moved the service_id to be specific attribute for IB and OPA SA Path Record, and thus wasn't assigned for RoCE. This caused to the following kernel panic in the CMA request handler flow: [ 27.074594] BUG: unable to handle kernel NULL pointer dereference at 0000000000000008 [ 27.074731] IP: __radix_tree_lookup+0x1d/0xe0 ... [ 27.075356] Workqueue: ib_cm cm_work_handler [ib_cm] [ 27.075401] task: ffff88022e3b8000 task.stack: ffffc90001298000 [ 27.075449] RIP: 0010:__radix_tree_lookup+0x1d/0xe0 ... [ 27.075979] Call Trace: [ 27.076015] radix_tree_lookup+0xd/0x10 [ 27.076055] cma_ps_find+0x59/0x70 [rdma_cm] [ 27.076097] cma_id_from_event+0xd2/0x470 [rdma_cm] [ 27.076144] ? ib_init_ah_from_path+0x39a/0x590 [ib_core] [ 27.076193] cma_req_handler+0x25/0x480 [rdma_cm] [ 27.076237] cm_process_work+0x25/0x120 [ib_cm] [ 27.076280] ? cm_get_bth_pkey.isra.62+0x3c/0xa0 [ib_cm] [ 27.076350] cm_req_handler+0xb03/0xd40 [ib_cm] [ 27.076430] ? sched_clock_cpu+0x11/0xb0 [ 27.076478] cm_work_handler+0x194/0x1588 [ib_cm] [ 27.076525] process_one_work+0x160/0x410 [ 27.076565] worker_thread+0x137/0x4a0 [ 27.076614] kthread+0x112/0x150 [ 27.076684] ? max_active_store+0x60/0x60 [ 27.077642] ? kthread_park+0x90/0x90 [ 27.078530] ret_from_fork+0x2c/0x40 This patch moves it back to the common SA Path Record structure and removes the redundant setter and getter. Tested on Connect-IB and Connect-X4 in Infiniband and RoCE respectively. Fixes: 9fdca4da4d8c (IB/SA: Split struct sa_path_rec based on IB ands ROCE specific fields) Signed-off-by: Majd Dibbiny <majd@mellanox.com> Reviewed-by: Parav Pandit <parav@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-05-22 00:09:54 +08:00
{ PATH_REC_FIELD(service_id),
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
.offset_words = 0,
.offset_bits = 0,
IB/sa: Add new QoS fields to path record The QoS annex defines new fields for path records. Add them to the ib_sa for consumers that want to use them. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Reviewed-by: Or Gerlitz <ogerlitz@voltaire.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2007-08-09 06:41:28 +08:00
.size_bits = 64 },
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
{ PATH_REC_FIELD(dgid),
.offset_words = 2,
.offset_bits = 0,
.size_bits = 128 },
{ PATH_REC_FIELD(sgid),
.offset_words = 6,
.offset_bits = 0,
.size_bits = 128 },
IB/SA: Split struct sa_path_rec based on IB and ROCE specific fields sa_path_rec now contains a union of sa_path_rec_ib and sa_path_rec_roce based on the type of the path record. Note that fields applicable to path record type ROCE v1 and ROCE v2 fall under sa_path_rec_roce. Accessor functions are added to these fields so the caller doesn't have to know the type. Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-28 07:06:00 +08:00
{ PATH_REC_FIELD(ib.dlid),
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
.offset_words = 10,
.offset_bits = 0,
.size_bits = 16 },
IB/SA: Split struct sa_path_rec based on IB and ROCE specific fields sa_path_rec now contains a union of sa_path_rec_ib and sa_path_rec_roce based on the type of the path record. Note that fields applicable to path record type ROCE v1 and ROCE v2 fall under sa_path_rec_roce. Accessor functions are added to these fields so the caller doesn't have to know the type. Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-28 07:06:00 +08:00
{ PATH_REC_FIELD(ib.slid),
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
.offset_words = 10,
.offset_bits = 16,
.size_bits = 16 },
IB/SA: Split struct sa_path_rec based on IB and ROCE specific fields sa_path_rec now contains a union of sa_path_rec_ib and sa_path_rec_roce based on the type of the path record. Note that fields applicable to path record type ROCE v1 and ROCE v2 fall under sa_path_rec_roce. Accessor functions are added to these fields so the caller doesn't have to know the type. Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-28 07:06:00 +08:00
{ PATH_REC_FIELD(ib.raw_traffic),
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
.offset_words = 11,
.offset_bits = 0,
.size_bits = 1 },
{ RESERVED,
.offset_words = 11,
.offset_bits = 1,
.size_bits = 3 },
{ PATH_REC_FIELD(flow_label),
.offset_words = 11,
.offset_bits = 4,
.size_bits = 20 },
{ PATH_REC_FIELD(hop_limit),
.offset_words = 11,
.offset_bits = 24,
.size_bits = 8 },
{ PATH_REC_FIELD(traffic_class),
.offset_words = 12,
.offset_bits = 0,
.size_bits = 8 },
{ PATH_REC_FIELD(reversible),
.offset_words = 12,
.offset_bits = 8,
.size_bits = 1 },
{ PATH_REC_FIELD(numb_path),
.offset_words = 12,
.offset_bits = 9,
.size_bits = 7 },
{ PATH_REC_FIELD(pkey),
.offset_words = 12,
.offset_bits = 16,
.size_bits = 16 },
IB/sa: Add new QoS fields to path record The QoS annex defines new fields for path records. Add them to the ib_sa for consumers that want to use them. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Reviewed-by: Or Gerlitz <ogerlitz@voltaire.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2007-08-09 06:41:28 +08:00
{ PATH_REC_FIELD(qos_class),
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
.offset_words = 13,
.offset_bits = 0,
.size_bits = 12 },
{ PATH_REC_FIELD(sl),
.offset_words = 13,
.offset_bits = 12,
.size_bits = 4 },
{ PATH_REC_FIELD(mtu_selector),
.offset_words = 13,
.offset_bits = 16,
.size_bits = 2 },
{ PATH_REC_FIELD(mtu),
.offset_words = 13,
.offset_bits = 18,
.size_bits = 6 },
{ PATH_REC_FIELD(rate_selector),
.offset_words = 13,
.offset_bits = 24,
.size_bits = 2 },
{ PATH_REC_FIELD(rate),
.offset_words = 13,
.offset_bits = 26,
.size_bits = 6 },
{ PATH_REC_FIELD(packet_life_time_selector),
.offset_words = 14,
.offset_bits = 0,
.size_bits = 2 },
{ PATH_REC_FIELD(packet_life_time),
.offset_words = 14,
.offset_bits = 2,
.size_bits = 6 },
{ PATH_REC_FIELD(preference),
.offset_words = 14,
.offset_bits = 8,
.size_bits = 8 },
{ RESERVED,
.offset_words = 14,
.offset_bits = 16,
.size_bits = 48 },
};
IB/SA: Add support to query OPA path records When the bit 26 of capmask2 field in OPA classport info query is set, SA will query for OPA path records instead of querying for IB path records. Note that OPA path records can only be queried by kernel ULPs. Userspace clients continue to query IB path records. Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-28 07:06:02 +08:00
#define OPA_PATH_REC_FIELD(field) \
.struct_offset_bytes = \
offsetof(struct sa_path_rec, field), \
.struct_size_bytes = \
RDMA/core: Use sizeof_field() helper Make use of the sizeof_field() helper instead of an open-coded version. Link: https://lore.kernel.org/r/20200527144152.GA22605@embeddedor Signed-off-by: Gustavo A. R. Silva <gustavoars@kernel.org> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-27 22:41:52 +08:00
sizeof_field(struct sa_path_rec, field), \
IB/SA: Add support to query OPA path records When the bit 26 of capmask2 field in OPA classport info query is set, SA will query for OPA path records instead of querying for IB path records. Note that OPA path records can only be queried by kernel ULPs. Userspace clients continue to query IB path records. Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-28 07:06:02 +08:00
.field_name = "sa_path_rec:" #field
static const struct ib_field opa_path_rec_table[] = {
RDMA/SA: Fix kernel panic in CMA request handler flow Commit 9fdca4da4d8c (IB/SA: Split struct sa_path_rec based on IB and ROCE specific fields) moved the service_id to be specific attribute for IB and OPA SA Path Record, and thus wasn't assigned for RoCE. This caused to the following kernel panic in the CMA request handler flow: [ 27.074594] BUG: unable to handle kernel NULL pointer dereference at 0000000000000008 [ 27.074731] IP: __radix_tree_lookup+0x1d/0xe0 ... [ 27.075356] Workqueue: ib_cm cm_work_handler [ib_cm] [ 27.075401] task: ffff88022e3b8000 task.stack: ffffc90001298000 [ 27.075449] RIP: 0010:__radix_tree_lookup+0x1d/0xe0 ... [ 27.075979] Call Trace: [ 27.076015] radix_tree_lookup+0xd/0x10 [ 27.076055] cma_ps_find+0x59/0x70 [rdma_cm] [ 27.076097] cma_id_from_event+0xd2/0x470 [rdma_cm] [ 27.076144] ? ib_init_ah_from_path+0x39a/0x590 [ib_core] [ 27.076193] cma_req_handler+0x25/0x480 [rdma_cm] [ 27.076237] cm_process_work+0x25/0x120 [ib_cm] [ 27.076280] ? cm_get_bth_pkey.isra.62+0x3c/0xa0 [ib_cm] [ 27.076350] cm_req_handler+0xb03/0xd40 [ib_cm] [ 27.076430] ? sched_clock_cpu+0x11/0xb0 [ 27.076478] cm_work_handler+0x194/0x1588 [ib_cm] [ 27.076525] process_one_work+0x160/0x410 [ 27.076565] worker_thread+0x137/0x4a0 [ 27.076614] kthread+0x112/0x150 [ 27.076684] ? max_active_store+0x60/0x60 [ 27.077642] ? kthread_park+0x90/0x90 [ 27.078530] ret_from_fork+0x2c/0x40 This patch moves it back to the common SA Path Record structure and removes the redundant setter and getter. Tested on Connect-IB and Connect-X4 in Infiniband and RoCE respectively. Fixes: 9fdca4da4d8c (IB/SA: Split struct sa_path_rec based on IB ands ROCE specific fields) Signed-off-by: Majd Dibbiny <majd@mellanox.com> Reviewed-by: Parav Pandit <parav@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-05-22 00:09:54 +08:00
{ OPA_PATH_REC_FIELD(service_id),
IB/SA: Add support to query OPA path records When the bit 26 of capmask2 field in OPA classport info query is set, SA will query for OPA path records instead of querying for IB path records. Note that OPA path records can only be queried by kernel ULPs. Userspace clients continue to query IB path records. Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-28 07:06:02 +08:00
.offset_words = 0,
.offset_bits = 0,
.size_bits = 64 },
{ OPA_PATH_REC_FIELD(dgid),
.offset_words = 2,
.offset_bits = 0,
.size_bits = 128 },
{ OPA_PATH_REC_FIELD(sgid),
.offset_words = 6,
.offset_bits = 0,
.size_bits = 128 },
{ OPA_PATH_REC_FIELD(opa.dlid),
.offset_words = 10,
.offset_bits = 0,
.size_bits = 32 },
{ OPA_PATH_REC_FIELD(opa.slid),
.offset_words = 11,
.offset_bits = 0,
.size_bits = 32 },
{ OPA_PATH_REC_FIELD(opa.raw_traffic),
.offset_words = 12,
.offset_bits = 0,
.size_bits = 1 },
{ RESERVED,
.offset_words = 12,
.offset_bits = 1,
.size_bits = 3 },
{ OPA_PATH_REC_FIELD(flow_label),
.offset_words = 12,
.offset_bits = 4,
.size_bits = 20 },
{ OPA_PATH_REC_FIELD(hop_limit),
.offset_words = 12,
.offset_bits = 24,
.size_bits = 8 },
{ OPA_PATH_REC_FIELD(traffic_class),
.offset_words = 13,
.offset_bits = 0,
.size_bits = 8 },
{ OPA_PATH_REC_FIELD(reversible),
.offset_words = 13,
.offset_bits = 8,
.size_bits = 1 },
{ OPA_PATH_REC_FIELD(numb_path),
.offset_words = 13,
.offset_bits = 9,
.size_bits = 7 },
{ OPA_PATH_REC_FIELD(pkey),
.offset_words = 13,
.offset_bits = 16,
.size_bits = 16 },
{ OPA_PATH_REC_FIELD(opa.l2_8B),
.offset_words = 14,
.offset_bits = 0,
.size_bits = 1 },
{ OPA_PATH_REC_FIELD(opa.l2_10B),
.offset_words = 14,
.offset_bits = 1,
.size_bits = 1 },
{ OPA_PATH_REC_FIELD(opa.l2_9B),
.offset_words = 14,
.offset_bits = 2,
.size_bits = 1 },
{ OPA_PATH_REC_FIELD(opa.l2_16B),
.offset_words = 14,
.offset_bits = 3,
.size_bits = 1 },
{ RESERVED,
.offset_words = 14,
.offset_bits = 4,
.size_bits = 2 },
{ OPA_PATH_REC_FIELD(opa.qos_type),
.offset_words = 14,
.offset_bits = 6,
.size_bits = 2 },
{ OPA_PATH_REC_FIELD(opa.qos_priority),
.offset_words = 14,
.offset_bits = 8,
.size_bits = 8 },
{ RESERVED,
.offset_words = 14,
.offset_bits = 16,
.size_bits = 3 },
{ OPA_PATH_REC_FIELD(sl),
.offset_words = 14,
.offset_bits = 19,
.size_bits = 5 },
{ RESERVED,
.offset_words = 14,
.offset_bits = 24,
.size_bits = 8 },
{ OPA_PATH_REC_FIELD(mtu_selector),
.offset_words = 15,
.offset_bits = 0,
.size_bits = 2 },
{ OPA_PATH_REC_FIELD(mtu),
.offset_words = 15,
.offset_bits = 2,
.size_bits = 6 },
{ OPA_PATH_REC_FIELD(rate_selector),
.offset_words = 15,
.offset_bits = 8,
.size_bits = 2 },
{ OPA_PATH_REC_FIELD(rate),
.offset_words = 15,
.offset_bits = 10,
.size_bits = 6 },
{ OPA_PATH_REC_FIELD(packet_life_time_selector),
.offset_words = 15,
.offset_bits = 16,
.size_bits = 2 },
{ OPA_PATH_REC_FIELD(packet_life_time),
.offset_words = 15,
.offset_bits = 18,
.size_bits = 6 },
{ OPA_PATH_REC_FIELD(preference),
.offset_words = 15,
.offset_bits = 24,
.size_bits = 8 },
};
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
#define MCMEMBER_REC_FIELD(field) \
.struct_offset_bytes = offsetof(struct ib_sa_mcmember_rec, field), \
RDMA/core: Use sizeof_field() helper Make use of the sizeof_field() helper instead of an open-coded version. Link: https://lore.kernel.org/r/20200527144152.GA22605@embeddedor Signed-off-by: Gustavo A. R. Silva <gustavoars@kernel.org> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-27 22:41:52 +08:00
.struct_size_bytes = sizeof_field(struct ib_sa_mcmember_rec, field), \
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
.field_name = "sa_mcmember_rec:" #field
static const struct ib_field mcmember_rec_table[] = {
{ MCMEMBER_REC_FIELD(mgid),
.offset_words = 0,
.offset_bits = 0,
.size_bits = 128 },
{ MCMEMBER_REC_FIELD(port_gid),
.offset_words = 4,
.offset_bits = 0,
.size_bits = 128 },
{ MCMEMBER_REC_FIELD(qkey),
.offset_words = 8,
.offset_bits = 0,
.size_bits = 32 },
{ MCMEMBER_REC_FIELD(mlid),
.offset_words = 9,
.offset_bits = 0,
.size_bits = 16 },
{ MCMEMBER_REC_FIELD(mtu_selector),
.offset_words = 9,
.offset_bits = 16,
.size_bits = 2 },
{ MCMEMBER_REC_FIELD(mtu),
.offset_words = 9,
.offset_bits = 18,
.size_bits = 6 },
{ MCMEMBER_REC_FIELD(traffic_class),
.offset_words = 9,
.offset_bits = 24,
.size_bits = 8 },
{ MCMEMBER_REC_FIELD(pkey),
.offset_words = 10,
.offset_bits = 0,
.size_bits = 16 },
{ MCMEMBER_REC_FIELD(rate_selector),
.offset_words = 10,
.offset_bits = 16,
.size_bits = 2 },
{ MCMEMBER_REC_FIELD(rate),
.offset_words = 10,
.offset_bits = 18,
.size_bits = 6 },
{ MCMEMBER_REC_FIELD(packet_life_time_selector),
.offset_words = 10,
.offset_bits = 24,
.size_bits = 2 },
{ MCMEMBER_REC_FIELD(packet_life_time),
.offset_words = 10,
.offset_bits = 26,
.size_bits = 6 },
{ MCMEMBER_REC_FIELD(sl),
.offset_words = 11,
.offset_bits = 0,
.size_bits = 4 },
{ MCMEMBER_REC_FIELD(flow_label),
.offset_words = 11,
.offset_bits = 4,
.size_bits = 20 },
{ MCMEMBER_REC_FIELD(hop_limit),
.offset_words = 11,
.offset_bits = 24,
.size_bits = 8 },
{ MCMEMBER_REC_FIELD(scope),
.offset_words = 12,
.offset_bits = 0,
.size_bits = 4 },
{ MCMEMBER_REC_FIELD(join_state),
.offset_words = 12,
.offset_bits = 4,
.size_bits = 4 },
{ MCMEMBER_REC_FIELD(proxy_join),
.offset_words = 12,
.offset_bits = 8,
.size_bits = 1 },
{ RESERVED,
.offset_words = 12,
.offset_bits = 9,
.size_bits = 23 },
};
IB/SA Agent: Add support for SA agent get ClassPortInfo New SA query function to return the ClassPortInfo struct from the SA. If the SM supports FullMemberSendOnly mode for MCG's, it sets a capability bit in the capability_mask2 field of the response. Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-05-26 03:02:05 +08:00
#define CLASSPORTINFO_REC_FIELD(field) \
.struct_offset_bytes = offsetof(struct ib_class_port_info, field), \
RDMA/core: Use sizeof_field() helper Make use of the sizeof_field() helper instead of an open-coded version. Link: https://lore.kernel.org/r/20200527144152.GA22605@embeddedor Signed-off-by: Gustavo A. R. Silva <gustavoars@kernel.org> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-27 22:41:52 +08:00
.struct_size_bytes = sizeof_field(struct ib_class_port_info, field), \
IB/SA Agent: Add support for SA agent get ClassPortInfo New SA query function to return the ClassPortInfo struct from the SA. If the SM supports FullMemberSendOnly mode for MCG's, it sets a capability bit in the capability_mask2 field of the response. Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-05-26 03:02:05 +08:00
.field_name = "ib_class_port_info:" #field
IB/SA: Add support to query opa classport info. For OPA devices, SA will query the OPA classport info instead of the IB defined classport info. opa classport info exposes additional information and capabilities that are specific to OPA devices. Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-29 07:20:26 +08:00
static const struct ib_field ib_classport_info_rec_table[] = {
IB/SA Agent: Add support for SA agent get ClassPortInfo New SA query function to return the ClassPortInfo struct from the SA. If the SM supports FullMemberSendOnly mode for MCG's, it sets a capability bit in the capability_mask2 field of the response. Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-05-26 03:02:05 +08:00
{ CLASSPORTINFO_REC_FIELD(base_version),
.offset_words = 0,
.offset_bits = 0,
.size_bits = 8 },
{ CLASSPORTINFO_REC_FIELD(class_version),
.offset_words = 0,
.offset_bits = 8,
.size_bits = 8 },
{ CLASSPORTINFO_REC_FIELD(capability_mask),
.offset_words = 0,
.offset_bits = 16,
.size_bits = 16 },
{ CLASSPORTINFO_REC_FIELD(cap_mask2_resp_time),
.offset_words = 1,
.offset_bits = 0,
.size_bits = 32 },
{ CLASSPORTINFO_REC_FIELD(redirect_gid),
.offset_words = 2,
.offset_bits = 0,
.size_bits = 128 },
{ CLASSPORTINFO_REC_FIELD(redirect_tcslfl),
.offset_words = 6,
.offset_bits = 0,
.size_bits = 32 },
{ CLASSPORTINFO_REC_FIELD(redirect_lid),
.offset_words = 7,
.offset_bits = 0,
.size_bits = 16 },
{ CLASSPORTINFO_REC_FIELD(redirect_pkey),
.offset_words = 7,
.offset_bits = 16,
.size_bits = 16 },
{ CLASSPORTINFO_REC_FIELD(redirect_qp),
.offset_words = 8,
.offset_bits = 0,
.size_bits = 32 },
{ CLASSPORTINFO_REC_FIELD(redirect_qkey),
.offset_words = 9,
.offset_bits = 0,
.size_bits = 32 },
{ CLASSPORTINFO_REC_FIELD(trap_gid),
.offset_words = 10,
.offset_bits = 0,
.size_bits = 128 },
{ CLASSPORTINFO_REC_FIELD(trap_tcslfl),
.offset_words = 14,
.offset_bits = 0,
.size_bits = 32 },
{ CLASSPORTINFO_REC_FIELD(trap_lid),
.offset_words = 15,
.offset_bits = 0,
.size_bits = 16 },
{ CLASSPORTINFO_REC_FIELD(trap_pkey),
.offset_words = 15,
.offset_bits = 16,
.size_bits = 16 },
{ CLASSPORTINFO_REC_FIELD(trap_hlqp),
.offset_words = 16,
.offset_bits = 0,
.size_bits = 32 },
{ CLASSPORTINFO_REC_FIELD(trap_qkey),
.offset_words = 17,
.offset_bits = 0,
.size_bits = 32 },
};
IB/SA: Add support to query opa classport info. For OPA devices, SA will query the OPA classport info instead of the IB defined classport info. opa classport info exposes additional information and capabilities that are specific to OPA devices. Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-29 07:20:26 +08:00
#define OPA_CLASSPORTINFO_REC_FIELD(field) \
.struct_offset_bytes =\
offsetof(struct opa_class_port_info, field), \
.struct_size_bytes = \
RDMA/core: Use sizeof_field() helper Make use of the sizeof_field() helper instead of an open-coded version. Link: https://lore.kernel.org/r/20200527144152.GA22605@embeddedor Signed-off-by: Gustavo A. R. Silva <gustavoars@kernel.org> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-27 22:41:52 +08:00
sizeof_field(struct opa_class_port_info, field), \
IB/SA: Add support to query opa classport info. For OPA devices, SA will query the OPA classport info instead of the IB defined classport info. opa classport info exposes additional information and capabilities that are specific to OPA devices. Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-29 07:20:26 +08:00
.field_name = "opa_class_port_info:" #field
static const struct ib_field opa_classport_info_rec_table[] = {
{ OPA_CLASSPORTINFO_REC_FIELD(base_version),
.offset_words = 0,
.offset_bits = 0,
.size_bits = 8 },
{ OPA_CLASSPORTINFO_REC_FIELD(class_version),
.offset_words = 0,
.offset_bits = 8,
.size_bits = 8 },
{ OPA_CLASSPORTINFO_REC_FIELD(cap_mask),
.offset_words = 0,
.offset_bits = 16,
.size_bits = 16 },
{ OPA_CLASSPORTINFO_REC_FIELD(cap_mask2_resp_time),
.offset_words = 1,
.offset_bits = 0,
.size_bits = 32 },
{ OPA_CLASSPORTINFO_REC_FIELD(redirect_gid),
.offset_words = 2,
.offset_bits = 0,
.size_bits = 128 },
{ OPA_CLASSPORTINFO_REC_FIELD(redirect_tc_fl),
.offset_words = 6,
.offset_bits = 0,
.size_bits = 32 },
{ OPA_CLASSPORTINFO_REC_FIELD(redirect_lid),
.offset_words = 7,
.offset_bits = 0,
.size_bits = 32 },
{ OPA_CLASSPORTINFO_REC_FIELD(redirect_sl_qp),
.offset_words = 8,
.offset_bits = 0,
.size_bits = 32 },
{ OPA_CLASSPORTINFO_REC_FIELD(redirect_qkey),
.offset_words = 9,
.offset_bits = 0,
.size_bits = 32 },
{ OPA_CLASSPORTINFO_REC_FIELD(trap_gid),
.offset_words = 10,
.offset_bits = 0,
.size_bits = 128 },
{ OPA_CLASSPORTINFO_REC_FIELD(trap_tc_fl),
.offset_words = 14,
.offset_bits = 0,
.size_bits = 32 },
{ OPA_CLASSPORTINFO_REC_FIELD(trap_lid),
.offset_words = 15,
.offset_bits = 0,
.size_bits = 32 },
{ OPA_CLASSPORTINFO_REC_FIELD(trap_hl_qp),
.offset_words = 16,
.offset_bits = 0,
.size_bits = 32 },
{ OPA_CLASSPORTINFO_REC_FIELD(trap_qkey),
.offset_words = 17,
.offset_bits = 0,
.size_bits = 32 },
{ OPA_CLASSPORTINFO_REC_FIELD(trap_pkey),
.offset_words = 18,
.offset_bits = 0,
.size_bits = 16 },
{ OPA_CLASSPORTINFO_REC_FIELD(redirect_pkey),
.offset_words = 18,
.offset_bits = 16,
.size_bits = 16 },
{ OPA_CLASSPORTINFO_REC_FIELD(trap_sl_rsvd),
.offset_words = 19,
.offset_bits = 0,
.size_bits = 8 },
{ RESERVED,
.offset_words = 19,
.offset_bits = 8,
.size_bits = 24 },
};
IB/sa: Add GuidInfoRecord query support This query is needed for SRIOV alias GUID support. The query is implemented per the IB Spec definition in section 15.2.5.18 (GuidInfoRecord). Signed-off-by: Erez Shitrit <erezsh@mellanox.co.il> Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il> Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com> Signed-off-by: Roland Dreier <roland@purestorage.com>
2012-06-19 16:21:38 +08:00
#define GUIDINFO_REC_FIELD(field) \
.struct_offset_bytes = offsetof(struct ib_sa_guidinfo_rec, field), \
RDMA/core: Use sizeof_field() helper Make use of the sizeof_field() helper instead of an open-coded version. Link: https://lore.kernel.org/r/20200527144152.GA22605@embeddedor Signed-off-by: Gustavo A. R. Silva <gustavoars@kernel.org> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-05-27 22:41:52 +08:00
.struct_size_bytes = sizeof_field(struct ib_sa_guidinfo_rec, field), \
IB/sa: Add GuidInfoRecord query support This query is needed for SRIOV alias GUID support. The query is implemented per the IB Spec definition in section 15.2.5.18 (GuidInfoRecord). Signed-off-by: Erez Shitrit <erezsh@mellanox.co.il> Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il> Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com> Signed-off-by: Roland Dreier <roland@purestorage.com>
2012-06-19 16:21:38 +08:00
.field_name = "sa_guidinfo_rec:" #field
static const struct ib_field guidinfo_rec_table[] = {
{ GUIDINFO_REC_FIELD(lid),
.offset_words = 0,
.offset_bits = 0,
.size_bits = 16 },
{ GUIDINFO_REC_FIELD(block_num),
.offset_words = 0,
.offset_bits = 16,
.size_bits = 8 },
{ GUIDINFO_REC_FIELD(res1),
.offset_words = 0,
.offset_bits = 24,
.size_bits = 8 },
{ GUIDINFO_REC_FIELD(res2),
.offset_words = 1,
.offset_bits = 0,
.size_bits = 32 },
{ GUIDINFO_REC_FIELD(guid_info_list),
.offset_words = 2,
.offset_bits = 0,
.size_bits = 512 },
};
RDMA/cma: Refactor the inbound/outbound path records process flow Refactors based on comments [1] of the multiple path records support patchset: - Return failure if not able to set inbound/outbound PRs; - Simplify the flow when receiving the PRs from netlink channel: When a good PR response is received, unpack it and call the path_query callback directly. This saves two memory allocations; - Define RDMA_PRIMARY_PATH_MAX_REC_NUM in a proper place. [1] https://lore.kernel.org/linux-rdma/Yyxp9E9pJtUids2o@nvidia.com/ Signed-off-by: Mark Zhang <markzhang@nvidia.com> Reviewed-by: Bart Van Assche <bvanassche@acm.org> #srp Link: https://lore.kernel.org/r/7610025d57342b8b6da0f19516c9612f9c3fdc37.1672819376.git.leonro@nvidia.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2023-01-04 16:03:41 +08:00
#define RDMA_PRIMARY_PATH_MAX_REC_NUM 3
IB/sa: Route SA pathrecord query through netlink This patch routes a SA pathrecord query to netlink first and processes the response appropriately. If a failure is returned, the request will be sent through IB. The decision whether to route the request to netlink first is determined by the presence of a listener for the local service netlink multicast group. If the user-space local service netlink multicast group listener is not present, the request will be sent through IB, just like what is currently being done. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: John Fleck <john.fleck@intel.com> Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-14 20:52:09 +08:00
static inline void ib_sa_disable_local_svc(struct ib_sa_query *query)
{
query->flags &= ~IB_SA_ENABLE_LOCAL_SERVICE;
}
static inline int ib_sa_query_cancelled(struct ib_sa_query *query)
{
return (query->flags & IB_SA_CANCEL);
}
static void ib_nl_set_path_rec_attrs(struct sk_buff *skb,
struct ib_sa_query *query)
{
IB/SA: Rename ib_sa_path_rec to sa_path_rec Rename ib_sa_path_rec to a more generic sa_path_rec. This is part of extending ib_sa to also support OPA path records in addition to the IB defined path records. Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-28 07:05:58 +08:00
struct sa_path_rec *sa_rec = query->mad_buf->context[1];
IB/sa: Route SA pathrecord query through netlink This patch routes a SA pathrecord query to netlink first and processes the response appropriately. If a failure is returned, the request will be sent through IB. The decision whether to route the request to netlink first is determined by the presence of a listener for the local service netlink multicast group. If the user-space local service netlink multicast group listener is not present, the request will be sent through IB, just like what is currently being done. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: John Fleck <john.fleck@intel.com> Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-14 20:52:09 +08:00
struct ib_sa_mad *mad = query->mad_buf->mad;
ib_sa_comp_mask comp_mask = mad->sa_hdr.comp_mask;
u16 val16;
u64 val64;
struct rdma_ls_resolve_header *header;
query->mad_buf->context[1] = NULL;
/* Construct the family header first */
networking: make skb_put & friends return void pointers It seems like a historic accident that these return unsigned char *, and in many places that means casts are required, more often than not. Make these functions (skb_put, __skb_put and pskb_put) return void * and remove all the casts across the tree, adding a (u8 *) cast only where the unsigned char pointer was used directly, all done with the following spatch: @@ expression SKB, LEN; typedef u8; identifier fn = { skb_put, __skb_put }; @@ - *(fn(SKB, LEN)) + *(u8 *)fn(SKB, LEN) @@ expression E, SKB, LEN; identifier fn = { skb_put, __skb_put }; type T; @@ - E = ((T *)(fn(SKB, LEN))) + E = fn(SKB, LEN) which actually doesn't cover pskb_put since there are only three users overall. A handful of stragglers were converted manually, notably a macro in drivers/isdn/i4l/isdn_bsdcomp.c and, oddly enough, one of the many instances in net/bluetooth/hci_sock.c. In the former file, I also had to fix one whitespace problem spatch introduced. Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2017-06-16 20:29:21 +08:00
header = skb_put(skb, NLMSG_ALIGN(sizeof(*header)));
RDMA/sa_query: Use strscpy_pad instead of memcpy to copy a string When copying the device name, the length of the data memcpy copied exceeds the length of the source buffer, which cause the KASAN issue below. Use strscpy_pad() instead. BUG: KASAN: slab-out-of-bounds in ib_nl_set_path_rec_attrs+0x136/0x320 [ib_core] Read of size 64 at addr ffff88811a10f5e0 by task rping/140263 CPU: 3 PID: 140263 Comm: rping Not tainted 5.15.0-rc1+ #1 Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS rel-1.13.0-0-gf21b5a4aeb02-prebuilt.qemu.org 04/01/2014 Call Trace: dump_stack_lvl+0x57/0x7d print_address_description.constprop.0+0x1d/0xa0 kasan_report+0xcb/0x110 kasan_check_range+0x13d/0x180 memcpy+0x20/0x60 ib_nl_set_path_rec_attrs+0x136/0x320 [ib_core] ib_nl_make_request+0x1c6/0x380 [ib_core] send_mad+0x20a/0x220 [ib_core] ib_sa_path_rec_get+0x3e3/0x800 [ib_core] cma_query_ib_route+0x29b/0x390 [rdma_cm] rdma_resolve_route+0x308/0x3e0 [rdma_cm] ucma_resolve_route+0xe1/0x150 [rdma_ucm] ucma_write+0x17b/0x1f0 [rdma_ucm] vfs_write+0x142/0x4d0 ksys_write+0x133/0x160 do_syscall_64+0x43/0x90 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x7f26499aa90f Code: 89 54 24 18 48 89 74 24 10 89 7c 24 08 e8 29 fd ff ff 48 8b 54 24 18 48 8b 74 24 10 41 89 c0 8b 7c 24 08 b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 77 31 44 89 c7 48 89 44 24 08 e8 5c fd ff ff 48 RSP: 002b:00007f26495f2dc0 EFLAGS: 00000293 ORIG_RAX: 0000000000000001 RAX: ffffffffffffffda RBX: 00000000000007d0 RCX: 00007f26499aa90f RDX: 0000000000000010 RSI: 00007f26495f2e00 RDI: 0000000000000003 RBP: 00005632a8315440 R08: 0000000000000000 R09: 0000000000000001 R10: 0000000000000000 R11: 0000000000000293 R12: 00007f26495f2e00 R13: 00005632a83154e0 R14: 00005632a8315440 R15: 00005632a830a810 Allocated by task 131419: kasan_save_stack+0x1b/0x40 __kasan_kmalloc+0x7c/0x90 proc_self_get_link+0x8b/0x100 pick_link+0x4f1/0x5c0 step_into+0x2eb/0x3d0 walk_component+0xc8/0x2c0 link_path_walk+0x3b8/0x580 path_openat+0x101/0x230 do_filp_open+0x12e/0x240 do_sys_openat2+0x115/0x280 __x64_sys_openat+0xce/0x140 do_syscall_64+0x43/0x90 entry_SYSCALL_64_after_hwframe+0x44/0xae Fixes: 2ca546b92a02 ("IB/sa: Route SA pathrecord query through netlink") Link: https://lore.kernel.org/r/72ede0f6dab61f7f23df9ac7a70666e07ef314b0.1635055496.git.leonro@nvidia.com Signed-off-by: Mark Zhang <markzhang@nvidia.com> Reviewed-by: Mark Bloch <mbloch@nvidia.com> Signed-off-by: Leon Romanovsky <leonro@nvidia.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
2021-10-24 14:08:20 +08:00
strscpy_pad(header->device_name,
dev_name(&query->port->agent->device->dev),
LS_DEVICE_NAME_MAX);
IB/sa: Route SA pathrecord query through netlink This patch routes a SA pathrecord query to netlink first and processes the response appropriately. If a failure is returned, the request will be sent through IB. The decision whether to route the request to netlink first is determined by the presence of a listener for the local service netlink multicast group. If the user-space local service netlink multicast group listener is not present, the request will be sent through IB, just like what is currently being done. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: John Fleck <john.fleck@intel.com> Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-14 20:52:09 +08:00
header->port_num = query->port->port_num;
if ((comp_mask & IB_SA_PATH_REC_REVERSIBLE) &&
sa_rec->reversible != 0)
RDMA/cma: Multiple path records support with netlink channel Support receiving inbound and outbound IB path records (along with GMP PathRecord) from user-space service through the RDMA netlink channel. The LIDs in these 3 PRs can be used in this way: 1. GMP PR: used as the standard local/remote LIDs; 2. DLID of outbound PR: Used as the "dlid" field for outbound traffic; 3. DLID of inbound PR: Used as the "dlid" field for outbound traffic in responder side. This is aimed to support adaptive routing. With current IB routing solution when a packet goes out it's assigned with a fixed DLID per target, meaning a fixed router will be used. The LIDs in inbound/outbound path records can be used to identify group of routers that allow communication with another subnet's entity. With them packets from an inter-subnet connection may travel through any router in the set to reach the target. As confirmed with Jason, when sending a netlink request, kernel uses LS_RESOLVE_PATH_USE_ALL so that the service knows kernel supports multiple PRs. Signed-off-by: Mark Zhang <markzhang@nvidia.com> Reviewed-by: Mark Bloch <mbloch@nvidia.com> Link: https://lore.kernel.org/r/2fa2b6c93c4c16c8915bac3cfc4f27be1d60519d.1662631201.git.leonro@nvidia.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2022-09-08 18:09:01 +08:00
query->path_use = LS_RESOLVE_PATH_USE_ALL;
IB/sa: Route SA pathrecord query through netlink This patch routes a SA pathrecord query to netlink first and processes the response appropriately. If a failure is returned, the request will be sent through IB. The decision whether to route the request to netlink first is determined by the presence of a listener for the local service netlink multicast group. If the user-space local service netlink multicast group listener is not present, the request will be sent through IB, just like what is currently being done. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: John Fleck <john.fleck@intel.com> Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-14 20:52:09 +08:00
else
query->path_use = LS_RESOLVE_PATH_USE_UNIDIRECTIONAL;
header->path_use = query->path_use;
/* Now build the attributes */
if (comp_mask & IB_SA_PATH_REC_SERVICE_ID) {
RDMA/SA: Fix kernel panic in CMA request handler flow Commit 9fdca4da4d8c (IB/SA: Split struct sa_path_rec based on IB and ROCE specific fields) moved the service_id to be specific attribute for IB and OPA SA Path Record, and thus wasn't assigned for RoCE. This caused to the following kernel panic in the CMA request handler flow: [ 27.074594] BUG: unable to handle kernel NULL pointer dereference at 0000000000000008 [ 27.074731] IP: __radix_tree_lookup+0x1d/0xe0 ... [ 27.075356] Workqueue: ib_cm cm_work_handler [ib_cm] [ 27.075401] task: ffff88022e3b8000 task.stack: ffffc90001298000 [ 27.075449] RIP: 0010:__radix_tree_lookup+0x1d/0xe0 ... [ 27.075979] Call Trace: [ 27.076015] radix_tree_lookup+0xd/0x10 [ 27.076055] cma_ps_find+0x59/0x70 [rdma_cm] [ 27.076097] cma_id_from_event+0xd2/0x470 [rdma_cm] [ 27.076144] ? ib_init_ah_from_path+0x39a/0x590 [ib_core] [ 27.076193] cma_req_handler+0x25/0x480 [rdma_cm] [ 27.076237] cm_process_work+0x25/0x120 [ib_cm] [ 27.076280] ? cm_get_bth_pkey.isra.62+0x3c/0xa0 [ib_cm] [ 27.076350] cm_req_handler+0xb03/0xd40 [ib_cm] [ 27.076430] ? sched_clock_cpu+0x11/0xb0 [ 27.076478] cm_work_handler+0x194/0x1588 [ib_cm] [ 27.076525] process_one_work+0x160/0x410 [ 27.076565] worker_thread+0x137/0x4a0 [ 27.076614] kthread+0x112/0x150 [ 27.076684] ? max_active_store+0x60/0x60 [ 27.077642] ? kthread_park+0x90/0x90 [ 27.078530] ret_from_fork+0x2c/0x40 This patch moves it back to the common SA Path Record structure and removes the redundant setter and getter. Tested on Connect-IB and Connect-X4 in Infiniband and RoCE respectively. Fixes: 9fdca4da4d8c (IB/SA: Split struct sa_path_rec based on IB ands ROCE specific fields) Signed-off-by: Majd Dibbiny <majd@mellanox.com> Reviewed-by: Parav Pandit <parav@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-05-22 00:09:54 +08:00
val64 = be64_to_cpu(sa_rec->service_id);
IB/sa: Route SA pathrecord query through netlink This patch routes a SA pathrecord query to netlink first and processes the response appropriately. If a failure is returned, the request will be sent through IB. The decision whether to route the request to netlink first is determined by the presence of a listener for the local service netlink multicast group. If the user-space local service netlink multicast group listener is not present, the request will be sent through IB, just like what is currently being done. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: John Fleck <john.fleck@intel.com> Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-14 20:52:09 +08:00
nla_put(skb, RDMA_NLA_F_MANDATORY | LS_NLA_TYPE_SERVICE_ID,
sizeof(val64), &val64);
}
if (comp_mask & IB_SA_PATH_REC_DGID)
nla_put(skb, RDMA_NLA_F_MANDATORY | LS_NLA_TYPE_DGID,
sizeof(sa_rec->dgid), &sa_rec->dgid);
if (comp_mask & IB_SA_PATH_REC_SGID)
nla_put(skb, RDMA_NLA_F_MANDATORY | LS_NLA_TYPE_SGID,
sizeof(sa_rec->sgid), &sa_rec->sgid);
if (comp_mask & IB_SA_PATH_REC_TRAFFIC_CLASS)
nla_put(skb, RDMA_NLA_F_MANDATORY | LS_NLA_TYPE_TCLASS,
sizeof(sa_rec->traffic_class), &sa_rec->traffic_class);
if (comp_mask & IB_SA_PATH_REC_PKEY) {
val16 = be16_to_cpu(sa_rec->pkey);
nla_put(skb, RDMA_NLA_F_MANDATORY | LS_NLA_TYPE_PKEY,
sizeof(val16), &val16);
}
if (comp_mask & IB_SA_PATH_REC_QOS_CLASS) {
val16 = be16_to_cpu(sa_rec->qos_class);
nla_put(skb, RDMA_NLA_F_MANDATORY | LS_NLA_TYPE_QOS_CLASS,
sizeof(val16), &val16);
}
}
static int ib_nl_get_path_rec_attrs_len(ib_sa_comp_mask comp_mask)
{
int len = 0;
if (comp_mask & IB_SA_PATH_REC_SERVICE_ID)
len += nla_total_size(sizeof(u64));
if (comp_mask & IB_SA_PATH_REC_DGID)
len += nla_total_size(sizeof(struct rdma_nla_ls_gid));
if (comp_mask & IB_SA_PATH_REC_SGID)
len += nla_total_size(sizeof(struct rdma_nla_ls_gid));
if (comp_mask & IB_SA_PATH_REC_TRAFFIC_CLASS)
len += nla_total_size(sizeof(u8));
if (comp_mask & IB_SA_PATH_REC_PKEY)
len += nla_total_size(sizeof(u16));
if (comp_mask & IB_SA_PATH_REC_QOS_CLASS)
len += nla_total_size(sizeof(u16));
/*
* Make sure that at least some of the required comp_mask bits are
* set.
*/
if (WARN_ON(len == 0))
return len;
/* Add the family header */
len += NLMSG_ALIGN(sizeof(struct rdma_ls_resolve_header));
return len;
}
IB/sa: Resolv use-after-free in ib_nl_make_request() There is a race condition where ib_nl_make_request() inserts the request data into the linked list but the timer in ib_nl_request_timeout() can see it and destroy it before ib_nl_send_msg() is done touching it. This could happen, for instance, if there is a long delay allocating memory during nlmsg_new() This causes a use-after-free in the send_mad() thread: [<ffffffffa02f43cb>] ? ib_pack+0x17b/0x240 [ib_core] [ <ffffffffa032aef1>] ib_sa_path_rec_get+0x181/0x200 [ib_sa] [<ffffffffa0379db0>] rdma_resolve_route+0x3c0/0x8d0 [rdma_cm] [<ffffffffa0374450>] ? cma_bind_port+0xa0/0xa0 [rdma_cm] [<ffffffffa040f850>] ? rds_rdma_cm_event_handler_cmn+0x850/0x850 [rds_rdma] [<ffffffffa040f22c>] rds_rdma_cm_event_handler_cmn+0x22c/0x850 [rds_rdma] [<ffffffffa040f860>] rds_rdma_cm_event_handler+0x10/0x20 [rds_rdma] [<ffffffffa037778e>] addr_handler+0x9e/0x140 [rdma_cm] [<ffffffffa026cdb4>] process_req+0x134/0x190 [ib_addr] [<ffffffff810a02f9>] process_one_work+0x169/0x4a0 [<ffffffff810a0b2b>] worker_thread+0x5b/0x560 [<ffffffff810a0ad0>] ? flush_delayed_work+0x50/0x50 [<ffffffff810a68fb>] kthread+0xcb/0xf0 [<ffffffff816ec49a>] ? __schedule+0x24a/0x810 [<ffffffff816ec49a>] ? __schedule+0x24a/0x810 [<ffffffff810a6830>] ? kthread_create_on_node+0x180/0x180 [<ffffffff816f25a7>] ret_from_fork+0x47/0x90 [<ffffffff810a6830>] ? kthread_create_on_node+0x180/0x180 The ownership rule is once the request is on the list, ownership transfers to the list and the local thread can't touch it any more, just like for the normal MAD case in send_mad(). Thus, instead of adding before send and then trying to delete after on errors, move the entire thing under the spinlock so that the send and update of the lists are atomic to the conurrent threads. Lightly reoganize things so spinlock safe memory allocations are done in the final NL send path and the rest of the setup work is done before and outside the lock. Fixes: 3ebd2fd0d011 ("IB/sa: Put netlink request into the request list before sending") Link: https://lore.kernel.org/r/1592964789-14533-1-git-send-email-divya.indi@oracle.com Signed-off-by: Divya Indi <divya.indi@oracle.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
2020-06-24 10:13:09 +08:00
static int ib_nl_make_request(struct ib_sa_query *query, gfp_t gfp_mask)
IB/sa: Route SA pathrecord query through netlink This patch routes a SA pathrecord query to netlink first and processes the response appropriately. If a failure is returned, the request will be sent through IB. The decision whether to route the request to netlink first is determined by the presence of a listener for the local service netlink multicast group. If the user-space local service netlink multicast group listener is not present, the request will be sent through IB, just like what is currently being done. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: John Fleck <john.fleck@intel.com> Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-14 20:52:09 +08:00
{
struct sk_buff *skb = NULL;
struct nlmsghdr *nlh;
void *data;
struct ib_sa_mad *mad;
int len;
IB/sa: Resolv use-after-free in ib_nl_make_request() There is a race condition where ib_nl_make_request() inserts the request data into the linked list but the timer in ib_nl_request_timeout() can see it and destroy it before ib_nl_send_msg() is done touching it. This could happen, for instance, if there is a long delay allocating memory during nlmsg_new() This causes a use-after-free in the send_mad() thread: [<ffffffffa02f43cb>] ? ib_pack+0x17b/0x240 [ib_core] [ <ffffffffa032aef1>] ib_sa_path_rec_get+0x181/0x200 [ib_sa] [<ffffffffa0379db0>] rdma_resolve_route+0x3c0/0x8d0 [rdma_cm] [<ffffffffa0374450>] ? cma_bind_port+0xa0/0xa0 [rdma_cm] [<ffffffffa040f850>] ? rds_rdma_cm_event_handler_cmn+0x850/0x850 [rds_rdma] [<ffffffffa040f22c>] rds_rdma_cm_event_handler_cmn+0x22c/0x850 [rds_rdma] [<ffffffffa040f860>] rds_rdma_cm_event_handler+0x10/0x20 [rds_rdma] [<ffffffffa037778e>] addr_handler+0x9e/0x140 [rdma_cm] [<ffffffffa026cdb4>] process_req+0x134/0x190 [ib_addr] [<ffffffff810a02f9>] process_one_work+0x169/0x4a0 [<ffffffff810a0b2b>] worker_thread+0x5b/0x560 [<ffffffff810a0ad0>] ? flush_delayed_work+0x50/0x50 [<ffffffff810a68fb>] kthread+0xcb/0xf0 [<ffffffff816ec49a>] ? __schedule+0x24a/0x810 [<ffffffff816ec49a>] ? __schedule+0x24a/0x810 [<ffffffff810a6830>] ? kthread_create_on_node+0x180/0x180 [<ffffffff816f25a7>] ret_from_fork+0x47/0x90 [<ffffffff810a6830>] ? kthread_create_on_node+0x180/0x180 The ownership rule is once the request is on the list, ownership transfers to the list and the local thread can't touch it any more, just like for the normal MAD case in send_mad(). Thus, instead of adding before send and then trying to delete after on errors, move the entire thing under the spinlock so that the send and update of the lists are atomic to the conurrent threads. Lightly reoganize things so spinlock safe memory allocations are done in the final NL send path and the rest of the setup work is done before and outside the lock. Fixes: 3ebd2fd0d011 ("IB/sa: Put netlink request into the request list before sending") Link: https://lore.kernel.org/r/1592964789-14533-1-git-send-email-divya.indi@oracle.com Signed-off-by: Divya Indi <divya.indi@oracle.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
2020-06-24 10:13:09 +08:00
unsigned long flags;
unsigned long delay;
gfp_t gfp_flag;
int ret;
INIT_LIST_HEAD(&query->list);
query->seq = (u32)atomic_inc_return(&ib_nl_sa_request_seq);
IB/sa: Route SA pathrecord query through netlink This patch routes a SA pathrecord query to netlink first and processes the response appropriately. If a failure is returned, the request will be sent through IB. The decision whether to route the request to netlink first is determined by the presence of a listener for the local service netlink multicast group. If the user-space local service netlink multicast group listener is not present, the request will be sent through IB, just like what is currently being done. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: John Fleck <john.fleck@intel.com> Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-14 20:52:09 +08:00
mad = query->mad_buf->mad;
len = ib_nl_get_path_rec_attrs_len(mad->sa_hdr.comp_mask);
if (len <= 0)
return -EMSGSIZE;
IB/sa: Put netlink request into the request list before sending It was found by Saurabh Sengar that the netlink code tried to allocate memory with GFP_KERNEL while holding a spinlock. While it is possible to fix the issue by replacing GFP_KERNEL with GFP_ATOMIC, it is better to get rid of the spinlock while sending the packet. However, in order to protect against a race condition that a quick response may be received before the request is put on the request list, we need to put the request on the list first. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reported-by: Saurabh Sengar <saurabh.truth@gmail.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-10-30 20:23:45 +08:00
skb = nlmsg_new(len, gfp_mask);
IB/sa: Route SA pathrecord query through netlink This patch routes a SA pathrecord query to netlink first and processes the response appropriately. If a failure is returned, the request will be sent through IB. The decision whether to route the request to netlink first is determined by the presence of a listener for the local service netlink multicast group. If the user-space local service netlink multicast group listener is not present, the request will be sent through IB, just like what is currently being done. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: John Fleck <john.fleck@intel.com> Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-14 20:52:09 +08:00
if (!skb)
return -ENOMEM;
/* Put nlmsg header only for now */
data = ibnl_put_msg(skb, &nlh, query->seq, 0, RDMA_NL_LS,
IB/sa: Fix rdma netlink message flags The flags to ibnl_put_msg should be NLM_F_REQUEST instead of GFP_KERNEL. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: John Fleck <john.fleck@intel.com> Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-21 02:20:42 +08:00
RDMA_NL_LS_OP_RESOLVE, NLM_F_REQUEST);
IB/sa: Route SA pathrecord query through netlink This patch routes a SA pathrecord query to netlink first and processes the response appropriately. If a failure is returned, the request will be sent through IB. The decision whether to route the request to netlink first is determined by the presence of a listener for the local service netlink multicast group. If the user-space local service netlink multicast group listener is not present, the request will be sent through IB, just like what is currently being done. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: John Fleck <john.fleck@intel.com> Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-14 20:52:09 +08:00
if (!data) {
IB/SA: Use correct free function Fixes a direct call to kfree_skb when nlmsg_free should be used. Fixes: 2ca546b92a02 ('IB/sa: Route SA pathrecord query through netlink') Signed-off-by: Mark Bloch <markb@mellanox.com> Reviewed-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Steve Wise <swise@opengridcomputing.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-05-07 03:45:27 +08:00
nlmsg_free(skb);
IB/sa: Route SA pathrecord query through netlink This patch routes a SA pathrecord query to netlink first and processes the response appropriately. If a failure is returned, the request will be sent through IB. The decision whether to route the request to netlink first is determined by the presence of a listener for the local service netlink multicast group. If the user-space local service netlink multicast group listener is not present, the request will be sent through IB, just like what is currently being done. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: John Fleck <john.fleck@intel.com> Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-14 20:52:09 +08:00
return -EMSGSIZE;
}
/* Add attributes */
ib_nl_set_path_rec_attrs(skb, query);
/* Repair the nlmsg header length */
nlmsg_end(skb, nlh);
IB/sa: Resolv use-after-free in ib_nl_make_request() There is a race condition where ib_nl_make_request() inserts the request data into the linked list but the timer in ib_nl_request_timeout() can see it and destroy it before ib_nl_send_msg() is done touching it. This could happen, for instance, if there is a long delay allocating memory during nlmsg_new() This causes a use-after-free in the send_mad() thread: [<ffffffffa02f43cb>] ? ib_pack+0x17b/0x240 [ib_core] [ <ffffffffa032aef1>] ib_sa_path_rec_get+0x181/0x200 [ib_sa] [<ffffffffa0379db0>] rdma_resolve_route+0x3c0/0x8d0 [rdma_cm] [<ffffffffa0374450>] ? cma_bind_port+0xa0/0xa0 [rdma_cm] [<ffffffffa040f850>] ? rds_rdma_cm_event_handler_cmn+0x850/0x850 [rds_rdma] [<ffffffffa040f22c>] rds_rdma_cm_event_handler_cmn+0x22c/0x850 [rds_rdma] [<ffffffffa040f860>] rds_rdma_cm_event_handler+0x10/0x20 [rds_rdma] [<ffffffffa037778e>] addr_handler+0x9e/0x140 [rdma_cm] [<ffffffffa026cdb4>] process_req+0x134/0x190 [ib_addr] [<ffffffff810a02f9>] process_one_work+0x169/0x4a0 [<ffffffff810a0b2b>] worker_thread+0x5b/0x560 [<ffffffff810a0ad0>] ? flush_delayed_work+0x50/0x50 [<ffffffff810a68fb>] kthread+0xcb/0xf0 [<ffffffff816ec49a>] ? __schedule+0x24a/0x810 [<ffffffff816ec49a>] ? __schedule+0x24a/0x810 [<ffffffff810a6830>] ? kthread_create_on_node+0x180/0x180 [<ffffffff816f25a7>] ret_from_fork+0x47/0x90 [<ffffffff810a6830>] ? kthread_create_on_node+0x180/0x180 The ownership rule is once the request is on the list, ownership transfers to the list and the local thread can't touch it any more, just like for the normal MAD case in send_mad(). Thus, instead of adding before send and then trying to delete after on errors, move the entire thing under the spinlock so that the send and update of the lists are atomic to the conurrent threads. Lightly reoganize things so spinlock safe memory allocations are done in the final NL send path and the rest of the setup work is done before and outside the lock. Fixes: 3ebd2fd0d011 ("IB/sa: Put netlink request into the request list before sending") Link: https://lore.kernel.org/r/1592964789-14533-1-git-send-email-divya.indi@oracle.com Signed-off-by: Divya Indi <divya.indi@oracle.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
2020-06-24 10:13:09 +08:00
gfp_flag = ((gfp_mask & GFP_ATOMIC) == GFP_ATOMIC) ? GFP_ATOMIC :
GFP_NOWAIT;
IB/sa: Route SA pathrecord query through netlink This patch routes a SA pathrecord query to netlink first and processes the response appropriately. If a failure is returned, the request will be sent through IB. The decision whether to route the request to netlink first is determined by the presence of a listener for the local service netlink multicast group. If the user-space local service netlink multicast group listener is not present, the request will be sent through IB, just like what is currently being done. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: John Fleck <john.fleck@intel.com> Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-14 20:52:09 +08:00
IB/sa: Resolv use-after-free in ib_nl_make_request() There is a race condition where ib_nl_make_request() inserts the request data into the linked list but the timer in ib_nl_request_timeout() can see it and destroy it before ib_nl_send_msg() is done touching it. This could happen, for instance, if there is a long delay allocating memory during nlmsg_new() This causes a use-after-free in the send_mad() thread: [<ffffffffa02f43cb>] ? ib_pack+0x17b/0x240 [ib_core] [ <ffffffffa032aef1>] ib_sa_path_rec_get+0x181/0x200 [ib_sa] [<ffffffffa0379db0>] rdma_resolve_route+0x3c0/0x8d0 [rdma_cm] [<ffffffffa0374450>] ? cma_bind_port+0xa0/0xa0 [rdma_cm] [<ffffffffa040f850>] ? rds_rdma_cm_event_handler_cmn+0x850/0x850 [rds_rdma] [<ffffffffa040f22c>] rds_rdma_cm_event_handler_cmn+0x22c/0x850 [rds_rdma] [<ffffffffa040f860>] rds_rdma_cm_event_handler+0x10/0x20 [rds_rdma] [<ffffffffa037778e>] addr_handler+0x9e/0x140 [rdma_cm] [<ffffffffa026cdb4>] process_req+0x134/0x190 [ib_addr] [<ffffffff810a02f9>] process_one_work+0x169/0x4a0 [<ffffffff810a0b2b>] worker_thread+0x5b/0x560 [<ffffffff810a0ad0>] ? flush_delayed_work+0x50/0x50 [<ffffffff810a68fb>] kthread+0xcb/0xf0 [<ffffffff816ec49a>] ? __schedule+0x24a/0x810 [<ffffffff816ec49a>] ? __schedule+0x24a/0x810 [<ffffffff810a6830>] ? kthread_create_on_node+0x180/0x180 [<ffffffff816f25a7>] ret_from_fork+0x47/0x90 [<ffffffff810a6830>] ? kthread_create_on_node+0x180/0x180 The ownership rule is once the request is on the list, ownership transfers to the list and the local thread can't touch it any more, just like for the normal MAD case in send_mad(). Thus, instead of adding before send and then trying to delete after on errors, move the entire thing under the spinlock so that the send and update of the lists are atomic to the conurrent threads. Lightly reoganize things so spinlock safe memory allocations are done in the final NL send path and the rest of the setup work is done before and outside the lock. Fixes: 3ebd2fd0d011 ("IB/sa: Put netlink request into the request list before sending") Link: https://lore.kernel.org/r/1592964789-14533-1-git-send-email-divya.indi@oracle.com Signed-off-by: Divya Indi <divya.indi@oracle.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
2020-06-24 10:13:09 +08:00
spin_lock_irqsave(&ib_nl_request_lock, flags);
ret = rdma_nl_multicast(&init_net, skb, RDMA_NL_GROUP_LS, gfp_flag);
IB/sa: Route SA pathrecord query through netlink This patch routes a SA pathrecord query to netlink first and processes the response appropriately. If a failure is returned, the request will be sent through IB. The decision whether to route the request to netlink first is determined by the presence of a listener for the local service netlink multicast group. If the user-space local service netlink multicast group listener is not present, the request will be sent through IB, just like what is currently being done. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: John Fleck <john.fleck@intel.com> Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-14 20:52:09 +08:00
IB/sa: Resolv use-after-free in ib_nl_make_request() There is a race condition where ib_nl_make_request() inserts the request data into the linked list but the timer in ib_nl_request_timeout() can see it and destroy it before ib_nl_send_msg() is done touching it. This could happen, for instance, if there is a long delay allocating memory during nlmsg_new() This causes a use-after-free in the send_mad() thread: [<ffffffffa02f43cb>] ? ib_pack+0x17b/0x240 [ib_core] [ <ffffffffa032aef1>] ib_sa_path_rec_get+0x181/0x200 [ib_sa] [<ffffffffa0379db0>] rdma_resolve_route+0x3c0/0x8d0 [rdma_cm] [<ffffffffa0374450>] ? cma_bind_port+0xa0/0xa0 [rdma_cm] [<ffffffffa040f850>] ? rds_rdma_cm_event_handler_cmn+0x850/0x850 [rds_rdma] [<ffffffffa040f22c>] rds_rdma_cm_event_handler_cmn+0x22c/0x850 [rds_rdma] [<ffffffffa040f860>] rds_rdma_cm_event_handler+0x10/0x20 [rds_rdma] [<ffffffffa037778e>] addr_handler+0x9e/0x140 [rdma_cm] [<ffffffffa026cdb4>] process_req+0x134/0x190 [ib_addr] [<ffffffff810a02f9>] process_one_work+0x169/0x4a0 [<ffffffff810a0b2b>] worker_thread+0x5b/0x560 [<ffffffff810a0ad0>] ? flush_delayed_work+0x50/0x50 [<ffffffff810a68fb>] kthread+0xcb/0xf0 [<ffffffff816ec49a>] ? __schedule+0x24a/0x810 [<ffffffff816ec49a>] ? __schedule+0x24a/0x810 [<ffffffff810a6830>] ? kthread_create_on_node+0x180/0x180 [<ffffffff816f25a7>] ret_from_fork+0x47/0x90 [<ffffffff810a6830>] ? kthread_create_on_node+0x180/0x180 The ownership rule is once the request is on the list, ownership transfers to the list and the local thread can't touch it any more, just like for the normal MAD case in send_mad(). Thus, instead of adding before send and then trying to delete after on errors, move the entire thing under the spinlock so that the send and update of the lists are atomic to the conurrent threads. Lightly reoganize things so spinlock safe memory allocations are done in the final NL send path and the rest of the setup work is done before and outside the lock. Fixes: 3ebd2fd0d011 ("IB/sa: Put netlink request into the request list before sending") Link: https://lore.kernel.org/r/1592964789-14533-1-git-send-email-divya.indi@oracle.com Signed-off-by: Divya Indi <divya.indi@oracle.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
2020-06-24 10:13:09 +08:00
if (ret)
goto out;
IB/sa: Route SA pathrecord query through netlink This patch routes a SA pathrecord query to netlink first and processes the response appropriately. If a failure is returned, the request will be sent through IB. The decision whether to route the request to netlink first is determined by the presence of a listener for the local service netlink multicast group. If the user-space local service netlink multicast group listener is not present, the request will be sent through IB, just like what is currently being done. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: John Fleck <john.fleck@intel.com> Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-14 20:52:09 +08:00
IB/sa: Resolv use-after-free in ib_nl_make_request() There is a race condition where ib_nl_make_request() inserts the request data into the linked list but the timer in ib_nl_request_timeout() can see it and destroy it before ib_nl_send_msg() is done touching it. This could happen, for instance, if there is a long delay allocating memory during nlmsg_new() This causes a use-after-free in the send_mad() thread: [<ffffffffa02f43cb>] ? ib_pack+0x17b/0x240 [ib_core] [ <ffffffffa032aef1>] ib_sa_path_rec_get+0x181/0x200 [ib_sa] [<ffffffffa0379db0>] rdma_resolve_route+0x3c0/0x8d0 [rdma_cm] [<ffffffffa0374450>] ? cma_bind_port+0xa0/0xa0 [rdma_cm] [<ffffffffa040f850>] ? rds_rdma_cm_event_handler_cmn+0x850/0x850 [rds_rdma] [<ffffffffa040f22c>] rds_rdma_cm_event_handler_cmn+0x22c/0x850 [rds_rdma] [<ffffffffa040f860>] rds_rdma_cm_event_handler+0x10/0x20 [rds_rdma] [<ffffffffa037778e>] addr_handler+0x9e/0x140 [rdma_cm] [<ffffffffa026cdb4>] process_req+0x134/0x190 [ib_addr] [<ffffffff810a02f9>] process_one_work+0x169/0x4a0 [<ffffffff810a0b2b>] worker_thread+0x5b/0x560 [<ffffffff810a0ad0>] ? flush_delayed_work+0x50/0x50 [<ffffffff810a68fb>] kthread+0xcb/0xf0 [<ffffffff816ec49a>] ? __schedule+0x24a/0x810 [<ffffffff816ec49a>] ? __schedule+0x24a/0x810 [<ffffffff810a6830>] ? kthread_create_on_node+0x180/0x180 [<ffffffff816f25a7>] ret_from_fork+0x47/0x90 [<ffffffff810a6830>] ? kthread_create_on_node+0x180/0x180 The ownership rule is once the request is on the list, ownership transfers to the list and the local thread can't touch it any more, just like for the normal MAD case in send_mad(). Thus, instead of adding before send and then trying to delete after on errors, move the entire thing under the spinlock so that the send and update of the lists are atomic to the conurrent threads. Lightly reoganize things so spinlock safe memory allocations are done in the final NL send path and the rest of the setup work is done before and outside the lock. Fixes: 3ebd2fd0d011 ("IB/sa: Put netlink request into the request list before sending") Link: https://lore.kernel.org/r/1592964789-14533-1-git-send-email-divya.indi@oracle.com Signed-off-by: Divya Indi <divya.indi@oracle.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
2020-06-24 10:13:09 +08:00
/* Put the request on the list.*/
IB/sa: Route SA pathrecord query through netlink This patch routes a SA pathrecord query to netlink first and processes the response appropriately. If a failure is returned, the request will be sent through IB. The decision whether to route the request to netlink first is determined by the presence of a listener for the local service netlink multicast group. If the user-space local service netlink multicast group listener is not present, the request will be sent through IB, just like what is currently being done. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: John Fleck <john.fleck@intel.com> Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-14 20:52:09 +08:00
delay = msecs_to_jiffies(sa_local_svc_timeout_ms);
query->timeout = delay + jiffies;
list_add_tail(&query->list, &ib_nl_request_list);
/* Start the timeout if this is the only request */
if (ib_nl_request_list.next == &query->list)
queue_delayed_work(ib_nl_wq, &ib_nl_timed_work, delay);
IB/sa: Resolv use-after-free in ib_nl_make_request() There is a race condition where ib_nl_make_request() inserts the request data into the linked list but the timer in ib_nl_request_timeout() can see it and destroy it before ib_nl_send_msg() is done touching it. This could happen, for instance, if there is a long delay allocating memory during nlmsg_new() This causes a use-after-free in the send_mad() thread: [<ffffffffa02f43cb>] ? ib_pack+0x17b/0x240 [ib_core] [ <ffffffffa032aef1>] ib_sa_path_rec_get+0x181/0x200 [ib_sa] [<ffffffffa0379db0>] rdma_resolve_route+0x3c0/0x8d0 [rdma_cm] [<ffffffffa0374450>] ? cma_bind_port+0xa0/0xa0 [rdma_cm] [<ffffffffa040f850>] ? rds_rdma_cm_event_handler_cmn+0x850/0x850 [rds_rdma] [<ffffffffa040f22c>] rds_rdma_cm_event_handler_cmn+0x22c/0x850 [rds_rdma] [<ffffffffa040f860>] rds_rdma_cm_event_handler+0x10/0x20 [rds_rdma] [<ffffffffa037778e>] addr_handler+0x9e/0x140 [rdma_cm] [<ffffffffa026cdb4>] process_req+0x134/0x190 [ib_addr] [<ffffffff810a02f9>] process_one_work+0x169/0x4a0 [<ffffffff810a0b2b>] worker_thread+0x5b/0x560 [<ffffffff810a0ad0>] ? flush_delayed_work+0x50/0x50 [<ffffffff810a68fb>] kthread+0xcb/0xf0 [<ffffffff816ec49a>] ? __schedule+0x24a/0x810 [<ffffffff816ec49a>] ? __schedule+0x24a/0x810 [<ffffffff810a6830>] ? kthread_create_on_node+0x180/0x180 [<ffffffff816f25a7>] ret_from_fork+0x47/0x90 [<ffffffff810a6830>] ? kthread_create_on_node+0x180/0x180 The ownership rule is once the request is on the list, ownership transfers to the list and the local thread can't touch it any more, just like for the normal MAD case in send_mad(). Thus, instead of adding before send and then trying to delete after on errors, move the entire thing under the spinlock so that the send and update of the lists are atomic to the conurrent threads. Lightly reoganize things so spinlock safe memory allocations are done in the final NL send path and the rest of the setup work is done before and outside the lock. Fixes: 3ebd2fd0d011 ("IB/sa: Put netlink request into the request list before sending") Link: https://lore.kernel.org/r/1592964789-14533-1-git-send-email-divya.indi@oracle.com Signed-off-by: Divya Indi <divya.indi@oracle.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
2020-06-24 10:13:09 +08:00
out:
spin_unlock_irqrestore(&ib_nl_request_lock, flags);
IB/sa: Put netlink request into the request list before sending It was found by Saurabh Sengar that the netlink code tried to allocate memory with GFP_KERNEL while holding a spinlock. While it is possible to fix the issue by replacing GFP_KERNEL with GFP_ATOMIC, it is better to get rid of the spinlock while sending the packet. However, in order to protect against a race condition that a quick response may be received before the request is put on the request list, we need to put the request on the list first. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reported-by: Saurabh Sengar <saurabh.truth@gmail.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-10-30 20:23:45 +08:00
IB/sa: Route SA pathrecord query through netlink This patch routes a SA pathrecord query to netlink first and processes the response appropriately. If a failure is returned, the request will be sent through IB. The decision whether to route the request to netlink first is determined by the presence of a listener for the local service netlink multicast group. If the user-space local service netlink multicast group listener is not present, the request will be sent through IB, just like what is currently being done. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: John Fleck <john.fleck@intel.com> Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-14 20:52:09 +08:00
return ret;
}
static int ib_nl_cancel_request(struct ib_sa_query *query)
{
unsigned long flags;
struct ib_sa_query *wait_query;
int found = 0;
spin_lock_irqsave(&ib_nl_request_lock, flags);
list_for_each_entry(wait_query, &ib_nl_request_list, list) {
/* Let the timeout to take care of the callback */
if (query == wait_query) {
query->flags |= IB_SA_CANCEL;
query->timeout = jiffies;
list_move(&query->list, &ib_nl_request_list);
found = 1;
mod_delayed_work(ib_nl_wq, &ib_nl_timed_work, 1);
break;
}
}
spin_unlock_irqrestore(&ib_nl_request_lock, flags);
return found;
}
static void send_handler(struct ib_mad_agent *agent,
struct ib_mad_send_wc *mad_send_wc);
static void ib_nl_process_good_resolve_rsp(struct ib_sa_query *query,
const struct nlmsghdr *nlh)
{
RDMA/cma: Refactor the inbound/outbound path records process flow Refactors based on comments [1] of the multiple path records support patchset: - Return failure if not able to set inbound/outbound PRs; - Simplify the flow when receiving the PRs from netlink channel: When a good PR response is received, unpack it and call the path_query callback directly. This saves two memory allocations; - Define RDMA_PRIMARY_PATH_MAX_REC_NUM in a proper place. [1] https://lore.kernel.org/linux-rdma/Yyxp9E9pJtUids2o@nvidia.com/ Signed-off-by: Mark Zhang <markzhang@nvidia.com> Reviewed-by: Bart Van Assche <bvanassche@acm.org> #srp Link: https://lore.kernel.org/r/7610025d57342b8b6da0f19516c9612f9c3fdc37.1672819376.git.leonro@nvidia.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2023-01-04 16:03:41 +08:00
struct sa_path_rec recs[RDMA_PRIMARY_PATH_MAX_REC_NUM];
RDMA/cma: Multiple path records support with netlink channel Support receiving inbound and outbound IB path records (along with GMP PathRecord) from user-space service through the RDMA netlink channel. The LIDs in these 3 PRs can be used in this way: 1. GMP PR: used as the standard local/remote LIDs; 2. DLID of outbound PR: Used as the "dlid" field for outbound traffic; 3. DLID of inbound PR: Used as the "dlid" field for outbound traffic in responder side. This is aimed to support adaptive routing. With current IB routing solution when a packet goes out it's assigned with a fixed DLID per target, meaning a fixed router will be used. The LIDs in inbound/outbound path records can be used to identify group of routers that allow communication with another subnet's entity. With them packets from an inter-subnet connection may travel through any router in the set to reach the target. As confirmed with Jason, when sending a netlink request, kernel uses LS_RESOLVE_PATH_USE_ALL so that the service knows kernel supports multiple PRs. Signed-off-by: Mark Zhang <markzhang@nvidia.com> Reviewed-by: Mark Bloch <mbloch@nvidia.com> Link: https://lore.kernel.org/r/2fa2b6c93c4c16c8915bac3cfc4f27be1d60519d.1662631201.git.leonro@nvidia.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2022-09-08 18:09:01 +08:00
struct ib_sa_path_query *path_query;
RDMA/cma: Refactor the inbound/outbound path records process flow Refactors based on comments [1] of the multiple path records support patchset: - Return failure if not able to set inbound/outbound PRs; - Simplify the flow when receiving the PRs from netlink channel: When a good PR response is received, unpack it and call the path_query callback directly. This saves two memory allocations; - Define RDMA_PRIMARY_PATH_MAX_REC_NUM in a proper place. [1] https://lore.kernel.org/linux-rdma/Yyxp9E9pJtUids2o@nvidia.com/ Signed-off-by: Mark Zhang <markzhang@nvidia.com> Reviewed-by: Bart Van Assche <bvanassche@acm.org> #srp Link: https://lore.kernel.org/r/7610025d57342b8b6da0f19516c9612f9c3fdc37.1672819376.git.leonro@nvidia.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2023-01-04 16:03:41 +08:00
struct ib_path_rec_data *rec_data;
IB/sa: Route SA pathrecord query through netlink This patch routes a SA pathrecord query to netlink first and processes the response appropriately. If a failure is returned, the request will be sent through IB. The decision whether to route the request to netlink first is determined by the presence of a listener for the local service netlink multicast group. If the user-space local service netlink multicast group listener is not present, the request will be sent through IB, just like what is currently being done. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: John Fleck <john.fleck@intel.com> Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-14 20:52:09 +08:00
struct ib_mad_send_wc mad_send_wc;
const struct nlattr *head, *curr;
RDMA/cma: Multiple path records support with netlink channel Support receiving inbound and outbound IB path records (along with GMP PathRecord) from user-space service through the RDMA netlink channel. The LIDs in these 3 PRs can be used in this way: 1. GMP PR: used as the standard local/remote LIDs; 2. DLID of outbound PR: Used as the "dlid" field for outbound traffic; 3. DLID of inbound PR: Used as the "dlid" field for outbound traffic in responder side. This is aimed to support adaptive routing. With current IB routing solution when a packet goes out it's assigned with a fixed DLID per target, meaning a fixed router will be used. The LIDs in inbound/outbound path records can be used to identify group of routers that allow communication with another subnet's entity. With them packets from an inter-subnet connection may travel through any router in the set to reach the target. As confirmed with Jason, when sending a netlink request, kernel uses LS_RESOLVE_PATH_USE_ALL so that the service knows kernel supports multiple PRs. Signed-off-by: Mark Zhang <markzhang@nvidia.com> Reviewed-by: Mark Bloch <mbloch@nvidia.com> Link: https://lore.kernel.org/r/2fa2b6c93c4c16c8915bac3cfc4f27be1d60519d.1662631201.git.leonro@nvidia.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2022-09-08 18:09:01 +08:00
struct ib_sa_mad *mad = NULL;
RDMA/cma: Refactor the inbound/outbound path records process flow Refactors based on comments [1] of the multiple path records support patchset: - Return failure if not able to set inbound/outbound PRs; - Simplify the flow when receiving the PRs from netlink channel: When a good PR response is received, unpack it and call the path_query callback directly. This saves two memory allocations; - Define RDMA_PRIMARY_PATH_MAX_REC_NUM in a proper place. [1] https://lore.kernel.org/linux-rdma/Yyxp9E9pJtUids2o@nvidia.com/ Signed-off-by: Mark Zhang <markzhang@nvidia.com> Reviewed-by: Bart Van Assche <bvanassche@acm.org> #srp Link: https://lore.kernel.org/r/7610025d57342b8b6da0f19516c9612f9c3fdc37.1672819376.git.leonro@nvidia.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2023-01-04 16:03:41 +08:00
int len, rem, status = -EIO;
unsigned int num_prs = 0;
IB/sa: Route SA pathrecord query through netlink This patch routes a SA pathrecord query to netlink first and processes the response appropriately. If a failure is returned, the request will be sent through IB. The decision whether to route the request to netlink first is determined by the presence of a listener for the local service netlink multicast group. If the user-space local service netlink multicast group listener is not present, the request will be sent through IB, just like what is currently being done. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: John Fleck <john.fleck@intel.com> Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-14 20:52:09 +08:00
u32 mask = 0;
RDMA/cma: Multiple path records support with netlink channel Support receiving inbound and outbound IB path records (along with GMP PathRecord) from user-space service through the RDMA netlink channel. The LIDs in these 3 PRs can be used in this way: 1. GMP PR: used as the standard local/remote LIDs; 2. DLID of outbound PR: Used as the "dlid" field for outbound traffic; 3. DLID of inbound PR: Used as the "dlid" field for outbound traffic in responder side. This is aimed to support adaptive routing. With current IB routing solution when a packet goes out it's assigned with a fixed DLID per target, meaning a fixed router will be used. The LIDs in inbound/outbound path records can be used to identify group of routers that allow communication with another subnet's entity. With them packets from an inter-subnet connection may travel through any router in the set to reach the target. As confirmed with Jason, when sending a netlink request, kernel uses LS_RESOLVE_PATH_USE_ALL so that the service knows kernel supports multiple PRs. Signed-off-by: Mark Zhang <markzhang@nvidia.com> Reviewed-by: Mark Bloch <mbloch@nvidia.com> Link: https://lore.kernel.org/r/2fa2b6c93c4c16c8915bac3cfc4f27be1d60519d.1662631201.git.leonro@nvidia.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2022-09-08 18:09:01 +08:00
if (!query->callback)
goto out;
IB/sa: Route SA pathrecord query through netlink This patch routes a SA pathrecord query to netlink first and processes the response appropriately. If a failure is returned, the request will be sent through IB. The decision whether to route the request to netlink first is determined by the presence of a listener for the local service netlink multicast group. If the user-space local service netlink multicast group listener is not present, the request will be sent through IB, just like what is currently being done. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: John Fleck <john.fleck@intel.com> Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-14 20:52:09 +08:00
RDMA/cma: Multiple path records support with netlink channel Support receiving inbound and outbound IB path records (along with GMP PathRecord) from user-space service through the RDMA netlink channel. The LIDs in these 3 PRs can be used in this way: 1. GMP PR: used as the standard local/remote LIDs; 2. DLID of outbound PR: Used as the "dlid" field for outbound traffic; 3. DLID of inbound PR: Used as the "dlid" field for outbound traffic in responder side. This is aimed to support adaptive routing. With current IB routing solution when a packet goes out it's assigned with a fixed DLID per target, meaning a fixed router will be used. The LIDs in inbound/outbound path records can be used to identify group of routers that allow communication with another subnet's entity. With them packets from an inter-subnet connection may travel through any router in the set to reach the target. As confirmed with Jason, when sending a netlink request, kernel uses LS_RESOLVE_PATH_USE_ALL so that the service knows kernel supports multiple PRs. Signed-off-by: Mark Zhang <markzhang@nvidia.com> Reviewed-by: Mark Bloch <mbloch@nvidia.com> Link: https://lore.kernel.org/r/2fa2b6c93c4c16c8915bac3cfc4f27be1d60519d.1662631201.git.leonro@nvidia.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2022-09-08 18:09:01 +08:00
path_query = container_of(query, struct ib_sa_path_query, sa_query);
mad = query->mad_buf->mad;
head = (const struct nlattr *) nlmsg_data(nlh);
len = nlmsg_len(nlh);
switch (query->path_use) {
case LS_RESOLVE_PATH_USE_UNIDIRECTIONAL:
mask = IB_PATH_PRIMARY | IB_PATH_OUTBOUND;
break;
case LS_RESOLVE_PATH_USE_ALL:
mask = IB_PATH_PRIMARY;
break;
case LS_RESOLVE_PATH_USE_GMP:
default:
mask = IB_PATH_PRIMARY | IB_PATH_GMP |
IB_PATH_BIDIRECTIONAL;
break;
}
nla_for_each_attr(curr, head, len, rem) {
if (curr->nla_type != LS_NLA_TYPE_PATH_RECORD)
continue;
RDMA/cma: Refactor the inbound/outbound path records process flow Refactors based on comments [1] of the multiple path records support patchset: - Return failure if not able to set inbound/outbound PRs; - Simplify the flow when receiving the PRs from netlink channel: When a good PR response is received, unpack it and call the path_query callback directly. This saves two memory allocations; - Define RDMA_PRIMARY_PATH_MAX_REC_NUM in a proper place. [1] https://lore.kernel.org/linux-rdma/Yyxp9E9pJtUids2o@nvidia.com/ Signed-off-by: Mark Zhang <markzhang@nvidia.com> Reviewed-by: Bart Van Assche <bvanassche@acm.org> #srp Link: https://lore.kernel.org/r/7610025d57342b8b6da0f19516c9612f9c3fdc37.1672819376.git.leonro@nvidia.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2023-01-04 16:03:41 +08:00
rec_data = nla_data(curr);
if ((rec_data->flags & mask) != mask)
RDMA/cma: Multiple path records support with netlink channel Support receiving inbound and outbound IB path records (along with GMP PathRecord) from user-space service through the RDMA netlink channel. The LIDs in these 3 PRs can be used in this way: 1. GMP PR: used as the standard local/remote LIDs; 2. DLID of outbound PR: Used as the "dlid" field for outbound traffic; 3. DLID of inbound PR: Used as the "dlid" field for outbound traffic in responder side. This is aimed to support adaptive routing. With current IB routing solution when a packet goes out it's assigned with a fixed DLID per target, meaning a fixed router will be used. The LIDs in inbound/outbound path records can be used to identify group of routers that allow communication with another subnet's entity. With them packets from an inter-subnet connection may travel through any router in the set to reach the target. As confirmed with Jason, when sending a netlink request, kernel uses LS_RESOLVE_PATH_USE_ALL so that the service knows kernel supports multiple PRs. Signed-off-by: Mark Zhang <markzhang@nvidia.com> Reviewed-by: Mark Bloch <mbloch@nvidia.com> Link: https://lore.kernel.org/r/2fa2b6c93c4c16c8915bac3cfc4f27be1d60519d.1662631201.git.leonro@nvidia.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2022-09-08 18:09:01 +08:00
continue;
RDMA/cma: Refactor the inbound/outbound path records process flow Refactors based on comments [1] of the multiple path records support patchset: - Return failure if not able to set inbound/outbound PRs; - Simplify the flow when receiving the PRs from netlink channel: When a good PR response is received, unpack it and call the path_query callback directly. This saves two memory allocations; - Define RDMA_PRIMARY_PATH_MAX_REC_NUM in a proper place. [1] https://lore.kernel.org/linux-rdma/Yyxp9E9pJtUids2o@nvidia.com/ Signed-off-by: Mark Zhang <markzhang@nvidia.com> Reviewed-by: Bart Van Assche <bvanassche@acm.org> #srp Link: https://lore.kernel.org/r/7610025d57342b8b6da0f19516c9612f9c3fdc37.1672819376.git.leonro@nvidia.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2023-01-04 16:03:41 +08:00
if ((query->flags & IB_SA_QUERY_OPA) ||
path_query->conv_pr) {
mad->mad_hdr.method |= IB_MGMT_METHOD_RESP;
memcpy(mad->data, rec_data->path_rec,
sizeof(rec_data->path_rec));
query->callback(query, 0, mad);
goto out;
IB/sa: Route SA pathrecord query through netlink This patch routes a SA pathrecord query to netlink first and processes the response appropriately. If a failure is returned, the request will be sent through IB. The decision whether to route the request to netlink first is determined by the presence of a listener for the local service netlink multicast group. If the user-space local service netlink multicast group listener is not present, the request will be sent through IB, just like what is currently being done. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: John Fleck <john.fleck@intel.com> Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-14 20:52:09 +08:00
}
RDMA/cma: Multiple path records support with netlink channel Support receiving inbound and outbound IB path records (along with GMP PathRecord) from user-space service through the RDMA netlink channel. The LIDs in these 3 PRs can be used in this way: 1. GMP PR: used as the standard local/remote LIDs; 2. DLID of outbound PR: Used as the "dlid" field for outbound traffic; 3. DLID of inbound PR: Used as the "dlid" field for outbound traffic in responder side. This is aimed to support adaptive routing. With current IB routing solution when a packet goes out it's assigned with a fixed DLID per target, meaning a fixed router will be used. The LIDs in inbound/outbound path records can be used to identify group of routers that allow communication with another subnet's entity. With them packets from an inter-subnet connection may travel through any router in the set to reach the target. As confirmed with Jason, when sending a netlink request, kernel uses LS_RESOLVE_PATH_USE_ALL so that the service knows kernel supports multiple PRs. Signed-off-by: Mark Zhang <markzhang@nvidia.com> Reviewed-by: Mark Bloch <mbloch@nvidia.com> Link: https://lore.kernel.org/r/2fa2b6c93c4c16c8915bac3cfc4f27be1d60519d.1662631201.git.leonro@nvidia.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2022-09-08 18:09:01 +08:00
RDMA/cma: Refactor the inbound/outbound path records process flow Refactors based on comments [1] of the multiple path records support patchset: - Return failure if not able to set inbound/outbound PRs; - Simplify the flow when receiving the PRs from netlink channel: When a good PR response is received, unpack it and call the path_query callback directly. This saves two memory allocations; - Define RDMA_PRIMARY_PATH_MAX_REC_NUM in a proper place. [1] https://lore.kernel.org/linux-rdma/Yyxp9E9pJtUids2o@nvidia.com/ Signed-off-by: Mark Zhang <markzhang@nvidia.com> Reviewed-by: Bart Van Assche <bvanassche@acm.org> #srp Link: https://lore.kernel.org/r/7610025d57342b8b6da0f19516c9612f9c3fdc37.1672819376.git.leonro@nvidia.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2023-01-04 16:03:41 +08:00
status = 0;
ib_unpack(path_rec_table, ARRAY_SIZE(path_rec_table),
rec_data->path_rec, &recs[num_prs]);
recs[num_prs].flags = rec_data->flags;
recs[num_prs].rec_type = SA_PATH_REC_TYPE_IB;
sa_path_set_dmac_zero(&recs[num_prs]);
RDMA/cma: Multiple path records support with netlink channel Support receiving inbound and outbound IB path records (along with GMP PathRecord) from user-space service through the RDMA netlink channel. The LIDs in these 3 PRs can be used in this way: 1. GMP PR: used as the standard local/remote LIDs; 2. DLID of outbound PR: Used as the "dlid" field for outbound traffic; 3. DLID of inbound PR: Used as the "dlid" field for outbound traffic in responder side. This is aimed to support adaptive routing. With current IB routing solution when a packet goes out it's assigned with a fixed DLID per target, meaning a fixed router will be used. The LIDs in inbound/outbound path records can be used to identify group of routers that allow communication with another subnet's entity. With them packets from an inter-subnet connection may travel through any router in the set to reach the target. As confirmed with Jason, when sending a netlink request, kernel uses LS_RESOLVE_PATH_USE_ALL so that the service knows kernel supports multiple PRs. Signed-off-by: Mark Zhang <markzhang@nvidia.com> Reviewed-by: Mark Bloch <mbloch@nvidia.com> Link: https://lore.kernel.org/r/2fa2b6c93c4c16c8915bac3cfc4f27be1d60519d.1662631201.git.leonro@nvidia.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2022-09-08 18:09:01 +08:00
num_prs++;
if (num_prs >= RDMA_PRIMARY_PATH_MAX_REC_NUM)
break;
IB/sa: Route SA pathrecord query through netlink This patch routes a SA pathrecord query to netlink first and processes the response appropriately. If a failure is returned, the request will be sent through IB. The decision whether to route the request to netlink first is determined by the presence of a listener for the local service netlink multicast group. If the user-space local service netlink multicast group listener is not present, the request will be sent through IB, just like what is currently being done. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: John Fleck <john.fleck@intel.com> Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-14 20:52:09 +08:00
}
RDMA/cma: Refactor the inbound/outbound path records process flow Refactors based on comments [1] of the multiple path records support patchset: - Return failure if not able to set inbound/outbound PRs; - Simplify the flow when receiving the PRs from netlink channel: When a good PR response is received, unpack it and call the path_query callback directly. This saves two memory allocations; - Define RDMA_PRIMARY_PATH_MAX_REC_NUM in a proper place. [1] https://lore.kernel.org/linux-rdma/Yyxp9E9pJtUids2o@nvidia.com/ Signed-off-by: Mark Zhang <markzhang@nvidia.com> Reviewed-by: Bart Van Assche <bvanassche@acm.org> #srp Link: https://lore.kernel.org/r/7610025d57342b8b6da0f19516c9612f9c3fdc37.1672819376.git.leonro@nvidia.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2023-01-04 16:03:41 +08:00
if (!status) {
RDMA/cma: Multiple path records support with netlink channel Support receiving inbound and outbound IB path records (along with GMP PathRecord) from user-space service through the RDMA netlink channel. The LIDs in these 3 PRs can be used in this way: 1. GMP PR: used as the standard local/remote LIDs; 2. DLID of outbound PR: Used as the "dlid" field for outbound traffic; 3. DLID of inbound PR: Used as the "dlid" field for outbound traffic in responder side. This is aimed to support adaptive routing. With current IB routing solution when a packet goes out it's assigned with a fixed DLID per target, meaning a fixed router will be used. The LIDs in inbound/outbound path records can be used to identify group of routers that allow communication with another subnet's entity. With them packets from an inter-subnet connection may travel through any router in the set to reach the target. As confirmed with Jason, when sending a netlink request, kernel uses LS_RESOLVE_PATH_USE_ALL so that the service knows kernel supports multiple PRs. Signed-off-by: Mark Zhang <markzhang@nvidia.com> Reviewed-by: Mark Bloch <mbloch@nvidia.com> Link: https://lore.kernel.org/r/2fa2b6c93c4c16c8915bac3cfc4f27be1d60519d.1662631201.git.leonro@nvidia.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2022-09-08 18:09:01 +08:00
mad->mad_hdr.method |= IB_MGMT_METHOD_RESP;
RDMA/cma: Refactor the inbound/outbound path records process flow Refactors based on comments [1] of the multiple path records support patchset: - Return failure if not able to set inbound/outbound PRs; - Simplify the flow when receiving the PRs from netlink channel: When a good PR response is received, unpack it and call the path_query callback directly. This saves two memory allocations; - Define RDMA_PRIMARY_PATH_MAX_REC_NUM in a proper place. [1] https://lore.kernel.org/linux-rdma/Yyxp9E9pJtUids2o@nvidia.com/ Signed-off-by: Mark Zhang <markzhang@nvidia.com> Reviewed-by: Bart Van Assche <bvanassche@acm.org> #srp Link: https://lore.kernel.org/r/7610025d57342b8b6da0f19516c9612f9c3fdc37.1672819376.git.leonro@nvidia.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2023-01-04 16:03:41 +08:00
path_query->callback(status, recs, num_prs,
path_query->context);
} else
query->callback(query, status, mad);
RDMA/cma: Multiple path records support with netlink channel Support receiving inbound and outbound IB path records (along with GMP PathRecord) from user-space service through the RDMA netlink channel. The LIDs in these 3 PRs can be used in this way: 1. GMP PR: used as the standard local/remote LIDs; 2. DLID of outbound PR: Used as the "dlid" field for outbound traffic; 3. DLID of inbound PR: Used as the "dlid" field for outbound traffic in responder side. This is aimed to support adaptive routing. With current IB routing solution when a packet goes out it's assigned with a fixed DLID per target, meaning a fixed router will be used. The LIDs in inbound/outbound path records can be used to identify group of routers that allow communication with another subnet's entity. With them packets from an inter-subnet connection may travel through any router in the set to reach the target. As confirmed with Jason, when sending a netlink request, kernel uses LS_RESOLVE_PATH_USE_ALL so that the service knows kernel supports multiple PRs. Signed-off-by: Mark Zhang <markzhang@nvidia.com> Reviewed-by: Mark Bloch <mbloch@nvidia.com> Link: https://lore.kernel.org/r/2fa2b6c93c4c16c8915bac3cfc4f27be1d60519d.1662631201.git.leonro@nvidia.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2022-09-08 18:09:01 +08:00
out:
IB/sa: Route SA pathrecord query through netlink This patch routes a SA pathrecord query to netlink first and processes the response appropriately. If a failure is returned, the request will be sent through IB. The decision whether to route the request to netlink first is determined by the presence of a listener for the local service netlink multicast group. If the user-space local service netlink multicast group listener is not present, the request will be sent through IB, just like what is currently being done. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: John Fleck <john.fleck@intel.com> Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-14 20:52:09 +08:00
mad_send_wc.send_buf = query->mad_buf;
mad_send_wc.status = IB_WC_SUCCESS;
send_handler(query->mad_buf->mad_agent, &mad_send_wc);
}
static void ib_nl_request_timeout(struct work_struct *work)
{
unsigned long flags;
struct ib_sa_query *query;
unsigned long delay;
struct ib_mad_send_wc mad_send_wc;
int ret;
spin_lock_irqsave(&ib_nl_request_lock, flags);
while (!list_empty(&ib_nl_request_list)) {
query = list_entry(ib_nl_request_list.next,
struct ib_sa_query, list);
if (time_after(query->timeout, jiffies)) {
delay = query->timeout - jiffies;
if ((long)delay <= 0)
delay = 1;
queue_delayed_work(ib_nl_wq, &ib_nl_timed_work, delay);
break;
}
list_del(&query->list);
ib_sa_disable_local_svc(query);
/* Hold the lock to protect against query cancellation */
if (ib_sa_query_cancelled(query))
ret = -1;
else
ret = ib_post_send_mad(query->mad_buf, NULL);
if (ret) {
mad_send_wc.send_buf = query->mad_buf;
mad_send_wc.status = IB_WC_WR_FLUSH_ERR;
spin_unlock_irqrestore(&ib_nl_request_lock, flags);
send_handler(query->port->agent, &mad_send_wc);
spin_lock_irqsave(&ib_nl_request_lock, flags);
}
}
spin_unlock_irqrestore(&ib_nl_request_lock, flags);
}
IB/core: Register SA ibnl client during ib_core initialization Move SA ibnl client registration to ib_core module init. This will allow us to register a single client to handle all RDMA_NL_LS operations and make it SA independent. Signed-off-by: Mark Bloch <markb@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-05-19 22:12:35 +08:00
int ib_nl_handle_set_timeout(struct sk_buff *skb,
RDMA/netlink: Convert LS to doit callback RDMA_NL_LS protocol is actually does not dump anything, but sets data and it should be handled by doit callback. This patch actually converts RDMA_NL_LS to doit callback, while preserving IWCM and RDMA_CM flows through netlink_dump_start(). Signed-off-by: Leon Romanovsky <leonro@mellanox.com> Reviewed-by: Steve Wise <swise@opengridcomputing.com>
2017-06-15 19:20:39 +08:00
struct nlmsghdr *nlh,
struct netlink_ext_ack *extack)
IB/sa: Route SA pathrecord query through netlink This patch routes a SA pathrecord query to netlink first and processes the response appropriately. If a failure is returned, the request will be sent through IB. The decision whether to route the request to netlink first is determined by the presence of a listener for the local service netlink multicast group. If the user-space local service netlink multicast group listener is not present, the request will be sent through IB, just like what is currently being done. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: John Fleck <john.fleck@intel.com> Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-14 20:52:09 +08:00
{
int timeout, delta, abs_delta;
const struct nlattr *attr;
unsigned long flags;
struct ib_sa_query *query;
long delay = 0;
struct nlattr *tb[LS_NLA_TYPE_MAX];
int ret;
IB/sa: Fix netlink local service GFP crash The rdma netlink local service registers a handler to handle RESOLVE response and another handler to handle SET_TIMEOUT request. The first thing these handlers do is to call netlink_capable() to check the access right of the received skb to make sure that the sender has root access. Under normal conditions, such responses and requests will be directly forwarded to the handlers without going through the netlink_dump pathway (see ibnl_rcv_msg() in drivers/infiniband/core/netlink.c). However, a user application could send a RESOLVE request (not response) to the local service, which will fall into the netlink_dump pathway, where a new skb will be created without initializing the control block. This new skb will be eventually forwarded to the local service RESOLVE response handler. Unfortunately, netlink_capable() will cause general protection fault if the skb's control block is not initialized. This patch will address the problem by checking the skb first. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Reported-by: Dmitry Vyukov <dvyukov@google.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-01-21 21:41:31 +08:00
if (!(nlh->nlmsg_flags & NLM_F_REQUEST) ||
RDMA/netlink: Add flag to consolidate common handling Add ability to provide flags to control RDMA netlink callbacks and convert addr.c and sa_query.c to be first users of such infrastructure. It allows to move their CAP_NET_ADMIN checks into netlink core. Signed-off-by: Leon Romanovsky <leonro@mellanox.com> Reviewed-by: Steve Wise <swise@opengridcomputing.com>
2017-06-12 21:00:19 +08:00
!(NETLINK_CB(skb).sk))
IB/sa: Route SA pathrecord query through netlink This patch routes a SA pathrecord query to netlink first and processes the response appropriately. If a failure is returned, the request will be sent through IB. The decision whether to route the request to netlink first is determined by the presence of a listener for the local service netlink multicast group. If the user-space local service netlink multicast group listener is not present, the request will be sent through IB, just like what is currently being done. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: John Fleck <john.fleck@intel.com> Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-14 20:52:09 +08:00
return -EPERM;
netlink: make validation more configurable for future strictness We currently have two levels of strict validation: 1) liberal (default) - undefined (type >= max) & NLA_UNSPEC attributes accepted - attribute length >= expected accepted - garbage at end of message accepted 2) strict (opt-in) - NLA_UNSPEC attributes accepted - attribute length >= expected accepted Split out parsing strictness into four different options: * TRAILING - check that there's no trailing data after parsing attributes (in message or nested) * MAXTYPE - reject attrs > max known type * UNSPEC - reject attributes with NLA_UNSPEC policy entries * STRICT_ATTRS - strictly validate attribute size The default for future things should be *everything*. The current *_strict() is a combination of TRAILING and MAXTYPE, and is renamed to _deprecated_strict(). The current regular parsing has none of this, and is renamed to *_parse_deprecated(). Additionally it allows us to selectively set one of the new flags even on old policies. Notably, the UNSPEC flag could be useful in this case, since it can be arranged (by filling in the policy) to not be an incompatible userspace ABI change, but would then going forward prevent forgetting attribute entries. Similar can apply to the POLICY flag. We end up with the following renames: * nla_parse -> nla_parse_deprecated * nla_parse_strict -> nla_parse_deprecated_strict * nlmsg_parse -> nlmsg_parse_deprecated * nlmsg_parse_strict -> nlmsg_parse_deprecated_strict * nla_parse_nested -> nla_parse_nested_deprecated * nla_validate_nested -> nla_validate_nested_deprecated Using spatch, of course: @@ expression TB, MAX, HEAD, LEN, POL, EXT; @@ -nla_parse(TB, MAX, HEAD, LEN, POL, EXT) +nla_parse_deprecated(TB, MAX, HEAD, LEN, POL, EXT) @@ expression NLH, HDRLEN, TB, MAX, POL, EXT; @@ -nlmsg_parse(NLH, HDRLEN, TB, MAX, POL, EXT) +nlmsg_parse_deprecated(NLH, HDRLEN, TB, MAX, POL, EXT) @@ expression NLH, HDRLEN, TB, MAX, POL, EXT; @@ -nlmsg_parse_strict(NLH, HDRLEN, TB, MAX, POL, EXT) +nlmsg_parse_deprecated_strict(NLH, HDRLEN, TB, MAX, POL, EXT) @@ expression TB, MAX, NLA, POL, EXT; @@ -nla_parse_nested(TB, MAX, NLA, POL, EXT) +nla_parse_nested_deprecated(TB, MAX, NLA, POL, EXT) @@ expression START, MAX, POL, EXT; @@ -nla_validate_nested(START, MAX, POL, EXT) +nla_validate_nested_deprecated(START, MAX, POL, EXT) @@ expression NLH, HDRLEN, MAX, POL, EXT; @@ -nlmsg_validate(NLH, HDRLEN, MAX, POL, EXT) +nlmsg_validate_deprecated(NLH, HDRLEN, MAX, POL, EXT) For this patch, don't actually add the strict, non-renamed versions yet so that it breaks compile if I get it wrong. Also, while at it, make nla_validate and nla_parse go down to a common __nla_validate_parse() function to avoid code duplication. Ultimately, this allows us to have very strict validation for every new caller of nla_parse()/nlmsg_parse() etc as re-introduced in the next patch, while existing things will continue to work as is. In effect then, this adds fully strict validation for any new command. Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-04-26 20:07:28 +08:00
ret = nla_parse_deprecated(tb, LS_NLA_TYPE_MAX - 1, nlmsg_data(nlh),
nlmsg_len(nlh), ib_nl_policy, NULL);
IB/sa: Route SA pathrecord query through netlink This patch routes a SA pathrecord query to netlink first and processes the response appropriately. If a failure is returned, the request will be sent through IB. The decision whether to route the request to netlink first is determined by the presence of a listener for the local service netlink multicast group. If the user-space local service netlink multicast group listener is not present, the request will be sent through IB, just like what is currently being done. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: John Fleck <john.fleck@intel.com> Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-14 20:52:09 +08:00
attr = (const struct nlattr *)tb[LS_NLA_TYPE_TIMEOUT];
if (ret || !attr)
goto settimeout_out;
timeout = *(int *) nla_data(attr);
if (timeout < IB_SA_LOCAL_SVC_TIMEOUT_MIN)
timeout = IB_SA_LOCAL_SVC_TIMEOUT_MIN;
if (timeout > IB_SA_LOCAL_SVC_TIMEOUT_MAX)
timeout = IB_SA_LOCAL_SVC_TIMEOUT_MAX;
delta = timeout - sa_local_svc_timeout_ms;
if (delta < 0)
abs_delta = -delta;
else
abs_delta = delta;
if (delta != 0) {
spin_lock_irqsave(&ib_nl_request_lock, flags);
sa_local_svc_timeout_ms = timeout;
list_for_each_entry(query, &ib_nl_request_list, list) {
if (delta < 0 && abs_delta > query->timeout)
query->timeout = 0;
else
query->timeout += delta;
/* Get the new delay from the first entry */
if (!delay) {
delay = query->timeout - jiffies;
if (delay <= 0)
delay = 1;
}
}
if (delay)
mod_delayed_work(ib_nl_wq, &ib_nl_timed_work,
(unsigned long)delay);
spin_unlock_irqrestore(&ib_nl_request_lock, flags);
}
settimeout_out:
RDMA/netlink: Do not always generate an ACK for some netlink operations In rdma_nl_rcv_skb(), the local variable err is assigned the return value of the supplied callback function, which could be one of ib_nl_handle_resolve_resp(), ib_nl_handle_set_timeout(), or ib_nl_handle_ip_res_resp(). These three functions all return skb->len on success. rdma_nl_rcv_skb() is merely a copy of netlink_rcv_skb(). The callback functions used by the latter have the convention: "Returns 0 on success or a negative error code". In particular, the statement (equal for both functions): if (nlh->nlmsg_flags & NLM_F_ACK || err) implies that rdma_nl_rcv_skb() always will ack a message, independent of the NLM_F_ACK being set in nlmsg_flags or not. The fix could be to change the above statement, but it is better to keep the two *_rcv_skb() functions equal in this respect and instead change the three callback functions in the rdma subsystem to the correct convention. Fixes: 2ca546b92a02 ("IB/sa: Route SA pathrecord query through netlink") Fixes: ae43f8286730 ("IB/core: Add IP to GID netlink offload") Link: https://lore.kernel.org/r/20191216120436.3204814-1-haakon.bugge@oracle.com Suggested-by: Mark Haywood <mark.haywood@oracle.com> Signed-off-by: Håkon Bugge <haakon.bugge@oracle.com> Tested-by: Mark Haywood <mark.haywood@oracle.com> Reviewed-by: Leon Romanovsky <leonro@mellanox.com> Reviewed-by: Jason Gunthorpe <jgg@mellanox.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2019-12-16 20:04:36 +08:00
return 0;
IB/sa: Route SA pathrecord query through netlink This patch routes a SA pathrecord query to netlink first and processes the response appropriately. If a failure is returned, the request will be sent through IB. The decision whether to route the request to netlink first is determined by the presence of a listener for the local service netlink multicast group. If the user-space local service netlink multicast group listener is not present, the request will be sent through IB, just like what is currently being done. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: John Fleck <john.fleck@intel.com> Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-14 20:52:09 +08:00
}
static inline int ib_nl_is_good_resolve_resp(const struct nlmsghdr *nlh)
{
struct nlattr *tb[LS_NLA_TYPE_MAX];
int ret;
if (nlh->nlmsg_flags & RDMA_NL_LS_F_ERR)
return 0;
netlink: make validation more configurable for future strictness We currently have two levels of strict validation: 1) liberal (default) - undefined (type >= max) & NLA_UNSPEC attributes accepted - attribute length >= expected accepted - garbage at end of message accepted 2) strict (opt-in) - NLA_UNSPEC attributes accepted - attribute length >= expected accepted Split out parsing strictness into four different options: * TRAILING - check that there's no trailing data after parsing attributes (in message or nested) * MAXTYPE - reject attrs > max known type * UNSPEC - reject attributes with NLA_UNSPEC policy entries * STRICT_ATTRS - strictly validate attribute size The default for future things should be *everything*. The current *_strict() is a combination of TRAILING and MAXTYPE, and is renamed to _deprecated_strict(). The current regular parsing has none of this, and is renamed to *_parse_deprecated(). Additionally it allows us to selectively set one of the new flags even on old policies. Notably, the UNSPEC flag could be useful in this case, since it can be arranged (by filling in the policy) to not be an incompatible userspace ABI change, but would then going forward prevent forgetting attribute entries. Similar can apply to the POLICY flag. We end up with the following renames: * nla_parse -> nla_parse_deprecated * nla_parse_strict -> nla_parse_deprecated_strict * nlmsg_parse -> nlmsg_parse_deprecated * nlmsg_parse_strict -> nlmsg_parse_deprecated_strict * nla_parse_nested -> nla_parse_nested_deprecated * nla_validate_nested -> nla_validate_nested_deprecated Using spatch, of course: @@ expression TB, MAX, HEAD, LEN, POL, EXT; @@ -nla_parse(TB, MAX, HEAD, LEN, POL, EXT) +nla_parse_deprecated(TB, MAX, HEAD, LEN, POL, EXT) @@ expression NLH, HDRLEN, TB, MAX, POL, EXT; @@ -nlmsg_parse(NLH, HDRLEN, TB, MAX, POL, EXT) +nlmsg_parse_deprecated(NLH, HDRLEN, TB, MAX, POL, EXT) @@ expression NLH, HDRLEN, TB, MAX, POL, EXT; @@ -nlmsg_parse_strict(NLH, HDRLEN, TB, MAX, POL, EXT) +nlmsg_parse_deprecated_strict(NLH, HDRLEN, TB, MAX, POL, EXT) @@ expression TB, MAX, NLA, POL, EXT; @@ -nla_parse_nested(TB, MAX, NLA, POL, EXT) +nla_parse_nested_deprecated(TB, MAX, NLA, POL, EXT) @@ expression START, MAX, POL, EXT; @@ -nla_validate_nested(START, MAX, POL, EXT) +nla_validate_nested_deprecated(START, MAX, POL, EXT) @@ expression NLH, HDRLEN, MAX, POL, EXT; @@ -nlmsg_validate(NLH, HDRLEN, MAX, POL, EXT) +nlmsg_validate_deprecated(NLH, HDRLEN, MAX, POL, EXT) For this patch, don't actually add the strict, non-renamed versions yet so that it breaks compile if I get it wrong. Also, while at it, make nla_validate and nla_parse go down to a common __nla_validate_parse() function to avoid code duplication. Ultimately, this allows us to have very strict validation for every new caller of nla_parse()/nlmsg_parse() etc as re-introduced in the next patch, while existing things will continue to work as is. In effect then, this adds fully strict validation for any new command. Signed-off-by: Johannes Berg <johannes.berg@intel.com> Signed-off-by: David S. Miller <davem@davemloft.net>
2019-04-26 20:07:28 +08:00
ret = nla_parse_deprecated(tb, LS_NLA_TYPE_MAX - 1, nlmsg_data(nlh),
nlmsg_len(nlh), ib_nl_policy, NULL);
IB/sa: Route SA pathrecord query through netlink This patch routes a SA pathrecord query to netlink first and processes the response appropriately. If a failure is returned, the request will be sent through IB. The decision whether to route the request to netlink first is determined by the presence of a listener for the local service netlink multicast group. If the user-space local service netlink multicast group listener is not present, the request will be sent through IB, just like what is currently being done. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: John Fleck <john.fleck@intel.com> Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-14 20:52:09 +08:00
if (ret)
return 0;
return 1;
}
IB/core: Register SA ibnl client during ib_core initialization Move SA ibnl client registration to ib_core module init. This will allow us to register a single client to handle all RDMA_NL_LS operations and make it SA independent. Signed-off-by: Mark Bloch <markb@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-05-19 22:12:35 +08:00
int ib_nl_handle_resolve_resp(struct sk_buff *skb,
RDMA/netlink: Convert LS to doit callback RDMA_NL_LS protocol is actually does not dump anything, but sets data and it should be handled by doit callback. This patch actually converts RDMA_NL_LS to doit callback, while preserving IWCM and RDMA_CM flows through netlink_dump_start(). Signed-off-by: Leon Romanovsky <leonro@mellanox.com> Reviewed-by: Steve Wise <swise@opengridcomputing.com>
2017-06-15 19:20:39 +08:00
struct nlmsghdr *nlh,
struct netlink_ext_ack *extack)
IB/sa: Route SA pathrecord query through netlink This patch routes a SA pathrecord query to netlink first and processes the response appropriately. If a failure is returned, the request will be sent through IB. The decision whether to route the request to netlink first is determined by the presence of a listener for the local service netlink multicast group. If the user-space local service netlink multicast group listener is not present, the request will be sent through IB, just like what is currently being done. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: John Fleck <john.fleck@intel.com> Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-14 20:52:09 +08:00
{
unsigned long flags;
IB/SA: Replace usage of found with dedicated list iterator variable To move the list iterator variable into the list_for_each_entry_*() macro in the future it should be avoided to use the list iterator variable after the loop body. To *never* use the list iterator variable after the loop it was concluded to use a separate iterator variable instead of a found boolean. This removes the need to use a found variable and simply checking if the variable was set, can determine if the break/goto was hit. Link: https://lore.kernel.org/r/20220331091634.644840-1-jakobkoschel@gmail.com Reviewed-by: Leon Romanovsky <leonro@nvidia.com> Signed-off-by: Jakob Koschel <jakobkoschel@gmail.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
2022-03-31 17:16:34 +08:00
struct ib_sa_query *query = NULL, *iter;
IB/sa: Route SA pathrecord query through netlink This patch routes a SA pathrecord query to netlink first and processes the response appropriately. If a failure is returned, the request will be sent through IB. The decision whether to route the request to netlink first is determined by the presence of a listener for the local service netlink multicast group. If the user-space local service netlink multicast group listener is not present, the request will be sent through IB, just like what is currently being done. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: John Fleck <john.fleck@intel.com> Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-14 20:52:09 +08:00
struct ib_mad_send_buf *send_buf;
struct ib_mad_send_wc mad_send_wc;
int ret;
IB/sa: Fix netlink local service GFP crash The rdma netlink local service registers a handler to handle RESOLVE response and another handler to handle SET_TIMEOUT request. The first thing these handlers do is to call netlink_capable() to check the access right of the received skb to make sure that the sender has root access. Under normal conditions, such responses and requests will be directly forwarded to the handlers without going through the netlink_dump pathway (see ibnl_rcv_msg() in drivers/infiniband/core/netlink.c). However, a user application could send a RESOLVE request (not response) to the local service, which will fall into the netlink_dump pathway, where a new skb will be created without initializing the control block. This new skb will be eventually forwarded to the local service RESOLVE response handler. Unfortunately, netlink_capable() will cause general protection fault if the skb's control block is not initialized. This patch will address the problem by checking the skb first. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Reported-by: Dmitry Vyukov <dvyukov@google.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-01-21 21:41:31 +08:00
if ((nlh->nlmsg_flags & NLM_F_REQUEST) ||
RDMA/netlink: Add flag to consolidate common handling Add ability to provide flags to control RDMA netlink callbacks and convert addr.c and sa_query.c to be first users of such infrastructure. It allows to move their CAP_NET_ADMIN checks into netlink core. Signed-off-by: Leon Romanovsky <leonro@mellanox.com> Reviewed-by: Steve Wise <swise@opengridcomputing.com>
2017-06-12 21:00:19 +08:00
!(NETLINK_CB(skb).sk))
IB/sa: Route SA pathrecord query through netlink This patch routes a SA pathrecord query to netlink first and processes the response appropriately. If a failure is returned, the request will be sent through IB. The decision whether to route the request to netlink first is determined by the presence of a listener for the local service netlink multicast group. If the user-space local service netlink multicast group listener is not present, the request will be sent through IB, just like what is currently being done. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: John Fleck <john.fleck@intel.com> Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-14 20:52:09 +08:00
return -EPERM;
spin_lock_irqsave(&ib_nl_request_lock, flags);
IB/SA: Replace usage of found with dedicated list iterator variable To move the list iterator variable into the list_for_each_entry_*() macro in the future it should be avoided to use the list iterator variable after the loop body. To *never* use the list iterator variable after the loop it was concluded to use a separate iterator variable instead of a found boolean. This removes the need to use a found variable and simply checking if the variable was set, can determine if the break/goto was hit. Link: https://lore.kernel.org/r/20220331091634.644840-1-jakobkoschel@gmail.com Reviewed-by: Leon Romanovsky <leonro@nvidia.com> Signed-off-by: Jakob Koschel <jakobkoschel@gmail.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
2022-03-31 17:16:34 +08:00
list_for_each_entry(iter, &ib_nl_request_list, list) {
IB/sa: Route SA pathrecord query through netlink This patch routes a SA pathrecord query to netlink first and processes the response appropriately. If a failure is returned, the request will be sent through IB. The decision whether to route the request to netlink first is determined by the presence of a listener for the local service netlink multicast group. If the user-space local service netlink multicast group listener is not present, the request will be sent through IB, just like what is currently being done. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: John Fleck <john.fleck@intel.com> Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-14 20:52:09 +08:00
/*
* If the query is cancelled, let the timeout routine
* take care of it.
*/
IB/SA: Replace usage of found with dedicated list iterator variable To move the list iterator variable into the list_for_each_entry_*() macro in the future it should be avoided to use the list iterator variable after the loop body. To *never* use the list iterator variable after the loop it was concluded to use a separate iterator variable instead of a found boolean. This removes the need to use a found variable and simply checking if the variable was set, can determine if the break/goto was hit. Link: https://lore.kernel.org/r/20220331091634.644840-1-jakobkoschel@gmail.com Reviewed-by: Leon Romanovsky <leonro@nvidia.com> Signed-off-by: Jakob Koschel <jakobkoschel@gmail.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
2022-03-31 17:16:34 +08:00
if (nlh->nlmsg_seq == iter->seq) {
if (!ib_sa_query_cancelled(iter)) {
list_del(&iter->list);
query = iter;
}
IB/sa: Route SA pathrecord query through netlink This patch routes a SA pathrecord query to netlink first and processes the response appropriately. If a failure is returned, the request will be sent through IB. The decision whether to route the request to netlink first is determined by the presence of a listener for the local service netlink multicast group. If the user-space local service netlink multicast group listener is not present, the request will be sent through IB, just like what is currently being done. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: John Fleck <john.fleck@intel.com> Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-14 20:52:09 +08:00
break;
}
}
IB/SA: Replace usage of found with dedicated list iterator variable To move the list iterator variable into the list_for_each_entry_*() macro in the future it should be avoided to use the list iterator variable after the loop body. To *never* use the list iterator variable after the loop it was concluded to use a separate iterator variable instead of a found boolean. This removes the need to use a found variable and simply checking if the variable was set, can determine if the break/goto was hit. Link: https://lore.kernel.org/r/20220331091634.644840-1-jakobkoschel@gmail.com Reviewed-by: Leon Romanovsky <leonro@nvidia.com> Signed-off-by: Jakob Koschel <jakobkoschel@gmail.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
2022-03-31 17:16:34 +08:00
if (!query) {
IB/sa: Route SA pathrecord query through netlink This patch routes a SA pathrecord query to netlink first and processes the response appropriately. If a failure is returned, the request will be sent through IB. The decision whether to route the request to netlink first is determined by the presence of a listener for the local service netlink multicast group. If the user-space local service netlink multicast group listener is not present, the request will be sent through IB, just like what is currently being done. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: John Fleck <john.fleck@intel.com> Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-14 20:52:09 +08:00
spin_unlock_irqrestore(&ib_nl_request_lock, flags);
goto resp_out;
}
send_buf = query->mad_buf;
if (!ib_nl_is_good_resolve_resp(nlh)) {
/* if the result is a failure, send out the packet via IB */
ib_sa_disable_local_svc(query);
ret = ib_post_send_mad(query->mad_buf, NULL);
spin_unlock_irqrestore(&ib_nl_request_lock, flags);
if (ret) {
mad_send_wc.send_buf = send_buf;
mad_send_wc.status = IB_WC_GENERAL_ERR;
send_handler(query->port->agent, &mad_send_wc);
}
} else {
spin_unlock_irqrestore(&ib_nl_request_lock, flags);
ib_nl_process_good_resolve_rsp(query, nlh);
}
resp_out:
RDMA/netlink: Do not always generate an ACK for some netlink operations In rdma_nl_rcv_skb(), the local variable err is assigned the return value of the supplied callback function, which could be one of ib_nl_handle_resolve_resp(), ib_nl_handle_set_timeout(), or ib_nl_handle_ip_res_resp(). These three functions all return skb->len on success. rdma_nl_rcv_skb() is merely a copy of netlink_rcv_skb(). The callback functions used by the latter have the convention: "Returns 0 on success or a negative error code". In particular, the statement (equal for both functions): if (nlh->nlmsg_flags & NLM_F_ACK || err) implies that rdma_nl_rcv_skb() always will ack a message, independent of the NLM_F_ACK being set in nlmsg_flags or not. The fix could be to change the above statement, but it is better to keep the two *_rcv_skb() functions equal in this respect and instead change the three callback functions in the rdma subsystem to the correct convention. Fixes: 2ca546b92a02 ("IB/sa: Route SA pathrecord query through netlink") Fixes: ae43f8286730 ("IB/core: Add IP to GID netlink offload") Link: https://lore.kernel.org/r/20191216120436.3204814-1-haakon.bugge@oracle.com Suggested-by: Mark Haywood <mark.haywood@oracle.com> Signed-off-by: Håkon Bugge <haakon.bugge@oracle.com> Tested-by: Mark Haywood <mark.haywood@oracle.com> Reviewed-by: Leon Romanovsky <leonro@mellanox.com> Reviewed-by: Jason Gunthorpe <jgg@mellanox.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2019-12-16 20:04:36 +08:00
return 0;
IB/sa: Route SA pathrecord query through netlink This patch routes a SA pathrecord query to netlink first and processes the response appropriately. If a failure is returned, the request will be sent through IB. The decision whether to route the request to netlink first is determined by the presence of a listener for the local service netlink multicast group. If the user-space local service netlink multicast group listener is not present, the request will be sent through IB, just like what is currently being done. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: John Fleck <john.fleck@intel.com> Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-14 20:52:09 +08:00
}
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
static void free_sm_ah(struct kref *kref)
{
struct ib_sa_sm_ah *sm_ah = container_of(kref, struct ib_sa_sm_ah, ref);
RDMA: Mark if destroy address handle is in a sleepable context Introduce a 'flags' field to destroy address handle callback and add a flag that marks whether the callback is executed in an atomic context or not. This will allow drivers to wait for completion instead of polling for it when it is allowed. Signed-off-by: Gal Pressman <galpress@amazon.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2018-12-12 17:09:06 +08:00
rdma_destroy_ah(sm_ah->ah, 0);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
kfree(sm_ah);
}
IB/sa: Require SA registration Require users to register with SA module, to prevent the sa_query module text from going away while an SA query callback is still running. Update all in-tree users for the new interface. Signed-off-by: Michael S. Tsirkin <mst@mellanox.co.il> Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2006-08-22 07:40:12 +08:00
void ib_sa_register_client(struct ib_sa_client *client)
{
atomic_set(&client->users, 1);
init_completion(&client->comp);
}
EXPORT_SYMBOL(ib_sa_register_client);
void ib_sa_unregister_client(struct ib_sa_client *client)
{
ib_sa_client_put(client);
wait_for_completion(&client->comp);
}
EXPORT_SYMBOL(ib_sa_unregister_client);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
/**
* ib_sa_cancel_query - try to cancel an SA query
* @id:ID of query to cancel
* @query:query pointer to cancel
*
* Try to cancel an SA query. If the id and query don't match up or
* the query has already completed, nothing is done. Otherwise the
* query is canceled and will complete with a status of -EINTR.
*/
void ib_sa_cancel_query(int id, struct ib_sa_query *query)
{
unsigned long flags;
[IB] Fix MAD layer DMA mappings to avoid touching data buffer once mapped The MAD layer was violating the DMA API by touching data buffers used for sends after the DMA mapping was done. This causes problems on non-cache-coherent architectures, because the device doing DMA won't see updates to the payload buffers that exist only in the CPU cache. Fix this by having all MAD consumers use ib_create_send_mad() to allocate their send buffers, and moving the DMA mapping into the MAD layer so it can be done just before calling send (and after any modifications of the send buffer by the MAD layer). Tested on a non-cache-coherent PowerPC 440SPe system. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-10-26 01:51:39 +08:00
struct ib_mad_send_buf *mad_buf;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
ib core: Convert query_idr to XArray Signed-off-by: Matthew Wilcox <willy@infradead.org> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2019-02-21 08:20:46 +08:00
xa_lock_irqsave(&queries, flags);
if (xa_load(&queries, id) != query) {
xa_unlock_irqrestore(&queries, flags);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
return;
}
[IB] Fix MAD layer DMA mappings to avoid touching data buffer once mapped The MAD layer was violating the DMA API by touching data buffers used for sends after the DMA mapping was done. This causes problems on non-cache-coherent architectures, because the device doing DMA won't see updates to the payload buffers that exist only in the CPU cache. Fix this by having all MAD consumers use ib_create_send_mad() to allocate their send buffers, and moving the DMA mapping into the MAD layer so it can be done just before calling send (and after any modifications of the send buffer by the MAD layer). Tested on a non-cache-coherent PowerPC 440SPe system. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-10-26 01:51:39 +08:00
mad_buf = query->mad_buf;
ib core: Convert query_idr to XArray Signed-off-by: Matthew Wilcox <willy@infradead.org> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2019-02-21 08:20:46 +08:00
xa_unlock_irqrestore(&queries, flags);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
IB/sa: Route SA pathrecord query through netlink This patch routes a SA pathrecord query to netlink first and processes the response appropriately. If a failure is returned, the request will be sent through IB. The decision whether to route the request to netlink first is determined by the presence of a listener for the local service netlink multicast group. If the user-space local service netlink multicast group listener is not present, the request will be sent through IB, just like what is currently being done. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: John Fleck <john.fleck@intel.com> Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-14 20:52:09 +08:00
/*
* If the query is still on the netlink request list, schedule
* it to be cancelled by the timeout routine. Otherwise, it has been
* sent to the MAD layer and has to be cancelled from there.
*/
if (!ib_nl_cancel_request(query))
IB/cm: Simplify ib_cancel_mad() and ib_modify_mad() calls The mad_agent parameter is redundant since the struct ib_mad_send_buf already has a pointer of it. Link: https://lore.kernel.org/r/0987c784b25f7bfa72f78691f50cff066de587e1.1622629024.git.leonro@nvidia.com Signed-off-by: Mark Zhang <markzhang@nvidia.com> Signed-off-by: Leon Romanovsky <leonro@nvidia.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
2021-06-02 18:27:06 +08:00
ib_cancel_mad(mad_buf);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
EXPORT_SYMBOL(ib_sa_cancel_query);
RDMA: Support more than 255 rdma ports Current code uses many different types when dealing with a port of a RDMA device: u8, unsigned int and u32. Switch to u32 to clean up the logic. This allows us to make (at least) the core view consistent and use the same type. Unfortunately not all places can be converted. Many uverbs functions expect port to be u8 so keep those places in order not to break UAPIs. HW/Spec defined values must also not be changed. With the switch to u32 we now can support devices with more than 255 ports. U32_MAX is reserved to make control logic a bit easier to deal with. As a device with U32_MAX ports probably isn't going to happen any time soon this seems like a non issue. When a device with more than 255 ports is created uverbs will report the RDMA device as having 255 ports as this is the max currently supported. The verbs interface is not changed yet because the IBTA spec limits the port size in too many places to be u8 and all applications that relies in verbs won't be able to cope with this change. At this stage, we are extending the interfaces that are using vendor channel solely Once the limitation is lifted mlx5 in switchdev mode will be able to have thousands of SFs created by the device. As the only instance of an RDMA device that reports more than 255 ports will be a representor device and it exposes itself as a RAW Ethernet only device CM/MAD/IPoIB and other ULPs aren't effected by this change and their sysfs/interfaces that are exposes to userspace can remain unchanged. While here cleanup some alignment issues and remove unneeded sanity checks (mainly in rdmavt), Link: https://lore.kernel.org/r/20210301070420.439400-1-leon@kernel.org Signed-off-by: Mark Bloch <mbloch@nvidia.com> Signed-off-by: Leon Romanovsky <leonro@nvidia.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
2021-03-01 15:04:20 +08:00
static u8 get_src_path_mask(struct ib_device *device, u32 port_num)
IB/sa: Set src_path_bits correctly in ib_init_ah_from_path() src_path_bits needs to mask off the base LID value. Signed-off-by: Sean Hefty <sean.hefty@intel.com>
2007-04-06 01:51:10 +08:00
{
struct ib_sa_device *sa_dev;
struct ib_sa_port *port;
unsigned long flags;
u8 src_path_mask;
sa_dev = ib_get_client_data(device, &sa_client);
if (!sa_dev)
return 0x7f;
port = &sa_dev->port[port_num - sa_dev->start_port];
spin_lock_irqsave(&port->ah_lock, flags);
src_path_mask = port->sm_ah ? port->sm_ah->src_path_mask : 0x7f;
spin_unlock_irqrestore(&port->ah_lock, flags);
return src_path_mask;
}
RDMA: Support more than 255 rdma ports Current code uses many different types when dealing with a port of a RDMA device: u8, unsigned int and u32. Switch to u32 to clean up the logic. This allows us to make (at least) the core view consistent and use the same type. Unfortunately not all places can be converted. Many uverbs functions expect port to be u8 so keep those places in order not to break UAPIs. HW/Spec defined values must also not be changed. With the switch to u32 we now can support devices with more than 255 ports. U32_MAX is reserved to make control logic a bit easier to deal with. As a device with U32_MAX ports probably isn't going to happen any time soon this seems like a non issue. When a device with more than 255 ports is created uverbs will report the RDMA device as having 255 ports as this is the max currently supported. The verbs interface is not changed yet because the IBTA spec limits the port size in too many places to be u8 and all applications that relies in verbs won't be able to cope with this change. At this stage, we are extending the interfaces that are using vendor channel solely Once the limitation is lifted mlx5 in switchdev mode will be able to have thousands of SFs created by the device. As the only instance of an RDMA device that reports more than 255 ports will be a representor device and it exposes itself as a RAW Ethernet only device CM/MAD/IPoIB and other ULPs aren't effected by this change and their sysfs/interfaces that are exposes to userspace can remain unchanged. While here cleanup some alignment issues and remove unneeded sanity checks (mainly in rdmavt), Link: https://lore.kernel.org/r/20210301070420.439400-1-leon@kernel.org Signed-off-by: Mark Bloch <mbloch@nvidia.com> Signed-off-by: Leon Romanovsky <leonro@nvidia.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
2021-03-01 15:04:20 +08:00
static int init_ah_attr_grh_fields(struct ib_device *device, u32 port_num,
IB/core: Refactor ib_init_ah_attr_from_path() for RoCE Resolving route for RoCE for a path record is needed only for the received CM requests. Therefore, (a) ib_init_ah_attr_from_path() is refactored first to isolate the code of resolving route. (b) Setting dlid, path bits is not needed for RoCE. Additionally ah attribute initialization is done from the path record entry, so it is better to refer to path record entry type for different link layer instead of ah attribute type while initializing ah attribute itself. Signed-off-by: Parav Pandit <parav@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2018-03-13 22:06:20 +08:00
struct sa_path_rec *rec,
IB/cm: Replace members of sa_path_rec with 'struct sgid_attr *' While processing a path record entry in CM messages the associated GID attribute is now also supplied. Currently for RoCE a netdevice's net namespace pointer and ifindex are stored in path record entry. Both of these fields of the netdev can change anytime while processing CM messages. Additionally storing net namespace without holding reference will lead to use-after-free crash. Therefore it is removed. Netdevice information for RoCE is instead provided via referenced gid attribute in ib_cm requests. Such a design leads to a situation where the kernel can crash when the net pointer becomes invalid. However today it is always initialized to init_net, which cannot become invalid. In order to support processing packets in any arbitrary namespace of the received packet, it is necessary to avoid such conditions. This patch removes the dependency on the net pointer and ifindex; instead it will rely on SGID attribute which contains a pointer to netdev. Signed-off-by: Parav Pandit <parav@mellanox.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com> Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
2018-06-19 15:59:19 +08:00
struct rdma_ah_attr *ah_attr,
const struct ib_gid_attr *gid_attr)
IB/core: Refactor ib_init_ah_attr_from_path() for RoCE Resolving route for RoCE for a path record is needed only for the received CM requests. Therefore, (a) ib_init_ah_attr_from_path() is refactored first to isolate the code of resolving route. (b) Setting dlid, path bits is not needed for RoCE. Additionally ah attribute initialization is done from the path record entry, so it is better to refer to path record entry type for different link layer instead of ah attribute type while initializing ah attribute itself. Signed-off-by: Parav Pandit <parav@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2018-03-13 22:06:20 +08:00
{
enum ib_gid_type type = sa_conv_pathrec_to_gid_type(rec);
RDMA/cm: Add RDMA CM support for IBoE devices Add support for IBoE device binding and IP --> GID resolution. Path resolving and multicast joining are implemented within cma.c by filling in the responses and running callbacks in the CMA work queue. IP --> GID resolution always yields IPv6 link local addresses; remote GIDs are derived from the destination MAC address of the remote port. Multicast GIDs are always mapped to multicast MACs as is done in IPv6. (IPv4 multicast is enabled by translating IPv4 multicast addresses to IPv6 multicast as described in <http://www.mail-archive.com/ipng@sunroof.eng.sun.com/msg02134.html>.) Some helper functions are added to ib_addr.h. Signed-off-by: Eli Cohen <eli@mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2010-10-14 03:26:51 +08:00
IB/cm: Replace members of sa_path_rec with 'struct sgid_attr *' While processing a path record entry in CM messages the associated GID attribute is now also supplied. Currently for RoCE a netdevice's net namespace pointer and ifindex are stored in path record entry. Both of these fields of the netdev can change anytime while processing CM messages. Additionally storing net namespace without holding reference will lead to use-after-free crash. Therefore it is removed. Netdevice information for RoCE is instead provided via referenced gid attribute in ib_cm requests. Such a design leads to a situation where the kernel can crash when the net pointer becomes invalid. However today it is always initialized to init_net, which cannot become invalid. In order to support processing packets in any arbitrary namespace of the received packet, it is necessary to avoid such conditions. This patch removes the dependency on the net pointer and ifindex; instead it will rely on SGID attribute which contains a pointer to netdev. Signed-off-by: Parav Pandit <parav@mellanox.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com> Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
2018-06-19 15:59:19 +08:00
if (!gid_attr) {
gid_attr = rdma_find_gid_by_port(device, &rec->sgid, type,
port_num, NULL);
if (IS_ERR(gid_attr))
return PTR_ERR(gid_attr);
} else
rdma_hold_gid_attr(gid_attr);
IB/core: Validate route when we init ah In order to make sure API users don't try to use SGIDs which don't conform to the routing table, validate the route before searching the RoCE GID table. Signed-off-by: Matan Barak <matanb@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-12-23 20:56:53 +08:00
IB: Make init_ah_attr_grh_fields set sgid_attr Use the sgid and other information from the path record to figure out the sgid_attrs. Store the selected table entry in the sgid_attr for everything else to use. Signed-off-by: Parav Pandit <parav@mellanox.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com> Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
2018-06-19 15:59:16 +08:00
rdma_move_grh_sgid_attr(ah_attr, &rec->dgid,
be32_to_cpu(rec->flow_label),
rec->hop_limit, rec->traffic_class,
gid_attr);
IB/core: Refactor ib_init_ah_attr_from_path() for RoCE Resolving route for RoCE for a path record is needed only for the received CM requests. Therefore, (a) ib_init_ah_attr_from_path() is refactored first to isolate the code of resolving route. (b) Setting dlid, path bits is not needed for RoCE. Additionally ah attribute initialization is done from the path record entry, so it is better to refer to path record entry type for different link layer instead of ah attribute type while initializing ah attribute itself. Signed-off-by: Parav Pandit <parav@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2018-03-13 22:06:20 +08:00
return 0;
}
IB/core: Validate route when we init ah In order to make sure API users don't try to use SGIDs which don't conform to the routing table, validate the route before searching the RoCE GID table. Signed-off-by: Matan Barak <matanb@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-12-23 20:56:53 +08:00
IB/cm: Replace members of sa_path_rec with 'struct sgid_attr *' While processing a path record entry in CM messages the associated GID attribute is now also supplied. Currently for RoCE a netdevice's net namespace pointer and ifindex are stored in path record entry. Both of these fields of the netdev can change anytime while processing CM messages. Additionally storing net namespace without holding reference will lead to use-after-free crash. Therefore it is removed. Netdevice information for RoCE is instead provided via referenced gid attribute in ib_cm requests. Such a design leads to a situation where the kernel can crash when the net pointer becomes invalid. However today it is always initialized to init_net, which cannot become invalid. In order to support processing packets in any arbitrary namespace of the received packet, it is necessary to avoid such conditions. This patch removes the dependency on the net pointer and ifindex; instead it will rely on SGID attribute which contains a pointer to netdev. Signed-off-by: Parav Pandit <parav@mellanox.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com> Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
2018-06-19 15:59:19 +08:00
/**
* ib_init_ah_attr_from_path - Initialize address handle attributes based on
* an SA path record.
* @device: Device associated ah attributes initialization.
* @port_num: Port on the specified device.
* @rec: path record entry to use for ah attributes initialization.
* @ah_attr: address handle attributes to initialization from path record.
infiniband: fix core/ kernel-doc notation Correct function parameter names (typos or renames). Add kernel-doc notation for missing function parameters. ../drivers/infiniband/core/sa_query.c:1263: warning: Function parameter or member 'gid_attr' not described in 'ib_init_ah_attr_from_path' ../drivers/infiniband/core/sa_query.c:1263: warning: Excess function parameter 'sgid_attr' description in 'ib_init_ah_attr_from_path' ../drivers/infiniband/core/device.c:145: warning: Function parameter or member 'dev' not described in 'rdma_dev_access_netns' ../drivers/infiniband/core/device.c:145: warning: Excess function parameter 'device' description in 'rdma_dev_access_netns' ../drivers/infiniband/core/device.c:1333: warning: Function parameter or member 'name' not described in 'ib_register_device' ../drivers/infiniband/core/device.c:1461: warning: Function parameter or member 'ib_dev' not described in 'ib_unregister_device' ../drivers/infiniband/core/device.c:1461: warning: Excess function parameter 'device' description in 'ib_unregister_device' ../drivers/infiniband/core/device.c:1483: warning: Function parameter or member 'ib_dev' not described in 'ib_unregister_device_and_put' ../drivers/infiniband/core/device.c:1550: warning: Function parameter or member 'ib_dev' not described in 'ib_unregister_device_queued' Link: https://lore.kernel.org/r/20191010035240.191542461@gmail.com Signed-off-by: Randy Dunlap <rd.dunlab@gmail.com> Reviewed-by: Jason Gunthorpe <jgg@mellanox.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2019-10-10 11:52:49 +08:00
* @gid_attr: SGID attribute to consider during initialization.
IB/cm: Replace members of sa_path_rec with 'struct sgid_attr *' While processing a path record entry in CM messages the associated GID attribute is now also supplied. Currently for RoCE a netdevice's net namespace pointer and ifindex are stored in path record entry. Both of these fields of the netdev can change anytime while processing CM messages. Additionally storing net namespace without holding reference will lead to use-after-free crash. Therefore it is removed. Netdevice information for RoCE is instead provided via referenced gid attribute in ib_cm requests. Such a design leads to a situation where the kernel can crash when the net pointer becomes invalid. However today it is always initialized to init_net, which cannot become invalid. In order to support processing packets in any arbitrary namespace of the received packet, it is necessary to avoid such conditions. This patch removes the dependency on the net pointer and ifindex; instead it will rely on SGID attribute which contains a pointer to netdev. Signed-off-by: Parav Pandit <parav@mellanox.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com> Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
2018-06-19 15:59:19 +08:00
*
* When ib_init_ah_attr_from_path() returns success,
* (a) for IB link layer it optionally contains a reference to SGID attribute
* when GRH is present for IB link layer.
* (b) for RoCE link layer it contains a reference to SGID attribute.
* User must invoke rdma_destroy_ah_attr() to release reference to SGID
* attributes which are initialized using ib_init_ah_attr_from_path().
*/
RDMA: Support more than 255 rdma ports Current code uses many different types when dealing with a port of a RDMA device: u8, unsigned int and u32. Switch to u32 to clean up the logic. This allows us to make (at least) the core view consistent and use the same type. Unfortunately not all places can be converted. Many uverbs functions expect port to be u8 so keep those places in order not to break UAPIs. HW/Spec defined values must also not be changed. With the switch to u32 we now can support devices with more than 255 ports. U32_MAX is reserved to make control logic a bit easier to deal with. As a device with U32_MAX ports probably isn't going to happen any time soon this seems like a non issue. When a device with more than 255 ports is created uverbs will report the RDMA device as having 255 ports as this is the max currently supported. The verbs interface is not changed yet because the IBTA spec limits the port size in too many places to be u8 and all applications that relies in verbs won't be able to cope with this change. At this stage, we are extending the interfaces that are using vendor channel solely Once the limitation is lifted mlx5 in switchdev mode will be able to have thousands of SFs created by the device. As the only instance of an RDMA device that reports more than 255 ports will be a representor device and it exposes itself as a RAW Ethernet only device CM/MAD/IPoIB and other ULPs aren't effected by this change and their sysfs/interfaces that are exposes to userspace can remain unchanged. While here cleanup some alignment issues and remove unneeded sanity checks (mainly in rdmavt), Link: https://lore.kernel.org/r/20210301070420.439400-1-leon@kernel.org Signed-off-by: Mark Bloch <mbloch@nvidia.com> Signed-off-by: Leon Romanovsky <leonro@nvidia.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
2021-03-01 15:04:20 +08:00
int ib_init_ah_attr_from_path(struct ib_device *device, u32 port_num,
IB/core: Refactor ib_init_ah_attr_from_path() for RoCE Resolving route for RoCE for a path record is needed only for the received CM requests. Therefore, (a) ib_init_ah_attr_from_path() is refactored first to isolate the code of resolving route. (b) Setting dlid, path bits is not needed for RoCE. Additionally ah attribute initialization is done from the path record entry, so it is better to refer to path record entry type for different link layer instead of ah attribute type while initializing ah attribute itself. Signed-off-by: Parav Pandit <parav@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2018-03-13 22:06:20 +08:00
struct sa_path_rec *rec,
IB/cm: Replace members of sa_path_rec with 'struct sgid_attr *' While processing a path record entry in CM messages the associated GID attribute is now also supplied. Currently for RoCE a netdevice's net namespace pointer and ifindex are stored in path record entry. Both of these fields of the netdev can change anytime while processing CM messages. Additionally storing net namespace without holding reference will lead to use-after-free crash. Therefore it is removed. Netdevice information for RoCE is instead provided via referenced gid attribute in ib_cm requests. Such a design leads to a situation where the kernel can crash when the net pointer becomes invalid. However today it is always initialized to init_net, which cannot become invalid. In order to support processing packets in any arbitrary namespace of the received packet, it is necessary to avoid such conditions. This patch removes the dependency on the net pointer and ifindex; instead it will rely on SGID attribute which contains a pointer to netdev. Signed-off-by: Parav Pandit <parav@mellanox.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com> Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
2018-06-19 15:59:19 +08:00
struct rdma_ah_attr *ah_attr,
const struct ib_gid_attr *gid_attr)
IB/core: Refactor ib_init_ah_attr_from_path() for RoCE Resolving route for RoCE for a path record is needed only for the received CM requests. Therefore, (a) ib_init_ah_attr_from_path() is refactored first to isolate the code of resolving route. (b) Setting dlid, path bits is not needed for RoCE. Additionally ah attribute initialization is done from the path record entry, so it is better to refer to path record entry type for different link layer instead of ah attribute type while initializing ah attribute itself. Signed-off-by: Parav Pandit <parav@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2018-03-13 22:06:20 +08:00
{
int ret = 0;
IB/core: Add netdev to path record In order to find the sgid_index, one could just query the IB cache with the correct GID and netdevice. Therefore, instead of storing the L2 attributes directly in the path, we only store the ifindex and net and use them later to get the sgid_index. The vlan_id and smac L2 attributes are removed in a later patch. Signed-off-by: Matan Barak <matanb@mellanox.com> Reviewed-By: Devesh Sharma <devesh.sharma@avagotech.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-10-15 23:38:47 +08:00
IB/core: Refactor ib_init_ah_attr_from_path() for RoCE Resolving route for RoCE for a path record is needed only for the received CM requests. Therefore, (a) ib_init_ah_attr_from_path() is refactored first to isolate the code of resolving route. (b) Setting dlid, path bits is not needed for RoCE. Additionally ah attribute initialization is done from the path record entry, so it is better to refer to path record entry type for different link layer instead of ah attribute type while initializing ah attribute itself. Signed-off-by: Parav Pandit <parav@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2018-03-13 22:06:20 +08:00
memset(ah_attr, 0, sizeof(*ah_attr));
ah_attr->type = rdma_ah_find_type(device, port_num);
rdma_ah_set_sl(ah_attr, rec->sl);
rdma_ah_set_port_num(ah_attr, port_num);
rdma_ah_set_static_rate(ah_attr, rec->rate);
IB/SA: Introduce path record specific types struct sa_path_rec has a gid_type field. This patch introduces a more generic path record specific type 'rec_type' which is either IB, ROCE v1 or ROCE v2. The patch also provides conversion functions to get a gid type from a path record type and vice versa Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-28 07:05:59 +08:00
IB/core: Refactor ib_init_ah_attr_from_path() for RoCE Resolving route for RoCE for a path record is needed only for the received CM requests. Therefore, (a) ib_init_ah_attr_from_path() is refactored first to isolate the code of resolving route. (b) Setting dlid, path bits is not needed for RoCE. Additionally ah attribute initialization is done from the path record entry, so it is better to refer to path record entry type for different link layer instead of ah attribute type while initializing ah attribute itself. Signed-off-by: Parav Pandit <parav@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2018-03-13 22:06:20 +08:00
if (sa_path_is_roce(rec)) {
IB/core: Simplify check for RoCE route resolve roce_resolve_route_from_path() resolves the route based on the netdevice of the GID attribute, therefore there is no point in checking again if the route is resolved matches the same interface it arrived. Signed-off-by: Parav Pandit <parav@mellanox.com> Signed-off-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2018-07-08 18:41:17 +08:00
ret = roce_resolve_route_from_path(rec, gid_attr);
IB/core: Refactor ib_init_ah_attr_from_path() for RoCE Resolving route for RoCE for a path record is needed only for the received CM requests. Therefore, (a) ib_init_ah_attr_from_path() is refactored first to isolate the code of resolving route. (b) Setting dlid, path bits is not needed for RoCE. Additionally ah attribute initialization is done from the path record entry, so it is better to refer to path record entry type for different link layer instead of ah attribute type while initializing ah attribute itself. Signed-off-by: Parav Pandit <parav@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2018-03-13 22:06:20 +08:00
if (ret)
IB/sa: Add ib_init_ah_from_path() Add a call to initialize address handle attributes given a path record. This is used by the CM, and would be useful for users of UD QPs. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2006-06-18 11:37:39 +08:00
return ret;
IB/core: Refactor ib_init_ah_attr_from_path() for RoCE Resolving route for RoCE for a path record is needed only for the received CM requests. Therefore, (a) ib_init_ah_attr_from_path() is refactored first to isolate the code of resolving route. (b) Setting dlid, path bits is not needed for RoCE. Additionally ah attribute initialization is done from the path record entry, so it is better to refer to path record entry type for different link layer instead of ah attribute type while initializing ah attribute itself. Signed-off-by: Parav Pandit <parav@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2018-03-13 22:06:20 +08:00
memcpy(ah_attr->roce.dmac, sa_path_get_dmac(rec), ETH_ALEN);
} else {
rdma_ah_set_dlid(ah_attr, be32_to_cpu(sa_path_get_dlid(rec)));
if (sa_path_is_opa(rec) &&
rdma_ah_get_dlid(ah_attr) == be16_to_cpu(IB_LID_PERMISSIVE))
rdma_ah_set_make_grd(ah_attr, true);
rdma_ah_set_path_bits(ah_attr,
be32_to_cpu(sa_path_get_slid(rec)) &
get_src_path_mask(device, port_num));
IB/SA: Split struct sa_path_rec based on IB and ROCE specific fields sa_path_rec now contains a union of sa_path_rec_ib and sa_path_rec_roce based on the type of the path record. Note that fields applicable to path record type ROCE v1 and ROCE v2 fall under sa_path_rec_roce. Accessor functions are added to these fields so the caller doesn't have to know the type. Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-28 07:06:00 +08:00
}
IB/core: Validate route when we init ah In order to make sure API users don't try to use SGIDs which don't conform to the routing table, validate the route before searching the RoCE GID table. Signed-off-by: Matan Barak <matanb@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-12-23 20:56:53 +08:00
IB/core: Refactor ib_init_ah_attr_from_path() for RoCE Resolving route for RoCE for a path record is needed only for the received CM requests. Therefore, (a) ib_init_ah_attr_from_path() is refactored first to isolate the code of resolving route. (b) Setting dlid, path bits is not needed for RoCE. Additionally ah attribute initialization is done from the path record entry, so it is better to refer to path record entry type for different link layer instead of ah attribute type while initializing ah attribute itself. Signed-off-by: Parav Pandit <parav@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2018-03-13 22:06:20 +08:00
if (rec->hop_limit > 0 || sa_path_is_roce(rec))
IB/cm: Replace members of sa_path_rec with 'struct sgid_attr *' While processing a path record entry in CM messages the associated GID attribute is now also supplied. Currently for RoCE a netdevice's net namespace pointer and ifindex are stored in path record entry. Both of these fields of the netdev can change anytime while processing CM messages. Additionally storing net namespace without holding reference will lead to use-after-free crash. Therefore it is removed. Netdevice information for RoCE is instead provided via referenced gid attribute in ib_cm requests. Such a design leads to a situation where the kernel can crash when the net pointer becomes invalid. However today it is always initialized to init_net, which cannot become invalid. In order to support processing packets in any arbitrary namespace of the received packet, it is necessary to avoid such conditions. This patch removes the dependency on the net pointer and ifindex; instead it will rely on SGID attribute which contains a pointer to netdev. Signed-off-by: Parav Pandit <parav@mellanox.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com> Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
2018-06-19 15:59:19 +08:00
ret = init_ah_attr_grh_fields(device, port_num,
rec, ah_attr, gid_attr);
IB/core: Refactor ib_init_ah_attr_from_path() for RoCE Resolving route for RoCE for a path record is needed only for the received CM requests. Therefore, (a) ib_init_ah_attr_from_path() is refactored first to isolate the code of resolving route. (b) Setting dlid, path bits is not needed for RoCE. Additionally ah attribute initialization is done from the path record entry, so it is better to refer to path record entry type for different link layer instead of ah attribute type while initializing ah attribute itself. Signed-off-by: Parav Pandit <parav@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2018-03-13 22:06:20 +08:00
return ret;
IB/sa: Add ib_init_ah_from_path() Add a call to initialize address handle attributes given a path record. This is used by the CM, and would be useful for users of UD QPs. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2006-06-18 11:37:39 +08:00
}
IB/{core, cm, cma, ipoib}: Rename ib_init_ah_from_path to ib_init_ah_attr_from_path Since ib_init_ah_from_path initializes the address handle attribute, it is renamed to reflect so. Signed-off-by: Parav Pandit <parav@mellanox.com> Reviewed-by: Daniel Jurgens <danielj@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2017-11-14 20:52:16 +08:00
EXPORT_SYMBOL(ib_init_ah_attr_from_path);
IB/sa: Add ib_init_ah_from_path() Add a call to initialize address handle attributes given a path record. This is used by the CM, and would be useful for users of UD QPs. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2006-06-18 11:37:39 +08:00
IB/sa: Make sure SA queries use default P_Key MADs sent to the SA should use the the default P_Key (0x7fff/0xffff). There's no requirement that the default P_Key is stored at index 0 in the local P_Key table, so add code to the sa_query module to look up the index of the default P_Key when creating an address handle for the SA (which is done any time the P_Key table might change), and use this index for all SA queries. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2007-06-19 02:03:58 +08:00
static int alloc_mad(struct ib_sa_query *query, gfp_t gfp_mask)
{
IB/SA: Check dlid before SA agent queries for ClassPortInfo SA queries SM for class port info when there is a LID_CHANGE event. When a base lid is configured before fm is started ie when smlid is not yet assigned, SA handles the LID_CHANGE event and tries query SM with lid 0. This will cause an hang. [ 1106.958820] INFO: task kworker/2:0:23 blocked for more than 120 seconds. [ 1106.965082] Tainted: G O 4.12.0+ #1 [ 1106.969602] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. [ 1106.977227] kworker/2:0 D 0 23 2 0x00000000 [ 1106.977250] Workqueue: infiniband update_ib_cpi [ib_core] [ 1106.977261] Call Trace: [ 1106.977273] __schedule+0x28e/0x860 [ 1106.977285] schedule+0x36/0x80 [ 1106.977298] schedule_timeout+0x1a3/0x2e0 [ 1106.977310] ? radix_tree_iter_tag_clear+0x1b/0x20 [ 1106.977322] ? idr_alloc+0x64/0x90 [ 1106.977334] wait_for_completion+0xe3/0x140 [ 1106.977347] ? wake_up_q+0x80/0x80 [ 1106.977369] update_ib_cpi+0x163/0x210 [ib_core] [ 1106.977381] process_one_work+0x147/0x370 [ 1106.977394] worker_thread+0x4a/0x390 [ 1106.977406] kthread+0x109/0x140 [ 1106.977418] ? process_one_work+0x370/0x370 [ 1106.977430] ? kthread_park+0x60/0x60 [ 1106.977443] ret_from_fork+0x22/0x30 Always ensure a proper smlid is assigned before querying SM for cpi. Fixes: ee1c60b1bff ("IB/SA: Modify SA to implicitly cache Class Port info") Reviewed-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Venkata Sandeep Dhanalakota <venkata.s.dhanalakota@intel.com> Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2017-12-19 11:26:58 +08:00
struct rdma_ah_attr ah_attr;
IB/sa: Make sure SA queries use default P_Key MADs sent to the SA should use the the default P_Key (0x7fff/0xffff). There's no requirement that the default P_Key is stored at index 0 in the local P_Key table, so add code to the sa_query module to look up the index of the default P_Key when creating an address handle for the SA (which is done any time the P_Key table might change), and use this index for all SA queries. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2007-06-19 02:03:58 +08:00
unsigned long flags;
spin_lock_irqsave(&query->port->ah_lock, flags);
IB/sa: Fail requests made while creating new SM AH This patch solves a race that occurs after an event occurs that causes the SA query module to flush its SM address handle (AH). When SM AH becomes invalid and needs an update it is handled by the global workqueue. On the other hand this event is also handled in the IPoIB driver by queuing work in the ipoib_workqueue that does multicast joins. Although queuing is in the right order, it is done to 2 different workqueues and so there is no guarantee that the first to be queued is the first to be executed. This causes a problem because IPoIB may end up sending an request to the old SM, which will take a long time to time out (since the old SM is gone); this leads to a much longer than necessary interruption in multicast traffer. The patch sets the SA query module's SM AH to NULL when the event occurs, and until update_sm_ah() is done, any request that needs sm_ah fails with -EAGAIN return status. For consumers, the patch doesn't make things worse. Before the patch, MADs are sent to the wrong SM so the request gets lost. Consumers can be improved if they examine the return code and respond to EAGAIN properly but even without an improvement the situation is not getting worse. Signed-off-by: Moni Levy <monil@voltaire.com> Signed-off-by: Moni Shoua <monis@voltaire.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2008-07-15 14:48:43 +08:00
if (!query->port->sm_ah) {
spin_unlock_irqrestore(&query->port->ah_lock, flags);
return -EAGAIN;
}
IB/sa: Make sure SA queries use default P_Key MADs sent to the SA should use the the default P_Key (0x7fff/0xffff). There's no requirement that the default P_Key is stored at index 0 in the local P_Key table, so add code to the sa_query module to look up the index of the default P_Key when creating an address handle for the SA (which is done any time the P_Key table might change), and use this index for all SA queries. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2007-06-19 02:03:58 +08:00
kref_get(&query->port->sm_ah->ref);
query->sm_ah = query->port->sm_ah;
spin_unlock_irqrestore(&query->port->ah_lock, flags);
IB/SA: Check dlid before SA agent queries for ClassPortInfo SA queries SM for class port info when there is a LID_CHANGE event. When a base lid is configured before fm is started ie when smlid is not yet assigned, SA handles the LID_CHANGE event and tries query SM with lid 0. This will cause an hang. [ 1106.958820] INFO: task kworker/2:0:23 blocked for more than 120 seconds. [ 1106.965082] Tainted: G O 4.12.0+ #1 [ 1106.969602] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message. [ 1106.977227] kworker/2:0 D 0 23 2 0x00000000 [ 1106.977250] Workqueue: infiniband update_ib_cpi [ib_core] [ 1106.977261] Call Trace: [ 1106.977273] __schedule+0x28e/0x860 [ 1106.977285] schedule+0x36/0x80 [ 1106.977298] schedule_timeout+0x1a3/0x2e0 [ 1106.977310] ? radix_tree_iter_tag_clear+0x1b/0x20 [ 1106.977322] ? idr_alloc+0x64/0x90 [ 1106.977334] wait_for_completion+0xe3/0x140 [ 1106.977347] ? wake_up_q+0x80/0x80 [ 1106.977369] update_ib_cpi+0x163/0x210 [ib_core] [ 1106.977381] process_one_work+0x147/0x370 [ 1106.977394] worker_thread+0x4a/0x390 [ 1106.977406] kthread+0x109/0x140 [ 1106.977418] ? process_one_work+0x370/0x370 [ 1106.977430] ? kthread_park+0x60/0x60 [ 1106.977443] ret_from_fork+0x22/0x30 Always ensure a proper smlid is assigned before querying SM for cpi. Fixes: ee1c60b1bff ("IB/SA: Modify SA to implicitly cache Class Port info") Reviewed-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Venkata Sandeep Dhanalakota <venkata.s.dhanalakota@intel.com> Signed-off-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2017-12-19 11:26:58 +08:00
/*
* Always check if sm_ah has valid dlid assigned,
* before querying for class port info
*/
if ((rdma_query_ah(query->sm_ah->ah, &ah_attr) < 0) ||
!rdma_is_valid_unicast_lid(&ah_attr)) {
kref_put(&query->sm_ah->ref, free_sm_ah);
return -EAGAIN;
}
IB/sa: Make sure SA queries use default P_Key MADs sent to the SA should use the the default P_Key (0x7fff/0xffff). There's no requirement that the default P_Key is stored at index 0 in the local P_Key table, so add code to the sa_query module to look up the index of the default P_Key when creating an address handle for the SA (which is done any time the P_Key table might change), and use this index for all SA queries. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2007-06-19 02:03:58 +08:00
query->mad_buf = ib_create_send_mad(query->port->agent, 1,
query->sm_ah->pkey_index,
0, IB_MGMT_SA_HDR, IB_MGMT_SA_DATA,
IB/mad: Support alternate Base Versions when creating MADs In preparation to support the new OPA MAD Base version, add a base version parameter to ib_create_send_mad and set it to IB_MGMT_BASE_VERSION for current users. Definition of the new base version and it's processing will occur in later patches. Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-06-07 02:38:28 +08:00
gfp_mask,
IB/SA: Add support to query opa classport info. For OPA devices, SA will query the OPA classport info instead of the IB defined classport info. opa classport info exposes additional information and capabilities that are specific to OPA devices. Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-29 07:20:26 +08:00
((query->flags & IB_SA_QUERY_OPA) ?
OPA_MGMT_BASE_VERSION :
IB_MGMT_BASE_VERSION));
IB/sa: Error handling thinko fix ib_create_send_mad() returns an error code pointer on error, not NULL. Signed-off-by: Michael S. Tsirkin <mst@dev.mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2007-09-09 19:55:11 +08:00
if (IS_ERR(query->mad_buf)) {
IB/sa: Make sure SA queries use default P_Key MADs sent to the SA should use the the default P_Key (0x7fff/0xffff). There's no requirement that the default P_Key is stored at index 0 in the local P_Key table, so add code to the sa_query module to look up the index of the default P_Key when creating an address handle for the SA (which is done any time the P_Key table might change), and use this index for all SA queries. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2007-06-19 02:03:58 +08:00
kref_put(&query->sm_ah->ref, free_sm_ah);
return -ENOMEM;
}
query->mad_buf->ah = query->sm_ah->ah;
return 0;
}
static void free_mad(struct ib_sa_query *query)
{
ib_free_send_mad(query->mad_buf);
kref_put(&query->sm_ah->ref, free_sm_ah);
}
IB/SA: Add support to query opa classport info. For OPA devices, SA will query the OPA classport info instead of the IB defined classport info. opa classport info exposes additional information and capabilities that are specific to OPA devices. Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-29 07:20:26 +08:00
static void init_mad(struct ib_sa_query *query, struct ib_mad_agent *agent)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
{
IB/SA: Add support to query opa classport info. For OPA devices, SA will query the OPA classport info instead of the IB defined classport info. opa classport info exposes additional information and capabilities that are specific to OPA devices. Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-29 07:20:26 +08:00
struct ib_sa_mad *mad = query->mad_buf->mad;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
unsigned long flags;
memset(mad, 0, sizeof *mad);
IB/SA: Add support to query opa classport info. For OPA devices, SA will query the OPA classport info instead of the IB defined classport info. opa classport info exposes additional information and capabilities that are specific to OPA devices. Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-29 07:20:26 +08:00
if (query->flags & IB_SA_QUERY_OPA) {
mad->mad_hdr.base_version = OPA_MGMT_BASE_VERSION;
mad->mad_hdr.class_version = OPA_SA_CLASS_VERSION;
} else {
mad->mad_hdr.base_version = IB_MGMT_BASE_VERSION;
mad->mad_hdr.class_version = IB_SA_CLASS_VERSION;
}
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
mad->mad_hdr.mgmt_class = IB_MGMT_CLASS_SUBN_ADM;
spin_lock_irqsave(&tid_lock, flags);
mad->mad_hdr.tid =
cpu_to_be64(((u64) agent->hi_tid) << 32 | tid++);
spin_unlock_irqrestore(&tid_lock, flags);
}
RDMA/core: Annotate timeout as unsigned long The ucma users supply timeout in u32 format, it means that any number with most significant bit set will be converted to negative value by various rdma_*, cma_* and sa_query functions, which treat timeout as int. In the lowest level, the timeout is converted back to be unsigned long. Remove this ambiguous conversion by updating all function signatures to receive unsigned long. Reported-by: Noa Osherovich <noaos@mellanox.com> Signed-off-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2018-10-11 22:30:05 +08:00
static int send_mad(struct ib_sa_query *query, unsigned long timeout_ms,
gfp_t gfp_mask)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
{
unsigned long flags;
[IB] Fix MAD layer DMA mappings to avoid touching data buffer once mapped The MAD layer was violating the DMA API by touching data buffers used for sends after the DMA mapping was done. This causes problems on non-cache-coherent architectures, because the device doing DMA won't see updates to the payload buffers that exist only in the CPU cache. Fix this by having all MAD consumers use ib_create_send_mad() to allocate their send buffers, and moving the DMA mapping into the MAD layer so it can be done just before calling send (and after any modifications of the send buffer by the MAD layer). Tested on a non-cache-coherent PowerPC 440SPe system. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-10-26 01:51:39 +08:00
int ret, id;
RDMA/core/sa_query: Retry SA queries A MAD packet is sent as an unreliable datagram (UD). SA requests are sent as MAD packets. As such, SA requests or responses may be silently dropped. IB Core's MAD layer has a timeout and retry mechanism, which amongst other, is used by RDMA CM. But it is not used by SA queries. The lack of retries of SA queries leads to long specified timeout, and error being returned in case of packet loss. The ULP or user-land process has to perform the retry. Fix this by taking advantage of the MAD layer's retry mechanism. First, a check against a zero timeout is added in rdma_resolve_route(). In send_mad(), we set the MAD layer timeout to one tenth of the specified timeout and the number of retries to 10. The special case when timeout is less than 10 is handled. With this fix: # ucmatose -c 1000 -S 1024 -C 1 runs stable on an Infiniband fabric. Without this fix, we see an intermittent behavior and it errors out with: cmatose: event: RDMA_CM_EVENT_ROUTE_ERROR, error: -110 (110 is ETIMEDOUT) Link: https://lore.kernel.org/r/1628784755-28316-1-git-send-email-haakon.bugge@oracle.com Signed-off-by: Håkon Bugge <haakon.bugge@oracle.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
2021-08-13 00:12:35 +08:00
const int nmbr_sa_query_retries = 10;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
ib core: Convert query_idr to XArray Signed-off-by: Matthew Wilcox <willy@infradead.org> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2019-02-21 08:20:46 +08:00
xa_lock_irqsave(&queries, flags);
ret = __xa_alloc(&queries, &id, query, xa_limit_32b, gfp_mask);
xa_unlock_irqrestore(&queries, flags);
if (ret < 0)
return ret;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
RDMA/core/sa_query: Retry SA queries A MAD packet is sent as an unreliable datagram (UD). SA requests are sent as MAD packets. As such, SA requests or responses may be silently dropped. IB Core's MAD layer has a timeout and retry mechanism, which amongst other, is used by RDMA CM. But it is not used by SA queries. The lack of retries of SA queries leads to long specified timeout, and error being returned in case of packet loss. The ULP or user-land process has to perform the retry. Fix this by taking advantage of the MAD layer's retry mechanism. First, a check against a zero timeout is added in rdma_resolve_route(). In send_mad(), we set the MAD layer timeout to one tenth of the specified timeout and the number of retries to 10. The special case when timeout is less than 10 is handled. With this fix: # ucmatose -c 1000 -S 1024 -C 1 runs stable on an Infiniband fabric. Without this fix, we see an intermittent behavior and it errors out with: cmatose: event: RDMA_CM_EVENT_ROUTE_ERROR, error: -110 (110 is ETIMEDOUT) Link: https://lore.kernel.org/r/1628784755-28316-1-git-send-email-haakon.bugge@oracle.com Signed-off-by: Håkon Bugge <haakon.bugge@oracle.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
2021-08-13 00:12:35 +08:00
query->mad_buf->timeout_ms = timeout_ms / nmbr_sa_query_retries;
query->mad_buf->retries = nmbr_sa_query_retries;
if (!query->mad_buf->timeout_ms) {
/* Special case, very small timeout_ms */
query->mad_buf->timeout_ms = 1;
query->mad_buf->retries = timeout_ms;
}
[IB] Fix MAD layer DMA mappings to avoid touching data buffer once mapped The MAD layer was violating the DMA API by touching data buffers used for sends after the DMA mapping was done. This causes problems on non-cache-coherent architectures, because the device doing DMA won't see updates to the payload buffers that exist only in the CPU cache. Fix this by having all MAD consumers use ib_create_send_mad() to allocate their send buffers, and moving the DMA mapping into the MAD layer so it can be done just before calling send (and after any modifications of the send buffer by the MAD layer). Tested on a non-cache-coherent PowerPC 440SPe system. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-10-26 01:51:39 +08:00
query->mad_buf->context[0] = query;
query->id = id;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
IB/SA: Add support to query OPA path records When the bit 26 of capmask2 field in OPA classport info query is set, SA will query for OPA path records instead of querying for IB path records. Note that OPA path records can only be queried by kernel ULPs. Userspace clients continue to query IB path records. Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-28 07:06:02 +08:00
if ((query->flags & IB_SA_ENABLE_LOCAL_SERVICE) &&
(!(query->flags & IB_SA_QUERY_OPA))) {
RDMA/netlink: Simplify netlink listener existence check All users of rdma_nl_chk_listeners() are interested to get boolean answer if netlink socket has listeners, so update all places to boolean function. Signed-off-by: Leon Romanovsky <leonro@mellanox.com> Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2018-10-02 16:49:24 +08:00
if (rdma_nl_chk_listeners(RDMA_NL_GROUP_LS)) {
IB/sa: Put netlink request into the request list before sending It was found by Saurabh Sengar that the netlink code tried to allocate memory with GFP_KERNEL while holding a spinlock. While it is possible to fix the issue by replacing GFP_KERNEL with GFP_ATOMIC, it is better to get rid of the spinlock while sending the packet. However, in order to protect against a race condition that a quick response may be received before the request is put on the request list, we need to put the request on the list first. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reported-by: Saurabh Sengar <saurabh.truth@gmail.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-10-30 20:23:45 +08:00
if (!ib_nl_make_request(query, gfp_mask))
IB/sa: Route SA pathrecord query through netlink This patch routes a SA pathrecord query to netlink first and processes the response appropriately. If a failure is returned, the request will be sent through IB. The decision whether to route the request to netlink first is determined by the presence of a listener for the local service netlink multicast group. If the user-space local service netlink multicast group listener is not present, the request will be sent through IB, just like what is currently being done. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: John Fleck <john.fleck@intel.com> Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-14 20:52:09 +08:00
return id;
}
ib_sa_disable_local_svc(query);
}
[IB] Fix MAD layer DMA mappings to avoid touching data buffer once mapped The MAD layer was violating the DMA API by touching data buffers used for sends after the DMA mapping was done. This causes problems on non-cache-coherent architectures, because the device doing DMA won't see updates to the payload buffers that exist only in the CPU cache. Fix this by having all MAD consumers use ib_create_send_mad() to allocate their send buffers, and moving the DMA mapping into the MAD layer so it can be done just before calling send (and after any modifications of the send buffer by the MAD layer). Tested on a non-cache-coherent PowerPC 440SPe system. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-10-26 01:51:39 +08:00
ret = ib_post_send_mad(query->mad_buf, NULL);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
if (ret) {
ib core: Convert query_idr to XArray Signed-off-by: Matthew Wilcox <willy@infradead.org> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2019-02-21 08:20:46 +08:00
xa_lock_irqsave(&queries, flags);
__xa_erase(&queries, id);
xa_unlock_irqrestore(&queries, flags);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
[PATCH] IB: Fix race in sa_query Use a copy of the id we'll return to the consumer so that we don't dereference query->sa_query after calling send_mad(). A completion may occur very quickly and end up freeing the query before we get to do anything after send_mad(). Signed-off-by: Roland Dreier <rolandd@cisco.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-06-28 05:36:46 +08:00
/*
* It's not safe to dereference query any more, because the
* send may already have completed and freed the query in
[IB] Fix MAD layer DMA mappings to avoid touching data buffer once mapped The MAD layer was violating the DMA API by touching data buffers used for sends after the DMA mapping was done. This causes problems on non-cache-coherent architectures, because the device doing DMA won't see updates to the payload buffers that exist only in the CPU cache. Fix this by having all MAD consumers use ib_create_send_mad() to allocate their send buffers, and moving the DMA mapping into the MAD layer so it can be done just before calling send (and after any modifications of the send buffer by the MAD layer). Tested on a non-cache-coherent PowerPC 440SPe system. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-10-26 01:51:39 +08:00
* another context.
[PATCH] IB: Fix race in sa_query Use a copy of the id we'll return to the consumer so that we don't dereference query->sa_query after calling send_mad(). A completion may occur very quickly and end up freeing the query before we get to do anything after send_mad(). Signed-off-by: Roland Dreier <rolandd@cisco.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-06-28 05:36:46 +08:00
*/
[IB] Fix MAD layer DMA mappings to avoid touching data buffer once mapped The MAD layer was violating the DMA API by touching data buffers used for sends after the DMA mapping was done. This causes problems on non-cache-coherent architectures, because the device doing DMA won't see updates to the payload buffers that exist only in the CPU cache. Fix this by having all MAD consumers use ib_create_send_mad() to allocate their send buffers, and moving the DMA mapping into the MAD layer so it can be done just before calling send (and after any modifications of the send buffer by the MAD layer). Tested on a non-cache-coherent PowerPC 440SPe system. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-10-26 01:51:39 +08:00
return ret ? ret : id;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
IB/SA: Rename ib_sa_path_rec to sa_path_rec Rename ib_sa_path_rec to a more generic sa_path_rec. This is part of extending ib_sa to also support OPA path records in addition to the IB defined path records. Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-28 07:05:58 +08:00
void ib_sa_unpack_path(void *attribute, struct sa_path_rec *rec)
RDMA/ucma: Add option to manually set IB path Export rdma_set_ib_paths to user space to allow applications to manually set the IB path used for connections. This allows alternative ways for a user space application or library to obtain path record information, including retrieving path information from cached data, avoiding direct interaction with the IB SA. The IB SA is a single, centralized entity that can limit scaling on large clusters running MPI applications. Future changes to the rdma cm can expand on this framework to support the full range of features allowed by the IB CM, such as separate forward and reverse paths and APM. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Reviewed-By: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2009-11-17 01:30:33 +08:00
{
ib_unpack(path_rec_table, ARRAY_SIZE(path_rec_table), attribute, rec);
}
EXPORT_SYMBOL(ib_sa_unpack_path);
IB/SA: Rename ib_sa_path_rec to sa_path_rec Rename ib_sa_path_rec to a more generic sa_path_rec. This is part of extending ib_sa to also support OPA path records in addition to the IB defined path records. Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-28 07:05:58 +08:00
void ib_sa_pack_path(struct sa_path_rec *rec, void *attribute)
IB/sa: Export function to pack a path record into wire format Allow converting from struct ib_sa_path_rec to the IB defined SA path record wire format. This will be used to report path data from the rdma cm into user space. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <roland@purestorage.com>
2013-05-30 01:09:26 +08:00
{
ib_pack(path_rec_table, ARRAY_SIZE(path_rec_table), rec, attribute);
}
EXPORT_SYMBOL(ib_sa_pack_path);
IB/SA: Add support to query OPA path records When the bit 26 of capmask2 field in OPA classport info query is set, SA will query for OPA path records instead of querying for IB path records. Note that OPA path records can only be queried by kernel ULPs. Userspace clients continue to query IB path records. Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-28 07:06:02 +08:00
static bool ib_sa_opa_pathrecord_support(struct ib_sa_client *client,
RDMA: Remove a few extra calls to ib_get_client_data() These four places already have easy access to the client data, just use that instead. Link: https://lore.kernel.org/r/0-v1-fae83f600b4a+68-less_get_client_data%25jgg@mellanox.com Acked-by: Ursula Braun <ubraun@linux.ibm.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-04-08 07:20:09 +08:00
struct ib_sa_device *sa_dev,
RDMA: Support more than 255 rdma ports Current code uses many different types when dealing with a port of a RDMA device: u8, unsigned int and u32. Switch to u32 to clean up the logic. This allows us to make (at least) the core view consistent and use the same type. Unfortunately not all places can be converted. Many uverbs functions expect port to be u8 so keep those places in order not to break UAPIs. HW/Spec defined values must also not be changed. With the switch to u32 we now can support devices with more than 255 ports. U32_MAX is reserved to make control logic a bit easier to deal with. As a device with U32_MAX ports probably isn't going to happen any time soon this seems like a non issue. When a device with more than 255 ports is created uverbs will report the RDMA device as having 255 ports as this is the max currently supported. The verbs interface is not changed yet because the IBTA spec limits the port size in too many places to be u8 and all applications that relies in verbs won't be able to cope with this change. At this stage, we are extending the interfaces that are using vendor channel solely Once the limitation is lifted mlx5 in switchdev mode will be able to have thousands of SFs created by the device. As the only instance of an RDMA device that reports more than 255 ports will be a representor device and it exposes itself as a RAW Ethernet only device CM/MAD/IPoIB and other ULPs aren't effected by this change and their sysfs/interfaces that are exposes to userspace can remain unchanged. While here cleanup some alignment issues and remove unneeded sanity checks (mainly in rdmavt), Link: https://lore.kernel.org/r/20210301070420.439400-1-leon@kernel.org Signed-off-by: Mark Bloch <mbloch@nvidia.com> Signed-off-by: Leon Romanovsky <leonro@nvidia.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
2021-03-01 15:04:20 +08:00
u32 port_num)
IB/SA: Add support to query OPA path records When the bit 26 of capmask2 field in OPA classport info query is set, SA will query for OPA path records instead of querying for IB path records. Note that OPA path records can only be queried by kernel ULPs. Userspace clients continue to query IB path records. Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-28 07:06:02 +08:00
{
struct ib_sa_port *port;
unsigned long flags;
bool ret = false;
port = &sa_dev->port[port_num - sa_dev->start_port];
spin_lock_irqsave(&port->classport_lock, flags);
if (!port->classport_info.valid)
goto ret;
if (port->classport_info.data.type == RDMA_CLASS_PORT_INFO_OPA)
ret = opa_get_cpi_capmask2(&port->classport_info.data.opa) &
OPA_CLASS_PORT_INFO_PR_SUPPORT;
ret:
spin_unlock_irqrestore(&port->classport_lock, flags);
return ret;
}
enum opa_pr_supported {
PR_NOT_SUPPORTED,
PR_OPA_SUPPORTED,
PR_IB_SUPPORTED
};
RDMA/core/sa_query: Demote non-conformant kernel-doc header Fixes the following W=1 kernel build warning(s): drivers/infiniband/core/sa_query.c:1449: warning: Function parameter or member 'client' not described in 'opa_pr_query_possible' drivers/infiniband/core/sa_query.c:1449: warning: Function parameter or member 'sa_dev' not described in 'opa_pr_query_possible' drivers/infiniband/core/sa_query.c:1449: warning: Function parameter or member 'device' not described in 'opa_pr_query_possible' drivers/infiniband/core/sa_query.c:1449: warning: Function parameter or member 'port_num' not described in 'opa_pr_query_possible' drivers/infiniband/core/sa_query.c:1449: warning: Function parameter or member 'rec' not described in 'opa_pr_query_possible' Link: https://lore.kernel.org/r/20210118223929.512175-13-lee.jones@linaro.org Cc: Doug Ledford <dledford@redhat.com> Cc: Jason Gunthorpe <jgg@ziepe.ca> Cc: linux-rdma@vger.kernel.org Signed-off-by: Lee Jones <lee.jones@linaro.org> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
2021-01-19 06:39:21 +08:00
/*
IB: Fix kernel-doc markups Some functions have different names between their prototypes and the kernel-doc markup. Others need to be fixed, as kernel-doc markups should use this format: identifier - description Link: https://lore.kernel.org/r/78b98c41a5a0f4c0106433d305b143028a4168b0.1606823973.git.mchehab+huawei@kernel.org Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
2020-12-01 20:08:55 +08:00
* opa_pr_query_possible - Check if current PR query can be an OPA query.
*
IB/SA: Add support to query OPA path records When the bit 26 of capmask2 field in OPA classport info query is set, SA will query for OPA path records instead of querying for IB path records. Note that OPA path records can only be queried by kernel ULPs. Userspace clients continue to query IB path records. Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-28 07:06:02 +08:00
* Retuns PR_NOT_SUPPORTED if a path record query is not
* possible, PR_OPA_SUPPORTED if an OPA path record query
* is possible and PR_IB_SUPPORTED if an IB path record
* query is possible.
*/
static int opa_pr_query_possible(struct ib_sa_client *client,
RDMA: Remove a few extra calls to ib_get_client_data() These four places already have easy access to the client data, just use that instead. Link: https://lore.kernel.org/r/0-v1-fae83f600b4a+68-less_get_client_data%25jgg@mellanox.com Acked-by: Ursula Braun <ubraun@linux.ibm.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-04-08 07:20:09 +08:00
struct ib_sa_device *sa_dev,
RDMA/core/sa_query: Remove unused argument Fixes:4c33bd1926cc ("IB/SA: Add support to query OPA path records") Link: https://lore.kernel.org/r/1624624257-3677-1-git-send-email-haakon.bugge@oracle.com Signed-off-by: Håkon Bugge <haakon.bugge@oracle.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
2021-06-25 20:30:57 +08:00
struct ib_device *device, u32 port_num)
IB/SA: Add support to query OPA path records When the bit 26 of capmask2 field in OPA classport info query is set, SA will query for OPA path records instead of querying for IB path records. Note that OPA path records can only be queried by kernel ULPs. Userspace clients continue to query IB path records. Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-28 07:06:02 +08:00
{
struct ib_port_attr port_attr;
if (ib_query_port(device, port_num, &port_attr))
return PR_NOT_SUPPORTED;
RDMA: Remove a few extra calls to ib_get_client_data() These four places already have easy access to the client data, just use that instead. Link: https://lore.kernel.org/r/0-v1-fae83f600b4a+68-less_get_client_data%25jgg@mellanox.com Acked-by: Ursula Braun <ubraun@linux.ibm.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-04-08 07:20:09 +08:00
if (ib_sa_opa_pathrecord_support(client, sa_dev, port_num))
IB/SA: Add support to query OPA path records When the bit 26 of capmask2 field in OPA classport info query is set, SA will query for OPA path records instead of querying for IB path records. Note that OPA path records can only be queried by kernel ULPs. Userspace clients continue to query IB path records. Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-28 07:06:02 +08:00
return PR_OPA_SUPPORTED;
if (port_attr.lid >= be16_to_cpu(IB_MULTICAST_LID_BASE))
return PR_NOT_SUPPORTED;
else
return PR_IB_SUPPORTED;
}
RDMA/cma: Refactor the inbound/outbound path records process flow Refactors based on comments [1] of the multiple path records support patchset: - Return failure if not able to set inbound/outbound PRs; - Simplify the flow when receiving the PRs from netlink channel: When a good PR response is received, unpack it and call the path_query callback directly. This saves two memory allocations; - Define RDMA_PRIMARY_PATH_MAX_REC_NUM in a proper place. [1] https://lore.kernel.org/linux-rdma/Yyxp9E9pJtUids2o@nvidia.com/ Signed-off-by: Mark Zhang <markzhang@nvidia.com> Reviewed-by: Bart Van Assche <bvanassche@acm.org> #srp Link: https://lore.kernel.org/r/7610025d57342b8b6da0f19516c9612f9c3fdc37.1672819376.git.leonro@nvidia.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2023-01-04 16:03:41 +08:00
static void ib_sa_path_rec_callback(struct ib_sa_query *sa_query,
int status, struct ib_sa_mad *mad)
RDMA/cma: Multiple path records support with netlink channel Support receiving inbound and outbound IB path records (along with GMP PathRecord) from user-space service through the RDMA netlink channel. The LIDs in these 3 PRs can be used in this way: 1. GMP PR: used as the standard local/remote LIDs; 2. DLID of outbound PR: Used as the "dlid" field for outbound traffic; 3. DLID of inbound PR: Used as the "dlid" field for outbound traffic in responder side. This is aimed to support adaptive routing. With current IB routing solution when a packet goes out it's assigned with a fixed DLID per target, meaning a fixed router will be used. The LIDs in inbound/outbound path records can be used to identify group of routers that allow communication with another subnet's entity. With them packets from an inter-subnet connection may travel through any router in the set to reach the target. As confirmed with Jason, when sending a netlink request, kernel uses LS_RESOLVE_PATH_USE_ALL so that the service knows kernel supports multiple PRs. Signed-off-by: Mark Zhang <markzhang@nvidia.com> Reviewed-by: Mark Bloch <mbloch@nvidia.com> Link: https://lore.kernel.org/r/2fa2b6c93c4c16c8915bac3cfc4f27be1d60519d.1662631201.git.leonro@nvidia.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2022-09-08 18:09:01 +08:00
{
RDMA/cma: Refactor the inbound/outbound path records process flow Refactors based on comments [1] of the multiple path records support patchset: - Return failure if not able to set inbound/outbound PRs; - Simplify the flow when receiving the PRs from netlink channel: When a good PR response is received, unpack it and call the path_query callback directly. This saves two memory allocations; - Define RDMA_PRIMARY_PATH_MAX_REC_NUM in a proper place. [1] https://lore.kernel.org/linux-rdma/Yyxp9E9pJtUids2o@nvidia.com/ Signed-off-by: Mark Zhang <markzhang@nvidia.com> Reviewed-by: Bart Van Assche <bvanassche@acm.org> #srp Link: https://lore.kernel.org/r/7610025d57342b8b6da0f19516c9612f9c3fdc37.1672819376.git.leonro@nvidia.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2023-01-04 16:03:41 +08:00
struct ib_sa_path_query *query =
container_of(sa_query, struct ib_sa_path_query, sa_query);
RDMA/cma: Multiple path records support with netlink channel Support receiving inbound and outbound IB path records (along with GMP PathRecord) from user-space service through the RDMA netlink channel. The LIDs in these 3 PRs can be used in this way: 1. GMP PR: used as the standard local/remote LIDs; 2. DLID of outbound PR: Used as the "dlid" field for outbound traffic; 3. DLID of inbound PR: Used as the "dlid" field for outbound traffic in responder side. This is aimed to support adaptive routing. With current IB routing solution when a packet goes out it's assigned with a fixed DLID per target, meaning a fixed router will be used. The LIDs in inbound/outbound path records can be used to identify group of routers that allow communication with another subnet's entity. With them packets from an inter-subnet connection may travel through any router in the set to reach the target. As confirmed with Jason, when sending a netlink request, kernel uses LS_RESOLVE_PATH_USE_ALL so that the service knows kernel supports multiple PRs. Signed-off-by: Mark Zhang <markzhang@nvidia.com> Reviewed-by: Mark Bloch <mbloch@nvidia.com> Link: https://lore.kernel.org/r/2fa2b6c93c4c16c8915bac3cfc4f27be1d60519d.1662631201.git.leonro@nvidia.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2022-09-08 18:09:01 +08:00
struct sa_path_rec rec = {};
RDMA/cma: Refactor the inbound/outbound path records process flow Refactors based on comments [1] of the multiple path records support patchset: - Return failure if not able to set inbound/outbound PRs; - Simplify the flow when receiving the PRs from netlink channel: When a good PR response is received, unpack it and call the path_query callback directly. This saves two memory allocations; - Define RDMA_PRIMARY_PATH_MAX_REC_NUM in a proper place. [1] https://lore.kernel.org/linux-rdma/Yyxp9E9pJtUids2o@nvidia.com/ Signed-off-by: Mark Zhang <markzhang@nvidia.com> Reviewed-by: Bart Van Assche <bvanassche@acm.org> #srp Link: https://lore.kernel.org/r/7610025d57342b8b6da0f19516c9612f9c3fdc37.1672819376.git.leonro@nvidia.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2023-01-04 16:03:41 +08:00
if (!mad) {
query->callback(status, NULL, 0, query->context);
return;
}
if (sa_query->flags & IB_SA_QUERY_OPA) {
ib_unpack(opa_path_rec_table, ARRAY_SIZE(opa_path_rec_table),
mad->data, &rec);
rec.rec_type = SA_PATH_REC_TYPE_OPA;
query->callback(status, &rec, 1, query->context);
return;
}
RDMA/cma: Multiple path records support with netlink channel Support receiving inbound and outbound IB path records (along with GMP PathRecord) from user-space service through the RDMA netlink channel. The LIDs in these 3 PRs can be used in this way: 1. GMP PR: used as the standard local/remote LIDs; 2. DLID of outbound PR: Used as the "dlid" field for outbound traffic; 3. DLID of inbound PR: Used as the "dlid" field for outbound traffic in responder side. This is aimed to support adaptive routing. With current IB routing solution when a packet goes out it's assigned with a fixed DLID per target, meaning a fixed router will be used. The LIDs in inbound/outbound path records can be used to identify group of routers that allow communication with another subnet's entity. With them packets from an inter-subnet connection may travel through any router in the set to reach the target. As confirmed with Jason, when sending a netlink request, kernel uses LS_RESOLVE_PATH_USE_ALL so that the service knows kernel supports multiple PRs. Signed-off-by: Mark Zhang <markzhang@nvidia.com> Reviewed-by: Mark Bloch <mbloch@nvidia.com> Link: https://lore.kernel.org/r/2fa2b6c93c4c16c8915bac3cfc4f27be1d60519d.1662631201.git.leonro@nvidia.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2022-09-08 18:09:01 +08:00
ib_unpack(path_rec_table, ARRAY_SIZE(path_rec_table),
mad->data, &rec);
rec.rec_type = SA_PATH_REC_TYPE_IB;
sa_path_set_dmac_zero(&rec);
if (query->conv_pr) {
struct sa_path_rec opa;
memset(&opa, 0, sizeof(struct sa_path_rec));
sa_convert_path_ib_to_opa(&opa, &rec);
query->callback(status, &opa, 1, query->context);
} else {
query->callback(status, &rec, 1, query->context);
}
}
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
static void ib_sa_path_rec_release(struct ib_sa_query *sa_query)
{
IB/SA: Add support to query OPA path records When the bit 26 of capmask2 field in OPA classport info query is set, SA will query for OPA path records instead of querying for IB path records. Note that OPA path records can only be queried by kernel ULPs. Userspace clients continue to query IB path records. Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-28 07:06:02 +08:00
struct ib_sa_path_query *query =
container_of(sa_query, struct ib_sa_path_query, sa_query);
kfree(query->conv_pr);
kfree(query);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
/**
* ib_sa_path_rec_get - Start a Path get query
IB/sa: Require SA registration Require users to register with SA module, to prevent the sa_query module text from going away while an SA query callback is still running. Update all in-tree users for the new interface. Signed-off-by: Michael S. Tsirkin <mst@mellanox.co.il> Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2006-08-22 07:40:12 +08:00
* @client:SA client
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
* @device:device to send query on
* @port_num: port number to send query on
* @rec:Path Record to send in query
* @comp_mask:component mask to send in query
* @timeout_ms:time to wait for response
* @gfp_mask:GFP mask to use for internal allocations
* @callback:function called when query completes, times out or is
* canceled
* @context:opaque user context passed to callback
* @sa_query:query context, used to cancel query
*
* Send a Path Record Get query to the SA to look up a path. The
* callback function will be called when the query completes (or
* fails); status is 0 for a successful response, -EINTR if the query
* is canceled, -ETIMEDOUT is the query timed out, or -EIO if an error
* occurred sending the query. The resp parameter of the callback is
* only valid if status is 0.
*
* If the return value of ib_sa_path_rec_get() is negative, it is an
* error code. Otherwise it is a query ID that can be used to cancel
* the query.
*/
IB/sa: Require SA registration Require users to register with SA module, to prevent the sa_query module text from going away while an SA query callback is still running. Update all in-tree users for the new interface. Signed-off-by: Michael S. Tsirkin <mst@mellanox.co.il> Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2006-08-22 07:40:12 +08:00
int ib_sa_path_rec_get(struct ib_sa_client *client,
RDMA: Support more than 255 rdma ports Current code uses many different types when dealing with a port of a RDMA device: u8, unsigned int and u32. Switch to u32 to clean up the logic. This allows us to make (at least) the core view consistent and use the same type. Unfortunately not all places can be converted. Many uverbs functions expect port to be u8 so keep those places in order not to break UAPIs. HW/Spec defined values must also not be changed. With the switch to u32 we now can support devices with more than 255 ports. U32_MAX is reserved to make control logic a bit easier to deal with. As a device with U32_MAX ports probably isn't going to happen any time soon this seems like a non issue. When a device with more than 255 ports is created uverbs will report the RDMA device as having 255 ports as this is the max currently supported. The verbs interface is not changed yet because the IBTA spec limits the port size in too many places to be u8 and all applications that relies in verbs won't be able to cope with this change. At this stage, we are extending the interfaces that are using vendor channel solely Once the limitation is lifted mlx5 in switchdev mode will be able to have thousands of SFs created by the device. As the only instance of an RDMA device that reports more than 255 ports will be a representor device and it exposes itself as a RAW Ethernet only device CM/MAD/IPoIB and other ULPs aren't effected by this change and their sysfs/interfaces that are exposes to userspace can remain unchanged. While here cleanup some alignment issues and remove unneeded sanity checks (mainly in rdmavt), Link: https://lore.kernel.org/r/20210301070420.439400-1-leon@kernel.org Signed-off-by: Mark Bloch <mbloch@nvidia.com> Signed-off-by: Leon Romanovsky <leonro@nvidia.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
2021-03-01 15:04:20 +08:00
struct ib_device *device, u32 port_num,
IB/SA: Rename ib_sa_path_rec to sa_path_rec Rename ib_sa_path_rec to a more generic sa_path_rec. This is part of extending ib_sa to also support OPA path records in addition to the IB defined path records. Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-28 07:05:58 +08:00
struct sa_path_rec *rec,
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
ib_sa_comp_mask comp_mask,
RDMA/core: Annotate timeout as unsigned long The ucma users supply timeout in u32 format, it means that any number with most significant bit set will be converted to negative value by various rdma_*, cma_* and sa_query functions, which treat timeout as int. In the lowest level, the timeout is converted back to be unsigned long. Remove this ambiguous conversion by updating all function signatures to receive unsigned long. Reported-by: Noa Osherovich <noaos@mellanox.com> Signed-off-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2018-10-11 22:30:05 +08:00
unsigned long timeout_ms, gfp_t gfp_mask,
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
void (*callback)(int status,
IB/SA: Rename ib_sa_path_rec to sa_path_rec Rename ib_sa_path_rec to a more generic sa_path_rec. This is part of extending ib_sa to also support OPA path records in addition to the IB defined path records. Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-28 07:05:58 +08:00
struct sa_path_rec *resp,
RDMA/cma: Refactor the inbound/outbound path records process flow Refactors based on comments [1] of the multiple path records support patchset: - Return failure if not able to set inbound/outbound PRs; - Simplify the flow when receiving the PRs from netlink channel: When a good PR response is received, unpack it and call the path_query callback directly. This saves two memory allocations; - Define RDMA_PRIMARY_PATH_MAX_REC_NUM in a proper place. [1] https://lore.kernel.org/linux-rdma/Yyxp9E9pJtUids2o@nvidia.com/ Signed-off-by: Mark Zhang <markzhang@nvidia.com> Reviewed-by: Bart Van Assche <bvanassche@acm.org> #srp Link: https://lore.kernel.org/r/7610025d57342b8b6da0f19516c9612f9c3fdc37.1672819376.git.leonro@nvidia.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2023-01-04 16:03:41 +08:00
unsigned int num_paths, void *context),
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
void *context,
struct ib_sa_query **sa_query)
{
struct ib_sa_path_query *query;
struct ib_sa_device *sa_dev = ib_get_client_data(device, &sa_client);
[IB] fail SA queries if device initialization failed If the SA query module's initialization fails for a device, then that device won't have a struct ib_sa_device associated. We should fail SA queries in that case, rather than blindly dereferencing the NULL pointer we get back from ib_get_client_data(). Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-10-14 01:45:02 +08:00
struct ib_sa_port *port;
struct ib_mad_agent *agent;
[IB] Fix MAD layer DMA mappings to avoid touching data buffer once mapped The MAD layer was violating the DMA API by touching data buffers used for sends after the DMA mapping was done. This causes problems on non-cache-coherent architectures, because the device doing DMA won't see updates to the payload buffers that exist only in the CPU cache. Fix this by having all MAD consumers use ib_create_send_mad() to allocate their send buffers, and moving the DMA mapping into the MAD layer so it can be done just before calling send (and after any modifications of the send buffer by the MAD layer). Tested on a non-cache-coherent PowerPC 440SPe system. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-10-26 01:51:39 +08:00
struct ib_sa_mad *mad;
IB/SA: Add support to query OPA path records When the bit 26 of capmask2 field in OPA classport info query is set, SA will query for OPA path records instead of querying for IB path records. Note that OPA path records can only be queried by kernel ULPs. Userspace clients continue to query IB path records. Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-28 07:06:02 +08:00
enum opa_pr_supported status;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
int ret;
[IB] fail SA queries if device initialization failed If the SA query module's initialization fails for a device, then that device won't have a struct ib_sa_device associated. We should fail SA queries in that case, rather than blindly dereferencing the NULL pointer we get back from ib_get_client_data(). Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-10-14 01:45:02 +08:00
if (!sa_dev)
return -ENODEV;
IB/SA: Add support to query OPA path records When the bit 26 of capmask2 field in OPA classport info query is set, SA will query for OPA path records instead of querying for IB path records. Note that OPA path records can only be queried by kernel ULPs. Userspace clients continue to query IB path records. Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-28 07:06:02 +08:00
if ((rec->rec_type != SA_PATH_REC_TYPE_IB) &&
(rec->rec_type != SA_PATH_REC_TYPE_OPA))
IB/SA: Introduce path record specific types struct sa_path_rec has a gid_type field. This patch introduces a more generic path record specific type 'rec_type' which is either IB, ROCE v1 or ROCE v2. The patch also provides conversion functions to get a gid type from a path record type and vice versa Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-28 07:05:59 +08:00
return -EINVAL;
[IB] fail SA queries if device initialization failed If the SA query module's initialization fails for a device, then that device won't have a struct ib_sa_device associated. We should fail SA queries in that case, rather than blindly dereferencing the NULL pointer we get back from ib_get_client_data(). Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-10-14 01:45:02 +08:00
port = &sa_dev->port[port_num - sa_dev->start_port];
agent = port->agent;
IB/sa: Allocate SA query with kzalloc Replace kmalloc with kzalloc so that all uninitialized fields in SA query will be zero-ed out to avoid unintentional consequence. This prepares the SA query structure to accept new fields in the future. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: John Fleck <john.fleck@intel.com> Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-14 20:52:08 +08:00
query = kzalloc(sizeof(*query), gfp_mask);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
if (!query)
return -ENOMEM;
[IB] Fix MAD layer DMA mappings to avoid touching data buffer once mapped The MAD layer was violating the DMA API by touching data buffers used for sends after the DMA mapping was done. This causes problems on non-cache-coherent architectures, because the device doing DMA won't see updates to the payload buffers that exist only in the CPU cache. Fix this by having all MAD consumers use ib_create_send_mad() to allocate their send buffers, and moving the DMA mapping into the MAD layer so it can be done just before calling send (and after any modifications of the send buffer by the MAD layer). Tested on a non-cache-coherent PowerPC 440SPe system. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-10-26 01:51:39 +08:00
IB/sa: Make sure SA queries use default P_Key MADs sent to the SA should use the the default P_Key (0x7fff/0xffff). There's no requirement that the default P_Key is stored at index 0 in the local P_Key table, so add code to the sa_query module to look up the index of the default P_Key when creating an address handle for the SA (which is done any time the P_Key table might change), and use this index for all SA queries. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2007-06-19 02:03:58 +08:00
query->sa_query.port = port;
IB/SA: Add support to query OPA path records When the bit 26 of capmask2 field in OPA classport info query is set, SA will query for OPA path records instead of querying for IB path records. Note that OPA path records can only be queried by kernel ULPs. Userspace clients continue to query IB path records. Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-28 07:06:02 +08:00
if (rec->rec_type == SA_PATH_REC_TYPE_OPA) {
RDMA/core/sa_query: Remove unused argument Fixes:4c33bd1926cc ("IB/SA: Add support to query OPA path records") Link: https://lore.kernel.org/r/1624624257-3677-1-git-send-email-haakon.bugge@oracle.com Signed-off-by: Håkon Bugge <haakon.bugge@oracle.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
2021-06-25 20:30:57 +08:00
status = opa_pr_query_possible(client, sa_dev, device, port_num);
IB/SA: Add support to query OPA path records When the bit 26 of capmask2 field in OPA classport info query is set, SA will query for OPA path records instead of querying for IB path records. Note that OPA path records can only be queried by kernel ULPs. Userspace clients continue to query IB path records. Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-28 07:06:02 +08:00
if (status == PR_NOT_SUPPORTED) {
ret = -EINVAL;
goto err1;
} else if (status == PR_OPA_SUPPORTED) {
query->sa_query.flags |= IB_SA_QUERY_OPA;
} else {
query->conv_pr =
kmalloc(sizeof(*query->conv_pr), gfp_mask);
if (!query->conv_pr) {
ret = -ENOMEM;
goto err1;
}
}
}
IB/sa: Make sure SA queries use default P_Key MADs sent to the SA should use the the default P_Key (0x7fff/0xffff). There's no requirement that the default P_Key is stored at index 0 in the local P_Key table, so add code to the sa_query module to look up the index of the default P_Key when creating an address handle for the SA (which is done any time the P_Key table might change), and use this index for all SA queries. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2007-06-19 02:03:58 +08:00
ret = alloc_mad(&query->sa_query, gfp_mask);
if (ret)
IB/SA: Add support to query OPA path records When the bit 26 of capmask2 field in OPA classport info query is set, SA will query for OPA path records instead of querying for IB path records. Note that OPA path records can only be queried by kernel ULPs. Userspace clients continue to query IB path records. Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-28 07:06:02 +08:00
goto err2;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
IB/sa: Require SA registration Require users to register with SA module, to prevent the sa_query module text from going away while an SA query callback is still running. Update all in-tree users for the new interface. Signed-off-by: Michael S. Tsirkin <mst@mellanox.co.il> Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2006-08-22 07:40:12 +08:00
ib_sa_client_get(client);
query->sa_query.client = client;
query->callback = callback;
query->context = context;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
[IB] Fix MAD layer DMA mappings to avoid touching data buffer once mapped The MAD layer was violating the DMA API by touching data buffers used for sends after the DMA mapping was done. This causes problems on non-cache-coherent architectures, because the device doing DMA won't see updates to the payload buffers that exist only in the CPU cache. Fix this by having all MAD consumers use ib_create_send_mad() to allocate their send buffers, and moving the DMA mapping into the MAD layer so it can be done just before calling send (and after any modifications of the send buffer by the MAD layer). Tested on a non-cache-coherent PowerPC 440SPe system. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-10-26 01:51:39 +08:00
mad = query->sa_query.mad_buf->mad;
IB/SA: Add support to query opa classport info. For OPA devices, SA will query the OPA classport info instead of the IB defined classport info. opa classport info exposes additional information and capabilities that are specific to OPA devices. Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-29 07:20:26 +08:00
init_mad(&query->sa_query, agent);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
[IB] Fix MAD layer DMA mappings to avoid touching data buffer once mapped The MAD layer was violating the DMA API by touching data buffers used for sends after the DMA mapping was done. This causes problems on non-cache-coherent architectures, because the device doing DMA won't see updates to the payload buffers that exist only in the CPU cache. Fix this by having all MAD consumers use ib_create_send_mad() to allocate their send buffers, and moving the DMA mapping into the MAD layer so it can be done just before calling send (and after any modifications of the send buffer by the MAD layer). Tested on a non-cache-coherent PowerPC 440SPe system. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-10-26 01:51:39 +08:00
query->sa_query.callback = callback ? ib_sa_path_rec_callback : NULL;
query->sa_query.release = ib_sa_path_rec_release;
mad->mad_hdr.method = IB_MGMT_METHOD_GET;
mad->mad_hdr.attr_id = cpu_to_be16(IB_SA_ATTR_PATH_REC);
mad->sa_hdr.comp_mask = comp_mask;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
IB/SA: Add support to query OPA path records When the bit 26 of capmask2 field in OPA classport info query is set, SA will query for OPA path records instead of querying for IB path records. Note that OPA path records can only be queried by kernel ULPs. Userspace clients continue to query IB path records. Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-28 07:06:02 +08:00
if (query->sa_query.flags & IB_SA_QUERY_OPA) {
ib_pack(opa_path_rec_table, ARRAY_SIZE(opa_path_rec_table),
rec, mad->data);
} else if (query->conv_pr) {
sa_convert_path_opa_to_ib(query->conv_pr, rec);
ib_pack(path_rec_table, ARRAY_SIZE(path_rec_table),
query->conv_pr, mad->data);
} else {
ib_pack(path_rec_table, ARRAY_SIZE(path_rec_table),
rec, mad->data);
}
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
*sa_query = &query->sa_query;
[PATCH] IB: Fix race in sa_query Use a copy of the id we'll return to the consumer so that we don't dereference query->sa_query after calling send_mad(). A completion may occur very quickly and end up freeing the query before we get to do anything after send_mad(). Signed-off-by: Roland Dreier <rolandd@cisco.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-06-28 05:36:46 +08:00
IB/sa: Route SA pathrecord query through netlink This patch routes a SA pathrecord query to netlink first and processes the response appropriately. If a failure is returned, the request will be sent through IB. The decision whether to route the request to netlink first is determined by the presence of a listener for the local service netlink multicast group. If the user-space local service netlink multicast group listener is not present, the request will be sent through IB, just like what is currently being done. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: John Fleck <john.fleck@intel.com> Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-14 20:52:09 +08:00
query->sa_query.flags |= IB_SA_ENABLE_LOCAL_SERVICE;
IB/SA: Add support to query OPA path records When the bit 26 of capmask2 field in OPA classport info query is set, SA will query for OPA path records instead of querying for IB path records. Note that OPA path records can only be queried by kernel ULPs. Userspace clients continue to query IB path records. Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-28 07:06:02 +08:00
query->sa_query.mad_buf->context[1] = (query->conv_pr) ?
query->conv_pr : rec;
IB/sa: Route SA pathrecord query through netlink This patch routes a SA pathrecord query to netlink first and processes the response appropriately. If a failure is returned, the request will be sent through IB. The decision whether to route the request to netlink first is determined by the presence of a listener for the local service netlink multicast group. If the user-space local service netlink multicast group listener is not present, the request will be sent through IB, just like what is currently being done. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: John Fleck <john.fleck@intel.com> Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-14 20:52:09 +08:00
[PATCH] IB/core: use correct gfp_mask in sa_query Avoid bogus out of memory errors: fix sa_query to actually pass gfp_mask supplied by the user to idr_pre_get. Signed-off-by: Michael S. Tsirkin <mst@mellanox.co.il> Acked-by: "Sean Hefty" <mshefty@ichips.intel.com> Acked-by: "Roland Dreier" <rdreier@cisco.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-07-14 15:23:56 +08:00
ret = send_mad(&query->sa_query, timeout_ms, gfp_mask);
[IB] Fix MAD layer DMA mappings to avoid touching data buffer once mapped The MAD layer was violating the DMA API by touching data buffers used for sends after the DMA mapping was done. This causes problems on non-cache-coherent architectures, because the device doing DMA won't see updates to the payload buffers that exist only in the CPU cache. Fix this by having all MAD consumers use ib_create_send_mad() to allocate their send buffers, and moving the DMA mapping into the MAD layer so it can be done just before calling send (and after any modifications of the send buffer by the MAD layer). Tested on a non-cache-coherent PowerPC 440SPe system. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-10-26 01:51:39 +08:00
if (ret < 0)
IB/SA: Add support to query OPA path records When the bit 26 of capmask2 field in OPA classport info query is set, SA will query for OPA path records instead of querying for IB path records. Note that OPA path records can only be queried by kernel ULPs. Userspace clients continue to query IB path records. Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-28 07:06:02 +08:00
goto err3;
[IB] Fix MAD layer DMA mappings to avoid touching data buffer once mapped The MAD layer was violating the DMA API by touching data buffers used for sends after the DMA mapping was done. This causes problems on non-cache-coherent architectures, because the device doing DMA won't see updates to the payload buffers that exist only in the CPU cache. Fix this by having all MAD consumers use ib_create_send_mad() to allocate their send buffers, and moving the DMA mapping into the MAD layer so it can be done just before calling send (and after any modifications of the send buffer by the MAD layer). Tested on a non-cache-coherent PowerPC 440SPe system. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-10-26 01:51:39 +08:00
return ret;
IB/SA: Add support to query OPA path records When the bit 26 of capmask2 field in OPA classport info query is set, SA will query for OPA path records instead of querying for IB path records. Note that OPA path records can only be queried by kernel ULPs. Userspace clients continue to query IB path records. Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-28 07:06:02 +08:00
err3:
[IB] Fix MAD layer DMA mappings to avoid touching data buffer once mapped The MAD layer was violating the DMA API by touching data buffers used for sends after the DMA mapping was done. This causes problems on non-cache-coherent architectures, because the device doing DMA won't see updates to the payload buffers that exist only in the CPU cache. Fix this by having all MAD consumers use ib_create_send_mad() to allocate their send buffers, and moving the DMA mapping into the MAD layer so it can be done just before calling send (and after any modifications of the send buffer by the MAD layer). Tested on a non-cache-coherent PowerPC 440SPe system. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-10-26 01:51:39 +08:00
*sa_query = NULL;
IB/sa: Require SA registration Require users to register with SA module, to prevent the sa_query module text from going away while an SA query callback is still running. Update all in-tree users for the new interface. Signed-off-by: Michael S. Tsirkin <mst@mellanox.co.il> Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2006-08-22 07:40:12 +08:00
ib_sa_client_put(query->sa_query.client);
IB/sa: Make sure SA queries use default P_Key MADs sent to the SA should use the the default P_Key (0x7fff/0xffff). There's no requirement that the default P_Key is stored at index 0 in the local P_Key table, so add code to the sa_query module to look up the index of the default P_Key when creating an address handle for the SA (which is done any time the P_Key table might change), and use this index for all SA queries. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2007-06-19 02:03:58 +08:00
free_mad(&query->sa_query);
IB/SA: Add support to query OPA path records When the bit 26 of capmask2 field in OPA classport info query is set, SA will query for OPA path records instead of querying for IB path records. Note that OPA path records can only be queried by kernel ULPs. Userspace clients continue to query IB path records. Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-28 07:06:02 +08:00
err2:
kfree(query->conv_pr);
[IB] Fix MAD layer DMA mappings to avoid touching data buffer once mapped The MAD layer was violating the DMA API by touching data buffers used for sends after the DMA mapping was done. This causes problems on non-cache-coherent architectures, because the device doing DMA won't see updates to the payload buffers that exist only in the CPU cache. Fix this by having all MAD consumers use ib_create_send_mad() to allocate their send buffers, and moving the DMA mapping into the MAD layer so it can be done just before calling send (and after any modifications of the send buffer by the MAD layer). Tested on a non-cache-coherent PowerPC 440SPe system. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-10-26 01:51:39 +08:00
err1:
kfree(query);
[PATCH] IB: Fix race in sa_query Use a copy of the id we'll return to the consumer so that we don't dereference query->sa_query after calling send_mad(). A completion may occur very quickly and end up freeing the query before we get to do anything after send_mad(). Signed-off-by: Roland Dreier <rolandd@cisco.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-06-28 05:36:46 +08:00
return ret;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
EXPORT_SYMBOL(ib_sa_path_rec_get);
static void ib_sa_mcmember_rec_callback(struct ib_sa_query *sa_query,
RDMA/cma: Refactor the inbound/outbound path records process flow Refactors based on comments [1] of the multiple path records support patchset: - Return failure if not able to set inbound/outbound PRs; - Simplify the flow when receiving the PRs from netlink channel: When a good PR response is received, unpack it and call the path_query callback directly. This saves two memory allocations; - Define RDMA_PRIMARY_PATH_MAX_REC_NUM in a proper place. [1] https://lore.kernel.org/linux-rdma/Yyxp9E9pJtUids2o@nvidia.com/ Signed-off-by: Mark Zhang <markzhang@nvidia.com> Reviewed-by: Bart Van Assche <bvanassche@acm.org> #srp Link: https://lore.kernel.org/r/7610025d57342b8b6da0f19516c9612f9c3fdc37.1672819376.git.leonro@nvidia.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2023-01-04 16:03:41 +08:00
int status, struct ib_sa_mad *mad)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
{
struct ib_sa_mcmember_query *query =
container_of(sa_query, struct ib_sa_mcmember_query, sa_query);
if (mad) {
struct ib_sa_mcmember_rec rec;
ib_unpack(mcmember_rec_table, ARRAY_SIZE(mcmember_rec_table),
mad->data, &rec);
query->callback(status, &rec, query->context);
} else
query->callback(status, NULL, query->context);
}
static void ib_sa_mcmember_rec_release(struct ib_sa_query *sa_query)
{
kfree(container_of(sa_query, struct ib_sa_mcmember_query, sa_query));
}
IB/sa: Require SA registration Require users to register with SA module, to prevent the sa_query module text from going away while an SA query callback is still running. Update all in-tree users for the new interface. Signed-off-by: Michael S. Tsirkin <mst@mellanox.co.il> Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2006-08-22 07:40:12 +08:00
int ib_sa_mcmember_rec_query(struct ib_sa_client *client,
RDMA: Support more than 255 rdma ports Current code uses many different types when dealing with a port of a RDMA device: u8, unsigned int and u32. Switch to u32 to clean up the logic. This allows us to make (at least) the core view consistent and use the same type. Unfortunately not all places can be converted. Many uverbs functions expect port to be u8 so keep those places in order not to break UAPIs. HW/Spec defined values must also not be changed. With the switch to u32 we now can support devices with more than 255 ports. U32_MAX is reserved to make control logic a bit easier to deal with. As a device with U32_MAX ports probably isn't going to happen any time soon this seems like a non issue. When a device with more than 255 ports is created uverbs will report the RDMA device as having 255 ports as this is the max currently supported. The verbs interface is not changed yet because the IBTA spec limits the port size in too many places to be u8 and all applications that relies in verbs won't be able to cope with this change. At this stage, we are extending the interfaces that are using vendor channel solely Once the limitation is lifted mlx5 in switchdev mode will be able to have thousands of SFs created by the device. As the only instance of an RDMA device that reports more than 255 ports will be a representor device and it exposes itself as a RAW Ethernet only device CM/MAD/IPoIB and other ULPs aren't effected by this change and their sysfs/interfaces that are exposes to userspace can remain unchanged. While here cleanup some alignment issues and remove unneeded sanity checks (mainly in rdmavt), Link: https://lore.kernel.org/r/20210301070420.439400-1-leon@kernel.org Signed-off-by: Mark Bloch <mbloch@nvidia.com> Signed-off-by: Leon Romanovsky <leonro@nvidia.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
2021-03-01 15:04:20 +08:00
struct ib_device *device, u32 port_num,
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
u8 method,
struct ib_sa_mcmember_rec *rec,
ib_sa_comp_mask comp_mask,
RDMA/core: Annotate timeout as unsigned long The ucma users supply timeout in u32 format, it means that any number with most significant bit set will be converted to negative value by various rdma_*, cma_* and sa_query functions, which treat timeout as int. In the lowest level, the timeout is converted back to be unsigned long. Remove this ambiguous conversion by updating all function signatures to receive unsigned long. Reported-by: Noa Osherovich <noaos@mellanox.com> Signed-off-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2018-10-11 22:30:05 +08:00
unsigned long timeout_ms, gfp_t gfp_mask,
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
void (*callback)(int status,
struct ib_sa_mcmember_rec *resp,
void *context),
void *context,
struct ib_sa_query **sa_query)
{
struct ib_sa_mcmember_query *query;
struct ib_sa_device *sa_dev = ib_get_client_data(device, &sa_client);
[IB] fail SA queries if device initialization failed If the SA query module's initialization fails for a device, then that device won't have a struct ib_sa_device associated. We should fail SA queries in that case, rather than blindly dereferencing the NULL pointer we get back from ib_get_client_data(). Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-10-14 01:45:02 +08:00
struct ib_sa_port *port;
struct ib_mad_agent *agent;
[IB] Fix MAD layer DMA mappings to avoid touching data buffer once mapped The MAD layer was violating the DMA API by touching data buffers used for sends after the DMA mapping was done. This causes problems on non-cache-coherent architectures, because the device doing DMA won't see updates to the payload buffers that exist only in the CPU cache. Fix this by having all MAD consumers use ib_create_send_mad() to allocate their send buffers, and moving the DMA mapping into the MAD layer so it can be done just before calling send (and after any modifications of the send buffer by the MAD layer). Tested on a non-cache-coherent PowerPC 440SPe system. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-10-26 01:51:39 +08:00
struct ib_sa_mad *mad;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
int ret;
[IB] fail SA queries if device initialization failed If the SA query module's initialization fails for a device, then that device won't have a struct ib_sa_device associated. We should fail SA queries in that case, rather than blindly dereferencing the NULL pointer we get back from ib_get_client_data(). Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-10-14 01:45:02 +08:00
if (!sa_dev)
return -ENODEV;
port = &sa_dev->port[port_num - sa_dev->start_port];
agent = port->agent;
IB/sa: Allocate SA query with kzalloc Replace kmalloc with kzalloc so that all uninitialized fields in SA query will be zero-ed out to avoid unintentional consequence. This prepares the SA query structure to accept new fields in the future. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: John Fleck <john.fleck@intel.com> Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-14 20:52:08 +08:00
query = kzalloc(sizeof(*query), gfp_mask);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
if (!query)
return -ENOMEM;
[IB] Fix MAD layer DMA mappings to avoid touching data buffer once mapped The MAD layer was violating the DMA API by touching data buffers used for sends after the DMA mapping was done. This causes problems on non-cache-coherent architectures, because the device doing DMA won't see updates to the payload buffers that exist only in the CPU cache. Fix this by having all MAD consumers use ib_create_send_mad() to allocate their send buffers, and moving the DMA mapping into the MAD layer so it can be done just before calling send (and after any modifications of the send buffer by the MAD layer). Tested on a non-cache-coherent PowerPC 440SPe system. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-10-26 01:51:39 +08:00
IB/sa: Make sure SA queries use default P_Key MADs sent to the SA should use the the default P_Key (0x7fff/0xffff). There's no requirement that the default P_Key is stored at index 0 in the local P_Key table, so add code to the sa_query module to look up the index of the default P_Key when creating an address handle for the SA (which is done any time the P_Key table might change), and use this index for all SA queries. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2007-06-19 02:03:58 +08:00
query->sa_query.port = port;
ret = alloc_mad(&query->sa_query, gfp_mask);
if (ret)
[IB] Fix MAD layer DMA mappings to avoid touching data buffer once mapped The MAD layer was violating the DMA API by touching data buffers used for sends after the DMA mapping was done. This causes problems on non-cache-coherent architectures, because the device doing DMA won't see updates to the payload buffers that exist only in the CPU cache. Fix this by having all MAD consumers use ib_create_send_mad() to allocate their send buffers, and moving the DMA mapping into the MAD layer so it can be done just before calling send (and after any modifications of the send buffer by the MAD layer). Tested on a non-cache-coherent PowerPC 440SPe system. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-10-26 01:51:39 +08:00
goto err1;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
IB/sa: Require SA registration Require users to register with SA module, to prevent the sa_query module text from going away while an SA query callback is still running. Update all in-tree users for the new interface. Signed-off-by: Michael S. Tsirkin <mst@mellanox.co.il> Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2006-08-22 07:40:12 +08:00
ib_sa_client_get(client);
query->sa_query.client = client;
query->callback = callback;
query->context = context;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
[IB] Fix MAD layer DMA mappings to avoid touching data buffer once mapped The MAD layer was violating the DMA API by touching data buffers used for sends after the DMA mapping was done. This causes problems on non-cache-coherent architectures, because the device doing DMA won't see updates to the payload buffers that exist only in the CPU cache. Fix this by having all MAD consumers use ib_create_send_mad() to allocate their send buffers, and moving the DMA mapping into the MAD layer so it can be done just before calling send (and after any modifications of the send buffer by the MAD layer). Tested on a non-cache-coherent PowerPC 440SPe system. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-10-26 01:51:39 +08:00
mad = query->sa_query.mad_buf->mad;
IB/SA: Add support to query opa classport info. For OPA devices, SA will query the OPA classport info instead of the IB defined classport info. opa classport info exposes additional information and capabilities that are specific to OPA devices. Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-29 07:20:26 +08:00
init_mad(&query->sa_query, agent);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
[IB] Fix MAD layer DMA mappings to avoid touching data buffer once mapped The MAD layer was violating the DMA API by touching data buffers used for sends after the DMA mapping was done. This causes problems on non-cache-coherent architectures, because the device doing DMA won't see updates to the payload buffers that exist only in the CPU cache. Fix this by having all MAD consumers use ib_create_send_mad() to allocate their send buffers, and moving the DMA mapping into the MAD layer so it can be done just before calling send (and after any modifications of the send buffer by the MAD layer). Tested on a non-cache-coherent PowerPC 440SPe system. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-10-26 01:51:39 +08:00
query->sa_query.callback = callback ? ib_sa_mcmember_rec_callback : NULL;
query->sa_query.release = ib_sa_mcmember_rec_release;
mad->mad_hdr.method = method;
mad->mad_hdr.attr_id = cpu_to_be16(IB_SA_ATTR_MC_MEMBER_REC);
mad->sa_hdr.comp_mask = comp_mask;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
ib_pack(mcmember_rec_table, ARRAY_SIZE(mcmember_rec_table),
[IB] Fix MAD layer DMA mappings to avoid touching data buffer once mapped The MAD layer was violating the DMA API by touching data buffers used for sends after the DMA mapping was done. This causes problems on non-cache-coherent architectures, because the device doing DMA won't see updates to the payload buffers that exist only in the CPU cache. Fix this by having all MAD consumers use ib_create_send_mad() to allocate their send buffers, and moving the DMA mapping into the MAD layer so it can be done just before calling send (and after any modifications of the send buffer by the MAD layer). Tested on a non-cache-coherent PowerPC 440SPe system. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-10-26 01:51:39 +08:00
rec, mad->data);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
*sa_query = &query->sa_query;
[PATCH] IB: Fix race in sa_query Use a copy of the id we'll return to the consumer so that we don't dereference query->sa_query after calling send_mad(). A completion may occur very quickly and end up freeing the query before we get to do anything after send_mad(). Signed-off-by: Roland Dreier <rolandd@cisco.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-06-28 05:36:46 +08:00
[PATCH] IB/core: use correct gfp_mask in sa_query Avoid bogus out of memory errors: fix sa_query to actually pass gfp_mask supplied by the user to idr_pre_get. Signed-off-by: Michael S. Tsirkin <mst@mellanox.co.il> Acked-by: "Sean Hefty" <mshefty@ichips.intel.com> Acked-by: "Roland Dreier" <rdreier@cisco.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-07-14 15:23:56 +08:00
ret = send_mad(&query->sa_query, timeout_ms, gfp_mask);
[IB] Fix MAD layer DMA mappings to avoid touching data buffer once mapped The MAD layer was violating the DMA API by touching data buffers used for sends after the DMA mapping was done. This causes problems on non-cache-coherent architectures, because the device doing DMA won't see updates to the payload buffers that exist only in the CPU cache. Fix this by having all MAD consumers use ib_create_send_mad() to allocate their send buffers, and moving the DMA mapping into the MAD layer so it can be done just before calling send (and after any modifications of the send buffer by the MAD layer). Tested on a non-cache-coherent PowerPC 440SPe system. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-10-26 01:51:39 +08:00
if (ret < 0)
goto err2;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
[PATCH] IB: Fix race in sa_query Use a copy of the id we'll return to the consumer so that we don't dereference query->sa_query after calling send_mad(). A completion may occur very quickly and end up freeing the query before we get to do anything after send_mad(). Signed-off-by: Roland Dreier <rolandd@cisco.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-06-28 05:36:46 +08:00
return ret;
[IB] Fix MAD layer DMA mappings to avoid touching data buffer once mapped The MAD layer was violating the DMA API by touching data buffers used for sends after the DMA mapping was done. This causes problems on non-cache-coherent architectures, because the device doing DMA won't see updates to the payload buffers that exist only in the CPU cache. Fix this by having all MAD consumers use ib_create_send_mad() to allocate their send buffers, and moving the DMA mapping into the MAD layer so it can be done just before calling send (and after any modifications of the send buffer by the MAD layer). Tested on a non-cache-coherent PowerPC 440SPe system. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-10-26 01:51:39 +08:00
err2:
*sa_query = NULL;
IB/sa: Require SA registration Require users to register with SA module, to prevent the sa_query module text from going away while an SA query callback is still running. Update all in-tree users for the new interface. Signed-off-by: Michael S. Tsirkin <mst@mellanox.co.il> Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2006-08-22 07:40:12 +08:00
ib_sa_client_put(query->sa_query.client);
IB/sa: Make sure SA queries use default P_Key MADs sent to the SA should use the the default P_Key (0x7fff/0xffff). There's no requirement that the default P_Key is stored at index 0 in the local P_Key table, so add code to the sa_query module to look up the index of the default P_Key when creating an address handle for the SA (which is done any time the P_Key table might change), and use this index for all SA queries. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2007-06-19 02:03:58 +08:00
free_mad(&query->sa_query);
[IB] Fix MAD layer DMA mappings to avoid touching data buffer once mapped The MAD layer was violating the DMA API by touching data buffers used for sends after the DMA mapping was done. This causes problems on non-cache-coherent architectures, because the device doing DMA won't see updates to the payload buffers that exist only in the CPU cache. Fix this by having all MAD consumers use ib_create_send_mad() to allocate their send buffers, and moving the DMA mapping into the MAD layer so it can be done just before calling send (and after any modifications of the send buffer by the MAD layer). Tested on a non-cache-coherent PowerPC 440SPe system. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-10-26 01:51:39 +08:00
err1:
kfree(query);
return ret;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
IB/sa: Add GuidInfoRecord query support This query is needed for SRIOV alias GUID support. The query is implemented per the IB Spec definition in section 15.2.5.18 (GuidInfoRecord). Signed-off-by: Erez Shitrit <erezsh@mellanox.co.il> Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il> Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com> Signed-off-by: Roland Dreier <roland@purestorage.com>
2012-06-19 16:21:38 +08:00
/* Support GuidInfoRecord */
static void ib_sa_guidinfo_rec_callback(struct ib_sa_query *sa_query,
RDMA/cma: Refactor the inbound/outbound path records process flow Refactors based on comments [1] of the multiple path records support patchset: - Return failure if not able to set inbound/outbound PRs; - Simplify the flow when receiving the PRs from netlink channel: When a good PR response is received, unpack it and call the path_query callback directly. This saves two memory allocations; - Define RDMA_PRIMARY_PATH_MAX_REC_NUM in a proper place. [1] https://lore.kernel.org/linux-rdma/Yyxp9E9pJtUids2o@nvidia.com/ Signed-off-by: Mark Zhang <markzhang@nvidia.com> Reviewed-by: Bart Van Assche <bvanassche@acm.org> #srp Link: https://lore.kernel.org/r/7610025d57342b8b6da0f19516c9612f9c3fdc37.1672819376.git.leonro@nvidia.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2023-01-04 16:03:41 +08:00
int status, struct ib_sa_mad *mad)
IB/sa: Add GuidInfoRecord query support This query is needed for SRIOV alias GUID support. The query is implemented per the IB Spec definition in section 15.2.5.18 (GuidInfoRecord). Signed-off-by: Erez Shitrit <erezsh@mellanox.co.il> Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il> Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com> Signed-off-by: Roland Dreier <roland@purestorage.com>
2012-06-19 16:21:38 +08:00
{
struct ib_sa_guidinfo_query *query =
container_of(sa_query, struct ib_sa_guidinfo_query, sa_query);
if (mad) {
struct ib_sa_guidinfo_rec rec;
ib_unpack(guidinfo_rec_table, ARRAY_SIZE(guidinfo_rec_table),
mad->data, &rec);
query->callback(status, &rec, query->context);
} else
query->callback(status, NULL, query->context);
}
static void ib_sa_guidinfo_rec_release(struct ib_sa_query *sa_query)
{
kfree(container_of(sa_query, struct ib_sa_guidinfo_query, sa_query));
}
int ib_sa_guid_info_rec_query(struct ib_sa_client *client,
RDMA: Support more than 255 rdma ports Current code uses many different types when dealing with a port of a RDMA device: u8, unsigned int and u32. Switch to u32 to clean up the logic. This allows us to make (at least) the core view consistent and use the same type. Unfortunately not all places can be converted. Many uverbs functions expect port to be u8 so keep those places in order not to break UAPIs. HW/Spec defined values must also not be changed. With the switch to u32 we now can support devices with more than 255 ports. U32_MAX is reserved to make control logic a bit easier to deal with. As a device with U32_MAX ports probably isn't going to happen any time soon this seems like a non issue. When a device with more than 255 ports is created uverbs will report the RDMA device as having 255 ports as this is the max currently supported. The verbs interface is not changed yet because the IBTA spec limits the port size in too many places to be u8 and all applications that relies in verbs won't be able to cope with this change. At this stage, we are extending the interfaces that are using vendor channel solely Once the limitation is lifted mlx5 in switchdev mode will be able to have thousands of SFs created by the device. As the only instance of an RDMA device that reports more than 255 ports will be a representor device and it exposes itself as a RAW Ethernet only device CM/MAD/IPoIB and other ULPs aren't effected by this change and their sysfs/interfaces that are exposes to userspace can remain unchanged. While here cleanup some alignment issues and remove unneeded sanity checks (mainly in rdmavt), Link: https://lore.kernel.org/r/20210301070420.439400-1-leon@kernel.org Signed-off-by: Mark Bloch <mbloch@nvidia.com> Signed-off-by: Leon Romanovsky <leonro@nvidia.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
2021-03-01 15:04:20 +08:00
struct ib_device *device, u32 port_num,
IB/sa: Add GuidInfoRecord query support This query is needed for SRIOV alias GUID support. The query is implemented per the IB Spec definition in section 15.2.5.18 (GuidInfoRecord). Signed-off-by: Erez Shitrit <erezsh@mellanox.co.il> Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il> Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com> Signed-off-by: Roland Dreier <roland@purestorage.com>
2012-06-19 16:21:38 +08:00
struct ib_sa_guidinfo_rec *rec,
ib_sa_comp_mask comp_mask, u8 method,
RDMA/core: Annotate timeout as unsigned long The ucma users supply timeout in u32 format, it means that any number with most significant bit set will be converted to negative value by various rdma_*, cma_* and sa_query functions, which treat timeout as int. In the lowest level, the timeout is converted back to be unsigned long. Remove this ambiguous conversion by updating all function signatures to receive unsigned long. Reported-by: Noa Osherovich <noaos@mellanox.com> Signed-off-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2018-10-11 22:30:05 +08:00
unsigned long timeout_ms, gfp_t gfp_mask,
IB/sa: Add GuidInfoRecord query support This query is needed for SRIOV alias GUID support. The query is implemented per the IB Spec definition in section 15.2.5.18 (GuidInfoRecord). Signed-off-by: Erez Shitrit <erezsh@mellanox.co.il> Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il> Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com> Signed-off-by: Roland Dreier <roland@purestorage.com>
2012-06-19 16:21:38 +08:00
void (*callback)(int status,
struct ib_sa_guidinfo_rec *resp,
void *context),
void *context,
struct ib_sa_query **sa_query)
{
struct ib_sa_guidinfo_query *query;
struct ib_sa_device *sa_dev = ib_get_client_data(device, &sa_client);
struct ib_sa_port *port;
struct ib_mad_agent *agent;
struct ib_sa_mad *mad;
int ret;
if (!sa_dev)
return -ENODEV;
if (method != IB_MGMT_METHOD_GET &&
method != IB_MGMT_METHOD_SET &&
method != IB_SA_METHOD_DELETE) {
return -EINVAL;
}
port = &sa_dev->port[port_num - sa_dev->start_port];
agent = port->agent;
IB/sa: Allocate SA query with kzalloc Replace kmalloc with kzalloc so that all uninitialized fields in SA query will be zero-ed out to avoid unintentional consequence. This prepares the SA query structure to accept new fields in the future. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: John Fleck <john.fleck@intel.com> Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-14 20:52:08 +08:00
query = kzalloc(sizeof(*query), gfp_mask);
IB/sa: Add GuidInfoRecord query support This query is needed for SRIOV alias GUID support. The query is implemented per the IB Spec definition in section 15.2.5.18 (GuidInfoRecord). Signed-off-by: Erez Shitrit <erezsh@mellanox.co.il> Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il> Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com> Signed-off-by: Roland Dreier <roland@purestorage.com>
2012-06-19 16:21:38 +08:00
if (!query)
return -ENOMEM;
query->sa_query.port = port;
ret = alloc_mad(&query->sa_query, gfp_mask);
if (ret)
goto err1;
ib_sa_client_get(client);
query->sa_query.client = client;
query->callback = callback;
query->context = context;
mad = query->sa_query.mad_buf->mad;
IB/SA: Add support to query opa classport info. For OPA devices, SA will query the OPA classport info instead of the IB defined classport info. opa classport info exposes additional information and capabilities that are specific to OPA devices. Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-29 07:20:26 +08:00
init_mad(&query->sa_query, agent);
IB/sa: Add GuidInfoRecord query support This query is needed for SRIOV alias GUID support. The query is implemented per the IB Spec definition in section 15.2.5.18 (GuidInfoRecord). Signed-off-by: Erez Shitrit <erezsh@mellanox.co.il> Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il> Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com> Signed-off-by: Roland Dreier <roland@purestorage.com>
2012-06-19 16:21:38 +08:00
query->sa_query.callback = callback ? ib_sa_guidinfo_rec_callback : NULL;
query->sa_query.release = ib_sa_guidinfo_rec_release;
mad->mad_hdr.method = method;
mad->mad_hdr.attr_id = cpu_to_be16(IB_SA_ATTR_GUID_INFO_REC);
mad->sa_hdr.comp_mask = comp_mask;
ib_pack(guidinfo_rec_table, ARRAY_SIZE(guidinfo_rec_table), rec,
mad->data);
*sa_query = &query->sa_query;
ret = send_mad(&query->sa_query, timeout_ms, gfp_mask);
if (ret < 0)
goto err2;
return ret;
err2:
*sa_query = NULL;
ib_sa_client_put(query->sa_query.client);
free_mad(&query->sa_query);
err1:
kfree(query);
return ret;
}
EXPORT_SYMBOL(ib_sa_guid_info_rec_query);
IB/SA: Modify SA to implicitly cache Class Port info SA will query and cache class port info as part of its initialization. SA will also invalidate and refresh the cache based on specific events. Callers such as IPoIB and CM can query the SA to get the classportinfo information. Apart from making the caller code much simpler, this change puts the onus on the SA to query and maintain classportinfo much like how it maitains the address handle to the SM. Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Don Hiatt <don.hiatt@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-03-21 07:38:08 +08:00
struct ib_classport_info_context {
struct completion done;
struct ib_sa_query *sa_query;
};
static void ib_classportinfo_cb(void *context)
{
struct ib_classport_info_context *cb_ctx = context;
complete(&cb_ctx->done);
}
IB/SA Agent: Add support for SA agent get ClassPortInfo New SA query function to return the ClassPortInfo struct from the SA. If the SM supports FullMemberSendOnly mode for MCG's, it sets a capability bit in the capability_mask2 field of the response. Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-05-26 03:02:05 +08:00
static void ib_sa_classport_info_rec_callback(struct ib_sa_query *sa_query,
RDMA/cma: Refactor the inbound/outbound path records process flow Refactors based on comments [1] of the multiple path records support patchset: - Return failure if not able to set inbound/outbound PRs; - Simplify the flow when receiving the PRs from netlink channel: When a good PR response is received, unpack it and call the path_query callback directly. This saves two memory allocations; - Define RDMA_PRIMARY_PATH_MAX_REC_NUM in a proper place. [1] https://lore.kernel.org/linux-rdma/Yyxp9E9pJtUids2o@nvidia.com/ Signed-off-by: Mark Zhang <markzhang@nvidia.com> Reviewed-by: Bart Van Assche <bvanassche@acm.org> #srp Link: https://lore.kernel.org/r/7610025d57342b8b6da0f19516c9612f9c3fdc37.1672819376.git.leonro@nvidia.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2023-01-04 16:03:41 +08:00
int status, struct ib_sa_mad *mad)
IB/SA Agent: Add support for SA agent get ClassPortInfo New SA query function to return the ClassPortInfo struct from the SA. If the SM supports FullMemberSendOnly mode for MCG's, it sets a capability bit in the capability_mask2 field of the response. Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-05-26 03:02:05 +08:00
{
IB/sa: Add cached attribute containing SM information to SA port Added a new SA port attribute containing SM ClassPortInfo fields, (ClassPortInfo fields: Table 126 IB Spec 1.3.). This is useful for checking SM support for specific features. The attribute is cached to avoid resending queries, caching is done when a successful ClassPortInfo reply is received on the port. Invalidation of the attribute is done on SM change events, SM re-registration events, and SM LID change events. The fields in ClassPortInfo should not change during SM runtime without an event. Signed-off-by: Alex Vesker <valex@mellanox.com> Reviewed by: Hal Rosenstock <hal@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-07-06 21:36:34 +08:00
unsigned long flags;
IB/SA Agent: Add support for SA agent get ClassPortInfo New SA query function to return the ClassPortInfo struct from the SA. If the SM supports FullMemberSendOnly mode for MCG's, it sets a capability bit in the capability_mask2 field of the response. Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-05-26 03:02:05 +08:00
struct ib_sa_classport_info_query *query =
container_of(sa_query, struct ib_sa_classport_info_query, sa_query);
IB/SA: Add support to query opa classport info. For OPA devices, SA will query the OPA classport info instead of the IB defined classport info. opa classport info exposes additional information and capabilities that are specific to OPA devices. Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-29 07:20:26 +08:00
struct ib_sa_classport_cache *info = &sa_query->port->classport_info;
IB/SA Agent: Add support for SA agent get ClassPortInfo New SA query function to return the ClassPortInfo struct from the SA. If the SM supports FullMemberSendOnly mode for MCG's, it sets a capability bit in the capability_mask2 field of the response. Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-05-26 03:02:05 +08:00
if (mad) {
IB/SA: Add support to query opa classport info. For OPA devices, SA will query the OPA classport info instead of the IB defined classport info. opa classport info exposes additional information and capabilities that are specific to OPA devices. Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-29 07:20:26 +08:00
if (sa_query->flags & IB_SA_QUERY_OPA) {
struct opa_class_port_info rec;
IB/SA Agent: Add support for SA agent get ClassPortInfo New SA query function to return the ClassPortInfo struct from the SA. If the SM supports FullMemberSendOnly mode for MCG's, it sets a capability bit in the capability_mask2 field of the response. Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-05-26 03:02:05 +08:00
IB/SA: Add support to query opa classport info. For OPA devices, SA will query the OPA classport info instead of the IB defined classport info. opa classport info exposes additional information and capabilities that are specific to OPA devices. Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-29 07:20:26 +08:00
ib_unpack(opa_classport_info_rec_table,
ARRAY_SIZE(opa_classport_info_rec_table),
mad->data, &rec);
spin_lock_irqsave(&sa_query->port->classport_lock,
flags);
if (!status && !info->valid) {
memcpy(&info->data.opa, &rec,
sizeof(info->data.opa));
info->valid = true;
info->data.type = RDMA_CLASS_PORT_INFO_OPA;
}
spin_unlock_irqrestore(&sa_query->port->classport_lock,
flags);
} else {
struct ib_class_port_info rec;
IB/sa: Add cached attribute containing SM information to SA port Added a new SA port attribute containing SM ClassPortInfo fields, (ClassPortInfo fields: Table 126 IB Spec 1.3.). This is useful for checking SM support for specific features. The attribute is cached to avoid resending queries, caching is done when a successful ClassPortInfo reply is received on the port. Invalidation of the attribute is done on SM change events, SM re-registration events, and SM LID change events. The fields in ClassPortInfo should not change during SM runtime without an event. Signed-off-by: Alex Vesker <valex@mellanox.com> Reviewed by: Hal Rosenstock <hal@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-07-06 21:36:34 +08:00
IB/SA: Add support to query opa classport info. For OPA devices, SA will query the OPA classport info instead of the IB defined classport info. opa classport info exposes additional information and capabilities that are specific to OPA devices. Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-29 07:20:26 +08:00
ib_unpack(ib_classport_info_rec_table,
ARRAY_SIZE(ib_classport_info_rec_table),
mad->data, &rec);
IB/sa: Add cached attribute containing SM information to SA port Added a new SA port attribute containing SM ClassPortInfo fields, (ClassPortInfo fields: Table 126 IB Spec 1.3.). This is useful for checking SM support for specific features. The attribute is cached to avoid resending queries, caching is done when a successful ClassPortInfo reply is received on the port. Invalidation of the attribute is done on SM change events, SM re-registration events, and SM LID change events. The fields in ClassPortInfo should not change during SM runtime without an event. Signed-off-by: Alex Vesker <valex@mellanox.com> Reviewed by: Hal Rosenstock <hal@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-07-06 21:36:34 +08:00
IB/SA: Add support to query opa classport info. For OPA devices, SA will query the OPA classport info instead of the IB defined classport info. opa classport info exposes additional information and capabilities that are specific to OPA devices. Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-29 07:20:26 +08:00
spin_lock_irqsave(&sa_query->port->classport_lock,
flags);
if (!status && !info->valid) {
memcpy(&info->data.ib, &rec,
sizeof(info->data.ib));
info->valid = true;
info->data.type = RDMA_CLASS_PORT_INFO_IB;
}
spin_unlock_irqrestore(&sa_query->port->classport_lock,
flags);
IB/sa: Add cached attribute containing SM information to SA port Added a new SA port attribute containing SM ClassPortInfo fields, (ClassPortInfo fields: Table 126 IB Spec 1.3.). This is useful for checking SM support for specific features. The attribute is cached to avoid resending queries, caching is done when a successful ClassPortInfo reply is received on the port. Invalidation of the attribute is done on SM change events, SM re-registration events, and SM LID change events. The fields in ClassPortInfo should not change during SM runtime without an event. Signed-off-by: Alex Vesker <valex@mellanox.com> Reviewed by: Hal Rosenstock <hal@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-07-06 21:36:34 +08:00
}
IB/SA Agent: Add support for SA agent get ClassPortInfo New SA query function to return the ClassPortInfo struct from the SA. If the SM supports FullMemberSendOnly mode for MCG's, it sets a capability bit in the capability_mask2 field of the response. Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-05-26 03:02:05 +08:00
}
IB/SA: Modify SA to implicitly cache Class Port info SA will query and cache class port info as part of its initialization. SA will also invalidate and refresh the cache based on specific events. Callers such as IPoIB and CM can query the SA to get the classportinfo information. Apart from making the caller code much simpler, this change puts the onus on the SA to query and maintain classportinfo much like how it maitains the address handle to the SM. Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Don Hiatt <don.hiatt@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-03-21 07:38:08 +08:00
query->callback(query->context);
IB/SA Agent: Add support for SA agent get ClassPortInfo New SA query function to return the ClassPortInfo struct from the SA. If the SM supports FullMemberSendOnly mode for MCG's, it sets a capability bit in the capability_mask2 field of the response. Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-05-26 03:02:05 +08:00
}
IB/SA: Modify SA to implicitly cache Class Port info SA will query and cache class port info as part of its initialization. SA will also invalidate and refresh the cache based on specific events. Callers such as IPoIB and CM can query the SA to get the classportinfo information. Apart from making the caller code much simpler, this change puts the onus on the SA to query and maintain classportinfo much like how it maitains the address handle to the SM. Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Don Hiatt <don.hiatt@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-03-21 07:38:08 +08:00
static void ib_sa_classport_info_rec_release(struct ib_sa_query *sa_query)
IB/SA Agent: Add support for SA agent get ClassPortInfo New SA query function to return the ClassPortInfo struct from the SA. If the SM supports FullMemberSendOnly mode for MCG's, it sets a capability bit in the capability_mask2 field of the response. Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-05-26 03:02:05 +08:00
{
kfree(container_of(sa_query, struct ib_sa_classport_info_query,
sa_query));
}
IB/SA: Modify SA to implicitly cache Class Port info SA will query and cache class port info as part of its initialization. SA will also invalidate and refresh the cache based on specific events. Callers such as IPoIB and CM can query the SA to get the classportinfo information. Apart from making the caller code much simpler, this change puts the onus on the SA to query and maintain classportinfo much like how it maitains the address handle to the SM. Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Don Hiatt <don.hiatt@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-03-21 07:38:08 +08:00
static int ib_sa_classport_info_rec_query(struct ib_sa_port *port,
RDMA/core: Annotate timeout as unsigned long The ucma users supply timeout in u32 format, it means that any number with most significant bit set will be converted to negative value by various rdma_*, cma_* and sa_query functions, which treat timeout as int. In the lowest level, the timeout is converted back to be unsigned long. Remove this ambiguous conversion by updating all function signatures to receive unsigned long. Reported-by: Noa Osherovich <noaos@mellanox.com> Signed-off-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2018-10-11 22:30:05 +08:00
unsigned long timeout_ms,
IB/SA: Modify SA to implicitly cache Class Port info SA will query and cache class port info as part of its initialization. SA will also invalidate and refresh the cache based on specific events. Callers such as IPoIB and CM can query the SA to get the classportinfo information. Apart from making the caller code much simpler, this change puts the onus on the SA to query and maintain classportinfo much like how it maitains the address handle to the SM. Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Don Hiatt <don.hiatt@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-03-21 07:38:08 +08:00
void (*callback)(void *context),
void *context,
struct ib_sa_query **sa_query)
IB/SA Agent: Add support for SA agent get ClassPortInfo New SA query function to return the ClassPortInfo struct from the SA. If the SM supports FullMemberSendOnly mode for MCG's, it sets a capability bit in the capability_mask2 field of the response. Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-05-26 03:02:05 +08:00
{
struct ib_mad_agent *agent;
IB/SA: Modify SA to implicitly cache Class Port info SA will query and cache class port info as part of its initialization. SA will also invalidate and refresh the cache based on specific events. Callers such as IPoIB and CM can query the SA to get the classportinfo information. Apart from making the caller code much simpler, this change puts the onus on the SA to query and maintain classportinfo much like how it maitains the address handle to the SM. Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Don Hiatt <don.hiatt@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-03-21 07:38:08 +08:00
struct ib_sa_classport_info_query *query;
IB/SA Agent: Add support for SA agent get ClassPortInfo New SA query function to return the ClassPortInfo struct from the SA. If the SM supports FullMemberSendOnly mode for MCG's, it sets a capability bit in the capability_mask2 field of the response. Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-05-26 03:02:05 +08:00
struct ib_sa_mad *mad;
IB/SA: Modify SA to implicitly cache Class Port info SA will query and cache class port info as part of its initialization. SA will also invalidate and refresh the cache based on specific events. Callers such as IPoIB and CM can query the SA to get the classportinfo information. Apart from making the caller code much simpler, this change puts the onus on the SA to query and maintain classportinfo much like how it maitains the address handle to the SM. Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Don Hiatt <don.hiatt@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-03-21 07:38:08 +08:00
gfp_t gfp_mask = GFP_KERNEL;
IB/SA Agent: Add support for SA agent get ClassPortInfo New SA query function to return the ClassPortInfo struct from the SA. If the SM supports FullMemberSendOnly mode for MCG's, it sets a capability bit in the capability_mask2 field of the response. Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-05-26 03:02:05 +08:00
int ret;
agent = port->agent;
query = kzalloc(sizeof(*query), gfp_mask);
if (!query)
return -ENOMEM;
query->sa_query.port = port;
IB/SA: Add support to query opa classport info. For OPA devices, SA will query the OPA classport info instead of the IB defined classport info. opa classport info exposes additional information and capabilities that are specific to OPA devices. Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-29 07:20:26 +08:00
query->sa_query.flags |= rdma_cap_opa_ah(port->agent->device,
port->port_num) ?
IB_SA_QUERY_OPA : 0;
IB/SA Agent: Add support for SA agent get ClassPortInfo New SA query function to return the ClassPortInfo struct from the SA. If the SM supports FullMemberSendOnly mode for MCG's, it sets a capability bit in the capability_mask2 field of the response. Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-05-26 03:02:05 +08:00
ret = alloc_mad(&query->sa_query, gfp_mask);
if (ret)
IB/SA: Modify SA to implicitly cache Class Port info SA will query and cache class port info as part of its initialization. SA will also invalidate and refresh the cache based on specific events. Callers such as IPoIB and CM can query the SA to get the classportinfo information. Apart from making the caller code much simpler, this change puts the onus on the SA to query and maintain classportinfo much like how it maitains the address handle to the SM. Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Don Hiatt <don.hiatt@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-03-21 07:38:08 +08:00
goto err_free;
IB/SA Agent: Add support for SA agent get ClassPortInfo New SA query function to return the ClassPortInfo struct from the SA. If the SM supports FullMemberSendOnly mode for MCG's, it sets a capability bit in the capability_mask2 field of the response. Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-05-26 03:02:05 +08:00
IB/SA: Modify SA to implicitly cache Class Port info SA will query and cache class port info as part of its initialization. SA will also invalidate and refresh the cache based on specific events. Callers such as IPoIB and CM can query the SA to get the classportinfo information. Apart from making the caller code much simpler, this change puts the onus on the SA to query and maintain classportinfo much like how it maitains the address handle to the SM. Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Don Hiatt <don.hiatt@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-03-21 07:38:08 +08:00
query->callback = callback;
query->context = context;
IB/SA Agent: Add support for SA agent get ClassPortInfo New SA query function to return the ClassPortInfo struct from the SA. If the SM supports FullMemberSendOnly mode for MCG's, it sets a capability bit in the capability_mask2 field of the response. Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-05-26 03:02:05 +08:00
mad = query->sa_query.mad_buf->mad;
IB/SA: Add support to query opa classport info. For OPA devices, SA will query the OPA classport info instead of the IB defined classport info. opa classport info exposes additional information and capabilities that are specific to OPA devices. Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-29 07:20:26 +08:00
init_mad(&query->sa_query, agent);
IB/SA Agent: Add support for SA agent get ClassPortInfo New SA query function to return the ClassPortInfo struct from the SA. If the SM supports FullMemberSendOnly mode for MCG's, it sets a capability bit in the capability_mask2 field of the response. Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-05-26 03:02:05 +08:00
IB/SA: Modify SA to implicitly cache Class Port info SA will query and cache class port info as part of its initialization. SA will also invalidate and refresh the cache based on specific events. Callers such as IPoIB and CM can query the SA to get the classportinfo information. Apart from making the caller code much simpler, this change puts the onus on the SA to query and maintain classportinfo much like how it maitains the address handle to the SM. Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Don Hiatt <don.hiatt@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-03-21 07:38:08 +08:00
query->sa_query.callback = ib_sa_classport_info_rec_callback;
query->sa_query.release = ib_sa_classport_info_rec_release;
IB/SA Agent: Add support for SA agent get ClassPortInfo New SA query function to return the ClassPortInfo struct from the SA. If the SM supports FullMemberSendOnly mode for MCG's, it sets a capability bit in the capability_mask2 field of the response. Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-05-26 03:02:05 +08:00
mad->mad_hdr.method = IB_MGMT_METHOD_GET;
mad->mad_hdr.attr_id = cpu_to_be16(IB_SA_ATTR_CLASS_PORTINFO);
mad->sa_hdr.comp_mask = 0;
*sa_query = &query->sa_query;
ret = send_mad(&query->sa_query, timeout_ms, gfp_mask);
if (ret < 0)
IB/SA: Modify SA to implicitly cache Class Port info SA will query and cache class port info as part of its initialization. SA will also invalidate and refresh the cache based on specific events. Callers such as IPoIB and CM can query the SA to get the classportinfo information. Apart from making the caller code much simpler, this change puts the onus on the SA to query and maintain classportinfo much like how it maitains the address handle to the SM. Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Don Hiatt <don.hiatt@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-03-21 07:38:08 +08:00
goto err_free_mad;
IB/SA Agent: Add support for SA agent get ClassPortInfo New SA query function to return the ClassPortInfo struct from the SA. If the SM supports FullMemberSendOnly mode for MCG's, it sets a capability bit in the capability_mask2 field of the response. Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-05-26 03:02:05 +08:00
return ret;
IB/SA: Modify SA to implicitly cache Class Port info SA will query and cache class port info as part of its initialization. SA will also invalidate and refresh the cache based on specific events. Callers such as IPoIB and CM can query the SA to get the classportinfo information. Apart from making the caller code much simpler, this change puts the onus on the SA to query and maintain classportinfo much like how it maitains the address handle to the SM. Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Don Hiatt <don.hiatt@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-03-21 07:38:08 +08:00
err_free_mad:
IB/SA Agent: Add support for SA agent get ClassPortInfo New SA query function to return the ClassPortInfo struct from the SA. If the SM supports FullMemberSendOnly mode for MCG's, it sets a capability bit in the capability_mask2 field of the response. Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-05-26 03:02:05 +08:00
*sa_query = NULL;
free_mad(&query->sa_query);
IB/SA: Modify SA to implicitly cache Class Port info SA will query and cache class port info as part of its initialization. SA will also invalidate and refresh the cache based on specific events. Callers such as IPoIB and CM can query the SA to get the classportinfo information. Apart from making the caller code much simpler, this change puts the onus on the SA to query and maintain classportinfo much like how it maitains the address handle to the SM. Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Don Hiatt <don.hiatt@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-03-21 07:38:08 +08:00
err_free:
IB/SA Agent: Add support for SA agent get ClassPortInfo New SA query function to return the ClassPortInfo struct from the SA. If the SM supports FullMemberSendOnly mode for MCG's, it sets a capability bit in the capability_mask2 field of the response. Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-05-26 03:02:05 +08:00
kfree(query);
return ret;
}
IB/SA: Modify SA to implicitly cache Class Port info SA will query and cache class port info as part of its initialization. SA will also invalidate and refresh the cache based on specific events. Callers such as IPoIB and CM can query the SA to get the classportinfo information. Apart from making the caller code much simpler, this change puts the onus on the SA to query and maintain classportinfo much like how it maitains the address handle to the SM. Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Don Hiatt <don.hiatt@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-03-21 07:38:08 +08:00
static void update_ib_cpi(struct work_struct *work)
{
struct ib_sa_port *port =
container_of(work, struct ib_sa_port, ib_cpi_work.work);
struct ib_classport_info_context *cb_context;
unsigned long flags;
int ret;
/* If the classport info is valid, nothing
* to do here.
*/
spin_lock_irqsave(&port->classport_lock, flags);
if (port->classport_info.valid) {
spin_unlock_irqrestore(&port->classport_lock, flags);
return;
}
spin_unlock_irqrestore(&port->classport_lock, flags);
cb_context = kmalloc(sizeof(*cb_context), GFP_KERNEL);
if (!cb_context)
goto err_nomem;
init_completion(&cb_context->done);
ret = ib_sa_classport_info_rec_query(port, 3000,
ib_classportinfo_cb, cb_context,
&cb_context->sa_query);
if (ret < 0)
goto free_cb_err;
wait_for_completion(&cb_context->done);
free_cb_err:
kfree(cb_context);
spin_lock_irqsave(&port->classport_lock, flags);
/* If the classport info is still not valid, the query should have
* failed for some reason. Retry issuing the query
*/
if (!port->classport_info.valid) {
port->classport_info.retry_cnt++;
if (port->classport_info.retry_cnt <=
IB_SA_CPI_MAX_RETRY_CNT) {
unsigned long delay =
msecs_to_jiffies(IB_SA_CPI_RETRY_WAIT);
queue_delayed_work(ib_wq, &port->ib_cpi_work, delay);
}
}
spin_unlock_irqrestore(&port->classport_lock, flags);
err_nomem:
return;
}
IB/SA Agent: Add support for SA agent get ClassPortInfo New SA query function to return the ClassPortInfo struct from the SA. If the SM supports FullMemberSendOnly mode for MCG's, it sets a capability bit in the capability_mask2 field of the response. Signed-off-by: Erez Shitrit <erezsh@mellanox.com> Reviewed-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-05-26 03:02:05 +08:00
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
static void send_handler(struct ib_mad_agent *agent,
struct ib_mad_send_wc *mad_send_wc)
{
[IB] Fix MAD layer DMA mappings to avoid touching data buffer once mapped The MAD layer was violating the DMA API by touching data buffers used for sends after the DMA mapping was done. This causes problems on non-cache-coherent architectures, because the device doing DMA won't see updates to the payload buffers that exist only in the CPU cache. Fix this by having all MAD consumers use ib_create_send_mad() to allocate their send buffers, and moving the DMA mapping into the MAD layer so it can be done just before calling send (and after any modifications of the send buffer by the MAD layer). Tested on a non-cache-coherent PowerPC 440SPe system. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-10-26 01:51:39 +08:00
struct ib_sa_query *query = mad_send_wc->send_buf->context[0];
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
unsigned long flags;
[PATCH] IB: allow NULL sa_query callbacks Check if a client passes a NULL callback into an SA query, and if so, never call back. This fixes an oops if someone unloads ib_ipoib and ib_sa in rapid succession. ib_ipoib does an MCMember delete with a NULL callback and 0 timeout on unload, which is usually fine since the delete completes successfully. However, if ib_sa is unloaded immediately afterwards, the delete will be canceled and ib_sa will try to call the (now already unloaded) ib_ipoib module back with the cancel completion, which triggers the oops. Signed-off-by: Roland Dreier <roland@topspin.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-05-26 03:31:29 +08:00
if (query->callback)
switch (mad_send_wc->status) {
case IB_WC_SUCCESS:
/* No callback -- already got recv */
break;
case IB_WC_RESP_TIMEOUT_ERR:
RDMA/cma: Refactor the inbound/outbound path records process flow Refactors based on comments [1] of the multiple path records support patchset: - Return failure if not able to set inbound/outbound PRs; - Simplify the flow when receiving the PRs from netlink channel: When a good PR response is received, unpack it and call the path_query callback directly. This saves two memory allocations; - Define RDMA_PRIMARY_PATH_MAX_REC_NUM in a proper place. [1] https://lore.kernel.org/linux-rdma/Yyxp9E9pJtUids2o@nvidia.com/ Signed-off-by: Mark Zhang <markzhang@nvidia.com> Reviewed-by: Bart Van Assche <bvanassche@acm.org> #srp Link: https://lore.kernel.org/r/7610025d57342b8b6da0f19516c9612f9c3fdc37.1672819376.git.leonro@nvidia.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2023-01-04 16:03:41 +08:00
query->callback(query, -ETIMEDOUT, NULL);
[PATCH] IB: allow NULL sa_query callbacks Check if a client passes a NULL callback into an SA query, and if so, never call back. This fixes an oops if someone unloads ib_ipoib and ib_sa in rapid succession. ib_ipoib does an MCMember delete with a NULL callback and 0 timeout on unload, which is usually fine since the delete completes successfully. However, if ib_sa is unloaded immediately afterwards, the delete will be canceled and ib_sa will try to call the (now already unloaded) ib_ipoib module back with the cancel completion, which triggers the oops. Signed-off-by: Roland Dreier <roland@topspin.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-05-26 03:31:29 +08:00
break;
case IB_WC_WR_FLUSH_ERR:
RDMA/cma: Refactor the inbound/outbound path records process flow Refactors based on comments [1] of the multiple path records support patchset: - Return failure if not able to set inbound/outbound PRs; - Simplify the flow when receiving the PRs from netlink channel: When a good PR response is received, unpack it and call the path_query callback directly. This saves two memory allocations; - Define RDMA_PRIMARY_PATH_MAX_REC_NUM in a proper place. [1] https://lore.kernel.org/linux-rdma/Yyxp9E9pJtUids2o@nvidia.com/ Signed-off-by: Mark Zhang <markzhang@nvidia.com> Reviewed-by: Bart Van Assche <bvanassche@acm.org> #srp Link: https://lore.kernel.org/r/7610025d57342b8b6da0f19516c9612f9c3fdc37.1672819376.git.leonro@nvidia.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2023-01-04 16:03:41 +08:00
query->callback(query, -EINTR, NULL);
[PATCH] IB: allow NULL sa_query callbacks Check if a client passes a NULL callback into an SA query, and if so, never call back. This fixes an oops if someone unloads ib_ipoib and ib_sa in rapid succession. ib_ipoib does an MCMember delete with a NULL callback and 0 timeout on unload, which is usually fine since the delete completes successfully. However, if ib_sa is unloaded immediately afterwards, the delete will be canceled and ib_sa will try to call the (now already unloaded) ib_ipoib module back with the cancel completion, which triggers the oops. Signed-off-by: Roland Dreier <roland@topspin.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-05-26 03:31:29 +08:00
break;
default:
RDMA/cma: Refactor the inbound/outbound path records process flow Refactors based on comments [1] of the multiple path records support patchset: - Return failure if not able to set inbound/outbound PRs; - Simplify the flow when receiving the PRs from netlink channel: When a good PR response is received, unpack it and call the path_query callback directly. This saves two memory allocations; - Define RDMA_PRIMARY_PATH_MAX_REC_NUM in a proper place. [1] https://lore.kernel.org/linux-rdma/Yyxp9E9pJtUids2o@nvidia.com/ Signed-off-by: Mark Zhang <markzhang@nvidia.com> Reviewed-by: Bart Van Assche <bvanassche@acm.org> #srp Link: https://lore.kernel.org/r/7610025d57342b8b6da0f19516c9612f9c3fdc37.1672819376.git.leonro@nvidia.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2023-01-04 16:03:41 +08:00
query->callback(query, -EIO, NULL);
[PATCH] IB: allow NULL sa_query callbacks Check if a client passes a NULL callback into an SA query, and if so, never call back. This fixes an oops if someone unloads ib_ipoib and ib_sa in rapid succession. ib_ipoib does an MCMember delete with a NULL callback and 0 timeout on unload, which is usually fine since the delete completes successfully. However, if ib_sa is unloaded immediately afterwards, the delete will be canceled and ib_sa will try to call the (now already unloaded) ib_ipoib module back with the cancel completion, which triggers the oops. Signed-off-by: Roland Dreier <roland@topspin.com> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2005-05-26 03:31:29 +08:00
break;
}
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
ib core: Convert query_idr to XArray Signed-off-by: Matthew Wilcox <willy@infradead.org> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2019-02-21 08:20:46 +08:00
xa_lock_irqsave(&queries, flags);
__xa_erase(&queries, query->id);
xa_unlock_irqrestore(&queries, flags);
[IB] Fix MAD layer DMA mappings to avoid touching data buffer once mapped The MAD layer was violating the DMA API by touching data buffers used for sends after the DMA mapping was done. This causes problems on non-cache-coherent architectures, because the device doing DMA won't see updates to the payload buffers that exist only in the CPU cache. Fix this by having all MAD consumers use ib_create_send_mad() to allocate their send buffers, and moving the DMA mapping into the MAD layer so it can be done just before calling send (and after any modifications of the send buffer by the MAD layer). Tested on a non-cache-coherent PowerPC 440SPe system. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-10-26 01:51:39 +08:00
IB/sa: Make sure SA queries use default P_Key MADs sent to the SA should use the the default P_Key (0x7fff/0xffff). There's no requirement that the default P_Key is stored at index 0 in the local P_Key table, so add code to the sa_query module to look up the index of the default P_Key when creating an address handle for the SA (which is done any time the P_Key table might change), and use this index for all SA queries. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2007-06-19 02:03:58 +08:00
free_mad(query);
IB/SA: Modify SA to implicitly cache Class Port info SA will query and cache class port info as part of its initialization. SA will also invalidate and refresh the cache based on specific events. Callers such as IPoIB and CM can query the SA to get the classportinfo information. Apart from making the caller code much simpler, this change puts the onus on the SA to query and maintain classportinfo much like how it maitains the address handle to the SM. Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Don Hiatt <don.hiatt@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-03-21 07:38:08 +08:00
if (query->client)
ib_sa_client_put(query->client);
[IB] Fix MAD layer DMA mappings to avoid touching data buffer once mapped The MAD layer was violating the DMA API by touching data buffers used for sends after the DMA mapping was done. This causes problems on non-cache-coherent architectures, because the device doing DMA won't see updates to the payload buffers that exist only in the CPU cache. Fix this by having all MAD consumers use ib_create_send_mad() to allocate their send buffers, and moving the DMA mapping into the MAD layer so it can be done just before calling send (and after any modifications of the send buffer by the MAD layer). Tested on a non-cache-coherent PowerPC 440SPe system. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-10-26 01:51:39 +08:00
query->release(query);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
static void recv_handler(struct ib_mad_agent *mad_agent,
IB/mad: pass ib_mad_send_buf explicitly to the recv_handler Stop abusing wr_id and just pass the parameter explicitly. Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Hal Rosenstock <hal@mellanox.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Sagi Grimberg <sagig@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-01-04 21:15:58 +08:00
struct ib_mad_send_buf *send_buf,
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
struct ib_mad_recv_wc *mad_recv_wc)
{
struct ib_sa_query *query;
IB/mad: pass ib_mad_send_buf explicitly to the recv_handler Stop abusing wr_id and just pass the parameter explicitly. Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Hal Rosenstock <hal@mellanox.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Sagi Grimberg <sagig@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-01-04 21:15:58 +08:00
if (!send_buf)
return;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
IB/mad: pass ib_mad_send_buf explicitly to the recv_handler Stop abusing wr_id and just pass the parameter explicitly. Signed-off-by: Christoph Hellwig <hch@lst.de> Reviewed-by: Hal Rosenstock <hal@mellanox.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Sagi Grimberg <sagig@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-01-04 21:15:58 +08:00
query = send_buf->context[0];
[IB] Fix MAD layer DMA mappings to avoid touching data buffer once mapped The MAD layer was violating the DMA API by touching data buffers used for sends after the DMA mapping was done. This causes problems on non-cache-coherent architectures, because the device doing DMA won't see updates to the payload buffers that exist only in the CPU cache. Fix this by having all MAD consumers use ib_create_send_mad() to allocate their send buffers, and moving the DMA mapping into the MAD layer so it can be done just before calling send (and after any modifications of the send buffer by the MAD layer). Tested on a non-cache-coherent PowerPC 440SPe system. Signed-off-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2005-10-26 01:51:39 +08:00
if (query->callback) {
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
if (mad_recv_wc->wc->status == IB_WC_SUCCESS)
query->callback(query,
mad_recv_wc->recv_buf.mad->mad_hdr.status ?
RDMA/cma: Refactor the inbound/outbound path records process flow Refactors based on comments [1] of the multiple path records support patchset: - Return failure if not able to set inbound/outbound PRs; - Simplify the flow when receiving the PRs from netlink channel: When a good PR response is received, unpack it and call the path_query callback directly. This saves two memory allocations; - Define RDMA_PRIMARY_PATH_MAX_REC_NUM in a proper place. [1] https://lore.kernel.org/linux-rdma/Yyxp9E9pJtUids2o@nvidia.com/ Signed-off-by: Mark Zhang <markzhang@nvidia.com> Reviewed-by: Bart Van Assche <bvanassche@acm.org> #srp Link: https://lore.kernel.org/r/7610025d57342b8b6da0f19516c9612f9c3fdc37.1672819376.git.leonro@nvidia.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2023-01-04 16:03:41 +08:00
-EINVAL : 0,
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
(struct ib_sa_mad *) mad_recv_wc->recv_buf.mad);
else
RDMA/cma: Refactor the inbound/outbound path records process flow Refactors based on comments [1] of the multiple path records support patchset: - Return failure if not able to set inbound/outbound PRs; - Simplify the flow when receiving the PRs from netlink channel: When a good PR response is received, unpack it and call the path_query callback directly. This saves two memory allocations; - Define RDMA_PRIMARY_PATH_MAX_REC_NUM in a proper place. [1] https://lore.kernel.org/linux-rdma/Yyxp9E9pJtUids2o@nvidia.com/ Signed-off-by: Mark Zhang <markzhang@nvidia.com> Reviewed-by: Bart Van Assche <bvanassche@acm.org> #srp Link: https://lore.kernel.org/r/7610025d57342b8b6da0f19516c9612f9c3fdc37.1672819376.git.leonro@nvidia.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2023-01-04 16:03:41 +08:00
query->callback(query, -EIO, NULL);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
ib_free_recv_mad(mad_recv_wc);
}
IB/SA: Move functions update_sm_ah() and ib_sa_event() Moving these will facilitate changes to these in the next patchs. This is strictly a move and there are no changes to the functions in any way. Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-03-21 07:38:07 +08:00
static void update_sm_ah(struct work_struct *work)
{
struct ib_sa_port *port =
container_of(work, struct ib_sa_port, update_task);
struct ib_sa_sm_ah *new_ah;
struct ib_port_attr port_attr;
IB/core: Rename struct ib_ah_attr to rdma_ah_attr This patch simply renames struct ib_ah_attr to rdma_ah_attr as these fields specify attributes that are not necessarily specific to IB. Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com> Reviewed-by: Sean Hefty <sean.hefty@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-30 02:41:18 +08:00
struct rdma_ah_attr ah_attr;
RDMA: Validate grh_required when handling AVs Extend the existing grh_required flag to check when AV's are handled that a GRH is present. Since we don't want to do query_port during the AV checks for performance reasons move the flag into the immutable_data. Signed-off-by: Artemy Kovalyov <artemyko@mellanox.com> Signed-off-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2018-07-04 20:57:50 +08:00
bool grh_required;
IB/SA: Move functions update_sm_ah() and ib_sa_event() Moving these will facilitate changes to these in the next patchs. This is strictly a move and there are no changes to the functions in any way. Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-03-21 07:38:07 +08:00
if (ib_query_port(port->agent->device, port->port_num, &port_attr)) {
pr_warn("Couldn't query port\n");
return;
}
new_ah = kmalloc(sizeof(*new_ah), GFP_KERNEL);
if (!new_ah)
return;
kref_init(&new_ah->ref);
new_ah->src_path_mask = (1 << port_attr.lmc) - 1;
new_ah->pkey_index = 0;
if (ib_find_pkey(port->agent->device, port->port_num,
IB_DEFAULT_PKEY_FULL, &new_ah->pkey_index))
pr_err("Couldn't find index for default PKey\n");
memset(&ah_attr, 0, sizeof(ah_attr));
IB/core: Define 'ib' and 'roce' rdma_ah_attr types rdma_ah_attr can now be either ib or roce allowing core components to use one type or the other and also to define attributes unique to a specific type. struct ib_ah is also initialized with the type when its first created. This ensures that calls such as modify_ah dont modify the type of the address handle attribute. Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Sean Hefty <sean.hefty@intel.com> Reviewed-by: Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-30 02:41:29 +08:00
ah_attr.type = rdma_ah_find_type(port->agent->device,
port->port_num);
IB/core: Use rdma_ah_attr accessor functions Modify core and driver components to use accessor functions introduced to access individual fields of rdma_ah_attr Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Sean Hefty <sean.hefty@intel.com> Reviewed-by: Niranjana Vishwanathapura <niranjana.vishwanathapura@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-04-30 02:41:28 +08:00
rdma_ah_set_dlid(&ah_attr, port_attr.sm_lid);
rdma_ah_set_sl(&ah_attr, port_attr.sm_sl);
rdma_ah_set_port_num(&ah_attr, port->port_num);
RDMA/hfi1: Move grh_required into update_sm_ah grh_required is intended to be a global setting where all AV's will require a GRH, not just the sm_lid. Move the special logic to the creation of the SM AH. Signed-off-by: Jason Gunthorpe <jgg@mellanox.com> Signed-off-by: Artemy Kovalyov <artemyko@mellanox.com> Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
2018-07-04 20:57:49 +08:00
RDMA: Validate grh_required when handling AVs Extend the existing grh_required flag to check when AV's are handled that a GRH is present. Since we don't want to do query_port during the AV checks for performance reasons move the flag into the immutable_data. Signed-off-by: Artemy Kovalyov <artemyko@mellanox.com> Signed-off-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2018-07-04 20:57:50 +08:00
grh_required = rdma_is_grh_required(port->agent->device,
port->port_num);
RDMA/hfi1: Move grh_required into update_sm_ah grh_required is intended to be a global setting where all AV's will require a GRH, not just the sm_lid. Move the special logic to the creation of the SM AH. Signed-off-by: Jason Gunthorpe <jgg@mellanox.com> Signed-off-by: Artemy Kovalyov <artemyko@mellanox.com> Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
2018-07-04 20:57:49 +08:00
/*
* The OPA sm_lid of 0xFFFF needs special handling so that it can be
* differentiated from a permissive LID of 0xFFFF. We set the
* grh_required flag here so the SA can program the DGID in the
* address handle appropriately
*/
if (ah_attr.type == RDMA_AH_ATTR_TYPE_OPA &&
RDMA: Validate grh_required when handling AVs Extend the existing grh_required flag to check when AV's are handled that a GRH is present. Since we don't want to do query_port during the AV checks for performance reasons move the flag into the immutable_data. Signed-off-by: Artemy Kovalyov <artemyko@mellanox.com> Signed-off-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2018-07-04 20:57:50 +08:00
(grh_required ||
RDMA/hfi1: Move grh_required into update_sm_ah grh_required is intended to be a global setting where all AV's will require a GRH, not just the sm_lid. Move the special logic to the creation of the SM AH. Signed-off-by: Jason Gunthorpe <jgg@mellanox.com> Signed-off-by: Artemy Kovalyov <artemyko@mellanox.com> Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
2018-07-04 20:57:49 +08:00
port_attr.sm_lid == be16_to_cpu(IB_LID_PERMISSIVE)))
rdma_ah_set_make_grd(&ah_attr, true);
RDMA: Validate grh_required when handling AVs Extend the existing grh_required flag to check when AV's are handled that a GRH is present. Since we don't want to do query_port during the AV checks for performance reasons move the flag into the immutable_data. Signed-off-by: Artemy Kovalyov <artemyko@mellanox.com> Signed-off-by: Leon Romanovsky <leonro@mellanox.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2018-07-04 20:57:50 +08:00
if (ah_attr.type == RDMA_AH_ATTR_TYPE_IB && grh_required) {
RDMA/hfi1: Move grh_required into update_sm_ah grh_required is intended to be a global setting where all AV's will require a GRH, not just the sm_lid. Move the special logic to the creation of the SM AH. Signed-off-by: Jason Gunthorpe <jgg@mellanox.com> Signed-off-by: Artemy Kovalyov <artemyko@mellanox.com> Signed-off-by: Leon Romanovsky <leonro@mellanox.com>
2018-07-04 20:57:49 +08:00
rdma_ah_set_ah_flags(&ah_attr, IB_AH_GRH);
rdma_ah_set_subnet_prefix(&ah_attr,
cpu_to_be64(port_attr.subnet_prefix));
rdma_ah_set_interface_id(&ah_attr,
cpu_to_be64(IB_SA_WELL_KNOWN_GUID));
IB/SA: Move functions update_sm_ah() and ib_sa_event() Moving these will facilitate changes to these in the next patchs. This is strictly a move and there are no changes to the functions in any way. Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-03-21 07:38:07 +08:00
}
RDMA: Mark if create address handle is in a sleepable context Introduce a 'flags' field to create address handle callback and add a flag that marks whether the callback is executed in an atomic context or not. This will allow drivers to wait for completion instead of polling for it when it is allowed. Signed-off-by: Gal Pressman <galpress@amazon.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2018-12-12 17:09:05 +08:00
new_ah->ah = rdma_create_ah(port->agent->qp->pd, &ah_attr,
RDMA_CREATE_AH_SLEEPABLE);
IB/SA: Move functions update_sm_ah() and ib_sa_event() Moving these will facilitate changes to these in the next patchs. This is strictly a move and there are no changes to the functions in any way. Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-03-21 07:38:07 +08:00
if (IS_ERR(new_ah->ah)) {
pr_warn("Couldn't create new SM AH\n");
kfree(new_ah);
return;
}
spin_lock_irq(&port->ah_lock);
if (port->sm_ah)
kref_put(&port->sm_ah->ref, free_sm_ah);
port->sm_ah = new_ah;
spin_unlock_irq(&port->ah_lock);
}
static void ib_sa_event(struct ib_event_handler *handler,
struct ib_event *event)
{
if (event->event == IB_EVENT_PORT_ERR ||
event->event == IB_EVENT_PORT_ACTIVE ||
event->event == IB_EVENT_LID_CHANGE ||
event->event == IB_EVENT_PKEY_CHANGE ||
event->event == IB_EVENT_SM_CHANGE ||
event->event == IB_EVENT_CLIENT_REREGISTER) {
unsigned long flags;
struct ib_sa_device *sa_dev =
container_of(handler, typeof(*sa_dev), event_handler);
RDMA: Support more than 255 rdma ports Current code uses many different types when dealing with a port of a RDMA device: u8, unsigned int and u32. Switch to u32 to clean up the logic. This allows us to make (at least) the core view consistent and use the same type. Unfortunately not all places can be converted. Many uverbs functions expect port to be u8 so keep those places in order not to break UAPIs. HW/Spec defined values must also not be changed. With the switch to u32 we now can support devices with more than 255 ports. U32_MAX is reserved to make control logic a bit easier to deal with. As a device with U32_MAX ports probably isn't going to happen any time soon this seems like a non issue. When a device with more than 255 ports is created uverbs will report the RDMA device as having 255 ports as this is the max currently supported. The verbs interface is not changed yet because the IBTA spec limits the port size in too many places to be u8 and all applications that relies in verbs won't be able to cope with this change. At this stage, we are extending the interfaces that are using vendor channel solely Once the limitation is lifted mlx5 in switchdev mode will be able to have thousands of SFs created by the device. As the only instance of an RDMA device that reports more than 255 ports will be a representor device and it exposes itself as a RAW Ethernet only device CM/MAD/IPoIB and other ULPs aren't effected by this change and their sysfs/interfaces that are exposes to userspace can remain unchanged. While here cleanup some alignment issues and remove unneeded sanity checks (mainly in rdmavt), Link: https://lore.kernel.org/r/20210301070420.439400-1-leon@kernel.org Signed-off-by: Mark Bloch <mbloch@nvidia.com> Signed-off-by: Leon Romanovsky <leonro@nvidia.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
2021-03-01 15:04:20 +08:00
u32 port_num = event->element.port_num - sa_dev->start_port;
IB/SA: Move functions update_sm_ah() and ib_sa_event() Moving these will facilitate changes to these in the next patchs. This is strictly a move and there are no changes to the functions in any way. Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-03-21 07:38:07 +08:00
struct ib_sa_port *port = &sa_dev->port[port_num];
if (!rdma_cap_ib_sa(handler->device, port->port_num))
return;
spin_lock_irqsave(&port->ah_lock, flags);
if (port->sm_ah)
kref_put(&port->sm_ah->ref, free_sm_ah);
port->sm_ah = NULL;
spin_unlock_irqrestore(&port->ah_lock, flags);
if (event->event == IB_EVENT_SM_CHANGE ||
event->event == IB_EVENT_CLIENT_REREGISTER ||
IB/SA: Modify SA to implicitly cache Class Port info SA will query and cache class port info as part of its initialization. SA will also invalidate and refresh the cache based on specific events. Callers such as IPoIB and CM can query the SA to get the classportinfo information. Apart from making the caller code much simpler, this change puts the onus on the SA to query and maintain classportinfo much like how it maitains the address handle to the SM. Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Don Hiatt <don.hiatt@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-03-21 07:38:08 +08:00
event->event == IB_EVENT_LID_CHANGE ||
event->event == IB_EVENT_PORT_ACTIVE) {
unsigned long delay =
msecs_to_jiffies(IB_SA_CPI_RETRY_WAIT);
IB/SA: Move functions update_sm_ah() and ib_sa_event() Moving these will facilitate changes to these in the next patchs. This is strictly a move and there are no changes to the functions in any way. Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-03-21 07:38:07 +08:00
spin_lock_irqsave(&port->classport_lock, flags);
port->classport_info.valid = false;
IB/SA: Modify SA to implicitly cache Class Port info SA will query and cache class port info as part of its initialization. SA will also invalidate and refresh the cache based on specific events. Callers such as IPoIB and CM can query the SA to get the classportinfo information. Apart from making the caller code much simpler, this change puts the onus on the SA to query and maintain classportinfo much like how it maitains the address handle to the SM. Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Don Hiatt <don.hiatt@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-03-21 07:38:08 +08:00
port->classport_info.retry_cnt = 0;
IB/SA: Move functions update_sm_ah() and ib_sa_event() Moving these will facilitate changes to these in the next patchs. This is strictly a move and there are no changes to the functions in any way. Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-03-21 07:38:07 +08:00
spin_unlock_irqrestore(&port->classport_lock, flags);
IB/SA: Modify SA to implicitly cache Class Port info SA will query and cache class port info as part of its initialization. SA will also invalidate and refresh the cache based on specific events. Callers such as IPoIB and CM can query the SA to get the classportinfo information. Apart from making the caller code much simpler, this change puts the onus on the SA to query and maintain classportinfo much like how it maitains the address handle to the SM. Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Don Hiatt <don.hiatt@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-03-21 07:38:08 +08:00
queue_delayed_work(ib_wq,
&port->ib_cpi_work, delay);
IB/SA: Move functions update_sm_ah() and ib_sa_event() Moving these will facilitate changes to these in the next patchs. This is strictly a move and there are no changes to the functions in any way. Reviewed-by: Don Hiatt <don.hiatt@intel.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-03-21 07:38:07 +08:00
}
queue_work(ib_wq, &sa_dev->port[port_num].update_task);
}
}
RDMA: Allow ib_client's to fail when add() is called When a client is added it isn't allowed to fail, but all the client's have various failure paths within their add routines. This creates the very fringe condition where the client was added, failed during add and didn't set the client_data. The core code will then still call other client_data centric ops like remove(), rename(), get_nl_info(), and get_net_dev_by_params() with NULL client_data - which is confusing and unexpected. If the add() callback fails, then do not call any more client ops for the device, even remove. Remove all the now redundant checks for NULL client_data in ops callbacks. Update all the add() callbacks to return error codes appropriately. EOPNOTSUPP is used for cases where the ULP does not support the ib_device - eg because it only works with IB. Link: https://lore.kernel.org/r/20200421172440.387069-1-leon@kernel.org Signed-off-by: Leon Romanovsky <leonro@mellanox.com> Acked-by: Ursula Braun <ubraun@linux.ibm.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-04-22 01:24:40 +08:00
static int ib_sa_add_one(struct ib_device *device)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
{
struct ib_sa_device *sa_dev;
int s, e, i;
IB/Verbs: Reform IB-core sa_query Use raw management helpers to reform IB-core sa_query. Signed-off-by: Michael Wang <yun.wang@profitbricks.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Tested-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Sean Hefty <sean.hefty@intel.com> Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Tested-by: Doug Ledford <dledford@redhat.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-05-05 20:50:22 +08:00
int count = 0;
RDMA: Allow ib_client's to fail when add() is called When a client is added it isn't allowed to fail, but all the client's have various failure paths within their add routines. This creates the very fringe condition where the client was added, failed during add and didn't set the client_data. The core code will then still call other client_data centric ops like remove(), rename(), get_nl_info(), and get_net_dev_by_params() with NULL client_data - which is confusing and unexpected. If the add() callback fails, then do not call any more client ops for the device, even remove. Remove all the now redundant checks for NULL client_data in ops callbacks. Update all the add() callbacks to return error codes appropriately. EOPNOTSUPP is used for cases where the ULP does not support the ib_device - eg because it only works with IB. Link: https://lore.kernel.org/r/20200421172440.387069-1-leon@kernel.org Signed-off-by: Leon Romanovsky <leonro@mellanox.com> Acked-by: Ursula Braun <ubraun@linux.ibm.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-04-22 01:24:40 +08:00
int ret;
RDMA: iWARP Core Changes. Modifications to the existing rdma header files, core files, drivers, and ulp files to support iWARP, including: - Hook iWARP CM into the build system and use it in rdma_cm. - Convert enum ib_node_type to enum rdma_node_type, which includes the possibility of RDMA_NODE_RNIC, and update everything for this. Signed-off-by: Tom Tucker <tom@opengridcomputing.com> Signed-off-by: Steve Wise <swise@opengridcomputing.com> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2006-08-04 05:02:42 +08:00
IB: Add rdma_cap_ib_switch helper and use where appropriate Persuant to Liran's comments on node_type on linux-rdma mailing list: In an effort to reform the RDMA core and ULPs to minimize use of node_type in struct ib_device, an additional bit is added to struct ib_device for is_switch (IB switch). This is needed to be initialized by any IB switch device driver. This is a NEW requirement on such device drivers which are all "out of tree". In addition, an ib_switch helper was added to ib_verbs.h based on the is_switch device bit rather than node_type (although those should be consistent). The RDMA core (MAD, SMI, agent, sa_query, multicast, sysfs) as well as (IPoIB and SRP) ULPs are updated where appropriate to use this new helper. In some cases, the helper is now used under the covers of using rdma_[start end]_port rather than the open coding previously used. Reviewed-by: Sean Hefty <sean.hefty@intel.com> Reviewed-By: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Tested-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Hal Rosenstock <hal@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-06-29 21:57:00 +08:00
s = rdma_start_port(device);
e = rdma_end_port(device);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
RDMA/core: Use size_{add,sub,mul}() in calls to struct_size() If, for any reason, the open-coded arithmetic causes a wraparound, the protection that `struct_size()` provides against potential integer overflows is defeated. Fix this by hardening calls to `struct_size()` with `size_add()`, `size_sub()` and `size_mul()`. Fixes: 467f432a521a ("RDMA/core: Split port and device counter sysfs attributes") Fixes: a4676388e2e2 ("RDMA/core: Simplify how the gid_attrs sysfs is created") Fixes: e9dd5daf884c ("IB/umad: Refactor code to use cdev_device_add()") Fixes: 324e227ea7c9 ("RDMA/device: Add ib_device_get_by_netdev()") Fixes: 5aad26a7eac5 ("IB/core: Use struct_size() in kzalloc()") Signed-off-by: Gustavo A. R. Silva <gustavoars@kernel.org> Link: https://lore.kernel.org/r/ZQdt4NsJFwwOYxUR@work Signed-off-by: Leon Romanovsky <leon@kernel.org>
2023-09-18 05:21:36 +08:00
sa_dev = kzalloc(struct_size(sa_dev, port,
size_add(size_sub(e, s), 1)),
GFP_KERNEL);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
if (!sa_dev)
RDMA: Allow ib_client's to fail when add() is called When a client is added it isn't allowed to fail, but all the client's have various failure paths within their add routines. This creates the very fringe condition where the client was added, failed during add and didn't set the client_data. The core code will then still call other client_data centric ops like remove(), rename(), get_nl_info(), and get_net_dev_by_params() with NULL client_data - which is confusing and unexpected. If the add() callback fails, then do not call any more client ops for the device, even remove. Remove all the now redundant checks for NULL client_data in ops callbacks. Update all the add() callbacks to return error codes appropriately. EOPNOTSUPP is used for cases where the ULP does not support the ib_device - eg because it only works with IB. Link: https://lore.kernel.org/r/20200421172440.387069-1-leon@kernel.org Signed-off-by: Leon Romanovsky <leonro@mellanox.com> Acked-by: Ursula Braun <ubraun@linux.ibm.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-04-22 01:24:40 +08:00
return -ENOMEM;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
sa_dev->start_port = s;
sa_dev->end_port = e;
for (i = 0; i <= e - s; ++i) {
IB/mad: IBoE supports only QP1 (no QP0) Since IBoE is using Ethernet as its link layer, there is no central management entity so there is need for QP0. QP1 is still needed since it handles communications between CM agents. This patch will skip QP0 and create only QP1 for IBoE ports. Signed-off-by: Eli Cohen <eli@mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2010-09-28 08:51:11 +08:00
spin_lock_init(&sa_dev->port[i].ah_lock);
IB/Verbs: Use management helper rdma_cap_ib_sa() Introduce helper rdma_cap_ib_sa() to help us check if the port of an IB device support Infiniband Subnet Administration. Signed-off-by: Michael Wang <yun.wang@profitbricks.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Tested-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Sean Hefty <sean.hefty@intel.com> Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Tested-by: Doug Ledford <dledford@redhat.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-05-05 20:50:36 +08:00
if (!rdma_cap_ib_sa(device, i + 1))
IB/mad: IBoE supports only QP1 (no QP0) Since IBoE is using Ethernet as its link layer, there is no central management entity so there is need for QP0. QP1 is still needed since it handles communications between CM agents. This patch will skip QP0 and create only QP1 for IBoE ports. Signed-off-by: Eli Cohen <eli@mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2010-09-28 08:51:11 +08:00
continue;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
sa_dev->port[i].sm_ah = NULL;
sa_dev->port[i].port_num = i + s;
IB/sa: Add cached attribute containing SM information to SA port Added a new SA port attribute containing SM ClassPortInfo fields, (ClassPortInfo fields: Table 126 IB Spec 1.3.). This is useful for checking SM support for specific features. The attribute is cached to avoid resending queries, caching is done when a successful ClassPortInfo reply is received on the port. Invalidation of the attribute is done on SM change events, SM re-registration events, and SM LID change events. The fields in ClassPortInfo should not change during SM runtime without an event. Signed-off-by: Alex Vesker <valex@mellanox.com> Reviewed by: Hal Rosenstock <hal@mellanox.com> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-07-06 21:36:34 +08:00
spin_lock_init(&sa_dev->port[i].classport_lock);
sa_dev->port[i].classport_info.valid = false;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
sa_dev->port[i].agent =
ib_register_mad_agent(device, i + s, IB_QPT_GSI,
NULL, 0, send_handler,
IB/mad: add new ioctl to ABI to support new registration options Registrations options are specified through flags. Definitions of flags will be in subsequent patches. Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Roland Dreier <roland@purestorage.com>
2014-08-09 07:00:55 +08:00
recv_handler, sa_dev, 0);
RDMA: Allow ib_client's to fail when add() is called When a client is added it isn't allowed to fail, but all the client's have various failure paths within their add routines. This creates the very fringe condition where the client was added, failed during add and didn't set the client_data. The core code will then still call other client_data centric ops like remove(), rename(), get_nl_info(), and get_net_dev_by_params() with NULL client_data - which is confusing and unexpected. If the add() callback fails, then do not call any more client ops for the device, even remove. Remove all the now redundant checks for NULL client_data in ops callbacks. Update all the add() callbacks to return error codes appropriately. EOPNOTSUPP is used for cases where the ULP does not support the ib_device - eg because it only works with IB. Link: https://lore.kernel.org/r/20200421172440.387069-1-leon@kernel.org Signed-off-by: Leon Romanovsky <leonro@mellanox.com> Acked-by: Ursula Braun <ubraun@linux.ibm.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-04-22 01:24:40 +08:00
if (IS_ERR(sa_dev->port[i].agent)) {
ret = PTR_ERR(sa_dev->port[i].agent);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
goto err;
RDMA: Allow ib_client's to fail when add() is called When a client is added it isn't allowed to fail, but all the client's have various failure paths within their add routines. This creates the very fringe condition where the client was added, failed during add and didn't set the client_data. The core code will then still call other client_data centric ops like remove(), rename(), get_nl_info(), and get_net_dev_by_params() with NULL client_data - which is confusing and unexpected. If the add() callback fails, then do not call any more client ops for the device, even remove. Remove all the now redundant checks for NULL client_data in ops callbacks. Update all the add() callbacks to return error codes appropriately. EOPNOTSUPP is used for cases where the ULP does not support the ib_device - eg because it only works with IB. Link: https://lore.kernel.org/r/20200421172440.387069-1-leon@kernel.org Signed-off-by: Leon Romanovsky <leonro@mellanox.com> Acked-by: Ursula Braun <ubraun@linux.ibm.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-04-22 01:24:40 +08:00
}
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
WorkStruct: make allyesconfig Fix up for make allyesconfig. Signed-Off-By: David Howells <dhowells@redhat.com>
2006-11-22 22:57:56 +08:00
INIT_WORK(&sa_dev->port[i].update_task, update_sm_ah);
IB/SA: Modify SA to implicitly cache Class Port info SA will query and cache class port info as part of its initialization. SA will also invalidate and refresh the cache based on specific events. Callers such as IPoIB and CM can query the SA to get the classportinfo information. Apart from making the caller code much simpler, this change puts the onus on the SA to query and maintain classportinfo much like how it maitains the address handle to the SM. Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Don Hiatt <don.hiatt@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-03-21 07:38:08 +08:00
INIT_DELAYED_WORK(&sa_dev->port[i].ib_cpi_work,
update_ib_cpi);
IB/Verbs: Reform IB-core sa_query Use raw management helpers to reform IB-core sa_query. Signed-off-by: Michael Wang <yun.wang@profitbricks.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Tested-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Sean Hefty <sean.hefty@intel.com> Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Tested-by: Doug Ledford <dledford@redhat.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-05-05 20:50:22 +08:00
count++;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
RDMA: Allow ib_client's to fail when add() is called When a client is added it isn't allowed to fail, but all the client's have various failure paths within their add routines. This creates the very fringe condition where the client was added, failed during add and didn't set the client_data. The core code will then still call other client_data centric ops like remove(), rename(), get_nl_info(), and get_net_dev_by_params() with NULL client_data - which is confusing and unexpected. If the add() callback fails, then do not call any more client ops for the device, even remove. Remove all the now redundant checks for NULL client_data in ops callbacks. Update all the add() callbacks to return error codes appropriately. EOPNOTSUPP is used for cases where the ULP does not support the ib_device - eg because it only works with IB. Link: https://lore.kernel.org/r/20200421172440.387069-1-leon@kernel.org Signed-off-by: Leon Romanovsky <leonro@mellanox.com> Acked-by: Ursula Braun <ubraun@linux.ibm.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-04-22 01:24:40 +08:00
if (!count) {
ret = -EOPNOTSUPP;
IB/Verbs: Reform IB-core sa_query Use raw management helpers to reform IB-core sa_query. Signed-off-by: Michael Wang <yun.wang@profitbricks.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Tested-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Sean Hefty <sean.hefty@intel.com> Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Tested-by: Doug Ledford <dledford@redhat.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-05-05 20:50:22 +08:00
goto free;
RDMA: Allow ib_client's to fail when add() is called When a client is added it isn't allowed to fail, but all the client's have various failure paths within their add routines. This creates the very fringe condition where the client was added, failed during add and didn't set the client_data. The core code will then still call other client_data centric ops like remove(), rename(), get_nl_info(), and get_net_dev_by_params() with NULL client_data - which is confusing and unexpected. If the add() callback fails, then do not call any more client ops for the device, even remove. Remove all the now redundant checks for NULL client_data in ops callbacks. Update all the add() callbacks to return error codes appropriately. EOPNOTSUPP is used for cases where the ULP does not support the ib_device - eg because it only works with IB. Link: https://lore.kernel.org/r/20200421172440.387069-1-leon@kernel.org Signed-off-by: Leon Romanovsky <leonro@mellanox.com> Acked-by: Ursula Braun <ubraun@linux.ibm.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-04-22 01:24:40 +08:00
}
IB/Verbs: Reform IB-core sa_query Use raw management helpers to reform IB-core sa_query. Signed-off-by: Michael Wang <yun.wang@profitbricks.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Tested-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Sean Hefty <sean.hefty@intel.com> Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Tested-by: Doug Ledford <dledford@redhat.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-05-05 20:50:22 +08:00
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
ib_set_client_data(device, &sa_client, sa_dev);
/*
* We register our event handler after everything is set up,
* and then update our cached info after the event handler is
* registered to avoid any problems if a port changes state
* during our initialization.
*/
INIT_IB_EVENT_HANDLER(&sa_dev->event_handler, device, ib_sa_event);
RDMA/(core, ulp): Convert register/unregister event handler to be void The functions ib_register_event_handler() and ib_unregister_event_handler() always returned success and they can't fail. Let's convert those functions to be void, remove redundant checks and cleanup tons of goto statements. Signed-off-by: Leon Romanovsky <leonro@mellanox.com> Reviewed-by: Dennis Dalessandro <dennis.dalessandro@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-08-17 20:50:36 +08:00
ib_register_event_handler(&sa_dev->event_handler);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
IB/Verbs: Reform IB-core sa_query Use raw management helpers to reform IB-core sa_query. Signed-off-by: Michael Wang <yun.wang@profitbricks.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Tested-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Sean Hefty <sean.hefty@intel.com> Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Tested-by: Doug Ledford <dledford@redhat.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-05-05 20:50:22 +08:00
for (i = 0; i <= e - s; ++i) {
IB/Verbs: Use management helper rdma_cap_ib_sa() Introduce helper rdma_cap_ib_sa() to help us check if the port of an IB device support Infiniband Subnet Administration. Signed-off-by: Michael Wang <yun.wang@profitbricks.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Tested-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Sean Hefty <sean.hefty@intel.com> Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Tested-by: Doug Ledford <dledford@redhat.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-05-05 20:50:36 +08:00
if (rdma_cap_ib_sa(device, i + 1))
IB/mad: IBoE supports only QP1 (no QP0) Since IBoE is using Ethernet as its link layer, there is no central management entity so there is need for QP0. QP1 is still needed since it handles communications between CM agents. This patch will skip QP0 and create only QP1 for IBoE ports. Signed-off-by: Eli Cohen <eli@mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2010-09-28 08:51:11 +08:00
update_sm_ah(&sa_dev->port[i].update_task);
IB/Verbs: Reform IB-core sa_query Use raw management helpers to reform IB-core sa_query. Signed-off-by: Michael Wang <yun.wang@profitbricks.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Tested-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Sean Hefty <sean.hefty@intel.com> Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Tested-by: Doug Ledford <dledford@redhat.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-05-05 20:50:22 +08:00
}
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
RDMA: Allow ib_client's to fail when add() is called When a client is added it isn't allowed to fail, but all the client's have various failure paths within their add routines. This creates the very fringe condition where the client was added, failed during add and didn't set the client_data. The core code will then still call other client_data centric ops like remove(), rename(), get_nl_info(), and get_net_dev_by_params() with NULL client_data - which is confusing and unexpected. If the add() callback fails, then do not call any more client ops for the device, even remove. Remove all the now redundant checks for NULL client_data in ops callbacks. Update all the add() callbacks to return error codes appropriately. EOPNOTSUPP is used for cases where the ULP does not support the ib_device - eg because it only works with IB. Link: https://lore.kernel.org/r/20200421172440.387069-1-leon@kernel.org Signed-off-by: Leon Romanovsky <leonro@mellanox.com> Acked-by: Ursula Braun <ubraun@linux.ibm.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-04-22 01:24:40 +08:00
return 0;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
err:
IB/Verbs: Reform IB-core sa_query Use raw management helpers to reform IB-core sa_query. Signed-off-by: Michael Wang <yun.wang@profitbricks.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Tested-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Sean Hefty <sean.hefty@intel.com> Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Tested-by: Doug Ledford <dledford@redhat.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-05-05 20:50:22 +08:00
while (--i >= 0) {
IB/Verbs: Use management helper rdma_cap_ib_sa() Introduce helper rdma_cap_ib_sa() to help us check if the port of an IB device support Infiniband Subnet Administration. Signed-off-by: Michael Wang <yun.wang@profitbricks.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Tested-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Sean Hefty <sean.hefty@intel.com> Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Tested-by: Doug Ledford <dledford@redhat.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-05-05 20:50:36 +08:00
if (rdma_cap_ib_sa(device, i + 1))
IB/mad: IBoE supports only QP1 (no QP0) Since IBoE is using Ethernet as its link layer, there is no central management entity so there is need for QP0. QP1 is still needed since it handles communications between CM agents. This patch will skip QP0 and create only QP1 for IBoE ports. Signed-off-by: Eli Cohen <eli@mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2010-09-28 08:51:11 +08:00
ib_unregister_mad_agent(sa_dev->port[i].agent);
IB/Verbs: Reform IB-core sa_query Use raw management helpers to reform IB-core sa_query. Signed-off-by: Michael Wang <yun.wang@profitbricks.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Tested-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Sean Hefty <sean.hefty@intel.com> Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Tested-by: Doug Ledford <dledford@redhat.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-05-05 20:50:22 +08:00
}
free:
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
kfree(sa_dev);
RDMA: Allow ib_client's to fail when add() is called When a client is added it isn't allowed to fail, but all the client's have various failure paths within their add routines. This creates the very fringe condition where the client was added, failed during add and didn't set the client_data. The core code will then still call other client_data centric ops like remove(), rename(), get_nl_info(), and get_net_dev_by_params() with NULL client_data - which is confusing and unexpected. If the add() callback fails, then do not call any more client ops for the device, even remove. Remove all the now redundant checks for NULL client_data in ops callbacks. Update all the add() callbacks to return error codes appropriately. EOPNOTSUPP is used for cases where the ULP does not support the ib_device - eg because it only works with IB. Link: https://lore.kernel.org/r/20200421172440.387069-1-leon@kernel.org Signed-off-by: Leon Romanovsky <leonro@mellanox.com> Acked-by: Ursula Braun <ubraun@linux.ibm.com> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2020-04-22 01:24:40 +08:00
return ret;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
IB/core: lock client data with lists_rwsem An ib_client callback that is called with the lists_rwsem locked only for read is protected from changes to the IB client lists, but not from ib_unregister_device() freeing its client data. This is because ib_unregister_device() will remove the device from the device list with lists_rwsem locked for write, but perform the rest of the cleanup, including the call to remove() without that lock. Mark client data that is undergoing de-registration with a new going_down flag in the client data context. Lock the client data list with lists_rwsem for write in addition to using the spinlock, so that functions calling the callback would be able to lock only lists_rwsem for read and let callbacks sleep. Since ib_unregister_client() now marks the client data context, no need for remove() to search the context again, so pass the client data directly to remove() callbacks. Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Signed-off-by: Haggai Eran <haggaie@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-07-30 22:50:14 +08:00
static void ib_sa_remove_one(struct ib_device *device, void *client_data)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
{
IB/core: lock client data with lists_rwsem An ib_client callback that is called with the lists_rwsem locked only for read is protected from changes to the IB client lists, but not from ib_unregister_device() freeing its client data. This is because ib_unregister_device() will remove the device from the device list with lists_rwsem locked for write, but perform the rest of the cleanup, including the call to remove() without that lock. Mark client data that is undergoing de-registration with a new going_down flag in the client data context. Lock the client data list with lists_rwsem for write in addition to using the spinlock, so that functions calling the callback would be able to lock only lists_rwsem for read and let callbacks sleep. Since ib_unregister_client() now marks the client data context, no need for remove() to search the context again, so pass the client data directly to remove() callbacks. Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Signed-off-by: Haggai Eran <haggaie@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-07-30 22:50:14 +08:00
struct ib_sa_device *sa_dev = client_data;
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
int i;
ib_unregister_event_handler(&sa_dev->event_handler);
RDMA: Update missed conversion of flush_scheduled_work() Commit f06267104dd9 ("RDMA: Update workqueue usage") introduced ib_wq and removed the use of flush_scheduled_work(); however, during the merge process one chunk was lost in ib_sa_remove_one(). Fix it. Signed-off-by: Tejun Heo <tj@kernel.org> Signed-off-by: Roland Dreier <roland@purestorage.com>
2011-01-24 19:06:54 +08:00
flush_workqueue(ib_wq);
IB/sa_query: Flush scheduled work before unloading module sa_query schedules work on IB asynchronous events. After unregistering the async event handler, make sure that this work has completed before releasing the IB device (and possibly allowing the sa_query module text to go away). Signed-off-by: Michael S. Tsirkin <mst@mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2006-01-18 01:53:51 +08:00
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
for (i = 0; i <= sa_dev->end_port - sa_dev->start_port; ++i) {
IB/Verbs: Use management helper rdma_cap_ib_sa() Introduce helper rdma_cap_ib_sa() to help us check if the port of an IB device support Infiniband Subnet Administration. Signed-off-by: Michael Wang <yun.wang@profitbricks.com> Reviewed-by: Ira Weiny <ira.weiny@intel.com> Tested-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Sean Hefty <sean.hefty@intel.com> Reviewed-by: Jason Gunthorpe <jgunthorpe@obsidianresearch.com> Tested-by: Doug Ledford <dledford@redhat.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-05-05 20:50:36 +08:00
if (rdma_cap_ib_sa(device, i + 1)) {
IB/SA: Modify SA to implicitly cache Class Port info SA will query and cache class port info as part of its initialization. SA will also invalidate and refresh the cache based on specific events. Callers such as IPoIB and CM can query the SA to get the classportinfo information. Apart from making the caller code much simpler, this change puts the onus on the SA to query and maintain classportinfo much like how it maitains the address handle to the SM. Reviewed-by: Ira Weiny <ira.weiny@intel.com> Reviewed-by: Don Hiatt <don.hiatt@intel.com> Signed-off-by: Dasaratharaman Chandramouli <dasaratharaman.chandramouli@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2017-03-21 07:38:08 +08:00
cancel_delayed_work_sync(&sa_dev->port[i].ib_cpi_work);
IB/mad: IBoE supports only QP1 (no QP0) Since IBoE is using Ethernet as its link layer, there is no central management entity so there is need for QP0. QP1 is still needed since it handles communications between CM agents. This patch will skip QP0 and create only QP1 for IBoE ports. Signed-off-by: Eli Cohen <eli@mellanox.co.il> Signed-off-by: Roland Dreier <rolandd@cisco.com>
2010-09-28 08:51:11 +08:00
ib_unregister_mad_agent(sa_dev->port[i].agent);
if (sa_dev->port[i].sm_ah)
kref_put(&sa_dev->port[i].sm_ah->ref, free_sm_ah);
}
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
kfree(sa_dev);
}
IB/SA: Integrate ib_sa module into ib_core module Consolidate ib_sa into ib_core, this commit eliminates ib_sa.ko and makes it part of ib_core.ko Signed-off-by: Mark Bloch <markb@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-05-19 22:12:33 +08:00
int ib_sa_init(void)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
{
int ret;
get_random_bytes(&tid, sizeof tid);
IB/sa: Route SA pathrecord query through netlink This patch routes a SA pathrecord query to netlink first and processes the response appropriately. If a failure is returned, the request will be sent through IB. The decision whether to route the request to netlink first is determined by the presence of a listener for the local service netlink multicast group. If the user-space local service netlink multicast group listener is not present, the request will be sent through IB, just like what is currently being done. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: John Fleck <john.fleck@intel.com> Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-14 20:52:09 +08:00
atomic_set(&ib_nl_sa_request_seq, 0);
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
ret = ib_register_client(&sa_client);
IB/sa: Track multicast join/leave requests The IB SA tracks multicast join/leave requests on a per port basis and does not do any reference counting: if two users of the same port join the same group, and one leaves that group, then the SA will remove the port from the group even though there is one user who wants to stay a member left. Therefore, in order to support multiple users of the same multicast group from the same port, we need to perform reference counting locally. To do this, add an multicast submodule to ib_sa to perform reference counting of multicast join/leave operations. Modify ib_ipoib (the only in-kernel user of multicast) to use the new interface. Signed-off-by: Roland Dreier <rolandd@cisco.com>
2007-02-16 09:00:17 +08:00
if (ret) {
IB/core: trivial prink cleanup. 1. Replaced printk with appropriate pr_warn, pr_err, pr_info. 2. Removed unnecessary prints around memory allocation failure which are not required, as reported by the checkpatch script. Signed-off-by: Parav Pandit <pandit.parav@gmail.com> Reviewed-by: Haggai Eran <haggaie@mellanox.com> Reviewed-by: Sagi Grimberg <sagig@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-03-02 03:20:29 +08:00
pr_err("Couldn't register ib_sa client\n");
IB/sa: Track multicast join/leave requests The IB SA tracks multicast join/leave requests on a per port basis and does not do any reference counting: if two users of the same port join the same group, and one leaves that group, then the SA will remove the port from the group even though there is one user who wants to stay a member left. Therefore, in order to support multiple users of the same multicast group from the same port, we need to perform reference counting locally. To do this, add an multicast submodule to ib_sa to perform reference counting of multicast join/leave operations. Modify ib_ipoib (the only in-kernel user of multicast) to use the new interface. Signed-off-by: Roland Dreier <rolandd@cisco.com>
2007-02-16 09:00:17 +08:00
goto err1;
}
ret = mcast_init();
if (ret) {
IB/core: trivial prink cleanup. 1. Replaced printk with appropriate pr_warn, pr_err, pr_info. 2. Removed unnecessary prints around memory allocation failure which are not required, as reported by the checkpatch script. Signed-off-by: Parav Pandit <pandit.parav@gmail.com> Reviewed-by: Haggai Eran <haggaie@mellanox.com> Reviewed-by: Sagi Grimberg <sagig@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-03-02 03:20:29 +08:00
pr_err("Couldn't initialize multicast handling\n");
IB/sa: Track multicast join/leave requests The IB SA tracks multicast join/leave requests on a per port basis and does not do any reference counting: if two users of the same port join the same group, and one leaves that group, then the SA will remove the port from the group even though there is one user who wants to stay a member left. Therefore, in order to support multiple users of the same multicast group from the same port, we need to perform reference counting locally. To do this, add an multicast submodule to ib_sa to perform reference counting of multicast join/leave operations. Modify ib_ipoib (the only in-kernel user of multicast) to use the new interface. Signed-off-by: Roland Dreier <rolandd@cisco.com>
2007-02-16 09:00:17 +08:00
goto err2;
}
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
IB/sa : Remove deprecated create_singlethread_workqueue alloc_ordered_workqueue() with WQ_MEM_RECLAIM set, replaces deprecated create_singlethread_workqueue(). This is the identity conversion. The workqueue "ib_nl" queues work item &ib_nl_timed_work. It has been identity converted. WQ_MEM_RECLAIM has been set to ensure forward progress under memory pressure. Signed-off-by: Bhaktipriya Shridhar <bhaktipriya96@gmail.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-08-16 01:57:46 +08:00
ib_nl_wq = alloc_ordered_workqueue("ib_nl_sa_wq", WQ_MEM_RECLAIM);
IB/sa: Route SA pathrecord query through netlink This patch routes a SA pathrecord query to netlink first and processes the response appropriately. If a failure is returned, the request will be sent through IB. The decision whether to route the request to netlink first is determined by the presence of a listener for the local service netlink multicast group. If the user-space local service netlink multicast group listener is not present, the request will be sent through IB, just like what is currently being done. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: John Fleck <john.fleck@intel.com> Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-14 20:52:09 +08:00
if (!ib_nl_wq) {
ret = -ENOMEM;
goto err3;
}
INIT_DELAYED_WORK(&ib_nl_timed_work, ib_nl_request_timeout);
IB/sa: Track multicast join/leave requests The IB SA tracks multicast join/leave requests on a per port basis and does not do any reference counting: if two users of the same port join the same group, and one leaves that group, then the SA will remove the port from the group even though there is one user who wants to stay a member left. Therefore, in order to support multiple users of the same multicast group from the same port, we need to perform reference counting locally. To do this, add an multicast submodule to ib_sa to perform reference counting of multicast join/leave operations. Modify ib_ipoib (the only in-kernel user of multicast) to use the new interface. Signed-off-by: Roland Dreier <rolandd@cisco.com>
2007-02-16 09:00:17 +08:00
return 0;
IB/core: Register SA ibnl client during ib_core initialization Move SA ibnl client registration to ib_core module init. This will allow us to register a single client to handle all RDMA_NL_LS operations and make it SA independent. Signed-off-by: Mark Bloch <markb@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-05-19 22:12:35 +08:00
IB/sa: Route SA pathrecord query through netlink This patch routes a SA pathrecord query to netlink first and processes the response appropriately. If a failure is returned, the request will be sent through IB. The decision whether to route the request to netlink first is determined by the presence of a listener for the local service netlink multicast group. If the user-space local service netlink multicast group listener is not present, the request will be sent through IB, just like what is currently being done. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: John Fleck <john.fleck@intel.com> Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-14 20:52:09 +08:00
err3:
mcast_cleanup();
IB/sa: Track multicast join/leave requests The IB SA tracks multicast join/leave requests on a per port basis and does not do any reference counting: if two users of the same port join the same group, and one leaves that group, then the SA will remove the port from the group even though there is one user who wants to stay a member left. Therefore, in order to support multiple users of the same multicast group from the same port, we need to perform reference counting locally. To do this, add an multicast submodule to ib_sa to perform reference counting of multicast join/leave operations. Modify ib_ipoib (the only in-kernel user of multicast) to use the new interface. Signed-off-by: Roland Dreier <rolandd@cisco.com>
2007-02-16 09:00:17 +08:00
err2:
ib_unregister_client(&sa_client);
err1:
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
return ret;
}
IB/SA: Integrate ib_sa module into ib_core module Consolidate ib_sa into ib_core, this commit eliminates ib_sa.ko and makes it part of ib_core.ko Signed-off-by: Mark Bloch <markb@mellanox.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2016-05-19 22:12:33 +08:00
void ib_sa_cleanup(void)
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
{
IB/sa: Route SA pathrecord query through netlink This patch routes a SA pathrecord query to netlink first and processes the response appropriately. If a failure is returned, the request will be sent through IB. The decision whether to route the request to netlink first is determined by the presence of a listener for the local service netlink multicast group. If the user-space local service netlink multicast group listener is not present, the request will be sent through IB, just like what is currently being done. Signed-off-by: Kaike Wan <kaike.wan@intel.com> Signed-off-by: John Fleck <john.fleck@intel.com> Signed-off-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Doug Ledford <dledford@redhat.com>
2015-08-14 20:52:09 +08:00
cancel_delayed_work(&ib_nl_timed_work);
destroy_workqueue(ib_nl_wq);
IB/sa: Track multicast join/leave requests The IB SA tracks multicast join/leave requests on a per port basis and does not do any reference counting: if two users of the same port join the same group, and one leaves that group, then the SA will remove the port from the group even though there is one user who wants to stay a member left. Therefore, in order to support multiple users of the same multicast group from the same port, we need to perform reference counting locally. To do this, add an multicast submodule to ib_sa to perform reference counting of multicast join/leave operations. Modify ib_ipoib (the only in-kernel user of multicast) to use the new interface. Signed-off-by: Roland Dreier <rolandd@cisco.com>
2007-02-16 09:00:17 +08:00
mcast_cleanup();
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
ib_unregister_client(&sa_client);
ib core: Convert query_idr to XArray Signed-off-by: Matthew Wilcox <willy@infradead.org> Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
2019-02-21 08:20:46 +08:00
WARN_ON(!xa_empty(&queries));
Linux-2.6.12-rc2 Initial git repository build. I'm not bothering with the full history, even though we have it. We can create a separate "historical" git archive of that later if we want to, and in the meantime it's about 3.2GB when imported into git - space that would just make the early git days unnecessarily complicated, when we don't have a lot of good infrastructure for it. Let it rip!
2005-04-17 06:20:36 +08:00
}
Reference in New Issue Copy Permalink