summaryrefslogtreecommitdiff
path: root/drivers/infiniband/sw/rxe
AgeCommit message (Collapse)Author
2025-06-08treewide, timers: Rename from_timer() to timer_container_of()Ingo Molnar
Move this API to the canonical timer_*() namespace. [ tglx: Redone against pre rc1 ] Signed-off-by: Ingo Molnar <mingo@kernel.org> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Link: https://lore.kernel.org/all/aB2X0jCKQO56WdMt@gmail.com
2025-05-26Merge tag 'v6.15' into rdma.git for-nextJason Gunthorpe
Following patches need the RDMA rc branch since we are past the RC cycle now. Merge conflicts resolved based on Linux-next: - For RXE odp changes keep for-next version and fixup new places that need to call is_odp_mr() https://lore.kernel.org/r/20250422143019.500201bd@canb.auug.org.au https://lore.kernel.org/r/20250514122455.3593b083@canb.auug.org.au - irdma is keeping the while/kfree bugfix from -rc and the pf/cdev_info change from for-next https://lore.kernel.org/r/20250513130630.280ee6c5@canb.auug.org.au Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
2025-05-22RDMA/rxe: Break endless pagefault loop for RO pagesLeon Romanovsky
RO pages has "perm" equal to 0, that caused to the situation where such pages were marked as needed to have fault and caused to infinite loop. Fixes: eedd5b1276e7 ("RDMA/umem: Store ODP access mask information in PFN") Reported-by: Daisuke Matsuda <dskmtsd@gmail.com> Closes: https://lore.kernel.org/all/3016329a-4edd-4550-862f-b298a1b79a39@gmail.com/ Link: https://patch.msgid.link/096fab178d48ed86942ee22eafe9be98e29092aa.1747913377.git.leonro@nvidia.com Tested-by: Daisuke Matsuda <dskmtsd@gmail.com> Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
2025-05-12RDMA/core: Convert UMEM ODP DMA mapping to caching IOVA and page linkageLeon Romanovsky
Reuse newly added DMA API to cache IOVA and only link/unlink pages in fast path for UMEM ODP flow. Tested-by: Jens Axboe <axboe@kernel.dk> Reviewed-by: Jason Gunthorpe <jgg@nvidia.com> Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
2025-05-12RDMA/umem: Store ODP access mask information in PFNLeon Romanovsky
As a preparation to remove dma_list, store access mask in PFN pointer and not in dma_addr_t. Tested-by: Jens Axboe <axboe@kernel.dk> Reviewed-by: Jason Gunthorpe <jgg@nvidia.com> Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
2025-04-21RDMA/rxe: Remove 32-bit architecture supportDaisuke Matsuda
Major linux distibutions have phased out support for 32-bit machines. Since rxe is primarily used for development and testing, the benefit of maintaining 32-bit support is minimal. This change simplifies ATOMIC WRITE implementations and improves maintainability of the driver. Signed-off-by: Daisuke Matsuda <matsuda-daisuke@fujitsu.com> Link: https://patch.msgid.link/20250421025101.3588139-1-matsuda-daisuke@fujitsu.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2025-04-20RDMA/rxe: Remove unused rxe_run_taskDr. David Alan Gilbert
rxe_run_task() has been unused since 2024's commit 23bc06af547f ("RDMA/rxe: Don't call direct between tasks") Remove it. Signed-off-by: Dr. David Alan Gilbert <linux@treblig.org> Link: https://patch.msgid.link/20250419132725.199785-1-linux@treblig.org Signed-off-by: Leon Romanovsky <leon@kernel.org>
2025-04-20RDMA/rxe: Fix "trying to register non-static key in rxe_qp_do_cleanup" bugZhu Yanjun
Call Trace: <TASK> __dump_stack lib/dump_stack.c:94 [inline] dump_stack_lvl+0x116/0x1f0 lib/dump_stack.c:120 assign_lock_key kernel/locking/lockdep.c:986 [inline] register_lock_class+0x4a3/0x4c0 kernel/locking/lockdep.c:1300 __lock_acquire+0x99/0x1ba0 kernel/locking/lockdep.c:5110 lock_acquire kernel/locking/lockdep.c:5866 [inline] lock_acquire+0x179/0x350 kernel/locking/lockdep.c:5823 __timer_delete_sync+0x152/0x1b0 kernel/time/timer.c:1644 rxe_qp_do_cleanup+0x5c3/0x7e0 drivers/infiniband/sw/rxe/rxe_qp.c:815 execute_in_process_context+0x3a/0x160 kernel/workqueue.c:4596 __rxe_cleanup+0x267/0x3c0 drivers/infiniband/sw/rxe/rxe_pool.c:232 rxe_create_qp+0x3f7/0x5f0 drivers/infiniband/sw/rxe/rxe_verbs.c:604 create_qp+0x62d/0xa80 drivers/infiniband/core/verbs.c:1250 ib_create_qp_kernel+0x9f/0x310 drivers/infiniband/core/verbs.c:1361 ib_create_qp include/rdma/ib_verbs.h:3803 [inline] rdma_create_qp+0x10c/0x340 drivers/infiniband/core/cma.c:1144 rds_ib_setup_qp+0xc86/0x19a0 net/rds/ib_cm.c:600 rds_ib_cm_initiate_connect+0x1e8/0x3d0 net/rds/ib_cm.c:944 rds_rdma_cm_event_handler_cmn+0x61f/0x8c0 net/rds/rdma_transport.c:109 cma_cm_event_handler+0x94/0x300 drivers/infiniband/core/cma.c:2184 cma_work_handler+0x15b/0x230 drivers/infiniband/core/cma.c:3042 process_one_work+0x9cc/0x1b70 kernel/workqueue.c:3238 process_scheduled_works kernel/workqueue.c:3319 [inline] worker_thread+0x6c8/0xf10 kernel/workqueue.c:3400 kthread+0x3c2/0x780 kernel/kthread.c:464 ret_from_fork+0x45/0x80 arch/x86/kernel/process.c:153 ret_from_fork_asm+0x1a/0x30 arch/x86/entry/entry_64.S:245 </TASK> The root cause is as below: In the function rxe_create_qp, the function rxe_qp_from_init is called to create qp, if this function rxe_qp_from_init fails, rxe_cleanup will be called to handle all the allocated resources, including the timers: retrans_timer and rnr_nak_timer. The function rxe_qp_from_init calls the function rxe_qp_init_req to initialize the timers: retrans_timer and rnr_nak_timer. But these timers are initialized in the end of rxe_qp_init_req. If some errors occur before the initialization of these timers, this problem will occur. The solution is to check whether these timers are initialized or not. If these timers are not initialized, ignore these timers. Fixes: 8700e3e7c485 ("Soft RoCE driver") Reported-by: syzbot+4edb496c3cad6e953a31@syzkaller.appspotmail.com Closes: https://syzkaller.appspot.com/bug?extid=4edb496c3cad6e953a31 Signed-off-by: Zhu Yanjun <yanjun.zhu@linux.dev> Link: https://patch.msgid.link/20250419080741.1515231-1-yanjun.zhu@linux.dev Signed-off-by: Leon Romanovsky <leon@kernel.org>
2025-04-20RDMA/rxe: Fix slab-use-after-free Read in rxe_queue_cleanup bugZhu Yanjun
Call Trace: <TASK> __dump_stack lib/dump_stack.c:94 [inline] dump_stack_lvl+0x7d/0xa0 lib/dump_stack.c:120 print_address_description mm/kasan/report.c:378 [inline] print_report+0xcf/0x610 mm/kasan/report.c:489 kasan_report+0xb5/0xe0 mm/kasan/report.c:602 rxe_queue_cleanup+0xd0/0xe0 drivers/infiniband/sw/rxe/rxe_queue.c:195 rxe_cq_cleanup+0x3f/0x50 drivers/infiniband/sw/rxe/rxe_cq.c:132 __rxe_cleanup+0x168/0x300 drivers/infiniband/sw/rxe/rxe_pool.c:232 rxe_create_cq+0x22e/0x3a0 drivers/infiniband/sw/rxe/rxe_verbs.c:1109 create_cq+0x658/0xb90 drivers/infiniband/core/uverbs_cmd.c:1052 ib_uverbs_create_cq+0xc7/0x120 drivers/infiniband/core/uverbs_cmd.c:1095 ib_uverbs_write+0x969/0xc90 drivers/infiniband/core/uverbs_main.c:679 vfs_write fs/read_write.c:677 [inline] vfs_write+0x26a/0xcc0 fs/read_write.c:659 ksys_write+0x1b8/0x200 fs/read_write.c:731 do_syscall_x64 arch/x86/entry/common.c:52 [inline] do_syscall_64+0xaa/0x1b0 arch/x86/entry/common.c:83 entry_SYSCALL_64_after_hwframe+0x77/0x7f In the function rxe_create_cq, when rxe_cq_from_init fails, the function rxe_cleanup will be called to handle the allocated resources. In fact, some memory resources have already been freed in the function rxe_cq_from_init. Thus, this problem will occur. The solution is to let rxe_cleanup do all the work. Fixes: 8700e3e7c485 ("Soft RoCE driver") Link: https://paste.ubuntu.com/p/tJgC42wDf6/ Tested-by: liuyi <liuy22@mails.tsinghua.edu.cn> Signed-off-by: Zhu Yanjun <yanjun.zhu@linux.dev> Link: https://patch.msgid.link/20250412075714.3257358-1-yanjun.zhu@linux.dev Reviewed-by: Daisuke Matsuda <matsuda-daisuke@fujitsu.com> Signed-off-by: Leon Romanovsky <leon@kernel.org>
2025-04-11RDMA/rxe: Fix mismatched type declarationsDaisuke Matsuda
Some functions return int values while they are defined as enum resp_states variables. This patch resolves the mismatches in rxe. Signed-off-by: Daisuke Matsuda <matsuda-daisuke@fujitsu.com> Link: https://patch.msgid.link/20250409102701.1275265-1-matsuda-daisuke@fujitsu.com Reviewed-by: Zhu Yanjun <yanjun.zhu@linux.dev> Signed-off-by: Leon Romanovsky <leon@kernel.org>
2025-04-09RDMA/rxe: Enable ODP in ATOMIC WRITE operationDaisuke Matsuda
Add rxe_odp_do_atomic_write() so that ODP specific steps are applied to ATOMIC WRITE requests. Signed-off-by: Daisuke Matsuda <matsuda-daisuke@fujitsu.com> Link: https://patch.msgid.link/20250324075649.3313968-3-matsuda-daisuke@fujitsu.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2025-04-08RDMA/rxe: Enable ODP in RDMA FLUSH operationDaisuke Matsuda
For persistent memories, add rxe_odp_flush_pmem_iova() so that ODP specific steps are executed. Otherwise, no additional consideration is required. Signed-off-by: Daisuke Matsuda <matsuda-daisuke@fujitsu.com> Link: https://patch.msgid.link/20250324075649.3313968-2-matsuda-daisuke@fujitsu.com Reviewed-by: Li Zhijian <lizhijian@fujitsu.com> Signed-off-by: Leon Romanovsky <leon@kernel.org>
2025-04-07RDMA/rxe: Fix null pointer dereference in ODP MR checkLi Zhijian
The blktests/rnbd reported a null pointer dereference as following. Similar to the mlx5, introduce a is_odp_mr() to check if the odp is enabled in this mr. Workqueue: rxe_wq do_work [rdma_rxe] RIP: 0010:rxe_mr_copy+0x57/0x210 [rdma_rxe] Code: 7c 04 48 89 f3 48 89 d5 41 89 cf 45 89 c4 0f 84 dc 00 00 00 89 ca e8 f8 f8 ff ff 85 c0 0f 85 75 01 00 00 49 8b 86 f0 00 00 00 <f6> 40 28 02 0f 85 98 01 00 00 41 8b 46 78 41 8b 8e 10 01 00 00 8d RSP: 0018:ffffa0aac02cfcf8 EFLAGS: 00010246 RAX: 0000000000000000 RBX: ffff9079cd440024 RCX: 0000000000000000 RDX: 000000000000003c RSI: ffff9079cd440060 RDI: ffff9079cd665600 RBP: ffff9079c0e5e45a R08: 0000000000000000 R09: 0000000000000000 R10: 000000003c000000 R11: 0000000000225510 R12: 0000000000000000 R13: 0000000000000000 R14: ffff9079cd665600 R15: 000000000000003c FS: 0000000000000000(0000) GS:ffff907ccfa80000(0000) knlGS:0000000000000000 CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 CR2: 0000000000000028 CR3: 0000000119498001 CR4: 00000000001726f0 Call Trace: <TASK> ? __die_body+0x1e/0x60 ? page_fault_oops+0x14f/0x4c0 ? rxe_mr_copy+0x57/0x210 [rdma_rxe] ? search_bpf_extables+0x5f/0x80 ? exc_page_fault+0x7e/0x180 ? asm_exc_page_fault+0x26/0x30 ? rxe_mr_copy+0x57/0x210 [rdma_rxe] ? rxe_mr_copy+0x48/0x210 [rdma_rxe] ? rxe_pool_get_index+0x50/0x90 [rdma_rxe] rxe_receiver+0x1d98/0x2530 [rdma_rxe] ? psi_task_switch+0x1ff/0x250 ? finish_task_switch+0x92/0x2d0 ? __schedule+0xbdf/0x17c0 do_task+0x65/0x1e0 [rdma_rxe] process_scheduled_works+0xaa/0x3f0 worker_thread+0x117/0x240 Fixes: d03fb5c6599e ("RDMA/rxe: Allow registering MRs for On-Demand Paging") Cc: stable@vger.kernel.org Link: https://patch.msgid.link/r/20250402032657.1762800-1-lizhijian@fujitsu.com Signed-off-by: Li Zhijian <lizhijian@fujitsu.com> Reviewed-by: Daisuke Matsuda <matsuda-daisuke@fujitsu.com> Reviewed-by: Zhu Yanjun <yanjun.zhu@linux.dev> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
2025-04-05treewide: Switch/rename to timer_delete[_sync]()Thomas Gleixner
timer_delete[_sync]() replaces del_timer[_sync](). Convert the whole tree over and remove the historical wrapper inlines. Conversion was done with coccinelle plus manual fixups where necessary. Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Ingo Molnar <mingo@kernel.org>
2025-03-29Merge tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rdma/rdmaLinus Torvalds
Pull rdma updates from Jason Gunthorpe: - Usual minor updates and fixes for bnxt_re, hfi1, rxe, mana, iser, mlx5, vmw_pvrdma, hns - Make rxe work on tun devices - mana gains more standard verbs as it moves toward supporting in-kernel verbs - DMABUF support for mana - Fix page size calculations when memory registration exceeds 4G - On Demand Paging support for rxe - mlx5 support for RDMA TRANSPORT flow tables and a new ucap mechanism to access control use of them - Optional RDMA_TX/RX counters per QP in mlx5 * tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rdma/rdma: (73 commits) IB/mad: Check available slots before posting receive WRs RDMA/mana_ib: Fix integer overflow during queue creation RDMA/mlx5: Fix calculation of total invalidated pages RDMA/mlx5: Fix mlx5_poll_one() cur_qp update flow RDMA/mlx5: Fix page_size variable overflow RDMA/mlx5: Drop access_flags from _mlx5_mr_cache_alloc() RDMA/mlx5: Fix cache entry update on dereg error RDMA/mlx5: Fix MR cache initialization error flow RDMA/mlx5: Support optional-counters binding for QPs RDMA/mlx5: Compile fs.c regardless of INFINIBAND_USER_ACCESS config RDMA/core: Pass port to counter bind/unbind operations RDMA/core: Add support to optional-counters binding configuration RDMA/core: Create and destroy rdma_counter using rdma_zalloc_drv_obj() RDMA/mlx5: Add optional counters for RDMA_TX/RX_packets/bytes RDMA/core: Fix use-after-free when rename device name RDMA/bnxt_re: Support perf management counters RDMA/rxe: Fix incorrect return value of rxe_odp_atomic_op() RDMA/uverbs: Propagate errors from rdma_lookup_get_uobject() RDMA/mana_ib: Handle net event for pointing to the current netdev net: mana: Change the function signature of mana_get_primary_netdev_rcu ...
2025-03-13RDMA/rxe: Fix incorrect return value of rxe_odp_atomic_op()Daisuke Matsuda
rxe_mr_do_atomic_op() returns enum resp_states numbers, so the ODP counterpart must not return raw errno codes. Signed-off-by: Daisuke Matsuda <matsuda-daisuke@fujitsu.com> Reviewed-by: Zhu Yanjun <yanjun.zhu@linux.dev> Link: https://patch.msgid.link/20250313064540.2619115-1-matsuda-daisuke@fujitsu.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2025-03-12RDMA/rxe: Improve readability of ODP pagefault interfaceDaisuke Matsuda
Use a meaningful constant explicitly instead of hard-coding a literal. Signed-off-by: Daisuke Matsuda <matsuda-daisuke@fujitsu.com> Link: https://patch.msgid.link/20250312065937.1787241-1-matsuda-daisuke@fujitsu.com Reviewed-by: Zhu Yanjun <yanjun.zhu@linux.dev> Signed-off-by: Leon Romanovsky <leon@kernel.org>
2025-03-03RDMA/rxe: Fix the failure of ibv_query_device() and ibv_query_device_ex() testsZhu Yanjun
In rdma-core, the following failures appear. " $ ./build/bin/run_tests.py -k device ssssssss....FF........s ====================================================================== FAIL: test_query_device (tests.test_device.DeviceTest.test_query_device) Test ibv_query_device() ---------------------------------------------------------------------- Traceback (most recent call last): File "/home/ubuntu/rdma-core/tests/test_device.py", line 63, in test_query_device self.verify_device_attr(attr, dev) File "/home/ubuntu/rdma-core/tests/test_device.py", line 200, in verify_device_attr assert attr.sys_image_guid != 0 ^^^^^^^^^^^^^^^^^^^^^^^^ AssertionError ====================================================================== FAIL: test_query_device_ex (tests.test_device.DeviceTest.test_query_device_ex) Test ibv_query_device_ex() ---------------------------------------------------------------------- Traceback (most recent call last): File "/home/ubuntu/rdma-core/tests/test_device.py", line 222, in test_query_device_ex self.verify_device_attr(attr_ex.orig_attr, dev) File "/home/ubuntu/rdma-core/tests/test_device.py", line 200, in verify_device_attr assert attr.sys_image_guid != 0 ^^^^^^^^^^^^^^^^^^^^^^^^ AssertionError " The root cause is: before a net device is set with rxe, this net device is used to generate a sys_image_guid. Fixes: 2ac5415022d1 ("RDMA/rxe: Remove the direct link to net_device") Signed-off-by: Zhu Yanjun <yanjun.zhu@linux.dev> Link: https://patch.msgid.link/20250302215444.3742072-1-yanjun.zhu@linux.dev Reviewed-by: Daisuke Matsuda <matsuda-daisuke@fujitsu.com> Tested-by: Daisuke Matsuda <matsuda-daisuke@fujitsu.com> Signed-off-by: Leon Romanovsky <leon@kernel.org>
2025-02-21RDMA/rxe: Add support for the traditional Atomic operations with ODPDaisuke Matsuda
Enable 'fetch and add' and 'compare and swap' operations to be used with ODP. This is comprised of the following steps: 1. Check the driver page table(umem_odp->dma_list) to see if the target page is both readable and writable. 2. If not, then trigger page fault to map the page. 3. Convert its user space address to a kernel logical address using PFNs in the driver page table(umem_odp->pfn_list). 4. Execute the operation. Link: https://patch.msgid.link/r/20241220100936.2193541-6-matsuda-daisuke@fujitsu.com Signed-off-by: Daisuke Matsuda <matsuda-daisuke@fujitsu.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
2025-02-21RDMA/rxe: Add support for Send/Recv/Write/Read with ODPDaisuke Matsuda
rxe_mr_copy() is used widely to copy data to/from a user MR. requester uses it to load payloads of requesting packets; responder uses it to process Send, Write, and Read operaetions; completer uses it to copy data from response packets of Read and Atomic operations to a user MR. Allow these operations to be used with ODP by adding a subordinate function rxe_odp_mr_copy(). It is comprised of the following steps: 1. Check the driver page table(umem_odp->dma_list) to see if pages being accessed are present with appropriate permission. 2. If necessary, trigger page fault to map the pages. 3. Convert their user space addresses to kernel logical addresses using PFNs in the driver page table(umem_odp->pfn_list). 4. Execute data copy to/from the pages. Link: https://patch.msgid.link/r/20241220100936.2193541-5-matsuda-daisuke@fujitsu.com Signed-off-by: Daisuke Matsuda <matsuda-daisuke@fujitsu.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
2025-02-21RDMA/rxe: Allow registering MRs for On-Demand PagingDaisuke Matsuda
Allow userspace to register an ODP-enabled MR, in which case the flag IB_ACCESS_ON_DEMAND is passed to rxe_reg_user_mr(). However, there is no RDMA operation enabled right now. They will be supported later in the subsequent two patches. rxe_odp_do_pagefault() is called to initialize an ODP-enabled MR. It syncs process address space from the CPU page table to the driver page table (dma_list/pfn_list in umem_odp) when called with RXE_PAGEFAULT_SNAPSHOT flag. Additionally, It can be used to trigger page fault when pages being accessed are not present or do not have proper read/write permissions, and possibly to prefetch pages in the future. Link: https://patch.msgid.link/r/20241220100936.2193541-4-matsuda-daisuke@fujitsu.com Signed-off-by: Daisuke Matsuda <matsuda-daisuke@fujitsu.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
2025-02-21RDMA/rxe: Add page invalidation supportDaisuke Matsuda
On page invalidation, an MMU notifier callback is invoked to unmap DMA addresses and update the driver page table(umem_odp->dma_list). The callback is registered when an ODP-enabled MR is created. Link: https://patch.msgid.link/r/20241220100936.2193541-3-matsuda-daisuke@fujitsu.com Signed-off-by: Daisuke Matsuda <matsuda-daisuke@fujitsu.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
2025-02-21RDMA/rxe: Move some code to rxe_loc.h in preparation for ODPDaisuke Matsuda
rxe_mr_init() and resp_states are going to be used in rxe_odp.c, which is to be created in the subsequent patch. Link: https://patch.msgid.link/r/20241220100936.2193541-2-matsuda-daisuke@fujitsu.com Signed-off-by: Daisuke Matsuda <matsuda-daisuke@fujitsu.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
2025-02-09RDMA/rxe: switch to using the crc32 libraryEric Biggers
Now that the crc32_le() library function takes advantage of architecture-specific optimizations, it is unnecessary to go through the crypto API. Just use crc32_le(). This is much simpler, and it improves performance due to eliminating the crypto API overhead. Signed-off-by: Eric Biggers <ebiggers@google.com> Link: https://patch.msgid.link/20250207032316.53941-1-ebiggers@kernel.org Reviewed-by: Zhu Yanjun <yanjun.zhu@linux.dev> Signed-off-by: Leon Romanovsky <leon@kernel.org>
2025-02-03RDMA/rxe: Add query_gid supportZhu Yanjun
The query_gid is not implemented in RXE. After the raw_gid is added, this query_gid should be implemented in RXE. Signed-off-by: Zhu Yanjun <yanjun.zhu@linux.dev> Link: https://patch.msgid.link/20250119172831.3123110-3-yanjun.zhu@linux.dev Signed-off-by: Leon Romanovsky <leon@kernel.org>
2025-02-03RDMA/rxe: Replace netdev dev addr with raw_gidZhu Yanjun
Because TUN device does not have dev_addr, but a gid in rdma is needed, as such, a raw_gid is generated to act as the gid. The similar commit is in SIW. This commit learns from the similar commit bad5b6e34ffb ("RDMA/siw: Fabricate a GID on tun and loopback devices") in SIW. Signed-off-by: Zhu Yanjun <yanjun.zhu@linux.dev> Link: https://patch.msgid.link/20250119172831.3123110-2-yanjun.zhu@linux.dev Signed-off-by: Leon Romanovsky <leon@kernel.org>
2025-01-24Merge tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rdma/rdmaLinus Torvalds
Pull rdma updates from Jason Gunthorpe: "Lighter that normal, but the now usual collection of driver fixes and small improvements: - Small fixes and minor improvements to cxgb4, bnxt_re, rxe, srp, efa, cxgb4 - Update mlx4 to use the new umem APIs, avoiding direct use of scatterlist - Support ROCEv2 in erdma - Remove various uncalled functions, constify bin_attribute - Provide core infrastructure to catch netdev events and route them to drivers, consolidating duplicated driver code - Fix rare race condition crashes in mlx5 ODP flows" * tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rdma/rdma: (63 commits) RDMA/mlx5: Fix implicit ODP use after free RDMA/mlx5: Fix a race for an ODP MR which leads to CQE with error RDMA/qib: Constify 'struct bin_attribute' RDMA/hfi1: Constify 'struct bin_attribute' RDMA/rxe: Fix the warning "__rxe_cleanup+0x12c/0x170 [rdma_rxe]" RDMA/cxgb4: Notify rdma stack for IB_EVENT_QP_LAST_WQE_REACHED event RDMA/bnxt_re: Allocate dev_attr information dynamically RDMA/bnxt_re: Pass the context for ulp_irq_stop RDMA/bnxt_re: Add support to handle DCB_CONFIG_CHANGE event RDMA/bnxt_re: Query firmware defaults of CC params during probe RDMA/bnxt_re: Add Async event handling support bnxt_en: Add ULP call to notify async events RDMA/mlx5: Fix indirect mkey ODP page count MAINTAINERS: Update the bnxt_re maintainers RDMA/hns: Clean up the legacy CONFIG_INFINIBAND_HNS RDMA/rtrs: Add missing deinit() call RDMA/efa: Align interrupt related fields to same type RDMA/bnxt_re: Fix to drop reference to the mmap entry in case of error RDMA/mlx5: Fix link status down event for MPV RDMA/erdma: Support create_ah/destroy_ah in non-sleepable contexts ...
2025-01-14RDMA/rxe: Fix the warning "__rxe_cleanup+0x12c/0x170 [rdma_rxe]"Zhu Yanjun
The Call Trace is as below: " <TASK> ? show_regs.cold+0x1a/0x1f ? __rxe_cleanup+0x12c/0x170 [rdma_rxe] ? __warn+0x84/0xd0 ? __rxe_cleanup+0x12c/0x170 [rdma_rxe] ? report_bug+0x105/0x180 ? handle_bug+0x46/0x80 ? exc_invalid_op+0x19/0x70 ? asm_exc_invalid_op+0x1b/0x20 ? __rxe_cleanup+0x12c/0x170 [rdma_rxe] ? __rxe_cleanup+0x124/0x170 [rdma_rxe] rxe_destroy_qp.cold+0x24/0x29 [rdma_rxe] ib_destroy_qp_user+0x118/0x190 [ib_core] rdma_destroy_qp.cold+0x43/0x5e [rdma_cm] rtrs_cq_qp_destroy.cold+0x1d/0x2b [rtrs_core] rtrs_srv_close_work.cold+0x1b/0x31 [rtrs_server] process_one_work+0x21d/0x3f0 worker_thread+0x4a/0x3c0 ? process_one_work+0x3f0/0x3f0 kthread+0xf0/0x120 ? kthread_complete_and_exit+0x20/0x20 ret_from_fork+0x22/0x30 </TASK> " When too many rdma resources are allocated, rxe needs more time to handle these rdma resources. Sometimes with the current timeout, rxe can not release the rdma resources correctly. Compared with other rdma drivers, a bigger timeout is used. Fixes: 215d0a755e1b ("RDMA/rxe: Stop lookup of partially built objects") Signed-off-by: Zhu Yanjun <yanjun.zhu@linux.dev> Link: https://patch.msgid.link/20250110160927.55014-1-yanjun.zhu@linux.dev Tested-by: Joe Klein <joe.klein812@gmail.com> Signed-off-by: Leon Romanovsky <leon@kernel.org>
2024-12-25RDMA/rxe: Remove deliver net device eventYuyu Li
Since the netdev events of link status is now handled in ib_core, remove the related code in drivers. In addition, remove the setting of port->attr.state in rxe_port_up() and rxe_port_down(), as it is only used in rxe_query_port(), and it can be replaced by ib_get_curr_port_state(). Signed-off-by: Yuyu Li <liyuyu6@huawei.com> Signed-off-by: Junxian Huang <huangjunxian6@hisilicon.com> Signed-off-by: Leon Romanovsky <leon@kernel.org>
2024-12-24RDMA/rxe: Remove the direct link to net_deviceZhu Yanjun
The similar patch in siw is in the link: https://git.kernel.org/rdma/rdma/c/16b87037b48889 This problem also occurred in RXE. The following analyze this problem. In the following Call Traces: " BUG: KASAN: slab-use-after-free in dev_get_flags+0x188/0x1d0 net/core/dev.c:8782 Read of size 4 at addr ffff8880554640b0 by task kworker/1:4/5295 CPU: 1 UID: 0 PID: 5295 Comm: kworker/1:4 Not tainted 6.12.0-rc3-syzkaller-00399-g9197b73fd7bb #0 Hardware name: Google Compute Engine/Google Compute Engine, BIOS Google 09/13/2024 Workqueue: infiniband ib_cache_event_task Call Trace: <TASK> __dump_stack lib/dump_stack.c:94 [inline] dump_stack_lvl+0x241/0x360 lib/dump_stack.c:120 print_address_description mm/kasan/report.c:377 [inline] print_report+0x169/0x550 mm/kasan/report.c:488 kasan_report+0x143/0x180 mm/kasan/report.c:601 dev_get_flags+0x188/0x1d0 net/core/dev.c:8782 rxe_query_port+0x12d/0x260 drivers/infiniband/sw/rxe/rxe_verbs.c:60 __ib_query_port drivers/infiniband/core/device.c:2111 [inline] ib_query_port+0x168/0x7d0 drivers/infiniband/core/device.c:2143 ib_cache_update+0x1a9/0xb80 drivers/infiniband/core/cache.c:1494 ib_cache_event_task+0xf3/0x1e0 drivers/infiniband/core/cache.c:1568 process_one_work kernel/workqueue.c:3229 [inline] process_scheduled_works+0xa65/0x1850 kernel/workqueue.c:3310 worker_thread+0x870/0xd30 kernel/workqueue.c:3391 kthread+0x2f2/0x390 kernel/kthread.c:389 ret_from_fork+0x4d/0x80 arch/x86/kernel/process.c:147 ret_from_fork_asm+0x1a/0x30 arch/x86/entry/entry_64.S:244 </TASK> " 1). In the link [1], " infiniband syz2: set down " This means that on 839.350575, the event ib_cache_event_task was sent andi queued in ib_wq. 2). In the link [1], " team0 (unregistering): Port device team_slave_0 removed " It indicates that before 843.251853, the net device should be freed. 3). In the link [1], " BUG: KASAN: slab-use-after-free in dev_get_flags+0x188/0x1d0 " This means that on 850.559070, this slab-use-after-free problem occurred. In all, on 839.350575, the event ib_cache_event_task was sent and queued in ib_wq, before 843.251853, the net device veth was freed. on 850.559070, this event was executed, and the mentioned freed net device was called. Thus, the above call trace occurred. [1] https://syzkaller.appspot.com/x/log.txt?x=12e7025f980000 Reported-by: syzbot+4b87489410b4efd181bf@syzkaller.appspotmail.com Closes: https://syzkaller.appspot.com/bug?extid=4b87489410b4efd181bf Fixes: 8700e3e7c485 ("Soft RoCE driver") Signed-off-by: Zhu Yanjun <yanjun.zhu@linux.dev> Link: https://patch.msgid.link/20241220222325.2487767-1-yanjun.zhu@linux.dev Signed-off-by: Leon Romanovsky <leon@kernel.org>
2024-12-17RDMA/rxe: Fix mismatched max_msg_szzhenwei pi
User mode queries max_msg_sz as 0x800000 by command 'ibv_devinfo -v', however ibv_post_send/ibv_post_recv has a limit of 2^31. Fix this mismatched information. Signed-off-by: zhenwei pi <pizhenwei@bytedance.com> Fixes: f605f26ea196 ("RDMA/rxe: Protect QP state with qp->state_lock") Fixes: 5bf944f24129 ("RDMA/rxe: Add error messages") Link: https://patch.msgid.link/20241216121953.765331-1-pizhenwei@bytedance.com Review-by: Zhu Yanjun <yanjun.zhu@linux.dev> Signed-off-by: Leon Romanovsky <leon@kernel.org>
2024-11-04RDMA/rxe: Set queue pair cur_qp_state when being queriedLiu Jian
Same with commit e375b9c92985 ("RDMA/cxgb4: Set queue pair state when being queried"). The API for ib_query_qp requires the driver to set cur_qp_state on return, add the missing set. Fixes: 8700e3e7c485 ("Soft RoCE driver") Signed-off-by: Liu Jian <liujian56@huawei.com> Link: https://patch.msgid.link/20241031092019.2138467-1-liujian56@huawei.com Reviewed-by: Zhu Yanjun <yanjun.zhu@linux.dev> Signed-off-by: Leon Romanovsky <leon@kernel.org>
2024-10-30RDMA/rxe: Fix the qp flush warnings in reqZhu Yanjun
When the qp is in error state, the status of WQEs in the queue should be set to error. Or else the following will appear. [ 920.617269] WARNING: CPU: 1 PID: 21 at drivers/infiniband/sw/rxe/rxe_comp.c:756 rxe_completer+0x989/0xcc0 [rdma_rxe] [ 920.617744] Modules linked in: rnbd_client(O) rtrs_client(O) rtrs_core(O) rdma_ucm rdma_cm iw_cm ib_cm crc32_generic rdma_rxe ip6_udp_tunnel udp_tunnel ib_uverbs ib_core loop brd null_blk ipv6 [ 920.618516] CPU: 1 PID: 21 Comm: ksoftirqd/1 Tainted: G O 6.1.113-storage+ #65 [ 920.618986] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04/01/2014 [ 920.619396] RIP: 0010:rxe_completer+0x989/0xcc0 [rdma_rxe] [ 920.619658] Code: 0f b6 84 24 3a 02 00 00 41 89 84 24 44 04 00 00 e9 2a f7 ff ff 39 ca bb 03 00 00 00 b8 0e 00 00 00 48 0f 45 d8 e9 15 f7 ff ff <0f> 0b e9 cb f8 ff ff 41 bf f5 ff ff ff e9 08 f8 ff ff 49 8d bc 24 [ 920.620482] RSP: 0018:ffff97b7c00bbc38 EFLAGS: 00010246 [ 920.620817] RAX: 0000000000000000 RBX: 000000000000000c RCX: 0000000000000008 [ 920.621183] RDX: ffff960dc396ebc0 RSI: 0000000000005400 RDI: ffff960dc4e2fbac [ 920.621548] RBP: 0000000000000000 R08: 0000000000000001 R09: ffffffffac406450 [ 920.621884] R10: ffffffffac4060c0 R11: 0000000000000001 R12: ffff960dc4e2f800 [ 920.622254] R13: ffff960dc4e2f928 R14: ffff97b7c029c580 R15: 0000000000000000 [ 920.622609] FS: 0000000000000000(0000) GS:ffff960ef7d00000(0000) knlGS:0000000000000000 [ 920.622979] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 920.623245] CR2: 00007fa056965e90 CR3: 00000001107f1000 CR4: 00000000000006e0 [ 920.623680] Call Trace: [ 920.623815] <TASK> [ 920.623933] ? __warn+0x79/0xc0 [ 920.624116] ? rxe_completer+0x989/0xcc0 [rdma_rxe] [ 920.624356] ? report_bug+0xfb/0x150 [ 920.624594] ? handle_bug+0x3c/0x60 [ 920.624796] ? exc_invalid_op+0x14/0x70 [ 920.624976] ? asm_exc_invalid_op+0x16/0x20 [ 920.625203] ? rxe_completer+0x989/0xcc0 [rdma_rxe] [ 920.625474] ? rxe_completer+0x329/0xcc0 [rdma_rxe] [ 920.625749] rxe_do_task+0x80/0x110 [rdma_rxe] [ 920.626037] rxe_requester+0x625/0xde0 [rdma_rxe] [ 920.626310] ? rxe_cq_post+0xe2/0x180 [rdma_rxe] [ 920.626583] ? do_complete+0x18d/0x220 [rdma_rxe] [ 920.626812] ? rxe_completer+0x1a3/0xcc0 [rdma_rxe] [ 920.627050] rxe_do_task+0x80/0x110 [rdma_rxe] [ 920.627285] tasklet_action_common.constprop.0+0xa4/0x120 [ 920.627522] handle_softirqs+0xc2/0x250 [ 920.627728] ? sort_range+0x20/0x20 [ 920.627942] run_ksoftirqd+0x1f/0x30 [ 920.628158] smpboot_thread_fn+0xc7/0x1b0 [ 920.628334] kthread+0xd6/0x100 [ 920.628504] ? kthread_complete_and_exit+0x20/0x20 [ 920.628709] ret_from_fork+0x1f/0x30 [ 920.628892] </TASK> Fixes: ae720bdb703b ("RDMA/rxe: Generate error completion for error requester QP state") Signed-off-by: Zhu Yanjun <yanjun.zhu@linux.dev> Link: https://patch.msgid.link/20241025152036.121417-1-yanjun.zhu@linux.dev Signed-off-by: Leon Romanovsky <leon@kernel.org>
2024-08-23RDMA/rxe: Fix __bth_set_resv6azhenwei pi
__bth_set_resv6a is used to clear BIT [24, 29] of rxe_bth::qpn, the wrong expression leads other BITs into 1. Link: https://patch.msgid.link/r/20240822065223.1117056-4-pizhenwei@bytedance.com Signed-off-by: zhenwei pi <pizhenwei@bytedance.com> Reviewed-by: Zhu Yanjun <yanjun.zhu@linux.dev> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
2024-08-23RDMA/rxe: Fix misspelling of 'rmda'zhenwei pi
Fix 'rmda' into 'RDMA'. Link: https://patch.msgid.link/r/20240822065223.1117056-3-pizhenwei@bytedance.com Signed-off-by: zhenwei pi <pizhenwei@bytedance.com> Reviewed-by: Zhu Yanjun <yanjun.zhu@linux.dev> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
2024-08-23RDMA/rxe: Use sizeof instead of hard code numberzhenwei pi
Use 'sizeof(union rdma_network_hdr)' instead of hard code GRH length for GSI and UD. Link: https://patch.msgid.link/r/20240822065223.1117056-2-pizhenwei@bytedance.com Signed-off-by: zhenwei pi <pizhenwei@bytedance.com> Reviewed-by: Zhu Yanjun <yanjun.zhu@linux.dev> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
2024-07-19Merge tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rdma/rdmaLinus Torvalds
Pull rdma updates from Jason Gunthorpe: "Usual collection of small improvements and fixes: - Bug fixes and minor improvments in efa, irdma, mlx4, mlx5, rxe, hf1, qib, ocrdma - bnxt_re support for MSN, which is a new retransmit logic - Initial mana support for RC qps - Use after free bug and cleanups in iwcm - Reduce resource usage in mlx5 when RDMA verbs features are not used - New verb to drain shared recieve queues, similar to normal recieve queues. This is necessary to allow ULPs a clean shutdown. Used in the iscsi rdma target - mlx5 support for more than 16 bits of doorbell indexes - Doorbell moderation support for bnxt_re - IB multi-plane support for mlx5 - New EFA adaptor PCI IDs - RDMA_NAME_ASSIGN_TYPE_USER to hint to userspace that it shouldn't rename the device - A collection of hns bugs - Fix long standing bug in bnxt_re with incorrect endian handling of immediate data" * tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rdma/rdma: (65 commits) IB/hfi1: Constify struct flag_table RDMA/mana_ib: Set correct device into ib bnxt_re: Fix imm_data endianness RDMA: Fix netdev tracker in ib_device_set_netdev RDMA/hns: Fix mbx timing out before CMD execution is completed RDMA/hns: Fix insufficient extend DB for VFs. RDMA/hns: Fix undifined behavior caused by invalid max_sge RDMA/hns: Fix shift-out-bounds when max_inline_data is 0 RDMA/hns: Fix missing pagesize and alignment check in FRMR RDMA/hns: Fix unmatch exception handling when init eq table fails RDMA/hns: Fix soft lockup under heavy CEQE load RDMA/hns: Check atomic wr length RDMA/ocrdma: Don't inline statistics functions RDMA/core: Introduce "name_assign_type" for an IB device RDMA/qib: Fix truncation compilation warnings in qib_verbs.c RDMA/qib: Fix truncation compilation warnings in qib_init.c RDMA/efa: Add EFA 0xefa3 PCI ID RDMA/mlx5: Support per-plane port IB counters by querying PPCNT register net/mlx5: mlx5_ifc update for accessing ppcnt register of plane ports RDMA/mlx5: Add plane index support when querying PTYS registers ...
2024-07-04RDMA/core: Introduce "name_assign_type" for an IB deviceMark Zhang
The name_assign_type indicates how the name is provided. Currently these types are supported: - RDMA_NAME_ASSIGN_TYPE_UNKNOWN: Unknown or not set; - RDMA_NAME_ASSIGN_TYPE_USER: Name is provided by the user; The user-created sub device, rxe and siw device has this type. When filling nl device info, it is set in the new attribute RDMA_NLDEV_ATTR_NAME_ASSIGN_TYPE. User-space tools like udev "rdma_rename" could check this attribute to determine if this device needs to be renamed or not. Signed-off-by: Mark Zhang <markzhang@nvidia.com> Link: https://lore.kernel.org/r/522591bef9a369cc8e5dcb77787e017bffee37fe.1719837610.git.leon@kernel.org Signed-off-by: Leon Romanovsky <leon@kernel.org>
2024-06-27RDMA: Pass entire uverbs attr bundle to create cq functionAkiva Goldberger
Changes the create_cq verb signature by sending the entire uverbs attr bundle as a parameter. This allows drivers to send driver specific attrs through ioctl for the create_cq verb and access them in their driver specific code. Also adds a new enum value for driver specific ioctl attributes for methods already supporting UHW. Link: https://lore.kernel.org/r/ed147343987c0d43fd391c1b2f85e2f425747387.1719512393.git.leon@kernel.org Signed-off-by: Akiva Goldberger <agoldberger@nvidia.com> Signed-off-by: Leon Romanovsky <leonro@nvidia.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
2024-06-26RDMA/rxe: Don't set BTH_ACK_MASK for UC or UD QPsHonggang LI
BTH_ACK_MASK bit is used to indicate that an acknowledge (for this packet) should be scheduled by the responder. Both UC and UD QPs are unacknowledged, so don't set BTH_ACK_MASK for UC or UD QPs. Fixes: 8700e3e7c485 ("Soft RoCE driver") Signed-off-by: Honggang LI <honggangli@163.com> Link: https://lore.kernel.org/r/20240624020348.494338-1-honggangli@163.com Reviewed-by: Zhu Yanjun <yanjun.zhu@linux.dev> Signed-off-by: Leon Romanovsky <leon@kernel.org> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
2024-06-09RDMA/rxe: Fix responder length checking for UD request packetsHonggang LI
According to the IBA specification: If a UD request packet is detected with an invalid length, the request shall be an invalid request and it shall be silently dropped by the responder. The responder then waits for a new request packet. commit 689c5421bfe0 ("RDMA/rxe: Fix incorrect responder length checking") defers responder length check for UD QPs in function `copy_data`. But it introduces a regression issue for UD QPs. When the packet size is too large to fit in the receive buffer. `copy_data` will return error code -EINVAL. Then `send_data_in` will return RESPST_ERR_MALFORMED_WQE. UD QP will transfer into ERROR state. Fixes: 689c5421bfe0 ("RDMA/rxe: Fix incorrect responder length checking") Signed-off-by: Honggang LI <honggangli@163.com> Link: https://lore.kernel.org/r/20240523094617.141148-1-honggangli@163.com Reviewed-by: Zhu Yanjun <yanjun.zhu@linux.dev> Signed-off-by: Leon Romanovsky <leon@kernel.org>
2024-05-30RDMA/rxe: Fix data copy for IB_SEND_INLINEHonggang LI
For RDMA Send and Write with IB_SEND_INLINE, the memory buffers specified in sge list will be placed inline in the Send Request. The data should be copied by CPU from the virtual addresses of corresponding sge list DMA addresses. Cc: stable@kernel.org Fixes: 8d7c7c0eeb74 ("RDMA: Add ib_virt_dma_to_page()") Signed-off-by: Honggang LI <honggangli@163.com> Link: https://lore.kernel.org/r/20240516095052.542767-1-honggangli@163.com Reviewed-by: Zhu Yanjun <yanjun.zhu@linux.dev> Reviewed-by: Li Zhijian <lizhijian@fujitsu.com> Reviewed-by: Jason Gunthorpe <jgg@nvidia.com> Signed-off-by: Leon Romanovsky <leon@kernel.org>
2024-05-18Merge tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rdma/rdmaLinus Torvalds
Pull rdma updates from Jason Gunthorpe: "Aside from the usual things this has an arch update for __iowrite64_copy() used by the RDMA drivers. This API was intended to generate large 64 byte MemWr TLPs on PCI. These days most processors had done this by just repeating writel() in a loop. S390 and some new ARM64 designs require a special helper to get this to generate. - Small improvements and fixes for erdma, efa, hfi1, bnxt_re - Fix a UAF crash after module unload on leaking restrack entry - Continue adding full RDMA support in mana with support for EQs, GID's and CQs - Improvements to the mkey cache in mlx5 - DSCP traffic class support in hns and several bug fixes - Cap the maximum number of MADs in the receive queue to avoid OOM - Another batch of rxe bug fixes from large scale testing - __iowrite64_copy() optimizations for write combining MMIO memory - Remove NULL checks before dev_put/hold() - EFA support for receive with immediate - Fix a recent memleaking regression in a cma error path" * tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rdma/rdma: (70 commits) RDMA/cma: Fix kmemleak in rdma_core observed during blktests nvme/rdma use siw RDMA/IPoIB: Fix format truncation compilation errors bnxt_re: avoid shift undefined behavior in bnxt_qplib_alloc_init_hwq RDMA/efa: Support QP with unsolicited write w/ imm. receive IB/hfi1: Remove generic .ndo_get_stats64 IB/hfi1: Do not use custom stat allocator RDMA/hfi1: Use RMW accessors for changing LNKCTL2 RDMA/mana_ib: implement uapi for creation of rnic cq RDMA/mana_ib: boundary check before installing cq callbacks RDMA/mana_ib: introduce a helper to remove cq callbacks RDMA/mana_ib: create and destroy RNIC cqs RDMA/mana_ib: create EQs for RNIC CQs RDMA/core: Remove NULL check before dev_{put, hold} RDMA/ipoib: Remove NULL check before dev_{put, hold} RDMA/mlx5: Remove NULL check before dev_{put, hold} RDMA/mlx5: Track DCT, DCI and REG_UMR QPs as diver_detail resources. RDMA/core: Add an option to display driver-specific QPs in the rdmatool RDMA/efa: Add shutdown notifier RDMA/mana_ib: Fix missing ret value IB/mlx5: Use __iowrite64_copy() for write combining stores ...
2024-04-22RDMA/rxe: Let destroy qp succeed with stuck packetBob Pearson
In some situations a sent packet may get queued in the NIC longer than than timeout of a ULP. Currently if this happens the ULP may try to reset the link by destroying the qp and setting up an alternate connection but will fail because the rxe driver is waiting for the packet to finish getting sent and be returned to the skb destructor function where the qp reference holding things up will be dropped. This patch modifies the way that the qp is passed to the destructor to pass the qp index and not a qp pointer. Then the destructor will attempt to lookup the qp from its index and if it fails exit early. This requires taking a reference on the struct sock rather than the qp allowing the qp to be destroyed while the sk is still around waiting for the packet to finish. Link: https://lore.kernel.org/r/20240329145513.35381-15-rpearsonhpe@gmail.com Signed-off-by: Bob Pearson <rpearsonhpe@gmail.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
2024-04-22RDMA/rxe: Get rid of pkt resend on errBob Pearson
Currently the rxe_driver detects packet drops by ip_local_out() which occur before the packet is sent on the wire and attempts to resend them. This is redundant with the usual retry mechanism which covers packets that get dropped in transit to or from the remote node. The way this is implemented is not robust since it sets need_req_skb and waits for the number of local skbs outstanding for this qp to drop below a low water mark. This is racy since the skb may be sent to the destructor before the requester can set the need_req_skb flag. This will cause a deadlock in the send path for that qp. This patch removes this mechanism since the normal retry path will correct the error and resend the packet and it makes no difference if the packet is dropped locally or later. Link: https://lore.kernel.org/r/20240329145513.35381-14-rpearsonhpe@gmail.com Signed-off-by: Bob Pearson <rpearsonhpe@gmail.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
2024-04-22RDMA/rxe: Make rxe_loopback match rxe_send behaviorBob Pearson
The rxe send path currently counts the number of skbs outstanding between the rxe driver and the ethernet driver to prevent too many packets to accumulate waiting to send. This patch makes the local loopback path behave the same way. The loopback path forwards the packets to the receive path which will eventually call kfree_skb on all packets and drop the qp references. This makes the loopback path more useful for software testing. Link: https://lore.kernel.org/r/20240329145513.35381-13-rpearsonhpe@gmail.com Signed-off-by: Bob Pearson <rpearsonhpe@gmail.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
2024-04-22RDMA/rxe: Fix incorrect rxe_put in error pathBob Pearson
In rxe_send() a ref is taken on the qp to keep it alive until the kfree_skb() has a chance to call the skb destructor rxe_skb_tx_dtor() which drops the reference. If the packet has an incorrect protocol the error path just calls kfree_skb() which will call the destructor which will drop the ref. Currently the driver also calls rxe_put() which is incorrect. Additionally since the packets sent to rxe_send() are under the control of the driver and it only ever produces IPV4 or IPV6 packets the simplest fix is to remove all the code in this block. Link: https://lore.kernel.org/r/20240329145513.35381-12-rpearsonhpe@gmail.com Signed-off-by: Bob Pearson <rpearsonhpe@gmail.com> Fixes: 9eb7f8e44d13 ("IB/rxe: Move refcounting earlier in rxe_send()") Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
2024-04-22RDMA/rxe: Don't call direct between tasksBob Pearson
Replace calls to rxe_run_task() with rxe_sched_task(). This prevents the tasks from all running on the same cpu. This change slightly reduces performance for single qp send and write benchmarks in loopback mode but greatly improves the performance with multiple qps because if run task is used all the work tends to be performed on one cpu. For actual on the wire benchmarks there is no noticeable performance change. Link: https://lore.kernel.org/r/20240329145513.35381-11-rpearsonhpe@gmail.com Signed-off-by: Bob Pearson <rpearsonhpe@gmail.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
2024-04-22RDMA/rxe: Don't call rxe_requester from rxe_completerBob Pearson
Instead of rescheduling rxe_requester from rxe_completer() just extend the duration of rxe_sender() by one pass. Setting run_requester_again forces rxe_completer() to return 0 which will cause rxe_sender() to be called at least one more time. Link: https://lore.kernel.org/r/20240329145513.35381-10-rpearsonhpe@gmail.com Signed-off-by: Bob Pearson <rpearsonhpe@gmail.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
2024-04-22RDMA/rxe: Don't schedule rxe_completer from rxe_requesterBob Pearson
Now that rxe_completer() is always called serially after rxe_requester() there is no reason to schedule rxe_completer() from rxe_requester(). Link: https://lore.kernel.org/r/20240329145513.35381-9-rpearsonhpe@gmail.com Signed-off-by: Bob Pearson <rpearsonhpe@gmail.com> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
generated by cgit (git 2.34.1) at 2025-07-25 18:27:44 +0000