- 04 Apr, 2023 2 commits
-
-
Selvin Xavier authored
bnxt_en driver does the queue mapping for RoCE traffic. Removing the queue mapping from RoCE driver. Signed-off-by:
Selvin Xavier <selvin.xavier@broadcom.com> Link: https://lore.kernel.org/r/1680169540-10029-3-git-send-email-selvin.xavier@broadcom.comSigned-off-by:
Leon Romanovsky <leon@kernel.org>
-
Selvin Xavier authored
Updating the HW structures to the latest version. This is copied from the code maintained internally. No functionality changes in this patch. Code is re-organized to match the file maintained in the internal tree. Also, New HW interface structures are added, which will be used by the drivers in future. CC: Michael Chan <michael.chan@broadcom.com> Signed-off-by:
-
- 03 Apr, 2023 7 commits
-
-
Tetsuo Handa authored
syzbot is reporting that siw_netdev_event(NETDEV_UNREGISTER) cannot destroy siw_device created after unshare(CLONE_NEWNET) due to net namespace check. It seems that this check was by error there and should be removed. Reported-by:
syzbot <syzbot+5e70d01ee8985ae62a3b@syzkaller.appspotmail.com> Link: https://syzkaller.appspot.com/bug?extid=5e70d01ee8985ae62a3bSuggested-by:
Jason Gunthorpe <jgg@ziepe.ca> Suggested-by:
Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp> Link: https://lore.kernel.org/r/a44e9ac5-44e2-d575-9e30-02483cc7ffd1@I-love.SAKURA.ne.jpReviewed-by:
Bernard Metzler <bmt@zurich.ibm.com> Signed-off-by:
-
Yang Li authored
The call netdev_{put, hold} of dev_{put, hold} will check NULL, so there is no need to check before using dev_{put, hold}, remove it to silence the warnings: ./drivers/infiniband/core/cma.c:713:2-9: WARNING: NULL check before dev_{put, hold} functions is not needed. ./drivers/infiniband/core/cma.c:2433:2-9: WARNING: NULL check before dev_{put, hold} functions is not needed. Reported-by:Abaci Robot <abaci@linux.alibaba.com> Link: https://bugzilla.openanolis.cn/show_bug.cgi?id=4668Signed-off-by:
Yang Li <yang.lee@linux.alibaba.com> Link: https://lore.kernel.org/r/20230331010633.63261-1-yang.lee@linux.alibaba.comSigned-off-by:
-
Tom Rix authored
clang with W=1 reports drivers/infiniband/hw/qib/qib_file_ops.c:487:20: error: variable 'cnt' set but not used [-Werror,-Wunused-but-set-variable] u32 tid, ctxttid, cnt, limit, tidcnt; ^ drivers/infiniband/hw/qib/qib_file_ops.c:1771:9: error: variable 'cnt' set but not used [-Werror,-Wunused-but-set-variable] int i, cnt = 0, maxtid = ctxt_tidbase + dd->rcvtidcnt; ^ This variable is not used so remove it. Signed-off-by:Tom Rix <trix@redhat.com> Link: https://lore.kernel.org/r/20230330235800.1845815-1-trix@redhat.comSigned-off-by:
-
Tom Rix authored
clang with W=1 reports drivers/infiniband/hw/mlx5/devx.c:1996:6: error: variable 'num_alloc_xa_entries' set but not used [-Werror,-Wunused-but-set-variable] int num_alloc_xa_entries = 0; ^ This variable is not used so remove it. Signed-off-by: -
Max Gurtovoy authored
This commit doesn't change any logic. Reviewed-by:
Sergey Gorenko <sergeygo@nvidia.com> Signed-off-by:
Max Gurtovoy <mgurtovoy@nvidia.com> Link: https://lore.kernel.org/r/20230330131333.37900-3-mgurtovoy@nvidia.comReviewed-by:
Sagi Grimberg <sagi@grimberg.me> Signed-off-by:
-
Max Gurtovoy authored
Move this common logic into iser_create_send_desc instead of duplicating the code. Reviewed-by:
-
Max Gurtovoy authored
The removed macros are old leftovers. Reviewed-by:
-
- 30 Mar, 2023 1 commit
-
-
Leon Romanovsky authored
There is no need to print any debug messages after failure to allocate memory, because kernel will print OOM dumps anyway. Together with removal of these messages, remove useless goto jumps. Fixes: 5bf944f2 ("RDMA/rxe: Add error messages") Reported-by:
Dan Carpenter <error27@gmail.com> Link: https://lore.kernel.org/all/ea43486f-43dd-4054-b1d5-3a0d202be621@kili.mountain Link: https://lore.kernel.org/r/d3cedf723b84e73e8062a67b7489d33802bafba2.1680113597.git.leon@kernel.orgReviewed-by:
Bob Pearson <rpearsonhpe@gmail.com> Signed-off-by:
Leon Romanovsky <leonro@nvidia.com>
-
- 29 Mar, 2023 8 commits
-
-
Bob Pearson authored
Remove the tasklet call in rxe_cq.c and also the is_dying in the cq struct. There is no reason for the rxe driver to defer the call to the cq completion handler by scheduling a tasklet. rxe_cq_post() is not called in a hard irq context. The rxe driver currently is incorrect because the tasklet call is made without protecting the cq pointer with a reference from having the underlying memory freed before the deferred routine is called. Executing the comp_handler inline fixes this problem. Fixes: 8700e3e7 ("Soft RoCE driver") Signed-off-by:
Zhu Yanjun <zyjzyj2000@gmail.com> Signed-off-by:
-
Tom Rix authored
clang with W=1 reports drivers/infiniband/hw/ocrdma/ocrdma_verbs.c:1592:6: error: variable 'discard_cnt' set but not used [-Werror,-Wunused-but-set-variable] int discard_cnt = 0; ^ This variable is not used so remove it. Signed-off-by: -
Tom Rix authored
clang with W=1 reports drivers/infiniband/hw/bnxt_re/qplib_fp.c:303:6: error: variable 'num_srqne_processed' set but not used [-Werror,-Wunused-but-set-variable] int num_srqne_processed = 0; ^ drivers/infiniband/hw/bnxt_re/qplib_fp.c:304:6: error: variable 'num_cqne_processed' set but not used [-Werror,-Wunused-but-set-variable] int num_cqne_processed = 0; ^ These variables are not used so remove them. Signed-off-by: -
Cai Huoqing authored
Remove pci_clear_master to simplify the code, the bus-mastering is also cleared in do_pci_disable_device, like this: ./drivers/pci/pci.c:2197 static void do_pci_disable_device(struct pci_dev *dev) { u16 pci_command; pci_read_config_word(dev, PCI_COMMAND, &pci_command); if (pci_command & PCI_COMMAND_MASTER) { pci_command &= ~PCI_COMMAND_MASTER; pci_write_config_word(dev, PCI_COMMAND, pci_command); } pcibios_disable_device(dev); }. And dev->is_busmaster is set to 0 in pci_disable_device. Signed-off-by:Cai Huoqing <cai.huoqing@linux.dev> Link: https://lore.kernel.org/r/20230323115742.13836-1-cai.huoqing@linux.devSigned-off-by:
-
Patrisious Haddad authored
Previously for switchdev only per device counters were supported. Currently we allocate counters for switchdev per port, which also includes the ports that belong to VF representors in order to expose them to users through the rdma tool, allowing the host to track the VFs statistics through their representors counters. Signed-off-by:
Patrisious Haddad <phaddad@nvidia.com> Link: https://lore.kernel.org/r/ea31e1103c125cd27931ba213f307cde30d2eaed.1679566038.git.leon@kernel.orgSigned-off-by:
-
Leon Romanovsky authored
* mlx5-next: net/mlx5: Introduce other vport query for Q-counters
-
Patrisious Haddad authored
These new fields in QUERY_Q_COUNTER command allow us to access another vport counters during the query command, which is specially useful to query representor vports. In addition also add the required caps to check if this capability is actually supported. Signed-off-by:
Michael Guralnik <michaelgur@nvidia.com> Link: https://lore.kernel.org/r/75c73a4a0e60f18c37b35a4a11ca2e2415e4a6f3.1679566038.git.leon@kernel.orgSigned-off-by:
-
Selvin Xavier authored
Add resize_cq verb support for user space CQs. Resize operation for kernel CQs are not supported now. Driver should free the current CQ only after user library polls for all the completions and switch to new CQ. So after the resize_cq is returned from the driver, user library polls for existing completions and store it as temporary data. Once library reaps all completions in the current CQ, it invokes the ibv_cmd_poll_cq to inform the driver about the resize_cq completion. Adding a check for user CQs in driver's poll_cq and complete the resize operation for user CQs. Updating uverbs_cmd_mask with poll_cq to support this. Signed-off-by:
-
- 24 Mar, 2023 12 commits
-
-
Cheng Xu authored
Hardware's page size is 4096, but the kernel's page size may vary. Driver should use hardware's page size when communicating with hardware. Fixes: 15505577 ("RDMA/erdma: Add verbs implementation") Link: https://lore.kernel.org/r/20230307102924.70577-2-chengyou@linux.alibaba.comSigned-off-by:
Cheng Xu <chengyou@linux.alibaba.com> Signed-off-by:
Jason Gunthorpe <jgg@nvidia.com>
-
Bob Pearson authored
This patch is a major rewrite of the tasklet routines in rxe_task.c. The main motivation for this is the realization that the code violates the safety of the qp pointer by correct reference counting. When a tasklet is scheduled from a verbs API the calling thread has a valid reference to the qp and schedules the tasklet to run at a later time carrying a pointer to the qp. Once the calling code returns however the qp can be destroyed at any time. In order to correct this a reference to the qp must be taken when the task is scheduled and held until it finishes running. This is complicated by the tasklet library not alwys running a task that is scheduled depending on whether someone else has scheduled it. This patch moves the logic for deciding whether to run or schedule a task outside of do_task() and guarantees that there is only one copy of the task scheduled or running at a time. Secondly the separate flags controlling teardown and draining of the task are included in the task state machine and all references to the state are protected by spinlocks to avoid consistency and memory barrier issues. Link: https://lore.kernel.org/r/20230304174533.11296-9-rpearsonhpe@gmail.comSigned-off-by:
-
Bob Pearson authored
Replace rxe_run_task() by rxe_sched_task() when tasks call each other. These are not performance critical and mainly involve error paths but they run the risk of causing deadlocks. Link: https://lore.kernel.org/r/20230304174533.11296-8-rpearsonhpe@gmail.comSigned-off-by:
Ian Ziemba <ian.ziemba@hpe.com> Signed-off-by:
-
Bob Pearson authored
The subroutine __rxe_do_task is not thread safe and it has no way to guarantee that the tasks, which are designed with the assumption that they are non-reentrant, are not reentered. All of its uses are non-performance critical. This patch replaces calls to __rxe_do_task with calls to rxe_sched_task. It also removes irrelevant or unneeded if tests. Instead of calling the task machinery a single call to the tasklet function (rxe_requester, etc.) is sufficient to draing the queues if task execution has been disabled or stopped. Together these changes allow the removal of __rxe_do_task. Link: https://lore.kernel.org/r/20230304174533.11296-7-rpearsonhpe@gmail.comSigned-off-by:
-
Bob Pearson authored
Currently each of the three tasklets requester, completer and responder in the rxe driver take and release a reference to the qp argument at the beginning and end of the subroutines. The caller passing in the qp argument should be responsible for holding a reference to qp so these are not required. Further doing so breaks the qp cleanup code in rxe_qp_do_cleanup which calls these routines after all the references have been dropped so they cannot drain the packet and work request queues as intended. In fact if these routines are deferred by calling tasklet_schedule there is no guarantee that the calling code does have a qp reference. That is a bug in rxe_task.c which will be fixed later in this series. Link: https://lore.kernel.org/r/20230304174533.11296-6-rpearsonhpe@gmail.comSigned-off-by:
-
Bob Pearson authored
Cleanup the handling of qp in the error state, reset state and during rxe_qp_do_cleanup. Make the same as rxe_resp.c Link: https://lore.kernel.org/r/20230304174533.11296-5-rpearsonhpe@gmail.comSigned-off-by:
-
Bob Pearson authored
Cleanup the handling of qp in the error state, reset state and during rxe_qp_do_cleanup. The error state does about the same thing as the others but has code spread all over. This patch combines them in a cleaner way. Link: https://lore.kernel.org/r/20230304174533.11296-4-rpearsonhpe@gmail.comSigned-off-by:
-
Bob Pearson authored
Originally is was thought that the tasklet machinery in rxe_task.c would be used in other applications but that has not happened for years. This patch replaces the 'void *arg' by struct 'rxe_qp *qp' in the parameters to the tasklet calls. This change will have no affect on performance but may make the code a little clearer. Link: https://lore.kernel.org/r/20230304174533.11296-2-rpearsonhpe@gmail.comSigned-off-by:
-
Bob Pearson authored
This patch adds error and debug messages so that every interaction with rdma-core through a verbs API call or a completion error return will generate at least one error message backed up by debug messages with more detail. With dynamic debugging one can follow up after seeing an error message by turning on the appropriate debug messages. Link: https://lore.kernel.org/r/20230303221623.8053-5-rpearsonhpe@gmail.comSigned-off-by:
-
Bob Pearson authored
Extend the dbg log messages (e.g. rxe_dbg_xxx) to include err and info types. rxe.c is modified to use these new log messages as examples. Link: https://lore.kernel.org/r/20230303221623.8053-4-rpearsonhpe@gmail.comSigned-off-by:
-
Bob Pearson authored
Replace the name rxe_dbg with rxe_dbg_dev which better matches the remaining rxe_dbg_xxx macros for debug messages with a rxe device parameter. Reuse the name rxe_dbg for debug messages which do not have a rxe device parameter. Link: https://lore.kernel.org/r/20230303221623.8053-3-rpearsonhpe@gmail.comSigned-off-by:
-
Bob Pearson authored
'exists' looks like a boolean. This patch replaces it by the normal name used for the rxe device, 'rxe', which should be a little less confusing. The second rxe_dbg() message is incorrect since rxe is known to be NULL and this will cause a seg fault if this message were ever sent. Replace it by pr_debug for the moment. Fixes: c6aba5ea ("RDMA/rxe: Replace pr_xxx by rxe_dbg_xxx in rxe.c") Link: https://lore.kernel.org/r/20230303221623.8053-2-rpearsonhpe@gmail.comSigned-off-by:
-
- 23 Mar, 2023 5 commits
-
-
Gustavo A. R. Silva authored
GCC-13 (and Clang)[1] does not like to access a partially allocated object, since it cannot reason about it for bounds checking. In this case 140 bytes are allocated for an object of type struct ib_umad_packet: packet = kzalloc(sizeof(*packet) + IB_MGMT_RMPP_HDR, GFP_KERNEL); However, notice that sizeof(*packet) is only 104 bytes: struct ib_umad_packet { struct ib_mad_send_buf * msg; /* 0 8 */ struct ib_mad_recv_wc * recv_wc; /* 8 8 */ struct list_head list; /* 16 16 */ int length; /* 32 4 */ /* XXX 4 bytes hole, try to pack */ struct ib_user_mad mad __attribute__((__aligned__(8))); /* 40 64 */ /* size: 104, cachelines: 2, members: 5 */ /* sum members: 100, holes: 1, sum holes: 4 */ /* forced alignments: 1, forced holes: 1, sum forced holes: 4 */ /* last cacheline: 40 bytes */ } __attribute__((__aligned__(8))); and 36 bytes extra bytes are allocated for a flexible-array member in struct ib_user_mad: include/rdma/ib_mad.h: 120 enum { ... 123 IB_MGMT_RMPP_HDR = 36, ... } struct ib_user_mad { struct ib_user_mad_hdr hdr; /* 0 64 */ /* --- cacheline 1 boundary (64 bytes) --- */ __u64 data[] __attribute__((__aligned__(8))); /* 64 0 */ /* size: 64, cachelines: 1, members: 2 */ /* forced alignments: 1 */ } __attribute__((__aligned__(8))); So we have sizeof(*packet) + IB_MGMT_RMPP_HDR == 140 bytes Then the address of the flex-array member (for which only 36 bytes were allocated) is casted and copied into a pointer to struct ib_rmpp_mad, which, in turn, is of size 256 bytes: rmpp_mad = (struct ib_rmpp_mad *) packet->mad.data; struct ib_rmpp_mad { struct ib_mad_hdr mad_hdr; /* 0 24 */ struct ib_rmpp_hdr rmpp_hdr; /* 24 12 */ u8 data[220]; /* 36 220 */ /* size: 256, cachelines: 4, members: 3 */ }; The thing is that those 36 bytes allocated for flex-array member data in struct ib_user_mad onlly account for the size of both struct ib_mad_hdr and struct ib_rmpp_hdr, but nothing is left for array u8 data[220]. So, the compiler is legitimately complaining about accessing an object for which not enough memory was allocated. Apparently, the only members of struct ib_rmpp_mad that are relevant (that are actually being used) in function ib_umad_write() are mad_hdr and rmpp_hdr. So, instead of casting packet->mad.data to (struct ib_rmpp_mad *) create a new structure struct ib_rmpp_mad_hdr { struct ib_mad_hdr mad_hdr; struct ib_rmpp_hdr rmpp_hdr; } __packed; and cast packet->mad.data to (struct ib_rmpp_mad_hdr *). Notice that IB_MGMT_RMPP_HDR == sizeof(struct ib_rmpp_mad_hdr) == 36 bytes Refactor the rest of the code, accordingly. Fix the following warnings seen under GCC-13 and -Warray-bounds: drivers/infiniband/core/user_mad.c:564:50: warning: array subscript ‘struct ib_rmpp_mad[0]’ is partly outside array bounds of ‘unsigned char[140]’ [-Warray-bounds=] drivers/infiniband/core/user_mad.c:566:42: warning: array subscript ‘struct ib_rmpp_mad[0]’ is partly outside array bounds of ‘unsigned char[140]’ [-Warray-bounds=] drivers/infiniband/core/user_mad.c:618:25: warning: array subscript ‘struct ib_rmpp_mad[0]’ is partly outside array bounds of ‘unsigned char[140]’ [-Warray-bounds=] drivers/infiniband/core/user_mad.c:622:44: warning: array subscript ‘struct ib_rmpp_mad[0]’ is partly outside array bounds of ‘unsigned char[140]’ [-Warray-bounds=] Link: https://github.com/KSPP/linux/issues/273 Link: https://godbolt.org/z/oYWaGM4Yb [1] Signed-off-by:Gustavo A. R. Silva <gustavoars@kernel.org> Link: https://lore.kernel.org/r/ZBpB91qQcB10m3Fw@workSigned-off-by:
-
Leon Romanovsky authored
This series from Or changes default of IB out-of-order feature and allows to the RDMA users to decide if they need to wait for completion for all segments or it is enough to wait for last segment completion only. Thanks Signed-off-by: -
Or Har-Toov authored
When FW supports ooo by default, enable the cap. Signed-off-by:
Or Har-Toov <ohartoov@nvidia.com> Signed-off-by:
-
Or Har-Toov authored
Set retry_mode to GO_BACK_N when qp is created with INTEGRITY_EN flag because out-of-order is not supported when doing HW offload of signature operations. Signed-off-by:
-
Or Har-Toov authored
Add needed HW bits for enabling out-of-order by default and use go_back_n when out-of-order is not needed. Signed-off-by:
-
- 22 Mar, 2023 4 commits
-
-
Yonatan Nachum authored
Add feature bit to existing device caps field. EFA supports data polling of 128 bytes blocks. The flag indicates that the NIC guarentees that a 128 byte aligned block is written in order, ie that observing the last 8 bits of the block mean the prior 127 bytes are also written. It is useful for "last data polling" acceleration techniques. Link: https://lore.kernel.org/r/20230219081328.10419-1-mrgolin@amazon.comReviewed-by:
Yehuda Yitschak <yehuday@amazon.com> Reviewed-by:
Yossi Leybovich <sleybo@amazon.com> Signed-off-by:
Yonatan Nachum <ynachum@amazon.com> Signed-off-by:
Michael Margolin <mrgolin@amazon.com> Acked-by:
Gal Pressman <gal.pressman@linux.dev> Signed-off-by:
-
Cheng Xu authored
After necessary configuration, driver should wait hardware finishing initialization. The wait sets at CMDQ related function though it has nothing to do with CMDQ. Refactor this part to make code cleaner. Signed-off-by:
-
Cheng Xu authored
Using void * to define EQ doorbell pointer can eliminate unnecessary casting when performing assignment. Also rename *db_addr* to *db* for a shorter name. Signed-off-by:
-
Cheng Xu authored
Replace __be32_to_cpu/__cpu_to_be16 with be32_to_cpu/cpu_to_be16. And use be32_to_cpu_array to copy and swap byte order to hide the loop. Signed-off-by:
-
- 19 Mar, 2023 1 commit
-
-
Linus Torvalds authored
-
