- 08 Apr, 2021 8 commits
-
-
Wenpeng Liang authored
The hardware supports only two types of abnormal interrupts. Fixes: a5073d60 ("RDMA/hns: Add eq support of hip08") Link: https://lore.kernel.org/r/1617354454-47840-5-git-send-email-liweihang@huawei.comSigned-off-by:
Wenpeng Liang <liangwenpeng@huawei.com> Signed-off-by:
Weihang Li <liweihang@huawei.com> Signed-off-by:
Jason Gunthorpe <jgg@nvidia.com>
-
Yangyang Li authored
The register value related to the eq interrupt depends only on enable_flag, so the redundant condition judgment is deleted. Fixes: a5073d60 ("RDMA/hns: Add eq support of hip08") Link: https://lore.kernel.org/r/1617354454-47840-4-git-send-email-liweihang@huawei.comSigned-off-by:
Yangyang Li <liyangyang20@huawei.com> Signed-off-by:
-
Weihang Li authored
When querying QP, the ULPs should be informed of the max length of inline data supported by the hardware. Fixes: 30b70788 ("RDMA/hns: Support inline data in extented sge space for RC") Link: https://lore.kernel.org/r/1617354454-47840-3-git-send-email-liweihang@huawei.comSigned-off-by:
-
Weihang Li authored
RQ inline is not supported on UD/GSI QP, it should be disabled in QPC. Link: https://lore.kernel.org/r/1617354454-47840-2-git-send-email-liweihang@huawei.comSigned-off-by:
-
Wenpeng Liang authored
Use ratelimited print in mbox and cmq. And print mailbox operation if mailbox fails because it's useful information for the user. Link: https://lore.kernel.org/r/1617262341-37571-4-git-send-email-liweihang@huawei.comSigned-off-by:
Lang Cheng <chenglang@huawei.com> Signed-off-by:
-
Lang Cheng authored
Add error code definition according to the return code from firmware to help find out more detailed reasons why a command fails to be sent. Link: https://lore.kernel.org/r/1617262341-37571-3-git-send-email-liweihang@huawei.comSigned-off-by:
-
Lang Cheng authored
Fix error of cmd's context number calculation algorithm to enable all of 32 cmd entries and support 32 concurrent accesses. Link: https://lore.kernel.org/r/1617262341-37571-2-git-send-email-liweihang@huawei.comSigned-off-by:
-
Bob Pearson authored
All responder errors from request packets that do not consume a receive WQE fail to generate acks for RC QPs. This patch corrects this behavior by making the flow follow the same path as request packets that do consume a WQE after the completion. Link: https://lore.kernel.org/r/20210402001016.3210-1-rpearson@hpe.com Link: https://lore.kernel.org/linux-rdma/1a7286ac-bcea-40fb-2267-480134dd301b@gmail.com/Signed-off-by:
Bob Pearson <rpearson@hpe.com> Signed-off-by:
-
- 07 Apr, 2021 11 commits
-
-
Yixian Liu authored
Local invalidate is also a kind of memory management operation, not only memory bind operation. Furthermore, as invalidate operations include local and remote, add prefix to the prompt message to make it clearer. Link: https://lore.kernel.org/r/1617698772-13871-1-git-send-email-liweihang@huawei.comSigned-off-by:
Yixian Liu <liuyixian@huawei.com> Signed-off-by:
-
Wei Yongjun authored
A spin lock is taken here so we should use GFP_ATOMIC. Fixes: f91696f2 ("RDMA/hns: Support congestion control type selection according to the FW") Link: https://lore.kernel.org/r/20210407154900.3486268-1-weiyongjun1@huawei.comReported-by:
Hulk Robot <hulkci@huawei.com> Signed-off-by:
Wei Yongjun <weiyongjun1@huawei.com> Signed-off-by:
-
Praveen Kumar Kannoju authored
To update xlt (during mlx5_ib_reg_user_mr()), the driver can request up to 1 MB (order-8) memory, depending on the size of the MR. This costly allocation can sometimes take very long to return (a few seconds). This causes the calling application to hang for a long time, especially when the system is fragmented. To avoid these long latency spikes, the calls the higher order allocations need to fail faster in case they are not available. In order to acheive this we need __GFP_NORETRY flag in the gfp_mask before during fetching the free pages. Allow the algorithm to automatically fall back to smaller page sizes. Link: https://lore.kernel.org/r/1617425635-35631-1-git-send-email-praveen.kannoju@oracle.comSigned-off-by:
Praveen Kumar Kannoju <praveen.kannoju@oracle.com> Acked-by:
Leon Romanovsky <leonro@nvidia.com> Signed-off-by:
-
Kaike Wan authored
Remove the unused function sdma_iowait_schedule(). Fixes: 77241056 ("IB/hfi1: add driver files") Link: https://lore.kernel.org/r/1617026791-89997-1-git-send-email-dennis.dalessandro@cornelisnetworks.comReviewed-by:
Mike Marciniszyn <mike.marciniszyn@cornelisnetworks.com> Signed-off-by:
Kaike Wan <kaike.wan@intel.com> Signed-off-by:
Dennis Dalessandro <dennis.dalessandro@cornelisnetworks.com> Signed-off-by:
-
Mike Marciniszyn authored
The code currently assumes that the mmu_notifier struct embedded in mmu_rb_handler only contains two fields. There are now extra fields: struct mmu_notifier { struct hlist_node hlist; const struct mmu_notifier_ops *ops; struct mm_struct *mm; struct rcu_head rcu; unsigned int users; }; Given that there in no init for the mmu_notifier, a kzalloc() should be used to insure that any newly added fields are given a predictable initial value of zero. Fixes: 06e0ffa6 ("IB/hfi1: Re-factor MMU notification code") Link: https://lore.kernel.org/r/1617026056-50483-9-git-send-email-dennis.dalessandro@cornelisnetworks.comReviewed-by:Adam Goldman <adam.goldman@intel.com> Signed-off-by:
-
Mike Marciniszyn authored
Add traces that were vital in isolating an issue with pq waitlist in commit fa8dac39 ("IB/hfi1: Fix another case where pq is left on waitlist") Link: https://lore.kernel.org/r/1617026056-50483-8-git-send-email-dennis.dalessandro@cornelisnetworks.comReviewed-by:
-
Mike Marciniszyn authored
hfi1_ipoib_send() directly calls hfi1_ipoib_send_dma() with no value add. Fix by renaming hfi1_ipoib_send_dma() to hfi1_ipoib_send(). Link: https://lore.kernel.org/r/1617026056-50483-6-git-send-email-dennis.dalessandro@cornelisnetworks.comReviewed-by:
-
Mike Marciniszyn authored
The call is from an ISR context and napi_schedule_irqoff() can be used. Change the call to the more efficient type. Link: https://lore.kernel.org/r/1617026056-50483-5-git-send-email-dennis.dalessandro@cornelisnetworks.comReviewed-by:
-
Mike Marciniszyn authored
The completion ring for tx is using the wrong size to size the ring, oversizing the ring by two orders of magniture. Correct the allocation size and use kcalloc_node() to allocate the ring. Fix mistaken GFP defines in similar allocations. Link: https://lore.kernel.org/r/1617026056-50483-4-git-send-email-dennis.dalessandro@cornelisnetworks.comReviewed-by:
-
Mike Marciniszyn authored
The current rdma_netdev handling in ipoib hooks the tx_timeout handler, but prints out a totally useless message that prevents effective debugging especially when multiple transmit queues are being used. Add a tx_timeout rdma_netdev hook and implement the callback in the hfi1 to print additional information. The existing non-helpful message is avoided when the driver has presented a callback. Link: https://lore.kernel.org/r/1617026056-50483-3-git-send-email-dennis.dalessandro@cornelisnetworks.comReviewed-by:
-
Mike Marciniszyn authored
Add traces to allow for debugging issues with AIP tx. Link: https://lore.kernel.org/r/1617026056-50483-2-git-send-email-dennis.dalessandro@cornelisnetworks.comReviewed-by:
-
- 01 Apr, 2021 21 commits
-
-
Yixian Liu authored
The doorbell update interfaces are very similar for different queues, such as SQ, RQ, SRQ, CQ and EQ. So reorganize these code and also fix some inappropriate naming. Link: https://lore.kernel.org/r/1616840738-7866-3-git-send-email-liweihang@huawei.comSigned-off-by:
-
Yixian Liu authored
HIP08 supports both normal and record doorbell mode for RQ and CQ, SQ record doorbell for userspace is also supported by the software for flushing CQE process. As now the capability of HIP08 are exposed to the user and are configurable, the support of normal doorbell should be added back. Note that, if switching to normal doorbell, the kernel will report "flush cqe is unsupported" if modify qp to error status as the flush is based on record doorbell. Link: https://lore.kernel.org/r/1616840738-7866-2-git-send-email-liweihang@huawei.comSigned-off-by:
-
Xi Wang authored
Use hr_reg_write() instead of roce_set_field() to simplify codes about configuring HEM BA. Link: https://lore.kernel.org/r/1616815294-13434-6-git-send-email-liweihang@huawei.comSigned-off-by:
Xi Wang <wangxi11@huawei.com> Signed-off-by:
-
Xi Wang authored
Encapsulate configuring GMV base address and other type of HEM table into two separate functions to make process of setting HEM clearer. Link: https://lore.kernel.org/r/1616815294-13434-5-git-send-email-liweihang@huawei.comSigned-off-by:
-
Xi Wang authored
The 'HNS_ROCE_OPC_QUERY_MB_ST' command will response the mailbox complete status and hardware busy flag, and the complete status is only valid when the busy flag is 0, so it's better to query these two fields at a time rather than separately. Link: https://lore.kernel.org/r/1616815294-13434-4-git-send-email-liweihang@huawei.comSigned-off-by:
-
Yixing Liu authored
Encapsulate two subprocesses into functions. Link: https://lore.kernel.org/r/1616815294-13434-3-git-send-email-liweihang@huawei.comSigned-off-by:
Yixing Liu <liuyixing1@huawei.com> Signed-off-by:
-
Weihang Li authored
Encapsulate the process of obtaining the current QP and filling WC as functions, also merge some duplicate code. Link: https://lore.kernel.org/r/1616815294-13434-2-git-send-email-liweihang@huawei.comSigned-off-by:
-
Weihang Li authored
Wei Hu(Xavier) has left Hisilicon and his email address is invalid now. I'd be glad to add him back with another address if he wants to continue maintain this module. Link: https://lore.kernel.org/r/1617007584-39842-1-git-send-email-liweihang@huawei.comSigned-off-by:
-
Jack Wang authored
Max io size is limited by both remote buffer size and the max fr pages per mr. Link: https://lore.kernel.org/r/20210325153308.1214057-20-gi-oh.kim@ionos.comSigned-off-by:
Jack Wang <jinpu.wang@ionos.com> Reviewed-by:
Md Haris Iqbal <haris.iqbal@ionos.com> Signed-off-by:
Gioh Kim <gi-oh.kim@ionos.com> Signed-off-by:
-
Jack Wang authored
Link: https://lore.kernel.org/r/20210325153308.1214057-18-gi-oh.kim@ionos.comSigned-off-by:
-
Gioh Kim authored
Before receiving the session name, the error message cannot include any information about which connection generates the error. This patch stores the addresses of source and target in the sessname field to show which generates the error. That field will be over-written when receiving the session name from client. Link: https://lore.kernel.org/r/20210325153308.1214057-17-gi-oh.kim@ionos.comSigned-off-by:
-
Gioh Kim authored
There is common code converting addresses of source machine and destination machine to a string. We already have a struct rtrs_addr to store two addresses. This patch introduces a new function that converts two addresses into one string with struct rtrs_addr. Link: https://lore.kernel.org/r/20210325153308.1214057-14-gi-oh.kim@ionos.comSigned-off-by:
-
Guoqing Jiang authored
Let these cases share the same path since all of them need to close session. Link: https://lore.kernel.org/r/20210325153308.1214057-11-gi-oh.kim@ionos.comSigned-off-by:
Guoqing Jiang <guoqing.jiang@ionos.com> Reviewed-by:
Danil Kipnis <danil.kipnis@ionos.com> Signed-off-by:
-
Guoqing Jiang authored
The two members are not used in the code, so remove them. Link: https://lore.kernel.org/r/20210325153308.1214057-10-gi-oh.kim@ionos.comSigned-off-by:
-
Guoqing Jiang authored
We can remove the label after move put_device to the right place. Link: https://lore.kernel.org/r/20210325153308.1214057-9-gi-oh.kim@ionos.comSigned-off-by:
-
Guoqing Jiang authored
There is no need to dereference 's' from 'sess', since we have "sess = to_clt_sess(s)" before. And we can deference 'dev' from 's' earlier. Link: https://lore.kernel.org/r/20210325153308.1214057-8-gi-oh.kim@ionos.comSigned-off-by:
-
Danil Kipnis authored
Danil will step down, Haris will take over. Also update to email address to ionos.com, cloud.ionos.com will still work for sometime. Link: https://lore.kernel.org/r/20210325153308.1214057-2-gi-oh.kim@ionos.comSigned-off-by:
Danil Kipnis <danil.kipnis@cloud.ionos.com> Acked-by:
Md Haris Iqbal <haris.iqbal@cloud.ionos.com> Signed-off-by:
Jack Wang <jinpu.wang@cloud.ionos.com> Signed-off-by:
-
YueHaibing authored
make W=1 warns this: In file included from drivers/infiniband/sw/rdmavt/mmap.c:51:0: ./include/rdma/uverbs_ioctl.h:937:1: warning: ‘_uverbs_get_const_unsigned’ defined but not used [-Wunused-function] _uverbs_get_const_unsigned(u64 *to, ^~~~~~~~~~~~~~~~~~~~~~~~~~ ./include/rdma/uverbs_ioctl.h:930:1: warning: ‘_uverbs_get_const_signed’ defined but not used [-Wunused-function] _uverbs_get_const_signed(s64 *to, const struct uverbs_attr_bundle *attrs_bundle, ^~~~~~~~~~~~~~~~~~~~~~~~ Make these functions inline to fix this warnings. Fixes: 2904bb37 ("IB/core: Split uverbs_get_const/default to consider target type") Link: https://lore.kernel.org/r/20210401021028.25720-1-yuehaibing@huawei.comSigned-off-by:
YueHaibing <yuehaibing@huawei.com> Signed-off-by:
-
Yangyang Li authored
The type of congestion control algorithm includes DCQCN, LDCP, HC3 and DIP. The driver will select one of them according to the firmware when querying PF capabilities, and then set the related configuration fields into QPC. Link: https://lore.kernel.org/r/1616679236-7795-3-git-send-email-liweihang@huawei.comSigned-off-by:
-
Wei Xu authored
Add a new type of command to query mac id of functions from the firmware, it is used to select the template of congestion algorithm. More info will be supported in the future. Link: https://lore.kernel.org/r/1616679236-7795-2-git-send-email-liweihang@huawei.comSigned-off-by:
Wei Xu <xuwei5@hisilicon.com> Signed-off-by:
Shengming Shu <shushengming1@huawei.com> Signed-off-by:
-
Håkon Bugge authored
On RoCE systems, a CM REQ contains a Primary Hop Limit > 1 and Primary Subnet Local is zero. In cm_req_handler(), the cm_process_routed_req() function is called. Since the Primary Subnet Local value is zero in the request, and since this is RoCE (Primary Local LID is permissive), the following statement will be executed: IBA_SET(CM_REQ_PRIMARY_SL, req_msg, wc->sl); This corrupts SL in req_msg if it was different from zero. In other words, a request to setup a connection using an SL != zero, will not be honored, and a connection using SL zero will be created instead. Fixed by not calling cm_process_routed_req() on RoCE systems, the cm_process_route_req() is only for IB anyhow. Fixes: 3971c9f6 ("IB/cm: Add interim support for routed paths") Link: https://lore.kernel.org/r/1616420132-31005-1-git-send-email-haakon.bugge@oracle.comSigned-off-by:Håkon Bugge <haakon.bugge@oracle.com> Signed-off-by:
-
