summaryrefslogtreecommitdiff
path: root/drivers/infiniband/hw
AgeCommit message (Collapse)Author
2024-12-30RDMA/erdma: Add missing fields to the erdma_device_ops_rocev2Boshi Yu
Set the query_ah field to the erdma_create_ah() function and set the size_ib_ah field to the size of struct erdma_ah. Reviewed-by: Cheng Xu <chengyou@linux.alibaba.com> Signed-off-by: Boshi Yu <boshiyu@linux.alibaba.com> Link: https://patch.msgid.link/20241226084141.74823-2-boshiyu@linux.alibaba.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2024-12-30RDMA/efa: Reset device on probe failureMichael Margolin
Make sure the device is being reset on driver exit whatever the reason is, to keep the device aligned and allow it to close shared resources (e.g. admin queue). Reviewed-by: Firas Jahjah <firasj@amazon.com> Reviewed-by: Yonatan Nachum <ynachum@amazon.com> Signed-off-by: Michael Margolin <mrgolin@amazon.com> Link: https://patch.msgid.link/20241225131548.15155-1-mrgolin@amazon.com Reviewed-by: Gal Pressman <gal.pressman@linux.dev> Signed-off-by: Leon Romanovsky <leon@kernel.org>
2024-12-25RDMA/hns: Support fast path for link-down events dispatchingYuyu Li
hns3 NIC driver can directly notify the RoCE driver about link status events bypassing the netdev notifier. This can provide more timely event dispatching for ULPs. Signed-off-by: Yuyu Li <liyuyu6@huawei.com> Signed-off-by: Junxian Huang <huangjunxian6@hisilicon.com> Signed-off-by: Leon Romanovsky <leon@kernel.org>
2024-12-25RDMA/mlx5: Handle link status event only for LAG deviceYuyu Li
The link status events of non-LAG devices are now handled in ib_core, so only LAG device events need to be handled in driver. Signed-off-by: Yuyu Li <liyuyu6@huawei.com> Signed-off-by: Junxian Huang <huangjunxian6@hisilicon.com> Signed-off-by: Leon Romanovsky <leon@kernel.org>
2024-12-25RDMA/pvrdma: Support report_port_event() opsYuyu Li
In addition to dispatching event, some private stuffs need to be done in this driver's link status event handler. Implement the new report_port_event() ops with the link status event codes. Signed-off-by: Yuyu Li <liyuyu6@huawei.com> Signed-off-by: Junxian Huang <huangjunxian6@hisilicon.com> Signed-off-by: Leon Romanovsky <leon@kernel.org>
2024-12-25RDMA/mlx4: Support report_port_event() opsYuyu Li
In addition to dispatching event, some private stuffs need to be done in this driver's link status event handler. Implement the new report_port_event() ops with the link status event codes. Signed-off-by: Yuyu Li <liyuyu6@huawei.com> Signed-off-by: Junxian Huang <huangjunxian6@hisilicon.com> Signed-off-by: Leon Romanovsky <leon@kernel.org>
2024-12-25RDMA/usnic: Support report_port_event() opsYuyu Li
In addition to dispatching event, some private stuffs need to be done in this driver's link status event handler. Implement the new report_port_event() ops with the link status event codes. Signed-off-by: Yuyu Li <liyuyu6@huawei.com> Signed-off-by: Junxian Huang <huangjunxian6@hisilicon.com> Signed-off-by: Leon Romanovsky <leon@kernel.org>
2024-12-25RDMA/irdma: 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. Signed-off-by: Yuyu Li <liyuyu6@huawei.com> Signed-off-by: Junxian Huang <huangjunxian6@hisilicon.com> Signed-off-by: Leon Romanovsky <leon@kernel.org>
2024-12-25RDMA/erdma: 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 dev->state as it is only used in erdma_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>
2024-12-25RDMA/bnxt_re: 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. 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/irdma: Remove unused irdma_cqp_*_fpm_val_cmd functionsDr. David Alan Gilbert
irdma_cqp_commit_fpm_val_cmd() and irdma_cqp_query_fpm_val_cmd() were added in 2021 by commit 915cc7ac0f8e ("RDMA/irdma: Add miscellaneous utility definitions") but haven't been used. Remove them. Signed-off-by: Dr. David Alan Gilbert <linux@treblig.org> Link: https://patch.msgid.link/20241223001613.307138-1-linux@treblig.org Signed-off-by: Leon Romanovsky <leon@kernel.org>
2024-12-23net/mlx5: fs, add counter object to flow destinationMoshe Shemesh
Currently mlx5_flow_destination includes counter_id which is assigned in case we use flow counter on the flow steering rule. However, counter_id is not enough data in case of using HW Steering. Thus, have mlx5_fc object as part of mlx5_flow_destination instead of counter_id and assign it where needed. In case counter_id is received from user space, create a local counter object to represent it. Signed-off-by: Moshe Shemesh <moshe@nvidia.com> Reviewed-by: Yevgeny Kliteynik <kliteyn@nvidia.com> Reviewed-by: Mark Bloch <mbloch@nvidia.com> Signed-off-by: Tariq Toukan <tariqt@nvidia.com> Link: https://patch.msgid.link/20241219175841.1094544-4-tariqt@nvidia.com Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2024-12-23RDMA/hns: Fix missing flush CQE for DWQEChengchang Tang
Flush CQE handler has not been called if QP state gets into errored mode in DWQE path. So, the new added outstanding WQEs will never be flushed. It leads to a hung task timeout when using NFS over RDMA: __switch_to+0x7c/0xd0 __schedule+0x350/0x750 schedule+0x50/0xf0 schedule_timeout+0x2c8/0x340 wait_for_common+0xf4/0x2b0 wait_for_completion+0x20/0x40 __ib_drain_sq+0x140/0x1d0 [ib_core] ib_drain_sq+0x98/0xb0 [ib_core] rpcrdma_xprt_disconnect+0x68/0x270 [rpcrdma] xprt_rdma_close+0x20/0x60 [rpcrdma] xprt_autoclose+0x64/0x1cc [sunrpc] process_one_work+0x1d8/0x4e0 worker_thread+0x154/0x420 kthread+0x108/0x150 ret_from_fork+0x10/0x18 Fixes: 01584a5edcc4 ("RDMA/hns: Add support of direct wqe") Signed-off-by: Chengchang Tang <tangchengchang@huawei.com> Signed-off-by: Junxian Huang <huangjunxian6@hisilicon.com> Link: https://patch.msgid.link/20241220055249.146943-5-huangjunxian6@hisilicon.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2024-12-23RDMA/hns: Fix warning storm caused by invalid input in IO pathChengchang Tang
WARN_ON() is called in the IO path. And it could lead to a warning storm. Use WARN_ON_ONCE() instead of WARN_ON(). Fixes: 12542f1de179 ("RDMA/hns: Refactor process about opcode in post_send()") Signed-off-by: Chengchang Tang <tangchengchang@huawei.com> Signed-off-by: Junxian Huang <huangjunxian6@hisilicon.com> Link: https://patch.msgid.link/20241220055249.146943-4-huangjunxian6@hisilicon.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2024-12-23RDMA/hns: Fix accessing invalid dip_ctx during destroying QPChengchang Tang
If it fails to modify QP to RTR, dip_ctx will not be attached. And during detroying QP, the invalid dip_ctx pointer will be accessed. Fixes: faa62440a577 ("RDMA/hns: Fix different dgids mapping to the same dip_idx") Signed-off-by: Chengchang Tang <tangchengchang@huawei.com> Signed-off-by: Junxian Huang <huangjunxian6@hisilicon.com> Link: https://patch.msgid.link/20241220055249.146943-3-huangjunxian6@hisilicon.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2024-12-23RDMA/hns: Fix mapping error of zero-hop WQE bufferwenglianfa
Due to HW limitation, the three region of WQE buffer must be mapped and set to HW in a fixed order: SQ buffer, SGE buffer, and RQ buffer. Currently when one region is zero-hop while the other two are not, the zero-hop region will not be mapped. This violate the limitation above and leads to address error. Fixes: 38389eaa4db1 ("RDMA/hns: Add mtr support for mixed multihop addressing") Signed-off-by: wenglianfa <wenglianfa@huawei.com> Signed-off-by: Junxian Huang <huangjunxian6@hisilicon.com> Link: https://patch.msgid.link/20241220055249.146943-2-huangjunxian6@hisilicon.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2024-12-19RDMA/erdma: Fix opcode conditional checkAdvait Dhamorikar
Fix conditional if else check by checking with wr->opcode. The indicated dead code may have performed some action; that action will never occur as op is pre-assigned a different value. Fixes: 999a0a2e9b87 ("RDMA/erdma: Support UD QPs and UD WRs") Signed-off-by: Advait Dhamorikar <advaitdhamorikar@gmail.com> Link: https://patch.msgid.link/20241219043939.10344-1-advaitdhamorikar@gmail.com Reviewed-by: Cheng Xu <chengyou@linux.alibaba.com> Signed-off-by: Leon Romanovsky <leon@kernel.org>
2024-12-19RDMA/bnxt_re: Fix the locking while accessing the QP tableSelvin Xavier
QP table handling is synchronized with destroy QP and Async event from the HW. The same needs to be synchronized during create_qp also. Use the same lock in create_qp also. Fixes: 76d3ddff7153 ("RDMA/bnxt_re: synchronize the qp-handle table array") Fixes: f218d67ef004 ("RDMA/bnxt_re: Allow posting when QPs are in error") Fixes: 84cf229f4001 ("RDMA/bnxt_re: Fix the qp table indexing") Signed-off-by: Selvin Xavier <selvin.xavier@broadcom.com> Link: https://patch.msgid.link/20241217102649.1377704-6-kalesh-anakkur.purayil@broadcom.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2024-12-19RDMA/bnxt_re: Fix MSN table size for variable wqe modeDamodharam Ammepalli
For variable size wqe mode, the MSN table size should be half the size of the SQ depth. Fixing this to avoid wrap around problems in the retransmission path. Fixes: de1d364c3815 ("RDMA/bnxt_re: Add support for Variable WQE in Genp7 adapters") Reviewed-by: Kashyap Desai <kashyap.desai@broadcom.com> Reviewed-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com> Signed-off-by: Damodharam Ammepalli <damodharam.ammepalli@broadcom.com> Signed-off-by: Selvin Xavier <selvin.xavier@broadcom.com> Link: https://patch.msgid.link/20241217102649.1377704-5-kalesh-anakkur.purayil@broadcom.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2024-12-19RDMA/bnxt_re: Add send queue size check for variable wqeDamodharam Ammepalli
For the fixed WQE case, HW supports 0xFFFF WQEs. For variable Size WQEs, HW treats this number as the 16 bytes slots. The maximum supported WQEs needs to be adjusted based on the number of slots. Set a maximum WQE limit for variable WQE scenario. Fixes: de1d364c3815 ("RDMA/bnxt_re: Add support for Variable WQE in Genp7 adapters") Reviewed-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com> Signed-off-by: Damodharam Ammepalli <damodharam.ammepalli@broadcom.com> Signed-off-by: Selvin Xavier <selvin.xavier@broadcom.com> Link: https://patch.msgid.link/20241217102649.1377704-4-kalesh-anakkur.purayil@broadcom.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2024-12-19RDMA/bnxt_re: Disable use of reserved wqesKalesh AP
Disabling the reserved wqes logic for Gen P5/P7 devices because this workaround is required only for legacy devices. Fixes: ecb53febfcad ("RDMA/bnxt_en: Enable RDMA driver support for 57500 chip") Signed-off-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com> Signed-off-by: Selvin Xavier <selvin.xavier@broadcom.com> Link: https://patch.msgid.link/20241217102649.1377704-3-kalesh-anakkur.purayil@broadcom.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2024-12-19RDMA/bnxt_re: Fix max_qp_wrs reportedSelvin Xavier
While creating qps, driver adds one extra entry to the sq size passed by the ULPs in order to avoid queue full condition. When ULPs creates QPs with max_qp_wr reported, driver creates QP with 1 more than the max_wqes supported by HW. Create QP fails in this case. To avoid this error, reduce 1 entry in max_qp_wqes and report it to the stack. Fixes: 1ac5a4047975 ("RDMA/bnxt_re: Add bnxt_re RoCE driver") Reviewed-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com> Signed-off-by: Selvin Xavier <selvin.xavier@broadcom.com> Link: https://patch.msgid.link/20241217102649.1377704-2-kalesh-anakkur.purayil@broadcom.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2024-12-17IB/hfi1: Remove unused hfi1_format_hwerrorsDr. David Alan Gilbert
hfi1_format_hwerrors() was added in 2015 by commit 7724105686e7 ("IB/hfi1: add driver files") but never used. Remove it. Signed-off-by: Dr. David Alan Gilbert <linux@treblig.org> Link: https://patch.msgid.link/20241216211914.745111-1-linux@treblig.org Signed-off-by: Leon Romanovsky <leon@kernel.org>
2024-12-16RDMA/bnxt_re: Fix reporting hw_ver in query_deviceKalesh AP
Driver currently populates subsystem_device id in the "hw_ver" field of ib_attr structure in query_device. Updated to populate PCI revision ID. Fixes: 1ac5a4047975 ("RDMA/bnxt_re: Add bnxt_re RoCE driver") Reviewed-by: Preethi G <preethi.gurusiddalingeswaraswamy@broadcom.com> Signed-off-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com> Signed-off-by: Selvin Xavier <selvin.xavier@broadcom.com> Link: https://patch.msgid.link/20241211083931.968831-6-kalesh-anakkur.purayil@broadcom.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2024-12-16RDMA/bnxt_re: Fix to export port num to ib_query_qpHongguang Gao
Current driver implementation doesn't populate the port_num field in query_qp. Adding the code to convert internal firmware port id to ibv defined port number and export it. Reviewed-by: Saravanan Vajravel <saravanan.vajravel@broadcom.com> Reviewed-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com> Signed-off-by: Hongguang Gao <hongguang.gao@broadcom.com> Signed-off-by: Selvin Xavier <selvin.xavier@broadcom.com> Link: https://patch.msgid.link/20241211083931.968831-5-kalesh-anakkur.purayil@broadcom.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2024-12-16RDMA/bnxt_re: Fix setting mandatory attributes for modify_qpDamodharam Ammepalli
Firmware expects "min_rnr_timer" as a mandatory attribute in MODIFY_QP command during the RTR-RTS transition. This needs to be enforced by the driver which is missing while setting bnxt_set_mandatory_attributes that sends these flags as part of modify_qp optimization. Fixes: 82c32d219272 ("RDMA/bnxt_re: Add support for optimized modify QP") Reviewed-by: Rukhsana Ansari <rukhsana.ansari@broadcom.com> Reviewed-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com> Signed-off-by: Damodharam Ammepalli <damodharam.ammepalli@broadcom.com> Signed-off-by: Selvin Xavier <selvin.xavier@broadcom.com> Link: https://patch.msgid.link/20241211083931.968831-4-kalesh-anakkur.purayil@broadcom.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2024-12-16RDMA/bnxt_re: Add check for path mtu in modify_qpSaravanan Vajravel
When RDMA app configures path MTU, add a check in modify_qp verb to make sure that it doesn't go beyond interface MTU. If this check fails, driver will fail the modify_qp verb. Fixes: 1ac5a4047975 ("RDMA/bnxt_re: Add bnxt_re RoCE driver") Reviewed-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com> Signed-off-by: Saravanan Vajravel <saravanan.vajravel@broadcom.com> Signed-off-by: Selvin Xavier <selvin.xavier@broadcom.com> Link: https://patch.msgid.link/20241211083931.968831-3-kalesh-anakkur.purayil@broadcom.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2024-12-16RDMA/bnxt_re: Fix the check for 9060 conditionKalesh AP
The check for 9060 condition should only be made for legacy chips. Fixes: 9152e0b722b2 ("RDMA/bnxt_re: HW workarounds for handling specific conditions") Reviewed-by: Kashyap Desai <kashyap.desai@broadcom.com> Signed-off-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com> Signed-off-by: Selvin Xavier <selvin.xavier@broadcom.com> Link: https://patch.msgid.link/20241211083931.968831-2-kalesh-anakkur.purayil@broadcom.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2024-12-16RDMA/bnxt_re: Remove unnecessary header file inclusionKalesh AP
There is no need to include bnxt_ulp.h in ib_verbs.c. Remove it. Also, fixed hw_counters.c to remove unwanted header file inclusions. Signed-off-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com> Signed-off-by: Selvin Xavier <selvin.xavier@broadcom.com> Link: https://patch.msgid.link/1733888745-30939-6-git-send-email-selvin.xavier@broadcom.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2024-12-16RDMA/bnxt_re: Eliminate need for some forward declarationsKalesh AP
Move the function definition of bnxt_re_shutdown() to avoid forward declarartion of bnxt_re_dev_uninit(). Move the function definition of bnxt_re_setup_cc() before bnxt_re_add_device() to avoid it's forward declarations. Also, forward declarartions of bnxt_re_stop_irq() and bnxt_re_dev_stop() are unnecessary. Signed-off-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com> Signed-off-by: Selvin Xavier <selvin.xavier@broadcom.com> Link: https://patch.msgid.link/1733888745-30939-5-git-send-email-selvin.xavier@broadcom.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2024-12-16RDMA/bnxt_re: Optimize error handling in bnxt_re_probeKalesh AP
Optimize error handling path in bnxt_re_probe by removing some duplicate code. Signed-off-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com> Signed-off-by: Selvin Xavier <selvin.xavier@broadcom.com> Link: https://patch.msgid.link/1733888745-30939-4-git-send-email-selvin.xavier@broadcom.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2024-12-16RDMA/bnxt_re: Remove unnecessary goto in bnxt_re_netdev_eventKalesh AP
Return directly in case of error without a goto label as there is no cleanup actions performed. Signed-off-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com> Signed-off-by: Selvin Xavier <selvin.xavier@broadcom.com> Link: https://patch.msgid.link/1733888745-30939-3-git-send-email-selvin.xavier@broadcom.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2024-12-16RDMA/bnxt_re: Remove extra new line in bnxt_re_netdev_eventKalesh AP
This is a purely cosmetic change. Signed-off-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com> Signed-off-by: Selvin Xavier <selvin.xavier@broadcom.com> Link: https://patch.msgid.link/1733888745-30939-2-git-send-email-selvin.xavier@broadcom.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2024-12-16RDMA/erdma: Support UD QPs and UD WRsBoshi Yu
The iWARP protocol supports only RC QPs previously. Now we add UD QPs and UD WRs support for the RoCEv2 protocol. Signed-off-by: Boshi Yu <boshiyu@linux.alibaba.com> Link: https://patch.msgid.link/20241211020930.68833-9-boshiyu@linux.alibaba.com Reviewed-by: Cheng Xu <chengyou@linux.alibaba.com> Signed-off-by: Leon Romanovsky <leon@kernel.org>
2024-12-16RDMA/erdma: Add the query_qp command to the cmdqBoshi Yu
Certian QP attributes, such as sq_draining, can only be obtained by querying the hardware on the erdma RoCEv2 device. To address this, we add the query_qp command to the cmdq and parse the response to retrieve corresponding QP attributes. Signed-off-by: Boshi Yu <boshiyu@linux.alibaba.com> Link: https://patch.msgid.link/20241211020930.68833-8-boshiyu@linux.alibaba.com Reviewed-by: Cheng Xu <chengyou@linux.alibaba.com> Signed-off-by: Leon Romanovsky <leon@kernel.org>
2024-12-16RDMA/erdma: Refactor the code of the modify_qp interfaceBoshi Yu
The procedure for modifying QP is similar for both the iWARP and RoCEv2 protocols. Therefore, we unify the code and provide the erdma_modify_qp() interface for both protocols. Signed-off-by: Boshi Yu <boshiyu@linux.alibaba.com> Link: https://patch.msgid.link/20241211020930.68833-7-boshiyu@linux.alibaba.com Reviewed-by: Cheng Xu <chengyou@linux.alibaba.com> Signed-off-by: Leon Romanovsky <leon@kernel.org>
2024-12-16RDMA/erdma: Add erdma_modify_qp_rocev2() interfaceBoshi Yu
The QP state machines in the RoCEv2 and iWARP protocols are different. To handle these differences for the erdma RoCEv2 device, we provide the erdma_modify_qp_rocev2() interface, which transitions the QP state and modifies QP attributes accordingly. Signed-off-by: Boshi Yu <boshiyu@linux.alibaba.com> Link: https://patch.msgid.link/20241211020930.68833-6-boshiyu@linux.alibaba.com Reviewed-by: Cheng Xu <chengyou@linux.alibaba.com> Signed-off-by: Leon Romanovsky <leon@kernel.org>
2024-12-16RDMA/erdma: Add address handle implementationBoshi Yu
The address handle contains the necessary information to transmit messages to a remote peer in the RoCEv2 protocol. This commit implements the erdma_create_ah(), erdma_destroy_ah(), and erdma_query_ah() interfaces, which are used to create, destroy, and query an address handle, respectively. Signed-off-by: Boshi Yu <boshiyu@linux.alibaba.com> Link: https://patch.msgid.link/20241211020930.68833-5-boshiyu@linux.alibaba.com Reviewed-by: Cheng Xu <chengyou@linux.alibaba.com> Signed-off-by: Leon Romanovsky <leon@kernel.org>
2024-12-16RDMA/erdma: Add the erdma_query_pkey() interfaceBoshi Yu
The erdma_query_pkey() interface queries the PKey at the specified index. Currently, erdma supports only one partition and returns the default PKey for each query. Besides, the correct length of the PKey table can be obtained by calling the erdma_query_port() and erdma_get_port_immutable() interfaces. Signed-off-by: Boshi Yu <boshiyu@linux.alibaba.com> Link: https://patch.msgid.link/20241211020930.68833-4-boshiyu@linux.alibaba.com Reviewed-by: Cheng Xu <chengyou@linux.alibaba.com> Signed-off-by: Leon Romanovsky <leon@kernel.org>
2024-12-16RDMA/erdma: Add GID table management interfacesBoshi Yu
The erdma_add_gid() interface inserts a GID entry at the specified index. The erdma_del_gid() interface deletes the GID entry at the specified index. Additionally, programs can invoke the erdma_query_port() and erdma_get_port_immutable() interfaces to query the GID table length. Signed-off-by: Boshi Yu <boshiyu@linux.alibaba.com> Link: https://patch.msgid.link/20241211020930.68833-3-boshiyu@linux.alibaba.com Reviewed-by: Cheng Xu <chengyou@linux.alibaba.com> Signed-off-by: Leon Romanovsky <leon@kernel.org>
2024-12-16RDMA/erdma: Probe the erdma RoCEv2 deviceBoshi Yu
Currently, the erdma driver supports both the iWARP and RoCEv2 protocols. The erdma driver reads the ERDMA_REGS_DEV_PROTO_REG register to identify the protocol used by the erdma device. Since each protocol requires different ib_device_ops, we introduce the erdma_device_ops_iwarp and erdma_device_ops_rocev2 for iWARP and RoCEv2 protocols, respectively. Signed-off-by: Boshi Yu <boshiyu@linux.alibaba.com> Link: https://patch.msgid.link/20241211020930.68833-2-boshiyu@linux.alibaba.com Reviewed-by: Cheng Xu <chengyou@linux.alibaba.com> Signed-off-by: Leon Romanovsky <leon@kernel.org>
2024-12-10rdma/cxgb4: Prevent potential integer overflow on 32bitDan Carpenter
The "gl->tot_len" variable is controlled by the user. It comes from process_responses(). On 32bit systems, the "gl->tot_len + sizeof(struct cpl_pass_accept_req) + sizeof(struct rss_header)" addition could have an integer wrapping bug. Use size_add() to prevent this. Fixes: 1cab775c3e75 ("RDMA/cxgb4: Fix LE hash collision bug for passive open connection") Link: https://patch.msgid.link/r/86b404e1-4a75-4a35-a34e-e3054fa554c7@stanley.mountain Signed-off-by: Dan Carpenter <dan.carpenter@linaro.org> Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
2024-12-10RDMA/mlx5: Extend ODP statistics with operation countChiara Meiohas
The current ODP counters represent the total number of pages handled, but it is not enough to understand the effectiveness of these operations. Extend the ODP counters to include the number of times page fault and invalidation events were handled. Example for a single page fault handling 512 pages: - page_fault: incremented by 512 (total pages) - page_fault_handled: incremented by 1 (operation count) The same example is applicable for page invalidation too. Previous output: $ rdma stat mr dev rocep8s0f0 mrn 8 page_faults 27 page_invalidations 0 page_prefetch 29 New output: $ rdma stat mr dev rocep8s0f0 mrn 21 page_faults 512 page_faults_handled 1 page_invalidations 0 page_invalidations_handled 0 page_prefetch 51200 Signed-off-by: Chiara Meiohas <cmeiohas@nvidia.com> Reviewed-by: Michael Guralnik <michaelgur@nvidia.com> Link: https://patch.msgid.link/b18f29ed1392996ade66e9e6c45f018925253f6a.1733234165.git.leonro@nvidia.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2024-12-10RDMA/mlx4: Use DMA iterator to write MTTLeon Romanovsky
Replace an open coding of rdma_umem_for_each_dma_block() with the proper function. Link: https://patch.msgid.link/0bf595962c964fb8918743405acf9103a5a85983.1733233299.git.leonro@nvidia.com Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
2024-12-10RDMA/mlx4: Use ib_umem_find_best_pgsz() to calculate MTT sizeLeon Romanovsky
Convert mlx4 to use ib_umem_find_best_pgsz() instead of open-coded variant to calculate MTT size. Link: https://patch.msgid.link/c39ec6f5d4664c439a72f2961728ebb5895a9f07.1733233299.git.leonro@nvidia.com Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
2024-12-10RDMA/mlx4: Avoid false error about access to uninitialized gids arrayLeon Romanovsky
Smatch generates the following false error report: drivers/infiniband/hw/mlx4/main.c:393 mlx4_ib_del_gid() error: uninitialized symbol 'gids'. Traditionally, we are not changing kernel code and asking people to fix the tools. However in this case, the fix can be done by simply rearranging the code to be more clear. Fixes: e26be1bfef81 ("IB/mlx4: Implement ib_device callbacks") Link: https://patch.msgid.link/6a3a1577463da16962463fcf62883a87506e9b62.1733233426.git.leonro@nvidia.com Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
2024-12-05RDMA/bnxt_re: Don't fail destroy QP and cleanup debugfs earlierKalesh AP
Change bnxt_re_destroy_qp to always return 0 and don't fail in case of error during destroy. In addition, delete debugfs QP to earlier stage. Fixes: d7d54769c042 ("RDMA/bnxt_re: Add debugfs hook in the driver") Reviewed-by: Kashyap Desai <kashyap.desai@broadcom.com> Signed-off-by: Kalesh AP <kalesh-anakkur.purayil@broadcom.com> Signed-off-by: Selvin Xavier <selvin.xavier@broadcom.com> Link: https://patch.msgid.link/20241204075416.478431-6-kalesh-anakkur.purayil@broadcom.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2024-12-05RDMA/bnxt_re: Avoid sending the modify QP workaround for latest adaptersKashyap Desai
The workaround to modify the UD QP from RTS to RTS is required only for older adapters. Issuing this for latest adapters can caus some unexpected behavior. Fix it Fixes: 1801d87b3598 ("RDMA/bnxt_re: Support new 5760X P7 devices") Signed-off-by: Kashyap Desai <kashyap.desai@broadcom.com> Signed-off-by: Selvin Xavier <selvin.xavier@broadcom.com> Link: https://patch.msgid.link/20241204075416.478431-4-kalesh-anakkur.purayil@broadcom.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2024-12-05RDMA/bnxt_re: Avoid initializing the software queue for user queuesSelvin Xavier
Software Queues to hold the WRs needs to be created for only kernel queues. Avoid allocating the unnecessary memory for user Queues. Fixes: 1ac5a4047975 ("RDMA/bnxt_re: Add bnxt_re RoCE driver") Fixes: 159fb4ceacd7 ("RDMA/bnxt_re: introduce a function to allocate swq") Signed-off-by: Selvin Xavier <selvin.xavier@broadcom.com> Link: https://patch.msgid.link/20241204075416.478431-3-kalesh-anakkur.purayil@broadcom.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
2024-12-05RDMA/bnxt_re: Fix max SGEs for the Work RequestKashyap Desai
Gen P7 supports up to 13 SGEs for now. WQE software structure can hold only 6 now. Since the max send sge is reported as 13, the stack can give requests up to 13 SGEs. This is causing traffic failures and system crashes. Use the define for max SGE supported for variable size. This will work for both static and variable WQEs. Fixes: 227f51743b61 ("RDMA/bnxt_re: Fix the max WQE size for static WQE support") Signed-off-by: Kashyap Desai <kashyap.desai@broadcom.com> Signed-off-by: Selvin Xavier <selvin.xavier@broadcom.com> Link: https://patch.msgid.link/20241204075416.478431-2-kalesh-anakkur.purayil@broadcom.com Signed-off-by: Leon Romanovsky <leon@kernel.org>
generated by cgit (git 2.34.1) at 2025-07-24 05:15:21 +0000