- 07 Dec, 2022 8 commits
-
-
Chengchang Tang authored
XRC caps has been set by default. But in fact, XRC is not supported in HIP08. Fixes: 32548870 ("RDMA/hns: Add support for XRC on HIP09") Link: https://lore.kernel.org/r/20221126102911.2921820-7-xuhaoyue1@hisilicon.comSigned-off-by:
Chengchang Tang <tangchengchang@huawei.com> Signed-off-by:
Haoyue Xu <xuhaoyue1@hisilicon.com> Signed-off-by:
Jason Gunthorpe <jgg@nvidia.com>
-
Chengchang Tang authored
The error code is fixed to EIO when CMD fails to excute. This patch converts the error status reported by firmware to linux errno. Fixes: a04ff739 ("RDMA/hns: Add command queue support for hip08 RoCE driver") Link: https://lore.kernel.org/r/20221126102911.2921820-6-xuhaoyue1@hisilicon.comSigned-off-by:
-
Chengchang Tang authored
Add verification to make sure the roce page size cap is supported by the system page size. Fixes: ba6bb7e9 ("RDMA/hns: Add interfaces to get pf capabilities from firmware") Link: https://lore.kernel.org/r/20221126102911.2921820-5-xuhaoyue1@hisilicon.comSigned-off-by:
-
Chengchang Tang authored
Now, The address of the first two pages in the MR will be searched, which use to speed up the lookup of the pbl table for hardware. An exception will occur when there is only one page in this MR. This patch fix the number of page to search. Fixes: 9b2cf76c ("RDMA/hns: Optimize PBL buffer allocation process") Link: https://lore.kernel.org/r/20221126102911.2921820-4-xuhaoyue1@hisilicon.comSigned-off-by:
-
Chengchang Tang authored
The queried AH attr is invalid. This patch fix it. This problem is found by rdma-core test test_mr_rereg_pd ERROR: test_mr_rereg_pd (tests.test_mr.MRTest) Test that cover rereg MR's PD with this flow: ---------------------------------------------------------------------- Traceback (most recent call last): File "./tests/test_mr.py", line 157, in test_mr_rereg_pd self.restate_qps() File "./tests/test_mr.py", line 113, in restate_qps self.server.qp.to_rts(self.server_qp_attr) File "qp.pyx", line 1137, in pyverbs.qp.QP.to_rts File "qp.pyx", line 1123, in pyverbs.qp.QP.to_rtr pyverbs.pyverbs_error.PyverbsRDMAError: Failed to modify QP state to RTR. Errno: 22, Invalid argument Fixes: 926a01dc ("RDMA/hns: Add QP operations support for hip08 SoC") Link: https://lore.kernel.org/r/20221126102911.2921820-3-xuhaoyue1@hisilicon.comSigned-off-by: -
Yixing Liu authored
After the hns roce driver is loaded, if you modify the mac address of the network port, the following error will appear: __ib_cache_gid_add: unable to add gid fe80:0000:0000:0000:4600:4dff:fe22:abb5 error=-28 hns3 0000:7d:00.0 hns_0: attr path_mtu(1) invalid while modify qp The reason for the error is that the gid being occupied will cause the failure to modify the gid. The gid is occupied by the loopback QP used by free mr. When the mac address is modified, the gid will change. If there is a busy QP at this time, the gid will not be released and the modification will fail. The QP of free mr is created using the ib interface. The ib interface will add a reference count to the gid, resulting in this error scenario. Considering that free mr is solving a bug in HIP08, not an actual business, it is not necessary to use ib interfaces. Fixes: 70f92521 ("RDMA/hns: Use the reserved loopback QPs to free MR before destroying MPT") Link: https://lore.kernel.org/r/20221126102911.2921820-2-xuhaoyue1@hisilicon.comSigned-off-by:Yixing Liu <liuyixing1@huawei.com> Signed-off-by:
-
Chao Leng authored
The ack timeout retransmission time is affected by the following two factors: one is packet life time, another is the HCA processing time. Now the default packet lifetime(CMA_IBOE_PACKET_LIFETIME) is 18. That means the minimum ack timeout is 2 seconds (2^(18+1)*4us=2.097seconds). The packet lifetime means the maximum transmission time of packets on the network, 2 seconds is too long. Assume the network is a clos topology with three layers, every packet will pass through five hops of switches. Assume the buffer of every switch is 128MB and the port transmission rate is 25 Gbit/s, the maximum transmission time of the packet is 200ms(128MB*5/25Gbit/s). Add double redundancy, it is less than 500ms. So change the CMA_IBOE_PACKET_LIFETIME to 16, the maximum transmission time of the packet will be about 500+ms, it is long enough. Link: https://lore.kernel.org/r/20221125010026.755-1-lengchao@huawei.comSigned-off-by:
Chao Leng <lengchao@huawei.com> Signed-off-by:
-
ye xingchen authored
Follow the advice of the Documentation/filesystems/sysfs.rst and show() should only use sysfs_emit() or sysfs_emit_at() when formatting the value to be returned to user space. Signed-off-by:
ye xingchen <ye.xingchen@zte.com.cn> Link: https://lore.kernel.org/r/202212071632188074249@zte.com.cnSigned-off-by:
Leon Romanovsky <leon@kernel.org>
-
- 04 Dec, 2022 4 commits
-
-
Li Zhijian authored
Goto label 'free' where it will kfree the 'in' is not needed though it's safe to kfree NULL. Return err code directly to simplify the code. 1973 free: 1974 kfree(in); 1975 return err; Signed-off-by:
Li Zhijian <lizhijian@fujitsu.com> Link: https://lore.kernel.org/r/20221203033714.25870-1-lizhijian@fujitsu.comSigned-off-by:
-
Wang Yufen authored
In the previous iteration of the while loop, the "ret" may have been assigned a value of 0, so the error return code -EINVAL may have been incorrectly set to 0. To fix set valid return code before calling to goto. Also investigate each case separately as Andy suggessted. Fixes: e711f968 ("IB/srp: replace custom implementation of hex2bin()") Fixes: 2a174df0 ("IB/srp: Use kstrtoull() instead of simple_strtoull()") Fixes: 19f31343 ("IB/srp: Add RDMA/CM support") Signed-off-by:
Wang Yufen <wangyufen@huawei.com> Link: https://lore.kernel.org/r/1669953638-11747-2-git-send-email-wangyufen@huawei.comReviewed-by:
Bart Van Assche <bvanassche@acm.org> Signed-off-by:
-
Wang Yufen authored
In the previous iteration of the while loop, the "ret" may have been assigned a value of 0, so the error return code -EINVAL may have been incorrectly set to 0. To fix set valid return code before calling to goto. Fixes: 97167e81 ("staging/rdma/hfi1: Tune for unknown channel if configuration file is absent") Signed-off-by:
-
Randy Dunlap authored
Disable all of drivers/infiniband/hw/ and rdmavt for UML builds until someone needs it and provides patches to support it. This prevents build errors in hw/qib/qib_wc_x86_64.c. Fixes: 68f5d3f3 ("um: add PCI over virtio emulation driver") Signed-off-by:
Randy Dunlap <rdunlap@infradead.org> Cc: Jason Gunthorpe <jgg@nvidia.com> Cc: Dennis Dalessandro <dennis.dalessandro@cornelisnetworks.com> Cc: Christoph Hellwig <hch@infradead.org> Cc: linux-rdma@vger.kernel.org Cc: Jeff Dike <jdike@addtoit.com> Cc: Richard Weinberger <richard@nod.at> Cc: Anton Ivanov <anton.ivanov@cambridgegreys.com> Cc: Johannes Berg <johannes@sipsolutions.net> Cc: linux-um@lists.infradead.org Link: https://lore.kernel.org/r/20221202211940.29111-1-rdunlap@infradead.orgSigned-off-by:
-
- 01 Dec, 2022 9 commits
-
-
Xiao Yang authored
The capability shows that rxe device supports atomic write operation. Link: https://lore.kernel.org/r/1669905568-62-4-git-send-email-yangx.jy@fujitsu.comSigned-off-by:
Xiao Yang <yangx.jy@fujitsu.com> Signed-off-by:
-
Xiao Yang authored
Generate an atomic write completion when the atomic write request has been finished. Link: https://lore.kernel.org/r/1669905568-62-3-git-send-email-yangx.jy@fujitsu.comSigned-off-by:
-
Xiao Yang authored
Make responder process an atomic write request and send a read response on RC service. Link: https://lore.kernel.org/r/1669905568-62-2-git-send-email-yangx.jy@fujitsu.comSigned-off-by:
-
Xiao Yang authored
Make requester process and send an atomic write request on RC service. Link: https://lore.kernel.org/r/1669905568-62-1-git-send-email-yangx.jy@fujitsu.comSigned-off-by:
-
Xiao Yang authored
Extend rxe_wr_opcode_info[] and rxe_opcode[] for new atomic write opcode. Link: https://lore.kernel.org/r/1669905432-14-5-git-send-email-yangx.jy@fujitsu.comSigned-off-by:
-
Xiao Yang authored
Define an atomic_wr array to store 8-byte value. Link: https://lore.kernel.org/r/1669905432-14-4-git-send-email-yangx.jy@fujitsu.comSigned-off-by:
-
Xiao Yang authored
1) Define new atomic write request/completion in kernel. 2) Define new atomic write capability in kernel. 3) Define new atomic write opcode for RC service in packet. Link: https://lore.kernel.org/r/1669905432-14-3-git-send-email-yangx.jy@fujitsu.comSigned-off-by:
-
Xiao Yang authored
1) Define new atomic write request/completion in userspace. 2) Define new atomic write capability in userspace. Link: https://lore.kernel.org/r/1669905432-14-2-git-send-email-yangx.jy@fujitsu.comSigned-off-by:
-
Yang Yang authored
There is no need to use netif_napi_add_weight() when the weight argument is 64. See "net: drop the weight argument from netif_napi_add". Signed-off-by:
Yang Yang <yang.yang29@zte.com.cn> Link: https://lore.kernel.org/r/202211301744378304494@zte.com.cnReviewed-by:
xu xin <xu.xin16@zte.com.cn> Reviewed-by:
Zhang Yunkai <zhang.yunkai@zte.com.cn> Signed-off-by:
-
- 30 Nov, 2022 2 commits
-
-
Mark Zhang authored
Return "-EMSGSIZE" instead of "-EINVAL" when filling a QP entry, so that new SKBs will be allocated if there's not enough room in current SKB. Fixes: 65959522 ("RDMA: Add support to dump resource tracker in RAW format") Signed-off-by:
Mark Zhang <markzhang@nvidia.com> Reviewed-by:
Patrisious Haddad <phaddad@nvidia.com> Link: https://lore.kernel.org/r/b5e9c62f6b8369acab5648b661bf539cbceeffdc.1669636336.git.leonro@nvidia.comSigned-off-by:
-
Or Har-Toov authored
Using nlmsg_put causes static analysis tools to many false positives of not checking the return value of nlmsg_put. In all uses in nldev.c, payload parameter is 0 so NULL will never be returned. So let's add useless checks to silence the warnings. Signed-off-by:
Or Har-Toov <ohartoov@nvidia.com> Reviewed-by:
Michael Guralnik <michaelgur@nvidia.com> Link: https://lore.kernel.org/r/bd924da89d5b4f5291a4a01d9b5ae47c0a9b6a3f.1669636336.git.leonro@nvidia.comSigned-off-by:
-
- 29 Nov, 2022 1 commit
-
-
zhang songyi 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}. Fix the following coccicheck warnings: /drivers/infiniband/hw/mlx4/main.c:1311:2-10: WARNING: WARNING NULL check before dev_{put, hold} functions is not needed. /drivers/infiniband/hw/mlx4/main.c:148:2-10: WARNING: WARNING NULL check before dev_{put, hold} functions is not needed. /drivers/infiniband/hw/mlx4/main.c:1959:3-11: WARNING: WARNING NULL check before dev_{put, hold} functions is not needed. /drivers/infiniband/hw/mlx4/main.c:1962:3-10: WARNING: WARNING NULL check before dev_{put, hold} functions is not needed. Signed-off-by:zhang songyi <zhang.songyi@zte.com.cn> Link: https://lore.kernel.org/r/202211291554079687539@zte.com.cnSigned-off-by:
-
- 28 Nov, 2022 2 commits
-
-
Yuan Can authored
As the nla_nest_start() may fail with NULL returned, the return value needs to be checked. Fixes: c4ffee7c ("RDMA/netlink: Implement counter dumpit calback") Signed-off-by:
Yuan Can <yuancan@huawei.com> Link: https://lore.kernel.org/r/20221126043410.85632-1-yuancan@huawei.comSigned-off-by:
-
Jason Gunthorpe authored
This will cause an informative backtrace to print if the user of ib_device_set_netdev() isn't careful about tearing down the ibdevice before its the netdevice parent is destroyed. Such as like this: unregister_netdevice: waiting for vlan0 to become free. Usage count = 2 leaked reference. ib_device_set_netdev+0x266/0x730 siw_newlink+0x4e0/0xfd0 nldev_newlink+0x35c/0x5c0 rdma_nl_rcv_msg+0x36d/0x690 rdma_nl_rcv+0x2ee/0x430 netlink_unicast+0x543/0x7f0 netlink_sendmsg+0x918/0xe20 sock_sendmsg+0xcf/0x120 ____sys_sendmsg+0x70d/0x8b0 ___sys_sendmsg+0x11d/0x1b0 __sys_sendmsg+0xfa/0x1d0 do_syscall_64+0x35/0xb0 entry_SYSCALL_64_after_hwframe+0x63/0xcd This will help debug the issues syzkaller is seeing. Signed-off-by:
-
- 24 Nov, 2022 4 commits
-
-
Cheng Xu authored
Firmware is responsible for flushing WRs in HW, and it's a little difficult for firmware to get the latest PI of QPs, especially for RQs after QP state being changed to ERROR. So we introduce a new CMDQ command, by which driver can notify to latest PI to FW, and then FW can flush all posted WRs. Link: https://lore.kernel.org/r/20221116023107.82835-4-chengyou@linux.alibaba.comSigned-off-by:
Cheng Xu <chengyou@linux.alibaba.com> Signed-off-by:
-
Cheng Xu authored
Each QP has a work for reflushing purpose. In the work, driver will report the latest pi to hardware. Link: https://lore.kernel.org/r/20221116023107.82835-3-chengyou@linux.alibaba.comSigned-off-by:
-
Cheng Xu authored
ERDMA driver use a workqueue for asynchronous reflush command posting. Implement the lifecycle of this workqueue. Link: https://lore.kernel.org/r/20221116023107.82835-2-chengyou@linux.alibaba.comSigned-off-by:
-
Cheng Xu authored
A non-ASCII character was wrongly put in a comment, use the ACSII version. Fixes: bee85e0e ("RDMA/erdma: Add main include file") Link: https://lore.kernel.org/r/20221124114933.77250-1-chengyou@linux.alibaba.comSigned-off-by:
-
- 22 Nov, 2022 5 commits
-
-
Zhang Xiaoxu authored
There is a null-ptr-deref when mount.cifs over rdma: BUG: KASAN: null-ptr-deref in rxe_qp_do_cleanup+0x2f3/0x360 [rdma_rxe] Read of size 8 at addr 0000000000000018 by task mount.cifs/3046 CPU: 2 PID: 3046 Comm: mount.cifs Not tainted 6.1.0-rc5+ #62 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.14.0-1.fc3 Call Trace: <TASK> dump_stack_lvl+0x34/0x44 kasan_report+0xad/0x130 rxe_qp_do_cleanup+0x2f3/0x360 [rdma_rxe] execute_in_process_context+0x25/0x90 __rxe_cleanup+0x101/0x1d0 [rdma_rxe] rxe_create_qp+0x16a/0x180 [rdma_rxe] create_qp.part.0+0x27d/0x340 ib_create_qp_kernel+0x73/0x160 rdma_create_qp+0x100/0x230 _smbd_get_connection+0x752/0x20f0 smbd_get_connection+0x21/0x40 cifs_get_tcp_session+0x8ef/0xda0 mount_get_conns+0x60/0x750 cifs_mount+0x103/0xd00 cifs_smb3_do_mount+0x1dd/0xcb0 smb3_get_tree+0x1d5/0x300 vfs_get_tree+0x41/0xf0 path_mount+0x9b3/0xdd0 __x64_sys_mount+0x190/0x1d0 do_syscall_64+0x35/0x80 entry_SYSCALL_64_after_hwframe+0x46/0xb0 The root cause of the issue is the socket create failed in rxe_qp_init_req(). So move the reset rxe_qp_do_cleanup() after the NULL ptr check. Fixes: 8700e3e7 ("Soft RoCE driver") Link: https://lore.kernel.org/r/20221122151437.1057671-1-zhangxiaoxu5@huawei.comSigned-off-by:
Zhang Xiaoxu <zhangxiaoxu5@huawei.com> Signed-off-by:
-
Jason Gunthorpe authored
Correct the mistake, mr is obviously NULL in this code path. Fixes: 2778b72b ("RDMA/rxe: Replace pr_xxx by rxe_dbg_xxx in rxe_mr.c") Link: https://lore.kernel.org/r/Y3eeJW0AdyJYhYyQ@kiliReported-by:
Dan Carpenter <error27@gmail.com> Signed-off-by:
-
Zhengchao Shao authored
When hns_roce_mr_enable() failed in hns_roce_alloc_mr(), mr_key is not released. Compiled test only. Fixes: 9b2cf76c ("RDMA/hns: Optimize PBL buffer allocation process") Signed-off-by:
Zhengchao Shao <shaozhengchao@huawei.com> Link: https://lore.kernel.org/r/20221119070834.48502-1-shaozhengchao@huawei.comSigned-off-by:
-
Mustafa Ismail authored
The av->net_type is not initialized before it is checked in irdma_modify_qp_roce. This leads to an incorrect update to the ARP cache and QP context. RoCEv2 connections might fail as result. Set the net_type using rdma_gid_attr_network_type. Fixes: 80005c43 ("RDMA/irdma: Use net_type to check network type") Signed-off-by:
Mustafa Ismail <mustafa.ismail@intel.com> Signed-off-by:
Shiraz Saleem <shiraz.saleem@intel.com> Link: https://lore.kernel.org/r/20221122004410.1471-1-shiraz.saleem@intel.comSigned-off-by:
-
Xiongfeng Wang authored
pci_get_device() will increase the reference count for the returned pci_dev, and also decrease the reference count for the input parameter *from* if it is not NULL. If we break out the loop in node_affinity_init() with 'dev' not NULL, we need to call pci_dev_put() to decrease the reference count. Add missing pci_dev_put() in error path. Fixes: c513de49 ("IB/hfi1: Invalid NUMA node information can cause a divide by zero") Signed-off-by:
Xiongfeng Wang <wangxiongfeng2@huawei.com> Link: https://lore.kernel.org/r/20221117131546.113280-1-wangxiongfeng2@huawei.comSigned-off-by:
-
- 21 Nov, 2022 1 commit
-
-
Maurizio Lombardi authored
Use the proper macro to get the current_stage value. Link: https://lore.kernel.org/r/20221116094535.138298-1-mlombard@redhat.comSigned-off-by:
Maurizio Lombardi <mlombard@redhat.com> Reviewed-by:
Mike Christie <michael.christie@oracle.com> Acked-by:
Sagi Grimberg <sagi@grimberg.me> Signed-off-by:
-
- 19 Nov, 2022 3 commits
-
-
Luoyouming authored
The user usually configures the number of sge through the max_send_sge parameter when creating qp, and configures the maximum size of inline data that can be sent through max_inline_data. Inline uses sge to fill data to send. Expect the following: 1) When the sge space cannot hold inline data, the sge space needs to be expanded to accommodate all inline data 2) When the sge space is enough to accommodate inline data, the upper limit of inline data can be increased so that users can send larger inline data Currently case one is not implemented. When the inline data is larger than the sge space, an error of insufficient sge space occurs. This part of the code needs to be reimplemented according to the expected rules. The calculation method of sge num is modified to take the maximum value of max_send_sge and the sge for max_inline_data to solve this problem. Fixes: 05201e01 ("RDMA/hns: Refactor process of setting extended sge") Fixes: 30b70788 ("RDMA/hns: Support inline data in extented sge space for RC") Link: https://lore.kernel.org/r/20221108133847.2304539-3-xuhaoyue1@hisilicon.comSigned-off-by:
Luoyouming <luoyouming@huawei.com> Signed-off-by:
-
Luoyouming authored
In the HNS ROCE driver, The sge is divided into standard sge and extended sge. There are 2 standard sge in RC/XRC, and the UD standard sge is 0. In the scenario of RC SQ inline, if the data does not exceed 32bytes, the standard sge will be used. If it exceeds, only the extended sge will be used to fill the data. Currently, when filling the extended sge, max_gs is directly used as the number of the extended sge, which did not subtract the number of standard sge. There is a logical error. The new algorithm subtracts the number of standard sge from max_gs to get the actual number of extended sge. Fixes: 30b70788 ("RDMA/hns: Support inline data in extented sge space for RC") Link: https://lore.kernel.org/r/20221108133847.2304539-2-xuhaoyue1@hisilicon.comSigned-off-by:
-
Li Zhijian authored
rxe_mr_cleanup() which tries to free mr->map again will be called when rxe_mr_init_user() fails: CPU: 0 PID: 4917 Comm: rdma_flush_serv Kdump: loaded Not tainted 6.1.0-rc1-roce-flush+ #25 Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.16.0-0-gd239552ce722-prebuilt.qemu.org 04/01/2014 Call Trace: <TASK> dump_stack_lvl+0x45/0x5d panic+0x19e/0x349 end_report.part.0+0x54/0x7c kasan_report.cold+0xa/0xf rxe_mr_cleanup+0x9d/0xf0 [rdma_rxe] __rxe_cleanup+0x10a/0x1e0 [rdma_rxe] rxe_reg_user_mr+0xb7/0xd0 [rdma_rxe] ib_uverbs_reg_mr+0x26a/0x480 [ib_uverbs] ib_uverbs_handler_UVERBS_METHOD_INVOKE_WRITE+0x1a2/0x250 [ib_uverbs] ib_uverbs_cmd_verbs+0x1397/0x15a0 [ib_uverbs] This issue was firstly exposed since commit b18c7da6 ("RDMA/rxe: Fix memory leak in error path code") and then we fixed it in commit 8ff5f5d9 ("RDMA/rxe: Prevent double freeing rxe_map_set()") but this fix was reverted together at last by commit 1e755506 (Revert "RDMA/rxe: Create duplicate mapping tables for FMRs") Simply let rxe_mr_cleanup() always handle freeing the mr->map once it is successfully allocated. Fixes: 1e755506 ("Revert "RDMA/rxe: Create duplicate mapping tables for FMRs"") Link: https://lore.kernel.org/r/1667099073-2-1-git-send-email-lizhijian@fujitsu.comSigned-off-by:
-
- 18 Nov, 2022 1 commit
-
-
Zhu Yanjun authored
The rdma_rxe driver does not actually support the reliable datagram transport but contains a variable with RD opcodes in driver code. And this variable is never used. So remove it. Link: https://lore.kernel.org/r/20221112023537.432912-1-yanjun.zhu@intel.comSigned-off-by:
Zhu Yanjun <yanjun.zhu@linux.dev> Signed-off-by:
-
