On Fri, Apr 29, 2022 at 05:31:04PM +0800, Wenpeng Liang wrote:
> From: Yangyang Li <liyangyang20@xxxxxxxxxx>
>
> CMDQ may fail during HNS ROCEE initialization. The following is the log
> when the execution fails:
>
> [ 481.424373] hns3 0000:bd:00.2: In reset process RoCE client reinit.
> [ 482.120830] hns3 0000:bd:00.2: CMDQ move tail from 840 to 839
> [ 482.129220] hns3 0000:bd:00.2 hns_2: failed to set gid, ret = -11!
> [ 482.184702] hns3 0000:bd:00.2: CMDQ move tail from 840 to 839
> <...>
> [ 485.540909] hns3 0000:bd:00.2: CMDQ move tail from 840 to 839
> [ 485.579958] hns3 0000:bd:00.2: CMDQ move tail from 840 to 0
> [ 495.694616] hns3 0000:bd:00.2: [cmd]token 14e mailbox 20 timeout.
> [ 495.700689] hns3 0000:bd:00.2 hns_2: set HEM step 0 failed!
> [ 495.706242] hns3 0000:bd:00.2 hns_2: set HEM address to HW failed!
> [ 495.712412] hns3 0000:bd:00.2 hns_2: failed to alloc mtpt, ret = -16.
> [ 495.718836] infiniband hns_2: Couldn't create ib_mad PD
> [ 495.724046] infiniband hns_2: Couldn't open port 1
> [ 495.729375] hns3 0000:bd:00.2: Reset done, RoCE client reinit finished.
>
> However, even if ib_mad client registration failed, ib_register_device()
> still returns success to the driver.
>
> In the device initialization process, CMDQ execution fails because HW/FW
> is abnormal. Therefore, if CMDQ fails, the initialization function should
> set CMDQ to a fatal error state and return a failure to the caller.
>
> Fixes: 9a4435375cd1 ("IB/hns: Add driver files for hns RoCE driver")
> Signed-off-by: Yangyang Li <liyangyang20@xxxxxxxxxx>
> Signed-off-by: Wenpeng Liang <liangwenpeng@xxxxxxxxxx>
> ---
> drivers/infiniband/hw/hns/hns_roce_device.h | 6 ++++++
> drivers/infiniband/hw/hns/hns_roce_hw_v2.c | 21 +++++++++++++++++++++
> 2 files changed, 27 insertions(+)
Applied to for-next, thanks
Jason
