This is a note to let you know that I've just added the patch titled RDMA/bnxt_re: Enhance the existing functions that wait for FW responses to the 6.4-stable tree which can be found at: http://www.kernel.org/git/?p=linux/kernel/git/stable/stable-queue.git;a=summary The filename of the patch is: rdma-bnxt_re-enhance-the-existing-functions-that-wai.patch and it can be found in the queue-6.4 subdirectory. If you, or anyone else, feels it should not be added to the stable tree, please let <stable@xxxxxxxxxxxxxxx> know about it. commit c7423534ba555e7cd7823b22a256536332186d1a Author: Kashyap Desai <kashyap.desai@xxxxxxxxxxxx> Date: Fri Jun 9 04:01:42 2023 -0700 RDMA/bnxt_re: Enhance the existing functions that wait for FW responses [ Upstream commit 8cf1d12ad56beb73d2439ccf334b7148e71de58e ] Use jiffies based timewait instead of counting iteration for commands that block for FW response. Also add a poll routine for control path commands. This is for polling completion if the waiting commands timeout. This avoids cases where the driver misses completion interrupts. Signed-off-by: Kashyap Desai <kashyap.desai@xxxxxxxxxxxx> Signed-off-by: Selvin Xavier <selvin.xavier@xxxxxxxxxxxx> Link: https://lore.kernel.org/r/1686308514-11996-6-git-send-email-selvin.xavier@xxxxxxxxxxxx Signed-off-by: Leon Romanovsky <leon@xxxxxxxxxx> Stable-dep-of: 29900bf351e1 ("RDMA/bnxt_re: Fix hang during driver unload") Signed-off-by: Sasha Levin <sashal@xxxxxxxxxx> diff --git a/drivers/infiniband/hw/bnxt_re/qplib_rcfw.c b/drivers/infiniband/hw/bnxt_re/qplib_rcfw.c index c11b8e708844c..918e588588885 100644 --- a/drivers/infiniband/hw/bnxt_re/qplib_rcfw.c +++ b/drivers/infiniband/hw/bnxt_re/qplib_rcfw.c @@ -53,37 +53,74 @@ static void bnxt_qplib_service_creq(struct tasklet_struct *t); -/* Hardware communication channel */ +/** + * __wait_for_resp - Don't hold the cpu context and wait for response + * @rcfw - rcfw channel instance of rdev + * @cookie - cookie to track the command + * + * Wait for command completion in sleepable context. + * + * Returns: + * 0 if command is completed by firmware. + * Non zero error code for rest of the case. + */ static int __wait_for_resp(struct bnxt_qplib_rcfw *rcfw, u16 cookie) { struct bnxt_qplib_cmdq_ctx *cmdq; u16 cbit; - int rc; + int ret; cmdq = &rcfw->cmdq; cbit = cookie % rcfw->cmdq_depth; - rc = wait_event_timeout(cmdq->waitq, - !test_bit(cbit, cmdq->cmdq_bitmap), - msecs_to_jiffies(RCFW_CMD_WAIT_TIME_MS)); - return rc ? 0 : -ETIMEDOUT; + + do { + /* Non zero means command completed */ + ret = wait_event_timeout(cmdq->waitq, + !test_bit(cbit, cmdq->cmdq_bitmap), + msecs_to_jiffies(10000)); + + if (!test_bit(cbit, cmdq->cmdq_bitmap)) + return 0; + + bnxt_qplib_service_creq(&rcfw->creq.creq_tasklet); + + if (!test_bit(cbit, cmdq->cmdq_bitmap)) + return 0; + + } while (true); }; +/** + * __block_for_resp - hold the cpu context and wait for response + * @rcfw - rcfw channel instance of rdev + * @cookie - cookie to track the command + * + * This function will hold the cpu (non-sleepable context) and + * wait for command completion. Maximum holding interval is 8 second. + * + * Returns: + * -ETIMEOUT if command is not completed in specific time interval. + * 0 if command is completed by firmware. + */ static int __block_for_resp(struct bnxt_qplib_rcfw *rcfw, u16 cookie) { - u32 count = RCFW_BLOCKED_CMD_WAIT_COUNT; - struct bnxt_qplib_cmdq_ctx *cmdq; + struct bnxt_qplib_cmdq_ctx *cmdq = &rcfw->cmdq; + unsigned long issue_time = 0; u16 cbit; - cmdq = &rcfw->cmdq; cbit = cookie % rcfw->cmdq_depth; - if (!test_bit(cbit, cmdq->cmdq_bitmap)) - goto done; + issue_time = jiffies; + do { udelay(1); + bnxt_qplib_service_creq(&rcfw->creq.creq_tasklet); - } while (test_bit(cbit, cmdq->cmdq_bitmap) && --count); -done: - return count ? 0 : -ETIMEDOUT; + if (!test_bit(cbit, cmdq->cmdq_bitmap)) + return 0; + + } while (time_before(jiffies, issue_time + (8 * HZ))); + + return -ETIMEDOUT; }; static int __send_message(struct bnxt_qplib_rcfw *rcfw,