git: 8d6c0743e36e - main - nvme: Introduce longer timeouts for admin queue

The branch main has been updated by mav:

URL: https://cgit.FreeBSD.org/src/commit/?id=8d6c0743e36e3cff9279c40468711a82db98df23

commit 8d6c0743e36e3cff9279c40468711a82db98df23
Author:     Alexander Motin <mav@FreeBSD.org>
AuthorDate: 2023-11-06 16:05:48 +0000
Commit:     Alexander Motin <mav@FreeBSD.org>
CommitDate: 2023-11-06 16:05:48 +0000

    nvme: Introduce longer timeouts for admin queue
    
    KIOXIA CD8 SSDs routinely take ~25 seconds to delete non-empty
    namespace.  In some cases like hot-plug it takes longer, triggering
    timeout and controller resets after just 30 seconds. Linux for many
    years has separate 60 seconds timeout for admin queue.  This patch
    does the same.  And it is good to be consistent.
    
    Sponsored by:   iXsystems, Inc.
    Reviewed by:    imp
    MFC after:      1 week
    Differential Revision:  https://reviews.freebsd.org/D42454
---
 sys/dev/nvme/nvme_ctrlr.c   |  6 ++++++
 sys/dev/nvme/nvme_private.h |  2 ++
 sys/dev/nvme/nvme_qpair.c   |  2 ++
 sys/dev/nvme/nvme_sysctl.c  | 15 ++++++++++-----
 4 files changed, 20 insertions(+), 5 deletions(-)

diff --git a/sys/dev/nvme/nvme_ctrlr.c b/sys/dev/nvme/nvme_ctrlr.c
index 90cd0b64a980..b72359a061da 100644
--- a/sys/dev/nvme/nvme_ctrlr.c
+++ b/sys/dev/nvme/nvme_ctrlr.c
@@ -1405,6 +1405,12 @@ nvme_ctrlr_construct(struct nvme_controller *ctrlr, device_t dev)
 	to = NVME_CAP_LO_TO(cap_lo) + 1;
 	ctrlr->ready_timeout_in_ms = to * 500;
 
+	timeout_period = NVME_ADMIN_TIMEOUT_PERIOD;
+	TUNABLE_INT_FETCH("hw.nvme.admin_timeout_period", &timeout_period);
+	timeout_period = min(timeout_period, NVME_MAX_TIMEOUT_PERIOD);
+	timeout_period = max(timeout_period, NVME_MIN_TIMEOUT_PERIOD);
+	ctrlr->admin_timeout_period = timeout_period;
+
 	timeout_period = NVME_DEFAULT_TIMEOUT_PERIOD;
 	TUNABLE_INT_FETCH("hw.nvme.timeout_period", &timeout_period);
 	timeout_period = min(timeout_period, NVME_MAX_TIMEOUT_PERIOD);
diff --git a/sys/dev/nvme/nvme_private.h b/sys/dev/nvme/nvme_private.h
index f6ad51795adb..1b6ea03c909a 100644
--- a/sys/dev/nvme/nvme_private.h
+++ b/sys/dev/nvme/nvme_private.h
@@ -86,6 +86,7 @@ MALLOC_DECLARE(M_NVME);
 #define NVME_MAX_CONSUMERS	(2)
 #define NVME_MAX_ASYNC_EVENTS	(8)
 
+#define NVME_ADMIN_TIMEOUT_PERIOD	(60)    /* in seconds */
 #define NVME_DEFAULT_TIMEOUT_PERIOD	(30)    /* in seconds */
 #define NVME_MIN_TIMEOUT_PERIOD		(5)
 #define NVME_MAX_TIMEOUT_PERIOD		(120)
@@ -279,6 +280,7 @@ struct nvme_controller {
 	uint32_t		int_coal_threshold;
 
 	/** timeout period in seconds */
+	uint32_t		admin_timeout_period;
 	uint32_t		timeout_period;
 
 	/** doorbell stride */
diff --git a/sys/dev/nvme/nvme_qpair.c b/sys/dev/nvme/nvme_qpair.c
index 569d54ab6aef..7345b1b59be5 100644
--- a/sys/dev/nvme/nvme_qpair.c
+++ b/sys/dev/nvme/nvme_qpair.c
@@ -1181,6 +1181,8 @@ nvme_qpair_submit_tracker(struct nvme_qpair *qpair, struct nvme_tracker *tr)
 	if (req->timeout) {
 		if (req->cb_fn == nvme_completion_poll_cb)
 			timeout = 1;
+		else if (qpair->id == 0)
+			timeout = ctrlr->admin_timeout_period;
 		else
 			timeout = ctrlr->timeout_period;
 		tr->deadline = getsbinuptime() + timeout * SBT_1S;
diff --git a/sys/dev/nvme/nvme_sysctl.c b/sys/dev/nvme/nvme_sysctl.c
index ac0d507e2337..d6452a2e5492 100644
--- a/sys/dev/nvme/nvme_sysctl.c
+++ b/sys/dev/nvme/nvme_sysctl.c
@@ -132,8 +132,8 @@ nvme_sysctl_int_coal_threshold(SYSCTL_HANDLER_ARGS)
 static int
 nvme_sysctl_timeout_period(SYSCTL_HANDLER_ARGS)
 {
-	struct nvme_controller *ctrlr = arg1;
-	uint32_t newval = ctrlr->timeout_period;
+	uint32_t *ptr = arg1;
+	uint32_t newval = *ptr;
 	int error = sysctl_handle_int(oidp, &newval, 0, req);
 
 	if (error || (req->newptr == NULL))
@@ -143,7 +143,7 @@ nvme_sysctl_timeout_period(SYSCTL_HANDLER_ARGS)
 	    newval < NVME_MIN_TIMEOUT_PERIOD) {
 		return (EINVAL);
 	} else {
-		ctrlr->timeout_period = newval;
+		*ptr = newval;
 	}
 
 	return (0);
@@ -352,10 +352,15 @@ nvme_sysctl_initialize_ctrlr(struct nvme_controller *ctrlr)
 	    nvme_sysctl_int_coal_threshold, "IU",
 	    "Interrupt coalescing threshold");
 
+	SYSCTL_ADD_PROC(ctrlr_ctx, ctrlr_list, OID_AUTO,
+	    "admin_timeout_period", CTLTYPE_UINT | CTLFLAG_RW | CTLFLAG_MPSAFE,
+	    &ctrlr->admin_timeout_period, 0, nvme_sysctl_timeout_period, "IU",
+	    "Timeout period for Admin queue (in seconds)");
+
 	SYSCTL_ADD_PROC(ctrlr_ctx, ctrlr_list, OID_AUTO,
 	    "timeout_period", CTLTYPE_UINT | CTLFLAG_RW | CTLFLAG_MPSAFE,
-	    ctrlr, 0, nvme_sysctl_timeout_period, "IU",
-	    "Timeout period (in seconds)");
+	    &ctrlr->timeout_period, 0, nvme_sysctl_timeout_period, "IU",
+	    "Timeout period for I/O queues (in seconds)");
 
 	SYSCTL_ADD_PROC(ctrlr_ctx, ctrlr_list, OID_AUTO,
 	    "num_cmds", CTLTYPE_S64 | CTLFLAG_RD | CTLFLAG_MPSAFE,