Linux-NVME Archive on lore.kernel.org
 help / color / mirror / Atom feed
From: Hannes Reinecke <hare@kernel.org>
To: Christoph Hellwig <hch@lst.de>
Cc: Sagi Grimberg <sagi@grimberg.me>, Keith Busch <kbusch@kernel.org>,
	linux-nvme@lists.infradead.org, Hannes Reinecke <hare@suse.de>,
	Hannes Reinecke <hare@kernel.org>
Subject: [PATCH 2/8] nvme-tcp: io_work stall debugging
Date: Tue, 16 Jul 2024 09:36:10 +0200	[thread overview]
Message-ID: <20240716073616.84417-3-hare@kernel.org> (raw)
In-Reply-To: <20240716073616.84417-1-hare@kernel.org>

From: Hannes Reinecke <hare@suse.de>

Add a debug message when the io workqueue exceeds the latency target
given by 'deadline'.

Signed-off-by: Hannes Reinecke <hare@kernel.org>
---
 drivers/nvme/host/tcp.c | 20 ++++++++++++++++----
 1 file changed, 16 insertions(+), 4 deletions(-)

diff --git a/drivers/nvme/host/tcp.c b/drivers/nvme/host/tcp.c
index 3cf9a9abb0e0..7876bf7d2fac 100644
--- a/drivers/nvme/host/tcp.c
+++ b/drivers/nvme/host/tcp.c
@@ -1282,13 +1282,14 @@ static void nvme_tcp_io_work(struct work_struct *w)
 {
 	struct nvme_tcp_queue *queue =
 		container_of(w, struct nvme_tcp_queue, io_work);
-	u64 start = ktime_to_us(ktime_get());
+	u64 start = ktime_to_us(ktime_get()), overrun;
 	u64 tx_deadline = start + deadline;
+	bool pending = false;
 
 	do {
-		bool pending = false;
 		int result;
 
+		pending = false;
 		if (mutex_trylock(&queue->send_mutex)) {
 			result = nvme_tcp_try_send(queue);
 			mutex_unlock(&queue->send_mutex);
@@ -1304,12 +1305,23 @@ static void nvme_tcp_io_work(struct work_struct *w)
 		else if (unlikely(result < 0))
 			return;
 
-		if (!pending || !queue->rd_enabled)
+		if (!queue->rd_enabled)
 			return;
+		if (!pending)
+			goto check;
 
 	} while (ktime_to_us(ktime_get()) < tx_deadline); /* quota is exhausted */
 
-	queue_work_on(queue->io_cpu, nvme_tcp_wq, &queue->io_work);
+check:
+	overrun = ktime_to_us(ktime_get()) - start;
+	if (overrun > 10 * deadline) {
+		dev_dbg(queue->ctrl->ctrl.device,
+			"queue %d: stall (%llu msecs)%s%s\n",
+			nvme_tcp_queue_id(queue), div_u64(overrun, 1000),
+			list_empty(&queue->send_list) ? " empty" : "", queue->request ? " pending" : "");
+	}
+	if (pending)
+		queue_work_on(queue->io_cpu, nvme_tcp_wq, &queue->io_work);
 }
 
 static void nvme_tcp_free_crypto(struct nvme_tcp_queue *queue)
-- 
2.35.3



  parent reply	other threads:[~2024-07-16  7:36 UTC|newest]

Thread overview: 24+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-07-16  7:36 [PATCHv3 0/8] nvme-tcp: improve scalability Hannes Reinecke
2024-07-16  7:36 ` [PATCH 1/8] nvme-tcp: switch TX deadline to microseconds and make it configurable Hannes Reinecke
2024-07-17 21:03   ` Sagi Grimberg
2024-07-18  6:30     ` Hannes Reinecke
2024-07-16  7:36 ` Hannes Reinecke [this message]
2024-07-17 21:05   ` [PATCH 2/8] nvme-tcp: io_work stall debugging Sagi Grimberg
2024-07-16  7:36 ` [PATCH 3/8] nvme-tcp: re-init request list entries Hannes Reinecke
2024-07-17 21:23   ` Sagi Grimberg
2024-07-16  7:36 ` [PATCH 4/8] nvme-tcp: improve stall debugging Hannes Reinecke
2024-07-17 21:11   ` Sagi Grimberg
2024-07-16  7:36 ` [PATCH 5/8] nvme-tcp: debugfs entries for latency statistics Hannes Reinecke
2024-07-17 21:14   ` Sagi Grimberg
2024-07-16  7:36 ` [PATCH 6/8] nvme-tcp: reduce callback lock contention Hannes Reinecke
2024-07-17 21:19   ` Sagi Grimberg
2024-07-18  6:42     ` Hannes Reinecke
2024-07-21 11:46       ` Sagi Grimberg
2024-07-16  7:36 ` [PATCH 7/8] nvme-tcp: check for SOCK_NOSPACE before sending Hannes Reinecke
2024-07-17 21:19   ` Sagi Grimberg
2024-07-16  7:36 ` [PATCH 8/8] nvme-tcp: align I/O cpu with blk-mq mapping Hannes Reinecke
2024-07-17 21:34   ` Sagi Grimberg
2024-08-13 19:36     ` Sagi Grimberg
2024-07-17 21:01 ` [PATCHv3 0/8] nvme-tcp: improve scalability Sagi Grimberg
2024-07-18  6:20   ` Hannes Reinecke
2024-07-21 12:05     ` Sagi Grimberg

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20240716073616.84417-3-hare@kernel.org \
    --to=hare@kernel.org \
    --cc=hare@suse.de \
    --cc=hch@lst.de \
    --cc=kbusch@kernel.org \
    --cc=linux-nvme@lists.infradead.org \
    --cc=sagi@grimberg.me \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox