Skip to contentSkip to conversation

nvme-tcp: fixup I/O stall on congested sockets

PATCH In review Hannes Reinecke 3 participants 6 messages Apr 29, 2025 NVMe

lore
Share & tools
lore
#0 lore raw

v5 cover nvme-tcp: fixup I/O stall on congested sockets

Hi all,

I have been chasing keep-alive timeouts with TLS enabled in the last few weeks (monthsA, even ...). On larger setups (eg with 32 queues) the connection never got established properly as I've been hitting keep-alive timeouts before the last queue got connected.
Turns out that occasionally we simply do not send the keep-alive request; it's been added to the request list but the io_work workqueue function is never restarted as it bails out after nvme_tcp_try_recv() returns -EAGAIN. During debugging I also found that we're quite lazy with the list handling of requests, so I've added a patche to ensure that all list elements are properly terminated.

As usual, comments and reviews are welcome.

Changes to v4:
- Drop check for 'queue->req' as noticed by Sagi

Changes to v3:
- Drop first patch as it already had been applied
- Include reviews from Sagi
- Check for sk_sock_is_writeable() to avoid requeing io_work when the socket is blocked

Changes to v2:
- Removed AEN patches again

Changes to the original submission:
- Include reviews from Chris Leech
- Add patch to requeue namespace scan
- Add patch to re-read ANA log page

Hannes Reinecke (2):
nvme-tcp: sanitize request list handling
nvme-tcp: fix I/O stalls on congested sockets

 drivers/nvme/host/tcp.c | 22 ++++++++++++++++++++--
 1 file changed, 20 insertions(+), 2 deletions(-)
Signature · 1 line
2.35.3
#3 lore raw

PATCH v5 2/2 nvme-tcp: fix I/O stalls on congested sockets

When the socket is busy processing nvme_tcp_try_recv() might return -EAGAIN, but this doesn't automatically imply that the sending side is blocked, too.
So check if there are pending requests once nvme_tcp_try_recv() returns -EAGAIN and continue with the sending loop to avoid I/O stalls.

---
 drivers/nvme/host/tcp.c | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)
drivers/nvme/host/tcp.c +6 -1
index 1c39a307456c..eb4222c5a77a 100644
@@ -1361,7 +1361,7 @@ static int nvme_tcp_try_recv(struct nvme_tcp_queue *queue)
 	queue->nr_cqe = 0;
 	consumed = sock->ops->read_sock(sk, &rd_desc, nvme_tcp_recv_skb);
 	release_sock(sk);
-	return consumed;
+	return consumed == -EAGAIN ? 0 : consumed;
 }
 
 static void nvme_tcp_io_work(struct work_struct *w)
@@ -1389,6 +1389,11 @@ static void nvme_tcp_io_work(struct work_struct *w)
 		else if (unlikely(result < 0))
 			return;
 
+		/* did we get some space after spending time in recv? */
+		if (nvme_tcp_queue_has_pending(queue) &&
+		    sk_stream_is_writeable(queue->sock->sk))
+			pending = true;
+
 		if (!pending || !queue->rd_enabled)
 			return;
 
Signature · 1 line
2.35.3
#1 lore raw

PATCH v5 1/2 nvme-tcp: sanitize request list handling

In reply to Hannes Reinecke

Validate the request in nvme_tcp_handle_r2t() to ensure it's not part of any list, otherwise a malicious R2T PDU might inject a loop in request list processing.

---
 drivers/nvme/host/tcp.c | 15 ++++++++++++++-
 1 file changed, 14 insertions(+), 1 deletion(-)
drivers/nvme/host/tcp.c +14 -1
index a9d455c39652..1c39a307456c 100644
@@ -452,7 +452,8 @@ nvme_tcp_fetch_request(struct nvme_tcp_queue *queue)
 			return NULL;
 	}
 
-	list_del(&req->entry);
+	list_del_init(&req->entry);
+	init_llist_node(&req->lentry);
 	return req;
 }
 
@@ -560,6 +561,8 @@ static int nvme_tcp_init_request(struct blk_mq_tag_set *set,
 	req->queue = queue;
 	nvme_req(rq)->ctrl = &ctrl->ctrl;
 	nvme_req(rq)->cmd = &pdu->cmd;
+	init_llist_node(&req->lentry);
+	INIT_LIST_HEAD(&req->entry);
 
 	return 0;
 }
@@ -764,6 +767,14 @@ static int nvme_tcp_handle_r2t(struct nvme_tcp_queue *queue,
 		return -EPROTO;
 	}
 
+	if (llist_on_list(&req->lentry) ||
+	    !list_empty(&req->entry)) {
+		dev_err(queue->ctrl->ctrl.device,
+			"req %d unexpected r2t while processing request\n",
+			rq->tag);
+		return -EPROTO;
+	}
+
 	req->pdu_len = 0;
 	req->h2cdata_left = r2t_length;
 	req->h2cdata_offset = r2t_offset;
@@ -2611,6 +2622,8 @@ static void nvme_tcp_submit_async_event(struct nvme_ctrl *arg)
 	ctrl->async_req.offset = 0;
 	ctrl->async_req.curr_bio = NULL;
 	ctrl->async_req.data_len = 0;
+	init_llist_node(&ctrl->async_req.lentry);
+	INIT_LIST_HEAD(&ctrl->async_req.entry);
 
 	nvme_tcp_queue_request(&ctrl->async_req, true);
 }
Signature · 1 line
2.35.3
#5 lore raw

Re: [PATCH 2/2] nvme-tcp: fix I/O stalls on congested sockets

Quoted 3 lines · Sagi Grimberg

I want kamaljit to give his Tested-by tag.

Sure.
I just wanted to push out the patches to give him a clean state for his testing.

Cheers,

Hannes

Signature · 4 lines
Dr. Hannes Reinecke                  Kernel Storage Architect
[email protected]                                +49 911 74053 688
SUSE Software Solutions GmbH, Frankenstr. 146, 90461 Nürnberg
HRB 36809 (AG Nürnberg), GF: I. Totev, A. McDonald, W. Knoblich