On 01/21/2016 03:46 PM, Stephen Rothwell wrote:
Hi Jens,
On Thu, 31 Dec 2015 14:34:57 +1100 Stephen Rothwell <s...@canb.auug.org.au>
wrote:
Today's linux-next merge of the block tree got a conflict in:
drivers/nvme/host/pci.c
between commit:
b5875222de2f ("NVMe: IO ending fixes on surprise removal")
from Linus' tree and commit:
5bae7f73d378 ("nvme: move namespace scanning to common code")
from the block tree.
I fixed it up (the code was moved - I added the fix patch below) and
can carry the fix as necessary (no action is required).
However, there was another part to the former patch that I could not
quite figure out how to reproduce - the fix to nvme_dev_remove().
From: Stephen Rothwell <s...@canb.auug.org.au>
Date: Thu, 31 Dec 2015 14:21:38 +1100
Subject: [PATCH] nvme: merge fix up for ns code movement
Signed-off-by: Stephen Rothwell <s...@canb.auug.org.au>
---
drivers/nvme/host/core.c | 11 ++++++++++-
1 file changed, 10 insertions(+), 1 deletion(-)
diff --git a/drivers/nvme/host/core.c b/drivers/nvme/host/core.c
index 1437ff36e91c..1375a83593b5 100644
--- a/drivers/nvme/host/core.c
+++ b/drivers/nvme/host/core.c
@@ -1118,8 +1118,17 @@ static void nvme_ns_remove(struct nvme_ns *ns)
bool kill = nvme_io_incapable(ns->ctrl) &&
!blk_queue_dying(ns->queue);
- if (kill)
+ if (kill) {
blk_set_queue_dying(ns->queue);
+
+ /*
+ * The controller was shutdown first if we got here through
+ * device removal. The shutdown may requeue outstanding
+ * requests. These need to be aborted immediately so
+ * del_gendisk doesn't block indefinitely for their completion.
+ */
+ blk_mq_abort_requeue_list(ns->queue);
+ }
if (ns->disk->flags & GENHD_FL_UP) {
if (blk_get_integrity(ns->disk))
blk_integrity_unregister(ns->disk);
--
2.6.4
So, I have been applying the above merge fix patch since Dec 31 and now
wonder if Linus needs to be told about it. Also noone every replied
about the nvme_dev_remove() part.
Linus is usually pretty damn good at figuring out, and seems to have fun
doing it. So I usually just defer to acking a merge resolution, but even
that is rarely needed. It was more of a mess this time around between
mainline and the nvme branch than I would have liked though, but mostly
due to timing of branching.
--
Jens Axboe
