]> git.ipfire.org Git - thirdparty/kernel/stable.git/commit
nvme: multipath: enable BLK_FEAT_ATOMIC_WRITES for multipathing
authorAlan Adamson <alan.adamson@oracle.com>
Thu, 8 May 2025 22:38:00 +0000 (15:38 -0700)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Wed, 4 Jun 2025 12:45:10 +0000 (14:45 +0200)
commit4235aa3629dc1ea851fb02a1cdfb0f91738b98e3
treeb11bd4a765a12f8e985db7e6c3cdd39570a78a02
parentc53f23f7075c9f63f14d7ec8f2cc3e33e118d986
nvme: multipath: enable BLK_FEAT_ATOMIC_WRITES for multipathing

[ Upstream commit a21675ee3b1ba094e229ae4cd8bddf7d215ab1b9 ]

A change to QEMU resulted in all nvme controllers (single and
multi-controller subsystems) to have its CMIC.MCTRS bit set which
indicates the subsystem supports multiple controllers and it is possible
a namespace can be shared between those multiple controllers in a
multipath configuration.

When a namespace of a CMIC.MCTRS enabled subsystem is allocated, a
multipath node is created.  The queue limits for this node are inherited
from the namespace being allocated. When inheriting queue limits, the
features being inherited need to be specified. The atomic write feature
(BLK_FEAT_ATOMIC_WRITES) was not specified so the atomic queue limits
were not inherited by the multipath disk node which resulted in the sysfs
atomic write attributes being zeroed. The fix is to include
BLK_FEAT_ATOMIC_WRITES in the list of features to be inherited.

Signed-off-by: Alan Adamson <alan.adamson@oracle.com>
Reviewed-by: John Garry <john.g.garry@oracle.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>
drivers/nvme/host/multipath.c