gluster: Align block-status tail
commite24154d878c11688202628755f8a47e3cba0f52a
authorMax Reitz <mreitz@redhat.com>
Thu, 5 Aug 2021 14:36:03 +0000 (5 16:36 +0200)
committerHanna Reitz <hreitz@redhat.com>
Wed, 15 Sep 2021 13:54:06 +0000 (15 15:54 +0200)
tree13a2bbb9a451827cb6f098aeaf65337395d0d8bd
parent0b6206b9c6825619cd721085fe082d7a0abc9af4
gluster: Align block-status tail

gluster's block-status implementation is basically a copy of that in
block/file-posix.c, there is only one thing missing, and that is
aligning trailing data extents to the request alignment (as added by
commit 9c3db310ff0).

Note that 9c3db310ff0 mentions that "there seems to be no other block
driver that sets request_alignment and [...]", but while block/gluster.c
does indeed not set request_alignment, block/io.c's
bdrv_refresh_limits() will still default to an alignment of 512 because
block/gluster.c does not provide a byte-aligned read function.
Therefore, unaligned tails can conceivably occur, and so we should apply
the change from 9c3db310ff0 to gluster's block-status implementation.

Reported-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Signed-off-by: Max Reitz <mreitz@redhat.com>
Message-Id: <20210805143603.59503-1-mreitz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Hanna Reitz <hreitz@redhat.com>
block/gluster.c