dm thin: update discard_granularity to reflect the thin-pool blocksize
authorLukas Czerner <lczerner@redhat.com>
Wed, 11 Jun 2014 16:28:43 +0000 (12:28 -0400)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Wed, 9 Jul 2014 18:18:26 +0000 (11:18 -0700)
commitdd7ba80f4a3933c025e6adcf5f5d730181d91ab0
treebc09ed28f4b67ebc944c932976dd71e889e908c3
parente4b08895ac3f0b5c5eae0f33c76a93a617cdfb63
dm thin: update discard_granularity to reflect the thin-pool blocksize

commit 09869de57ed2728ae3c619803932a86cb0e2c4f8 upstream.

DM thinp already checks whether the discard_granularity of the data
device is a factor of the thin-pool block size.  But when using the
dm-thin-pool's discard passdown support, DM thinp was not selecting the
max of the underlying data device's discard_granularity and the
thin-pool's block size.

Update set_discard_limits() to set discard_granularity to the max of
these values.  This enables blkdev_issue_discard() to properly align the
discards that are sent to the DM thin device on a full block boundary.
As such each discard will now cover an entire DM thin-pool block and the
block will be reclaimed.

Reported-by: Zdenek Kabelac <zkabelac@redhat.com>
Signed-off-by: Lukas Czerner <lczerner@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
drivers/md/dm-thin.c