diff options
author | NeilBrown <neilb@suse.de> | 2010-08-16 18:09:31 +1000 |
---|---|---|
committer | NeilBrown <neilb@suse.de> | 2010-08-18 11:39:38 +1000 |
commit | 3a3a5ddb7a0f43c3dd0f98673f3d930a456725f8 (patch) | |
tree | fa1e33c011c9f6350ce5118b89deafacbb1e0f2b /drivers/usb | |
parent | da5cabf80e2433131bf0ed8993abc0f7ea618c73 (diff) | |
download | op-kernel-dev-3a3a5ddb7a0f43c3dd0f98673f3d930a456725f8.zip op-kernel-dev-3a3a5ddb7a0f43c3dd0f98673f3d930a456725f8.tar.gz |
Update recovery_offset even when external metadata is used.
The update of ->recovery_offset in sync_sbs is appropriate even then external
metadata is in use. However sync_sbs is only called when native
metadata is used.
So move that update in to the top of md_update_sb (which is the only
caller of sync_sbs) before the test on ->external.
This moves the update out of ->write_lock protection, but those fields
only need ->reconfig_mutex protection which they still have.
Also move the test on ->persistent up to where ->external is set as
for metadata update purposes they are the same.
Clear MD_CHANGE_DEVS and MD_CHANGE_CLEAN as they can only be confusing
if ->external is set or ->persistent isn't.
Finally move the update of ->utime down as it is only relevent (like
the ->events update) for native metadata.
Signed-off-by: NeilBrown <neilb@suse.de>
Reported-by: "Kwolek, Adam" <adam.kwolek@intel.com>
Diffstat (limited to 'drivers/usb')
0 files changed, 0 insertions, 0 deletions