| Message ID | x49610gq180.fsf@segfault.boston.devel.redhat.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Wed, Dec 02, 2015 at 11:11:59AM -0500, Jeff Moyer wrote: > Hi, > > In testing with MOUNT_OPTIONS="-o dax" ./check ..., I ran into issues > with every single test that uses dmflakey. The problem is that the > mount will fail, since we're layering a stacking driver on top of the > pmem device, and that stacking driver does not support direct access. > To fix that, I decided to filter out the dax mount option from dmflakey > mounts. If there are suggestions for a better approach, I'm all ears. > > Signed-off-by: Jeff Moyer <jmoyer@redhat.com> These tests don't run on a ram disk because of the _require_sane_bdev_flush() check in each of these dm-flakey tests and so none of my DAX testing has tripped over this... Because dm-flakey will not alow you to use DAX, there's not much point in running these tests if the DAX mount option is enabled. Hence I'd just add this to common/dm-flakey: echo $MOUNT_OPTIONS | grep -q dax if [ $? -ne 0 ]; then _notrun "Cannot run tests with DAX on dmflakey devices" fi That way we have a clear reminder that these tests do not run with DAX enabled and hence we are not doing crash integrity testing using dm-flakey on DAX devices.... Cheers, Dave.
Dave Chinner <david@fromorbit.com> writes: > Because dm-flakey will not alow you to use DAX, there's not much > point in running these tests if the DAX mount option is enabled. > Hence I'd just add this to common/dm-flakey: > > echo $MOUNT_OPTIONS | grep -q dax > if [ $? -ne 0 ]; then > _notrun "Cannot run tests with DAX on dmflakey devices" > fi > > That way we have a clear reminder that these tests do not run with > DAX enabled and hence we are not doing crash integrity testing using > dm-flakey on DAX devices.... Yes, that makes more sense. Or, I guess I could add dax support to dm-flakey. -Jeff -- To unsubscribe from this list: send the line "unsubscribe fstests" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
diff --git a/common/dmflakey b/common/dmflakey index e55aefd..693d231 100644 --- a/common/dmflakey +++ b/common/dmflakey @@ -21,6 +21,47 @@ FLAKEY_ALLOW_WRITES=0 FLAKEY_DROP_WRITES=1 +_filter_mount_opt() +{ + local opts='' + local old_opts=$1 + local filter=$2 + + # If there are no options, or if the filter isn't present, + # just return. + if [ -z "$old_opts" ]; then + echo -n "" + return + fi + + echo $old_opts | grep -q $filter + if [ $? -ne 0 ]; then + echo -n $old_opts + return + fi + + # -o, --options opts + # Options are specified with a -o flag followed by a comma sepa- + # rated string of options. For example: + # mount LABEL=mydisk -o noatime,nouser + # + # There are several cases we have to handle: + # -o dax; -o,dax + # -o,ro,dax; -o,dax,ro + # -o,ro,dax,discard + # + opts=$(echo $old_opts | sed -e "s/$filter//g") + opts=$(echo $opts | sed -e 's/,,/,/g') + opts=$(echo $opts | sed -e 's/ ,/ /g') + # get rid of trailing whitespace or comma + opts=$(echo $opts | sed -e 's/[[:space:]]$//g') + opts=$(echo $opts | sed -e 's/,$//g') + # finally, if there's nothing left, get rid of the -o + opts=$(echo $opts | sed -e 's/-o$//g') + + echo -n $opts +} + _init_flakey() { local BLK_DEV_SIZE=`blockdev --getsz $SCRATCH_DEV` @@ -34,7 +75,8 @@ _init_flakey() _mount_flakey() { - mount -t $FSTYP $MOUNT_OPTIONS $FLAKEY_DEV $SCRATCH_MNT + mount_options=$(_filter_mount_opt "$MOUNT_OPTIONS" dax) + mount -t $FSTYP $mount_options $FLAKEY_DEV $SCRATCH_MNT } _unmount_flakey()
Hi, In testing with MOUNT_OPTIONS="-o dax" ./check ..., I ran into issues with every single test that uses dmflakey. The problem is that the mount will fail, since we're layering a stacking driver on top of the pmem device, and that stacking driver does not support direct access. To fix that, I decided to filter out the dax mount option from dmflakey mounts. If there are suggestions for a better approach, I'm all ears. Signed-off-by: Jeff Moyer <jmoyer@redhat.com> -- To unsubscribe from this list: send the line "unsubscribe fstests" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html