On Thu, Jan 26, 2023 at 06:21:43AM +0100, Christoph Hellwig wrote: > On Wed, Jan 25, 2023 at 04:51:16PM -0500, Josef Bacik wrote: > > This is causing a panic in btrfs/125 because you set bbio to > > btrfs_bio(split_bio), which has a NULL end_io. You need something like the > > following so that we're ending the correct bbio. Thanks, > > Just curious, what are the other configuration details as I've never been > able to hit it? > > The fix itself looks good. I reproduced it on the CI setup I've got for us, this was the config [btrfs_normal_freespacetree] TEST_DIR=/mnt/test TEST_DEV=/dev/mapper/vg0-lv0 SCRATCH_DEV_POOL="/dev/mapper/vg0-lv7 /dev/mapper/vg0-lv6 /dev/mapper/vg0-lv5 /dev/mapper/vg0-lv4 /dev/mapper/vg0-lv3 /dev/mapper/vg0-lv2 /dev/mapper/vg0-lv1 " SCRATCH_MNT=/mnt/scratch LOGWRITES_DEV=/dev/mapper/vg0-lv8 PERF_CONFIGNAME=jbacik MKFS_OPTIONS="-K -f -O ^no-holes" MOUNT_OPTIONS="-o space_cache=v2" FSTYP=btrfs I actually hadn't been running 125 because it wasn't in the auto group, Dave noticed it, I just tried it on this VM and hit it right away. No worries, that's why we have the CI stuff, sometimes it just doesn't trigger for us but will trigger with the CI setup. Thanks, Josef
