Hello, I have a broken btrfs file system and as far as I know, you're interested in bug reports since btrfs is still in development. If not, let me know. First, basic system information: OS: Arch Linux Kernel: 3.8.11 System layout: / btrfs /boot ext2 /home btrfs on LUKS subvolumes of /: btrfs subvolume list -a / ID 256 gen 2629 top level 5 path <FS_TREE>/__active ID 282 gen 2626 top level 5 path <FS_TREE>/__active/__backup ID 285 gen 5494 top level 5 path <FS_TREE>/__active/__restore I upgraded my system and installed linux 3.8.11 last week. Before the upgrade, I took a snapshot of / to be able to roll back. After upgrading, my system wouldn't reboot. I booted a live system, copied my ro snapshot to a new subvolume and set it as default. I upgraded again, which seemed to work fine. After booting an LTS kernel (3.0.75) the system wouldn't boot anymore. So basically, I have two broken snapshots with different errors: The original snapshot (__active) doesn't mount anymore, neither during boot nor from a live system. I can mount it ro though. I get the following error when I mount with -o recovery,rw: [ 233.836936] kernel BUG at fs/btrfs/extent-tree.c:5234! [ 233.837834] invalid opcode: 0000 [#1] PREEMPT SMP ... [ 233.882508] BUG: scheduling while atomic: ls/608/0x10000003 (Full dmesg see attachment dmesg.mount-with-recovery,rw,subvolid=5) When I boot (and mount #285) I get the following error: RIP: __write_lock_failed+0xd0x20 (note that I only have a screenshot because the system freezes). If I mount from a live system, I don't get any errors. The bug seems somewhat reproducible because it happened twice after uprading to 3.8.11. I hope this bug report is helpful in a way, please let me know if you need additional information. If not, I will play back a backup of the partition, so I'm not too worried about data recovery. Any hints are still appreciated. Greetings
Attachment:
dmesg.mount-with-recovery,rw,subvolid=5
Description: Binary data
