5fbc7c59fd22c5a6531b40b0759624b680a95e52 ubuntu_btrfs_kernel_fixes failed with J-oem-6.5

[Po-Hsu Lin]

This issue can be found since 6.5.0-1003.3 when we start testing this kernel with bare-metal resources. (This test was not run on openstack VM)

Test failed with:
$ sudo ./5fbc7c59fd22c5a6531b40b0759624b680a95e52.sh
fix 5fbc7c59fd22c5a6531b40b0759624b680a95e52

    Btrfs: fix unfinished readahead thread for raid5/6 degraded mounting

    Steps to reproduce:

     # mkfs.btrfs -f /dev/sd[b-f] -m raid5 -d raid5
     # mkfs.ext4 /dev/sdc --->corrupt one of btrfs device
     # mount /dev/sdb /mnt -o degraded
     # btrfs scrub start -BRd /mnt

    This is because readahead would skip missing device, this is not true
    for RAID5/6, because REQ_GET_READ_MIRRORS return 1 for RAID5/6 block
    mapping. If expected data locates in missing device, readahead thread
    would not call __readahead_hook() which makes event @rc->elems=0
    wait forever.

    Fix this problem by checking return value of btrfs_map_block(),we
    can only skip missing device safely if there are several mirrors.

mke2fs 1.46.5 (30-Dec-2021)
/dev/loop8 contains a btrfs file system
Discarding device blocks: done
Creating filesystem with 524288 4k blocks and 131072 inodes
Filesystem UUID: 98040e90-cb46-4563-877f-21c2c2c16449
Superblock backups stored on blocks:
 32768, 98304, 163840, 229376, 294912

Allocating group tables: done
Writing inode tables: done
Creating journal (16384 blocks): done
Writing superblocks and filesystem accounting information: done

mount: /dev/loop7: can't find in /etc/fstab.
mount /dev/loop7 on degraded raid5 failed

dmesg output:
[67744.195644] loop7: detected capacity change from 0 to 4194304
[67744.204713] loop8: detected capacity change from 0 to 4194304
[67744.213210] loop9: detected capacity change from 0 to 4194304
[67744.220870] loop10: detected capacity change from 0 to 4194304
[67744.227971] loop11: detected capacity change from 0 to 4194304
[67747.181182] BTRFS: device fsid 140973c6-10a9-4557-9724-f73b8f7c11ed devid 1 transid 6 /dev/loop7 scanned by systemd-udevd (13456)
[67747.222225] BTRFS: device fsid 140973c6-10a9-4557-9724-f73b8f7c11ed devid 2 transid 6 /dev/loop8 scanned by systemd-udevd (13449)
[67747.253234] BTRFS: device fsid 140973c6-10a9-4557-9724-f73b8f7c11ed devid 3 transid 6 /dev/loop9 scanned by systemd-udevd (13456)
[67747.279864] BTRFS: device fsid 140973c6-10a9-4557-9724-f73b8f7c11ed devid 4 transid 6 /dev/loop10 scanned by systemd-udevd (13449)
[67747.294655] BTRFS: device fsid 140973c6-10a9-4557-9724-f73b8f7c11ed devid 5 transid 6 /dev/loop11 scanned by mkfs.btrfs (13448)
[67795.394565] systemd-journald[271]: Sent WATCHDOG=1 notification.
[67867.395505] systemd-journald[271]: Sent WATCHDOG=1 notification.

[Po-Hsu Lin (cypressyew)]

The same failure can be found in Mantic 6.5.0-1005 as well:
ubuntu@autovm4:~/autotest-client-tests/ubuntu_btrfs_kernel_fixes/fixes$ sudo ./5fbc7c59fd22c5a6531b40b0759624b680a95e52.sh
fix 5fbc7c59fd22c5a6531b40b0759624b680a95e52

    Btrfs: fix unfinished readahead thread for raid5/6 degraded mounting

    Steps to reproduce:

     # mkfs.btrfs -f /dev/sd[b-f] -m raid5 -d raid5
     # mkfs.ext4 /dev/sdc --->corrupt one of btrfs device
     # mount /dev/sdb /mnt -o degraded
     # btrfs scrub start -BRd /mnt

    This is because readahead would skip missing device, this is not true
    for RAID5/6, because REQ_GET_READ_MIRRORS return 1 for RAID5/6 block
    mapping. If expected data locates in missing device, readahead thread
    would not call __readahead_hook() which makes event @rc->elems=0
    wait forever.

    Fix this problem by checking return value of btrfs_map_block(),we
    can only skip missing device safely if there are several mirrors.

mke2fs 1.47.0 (5-Feb-2023)
/dev/loop4 contains a btrfs file system
Discarding device blocks: done
Creating filesystem with 524288 4k blocks and 131072 inodes
Filesystem UUID: 2d62988f-a1b2-425d-9dd9-1695390d9317
Superblock backups stored on blocks:
 32768, 98304, 163840, 229376, 294912

Allocating group tables: done
Writing inode tables: done
Creating journal (16384 blocks): done
Writing superblocks and filesystem accounting information: done

mount: /dev/loop3: can't find in /etc/fstab.
mount /dev/loop3 on degraded raid5 failed
ubuntu@autovm4:~/autotest-client-tests/ubuntu_btrfs_kernel_fixes/fixes$ uname -a
Linux autovm4 6.5.0-5-generic #5-Ubuntu SMP PREEMPT_DYNAMIC Wed Sep 6 15:11:07 UTC 2023 x86_64 x86_64 x86_64 GNU/Linux