ACK/Cmnt: [SRU][J][PATCH 0/4] Fix regression in ext4 online resize

[Tim Gardner]

On 1/24/23 5:05 PM, Khalid Elmously wrote:
> BugLink: https://bugs.launchpad.net/bugs/2003816
> 
> A regression was introduced with de394a8665 "ext4: update s_overhead_clusters in the superblock during an on-line resize" in the 5.15 kernels. The issue was reported on AWS and on GCP but I believe all Jammy 5.15 kernels are affected.
> 
> 3 ext4 fixes were applied at the same time upstream. The actual fix for the reported issue is "ext4: fix bad checksum after online resize" change. I have not reproduced/tested the bigalloc or group descriptors issues but they fix regressions introduced by the same commit and seemed simple enough so I grabbed them as well. The group descriptors fix required a complicated pre-requisite patch which would have brought in a lot of unnecessary code, so I applied a much smaller version of it as a SAUCE patch instead.
> 
> 
> Confirmed that the reported issue is no longer reproducible with these changes.
> 
> 
> 
> Baokun Li (3):
>    ext4: fix bad checksum after online resize
>    ext4: fix corruption when online resizing a 1K bigalloc fs
>    ext4: fix corrupt backup group descriptors after online resize
> 
> Khalid Elmously (1):
>    UBUNTU: SAUCE: Export ext4_superblock_csum function
> 
>   fs/ext4/ext4.h   |  2 ++
>   fs/ext4/resize.c | 27 ++++++++++++++++++++-------
>   fs/ext4/super.c  |  2 +-
>   3 files changed, 23 insertions(+), 8 deletions(-)
> 
Acked-by: Tim Gardner <tim.gardner at canonical.com>

I think patch 3 could be folded into patch 4, but meh...

-- 
-----------
Tim Gardner
Canonical, Inc