reassign 1104677 src:linux-signed-amd64 6.1.135+1
thanks
If the mkfs.ext4 is getting killed when trying to issue the discard
commands to the RAID ARRAY with the 6.1.0-34 kernel[1] (note: 6.1.0-34
is built from upstream kernel version 6.1.135), but it works fine with
the 6.1.0-32 kernel, then this is clearly a kernel bug, not an
e2fsprogs bug.
Note to the linux kernel Debian maintainer: the user can avoid the
problem by either (a) disabling the use of discard commands, or (b)
not trying to format using a RAID block device, so the bug is almost
certainly either (1) in the software RAID kernel driver, or (2) in the
underlying hardware driver. Given that the user is using NVMe
devices, and if there was this kind of bug in the NVMe driver, it's
likely that many users would have complained --- my guess is that it's
probably (1).
I've also checked upstream for 6.1 LTS, and the latest upstream LTS
version is 6.1.137, released today, and I don't see any relevant bug
fixes in drivers/md between 6.1.135 and 6.1.137.
Note to the bug reporter. You might want to check your console
messages or the output of the dmesg command immediately after trying
to run mkfs.ext4 and seeing it get killed, since it's likely that the
kernel issued a kernel OOPS message, and the stack trace would likely
be helpful.
If you have the time / ability, if you could try reproducing the
problem using upstream 6.1.135, and bisecting it down (sorry,
packages.debian.org is currently very sad, so I can't find the
upstream kernel version that was used to build 6.1.0-32), and could
report it to sta...@kernel.org and cc: linux-ker...@vger.kernel.org
and linux-r...@vger.kernel.org, that would allow you to engage the
upstream Linux kernel maintainers.
Cheers,
- Ted
