linux-kernel - Linux regressions report for mainline [2024-03-10]

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [day] [month] [year] [list]

Message-ID: <053fb024-0fbf-4bdf-ac3a-26493ca08acd@leemhuis.info>
Date: Sun, 10 Mar 2024 16:18:33 +0100
From: Thorsten Leemhuis <regressions@...mhuis.info>
To: Linus Torvalds <torvalds@...ux-foundation.org>
Cc: LKML <linux-kernel@...r.kernel.org>,
 Linux kernel regressions list <regressions@...ts.linux.dev>
Subject: Linux regressions report for mainline [2024-03-10]

Hi Linus, here is a quick status update wrt. to regressions introduced
in this cycle. This time hand-complied instead of from regzbot (I just
broke something and lack time to fix it):

# Things that might or might not be a good reason for a -rc8

* Btrfs since v6.8-rc6 has a fiemap regression that "can actually have
serious consequences", as mentioned in
https://lore.kernel.org/linux-btrfs/cover.1709202499.git.fdmanana@suse.com/
The fixes in reply to that mail since Wednesday are in -next as
978b63f7464abc ("btrfs: fix race when detecting delalloc ranges during
fiemap") and 1cab1375ba6d53 ("btrfs: reuse cloned extent buffer during
fiemap to avoid re-allocations"), but not on the way to you afaics.

* A fix for Johan Hovold's last big Lenovo ThinkPad X13s regression
(hard resets disconnecting external displays) is currently up for review
and testing:
https://lore.kernel.org/lkml/ZeyOmJLlBbwnmaJN@hovoldconsulting.com/

# Fixes for regressions from this cycle afaics unlikely to reach 6.8

* -next commit 3445139e3a594b ("Revert "Revert "md/raid5: Wait for
MD_SB_CHANGE_PENDING in raid5d"""), which reverts a commit from
6.8-rc1; Song thinks it's better to merge it for 6.9 and then backport
to 6.8:
https://lore.kernel.org/all/CAPhsuW51ND4qH4My8Uz1XaZSdvAjDR7eL7O-RLr5wKmFJA0XMQ@mail.gmail.com/
Report: https://lore.kernel.org/all/20240123005700.9302-1-dan@danm.net/

* -next commit 0d7dfc79fb9b4b ("drm/msm/a6xx: specify UBWC config for
sc7180" No reply to my recent inquiry why this fix for a commit from
this cycle was queued for 6.9:
https://lore.kernel.org/all/4e55f6fd-8875-4a61-9070-f0ecad8082ea@leemhuis.info/
Report: https://gitlab.freedesktop.org/drm/msm/-/issues/49 and
https://lore.kernel.org/all/85581fad-da8f-4550-a1c8-8f2996425dcd@lausen.nl/

# Regressions fixes sitting on the list awaiting review

* "drm/nouveau: keep DMA buffers required for suspend/resume", which
fixes rendering freezes with multi-GPU setups:
https://lore.kernel.org/dri-devel/20240229175822.30613-1-sidpranjale127@protonmail.com/

* "Fix divide-by-zero on DP unplug with nouveau" which fixes a divide error:
https://lore.kernel.org/regressions/ZeoAPFIF6NClUl4P@debian.local/

* "u64_stats: fix u64_stats_init() for lockdep when used repeatedly in
one file" fixes a lockdep splat
https://lore.kernel.org/lkml/20240306111157.29327-1-petr@tesarici.cz/

# Regression reports where no fix is in sight afaics

* dmaengine: CPU stalls while loading bluetooth module
https://lore.kernel.org/lkml/ZYhQ2-OnjDgoqjvt@wens.tw/ and
https://lore.kernel.org/lkml/1553a526-6f28-4a68-88a8-f35bd22d9894@linumiz.com/
See also:
https://lore.kernel.org/all/000001da7140%246a0f1570%243e2d4050%24@samsung.com/
Revert might be counter-productive.

* x86/mm/ident_map: kexec now leads to reboot
https://lore.kernel.org/stable/3a1b9909-45ac-4f97-ad68-d16ef1ce99db@pavinjoseph.com/
https://lore.kernel.org/all/CAJbxNHe3EJ88ABB1aZ8bYZq=a36F0TFST1Fqu4fkugvyU_fjhw@mail.gmail.com/
A revert apparently could fix this.

Ciao, Thorsten