regressions report as of Linux 7.1-rc5 post (was: Linux 7.1-rc5)

[Thorsten Leemhuis]

Hi Linus!

On 5/24/26 23:07, Linus Torvalds wrote:
>
> So I think I'll start being a bit more hardnosed about this kind of
> unnecessary churn this late in the game. We are supposed to look for
> *regressions*. Non-critical fixes to long-standing issues are simply
> not appropriate for this late in the release cycle.

Speaking of regressions: I got the feeling the weekly regression reports
made way more work then they were worth it up to -rc5. But we are
approaching -rc6, so let me give you an overview about some unresolved
regressions.


###########################################################################
Unresolved regressions from this cycle that could easily get fixed for
-rc6 -- but might or might not:

---
media: renesas: vsp1: brx: configuration of the DRM pipeline broken
https://lore.kernel.org/linux-media/CA%2BV-a8t481xuwava0nb7uY9CUPqFWZ_8EP0xrK3BgumP7HDcLg@xxxxxxxxxxxxxx/

Known for 6 weeks. A patch-series to fix this with reverts is sitting on
the list for three weeks now. Prodded there Wednesday, but maintainers
did not yet pick things up. Laurent sent a PR on Friday to get them to
act, maybe that will do the trick:
https://lore.kernel.org/linux-media/20260529102955.GA675012@xxxxxxxxxxxxxxxxxxxxxxxxxx/

---
drm/i915: backlight problems
https://lore.kernel.org/lkml/CADo9pHjr-zZ9C3%2B026y5%2BXOGPSeRzSJMCHof27TVPtAUNgTc8A@xxxxxxxxxxxxxx/
aka https://gitlab.freedesktop.org/drm/i915/kernel/-/work_items/16015
https://gitlab.freedesktop.org/drm/i915/kernel/-/issues/16043
https://lore.kernel.org/lkml/d2de7933-e650-4b19-8d88-90d66693dcfc@xxxxxxxxxxxxxxxxxxxxxxxxx/
aka https://gitlab.freedesktop.org/drm/i915/kernel/-/work_items/16097

Known since -rc1 and fixed by a revert that sits in -next as
f30fddb4402313 ("Revert "drm/i915/backlight: Remove
try_vesa_interface"") [next-20260520] -- looks a bit like it's destined
to be merged in the next merge window, but with all those cherry-picking
in the drm subsystem I might easily be wrong. Asked for the plans on
Thursday, no reply yet:
https://gitlab.freedesktop.org/drm/i915/kernel/-/work_items/16043#note_3492668

---
soc: imx8m: warning and serial id data missing
https://lore.kernel.org/lkml/6593091.DvuYhMxLoT@steina-w/

Known for 9 weeks. The fix was sitting on the list and in next
(2c6821657ce3b3 ("soc: imx8m: Fix match data lookup for soc device"))
for a while; I prodded on Thursday and it's supposed to be en route to
mainline now.


###########################################################################
Unresolved regressions from this cycle that are unlikely to get fixed
for -rc6:

---
drm: amdgpu: handling FreeSync via MCCS broke VRR
https://gitlab.freedesktop.org/drm/amd/-/issues/5286

Known for 17 days and reporter proposed a fix shortly afterwards; I
prodded the maintainers on Thursday and they want to look into things
and get something into next weeks -fixes pull:
https://lore.kernel.org/amd-gfx/20260518154809.660061-1-michel@xxxxxxxxxxx/
https://lore.kernel.org/amd-gfx/BL1PR12MB51447AB016BB28843BAD0216F7092@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx/

---
cpufreq/amd-pstate: switching to performance profile broke
https://bugzilla.kernel.org/show_bug.cgi?id=221505

Known for 18 days, culprit unknown. Mario and colleagues fixed a few
bugs in this area in the past few weeks, but seems one problem that
affects at least two people on Asus laptops remains -- a fix or a revert
to resolve this is sadly not in sight
(https://bugzilla.kernel.org/show_bug.cgi?id=221505#c22).

---
loop: loop drive broken due to some race exposed by an unknown change
https://lore.kernel.org/lkml/69e2ca14.a00a0220.1bd0ca.0031.GAE@xxxxxxxxxx/

Known since 6 weeks. People are busy trying to get this resolved.


#########################################################################
Unresolved regressions from the previous cycle that could easily get
fixed for -rc6 -- but might or might not:

---
ARM: at91: finding rootfs on raw nand on sam9x60 broken
https://lore.kernel.org/lkml/20260423-stoke-ocean-139dbe306256@xxxxxxxxxxx/

Known for 5 weeks. Fix was sitting in -next as 754d60ad1c9189 ("memory:
atmel-ebi: Allow deferred probing") since next-20260504; prodded on
Thursday, now en route to mainline.

---
drm/amd/display: screen goes black halfway through the booting proces
https://gitlab.freedesktop.org/drm/amd/-/issues/5272

Known for 19 days. Hope the fix will make it to mainline soon:
08236c3ef284cd ("drm/amd/display: Write REFCLK to 48MHz on DCN21")
[next-20260528]

---
serial: 8250: BREAK + SysRq dispatch broken
https://lore.kernel.org/lkml/5efe9e03-4d86-43a0-9ec2-e610ff31095d@xxxxxxx/

Known for 17 days. Hope the fix will make it to mainline soon:
71f42b2149a130 ("serial: 8250: dispatch SysRq character in
serial8250_handle_irq()") [next-20260525 (pending-fixes)]

---
accel/amdxdna: AMD PMC causes hard hang on s2idle suspend
https://gitlab.freedesktop.org/drm/amd/-/issues/5319

Known for 10 days. Hope the fix will make it to mainline soon:
a8878e19d2f520 ("accel/amdxdna: Block running when IOMMU is off")
[next-20260522 (pending-fixes)]

---
ipv6: SSH session gets dropped, usually several times, in the first hour
or so https://lore.kernel.org/all/20260521135310.GC977@xxxxxxxxxxx/

Known for 8 days. Revert under lively discussion, see thread above and:
https://lore.kernel.org/all/20260529112357.5079-1-fmancera@xxxxxxx/
https://lore.kernel.org/all/20260528111123.4667-1-fmancera@xxxxxxx/


#########################################################################
Unresolved regressions from even earlier cycles that could easily get
fixed for -rc6, but that's unlikely to happen from what I see:

---
e1000e: "Unstable ethernet connection on Dell Pro 16" as well as "PC10
in runtime idle in newer systems prevented"
https://lore.kernel.org/all/20260417104330.3031987-1-dima.ruinskiy@xxxxxxxxx/
https://lists.osuosl.org/pipermail/intel-wired-lan/Week-of-Mon-20250623/048860.html

Known for 5 months, and a fix exists, but Intel claims it needs more time:
https://lore.kernel.org/all/20260417104330.3031987-1-dima.ruinskiy@xxxxxxxxx/
https://lore.kernel.org/all/56f5f3da-31b8-4847-b274-ce140bf39617@xxxxxxxxx/

---
scsi: massive resource usage that can lead to kernel crashes
https://lore.kernel.org/lkml/773ba972-433b-44b4-89d2-295bd9f5de38@xxxxxxxxxx/
https://lore.kernel.org/linux-scsi/20260415204850.799431-1-mwilck@xxxxxxxx/

Known for 9 weeks. Fixes exists for weeks, but discussion is slow. hch
recently ACKed it:
https://lore.kernel.org/linux-scsi/20260513174236.430465-1-mwilck@xxxxxxxx/

---
rtl8xxxu: RTL8188FTV (0bda:f179) fails to authenticate/configure since 6.19
https://bugzilla.kernel.org/show_bug.cgi?id=221394
https://lore.kernel.org/linux-wireless/8a4c664f-231f-437e-b010-b741c7ed7be8@xxxxxxxxxxxxx/

Known for 6 weeks. A fix is sitting in a sub-sub-subsystem tree for a
while now that is not even in -next -- and it looks like the developers
are unwilling to merge the fix this cycle:
https://lore.kernel.org/linux-wireless/c57de68e-5d57-4c26-898f-8a284bb25381@xxxxxxxxx/

---
x86/fpu: applications that construct signal frames in userspace broke
(such as CRIU and gVisor)
https://lore.kernel.org/lkml/20260429000623.3356606-1-avagin@xxxxxxxxxx/

Known for 4 weeks. A revert was proposed a month ago (see above), but
didn't make any progress; a newer revert was posted this week:
https://lore.kernel.org/lkml/20260526205047.3339490-2-avagin@xxxxxxxxxx/

---
NFSv4: mkdir(2) on an NFSv4 mount occasional returns -EEXIST
https://lore.kernel.org/lkml/20260429104938.1776671-1-igor.raits@xxxxxxxxx/

Known for 4 weeks. A fix exists, but maintainers didn't reply yet and
recent prodding by yours truly did not help:
https://lore.kernel.org/lkml/20260429104938.1776671-1-igor.raits@xxxxxxxxx/


#########################################################################
Unresolved regressions from even earlier cycles that could easily get
fixed for -rc6 -- but might or might not:

---
parport: race leading to a NULL pointer dereference and a panic
https://lore.kernel.org/lkml/6ba903ad-9897-42bb-8c2d-337385cc3746@xxxxxxxxxxxxx/
https://bugzilla.kernel.org/show_bug.cgi?id=221526

Known for 6 months now. Culprit is very old, but seems the problem
somehow happens more often since some time ago. Hope the fix will make
it to mainline soon: ef15ccbb3e8640 ("parport: Fix race between port and
client registration") [next-20260525 (pending-fixes)]

---
usb: typec: ucsi: laptop does not enter suspend anymore
https://lore.kernel.org/lkml/agvqqMt9x0df-WXS@xxxxxxxxxx/

Known for 10 days. Hope the fix will make it to mainline soon:
d98d413ca65d07 ("usb: typec: ucsi: Don't update power_supply on power
role change if not connected")