I'm hitting this on different hardware, a Framework Laptop 16 (Ryzen
7040, Phoenix) on Ubuntu 24.04 with kernel 7.0.0-31. These finds are
from research with Claude Fable. My trigger is a USB-C dock unplug
rather than an eDP screen-off or wake, but it's the same blob and the
same underlying regression.

One thing worth flagging first: I don't think the test build in comment
#2 includes a fix for this. Its dcn_3_1_4_dmcub.bin.zst is identical to
the broken one:

  1dec0614bd16c87958ca2e2874a18cc0  candidate build (lp2163303)
  1dec0614bd16c87958ca2e2874a18cc0  0ubuntu3.2   (broken, 0x08004800)
  edbf69ba73444503f6b331ecf34a790d  0ubuntu2.26  (works,  0x08005300)

Since that build is named for LP #2163303, I suspect it leaves DCN 3.1.4
untouched, which would mean testing it shows no change with nothing to
explain why. It may just need a rebuild that picks up the DCN 3.1.4
blob.

On the evidence side, I went back through 21 boots, and the DMUB
firmware version is the only variable that predicts failure. On
0x08005300 I have 56 dock unplugs with zero DMCUB errors. On 0x08004800
every boot that saw an unplug logged errors, somewhere between 12 and
14926 of them.

Kernel 7.0.0-28 and 7.0.0-31 both show up on the working and failing
sides, and both of my BIOS versions appear among the failures, so
neither looks involved. The clearest test was booting 7.0.0-28, which
had been clean across more than 50 unplugs on the old firmware. It still
failed on 0x08004800, which points at the firmware on its own.

It's intermittent, too. One boot logged three unplugs but only one hang.

The signature starts a few seconds after the dock disconnects and then
repeats indefinitely:

  amdgpu: [drm] *ERROR* dc_dmub_srv_log_diagnostic_data: DMCUB error - 
collecting diagnostic data
  amdgpu: [drm] *ERROR* Error queueing DMUB command: status=2

The desktop drops to roughly 0.1 FPS while CPU and GPU sit idle, since
the compositor blocks on each failed command. Suspend and resume clears
it, as does a reboot.

I also looked at the dc/dce/dmub_replay.c warnings that have come up
elsewhere. They seem to be downstream symptoms, since they fire a couple
of seconds after the DMCUB is already dead. Booting with
amdgpu.dcdebugmask=0xc10, which turns off PSR, Replay and IPS, silenced
those warnings but didn't prevent the hang.

-- 
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/2163554

Title:
  DCN314/Phoenix DMCUB 0x08004800 page-faults at GART 0x6026000 during
  eDP power transition, wedging display (2.29 regression)

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/linux-firmware/+bug/2163554/+subscriptions


-- 
ubuntu-bugs mailing list
[email protected]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs

Reply via email to