Nouveau - May 2024 - [REGRESSION] v6.9-rc7: nouveau: init failed, no display output from kernel; successfully bisected

On 06.05.24 20:23, Dan Moulding wrote:
> After upgrading to rc7 from rc6 on a system with NVIDIA GP104 using
> the nouveau driver, I get no display output from the kernel (only the
> output from GRUB shows on the primary display). Nonetheless, I was
> able to SSH to the system and get the kernel log from dmesg. I found
> errors from nouveau in it. Grepping it for nouveau gives me this:
> 
> [    0.367379] nouveau 0000:01:00.0: NVIDIA GP104 (134000a1)
> [    0.474499] nouveau 0000:01:00.0: bios: version 86.04.50.80.13
> [    0.474620] nouveau 0000:01:00.0: pmu: firmware unavailable
> [    0.474977] nouveau 0000:01:00.0: fb: 8192 MiB GDDR5
> [    0.484371] nouveau 0000:01:00.0: sec2(acr): mbox 00000001 00000000
> [    0.484377] nouveau 0000:01:00.0: sec2(acr):load: boot failed: -5
> [    0.484379] nouveau 0000:01:00.0: acr: init failed, -5
> [    0.484466] nouveau 0000:01:00.0: init failed with -5
> [    0.484468] nouveau: DRM-master:00000000:00000080: init failed with -5
> [    0.484470] nouveau 0000:01:00.0: DRM-master: Device allocation failed:
-5
> [    0.485078] nouveau 0000:01:00.0: probe with driver nouveau failed with
error -50
> 
> I bisected between v6.9-rc6 and v6.9-rc7 and that identified commit
> 52a6947bf576 ("drm/nouveau/firmware: Fix SG_DEBUG error with
> nvkm_firmware_ctor()") as the first bad commit.
Lyude, that's a commit of yours.

Given that 6.9 is due a quick question: I assume there is no easy fix
for this in sight? Or is a quick revert something that might be
appropriate to prevent this from entering 6.9?

Ciao, Thorsten
> I then rebuilt
> v6.9-rc7 with just that commit reverted and the problem does not
> occur.
> 
> Please let me know if there are any additional details I can provide
> that would be helpful, or if I should reproduce the failure with
> additional debugging options enabled, etc.
> 
> Cheers,
> 
> -- Dan

On Thu, May 9, 2024 at 8:31?PM Linux regression tracking (Thorsten
Leemhuis) <regressions at leemhuis.info> wrote:
>
> On 06.05.24 20:23, Dan Moulding wrote:
> > After upgrading to rc7 from rc6 on a system with NVIDIA GP104 using
> > the nouveau driver, I get no display output from the kernel (only the
> > output from GRUB shows on the primary display). Nonetheless, I was
> > able to SSH to the system and get the kernel log from dmesg. I found
> > errors from nouveau in it. Grepping it for nouveau gives me this:
> >
> > [    0.367379] nouveau 0000:01:00.0: NVIDIA GP104 (134000a1)
> > [    0.474499] nouveau 0000:01:00.0: bios: version 86.04.50.80.13
> > [    0.474620] nouveau 0000:01:00.0: pmu: firmware unavailable
> > [    0.474977] nouveau 0000:01:00.0: fb: 8192 MiB GDDR5
> > [    0.484371] nouveau 0000:01:00.0: sec2(acr): mbox 00000001 00000000
> > [    0.484377] nouveau 0000:01:00.0: sec2(acr):load: boot failed: -5
> > [    0.484379] nouveau 0000:01:00.0: acr: init failed, -5
> > [    0.484466] nouveau 0000:01:00.0: init failed with -5
> > [    0.484468] nouveau: DRM-master:00000000:00000080: init failed with
-5
> > [    0.484470] nouveau 0000:01:00.0: DRM-master: Device allocation
failed: -5
> > [    0.485078] nouveau 0000:01:00.0: probe with driver nouveau failed
with error -50
> >
> > I bisected between v6.9-rc6 and v6.9-rc7 and that identified commit
> > 52a6947bf576 ("drm/nouveau/firmware: Fix SG_DEBUG error with
> > nvkm_firmware_ctor()") as the first bad commit.
>
> Lyude, that's a commit of yours.
>
> Given that 6.9 is due a quick question: I assume there is no easy fix
> for this in sight? Or is a quick revert something that might be
> appropriate to prevent this from entering 6.9?
I'll take a look today and see if I can reproduce it and revert it if
needed.

Dave.