Re: [PATCH] Revert "nouveau: pci: quiesce GPU on shutdown"

["Uwe Kleine-König" <ukleinek@debian.org>]

[-- Attachment #1: Type: text/plain, Size: 3616 bytes --]

Control: forwarded -1 https://lore.kernel.org/nouveau/aviIPk4xbKhgqcTy@monoceros

Hello,

On Sun, Oct 25, 2026 at 10:30:06PM +0100, Marek Vasut wrote:
> This reverts commit 310326bb7df4bba094a3fc60364c641c547fd923.
> 
> This breaks shutdown on ThinkPad W510, without the offending commit
> the system normally powers off, with the offending commit the system
> ends up spinning indefinitely and the CPU fan goes full throttle on
> shutdown.
> 
> The behavior was already reported in debian bugtracker multiple times
> for various other computers too, and git bisect lands as this commit.
> https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1141969

I think the original bug #1141969 is a different issue.

> https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1146033

Here it affects a Lenovo ThinkPad T510 and I think that's the same issue
that Marek has.


> I tested the revert with Linux 7.2.12 from debian unstable, not next
> or drm trees, mainly because it was easier with debian test-patches
> script on that machine.
> 
> Signed-off-by: Marek Vasut <marex@nabladev.com>
> ---
> Cc: Danilo Krummrich <dakr@kernel.org>
> Cc: David Airlie <airlied@gmail.com>
> Cc: Li Chen <me@linux.beauty>
> Cc: Lyude Paul <lyude@redhat.com>
> Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
> Cc: Maxime Ripard <mripard@kernel.org>
> Cc: Ole Carlsen <ole.carlsen@tutamail.com>
> Cc: Simona Vetter <simona@ffwll.ch>
> Cc: Thomas Zimmermann <tzimmermann@suse.de>
> Cc: Uwe Kleine-König <ukleinek@debian.org>
> Cc: dri-devel@lists.freedesktop.org
> Cc: linux-kernel@vger.kernel.org
> Cc: nouveau@lists.freedesktop.org
> ---
> NOTE: Maybe the revert is not the ideal solution, but it does fix my
>       problem, and I think being able to shut computer down is a good
>       functionality to have.
> ---
>  drivers/gpu/drm/nouveau/nouveau_drm.c | 32 ---------------------------
>  1 file changed, 32 deletions(-)
> 
> diff --git a/drivers/gpu/drm/nouveau/nouveau_drm.c b/drivers/gpu/drm/nouveau/nouveau_drm.c
> index 5e1336615cf6d..36c74f9edaa83 100644
> --- a/drivers/gpu/drm/nouveau/nouveau_drm.c
> +++ b/drivers/gpu/drm/nouveau/nouveau_drm.c
> @@ -1090,37 +1090,6 @@ nouveau_pmops_resume(struct device *dev)
>  	return ret;
>  }
>  
> -static void
> -nouveau_drm_shutdown(struct pci_dev *pdev)
> -{
> -	struct nouveau_drm *drm = pci_get_drvdata(pdev);
> -	int ret;
> -
> -	if (!drm)
> -		return;
> -
> -	if (drm->dev->switch_power_state == DRM_SWITCH_POWER_OFF ||
> -	    drm->dev->switch_power_state == DRM_SWITCH_POWER_DYNAMIC_OFF)
> -		return;
> -
> -	ret = nouveau_do_suspend(drm, false);
> -	if (ret)
> -		NV_ERROR(drm, "shutdown suspend failed with: %d\n", ret);
> -
> -	pci_save_state(pdev);
> -	pci_disable_device(pdev);
> -	pci_set_power_state(pdev, PCI_D3hot);
> -	/*
> -	 *  This is just to give the pci power transition time to settle
> -	 *  before an immediate kexec jump. it’s mirroring the existing
> -	 *  nouveau_pmops_suspend() behavior, which already does
> -	 *  udelay(200) right after pci_set_power_state(..., pci_d3hot). In
> -	 *  ->shutdown() we’re allowed to sleep, so I used usleep_range()
> -	 *  instead of a busy-wait udelay().
> -	 */
> -	usleep_range(200, 400);
> -}
> -

I cannot reproduce it, but maybe it's helpful to find out where/how the
driver hangs. Adding printks might be enough, and if not, remove stuff
from nouveau_drm_shutdown() until it works. I'd guess this gives the
needed insights to determine a fix without reverting the identified
commit.

Best regards
Uwe

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 488 bytes --]