Re: [PATCH v5 5/5] drm/xe/ras: Move xe_ras_process_errors() to xe_ras

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Medium] xe_ras: boot-time wedge uevents are dropped before device registration
--

commit 7cc17054a629b3feb51763fdbe477fcbb6bc6ef3
Author: Raag Jadav <raag.jadav@intel.com>

drm/xe/ras: Move xe_ras_process_errors() to xe_ras

This commit moves xe_ras_process_errors() from the end of the probe
sequence back to xe_ras_init(). This was done to revert a previous change
and process hardware errors detected before the driver was loaded.

> diff --git a/drivers/gpu/drm/xe/xe_device.c b/drivers/gpu/drm/xe/xe_device.c
> index 23dfff89a4aa0..8d23e9cac3c67 100644
> --- a/drivers/gpu/drm/xe/xe_device.c
> +++ b/drivers/gpu/drm/xe/xe_device.c
[ ... ]
> diff --git a/drivers/gpu/drm/xe/xe_ras.c b/drivers/gpu/drm/xe/xe_ras.c
> index 922e5e793b135..8e399ffec9856 100644
> --- a/drivers/gpu/drm/xe/xe_ras.c
> +++ b/drivers/gpu/drm/xe/xe_ras.c
[ ... ]
> @@ -932,6 +932,13 @@ void xe_ras_init(struct xe_device *xe)
>  	if (IS_ENABLED(CONFIG_PCIEAER))
>  		ras_usp_aer_init(xe);
>  
> +	/*
> +	 * During probe, process and log any errors detected by firmware while the driver was not
> +	 * loaded. Critical errors such as Punit and CSC are reported through Pcode init failure,
> +	 * causing the driver to enter survivability mode.
> +	 */
> +	xe_ras_process_errors(xe);
        ^^^^^^^^^^^^^^^^^^^^^^^^^^

[Severity: Medium]
Will this silently drop boot-time wedge uevents?

Moving xe_ras_process_errors() to xe_ras_init() causes it to execute before
drm_dev_register() is called in the probe sequence.

If a hardware error is detected during early probe, xe_wedge_declare() in
drivers/gpu/drm/xe/xe_wedge.c will broadcast a wedge event to userspace:

drivers/gpu/drm/xe/xe_wedge.c:xe_wedge_declare()
  drivers/gpu/drm/drm_drv.c:drm_dev_wedged_event()
    kobject_uevent_env()

Since the DRM device has not been added to sysfs yet, the KOBJ_CHANGE uevent
is silently dropped by the kobject core because dev->primary->kdev is not
fully registered. This means userspace automation will miss the WEDGED uevent.

> +
>  	ret = devm_device_add_group(xe->drm.dev, &gpu_health_group);
>  	if (ret)
>  		xe_err(xe, "Failed to create GPU health sysfs, err=%d\n", ret);

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20261013110439.181447-1-raag.jadav@intel.com?part=5