Re: [PATCH v2 2/2] drm/xe/xe_drm_ras: Add error-event support in XE drm_ras

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Medium] Non-ratelimited `drm_err` in atomic context can cause console flood and soft lockups under memory pressure.
--

commit fadf160f443d10205c2dcc1f2b3421344e7bb067
Author: Riana Tauro <riana.tauro@intel.com>

drm/xe/xe_drm_ras: Add error-event support in XE drm_ras

This commit adds error-event support in the XE drm_ras subsystem. It notifies
userspace via netlink when a hardware error occurs, allowing tools to
subscribe to these events.

> diff --git a/drivers/gpu/drm/xe/xe_drm_ras.c b/drivers/gpu/drm/xe/xe_drm_ras.c
> index cd236f53699e0..09588e3fd984d 100644
> --- a/drivers/gpu/drm/xe/xe_drm_ras.c
> +++ b/drivers/gpu/drm/xe/xe_drm_ras.c

[ ... ]

> +void xe_drm_ras_event(struct xe_device *xe, u32 component, u32 severity, u32 value, gfp_t flags)
> +{
> +	struct xe_drm_ras *ras = &xe->ras;
> +	struct xe_drm_ras_counter *info = ras->info[severity];
> +	struct drm_ras_node *node = &ras->node[severity];
> +	int ret;
> +
> +	if (!info || !info[component].name)
> +		return;
> +
> +	ret = drm_ras_nl_error_event(node, component, info[component].name, value, flags);
> +	if (ret)
> +		drm_err(&xe->drm, "RAS error-event failed: %d for %s %s\n", ret,
> +			info[component].name, error_severity[severity]);

[Severity: Medium]
Could this un-ratelimited drm_err() cause console flooding and soft lockups
under memory pressure?

This function is called by hw_error_source_handler(), which iterates over
hardware errors while holding xe->irq.lock with interrupts disabled. Since
the netlink message allocation uses GFP_ATOMIC, it can fail repeatedly under
low memory conditions.

If an interrupt storm of hardware errors occurs, repeatedly printing to the
console from within a hard IRQ spinlock could heavily stall the CPU and
trigger RCU stalls or watchdogs.

Would it be safer to use drm_err_ratelimited() here, matching the behavior
of the other error logging in hw_error_source_handler()?

> +}

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260611052144.784969-4-riana.tauro@intel.com?part=2