On 03.12.2024 15:54, Eugene Kobyak wrote:
> When the intel_context structure contains NULL,
> it raises a NULL pointer dereference error in drm_info().
>
> Fixes: e8a3319c31a1 ("drm/i915: Allow error capture without a request")
> Closes: https://gitlab.freedesktop.org/drm/i915/kernel/-/issues/12309
> Reviewed-by: Andi Shyti <andi.shyti@xxxxxxxxxxxxxxx>
> Cc: John Harrison <John.C.Harrison@xxxxxxxxx>
> Cc: <stable@xxxxxxxxxxxxxxx> # v6.3+
> Signed-off-by: Eugene Kobyak <eugene.kobyak@xxxxxxxxx>
> ---
> v2:
> - return drm_info to separate condition
> v3:
> - create separate condition which generate string if intel_context exist
> v4:
> - rollback and add check intel_context in log condition
> v5:
> - create separate string with guc_id if intel_context exist
> v6:
> - print changed log if intel_context exist
>
> drivers/gpu/drm/i915/i915_gpu_error.c | 18 +++++++++++++++---
> 1 file changed, 15 insertions(+), 3 deletions(-)
>
> diff --git a/drivers/gpu/drm/i915/i915_gpu_error.c b/drivers/gpu/drm/i915/i915_gpu_error.c
> index 135ded17334e..d88cefb889c3 100644
> --- a/drivers/gpu/drm/i915/i915_gpu_error.c
> +++ b/drivers/gpu/drm/i915/i915_gpu_error.c
> @@ -1643,9 +1643,21 @@ capture_engine(struct intel_engine_cs *engine,
> return NULL;
>
> intel_engine_get_hung_entity(engine, &ce, &rq);
> - if (rq && !i915_request_started(rq))
> - drm_info(&engine->gt->i915->drm, "Got hung context on %s with active request %lld:%lld [0x%04X] not yet started\n",
> - engine->name, rq->fence.context, rq->fence.seqno, ce->guc_id.id);
> + if (rq && !i915_request_started(rq)) {
> + /*
> + * We want to know also what is the gcu_id of the context,
typo: guc_id
> + * but if we don't have the context reference, then skip
> + * printing it.
> + */
but IMO this comment is redundant as it's quite obvious that without
context pointer you can't print guc_id member
> + if (ce)
> + drm_info(&engine->gt->i915->drm,
> + "Got hung context on %s with active request %lld:%lld [0x%04X] not yet started\n",
> + engine->name, rq->fence.context, rq->fence.seqno, ce->guc_id.id);
> + else
> + drm_info(&engine->gt->i915->drm,
> + "Got hung context on %s with active request %lld:%lld not yet started\n",
> + engine->name, rq->fence.context, rq->fence.seqno);
since you are touching drm_info() where we use engine->gt then maybe
it's good time to switch to gt_info() to get better per-GT message?
> + }
>
> if (rq) {
> capture = intel_engine_coredump_add_request(ee, rq, ATOMIC_MAYFAIL);
