[Intel-gfx] [CI 1/2] drm/i915/gt: Check for arbitration after writing start seqno
chris at chris-wilson.co.uk
Tue Jan 12 10:07:58 UTC 2021
On the off chance that we need to arbitrate before launching the
payload, perform the check after we signal the request is ready to
start. Assuming instantaneous processing of the CS event, the request
will then be treated as having started when we make the decisions as to
how to process that CS event.
v2: More commentary about the users of i915_request_started() as a
reminder about why we are marking the initial breadcrumb.
Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin at intel.com>
drivers/gpu/drm/i915/gt/gen8_engine_cs.c | 23 +++++++++++++++++------
1 file changed, 17 insertions(+), 6 deletions(-)
diff --git a/drivers/gpu/drm/i915/gt/gen8_engine_cs.c b/drivers/gpu/drm/i915/gt/gen8_engine_cs.c
index 2e36e0a9d8a6..1ed9f572c8a4 100644
@@ -361,19 +361,30 @@ int gen8_emit_init_breadcrumb(struct i915_request *rq)
+ *cs++ = MI_STORE_DWORD_IMM_GEN4 | MI_USE_GGTT;
+ *cs++ = hwsp_offset(rq);
+ *cs++ = 0;
+ *cs++ = rq->fence.seqno - 1;
* Check if we have been preempted before we even get started.
* After this point i915_request_started() reports true, even if
* we get preempted and so are no longer running.
+ * i915_request_started() is used during preemption processing
+ * to decide if the request is currently inside the user payload
+ * or spinning on a kernel semaphore (or earlier). For no-preemption
+ * requests, we do allow preemption on the semaphore before the user
+ * payload, but do not allow preemption once the request is started.
+ * i915_request_started() is similarly used during GPU hangs to
+ * determine if the user's payload was guilty, and if so, the
+ * request is banned. Before the request is started, it is assumed
+ * to be unharmed and an innocent victim of another's hang.
- *cs++ = MI_ARB_CHECK;
*cs++ = MI_NOOP;
- *cs++ = MI_STORE_DWORD_IMM_GEN4 | MI_USE_GGTT;
- *cs++ = hwsp_offset(rq);
- *cs++ = 0;
- *cs++ = rq->fence.seqno - 1;
+ *cs++ = MI_ARB_CHECK;
More information about the Intel-gfx