Lines Matching refs:semaphores
643 * Are we using semaphores when the gpu is already saturated?
645 * Using semaphores incurs a cost in having the GPU poll a
653 * system is overloaded and using semaphores at this time only
655 * further use of semaphores until we are idle again, whence we
658 if (request->sched.semaphores &&
660 engine->saturated |= request->sched.semaphores;
737 if (request->sched.semaphores && __i915_request_has_started(request))
738 request->sched.semaphores = 0;
1140 return rq->sched.semaphores | READ_ONCE(rq->engine->saturated);
1220 * semaphores as they bypass the fence signaling metadata, and we
1240 to->sched.semaphores |= mask;
1296 * Ensure both start together [after all semaphores in signal]
1338 * The downside of using semaphores is that we lose metadata passing
1752 * precludes optimising to use semaphores serialisation of a single
2207 * - a ready request may still need to wait on semaphores