To complete the idle worker, we must complete 2 passes of wait-for-idle.
At the end of the first pass, we queue a switch-to-kernel-context and
may only idle after waiting for its completion. Speed up the flush_work
by doing the wait explicitly, which then allows us to remove the
unbounded loop trying to complete the flush_work in the next patch.
References: 79ffac8599c4 ("drm/i915: Invert the GEM wakeref hierarchy")
Testcase: igt/gem_ppgtt/flind-and-close-vma-leak
Signed-off-by: Chris Wilson <chris@xxxxxxxxxxxxxxxxxx>
Cc: Tvrtko Ursulin <tvrtko.ursulin@xxxxxxxxx>
---
drivers/gpu/drm/i915/i915_debugfs.c | 16 ++++++++++++++--
1 file changed, 14 insertions(+), 2 deletions(-)
diff --git a/drivers/gpu/drm/i915/i915_debugfs.c b/drivers/gpu/drm/i915/i915_debugfs.c
index 14cd83e9ea8b..f60aed7747e5 100644
--- a/drivers/gpu/drm/i915/i915_debugfs.c
+++ b/drivers/gpu/drm/i915/i915_debugfs.c
@@ -3901,14 +3901,26 @@ i915_drop_caches_set(void *data, u64 val)
/* No need to check and wait for gpu resets, only libdrm auto-restarts
* on ioctls on -EAGAIN. */
- if (val & (DROP_ACTIVE | DROP_RETIRE | DROP_RESET_SEQNO)) {
+ if (val & (DROP_ACTIVE | DROP_IDLE | DROP_RETIRE | DROP_RESET_SEQNO)) {
int ret;
ret = mutex_lock_interruptible(&i915->drm.struct_mutex);
if (ret)
return ret;
- if (val & DROP_ACTIVE)
+ /*
+ * To finish the flush of the idle_worker, we must complete
+ * the switch-to-kernel-context, which requires a double
+ * pass through wait_for_idle: first queues the switch,
+ * second waits for the switch.
+ */
+ if (ret == 0 && val & (DROP_IDLE | DROP_ACTIVE))
+ ret = i915_gem_wait_for_idle(i915,
+ I915_WAIT_INTERRUPTIBLE |
+ I915_WAIT_LOCKED,
+ MAX_SCHEDULE_TIMEOUT);
+
+ if (ret == 0 && val & DROP_IDLE)
ret = i915_gem_wait_for_idle(i915,
I915_WAIT_INTERRUPTIBLE |
I915_WAIT_LOCKED,
--
2.20.1
_______________________________________________
Intel-gfx mailing list
Intel-gfx@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
