On 05/07/2017 14:18, Chris Wilson wrote:
We need to reap the stale contexts for all new contexts, be they created
by user in i915_gem_context_ioctl or from opening a new file in
i915_gem_context_open. Both paths may be called very frequently
accumulating many stale contexts before any worker has a chance to run
and free their memory.
Fixes: 1acfc104cdf8 ("drm/i915: Enable rcu-only context lookups")
Signed-off-by: Chris Wilson <ch...@chris-wilson.co.uk>
Cc: Joonas Lahtinen <joonas.lahti...@linux.intel.com>
Cc: Tvrtko Ursulin <tvrtko.ursu...@intel.com>
---
drivers/gpu/drm/i915/i915_gem_context.c | 8 ++++----
1 file changed, 4 insertions(+), 4 deletions(-)
diff --git a/drivers/gpu/drm/i915/i915_gem_context.c
b/drivers/gpu/drm/i915/i915_gem_context.c
index 98d2ce98f467..c58a95c33c25 100644
--- a/drivers/gpu/drm/i915/i915_gem_context.c
+++ b/drivers/gpu/drm/i915/i915_gem_context.c
@@ -383,6 +383,10 @@ i915_gem_create_context(struct drm_i915_private *dev_priv,
lockdep_assert_held(&dev_priv->drm.struct_mutex);
+ /* Reap stale contexts */
+ i915_gem_retire_requests(dev_priv);
+ contexts_free(dev_priv);
+
ctx = __create_hw_context(dev_priv, file_priv);
if (IS_ERR(ctx))
return ctx;
@@ -989,10 +993,6 @@ int i915_gem_context_create_ioctl(struct drm_device *dev,
void *data,
if (ret)
return ret;
- /* Reap stale contexts */
- i915_gem_retire_requests(dev_priv);
- contexts_free(dev_priv);
-
ctx = i915_gem_create_context(dev_priv, file_priv);
mutex_unlock(&dev->struct_mutex);
if (IS_ERR(ctx))
Can't make myself concerned about the context creation slowdown with
this solution. I guess it highlights the same old RCU/delayed worker
issues, but it could be an option to somehow generalize these things
based on some threshold, like:
maybe_pruge_deleted_contexts()
{
if (nr_contexts_on_free_list > some_threshold) {
retire...
free...
}
}
Anyway:
Reviewed-by: Tvrtko Ursulin <tvrtko.ursu...@intel.com>
Regards,
Tvrtko
_______________________________________________
Intel-gfx mailing list
Intel-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
