[Intel-gfx] [PATCH 3/3] drm/i915: Ratelimit request allocation under oom

[Chris Wilson]

If we fail to allocate a request, we can reap the outstanding requests
and push them to the request's slab's freelist before trying again. This
forces us to ratelimit malicious clients that tie up all of the system
resources in requests, instead of causing a system-wide oom.

Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
Cc: Joonas Lahtinen <joonas.lahtinen at linux.intel.com>
---
 drivers/gpu/drm/i915/i915_gem_request.c | 19 +++++++++++++++----
 1 file changed, 15 insertions(+), 4 deletions(-)

diff --git a/drivers/gpu/drm/i915/i915_gem_request.c b/drivers/gpu/drm/i915/i915_gem_request.c
index 4d5e2b714382..59f023bb7015 100644
--- a/drivers/gpu/drm/i915/i915_gem_request.c
+++ b/drivers/gpu/drm/i915/i915_gem_request.c
@@ -677,10 +677,21 @@ i915_gem_request_alloc(struct intel_engine_cs *engine,
 	 *
 	 * Do not use kmem_cache_zalloc() here!
 	 */
-	req = kmem_cache_alloc(dev_priv->requests, GFP_KERNEL);
-	if (!req) {
-		ret = -ENOMEM;
-		goto err_unreserve;
+	req = kmem_cache_alloc(dev_priv->requests,
+			       GFP_KERNEL | __GFP_RETRY_MAYFAIL | __GFP_NOWARN);
+	if (unlikely(!req)) {
+		/* Ratelimit ourselves to prevent oom from malicious clients */
+		ret = i915_gem_wait_for_idle(dev_priv,
+					     I915_WAIT_LOCKED |
+					     I915_WAIT_INTERRUPTIBLE);
+		if (ret)
+			goto err_unreserve;
+
+		req = kmem_cache_alloc(dev_priv->requests, GFP_KERNEL);
+		if (!req) {
+			ret = -ENOMEM;
+			goto err_unreserve;
+		}
 	}
 
 	req->timeline = i915_gem_context_lookup_timeline(ctx, engine);
-- 
2.15.1