drm/i915: Fixup the free space logic in ring_prepare
authorAkash Goel <akash.goel@intel.com>
Fri, 11 Mar 2016 09:26:42 +0000 (14:56 +0530)
committerChris Wilson <chris@chris-wilson.co.uk>
Sat, 9 Apr 2016 10:37:48 +0000 (11:37 +0100)
Currently for the case where there is enough space at the end of Ring
buffer for accommodating only the base request, the wrapround is done
immediately and as a result the base request gets added at the start
of Ring buffer. But there may not be enough free space at the beginning
to accommodate the base request, as before the wraparound, the wait was
effectively done for the reserved_size free space from the start of
Ring buffer. In such a case there is a potential of Ring buffer overflow,
the instructions at the head of Ring (ACTHD) can get overwritten.

Since the base request can fit in the remaining space, there is no need
to wraparound immediately. The wraparound will anyway happen later when
the reserved part starts getting used.

Cc: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Akash Goel <akash.goel@intel.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Link: http://patchwork.freedesktop.org/patch/msgid/1457688402-10411-1-git-send-email-akash.goel@intel.com
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: stable@vger.kernel.org
drivers/gpu/drm/i915/intel_lrc.c
drivers/gpu/drm/i915/intel_ringbuffer.c

index a1db6a02cf23592d72d085109c42585f362fbd45..67d48d759b46397bd24604242f986145e1d8b7cd 100644 (file)
@@ -854,11 +854,11 @@ static int logical_ring_prepare(struct drm_i915_gem_request *req, int bytes)
                if (unlikely(total_bytes > remain_usable)) {
                        /*
                         * The base request will fit but the reserved space
-                        * falls off the end. So only need to to wait for the
-                        * reserved size after flushing out the remainder.
+                        * falls off the end. So don't need an immediate wrap
+                        * and only need to effectively wait for the reserved
+                        * size space from the start of ringbuffer.
                         */
                        wait_bytes = remain_actual + ringbuf->reserved_size;
-                       need_wrap = true;
                } else if (total_bytes > ringbuf->space) {
                        /* No wrapping required, just waiting. */
                        wait_bytes = total_bytes;
index 6b4952031e30ce8a6d3fe35bfbf45fd8f42ae5ad..39aa318ad77917923ac24e09ec86a7d2925d2af8 100644 (file)
@@ -2478,11 +2478,11 @@ static int __intel_ring_prepare(struct intel_engine_cs *engine, int bytes)
                if (unlikely(total_bytes > remain_usable)) {
                        /*
                         * The base request will fit but the reserved space
-                        * falls off the end. So only need to to wait for the
-                        * reserved size after flushing out the remainder.
+                        * falls off the end. So don't need an immediate wrap
+                        * and only need to effectively wait for the reserved
+                        * size space from the start of ringbuffer.
                         */
                        wait_bytes = remain_actual + ringbuf->reserved_size;
-                       need_wrap = true;
                } else if (total_bytes > ringbuf->space) {
                        /* No wrapping required, just waiting. */
                        wait_bytes = total_bytes;