drm/i915: Only pwrite through the GTT if there is space in the aperture

[deliverable/linux.git] / drivers / gpu / drm / i915 / intel_ringbuffer.c

diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.c b/drivers/gpu/drm/i915/intel_ringbuffer.c
index 8733da529edf43fdf50aa3e33a531c7397356767..ac93643731aad1947d9ea1e2c7cc20fb19f69105 100644 (file)
--- a/drivers/gpu/drm/i915/intel_ringbuffer.c
+++ b/drivers/gpu/drm/i915/intel_ringbuffer.c
@@ -214,26 +214,35 @@ gen6_render_ring_flush(struct intel_ring_buffer *ring,
                          u32 invalidate_domains, u32 flush_domains)
 {
        u32 flags = 0;
+       struct pipe_control *pc = ring->private;
+       u32 scratch_addr = pc->gtt_offset + 128;
        int ret;
 
        /* Just flush everything.  Experiments have shown that reducing the
         * number of bits based on the write domains has little performance
         * impact.
         */
-       flags |= PIPE_CONTROL_RENDER_TARGET_CACHE_FLUSH;
-       flags |= PIPE_CONTROL_TLB_INVALIDATE;
-       flags |= PIPE_CONTROL_INSTRUCTION_CACHE_INVALIDATE;
-       flags |= PIPE_CONTROL_TEXTURE_CACHE_INVALIDATE;
-       flags |= PIPE_CONTROL_DEPTH_CACHE_FLUSH;
-       flags |= PIPE_CONTROL_VF_CACHE_INVALIDATE;
-       flags |= PIPE_CONTROL_CONST_CACHE_INVALIDATE;
-       flags |= PIPE_CONTROL_STATE_CACHE_INVALIDATE;
-       /*
-        * Ensure that any following seqno writes only happen when the render
-        * cache is indeed flushed (but only if the caller actually wants that).
-        */
-       if (flush_domains)
+       if (flush_domains) {
+               flags |= PIPE_CONTROL_RENDER_TARGET_CACHE_FLUSH;
+               flags |= PIPE_CONTROL_DEPTH_CACHE_FLUSH;
+               /*
+                * Ensure that any following seqno writes only happen
+                * when the render cache is indeed flushed.
+                */
                flags |= PIPE_CONTROL_CS_STALL;
+       }
+       if (invalidate_domains) {
+               flags |= PIPE_CONTROL_TLB_INVALIDATE;
+               flags |= PIPE_CONTROL_INSTRUCTION_CACHE_INVALIDATE;
+               flags |= PIPE_CONTROL_TEXTURE_CACHE_INVALIDATE;
+               flags |= PIPE_CONTROL_VF_CACHE_INVALIDATE;
+               flags |= PIPE_CONTROL_CONST_CACHE_INVALIDATE;
+               flags |= PIPE_CONTROL_STATE_CACHE_INVALIDATE;
+               /*
+                * TLB invalidate requires a post-sync write.
+                */
+               flags |= PIPE_CONTROL_QW_WRITE;
+       }
 
        ret = intel_ring_begin(ring, 4);
        if (ret)
@@ -241,7 +250,7 @@ gen6_render_ring_flush(struct intel_ring_buffer *ring,
 
        intel_ring_emit(ring, GFX_OP_PIPE_CONTROL(4));
        intel_ring_emit(ring, flags);
-       intel_ring_emit(ring, 0);
+       intel_ring_emit(ring, scratch_addr | PIPE_CONTROL_GLOBAL_GTT);
        intel_ring_emit(ring, 0);
        intel_ring_advance(ring);
 
@@ -294,8 +303,6 @@ static int init_ring_common(struct intel_ring_buffer *ring)
        I915_WRITE_HEAD(ring, 0);
        ring->write_tail(ring, 0);
 
-       /* Initialize the ring. */
-       I915_WRITE_START(ring, obj->gtt_offset);
        head = I915_READ_HEAD(ring) & HEAD_ADDR;
 
        /* G45 ring initialization fails to reset head to zero */
@@ -321,6 +328,11 @@ static int init_ring_common(struct intel_ring_buffer *ring)
                }
        }
 
+       /* Initialize the ring. This must happen _after_ we've cleared the ring
+        * registers with the above sequence (the readback of the HEAD registers
+        * also enforces ordering), otherwise the hw might lose the new ring
+        * register values. */
+       I915_WRITE_START(ring, obj->gtt_offset);
        I915_WRITE_CTL(ring,
                        ((ring->size - PAGE_SIZE) & RING_NR_PAGES)
                        | RING_VALID);
@@ -379,7 +391,7 @@ init_pipe_control(struct intel_ring_buffer *ring)
 
        i915_gem_object_set_cache_level(obj, I915_CACHE_LLC);
 
-       ret = i915_gem_object_pin(obj, 4096, true);
+       ret = i915_gem_object_pin(obj, 4096, true, false);
        if (ret)
                goto err_unref;
 
@@ -625,26 +637,24 @@ pc_render_add_request(struct intel_ring_buffer *ring,
 }
 
 static u32
-gen6_ring_get_seqno(struct intel_ring_buffer *ring)
+gen6_ring_get_seqno(struct intel_ring_buffer *ring, bool lazy_coherency)
 {
-       struct drm_device *dev = ring->dev;
-
        /* Workaround to force correct ordering between irq and seqno writes on
         * ivb (and maybe also on snb) by reading from a CS register (like
         * ACTHD) before reading the status page. */
-       if (IS_GEN6(dev) || IS_GEN7(dev))
+       if (!lazy_coherency)
                intel_ring_get_active_head(ring);
        return intel_read_status_page(ring, I915_GEM_HWS_INDEX);
 }
 
 static u32
-ring_get_seqno(struct intel_ring_buffer *ring)
+ring_get_seqno(struct intel_ring_buffer *ring, bool lazy_coherency)
 {
        return intel_read_status_page(ring, I915_GEM_HWS_INDEX);
 }
 
 static u32
-pc_render_get_seqno(struct intel_ring_buffer *ring)
+pc_render_get_seqno(struct intel_ring_buffer *ring, bool lazy_coherency)
 {
        struct pipe_control *pc = ring->private;
        return pc->cpu_page[0];
@@ -969,7 +979,7 @@ static int init_status_page(struct intel_ring_buffer *ring)
 
        i915_gem_object_set_cache_level(obj, I915_CACHE_LLC);
 
-       ret = i915_gem_object_pin(obj, 4096, true);
+       ret = i915_gem_object_pin(obj, 4096, true, false);
        if (ret != 0) {
                goto err_unref;
        }
@@ -1026,7 +1036,7 @@ static int intel_init_ring_buffer(struct drm_device *dev,
 
        ring->obj = obj;
 
-       ret = i915_gem_object_pin(obj, PAGE_SIZE, true);
+       ret = i915_gem_object_pin(obj, PAGE_SIZE, true, false);
        if (ret)
                goto err_unref;