[intel] Clean up cliprect handling in intel drivers.

In particular, batch buffers are no longer flushed when switching from
CLIPRECTS to NO_CLIPRECTS or vice versa, and 965 just uses DRM cliprect
handling for primitives instead of trying to sneak in its own to avoid the
DRM stuff. The disadvantage is that we will re-execute state updates per
cliprect, but the advantage is that we will be able to accumulate larger
batch buffers, which were proving to be a major overhead.

1 files changed, 24 insertions, 12 deletions

diff --git a/src/mesa/drivers/dri/intel/intel_batchbuffer.h b/src/mesa/drivers/dri/intel/intel_batchbuffer.h
index 1aa86ae3f0..5b6e0a19a5 100644
--- a/src/mesa/drivers/dri/intel/intel_batchbuffer.h
+++ b/src/mesa/drivers/dri/intel/intel_batchbuffer.h
@@ -10,8 +10,10 @@ struct intel_context;
 #define BATCH_SZ 16384
 #define BATCH_RESERVED 16
 
-#define INTEL_BATCH_NO_CLIPRECTS 0x1
-#define INTEL_BATCH_CLIPRECTS    0x2
+enum cliprects_enable {
+   INTEL_BATCH_CLIPRECTS = 0,
+   INTEL_BATCH_NO_CLIPRECTS = 1
+};
 
 struct intel_batchbuffer
 {
@@ -19,11 +21,12 @@ struct intel_batchbuffer
 
    dri_bo *buf;
    dri_fence *last_fence;
-   GLuint flags;
 
    GLubyte *map;
    GLubyte *ptr;
 
+   enum cliprects_enable cliprects_enable;
+
    GLuint size;
 
    GLuint dirty_state;
@@ -48,7 +51,8 @@ void intel_batchbuffer_reset(struct intel_batchbuffer *batch);
  * intel_buffer_dword() calls.
  */
 void intel_batchbuffer_data(struct intel_batchbuffer *batch,
-                            const void *data, GLuint bytes, GLuint flags);
+                            const void *data, GLuint bytes,
+			    enum cliprects_enable cliprects_enable);
 
 void intel_batchbuffer_release_space(struct intel_batchbuffer *batch,
                                      GLuint bytes);
@@ -80,29 +84,37 @@ intel_batchbuffer_emit_dword(struct intel_batchbuffer *batch, GLuint dword)
 
 static INLINE void
 intel_batchbuffer_require_space(struct intel_batchbuffer *batch,
-                                GLuint sz, GLuint flags)
+                                GLuint sz,
+				enum cliprects_enable cliprects_enable)
 {
    assert(sz < batch->size - 8);
-   if (intel_batchbuffer_space(batch) < sz ||
-       (batch->flags != 0 && flags != 0 && batch->flags != flags))
+   if (intel_batchbuffer_space(batch) < sz)
       intel_batchbuffer_flush(batch);
 
-   batch->flags |= flags;
+   /* Upgrade the buffer to being looped over per cliprect if this batch
+    * emit needs it.  The code used to emit a batch whenever the
+    * cliprects_enable was changed, but reducing the overhead of frequent
+    * batch flushing is more important than reducing state parsing,
+    * particularly as we move towards private backbuffers and number
+    * cliprects always being 1 except at swap.
+    */
+   if (cliprects_enable == INTEL_BATCH_CLIPRECTS)
+      batch->cliprects_enable = INTEL_BATCH_CLIPRECTS;
 }
 
 /* Here are the crusty old macros, to be removed:
  */
 #define BATCH_LOCALS
 
-#define BEGIN_BATCH(n, flags) do {				\
-   intel_batchbuffer_require_space(intel->batch, (n)*4, flags);	\
+#define BEGIN_BATCH(n, cliprects_enable) do {				\
+   intel_batchbuffer_require_space(intel->batch, (n)*4, cliprects_enable); \
 } while (0)
 
 #define OUT_BATCH(d)  intel_batchbuffer_emit_dword(intel->batch, d)
 
-#define OUT_RELOC(buf, flags, delta) do { 				\
+#define OUT_RELOC(buf, cliprects_enable, delta) do { 			\
    assert((delta) >= 0);						\
-   intel_batchbuffer_emit_reloc(intel->batch, buf, flags, delta);	\
+   intel_batchbuffer_emit_reloc(intel->batch, buf, cliprects_enable, delta); \
 } while (0)
 
 #define ADVANCE_BATCH() do { } while(0)