intel: Annotate debug printout checks with unlikely().

This provides the optimizer with hints about code hotness, which we're
quite certain about for debug printouts (or, rather, while we
developers often hit the checks for debug printouts, we don't care
about performance while doing so).

10 files changed, 67 insertions, 75 deletions

diff --git a/src/mesa/drivers/dri/intel/intel_batchbuffer.c b/src/mesa/drivers/dri/intel/intel_batchbuffer.c
index aaf3a573cf7..4b498f8c5b2 100644
--- a/src/mesa/drivers/dri/intel/intel_batchbuffer.c
+++ b/src/mesa/drivers/dri/intel/intel_batchbuffer.c
@@ -97,7 +97,7 @@ do_flush_locked(struct intel_batchbuffer *batch, GLuint used)
 			(x_off & 0xffff) | (y_off << 16));
    }
 
-   if (INTEL_DEBUG & DEBUG_BATCH) {
+   if (unlikely(INTEL_DEBUG & DEBUG_BATCH)) {
       drm_intel_bo_map(batch->buf, GL_FALSE);
       intel_decode(batch->buf->virtual, used / 4, batch->buf->offset,
 		   intel->intelScreen->deviceID, GL_TRUE);
@@ -128,7 +128,7 @@ _intel_batchbuffer_flush(struct intel_batchbuffer *batch, const char *file,
    if (used == 0)
       return;
 
-   if (INTEL_DEBUG & DEBUG_BATCH)
+   if (unlikely(INTEL_DEBUG & DEBUG_BATCH))
       fprintf(stderr, "%s:%d: Batchbuffer flush with %db used\n", file, line,
 	      used);
 
@@ -172,7 +172,7 @@ _intel_batchbuffer_flush(struct intel_batchbuffer *batch, const char *file,
 
    do_flush_locked(batch, used);
 
-   if (INTEL_DEBUG & DEBUG_SYNC) {
+   if (unlikely(INTEL_DEBUG & DEBUG_SYNC)) {
       fprintf(stderr, "waiting for idle\n");
       drm_intel_bo_map(batch->buf, GL_TRUE);
       drm_intel_bo_unmap(batch->buf);
diff --git a/src/mesa/drivers/dri/intel/intel_clear.c b/src/mesa/drivers/dri/intel/intel_clear.c
index d7814635b72..fa451f0045e 100644
--- a/src/mesa/drivers/dri/intel/intel_clear.c
+++ b/src/mesa/drivers/dri/intel/intel_clear.c
@@ -58,6 +58,21 @@ static const char *buffer_names[] = {
    [BUFFER_COLOR7] = "color7",
 };
 
+static void
+debug_mask(const char *name, GLbitfield mask)
+{
+   GLuint i;
+
+   if (unlikely(INTEL_DEBUG & DEBUG_BLIT)) {
+      DBG("%s clear:", name);
+      for (i = 0; i < BUFFER_COUNT; i++) {
+	 if (mask & (1 << i))
+	    DBG(" %s", buffer_names[i]);
+      }
+      DBG("\n");
+   }
+}
+
 /**
  * Called by ctx->Driver.Clear.
  */
@@ -70,7 +85,6 @@ intelClear(struct gl_context *ctx, GLbitfield mask)
    GLbitfield blit_mask = 0;
    GLbitfield swrast_mask = 0;
    struct gl_framebuffer *fb = ctx->DrawBuffer;
-   GLuint i;
 
    if (mask & (BUFFER_BIT_FRONT_LEFT | BUFFER_BIT_FRONT_RIGHT)) {
       intel->front_buffer_dirty = GL_TRUE;
@@ -162,39 +176,17 @@ intelClear(struct gl_context *ctx, GLbitfield mask)
    }
 
    if (blit_mask) {
-      if (INTEL_DEBUG & DEBUG_BLIT) {
-	 DBG("blit clear:");
-	 for (i = 0; i < BUFFER_COUNT; i++) {
-	    if (blit_mask & (1 << i))
-	       DBG(" %s", buffer_names[i]);
-	 }
-	 DBG("\n");
-      }
+      debug_mask("blit", blit_mask);
       intelClearWithBlit(ctx, blit_mask);
    }
 
    if (tri_mask) {
-      if (INTEL_DEBUG & DEBUG_BLIT) {
-	 DBG("tri clear:");
-	 for (i = 0; i < BUFFER_COUNT; i++) {
-	    if (tri_mask & (1 << i))
-	       DBG(" %s", buffer_names[i]);
-	 }
-	 DBG("\n");
-      }
-
+      debug_mask("tri", tri_mask);
       _mesa_meta_Clear(&intel->ctx, tri_mask);
    }
 
    if (swrast_mask) {
-      if (INTEL_DEBUG & DEBUG_BLIT) {
-	 DBG("swrast clear:");
-	 for (i = 0; i < BUFFER_COUNT; i++) {
-	    if (swrast_mask & (1 << i))
-	       DBG(" %s", buffer_names[i]);
-	 }
-	 DBG("\n");
-      }
+      debug_mask("swrast", swrast_mask);
       _swrast_Clear(ctx, swrast_mask);
    }
 }
diff --git a/src/mesa/drivers/dri/intel/intel_context.c b/src/mesa/drivers/dri/intel/intel_context.c
index 7ace50bde97..de4bcde20b6 100644
--- a/src/mesa/drivers/dri/intel/intel_context.c
+++ b/src/mesa/drivers/dri/intel/intel_context.c
@@ -249,7 +249,7 @@ intel_update_renderbuffers(__DRIcontext *context, __DRIdrawable *drawable)
     * thus ignore the invalidate. */
    drawable->lastStamp = drawable->dri2.stamp;
 
-   if (INTEL_DEBUG & DEBUG_DRI)
+   if (unlikely(INTEL_DEBUG & DEBUG_DRI))
       fprintf(stderr, "enter %s, drawable %p\n", __func__, drawable);
 
    screen = intel->intelScreen->driScrnPriv;
@@ -378,14 +378,14 @@ intel_update_renderbuffers(__DRIcontext *context, __DRIdrawable *drawable)
        if (rb->region && rb->region->name == buffers[i].name)
 	     continue;
 
-       if (INTEL_DEBUG & DEBUG_DRI)
+       if (unlikely(INTEL_DEBUG & DEBUG_DRI))
 	  fprintf(stderr,
 		  "attaching buffer %d, at %d, cpp %d, pitch %d\n",
 		  buffers[i].name, buffers[i].attachment,
 		  buffers[i].cpp, buffers[i].pitch);
        
        if (buffers[i].attachment == __DRI_BUFFER_STENCIL && depth_region) {
-	  if (INTEL_DEBUG & DEBUG_DRI)
+	  if (unlikely(INTEL_DEBUG & DEBUG_DRI))
 	     fprintf(stderr, "(reusing depth buffer as stencil)\n");
 	  intel_region_reference(&region, depth_region);
        }
diff --git a/src/mesa/drivers/dri/intel/intel_context.h b/src/mesa/drivers/dri/intel/intel_context.h
index 46d10d74ba3..2ec65b83026 100644
--- a/src/mesa/drivers/dri/intel/intel_context.h
+++ b/src/mesa/drivers/dri/intel/intel_context.h
@@ -98,6 +98,16 @@ extern void intelFallback(struct intel_context *intel, GLbitfield bit,
 
 #define INTEL_MAX_FIXUP 64
 
+#ifndef likely
+#ifdef __GNUC__
+#define likely(expr) (__builtin_expect(expr, 1))
+#define unlikely(expr) (__builtin_expect(expr, 0))
+#else
+#define likely(expr) (expr)
+#define unlikely(expr) (expr)
+#endif
+#endif
+
 struct intel_sync_object {
    struct gl_sync_object Base;
 
@@ -359,10 +369,15 @@ extern int INTEL_DEBUG;
 #define DEBUG_CLIP      0x8000000
 
 #define DBG(...) do {						\
-	if (INTEL_DEBUG & FILE_DEBUG_FLAG)			\
+	if (unlikely(INTEL_DEBUG & FILE_DEBUG_FLAG))		\
 		printf(__VA_ARGS__);			\
 } while(0)
 
+#define fallback_debug(...) do {				\
+	if (unlikely(INTEL_DEBUG & DEBUG_FALLBACKS))		\
+		printf(__VA_ARGS__);				\
+} while(0)
+
 #define PCI_CHIP_845_G			0x2562
 #define PCI_CHIP_I830_M			0x3577
 #define PCI_CHIP_I855_GM		0x3582
diff --git a/src/mesa/drivers/dri/intel/intel_pixel.c b/src/mesa/drivers/dri/intel/intel_pixel.c
index 60583ef4c0d..d5c35775ce4 100644
--- a/src/mesa/drivers/dri/intel/intel_pixel.c
+++ b/src/mesa/drivers/dri/intel/intel_pixel.c
@@ -147,10 +147,9 @@ intel_check_blit_format(struct intel_region * region,
       return GL_TRUE;
    }
 
-   if (INTEL_DEBUG & DEBUG_PIXEL)
-      fprintf(stderr, "%s: bad format for blit (cpp %d, type %s format %s)\n",
-              __FUNCTION__, region->cpp,
-              _mesa_lookup_enum_by_nr(type), _mesa_lookup_enum_by_nr(format));
+   DBG("%s: bad format for blit (cpp %d, type %s format %s)\n",
+       __FUNCTION__, region->cpp,
+       _mesa_lookup_enum_by_nr(type), _mesa_lookup_enum_by_nr(format));
 
    return GL_FALSE;
 }
diff --git a/src/mesa/drivers/dri/intel/intel_pixel_bitmap.c b/src/mesa/drivers/dri/intel/intel_pixel_bitmap.c
index 63fb4b37b18..e7356a6da0d 100644
--- a/src/mesa/drivers/dri/intel/intel_pixel_bitmap.c
+++ b/src/mesa/drivers/dri/intel/intel_pixel_bitmap.c
@@ -113,9 +113,8 @@ static GLuint get_bitmap_rect(GLsizei width, GLsizei height,
    GLint incr;
    GLuint count = 0;
 
-   if (INTEL_DEBUG & DEBUG_PIXEL)
-      printf("%s %d,%d %dx%d bitmap %dx%d skip %d src_offset %d mask %d\n",
-		   __FUNCTION__, x,y,w,h,width,height,unpack->SkipPixels, src_offset, mask);
+   DBG("%s %d,%d %dx%d bitmap %dx%d skip %d src_offset %d mask %d\n",
+       __FUNCTION__, x,y,w,h,width,height,unpack->SkipPixels, src_offset, mask);
 
    if (invert) {
       first = h-1;
@@ -285,7 +284,7 @@ do_blit_bitmap( struct gl_context *ctx,
    }
 out:
 
-   if (INTEL_DEBUG & DEBUG_SYNC)
+   if (unlikely(INTEL_DEBUG & DEBUG_SYNC))
       intel_batchbuffer_flush(intel->batch);
 
    if (_mesa_is_bufferobj(unpack->BufferObj)) {
@@ -299,6 +298,7 @@ out:
    return GL_TRUE;
 }
 
+
 /* There are a large number of possible ways to implement bitmap on
  * this hardware, most of them have some sort of drawback.  Here are a
  * few that spring to mind:
diff --git a/src/mesa/drivers/dri/intel/intel_pixel_copy.c b/src/mesa/drivers/dri/intel/intel_pixel_copy.c
index c6b36ed4291..a7ca780e944 100644
--- a/src/mesa/drivers/dri/intel/intel_pixel_copy.c
+++ b/src/mesa/drivers/dri/intel/intel_pixel_copy.c
@@ -119,8 +119,7 @@ do_blit_copypixels(struct gl_context * ctx,
    GLboolean flip = GL_FALSE;
 
    if (type == GL_DEPTH || type == GL_STENCIL) {
-      if (INTEL_DEBUG & DEBUG_FALLBACKS)
-	 fprintf(stderr, "glCopyPixels() fallback: GL_DEPTH || GL_STENCIL\n");
+      fallback_debug("glCopyPixels() fallback: GL_DEPTH || GL_STENCIL\n");
       return GL_FALSE;
    }
 
@@ -203,8 +202,7 @@ intelCopyPixels(struct gl_context * ctx,
                 GLsizei width, GLsizei height,
                 GLint destx, GLint desty, GLenum type)
 {
-   if (INTEL_DEBUG & DEBUG_PIXEL)
-      fprintf(stderr, "%s\n", __FUNCTION__);
+   DBG("%s\n", __FUNCTION__);
 
    if (do_blit_copypixels(ctx, srcx, srcy, width, height, destx, desty, type))
       return;
diff --git a/src/mesa/drivers/dri/intel/intel_pixel_read.c b/src/mesa/drivers/dri/intel/intel_pixel_read.c
index b249f9a5a0b..54da29236d2 100644
--- a/src/mesa/drivers/dri/intel/intel_pixel_read.c
+++ b/src/mesa/drivers/dri/intel/intel_pixel_read.c
@@ -42,6 +42,8 @@
 #include "intel_pixel.h"
 #include "intel_buffer_objects.h"
 
+#define FILE_DEBUG_FLAG DEBUG_PIXEL
+
 /* For many applications, the new ability to pull the source buffers
  * back out of the GTT and then do the packing/conversion operations
  * in software will be as much of an improvement as trying to get the
@@ -79,8 +81,7 @@ do_blit_readpixels(struct gl_context * ctx,
    GLboolean all;
    GLint dst_x, dst_y;
 
-   if (INTEL_DEBUG & DEBUG_PIXEL)
-      printf("%s\n", __FUNCTION__);
+   DBG("%s\n", __FUNCTION__);
 
    if (!src)
       return GL_FALSE;
@@ -88,22 +89,19 @@ do_blit_readpixels(struct gl_context * ctx,
    if (!_mesa_is_bufferobj(pack->BufferObj)) {
       /* PBO only for now:
        */
-      if (INTEL_DEBUG & DEBUG_PIXEL)
-         printf("%s - not PBO\n", __FUNCTION__);
+      DBG("%s - not PBO\n", __FUNCTION__);
       return GL_FALSE;
    }
 
 
    if (ctx->_ImageTransferState ||
        !intel_check_blit_format(src, format, type)) {
-      if (INTEL_DEBUG & DEBUG_PIXEL)
-         printf("%s - bad format for blit\n", __FUNCTION__);
+      DBG("%s - bad format for blit\n", __FUNCTION__);
       return GL_FALSE;
    }
 
    if (pack->Alignment != 1 || pack->SwapBytes || pack->LsbFirst) {
-      if (INTEL_DEBUG & DEBUG_PIXEL)
-         printf("%s: bad packing params\n", __FUNCTION__);
+      DBG("%s: bad packing params\n", __FUNCTION__);
       return GL_FALSE;
    }
 
@@ -113,8 +111,7 @@ do_blit_readpixels(struct gl_context * ctx,
       rowLength = width;
 
    if (pack->Invert) {
-      if (INTEL_DEBUG & DEBUG_PIXEL)
-         printf("%s: MESA_PACK_INVERT not done yet\n", __FUNCTION__);
+      DBG("%s: MESA_PACK_INVERT not done yet\n", __FUNCTION__);
       return GL_FALSE;
    }
    else {
@@ -158,8 +155,7 @@ do_blit_readpixels(struct gl_context * ctx,
       return GL_FALSE;
    }
 
-   if (INTEL_DEBUG & DEBUG_PIXEL)
-      printf("%s - DONE\n", __FUNCTION__);
+   DBG("%s - DONE\n", __FUNCTION__);
 
    return GL_TRUE;
 }
@@ -173,8 +169,7 @@ intelReadPixels(struct gl_context * ctx,
    struct intel_context *intel = intel_context(ctx);
    GLboolean dirty;
 
-   if (INTEL_DEBUG & DEBUG_PIXEL)
-      fprintf(stderr, "%s\n", __FUNCTION__);
+   DBG("%s\n", __FUNCTION__);
 
    intel_flush(ctx);
 
@@ -188,8 +183,7 @@ intelReadPixels(struct gl_context * ctx,
        (ctx, x, y, width, height, format, type, pack, pixels))
       return;
 
-   if (INTEL_DEBUG & DEBUG_PIXEL)
-      printf("%s: fallback to swrast\n", __FUNCTION__);
+   fallback_debug("%s: fallback to swrast\n", __FUNCTION__);
 
    /* Update Mesa state before calling down into _swrast_ReadPixels, as
     * the spans code requires the computed buffer states to be up to date,
diff --git a/src/mesa/drivers/dri/intel/intel_tex.c b/src/mesa/drivers/dri/intel/intel_tex.c
index 3d9a2549db0..743a0432e43 100644
--- a/src/mesa/drivers/dri/intel/intel_tex.c
+++ b/src/mesa/drivers/dri/intel/intel_tex.c
@@ -158,8 +158,7 @@ intelGenerateMipmap(struct gl_context *ctx, GLenum target,
       struct intel_context *intel = intel_context(ctx);
       struct intel_texture_object *intelObj = intel_texture_object(texObj);
 
-      if (INTEL_DEBUG & DEBUG_FALLBACKS)
-	 fprintf(stderr, "%s - fallback to swrast\n", __FUNCTION__);
+      fallback_debug("%s - fallback to swrast\n", __FUNCTION__);
 
       intel_tex_map_level_images(intel, intelObj, texObj->BaseLevel);
       _mesa_generate_mipmap(ctx, target, texObj);
diff --git a/src/mesa/drivers/dri/intel/intel_tex_copy.c b/src/mesa/drivers/dri/intel/intel_tex_copy.c
index 2d046fd52d9..284ba19e8a3 100644
--- a/src/mesa/drivers/dri/intel/intel_tex_copy.c
+++ b/src/mesa/drivers/dri/intel/intel_tex_copy.c
@@ -105,16 +105,15 @@ do_copy_texsubimage(struct intel_context *intel,
    const struct intel_region *src = get_teximage_source(intel, internalFormat);
 
    if (!intelImage->mt || !src || !src->buffer) {
-      if (INTEL_DEBUG & DEBUG_FALLBACKS)
+      if (unlikely(INTEL_DEBUG & DEBUG_FALLBACKS))
 	 fprintf(stderr, "%s fail %p %p (0x%08x)\n",
 		 __FUNCTION__, intelImage->mt, src, internalFormat);
       return GL_FALSE;
    }
 
    if (intelImage->mt->cpp != src->cpp) {
-      if (INTEL_DEBUG & DEBUG_FALLBACKS)
-	 fprintf(stderr, "%s fail %d vs %d cpp\n",
-		 __FUNCTION__, intelImage->mt->cpp, src->cpp);
+      fallback_debug("%s fail %d vs %d cpp\n",
+		     __FUNCTION__, intelImage->mt->cpp, src->cpp);
       return GL_FALSE;
    }
 
@@ -212,8 +211,7 @@ intelCopyTexImage1D(struct gl_context * ctx, GLenum target, GLint level,
    return;
 
  fail:
-   if (INTEL_DEBUG & DEBUG_FALLBACKS)
-      fprintf(stderr, "%s - fallback to swrast\n", __FUNCTION__);
+   fallback_debug("%s - fallback to swrast\n", __FUNCTION__);
    _mesa_meta_CopyTexImage1D(ctx, target, level, internalFormat, x, y,
                              width, border);
 }
@@ -261,8 +259,7 @@ intelCopyTexImage2D(struct gl_context * ctx, GLenum target, GLint level,
    return;
 
  fail:
-   if (INTEL_DEBUG & DEBUG_FALLBACKS)
-      fprintf(stderr, "%s - fallback to swrast\n", __FUNCTION__);
+   fallback_debug("%s - fallback to swrast\n", __FUNCTION__);
    _mesa_meta_CopyTexImage2D(ctx, target, level, internalFormat, x, y,
                              width, height, border);
 }
@@ -287,8 +284,7 @@ intelCopyTexSubImage1D(struct gl_context * ctx, GLenum target, GLint level,
    if (!do_copy_texsubimage(intel_context(ctx), target,
                             intel_texture_image(texImage),
                             internalFormat, xoffset, 0, x, y, width, 1)) {
-      if (INTEL_DEBUG & DEBUG_FALLBACKS)
-         fprintf(stderr, "%s - fallback to swrast\n", __FUNCTION__);
+      fallback_debug("%s - fallback to swrast\n", __FUNCTION__);
       _mesa_meta_CopyTexSubImage1D(ctx, target, level, xoffset, x, y, width);
    }
 }
@@ -314,8 +310,7 @@ intelCopyTexSubImage2D(struct gl_context * ctx, GLenum target, GLint level,
                             internalFormat,
                             xoffset, yoffset, x, y, width, height)) {
 
-      if (INTEL_DEBUG & DEBUG_FALLBACKS)
-         fprintf(stderr, "%s - fallback to swrast\n", __FUNCTION__);
+      fallback_debug("%s - fallback to swrast\n", __FUNCTION__);
       _mesa_meta_CopyTexSubImage2D(ctx, target, level,
                                    xoffset, yoffset, x, y, width, height);
    }