Merge branch 'xa_branch'
[mesa.git] / src / mesa / drivers / dri / intel / intel_screen.c
index e240957197d5fb596d9a5ec32bc3768b363d0931..2a3a601ddbad721bdd5631f4a3179ab10cc00818 100644 (file)
  * 
  **************************************************************************/
 
+#include <errno.h>
 #include "main/glheader.h"
 #include "main/context.h"
 #include "main/framebuffer.h"
 #include "main/renderbuffer.h"
+#include "main/hash.h"
+#include "main/fbobject.h"
+#include "main/mfeatures.h"
 
 #include "utils.h"
 #include "xmlpool.h"
 
-#include "intel_batchbuffer.h"
-#include "intel_buffers.h"
-#include "intel_bufmgr.h"
-#include "intel_chipset.h"
-#include "intel_extensions.h"
-#include "intel_fbo.h"
-#include "intel_regions.h"
-#include "intel_screen.h"
-#include "intel_span.h"
-#include "intel_tex.h"
-
-#include "i915_drm.h"
-#include "i830_dri.h"
-
-#define DRI_CONF_TEXTURE_TILING(def) \
-       DRI_CONF_OPT_BEGIN(texture_tiling, bool, def)           \
-               DRI_CONF_DESC(en, "Enable texture tiling")      \
-       DRI_CONF_OPT_END                                        \
-
 PUBLIC const char __driConfigOptions[] =
    DRI_CONF_BEGIN
    DRI_CONF_SECTION_PERFORMANCE
@@ -66,17 +51,15 @@ PUBLIC const char __driConfigOptions[] =
         DRI_CONF_DESC_END
       DRI_CONF_OPT_END
 
-#ifdef I915
-     DRI_CONF_TEXTURE_TILING(false)
-#else
-     DRI_CONF_TEXTURE_TILING(true)
-#endif
+      DRI_CONF_OPT_BEGIN(texture_tiling, bool, true)
+        DRI_CONF_DESC(en, "Enable texture tiling")
+      DRI_CONF_OPT_END
 
       DRI_CONF_OPT_BEGIN(early_z, bool, false)
         DRI_CONF_DESC(en, "Enable early Z in classic mode (unstable, 945-only).")
       DRI_CONF_OPT_END
 
-      DRI_CONF_OPT_BEGIN(fragment_shader, bool, false)
+      DRI_CONF_OPT_BEGIN(fragment_shader, bool, true)
         DRI_CONF_DESC(en, "Enable limited ARB_fragment_shader support on 915/945.")
       DRI_CONF_OPT_END
 
@@ -98,15 +81,21 @@ DRI_CONF_END;
 
 const GLuint __driNConfigOptions = 11;
 
+#include "intel_batchbuffer.h"
+#include "intel_buffers.h"
+#include "intel_bufmgr.h"
+#include "intel_chipset.h"
+#include "intel_fbo.h"
+#include "intel_screen.h"
+#include "intel_tex.h"
+#include "intel_regions.h"
+
+#include "i915_drm.h"
+
 #ifdef USE_NEW_INTERFACE
 static PFNGLXCREATECONTEXTMODES create_context_modes = NULL;
 #endif /*USE_NEW_INTERFACE */
 
-static const __DRItexOffsetExtension intelTexOffsetExtension = {
-   { __DRI_TEX_OFFSET },
-   intelSetTexOffset,
-};
-
 static const __DRItexBufferExtension intelTexBufferExtension = {
     { __DRI_TEX_BUFFER, __DRI_TEX_BUFFER_VERSION },
    intelSetTexBuffer,
@@ -116,33 +105,223 @@ static const __DRItexBufferExtension intelTexBufferExtension = {
 static void
 intelDRI2Flush(__DRIdrawable *drawable)
 {
-   struct intel_context *intel = drawable->driContextPriv->driverPrivate;
+   GET_CURRENT_CONTEXT(ctx);
+   struct intel_context *intel = intel_context(ctx);
 
    if (intel->gen < 4)
       INTEL_FIREVERTICES(intel);
 
-   if (intel->batch->map != intel->batch->ptr)
-      intel_batchbuffer_flush(intel->batch);
-}
+   intel->need_throttle = GL_TRUE;
 
-static void
-intelDRI2FlushInvalidate(__DRIdrawable *drawable)
-{
-   intelDRI2Flush(drawable);
-   drawable->validBuffers = GL_FALSE;
+   if (intel->batch.used)
+      intel_batchbuffer_flush(intel);
 }
 
 static const struct __DRI2flushExtensionRec intelFlushExtension = {
     { __DRI2_FLUSH, __DRI2_FLUSH_VERSION },
     intelDRI2Flush,
-    intelDRI2FlushInvalidate,
+    dri2InvalidateDrawable,
+};
+
+static __DRIimage *
+intel_create_image_from_name(__DRIscreen *screen,
+                            int width, int height, int format,
+                            int name, int pitch, void *loaderPrivate)
+{
+    struct intel_screen *intelScreen = screen->private;
+    __DRIimage *image;
+    int cpp;
+
+    image = CALLOC(sizeof *image);
+    if (image == NULL)
+       return NULL;
+
+    switch (format) {
+    case __DRI_IMAGE_FORMAT_RGB565:
+       image->format = MESA_FORMAT_RGB565;
+       image->internal_format = GL_RGB;
+       image->data_type = GL_UNSIGNED_BYTE;
+       break;
+    case __DRI_IMAGE_FORMAT_XRGB8888:
+       image->format = MESA_FORMAT_XRGB8888;
+       image->internal_format = GL_RGB;
+       image->data_type = GL_UNSIGNED_BYTE;
+       break;
+    case __DRI_IMAGE_FORMAT_ARGB8888:
+       image->format = MESA_FORMAT_ARGB8888;
+       image->internal_format = GL_RGBA;
+       image->data_type = GL_UNSIGNED_BYTE;
+       break;
+    default:
+       free(image);
+       return NULL;
+    }
+
+    image->data = loaderPrivate;
+    cpp = _mesa_get_format_bytes(image->format);
+
+    image->region = intel_region_alloc_for_handle(intelScreen,
+                                                 cpp, width, height,
+                                                 pitch, name, "image");
+    if (image->region == NULL) {
+       FREE(image);
+       return NULL;
+    }
+
+    return image;      
+}
+
+static __DRIimage *
+intel_create_image_from_renderbuffer(__DRIcontext *context,
+                                    int renderbuffer, void *loaderPrivate)
+{
+   __DRIimage *image;
+   struct intel_context *intel = context->driverPrivate;
+   struct gl_renderbuffer *rb;
+   struct intel_renderbuffer *irb;
+
+   rb = _mesa_lookup_renderbuffer(&intel->ctx, renderbuffer);
+   if (!rb) {
+      _mesa_error(&intel->ctx,
+                 GL_INVALID_OPERATION, "glRenderbufferExternalMESA");
+      return NULL;
+   }
+
+   irb = intel_renderbuffer(rb);
+   image = CALLOC(sizeof *image);
+   if (image == NULL)
+      return NULL;
+
+   image->internal_format = rb->InternalFormat;
+   image->format = rb->Format;
+   image->data_type = rb->DataType;
+   image->data = loaderPrivate;
+   intel_region_reference(&image->region, irb->region);
+
+   return image;
+}
+
+static void
+intel_destroy_image(__DRIimage *image)
+{
+    intel_region_release(&image->region);
+    FREE(image);
+}
+
+static __DRIimage *
+intel_create_image(__DRIscreen *screen,
+                  int width, int height, int format,
+                  unsigned int use,
+                  void *loaderPrivate)
+{
+   __DRIimage *image;
+   struct intel_screen *intelScreen = screen->private;
+   uint32_t tiling;
+   int cpp;
+
+   tiling = I915_TILING_X;
+   if (use & __DRI_IMAGE_USE_CURSOR) {
+      if (width != 64 || height != 64)
+        return NULL;
+      tiling = I915_TILING_NONE;
+   }
+
+   image = CALLOC(sizeof *image);
+   if (image == NULL)
+      return NULL;
+
+   switch (format) {
+   case __DRI_IMAGE_FORMAT_RGB565:
+      image->format = MESA_FORMAT_RGB565;
+      image->internal_format = GL_RGB;
+      image->data_type = GL_UNSIGNED_BYTE;
+      break;
+   case __DRI_IMAGE_FORMAT_XRGB8888:
+      image->format = MESA_FORMAT_XRGB8888;
+      image->internal_format = GL_RGB;
+      image->data_type = GL_UNSIGNED_BYTE;
+      break;
+   case __DRI_IMAGE_FORMAT_ARGB8888:
+      image->format = MESA_FORMAT_ARGB8888;
+      image->internal_format = GL_RGBA;
+      image->data_type = GL_UNSIGNED_BYTE;
+      break;
+   default:
+      free(image);
+      return NULL;
+   }
+
+   image->data = loaderPrivate;
+   cpp = _mesa_get_format_bytes(image->format);
+
+   image->region =
+      intel_region_alloc(intelScreen, tiling,
+                        cpp, width, height, GL_TRUE);
+   if (image->region == NULL) {
+      FREE(image);
+      return NULL;
+   }
+   
+   return image;
+}
+
+static GLboolean
+intel_query_image(__DRIimage *image, int attrib, int *value)
+{
+   switch (attrib) {
+   case __DRI_IMAGE_ATTRIB_STRIDE:
+      *value = image->region->pitch * image->region->cpp;
+      return GL_TRUE;
+   case __DRI_IMAGE_ATTRIB_HANDLE:
+      *value = image->region->buffer->handle;
+      return GL_TRUE;
+   case __DRI_IMAGE_ATTRIB_NAME:
+      return intel_region_flink(image->region, (uint32_t *) value);
+   default:
+      return GL_FALSE;
+   }
+}
+
+static __DRIimage *
+intel_dup_image(__DRIimage *orig_image, void *loaderPrivate)
+{
+   __DRIimage *image;
+
+   image = CALLOC(sizeof *image);
+   if (image == NULL)
+      return NULL;
+
+   image->region = NULL;
+   intel_region_reference(&image->region, orig_image->region);
+   if (image->region == NULL) {
+      FREE(image);
+      return NULL;
+   }
+
+   image->internal_format = orig_image->internal_format;
+   image->format          = orig_image->format;
+   image->data_type       = orig_image->data_type;
+   image->data            = loaderPrivate;
+   
+   return image;
+}
+
+static struct __DRIimageExtensionRec intelImageExtension = {
+    { __DRI_IMAGE, __DRI_IMAGE_VERSION },
+    intel_create_image_from_name,
+    intel_create_image_from_renderbuffer,
+    intel_destroy_image,
+    intel_create_image,
+    intel_query_image,
+    intel_dup_image
 };
 
 static const __DRIextension *intelScreenExtensions[] = {
     &driReadDrawableExtension,
-    &intelTexOffsetExtension.base,
     &intelTexBufferExtension.base,
     &intelFlushExtension.base,
+    &intelImageExtension.base,
+    &dri2ConfigQueryExtension.base,
     NULL
 };
 
@@ -157,21 +336,40 @@ intel_get_param(__DRIscreen *psp, int param, int *value)
 
    ret = drmCommandWriteRead(psp->fd, DRM_I915_GETPARAM, &gp, sizeof(gp));
    if (ret) {
-      _mesa_warning(NULL, "drm_i915_getparam: %d", ret);
+      if (ret != -EINVAL)
+        _mesa_warning(NULL, "drm_i915_getparam: %d", ret);
       return GL_FALSE;
    }
 
    return GL_TRUE;
 }
 
+static GLboolean
+intel_get_boolean(__DRIscreen *psp, int param)
+{
+   int value = 0;
+   return intel_get_param(psp, param, &value) && value;
+}
+
+static void
+nop_callback(GLuint key, void *data, void *userData)
+{
+}
+
 static void
 intelDestroyScreen(__DRIscreen * sPriv)
 {
-   intelScreenPrivate *intelScreen = (intelScreenPrivate *) sPriv->private;
+   struct intel_screen *intelScreen = sPriv->private;
 
    dri_bufmgr_destroy(intelScreen->bufmgr);
    driDestroyOptionInfo(&intelScreen->optionCache);
 
+   /* Some regions may still have references to them at this point, so
+    * flush the hash table to prevent _mesa_DeleteHashTable() from
+    * complaining about the hash not being empty; */
+   _mesa_HashDeleteAll(intelScreen->named_regions, nop_callback, NULL);
+   _mesa_DeleteHashTable(intelScreen->named_regions);
+
    FREE(intelScreen);
    sPriv->private = NULL;
 }
@@ -183,16 +381,15 @@ intelDestroyScreen(__DRIscreen * sPriv)
 static GLboolean
 intelCreateBuffer(__DRIscreen * driScrnPriv,
                   __DRIdrawable * driDrawPriv,
-                  const __GLcontextModes * mesaVis, GLboolean isPixmap)
+                  const struct gl_config * mesaVis, GLboolean isPixmap)
 {
    struct intel_renderbuffer *rb;
+   struct intel_screen *screen = (struct intel_screen*) driScrnPriv->private;
 
    if (isPixmap) {
       return GL_FALSE;          /* not implemented */
    }
    else {
-      GLboolean swStencil = (mesaVis->stencilBits > 0 &&
-                             mesaVis->depthBits != 24);
       gl_format rgbFormat;
 
       struct gl_framebuffer *fb = CALLOC_STRUCT(gl_framebuffer);
@@ -200,7 +397,7 @@ intelCreateBuffer(__DRIscreen * driScrnPriv,
       if (!fb)
         return GL_FALSE;
 
-      _mesa_initialize_framebuffer(fb, mesaVis);
+      _mesa_initialize_window_framebuffer(fb, mesaVis);
 
       if (mesaVis->redBits == 5)
         rgbFormat = MESA_FORMAT_RGB565;
@@ -218,32 +415,53 @@ intelCreateBuffer(__DRIscreen * driScrnPriv,
          _mesa_add_renderbuffer(fb, BUFFER_BACK_LEFT, &rb->Base);
       }
 
+      /*
+       * Assert here that the gl_config has an expected depth/stencil bit
+       * combination: one of d24/s8, d16/s0, d0/s0. (See intelInitScreen2(),
+       * which constructs the advertised configs.)
+       */
       if (mesaVis->depthBits == 24) {
-        if (mesaVis->stencilBits == 8) {
-           /* combined depth/stencil buffer */
-           struct intel_renderbuffer *depthStencilRb
-              = intel_create_renderbuffer(MESA_FORMAT_S8_Z24);
-           /* note: bind RB to two attachment points */
-           _mesa_add_renderbuffer(fb, BUFFER_DEPTH, &depthStencilRb->Base);
-           _mesa_add_renderbuffer(fb, BUFFER_STENCIL, &depthStencilRb->Base);
+        assert(mesaVis->stencilBits == 8);
+
+        if (screen->hw_has_separate_stencil
+            && screen->dri2_has_hiz != INTEL_DRI2_HAS_HIZ_FALSE) {
+           /*
+            * Request a separate stencil buffer even if we do not yet know if
+            * the screen supports it. (See comments for
+            * enum intel_dri2_has_hiz).
+            */
+           rb = intel_create_renderbuffer(MESA_FORMAT_X8_Z24);
+           _mesa_add_renderbuffer(fb, BUFFER_DEPTH, &rb->Base);
+           rb = intel_create_renderbuffer(MESA_FORMAT_S8);
+           _mesa_add_renderbuffer(fb, BUFFER_STENCIL, &rb->Base);
         } else {
-           struct intel_renderbuffer *depthRb
-              = intel_create_renderbuffer(MESA_FORMAT_X8_Z24);
-           _mesa_add_renderbuffer(fb, BUFFER_DEPTH, &depthRb->Base);
+           /*
+            * Use combined depth/stencil. Note that the renderbuffer is
+            * attached to two attachment points.
+            */
+           rb = intel_create_renderbuffer(MESA_FORMAT_S8_Z24);
+           _mesa_add_renderbuffer(fb, BUFFER_DEPTH, &rb->Base);
+           _mesa_add_renderbuffer(fb, BUFFER_STENCIL, &rb->Base);
         }
       }
       else if (mesaVis->depthBits == 16) {
+        assert(mesaVis->stencilBits == 0);
          /* just 16-bit depth buffer, no hw stencil */
          struct intel_renderbuffer *depthRb
            = intel_create_renderbuffer(MESA_FORMAT_Z16);
          _mesa_add_renderbuffer(fb, BUFFER_DEPTH, &depthRb->Base);
       }
+      else {
+        assert(mesaVis->depthBits == 0);
+        assert(mesaVis->stencilBits == 0);
+      }
 
       /* now add any/all software-based renderbuffers we may need */
       _mesa_add_soft_renderbuffers(fb,
                                    GL_FALSE, /* never sw color */
                                    GL_FALSE, /* never sw depth */
-                                   swStencil, mesaVis->accumRedBits > 0,
+                                   GL_FALSE, /* never sw stencil */
+                                   mesaVis->accumRedBits > 0,
                                    GL_FALSE, /* never sw alpha */
                                    GL_FALSE  /* never sw aux */ );
       driDrawPriv->driverPrivate = fb;
@@ -264,29 +482,32 @@ intelDestroyBuffer(__DRIdrawable * driDrawPriv)
  * init-designated function to register chipids and createcontext
  * functions.
  */
-extern GLboolean i830CreateContext(const __GLcontextModes * mesaVis,
+extern GLboolean i830CreateContext(const struct gl_config * mesaVis,
                                    __DRIcontext * driContextPriv,
                                    void *sharedContextPrivate);
 
-extern GLboolean i915CreateContext(const __GLcontextModes * mesaVis,
+extern GLboolean i915CreateContext(int api,
+                                  const struct gl_config * mesaVis,
                                    __DRIcontext * driContextPriv,
                                    void *sharedContextPrivate);
-extern GLboolean brwCreateContext(const __GLcontextModes * mesaVis,
+extern GLboolean brwCreateContext(int api,
+                                 const struct gl_config * mesaVis,
                                  __DRIcontext * driContextPriv,
                                  void *sharedContextPrivate);
 
 static GLboolean
-intelCreateContext(const __GLcontextModes * mesaVis,
+intelCreateContext(gl_api api,
+                  const struct gl_config * mesaVis,
                    __DRIcontext * driContextPriv,
                    void *sharedContextPrivate)
 {
    __DRIscreen *sPriv = driContextPriv->driScreenPriv;
-   intelScreenPrivate *intelScreen = (intelScreenPrivate *) sPriv->private;
+   struct intel_screen *intelScreen = sPriv->private;
 
 #ifdef I915
    if (IS_9XX(intelScreen->deviceID)) {
       if (!IS_965(intelScreen->deviceID)) {
-        return i915CreateContext(mesaVis, driContextPriv,
+        return i915CreateContext(api, mesaVis, driContextPriv,
                                  sharedContextPrivate);
       }
    } else {
@@ -295,59 +516,118 @@ intelCreateContext(const __GLcontextModes * mesaVis,
    }
 #else
    if (IS_965(intelScreen->deviceID))
-      return brwCreateContext(mesaVis, driContextPriv, sharedContextPrivate);
+      return brwCreateContext(api, mesaVis,
+                             driContextPriv, sharedContextPrivate);
 #endif
-   fprintf(stderr, "Unrecognized deviceID %x\n", intelScreen->deviceID);
+   fprintf(stderr, "Unrecognized deviceID 0x%x\n", intelScreen->deviceID);
    return GL_FALSE;
 }
 
 static GLboolean
-intel_init_bufmgr(intelScreenPrivate *intelScreen)
+intel_init_bufmgr(struct intel_screen *intelScreen)
 {
    __DRIscreen *spriv = intelScreen->driScrnPriv;
    int num_fences = 0;
 
-   intelScreen->no_hw = getenv("INTEL_NO_HW") != NULL;
+   intelScreen->no_hw = (getenv("INTEL_NO_HW") != NULL ||
+                        getenv("INTEL_DEVID_OVERRIDE") != NULL);
 
    intelScreen->bufmgr = intel_bufmgr_gem_init(spriv->fd, BATCH_SZ);
-   /* Otherwise, use the classic buffer manager. */
    if (intelScreen->bufmgr == NULL) {
       fprintf(stderr, "[%s:%u] Error initializing buffer manager.\n",
              __func__, __LINE__);
       return GL_FALSE;
    }
 
-   if (intel_get_param(spriv, I915_PARAM_NUM_FENCES_AVAIL, &num_fences))
-      intelScreen->kernel_exec_fencing = !!num_fences;
-   else
-      intelScreen->kernel_exec_fencing = GL_FALSE;
+   if (!intel_get_param(spriv, I915_PARAM_NUM_FENCES_AVAIL, &num_fences) ||
+       num_fences == 0) {
+      fprintf(stderr, "[%s: %u] Kernel 2.6.29 required.\n", __func__, __LINE__);
+      return GL_FALSE;
+   }
+
+   drm_intel_bufmgr_gem_enable_fenced_relocs(intelScreen->bufmgr);
+
+   intelScreen->named_regions = _mesa_NewHashTable();
+
+   intelScreen->relaxed_relocations = 0;
+   intelScreen->relaxed_relocations |=
+      intel_get_boolean(spriv, I915_PARAM_HAS_RELAXED_DELTA) << 0;
 
    return GL_TRUE;
 }
 
+/**
+ * Override intel_screen.hw_has_hiz with environment variable INTEL_HIZ.
+ *
+ * Valid values for INTEL_HIZ are "0" and "1". If an invalid valid value is
+ * encountered, a warning is emitted and INTEL_HIZ is ignored.
+ */
+static void
+intel_override_hiz(struct intel_screen *intel)
+{
+   const char *s = getenv("INTEL_HIZ");
+   if (!s) {
+      return;
+   } else if (!strncmp("0", s, 2)) {
+      intel->hw_has_hiz = false;
+   } else if (!strncmp("1", s, 2)) {
+      intel->hw_has_hiz = true;
+   } else {
+      fprintf(stderr,
+             "warning: env variable INTEL_HIZ=\"%s\" has invalid value "
+             "and is ignored", s);
+   }
+}
+
+/**
+ * Override intel_screen.hw_has_separate_stencil with environment variable
+ * INTEL_SEPARATE_STENCIL.
+ *
+ * Valid values for INTEL_SEPARATE_STENCIL are "0" and "1". If an invalid
+ * valid value is encountered, a warning is emitted and INTEL_SEPARATE_STENCIL
+ * is ignored.
+ */
+static void
+intel_override_separate_stencil(struct intel_screen *screen)
+{
+   const char *s = getenv("INTEL_SEPARATE_STENCIL");
+   if (!s) {
+      return;
+   } else if (!strncmp("0", s, 2)) {
+      screen->hw_has_separate_stencil = false;
+   } else if (!strncmp("1", s, 2)) {
+      screen->hw_has_separate_stencil = true;
+   } else {
+      fprintf(stderr,
+             "warning: env variable INTEL_SEPARATE_STENCIL=\"%s\" has "
+             "invalid value and is ignored", s);
+   }
+}
+
 /**
  * This is the driver specific part of the createNewScreen entry point.
  * Called when using DRI2.
  *
- * \return the __GLcontextModes supported by this driver
+ * \return the struct gl_config supported by this driver
  */
 static const
 __DRIconfig **intelInitScreen2(__DRIscreen *psp)
 {
-   intelScreenPrivate *intelScreen;
+   struct intel_screen *intelScreen;
    GLenum fb_format[3];
    GLenum fb_type[3];
+   unsigned int api_mask;
+   char *devid_override;
 
    static const GLenum back_buffer_modes[] = {
-       GLX_NONE, GLX_SWAP_UNDEFINED_OML,
-       GLX_SWAP_EXCHANGE_OML, GLX_SWAP_COPY_OML
+       GLX_NONE, GLX_SWAP_UNDEFINED_OML, GLX_SWAP_COPY_OML
    };
    uint8_t depth_bits[4], stencil_bits[4], msaa_samples_array[1];
    int color;
    __DRIconfig **configs = NULL;
 
    /* Allocate the private area */
-   intelScreen = (intelScreenPrivate *) CALLOC(sizeof(intelScreenPrivate));
+   intelScreen = CALLOC(sizeof *intelScreen);
    if (!intelScreen) {
       fprintf(stderr, "\nERROR!  Allocating private area failed\n");
       return GL_FALSE;
@@ -359,28 +639,63 @@ __DRIconfig **intelInitScreen2(__DRIscreen *psp)
    intelScreen->driScrnPriv = psp;
    psp->private = (void *) intelScreen;
 
-   intelScreen->drmMinor = psp->drm_version.minor;
-
    /* Determine chipset ID */
    if (!intel_get_param(psp, I915_PARAM_CHIPSET_ID,
                        &intelScreen->deviceID))
       return GL_FALSE;
 
+   /* Allow an override of the device ID for the purpose of making the
+    * driver produce dumps for debugging of new chipset enablement.
+    * This implies INTEL_NO_HW, to avoid programming your actual GPU
+    * incorrectly.
+    */
+   devid_override = getenv("INTEL_DEVID_OVERRIDE");
+   if (devid_override) {
+      intelScreen->deviceID = strtod(devid_override, NULL);
+   }
+
+   if (IS_GEN7(intelScreen->deviceID)) {
+      intelScreen->gen = 7;
+   } else if (IS_GEN6(intelScreen->deviceID)) {
+      intelScreen->gen = 6;
+   } else if (IS_GEN5(intelScreen->deviceID)) {
+      intelScreen->gen = 5;
+   } else if (IS_965(intelScreen->deviceID)) {
+      intelScreen->gen = 4;
+   } else if (IS_9XX(intelScreen->deviceID)) {
+      intelScreen->gen = 3;
+   } else {
+      intelScreen->gen = 2;
+   }
+
+   /*
+    * FIXME: The hiz and separate stencil fields need updating once the
+    * FIXME: features are completely implemented for a given chipset.
+    */
+   intelScreen->hw_has_separate_stencil = intelScreen->gen >= 7;
+   intelScreen->hw_must_use_separate_stencil = intelScreen->gen >= 7;
+   intelScreen->hw_has_hiz = false;
+   intelScreen->dri2_has_hiz = INTEL_DRI2_HAS_HIZ_UNKNOWN;
+
+   intel_override_hiz(intelScreen);
+   intel_override_separate_stencil(intelScreen);
+
+   api_mask = (1 << __DRI_API_OPENGL);
+#if FEATURE_ES1
+   api_mask |= (1 << __DRI_API_GLES);
+#endif
+#if FEATURE_ES2
+   api_mask |= (1 << __DRI_API_GLES2);
+#endif
+
+   if (IS_9XX(intelScreen->deviceID) || IS_965(intelScreen->deviceID))
+      psp->api_mask = api_mask;
+
    if (!intel_init_bufmgr(intelScreen))
        return GL_FALSE;
 
-   intelScreen->irq_active = 1;
    psp->extensions = intelScreenExtensions;
 
-   depth_bits[0] = 0;
-   stencil_bits[0] = 0;
-   depth_bits[1] = 16;
-   stencil_bits[1] = 0;
-   depth_bits[2] = 24;
-   stencil_bits[2] = 0;
-   depth_bits[3] = 24;
-   stencil_bits[3] = 8;
-
    msaa_samples_array[0] = 0;
 
    fb_format[0] = GL_RGB;
@@ -395,27 +710,27 @@ __DRIconfig **intelInitScreen2(__DRIscreen *psp)
    depth_bits[0] = 0;
    stencil_bits[0] = 0;
 
+   /* Generate a rich set of useful configs that do not include an
+    * accumulation buffer.
+    */
    for (color = 0; color < ARRAY_SIZE(fb_format); color++) {
       __DRIconfig **new_configs;
       int depth_factor;
 
-      /* With DRI2 right now, GetBuffers always returns a depth/stencil buffer
-       * with the same cpp as the drawable.  So we can't support depth cpp !=
-       * color cpp currently.
+      /* Starting with DRI2 protocol version 1.1 we can request a depth/stencil
+       * buffer that has a diffferent number of bits per pixel than the color
+       * buffer.  This isn't yet supported here.
        */
       if (fb_type[color] == GL_UNSIGNED_SHORT_5_6_5) {
         depth_bits[1] = 16;
         stencil_bits[1] = 0;
-
-        depth_factor = 2;
       } else {
         depth_bits[1] = 24;
-        stencil_bits[1] = 0;
-        depth_bits[2] = 24;
-        stencil_bits[2] = 8;
-
-        depth_factor = 3;
+        stencil_bits[1] = 8;
       }
+
+      depth_factor = 2;
+
       new_configs = driCreateConfigs(fb_format[color], fb_type[color],
                                     depth_bits,
                                     stencil_bits,
@@ -423,7 +738,33 @@ __DRIconfig **intelInitScreen2(__DRIscreen *psp)
                                     back_buffer_modes,
                                     ARRAY_SIZE(back_buffer_modes),
                                     msaa_samples_array,
-                                    ARRAY_SIZE(msaa_samples_array));
+                                    ARRAY_SIZE(msaa_samples_array),
+                                    GL_FALSE);
+      if (configs == NULL)
+        configs = new_configs;
+      else
+        configs = driConcatConfigs(configs, new_configs);
+   }
+
+   /* Generate the minimum possible set of configs that include an
+    * accumulation buffer.
+    */
+   for (color = 0; color < ARRAY_SIZE(fb_format); color++) {
+      __DRIconfig **new_configs;
+
+      if (fb_type[color] == GL_UNSIGNED_SHORT_5_6_5) {
+        depth_bits[0] = 16;
+        stencil_bits[0] = 0;
+      } else {
+        depth_bits[0] = 24;
+        stencil_bits[0] = 8;
+      }
+
+      new_configs = driCreateConfigs(fb_format[color], fb_type[color],
+                                    depth_bits, stencil_bits, 1,
+                                    back_buffer_modes + 1, 1,
+                                    msaa_samples_array, 1,
+                                    GL_TRUE);
       if (configs == NULL)
         configs = new_configs;
       else
@@ -439,6 +780,60 @@ __DRIconfig **intelInitScreen2(__DRIscreen *psp)
    return (const __DRIconfig **)configs;
 }
 
+struct intel_buffer {
+   __DRIbuffer base;
+   struct intel_region *region;
+};
+
+static __DRIbuffer *
+intelAllocateBuffer(__DRIscreen *screen,
+                   unsigned attachment, unsigned format,
+                   int width, int height)
+{
+   struct intel_buffer *intelBuffer;
+   struct intel_screen *intelScreen = screen->private;
+   uint32_t tiling;
+
+   intelBuffer = CALLOC(sizeof *intelBuffer);
+   if (intelBuffer == NULL)
+      return NULL;
+
+   if ((attachment == __DRI_BUFFER_DEPTH ||
+       attachment == __DRI_BUFFER_STENCIL ||
+       attachment == __DRI_BUFFER_DEPTH_STENCIL) &&
+       intelScreen->gen >= 4)
+      tiling = I915_TILING_Y;
+   else
+      tiling = I915_TILING_X;
+
+   intelBuffer->region = intel_region_alloc(intelScreen, tiling,
+                                           format / 8, width, height, GL_TRUE);
+   
+   if (intelBuffer->region == NULL) {
+          FREE(intelBuffer);
+          return NULL;
+   }
+   
+   intel_region_flink(intelBuffer->region, &intelBuffer->base.name);
+
+   intelBuffer->base.attachment = attachment;
+   intelBuffer->base.cpp = intelBuffer->region->cpp;
+   intelBuffer->base.pitch =
+         intelBuffer->region->pitch * intelBuffer->region->cpp;
+
+   return &intelBuffer->base;
+}
+
+static void
+intelReleaseBuffer(__DRIscreen *screen, __DRIbuffer *buffer)
+{
+   struct intel_buffer *intelBuffer = (struct intel_buffer *) buffer;
+
+   intel_region_release(&intelBuffer->region);
+   free(intelBuffer);
+}
+
+
 const struct __DriverAPIRec driDriverAPI = {
    .DestroyScreen       = intelDestroyScreen,
    .CreateContext       = intelCreateContext,
@@ -448,6 +843,8 @@ const struct __DriverAPIRec driDriverAPI = {
    .MakeCurrent                 = intelMakeCurrent,
    .UnbindContext       = intelUnbindContext,
    .InitScreen2                 = intelInitScreen2,
+   .AllocateBuffer       = intelAllocateBuffer,
+   .ReleaseBuffer        = intelReleaseBuffer
 };
 
 /* This is the table of extensions that the loader will dlsym() for. */