Merge remote-tracking branch 'origin/released'

Conflicts:
	libxcb/src/xcb_util.c
	mesalib/src/mesa/drivers/dri/common/dri_util.c
	xorg-server/os/utils.c

33 files changed, 1106 insertions, 828 deletions

diff --git a/mesalib/src/mesa/drivers/common/meta.c b/mesalib/src/mesa/drivers/common/meta.c
index ac27abba6..ab86f9c96 100644
--- a/mesalib/src/mesa/drivers/common/meta.c
+++ b/mesalib/src/mesa/drivers/common/meta.c
@@ -2328,7 +2328,7 @@ _mesa_meta_Bitmap(struct gl_context *ctx,
    if (ctx->_ImageTransferState ||
        ctx->FragmentProgram._Enabled ||
        ctx->Fog.Enabled ||
-       ctx->Texture._EnabledUnits ||
+       ctx->Texture._MaxEnabledTexImageUnit != -1 ||
        width > tex->MaxSize ||
        height > tex->MaxSize) {
       _swrast_Bitmap(ctx, x, y, width, height, unpack, bitmap1);
@@ -3265,7 +3265,7 @@ _mesa_meta_DrawTex(struct gl_context *ctx, GLfloat x, GLfloat y, GLfloat z,
          GLfloat s, t, s1, t1;
          GLuint tw, th;
 
-         if (!ctx->Texture.Unit[i]._ReallyEnabled) {
+         if (!ctx->Texture.Unit[i]._Current) {
             GLuint j;
             for (j = 0; j < 4; j++) {
                verts[j].st[i][0] = 0.0f;
diff --git a/mesalib/src/mesa/drivers/dri/common/SConscript b/mesalib/src/mesa/drivers/dri/common/SConscript
index 8b15532e3..41f6356c3 100644
--- a/mesalib/src/mesa/drivers/dri/common/SConscript
+++ b/mesalib/src/mesa/drivers/dri/common/SConscript
@@ -30,6 +30,7 @@ drienv.Replace(CPPPATH = [
 driswenv = drienv.Clone()
 driswenv.Append(CPPDEFINES = [
     '__NOT_HAVE_DRM_H',
+    'HAVE_DLADDR',
 ])
 
 drienv.PkgUseModules('DRM')
diff --git a/mesalib/src/mesa/drivers/dri/common/dri_util.c b/mesalib/src/mesa/drivers/dri/common/dri_util.c
index 78cfcf536..bb07a9541 100644..100755
--- a/mesalib/src/mesa/drivers/dri/common/dri_util.c
+++ b/mesalib/src/mesa/drivers/dri/common/dri_util.c
@@ -453,10 +453,7 @@ driContextSetFlags(struct gl_context *ctx, uint32_t flags)
     if ((flags & __DRI_CTX_FLAG_FORWARD_COMPATIBLE) != 0)
         ctx->Const.ContextFlags |= GL_CONTEXT_FLAG_FORWARD_COMPATIBLE_BIT;
     if ((flags & __DRI_CTX_FLAG_DEBUG) != 0) {
-        struct gl_debug_state *debug = _mesa_get_debug_state(ctx);
-        if (debug) {
-            debug->DebugOutput = GL_TRUE;
-        }
+       _mesa_set_debug_state_int(ctx, GL_DEBUG_OUTPUT, GL_TRUE);
         ctx->Const.ContextFlags |= GL_CONTEXT_FLAG_DEBUG_BIT;
     }
 }
@@ -782,7 +779,7 @@ const __DRIswrastExtension driSWRastExtension = {
 };
 
 const __DRI2configQueryExtension dri2ConfigQueryExtension = {
-   /*.base =*/ { __DRI2_CONFIG_QUERY, __DRI2_CONFIG_QUERY_VERSION },
+   /*.base =*/ { __DRI2_CONFIG_QUERY, 1 },
 
    /*.configQueryb        =*/ dri2ConfigQueryb,
    /*.configQueryi        =*/ dri2ConfigQueryi,
diff --git a/mesalib/src/mesa/drivers/dri/common/dri_util.h b/mesalib/src/mesa/drivers/dri/common/dri_util.h
index a37a0bbbc..1138bf106 100644
--- a/mesalib/src/mesa/drivers/dri/common/dri_util.h
+++ b/mesalib/src/mesa/drivers/dri/common/dri_util.h
@@ -174,13 +174,13 @@ struct __DRIscreenRec {
     struct {
 	/* Flag to indicate that this is a DRI2 screen.  Many of the above
 	 * fields will not be valid or initializaed in that case. */
-	__DRIdri2LoaderExtension *loader;
-	__DRIimageLookupExtension *image;
-	__DRIuseInvalidateExtension *useInvalidate;
+	const __DRIdri2LoaderExtension *loader;
+	const __DRIimageLookupExtension *image;
+	const __DRIuseInvalidateExtension *useInvalidate;
     } dri2;
 
     struct {
-        __DRIimageLoaderExtension *loader;
+        const __DRIimageLoaderExtension *loader;
     } image;
 
     driOptionCache optionInfo;
diff --git a/mesalib/src/mesa/drivers/dri/common/megadriver_stub.c b/mesalib/src/mesa/drivers/dri/common/megadriver_stub.c
index 4c3cc6776..f0c8878a8 100644
--- a/mesalib/src/mesa/drivers/dri/common/megadriver_stub.c
+++ b/mesalib/src/mesa/drivers/dri/common/megadriver_stub.c
@@ -28,10 +28,9 @@
 
 /* We need GNU extensions to dlfcn.h in order to provide backward
  * compatibility for the older DRI driver loader mechanism. (dladdr,
- * Dl_info, and RTLD_DEFAULT are only defined when _GNU_SOURCE is
- * defined.)
+ * Dl_info, and RTLD_DEFAULT)
  */
-#ifdef RTLD_DEFAULT
+#if defined(RTLD_DEFAULT) && defined(HAVE_DLADDR)
 
 #define MEGADRIVER_STUB_MAX_EXTENSIONS 10
 #define LIB_PATH_SUFFIX "_dri.so"
@@ -148,7 +147,7 @@ megadriver_stub_init(void)
    }
 }
 
-#endif /* RTLD_DEFAULT */
+#endif /* RTLD_DEFAULT && HAVE_DLADDR */
 
 static const
 __DRIconfig **stub_error_init_screen(__DRIscreen *psp)
diff --git a/mesalib/src/mesa/drivers/dri/swrast/swrast.c b/mesalib/src/mesa/drivers/dri/swrast/swrast.c
index 3d19a52dc..b9a5d71ef 100644
--- a/mesalib/src/mesa/drivers/dri/swrast/swrast.c
+++ b/mesalib/src/mesa/drivers/dri/swrast/swrast.c
@@ -115,9 +115,11 @@ static void swrastSetTexBuffer(__DRIcontext *pDRICtx, GLint target,
 }
 
 static const __DRItexBufferExtension swrastTexBufferExtension = {
-    { __DRI_TEX_BUFFER, __DRI_TEX_BUFFER_VERSION },
-    swrastSetTexBuffer,
-    swrastSetTexBuffer2,
+   .base = { __DRI_TEX_BUFFER, 3 },
+
+   .setTexBuffer        = swrastSetTexBuffer,
+   .setTexBuffer2       = swrastSetTexBuffer2,
+   .releaseTexBuffer    = NULL,
 };
 
 static const __DRIextension *dri_screen_extensions[] = {
diff --git a/mesalib/src/mesa/main/enable.c b/mesalib/src/mesa/main/enable.c
index edd4751e1..0f3bcf0a9 100644
--- a/mesalib/src/mesa/main/enable.c
+++ b/mesalib/src/mesa/main/enable.c
@@ -368,26 +368,11 @@ _mesa_set_enable(struct gl_context *ctx, GLenum cap, GLboolean state)
          ctx->Depth.Test = state;
          break;
       case GL_DEBUG_OUTPUT:
-         if (!_mesa_is_desktop_gl(ctx)) {
-            goto invalid_enum_error;
-         }
-         else {
-            struct gl_debug_state *debug = _mesa_get_debug_state(ctx);
-            if (debug) {
-               debug->DebugOutput = state;
-            }
-         }
-         break;
       case GL_DEBUG_OUTPUT_SYNCHRONOUS_ARB:
-         if (!_mesa_is_desktop_gl(ctx)) {
+         if (!_mesa_is_desktop_gl(ctx))
             goto invalid_enum_error;
-         }
-         else {
-            struct gl_debug_state *debug = _mesa_get_debug_state(ctx);
-            if (debug) {
-               debug->SyncOutput = state;
-            }
-         }
+         else
+            _mesa_set_debug_state_int(ctx, cap, state);
          break;
       case GL_DITHER:
          if (ctx->Color.DitherFlag == state)
@@ -1239,21 +1224,11 @@ _mesa_IsEnabled( GLenum cap )
       case GL_CULL_FACE:
          return ctx->Polygon.CullFlag;
       case GL_DEBUG_OUTPUT:
-         if (!_mesa_is_desktop_gl(ctx))
-            goto invalid_enum_error;
-         if (ctx->Debug) {
-            return ctx->Debug->DebugOutput;
-         } else {
-            return GL_FALSE;
-         }
       case GL_DEBUG_OUTPUT_SYNCHRONOUS_ARB:
          if (!_mesa_is_desktop_gl(ctx))
             goto invalid_enum_error;
-         if (ctx->Debug) {
-            return ctx->Debug->SyncOutput;
-         } else {
-            return GL_FALSE;
-         }
+         else
+            return (GLboolean) _mesa_get_debug_state_int(ctx, cap);
       case GL_DEPTH_TEST:
          return ctx->Depth.Test;
       case GL_DITHER:
diff --git a/mesalib/src/mesa/main/errors.c b/mesalib/src/mesa/main/errors.c
index d80fda0db..30a867298 100644
--- a/mesalib/src/mesa/main/errors.c
+++ b/mesalib/src/mesa/main/errors.c
@@ -41,10 +41,63 @@
 static mtx_t DynamicIDMutex = _MTX_INITIALIZER_NP;
 static GLuint NextDynamicID = 1;
 
-struct gl_debug_severity
+/**
+ * A namespace element.
+ */
+struct gl_debug_element
 {
    struct simple_node link;
+
    GLuint ID;
+   /* at which severity levels (mesa_debug_severity) is the message enabled */
+   GLbitfield State;
+};
+
+struct gl_debug_namespace
+{
+   struct simple_node Elements;
+   GLbitfield DefaultState;
+};
+
+struct gl_debug_group {
+   struct gl_debug_namespace Namespaces[MESA_DEBUG_SOURCE_COUNT][MESA_DEBUG_TYPE_COUNT];
+};
+
+/**
+ * An error, warning, or other piece of debug information for an application
+ * to consume via GL_ARB_debug_output/GL_KHR_debug.
+ */
+struct gl_debug_message
+{
+   enum mesa_debug_source source;
+   enum mesa_debug_type type;
+   GLuint id;
+   enum mesa_debug_severity severity;
+   GLsizei length;
+   GLcharARB *message;
+};
+
+/**
+ * Debug message log.  It works like a ring buffer.
+ */
+struct gl_debug_log {
+   struct gl_debug_message Messages[MAX_DEBUG_LOGGED_MESSAGES];
+   GLint NextMessage;
+   GLint NumMessages;
+};
+
+struct gl_debug_state
+{
+   GLDEBUGPROC Callback;
+   const void *CallbackData;
+   GLboolean SyncOutput;
+   GLboolean DebugOutput;
+
+   struct gl_debug_group *Groups[MAX_DEBUG_GROUP_STACK_DEPTH];
+   struct gl_debug_message GroupMessages[MAX_DEBUG_GROUP_STACK_DEPTH];
+   GLint GroupStackDepth;
+
+   struct gl_debug_log Log;
 };
 
 static char out_of_memory[] = "Debugging error: out of memory";
@@ -139,235 +192,399 @@ debug_get_id(GLuint *id)
    }
 }
 
+static void
+debug_message_clear(struct gl_debug_message *msg)
+{
+   if (msg->message != (char*)out_of_memory)
+      free(msg->message);
+   msg->message = NULL;
+   msg->length = 0;
+}
+
+static void
+debug_message_store(struct gl_debug_message *msg,
+                    enum mesa_debug_source source,
+                    enum mesa_debug_type type, GLuint id,
+                    enum mesa_debug_severity severity,
+                    GLsizei len, const char *buf)
+{
+   assert(!msg->message && !msg->length);
+
+   msg->message = malloc(len+1);
+   if (msg->message) {
+      (void) strncpy(msg->message, buf, (size_t)len);
+      msg->message[len] = '\0';
+
+      msg->length = len+1;
+      msg->source = source;
+      msg->type = type;
+      msg->id = id;
+      msg->severity = severity;
+   } else {
+      static GLuint oom_msg_id = 0;
+      debug_get_id(&oom_msg_id);
+
+      /* malloc failed! */
+      msg->message = out_of_memory;
+      msg->length = strlen(out_of_memory)+1;
+      msg->source = MESA_DEBUG_SOURCE_OTHER;
+      msg->type = MESA_DEBUG_TYPE_ERROR;
+      msg->id = oom_msg_id;
+      msg->severity = MESA_DEBUG_SEVERITY_HIGH;
+   }
+}
+
+static void
+debug_namespace_init(struct gl_debug_namespace *ns)
+{
+   make_empty_list(&ns->Elements);
 
-/*
- * We store a bitfield in the hash table, with five possible values total.
- *
- * The ENABLED_BIT's purpose is self-explanatory.
- *
- * The FOUND_BIT is needed to differentiate the value of DISABLED from
- * the value returned by HashTableLookup() when it can't find the given key.
- *
- * The KNOWN_SEVERITY bit is a bit complicated:
- *
- * A client may call Control() with an array of IDs, then call Control()
- * on all message IDs of a certain severity, then Insert() one of the
- * previously specified IDs, giving us a known severity level, then call
- * Control() on all message IDs of a certain severity level again.
- *
- * After the first call, those IDs will have a FOUND_BIT, but will not
- * exist in any severity-specific list, so the second call will not
- * impact them. This is undesirable but unavoidable given the API:
- * The only entrypoint that gives a severity for a client-defined ID
- * is the Insert() call.
- *
- * For the sake of Control(), we want to maintain the invariant
- * that an ID will either appear in none of the three severity lists,
- * or appear once, to minimize pointless duplication and potential surprises.
- *
- * Because Insert() is the only place that will learn an ID's severity,
- * it should insert an ID into the appropriate list, but only if the ID
- * doesn't exist in it or any other list yet. Because searching all three
- * lists at O(n) is needlessly expensive, we store KNOWN_SEVERITY.
- */
-enum {
-   FOUND_BIT = 1 << 0,
-   ENABLED_BIT = 1 << 1,
-   KNOWN_SEVERITY = 1 << 2,
-
-   /* HashTable reserves zero as a return value meaning 'not found' */
-   NOT_FOUND = 0,
-   DISABLED = FOUND_BIT,
-   ENABLED = ENABLED_BIT | FOUND_BIT
-};
+   /* Enable all the messages with severity HIGH or MEDIUM by default */
+   ns->DefaultState = (1 << MESA_DEBUG_SEVERITY_HIGH) |
+                      (1 << MESA_DEBUG_SEVERITY_MEDIUM);
+}
+
+static void
+debug_namespace_clear(struct gl_debug_namespace *ns)
+{
+   struct simple_node *node, *tmp;
+
+   foreach_s(node, tmp, &ns->Elements)
+      free(node);
+}
+
+static bool
+debug_namespace_copy(struct gl_debug_namespace *dst,
+                     const struct gl_debug_namespace *src)
+{
+   struct simple_node *node;
+
+   dst->DefaultState = src->DefaultState;
 
+   make_empty_list(&dst->Elements);
+   foreach(node, &src->Elements) {
+      const struct gl_debug_element *elem =
+         (const struct gl_debug_element *) node;
+      struct gl_debug_element *copy;
+
+      copy = malloc(sizeof(*copy));
+      if (!copy) {
+         debug_namespace_clear(dst);
+         return false;
+      }
+
+      copy->ID = elem->ID;
+      copy->State = elem->State;
+      insert_at_tail(&dst->Elements, &copy->link);
+   }
+
+   return true;
+}
 
 /**
- * Return debug state for the context.  The debug state will be allocated
- * and initialized upon the first call.
+ * Set the state of \p id in the namespace.
  */
-struct gl_debug_state *
-_mesa_get_debug_state(struct gl_context *ctx)
+static bool
+debug_namespace_set(struct gl_debug_namespace *ns,
+                    GLuint id, bool enabled)
 {
-   if (!ctx->Debug) {
-      ctx->Debug = CALLOC_STRUCT(gl_debug_state);
-      if (!ctx->Debug) {
-         _mesa_error(ctx, GL_OUT_OF_MEMORY, "allocating debug state");
+   const uint32_t state = (enabled) ?
+      ((1 << MESA_DEBUG_SEVERITY_COUNT) - 1) : 0;
+   struct gl_debug_element *elem = NULL;
+   struct simple_node *node;
+
+   /* find the element */
+   foreach(node, &ns->Elements) {
+      struct gl_debug_element *tmp = (struct gl_debug_element *) node;
+      if (tmp->ID == id) {
+         elem = tmp;
+         break;
       }
-      else {
-         struct gl_debug_state *debug = ctx->Debug;
-         int s, t, sev;
-
-         /* Enable all the messages with severity HIGH or MEDIUM by default. */
-         memset(debug->Defaults[0][MESA_DEBUG_SEVERITY_HIGH], GL_TRUE,
-                sizeof debug->Defaults[0][MESA_DEBUG_SEVERITY_HIGH]);
-         memset(debug->Defaults[0][MESA_DEBUG_SEVERITY_MEDIUM], GL_TRUE,
-                sizeof debug->Defaults[0][MESA_DEBUG_SEVERITY_MEDIUM]);
-         memset(debug->Defaults[0][MESA_DEBUG_SEVERITY_LOW], GL_FALSE,
-                sizeof debug->Defaults[0][MESA_DEBUG_SEVERITY_LOW]);
-
-         /* Initialize state for filtering known debug messages. */
-         for (s = 0; s < MESA_DEBUG_SOURCE_COUNT; s++) {
-            for (t = 0; t < MESA_DEBUG_TYPE_COUNT; t++) {
-               debug->Namespaces[0][s][t].IDs = _mesa_NewHashTable();
-               assert(debug->Namespaces[0][s][t].IDs);
-
-               for (sev = 0; sev < MESA_DEBUG_SEVERITY_COUNT; sev++) {
-                  make_empty_list(&debug->Namespaces[0][s][t].Severity[sev]);
-               }
-            }
-         }
+   }
+
+   /* we do not need the element if it has the default state */
+   if (ns->DefaultState == state) {
+      if (elem) {
+         remove_from_list(&elem->link);
+         free(elem);
       }
+      return true;
    }
 
-   return ctx->Debug;
+   if (!elem) {
+      elem = malloc(sizeof(*elem));
+      if (!elem)
+         return false;
+
+      elem->ID = id;
+      insert_at_tail(&ns->Elements, &elem->link);
+   }
+
+   elem->State = state;
+
+   return true;
 }
 
+/**
+ * Set the default state of the namespace for \p severity.  When \p severity
+ * is MESA_DEBUG_SEVERITY_COUNT, the default values for all severities are
+ * updated.
+ */
+static void
+debug_namespace_set_all(struct gl_debug_namespace *ns,
+                        enum mesa_debug_severity severity,
+                        bool enabled)
+{
+   struct simple_node *node, *tmp;
+   uint32_t mask, val;
+
+   /* set all elements to the same state */
+   if (severity == MESA_DEBUG_SEVERITY_COUNT) {
+      ns->DefaultState = (enabled) ? ((1 << severity) - 1) : 0;
+      debug_namespace_clear(ns);
+      make_empty_list(&ns->Elements);
+      return;
+   }
+
+   mask = 1 << severity;
+   val = (enabled) ? mask : 0;
 
+   ns->DefaultState = (ns->DefaultState & ~mask) | val;
+
+   foreach_s(node, tmp, &ns->Elements) {
+      struct gl_debug_element *elem = (struct gl_debug_element *) node;
+
+      elem->State = (elem->State & ~mask) | val;
+      if (elem->State == ns->DefaultState) {
+         remove_from_list(node);
+         free(node);
+      }
+   }
+}
 
 /**
- * Returns the state of the given message source/type/ID tuple.
+ * Get the state of \p id in the namespace.
  */
-static GLboolean
-should_log(struct gl_context *ctx,
-           enum mesa_debug_source source,
-           enum mesa_debug_type type,
-           GLuint id,
-           enum mesa_debug_severity severity)
+static bool
+debug_namespace_get(const struct gl_debug_namespace *ns, GLuint id,
+                    enum mesa_debug_severity severity)
 {
-   struct gl_debug_state *debug;
-   uintptr_t state = 0;
+   struct simple_node *node;
+   uint32_t state;
 
-   if (!ctx->Debug) {
-      /* no debug state set so far */
-      return GL_FALSE;
+   state = ns->DefaultState;
+   foreach(node, &ns->Elements) {
+      struct gl_debug_element *elem = (struct gl_debug_element *) node;
+
+      if (elem->ID == id) {
+         state = elem->State;
+         break;
+      }
    }
 
-   debug = _mesa_get_debug_state(ctx);
-   if (debug) {
-      const GLint gstack = debug->GroupStackDepth;
-      struct gl_debug_namespace *nspace =
-         &debug->Namespaces[gstack][source][type];
+   return (state & (1 << severity));
+}
 
-      if (!debug->DebugOutput)
-         return GL_FALSE;
+/**
+ * Allocate and initialize context debug state.
+ */
+static struct gl_debug_state *
+debug_create(void)
+{
+   struct gl_debug_state *debug;
+   int s, t;
 
-      /* In addition to not being able to store zero as a value, HashTable also
-       * can't use zero as a key.
-       */
-      if (id)
-         state = (uintptr_t)_mesa_HashLookup(nspace->IDs, id);
-      else
-         state = nspace->ZeroID;
+   debug = CALLOC_STRUCT(gl_debug_state);
+   if (!debug)
+      return NULL;
 
-      /* Only do this once for each ID. This makes sure the ID exists in,
-       * at most, one list, and does not pointlessly appear multiple times.
-       */
-      if (!(state & KNOWN_SEVERITY)) {
-         struct gl_debug_severity *entry;
-
-         if (state == NOT_FOUND) {
-            if (debug->Defaults[gstack][severity][source][type])
-               state = ENABLED;
-            else
-               state = DISABLED;
-         }
+   debug->Groups[0] = malloc(sizeof(*debug->Groups[0]));
+   if (!debug->Groups[0]) {
+      free(debug);
+      return NULL;
+   }
+
+   /* Initialize state for filtering known debug messages. */
+   for (s = 0; s < MESA_DEBUG_SOURCE_COUNT; s++) {
+      for (t = 0; t < MESA_DEBUG_TYPE_COUNT; t++)
+         debug_namespace_init(&debug->Groups[0]->Namespaces[s][t]);
+   }
+
+   return debug;
+}
 
-         entry = malloc(sizeof *entry);
-         if (!entry)
-            goto out;
+/**
+ * Return true if the top debug group points to the group below it.
+ */
+static bool
+debug_is_group_read_only(const struct gl_debug_state *debug)
+{
+   const GLint gstack = debug->GroupStackDepth;
+   return (gstack > 0 && debug->Groups[gstack] == debug->Groups[gstack - 1]);
+}
+
+/**
+ * Make the top debug group writable.
+ */
+static bool
+debug_make_group_writable(struct gl_debug_state *debug)
+{
+   const GLint gstack = debug->GroupStackDepth;
+   const struct gl_debug_group *src = debug->Groups[gstack];
+   struct gl_debug_group *dst;
+   int s, t;
 
-         state |= KNOWN_SEVERITY;
+   if (!debug_is_group_read_only(debug))
+      return true;
 
-         if (id)
-            _mesa_HashInsert(nspace->IDs, id, (void*)state);
-         else
-            nspace->ZeroID = state;
+   dst = malloc(sizeof(*dst));
+   if (!dst)
+      return false;
 
-         entry->ID = id;
-         insert_at_tail(&nspace->Severity[severity], &entry->link);
+   for (s = 0; s < MESA_DEBUG_SOURCE_COUNT; s++) {
+      for (t = 0; t < MESA_DEBUG_TYPE_COUNT; t++) {
+         if (!debug_namespace_copy(&dst->Namespaces[s][t],
+                                   &src->Namespaces[s][t])) {
+            /* error path! */
+            for (t = t - 1; t >= 0; t--)
+               debug_namespace_clear(&dst->Namespaces[s][t]);
+            for (s = s - 1; s >= 0; s--) {
+               for (t = 0; t < MESA_DEBUG_TYPE_COUNT; t++)
+                  debug_namespace_clear(&dst->Namespaces[s][t]);
+            }
+            free(dst);
+            return false;
+         }
       }
    }
-out:
-   return !!(state & ENABLED_BIT);
-}
 
+   debug->Groups[gstack] = dst;
+
+   return true;
+}
 
 /**
- * Sets the state of the given message source/type/ID tuple.
+ * Free the top debug group.
  */
 static void
-set_message_state(struct gl_context *ctx,
-                  enum mesa_debug_source source,
-                  enum mesa_debug_type type,
-                  GLuint id, GLboolean enabled)
+debug_clear_group(struct gl_debug_state *debug)
 {
-   struct gl_debug_state *debug = _mesa_get_debug_state(ctx);
+   const GLint gstack = debug->GroupStackDepth;
 
-   if (debug) {
-      GLint gstack = debug->GroupStackDepth;
-      struct gl_debug_namespace *nspace =
-         &debug->Namespaces[gstack][source][type];
-      uintptr_t state;
+   if (!debug_is_group_read_only(debug)) {
+      struct gl_debug_group *grp = debug->Groups[gstack];
+      int s, t;
 
-      /* In addition to not being able to store zero as a value, HashTable also
-       * can't use zero as a key.
-       */
-      if (id)
-         state = (uintptr_t)_mesa_HashLookup(nspace->IDs, id);
-      else
-         state = nspace->ZeroID;
-
-      if (state == NOT_FOUND)
-         state = enabled ? ENABLED : DISABLED;
-      else {
-         if (enabled)
-            state |= ENABLED_BIT;
-         else
-            state &= ~ENABLED_BIT;
+      for (s = 0; s < MESA_DEBUG_SOURCE_COUNT; s++) {
+         for (t = 0; t < MESA_DEBUG_TYPE_COUNT; t++)
+            debug_namespace_clear(&grp->Namespaces[s][t]);
       }
 
-      if (id)
-         _mesa_HashInsert(nspace->IDs, id, (void*)state);
-      else
-         nspace->ZeroID = state;
+      free(grp);
    }
+
+   debug->Groups[gstack] = NULL;
 }
 
+/**
+ * Loop through debug group stack tearing down states for
+ * filtering debug messages.  Then free debug output state.
+ */
+static void
+debug_destroy(struct gl_debug_state *debug)
+{
+   while (debug->GroupStackDepth > 0) {
+      debug_clear_group(debug);
+      debug->GroupStackDepth--;
+   }
 
+   debug_clear_group(debug);
+   free(debug);
+}
+
+/**
+ * Sets the state of the given message source/type/ID tuple.
+ */
 static void
-store_message_details(struct gl_debug_msg *emptySlot,
-                      enum mesa_debug_source source,
-                      enum mesa_debug_type type, GLuint id,
-                      enum mesa_debug_severity severity, GLint len,
-                      const char *buf)
+debug_set_message_enable(struct gl_debug_state *debug,
+                         enum mesa_debug_source source,
+                         enum mesa_debug_type type,
+                         GLuint id, GLboolean enabled)
 {
-   assert(!emptySlot->message && !emptySlot->length);
-
-   emptySlot->message = malloc(len+1);
-   if (emptySlot->message) {
-      (void) strncpy(emptySlot->message, buf, (size_t)len);
-      emptySlot->message[len] = '\0';
-
-      emptySlot->length = len+1;
-      emptySlot->source = source;
-      emptySlot->type = type;
-      emptySlot->id = id;
-      emptySlot->severity = severity;
+   const GLint gstack = debug->GroupStackDepth;
+   struct gl_debug_namespace *ns;
+
+   debug_make_group_writable(debug);
+   ns = &debug->Groups[gstack]->Namespaces[source][type];
+
+   debug_namespace_set(ns, id, enabled);
+}
+
+/*
+ * Set the state of all message IDs found in the given intersection of
+ * 'source', 'type', and 'severity'.  The _COUNT enum can be used for
+ * GL_DONT_CARE (include all messages in the class).
+ *
+ * This requires both setting the state of all previously seen message
+ * IDs in the hash table, and setting the default state for all
+ * applicable combinations of source/type/severity, so that all the
+ * yet-unknown message IDs that may be used in the future will be
+ * impacted as if they were already known.
+ */
+static void
+debug_set_message_enable_all(struct gl_debug_state *debug,
+                             enum mesa_debug_source source,
+                             enum mesa_debug_type type,
+                             enum mesa_debug_severity severity,
+                             GLboolean enabled)
+{
+   const GLint gstack = debug->GroupStackDepth;
+   int s, t, smax, tmax;
+
+   if (source == MESA_DEBUG_SOURCE_COUNT) {
+      source = 0;
+      smax = MESA_DEBUG_SOURCE_COUNT;
    } else {
-      static GLuint oom_msg_id = 0;
-      debug_get_id(&oom_msg_id);
+      smax = source+1;
+   }
 
-      /* malloc failed! */
-      emptySlot->message = out_of_memory;
-      emptySlot->length = strlen(out_of_memory)+1;
-      emptySlot->source = MESA_DEBUG_SOURCE_OTHER;
-      emptySlot->type = MESA_DEBUG_TYPE_ERROR;
-      emptySlot->id = oom_msg_id;
-      emptySlot->severity = MESA_DEBUG_SEVERITY_HIGH;
+   if (type == MESA_DEBUG_TYPE_COUNT) {
+      type = 0;
+      tmax = MESA_DEBUG_TYPE_COUNT;
+   } else {
+      tmax = type+1;
+   }
+
+   debug_make_group_writable(debug);
+
+   for (s = source; s < smax; s++) {
+      for (t = type; t < tmax; t++) {
+         struct gl_debug_namespace *nspace =
+            &debug->Groups[gstack]->Namespaces[s][t];
+         debug_namespace_set_all(nspace, severity, enabled);
+      }
    }
 }
 
+/**
+ * Returns if the given message source/type/ID tuple is enabled.
+ */
+static bool
+debug_is_message_enabled(const struct gl_debug_state *debug,
+                         enum mesa_debug_source source,
+                         enum mesa_debug_type type,
+                         GLuint id,
+                         enum mesa_debug_severity severity)
+{
+   const GLint gstack = debug->GroupStackDepth;
+   struct gl_debug_group *grp = debug->Groups[gstack];
+   struct gl_debug_namespace *nspace = &grp->Namespaces[source][type];
+
+   if (!debug->DebugOutput)
+      return false;
+
+   return debug_namespace_get(nspace, id, severity);
+}
 
 /**
  * 'buf' is not necessarily a null-terminated string. When logging, copy
@@ -376,108 +593,229 @@ store_message_details(struct gl_debug_msg *emptySlot,
  * the null terminator this time.
  */
 static void
-log_msg(struct gl_context *ctx, enum mesa_debug_source source,
-        enum mesa_debug_type type, GLuint id,
-        enum mesa_debug_severity severity, GLint len, const char *buf)
+debug_log_message(struct gl_debug_state *debug,
+                  enum mesa_debug_source source,
+                  enum mesa_debug_type type, GLuint id,
+                  enum mesa_debug_severity severity,
+                  GLsizei len, const char *buf)
 {
-   struct gl_debug_state *debug = _mesa_get_debug_state(ctx);
+   struct gl_debug_log *log = &debug->Log;
    GLint nextEmpty;
-   struct gl_debug_msg *emptySlot;
-
-   if (!debug)
-      return;
+   struct gl_debug_message *emptySlot;
 
    assert(len >= 0 && len < MAX_DEBUG_MESSAGE_LENGTH);
 
-   if (!should_log(ctx, source, type, id, severity))
+   if (log->NumMessages == MAX_DEBUG_LOGGED_MESSAGES)
       return;
 
-   if (debug->Callback) {
-       GLenum gl_type = debug_type_enums[type];
-       GLenum gl_severity = debug_severity_enums[severity];
+   nextEmpty = (log->NextMessage + log->NumMessages)
+      % MAX_DEBUG_LOGGED_MESSAGES;
+   emptySlot = &log->Messages[nextEmpty];
 
-      debug->Callback(debug_source_enums[source], gl_type, id, gl_severity,
-                      len, buf, debug->CallbackData);
-      return;
-   }
+   debug_message_store(emptySlot, source, type,
+                       id, severity, len, buf);
 
-   if (debug->NumMessages == MAX_DEBUG_LOGGED_MESSAGES)
-      return;
+   log->NumMessages++;
+}
+
+/**
+ * Return the oldest debug message out of the log.
+ */
+static const struct gl_debug_message *
+debug_fetch_message(const struct gl_debug_state *debug)
+{
+   const struct gl_debug_log *log = &debug->Log;
+
+   return (log->NumMessages) ? &log->Messages[log->NextMessage] : NULL;
+}
+
+/**
+ * Delete the oldest debug messages out of the log.
+ */
+static void
+debug_delete_messages(struct gl_debug_state *debug, unsigned count)
+{
+   struct gl_debug_log *log = &debug->Log;
 
-   nextEmpty = (debug->NextMsg + debug->NumMessages)
-                          % MAX_DEBUG_LOGGED_MESSAGES;
-   emptySlot = &debug->Log[nextEmpty];
+   if (count > log->NumMessages)
+      count = log->NumMessages;
 
-   store_message_details(emptySlot, source, type, id, severity, len, buf);
+   while (count--) {
+      struct gl_debug_message *msg = &log->Messages[log->NextMessage];
 
-   if (debug->NumMessages == 0)
-      debug->NextMsgLength = debug->Log[debug->NextMsg].length;
+      debug_message_clear(msg);
 
-   debug->NumMessages++;
+      log->NumMessages--;
+      log->NextMessage++;
+      log->NextMessage %= MAX_DEBUG_LOGGED_MESSAGES;
+   }
+}
+
+static struct gl_debug_message *
+debug_get_group_message(struct gl_debug_state *debug)
+{
+   return &debug->GroupMessages[debug->GroupStackDepth];
+}
+
+static void
+debug_push_group(struct gl_debug_state *debug)
+{
+   const GLint gstack = debug->GroupStackDepth;
+
+   /* just point to the previous stack */
+   debug->Groups[gstack + 1] = debug->Groups[gstack];
+   debug->GroupStackDepth++;
+}
+
+static void
+debug_pop_group(struct gl_debug_state *debug)
+{
+   debug_clear_group(debug);
+   debug->GroupStackDepth--;
 }
 
 
 /**
- * Pop the oldest debug message out of the log.
- * Writes the message string, including the null terminator, into 'buf',
- * using up to 'bufSize' bytes. If 'bufSize' is too small, or
- * if 'buf' is NULL, nothing is written.
- *
- * Returns the number of bytes written on success, or when 'buf' is NULL,
- * the number that would have been written. A return value of 0
- * indicates failure.
+ * Return debug state for the context.  The debug state will be allocated
+ * and initialized upon the first call.
  */
-static GLsizei
-get_msg(struct gl_context *ctx, GLenum *source, GLenum *type,
-        GLuint *id, GLenum *severity, GLsizei bufSize, char *buf)
+static struct gl_debug_state *
+_mesa_get_debug_state(struct gl_context *ctx)
+{
+   if (!ctx->Debug) {
+      ctx->Debug = debug_create();
+      if (!ctx->Debug) {
+         _mesa_error(ctx, GL_OUT_OF_MEMORY, "allocating debug state");
+      }
+   }
+
+   return ctx->Debug;
+}
+
+/**
+ * Set the integer debug state specified by \p pname.  This can be called from
+ * _mesa_set_enable for example.
+ */
+bool
+_mesa_set_debug_state_int(struct gl_context *ctx, GLenum pname, GLint val)
 {
    struct gl_debug_state *debug = _mesa_get_debug_state(ctx);
-   struct gl_debug_msg *msg;
-   GLsizei length;
 
-   if (!debug || debug->NumMessages == 0)
-      return 0;
+   if (!debug)
+      return false;
 
-   msg = &debug->Log[debug->NextMsg];
-   length = msg->length;
+   switch (pname) {
+   case GL_DEBUG_OUTPUT:
+      debug->DebugOutput = (val != 0);
+      break;
+   case GL_DEBUG_OUTPUT_SYNCHRONOUS_ARB:
+      debug->SyncOutput = (val != 0);
+      break;
+   default:
+      assert(!"unknown debug output param");
+      break;
+   }
 
-   assert(length > 0 && length == debug->NextMsgLength);
+   return true;
+}
 
-   if (bufSize < length && buf != NULL)
+/**
+ * Query the integer debug state specified by \p pname.  This can be called
+ * _mesa_GetIntegerv for example.
+ */
+GLint
+_mesa_get_debug_state_int(struct gl_context *ctx, GLenum pname)
+{
+   struct gl_debug_state *debug;
+   GLint val;
+
+   debug = ctx->Debug;
+   if (!debug)
       return 0;
 
-   if (severity) {
-      *severity = debug_severity_enums[msg->severity];
+   switch (pname) {
+   case GL_DEBUG_OUTPUT:
+      val = debug->DebugOutput;
+      break;
+   case GL_DEBUG_OUTPUT_SYNCHRONOUS_ARB:
+      val = debug->SyncOutput;
+      break;
+   case GL_DEBUG_LOGGED_MESSAGES:
+      val = debug->Log.NumMessages;
+      break;
+   case GL_DEBUG_NEXT_LOGGED_MESSAGE_LENGTH:
+      val = (debug->Log.NumMessages) ?
+         debug->Log.Messages[debug->Log.NextMessage].length : 0;
+      break;
+   case GL_DEBUG_GROUP_STACK_DEPTH:
+      val = debug->GroupStackDepth;
+      break;
+   default:
+      assert(!"unknown debug output param");
+      val = 0;
+      break;
    }
 
-   if (source) {
-      *source = debug_source_enums[msg->source];
-   }
+   return val;
+}
 
-   if (type) {
-      *type = debug_type_enums[msg->type];
-   }
+/**
+ * Query the pointer debug state specified by \p pname.  This can be called
+ * _mesa_GetPointerv for example.
+ */
+void *
+_mesa_get_debug_state_ptr(struct gl_context *ctx, GLenum pname)
+{
+   struct gl_debug_state *debug;
+   void *val;
 
-   if (id) {
-      *id = msg->id;
-   }
+   debug = ctx->Debug;
+   if (!debug)
+      return NULL;
 
-   if (buf) {
-      assert(msg->message[length-1] == '\0');
-      (void) strncpy(buf, msg->message, (size_t)length);
+   switch (pname) {
+   case GL_DEBUG_CALLBACK_FUNCTION_ARB:
+      val = (void *) debug->Callback;
+      break;
+   case GL_DEBUG_CALLBACK_USER_PARAM_ARB:
+      val = (void *) debug->CallbackData;
+      break;
+   default:
+      assert(!"unknown debug output param");
+      val = NULL;
+      break;
    }
 
-   if (msg->message != (char*)out_of_memory)
-      free(msg->message);
-   msg->message = NULL;
-   msg->length = 0;
+   return val;
+}
 
-   debug->NumMessages--;
-   debug->NextMsg++;
-   debug->NextMsg %= MAX_DEBUG_LOGGED_MESSAGES;
-   debug->NextMsgLength = debug->Log[debug->NextMsg].length;
 
-   return length;
+/**
+ * Log a client or driver debug message.
+ */
+static void
+log_msg(struct gl_context *ctx, enum mesa_debug_source source,
+        enum mesa_debug_type type, GLuint id,
+        enum mesa_debug_severity severity, GLint len, const char *buf)
+{
+   struct gl_debug_state *debug = _mesa_get_debug_state(ctx);
+
+   if (!debug)
+      return;
+
+   if (!debug_is_message_enabled(debug, source, type, id, severity))
+      return;
+
+   if (debug->Callback) {
+       GLenum gl_type = debug_type_enums[type];
+       GLenum gl_severity = debug_severity_enums[severity];
+
+      debug->Callback(debug_source_enums[source], gl_type, id, gl_severity,
+                      len, buf, debug->CallbackData);
+      return;
+   }
+
+   debug_log_message(debug, source, type, id, severity, len, buf);
 }
 
 
@@ -560,172 +898,18 @@ error:
 }
 
 
-/**
- * Set the state of all message IDs found in the given intersection of
- * 'source', 'type', and 'severity'.  The _COUNT enum can be used for
- * GL_DONT_CARE (include all messages in the class).
- *
- * This requires both setting the state of all previously seen message
- * IDs in the hash table, and setting the default state for all
- * applicable combinations of source/type/severity, so that all the
- * yet-unknown message IDs that may be used in the future will be
- * impacted as if they were already known.
- */
-static void
-control_messages(struct gl_context *ctx,
-                 enum mesa_debug_source source,
-                 enum mesa_debug_type type,
-                 enum mesa_debug_severity severity,
-                 GLboolean enabled)
-{
-   struct gl_debug_state *debug = _mesa_get_debug_state(ctx);
-   int s, t, sev, smax, tmax, sevmax;
-   const GLint gstack = debug ? debug->GroupStackDepth : 0;
-
-   if (!debug)
-      return;
-
-   if (source == MESA_DEBUG_SOURCE_COUNT) {
-      source = 0;
-      smax = MESA_DEBUG_SOURCE_COUNT;
-   } else {
-      smax = source+1;
-   }
-
-   if (type == MESA_DEBUG_TYPE_COUNT) {
-      type = 0;
-      tmax = MESA_DEBUG_TYPE_COUNT;
-   } else {
-      tmax = type+1;
-   }
-
-   if (severity == MESA_DEBUG_SEVERITY_COUNT) {
-      severity = 0;
-      sevmax = MESA_DEBUG_SEVERITY_COUNT;
-   } else {
-      sevmax = severity+1;
-   }
-
-   for (sev = severity; sev < sevmax; sev++) {
-      for (s = source; s < smax; s++) {
-         for (t = type; t < tmax; t++) {
-            struct simple_node *node;
-            struct gl_debug_severity *entry;
-
-            /* change the default for IDs we've never seen before. */
-            debug->Defaults[gstack][sev][s][t] = enabled;
-
-            /* Now change the state of IDs we *have* seen... */
-            foreach(node, &debug->Namespaces[gstack][s][t].Severity[sev]) {
-               entry = (struct gl_debug_severity *)node;
-               set_message_state(ctx, s, t, entry->ID, enabled);
-            }
-         }
-      }
-   }
-}
-
-
-/**
- * Debugging-message namespaces with the source APPLICATION or THIRD_PARTY
- * require special handling, since the IDs in them are controlled by clients,
- * not the OpenGL implementation.
- *
- * 'count' is the length of the array 'ids'. If 'count' is nonzero, all
- * the given IDs in the namespace defined by 'esource' and 'etype'
- * will be affected.
- *
- * If 'count' is zero, this sets the state of all IDs that match
- * the combination of 'esource', 'etype', and 'eseverity'.
- */
-static void
-control_app_messages(struct gl_context *ctx, GLenum esource, GLenum etype,
-                     GLenum eseverity, GLsizei count, const GLuint *ids,
-                     GLboolean enabled)
-{
-   GLsizei i;
-   enum mesa_debug_source source = gl_enum_to_debug_source(esource);
-   enum mesa_debug_type type = gl_enum_to_debug_type(etype);
-   enum mesa_debug_severity severity = gl_enum_to_debug_severity(eseverity);
-
-   for (i = 0; i < count; i++)
-      set_message_state(ctx, source, type, ids[i], enabled);
-
-   if (count)
-      return;
-
-   control_messages(ctx, source, type, severity, enabled);
-}
-
-
-/**
- * This is a generic message insert function.
- * Validation of source, type and severity parameters should be done
- * before calling this funtion.
- */
-static void
-message_insert(GLenum source, GLenum type, GLuint id,
-               GLenum severity, GLint length, const GLchar *buf,
-               const char *callerstr)
+static GLboolean
+validate_length(struct gl_context *ctx, const char *callerstr, GLsizei length)
 {
-   GET_CURRENT_CONTEXT(ctx);
-
-   if (length < 0)
-      length = strlen(buf);
-
    if (length >= MAX_DEBUG_MESSAGE_LENGTH) {
       _mesa_error(ctx, GL_INVALID_VALUE,
                  "%s(length=%d, which is not less than "
                  "GL_MAX_DEBUG_MESSAGE_LENGTH=%d)", callerstr, length,
                  MAX_DEBUG_MESSAGE_LENGTH);
-      return;
+      return GL_FALSE;
    }
 
-   log_msg(ctx,
-           gl_enum_to_debug_source(source),
-           gl_enum_to_debug_type(type), id,
-           gl_enum_to_debug_severity(severity), length, buf);
-}
-
-
-static void
-do_nothing(GLuint key, void *data, void *userData)
-{
-}
-
-
-/**
- * Free context state pertaining to error/debug state for the given stack
- * depth.
- */
-static void
-free_errors_data(struct gl_context *ctx, GLint gstack)
-{
-   struct gl_debug_state *debug = ctx->Debug;
-   enum mesa_debug_type t;
-   enum mesa_debug_source s;
-   enum mesa_debug_severity sev;
-
-   assert(debug);
-
-   /* Tear down state for filtering debug messages. */
-   for (s = 0; s < MESA_DEBUG_SOURCE_COUNT; s++) {
-      for (t = 0; t < MESA_DEBUG_TYPE_COUNT; t++) {
-         _mesa_HashDeleteAll(debug->Namespaces[gstack][s][t].IDs,
-                             do_nothing, NULL);
-         _mesa_DeleteHashTable(debug->Namespaces[gstack][s][t].IDs);
-         for (sev = 0; sev < MESA_DEBUG_SEVERITY_COUNT; sev++) {
-            struct simple_node *node, *tmp;
-            struct gl_debug_severity *entry;
-
-            foreach_s(node, tmp,
-                      &debug->Namespaces[gstack][s][t].Severity[sev]) {
-               entry = (struct gl_debug_severity *)node;
-               free(entry);
-            }
-         }
-      }
-   }
+   return GL_TRUE;
 }
 
 
@@ -741,7 +925,15 @@ _mesa_DebugMessageInsert(GLenum source, GLenum type, GLuint id,
    if (!validate_params(ctx, INSERT, callerstr, source, type, severity))
       return; /* GL_INVALID_ENUM */
 
-   message_insert(source, type, id, severity, length, buf, callerstr);
+   if (length < 0)
+      length = strlen(buf);
+   if (!validate_length(ctx, callerstr, length))
+      return; /* GL_INVALID_VALUE */
+
+   log_msg(ctx, gl_enum_to_debug_source(source),
+           gl_enum_to_debug_type(type), id,
+           gl_enum_to_debug_severity(severity),
+           length, buf);
 }
 
 
@@ -751,6 +943,7 @@ _mesa_GetDebugMessageLog(GLuint count, GLsizei logSize, GLenum *sources,
                          GLsizei *lengths, GLchar *messageLog)
 {
    GET_CURRENT_CONTEXT(ctx);
+   struct gl_debug_state *debug;
    GLuint ret;
 
    if (!messageLog)
@@ -763,29 +956,39 @@ _mesa_GetDebugMessageLog(GLuint count, GLsizei logSize, GLenum *sources,
       return 0;
    }
 
+   debug = _mesa_get_debug_state(ctx);
+   if (!debug)
+      return 0;
+
    for (ret = 0; ret < count; ret++) {
-      GLsizei written = get_msg(ctx, sources, types, ids, severities,
-                                logSize, messageLog);
-      if (!written)
+      const struct gl_debug_message *msg = debug_fetch_message(debug);
+
+      if (!msg)
+         break;
+
+      if (logSize < msg->length && messageLog != NULL)
          break;
 
       if (messageLog) {
-         messageLog += written;
-         logSize -= written;
-      }
-      if (lengths) {
-         *lengths = written;
-         lengths++;
+         assert(msg->message[msg->length-1] == '\0');
+         (void) strncpy(messageLog, msg->message, (size_t)msg->length);
+
+         messageLog += msg->length;
+         logSize -= msg->length;
       }
 
+      if (lengths)
+         *lengths++ = msg->length;
       if (severities)
-         severities++;
+         *severities++ = debug_severity_enums[msg->severity];
       if (sources)
-         sources++;
+         *sources++ = debug_source_enums[msg->source];
       if (types)
-         types++;
+         *types++ = debug_type_enums[msg->type];
       if (ids)
-         ids++;
+         *ids++ = msg->id;
+
+      debug_delete_messages(debug, 1);
    }
 
    return ret;
@@ -797,9 +1000,12 @@ _mesa_DebugMessageControl(GLenum gl_source, GLenum gl_type,
                           GLenum gl_severity, GLsizei count,
                           const GLuint *ids, GLboolean enabled)
 {
-   const char *callerstr = "glDebugMessageControl";
-
    GET_CURRENT_CONTEXT(ctx);
+   enum mesa_debug_source source = gl_enum_to_debug_source(gl_source);
+   enum mesa_debug_type type = gl_enum_to_debug_type(gl_type);
+   enum mesa_debug_severity severity = gl_enum_to_debug_severity(gl_severity);
+   const char *callerstr = "glDebugMessageControl";
+   struct gl_debug_state *debug;
 
    if (count < 0) {
       _mesa_error(ctx, GL_INVALID_VALUE,
@@ -821,8 +1027,18 @@ _mesa_DebugMessageControl(GLenum gl_source, GLenum gl_type,
       return;
    }
 
-   control_app_messages(ctx, gl_source, gl_type, gl_severity,
-                        count, ids, enabled);
+   debug = _mesa_get_debug_state(ctx);
+   if (!debug)
+      return;
+
+   if (count) {
+      GLsizei i;
+      for (i = 0; i < count; i++)
+         debug_set_message_enable(debug, source, type, ids[i], enabled);
+   }
+   else {
+      debug_set_message_enable_all(debug, source, type, severity, enabled);
+   }
 }
 
 
@@ -845,10 +1061,7 @@ _mesa_PushDebugGroup(GLenum source, GLuint id, GLsizei length,
    GET_CURRENT_CONTEXT(ctx);
    struct gl_debug_state *debug = _mesa_get_debug_state(ctx);
    const char *callerstr = "glPushDebugGroup";
-   int s, t, sev;
-   GLint prevStackDepth;
-   GLint currStackDepth;
-   struct gl_debug_msg *emptySlot;
+   struct gl_debug_message *emptySlot;
 
    if (!debug)
       return;
@@ -868,55 +1081,26 @@ _mesa_PushDebugGroup(GLenum source, GLuint id, GLsizei length,
       return;
    }
 
-   message_insert(source, GL_DEBUG_TYPE_PUSH_GROUP, id,
-                  GL_DEBUG_SEVERITY_NOTIFICATION, length,
-                  message, callerstr);
+   if (length < 0)
+      length = strlen(message);
+   if (!validate_length(ctx, callerstr, length))
+      return; /* GL_INVALID_VALUE */
 
-   prevStackDepth = debug->GroupStackDepth;
-   debug->GroupStackDepth++;
-   currStackDepth = debug->GroupStackDepth;
+   log_msg(ctx, gl_enum_to_debug_source(source),
+           MESA_DEBUG_TYPE_PUSH_GROUP, id,
+           MESA_DEBUG_SEVERITY_NOTIFICATION, length,
+           message);
 
    /* pop reuses the message details from push so we store this */
-   if (length < 0)
-      length = strlen(message);
-   emptySlot = &debug->DebugGroupMsgs[debug->GroupStackDepth];
-   store_message_details(emptySlot, gl_enum_to_debug_source(source),
-                         gl_enum_to_debug_type(GL_DEBUG_TYPE_PUSH_GROUP),
-                         id,
-                   gl_enum_to_debug_severity(GL_DEBUG_SEVERITY_NOTIFICATION),
-                         length, message);
-
-   /* inherit the control volume of the debug group previously residing on
-    * the top of the debug group stack
-    */
-   for (s = 0; s < MESA_DEBUG_SOURCE_COUNT; s++) {
-      for (t = 0; t < MESA_DEBUG_TYPE_COUNT; t++) {
-         /* copy id settings */
-         debug->Namespaces[currStackDepth][s][t].IDs =
-            _mesa_HashClone(debug->Namespaces[prevStackDepth][s][t].IDs);
-
-         for (sev = 0; sev < MESA_DEBUG_SEVERITY_COUNT; sev++) {
-            struct gl_debug_severity *entry, *prevEntry;
-            struct simple_node *node;
-
-            /* copy default settings for unknown ids */
-            debug->Defaults[currStackDepth][sev][s][t] =
-               debug->Defaults[prevStackDepth][sev][s][t];
-
-            /* copy known id severity settings */
-            make_empty_list(&debug->Namespaces[currStackDepth][s][t].Severity[sev]);
-            foreach(node, &debug->Namespaces[prevStackDepth][s][t].Severity[sev]) {
-               prevEntry = (struct gl_debug_severity *)node;
-               entry = malloc(sizeof *entry);
-               if (!entry)
-                  return;
-
-               entry->ID = prevEntry->ID;
-               insert_at_tail(&debug->Namespaces[currStackDepth][s][t].Severity[sev], &entry->link);
-            }
-         }
-      }
-   }
+   emptySlot = debug_get_group_message(debug);
+   debug_message_store(emptySlot,
+                       gl_enum_to_debug_source(source),
+                       gl_enum_to_debug_type(GL_DEBUG_TYPE_PUSH_GROUP),
+                       id,
+                       gl_enum_to_debug_severity(GL_DEBUG_SEVERITY_NOTIFICATION),
+                       length, message);
+
+   debug_push_group(debug);
 }
 
 
@@ -926,8 +1110,7 @@ _mesa_PopDebugGroup(void)
    GET_CURRENT_CONTEXT(ctx);
    struct gl_debug_state *debug = _mesa_get_debug_state(ctx);
    const char *callerstr = "glPopDebugGroup";
-   struct gl_debug_msg *gdmessage;
-   GLint prevStackDepth;
+   struct gl_debug_message *gdmessage;
 
    if (!debug)
       return;
@@ -937,26 +1120,16 @@ _mesa_PopDebugGroup(void)
       return;
    }
 
-   prevStackDepth = debug->GroupStackDepth;
-   debug->GroupStackDepth--;
+   debug_pop_group(debug);
 
-   gdmessage = &debug->DebugGroupMsgs[prevStackDepth];
-   /* using log_msg() directly here as verification of parameters
-    * already done in push
-    */
+   gdmessage = debug_get_group_message(debug);
    log_msg(ctx, gdmessage->source,
            gl_enum_to_debug_type(GL_DEBUG_TYPE_POP_GROUP),
            gdmessage->id,
            gl_enum_to_debug_severity(GL_DEBUG_SEVERITY_NOTIFICATION),
            gdmessage->length, gdmessage->message);
 
-   if (gdmessage->message != (char*)out_of_memory)
-      free(gdmessage->message);
-   gdmessage->message = NULL;
-   gdmessage->length = 0;
-
-   /* free popped debug group data */
-   free_errors_data(ctx, prevStackDepth);
+   debug_message_clear(gdmessage);
 }
 
 
@@ -967,20 +1140,11 @@ _mesa_init_errors(struct gl_context *ctx)
 }
 
 
-/**
- * Loop through debug group stack tearing down states for
- * filtering debug messages.  Then free debug output state.
- */
 void
 _mesa_free_errors_data(struct gl_context *ctx)
 {
    if (ctx->Debug) {
-      GLint i;
-
-      for (i = 0; i <= ctx->Debug->GroupStackDepth; i++) {
-         free_errors_data(ctx, i);
-      }
-      free(ctx->Debug);
+      debug_destroy(ctx->Debug);
       /* set to NULL just in case it is used before context is completely gone. */
       ctx->Debug = NULL;
    }
@@ -1198,11 +1362,16 @@ _mesa_error( struct gl_context *ctx, GLenum error, const char *fmtString, ... )
    debug_get_id(&error_msg_id);
 
    do_output = should_output(ctx, error, fmtString);
-   do_log = should_log(ctx,
-                       MESA_DEBUG_SOURCE_API,
-                       MESA_DEBUG_TYPE_ERROR,
-                       error_msg_id,
-                       MESA_DEBUG_SEVERITY_HIGH);
+   if (ctx->Debug) {
+      do_log = debug_is_message_enabled(ctx->Debug,
+                                        MESA_DEBUG_SOURCE_API,
+                                        MESA_DEBUG_TYPE_ERROR,
+                                        error_msg_id,
+                                        MESA_DEBUG_SEVERITY_HIGH);
+   }
+   else {
+      do_log = GL_FALSE;
+   }
 
    if (do_output || do_log) {
       char s[MAX_DEBUG_MESSAGE_LENGTH], s2[MAX_DEBUG_MESSAGE_LENGTH];
diff --git a/mesalib/src/mesa/main/errors.h b/mesalib/src/mesa/main/errors.h
index e0706e5b9..06d0b21fc 100644
--- a/mesalib/src/mesa/main/errors.h
+++ b/mesalib/src/mesa/main/errors.h
@@ -83,8 +83,14 @@ _mesa_gl_debug(struct gl_context *ctx,
    }                                                                      \
 } while (0)
 
-struct gl_debug_state *
-_mesa_get_debug_state(struct gl_context *ctx);
+bool
+_mesa_set_debug_state_int(struct gl_context *ctx, GLenum pname, GLint val);
+
+GLint
+_mesa_get_debug_state_int(struct gl_context *ctx, GLenum pname);
+
+void *
+_mesa_get_debug_state_ptr(struct gl_context *ctx, GLenum pname);
 
 extern void
 _mesa_shader_debug(struct gl_context *ctx, GLenum type, GLuint *id,
diff --git a/mesalib/src/mesa/main/ff_fragment_shader.cpp b/mesalib/src/mesa/main/ff_fragment_shader.cpp
index 66c18fa16..605f3713e 100644
--- a/mesalib/src/mesa/main/ff_fragment_shader.cpp
+++ b/mesalib/src/mesa/main/ff_fragment_shader.cpp
@@ -42,6 +42,7 @@ extern "C" {
 #include "program/prog_statevars.h"
 #include "program/programopt.h"
 #include "texenvprogram.h"
+#include "texobj.h"
 }
 #include "main/uniforms.h"
 #include "../glsl/glsl_types.h"
@@ -290,18 +291,6 @@ need_saturate( GLuint mode )
    }
 }
 
-
-
-/**
- * Translate TEXTURE_x_BIT to TEXTURE_x_INDEX.
- */
-static GLuint translate_tex_src_bit( GLbitfield bit )
-{
-   ASSERT(bit);
-   return ffs(bit) - 1;
-}
-
-
 #define VERT_BIT_TEX_ANY    (0xff << VERT_ATTRIB_TEX0)
 
 /**
@@ -430,7 +419,7 @@ static GLuint make_state_key( struct gl_context *ctx,  struct state_key *key )
       const struct gl_sampler_object *samp;
       GLenum format;
 
-      if (!texUnit->_ReallyEnabled || !texUnit->Enabled)
+      if (!texUnit->_Current || !texUnit->Enabled)
          continue;
 
       samp = _mesa_get_samplerobj(ctx, i);
@@ -441,8 +430,8 @@ static GLuint make_state_key( struct gl_context *ctx,  struct state_key *key )
       key->nr_enabled_units = i + 1;
       inputs_referenced |= VARYING_BIT_TEX(i);
 
-      key->unit[i].source_index =
-         translate_tex_src_bit(texUnit->_ReallyEnabled);
+      key->unit[i].source_index = _mesa_tex_target_to_index(ctx,
+                                                            texObj->Target);
 
       key->unit[i].shadow =
          ((samp->CompareMode == GL_COMPARE_R_TO_TEXTURE) &&
diff --git a/mesalib/src/mesa/main/ffvertex_prog.c b/mesalib/src/mesa/main/ffvertex_prog.c
index c5583c965..728cf968b 100644
--- a/mesalib/src/mesa/main/ffvertex_prog.c
+++ b/mesalib/src/mesa/main/ffvertex_prog.c
@@ -233,13 +233,13 @@ static void make_state_key( struct gl_context *ctx, struct state_key *key )
 
    if (ctx->Texture._TexGenEnabled ||
        ctx->Texture._TexMatEnabled ||
-       ctx->Texture._EnabledUnits)
+       ctx->Texture._MaxEnabledTexImageUnit != -1)
       key->texture_enabled_global = 1;
 
    for (i = 0; i < MAX_TEXTURE_COORD_UNITS; i++) {
       struct gl_texture_unit *texUnit = &ctx->Texture.Unit[i];
 
-      if (texUnit->_ReallyEnabled)
+      if (texUnit->_Current)
 	 key->unit[i].texunit_really_enabled = 1;
 
       if (ctx->Point.PointSprite)
diff --git a/mesalib/src/mesa/main/get.c b/mesalib/src/mesa/main/get.c
index 6d9579008..fe35ff3ef 100644
--- a/mesalib/src/mesa/main/get.c
+++ b/mesalib/src/mesa/main/get.c
@@ -989,24 +989,10 @@ find_custom_value(struct gl_context *ctx, const struct value_desc *d, union valu
       break;
    /* GL_KHR_DEBUG */
    case GL_DEBUG_LOGGED_MESSAGES:
-      {
-         struct gl_debug_state *debug = _mesa_get_debug_state(ctx);
-         v->value_int = debug ? debug->NumMessages : 0;
-      }
-      break;
    case GL_DEBUG_NEXT_LOGGED_MESSAGE_LENGTH:
-      {
-         struct gl_debug_state *debug = _mesa_get_debug_state(ctx);
-         v->value_int = debug ? debug->NextMsgLength : 0;
-      }
-      break;
    case GL_DEBUG_GROUP_STACK_DEPTH:
-      {
-         struct gl_debug_state *debug = _mesa_get_debug_state(ctx);
-         v->value_int = debug ? debug->GroupStackDepth : 0;
-      }
+      v->value_int = _mesa_get_debug_state_int(ctx, d->pname);
       break;
-
    /* GL_ARB_shader_atomic_counters */
    case GL_ATOMIC_COUNTER_BUFFER_BINDING:
       v->value_int = ctx->AtomicBuffer->Name;
diff --git a/mesalib/src/mesa/main/getstring.c b/mesalib/src/mesa/main/getstring.c
index b0bd3190b..431d60b03 100644
--- a/mesalib/src/mesa/main/getstring.c
+++ b/mesalib/src/mesa/main/getstring.c
@@ -253,22 +253,11 @@ _mesa_GetPointerv( GLenum pname, GLvoid **params )
          *params = (GLvoid *) ctx->Array.VAO->VertexAttrib[VERT_ATTRIB_POINT_SIZE].Ptr;
          break;
       case GL_DEBUG_CALLBACK_FUNCTION_ARB:
-         if (!_mesa_is_desktop_gl(ctx)) {
-            goto invalid_pname;
-         }
-         else {
-            struct gl_debug_state *debug = _mesa_get_debug_state(ctx);
-            *params = debug ? (void *) debug->Callback : NULL;
-         }
-         break;
       case GL_DEBUG_CALLBACK_USER_PARAM_ARB:
-         if (!_mesa_is_desktop_gl(ctx)) {
+         if (!_mesa_is_desktop_gl(ctx))
             goto invalid_pname;
-         }
-         else {
-            struct gl_debug_state *debug = _mesa_get_debug_state(ctx);
-            *params = debug ? (void *) debug->CallbackData : NULL;
-         }
+         else
+            *params = _mesa_get_debug_state_ptr(ctx, pname);
          break;
       default:
          goto invalid_pname;
diff --git a/mesalib/src/mesa/main/mtypes.h b/mesalib/src/mesa/main/mtypes.h
index 66943836c..5fbfffe98 100644
--- a/mesalib/src/mesa/main/mtypes.h
+++ b/mesalib/src/mesa/main/mtypes.h
@@ -76,6 +76,7 @@ struct gl_list_extensions;
 struct gl_meta_state;
 struct gl_program_cache;
 struct gl_texture_object;
+struct gl_debug_state;
 struct gl_context;
 struct st_context;
 struct gl_uniform_storage;
@@ -1085,7 +1086,6 @@ typedef enum
 
 /**
  * Bit flags for each type of texture object
- * Used for Texture.Unit[]._ReallyEnabled flags.
  */
 /*@{*/
 #define TEXTURE_2D_MULTISAMPLE_BIT (1 << TEXTURE_2D_MULTISAMPLE_INDEX)
@@ -1327,7 +1327,6 @@ struct gl_texgen
 struct gl_texture_unit
 {
    GLbitfield Enabled;          /**< bitmask of TEXTURE_*_BIT flags */
-   GLbitfield _ReallyEnabled;   /**< 0 or exactly one of TEXTURE_*_BIT flags */
 
    GLenum EnvMode;              /**< GL_MODULATE, GL_DECAL, GL_BLEND, etc. */
    GLclampf EnvColor[4];
@@ -1388,9 +1387,6 @@ struct gl_texture_attrib
    /** GL_ARB_seamless_cubemap */
    GLboolean CubeMapSeamless;
 
-   /** Texture units/samplers used by vertex or fragment texturing */
-   GLbitfield _EnabledUnits;
-
    /** Texture coord units/sets used for fragment texturing */
    GLbitfield _EnabledCoordUnits;
 
@@ -1403,8 +1399,11 @@ struct gl_texture_attrib
    /** Bitwise-OR of all Texture.Unit[i]._GenFlags */
    GLbitfield _GenFlags;
 
-   /** Upper bound on _ReallyEnabled texunits. */
+   /** Largest index of a texture unit with _Current != NULL. */
    GLint _MaxEnabledTexImageUnit;
+
+   /** Largest index + 1 of texture units that have had any CurrentTex set. */
+   GLint NumCurrentTexUsed;
 };
 
 
@@ -3822,45 +3821,6 @@ enum mesa_debug_severity {
 /** @} */
 
 /**
- * An error, warning, or other piece of debug information for an application
- * to consume via GL_ARB_debug_output/GL_KHR_debug.
- */
-struct gl_debug_msg
-{
-   enum mesa_debug_source source;
-   enum mesa_debug_type type;
-   GLuint id;
-   enum mesa_debug_severity severity;
-   GLsizei length;
-   GLcharARB *message;
-};
-
-struct gl_debug_namespace
-{
-   struct _mesa_HashTable *IDs;
-   unsigned ZeroID; /* a HashTable won't take zero, so store its state here */
-   /** lists of IDs in the hash table at each severity */
-   struct simple_node Severity[MESA_DEBUG_SEVERITY_COUNT];
-};
-
-struct gl_debug_state
-{
-   GLDEBUGPROC Callback;
-   const void *CallbackData;
-   GLboolean SyncOutput;
-   GLboolean DebugOutput;
-   GLboolean Defaults[MAX_DEBUG_GROUP_STACK_DEPTH][MESA_DEBUG_SEVERITY_COUNT][MESA_DEBUG_SOURCE_COUNT][MESA_DEBUG_TYPE_COUNT];
-   struct gl_debug_namespace Namespaces[MAX_DEBUG_GROUP_STACK_DEPTH][MESA_DEBUG_SOURCE_COUNT][MESA_DEBUG_TYPE_COUNT];
-   struct gl_debug_msg Log[MAX_DEBUG_LOGGED_MESSAGES];
-   struct gl_debug_msg DebugGroupMsgs[MAX_DEBUG_GROUP_STACK_DEPTH];
-   GLint GroupStackDepth;
-   GLint NumMessages;
-   GLint NextMsg;
-   GLint NextMsgLength; /* redundant, but copied here from Log[NextMsg].length
-                           for the sake of the offsetof() code in get.c */
-};
-
-/**
  * Enum for the OpenGL APIs we know about and may support.
  *
  * NOTE: This must match the api_enum table in
diff --git a/mesalib/src/mesa/main/texobj.c b/mesalib/src/mesa/main/texobj.c
index 918dd59ed..85246c8ab 100644
--- a/mesalib/src/mesa/main/texobj.c
+++ b/mesalib/src/mesa/main/texobj.c
@@ -1094,7 +1094,7 @@ unbind_texobj_from_texunits(struct gl_context *ctx,
 {
    GLuint u, tex;
 
-   for (u = 0; u < Elements(ctx->Texture.Unit); u++) {
+   for (u = 0; u < ctx->Texture.NumCurrentTexUsed; u++) {
       struct gl_texture_unit *unit = &ctx->Texture.Unit[u];
       for (tex = 0; tex < NUM_TEXTURE_TARGETS; tex++) {
          if (texObj == unit->CurrentTex[tex]) {
@@ -1353,6 +1353,8 @@ _mesa_BindTexture( GLenum target, GLuint texName )
     * count hits zero.
     */
    _mesa_reference_texobj(&texUnit->CurrentTex[targetIndex], newTexObj);
+   ctx->Texture.NumCurrentTexUsed = MAX2(ctx->Texture.NumCurrentTexUsed,
+                                         ctx->Texture.CurrentUnit + 1);
    ASSERT(texUnit->CurrentTex[targetIndex]);
 
    /* Pass BindTexture call to device driver */
diff --git a/mesalib/src/mesa/main/texparam.c b/mesalib/src/mesa/main/texparam.c
index 40790ff0e..dc17ea584 100644
--- a/mesalib/src/mesa/main/texparam.c
+++ b/mesalib/src/mesa/main/texparam.c
@@ -485,7 +485,7 @@ set_tex_parameteri(struct gl_context *ctx,
          const GLuint comp = pname - GL_TEXTURE_SWIZZLE_R_EXT;
          const GLint swz = comp_to_swizzle(params[0]);
          if (swz < 0) {
-            _mesa_error(ctx, GL_INVALID_OPERATION,
+            _mesa_error(ctx, GL_INVALID_ENUM,
                         "glTexParameter(swizzle 0x%x)", params[0]);
             return GL_FALSE;
          }
@@ -510,7 +510,7 @@ set_tex_parameteri(struct gl_context *ctx,
                set_swizzle_component(&texObj->_Swizzle, comp, swz);
             }
             else {
-               _mesa_error(ctx, GL_INVALID_OPERATION,
+               _mesa_error(ctx, GL_INVALID_ENUM,
                            "glTexParameter(swizzle 0x%x)", params[comp]);
                return GL_FALSE;
             }
diff --git a/mesalib/src/mesa/main/texstate.c b/mesalib/src/mesa/main/texstate.c
index b68920ce1..91b290691 100644
--- a/mesalib/src/mesa/main/texstate.c
+++ b/mesalib/src/mesa/main/texstate.c
@@ -40,7 +40,7 @@
 #include "teximage.h"
 #include "texstate.h"
 #include "mtypes.h"
-
+#include "bitset.h"
 
 
 /**
@@ -108,6 +108,10 @@ _mesa_copy_texture_state( const struct gl_context *src, struct gl_context *dst )
          for (tex = 0; tex < NUM_TEXTURE_TARGETS; tex++) {
             _mesa_reference_texobj(&dst->Texture.Unit[u].CurrentTex[tex],
                                    src->Texture.Unit[u].CurrentTex[tex]);
+            if (src->Texture.Unit[u].CurrentTex[tex]) {
+               dst->Texture.NumCurrentTexUsed =
+                  MAX2(dst->Texture.NumCurrentTexUsed, u + 1);
+            }
          }
          _mesa_unlock_context_textures(dst);
       }
@@ -371,7 +375,7 @@ update_texture_matrices( struct gl_context *ctx )
       if (_math_matrix_is_dirty(ctx->TextureMatrixStack[u].Top)) {
 	 _math_matrix_analyse( ctx->TextureMatrixStack[u].Top );
 
-	 if (ctx->Texture.Unit[u]._ReallyEnabled &&
+	 if (ctx->Texture.Unit[u]._Current &&
 	     ctx->TextureMatrixStack[u].Top->type != MATRIX_IDENTITY)
 	    ctx->Texture._TexMatEnabled |= ENABLE_TEXMAT(u);
       }
@@ -515,83 +519,151 @@ update_texgen(struct gl_context *ctx)
    }
 }
 
-/**
- * \note This routine refers to derived texture matrix values to
- * compute the ENABLE_TEXMAT flags, but is only called on
- * _NEW_TEXTURE.  On changes to _NEW_TEXTURE_MATRIX, the ENABLE_TEXMAT
- * flags are updated by _mesa_update_texture_matrices, above.
- *
- * \param ctx GL context.
- */
+static struct gl_texture_object *
+update_single_program_texture(struct gl_context *ctx, struct gl_program *prog,
+                              int s)
+{
+   gl_texture_index target_index;
+   struct gl_texture_unit *texUnit;
+   struct gl_texture_object *texObj;
+   struct gl_sampler_object *sampler;
+   int unit;
+
+   if (!(prog->SamplersUsed & (1 << s)))
+      return NULL;
+
+   unit = prog->SamplerUnits[s];
+   texUnit = &ctx->Texture.Unit[unit];
+
+   /* Note: If more than one bit was set in TexturesUsed[unit], then we should
+    * have had the draw call rejected already.  From the GL 4.4 specification,
+    * section 7.10 ("Samplers"):
+    *
+    *     "It is not allowed to have variables of different sampler types
+    *      pointing to the same texture image unit within a program
+    *      object. This situation can only be detected at the next rendering
+    *      command issued which triggers shader invocations, and an
+    *      INVALID_OPERATION error will then be generated."
+    */
+   target_index = ffs(prog->TexturesUsed[unit]) - 1;
+   texObj = texUnit->CurrentTex[target_index];
+
+   sampler = texUnit->Sampler ?
+      texUnit->Sampler : &texObj->Sampler;
+
+   if (likely(texObj)) {
+      if (_mesa_is_texture_complete(texObj, sampler))
+         return texObj;
+
+      _mesa_test_texobj_completeness(ctx, texObj);
+      if (_mesa_is_texture_complete(texObj, sampler))
+         return texObj;
+   }
+
+   /* If we've reached this point, we didn't find a complete texture of the
+    * shader's target.  From the GL 4.4 core specification, section 11.1.3.5
+    * ("Texture Access"):
+    *
+    *     "If a sampler is used in a shader and the sampler’s associated
+    *      texture is not complete, as defined in section 8.17, (0, 0, 0, 1)
+    *      will be returned for a non-shadow sampler and 0 for a shadow
+    *      sampler."
+    *
+    * Mesa implements this by creating a hidden texture object with a pixel of
+    * that value.
+    */
+   texObj = _mesa_get_fallback_texture(ctx, target_index);
+   assert(texObj);
+
+   return texObj;
+}
+
 static void
-update_texture_state( struct gl_context *ctx )
+update_program_texture_state(struct gl_context *ctx, struct gl_program **prog,
+                             BITSET_WORD *enabled_texture_units)
 {
-   GLuint unit;
-   struct gl_program *prog[MESA_SHADER_STAGES];
-   GLbitfield enabledFragUnits = 0x0;
    int i;
 
    for (i = 0; i < MESA_SHADER_STAGES; i++) {
-      if (ctx->_Shader->CurrentProgram[i] &&
-          ctx->_Shader->CurrentProgram[i]->LinkStatus) {
-         prog[i] = ctx->_Shader->CurrentProgram[i]->_LinkedShaders[i]->Program;
-      } else {
-         if (i == MESA_SHADER_FRAGMENT && ctx->FragmentProgram._Enabled)
-            prog[i] = &ctx->FragmentProgram.Current->Base;
-         else
-            prog[i] = NULL;
+      int s;
+
+      if (!prog[i])
+         continue;
+
+      /* We can't only do the shifting trick as the loop condition because if
+       * sampler 31 is active, the next iteration tries to shift by 32, which is
+       * undefined.
+       */
+      for (s = 0; s < MAX_SAMPLERS && (1 << s) <= prog[i]->SamplersUsed; s++) {
+         struct gl_texture_object *texObj;
+
+         texObj = update_single_program_texture(ctx, prog[i], s);
+         if (texObj) {
+            int unit = prog[i]->SamplerUnits[s];
+            _mesa_reference_texobj(&ctx->Texture.Unit[unit]._Current, texObj);
+            BITSET_SET(enabled_texture_units, unit);
+            ctx->Texture._MaxEnabledTexImageUnit =
+               MAX2(ctx->Texture._MaxEnabledTexImageUnit, (int)unit);
+         }
       }
    }
 
-   /* TODO: only set this if there are actual changes */
-   ctx->NewState |= _NEW_TEXTURE;
+   if (prog[MESA_SHADER_FRAGMENT]) {
+      const GLuint coordMask = (1 << MAX_TEXTURE_COORD_UNITS) - 1;
+      ctx->Texture._EnabledCoordUnits |=
+         (prog[MESA_SHADER_FRAGMENT]->InputsRead >> VARYING_SLOT_TEX0) &
+         coordMask;
+   }
+}
 
-   ctx->Texture._EnabledUnits = 0x0;
-   ctx->Texture._GenFlags = 0x0;
-   ctx->Texture._TexMatEnabled = 0x0;
-   ctx->Texture._TexGenEnabled = 0x0;
-   ctx->Texture._MaxEnabledTexImageUnit = -1;
+static void
+update_ff_texture_state(struct gl_context *ctx,
+                        BITSET_WORD *enabled_texture_units)
+{
+   int unit;
 
-   /*
-    * Update texture unit state.
-    */
-   for (unit = 0; unit < ctx->Const.MaxCombinedTextureImageUnits; unit++) {
+   for (unit = 0; unit < ctx->Const.MaxTextureUnits; unit++) {
       struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit];
-      GLbitfield enabledTargetsByStage[MESA_SHADER_STAGES];
-      GLbitfield enabledTargets = 0x0;
       GLuint texIndex;
 
-      /* Get the bitmask of texture target enables.
-       * enableBits will be a mask of the TEXTURE_*_BIT flags indicating
-       * which texture targets are enabled (fixed function) or referenced
-       * by a fragment program/program.  When multiple flags are set, we'll
-       * settle on the one with highest priority (see below).
-       */
-      for (i = 0; i < MESA_SHADER_STAGES; i++) {
-         if (prog[i])
-            enabledTargetsByStage[i] = prog[i]->TexturesUsed[unit];
-         else if (i == MESA_SHADER_FRAGMENT)
-            enabledTargetsByStage[i] = texUnit->Enabled;
-         else
-            enabledTargetsByStage[i] = 0;
-         enabledTargets |= enabledTargetsByStage[i];
-      }
-
-      texUnit->_ReallyEnabled = 0x0;
+      if (texUnit->Enabled == 0x0)
+         continue;
 
-      if (enabledTargets == 0x0) {
-         /* neither vertex nor fragment processing uses this unit */
+      /* If a shader already dictated what texture target was used for this
+       * unit, just go along with it.
+       */
+      if (BITSET_TEST(enabled_texture_units, unit))
          continue;
-      }
 
-      /* Look for the highest priority texture target that's enabled (or used
-       * by the vert/frag shaders) and "complete".  That's the one we'll use
-       * for texturing.
+      /* From the GL 4.4 compat specification, section 16.2 ("Texture Application"):
+       *
+       *     "Texturing is enabled or disabled using the generic Enable and
+       *      Disable commands, respectively, with the symbolic constants
+       *      TEXTURE_1D, TEXTURE_2D, TEXTURE_RECTANGLE, TEXTURE_3D, or
+       *      TEXTURE_CUBE_MAP to enable the one-, two-, rectangular,
+       *      three-dimensional, or cube map texture, respectively. If more
+       *      than one of these textures is enabled, the first one enabled
+       *      from the following list is used:
+       *
+       *      • cube map texture
+       *      • three-dimensional texture
+       *      • rectangular texture
+       *      • two-dimensional texture
+       *      • one-dimensional texture"
        *
        * Note that the TEXTURE_x_INDEX values are in high to low priority.
+       * Also:
+       *
+       *     "If a texture unit is disabled or has an invalid or incomplete
+       *      texture (as defined in section 8.17) bound to it, then blending
+       *      is disabled for that texture unit. If the texture environment
+       *      for a given enabled texture unit references a disabled texture
+       *      unit, or an invalid or incomplete texture that is bound to
+       *      another unit, then the results of texture blending are
+       *      undefined."
        */
       for (texIndex = 0; texIndex < NUM_TEXTURE_TARGETS; texIndex++) {
-         if (enabledTargets & (1 << texIndex)) {
+         if (texUnit->Enabled & (1 << texIndex)) {
             struct gl_texture_object *texObj = texUnit->CurrentTex[texIndex];
             struct gl_sampler_object *sampler = texUnit->Sampler ?
                texUnit->Sampler : &texObj->Sampler;
@@ -600,62 +672,84 @@ update_texture_state( struct gl_context *ctx )
                _mesa_test_texobj_completeness(ctx, texObj);
             }
             if (_mesa_is_texture_complete(texObj, sampler)) {
-               texUnit->_ReallyEnabled = 1 << texIndex;
                _mesa_reference_texobj(&texUnit->_Current, texObj);
                break;
             }
          }
       }
 
-      if (!texUnit->_ReallyEnabled) {
-         if (prog[MESA_SHADER_FRAGMENT]) {
-            /* If we get here it means the shader is expecting a texture
-             * object, but there isn't one (or it's incomplete).  Use the
-             * fallback texture.
-             */
-            struct gl_texture_object *texObj;
-            gl_texture_index texTarget;
-
-            texTarget = (gl_texture_index) (ffs(enabledTargets) - 1);
-            texObj = _mesa_get_fallback_texture(ctx, texTarget);
-            
-            assert(texObj);
-            if (!texObj) {
-               /* invalid fallback texture: don't enable the texture unit */
-               continue;
-            }
-
-            _mesa_reference_texobj(&texUnit->_Current, texObj);
-            texUnit->_ReallyEnabled = 1 << texTarget;
-         }
-         else {
-            /* fixed-function: texture unit is really disabled */
-            continue;
-         }
-      }
+      if (texIndex == NUM_TEXTURE_TARGETS)
+         continue;
 
       /* if we get here, we know this texture unit is enabled */
+      BITSET_SET(enabled_texture_units, unit);
+      ctx->Texture._MaxEnabledTexImageUnit =
+         MAX2(ctx->Texture._MaxEnabledTexImageUnit, (int)unit);
 
-      ctx->Texture._EnabledUnits |= (1 << unit);
-      ctx->Texture._MaxEnabledTexImageUnit = unit;
+      ctx->Texture._EnabledCoordUnits |= 1 << unit;
 
-      if (enabledTargetsByStage[MESA_SHADER_FRAGMENT])
-         enabledFragUnits |= (1 << unit);
+      update_tex_combine(ctx, texUnit);
+   }
+}
+
+/**
+ * \note This routine refers to derived texture matrix values to
+ * compute the ENABLE_TEXMAT flags, but is only called on
+ * _NEW_TEXTURE.  On changes to _NEW_TEXTURE_MATRIX, the ENABLE_TEXMAT
+ * flags are updated by _mesa_update_texture_matrices, above.
+ *
+ * \param ctx GL context.
+ */
+static void
+update_texture_state( struct gl_context *ctx )
+{
+   struct gl_program *prog[MESA_SHADER_STAGES];
+   int i;
+   int old_max_unit = ctx->Texture._MaxEnabledTexImageUnit;
+   BITSET_DECLARE(enabled_texture_units, MAX_COMBINED_TEXTURE_IMAGE_UNITS);
 
-      if (!prog[MESA_SHADER_FRAGMENT])
-         update_tex_combine(ctx, texUnit);
+   for (i = 0; i < MESA_SHADER_STAGES; i++) {
+      if (ctx->_Shader->CurrentProgram[i] &&
+          ctx->_Shader->CurrentProgram[i]->LinkStatus) {
+         prog[i] = ctx->_Shader->CurrentProgram[i]->_LinkedShaders[i]->Program;
+      } else {
+         if (i == MESA_SHADER_FRAGMENT && ctx->FragmentProgram._Enabled)
+            prog[i] = &ctx->FragmentProgram.Current->Base;
+         else
+            prog[i] = NULL;
+      }
    }
 
+   /* TODO: only set this if there are actual changes */
+   ctx->NewState |= _NEW_TEXTURE;
 
-   /* Determine which texture coordinate sets are actually needed */
-   if (prog[MESA_SHADER_FRAGMENT]) {
-      const GLuint coordMask = (1 << MAX_TEXTURE_COORD_UNITS) - 1;
-      ctx->Texture._EnabledCoordUnits
-         = (prog[MESA_SHADER_FRAGMENT]->InputsRead >> VARYING_SLOT_TEX0) &
-         coordMask;
+   ctx->Texture._GenFlags = 0x0;
+   ctx->Texture._TexMatEnabled = 0x0;
+   ctx->Texture._TexGenEnabled = 0x0;
+   ctx->Texture._MaxEnabledTexImageUnit = -1;
+   ctx->Texture._EnabledCoordUnits = 0x0;
+
+   memset(&enabled_texture_units, 0, sizeof(enabled_texture_units));
+
+   /* First, walk over our programs pulling in all the textures for them.
+    * Programs dictate specific texture targets to be enabled, and for a draw
+    * call to be valid they can't conflict about which texture targets are
+    * used.
+    */
+   update_program_texture_state(ctx, prog, enabled_texture_units);
+
+   /* Also pull in any textures necessary for fixed function fragment shading.
+    */
+   if (!prog[MESA_SHADER_FRAGMENT])
+      update_ff_texture_state(ctx, enabled_texture_units);
+
+   /* Now, clear out the _Current of any disabled texture units. */
+   for (i = 0; i <= ctx->Texture._MaxEnabledTexImageUnit; i++) {
+      if (!BITSET_TEST(enabled_texture_units, i))
+         _mesa_reference_texobj(&ctx->Texture.Unit[i]._Current, NULL);
    }
-   else {
-      ctx->Texture._EnabledCoordUnits = enabledFragUnits;
+   for (i = ctx->Texture._MaxEnabledTexImageUnit + 1; i <= old_max_unit; i++) {
+      _mesa_reference_texobj(&ctx->Texture.Unit[i]._Current, NULL);
    }
 
    if (!prog[MESA_SHADER_FRAGMENT] || !prog[MESA_SHADER_VERTEX])
@@ -796,7 +890,6 @@ _mesa_init_texture(struct gl_context *ctx)
 
    /* Texture group */
    ctx->Texture.CurrentUnit = 0;      /* multitexture */
-   ctx->Texture._EnabledUnits = 0x0;
 
    /* Appendix F.2 of the OpenGL ES 3.0 spec says:
     *
@@ -824,6 +917,8 @@ _mesa_init_texture(struct gl_context *ctx)
    _mesa_reference_buffer_object(ctx, &ctx->Texture.BufferObject,
                                  ctx->Shared->NullBufferObj);
 
+   ctx->Texture.NumCurrentTexUsed = 0;
+
    return GL_TRUE;
 }
 
diff --git a/mesalib/src/mesa/state_tracker/st_atom.c b/mesalib/src/mesa/state_tracker/st_atom.c
index 99e9df26b..625ea2965 100644
--- a/mesalib/src/mesa/state_tracker/st_atom.c
+++ b/mesalib/src/mesa/state_tracker/st_atom.c
@@ -61,6 +61,7 @@ static const struct st_tracked_state *atoms[] =
    &st_update_sampler, /* depends on update_*_texture for swizzle */
    &st_update_framebuffer,
    &st_update_msaa,
+   &st_update_sample_shading,
    &st_update_vs_constants,
    &st_update_gs_constants,
    &st_update_fs_constants,
diff --git a/mesalib/src/mesa/state_tracker/st_atom.h b/mesalib/src/mesa/state_tracker/st_atom.h
index 60d89d75f..c50111d50 100644
--- a/mesalib/src/mesa/state_tracker/st_atom.h
+++ b/mesalib/src/mesa/state_tracker/st_atom.h
@@ -59,6 +59,7 @@ extern const struct st_tracked_state st_update_viewport;
 extern const struct st_tracked_state st_update_scissor;
 extern const struct st_tracked_state st_update_blend;
 extern const struct st_tracked_state st_update_msaa;
+extern const struct st_tracked_state st_update_sample_shading;
 extern const struct st_tracked_state st_update_sampler;
 extern const struct st_tracked_state st_update_fragment_texture;
 extern const struct st_tracked_state st_update_vertex_texture;
diff --git a/mesalib/src/mesa/state_tracker/st_atom_framebuffer.c b/mesalib/src/mesa/state_tracker/st_atom_framebuffer.c
index a17417c35..b195c55b3 100644
--- a/mesalib/src/mesa/state_tracker/st_atom_framebuffer.c
+++ b/mesalib/src/mesa/state_tracker/st_atom_framebuffer.c
@@ -113,6 +113,12 @@ update_framebuffer_state( struct st_context *st )
       pipe_surface_reference(&framebuffer->cbufs[i], NULL);
    }
 
+   /* Remove trailing GL_NONE draw buffers. */
+   while (framebuffer->nr_cbufs &&
+          !framebuffer->cbufs[framebuffer->nr_cbufs-1]) {
+      framebuffer->nr_cbufs--;
+   }
+
    /*
     * Depth/Stencil renderbuffer/surface.
     */
@@ -147,11 +153,10 @@ update_framebuffer_state( struct st_context *st )
    }
 #endif
 
-   /* _mesa_test_framebuffer_completeness refuses framebuffers with no
-    * attachments, so this should never happen.
-    */
-   assert(framebuffer->width  != UINT_MAX);
-   assert(framebuffer->height != UINT_MAX);
+   if (framebuffer->width == UINT_MAX)
+      framebuffer->width = 0;
+   if (framebuffer->height == UINT_MAX)
+      framebuffer->height = 0;
 
    cso_set_framebuffer(st->cso_context, framebuffer);
 }
diff --git a/mesalib/src/mesa/state_tracker/st_atom_msaa.c b/mesalib/src/mesa/state_tracker/st_atom_msaa.c
index 2f3a42e02..703cb5e75 100644
--- a/mesalib/src/mesa/state_tracker/st_atom_msaa.c
+++ b/mesalib/src/mesa/state_tracker/st_atom_msaa.c
@@ -27,8 +27,10 @@
 
 
 #include "st_context.h"
+#include "pipe/p_screen.h"
 #include "pipe/p_context.h"
 #include "st_atom.h"
+#include "st_program.h"
 
 #include "cso_cache/cso_context.h"
 #include "util/u_framebuffer.h"
@@ -70,6 +72,18 @@ static void update_sample_mask( struct st_context *st )
    }
 }
 
+static void update_sample_shading( struct st_context *st )
+{
+   if (!st->fp)
+      return;
+
+   if (!st->ctx->Extensions.ARB_sample_shading)
+      return;
+
+   cso_set_min_samples(
+	 st->cso_context,
+         _mesa_get_min_invocations_per_fragment(st->ctx, &st->fp->Base, false));
+}
 
 const struct st_tracked_state st_update_msaa = {
    "st_update_msaa",					/* name */
@@ -79,3 +93,12 @@ const struct st_tracked_state st_update_msaa = {
    },
    update_sample_mask					/* update */
 };
+
+const struct st_tracked_state st_update_sample_shading = {
+   "st_update_sample_shading",				/* name */
+   {							/* dirty */
+      (_NEW_MULTISAMPLE | _NEW_PROGRAM | _NEW_BUFFERS),	/* mesa */
+      ST_NEW_FRAGMENT_PROGRAM | ST_NEW_FRAMEBUFFER,	/* st */
+   },
+   update_sample_shading				/* update */
+};
diff --git a/mesalib/src/mesa/state_tracker/st_atom_pixeltransfer.c b/mesalib/src/mesa/state_tracker/st_atom_pixeltransfer.c
index 306eeb153..a04163cc1 100644
--- a/mesalib/src/mesa/state_tracker/st_atom_pixeltransfer.c
+++ b/mesalib/src/mesa/state_tracker/st_atom_pixeltransfer.c
@@ -121,7 +121,7 @@ load_color_map_texture(struct gl_context *ctx, struct pipe_resource *pt)
          rgba[2] = ctx->PixelMaps.BtoB.Map[j * bSize / texSize];
          rgba[3] = ctx->PixelMaps.AtoA.Map[i * aSize / texSize];
          util_pack_color(rgba, pt->format, &uc);
-         *(dest + k) = uc.ui;
+         *(dest + k) = uc.ui[0];
       }
    }
 
diff --git a/mesalib/src/mesa/state_tracker/st_cb_clear.c b/mesalib/src/mesa/state_tracker/st_cb_clear.c
index fcd7e1382..371f7fcda 100644
--- a/mesalib/src/mesa/state_tracker/st_cb_clear.c
+++ b/mesalib/src/mesa/state_tracker/st_cb_clear.c
@@ -240,6 +240,7 @@ clear_with_quad(struct gl_context *ctx, unsigned clear_buffers)
    cso_save_depth_stencil_alpha(st->cso_context);
    cso_save_rasterizer(st->cso_context);
    cso_save_sample_mask(st->cso_context);
+   cso_save_min_samples(st->cso_context);
    cso_save_viewport(st->cso_context);
    cso_save_fragment_shader(st->cso_context);
    cso_save_stream_outputs(st->cso_context);
@@ -309,6 +310,7 @@ clear_with_quad(struct gl_context *ctx, unsigned clear_buffers)
    cso_set_vertex_elements(st->cso_context, 2, st->velems_util_draw);
    cso_set_stream_outputs(st->cso_context, 0, NULL, NULL);
    cso_set_sample_mask(st->cso_context, ~0);
+   cso_set_min_samples(st->cso_context, 1);
    cso_set_rasterizer(st->cso_context, &st->clear.raster);
 
    /* viewport state: viewport matching window dims */
@@ -348,6 +350,7 @@ clear_with_quad(struct gl_context *ctx, unsigned clear_buffers)
    cso_restore_depth_stencil_alpha(st->cso_context);
    cso_restore_rasterizer(st->cso_context);
    cso_restore_sample_mask(st->cso_context);
+   cso_restore_min_samples(st->cso_context);
    cso_restore_viewport(st->cso_context);
    cso_restore_fragment_shader(st->cso_context);
    cso_restore_vertex_shader(st->cso_context);
diff --git a/mesalib/src/mesa/state_tracker/st_cb_drawtex.c b/mesalib/src/mesa/state_tracker/st_cb_drawtex.c
index b0a44fd90..f997e6b00 100644
--- a/mesalib/src/mesa/state_tracker/st_cb_drawtex.c
+++ b/mesalib/src/mesa/state_tracker/st_cb_drawtex.c
@@ -123,7 +123,8 @@ st_DrawTex(struct gl_context *ctx, GLfloat x, GLfloat y, GLfloat z,
    /* determine how many enabled sets of texcoords */
    numTexCoords = 0;
    for (i = 0; i < ctx->Const.MaxTextureUnits; i++) {
-      if (ctx->Texture.Unit[i]._ReallyEnabled & TEXTURE_2D_BIT) {
+      if (ctx->Texture.Unit[i]._Current &&
+          ctx->Texture.Unit[i]._Current->Target == GL_TEXTURE_2D) {
          numTexCoords++;
       }
    }
@@ -192,7 +193,8 @@ st_DrawTex(struct gl_context *ctx, GLfloat x, GLfloat y, GLfloat z,
 
       /* texcoords */
       for (i = 0; i < ctx->Const.MaxTextureUnits; i++) {
-         if (ctx->Texture.Unit[i]._ReallyEnabled & TEXTURE_2D_BIT) {
+         if (ctx->Texture.Unit[i]._Current &&
+             ctx->Texture.Unit[i]._Current->Target == GL_TEXTURE_2D) {
             struct gl_texture_object *obj = ctx->Texture.Unit[i]._Current;
             struct gl_texture_image *img = obj->Image[0][obj->BaseLevel];
             const GLfloat wt = (GLfloat) img->Width;
diff --git a/mesalib/src/mesa/state_tracker/st_extensions.c b/mesalib/src/mesa/state_tracker/st_extensions.c
index 70726886e..a245fdf18 100644
--- a/mesalib/src/mesa/state_tracker/st_extensions.c
+++ b/mesalib/src/mesa/state_tracker/st_extensions.c
@@ -425,7 +425,8 @@ void st_init_extensions(struct st_context *st)
       { o(OES_standard_derivatives),         PIPE_CAP_SM3                              },
       { o(ARB_texture_cube_map_array),       PIPE_CAP_CUBE_MAP_ARRAY                   },
       { o(ARB_texture_multisample),          PIPE_CAP_TEXTURE_MULTISAMPLE              },
-      { o(ARB_texture_query_lod),            PIPE_CAP_TEXTURE_QUERY_LOD                }
+      { o(ARB_texture_query_lod),            PIPE_CAP_TEXTURE_QUERY_LOD                },
+      { o(ARB_sample_shading),               PIPE_CAP_SAMPLE_SHADING                   },
    };
 
    /* Required: render target and sampler support */
diff --git a/mesalib/src/mesa/state_tracker/st_glsl_to_tgsi.cpp b/mesalib/src/mesa/state_tracker/st_glsl_to_tgsi.cpp
index 15b5279b8..d1c3856b5 100644
--- a/mesalib/src/mesa/state_tracker/st_glsl_to_tgsi.cpp
+++ b/mesalib/src/mesa/state_tracker/st_glsl_to_tgsi.cpp
@@ -241,7 +241,7 @@ public:
 
    unsigned op;
    st_dst_reg dst;
-   st_src_reg src[3];
+   st_src_reg src[4];
    /** Pointer to the ir source this tree came from for debugging */
    ir_instruction *ir;
    GLboolean cond_update;
@@ -411,7 +411,12 @@ public:
    glsl_to_tgsi_instruction *emit(ir_instruction *ir, unsigned op,
         		        st_dst_reg dst,
         		        st_src_reg src0, st_src_reg src1, st_src_reg src2);
-   
+
+   glsl_to_tgsi_instruction *emit(ir_instruction *ir, unsigned op,
+                                  st_dst_reg dst,
+                                  st_src_reg src0, st_src_reg src1,
+                                  st_src_reg src2, st_src_reg src3);
+
    unsigned get_opcode(ir_instruction *ir, unsigned op,
                     st_dst_reg dst,
                     st_src_reg src0, st_src_reg src1);
@@ -524,8 +529,9 @@ num_inst_src_regs(unsigned opcode)
 
 glsl_to_tgsi_instruction *
 glsl_to_tgsi_visitor::emit(ir_instruction *ir, unsigned op,
-        		 st_dst_reg dst,
-        		 st_src_reg src0, st_src_reg src1, st_src_reg src2)
+                           st_dst_reg dst,
+                           st_src_reg src0, st_src_reg src1,
+                           st_src_reg src2, st_src_reg src3)
 {
    glsl_to_tgsi_instruction *inst = new(mem_ctx) glsl_to_tgsi_instruction();
    int num_reladdr = 0, i;
@@ -540,7 +546,9 @@ glsl_to_tgsi_visitor::emit(ir_instruction *ir, unsigned op,
    num_reladdr += src0.reladdr != NULL || src0.reladdr2 != NULL;
    num_reladdr += src1.reladdr != NULL || src1.reladdr2 != NULL;
    num_reladdr += src2.reladdr != NULL || src2.reladdr2 != NULL;
+   num_reladdr += src3.reladdr != NULL || src3.reladdr2 != NULL;
 
+   reladdr_to_temp(ir, &src3, &num_reladdr);
    reladdr_to_temp(ir, &src2, &num_reladdr);
    reladdr_to_temp(ir, &src1, &num_reladdr);
    reladdr_to_temp(ir, &src0, &num_reladdr);
@@ -556,6 +564,7 @@ glsl_to_tgsi_visitor::emit(ir_instruction *ir, unsigned op,
    inst->src[0] = src0;
    inst->src[1] = src1;
    inst->src[2] = src2;
+   inst->src[3] = src3;
    inst->ir = ir;
    inst->dead_mask = 0;
 
@@ -577,7 +586,7 @@ glsl_to_tgsi_visitor::emit(ir_instruction *ir, unsigned op,
       }
    }
    else {
-      for (i=0; i<3; i++) {
+      for (i=0; i<4; i++) {
          if(inst->src[i].reladdr) {
             switch(inst->src[i].file) {
             case PROGRAM_STATE_VAR:
@@ -600,12 +609,19 @@ glsl_to_tgsi_visitor::emit(ir_instruction *ir, unsigned op,
    return inst;
 }
 
+glsl_to_tgsi_instruction *
+glsl_to_tgsi_visitor::emit(ir_instruction *ir, unsigned op,
+                           st_dst_reg dst, st_src_reg src0,
+                           st_src_reg src1, st_src_reg src2)
+{
+   return emit(ir, op, dst, src0, src1, src2, undef_src);
+}
 
 glsl_to_tgsi_instruction *
 glsl_to_tgsi_visitor::emit(ir_instruction *ir, unsigned op,
         		 st_dst_reg dst, st_src_reg src0, st_src_reg src1)
 {
-   return emit(ir, op, dst, src0, src1, undef_src);
+   return emit(ir, op, dst, src0, src1, undef_src, undef_src);
 }
 
 glsl_to_tgsi_instruction *
@@ -613,13 +629,13 @@ glsl_to_tgsi_visitor::emit(ir_instruction *ir, unsigned op,
         		 st_dst_reg dst, st_src_reg src0)
 {
    assert(dst.writemask != 0);
-   return emit(ir, op, dst, src0, undef_src, undef_src);
+   return emit(ir, op, dst, src0, undef_src, undef_src, undef_src);
 }
 
 glsl_to_tgsi_instruction *
 glsl_to_tgsi_visitor::emit(ir_instruction *ir, unsigned op)
 {
-   return emit(ir, op, undef_dst, undef_src, undef_src, undef_src);
+   return emit(ir, op, undef_dst, undef_src, undef_src, undef_src, undef_src);
 }
 
 /**
@@ -690,7 +706,10 @@ glsl_to_tgsi_visitor::get_opcode(ir_instruction *ir, unsigned op,
 
       case2fi(SSG, ISSG);
       case3(ABS, IABS, IABS);
-      
+
+      case2iu(IBFE, UBFE);
+      case2iu(IMSB, UMSB);
+      case2iu(IMUL_HI, UMUL_HI);
       default: break;
    }
    
@@ -1934,6 +1953,33 @@ glsl_to_tgsi_visitor::visit(ir_expression *ir)
          emit(ir, TGSI_OPCODE_CMP, result_dst, op[0], op[1], op[2]);
       }
       break;
+   case ir_triop_bitfield_extract:
+      emit(ir, TGSI_OPCODE_IBFE, result_dst, op[0], op[1], op[2]);
+      break;
+   case ir_quadop_bitfield_insert:
+      emit(ir, TGSI_OPCODE_BFI, result_dst, op[0], op[1], op[2], op[3]);
+      break;
+   case ir_unop_bitfield_reverse:
+      emit(ir, TGSI_OPCODE_BREV, result_dst, op[0]);
+      break;
+   case ir_unop_bit_count:
+      emit(ir, TGSI_OPCODE_POPC, result_dst, op[0]);
+      break;
+   case ir_unop_find_msb:
+      emit(ir, TGSI_OPCODE_IMSB, result_dst, op[0]);
+      break;
+   case ir_unop_find_lsb:
+      emit(ir, TGSI_OPCODE_LSB, result_dst, op[0]);
+      break;
+   case ir_binop_imul_high:
+      emit(ir, TGSI_OPCODE_IMUL_HI, result_dst, op[0], op[1]);
+      break;
+   case ir_triop_fma:
+      /* NOTE: Perhaps there should be a special opcode that enforces fused
+       * mul-add. Just use MAD for now.
+       */
+      emit(ir, TGSI_OPCODE_MAD, result_dst, op[0], op[1], op[2]);
+      break;
    case ir_unop_pack_snorm_2x16:
    case ir_unop_pack_unorm_2x16:
    case ir_unop_pack_half_2x16:
@@ -1947,22 +1993,14 @@ glsl_to_tgsi_visitor::visit(ir_expression *ir)
    case ir_unop_unpack_snorm_4x8:
    case ir_unop_unpack_unorm_4x8:
    case ir_binop_pack_half_2x16_split:
-   case ir_unop_bitfield_reverse:
-   case ir_unop_bit_count:
-   case ir_unop_find_msb:
-   case ir_unop_find_lsb:
    case ir_binop_bfm:
-   case ir_triop_fma:
    case ir_triop_bfi:
-   case ir_triop_bitfield_extract:
-   case ir_quadop_bitfield_insert:
    case ir_quadop_vector:
    case ir_binop_vector_extract:
    case ir_triop_vector_insert:
    case ir_binop_ldexp:
    case ir_binop_carry:
    case ir_binop_borrow:
-   case ir_binop_imul_high:
       /* This operation is not supported, or should have already been handled.
        */
       assert(!"Invalid ir opcode in glsl_to_tgsi_visitor::visit()");
@@ -4163,7 +4201,11 @@ struct st_translate {
 static unsigned mesa_sysval_to_semantic[SYSTEM_VALUE_MAX] = {
    TGSI_SEMANTIC_FACE,
    TGSI_SEMANTIC_VERTEXID,
-   TGSI_SEMANTIC_INSTANCEID
+   TGSI_SEMANTIC_INSTANCEID,
+   TGSI_SEMANTIC_SAMPLEID,
+   TGSI_SEMANTIC_SAMPLEPOS,
+   TGSI_SEMANTIC_SAMPLEMASK,
+   TGSI_SEMANTIC_INVOCATIONID,
 };
 
 /**
@@ -4381,7 +4423,8 @@ translate_dst(struct st_translate *t,
          break;
 
       case TGSI_PROCESSOR_FRAGMENT:
-         if (dst_reg->index >= FRAG_RESULT_COLOR) {
+         if (dst_reg->index == FRAG_RESULT_COLOR ||
+             dst_reg->index >= FRAG_RESULT_DATA0) {
             dst = ureg_saturate(dst);
          }
          break;
@@ -4876,6 +4919,15 @@ st_translate_program(
                                              TGSI_SEMANTIC_COLOR,
                                              outputSemanticIndex[i]);
             break;
+         case TGSI_SEMANTIC_SAMPLEMASK:
+            t->outputs[i] = ureg_DECL_output(ureg,
+                                             TGSI_SEMANTIC_SAMPLEMASK,
+                                             outputSemanticIndex[i]);
+            /* TODO: If we ever support more than 32 samples, this will have
+             * to become an array.
+             */
+            t->outputs[i] = ureg_writemask(t->outputs[i], TGSI_WRITEMASK_X);
+            break;
          default:
             assert(!"fragment shader outputs must be POSITION/STENCIL/COLOR");
             ret = PIPE_ERROR_BAD_INPUT;
@@ -5343,6 +5395,7 @@ st_link_shader(struct gl_context *ctx, struct gl_shader_program *prog)
                          DIV_TO_MUL_RCP |
                          EXP_TO_EXP2 |
                          LOG_TO_LOG2 |
+                         LDEXP_TO_ARITH |
                          (options->EmitNoPow ? POW_TO_EXP2 : 0) |
                          (!ctx->Const.NativeIntegers ? INT_DIV_TO_MUL_RCP : 0));
 
diff --git a/mesalib/src/mesa/state_tracker/st_manager.c b/mesalib/src/mesa/state_tracker/st_manager.c
index 314d34232..706af7fd1 100644
--- a/mesalib/src/mesa/state_tracker/st_manager.c
+++ b/mesalib/src/mesa/state_tracker/st_manager.c
@@ -183,8 +183,13 @@ st_framebuffer_validate(struct st_framebuffer *stfb,
    uint width, height;
    unsigned i;
    boolean changed = FALSE;
-   int32_t new_stamp = p_atomic_read(&stfb->iface->stamp);
+   int32_t new_stamp;
 
+   /* Check for incomplete framebuffers (e.g. EGL_KHR_surfaceless_context) */
+   if (!stfb->iface)
+      return;
+
+   new_stamp = p_atomic_read(&stfb->iface->stamp);
    if (stfb->iface_stamp == new_stamp)
       return;
 
@@ -663,13 +668,11 @@ st_api_create_context(struct st_api *stapi, struct st_manager *smapi,
    }
 
    if (attribs->flags & ST_CONTEXT_FLAG_DEBUG){
-      struct gl_debug_state *debug = _mesa_get_debug_state(st->ctx);
-      if (!debug) {
+      if (!_mesa_set_debug_state_int(st->ctx, GL_DEBUG_OUTPUT, GL_TRUE)) {
          *error = ST_CONTEXT_ERROR_NO_MEMORY;
          return NULL;
       }
       st->ctx->Const.ContextFlags |= GL_CONTEXT_FLAG_DEBUG_BIT;
-      debug->DebugOutput = GL_TRUE;
    }
 
    if (attribs->flags & ST_CONTEXT_FLAG_FORWARD_COMPATIBLE)
diff --git a/mesalib/src/mesa/state_tracker/st_program.c b/mesalib/src/mesa/state_tracker/st_program.c
index 2f0ec32f5..f9887a2c4 100644
--- a/mesalib/src/mesa/state_tracker/st_program.c
+++ b/mesalib/src/mesa/state_tracker/st_program.c
@@ -682,6 +682,7 @@ st_translate_fragment_program(struct st_context *st,
       /* handle remaining outputs (color) */
       for (attr = 0; attr < FRAG_RESULT_MAX; attr++) {
          if (outputsWritten & BITFIELD64_BIT(attr)) {
+            int semantic = TGSI_SEMANTIC_COLOR;
             switch (attr) {
             case FRAG_RESULT_DEPTH:
             case FRAG_RESULT_STENCIL:
@@ -689,17 +690,21 @@ st_translate_fragment_program(struct st_context *st,
                assert(0);
                break;
             case FRAG_RESULT_COLOR:
-               write_all = GL_TRUE; /* fallthrough */
-            default:
-               assert(attr == FRAG_RESULT_COLOR ||
-                      (FRAG_RESULT_DATA0 <= attr && attr < FRAG_RESULT_MAX));
-               fs_output_semantic_name[fs_num_outputs] = TGSI_SEMANTIC_COLOR;
-               fs_output_semantic_index[fs_num_outputs] = numColors;
-               outputMapping[attr] = fs_num_outputs;
-               numColors++;
+               write_all = GL_TRUE;
+               break;
+            case FRAG_RESULT_SAMPLE_MASK:
+               semantic = TGSI_SEMANTIC_SAMPLEMASK;
                break;
             }
 
+            assert(attr == FRAG_RESULT_COLOR ||
+                   attr == FRAG_RESULT_SAMPLE_MASK ||
+                   (FRAG_RESULT_DATA0 <= attr && attr < FRAG_RESULT_MAX));
+            fs_output_semantic_name[fs_num_outputs] = semantic;
+            fs_output_semantic_index[fs_num_outputs] = numColors;
+            outputMapping[attr] = fs_num_outputs;
+            numColors++;
+
             fs_num_outputs++;
          }
       }
diff --git a/mesalib/src/mesa/swrast/s_context.c b/mesalib/src/mesa/swrast/s_context.c
index 07485006c..ad11aa2c1 100644
--- a/mesalib/src/mesa/swrast/s_context.c
+++ b/mesalib/src/mesa/swrast/s_context.c
@@ -73,7 +73,7 @@ _swrast_update_rasterflags( struct gl_context *ctx )
       }
    }
    if (ctx->Color.ColorLogicOpEnabled) rasterMask |= LOGIC_OP_BIT;
-   if (ctx->Texture._EnabledUnits)     rasterMask |= TEXTURE_BIT;
+   if (ctx->Texture._MaxEnabledTexImageUnit >= 0) rasterMask |= TEXTURE_BIT;
    if (   ctx->ViewportArray[0].X < 0
        || ctx->ViewportArray[0].X + ctx->ViewportArray[0].Width > (GLfloat) ctx->DrawBuffer->Width
        || ctx->ViewportArray[0].Y < 0
@@ -286,7 +286,7 @@ _swrast_update_specular_vertex_add(struct gl_context *ctx)
        ctx->Light.Model.ColorControl == GL_SEPARATE_SPECULAR_COLOR);
 
    swrast->SpecularVertexAdd = (separateSpecular
-                                && ctx->Texture._EnabledUnits == 0x0
+                                && ctx->Texture._MaxEnabledTexImageUnit == -1
                                 && !_swrast_use_fragment_program(ctx)
                                 && !ctx->ATIFragmentShader._Enabled);
 }
@@ -523,7 +523,7 @@ _swrast_update_active_attribs(struct gl_context *ctx)
       if (swrast->_FogEnabled)
          attribsMask |= VARYING_BIT_FOGC;
 
-      attribsMask |= (ctx->Texture._EnabledUnits << VARYING_SLOT_TEX0);
+      attribsMask |= (ctx->Texture._EnabledCoordUnits << VARYING_SLOT_TEX0);
    }
 
    swrast->_ActiveAttribMask = attribsMask;
@@ -923,7 +923,7 @@ _swrast_print_vertex( struct gl_context *ctx, const SWvertex *v )
                   v->attrib[VARYING_SLOT_POS][3]);
 
       for (i = 0 ; i < ctx->Const.MaxTextureCoordUnits ; i++)
-	 if (ctx->Texture.Unit[i]._ReallyEnabled)
+	 if (ctx->Texture.Unit[i]._Current)
 	    _mesa_debug(ctx, "texcoord[%d] %f %f %f %f\n", i,
                         v->attrib[VARYING_SLOT_TEX0 + i][0],
                         v->attrib[VARYING_SLOT_TEX0 + i][1],
diff --git a/mesalib/src/mesa/swrast/s_texcombine.c b/mesalib/src/mesa/swrast/s_texcombine.c
index 32d7c0944..c1a152aa8 100644
--- a/mesalib/src/mesa/swrast/s_texcombine.c
+++ b/mesalib/src/mesa/swrast/s_texcombine.c
@@ -189,7 +189,7 @@ texture_combine( struct gl_context *ctx, GLuint unit,
             {
                const GLuint srcUnit = srcRGB - GL_TEXTURE0;
                ASSERT(srcUnit < ctx->Const.MaxTextureUnits);
-               if (!ctx->Texture.Unit[srcUnit]._ReallyEnabled)
+               if (!ctx->Texture.Unit[srcUnit]._Current)
                   goto end;
                argRGB[term] = get_texel_array(swrast, srcUnit);
             }
@@ -279,7 +279,7 @@ texture_combine( struct gl_context *ctx, GLuint unit,
             {
                const GLuint srcUnit = srcA - GL_TEXTURE0;
                ASSERT(srcUnit < ctx->Const.MaxTextureUnits);
-               if (!ctx->Texture.Unit[srcUnit]._ReallyEnabled)
+               if (!ctx->Texture.Unit[srcUnit]._Current)
                   goto end;
                argA[term] = get_texel_array(swrast, srcUnit);
             }
@@ -657,7 +657,7 @@ _swrast_texture_span( struct gl_context *ctx, SWspan *span )
    for (unit = 0; unit < ctx->Const.MaxTextureUnits; unit++) {
       const struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit];
 
-      if (texUnit->_ReallyEnabled &&
+      if (texUnit->_Current &&
          texUnit->_CurrentCombine->ModeRGB == GL_BUMP_ENVMAP_ATI) {
          const GLfloat (*texcoords)[4] = (const GLfloat (*)[4])
             span->array->attribs[VARYING_SLOT_TEX0 + unit];
@@ -723,7 +723,7 @@ _swrast_texture_span( struct gl_context *ctx, SWspan *span )
     */
    for (unit = 0; unit < ctx->Const.MaxTextureUnits; unit++) {
       const struct gl_texture_unit *texUnit = &ctx->Texture.Unit[unit];
-      if (texUnit->_ReallyEnabled &&
+      if (texUnit->_Current &&
           texUnit->_CurrentCombine->ModeRGB != GL_BUMP_ENVMAP_ATI) {
          const GLfloat (*texcoords)[4] = (const GLfloat (*)[4])
             span->array->attribs[VARYING_SLOT_TEX0 + unit];
@@ -787,7 +787,7 @@ _swrast_texture_span( struct gl_context *ctx, SWspan *span )
     * We modify the span->color.rgba values.
     */
    for (unit = 0; unit < ctx->Const.MaxTextureUnits; unit++) {
-      if (ctx->Texture.Unit[unit]._ReallyEnabled)
+      if (ctx->Texture.Unit[unit]._Current)
          texture_combine(ctx, unit, primary_rgba, swrast->TexelBuffer, span);
    }
 
diff --git a/mesalib/src/mesa/swrast/s_texture.c b/mesalib/src/mesa/swrast/s_texture.c
index c08a4e9d1..9ccd0e347 100644
--- a/mesalib/src/mesa/swrast/s_texture.c
+++ b/mesalib/src/mesa/swrast/s_texture.c
@@ -177,8 +177,8 @@ _swrast_free_texture_image_buffer(struct gl_context *ctx,
  * Error checking for debugging only.
  */
 static void
-_mesa_check_map_teximage(struct gl_texture_image *texImage,
-                         GLuint slice, GLuint x, GLuint y, GLuint w, GLuint h)
+check_map_teximage(const struct gl_texture_image *texImage,
+                   GLuint slice, GLuint x, GLuint y, GLuint w, GLuint h)
 {
 
    if (texImage->TexObject->Target == GL_TEXTURE_1D)
@@ -216,7 +216,16 @@ _swrast_map_teximage(struct gl_context *ctx,
    GLint stride, texelSize;
    GLuint bw, bh;
 
-   _mesa_check_map_teximage(texImage, slice, x, y, w, h);
+   check_map_teximage(texImage, slice, x, y, w, h);
+
+   if (!swImage->Buffer) {
+      /* Either glTexImage was called with a NULL <pixels> argument or
+       * we ran out of memory when allocating texture memory,
+       */
+      *mapOut = NULL;
+      *rowStrideOut = 0;
+      return;
+   }
 
    texelSize = _mesa_get_format_bytes(texImage->TexFormat);
    stride = _mesa_format_row_stride(texImage->TexFormat, texImage->Width);
@@ -225,12 +234,6 @@ _swrast_map_teximage(struct gl_context *ctx,
    assert(x % bw == 0);
    assert(y % bh == 0);
 
-   if (!swImage->Buffer) {
-      /* probably ran out of memory when allocating tex mem */
-      *mapOut = NULL;
-      return;
-   }
-
    /* This function can only be used with a swrast-allocated buffer, in which
     * case ImageSlices is populated with pointers into Buffer.
     */
@@ -266,7 +269,7 @@ _swrast_map_texture(struct gl_context *ctx, struct gl_texture_object *texObj)
       for (level = texObj->BaseLevel; level < MAX_TEXTURE_LEVELS; level++) {
          struct gl_texture_image *texImage = texObj->Image[face][level];
          struct swrast_texture_image *swImage = swrast_texture_image(texImage);
-         unsigned int i;
+         unsigned int i, slices;
 
          if (!texImage)
             continue;
@@ -279,7 +282,16 @@ _swrast_map_texture(struct gl_context *ctx, struct gl_texture_object *texObj)
             continue;
          }
 
-         for (i = 0; i < texture_slices(texImage); i++) {
+         if (!swImage->ImageSlices) {
+            swImage->ImageSlices =
+               calloc(texture_slices(texImage), sizeof(void *));
+            if (!swImage->ImageSlices)
+               continue;
+         }
+
+         slices = texture_slices(texImage);
+
+         for (i = 0; i < slices; i++) {
             GLubyte *map;
             GLint rowStride;
 
@@ -317,7 +329,7 @@ _swrast_unmap_texture(struct gl_context *ctx, struct gl_texture_object *texObj)
       for (level = texObj->BaseLevel; level < MAX_TEXTURE_LEVELS; level++) {
          struct gl_texture_image *texImage = texObj->Image[face][level];
          struct swrast_texture_image *swImage = swrast_texture_image(texImage);
-         unsigned int i;
+         unsigned int i, slices;
 
          if (!texImage)
             continue;
@@ -325,7 +337,12 @@ _swrast_unmap_texture(struct gl_context *ctx, struct gl_texture_object *texObj)
          if (swImage->Buffer)
             return;
 
-         for (i = 0; i < texture_slices(texImage); i++) {
+         if (!swImage->ImageSlices)
+            continue;
+
+         slices = texture_slices(texImage);
+
+         for (i = 0; i < slices; i++) {
             if (swImage->ImageSlices[i]) {
                ctx->Driver.UnmapTextureImage(ctx, texImage, i);
                swImage->ImageSlices[i] = NULL;
@@ -342,16 +359,13 @@ _swrast_unmap_texture(struct gl_context *ctx, struct gl_texture_object *texObj)
 void
 _swrast_map_textures(struct gl_context *ctx)
 {
-   GLbitfield enabledUnits = ctx->Texture._EnabledUnits;
+   int unit;
 
-   /* loop over enabled texture units */
-   while (enabledUnits) {
-      GLuint unit = ffs(enabledUnits) - 1;
+   for (unit = 0; unit <= ctx->Texture._MaxEnabledTexImageUnit; unit++) {
       struct gl_texture_object *texObj = ctx->Texture.Unit[unit]._Current;
-      
-      _swrast_map_texture(ctx, texObj);
 
-      enabledUnits &= ~(1 << unit);
+      if (texObj)
+         _swrast_map_texture(ctx, texObj);
    }
 }
 
@@ -362,15 +376,11 @@ _swrast_map_textures(struct gl_context *ctx)
 void
 _swrast_unmap_textures(struct gl_context *ctx)
 {
-   GLbitfield enabledUnits = ctx->Texture._EnabledUnits;
-
-   /* loop over enabled texture units */
-   while (enabledUnits) {
-      GLuint unit = ffs(enabledUnits) - 1;
+   int unit;
+   for (unit = 0; unit <= ctx->Texture._MaxEnabledTexImageUnit; unit++) {
       struct gl_texture_object *texObj = ctx->Texture.Unit[unit]._Current;
-      
-      _swrast_unmap_texture(ctx, texObj);
 
-      enabledUnits &= ~(1 << unit);
+      if (texObj)
+         _swrast_unmap_texture(ctx, texObj);
    }
 }
diff --git a/mesalib/src/mesa/swrast/s_triangle.c b/mesalib/src/mesa/swrast/s_triangle.c
index 00383fb84..164906643 100644
--- a/mesalib/src/mesa/swrast/s_triangle.c
+++ b/mesalib/src/mesa/swrast/s_triangle.c
@@ -1071,8 +1071,8 @@ _swrast_choose_triangle( struct gl_context *ctx )
          if (ctx->Texture._EnabledCoordUnits == 0x1
              && !_swrast_use_fragment_program(ctx)
              && !ctx->ATIFragmentShader._Enabled
-             && ctx->Texture._EnabledUnits == 0x1
-             && ctx->Texture.Unit[0]._ReallyEnabled == TEXTURE_2D_BIT
+             && ctx->Texture._MaxEnabledTexImageUnit == 0
+             && ctx->Texture.Unit[0]._Current->Target == GL_TEXTURE_2D
              && samp->WrapS == GL_REPEAT
              && samp->WrapT == GL_REPEAT
              && texObj2D->_Swizzle == SWIZZLE_NOOP
diff --git a/mesalib/src/mesa/swrast_setup/ss_context.c b/mesalib/src/mesa/swrast_setup/ss_context.c
index 12a47358f..0b3b9e4df 100644
--- a/mesalib/src/mesa/swrast_setup/ss_context.c
+++ b/mesalib/src/mesa/swrast_setup/ss_context.c
@@ -287,7 +287,8 @@ _swsetup_Translate( struct gl_context *ctx, const void *vertex, SWvertex *dest )
 
    _tnl_get_attr( ctx, vertex, _TNL_ATTRIB_COLOR0,
                   dest->attrib[VARYING_SLOT_COL0] );
-   UNCLAMPED_FLOAT_TO_RGBA_CHAN( dest->color, tmp );
+
+   UNCLAMPED_FLOAT_TO_RGBA_CHAN(dest->color, dest->attrib[VARYING_SLOT_COL0]);
 
    _tnl_get_attr( ctx, vertex, _TNL_ATTRIB_COLOR1,
                   dest->attrib[VARYING_SLOT_COL1]);