i965: Implement threaded GL support.

Now i965 supports mesa_glthread=true like Gallium drivers do.

According to Markus (degasus), the Citra emulator now runs ~30% faster.
Emmanuel (linkmauve) also reported that the Dolphin emulator improved
by 2.8x on one game.  (Both of those still need to be added to drirc.)

An Intel Mesa CI run with mesa_glthread=true appears to be happy.

Bioshock Infinite's benchmark mode seems to be around 15-20% faster
on my Skylake GT4 at 1920x1080.

Tested-by: Markus Wick <markus@selfnet.de>
Tested-by: Emmanuel Gil Peyrot <linkmauve@linkmauve.fr>
Tested-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>

src/mesa/drivers/dri/i965/brw_context.c

@@ -36,6 +36,7 @@
 #include "main/context.h"
 #include "main/fbobject.h"
 #include "main/extensions.h"
+#include "main/glthread.h"
 #include "main/imports.h"
 #include "main/macros.h"
 #include "main/points.h"
@@ -147,6 +148,24 @@ intel_get_string(struct gl_context * ctx, GLenum name)
    }
 }
 
+static void
+brw_set_background_context(struct gl_context *ctx,
+                           struct util_queue_monitoring *queue_info)
+{
+   struct brw_context *brw = brw_context(ctx);
+   __DRIcontext *driContext = brw->driContext;
+   __DRIscreen *driScreen = driContext->driScreenPriv;
+   const __DRIbackgroundCallableExtension *backgroundCallable =
+      driScreen->dri2.backgroundCallable;
+
+   /* Note: Mesa will only call this function if we've called
+    * _mesa_enable_multithreading().  We only do that if the loader exposed
+    * the __DRI_BACKGROUND_CALLABLE extension.  So we know that
+    * backgroundCallable is not NULL.
+    */
+   backgroundCallable->setBackgroundContext(driContext->loaderPrivate);
+}
+
 static void
 intel_viewport(struct gl_context *ctx)
 {
@@ -376,6 +395,8 @@ brw_init_driver_functions(struct brw_context *brw,
    if (brw->screen->disk_cache) {
       functions->ShaderCacheSerializeDriverBlob = brw_program_serialize_nir;
    }
+
+   functions->SetBackgroundContext = brw_set_background_context;
 }
 
 static void
@@ -1126,6 +1147,12 @@ brwCreateContext(gl_api api,
 
    brw->ctx.Cache = brw->screen->disk_cache;
 
+   if (driContextPriv->driScreenPriv->dri2.backgroundCallable &&
+       driQueryOptionb(&screen->optionCache, "mesa_glthread")) {
+      /* Loader supports multithreading, and so do we. */
+      _mesa_glthread_init(ctx);
+   }
+
    return true;
 }
 
@@ -1136,6 +1163,18 @@ intelDestroyContext(__DRIcontext * driContextPriv)
       (struct brw_context *) driContextPriv->driverPrivate;
    struct gl_context *ctx = &brw->ctx;
 
+   GET_CURRENT_CONTEXT(curctx);
+
+   if (curctx == NULL) {
+      /* No current context, but we need one to release
+       * renderbuffer surface when we release framebuffer.
+       * So temporarily bind the context.
+       */
+      _mesa_make_current(ctx, NULL, NULL);
+   }
+
+   _mesa_glthread_destroy(&brw->ctx);
+
    _mesa_meta_free(&brw->ctx);
 
    if (INTEL_DEBUG & DEBUG_SHADER_TIME) {
@@ -1196,6 +1235,9 @@ intelDestroyContext(__DRIcontext * driContextPriv)
 GLboolean
 intelUnbindContext(__DRIcontext * driContextPriv)
 {
+   GET_CURRENT_CONTEXT(ctx);
+   _mesa_glthread_finish(ctx);
+
    /* Unset current context and dispath table */
    _mesa_make_current(NULL, NULL, NULL);
 
@@ -1299,6 +1341,8 @@ intelMakeCurrent(__DRIcontext * driContextPriv,
 
       _mesa_make_current(ctx, fb, readFb);
    } else {
+      GET_CURRENT_CONTEXT(ctx);
+      _mesa_glthread_finish(ctx);
       _mesa_make_current(NULL, NULL, NULL);
    }
 

src/mesa/drivers/dri/i965/intel_screen.c

@@ -34,6 +34,7 @@
 #include "main/hash.h"
 #include "main/fbobject.h"
 #include "main/version.h"
+#include "main/glthread.h"
 #include "swrast/s_renderbuffer.h"
 #include "util/ralloc.h"
 #include "util/disk_cache.h"
@@ -62,6 +63,7 @@ DRI_CONF_BEGIN
 	 DRI_CONF_DESC_END
       DRI_CONF_OPT_END
       DRI_CONF_MESA_NO_ERROR("false")
+      DRI_CONF_MESA_GLTHREAD("false")
    DRI_CONF_SECTION_END
 
    DRI_CONF_SECTION_QUALITY
@@ -148,6 +150,8 @@ intel_dri2_flush_with_flags(__DRIcontext *cPriv,
 
    struct gl_context *ctx = &brw->ctx;
 
+   _mesa_glthread_finish(ctx);
+
    FLUSH_VERTICES(ctx, 0);
 
    if (flags & __DRI2_FLUSH_DRAWABLE)

src/mesa/drivers/dri/i965/intel_tex_image.c

@@ -14,6 +14,7 @@
 #include "main/texobj.h"
 #include "main/teximage.h"
 #include "main/texstore.h"
+#include "main/glthread.h"
 
 #include "drivers/common/meta.h"
 
@@ -442,6 +443,8 @@ intelSetTexBuffer2(__DRIcontext *pDRICtx, GLint target,
    mesa_format texFormat = MESA_FORMAT_NONE;
    GLenum internal_format = 0;
 
+   _mesa_glthread_finish(ctx);
+
    texObj = _mesa_get_current_tex_object(ctx, target);
 
    if (!texObj)