mesa: Add infrastructure for a worker thread to process GL commands.

v2: Keep an allocated buffer around instead of checking for one at the
    start of every GL command.  Inline the now-small space allocation
    function.
v3: Remove duplicate !glthread->shutdown check, process remaining work
    before shutdown.
v4: Fix leaks on destroy.
V5: (Timothy Arceri) fix order of source files in makefile

Acked-by: Timothy Arceri <[email protected]>
Acked-by: Marek Olšák <[email protected]>
Tested-by: Dieter Nützel <[email protected]>
Tested-by: Mike Lothian <[email protected]>

1 files changed, 228 insertions, 0 deletions

diff --git a/src/mesa/main/glthread.c b/src/mesa/main/glthread.c
new file mode 100644
index 00000000000..76eb0cf7dac
--- /dev/null
+++ b/src/mesa/main/glthread.c
@@ -0,0 +1,228 @@
+/*
+ * Copyright © 2012 Intel Corporation
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice (including the next
+ * paragraph) shall be included in all copies or substantial portions of the
+ * Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
+ * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+ * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
+ * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
+ * IN THE SOFTWARE.
+ */
+
+/** @file glthread.c
+ *
+ * Support functions for the glthread feature of Mesa.
+ *
+ * In multicore systems, many applications end up CPU-bound with about half
+ * their time spent inside their rendering thread and half inside Mesa.  To
+ * alleviate this, we put a shim layer in Mesa at the GL dispatch level that
+ * quickly logs the GL commands to a buffer to be processed by a worker
+ * thread.
+ */
+
+#include "main/mtypes.h"
+#include "main/glthread.h"
+#include "main/marshal.h"
+#include "main/marshal_generated.h"
+
+#ifdef HAVE_PTHREAD
+
+static void
+glthread_allocate_batch(struct gl_context *ctx)
+{
+   struct glthread_state *glthread = ctx->GLThread;
+
+   /* TODO: handle memory allocation failure. */
+   glthread->batch = calloc(1, sizeof(*glthread->batch));
+   if (!glthread->batch)
+      return;
+   glthread->batch->buffer = malloc(MARSHAL_MAX_CMD_SIZE);
+}
+
+static void
+glthread_unmarshal_batch(struct gl_context *ctx, struct glthread_batch *batch)
+{
+   free(batch->buffer);
+   free(batch);
+}
+
+static void *
+glthread_worker(void *data)
+{
+   struct gl_context *ctx = data;
+   struct glthread_state *glthread = ctx->GLThread;
+
+   ctx->Driver.SetBackgroundContext(ctx);
+   _glapi_set_context(ctx);
+
+   pthread_mutex_lock(&glthread->mutex);
+
+   while (true) {
+      struct glthread_batch *batch;
+
+      /* Block (dropping the lock) until new work arrives for us. */
+      while (!glthread->batch_queue && !glthread->shutdown) {
+         pthread_cond_broadcast(&glthread->work_done);
+         pthread_cond_wait(&glthread->new_work, &glthread->mutex);
+      }
+
+      batch = glthread->batch_queue;
+
+      if (glthread->shutdown && !batch) {
+         pthread_cond_broadcast(&glthread->work_done);
+         pthread_mutex_unlock(&glthread->mutex);
+         return NULL;
+      }
+      glthread->batch_queue = batch->next;
+      if (glthread->batch_queue_tail == &batch->next)
+         glthread->batch_queue_tail = &glthread->batch_queue;
+
+      glthread->busy = true;
+      pthread_mutex_unlock(&glthread->mutex);
+
+      glthread_unmarshal_batch(ctx, batch);
+
+      pthread_mutex_lock(&glthread->mutex);
+      glthread->busy = false;
+   }
+
+   /* UNREACHED */
+   return NULL;
+}
+
+void
+_mesa_glthread_init(struct gl_context *ctx)
+{
+   struct glthread_state *glthread = calloc(1, sizeof(*glthread));
+
+   if (!glthread)
+      return;
+
+   pthread_mutex_init(&glthread->mutex, NULL);
+   pthread_cond_init(&glthread->new_work, NULL);
+   pthread_cond_init(&glthread->work_done, NULL);
+
+   glthread->batch_queue_tail = &glthread->batch_queue;
+   ctx->GLThread = glthread;
+
+   glthread_allocate_batch(ctx);
+
+   pthread_create(&glthread->thread, NULL, glthread_worker, ctx);
+}
+
+void
+_mesa_glthread_destroy(struct gl_context *ctx)
+{
+   struct glthread_state *glthread = ctx->GLThread;
+
+   if (!glthread)
+      return;
+
+   _mesa_glthread_flush_batch(ctx);
+
+   pthread_mutex_lock(&glthread->mutex);
+   glthread->shutdown = true;
+   pthread_cond_broadcast(&glthread->new_work);
+   pthread_mutex_unlock(&glthread->mutex);
+
+   /* Since this waits for the thread to exit, it means that all queued work
+    * will have been completed.
+    */
+   pthread_join(glthread->thread, NULL);
+
+   pthread_cond_destroy(&glthread->new_work);
+   pthread_cond_destroy(&glthread->work_done);
+   pthread_mutex_destroy(&glthread->mutex);
+
+   /* Due to the join above, there should be one empty batch allocated at this
+    * point, and no batches queued.
+    */
+   assert(!glthread->batch->used);
+   assert(!glthread->batch->next);
+   free(glthread->batch);
+   assert(!glthread->batch_queue);
+
+   free(glthread);
+   ctx->GLThread = NULL;
+}
+
+void
+_mesa_glthread_flush_batch(struct gl_context *ctx)
+{
+   struct glthread_state *glthread = ctx->GLThread;
+   struct glthread_batch *batch;
+
+   if (!glthread)
+      return;
+
+   batch = glthread->batch;
+   if (!batch->used)
+      return;
+
+   /* Immediately reallocate a new batch, since the next marshalled call would
+    * just do it.
+    */
+   glthread_allocate_batch(ctx);
+
+   /* Debug: execute the batch immediately from this thread.
+    *
+    * Note that glthread_unmarshal_batch() changes the dispatch table so we'll
+    * need to restore it when it returns.
+    */
+   if (false) {
+      glthread_unmarshal_batch(ctx, batch);
+      return;
+   }
+
+   pthread_mutex_lock(&glthread->mutex);
+   *glthread->batch_queue_tail = batch;
+   glthread->batch_queue_tail = &batch->next;
+   pthread_cond_broadcast(&glthread->new_work);
+   pthread_mutex_unlock(&glthread->mutex);
+}
+
+/**
+ * Waits for all pending batches have been unmarshaled.
+ *
+ * This can be used by the main thread to synchronize access to the context,
+ * since the worker thread will be idle after this.
+ */
+void
+_mesa_glthread_finish(struct gl_context *ctx)
+{
+   struct glthread_state *glthread = ctx->GLThread;
+
+   if (!glthread)
+      return;
+
+   /* If this is called from the worker thread, then we've hit a path that
+    * might be called from either the main thread or the worker (such as some
+    * dri interface entrypoints), in which case we don't need to actually
+    * synchronize against ourself.
+    */
+   if (pthread_self() == glthread->thread)
+      return;
+
+   _mesa_glthread_flush_batch(ctx);
+
+   pthread_mutex_lock(&glthread->mutex);
+
+   while (glthread->batch_queue || glthread->busy)
+      pthread_cond_wait(&glthread->work_done, &glthread->mutex);
+
+   pthread_mutex_unlock(&glthread->mutex);
+}
+
+#endif