r600g/radeonsi: Use write-combined CPU mappings of some BOs in GTT

Reviewed-by: Marek Olšák <[email protected]>

6 files changed, 43 insertions, 6 deletions

diff --git a/src/gallium/winsys/radeon/drm/radeon_drm_bo.c b/src/gallium/winsys/radeon/drm/radeon_drm_bo.c
index d06bb344c73..73f8d385d7b 100644
--- a/src/gallium/winsys/radeon/drm/radeon_drm_bo.c
+++ b/src/gallium/winsys/radeon/drm/radeon_drm_bo.c
@@ -477,6 +477,10 @@ const struct pb_vtbl radeon_bo_vtbl = {
     radeon_bo_get_base_buffer,
 };
 
+#ifndef RADEON_GEM_GTT_WC
+#define RADEON_GEM_GTT_WC (1 << 2)
+#endif
+
 static struct pb_buffer *radeon_bomgr_create_bo(struct pb_manager *_mgr,
                                                 pb_size size,
                                                 const struct pb_desc *desc)
@@ -497,6 +501,10 @@ static struct pb_buffer *radeon_bomgr_create_bo(struct pb_manager *_mgr,
     args.size = size;
     args.alignment = desc->alignment;
     args.initial_domain = rdesc->initial_domains;
+    args.flags = 0;
+
+    if (rdesc->flags & RADEON_FLAG_GTT_WC)
+        args.flags |= RADEON_GEM_GTT_WC;
 
     if (drmCommandWriteRead(rws->fd, DRM_RADEON_GEM_CREATE,
                             &args, sizeof(args))) {
@@ -504,6 +512,7 @@ static struct pb_buffer *radeon_bomgr_create_bo(struct pb_manager *_mgr,
         fprintf(stderr, "radeon:    size      : %d bytes\n", size);
         fprintf(stderr, "radeon:    alignment : %d bytes\n", desc->alignment);
         fprintf(stderr, "radeon:    domains   : %d\n", args.initial_domain);
+        fprintf(stderr, "radeon:    flags     : %d\n", args.flags);
         return NULL;
     }
 
@@ -784,7 +793,8 @@ radeon_winsys_bo_create(struct radeon_winsys *rws,
                         unsigned size,
                         unsigned alignment,
                         boolean use_reusable_pool,
-                        enum radeon_bo_domain domain)
+                        enum radeon_bo_domain domain,
+                        enum radeon_bo_flag flags)
 {
     struct radeon_drm_winsys *ws = radeon_drm_winsys(rws);
     struct radeon_bomgr *mgr = radeon_bomgr(ws->kman);
@@ -798,13 +808,20 @@ radeon_winsys_bo_create(struct radeon_winsys *rws,
     /* Additional criteria for the cache manager. */
     desc.base.usage = domain;
     desc.initial_domains = domain;
+    desc.flags = flags;
 
     /* Assign a buffer manager. */
     if (use_reusable_pool) {
-        if (domain == RADEON_DOMAIN_VRAM)
-            provider = ws->cman_vram;
-        else
+        if (domain == RADEON_DOMAIN_VRAM) {
+            if (flags & RADEON_FLAG_GTT_WC)
+                provider = ws->cman_vram_gtt_wc;
+            else
+                provider = ws->cman_vram;
+        } else if (flags & RADEON_FLAG_GTT_WC) {
+            provider = ws->cman_gtt_wc;
+        } else {
             provider = ws->cman_gtt;
+        }
     } else {
         provider = ws->kman;
     }
diff --git a/src/gallium/winsys/radeon/drm/radeon_drm_bo.h b/src/gallium/winsys/radeon/drm/radeon_drm_bo.h
index f5b122f17c7..1c00a13c1e0 100644
--- a/src/gallium/winsys/radeon/drm/radeon_drm_bo.h
+++ b/src/gallium/winsys/radeon/drm/radeon_drm_bo.h
@@ -42,6 +42,7 @@ struct radeon_bo_desc {
     struct pb_desc base;
 
     unsigned initial_domains;
+    unsigned flags;
 };
 
 struct radeon_bo {
diff --git a/src/gallium/winsys/radeon/drm/radeon_drm_cs.c b/src/gallium/winsys/radeon/drm/radeon_drm_cs.c
index 67375dccc1d..3596f8d1cd0 100644
--- a/src/gallium/winsys/radeon/drm/radeon_drm_cs.c
+++ b/src/gallium/winsys/radeon/drm/radeon_drm_cs.c
@@ -606,7 +606,7 @@ radeon_cs_create_fence(struct radeon_winsys_cs *rcs)
 
     /* Create a fence, which is a dummy BO. */
     fence = cs->ws->base.buffer_create(&cs->ws->base, 1, 1, TRUE,
-                                       RADEON_DOMAIN_GTT);
+                                       RADEON_DOMAIN_GTT, 0);
     /* Add the fence as a dummy relocation. */
     cs->ws->base.cs_add_reloc(rcs, cs->ws->base.buffer_get_cs_handle(fence),
                               RADEON_USAGE_READWRITE, RADEON_DOMAIN_GTT,
diff --git a/src/gallium/winsys/radeon/drm/radeon_drm_winsys.c b/src/gallium/winsys/radeon/drm/radeon_drm_winsys.c
index 9e60de3ab77..910d06b5dba 100644
--- a/src/gallium/winsys/radeon/drm/radeon_drm_winsys.c
+++ b/src/gallium/winsys/radeon/drm/radeon_drm_winsys.c
@@ -424,7 +424,9 @@ static void radeon_winsys_destroy(struct radeon_winsys *rws)
     pipe_mutex_destroy(ws->cs_stack_lock);
 
     ws->cman_vram->destroy(ws->cman_vram);
+    ws->cman_vram_gtt_wc->destroy(ws->cman_vram_gtt_wc);
     ws->cman_gtt->destroy(ws->cman_gtt);
+    ws->cman_gtt_wc->destroy(ws->cman_gtt_wc);
     ws->kman->destroy(ws->kman);
     if (ws->gen >= DRV_R600) {
         radeon_surface_manager_free(ws->surf_man);
@@ -642,9 +644,15 @@ radeon_drm_winsys_create(int fd, radeon_screen_create_t screen_create)
     ws->cman_vram = pb_cache_manager_create(ws->kman, 1000000, 2.0f, 0);
     if (!ws->cman_vram)
         goto fail;
+    ws->cman_vram_gtt_wc = pb_cache_manager_create(ws->kman, 1000000, 2.0f, 0);
+    if (!ws->cman_vram_gtt_wc)
+        goto fail;
     ws->cman_gtt = pb_cache_manager_create(ws->kman, 1000000, 2.0f, 0);
     if (!ws->cman_gtt)
         goto fail;
+    ws->cman_gtt_wc = pb_cache_manager_create(ws->kman, 1000000, 2.0f, 0);
+    if (!ws->cman_gtt_wc)
+        goto fail;
 
     if (ws->gen >= DRV_R600) {
         ws->surf_man = radeon_surface_manager_new(fd);
@@ -701,8 +709,12 @@ fail:
     pipe_mutex_unlock(fd_tab_mutex);
     if (ws->cman_gtt)
         ws->cman_gtt->destroy(ws->cman_gtt);
+    if (ws->cman_gtt_wc)
+        ws->cman_gtt_wc->destroy(ws->cman_gtt_wc);
     if (ws->cman_vram)
         ws->cman_vram->destroy(ws->cman_vram);
+    if (ws->cman_vram_gtt_wc)
+        ws->cman_vram_gtt_wc->destroy(ws->cman_vram_gtt_wc);
     if (ws->kman)
         ws->kman->destroy(ws->kman);
     if (ws->surf_man)
diff --git a/src/gallium/winsys/radeon/drm/radeon_drm_winsys.h b/src/gallium/winsys/radeon/drm/radeon_drm_winsys.h
index fc6f53b7231..ea6f7f01f14 100644
--- a/src/gallium/winsys/radeon/drm/radeon_drm_winsys.h
+++ b/src/gallium/winsys/radeon/drm/radeon_drm_winsys.h
@@ -58,7 +58,9 @@ struct radeon_drm_winsys {
 
     struct pb_manager *kman;
     struct pb_manager *cman_vram;
+    struct pb_manager *cman_vram_gtt_wc;
     struct pb_manager *cman_gtt;
+    struct pb_manager *cman_gtt_wc;
     struct radeon_surface_manager *surf_man;
 
     uint32_t num_cpus;      /* Number of CPUs. */
diff --git a/src/gallium/winsys/radeon/drm/radeon_winsys.h b/src/gallium/winsys/radeon/drm/radeon_winsys.h
index a63a50b01e8..9aea1e89f06 100644
--- a/src/gallium/winsys/radeon/drm/radeon_winsys.h
+++ b/src/gallium/winsys/radeon/drm/radeon_winsys.h
@@ -65,6 +65,10 @@ enum radeon_bo_domain { /* bitfield */
     RADEON_DOMAIN_VRAM_GTT = RADEON_DOMAIN_VRAM | RADEON_DOMAIN_GTT
 };
 
+enum radeon_bo_flag { /* bitfield */
+   RADEON_FLAG_GTT_WC = (1 << 0)
+};
+
 enum radeon_bo_usage { /* bitfield */
     RADEON_USAGE_READ = 2,
     RADEON_USAGE_WRITE = 4,
@@ -287,7 +291,8 @@ struct radeon_winsys {
                                        unsigned size,
                                        unsigned alignment,
                                        boolean use_reusable_pool,
-                                       enum radeon_bo_domain domain);
+                                       enum radeon_bo_domain domain,
+                                       enum radeon_bo_flag flags);
 
     struct radeon_winsys_cs_handle *(*buffer_get_cs_handle)(
             struct pb_buffer *buf);