[PATCH v3 0/1] MR970: wined3d: Do not enforce map access for resources with WINED3D_RESOURCE_ACCESS_CPU.

[Zebediah Figura (＠zfigura)]

--
  v3: wined3d: Do not enforce GL map access for resources with WINED3D_RESOURCE_ACCESS_CPU.

https://gitlab.winehq.org/wine/wine/-/merge_requests/970

[PATCH v3 1/1] wined3d: Do not enforce GL map access for resources with WINED3D_RESOURCE_ACCESS_CPU.

[Zebediah Figura]

From: Zebediah Figura

d3d maps of such resources will map the CPU copy, and uploads and downloads use
glBufferSubData() and glGetBufferSubData() respectively. There is no need to map
the BO.

This improves performance of Indivisible on NVidia GPUs. The game uses a d3d9
MANAGED buffer for streaming vertex data, which results in poor performance on
NVidia when using GL_MAP_READ_BIT | GL_MAP_PERSISTENT_BIT. With this change we
use neither.

This change should only affect managed resources, i.e. those with both CPU and
GPU access. We never create a BO for CPU-only resources.
---
 dlls/wined3d/resource.c | 11 +++++++----
 1 file changed, 7 insertions(+), 4 deletions(-)

diff --git a/dlls/wined3d/resource.c b/dlls/wined3d/resource.c
index 08edce3630c..c32c9580ae9 100644
--- a/dlls/wined3d/resource.c
+++ b/dlls/wined3d/resource.c
@@ -379,10 +379,13 @@ GLbitfield wined3d_resource_gl_storage_flags(const struct wined3d_resource *reso

if (resource->usage & WINED3DUSAGE_DYNAMIC)
         flags |= GL_CLIENT_STORAGE_BIT;
-    if (access & WINED3D_RESOURCE_ACCESS_MAP_W)
-        flags |= GL_MAP_WRITE_BIT;
-    if (access & WINED3D_RESOURCE_ACCESS_MAP_R)
-        flags |= GL_MAP_READ_BIT;
+    if (!(access & WINED3D_RESOURCE_ACCESS_CPU))
+    {
+        if (access & WINED3D_RESOURCE_ACCESS_MAP_W)
+            flags |= GL_MAP_WRITE_BIT;
+        if (access & WINED3D_RESOURCE_ACCESS_MAP_R)
+            flags |= GL_MAP_READ_BIT;
+    }

return flags;
 }

[PATCH v3 1/1] wined3d: Do not enforce GL map access for resources with WINED3D_RESOURCE_ACCESS_CPU.

[Marvin]

Hi,

It looks like your patch introduced the new failures shown below.
Please investigate and fix them before resubmitting your patch.
If they are not new, fixing them anyway would help a lot. Otherwise
please ask for the known failures list to be updated.

The tests also ran into some preexisting test failures. If you know how
to fix them that would be helpful. See the TestBot job for the details:

The full results can be found at:
https://testbot.winehq.org/JobDetails.pl?Key=125156

Your paranoid android.

=== debian11 (32 bit report) ===

ddraw:
ddraw7.c:15663: Test failed: Expected unsynchronised map for flags 0x1000.
ddraw7.c:15663: Test failed: Expected unsynchronised map for flags 0x3000.

=== debian11 (build log) ===

Use of uninitialized value $Flaky in addition (+) at /home/testbot/lib/WineTestBot/LogUtils.pm line 720, <$LogFile> line 24729.
Use of uninitialized value $Flaky in addition (+) at /home/testbot/lib/WineTestBot/LogUtils.pm line 720, <$LogFile> line 24729.
Use of uninitialized value $Flaky in addition (+) at /home/testbot/lib/WineTestBot/LogUtils.pm line 720, <$LogFile> line 24729.

[Zebediah Figura (＠zfigura)]

On Wed Oct 19 09:39:34 2022 +0000, Jan Sikorski wrote:
> Right, maybe it's fine, but to me it's not immediately clear what are
> the consequences for D3D11. It would be helpful to explain it in the
> commit message.

I've pushed a new version that clarifies that this should only affect managed resources.