Re: [PATCH] wined3d: correctly guess_card_vendor() for gallium r600 drivers

[Austin English]

On Sun, Nov 7, 2010 at 2:08 PM, Brian Paterni  wrote:
> Hello, I've been receiving a couple "unrecognized GL_VENDOR" fixme's lately and
> decided to poke around wine's source. I've found that wined3d_guess_card_vendor
> does not return the correct enum for people running mesa's r600g driver, and now
> that r600g returns "AMD <gpu chip>" as part of openGL's renderer string, we can
> easily use that to check for AMD (ATI) hardware:
>
> dlls/wined3d/directx.c |    5 ++---
>  1 files changed, 2 insertions(+), 3 deletions(-)
>
> diff --git a/dlls/wined3d/directx.c b/dlls/wined3d/directx.c
> index 32d2431..3908c5f 100644
> --- a/dlls/wined3d/directx.c
> +++ b/dlls/wined3d/directx.c
> @@ -1373,11 +1373,10 @@ static enum wined3d_pci_vendor wined3d_guess_card_vendor(const char *gl_vendor_s
>     if (strstr(gl_vendor_string, "ATI")
>             || strstr(gl_vendor_string, "Advanced Micro Devices, Inc.")
>             || strstr(gl_vendor_string, "X.Org R300 Project")
> - || (strstr(gl_vendor_string, "X.Org") && strstr(gl_renderer, "AMD"))
>
> || strstr(gl_renderer, "R100")
>             || strstr(gl_renderer, "R200")
> - || strstr(gl_renderer, "R300")
> - || strstr(gl_renderer, "R600")
> - || strstr(gl_renderer, "R700"))
> - || strstr(gl_renderer, "R300"))
>
> return HW_VENDOR_ATI;
> if (strstr(gl_vendor_string, "Intel(R)")
>
> 1.7.2.3

Howdy Brian,

You're mixing tabs and spaces. Please use consistent spacing, as the
rest of the file does.

[PATCH] wined3d: correctly guess_card_vendor() for gallium r600 drivers

[Brian Paterni]

On Sun, Nov 07, 2010 at 06:48:22PM -0800, Austin English wrote:
> Howdy Brian,
>
> You're mixing tabs and spaces. Please use consistent spacing, as the
> rest of the file does.

Indeed, I was wondering about that... Also, I realized after submitting my
previous email that checking for "X.Org" in gl_vendor_string is probably
unnecessary as well, so I removed that portion from the following:

---
 dlls/wined3d/directx.c |    5 ++---
 1 files changed, 2 insertions(+), 3 deletions(-)

diff --git a/dlls/wined3d/directx.c b/dlls/wined3d/directx.c
index 32d2431..fa5b5cb 100644
--- a/dlls/wined3d/directx.c
+++ b/dlls/wined3d/directx.c
@@ -1373,11 +1373,10 @@ static enum wined3d_pci_vendor wined3d_guess_card_vendor(const char *gl_vendor_s
     if (strstr(gl_vendor_string, "ATI")
             || strstr(gl_vendor_string, "Advanced Micro Devices, Inc.")
             || strstr(gl_vendor_string, "X.Org R300 Project")
+            || strstr(gl_renderer, "AMD")
             || strstr(gl_renderer, "R100")
             || strstr(gl_renderer, "R200")
-            || strstr(gl_renderer, "R300")
-            || strstr(gl_renderer, "R600")
-            || strstr(gl_renderer, "R700"))
+            || strstr(gl_renderer, "R300"))
         return HW_VENDOR_ATI;

if (strstr(gl_vendor_string, "Intel(R)")

[PATCH] wined3d: correctly guess_card_vendor() for gallium r600 drivers

[Henri Verbeet]

On 8 November 2010 05:52, Brian Paterni  wrote:
> Indeed, I was wondering about that... Also, I realized after submitting my
> previous email that checking for "X.Org" in gl_vendor_string is probably
> unnecessary as well, so I removed that portion from the following:

Couple of points:
- Killing the "R600" and "R700" matches is going to kill detection on
older versions of the driver. It may make sense to do that as a
cleanup at some point, especially considering that r600g is a
relatively new driver, but for the moment I think it's still a bit too
early.
- Patches aren't picked up from wine-devel, just wine-patches.
- I think it should be obvious that I wrote a wined3d patch for
testing before pushing 9f064116458ff68b92f179a19be4615f003eaa3e.