On 12/13/2017 05:27 PM, Marek Olšák wrote:
Mario, can we push these patches?

Marek


Sorry for the late response and thanks for the reviews! Was a bit sick the last days, so couldn't think or do anything about the recent comments so far.

I think all patches have a r-b by you and/or Tapani, except the nouveau patch 20/22 on which Ilia commented. Not sure about that one. If we leave it out then stuff breaks on nouveau. If we leave it in, then as far as i understand Ilia's comments, we have a slight violation of spec atm. - If a app asks for a framebuffer without alpha channel (xrgb2101010), but then enables blending with DST_ALPHA it would potentially get random 2 x-bit trash instead of a well defined 1.0 DST_ALPHA. Otoh. it would be legal but a bit non-sensical for an app to ask for no alpha channel, but then choose blend functions that only make sense with an alpha channel?

The third option would be to replace that patch with one that disables the rgb10 support for visuals on nouveau completely until we have a solution for properly emulating xrgb201010 on nouveau to cover such corner cases?

-mario


On Wed, Nov 29, 2017 at 5:21 AM, Mario Kleiner
<mario.kleiner...@gmail.com> wrote:
Some clients may not like rgb10 fbconfigs and visuals.
Support driconf option 'allow_rgb10_configs' on gallium
to allow per application enable/disable.

The option defaults to enabled.

Signed-off-by: Mario Kleiner <mario.kleiner...@gmail.com>
---
  src/gallium/auxiliary/pipe-loader/driinfo_gallium.h | 1 +
  src/gallium/state_trackers/dri/dri_screen.c         | 8 ++++++++
  2 files changed, 9 insertions(+)

diff --git a/src/gallium/auxiliary/pipe-loader/driinfo_gallium.h 
b/src/gallium/auxiliary/pipe-loader/driinfo_gallium.h
index d2d2c9d..db0d633 100644
--- a/src/gallium/auxiliary/pipe-loader/driinfo_gallium.h
+++ b/src/gallium/auxiliary/pipe-loader/driinfo_gallium.h
@@ -31,4 +31,5 @@ DRI_CONF_SECTION_END
  DRI_CONF_SECTION_MISCELLANEOUS
     DRI_CONF_ALWAYS_HAVE_DEPTH_BUFFER("false")
     DRI_CONF_GLSL_ZERO_INIT("false")
+   DRI_CONF_ALLOW_RGB10_CONFIGS("true")
  DRI_CONF_SECTION_END
diff --git a/src/gallium/state_trackers/dri/dri_screen.c 
b/src/gallium/state_trackers/dri/dri_screen.c
index 04afe71..d307b4f 100644
--- a/src/gallium/state_trackers/dri/dri_screen.c
+++ b/src/gallium/state_trackers/dri/dri_screen.c
@@ -156,6 +156,7 @@ dri_fill_in_modes(struct dri_screen *screen)
     struct pipe_screen *p_screen = screen->base.screen;
     boolean pf_z16, pf_x8z24, pf_z24x8, pf_s8z24, pf_z24s8, pf_z32;
     boolean mixed_color_depth;
+   boolean allow_rgb10;

     static const GLenum back_buffer_modes[] = {
        __DRI_ATTRIB_SWAP_NONE, __DRI_ATTRIB_SWAP_UNDEFINED,
@@ -172,6 +173,8 @@ dri_fill_in_modes(struct dri_screen *screen)
        depth_buffer_factor = 1;
     }

+   allow_rgb10 = driQueryOptionb(&screen->dev->option_cache, 
"allow_rgb10_configs");
+
     msaa_samples_max = (screen->st_api->feature_mask & 
ST_API_FEATURE_MS_VISUALS_MASK)
        ? MSAA_VISUAL_MAX_SAMPLES : 1;

@@ -231,6 +234,11 @@ dri_fill_in_modes(struct dri_screen *screen)
        unsigned num_msaa_modes = 0; /* includes a single-sample mode */
        uint8_t msaa_modes[MSAA_VISUAL_MAX_SAMPLES];

+      if (!allow_rgb10 &&
+          (mesa_formats[format] == MESA_FORMAT_B10G10R10A2_UNORM ||
+           mesa_formats[format] == MESA_FORMAT_B10G10R10X2_UNORM))
+         continue;
+
        if (!p_screen->is_format_supported(p_screen, pipe_formats[format],
                                           PIPE_TEXTURE_2D, 0,
                                           PIPE_BIND_RENDER_TARGET))
--
2.7.4

_______________________________________________
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev
_______________________________________________
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev

Reply via email to