On 2025-10-16 18:39, Jani Nikula wrote:
On Thu, 16 Oct 2025, Yaroslav Bolyukin <[email protected]> wrote:
VESA vendor header from DisplayID spec may contain fixed bit per pixel
rate, it should be respected by drm driver
Signed-off-by: Yaroslav Bolyukin <[email protected]>
---
drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c | 14 +++++++++++++-
1 file changed, 13 insertions(+), 1 deletion(-)
diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
index 0d03e324d5b9..ebe5bb4eecf8 100644
--- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
+++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c
@@ -6521,6 +6521,11 @@ static void fill_stream_properties_from_drm_display_mode(
stream->output_color_space = get_output_color_space(timing_out, connector_state);
stream->content_type = get_output_content_type(connector_state);
+
+ /* DisplayID Type VII pass-through timings. */
+ if (mode_in->dsc_passthrough_timings_support && info->dp_dsc_bpp != 0) {
+ stream->timing.dsc_fixed_bits_per_pixel_x16 = info->dp_dsc_bpp;
+ }
If we had mode->dp_dsc_bpp_x16 directly, or something better named, this
would be simpler. This will eventually be replicated in a bunch of
drivers.
BR,
Jani.
Unfortunately, that won't solve the problem that there is no good way to
pass this value from display_modes created from e.g
DRM_IOCTL_MODE_SETCRTC. I believe it would be better if there was a
syscall to inherit from a native mode instead of creating a new one, but
this is too far fetched.
I'll check how this was implemented by nvidia open driver, as they have
added support for this value in their driver somehow.
>> }
static void fill_audio_info(struct audio_info *audio_info,
@@ -7067,6 +7072,13 @@ create_stream_for_sink(struct drm_connector *connector,
&mode, preferred_mode, scale);
preferred_refresh = drm_mode_vrefresh(preferred_mode);
+
+ /*
+ * HACK: In case of multiple supported modes, we should
look at the matching mode to decide this flag.
+ * But what is matching mode, how should it be decided?
+ * Assuming that only preferred mode would have this
flag.
+ */
+ mode.dsc_passthrough_timings_support =
preferred_mode->dsc_passthrough_timings_support;
}
}
@@ -7756,7 +7768,7 @@ create_validate_stream_for_sink(struct drm_connector *connector,
drm_dbg_kms(connector->dev, "%s:%d Validation failed with
%d, retrying w/ YUV420\n",
__func__, __LINE__, dc_result);
aconnector->force_yuv420_output = true;
- }
+}>> stream = create_validate_stream_for_sink(connector, drm_mode,
dm_state, old_stream);
aconnector->force_yuv422_output = false;
Regards,
Lach
