[Bug 111306] gbm creates BO with wrong pitch when dri3_get_modifiers returns modifiers, causing drmModeAddFB2WithModifiers to fail

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=111306

--- Comment #19 from Lionel Landwerlin <lionel.g.landwerlin at linux.intel.com> ---
(In reply to Hans de Goede from comment #17)
> That function builds a mapping from formats to supported-modifiers for each
> format.
> 
> I've added the following debug:
> 
> diff --git a/hw/xfree86/drivers/modesetting/drmmode_display.c
> b/hw/xfree86/drivers/modesetting/drmmode_display.c
> index e0d7cfb5c..249661985 100644
> --- a/hw/xfree86/drivers/modesetting/drmmode_display.c
> +++ b/hw/xfree86/drivers/modesetting/drmmode_display.c
> @@ -2004,6 +2004,9 @@ populate_format_modifiers(xf86CrtcPtr crtc, const
> drmModePlane *kplane,
>          for (j = 0; j < fmt_mod_blob->count_modifiers; j++) {
>              struct drm_format_modifier *mod = &blob_modifiers[j];
>  
> +            if (mod->modifier == I915_FORMAT_MOD_Y_TILED_CCS)
> +                ErrorF("populate_format_modifiers i %d format %x formats
> %llx\n", i, blob_formats[i], mod->formats);
> +
>              if ((i < mod->offset) || (i > mod->offset + 63))
>                  continue;
>              if (!(mod->formats & (1 << (i - mod->offset))))
> 
> Which for me generates:
> 
> [  3860.889] populate_format_modifiers i 0 format 20203843 formats 3c
> [  3860.889] populate_format_modifiers i 1 format 36314752 formats 3c
> [  3860.889] populate_format_modifiers i 2 format 34325258 formats 3c
> [  3860.889] populate_format_modifiers i 3 format 34324258 formats 3c
> [  3860.889] populate_format_modifiers i 4 format 34325241 formats 3c
> [  3860.889] populate_format_modifiers i 5 format 34324241 formats 3c
> [  3860.889] populate_format_modifiers i 6 format 30335258 formats 3c
> [  3860.889] populate_format_modifiers i 7 format 30334258 formats 3c
> [  3860.889] populate_format_modifiers i 8 format 56595559 formats 3c
> [  3860.889] populate_format_modifiers i 9 format 55595659 formats 3c
> [  3860.889] populate_format_modifiers i 10 format 59565955 formats 3c
> [  3860.889] populate_format_modifiers i 11 format 59555956 formats 3c
> 
> Note mod->offset always is 0 for me, I had that in the debug before but I
> dropped it.
> 
> IOW for formats 2-5 in the format list part of the retreived blob, the
> I915_FORMAT_MOD_Y_TILED_CCS modifier is supported. 
> 
> Notice that those 4 formats are:
> DRM_FORMAT_XRGB8888 (XR24)
> DRM_FORMAT_XBGR8888 (XB24)
> DRM_FORMAT_ARGB8888 (AR24)
> DRM_FORMAT_ABGR8888 (AB24)
> 
> So the fb ending up in one of these seems reasonable.
> 
> I guess that for some reason the blob property describing the possible
> modifiers on the CRTC uses in your case is different.
> 
> Are you using a skylake GPU? Perhaps things are different for you because
> you are using a (e)DP monitor/panel instead of the HDMI monitors I'm using?

Right yeah, I'm basically on DVI :) But Skylake GPU.

-- 
You are receiving this mail because:
You are the assignee for the bug.
You are the QA Contact for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/intel-3d-bugs/attachments/20190808/33efc81a/attachment.html>