On Wed, Jun 6, 2018 at 1:25 PM, Daniel Stone <dan...@fooishbar.org> wrote:
> We definitely do, but I assume it's not been tested recently ... Ok, Lionel's patch it is then. :) > (Sorry about mangled formatting) > Wow, you managed to mangle it bad... No worries though; I have to talk to plenty of Outlook users every day. :P --Jason > On Wed, 6 Jun 2018, 8:42 pm Jason Ekstrand, <ja...@jlekstrand.net> wrote: > >> On June 6, 2018 12:37:58 Daniel Stone <dan...@fooishbar.org> wrote: >> >>> Right, it's a feature we use, because we do all import them as separate >>> EGLImages ... and we won't if it's not advertised. >> >> >> I'm a bit skeptical given that it doesn't actually work today because the >> DRI format to Mesa format conversation function doesn't handle R it RG >> formats today. Maybe it goes through some other path? >> >> In any case, I'm happy to drop this patch in favor of Lionel's patch to >> make the DRI format to Mesa format conversation function actually work for >> these formats. >> >> --Jason >> >> >> >>> On Wed, 6 Jun 2018, 7:05 pm Jason Ekstrand, <ja...@jlekstrand.net> >>> wrote: >>> >>>> On Wed, Jun 6, 2018 at 11:03 AM, Jason Ekstrand <ja...@jlekstrand.net> >>>> wrote: >>>> >>>>> On Wed, Jun 6, 2018 at 11:00 AM, Daniel Stone <dan...@fooishbar.org> >>>>> wrote: >>>>> >>>>>> Sorry, but as written this will regress ability to import NV12 images >>>>>> as separately-addressed planes with shader conversion to RGB; Kodi, >>>>>> Mutter >>>>>> and Weston all use this. >>>>>> >>>>> >>>>> I don't believe it will. It only makes it so that we don't advertise >>>>> R and RG formats through eglQueryDmaBufFormatsEXT. This means that you >>>>> can't import the planes each as separate images but you can still import a >>>>> planar image. >>>>> >>>> >>>> Arguably, though, importing the planes as separate images does sound >>>> like a feature... >>>> >>>> >>>> >>>>> --Jason >>>>> >>>>> >>>>> >>>>>> On Wed, 6 Jun 2018, 6:48 pm Jason Ekstrand, <ja...@jlekstrand.net> >>>>>> wrote: >>>>>> >>>>>>> Cc: mesa-sta...@lists.freedesktop.org >>>>>>> --- >>>>>>> src/mesa/drivers/dri/i965/intel_screen.c | 12 ++++++++++++ >>>>>>> 1 file changed, 12 insertions(+) >>>>>>> >>>>>>> diff --git a/src/mesa/drivers/dri/i965/intel_screen.c >>>>>>> b/src/mesa/drivers/dri/i965/intel_screen.c >>>>>>> index 5f0eeb41779..f681b221e7b 100644 >>>>>>> --- a/src/mesa/drivers/dri/i965/intel_screen.c >>>>>>> +++ b/src/mesa/drivers/dri/i965/intel_screen.c >>>>>>> @@ -1269,6 +1269,18 @@ intel_image_format_is_supported(const struct >>>>>>> intel_image_format *fmt) >>>>>>> fmt->fourcc == __DRI_IMAGE_FOURCC_SABGR8888) >>>>>>> return false; >>>>>>> >>>>>>> + /* The dri_interface.h file says: >>>>>>> + * >>>>>>> + * "R8, GR88 and NONE should not be used with >>>>>>> createImageFromName or >>>>>>> + * createImage, and are returned by query from sub images >>>>>>> created with >>>>>>> + * createImageFromNames (NONE, see above) and fromPlane (R8 & >>>>>>> GR88)." >>>>>>> + * >>>>>>> + * Let's not advertise support for R or RG formats. >>>>>>> + */ >>>>>>> + if (fmt->components == __DRI_IMAGE_COMPONENTS_R || >>>>>>> + fmt->components == __DRI_IMAGE_COMPONENTS_RG) >>>>>>> + return false; >>>>>>> + >>>>>>> return true; >>>>>>> } >>>>>>> >>>>>>> -- >>>>>>> 2.17.1 >>>>>>> >>>>>>> _______________________________________________ >>>>>>> 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
