2014-07-02 18:48 GMT+02:00 Emil Velikov <emil.l.veli...@gmail.com>:
> On 02/07/14 13:11, Andreas Pokorny wrote:
> > Hi,
> > Still trying the patch. Meanwhile I found two more things here:
> >
> > [...]
> >
> > The driver_name is later freed with free, so to avoid abort strdup should
> > be necessary.
> >
> The free snuck in shortly before Giovanni sent out the last revision. It
> is on
> my todo list, as I rebase the patches next week.
>
Good. There was another case with "kms_swrast" in the other patch.
>
>
>
> >> + if (ret) {
> >> + fprintf(stderr, "failed to load swrast driver\n");
> >> + return ret;
> >> + }
> >> +
> >> + dri->extensions = gbm_dri_screen_extensions;
> >> +
> >> + if (dri->swrast == NULL)
> >> + return -1;
> >> +
> >> + if (dri->swrast->base.version >= 4) {
> >> + dri->screen = dri->swrast->createNewScreen2(0, dri->extensions,
> >> +
> dri->driver_extensions,
> >> + &dri->driver_configs,
> >> dri);
> >> + } else {
> >> + dri->screen = dri->swrast->createNewScreen(0, dri->extensions,
> >> + &dri->driver_configs,
> >> dri);
> >> + }
> >>
> >
> > Is there any reason for not binding the gbm_dri_core_extensions here?
> > If there isnt I think you could easily combine that function with
> > dri_screen_create_dri2.
> >
> gbm_dri_core_extensions seems to be missing in here. If you're thinking
> about
> gbm_dri_device_extensions, then things go a bit different. Those extensions
> are present for hw backed drivers only, thus it would make little sense to
> try
> and bind them in here.
>
> A minor split in dri_load_driver() to handle only the required one will
> come
> with the rebase.
>
>
I think am mostly interested in __DRI_IMAGE extension. This part of the
code is still confusing for me.
regards
Andreas
_______________________________________________
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/mesa-dev
