On Tue, Feb 14, 2017 at 11:34 AM, Chad Versace <chadvers...@chromium.org> wrote:
> On Tue 14 Feb 2017, Matt Turner wrote: > > On Tue, Feb 14, 2017 at 11:21 AM, Matt Turner <matts...@gmail.com> > wrote: > > > On Tue, Feb 14, 2017 at 11:18 AM, Matt Turner <matts...@gmail.com> > wrote: > > >> On Tue, Feb 14, 2017 at 10:59 AM, Jason Ekstrand < > ja...@jlekstrand.net> wrote: > > >>> I'm not sure how I feel about the silent fall-backs. At least in > the Vulkan > > >>> driver, we should fail to compile if we can't get build-id. > Otherwise, > > >>> you'll end up compiling a driver that will always fail device > creation. > > >> > > >> That was really an attempt to preempt questions about Windows. > > >> > > >> I am happy to drop it. > > > > > > Think-o. No, it's necessary for systems that don't have > > > dl_iterate_phdr (Windows, AFAIK). > > > > I promise I'll stop replying to myself after this... > > > > I guess dropping the fallback and simply wrapping build_id.c in > > HAVE_DL_ITERATE_PHDR is sufficient? Windows will build an empty source > > file, which is fine because no code should ever attempt to use it on > > Windows; and there would be no chance of the Vulkan driver calling > > (non-existent) fallback code. > > I'm in favor of dropping the fallback stubs. > I don't really care too much how it works so long as anv fails to compile. I could see the stubs maybe being useful if you wanted to have the shader cache stuff silently and automatically fall back to disabled. Or, the shader cache stuff could have a #if. It doesn't matter to me.
_______________________________________________ mesa-dev mailing list mesa-dev@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/mesa-dev
