On Thu, Mar 8, 2012 at 8:31 PM, Jeremy Huddleston <[email protected]> wrote: > > On Mar 8, 2012, at 09:51, Matt Turner wrote: > >> On Thu, Mar 8, 2012 at 5:41 PM, Jeremy Huddleston <[email protected]> wrote: >>> Signed-off-by: Jeremy Huddleston <[email protected]> >>> --- >>> configure.ac | 9 +++++++++ >>> 1 files changed, 9 insertions(+), 0 deletions(-) >>> >>> diff --git a/configure.ac b/configure.ac >>> index c3c711c..1ca3c02 100644 >>> --- a/configure.ac >>> +++ b/configure.ac >>> @@ -297,6 +297,15 @@ AC_COMPILE_IFELSE([AC_LANG_SOURCE([[ >>> #include <mmintrin.h> >>> int main () { >>> __m64 v = _mm_cvtsi32_si64 (1); >>> + __m64 w; >>> + int8_t x; >>> + >>> + /* Some versions of clang will choke on K */ >>> + asm ("pshufw %2, %1, %0\n\t" >>> + : "=y" (v) >>> + : "y" (w), "K" (x) >>> + ); >>> + >>> return _mm_cvtsi64_si32 (v); >>> }]])], have_mmx_intrinsics=yes) >>> CFLAGS=$xserver_save_CFLAGS >>> -- >>> 1.7.9 >> >> Newer clangs support the "K" constraint? Fixed in response to its use >> in pixman? :) > > Unfortunately, I'm just being optimistic at this point.
Is there at least a bug reported in LLVM bugtracker? IMHO a good practice is to never add any workarounds without having a bug reported against the offending tool or library. -- Best regards, Siarhei Siamashka _______________________________________________ Pixman mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/pixman
