On Fri, Jun 05, 2015 at 01:40:31PM +0200, Andreas Cadhalpun wrote: > On 05.06.2015 00:47, Michael Niedermayer wrote: > > On Thu, Jun 04, 2015 at 11:17:45PM +0200, Andreas Cadhalpun wrote: > >> Without this check it causes SIGILL crashes on ARMv5. > >> > >> Signed-off-by: Andreas Cadhalpun <andreas.cadhal...@googlemail.com> > >> --- > >> libavutil/arm/cpu.c | 2 +- > >> 1 file changed, 1 insertion(+), 1 deletion(-) > >> > >> diff --git a/libavutil/arm/cpu.c b/libavutil/arm/cpu.c > >> index f1683e8..ac42a05 100644 > >> --- a/libavutil/arm/cpu.c > >> +++ b/libavutil/arm/cpu.c > >> @@ -128,7 +128,7 @@ int ff_get_cpu_flags_arm(void) > >> trickle down. */ > >> if (flags & (AV_CPU_FLAG_VFPV3 | AV_CPU_FLAG_NEON)) > >> flags |= AV_CPU_FLAG_ARMV6T2; > >> - else > >> + else if (flags & AV_CPU_FLAG_ARMV6) > >> /* Some functions use the 'setend' instruction which is deprecated on > >> ARMv8 > >> * and serializing on some ARMv7 cores. This ensures such functions > >> * are only enabled on ARMv6. */ > > > > shouldnt this be after > > "- if (flags & AV_CPU_FLAG_ARMV6T2) > > flags |= AV_CPU_FLAG_ARMV6; > > " > > That would require duplicating the !(flags & (AV_CPU_FLAG_VFPV3 | > AV_CPU_FLAG_NEON) > check. > > > or (AV_CPU_FLAG_ARMV6T2 | AV_CPU_FLAG_ARMV6) > > That seems better, assuming setend is useful on ARMv6t2.
AFAIK its even useful on some later cpus that have NEON (that is IIRC the one i tested long ago) > Updated patch attached. LGTM [...] -- Michael GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB Its not that you shouldnt use gotos but rather that you should write readable code and code with gotos often but not always is less readable
signature.asc
Description: Digital signature
_______________________________________________ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
