Martin Storsjö <[email protected]> writes:

> On Thu, 17 Feb 2011, Måns Rullgård wrote:
>
>> Martin Storsjö <[email protected]> writes:
>> 
>> > On Fri, 11 Feb 2011, Mans Rullgard  wrote:
>> >
>> >> Module: ffmpeg
>> >> Branch: master
>> >> Commit: 4b884207eb7319577f8ceb94a56f42c5ac0ab294
>> >> 
>> >> Author: Mans Rullgard <[email protected]>
>> >> Date:   Fri Feb 11 18:28:12 2011 +0000
>> >> 
>> >> configure: remove early check_deps $ARCH_EXT_LIST
>> >> 
>> >> The early disabling of irrelevant arch extensions is no longer
>> >> required, and removing it makes dependencies involving these
>> >> work as expected.
>> >> 
>> >> Signed-off-by: Mans Rullgard <[email protected]>
>> >> 
>> >> ---
>> >> 
>> >>  configure |    2 --
>> >>  1 files changed, 0 insertions(+), 2 deletions(-)
>> >> 
>> >> diff --git a/configure b/configure
>> >> index a3ca15d..a310d25 100755
>> >> --- a/configure
>> >> +++ b/configure
>> >> @@ -2494,8 +2494,6 @@ die_license_disabled version3 libopencore_amrwb
>> >>  
>> >>  enabled version3 && { enabled gpl && enable gplv3 || enable lgplv3; }
>> >>  
>> >> -check_deps $ARCH_EXT_LIST
>> >> -
>> >>  disabled optimizations || check_cflags -fomit-frame-pointer
>> >>  
>> >>  enable_pic() {
>> >
>> > This change broke compilation for Windows CE, where configure now errors 
>> > out with this error:
>> >
>> > Error, no aligned memory allocator but SSE enabled, disable it or use 
>> > --enable-memalign-hack.
>> >
>> > This is checked at this point:
>> >
>> > if ! enabled_any memalign memalign_hack posix_memalign malloc_aligned &&
>> >      enabled_any $need_memalign ; then
>> >     die "Error, no aligned memory allocator but SSE enabled, disable it or 
>> > use --enable-memalign-hack."
>> > fi
>> >
>> > Here, enabled_any $need_memalign (which is "altivec neon sse") triggers, 
>> > believing altivec and sse to be enabled (while neon correctly seems to be 
>> > disabled).
>> 
>> Now I remember why that check_deps was there...
>> 
>> Can you live with it while I come up with a proper solution?
>
> Moving the thread over to this ML (I just ran into this issue in another 
> build configuration, too).

A simple fix would be to move that memalign error after the check_deps
call.  In fact, that should be more correct in general.

-- 
Måns Rullgård
[email protected]
_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel

Reply via email to