[FFmpeg-devel] [PATCH] Workaround Mac-OS x86_64 dynamic loader bug for Yasm generated code

[Art Clarke]

On Wed, Feb 4, 2009 at 12:53 PM, Art Clarke <aclarke at xuggle.com> wrote:
> On Wed, Feb 4, 2009 at 12:27 PM, Art Clarke <aclarke at xuggle.com> wrote:
>> See this chain for details on the problem:
>> http://lists.mplayerhq.hu/pipermail/ffmpeg-devel/2009-February/061189.html
>>
>> Tested on Mac and Linux in 32 and 64 bit builds and everything seems to work.
>

Ping.  [insert summary message here.]

[insert witty comment to get attention of ffmpeg developers].

[insert follow up witty comment to get slight chuckle]

[insert final follow up comment that i think is funny, but ultimately
fails a basic humor test and forces immense resentment to rise up in
ffmpeg developers, leading to death match like Ivan vs. Diego except
actually Art vs. Rhino to see who can best take a headbutt to stomach]

[insert final plea for patch to be cleanly reviewed or rejected]

- Art

p.s. note to self; remember to fill in template sections on this
e-mail before sending.

------------

Index: libavcodec/x86/fft_mmx.asm
===================================================================
--- libavcodec/x86/fft_mmx.asm	(revision 16989)
+++ libavcodec/x86/fft_mmx.asm	(working copy)
@@ -446,8 +446,16 @@
 %endrep
 %undef n

+%ifidn __OUTPUT_FORMAT__,macho64
+; Putting this section in .text on x86_64 darwin causes problems
+section .rodata align=16
+%endif
 align 8
 dispatch_tab%3%2: pointer list_of_fft
+%ifidn __OUTPUT_FORMAT__,macho64
+; Set section back to .text
+section .text align=16
+%endif

 ; On x86_32, this function does the register saving and restoring for
all of fft.
 ; The others pass args in registers and don't spill anything.