[Ffmpeg-devel] Re: [PATCH] Fix compilation when using --disable-opts

Reimar Döffinger Reimar.Doeffinger
Sun Apr 1 11:28:40 CEST 2007


Hello,
On Sun, Apr 01, 2007 at 11:21:05AM +0200, Michael Niedermayer wrote:
> On Sun, Apr 01, 2007 at 10:09:16AM +0200, Reimar D?ffinger wrote:
> > On Sun, Apr 01, 2007 at 02:23:48AM +0200, Alexis Ballier wrote:
> > > >It might compile but that's because it ends up disabling the asm.
> > > 
> > > which asm was disabled when building pic before
> > > http://svn.mplayerhq.hu/ffmpeg?view=rev&revision=8549
> > > 
> > > that changed from :
> > > #if defined(ARCH_X86) && !(defined(PIC) && defined(__GNUC__))
> > > to :
> > > #if defined(ARCH_X86) && defined(CONFIG_7REGS) && 
> > > defined(CONFIG_EBX_AVAILABLE)
> > 
> > It seems that for some of the code the right condition is
> > #if defined(ARCH_X86) && defined(PIC)
> > since due to the addressing modes used not allowing an extra offset due
> > to PIC, no matter how many registers are available.
> 
> which code are you talking about?
> and dont forget that data can be copied onto the stack that is nothing in
> the gcc asm docs forbids it

at least cabac.h, line 388
"movzbl "MANGLE(ff_h264_lps_range)"(%0, %%ebx, 2), %%esi\n\t"
and line 395
"movzbl "MANGLE(ff_h264_lps_range)"("ret", "range", 2), "range" \n\t"\

No way to make ff_h264_lps_range RIP-relative as needed to allow
for PIC at least on AMD64 it seems. At least that is my understanding of
the issue so far.

Greetings,
Reimar D?ffinger




More information about the ffmpeg-devel mailing list