[FFmpeg-devel] Patch: Inline asm fixes for Intel compiler on Windows

[Matt Oliver]

On 9 February 2014 19:43, Carl Eugen Hoyos <cehoyos at ag.or.at> wrote:

> Matt Oliver <protogonoi <at> gmail.com> writes:
>
> > > s/swscale_mmx.c/swscale.c
> >
> > Sorry I dont understand what your comment is
> > referring to here. Do you want me to remove that
> > comment line from the patch?
>
> No, please replace "swscale_mmx.c" in the comment
> with "swscale.c" (or confirm that I am wrong).
>
> (And please fix your quoting to make my answer to
> your mail easier to read.)
>
> > > Please simplify as discussed.
> >
> > OK, as i said I was trying to be forward thinking
> > in avoiding future issues but if youd prefer it
> > simplified then I will do it that way.
>
> Or consider waiting for a comment from somebody else.
>
> Thank you for the other explanations, Carl Eugen
>
> _______________________________________________
> ffmpeg-devel mailing list
> ffmpeg-devel at ffmpeg.org
> http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
>

> No, please replace "swscale_mmx.c" in the comment
> with "swscale.c" (or confirm that I am wrong).

No need you were correct (although its specifically in x86/swscale.c). And
ive used sed enough that I should have understood what you meant (also if
the formatting of my emails is off then I do apologise as its fine on my
end so i havnt noticed). Anyway I updated the patch1 so that the comment
now is for x86/swscale.c.

Patch4 was updated so the second BROKEN_REGISTER_ALLOCATION define has been
renamed so that it is now much clearer it is a 64bit register allocation
error causing mismatches as compared to the other one which is a register
allocation issue that occurs all the time.

> Or consider waiting for a comment from somebody else.

Ill save sometime on this. I had actually tested this (and then forgot) but
the check for inline_asm_direct_symbol_refs has to be done the way it is
currently. Just using check_inline_asm doesnt declare the 'int test'
variable and does cause issues. Even gcc will fail the test if done that
way so it has to stay as it stands currently. Gcc is more forgiving about
extern variables than intel but it still needs them to be somewhere.

Anyway thanks for the feedback Carl Eugen. Given the last few fixes have
just been comments and small things hopefully this is close to being ready
;)
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 0001-1-6-Add-missing-external-declarations.patch
Type: application/octet-stream
Size: 2408 bytes
Desc: not available
URL: <http://ffmpeg.org/pipermail/ffmpeg-devel/attachments/20140209/3ee63cd2/attachment.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 0004-4-6-Fix-for-broken-register-allocation-issues-with-I.patch
Type: application/octet-stream
Size: 3013 bytes
Desc: not available
URL: <http://ffmpeg.org/pipermail/ffmpeg-devel/attachments/20140209/3ee63cd2/attachment-0001.obj>