[FFmpeg-devel] [PATCH] libavcodec/x86/vp8dsp.asm: Revert syntax causing assembly errors in OSS-Fuzz

[FFmpeg-devel] [PATCH] libavcodec/x86/vp8dsp.asm: Revert syntax causing assembly errors in OSS-Fuzz

[xingyaner via ffmpeg-devel]

In commit 99fb257 and commit 3135bc0 of FFmpeg, New changes introduce syntax that relies on the assembler preprocessor to evaluate Boolean logic operations (== and !!).
 However, the assemblers (NASM/YASM) within the OSS-Fuzz environment, or the x86inc.  asm macro expansion mechanism, cannot correctly parse these complex expressions,
resulting in an expecting ')' error.
To fix the fuzzing build error, I suggest keeping these two lines of code in their original version.
'cglobal put_vp8_epel%1_h6, 6, 6 + npicregs, 8, dst, dststride, src, srcstride, height, mx, picreg'
'cglobal put_vp8_epel%1_h4, 6, 6 + npicregs, 7, dst, dststride, src, srcstride, height, mx, picreg'␋

Signed-off-by: xingyaner <longfor2025@gmail.com>
---
 libavcodec/x86/vp8dsp.asm | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/libavcodec/x86/vp8dsp.asm b/libavcodec/x86/vp8dsp.asm
index 22356f687b..d758b84afa 100644
--- a/libavcodec/x86/vp8dsp.asm
+++ b/libavcodec/x86/vp8dsp.asm
@@ -173,7 +173,7 @@ SECTION .text
 %define MOV movq
 %endif
 
-cglobal put_vp8_epel%1_h6, 6, 6 + npicregs, 6+2*(%1==8), dst, dststride, src, srcstride, height, mx, picreg
+cglobal put_vp8_epel%1_h6, 6, 6 + npicregs, 8, dst, dststride, src, srcstride, height, mx, picreg
 %if %1 == 4
     mova      m3, [filter4_h6_shuf]
 %if PIC
@@ -228,7 +228,7 @@ cglobal put_vp8_epel%1_h6, 6, 6 + npicregs, 6+2*(%1==8), dst, dststride, src, sr
     jg .nextrow
     RET
 
-cglobal put_vp8_epel%1_h4, 6, 6 + npicregs, 6+!!(%1 == 8), dst, dststride, src, srcstride, height, mx, picreg
+cglobal put_vp8_epel%1_h4, 6, 6 + npicregs, 7, dst, dststride, src, srcstride, height, mx, picreg
     mova      m2, [pw_256]
 %if %1 == 8
     shl      mxd, 4
-- 
2.34.1

_______________________________________________
ffmpeg-devel mailing list -- ffmpeg-devel@ffmpeg.org
To unsubscribe send an email to ffmpeg-devel-leave@ffmpeg.org

[FFmpeg-devel] Re: [PATCH] libavcodec/x86/vp8dsp.asm: Revert syntax causing assembly errors in OSS-Fuzz

[Tobias Rapp via ffmpeg-devel]

On 10/12/2025 13:58, xingyaner via ffmpeg-devel wrote:

> In commit 99fb257 and commit 3135bc0 of FFmpeg, New changes introduce syntax that relies on the assembler preprocessor to evaluate Boolean logic operations (== and !!).
>   However, the assemblers (NASM/YASM) within the OSS-Fuzz environment, or the x86inc.  asm macro expansion mechanism, cannot correctly parse these complex expressions,
> resulting in an expecting ')' error.
> To fix the fuzzing build error, I suggest keeping these two lines of code in their original version.
> 'cglobal put_vp8_epel%1_h6, 6, 6 + npicregs, 8, dst, dststride, src, srcstride, height, mx, picreg'
> 'cglobal put_vp8_epel%1_h4, 6, 6 + npicregs, 7, dst, dststride, src, srcstride, height, mx, picreg'␋
>
> Signed-off-by: xingyaner <longfor2025@gmail.com>
> ---
>   libavcodec/x86/vp8dsp.asm | 4 ++--
>   1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/libavcodec/x86/vp8dsp.asm b/libavcodec/x86/vp8dsp.asm
> index 22356f687b..d758b84afa 100644
> --- a/libavcodec/x86/vp8dsp.asm
> +++ b/libavcodec/x86/vp8dsp.asm
> @@ -173,7 +173,7 @@ SECTION .text
>   %define MOV movq
>   %endif
>   
> -cglobal put_vp8_epel%1_h6, 6, 6 + npicregs, 6+2*(%1==8), dst, dststride, src, srcstride, height, mx, picreg
> +cglobal put_vp8_epel%1_h6, 6, 6 + npicregs, 8, dst, dststride, src, srcstride, height, mx, picreg
>   %if %1 == 4
>       mova      m3, [filter4_h6_shuf]
>   %if PIC
> @@ -228,7 +228,7 @@ cglobal put_vp8_epel%1_h6, 6, 6 + npicregs, 6+2*(%1==8), dst, dststride, src, sr
>       jg .nextrow
>       RET
>   
> -cglobal put_vp8_epel%1_h4, 6, 6 + npicregs, 6+!!(%1 == 8), dst, dststride, src, srcstride, height, mx, picreg
> +cglobal put_vp8_epel%1_h4, 6, 6 + npicregs, 7, dst, dststride, src, srcstride, height, mx, picreg
>       mova      m2, [pw_256]
>   %if %1 == 8
>       shl      mxd, 4

I can confirm that this fixes NASM errors during builds on my computer.

BTW: There seems to be some stray Unicode character in the commit 
message, after the quoted lines. There should be no need to quote the 
changed lines as they are obvious from the commit diff.

Regards, Tobias

_______________________________________________
ffmpeg-devel mailing list -- ffmpeg-devel@ffmpeg.org
To unsubscribe send an email to ffmpeg-devel-leave@ffmpeg.org

[FFmpeg-devel] Re: [PATCH] libavcodec/x86/vp8dsp.asm: Revert syntax causing assembly errors in OSS-Fuzz

[xing xing via ffmpeg-devel]

Correct!

Thank you for your reply.

Regards,xingyaner


On Thu, Dec 11, 2025 at 5:50 PM Tobias Rapp <t.rapp@noa-archive.com> wrote:

> On 10/12/2025 13:58, xingyaner via ffmpeg-devel wrote:
>
> > In commit 99fb257 and commit 3135bc0 of FFmpeg, New changes introduce
> syntax that relies on the assembler preprocessor to evaluate Boolean logic
> operations (== and !!).
> >   However, the assemblers (NASM/YASM) within the OSS-Fuzz environment,
> or the x86inc.  asm macro expansion mechanism, cannot correctly parse these
> complex expressions,
> > resulting in an expecting ')' error.
> > To fix the fuzzing build error, I suggest keeping these two lines of
> code in their original version.
> > 'cglobal put_vp8_epel%1_h6, 6, 6 + npicregs, 8, dst, dststride, src,
> srcstride, height, mx, picreg'
> > 'cglobal put_vp8_epel%1_h4, 6, 6 + npicregs, 7, dst, dststride, src,
> srcstride, height, mx, picreg'␋
> >
> > Signed-off-by: xingyaner <longfor2025@gmail.com>
> > ---
> >   libavcodec/x86/vp8dsp.asm | 4 ++--
> >   1 file changed, 2 insertions(+), 2 deletions(-)
> >
> > diff --git a/libavcodec/x86/vp8dsp.asm b/libavcodec/x86/vp8dsp.asm
> > index 22356f687b..d758b84afa 100644
> > --- a/libavcodec/x86/vp8dsp.asm
> > +++ b/libavcodec/x86/vp8dsp.asm
> > @@ -173,7 +173,7 @@ SECTION .text
> >   %define MOV movq
> >   %endif
> >
> > -cglobal put_vp8_epel%1_h6, 6, 6 + npicregs, 6+2*(%1==8), dst,
> dststride, src, srcstride, height, mx, picreg
> > +cglobal put_vp8_epel%1_h6, 6, 6 + npicregs, 8, dst, dststride, src,
> srcstride, height, mx, picreg
> >   %if %1 == 4
> >       mova      m3, [filter4_h6_shuf]
> >   %if PIC
> > @@ -228,7 +228,7 @@ cglobal put_vp8_epel%1_h6, 6, 6 + npicregs,
> 6+2*(%1==8), dst, dststride, src, sr
> >       jg .nextrow
> >       RET
> >
> > -cglobal put_vp8_epel%1_h4, 6, 6 + npicregs, 6+!!(%1 == 8), dst,
> dststride, src, srcstride, height, mx, picreg
> > +cglobal put_vp8_epel%1_h4, 6, 6 + npicregs, 7, dst, dststride, src,
> srcstride, height, mx, picreg
> >       mova      m2, [pw_256]
> >   %if %1 == 8
> >       shl      mxd, 4
>
> I can confirm that this fixes NASM errors during builds on my computer.
>
> BTW: There seems to be some stray Unicode character in the commit
> message, after the quoted lines. There should be no need to quote the
> changed lines as they are obvious from the commit diff.
>
> Regards, Tobias
>
>
_______________________________________________
ffmpeg-devel mailing list -- ffmpeg-devel@ffmpeg.org
To unsubscribe send an email to ffmpeg-devel-leave@ffmpeg.org

[FFmpeg-devel] Re: [PATCH] libavcodec/x86/vp8dsp.asm: Revert syntax causing assembly errors in OSS-Fuzz

[Michael Niedermayer via ffmpeg-devel]

[-- Attachment #1.1: Type: text/plain, Size: 1537 bytes --]

Hi

On Wed, Dec 10, 2025 at 08:58:03PM +0800, xingyaner via ffmpeg-devel wrote:
> In commit 99fb257 and commit 3135bc0 of FFmpeg, New changes introduce syntax that relies on the assembler preprocessor to evaluate Boolean logic operations (== and !!).
>  However, the assemblers (NASM/YASM) within the OSS-Fuzz environment, or the x86inc.  asm macro expansion mechanism, cannot correctly parse these complex expressions,
> resulting in an expecting ')' error.
> To fix the fuzzing build error, I suggest keeping these two lines of code in their original version.
> 'cglobal put_vp8_epel%1_h6, 6, 6 + npicregs, 8, dst, dststride, src, srcstride, height, mx, picreg'
> 'cglobal put_vp8_epel%1_h4, 6, 6 + npicregs, 7, dst, dststride, src, srcstride, height, mx, picreg'␋
> 
> Signed-off-by: xingyaner <longfor2025@gmail.com>
> ---
>  libavcodec/x86/vp8dsp.asm | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)

git am seems not to like this, maybe related to the stray unicode chars, didnt investigate

Applying: libavcodec/x86/vp8dsp.asm: Revert syntax causing assembly errors in OSS-Fuzz
Using index info to reconstruct a base tree...
error: patch failed: libavcodec/x86/vp8dsp.asm:173
error: libavcodec/x86/vp8dsp.asm: patch does not apply
error: Did you hand edit your patch?
It does not apply to blobs recorded in its index.

[...]
-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

Everything should be made as simple as possible, but not simpler.
-- Albert Einstein

[-- Attachment #1.2: signature.asc --]
[-- Type: application/pgp-signature, Size: 195 bytes --]

[-- Attachment #2: Type: text/plain, Size: 163 bytes --]

_______________________________________________
ffmpeg-devel mailing list -- ffmpeg-devel@ffmpeg.org
To unsubscribe send an email to ffmpeg-devel-leave@ffmpeg.org

[FFmpeg-devel] [PATCH] libavcodec/x86/vp8dsp.asm: Revert syntax causing assembly errors in OSS-Fuzz

[xingyaner via ffmpeg-devel]

In commit 99fb257 and commit 3135bc0 of FFmpeg, New changes introduce syntax that relies on the assembler preprocessor to evaluate Boolean logic operations (== and !!).
 However, the assemblers (NASM/YASM) within the OSS-Fuzz environment, or the x86inc.  asm macro expansion mechanism, cannot correctly parse these complex expressions,
resulting in an expecting ')' error.
To fix the fuzzing build error, I suggest keeping these two lines of code in their original version.
'cglobal put_vp8_epel%1_h6, 6, 6 + npicregs, 8, dst, dststride, src, srcstride, height, mx, picreg'
'cglobal put_vp8_epel%1_h4, 6, 6 + npicregs, 7, dst, dststride, src, srcstride, height, mx, picreg'␋

Signed-off-by: xingyaner <longfor2025@gmail.com>
---
 libavcodec/x86/vp8dsp.asm | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/libavcodec/x86/vp8dsp.asm b/libavcodec/x86/vp8dsp.asm
index 22356f687b..d758b84afa 100644
--- a/libavcodec/x86/vp8dsp.asm
+++ b/libavcodec/x86/vp8dsp.asm
@@ -173,7 +173,7 @@ SECTION .text
 %define MOV movq
 %endif
 
-cglobal put_vp8_epel%1_h6, 6, 6 + npicregs, 6+2*(%1==8), dst, dststride, src, srcstride, height, mx, picreg
+cglobal put_vp8_epel%1_h6, 6, 6 + npicregs, 8, dst, dststride, src, srcstride, height, mx, picreg
 %if %1 == 4
     mova      m3, [filter4_h6_shuf]
 %if PIC
@@ -228,7 +228,7 @@ cglobal put_vp8_epel%1_h6, 6, 6 + npicregs, 6+2*(%1==8), dst, dststride, src, sr
     jg .nextrow
     RET
 
-cglobal put_vp8_epel%1_h4, 6, 6 + npicregs, 6+!!(%1 == 8), dst, dststride, src, srcstride, height, mx, picreg
+cglobal put_vp8_epel%1_h4, 6, 6 + npicregs, 7, dst, dststride, src, srcstride, height, mx, picreg
     mova      m2, [pw_256]
 %if %1 == 8
     shl      mxd, 4
-- 
2.34.1

_______________________________________________
ffmpeg-devel mailing list -- ffmpeg-devel@ffmpeg.org
To unsubscribe send an email to ffmpeg-devel-leave@ffmpeg.org