[FFmpeg-devel] [PATCH 1/2] avio: add avio_skip macro

[Måns Rullgård]

"Ronald S. Bultje" <rsbultje at gmail.com> writes:

> Hi,
>
> On Thu, Mar 3, 2011 at 9:16 AM, Michael Niedermayer <michaelni at gmx.at> wrote:
>> On Thu, Mar 03, 2011 at 08:26:50AM -0500, Ronald S. Bultje wrote:
>>> Hi,
>>>
>>> On Thu, Mar 3, 2011 at 6:37 AM, Peter Ross <pross at xvid.org> wrote:
>>> > This is a substitute for the url_fskip function that was deprecated by
>>> > commit 0300db8ad778a194b4a8ec98f6da3de5b41c46ee. avio_fskip is provided to
>>> > improve demuxer code readability. It distinguishes the act of skipping over
>>> > unknown or irrelevant bytes from the standard avio_seek operation.
>>> > ---
>>> > ?libavformat/avio.h | ? ?6 ++++++
>>> > ?1 files changed, 6 insertions(+), 0 deletions(-)
>>> >
>>> > diff --git a/libavformat/avio.h b/libavformat/avio.h
>>> > index cc46ad7..fc9f618 100644
>>> > --- a/libavformat/avio.h
>>> > +++ b/libavformat/avio.h
>>> > @@ -470,6 +470,12 @@ int avio_put_str16le(AVIOContext *s, const char *str);
>>> > ?int64_t avio_seek(AVIOContext *s, int64_t offset, int whence);
>>> >
>>> > ?/**
>>> > + * Skip given number of bytes forward
>>> > + * @return new position or AVERROR.
>>> > + */
>>> > +#define avio_skip(s, offset) avio_seek(s, offset, SEEK_CUR)
>>> > +
>>> > +/**
>>>
>>> OK with me.
>>
>> for the record id like to repeat that a function is better for this
>
> int url_fskip(ByteIOContext *s, int64_t offset)
> {
>     int64_t ret = url_fseek(s, offset, SEEK_CUR);
>     return ret < 0 ? ret : 0;
> }
>
> No it is not.
>
> Dump of assembler code for function url_fskip:
> 0x00000001000304c0 <url_fskip+0>:	sub    $0x8,%rsp
> 0x00000001000304c4 <url_fskip+4>:	mov    $0x1,%edx
> 0x00000001000304c9 <url_fskip+9>:	callq  0x100030290 <url_fseek>
> 0x00000001000304ce <url_fskip+14>:	mov    %rax,%rdx
> 0x00000001000304d1 <url_fskip+17>:	xor    %eax,%eax
> 0x00000001000304d3 <url_fskip+19>:	test   %rdx,%rdx
> 0x00000001000304d6 <url_fskip+22>:	cmovle %rdx,%rax
> 0x00000001000304da <url_fskip+26>:	add    $0x8,%rsp
> 0x00000001000304de <url_fskip+30>:	retq
> 0x00000001000304df <url_fskip+31>:	nop
> End of assembler dump.
>
> And no it is not inlined.

It is also useless ABI bloat.

-- 
M?ns Rullg?rd
mans at mansr.com