[FFmpeg-devel] [PATCH] avio: fix fourcc if any character is >=0x80.

[Baptiste Coudurier]

On 3/4/11 1:19 PM, Ronald S. Bultje wrote:
> Hi,
> 
> On Fri, Mar 4, 2011 at 4:14 PM, Baptiste Coudurier
> <baptiste.coudurier at gmail.com> wrote:
>> On 3/4/11 7:50 AM, Ronald S. Bultje wrote:
>>> Hi,
>>>
>>> 2011/3/4 M?ns Rullg?rd <mans at mansr.com>:
>>>> "Ronald S. Bultje" <rsbultje at gmail.com> writes:
>>>>
>>>>> Fixes issue 2638.
>>>>> ---
>>>>>  libavformat/avio_internal.h |    5 ++++-
>>>>>  1 files changed, 4 insertions(+), 1 deletions(-)
>>>>>
>>>>> diff --git a/libavformat/avio_internal.h b/libavformat/avio_internal.h
>>>>> index 3b38990..279c7f6 100644
>>>>> --- a/libavformat/avio_internal.h
>>>>> +++ b/libavformat/avio_internal.h
>>>>> @@ -42,6 +42,9 @@ int ffio_read_partial(AVIOContext *s, unsigned char *buf, int size);
>>>>>
>>>>>  void ffio_fill(AVIOContext *s, int b, int count);
>>>>>
>>>>> -#define ffio_wfourcc(pb, str) avio_wl32(pb, MKTAG((str)[0], (str)[1], (str)[2], (str)[3]))
>>>>> +static av_always_inline void ffio_wfourcc(AVIOContext *pb, const uint8_t *s)
>>>>> +{
>>>>> +    avio_wl32(pb, MKTAG(s[0], s[1], s[2], s[3]));
>>>>> +}
>>
>> avio_write(pb, s, 4) is faster and better IMHO.
>> Why not using that ?
> 
> I'd like to eventually change avio_wl32() into doing that if
> endianness is suitable, just didn't get to it yet.
> 
> Also, I believe it should actually use AV_WL32() instead of a plain
> avio_write() (which, on x86-32, expands to the same ASM).

That won't eliminate the MKTAG which is slower, and in the context of a
"fourcc"/"tag", avio_write is better in every case here.

-- 
Baptiste COUDURIER
Key fingerprint                 8D77134D20CC9220201FC5DB0AC9325C5C1ABAAA
FFmpeg maintainer                                  http://www.ffmpeg.org