[MPlayer-dev-eng] [PATCH] gcc 4.4.0 and liba52 internal

[KO Myung-Hun]

Hi/2.

Andrew Savchenko wrote:
> Hi,
>
> On Sunday 10 May 2009, KO Myung-Hun wrote:
>   
>> Hi/2.
>>
>> Andrew Savchenko wrote:
>>     
>>> Hi,
>>>
>>> On Sunday 10 May 2009, KO Myung-Hun wrote:
>>>       
>>>> Hi/2.
>>>>
>>>> This patch fixes the error as the attachment when compiling 
>>>>         
> liba52
>   
>>>> internal version with gcc 4.4.0 with -O4.
>>>>
>>>> We can avoid the error using -O2 instead of -O4.
>>>>         
>>> -Ox is set of flags. Can you tell which flag exactly triggers 
>>>       
> this?
>   
>>> It is important to understand whether this is gcc bug or bug in
>>> liba52.
>>>       
>> I'm sorry what I tried are just -O2, -O3 and -O4. Of them, only -O2
>> succeeded.
>>     
>
> Ok, but you misunderstood me. As of gcc-4.3.3 -O3 differs from -O2 by 
> the following flags: -finline-functions, -funswitch-loops,
> -fpredictive-commoning, -fgcse-after-reload, -ftree-vectorize.
> For gcc-4.4.0 this may differ, so please refer to gcc manual. -O4 has 
> no difference from -O3 in gcc for now.
>
>   

gcc 4.3.3 is also different from gcc 4.4.0 when interpreting -O4 ?
Because gcc 4.3.3 compiles MPlayer well with -O4.

> Then, please, try to compile with -O2 and wiht one of the -O3-specific 
> flags consecutively in order to find flag causing this problem. In 
> theory it is possible that not a single flag, but combination of some 
> flags causes this issue, but likelihood of latter event is rather 
> low.
>
>   

Ok. I'll try.

-- 
KO Myung-Hun

Using Mozilla SeaMonkey 1.1.16
Under OS/2 Warp 4 for Korean with FixPak #15
On AMD ThunderBird 1 GHz with 512 MB RAM

Korean OS/2 User Community : http://www.ecomstation.co.kr