Pierre Habouzit ha scritto:
> On Wed, Dec 13, 2006 at 05:53:03PM +0100, A Mennucc wrote:
>> here is some more info:
>>
>> $ gdb ./mplayer
>> This GDB was configured as "x86_64-linux-gnu"...Using host libthread_db
>> library "/lib/libthread_db.so.1".
>>
>> (gdb) run ~/mplayer/bench/foobar.mpeg
>>
>> Program received signal SIGSEGV, Segmentation fault.
>> [Switching to Thread 47190863550720 (LWP 1368)]
>> MC_put_o_16_mmxext (dest=0xcb5f00 "", ref=0x0, stride=304, height=16)
>> at motion_comp_mmx.c:546
>> 546 movq_m2r (*ref, mm0);
>> (gdb) li
>> 541
>> 542 static inline void MC_put1_16 (int height, uint8_t * dest, const
>> uint8_t * ref,
>> 543 const int stride)
>> 544 {
>> 545 do {
>> 546 movq_m2r (*ref, mm0);
>> 547 movq_m2r (*(ref+8), mm1);
>> 548 ref += stride;
>> 549 movq_r2m (mm0, *dest);
>> 550 movq_r2m (mm1, *(dest+8));
>>
>> we should understand why ref==0
>>
>> anyway I will add an assert
>
> O_o *blink* *blink*
>
> do you know that assert is a macro that may be disabled if for some
> reason the build system defines NDEBUG ?!
you are criticizing my way of solving this bug *even before I sit down
and start writing the code*
even supposing that
- I would use assert(ref)
- the code is compiled with NDEBUG
did you consider the fact that I will check my patch against your
foobar.mpeg, in the end, to make sure that it will work?
congratulations! you really know how to go along and collaborate with
other people
(yes that is sarcasm)
a.
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
