[MPlayer-users] configure --disable-network fails on make

Alexander Bokovikov openworld at uralweb.ru
Mon Mar 17 12:31:36 CET 2008


John Brown wrote:

> Anyway, why don't you simply build MPlayer on the PIII?

LOL :) Did you ever try to do it? Do you imagin how much time it may take? 
Even if I'd compile it only once (after all possible debugging done on P-IV) 
I could take a dinner + coffee + ... until it returns from making... I tried 
it some time ago and have no wishes to repeat it.

>> Exception Code: 0xC000001D
>> Address: 0x0040DE5C

> That means EXCEPTION_ILLEGAL_INSTRUCTION.

Yes, that's a puzzle. I've tried to debug your EXE in external debugger and 
have found particular line, which calls an "address", located BETWEEN two 
subsequent assembler lines. This is the error reason. When such call is 
made, a completely invalid opcode occurs at CPU input. Somebody more skilled 
definitely is required here to investigate why it so happen... Unfortunately 
I even can't tell in which source module it occurs. Though I can try your 
advice concerning gdb...

Nevertheless, it looks like I've found how to overcome this error (Exception 
0xC000001D). I just need to assign the target. It works OK if I 
use --target=i686-MINGW32 + --disable-ssse3. For some reason MEncoder does 
not indicate SSSE3 state among other flags, when we run it. Nevertheless it 
is important to explicitly disable it, as I've found. So, my 
"P-III-compatible" configure string is here:

./configure --enable-static --disable-network --disable-runtime-cpudetection 
 --enable-mmx --enable-mmxext --enable-sse --disable-sse2 --disable-ssse3 --enable-largefiles 
 --target=i686-MINGW32

Of course, I've fixed the Assert() bug in swscale.c, as I wrote in another 
thread. After this my build is working nice... already for several hours... 
will see how it will behave for a longer testing with several files...

Thank you for your support.




More information about the MPlayer-users mailing list