[xine-devel] Re: [xine-user] CFLAGS breaking builds and fucking up optimalisations

[Michael R. <mr...@us...>]

Hi Torsten,

> > Your compiler seems to be in deep trouble. My gcc 3.2 compiles xine
> > flawlessly.
>
> Did you try running
> gcc -DHAVE_CONFIG_H -I. -I. -I../.. -I../.. -I../../include
> -I../../include -I../../src -I../../src/xine-engine
> -I../../src/xine-engine -I../../src/xine-utils -O2 -march=pentium4
> -Wall -D_REENTRANT -D_FILE_OFFSET_BITS=64 -DXINE_COMPILE -O3 -pipe
> -fomit-frame-pointer -falign-functions=4 -falign-loops=4
> -falign-jumps=4 -mpreferred-stack-boundary=2
> -fexpensive-optimizations -fschedule-insns2 -fno-strict-aliasing
> -ffast-math -funroll-loops -finline-functions -mcpu=pentiumpro
> -I/usr/include/kde/artsc -c cfft.c -MT cfft.lo -MD -MP -MF
> .deps/cfft.TPlo  -DPIC -o cfft.lo
> in src/libfaad? I receive the exactly same error message, and I do
> not have some "buggy redhat compiler", I use gcc 3.2.2 compiled from
> source with SuSE 8.1.

Finally some light is shed on this. Thanks for this useful contribution.

No, I did not try that, because I do not have a Pentium4 here. You are 
indeed right that compilation bails out on my gcc as well, but only 
with the "-O2 -march=pentium4" which are _NOT_ set by xine's configure.
When I remove them and go with standard xine CFLAGS, everything works 
nicely. It even works, when I replace the "-mcpu=pentiumpro" with 
"-mcpu=pentium4", which will be used, when you configure with 
"--host=pentium4-suse-linux" which is the recommended course of action 
when you want pentium4 optimization (see ./configure --help).

So to draw the conclusion: Some users have compilation problems, because 
the gcc chokes on additional CFLAGS the users specified. I cannot see 
any mistake on xine's side. When you add your own CFLAGS and the 
compiler doesn't like them, we cannot help you much apart from 
recommending to use the defaults suggested by ./configure.

But it has at least been shown, that it is not a RedHat-specific 
problem, so my appologies to RedHat and its users for accusing them of 
breaking gcc.

Michael

-- 
panic("Unable to find empty mailbox for aha1542.\n");
	2.2.16 /usr/src/linux/drivers/scsi/aha1542.c