XFree86-4.3.0-44 build fails in xc/programs/bitmap

Mike A. Harris mharris at www.linux.org.uk
Sat Nov 29 01:58:39 UTC 2003

On Thu, 27 Nov 2003, Alexander Dalloz wrote:

>> I'm trying to build XFree from development. It fails here:
>> gcc -m32 -o bitmap -march=athlon-xp -m3dnow -msse -mmmx -mfpmath=sse -Os  
>> -fno-strict-aliasing -pip e -ansi -pedantic -Wall -Wpointer-arith -Wundef    
>>   -L../../exports/lib   BitEdit.o CutPaste.o Grap hics.o ReqMach.o Bitmap.o  
>>                   Dialog.o Handlers.o -lXaw -lXmu -lXt -lSM -lICE -lXpm - 
>> lXext -lX11     -lm -Wl,-rpath-link,../../exports/lib
>> ../../exports/lib/libXaw.so: undefined reference to `.L91'
>> collect2: ld returned 1 exit status
>> make[4]: *** [bitmap] Error 1
>> make[4]: Leaving directory 
>> `/usr/src/redhat/BUILD/XFree86-4.3.0/xc/programs/bitmap'
>> Is this the correct list for build issues?
>If you build test packages I think so.
>> thanks
>> sean
>Btw. to set "-march=athlon-xp -m3dnow -msse -mmmx -mfpmath=sse" as gcc
>3.3 compiler options the 3dnow, sse, mmx and -mfemath=sse calls are
>redundant as the -march setting to athlon-xp will choose that all.

Athlon reorders instructions internally anyway, the net result
generally being that it makes no difference what CPU instruction
scheduling is chosen, as the chip does it's own thing internally.

I'd be kind of surprised if compiling with all those options 
yields any real world performance gains anyway.  Placebo effect 
perhaps...  ;o)

Mike A. Harris

More information about the test mailing list