gcc ch5 problem - still

Robert Connolly robert at linuxfromscratch.org
Wed Dec 14 14:54:36 PST 2005

You don't have ssp guard functions in libc. Either you're still linking 
to /lib because you didn't use the gcc x86_specs patch, or you disabled it in 
uclibc or didn't use the uclibc config patch. Check the test in the adjusting 
toolchain page and double check you're linking to /tools/lib. If you are then 
double check that you used all the uclibc patches.


On December 14, 2005 11:56 am, Declan Moriarty wrote:
> Now I can reliably get to the build of test.c with everythinhg
> looking right. test.c fails as follows
> hlfs:/mnt/hlfs/sources/gcc-build$ gcc -o test test.c
> Aborted
> hlfs:/mnt/hlfs/sources/gcc-build$
> Or, when done as root....
> root:/mnt/hlfs/sources/gcc-build#gcc -o test test.c
> /tmp/cctsbP82.o(.text+0x3d): In function `main':
> : undefined reference to `__guard'
> collect2: ld returned 1 exit status
> root:/mnt/hlfs/sources/gcc-build#
> and I get these things left over in /tmp
> cc2yST6J.s  cca5VIid.o  cciIpf8f.o  ccMkbfVd.o  ccSQ6hBr.o
> cc5T6rUE.s  ccDkVPLa.s  ccKElpYl.o  ccQ5pVVh.s  ccYpiZmI.s
> apparently made by the compiler. They are recent.
> I grepped the patches for __guard and found it in gcc and uClibc
> and went back over them, but all remains. I am pretty sure all
> sections of all patches applied.
> /tmp/cctsbP82.o was obviously deleted, but some others were not,
> which indicates some malfunction along the way?
> --
> 	With best Regards,
> 	Declan Moriarty.

More information about the hlfs-dev mailing list