Glibc-2.3.3-lfs-5.1? (was Re: QLFS project announcement)

Erik-Jan ej.lfs at
Thu Sep 9 11:35:27 PDT 2004

Robert Connolly wrote:
> I haven't had anything good happen with gcc-3.4.1, I'm not sure why its so 
> picky... glibc doesn't like it. I think switching back to hjl binutils would 

I had a problem with the glibc-2.3.3-ssp_frandom-2.patch. It worked ok 
with gcc-3.3, but with gcc-3.4 it didn't define __guard and 
__stack_smash_handler as global but as local symbols.

make gcc-ch5-pass2 then barfed with "checking for library containing 
gettext... configure: error: Link tests are not allowed after 

Looking at config.log showed that the linker failed because it couldn't 
find __guard etc, which made configure think it was building 
non-executable binaries (or something like that) and therefor configure 
gave an error.

After a long search I found something: in the 
glibc-2.3.3-ssp_frandom-2.patch, there is a hunk patching 

diff -Naur glibc-2.3-20040418.orig/sysdeps/unix/sysv/linux/Versions 
--- glibc-2.3-20040418.orig/sysdeps/unix/sysv/linux/Versions	2004-03-19 
00:13:55.000000000 +0000
+++ glibc-2.3-20040418.ssp-frandom/sysdeps/unix/sysv/linux/Versions 
2004-04-25 07:08:58.000000000 +0000
@@ -108,6 +108,7 @@
    GLIBC_2.3.2 {
      # New kernel interfaces.
      epoll_create; epoll_ctl; epoll_wait;
+	__guard; __guard_setup; __stack_smash_handler;
    GLIBC_2.3.3 {
      gnu_dev_major; gnu_dev_minor; gnu_dev_makedev;

So the problem seemed to be here. There was only one thing I could find: 
  the indention before __guard was a tab. I replaced it with some spaces 
and after that, gcc-3.4 could build a ssp-glibc. Doing the checks for 
ssp, it even worked! :))

I'm not familiar with all the details of ssp, and I have no idea why the 
tab didn't work and the spaces did. But for anyone running into the same 
problem, I've created glibc-2.3.3-ssp_frandom-3.patch, with my tiny fix.


-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: glibc-2.3.3-ssp_frandom-3.patch
URL: <>

More information about the hlfs-dev mailing list