Arrgh! Broken compiler, I think. How to fix?

Jeremy Henty jeremy at chaos.org.uk
Sat Jan 3 13:05:45 PST 2004


No reply from the list so I ploughed on ahead, recompiled gcc and
things *seem* to work now.  I'd still appreciate advice in case I have
merely stored up more trouble for myself later.

I wrote: 

> All looked fine until I tried to compile my first Qt application
> (Muse) when uic started segfaulting during the build.  I searched
> the archives for "qt uic segfault" and it looks like my compiler is
> broken.  If I try the "readelf -s crtbeginS.o | grep HIDDEN" test
> described in
> http://archives.linuxfromscratch.org/mail-archives/blfs-support/2003-January/034740.html
> it fails: I get no output instead of a reference to __dso_handle .

I suspect I did chapter 5 of LFS 4.1 with binutils 2.13.2.1 (I didn't
know to FBBG then) and fell foul of the bug Gerard identified last
January (gcc's configure mis-parses the version string).  I have
binutils 2.13.2 now, but unfortunately chapter 6 recompiles gcc first,
so it didn't get fixed.  This time round the configure detects hidden
support correctly and "readelf -s crtbeginS.o" reports a __dso_handle
symbol.  Qt and MusE now compile OK.

So am I in the clear now or is there more fixing to do?

Regards, 

Jeremy Henty



More information about the lfs-support mailing list