On Thu, 2004-01-29 at 16:19, Andrew Cagney wrote:
> > Andrew Cagney writes:
> >  > Hello,
> >  > 
> >  > Ref: RFA symtab: Fix for PR c++/1267 ("next" and shared libraries)
> >  > http://sources.redhat.com/ml/gdb-patches/2003-07/msg00354.html
> >  > 
> >  > The change unfortunatly broke IRIX 6.5's host compiler which is using 
> >  > mdebugread :-(  That code was looking for a symbol in the absolute 
> >  > section "*ABS*" but the PR/1267 change was causing *ABS* symbols to be 
> >  > ignored (find_pc_section didn't return an absolute section).
> >  > 
> >  > Since the underlying problem with PR/1267 was with the frame code 
> >  > needing a minimal symbol that was in the same section as the frame's PC, 
> >  > and that code [indirectly] calls lookup_minimal_symbol_by_pc, I moved 
> >  > the find_pc_section call to that function.
> >  > 
> >  > Tested on i386 GNU/Linux (dwarf 2) with no regressions.
> >  > Tested on PPC NetBSD (stabs) with no regressions.
> >  > Tested on IRIX and all the warnings, and many failures, disappeared.
> >  > See gdb/1519 for how to exercise the bug.
> >  > 
> >  > ok?
> > 
> > ok. Does it still fix shlibs/1237 and shlibs/1280 too? Adam, could you
> > check please?
> 
> Isn't that covered by the testsuite?  Sigh.
> 


This is probably ugly, but it shows the basic idea...