Re: gdb backtrace for g++ programs showing ?? symbol

[Paul Pluzhnikov]

naam.yaad.nahin@gmail.com writes:

> The debug versions are installed under /usr/lib/debug.

There are several possibilities:

1. You may have installed a version of libstdc++.so.N compiled with
   '-g' into /usr/lib/debug.

   In that case, you should be able to use that version with 
   'export LD_LIBRARY_PATH=/usr/lib/debug'

2. You may have installed libstdc++.N.debug debug-info file into
   /usr/lib/debug/usr/lib/...

   In that case, all you should need is a version of gdb that
   understands debuginfo files.

Since you referred to RH-7.2 errata, it sounds like you are using
an ancient system, so option 1 is the more likely one.

>>  http://catb.org/esr/faqs/smart-questions.html
>
> And here I was asking questions after reading the above ... :-)

;-)

> Anyway, I recognize that my questions were too vague. The full details
> are too complicated (at least for me) and I am therefore trying to
> reproduce the behavior using a simpler test case.

You should at least post the gdb backtrace -- it may not give you any
clues, but may give someone else a good idea of what you are seeing.

> However, I had seen that gcc 2.96 had some problems with handling
> exceptions. I was therefore wondering what were the exact problems
> (since the Redhat Errata is too vague) and whether anyone in this
> group had seen/known the specific gcc 2.96 problem.

The 'gcc-2.96' generally works fine, but there are weird corner
cases where it doesn't. [I know such corner cases exist, but I've
never seen one myself.] Even if someone reading this group has seen
such a corner case, you didn't give any details which would enable
that person to tell you that you are likely seeing the same corner
case, and point you to a bug report or a workaround.

Cheers,
-- 
In order to understand recursion you must first understand recursion.
Remove /-nsp/ for email.