Re: gdb runs multi-threaded with non-threaded app and gets SEGV

[Larry Martell <larry at software-horizons dot com>]

--- "Min Xu (Hsu)" <xu@cs.wisc.edu> wrote:

> Larry,
> 
> How about you run the program and let the program
> pass the
> lib call without using GDB. And spin there until gdb
> is attached?

I tried this, but after it attaches to the process,
as then I try to continue it, it immediatly gets
a SEGV. And it sounded like such a good idea too :-(

Thanks!
-larry

> On Fri, 15 Jul 2005 Larry Martell wrote :
> > --- Ramana Radhakrishnan
> > <ramana.radhakrishnan@codito.com> wrote:
> > > On Fri, 2005-07-15 at 11:25 -0700, Larry Martell
> > > wrote:
> > > > I have a program that has 1 line of code - a
> call
> > > to
> > > > a third party libary function:
> > > >
> > > > #include "ni488.h"
> > > >
> > > > int main() {
> > > >         int handle = ibdev(0, 7, 0, 10, 1, 0);
> > > > }
> > >
> > >
> > > That is not enough info for us to help you with
> .
> > > Try CVS head or a more
> > > recent version of gdb and see if that works.
> >
> > OK, I've upgraded to gdb 6.3
> >
> > > Whats the version of libc ,
> >
> > 2.3.3
> >
> > > gcc
> >
> > 3.4.1
> >
> > > Also maybe your 3rd party
> > > library might be multithreaded for all we know.
> Do a
> > > ldd on the library if its a .so and check .
> >
> > OK, an ldd showed:
> >
> > libpthread.so.0 => /lib/tls/libpthread.so.0
> > (0x40278000)
> >
> > So apparently the 3rd party lib is multithreaded.
> >
> > So that answers 1 question. But the more important
> one
> > is why I get the SEGV in gdb, but not when I run
> the
> > program without gdb?
> >
> > Here's the deal - I am developing an app (that is
> > not getting a SEGV) and I wanted to debug it. When
> > I ran the app in gdb it got the SEGV on my call to
> > ibdev. Thinking I probably had a memory corruption
> > somewhere in my application I ran valgrind on it.
> > It said the 3rd party lib was attempting to jump
> > to address 0. Still thinking it was my issue, I
> > created the single line test program (shown at the
> > top of this message). And that's where I am now -
> I
> > need gdb to get past this call so I can debug my
> > actaul app.
> >
> > Thanks!
> > -larry
> >
> >
> > > > The program is not multi-threaded nor does not
> > > link
> > > > with any multi-threaded libs. It's built like
> > > this:
> > > >
> > > > gcc -g -o nitest nitest.c -lgpibapi
> > > >
> > > > If I run the program outside of the debugger
> it
> > > does
> > > > not crash.
> > > >
> > > > If I run it from within gdb here's what I get:
> > > >
> > > > $ gdb nitest
> > > > GNU gdb 6.2-2mdk (Mandrakelinux)
> > > > Copyright 2004 Free Software Foundation, Inc.
> > > > GDB is free software, covered by the GNU
> General
> > > > Public License, and you are
> > > > welcome to change it and/or distribute copies
> of
> > > it
> > > > under certain conditions.
> > > > Type "show copying" to see the conditions.
> > > > There is absolutely no warranty for GDB.  Type
> > > "show
> > > > warranty" for details.
> > > > This GDB was configured as
> > > > "i586-mandrake-linux-gnu"...Using host
> > > libthread_db
> > > > library "/lib/tls/libthread_db.so.1".
> > > >
> > > > (gdb) start
> > > > Breakpoint 1 at 0x8048398: file nitest.c, line
> 4.
> > > > Starting program:
> /home/martell/src/nitest2/nitest
> > > > [Thread debugging using libthread_db enabled]
> > > > [New Thread 1076484320 (LWP 9873)]
> > > > [Switching to Thread 1076484320 (LWP 9873)]
> > > > main () at nitest.c:4
> > > > 4               int handle = ibdev(0, 7, 0,
> 10, 1,
> > > 0);
> > > > (gdb) c
> > > > Continuing.
> > > > [New Thread 1085078448 (LWP 9876)]
> > > > [New Thread 1093471152 (LWP 9877)]
> > > >
> > > > Program received signal SIGSEGV, Segmentation
> > > fault.
> > > > [Switching to Thread 1085078448 (LWP 9876)]
> > > > 0x4005d609 in ?? () from
> > > > /usr/local/lib/libgpibapi.so.2
> > > >
> > > > Can anyone tell me why gdb runs this as
> > > multi-threaded
> > > > and why I get a SEGV in gdb, but I do not when
> I
> > > run
> > > > it outside of gdb?
> > > >
> > > > Thanks!
> > > > -larry
> > >
> > >
> > >
> >
> 
>