LD_PRELOAD, SIGTRAP and step/next

LD_PRELOAD, SIGTRAP and step/next

[David Lecomber]

Dear all,

I am debugging using a memory checking library -- dmalloc -- this does
the usual things via an LD_PRELOAD over the malloc.  I am getting
SIGTRAP when trying to call a function in this library -- but only after
a step or a next; after a user-defined breakpoint behaviour is correct!

Can anyone explain/resolve the behaviour:

(gdb) set environment LD_PRELOAD=lib/libdmalloc.so
(gdb) b main
(gdb) run
Starting program: /home/david/projects/code/ddt/examples/a.out

Breakpoint 1, main (argc=1, argv=0xbffff604, environ=0xbffff60c) at
hello.c:65
65              int tag = 50;           /* Tag for messages */
(gdb) print dmalloc_message("hello")ne}
void
(gdb) step
79              t2 = malloc(sizeof(typeThree));
(gdb) p dmalloc_message ("ho")
Program received signal SIGTRAP, Trace/breakpoint trap.
0xb7f831c3 in dmalloc_message (format=0x0) at malloc.c:2079
2079    {
The program being debugged was signaled while in a function called from
GDB.
GDB remains in the frame where the signal was received.
To change this behavior use "set unwindonsignal on"
Evaluation of the expression containing the function (dmalloc_message) 
[..]

Very odd indeed!


d.

-- 
David Lecomber <david@allinea.com>

Re: LD_PRELOAD, SIGTRAP and step/next

[Daniel Jacobowitz]

On Thu, Mar 17, 2005 at 12:44:15PM +0000, David Lecomber wrote:
> Dear all,
> 
> I am debugging using a memory checking library -- dmalloc -- this does
> the usual things via an LD_PRELOAD over the malloc.  I am getting
> SIGTRAP when trying to call a function in this library -- but only after
> a step or a next; after a user-defined breakpoint behaviour is correct!

Kernel version?  This sounds like a Linux kernel bug that has since
been fixed...

-- 
Daniel Jacobowitz
CodeSourcery, LLC

Re: LD_PRELOAD, SIGTRAP and step/next

[David Lecomber]

On Thu, 2005-03-17 at 09:12 -0500, Daniel Jacobowitz wrote:
> On Thu, Mar 17, 2005 at 12:44:15PM +0000, David Lecomber wrote:
> > Dear all,
> > 
> > I am debugging using a memory checking library -- dmalloc -- this does
> > the usual things via an LD_PRELOAD over the malloc.  I am getting
> > SIGTRAP when trying to call a function in this library -- but only after
> > a step or a next; after a user-defined breakpoint behaviour is correct!
> 
> Kernel version?  This sounds like a Linux kernel bug that has since
> been fixed...

2.6.10-1.770_FC3smp

so pretty recent!

d,
-- 
David Lecomber <david@allinea.com>

Re: LD_PRELOAD, SIGTRAP and step/next

[Daniel Jacobowitz]

On Thu, Mar 17, 2005 at 02:15:22PM +0000, David Lecomber wrote:
> On Thu, 2005-03-17 at 09:12 -0500, Daniel Jacobowitz wrote:
> > On Thu, Mar 17, 2005 at 12:44:15PM +0000, David Lecomber wrote:
> > > Dear all,
> > > 
> > > I am debugging using a memory checking library -- dmalloc -- this does
> > > the usual things via an LD_PRELOAD over the malloc.  I am getting
> > > SIGTRAP when trying to call a function in this library -- but only after
> > > a step or a next; after a user-defined breakpoint behaviour is correct!
> > 
> > Kernel version?  This sounds like a Linux kernel bug that has since
> > been fixed...
> 
> 2.6.10-1.770_FC3smp
> 
> so pretty recent!

That's the one.  Try an even more recent kernel.

(The bug wasn't very long lived; it was introduced and fixed between
2.6.10 and 2.6.11.  I'm getting the feeling that FC3 picked it up
however.)

-- 
Daniel Jacobowitz
CodeSourcery, LLC