RE: gdbserver multithreaded debugging problem

RE: gdbserver multithreaded debugging problem

[nak26]

>===== Original Message From cerickson@mvista.com =====
>Nik,
>The files you need are in the target/lib directory
>
>libthread_db.so.1  link to libthread_db-1.0.so
>libthread_db-1.0.so
>
>and the target/usr/lib directory
>
>libthread_bd.a
>libthread_db.so  link to ../../lib/libthread_db.so.1
>libthread_db_p.a
>
>Then things should just be AOK. Now if you stripped your executables
>that are running on your target then you have another problem. So
>be advised that while in debug do not strip any threaded programs.
>
>Cal

Cal:

Sorry to be bugging you again, but I still haven't had success resolving the 
issue. I have the links set up as you perscribed, and my lib path includes 
/usr/lib (I also tried it including both /usr/lib and /lib in this order and 
vice versa) but I'm still seeing the SIG32  messages.

----------------------------------------------------------
Here is my setup:
------------------------------------------------------------------------------
-------
lab@(none):/usr/lib$ ls -l | grep libthread
-rwxr-xr-x    1 root     root        95174 Mar 21 05:19 libthread.so
-rw-r--r--    1 root     root      2197690 Mar 21 04:06 libthread_db.a
lrwxrwxrwx    1 root     root           27 Jun 24 13:36 libthread_db.so -> 
../../lib/libthread_db.so.1
-rwxr-xr-x    1 root     root        25911 Mar 21 07:33 libthread_db.so.1.0.20
-rw-r--r--    1 root     root      2199126 Mar 21 04:06 libthread_db_p.a
-rwxr-xr-x    1 root     root       301667 Mar 21 05:20 libthreaddbg.so
lab@(none):/usr/lib$

lab@(none):/lib$ ls -l | grep libthread
-rw-r--r--    1 root     root       935480 Mar 21 04:06 libthread_db-1.0.so
lrwxrwxrwx    1 root     root           19 May  2 16:00 libthread_db.so.1 -> 
libthread_db-1.0.so
lab@(none):/lib$

lab@(none):~$ env | grep LD_LIB
LD_LIBRARY_PATH=/usr/lib:/lib:/opt/intel/mkl/lib/32
lab@(none):~$

--------------------------------------------------------------------

Per Daniel's suggestion I looked at the config.log, and multithreading seems 
to be set up correctly prior to the build.

I guess I can try debugging a multithreaded toy example, or hopefully you'll 
see something else that I'm doing wrong.

Thanks,
Nik

Re: gdbserver multithreaded debugging problem

[Daniel Jacobowitz]

On Mon, Jun 23, 2003 at 03:39:06PM -0400, nak26 wrote:
> Hi:
> 
> I am trying to use gdbserver to remotely debug a multithreaded application.
> 
> Remote host is a IA32, running MontaVista Linux 3.0 kernel v 2.4.18; gdbserver 
> from the 5.3 release tarball, patched with the PREPARE_TO_PROCEED patch, and 
> built on the remote host.
> 
> Local host is an IA32 machine, running RedHat 8.0, gdb v 5.2.1-4.
> 
> I get a "Real-time event 32"-signal as each thread is spun up, execution 
> stops, and I have to re-continue. After all the threads are created and I hit 
> a breakpoint, info thread only displays the main thread.
> 
> I don't see this problem while debugging with gdb (5.2.1) on the remote (and 
> the local) target
> .
> Is there anything I can do to solve the problem I am experiencing with 
> gdbserver?

Did gdbserver detect thread support when you built it?  Check the output
of configure.

-- 
Daniel Jacobowitz
MontaVista Software                         Debian GNU/Linux Developer

Re: gdbserver multithreaded debugging problem

[Cal Erickson]

Nik,
You are missing libthread_db.* in your LIBRAY PATH. If you had these
then you would not see the the SIG 32 messages.

Cal Erickson

nak26 wrote:

> Hi:
> 
> I am trying to use gdbserver to remotely debug a multithreaded application.
> 
> Remote host is a IA32, running MontaVista Linux 3.0 kernel v 2.4.18; gdbserver 
> from the 5.3 release tarball, patched with the PREPARE_TO_PROCEED patch, and 
> built on the remote host.
> 
> Local host is an IA32 machine, running RedHat 8.0, gdb v 5.2.1-4.
> 
> I get a "Real-time event 32"-signal as each thread is spun up, execution 
> stops, and I have to re-continue. After all the threads are created and I hit 
> a breakpoint, info thread only displays the main thread.
> 
> I don't see this problem while debugging with gdb (5.2.1) on the remote (and 
> the local) target
> .
> Is there anything I can do to solve the problem I am experiencing with 
> gdbserver?
> 
> Here's gdb's output when debugging remotely:
> --------------------------------------------
> 
> (gdb) target remote 192.168.139.14:8888
> Remote debugging using 192.168.139.14:8888
> 0x40001390 in ?? ()
> (gdb) b osevent.cxx:215
> Breakpoint 1 at 0x826dd03: file ../../common/osevent.cxx, line 215.
> (gdb) cont
> Continuing.
> 
> Program received signal SIG32, Real-time event 32.
> 0x404b724d in ?? ()
> (gdb) c
> Continuing.
> 
> Program received signal SIG32, Real-time event 32.
> 0x404b724d in ?? ()
> (gdb)
> Continuing.
> 
> Program received signal SIG32, Real-time event 32.
> 0x404b724d in ?? ()
> (gdb)
> Continuing.
> 
> Program received signal SIG32, Real-time event 32.
> 0x404b724d in ?? ()
> (gdb)
> Continuing.
> 
> Program received signal SIG32, Real-time event 32.
> 0x404b724d in ?? ()
> (gdb)
> Continuing.
> 
> Program received signal SIG32, Real-time event 32.
> 0x404b724d in ?? ()
> (gdb)
> Continuing.
> 
> Program received signal SIG32, Real-time event 32.
> 0x404b724d in ?? ()
> (gdb)
> Continuing.
> 
> Breakpoint 1, OsEventManagerClass::eventLoop() (this=0x89caf78) at 
> ../../common/osevent.cxx:215
> 215           if (pTimer) {
> (gdb) info thread
>   1 Thread 3860  OsEventManagerClass::eventLoop() (this=0x89caf78) at 
> ../../common/osevent.cxx:215
> 
> ----------------------------------------------
> 
> Thanks,
> --Nik
> 
> 
> 


-- 
===========================================================================
Cal Erickson                 MontaVista Software Inc.
Linux Consultant             1237 E. Arques Ave.
Phone (408) 328-0304         Sunnyvale CA 94085
Fax   (408) 328-9204         web http://www.mvista.com
===========================================================================

gdbserver multithreaded debugging problem

[nak26]

Hi:

I am trying to use gdbserver to remotely debug a multithreaded application.

Remote host is a IA32, running MontaVista Linux 3.0 kernel v 2.4.18; gdbserver 
from the 5.3 release tarball, patched with the PREPARE_TO_PROCEED patch, and 
built on the remote host.

Local host is an IA32 machine, running RedHat 8.0, gdb v 5.2.1-4.

I get a "Real-time event 32"-signal as each thread is spun up, execution 
stops, and I have to re-continue. After all the threads are created and I hit 
a breakpoint, info thread only displays the main thread.

I don't see this problem while debugging with gdb (5.2.1) on the remote (and 
the local) target
.
Is there anything I can do to solve the problem I am experiencing with 
gdbserver?

Here's gdb's output when debugging remotely:
--------------------------------------------

(gdb) target remote 192.168.139.14:8888
Remote debugging using 192.168.139.14:8888
0x40001390 in ?? ()
(gdb) b osevent.cxx:215
Breakpoint 1 at 0x826dd03: file ../../common/osevent.cxx, line 215.
(gdb) cont
Continuing.

Program received signal SIG32, Real-time event 32.
0x404b724d in ?? ()
(gdb) c
Continuing.

Program received signal SIG32, Real-time event 32.
0x404b724d in ?? ()
(gdb)
Continuing.

Program received signal SIG32, Real-time event 32.
0x404b724d in ?? ()
(gdb)
Continuing.

Program received signal SIG32, Real-time event 32.
0x404b724d in ?? ()
(gdb)
Continuing.

Program received signal SIG32, Real-time event 32.
0x404b724d in ?? ()
(gdb)
Continuing.

Program received signal SIG32, Real-time event 32.
0x404b724d in ?? ()
(gdb)
Continuing.

Program received signal SIG32, Real-time event 32.
0x404b724d in ?? ()
(gdb)
Continuing.

Breakpoint 1, OsEventManagerClass::eventLoop() (this=0x89caf78) at 
../../common/osevent.cxx:215
215           if (pTimer) {
(gdb) info thread
  1 Thread 3860  OsEventManagerClass::eventLoop() (this=0x89caf78) at 
../../common/osevent.cxx:215

----------------------------------------------

Thanks,
--Nik