Can threaded program be remotely debugged using Insight ?

Can threaded program be remotely debugged using Insight ?

[Masahide Tomita]

Hi,

This may be FAQ, but I couldn't really determine what is the solution for
this problem, when I searched on the web.

I'm trying to use Insight to remotely debug multi-threaded program.
Host machine and target board are both Redhat 7.2 system running on Intel
CPU.
I tryied to debug a program using thread, so I started gdbserver and gdb
(Insight) like below:

[target]
gdbserver ***.***.***.***:1001 program program.c

[host]
/usr/local/bin/gdb program

They start normally, and run normally, but when program gets to the function
pthread_create, insight on host machine freezes.
Well, it looks like still working since mouse cursol get "clock" icon, but
cpu load level is almost zero, and no matter how long you wait you get no
response from insight.

And I get the message like below on target console.

GDBbug : target.c (target_signal_from_host) : Unrecognized real-time signal


So, my real questions are:

1. Is this a problem of insight, or gdb?
2. Is there any way to debug threded program by using gdb, or insight?

Any help or information is appreceated.

Thank you.


--

Masahide Tomita

fsi_tomi@bcpc601.neec.abk.nec.co.jp

Re: Can threaded program be remotely debugged using Insight ?

[Keith Seitz]

On Wed, 9 Oct 2002, Masahide Tomita wrote:

> So, my real questions are:
> 
> 1. Is this a problem of insight, or gdb?

Can you run this with just GDB (gdb -nw)? It sounds like a gdb bug, but 
running "-nw" will tell us for sure.

> 2. Is there any way to debug threded program by using gdb, or insight?

I think that the real problem may be the warning you saw, but we'll get to 
that after we determine whether Insight is at fault.

Keith

Re: Can threaded program be remotely debugged using Insight ?

[Masahide Tomita]

Hi Keith,

Thank you for the reply.
Anyway, how come mail from you everytime arrive twice?



> On Wed, 9 Oct 2002, Masahide Tomita wrote:
>
> > So, my real questions are:
> >
> > 1. Is this a problem of insight, or gdb?
>
> Can you run this with just GDB (gdb -nw)? It sounds like a gdb bug, but
> running "-nw" will tell us for sure.

I tried this one, and got same result, as long as I try to debug remotely.
Debugging locally goes more than just fine, even with GUI mode.


> > 2. Is there any way to debug threded program by using gdb, or insight?
>
> I think that the real problem may be the warning you saw, but we'll get to
> that after we determine whether Insight is at fault.

I locally run gdb (not Insight) on target, to see if this is target oriented
issue, and I get thread created and program going, but I get the message
below now:

Program received signal SIG32, Real-time event 32.
0x4005db75 in sigsuspend( ) from /lib/libc.so.6

These, I don't get from gdb on PC.
GDB versions of both PC and target are same.

I can't even guess what the problem is now.

Somebody please help me.


Thank you.

--
Masahide Tomita

Re: Can threaded program be remotely debugged using Insight ?

[Keith Seitz]

On Thu, 10 Oct 2002, Masahide Tomita wrote:

> Thank you for the reply.
> Anyway, how come mail from you everytime arrive twice?

If you're subscribed to the mailing list, you'd get my reply via that 
route as well as from me directly.

> I locally run gdb (not Insight) on target, to see if this is target oriented
> issue, and I get thread created and program going, but I get the message
> below now:
> 
> Program received signal SIG32, Real-time event 32.
> 0x4005db75 in sigsuspend( ) from /lib/libc.so.6

Ok, this is definitely a gdb or gdbserver issue, then. This subject seems 
vaguely familiar: I've seen this discussed recently somewhere. I would 
recommend sending a gdb session showing the error to 
gdb@sources.redhat.com, where the gdb experts hang out.

Keith

Re: Can threaded program be remotely debugged using Insight ?

[Cal Erickson]

Current versions of gdbserver are incapable of debugging threaded 
programs. They do not have enough knowledge of how threads are set up 
and used. There are projects which are trying to remedy this. An 
interesting feature in the 5.2 and later versions of gdbserver is the 
ability to attach to a running process. However make note of the fact 
that I used the word process and not thread.

Cal Erickson

Keith Seitz wrote:

> On Thu, 10 Oct 2002, Masahide Tomita wrote:
> 
> 
>>Thank you for the reply.
>>Anyway, how come mail from you everytime arrive twice?
>>
> 
> If you're subscribed to the mailing list, you'd get my reply via that 
> route as well as from me directly.
> 
> 
>>I locally run gdb (not Insight) on target, to see if this is target oriented
>>issue, and I get thread created and program going, but I get the message
>>below now:
>>
>>Program received signal SIG32, Real-time event 32.
>>0x4005db75 in sigsuspend( ) from /lib/libc.so.6
>>
> 
> Ok, this is definitely a gdb or gdbserver issue, then. This subject seems 
> vaguely familiar: I've seen this discussed recently somewhere. I would 
> recommend sending a gdb session showing the error to 
> gdb@sources.redhat.com, where the gdb experts hang out.
> 
> Keith
> 
> 
> 
> 


-- 
===========================================================================
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
===========================================================================