GDB is not trapping SIGINT

GDB is not trapping SIGINT

[Andy Howell]

Hello,

I'm having problems getting GDB to not pass SIGINT to my program. In my 
main thread I do a sigwait for SIGINT. When I'm debugging, I don't want 
this called. "info handle" shows that SIGINT should not be being passed 
to my app.

I using GDB 6.3 on RH Linux.

Any ideas?

	Andy

Re: GDB is not trapping SIGINT

[Daniel Jacobowitz]

On Sun, Apr 02, 2006 at 11:58:59PM -0500, Andy Howell wrote:
> Hello,
> 
> I'm having problems getting GDB to not pass SIGINT to my program. In my 
> main thread I do a sigwait for SIGINT. When I'm debugging, I don't want 
> this called. "info handle" shows that SIGINT should not be being passed 
> to my app.
> 
> I using GDB 6.3 on RH Linux.

But how are you using it? :-)  An example, please.

You also did not mention what kernel version you are using; at least
one sigwait-related kernel bug has been reported.  And it looks like
it is still present in current 2.6 kernels, so you may want to report
this to the kernel developers.  sys_rt_sigtimedwait calls
dequeue_signal, but never lets ptrace see it.  Maybe the bits from
get_signal_to_deliver can be moved into dequeue_signal now.

-- 
Daniel Jacobowitz
CodeSourcery

Re: GDB is not trapping SIGINT

[Andy Howell]

Daniel Jacobowitz wrote:
> On Sun, Apr 02, 2006 at 11:58:59PM -0500, Andy Howell wrote:
>> Hello,
>>
>> I'm having problems getting GDB to not pass SIGINT to my program. In my 
>> main thread I do a sigwait for SIGINT. When I'm debugging, I don't want 
>> this called. "info handle" shows that SIGINT should not be being passed 
>> to my app.
>>
>> I using GDB 6.3 on RH Linux.
> 
> But how are you using it? :-)  An example, please.

 > You also did not mention what kernel version you are using; at least
 > one sigwait-related kernel bug has been reported.  And it looks like
 > it is still present in current 2.6 kernels, so you may want to report
 > this to the kernel developers.  sys_rt_sigtimedwait calls
 > dequeue_signal, but never lets ptrace see it.  Maybe the bits from
 > get_signal_to_deliver can be moved into dequeue_signal now.
 >

Daniel,

Sorry for previous lack of details. It does seem to be Linux specific. I 
see the same behavior in Redhat Linux 2.4.21-4.EL and 2.6.9-22.0.2.EL, 
as well as Fedora 3 2.6.12. On solaris 2.8 it works fine.

It sounds like without a kernel patch, I'm out of luck. Thanks for your 
help.

For what its worth, my program first creates a signal set, block new 
threads from seeing them, and then the main thread waits for a signal:

	::sigemptyset( &set );
	::sigaddset( &set, SIGINT );
	::sigaddset( &set, SIGQUIT );
	::sigaddset( &set, SIGUSR2 );
	.... more signals

	// Block new threads from getting signals
	pthread_sigmask( SIG_BLOCK &set, NULL );

	// Create a bunch of threads
	.....

	// wait for signal
	
	do {
	  sigErr = sigwait( &set, &sig );
  	  if( sig == SIGUSR2 ) {
	    doSomething(...)
	  }
         while( sig = SIGUSR2 )

	// Do cleanup and exit

Thanks,

	Andy