gdbserver, pthreads and SIGPWR

[Daniel Jacobowitz]

On Thu, Mar 07, 2002 at 12:54:40PM -0500, Andrew Cagney wrote:
> >Hello,
> >i am using gdb 5.1.1 to remote debug my programs.
> >whenever i create a new thread with pthread_create(), a SIGPWR is caught.
> >
> >Why does this happen when i do remote debugging, but not, when i use gdb
> >directly?
> >What should i do when the signal arrives?
> >Can i continue savely?
> 
> (I'm guessing a remote GNU/Linux target).
> 
> Try:
> 	(gdb) help handle
> 	(gdb) handle SIGPWR pass nostop noprint
> I suspect the SIGPWR is being used as part of the thread implementation. 
>  You're probably also going to find that threads don't work very well 
> remotely, I don't know of many debug agent implementations that support 
> threads.

SIGPWR is some random low-numbered signal.

TARGET_SIGNAL_SIGPWR is 32.

This is actually SIG32, the thread management signal.


And what Andrew said about debugging threads remotely is true, at least
for now.  It will not work well.

-- 
Daniel Jacobowitz                           Carnegie Mellon University
MontaVista Software                         Debian GNU/Linux Developer