Re: cli/1056: divide by zero hangs gdb

[Michael Elizabeth Chastain <mec at shout dot net>]

The following reply was made to PR cli/1056; it has been noted by GNATS.

From: Michael Elizabeth Chastain <mec@shout.net>
To: gdb-gnats@sources.redhat.com
Cc:  
Subject: Re: cli/1056: divide by zero hangs gdb
Date: Thu, 28 Aug 2003 04:07:22 -0400

 I tried messing with handle_sigfpe:
 
   /* old */
   static void
   handle_sigfpe (int sig)
   {
     mark_async_signal_handler_wrapper (sigfpe_token);
     signal (sig, handle_sigfpe);
   }
 
   /* new */
   static void
   handle_sigfpe (int sig)
   {
     signal (sig, handle_sigfpe);
     error ("Erroneous arithmetic operation.");
   }
 
 Works for me.  But I don't know if it's architecturally okay to
 call 'error' from a signal handler.  We really have little choice,
 though.  We can either:
 
   . not have a signal handler for handle_sigfpe
   . longjmp out of the signal handler
   . do some nasty os-specific and machine-instruction-specific code reading
     to advance the program counter past the faulty instruction
 
 I need to talk to someone about these async signal handler wrapper
 thingies and why gdb can't use one for SIGFPE.
 
 Michael C