infttrace vs add_thread

infttrace vs add_thread

[law]

When add_thread was changed to accept a ptid_t type instead of just an
integer, nobody updated infttrace.c to reflect those changes.

The net result is a segfault if you try to use GDB on hpux11.

Again, since I'm no longer a GDB maintainer, I'll hold off installing the
fix until approved by a GDB maintainer.


	* infttrate.c (child_acknowledge_created_inferior): Pass
	correct argument to add_thread.
	(update_thread_state_after_attach): Likewise.
Index: infttrace.c
===================================================================
RCS file: /cvs/cvsfiles/devo/gdb/infttrace.c,v
retrieving revision 2.22
diff -c -3 -p -r2.22 infttrace.c
*** infttrace.c	2001/09/30 19:16:20	2.22
--- infttrace.c	2001/12/18 17:50:01
*************** child_acknowledge_created_inferior (int 
*** 3123,3129 ****
     *    the process safely to ask what it is.  Anyway, we'll
     *    add it when it gets the EXEC event.
     */
!   add_thread (pid);		/* in thread.c */
  
    /* We can now set the child's ttrace event mask.
     */
--- 3123,3129 ----
     *    the process safely to ask what it is.  Anyway, we'll
     *    add it when it gets the EXEC event.
     */
!   add_thread (pid_to_ptid (pid));		/* in thread.c */
  
    /* We can now set the child's ttrace event mask.
     */
*************** update_thread_state_after_attach (int pi
*** 4764,4770 ****
  	    }
  	}
  
!       add_thread (tid);		/* in thread.c */
      }
  
  #ifdef PARANOIA
--- 4764,4770 ----
  	    }
  	}
  
!       add_thread (pid_to_ptid (pid));		/* in thread.c */
      }
  
  #ifdef PARANOIA

Re: infttrace vs add_thread

[Andrew Cagney]

> When add_thread was changed to accept a ptid_t type instead of just an
> integer, nobody updated infttrace.c to reflect those changes.
> 
> The net result is a segfault if you try to use GDB on hpux11.
> 
> Again, since I'm no longer a GDB maintainer, I'll hold off installing the
> fix until approved by a GDB maintainer.


But you do have write after approval.  Cleanups like this are considered 
obvious.

Andrew

Re: infttrace vs add_thread

[law]

  > > Again, since I'm no longer a GDB maintainer, I'll hold off installing the
  > > fix until approved by a GDB maintainer.
  > 
  > 
  > But you do have write after approval.  Cleanups like this are considered 
  > obvious.
OK.  I've been out of the GDB development process for a long time -- I don't
want to step on any toes.  Meaning I'll probably be a little over-conservative
about this kind of stuff.

jeff