This is the mail archive of the gdb-patches@sources.redhat.com mailing list for the GDB project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

[PATCH RFA] lin-lwp.c pending events.


Mark, even though we are 'co-maintainers' of this code, 
I do like to get your approval and feedback.  Linus threads
are so horrendously complicated, it's good to have two 
pairs of eyes.

This change catches the somewhat rare circumstance where GDB 
calls target_resume with a specific PTID, but then calls 
target_wait with a wild-card.  You could argue that GDB
shouldn't do that, but it turns out that it would be very
difficult to prevent it.  When it happens, we want to 
make sure that we don't try to handle a 'pending' event
from a different LWP, because breakpoints may not be inserted
and that different LWP will run away.
2001-06-04  Michael Snyder  <msnyder@redhat.com>

	* lin-lwp.c (lin_lwp_resume): Set global 'solo_resume_pid' if
	resuming only one LWP.
	(lin_lwp_wait): Use global 'solo_resume_pid' to reject pending
	events if they belong to the wrong LWP.

Index: lin-lwp.c
===================================================================
RCS file: /cvs/src/src/gdb/lin-lwp.c,v
retrieving revision 1.20
diff -c -3 -p -r1.20 lin-lwp.c
*** lin-lwp.c	2001/05/15 00:13:47	1.20
--- lin-lwp.c	2001/06/05 02:58:29
*************** resume_callback (struct lwp_info *lp, vo
*** 468,473 ****
--- 468,475 ----
    return 0;
  }
  
+ static pid_t solo_resume_pid = -1;
+ 
  static void
  lin_lwp_resume (ptid_t ptid, int step, enum target_signal signo)
  {
*************** lin_lwp_resume (ptid_t ptid, int step, e
*** 479,484 ****
--- 481,490 ----
       id'.  But if STEP is zero, then PID means `continue *all*
       processes, but give the signal only to this one'.  */
    resume_all = (PIDGET (ptid) == -1) || !step;
+   if (resume_all)
+     solo_resume_pid = -1;
+   else
+     solo_resume_pid = GET_LWP (ptid);
  
    /* If PID is -1, it's the current inferior that should be
       handled special.  */
*************** lin_lwp_wait (ptid_t ptid, struct target
*** 670,678 ****
    /* First check if there is a LWP with a wait status pending.  */
    if (pid == -1)
      {
!       /* Any LWP will do.  */
        lp = iterate_over_lwps (status_callback, NULL);
!       if (lp)
  	{
  	  if (debug_lin_lwp)
  	    fprintf_unfiltered (gdb_stdlog, 
--- 676,688 ----
    /* First check if there is a LWP with a wait status pending.  */
    if (pid == -1)
      {
!       /* We are asked to wait for any LWP.  */
        lp = iterate_over_lwps (status_callback, NULL);
!       /* However, if we resumed only one LWP, then we should accept 
! 	 a pending wait status only for the same LWP.  Otherwise we may 
!          really confuse GDB.  */
!       if (lp &&
! 	  (solo_resume_pid == -1 || solo_resume_pid == GET_LWP (lp->ptid)))
  	{
  	  if (debug_lin_lwp)
  	    fprintf_unfiltered (gdb_stdlog, 

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]