[exim-cvs] cvs commit: exim/exim-doc/doc-txt ChangeLog exim…

Kezdőlap
Üzenet törlése
Válasz az üzenetre
Szerző: Philip Hazel
Dátum:  
Címzett: exim-cvs
Tárgy: [exim-cvs] cvs commit: exim/exim-doc/doc-txt ChangeLog exim/exim-src/src daemon.c
ph10 2004/11/11 16:03:47 GMT

  Modified files:
    exim-doc/doc-txt     ChangeLog 
    exim-src/src         daemon.c 
  Log:
  (a) Changed to using os_restarting_signal() for setting the SIGCHLD
  handler in the daemon; this may fix Tony's obscure occasional crashes.
  (b) Reduced the size of the race window for noticing SIGCHLDs (note: it
  is not *essential* for Exim to see them).


  Revision  Changes    Path
  1.23      +8 -0      exim/exim-doc/doc-txt/ChangeLog
  1.3       +25 -8     exim/exim-src/src/daemon.c


  Index: ChangeLog
  ===================================================================
  RCS file: /home/cvs/exim/exim-doc/doc-txt/ChangeLog,v
  retrieving revision 1.22
  retrieving revision 1.23
  diff -u -r1.22 -r1.23
  --- ChangeLog    11 Nov 2004 11:40:36 -0000    1.22
  +++ ChangeLog    11 Nov 2004 16:03:47 -0000    1.23
  @@ -1,4 +1,4 @@
  -$Cambridge: exim/exim-doc/doc-txt/ChangeLog,v 1.22 2004/11/11 11:40:36 ph10 Exp $
  +$Cambridge: exim/exim-doc/doc-txt/ChangeLog,v 1.23 2004/11/11 16:03:47 ph10 Exp $


   Change log file for Exim from version 4.21
   -------------------------------------------
  @@ -85,6 +85,14 @@
       matching the list had actually called for a host name (not very likely,
       since this list is usually IP addresses). A host name is now passed as
       "[x.x.x.x]".
  +
  +24. Changed the calls that set up the SIGCHLD handler in the daemon to use the
  +    code that specifies a non-restarting handler (typically sigaction() in
  +    modern systems) in an attempt to fix a rare and obscure crash bug.
  +
  +25. Narrowed the window for a race in the daemon that could cause it to ignore
  +    SIGCHLD signals. This is not a major problem, because they are used only to
  +    wake it up if nothing else does.



Exim version 4.43

  Index: daemon.c
  ===================================================================
  RCS file: /home/cvs/exim/exim-src/src/daemon.c,v
  retrieving revision 1.2
  retrieving revision 1.3
  diff -u -r1.2 -r1.3
  --- daemon.c    10 Nov 2004 10:29:56 -0000    1.2
  +++ daemon.c    11 Nov 2004 16:03:47 -0000    1.3
  @@ -1,4 +1,4 @@
  -/* $Cambridge: exim/exim-src/src/daemon.c,v 1.2 2004/11/10 10:29:56 ph10 Exp $ */
  +/* $Cambridge: exim/exim-src/src/daemon.c,v 1.3 2004/11/11 16:03:47 ph10 Exp $ */


   /*************************************************
   *     Exim - an Internet mail transport agent    *
  @@ -84,8 +84,8 @@
   main_sigchld_handler(int sig)
   {
   sig = sig;    /* Keep picky compilers happy */
  +os_non_restarting_signal(SIGCHLD, SIG_DFL);
   sigchld_seen = TRUE;
  -signal(SIGCHLD, SIG_DFL);
   }



@@ -1377,7 +1377,7 @@
/* Set up the handler for termination of child processes. */

sigchld_seen = FALSE;
-signal(SIGCHLD, main_sigchld_handler);
+os_non_restarting_signal(SIGCHLD, main_sigchld_handler);

   /* If we are to run the queue periodically, pretend the alarm has just gone
   off. This will cause the first queue-runner to get kicked off straight away. */
  @@ -1624,17 +1624,34 @@
         }


       DEBUG(D_any) debug_printf("Listening...\n");
  +    
  +    /* In rare cases we may have had a SIGCHLD signal in the time between 
  +    setting the handler (below) and getting back here. If so, pretend that the 
  +    select() was interrupted so that we reap the child. This might still leave
  +    a small window when a SIGCHLD could get lost. However, since we use SIGCHLD 
  +    only to do the reaping more quickly, it shouldn't result in anything other
  +    than a delay until something else causes a wake-up. */


  -    if ((lcount = select(max_socket + 1, (SELECT_ARG2_TYPE *)&select_listen,
  -         NULL, NULL, NULL)) < 0)
  +    if (sigchld_seen)
  +      {
  +      lcount = -1;
  +      errno = EINTR;  
  +      }
  +    else
  +      {      
  +      lcount = select(max_socket + 1, (SELECT_ARG2_TYPE *)&select_listen,
  +        NULL, NULL, NULL);
  +      }   
  +
  +    if (lcount < 0)
         {
         select_failed = TRUE;
         lcount = 1;
         }
  -
  +      
       /* Loop for all the sockets that are currently ready to go. If select
  -    actually failed, we have set the count to 1 and a flag, so as to use the
  -    common error code for select/accept below. */
  +    actually failed, we have set the count to 1 and select_failed=TRUE, so as
  +    to use the common error code for select/accept below. */


       while (lcount-- > 0)
         {
  @@ -1785,7 +1802,7 @@
     if (sigchld_seen)
       {
       sigchld_seen = FALSE;
  -    signal(SIGCHLD, main_sigchld_handler);
  +    os_non_restarting_signal(SIGCHLD, main_sigchld_handler);
       }


     /* Handle being woken by SIGHUP. We know at this point that the result