[Boost-commit] svn:boost r39467 - trunk/tools/jam/src

From: kbelco_at_[hidden]
Date: 2007-09-21 18:38:19

Author: noel_belcourt
Date: 2007-09-21 18:38:17 EDT (Fri, 21 Sep 2007)
New Revision: 39467
URL: http://svn.boost.org/trac/boost/changeset/39467

Log:
Another patch to fix the -lx timeout code.  Some actions spawn
sub-processes after bjam forks a new process (for example, after
g++ is forked by bjam, g++ then forks sub-processes like cc1plus).
The timeout code would kill the g++ process, but might not kill
the subprocesses spawned by g++.

I fixed this problem by making the bjam fork'ed process (g++) a 
session leader by calling setsid() before calling exec.  The setsid 
call, in essence, gives all child processes a parent process id 
(ppid) of the g++ process id.  This guarantees that killing g++ 
will kill all child processes spawned by g++ as well.

One last comment on the maximum process time before a process is actually
killed.  The worst case process elapsed time is 2x seconds if -lx is
given.  The reason is that a process might be one second away from being
killed and, if there's no other signal activity, the select function will
wait x seconds before timing out and killing any active processes.  So
if you say -lx and monitor a build known to have lengthy processes, you 
may see a process with up to 2x seconds of time before it is killed.

Text files modified: 
   trunk/tools/jam/src/execunix.c |    23 ++++++++++++-----------                 
   1 files changed, 12 insertions(+), 11 deletions(-)

Modified: trunk/tools/jam/src/execunix.c
==============================================================================
--- trunk/tools/jam/src/execunix.c	(original)
+++ trunk/tools/jam/src/execunix.c	2007-09-21 18:38:17 EDT (Fri, 21 Sep 2007)
@@ -211,6 +211,17 @@
             else
                 dup2(err[1], STDERR_FILENO);
 
+            /* Make this process a session leader
+             * so that when we kill it, all child
+             * processes of this process are terminated
+             * as well.
+             *
+             * we use kill(-pid, SIGKILL) to kill the
+             * session leader and all children of this
+             * session.
+             */
+            setsid();
+
             execvp( argv[0], argv );
             _exit(127);
         }
@@ -367,7 +378,7 @@
             struct tms buf;
             clock_t current = times(&buf);
             if (globs.timeout <= (current-cmdtab[i].start_time)/tps) {
-                kill(cmdtab[i].pid, SIGKILL);
+                kill(-cmdtab[i].pid, SIGKILL);
                 cmdtab[i].exit_reason = EXIT_TIMEOUT;
             }
         }
@@ -408,16 +419,6 @@
 
             /* select will wait until io on a descriptor or a signal */
             ret = select(fd_max+1, &fds, 0, 0, &tv);
-
-            if (0 == ret) {
-                /* select timed out, all processes have expired, kill them */
-                for (i=0; i<globs.jobs; ++i) {
-                    cmdtab[i].start_time = 0;
-                }
-                /* select will wait until io on a descriptor or a signal */
-                populate_file_descriptors(&fd_max, &fds);
-                ret = select(fd_max+1, &fds, 0, 0, 0);
-            }
         }
         else {
             /* select will wait until io on a descriptor or a signal */