[RFC][PATCH 12/16] Terminate processes in a ns when reaper is exiting.

sukadev at us.ibm.com sukadev at us.ibm.com
Wed May 23 18:13:54 PDT 2007


Subject: Terminate processes in a ns when reaper is exiting.

From: Sukadev Bhattiprolu <sukadev at us.ibm.com>

This should actually be considered a part of the previous patch which
enables cloning of pid namespace. Its been separated out for easier
review.

Terminate all processes in a namespace when the reaper of the namespace
is exiting. We do this by walking the pidmap of the namespace and sending
SIGKILL to all processes.

TODO:
	- Consider maintaining a per-pid namespace tasklist. Use that list
	  to terminate processes in the namespace more efficiently. Such a 
	  tasklist may also be useful to freeze or checkpoint an application.

Signed-off-by: Sukadev Bhattiprolu <sukadev at us.ibm.com>
---
 include/linux/pid.h           |    1 +
 include/linux/pid_namespace.h |    1 +
 kernel/exit.c                 |    5 +++--
 kernel/fork.c                 |   19 +++++++++++++++++++
 kernel/pid.c                  |   42 ++++++++++++++++++++++++++++++++++++++++++
 5 files changed, 66 insertions(+), 2 deletions(-)

Index: lx26-21-mm2/include/linux/pid.h
===================================================================
--- lx26-21-mm2.orig/include/linux/pid.h	2007-05-22 16:59:52.000000000 -0700
+++ lx26-21-mm2/include/linux/pid.h	2007-05-22 16:59:53.000000000 -0700
@@ -124,6 +124,7 @@ extern void FASTCALL(free_pid(struct pid
 
 extern pid_t pid_to_nr_in_ns(struct pid_namespace *ns, struct pid *pid);
 extern pid_t pid_to_nr(struct pid *pid);
+extern void zap_pid_ns_processes(struct pid_namespace *pid_ns);
 
 #define do_each_pid_task(pid, type, task)				\
 	do {								\
Index: lx26-21-mm2/include/linux/pid_namespace.h
===================================================================
--- lx26-21-mm2.orig/include/linux/pid_namespace.h	2007-05-22 16:59:52.000000000 -0700
+++ lx26-21-mm2/include/linux/pid_namespace.h	2007-05-22 16:59:53.000000000 -0700
@@ -32,6 +32,7 @@ struct pid_namespace {
 	struct pidmap pidmap[PIDMAP_ENTRIES];
 	int last_pid;
 	struct task_struct *child_reaper;
+	atomic_t terminating;
 };
 
 extern struct pid_namespace init_pid_ns;
Index: lx26-21-mm2/kernel/exit.c
===================================================================
--- lx26-21-mm2.orig/kernel/exit.c	2007-05-22 16:59:52.000000000 -0700
+++ lx26-21-mm2/kernel/exit.c	2007-05-22 16:59:53.000000000 -0700
@@ -883,9 +883,10 @@ fastcall NORET_TYPE void do_exit(long co
 	 * the reaper of the active pid namespace.
 	 */
 	if (unlikely(tsk == pid_ns->child_reaper)) {
-		if (pid_ns != &init_pid_ns)
+		if (pid_ns != &init_pid_ns) {
+			zap_pid_ns_processes(pid_ns);
 			pid_ns->child_reaper = init_pid_ns.child_reaper;
-		else
+		} else
 			panic("Attempted to kill init!");
 	}
 
Index: lx26-21-mm2/kernel/fork.c
===================================================================
--- lx26-21-mm2.orig/kernel/fork.c	2007-05-22 16:59:52.000000000 -0700
+++ lx26-21-mm2/kernel/fork.c	2007-05-22 16:59:53.000000000 -0700
@@ -1397,6 +1397,7 @@ long do_fork(unsigned long clone_flags,
 	 */
 	if (!IS_ERR(p)) {
 		struct completion vfork;
+		struct pid_namespace *pid_ns;
 
 		nr = pid_to_nr(task_pid(p));
 
@@ -1405,6 +1406,24 @@ long do_fork(unsigned long clone_flags,
 			init_completion(&vfork);
 		}
 
+		/*
+		 * If our pid namespace was asked to terminate while we were
+		 * being created, exit immediately.
+		 *
+		 * We need the check here for the case where the termination
+		 * request was received after a pid_t was allocated in
+		 * alloc_pid_nr() and before the process was fully created
+		 * ("old enough") to receive the signal sent from do_exit().
+		 *
+		 * TODO: Implement a task list per pid namespace to simplify
+		 * terminating the pid namespace and remove this check.
+		 */
+		pid_ns = task_active_pid_ns(p);
+		if (atomic_read(&pid_ns->terminating)) {
+			sigaddset(&p->pending.signal, SIGKILL);
+			set_tsk_thread_flag(p, TIF_SIGPENDING);
+		}
+
 		if ((p->ptrace & PT_PTRACED) || (clone_flags & CLONE_STOPPED)) {
 			/*
 			 * We'll start up with an immediate SIGSTOP.
Index: lx26-21-mm2/kernel/pid.c
===================================================================
--- lx26-21-mm2.orig/kernel/pid.c	2007-05-22 16:59:52.000000000 -0700
+++ lx26-21-mm2/kernel/pid.c	2007-05-22 16:59:53.000000000 -0700
@@ -124,6 +124,9 @@ static int alloc_pidmap(struct pid_names
 	int i, offset, max_scan, pid, last = pid_ns->last_pid;
 	struct pidmap *map;
 
+	if (atomic_read(&pid_ns->terminating))
+		return -1;
+
 	pid = last + 1;
 	if (pid >= pid_max)
 		pid = RESERVED_PIDS;
@@ -405,6 +408,39 @@ static struct pid_namespace *alloc_pid_n
 	return ns;
 }
 
+/*
+ * When child reaper of the pid namespace @pid_ns is itself terminating,
+ * we need to terminate all processes in the pid namespace since /proc
+ * has a reference to the child reaper of the pid namespace.
+ *
+ * Send SIGKILL to all processes in the pid namespace. Set the 'terminating'
+ * flag in pid_ns to prevent any new processes from getting created in the
+ * pid namespace.
+ *
+ * Note that we will also be terminating all our child pid namespaces
+ * (if any) since we send SIGKILL their reapers as well.
+ *
+ * TODO: It maybe more efficient to maintain a list of tasks in the
+ *       pid namespace and walk that list.
+ */
+void zap_pid_ns_processes(struct pid_namespace *pid_ns)
+{
+	int nr;
+
+	atomic_set(&pid_ns->terminating, 1);
+
+	/*
+	 * We know pid == 1 is terminating. Find remaining pid_ts
+	 * in the namespace and terminate them.
+	 */
+	nr = next_pidmap(pid_ns, 1);
+	while (nr > 0) {
+		kill_proc(nr, SIGKILL, 1);
+		nr = next_pidmap(pid_ns, nr);
+	}
+	return;
+}
+
 #else
 
 static int alloc_pid_ns()
@@ -417,6 +453,12 @@ static int alloc_pid_ns()
 	}
 	return 0;
 }
+
+void zap_pid_ns_processes(struct pid_namespace *pid_ns)
+{
+	/* Nothing to do when we don't have multiple pid namespaces */
+	return;
+}
 #endif 	/*CONFIG_PID_NS*/
 
 void toss_pid(struct pid *pid)


More information about the Containers mailing list