um: fix possible race on task->mm

Checking for task->mm is dangerous as ->mm might disappear (exit_mm()
assigns NULL under task_lock(), so tasklist lock is not enough).

We can't use get_task_mm()/mmput() pair as mmput() might sleep, so let's
take the task lock while we care about its mm.

Note that we should also use find_lock_task_mm() to check all process'
threads for a valid mm, but for uml we'll do it in a separate patch.

Signed-off-by: Anton Vorontsov <anton.vorontsov@linaro.org>
Cc: Richard Weinberger <richard@nod.at>
Cc: Oleg Nesterov <oleg@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

1 files changed, 5 insertions, 2 deletions

diff --git a/arch/um/kernel/reboot.c b/arch/um/kernel/reboot.c
index 66d754c..1411f4e 100644
--- a/arch/um/kernel/reboot.c
+++ b/arch/um/kernel/reboot.c
@@ -25,10 +25,13 @@ static void kill_off_processes(void)
 
 		read_lock(&tasklist_lock);
 		for_each_process(p) {
-			if (p->mm == NULL)
+			task_lock(p);
+			if (!p->mm) {
+				task_unlock(p);
 				continue;
-
+			}
 			pid = p->mm->context.id.u.pid;
+			task_unlock(p);
 			os_kill_ptraced_process(pid, 1);
 		}
 		read_unlock(&tasklist_lock);