Commit d73437ade6b00e559b73f805e272446e2afdd3b3

Authored by Oleg Nesterov
Committed by Greg Kroah-Hartman
1 parent 0324896e2e

exit: fix race between wait_consider_task() and wait_task_zombie()

commit 3245d6acab981a2388ffb877c7ecc97e763c59d4 upstream.

wait_consider_task() checks EXIT_ZOMBIE after EXIT_DEAD/EXIT_TRACE and
both checks can fail if we race with EXIT_ZOMBIE -> EXIT_DEAD/EXIT_TRACE
change in between, gcc needs to reload p->exit_state after
security_task_wait().  In this case ->notask_error will be wrongly
cleared and do_wait() can hang forever if it was the last eligible
child.

Many thanks to Arne who carefully investigated the problem.

Note: this bug is very old but it was pure theoretical until commit
b3ab03160dfa ("wait: completely ignore the EXIT_DEAD tasks").  Before
this commit "-O2" was probably enough to guarantee that compiler won't
read ->exit_state twice.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Reported-by: Arne Goedeke <el@laramies.com>
Tested-by: Arne Goedeke <el@laramies.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Showing 1 changed file with 9 additions and 3 deletions Side-by-side Diff

... ... @@ -1302,9 +1302,15 @@
1302 1302 static int wait_consider_task(struct wait_opts *wo, int ptrace,
1303 1303 struct task_struct *p)
1304 1304 {
  1305 + /*
  1306 + * We can race with wait_task_zombie() from another thread.
  1307 + * Ensure that EXIT_ZOMBIE -> EXIT_DEAD/EXIT_TRACE transition
  1308 + * can't confuse the checks below.
  1309 + */
  1310 + int exit_state = ACCESS_ONCE(p->exit_state);
1305 1311 int ret;
1306 1312  
1307   - if (unlikely(p->exit_state == EXIT_DEAD))
  1313 + if (unlikely(exit_state == EXIT_DEAD))
1308 1314 return 0;
1309 1315  
1310 1316 ret = eligible_child(wo, p);
... ... @@ -1325,7 +1331,7 @@
1325 1331 return 0;
1326 1332 }
1327 1333  
1328   - if (unlikely(p->exit_state == EXIT_TRACE)) {
  1334 + if (unlikely(exit_state == EXIT_TRACE)) {
1329 1335 /*
1330 1336 * ptrace == 0 means we are the natural parent. In this case
1331 1337 * we should clear notask_error, debugger will notify us.
... ... @@ -1352,7 +1358,7 @@
1352 1358 }
1353 1359  
1354 1360 /* slay zombie? */
1355   - if (p->exit_state == EXIT_ZOMBIE) {
  1361 + if (exit_state == EXIT_ZOMBIE) {
1356 1362 /* we don't reap group leaders with subthreads */
1357 1363 if (!delay_group_leader(p)) {
1358 1364 /*