Commit 19e274630c9e23a84d5940af83cf5db35103f968

Authored by Tejun Heo
Committed by Oleg Nesterov
1 parent 40ae717d1e

job control: reorganize wait_task_stopped()

wait_task_stopped() tested task_stopped_code() without acquiring
siglock and, if stop condition existed, called wait_task_stopped() and
directly returned the result.  This patch moves the initial
task_stopped_code() testing into wait_task_stopped() and make
wait_consider_task() fall through to wait_task_continue() on 0 return.

This is for the following two reasons.

* Because the initial task_stopped_code() test is done without
  acquiring siglock, it may race against SIGCONT generation.  The
  stopped condition might have been replaced by continued state by the
  time wait_task_stopped() acquired siglock.  This may lead to
  unexpected failure of WNOHANG waits.

  This reorganization addresses this single race case but there are
  other cases - TASK_RUNNING -> TASK_STOPPED transition and EXIT_*
  transitions.

* Scheduled ptrace updates require changes to the initial test which
  would fit better inside wait_task_stopped().

Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Oleg Nesterov <oleg@redhat.com>
Signed-off-by: Oleg Nesterov <oleg@redhat.com>

Showing 1 changed file with 23 additions and 7 deletions Side-by-side Diff

... ... @@ -1377,11 +1377,23 @@
1377 1377 return NULL;
1378 1378 }
1379 1379  
1380   -/*
1381   - * Handle sys_wait4 work for one task in state TASK_STOPPED. We hold
1382   - * read_lock(&tasklist_lock) on entry. If we return zero, we still hold
1383   - * the lock and this task is uninteresting. If we return nonzero, we have
1384   - * released the lock and the system call should return.
  1380 +/**
  1381 + * wait_task_stopped - Wait for %TASK_STOPPED or %TASK_TRACED
  1382 + * @wo: wait options
  1383 + * @ptrace: is the wait for ptrace
  1384 + * @p: task to wait for
  1385 + *
  1386 + * Handle sys_wait4() work for %p in state %TASK_STOPPED or %TASK_TRACED.
  1387 + *
  1388 + * CONTEXT:
  1389 + * read_lock(&tasklist_lock), which is released if return value is
  1390 + * non-zero. Also, grabs and releases @p->sighand->siglock.
  1391 + *
  1392 + * RETURNS:
  1393 + * 0 if wait condition didn't exist and search for other wait conditions
  1394 + * should continue. Non-zero return, -errno on failure and @p's pid on
  1395 + * success, implies that tasklist_lock is released and wait condition
  1396 + * search should terminate.
1385 1397 */
1386 1398 static int wait_task_stopped(struct wait_opts *wo,
1387 1399 int ptrace, struct task_struct *p)
... ... @@ -1397,6 +1409,9 @@
1397 1409 if (!ptrace && !(wo->wo_flags & WUNTRACED))
1398 1410 return 0;
1399 1411  
  1412 + if (!task_stopped_code(p, ptrace))
  1413 + return 0;
  1414 +
1400 1415 exit_code = 0;
1401 1416 spin_lock_irq(&p->sighand->siglock);
1402 1417  
... ... @@ -1607,8 +1622,9 @@
1607 1622 * Wait for stopped. Depending on @ptrace, different stopped state
1608 1623 * is used and the two don't interact with each other.
1609 1624 */
1610   - if (task_stopped_code(p, ptrace))
1611   - return wait_task_stopped(wo, ptrace, p);
  1625 + ret = wait_task_stopped(wo, ptrace, p);
  1626 + if (ret)
  1627 + return ret;
1612 1628  
1613 1629 /*
1614 1630 * Wait for continued. There's only one continued state and the