From 051f5d223dfc1806e216f60e4b29c1bf35f5c2d3 Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Sun, 5 Jul 2026 11:05:17 +0200 Subject: [PATCH] lockdep: Enable the printing of held locks of remote running tasks and print task CPU MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Background: ========== Currently lockdep does not print out the held locks of non-current tasks that are running on some other CPU, due to the fact that the held locks array is in flux and may be unreliable to print. Syzkaller on the other hand found it that the analysis of locking bugs is easier if we print this information too, because the more locking information the merrier. In particular races are bound to have multiple tasks running on different CPUs, and the exclusion of their held locks information is unnecessarily limiting. So while it's still true that printing out their held locks array is racy, it's not as bad as it seems. There's 16 internal callers to lockdep_print_held_locks(): - 14 callers call it with the current task, which should be safe out of box. - 1 caller, debug_show_all_locks(), calls it with RCU held, which should guarantee that 'p' cannot go away under us. - 1 caller, debug_show_held_locks(), exposes the internal API with the constraint that it should only be called by drivers or platform code if the task isn't actively running - we can assume that if it nevertheless does, it will be Their Problem™. As for held locks being changed from under debug_show_held_locks(), while the task cannot go away, so the held-locks array itself is safe (although potentially non-stable), AFAICS the worst-case race can be garbage printed out by print_lock(), not any actual crashes. In particular: unsigned int class_idx = hlock->class_idx; may be stale (belong to a lock that already got released on another CPU), but it should still be a valid class index bound by MAX_LOCKDEP_KEYS, and thus the lock_classes_in_use bitmap use should be safe. The other two accesses are ::acquire_ip and ::instance: printk(KERN_CONT "%px", hlock->instance); print_lock_name(hlock, lock); printk(KERN_CONT ", at: %pS\n", (void *)hlock->acquire_ip); But both are printed out as pointers, so no risk of dereference of a dangling pointer. We may print a garbage pointer. Also note that the check itself doesn't protect debug_show_held_locks() from printing garbage, as there's nothing that keeps a task from becoming runnable a nanosecond after we've run the task_is_running() check. In fact I'd argue that it's better to make this function *more* racy, for the simple robustness reason that we absolutely do not want it to crash even in the racy case. TL;DR: it should be fine to print the held locks of running tasks too, as long as we print out the information as well that a task is running, so that users are aware of any racy output. Implementation: ============== Implement that change. Also re-flow the function and streamline the printout into a single statement for all cases, which changes the 'no locks held by' / '%d lock[s] held by' phrasing that had a dependency on English spelling of plurals, to a uniform: locks held by bash/1234: %d Which spells correctly for 0, 1 and higher values, and should also be easier to parse both for humans and for scripts. Finally, print out the last CPU a task has ran on. This is very useful information for races and for locking bugs in particular. This basically extends the 'on CPU#%d' message we print for running tasks to all tasks we print. Reported-by: Tetsuo Handa Suggested-by: Tetsuo Handa Tested-by: Tetsuo Handa Signed-off-by: Ingo Molnar Cc: Boqun Feng Cc: Gary Guo Cc: Mark Brown Cc: Theodore Tso Cc: Miguel Ojeda Cc: Linus Torvalds Cc: Peter Zijlstra Cc: Will Deacon Cc: Greg Kroah-Hartman Cc: Waiman Long Link: https://patch.msgid.link/akoeSIQGwqd9cZwd@gmail.com --- kernel/locking/lockdep.c | 34 +++++++++++++++++++++++++--------- 1 file changed, 25 insertions(+), 9 deletions(-) diff --git a/kernel/locking/lockdep.c b/kernel/locking/lockdep.c index 2d4c5bab5af8..ff4bbf14665e 100644 --- a/kernel/locking/lockdep.c +++ b/kernel/locking/lockdep.c @@ -787,17 +787,33 @@ static void lockdep_print_held_locks(struct task_struct *p) { int i, depth = READ_ONCE(p->lockdep_depth); - if (!depth) - printk("no locks held by %s/%d.\n", p->comm, task_pid_nr(p)); - else - printk("%d lock%s held by %s/%d:\n", depth, - str_plural(depth), p->comm, task_pid_nr(p)); /* - * It's not reliable to print a task's held locks if it's not sleeping - * and it's not the current task. + * Note that it's always somewhat unreliable to print held locks + * of a task that is running on another CPU, but we cannot guarantee + * the stability of ->held_locks without actually stopping all active + * remote CPUs, which we absolutely do not want to do because it's + * very intrusive and thus slow. + * + * So we do the next best thing here: we print out the held lock + * array on a best-effort basis, without crashing even if the + * fields are being modified on another CPU. Note the careful + * construction of print_lock() so that it never crashes. + * + * We also print out the CPU the task is or was last running on, with + * the message saying 'on CPU...' if the task is running, and + * 'last CPU' if it's not. + * + * Also note that the task_is_running(p) information is fundamentally + * racy: even if the message says the task is 'on CPU', the task may + * have scheduled out already, or if it says 'last CPU', it may just + * have scheduled in on another CPU. But even with these limitations + * it's still useful debuggining information. */ - if (p != current && task_is_running(p)) - return; + printk("locks held by %s/%d: %d, %s CPU#%d%s\n", + p->comm, task_pid_nr(p), depth, + task_is_running(p) ? "last" : "on", task_cpu(p), + depth > 0 ? ":" : ""); + for (i = 0; i < depth; i++) { printk(" #%d: ", i); print_lock(p->held_locks + i);