Re: [syzbot] possible deadlock in vm_mmap_pgoff

From: Steven Rostedt
Date: Wed Apr 28 2021 - 09:30:34 EST

On Wed, 28 Apr 2021 14:15:22 +0800
Hillf Danton <hdanton@xxxxxxxx> wrote:

> On Tue, 27 Apr 2021 08:18:27
> > Hello,
> >
> > syzbot found the following issue on:
> >
> > HEAD commit: 7af08140 Revert "gcov: clang: fix clang-11+ build"
> > git tree: upstream
> > console output: https://syzkaller.appspot.com/x/log.txt?x=17d57dfed00000
> > kernel config: https://syzkaller.appspot.com/x/.config?x=76c0382ceab56d34
> > dashboard link: https://syzkaller.appspot.com/bug?extid=f619f7c0a2a5f87694e6
> >
> > Unfortunately, I don't have any reproducer for this issue yet.
> >
> > IMPORTANT: if you fix the issue, please add the following tag to the commit:
> > Reported-by: syzbot+f619f7c0a2a5f87694e6@xxxxxxxxxxxxxxxxxxxxxxxxx
> >
> > ======================================================
> > WARNING: possible circular locking dependency detected
> > 5.12.0-rc8-syzkaller #0 Not tainted
> > ------------------------------------------------------
> > syz-executor.1/16055 is trying to acquire lock:
> > ffffffff8bfe2bc8 (event_mutex){+.+.}-{3:3}, at: perf_trace_destroy+0x23/0xf0 kernel/trace/trace_event_perf.c:241
> >
> > but task is already holding lock:
> > ffff88801b887258 (&mm->mmap_lock#2){++++}-{3:3}, at: mmap_write_lock_killable include/linux/mmap_lock.h:87 [inline]
> > ffff88801b887258 (&mm->mmap_lock#2){++++}-{3:3}, at: vm_mmap_pgoff+0x15c/0x290 mm/util.c:517
> >
> > which lock already depends on the new lock.
> >
> >
> > the existing dependency chain (in reverse order) is:
> >
> > -> #3 (&mm->mmap_lock#2){++++}-{3:3}:
> > down_write_killable+0x95/0x170 kernel/locking/rwsem.c:1417
> > mmap_write_lock_killable include/linux/mmap_lock.h:87 [inline]
> > dup_mmap kernel/fork.c:480 [inline]
> > dup_mm+0x12e/0x1380 kernel/fork.c:1368
> > copy_mm kernel/fork.c:1424 [inline]
> > copy_process+0x2bc8/0x71a0 kernel/fork.c:2113
> > kernel_clone+0xe7/0xab0 kernel/fork.c:2500
> > __do_sys_clone+0xc8/0x110 kernel/fork.c:2617
> > do_syscall_64+0x2d/0x70 arch/x86/entry/common.c:46
> > entry_SYSCALL_64_after_hwframe+0x44/0xae
> >
> > -> #2 (dup_mmap_sem){++++}-{0:0}:
> > percpu_down_write+0x95/0x440 kernel/locking/percpu-rwsem.c:217
> > register_for_each_vma+0x2c/0xc10 kernel/events/uprobes.c:1040
> > __uprobe_register+0x5c2/0x850 kernel/events/uprobes.c:1181
> > trace_uprobe_enable kernel/trace/trace_uprobe.c:1065 [inline]
> > probe_event_enable+0x357/0xa00 kernel/trace/trace_uprobe.c:1134
> > trace_uprobe_register+0x443/0x880 kernel/trace/trace_uprobe.c:1461
> > perf_trace_event_reg kernel/trace/trace_event_perf.c:129 [inline]
> > perf_trace_event_init+0x549/0xa20 kernel/trace/trace_event_perf.c:204
> > perf_uprobe_init+0x16f/0x210 kernel/trace/trace_event_perf.c:336
> > perf_uprobe_event_init+0xff/0x1c0 kernel/events/core.c:9754
> > perf_try_init_event+0x12a/0x560 kernel/events/core.c:11071
> > perf_init_event kernel/events/core.c:11123 [inline]
> > perf_event_alloc.part.0+0xe3b/0x3960 kernel/events/core.c:11403
> > perf_event_alloc kernel/events/core.c:11785 [inline]
> > __do_sys_perf_event_open+0x647/0x2e60 kernel/events/core.c:11883
> > do_syscall_64+0x2d/0x70 arch/x86/entry/common.c:46
> > entry_SYSCALL_64_after_hwframe+0x44/0xae
> >
> > -> #1 (&uprobe->register_rwsem){+.+.}-{3:3}:
> > down_write+0x92/0x150 kernel/locking/rwsem.c:1406
> > __uprobe_register+0x531/0x850 kernel/events/uprobes.c:1177
> > trace_uprobe_enable kernel/trace/trace_uprobe.c:1065 [inline]
> > probe_event_enable+0x357/0xa00 kernel/trace/trace_uprobe.c:1134
> > trace_uprobe_register+0x443/0x880 kernel/trace/trace_uprobe.c:1461
> > perf_trace_event_reg kernel/trace/trace_event_perf.c:129 [inline]
> > perf_trace_event_init+0x549/0xa20 kernel/trace/trace_event_perf.c:204
> > perf_uprobe_init+0x16f/0x210 kernel/trace/trace_event_perf.c:336
> > perf_uprobe_event_init+0xff/0x1c0 kernel/events/core.c:9754
> > perf_try_init_event+0x12a/0x560 kernel/events/core.c:11071
> > perf_init_event kernel/events/core.c:11123 [inline]
> > perf_event_alloc.part.0+0xe3b/0x3960 kernel/events/core.c:11403
> > perf_event_alloc kernel/events/core.c:11785 [inline]
> > __do_sys_perf_event_open+0x647/0x2e60 kernel/events/core.c:11883
> > do_syscall_64+0x2d/0x70 arch/x86/entry/common.c:46
> > entry_SYSCALL_64_after_hwframe+0x44/0xae
> >
> > -> #0 (event_mutex){+.+.}-{3:3}:
> > check_prev_add kernel/locking/lockdep.c:2937 [inline]
> > check_prevs_add kernel/locking/lockdep.c:3060 [inline]
> > validate_chain kernel/locking/lockdep.c:3675 [inline]
> > __lock_acquire+0x2b14/0x54c0 kernel/locking/lockdep.c:4901
> > lock_acquire kernel/locking/lockdep.c:5511 [inline]
> > lock_acquire+0x1ab/0x740 kernel/locking/lockdep.c:5476
> > __mutex_lock_common kernel/locking/mutex.c:949 [inline]
> > __mutex_lock+0x139/0x1120 kernel/locking/mutex.c:1096
> > perf_trace_destroy+0x23/0xf0 kernel/trace/trace_event_perf.c:241
> > _free_event+0x2ee/0x1380 kernel/events/core.c:4863
> > put_event kernel/events/core.c:4957 [inline]
> > perf_mmap_close+0x572/0xe10 kernel/events/core.c:6002
> > remove_vma+0xae/0x170 mm/mmap.c:180
> > remove_vma_list mm/mmap.c:2653 [inline]
> > __do_munmap+0x74f/0x11a0 mm/mmap.c:2909
> > do_munmap mm/mmap.c:2917 [inline]
> > munmap_vma_range mm/mmap.c:598 [inline]
> > mmap_region+0x85a/0x1730 mm/mmap.c:1750
> > do_mmap+0xcff/0x11d0 mm/mmap.c:1581
> > vm_mmap_pgoff+0x1b7/0x290 mm/util.c:519
> > ksys_mmap_pgoff+0x49c/0x620 mm/mmap.c:1632
> > do_syscall_64+0x2d/0x70 arch/x86/entry/common.c:46
> > entry_SYSCALL_64_after_hwframe+0x44/0xae
> >
> > other info that might help us debug this:
> >
> > Chain exists of:
> > event_mutex --> dup_mmap_sem --> &mm->mmap_lock#2
> >
> > Possible unsafe locking scenario:
> >
> > CPU0 CPU1
> > ---- ----
> > lock(&mm->mmap_lock#2);
> > lock(dup_mmap_sem);
> > lock(&mm->mmap_lock#2);
> > lock(event_mutex);
> >
> > *** DEADLOCK ***
>
> Break the lock chain by asking kworker to destroy perf event.
> Thoughts other than workqueue are appreciated.

I think this could cause a problem with the delay.

This is called from event->destroy()


if (event->destroy)
event->destroy(event);
module_put(pmu->module);

What if this event is in that module, and we just unloaded it?

Then the workqueue would try to modify the module text that no longer
exists. Right?

I Cc'd some people that understand this code better.

-- Steve


>
> +++ x/kernel/trace/trace_event_perf.c
> @@ -236,14 +236,22 @@ int perf_trace_init(struct perf_event *p
> return ret;
> }
>
> -void perf_trace_destroy(struct perf_event *p_event)
> +static void perf_trace_destroy_work_fn(struct work_struct *w)
> {
> + struct perf_event *p_event = container_of(w, struct perf_event,
> + destroy_work);
> mutex_lock(&event_mutex);
> perf_trace_event_close(p_event);
> perf_trace_event_unreg(p_event);
> mutex_unlock(&event_mutex);
> }
>
> +void perf_trace_destroy(struct perf_event *p_event)
> +{
> + INIT_WORK(&p_event->destroy_work, perf_trace_destroy_work_fn);
> + queue_work(system_unbound_wq, &p_event->destroy_work);
> +}
> +
> #ifdef CONFIG_KPROBE_EVENTS
> int perf_kprobe_init(struct perf_event *p_event, bool is_retprobe)
> {