From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S935211AbcIFNKC (ORCPT ); Tue, 6 Sep 2016 09:10:02 -0400 Received: from mail-lf0-f50.google.com ([209.85.215.50]:36008 "EHLO mail-lf0-f50.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1756043AbcIFNJz (ORCPT ); Tue, 6 Sep 2016 09:09:55 -0400 MIME-Version: 1.0 In-Reply-To: References:

From: Dmitry Vyukov Date: Tue, 6 Sep 2016 15:09:25 +0200 Message-ID: Subject: Re: sound: divide by 0 in snd_hrtimer_callback (or hang) To: Takashi Iwai Cc: Vegard Nossum , Vegard Nossum , Jaroslav Kysela , Kangjie Lu , alsa-devel@alsa-project.org, LKML , syzkaller Content-Type: text/plain; charset=UTF-8 X-ccpol: medium Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Tue, Sep 6, 2016 at 3:00 PM, Takashi Iwai wrote: > On Sat, 03 Sep 2016 14:35:15 +0200, > Dmitry Vyukov wrote: >> >> Hello, >> >> The following program causes either division error or hangs kernel: >> >> https://gist.githubusercontent.com/dvyukov/b1f33be3bfcc15d629e10db483bee1e4/raw/1574b407e0456bf8277bee423e7731641f5f2c14/gistfile1.txt >> >> >> divide error: 0000 [#1] SMP DEBUG_PAGEALLOC KASAN >> Dumping ftrace buffer: >> (ftrace buffer empty) >> Modules linked in: >> CPU: 1 PID: 27320 Comm: syz-executor Not tainted 4.8.0-rc3-next-20160825+ #8 >> Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Bochs 01/01/2011 >> task: ffff88003c9b2280 task.stack: ffff880027280000 >> RIP: 0010:[] [< inline >] ktime_divns >> include/linux/ktime.h:195 >> RIP: 0010:[] [] >> snd_hrtimer_callback+0x1bc/0x3c0 sound/core/hrtimer.c:62 >> RSP: 0018:ffff88003ed07dd8 EFLAGS: 00010006 >> RAX: 0000000000004801 RBX: ffff88003d3a0a88 RCX: 0000000000000000 >> RDX: 0000000000000000 RSI: ffff88003d3a0aa0 RDI: ffffffff886b69a8 >> RBP: ffff88003ed07e30 R08: 0000010e85417665 R09: ffff88007fff7048 >> R10: ffff88007fff7058 R11: ffff88007fff7050 R12: ffff88003d00f840 >> R13: ffff88003d00f8f4 R14: ffff88003d00f970 R15: ffff88003d3a0a80 >> FS: 0000000000000000(0000) GS:ffff88003ed00000(0000) knlGS:0000000000000000 >> CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 >> CR2: 00007f4152126db8 CR3: 000000006b896000 CR4: 00000000000006e0 >> DR0: 000000000000001e DR1: 000000000000001e DR2: 0000000000000000 >> DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000600 >> Stack: >> 0000000000000000 ffff88003d3a0aa0 ffff88003d00f8f8 0000000000004801 >> ffff88003d3a0ae0 0000000000000000 ffff88003d3a0a88 0000000000000001 >> ffff88003ed1b400 dffffc0000000000 0000000000000001 ffff88003ed07f28 >> Call Trace: >> >> [< inline >] __run_hrtimer kernel/time/hrtimer.c:1238 >> [] __hrtimer_run_queues+0x325/0xe70 >> kernel/time/hrtimer.c:1302 >> [] hrtimer_interrupt+0x18b/0x420 kernel/time/hrtimer.c:1336 >> [] local_apic_timer_interrupt+0x6f/0xe0 >> arch/x86/kernel/apic/apic.c:933 >> [] smp_apic_timer_interrupt+0x76/0xa0 >> arch/x86/kernel/apic/apic.c:957 >> [] apic_timer_interrupt+0x8c/0xa0 >> arch/x86/entry/entry_64.S:487 >> >> [< inline >] rcu_lock_release include/linux/rcupdate.h:494 >> [< inline >] rcu_read_unlock include/linux/rcupdate.h:927 >> [] unlock_page_memcg+0x95/0x130 mm/memcontrol.c:1680 >> [< inline >] page_remove_file_rmap mm/rmap.c:1343 >> [] page_remove_rmap+0x1cc/0x8e0 mm/rmap.c:1394 >> [< inline >] zap_pte_range mm/memory.c:1170 >> [< inline >] zap_pmd_range mm/memory.c:1257 >> [< inline >] zap_pud_range mm/memory.c:1278 >> [] unmap_page_range+0xf89/0x1ba0 mm/memory.c:1299 >> [] unmap_single_vma+0x101/0x260 mm/memory.c:1344 >> [] unmap_vmas+0xf1/0x1b0 >> [] exit_mmap+0x22b/0x420 mm/mmap.c:2787 >> [< inline >] __mmput kernel/fork.c:770 >> [] mmput+0xd6/0x3c0 kernel/fork.c:790 >> [< inline >] exit_mm kernel/exit.c:512 >> [] do_exit+0x738/0x2e70 kernel/exit.c:815 >> [] do_group_exit+0x108/0x330 kernel/exit.c:958 >> [] get_signal+0x62a/0x15d0 kernel/signal.c:2307 >> [] do_signal+0x83/0x1f60 arch/x86/kernel/signal.c:805 >> [] exit_to_usermode_loop+0x1a5/0x210 >> arch/x86/entry/common.c:163 >> [< inline >] prepare_exit_to_usermode arch/x86/entry/common.c:198 >> [< inline >] syscall_return_slowpath arch/x86/entry/common.c:267 >> [] do_syscall_64+0x4c0/0x640 arch/x86/entry/common.c:293 >> [] entry_SYSCALL64_slow_path+0x25/0x25 >> Code: cb a2 ce fb 8b 0d f5 21 70 05 48 0f af 4d d0 48 85 c9 0f 88 6a >> 01 00 00 48 89 4d a8 e8 ae a2 ce fb 48 8b 45 c0 48 8b 4d a8 48 99 <48> >> f7 f9 48 01 45 d0 e8 98 a2 ce fb 4c 89 fa 48 b8 00 00 00 00 >> RIP [< inline >] ktime_divns include/linux/ktime.h:195 >> RIP [] snd_hrtimer_callback+0x1bc/0x3c0 >> sound/core/hrtimer.c:62 >> RSP >> ---[ end trace e1a6d1e5a6136d50 ]--- >> >> >> or: >> >> INFO: rcu_sched detected stalls on CPUs/tasks: >> 0-...: (1 GPs behind) idle=203/140000000000001/0 softirq=923090/923091 fqs=6490 >> (detected by 1, t=26002 jiffies, g=508022, c=508021, q=20) >> Task dump for CPU 0: >> a.out R running task 28848 31756 4047 0x0000000a >> 0000000000000000 0000000000000282 0000000000000000 ffff880066a9b430 >> ffff880066a9b430 ffff8800667cfae0 0000000000000282 0000000000000000 >> 0000000000000000 0000000000000000 000003a0184674dd dffffc0000000000 >> Call Trace: >> [< inline >] spin_unlock_irqrestore ./include/linux/spinlock.h:362 >> [] snd_timer_start1+0xdd/0x740 sound/core/timer.c:477 >> [] snd_timer_continue+0x45/0x80 sound/core/timer.c:606 >> [< inline >] snd_timer_user_continue sound/core/timer.c:1835 >> [< inline >] __snd_timer_user_ioctl sound/core/timer.c:1902 >> [] snd_timer_user_ioctl+0xe99/0x2470 sound/core/timer.c:1917 >> [< inline >] vfs_ioctl fs/ioctl.c:43 >> [] do_vfs_ioctl+0x18c/0x1080 fs/ioctl.c:675 >> [< inline >] SYSC_ioctl fs/ioctl.c:690 >> [] SyS_ioctl+0x8f/0xc0 fs/ioctl.c:681 >> [] entry_SYSCALL_64_fastpath+0x23/0xc1 >> arch/x86/entry/entry_64.S:208 >> >> >> >> I am on 0f98f121e1670eaa2a2fbb675e07d6ba7f0e146f of linux-next with >> the following diff in sound/core/timer.c (one is a fix for another >> division by 0): >> >> diff --git a/sound/core/timer.c b/sound/core/timer.c >> index 9a6157e..3e55c6d 100644 >> --- a/sound/core/timer.c >> +++ b/sound/core/timer.c >> @@ -813,6 +813,7 @@ int snd_timer_new(struct snd_card *card, char *id, >> struct snd_timer_id *tid, >> timer->tmr_subdevice = tid->subdevice; >> if (id) >> strlcpy(timer->id, id, sizeof(timer->id)); >> + timer->sticks = 1; >> INIT_LIST_HEAD(&timer->device_list); >> INIT_LIST_HEAD(&timer->open_list_head); >> INIT_LIST_HEAD(&timer->active_list_head); >> @@ -1958,6 +1959,7 @@ static ssize_t snd_timer_user_read(struct file >> *file, char __user *buffer, >> tu->qused--; >> spin_unlock_irq(&tu->qlock); >> >> + mutex_lock(&tu->ioctl_lock); >> if (tu->tread) { >> if (copy_to_user(buffer, &tu->tqueue[qhead], >> sizeof(struct snd_timer_tread))) >> @@ -1967,6 +1969,7 @@ static ssize_t snd_timer_user_read(struct file >> *file, char __user *buffer, >> sizeof(struct snd_timer_read))) >> err = -EFAULT; >> } >> + mutex_unlock(&tu->ioctl_lock); >> >> spin_lock_irq(&tu->qlock); >> if (err < 0) >> > > I couldn't trigger the bug any longer after merging for-linus branch > of sound.git tree: > git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound.git > > But I thought it got fixed by sticks=1 in the above. Maybe it's a > combination, as it contains yet another timer fix by Vegard. You mean that my local changes are a combination of two patches? But I would not expect that addition of mutex_lock/unlock can affect division by 0. Or you mean that there some other fix by Vegard? What is it? > The pull request will be sent in this week for 4.8-rc5.