On 2018/03/15 17:58, Kirill Tkhai wrote: > On 15.03.2018 01:22, Andrew Morton wrote: >> On Wed, 14 Mar 2018 15:09:09 -0700 Tejun Heo <tj@xxxxxxxxxx> wrote: >> >>> Hello, Andrew. >>> >>> On Wed, Mar 14, 2018 at 01:56:31PM -0700, Andrew Morton wrote: >>>> It would benefit from a comment explaining why we're doing this (it's >>>> for the oom-killer). >>> >>> Will add. >>> >>>> My memory is weak and our documentation is awful. What does >>>> mutex_lock_killable() actually do and how does it differ from >>>> mutex_lock_interruptible()? Userspace tasks can run pcpu_alloc() and I >>> >>> IIRC, killable listens only to SIGKILL. I think that killable listens to any signal which results in termination of that process. For example, if a process is configured to terminate upon SIGINT, fatal_signal_pending() becomes true upon SIGINT. >>> >>>> wonder if there's any way in which a userspace-delivered signal can >>>> disrupt another userspace task's memory allocation attempt? >>> >>> Hmm... maybe. Just honoring SIGKILL *should* be fine but the alloc >>> failure paths might be broken, so there are some risks. Given that >>> the cases where userspace tasks end up allocation percpu memory is >>> pretty limited and/or priviledged (like mount, bpf), I don't think the >>> risks are high tho. >> >> hm. spose so. Maybe. Are there other ways? I assume the time is >> being spent in pcpu_create_chunk()? We could drop the mutex while >> running that stuff and take the appropriate did-we-race-with-someone >> testing after retaking it. Or similar. > > The balance work spends its time in pcpu_populate_chunk(). There are > two stacks of this problem: Will you show me more contexts? Unless CONFIG_MMU=n kernels, the OOM reaper reclaims memory from the OOM victim. Therefore, "If tasks doing pcpu_alloc() are choosen by OOM killer, they can't exit, because they are waiting for the mutex." should not cause problems. Of course, giving up upon SIGKILL is nice regardless. > > [ 106.313267] kworker/2:2 D13832 936 2 0x80000000 > [ 106.313740] Workqueue: events pcpu_balance_workfn > [ 106.314109] Call Trace: > [ 106.314293] ? __schedule+0x267/0x750 > [ 106.314570] schedule+0x2d/0x90 > [ 106.314803] schedule_timeout+0x17f/0x390 > [ 106.315106] ? __next_timer_interrupt+0xc0/0xc0 > [ 106.315429] __alloc_pages_slowpath+0xb73/0xd90 > [ 106.315792] __alloc_pages_nodemask+0x16a/0x210 > [ 106.316148] pcpu_populate_chunk+0xce/0x300 > [ 106.316479] pcpu_balance_workfn+0x3f3/0x580 > [ 106.316853] ? _raw_spin_unlock_irq+0xe/0x30 > [ 106.317227] ? finish_task_switch+0x8d/0x250 > [ 106.317632] process_one_work+0x1b7/0x410 > [ 106.317970] worker_thread+0x26/0x3d0 > [ 106.318304] ? process_one_work+0x410/0x410 > [ 106.318649] kthread+0x10e/0x130 > [ 106.318916] ? __kthread_create_worker+0x120/0x120 > [ 106.319360] ret_from_fork+0x35/0x40 > > [ 106.453375] a.out D13400 3670 1 0x00100004 > [ 106.453880] Call Trace: > [ 106.454114] ? __schedule+0x267/0x750 > [ 106.454427] schedule+0x2d/0x90 > [ 106.454829] schedule_preempt_disabled+0xf/0x20 > [ 106.455422] __mutex_lock.isra.2+0x181/0x4d0 > [ 106.455988] ? pcpu_alloc+0x3c4/0x670 > [ 106.456465] pcpu_alloc+0x3c4/0x670 > [ 106.456973] ? preempt_count_add+0x63/0x90 > [ 106.457401] ? __local_bh_enable_ip+0x2e/0x60 > [ 106.457882] ipv6_add_dev+0x121/0x490 > [ 106.458330] addrconf_notify+0x27b/0x9a0 > [ 106.458823] ? inetdev_init+0xd7/0x150 > [ 106.459270] ? inetdev_event+0x339/0x4b0 > [ 106.459738] ? preempt_count_add+0x63/0x90 > [ 106.460243] ? _raw_spin_lock_irq+0xf/0x30 > [ 106.460747] ? notifier_call_chain+0x42/0x60 > [ 106.461271] notifier_call_chain+0x42/0x60 > [ 106.461819] register_netdevice+0x415/0x530 > [ 106.462364] register_netdev+0x11/0x20 > [ 106.462849] loopback_net_init+0x43/0x90 > [ 106.463216] ops_init+0x3b/0x100 > [ 106.463516] setup_net+0x7d/0x150 > [ 106.463831] copy_net_ns+0x14b/0x180 > [ 106.464134] create_new_namespaces+0x117/0x1b0 > [ 106.464481] unshare_nsproxy_namespaces+0x5b/0x90 > [ 106.464864] SyS_unshare+0x1b0/0x300 > > [ 106.536845] Kernel panic - not syncing: Out of memory and no killable processes... These two stacks of this problem are not blocked at mutex_lock(). Why all OOM-killable threads were killed? There were only few? Does pcpu_alloc() allocate so much enough to deplete memory reserves?
