On 2022/3/5 4:28 上午, Zhouyi Zhou wrote:
> Hi Paul
>
> Yes, your suggestion works!
> Thank you for your guidance, this is the first time I ever did a
> bisection, I gained greatly during this process!
Hi Zhouyi, thanks for the bisection test.
Could this patch solve RCU warning? I just tested on my machine and there
is no RCU warning any more.
diff --git a/include/linux/cgroup.h b/include/linux/cgroup.h
index 1e356c222756..0d1ada8968d7 100644
--- a/include/linux/cgroup.h
+++ b/include/linux/cgroup.h
@@ -450,6 +450,7 @@ extern struct mutex cgroup_mutex;
extern spinlock_t css_set_lock;
#define task_css_set_check(task, __c) \
rcu_dereference_check((task)->cgroups, \
+ rcu_read_lock_sched_held() || \
lockdep_is_held(&cgroup_mutex) || \
lockdep_is_held(&css_set_lock) || \
((task)->flags & PF_EXITING) || (__c))
Thanks.
>
> On Sat, Mar 5, 2022 at 3:44 AM Paul E. McKenney <paulmck@xxxxxxxxxx> wrote:
>>
>> On Sat, Mar 05, 2022 at 03:38:33AM +0800, Zhouyi Zhou wrote:
>>> Hi,
>>> I can reproduce these warnings on my X86 VM.
>>> Following the my backtrace:
>>>
>>> (gdb) bt
>>> #0 lockdep_rcu_suspicious (file=file@entry=0xffffffff825ea8ff
>>> "include/linux/cgroup.h", line=line@entry=494,
>>> s=s@entry=0xffffffff825c5c50 "suspicious rcu_dereference_check()
>>> usage") at ./arch/x86/include/asm/current.h:15
>>> #1 0xffffffff81183103 in task_css (subsys_id=2,
>>> task=0xffff888100804080) at ./include/linux/cgroup.h:494
>>> #2 task_ca (tsk=0xffff888100804080) at kernel/sched/cpuacct.c:40
>>> #3 cpuacct_charge (tsk=tsk@entry=0xffff888100804080,
>>> cputime=cputime@entry=3344803) at kernel/sched/cpuacct.c:342
>>> #4 0xffffffff81162655 in cgroup_account_cputime (delta_exec=3344803,
>>> task=0xffff888100804080) at ./include/linux/cgroup.h:792
>>> #5 update_curr (cfs_rq=cfs_rq@entry=0xffff88813b63f500) at
>>> kernel/sched/fair.c:907
>>> #6 0xffffffff81164797 in dequeue_entity (flags=10,
>>> se=0xffff888100804100, cfs_rq=0xffff88813b63f500) at
>>> kernel/sched/fair.c:5771
>>> #7 dequeue_task_fair (rq=0xffff88813b63f440, p=0xffff888100804080,
>>> flags=10) at kernel/sched/fair.c:5771
>>> #8 0xffffffff8115412a in dequeue_task (flags=10,
>>> p=0xffff888100804080, rq=0xffff88813b63f440) at
>>> kernel/sched/core.c:2019
>>> #9 __do_set_cpus_allowed (p=0xffff888100804080,
>>> new_mask=0xffffffff831b4d40 <housekeeping+512>, flags=0) at
>>> kernel/sched/core.c:2508
>>> #10 0xffffffff811564ca in __set_cpus_allowed_ptr_locked
>>> (p=p@entry=0xffff888100804080,
>>> new_mask=new_mask@entry=0xffffffff831b4d40 <housekeeping+512>,
>>> flags=flags@entry=0, rq=0xffff88813b63f440,
>>> rf=rf@entry=0xffffc9000012bee8) at kernel/sched/core.c:2841
>>> #11 0xffffffff81156573 in __set_cpus_allowed_ptr
>>> (p=p@entry=0xffff888100804080, new_mask=0xffffffff831b4d40
>>> <housekeeping+512>, flags=flags@entry=0)
>>> at kernel/sched/core.c:2874
>>> #12 0xffffffff8115664c in set_cpus_allowed_ptr
>>> (p=p@entry=0xffff888100804080, new_mask=<optimized out>) at
>>> kernel/sched/core.c:2879
>>> #13 0xffffffff81144676 in kthreadd (unused=<optimized out>) at
>>> kernel/kthread.c:724
>>> #14 0xffffffff810019df in ret_from_fork () at arch/x86/entry/entry_64.S:295
>>> #15 0x0000000000000000 in ?? ()
>>>
>>> Do the warnings have something to do with commit
>>> dc6e0818bc9a0336d9accf3ea35d146d72aa7a18 (sched/cpuacct: Optimize away
>>> RCU read lock) ?
>>
>> If you have not already done so, could you please try running on this
>> commit and then on the commit immediately preceding it? Just as a
>> authoritative way to answer your question. ;-)
>>
>> Thanx, Paul
>>
> git reset --hard dc6e0818bc9a0336d9accf3ea35d146d72aa7a18
> make -j 16 bindeb-pkg
> there are RCU WARNINGs:
> http://154.223.142.244/logs/20220305/log.dc6e0818bc9a0336d9accf3ea35d146d72aa7a18.txt
>
> then I running on the commit immediately preceding it
> git reset --hard 248cc9993d1cc12b8e9ed716cc3fc09f6c3517dd
> make -j 16 bindeb-pkg
> the WARNINGs are gone with the wind ;-)
> http://154.223.142.244/logs/20220305/log.248cc9993d1cc12b8e9ed716cc3fc09f6c3517dd.txt
>
> Many thanks
> Zhouyi
>>> Many thanks
>>> Zhouyi
>>>
>>> On Fri, Mar 4, 2022 at 10:43 PM Naresh Kamboju
>>> <naresh.kamboju@xxxxxxxxxx> wrote:
>>>>
>>>> While booting x86_64 with linux next-20220304 kernel the following kernel
>>>> warning reported [1].
>>>>
>>>> metadata:
>>>> git_ref: master
>>>> git_repo: https://gitlab.com/Linaro/lkft/mirrors/next/linux-next
>>>> git_sha: 6d284ba80c0c485bcaa7bdf67f232d572640edc4
>>>> git_describe: next-20220304
>>>> kernel-config: https://builds.tuxbuild.com/25uuYdm2vTO275aZ3IewZY5sKbU/config
>>>>
>>>>
>>>> warning log:
>>>> ---------
>>>> [ 1.482171] MDS: Mitigation: Clear CPU buffers
>>>> [ 1.485680] Freeing SMP alternatives memory: 52K
>>>> [ 1.487341]
>>>> [ 1.488169] =============================
>>>> [ 1.488169] WARNING: suspicious RCU usage
>>>> [ 1.488169] 5.17.0-rc6-next-20220304 #1 Not tainted
>>>> [ 1.488169] -----------------------------
>>>> [ 1.488169] include/linux/cgroup.h:494 suspicious
>>>> rcu_dereference_check() usage!
>>>> [ 1.488169]
>>>> [ 1.488169] other info that might help us debug this:
>>>> [ 1.488169]
>>>> [ 1.488169]
>>>> [ 1.488169] rcu_scheduler_active = 1, debug_locks = 1
>>>> [ 1.488169] 2 locks held by kthreadd/2:
>>>> [ 1.488169] #0: ffff9ba440352330 (&p->pi_lock){....}-{2:2}, at:
>>>> task_rq_lock+0x2e/0x130
>>>> [ 1.488169] #1: ffff9ba7a7a2d058 (&rq->__lock){-...}-{2:2}, at:
>>>> task_rq_lock+0x5d/0x130
>>>> [ 1.488169]
>>>> [ 1.488169] stack backtrace:
>>>> [ 1.488169] CPU: 0 PID: 2 Comm: kthreadd Not tainted
>>>> 5.17.0-rc6-next-20220304 #1
>>>> [ 1.488169] Hardware name: Supermicro SYS-5019S-ML/X11SSH-F, BIOS
>>>> 2.5 11/26/2020
>>>> [ 1.488169] Call Trace:
>>>> [ 1.488169] <TASK>
>>>> [ 1.488169] dump_stack_lvl+0x49/0x5e
>>>> [ 1.488169] dump_stack+0x10/0x12
>>>> [ 1.488169] lockdep_rcu_suspicious+0xed/0xf8
>>>> [ 1.488169] cpuacct_charge+0x10c/0x120
>>>> [ 1.488169] update_curr+0x165/0x340
>>>> [ 1.488169] dequeue_entity+0x23/0x430
>>>> [ 1.488169] dequeue_task_fair+0xba/0x3b0
>>>> [ 1.488169] __do_set_cpus_allowed+0xca/0x2c0
>>>> [ 1.488169] __set_cpus_allowed_ptr_locked+0x128/0x1b0
>>>> [ 1.488169] set_cpus_allowed_ptr+0x43/0x70
>>>> [ 1.488169] kthreadd+0x49/0x230
>>>> [ 1.488169] ? kthread_is_per_cpu+0x30/0x30
>>>> [ 1.488169] ret_from_fork+0x22/0x30
>>>> [ 1.488169] </TASK>
>>>> [ 1.488169]
>>>> [ 1.488169] =============================
>>>> [ 1.488169] WARNING: suspicious RCU usage
>>>> [ 1.488169] 5.17.0-rc6-next-20220304 #1 Not tainted
>>>> [ 1.488169] -----------------------------
>>>> [ 1.488169] include/linux/cgroup.h:481 suspicious
>>>> rcu_dereference_check() usage!
>>>> [ 1.488169]
>>>> [ 1.488169] other info that might help us debug this:
>>>> [ 1.488169]
>>>> [ 1.488169]
>>>> [ 1.488169] rcu_scheduler_active = 1, debug_locks = 1
>>>> [ 1.488169] 2 locks held by kthreadd/2:
>>>> [ 1.488169] #0: ffff9ba440352330 (&p->pi_lock){....}-{2:2}, at:
>>>> task_rq_lock+0x2e/0x130
>>>> [ 1.488169] #1: ffff9ba7a7a2d058 (&rq->__lock){-...}-{2:2}, at:
>>>> task_rq_lock+0x5d/0x130
>>>> [ 1.488169]
>>>> [ 1.488169] stack backtrace:
>>>> [ 1.488169] CPU: 0 PID: 2 Comm: kthreadd Not tainted
>>>> 5.17.0-rc6-next-20220304 #1
>>>> [ 1.488169] Hardware name: Supermicro SYS-5019S-ML/X11SSH-F, BIOS
>>>> 2.5 11/26/2020
>>>> [ 1.488169] Call Trace:
>>>> [ 1.488169] <TASK>
>>>> [ 1.488169] dump_stack_lvl+0x49/0x5e
>>>> [ 1.488169] dump_stack+0x10/0x12
>>>> [ 1.488169] lockdep_rcu_suspicious+0xed/0xf8
>>>> [ 1.488169] update_curr+0x2b7/0x340
>>>> [ 1.488169] dequeue_entity+0x23/0x430
>>>> [ 1.488169] dequeue_task_fair+0xba/0x3b0
>>>> [ 1.488169] __do_set_cpus_allowed+0xca/0x2c0
>>>> [ 1.488169] __set_cpus_allowed_ptr_locked+0x128/0x1b0
>>>> [ 1.488169] set_cpus_allowed_ptr+0x43/0x70
>>>> [ 1.488169] kthreadd+0x49/0x230
>>>> [ 1.488169] ? kthread_is_per_cpu+0x30/0x30
>>>> [ 1.488169] ret_from_fork+0x22/0x30
>>>> [ 1.488169] </TASK>
>>>> [ 1.488169]
>>>> [ 1.488169] =============================
>>>> [ 1.488169] WARNING: suspicious RCU usage
>>>> [ 1.488169] 5.17.0-rc6-next-20220304 #1 Not tainted
>>>> [ 1.488169] -----------------------------
>>>> [ 1.488169] include/linux/cgroup.h:481 suspicious
>>>> rcu_dereference_check() usage!
>>>> [ 1.488169]
>>>> [ 1.488169] other info that might help us debug this:
>>>> [ 1.488169]
>>>> [ 1.488169]
>>>> [ 1.488169] rcu_scheduler_active = 1, debug_locks = 1
>>>> [ 1.488169] no locks held by kthreadd/2.
>>>> [ 1.488169]
>>>> [ 1.488169] stack backtrace:
>>>> [ 1.488169] CPU: 0 PID: 2 Comm: kthreadd Not tainted
>>>> 5.17.0-rc6-next-20220304 #1
>>>> [ 1.488169] Hardware name: Supermicro SYS-5019S-ML/X11SSH-F, BIOS
>>>> 2.5 11/26/2020
>>>> [ 1.488169] Call Trace:
>>>> [ 1.488169] <IRQ>
>>>> [ 1.488169] dump_stack_lvl+0x49/0x5e
>>>> [ 1.488169] dump_stack+0x10/0x12
>>>> [ 1.488169] lockdep_rcu_suspicious+0xed/0xf8
>>>> [ 1.488169] account_system_index_time+0x127/0x130
>>>> [ 1.488169] account_system_time+0x50/0x60
>>>> [ 1.488169] account_process_tick+0x59/0x80
>>>> [ 1.488169] update_process_times+0x58/0xe0
>>>> [ 1.488169] tick_periodic+0x37/0xf0
>>>> [ 1.488169] tick_handle_periodic+0x24/0x70
>>>> [ 1.488169] timer_interrupt+0x18/0x20
>>>> [ 1.488169] __handle_irq_event_percpu+0x95/0x2f0
>>>> [ 1.488169] handle_irq_event+0x39/0x80
>>>> [ 1.488169] handle_edge_irq+0xa4/0x240
>>>> [ 1.488169] __common_interrupt+0x8d/0x170
>>>> [ 1.488169] common_interrupt+0xbd/0xe0
>>>> [ 1.488169] </IRQ>
>>>> [ 1.488169] <TASK>
>>>> [ 1.488169] asm_common_interrupt+0x1e/0x40
>>>> [ 1.488169] RIP: 0010:_raw_spin_unlock_irqrestore+0x37/0x60
>>>> [ 1.488169] Code: fc 48 83 c7 18 53 48 89 f3 48 8b 75 08 e8 21 83
>>>> ed fe 4c 89 e7 e8 19 b0 ed fe 80 e7 02 74 06 e8 6f 77 fa fe fb bf 01
>>>> 00 00 00 <e8> 24 6b e9 fe 65 8b 05 bd e1 28 66 85 c0 74 05 5b 41 5c 5d
>>>> c3 0f
>>>> [ 1.488169] RSP: 0000:ffffb75b0002fd50 EFLAGS: 00000202
>>>> [ 1.488169] RAX: 0000000000000007 RBX: 0000000000000246 RCX: 0000000000000000
>>>> [ 1.488169] RDX: 0000000000000000 RSI: ffffffff9a800a7c RDI: 0000000000000001
>>>> [ 1.488169] RBP: ffffb75b0002fd60 R08: 0000000000000001 R09: 0000000000000001
>>>> [ 1.488169] R10: 0000000000000002 R11: 0000000000000001 R12: ffff9ba440352318
>>>> [ 1.488169] R13: 0000000000000000 R14: 0000000000000000 R15: ffff9ba7a7a2d040
>>>> [ 1.488169] affine_move_task+0x2d4/0x5d0
>>>> [ 1.488169] ? __this_cpu_preempt_check+0x13/0x20
>>>> [ 1.488169] ? lock_is_held_type+0xdd/0x130
>>>> [ 1.488169] ? enqueue_entity+0x1b8/0x520
>>>> [ 1.488169] __set_cpus_allowed_ptr_locked+0x15c/0x1b0
>>>> [ 1.488169] set_cpus_allowed_ptr+0x43/0x70
>>>> [ 1.488169] kthreadd+0x49/0x230
>>>> [ 1.488169] ? kthread_is_per_cpu+0x30/0x30
>>>> [ 1.488169] ret_from_fork+0x22/0x30
>>>> [ 1.488169] </TASK>
>>>> [ 1.489211] smpboot: Estimated ratio of average max frequency by
>>>> base frequency (times 1024): 1126
>>>> [ 1.490189] smpboot: CPU0: Intel(R) Xeon(R) CPU E3-1220 v6 @
>>>> 3.00GHz (family: 0x6, model: 0x9e, stepping: 0x9)
>>>> [ 1.491635] cblist_init_generic: Setting adjustable number of
>>>> callback queues.
>>>> [ 1.492171] cblist_init_generic: Setting shift to 2 and lim to 1.
>>>> [ 1.493226] cblist_init_generic: Setting shift to 2 and lim to 1.
>>>> [ 1.494226] cblist_init_generic: Setting shift to 2 and lim to 1.
>>>> [ 1.495219] Running RCU-tasks wait API self tests
>>>> [ 1.598317] Performance Events: PEBS fmt3+, Skylake events, 32-deep
>>>> LBR, full-width counters, Intel PMU driver.
>>>> [ 1.599176] ... version: 4
>>>> [ 1.600171] ... bit width: 48
>>>> [ 1.601171] ... generic registers: 8
>>>> [ 1.602171] ... value mask: 0000ffffffffffff
>>>> [ 1.603171] ... max period: 00007fffffffffff
>>>> [ 1.604176] ... fixed-purpose events: 3
>>>> [ 1.605171] ... event mask: 00000007000000ff
>>>> [ 1.606188] Callback from call_rcu_tasks_trace() invoked.
>>>> [ 1.607450] rcu: Hierarchical SRCU implementation.
>>>> [ 1.608310]
>>>> [ 1.609169] =============================
>>>> [ 1.609169] WARNING: suspicious RCU usage
>>>> [ 1.609169] 5.17.0-rc6-next-20220304 #1 Not tainted
>>>> [ 1.609169] -----------------------------
>>>> [ 1.609169] include/linux/cgroup.h:481 suspicious
>>>> rcu_dereference_check() usage!
>>>> [ 1.609169]
>>>> [ 1.609169] other info that might help us debug this:
>>>> [ 1.609169]
>>>> [ 1.609169]
>>>> [ 1.609169] rcu_scheduler_active = 1, debug_locks = 1
>>>> [ 1.609169] 1 lock held by migration/0/17:
>>>> [ 1.609169] #0: ffff9ba7a7a2d058 (&rq->__lock){-.-.}-{2:2}, at:
>>>> __schedule+0x12d/0xcb0
>>>> [ 1.609169]
>>>> [ 1.609169] stack backtrace:
>>>> [ 1.609169] CPU: 0 PID: 17 Comm: migration/0 Not tainted
>>>> 5.17.0-rc6-next-20220304 #1
>>>> [ 1.609169] Hardware name: Supermicro SYS-5019S-ML/X11SSH-F, BIOS
>>>> 2.5 11/26/2020
>>>> [ 1.609169] Stopper: 0x0 <- 0x0
>>>> [ 1.609169] Call Trace:
>>>> [ 1.609169] <TASK>
>>>> [ 1.609169] dump_stack_lvl+0x49/0x5e
>>>> [ 1.609169] dump_stack+0x10/0x12
>>>> [ 1.609169] lockdep_rcu_suspicious+0xed/0xf8
>>>> [ 1.609169] put_prev_task_stop+0x1dc/0x240
>>>> [ 1.609169] __schedule+0x751/0xcb0
>>>> [ 1.609169] ? trace_preempt_off+0x29/0xc0
>>>> [ 1.609169] ? smpboot_thread_fn+0x33/0x290
>>>> [ 1.609169] schedule+0x58/0xc0
>>>> [ 1.609169] smpboot_thread_fn+0xec/0x290
>>>> [ 1.609169] ? sort_range+0x30/0x30
>>>> [ 1.609169] kthread+0x107/0x130
>>>> [ 1.609169] ? kthread_complete_and_exit+0x20/0x20
>>>> [ 1.609169] ret_from_fork+0x22/0x30
>>>> [ 1.609169] </TASK>
>>>> [ 1.610276] smp: Bringing up secondary CPUs ...
>>>> [ 1.611596] x86: Booting SMP configuration:
>>>> [ 1.612179] .... node #0, CPUs: #1 #2 #3
>>>> [ 1.614396] smp: Brought up 1 node, 4 CPUs
>>>>
>>>> Reported-by: Linux Kernel Functional Testing <lkft@xxxxxxxxxx>
>>>>
>>>> --
>>>> Linaro LKFT
>>>> https://lkft.linaro.org
>>>>
>>>> [1] https://qa-reports.linaro.org/lkft/linux-next-master/build/next-20220304/testrun/8235955/suite/linux-log-parser/test/check-kernel-warning-4655400/log
