On 1/25/2021 5:29 PM, Hillf Danton wrote:
> On 25 Jan 2021 16:31:32 +0800 Xing Zhengjun wrote:
>> On 1/22/2021 3:59 PM, Hillf Danton wrote:
>>> On Fri, 22 Jan 2021 09:48:32 +0800 Xing Zhengjun wrote:
>>>> On 1/21/2021 12:00 PM, Hillf Danton wrote:
>>>>> On Wed, 20 Jan 2021 21:46:33 +0800 Oliver Sang wrote:
>>>>>> On Fri, Jan 15, 2021 at 03:24:32PM +0800, Hillf Danton wrote:
>>>>>>> Thu, 14 Jan 2021 15:45:11 +0800
>>>>>>>>
>>>>>>>> FYI, we noticed the following commit (built with gcc-9):
>>>>>>>>
>>>>>>>> commit: d5bff968ea9cc005e632d9369c26cbd8148c93d5 ("workqueue: break affinity initiatively")
>>>>>>>> https://git.kernel.org/cgit/linux/kernel/git/paulmck/linux-rcu.git dev.2021.01.11b
>>>>>>>>
>>>>>>> [...]
>>>>>>>>
>>>>>>>> [   73.794288] WARNING: CPU: 0 PID: 22 at kernel/workqueue.c:2192 process_one_work
>>>>>>>
>>>>>>> Thanks for your report.
>>>>>>>
>>>>>>> We can also break CPU affinity by checking POOL_DISASSOCIATED at attach
>>>>>>> time without extra cost paid; that way we have the same behavior as at
>>>>>>> the unbind time.
>>>>>>>
>>>>>>> What is more the change that makes kworker pcpu is cut because they are
>>>>>>> going to not help either hotplug or the mechanism of stop machine.
>>>>>>
>>>>>> hi, by applying below patch, the issue still happened.
>>>>>
>>>>> Thanks for your report.
>>>>>>
>>>>>> [ 4.574467] pci 0000:00:00.0: Limiting direct PCI/PCI transfers
>>>>>> [ 4.575651] pci 0000:00:01.0: Activating ISA DMA hang workarounds
>>>>>> [ 4.576900] pci 0000:00:02.0: Video device with shadowed ROM at [mem 0x000c0000-0x000dffff]
>>>>>> [ 4.578648] PCI: CLS 0 bytes, default 64
>>>>>> [ 4.579685] Unpacking initramfs...
>>>>>> [ 8.878031] -----------[ cut here ]-----------
>>>>>> [ 8.879083] WARNING: CPU: 0 PID: 22 at kernel/workqueue.c:2187 process_one_work+0x92/0x9e0
>>>>>> [ 8.880688] Modules linked in:
>>>>>> [ 8.881274] CPU: 0 PID: 22 Comm: kworker/1:0 Not tainted 5.11.0-rc3-gc213503139bb #2
>>>>>
>>>>> The kworker bond to CPU1 runs on CPU0 and triggers the warning, which
>>>>> shows that scheduler breaks CPU affinity, after 06249738a41a
>>>>> ("workqueue: Manually break affinity on hotplug"), though quite likely
>>>>> by kworker/1:0 for the initial workers.
>>>>>
>>>>>> [ 8.882518] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.12.0-1 04/01/2014
>>>>>> [ 8.887539] Workqueue: 0x0 (events)
>>>>>> [ 8.887838] EIP: process_one_work+0x92/0x9e0
>>>>>> [ 8.887838] Code: 37 64 a1 58 54 4c 43 39 45 24 74 2c 31 c9 ba 01 00 00 00 c7 04 24 01 00 00 00 b8 08 1d f5 42 e8 74 85 13 00 ff 05 b8 30 04 43 <0f> 0b ba 01 00 00 00 eb 22 8d 74 26 00 90 c7 04 24 01 00 00 00 31
>>>>>> [ 8.887838] EAX: 42f51d08 EBX: 00000000 ECX: 00000000 EDX: 00000001
>>>>>> [ 8.887838] ESI: 43c04720 EDI: 42e45620 EBP: de7f23c0 ESP: 43d7bf08
>>>>>> [ 8.887838] DS: 007b ES: 007b FS: 00d8 GS: 0000 SS: 0068 EFLAGS: 00010002
>>>>>> [ 8.887838] CR0: 80050033 CR2: 00000000 CR3: 034e3000 CR4: 000406d0
>>>>>> [ 8.887838] Call Trace:
>>>>>> [ 8.887838] ? worker_thread+0x98/0x6a0
>>>>>> [ 8.887838] ? worker_thread+0x2dd/0x6a0
>>>>>> [ 8.887838] ? kthread+0x1ba/0x1e0
>>>>>> [ 8.887838] ? create_worker+0x1e0/0x1e0
>>>>>> [ 8.887838] ? kzalloc+0x20/0x20
>>>>>> [ 8.887838] ? ret_from_fork+0x1c/0x28
>>>>>> [ 8.887838] _warn_unseeded_randomness: 63 callbacks suppressed
>>>>>> [ 8.887838] random: get_random_bytes called from init_oops_id+0x2b/0x60 with crng_init=0
>>>>>> [ 8.887838] --[ end trace ac461b4d54c37cfa ]--
>>>>>
>>>>>
>>>>> Instead of creating the initial workers only on the active CPUS, rebind
>>>>> them (labeled pcpu) and jump to the right CPU at bootup time.
>>>>>
>>>>> --- a/kernel/workqueue.c
>>>>> +++ b/kernel/workqueue.c
>>>>> @@ -2385,6 +2385,16 @@ woke_up:
>>>>>     		return 0;
>>>>>     	}
>>>>>     
>>>>> +	if (!(pool->flags & POOL_DISASSOCIATED) && smp_processor_id() !=
>>>>> +								pool->cpu) {
>>>>> +		/* scheduler breaks CPU affinity for us, rebind it */
>>>>> +		raw_spin_unlock_irq(&pool->lock);
>>>>> +		set_cpus_allowed_ptr(worker->task, pool->attrs->cpumask);
>>>>> +		/* and jump to the right seat */
>>>>> +		schedule_timeout_interruptible(1);
>>>>> +		goto woke_up;
>>>>> +	}
>>>>> +
>>>>>     	worker_leave_idle(worker);
>>>>>     recheck:
>>>>>     	/* no more worker necessary? */
>>>>> --
>>>>>
>>>> I test the patch, the warning still appears in the kernel log.
>>>
>>> Thanks for your report.
>>>>
>>>> [  230.356503] smpboot: CPU 1 is now offline
>>>> [  230.544652] x86: Booting SMP configuration:
>>>> [  230.545077] smpboot: Booting Node 0 Processor 1 APIC 0x1
>>>> [  230.545640] kvm-clock: cpu 1, msr 34f6021, secondary cpu clock
>>>> [  230.545675] masked ExtINT on CPU#1
>>>> [  230.593829] ------------[ cut here ]------------
>>>> [  230.594257] WARNING: CPU: 0 PID: 257 at kernel/workqueue.c:2192 process_one_work+0x92/0x9e0
>>>> [  230.594990] Modules linked in: rcutorture torture mousedev input_leds
>>>> led_class pcspkr psmouse evbug tiny_power_button button
>>>> [  230.595961] CPU: 0 PID: 257 Comm: kworker/1:3 Not tainted 5.11.0-rc3-gdcba55d9080f #2
>>>
>>> Like what was reported, kworker bond to CPU1 runs on CPU0 and triggers
>>> warning, due to scheduler breaking CPU affinity for us. What is new, the
>>> affinity was broken at offline time instead of bootup.
>>>
>>>> [  230.596621] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.12.0-1 04/01/2014
>>>> [  230.597322] Workqueue:  0x0 (rcu_gp)
>>>> [  230.597636] EIP: process_one_work+0x92/0x9e0
>>>> [  230.598005] Code: 37 64 a1 58 54 4c 43 39 45 24 74 2c 31 c9 ba 01 00
>>>> 00 00 c7 04 24 01 00 00 00 b8 08 1d f5 42 e8 f4 85 13 00 ff 05 cc 30 04
>>>> 43 <0f> 0b ba 01 00 00 00 eb 22 8d 74 26 00 90 c7 04 24 01 00 00 00 31
>>>> [  230.599569] EAX: 42f51d08 EBX: 00000000 ECX: 00000000 EDX: 00000001
>>>> [  230.600100] ESI: 43d94240 EDI: df4040f4 EBP: de7f23c0 ESP: bf5f1f08
>>>> [  230.600629] DS: 007b ES: 007b FS: 00d8 GS: 0000 SS: 0068 EFLAGS: 00010002
>>>> [  230.601203] CR0: 80050033 CR2: 01bdecbc CR3: 04e2c000 CR4: 000406d0
>>>> [  230.601735] DR0: 00000000 DR1: 00000000 DR2: 00000000 DR3: 00000000
>>>> [  230.602265] DR6: fffe0ff0 DR7: 00000400
>>>> [  230.602594] Call Trace:
>>>> [  230.602813]  ? process_one_work+0x20e/0x9e0
>>>> [  230.603181]  ? worker_thread+0x32d/0x700
>>>> [  230.603522]  ? kthread+0x1ba/0x1e0
>>>> [  230.603818]  ? create_worker+0x1e0/0x1e0
>>>> [  230.604157]  ? kzalloc+0x20/0x20
>>>> [  230.604524]  ? ret_from_fork+0x1c/0x28
>>>> [  230.604850] ---[ end trace 06b1e66b5e17fa85 ]---
>>>> [  230.605504] kvm-guest: stealtime: cpu 1, msr 9e7e6ec0
>>>> [  230.766960] smpboot: CPU 1 is now offline
>>>> [  230.814803] x86: Booting SMP configuration:
>>>> [  230.815306] smpboot: Booting Node 0 Processor 1 APIC 0x1
>>>> [  230.815964] kvm-clock: cpu 1, msr 34f6021, secondary cpu clock
>>>
>>>
>>> Unlike the above diff that is at most papering over the problem
>>> sitting somewhere in the scheduler, add change to creating worker
>>> by skipping set_cpus_allowed_ptr() because we will wake it up after
>>> attaching it to worker pool.
>>>
>>> If we can ignore rescuer for now, then the allowed ptr is only
>>> updated at on/offline time; lets see the difference at boot time.
>>>
>>>
>>> --- a/kernel/workqueue.c
>>> +++ b/kernel/workqueue.c
>>> @@ -1844,16 +1844,10 @@ static struct worker *alloc_worker(int n
>>>     * cpu-[un]hotplugs.
>>>     */
>>>    static void worker_attach_to_pool(struct worker *worker,
>>> -				   struct worker_pool *pool)
>>> +				   struct worker_pool *pool,
>>> +				   int update_cpus_allowed)
>>>    {
>>>    	mutex_lock(&wq_pool_attach_mutex);
>>> -
>>> -	/*
>>> -	 * set_cpus_allowed_ptr() will fail if the cpumask doesn't have any
>>> -	 * online CPUs.  It'll be re-applied when any of the CPUs come up.
>>> -	 */
>>> -	set_cpus_allowed_ptr(worker->task, pool->attrs->cpumask);
>>> -
>>>    	/*
>>>    	 * The wq_pool_attach_mutex ensures %POOL_DISASSOCIATED remains
>>>    	 * stable across this function.  See the comments above the flag
>>> @@ -1867,6 +1861,9 @@ static void worker_attach_to_pool(struct
>>>    	list_add_tail(&worker->node, &pool->workers);
>>>    	worker->pool = pool;
>>>    
>>> +	if (update_cpus_allowed)
>>> +		set_cpus_allowed_ptr(worker->task, pool->attrs->cpumask);
>>> +
>>>    	mutex_unlock(&wq_pool_attach_mutex);
>>>    }
>>>    
>>> @@ -1942,8 +1939,11 @@ static struct worker *create_worker(stru
>>>    	set_user_nice(worker->task, pool->attrs->nice);
>>>    	kthread_bind_mask(worker->task, pool->attrs->cpumask);
>>>    
>>> -	/* successful, attach the worker to the pool */
>>> -	worker_attach_to_pool(worker, pool);
>>> +	/*
>>> +	 * attach the worker to the pool without asking scheduler to
>>> +	 * update CPUs allowed
>>> +	 */
>>> +	worker_attach_to_pool(worker, pool, 0);
>>>    
>>>    	/* start the newly created worker */
>>>    	raw_spin_lock_irq(&pool->lock);
>>> @@ -2508,7 +2508,7 @@ repeat:
>>>    
>>>    		raw_spin_unlock_irq(&wq_mayday_lock);
>>>    
>>> -		worker_attach_to_pool(rescuer, pool);
>>> +		worker_attach_to_pool(rescuer, pool, 1);
>>>    
>>>    		raw_spin_lock_irq(&pool->lock);
>>>    
>>> --
>>>
>> I test the patch, the warning still appears in the kernel log.
> 
> Thanks.
>>
>> [   55.754187] smpboot: Booting Node 0 Processor 1 APIC 0x1
>> [   55.785594] kvm-clock: cpu 1, msr 34f6021, secondary cpu clock
>> [   55.785646] masked ExtINT on CPU#1
>> [   55.920602] ------------[ cut here ]------------
>> [   55.921355] WARNING: CPU: 0 PID: 160 at kernel/workqueue.c:2192 process_one_work+0x92/0x9e0
>> [   55.922583] Modules linked in: rcutorture torture mousedev evbug
>> input_leds led_class tiny_power_button psmouse pcspkr button
>> [   55.924294] CPU: 0 PID: 160 Comm: kworker/1:2 Not tainted 5.11.0-rc3-00186-g77bf4e461cfa #2
> 
> Same issue as before.
> 
>> [   55.925552] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.12.0-1 04/01/2014
>> [   55.926763] Workqueue:  0x0 (rcu_gp)
>> [   55.927298] EIP: process_one_work+0x92/0x9e0
>> [   55.927950] Code: 37 64 a1 58 54 4c 43 39 45 24 74 2c 31 c9 ba 01 00
>> 00 00 c7 04 24 01 00 00 00 b8 08 1d f5 42 e8 94 85 13 00 ff 05 b8 30 04
>> 43 <0f> 0b ba 01 00 00 00 eb 22 8d 74 26 00 90 c7 04 24 01 00 00 00 31
>> [   55.930726] EAX: 42f51d08 EBX: 00000000 ECX: 00000000 EDX: 00000001
>> [   55.931642] ESI: 43d90540 EDI: df48c0f4 EBP: de7f23c0 ESP: bfb47f08
>> [   55.932590] DS: 007b ES: 007b FS: 00d8 GS: 0000 SS: 0068 EFLAGS: 00010002
>> [   55.933609] CR0: 80050033 CR2: 024e994c CR3: 7fd80000 CR4: 000406d0
>> [   55.934555] DR0: 00000000 DR1: 00000000 DR2: 00000000 DR3: 00000000
>> [   55.935457] DR6: fffe0ff0 DR7: 00000400
>> [   55.936041] Call Trace:
>> [   55.936534]  ? process_one_work+0x20e/0x9e0
>> [   55.937305]  ? worker_thread+0x2dd/0x6a0
>> [   55.938018]  ? kthread+0x1ba/0x1e0
>> [   55.938598]  ? create_worker+0x1e0/0x1e0
>> [   55.939315]  ? kzalloc+0x20/0x20
>> [   55.940000]  ? ret_from_fork+0x1c/0x28
>> [   55.940627] ---[ end trace d155e9e6402de179 ]---
>> [   55.941641] kvm-guest: stealtime: cpu 1, msr 9e7e6ec0
>> [   56.155271] smpboot: CPU 1 is now offline
>> [   56.193613] x86: Booting SMP configuration:
>> [   56.194400] smpboot: Booting Node 0 Processor 1 APIC 0x1
> 
> The changes in the diff below are
> 
> 1/ at rescue time, change CPU affinity only if POOL_DISASSOCIATED
> is not set, and print warning the same way as offline time.
> 
> 2/ at offine time, dont update allowed CPUs after setting
> POOL_DISASSOCIATED because we no longer have interest in affinity.
> 
> 3/ at online time, mark pcpu before binding affinity.
> 
> Though one change a diff is appreciated, by the WARNs, we can tell
> which is what if any warning goes into dmesg.
> 
> --- a/kernel/workqueue.c
> +++ b/kernel/workqueue.c
> @@ -1844,25 +1844,23 @@ static struct worker *alloc_worker(int n
>    * cpu-[un]hotplugs.
>    */
>   static void worker_attach_to_pool(struct worker *worker,
> -				   struct worker_pool *pool)
> +				   struct worker_pool *pool, int set)
>   {
>   	mutex_lock(&wq_pool_attach_mutex);
>   
>   	/*
> -	 * set_cpus_allowed_ptr() will fail if the cpumask doesn't have any
> -	 * online CPUs.  It'll be re-applied when any of the CPUs come up.
> -	 */
> -	set_cpus_allowed_ptr(worker->task, pool->attrs->cpumask);
> -
> -	/*
>   	 * The wq_pool_attach_mutex ensures %POOL_DISASSOCIATED remains
>   	 * stable across this function.  See the comments above the flag
>   	 * definition for details.
>   	 */
>   	if (pool->flags & POOL_DISASSOCIATED)
>   		worker->flags |= WORKER_UNBOUND;
> -	else
> +	else {
>   		kthread_set_per_cpu(worker->task, true);
> +		if (set)
> +			WARN_ON_ONCE(set_cpus_allowed_ptr(worker->task,
> +						pool->attrs->cpumask) < 0);
> +	}
>   
>   	list_add_tail(&worker->node, &pool->workers);
>   	worker->pool = pool;
> @@ -1943,7 +1941,7 @@ static struct worker *create_worker(stru
>   	kthread_bind_mask(worker->task, pool->attrs->cpumask);
>   
>   	/* successful, attach the worker to the pool */
> -	worker_attach_to_pool(worker, pool);
> +	worker_attach_to_pool(worker, pool, 0);
>   
>   	/* start the newly created worker */
>   	raw_spin_lock_irq(&pool->lock);
> @@ -2508,7 +2506,7 @@ repeat:
>   
>   		raw_spin_unlock_irq(&wq_mayday_lock);
>   
> -		worker_attach_to_pool(rescuer, pool);
> +		worker_attach_to_pool(rescuer, pool, 1);
>   
>   		raw_spin_lock_irq(&pool->lock);
>   
> @@ -4923,7 +4921,6 @@ static void unbind_workers(int cpu)
>   
>   		for_each_pool_worker(worker, pool) {
>   			kthread_set_per_cpu(worker->task, false);
> -			WARN_ON_ONCE(set_cpus_allowed_ptr(worker->task, cpu_possible_mask) < 0);
>   		}
>   
>   		mutex_unlock(&wq_pool_attach_mutex);
> @@ -4977,9 +4974,9 @@ static void rebind_workers(struct worker
>   	 * from CPU_ONLINE, the following shouldn't fail.
>   	 */
>   	for_each_pool_worker(worker, pool) {
> +		kthread_set_per_cpu(worker->task, true);
>   		WARN_ON_ONCE(set_cpus_allowed_ptr(worker->task,
>   						  pool->attrs->cpumask) < 0);
> -		kthread_set_per_cpu(worker->task, true);
>   	}
>   
>   	raw_spin_lock_irq(&pool->lock);
> --
> 
I test the patch, the warning still appears in the kernel log, but the 
warning is different from before.

[    0.054803] kvm-clock: cpu 1, msr 34f6021, secondary cpu clock
[    0.054803] masked ExtINT on CPU#1
[    0.054803] smpboot: CPU 1 Converting physical 0 to logical die 1
[    1.890338] ------------[ cut here ]------------
[    1.890338] WARNING: CPU: 1 PID: 18 at kernel/kthread.c:508 
kthread_set_per_cpu+0x156/0x180
[    1.890338] Modules linked in:
[    1.890338] CPU: 1 PID: 18 Comm: cpuhp/1 Not tainted 
5.11.0-rc3-00186-ged03082352b2 #2
[    1.890338] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), 
BIOS 1.12.0-1 04/01/2014
[    1.890338] EIP: kthread_set_per_cpu+0x156/0x180
[    1.890338] Code: 00 00 00 00 ff 05 68 4e 04 43 83 c4 08 5b 5e 5f c3 
8d 76 00 ff 05 34 50 04 43 0f 0b e9 f9 fe ff ff 8d 76 00 ff 05 2c 4e 04 
43 <0f> 0b eb 9d 8d b6 00 00 00 00 ff 05 40 4e 04 43 0f 0b e9 45 ff ff
[    1.890338] EAX: 42f52ce0 EBX: 00000001 ECX: 00000000 EDX: 00000001
[    1.890338] ESI: 43d76300 EDI: 43c0de00 EBP: de7f2564 ESP: 43d6beb8
[    1.900350] DS: 007b ES: 007b FS: 00d8 GS: 0000 SS: 0068 EFLAGS: 00010202
[    1.901303] CR0: 80050033 CR2: 00000000 CR3: 034e3000 CR4: 000406d0
[    1.902280] Call Trace:
[    1.902682]  ? workqueue_online_cpu+0x12b/0x640
[    1.903415]  ? workqueue_prepare_cpu+0xa0/0xa0
[    1.904155]  ? cpuhp_invoke_callback+0x1ed/0x1340
[    1.904941]  ? cpuhp_thread_fun+0x28f/0x460
[    1.905630]  ? cpuhp_thread_fun+0x49/0x460
[    1.906298]  ? smpboot_thread_fn+0x446/0x620
[    1.910275]  ? kthread+0x1ba/0x1e0
[    1.910857]  ? __smpboot_create_thread+0x260/0x260
[    1.911659]  ? kzalloc+0x20/0x20
[    1.912368]  ? ret_from_fork+0x1c/0x28
[    1.913016] ---[ end trace 6f6c005278241eba ]---
[    1.913971] kvm-guest: stealtime: cpu 1, msr 9e7e6ec0
[    1.920012] smp: Brought up 1 node, 2 CPUs
[    1.920299] smpboot: Max logical packages: 2
[    1.921019] smpboot: Total of 2 processors activated (10774.03 BogoMIPS)



-- 
Zhengjun Xing