Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.kernel > #1725457 > unrolled thread

[PATCH v2] sched: check user input value of sysctl_sched_time_avg

Back to article view | Back to linux.kernel

Contents

  [PATCH v2] sched: check user input value of sysctl_sched_time_avg Ethan Zhao <ethan.zhao@oracle.com> - 2017-09-02 09:00 +0200
    Re: [PATCH v2] sched: check user input value of  sysctl_sched_time_avg Mike Galbraith <efault@gmx.de> - 2017-09-02 09:20 +0200
    Re: [PATCH v2] sched: check user input value of sysctl_sched_time_avg Peter Zijlstra <peterz@infradead.org> - 2017-09-04 09:50 +0200
      Re: [PATCH v2] sched: check user input value of sysctl_sched_time_avg Ethan Zhao <ethan.zhao@oracle.com> - 2017-09-04 10:00 +0200
        Re: [PATCH v2] sched: check user input value of sysctl_sched_time_avg "Luis R. Rodriguez" <mcgrof@kernel.org> - 2017-09-06 22:00 +0200
          Re: [PATCH v2] sched: check user input value of sysctl_sched_time_avg Ethan Zhao <ethan.zhao@oracle.com> - 2017-09-07 06:30 +0200

#1725457 — [PATCH v2] sched: check user input value of sysctl_sched_time_avg

System will hang if user set sysctl_sched_time_avg to 0 by

[root@XXX ~]# sysctl kernel.sched_time_avg_ms=0

Stack traceback for pid 0
0xffff883f6406c600 0 0 1 3 R 0xffff883f6406cf50 *swapper/3
ffff883f7ccc3ae8 0000000000000018 ffffffff810c4dd0 0000000000000000
0000000000017800 ffff883f7ccc3d78 0000000000000003 ffff883f7ccc3bf8
ffffffff810c4fc9 ffff883f7ccc3c08 00000000810c5043 ffff883f7ccc3c08
Call Trace:
<IRQ> [<ffffffff810c4dd0>] ? update_group_capacity+0x110/0x200
[<ffffffff810c4fc9>] ? update_sd_lb_stats+0x109/0x600
[<ffffffff810c5507>] ? find_busiest_group+0x47/0x530
[<ffffffff810c5b84>] ? load_balance+0x194/0x900
[<ffffffff810ad5ca>] ? update_rq_clock.part.83+0x1a/0xe0
[<ffffffff810c6d42>] ? rebalance_domains+0x152/0x290
[<ffffffff810c6f5c>] ? run_rebalance_domains+0xdc/0x1d0
[<ffffffff8108a75b>] ? __do_softirq+0xfb/0x320
[<ffffffff8108ac85>] ? irq_exit+0x125/0x130
[<ffffffff810b3a17>] ? scheduler_ipi+0x97/0x160
[<ffffffff81052709>] ? smp_reschedule_interrupt+0x29/0x30
[<ffffffff8173a1be>] ? reschedule_interrupt+0x6e/0x80
 <EOI> [<ffffffff815bc83c>] ? cpuidle_enter_state+0xcc/0x230
[<ffffffff815bc80c>] ? cpuidle_enter_state+0x9c/0x230
[<ffffffff815bc9d7>] ? cpuidle_enter+0x17/0x20
[<ffffffff810cd6dc>] ? cpu_startup_entry+0x38c/0x420
[<ffffffff81053373>] ? start_secondary+0x173/0x1e0

Because divide-by-zero error happens in function

update_group_capacity()
  update_cpu_capacity()
    scale_rt_capacity()
    {
        ...
        total = sched_avg_period() + delta;
        used = div_u64(avg, total);
        ...
    }

Seems this issue could be reproduced on all I tried stable 4.1 - last
kernel.

To fix this issue, check user input value of sysctl_sched_time_avg, keep
it unchanged when hit invalid input.

Reported-by: James Puthukattukaran <james.puthukattukaran@oracle.com>
Signed-off-by: Ethan Zhao <ethan.zhao@oracle.com>
---
 v2: check it in sysctl table (input side) as Peter suggested.
 Tested on stable 4.1, applied on stable 4.13-rc5 okay.

 include/linux/sched/sysctl.h |  3 +++
 kernel/sched/fair.c          | 26 ++++++++++++++++++++++++++
 kernel/sysctl.c              |  2 +-
 3 files changed, 30 insertions(+), 1 deletion(-)

diff --git a/include/linux/sched/sysctl.h b/include/linux/sched/sysctl.h
index 0f5ecd4..cc25807 100644
--- a/include/linux/sched/sysctl.h
+++ b/include/linux/sched/sysctl.h
@@ -44,6 +44,9 @@ enum sched_tunable_scaling {
 int sched_proc_update_handler(struct ctl_table *table, int write,
 		void __user *buffer, size_t *length,
 		loff_t *ppos);
+int sched_time_avg_handler(struct ctl_table *table, int write,
+		void __user *buffer, size_t *lenp,
+		loff_t *ppos);
 #endif
 
 /*
diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index c95880e..61155dc 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -650,6 +650,32 @@ int sched_proc_update_handler(struct ctl_table *table, int write,
 
 	return 0;
 }
+
+int sched_time_avg_handler(struct ctl_table *table, int write,
+		void __user *buffer, size_t *lenp,
+		loff_t *ppos)
+{
+	struct ctl_table tbl;
+	int new_value, ret;
+
+	memset(&tbl, 0, sizeof(struct ctl_table));
+	tbl.maxlen = sizeof(unsigned int);
+
+	if (write)
+		tbl.data = &new_value;
+	else
+		tbl.data = &sysctl_sched_time_avg;
+
+	ret = proc_dointvec(&tbl, write, buffer, lenp, ppos);
+	if (write && ret == 0) {
+		if (new_value <= 0)
+			ret =  -EINVAL;
+		else
+			sysctl_sched_time_avg = new_value;
+	}
+
+	return ret;
+}
 #endif
 
 /*
diff --git a/kernel/sysctl.c b/kernel/sysctl.c
index 6648fbb..609bed2 100644
--- a/kernel/sysctl.c
+++ b/kernel/sysctl.c
@@ -367,7 +367,7 @@ static int sysrq_sysctl_handler(struct ctl_table *table, int write,
 		.data		= &sysctl_sched_time_avg,
 		.maxlen		= sizeof(unsigned int),
 		.mode		= 0644,
-		.proc_handler	= proc_dointvec,
+		.proc_handler	= sched_time_avg_handler,
 	},
 #ifdef CONFIG_SCHEDSTATS
 	{
-- 
1.8.3.1

[toc] | [next] | [standalone]

#1725462 — Re: [PATCH v2] sched: check user input value of sysctl_sched_time_avg

On Sat, 2017-09-02 at 14:57 +0800, Ethan Zhao wrote:
> System will hang if user set sysctl_sched_time_avg to 0 by
> 
> [root@XXX ~]# sysctl kernel.sched_time_avg_ms=0

What's wrong with using proc_dointvec_minmax?

	-Mike

[toc] | [prev] | [next] | [standalone]

#1725898

On Sat, Sep 02, 2017 at 02:57:32PM +0800, Ethan Zhao wrote:
> diff --git a/kernel/sysctl.c b/kernel/sysctl.c
> index 6648fbb..609bed2 100644
> --- a/kernel/sysctl.c
> +++ b/kernel/sysctl.c
> @@ -367,7 +367,7 @@ static int sysrq_sysctl_handler(struct ctl_table *table, int write,
>  		.data		= &sysctl_sched_time_avg,
>  		.maxlen		= sizeof(unsigned int),
>  		.mode		= 0644,
> -		.proc_handler	= proc_dointvec,
> +		.proc_handler	= sched_time_avg_handler,

*sigh*, what's wrong with the below? Too easy?

diff --git a/kernel/sysctl.c b/kernel/sysctl.c
index 6648fbbb8157..bbbc6a17c15e 100644
--- a/kernel/sysctl.c
+++ b/kernel/sysctl.c
@@ -367,7 +367,8 @@ static struct ctl_table kern_table[] = {
 		.data		= &sysctl_sched_time_avg,
 		.maxlen		= sizeof(unsigned int),
 		.mode		= 0644,
-		.proc_handler	= proc_dointvec,
+		.proc_handler	= proc_dointvec_min_max,
+		.extra1		= &one,
 	},
 #ifdef CONFIG_SCHEDSTATS
 	{

[toc] | [prev] | [next] | [standalone]

#1725906

Peter,


On 2017/9/4 15:49, Peter Zijlstra wrote:
> On Sat, Sep 02, 2017 at 02:57:32PM +0800, Ethan Zhao wrote:
>> diff --git a/kernel/sysctl.c b/kernel/sysctl.c
>> index 6648fbb..609bed2 100644
>> --- a/kernel/sysctl.c
>> +++ b/kernel/sysctl.c
>> @@ -367,7 +367,7 @@ static int sysrq_sysctl_handler(struct ctl_table *table, int write,
>>   		.data		= &sysctl_sched_time_avg,
>>   		.maxlen		= sizeof(unsigned int),
>>   		.mode		= 0644,
>> -		.proc_handler	= proc_dointvec,
>> +		.proc_handler	= sched_time_avg_handler,
> *sigh*, what's wrong with the below? Too easy?
  :),  seems I walked zigzag several cycles to get the right point.

  Thanks,
  Ethan
>
> diff --git a/kernel/sysctl.c b/kernel/sysctl.c
> index 6648fbbb8157..bbbc6a17c15e 100644
> --- a/kernel/sysctl.c
> +++ b/kernel/sysctl.c
> @@ -367,7 +367,8 @@ static struct ctl_table kern_table[] = {
>   		.data		= &sysctl_sched_time_avg,
>   		.maxlen		= sizeof(unsigned int),
>   		.mode		= 0644,
> -		.proc_handler	= proc_dointvec,
> +		.proc_handler	= proc_dointvec_min_max,
> +		.extra1		= &one,
>   	},
>   #ifdef CONFIG_SCHEDSTATS
>   	{
>
>

[toc] | [prev] | [next] | [standalone]

#1727699

On Mon, Sep 04, 2017 at 03:54:23PM +0800, Ethan Zhao wrote:
> Peter,
> 
> 
> On 2017/9/4 15:49, Peter Zijlstra wrote:
> > On Sat, Sep 02, 2017 at 02:57:32PM +0800, Ethan Zhao wrote:
> > > diff --git a/kernel/sysctl.c b/kernel/sysctl.c
> > > index 6648fbb..609bed2 100644
> > > --- a/kernel/sysctl.c
> > > +++ b/kernel/sysctl.c
> > > @@ -367,7 +367,7 @@ static int sysrq_sysctl_handler(struct ctl_table *table, int write,
> > >   		.data		= &sysctl_sched_time_avg,
> > >   		.maxlen		= sizeof(unsigned int),
> > >   		.mode		= 0644,
> > > -		.proc_handler	= proc_dointvec,
> > > +		.proc_handler	= sched_time_avg_handler,
> > *sigh*, what's wrong with the below? Too easy?
>  :),  seems I walked zigzag several cycles to get the right point.

It sounds like negative values are not possible too, right? If so then
you can use proc_douintvec_minmax().

  Luis

[toc] | [prev] | [next] | [standalone]

#1727902

On 2017/9/7 3:50, Luis R. Rodriguez wrote:
> On Mon, Sep 04, 2017 at 03:54:23PM +0800, Ethan Zhao wrote:
>> Peter,
>>
>>
>> On 2017/9/4 15:49, Peter Zijlstra wrote:
>>> On Sat, Sep 02, 2017 at 02:57:32PM +0800, Ethan Zhao wrote:
>>>> diff --git a/kernel/sysctl.c b/kernel/sysctl.c
>>>> index 6648fbb..609bed2 100644
>>>> --- a/kernel/sysctl.c
>>>> +++ b/kernel/sysctl.c
>>>> @@ -367,7 +367,7 @@ static int sysrq_sysctl_handler(struct ctl_table *table, int write,
>>>>    		.data		= &sysctl_sched_time_avg,
>>>>    		.maxlen		= sizeof(unsigned int),
>>>>    		.mode		= 0644,
>>>> -		.proc_handler	= proc_dointvec,
>>>> +		.proc_handler	= sched_time_avg_handler,
>>> *sigh*, what's wrong with the below? Too easy?
>>   :),  seems I walked zigzag several cycles to get the right point.
> It sounds like negative values are not possible too, right? If so then
> you can use proc_douintvec_minmax().
  V4 has been sent days ago, please see v4.

  Thank,
  Ethan
>
>    Luis
>

[toc] | [prev] | [standalone]

Back to top | Article view | linux.kernel

csiph-web