[PATCH] tick/sched: check skew_tick kernel parameter's validity

[PATCH] tick/sched: check skew_tick kernel parameter's validity

[Sergey Shtylyov]

In skew_tick(), the result of get_option() call is ignored, so even if a
value of the skew_tick kernel parameter was not specified (or specified
as a list or range), the kernel won't complain. Add the missing check...

Found by Linux Verification Center (linuxtesting.org) with the Svace static
analysis tool.

Signed-off-by: Sergey Shtylyov <s.shtylyov@auroraos.dev>

---
The patch is against the timers/nohz branch of the tip.git repo...

 kernel/time/tick-sched.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/kernel/time/tick-sched.c b/kernel/time/tick-sched.c
index c1ee0b256445..34ceccdd49b5 100644
--- a/kernel/time/tick-sched.c
+++ b/kernel/time/tick-sched.c
@@ -1483,7 +1483,8 @@ static int sched_skew_tick;
 
 static int __init skew_tick(char *str)
 {
-	get_option(&str, &sched_skew_tick);
+	if (get_option(&str, &sched_skew_tick) != 1)
+		return -EINVAL;
 
 	return 0;
 }
-- 
2.55.0

Re: [PATCH] tick/sched: check skew_tick kernel parameter's validity

[Frederic Weisbecker]

Hi,

Le Sat, Aug 29, 2026 at 11:25:35PM +0300, Sergey Shtylyov a écrit :
> In skew_tick(), the result of get_option() call is ignored, so even if a
> value of the skew_tick kernel parameter was not specified (or specified
> as a list or range), the kernel won't complain. Add the missing check...
> 
> Found by Linux Verification Center (linuxtesting.org) with the Svace static
> analysis tool.
> 
> Signed-off-by: Sergey Shtylyov <s.shtylyov@auroraos.dev>
> 
> ---
> The patch is against the timers/nohz branch of the tip.git repo...
> 
>  kernel/time/tick-sched.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/kernel/time/tick-sched.c b/kernel/time/tick-sched.c
> index c1ee0b256445..34ceccdd49b5 100644
> --- a/kernel/time/tick-sched.c
> +++ b/kernel/time/tick-sched.c
> @@ -1483,7 +1483,8 @@ static int sched_skew_tick;
>  
>  static int __init skew_tick(char *str)
>  {
> -	get_option(&str, &sched_skew_tick);
> +	if (get_option(&str, &sched_skew_tick) != 1)
> +		return -EINVAL;
>  
>  	return 0;

I'm not sure that matters. Initcall return values seem to be ignored on boot.
Looks like only modules do care.

Thanks.

>  }
> -- 
> 2.55.0

-- 
Frederic Weisbecker
SUSE Labs

Re: [PATCH] tick/sched: check skew_tick kernel parameter's validity

[Sergey Shtylyov]

On 9/1/26 3:55 PM, Frederic Weisbecker wrote:
[...]

>> In skew_tick(), the result of get_option() call is ignored, so even if a
>> value of the skew_tick kernel parameter was not specified (or specified
>> as a list or range), the kernel won't complain. Add the missing check...
>>
>> Found by Linux Verification Center (linuxtesting.org) with the Svace static
>> analysis tool.
>>
>> Signed-off-by: Sergey Shtylyov <s.shtylyov@auroraos.dev>
>>
>> ---
>> The patch is against the timers/nohz branch of the tip.git repo...
>>
>>  kernel/time/tick-sched.c | 3 ++-
>>  1 file changed, 2 insertions(+), 1 deletion(-)
>>
>> diff --git a/kernel/time/tick-sched.c b/kernel/time/tick-sched.c
>> index c1ee0b256445..34ceccdd49b5 100644
>> --- a/kernel/time/tick-sched.c
>> +++ b/kernel/time/tick-sched.c
>> @@ -1483,7 +1483,8 @@ static int sched_skew_tick;
>>
>>  static int __init skew_tick(char *str)
>>  {
>> -     get_option(&str, &sched_skew_tick);
>> +     if (get_option(&str, &sched_skew_tick) != 1)
>> +             return -EINVAL;
>>
>>       return 0;
> 
> I'm not sure that matters. Initcall return values seem to be ignored on boot.

   But that's not an initcall, IIUC.
   Looking at do_early_param(), the kernel would call pr_warn() if the setup
fucntion fails...

> Looks like only modules do care.

[...]

MBR, Sergey

Re: [PATCH] tick/sched: check skew_tick kernel parameter's validity

[Thomas Gleixner]

On Sat, Aug 29 2026 at 23:25, Sergey Shtylyov wrote:
> In skew_tick(), the result of get_option() call is ignored, so even if a
> value of the skew_tick kernel parameter was not specified (or specified
> as a list or range), the kernel won't complain. Add the missing check...
>
> Found by Linux Verification Center (linuxtesting.org) with the Svace static
> analysis tool.
>
> Signed-off-by: Sergey Shtylyov <s.shtylyov@auroraos.dev>
>
> ---
> The patch is against the timers/nohz branch of the tip.git repo...
>
>  kernel/time/tick-sched.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
>
> diff --git a/kernel/time/tick-sched.c b/kernel/time/tick-sched.c
> index c1ee0b256445..34ceccdd49b5 100644
> --- a/kernel/time/tick-sched.c
> +++ b/kernel/time/tick-sched.c
> @@ -1483,7 +1483,8 @@ static int sched_skew_tick;
>  
>  static int __init skew_tick(char *str)
>  {
> -	get_option(&str, &sched_skew_tick);
> +	if (get_option(&str, &sched_skew_tick) != 1)
> +		return -EINVAL;
>  
>  	return 0;
>  }

sched_skew_tick should be converted to bool and this function should
simply do:

       return kstrtobool(str, &sched_skew_tick);

And to make it complete, sched_skew_tick should be annotated with
ro_after_init.

Thanks,

        tglx

Re: [PATCH] tick/sched: check skew_tick kernel parameter's validity

[Sergey Shtylyov]

On 9/29/26 11:01 PM, Thomas Gleixner wrote:
[...]>> In skew_tick(), the result of get_option() call is ignored, so even if a
>> value of the skew_tick kernel parameter was not specified (or specified
>> as a list or range), the kernel won't complain. Add the missing check...
>>
>> Found by Linux Verification Center (linuxtesting.org) with the Svace static
>> analysis tool.
>>
>> Signed-off-by: Sergey Shtylyov <s.shtylyov@auroraos.dev>
>>
>> ---
>> The patch is against the timers/nohz branch of the tip.git repo...
>>
>>  kernel/time/tick-sched.c | 3 ++-
>>  1 file changed, 2 insertions(+), 1 deletion(-)
>>
>> diff --git a/kernel/time/tick-sched.c b/kernel/time/tick-sched.c
>> index c1ee0b256445..34ceccdd49b5 100644
>> --- a/kernel/time/tick-sched.c
>> +++ b/kernel/time/tick-sched.c
>> @@ -1483,7 +1483,8 @@ static int sched_skew_tick;
>>
>>  static int __init skew_tick(char *str)
>>  {
>> -     get_option(&str, &sched_skew_tick);
>> +     if (get_option(&str, &sched_skew_tick) != 1)
>> +             return -EINVAL;
>>
>>       return 0;
>>  }
> 
> sched_skew_tick should be converted to bool and this function should
> simply do:
> 
>        return kstrtobool(str, &sched_skew_tick);

   Thanks for your comment!
   OK, I'll do that (and add the Suggested-by: tag, of/c)...

> And to make it complete, sched_skew_tick should be annotated with
> ro_after_init.

   I guess you meant __ro_after_init? Would it then be OK adding it
in the same patch?

> Thanks,
> 
>         tglx

MBR, Sergey

Re: [PATCH] tick/sched: check skew_tick kernel parameter's validity

[Thomas Gleixner]

On Wed, Sep 30 2026 at 22:01, Sergey Shtylyov wrote:
> On 9/29/26 11:01 PM, Thomas Gleixner wrote:
> [...]>> In skew_tick(), the result of get_option() call is ignored, so even if a
>>> value of the skew_tick kernel parameter was not specified (or specified
>>> as a list or range), the kernel won't complain. Add the missing check...
>>>
>>> Found by Linux Verification Center (linuxtesting.org) with the Svace static
>>> analysis tool.
>>>
>>> Signed-off-by: Sergey Shtylyov <s.shtylyov@auroraos.dev>
>>>
>>> ---
>>> The patch is against the timers/nohz branch of the tip.git repo...
>>>
>>>  kernel/time/tick-sched.c | 3 ++-
>>>  1 file changed, 2 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/kernel/time/tick-sched.c b/kernel/time/tick-sched.c
>>> index c1ee0b256445..34ceccdd49b5 100644
>>> --- a/kernel/time/tick-sched.c
>>> +++ b/kernel/time/tick-sched.c
>>> @@ -1483,7 +1483,8 @@ static int sched_skew_tick;
>>>
>>>  static int __init skew_tick(char *str)
>>>  {
>>> -     get_option(&str, &sched_skew_tick);
>>> +     if (get_option(&str, &sched_skew_tick) != 1)
>>> +             return -EINVAL;
>>>
>>>       return 0;
>>>  }
>> 
>> sched_skew_tick should be converted to bool and this function should
>> simply do:
>> 
>>        return kstrtobool(str, &sched_skew_tick);
>
>    Thanks for your comment!
>    OK, I'll do that (and add the Suggested-by: tag, of/c)...
>
>> And to make it complete, sched_skew_tick should be annotated with
>> ro_after_init.
>
>    I guess you meant __ro_after_init? Would it then be OK adding it
> in the same patch?

As you change the data type of the variable anyway that's fine. Just
mention it in the changelog.

Thanks,

        tglx

Re: [PATCH] tick/sched: check skew_tick kernel parameter's validity

[Sergey Shtylyov]

On 9/30/26 10:36 PM, Thomas Gleixner wrote:

[...]

>>> And to make it complete, sched_skew_tick should be annotated with
>>> ro_after_init.
>>
>>    I guess you meant __ro_after_init? Would it then be OK adding it
>> in the same patch?
> 
> As you change the data type of the variable anyway that's fine. Just
> mention it in the changelog.

   Oh, I know the drill! :-) 
> Thanks,
> 
>         tglx

MBR, Sergey

Re: [PATCH] tick/sched: check skew_tick kernel parameter's validity

[Sergey Shtylyov]

On 9/30/26 10:01 PM, Sergey Shtylyov wrote:
[...]

>>> In skew_tick(), the result of get_option() call is ignored, so even if a
>>> value of the skew_tick kernel parameter was not specified (or specified
>>> as a list or range), the kernel won't complain. Add the missing check...
>>>
>>> Found by Linux Verification Center (linuxtesting.org) with the Svace static
>>> analysis tool.
>>>
>>> Signed-off-by: Sergey Shtylyov <s.shtylyov@auroraos.dev>
>>>
>>> ---
>>> The patch is against the timers/nohz branch of the tip.git repo...
>>>
>>>  kernel/time/tick-sched.c | 3 ++-
>>>  1 file changed, 2 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/kernel/time/tick-sched.c b/kernel/time/tick-sched.c
>>> index c1ee0b256445..34ceccdd49b5 100644
>>> --- a/kernel/time/tick-sched.c
>>> +++ b/kernel/time/tick-sched.c
>>> @@ -1483,7 +1483,8 @@ static int sched_skew_tick;
>>>
>>>  static int __init skew_tick(char *str)
>>>  {
>>> -     get_option(&str, &sched_skew_tick);
>>> +     if (get_option(&str, &sched_skew_tick) != 1)
>>> +             return -EINVAL;
>>>
>>>       return 0;
>>>  }
>>
>> sched_skew_tick should be converted to bool and this function should
>> simply do:
>>
>>        return kstrtobool(str, &sched_skew_tick);
> 
>    Thanks for your comment!
>    OK, I'll do that (and add the Suggested-by: tag, of/c)...

   One thing I did forget to ask: we still don't need the Fixes tag here?

[...]

>> Thanks,
>>
>>         tglx

MBR, Sergey

Re: [PATCH] tick/sched: check skew_tick kernel parameter's validity

[Thomas Gleixner]

On Wed, Sep 30 2026 at 23:04, Sergey Shtylyov wrote:
> On 9/30/26 10:01 PM, Sergey Shtylyov wrote:
>>    Thanks for your comment!
>>    OK, I'll do that (and add the Suggested-by: tag, of/c)...
>
>    One thing I did forget to ask: we still don't need the Fixes tag here?

I have no strong opinion about it, but it's not really fixing a real
world problem aside of preventing a sysadmin to shoot himself itno the
foot. :)