[PATCH] sys_prctl(): simplify arg2 judgment when calling PR_SET_TIMERSLACK

[PATCH] sys_prctl(): simplify arg2 judgment when calling PR_SET_TIMERSLACK

[Yang Xu]

arg2 will never < 0, for its type is 'unsigned long'. So negative
judgment is meaningless.

Signed-off-by: Yang Xu <xuyang2018.jy@cn.fujitsu.com>
---
 kernel/sys.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/kernel/sys.c b/kernel/sys.c
index 2969304c29fe..399457d26bef 100644
--- a/kernel/sys.c
+++ b/kernel/sys.c
@@ -2372,11 +2372,11 @@ SYSCALL_DEFINE5(prctl, int, option, unsigned long, arg2, unsigned long, arg3,
 			error = current->timer_slack_ns;
 		break;
 	case PR_SET_TIMERSLACK:
-		if (arg2 <= 0)
+		if (arg2)
+			current->timer_slack_ns = arg2;
+		else
 			current->timer_slack_ns =
 					current->default_timer_slack_ns;
-		else
-			current->timer_slack_ns = arg2;
 		break;
 	case PR_MCE_KILL:
 		if (arg4 | arg5)
-- 
2.18.1

Re: [PATCH] sys_prctl(): simplify arg2 judgment when calling PR_SET_TIMERSLACK

[Cyrill Gorcunov]

On Tue, Jul 23, 2019 at 11:30:53AM +0800, Yang Xu wrote:
> arg2 will never < 0, for its type is 'unsigned long'. So negative
> judgment is meaningless.
> 
> Signed-off-by: Yang Xu <xuyang2018.jy@cn.fujitsu.com>
> ---
>  kernel/sys.c | 6 +++---
>  1 file changed, 3 insertions(+), 3 deletions(-)
> 
> diff --git a/kernel/sys.c b/kernel/sys.c
> index 2969304c29fe..399457d26bef 100644
> --- a/kernel/sys.c
> +++ b/kernel/sys.c
> @@ -2372,11 +2372,11 @@ SYSCALL_DEFINE5(prctl, int, option, unsigned long, arg2, unsigned long, arg3,
>  			error = current->timer_slack_ns;
>  		break;
>  	case PR_SET_TIMERSLACK:
> -		if (arg2 <= 0)
> +		if (arg2)
> +			current->timer_slack_ns = arg2;
> +		else
>  			current->timer_slack_ns =
>  					current->default_timer_slack_ns;
> -		else
> -			current->timer_slack_ns = arg2;
>  		break;
>  	case PR_MCE_KILL:
>  		if (arg4 | arg5)

From a glance it looks correct to me, but then...

1) you might simply compare with zero, iow if (arg2 == 0)
   instead of changing 7 lines
2) according to man page passing negative value should be acceptable,
   though it never worked as expected. I've been grepping "git log"
   for this file and the former API is coming from

commit 6976675d94042fbd446231d1bd8b7de71a980ada
Author: Arjan van de Ven <arjan@linux.intel.com>
Date:   Mon Sep 1 15:52:40 2008 -0700

    hrtimer: create a "timer_slack" field in the task struct

which is 11 years old by now. Nobody complained so far even when man
page is saying pretty obviously

       PR_SET_TIMERSLACK (since Linux 2.6.28)
              Each thread has two associated timer slack values:  a  "default"
              value, and a "current" value.  This operation sets the "current"
              timer slack value for the calling  thread.   If  the  nanosecond
              value  supplied in arg2 is greater than zero, then the "current"
              value is set to this value.  If arg2 is less than  or  equal  to
              zero,  the  "current"  timer  slack  is  reset  to  the thread's
              "default" timer slack value.

So i think to match the man page (and assuming that accepting negative value
has been supposed) we should rather do

	if ((long)arg2 < 0)

Thoughts?

Re: [PATCH] sys_prctl(): simplify arg2 judgment when calling PR_SET_TIMERSLACK

[Yang Xu]

on 2019/07/23 15:23, Cyrill Gorcunov wrote:

> On Tue, Jul 23, 2019 at 11:30:53AM +0800, Yang Xu wrote:
>> arg2 will never<  0, for its type is 'unsigned long'. So negative
>> judgment is meaningless.
>>
>> Signed-off-by: Yang Xu<xuyang2018.jy@cn.fujitsu.com>
>> ---
>>   kernel/sys.c | 6 +++---
>>   1 file changed, 3 insertions(+), 3 deletions(-)
>>
>> diff --git a/kernel/sys.c b/kernel/sys.c
>> index 2969304c29fe..399457d26bef 100644
>> --- a/kernel/sys.c
>> +++ b/kernel/sys.c
>> @@ -2372,11 +2372,11 @@ SYSCALL_DEFINE5(prctl, int, option, unsigned long, arg2, unsigned long, arg3,
>>   			error = current->timer_slack_ns;
>>   		break;
>>   	case PR_SET_TIMERSLACK:
>> -		if (arg2<= 0)
>> +		if (arg2)
>> +			current->timer_slack_ns = arg2;
>> +		else
>>   			current->timer_slack_ns =
>>   					current->default_timer_slack_ns;
>> -		else
>> -			current->timer_slack_ns = arg2;
>>   		break;
>>   	case PR_MCE_KILL:
>>   		if (arg4 | arg5)
> > From a glance it looks correct to me, but then...
>
> 1) you might simply compare with zero, iow if (arg2 == 0)
>     instead of changing 7 lines
Hi Cyril

Indeed.  simply compare with zero might be better.

> 2) according to man page passing negative value should be acceptable,
>     though it never worked as expected. I've been grepping "git log"
>     for this file and the former API is coming from
>
> commit 6976675d94042fbd446231d1bd8b7de71a980ada
> Author: Arjan van de Ven<arjan@linux.intel.com>
> Date:   Mon Sep 1 15:52:40 2008 -0700
>
>      hrtimer: create a "timer_slack" field in the task struct
>
> which is 11 years old by now. Nobody complained so far even when man
> page is saying pretty obviously
>
>         PR_SET_TIMERSLACK (since Linux 2.6.28)
>                Each thread has two associated timer slack values:  a  "default"
>                value, and a "current" value.  This operation sets the "current"
>                timer slack value for the calling  thread.   If  the  nanosecond
>                value  supplied in arg2 is greater than zero, then the "current"
>                value is set to this value.  If arg2 is less than  or  equal  to
>                zero,  the  "current"  timer  slack  is  reset  to  the thread's
>                "default" timer slack value.
>
> So i think to match the man page (and assuming that accepting negative value
> has been supposed) we should rather do
>
> 	if ((long)arg2<  0)
Looks correct. But if we set a ULONG_MAX(PR_GET_TIMERSLACK also limits ULONG_MAX)
value(about 4s) on 32bit machine, this code will think this value is a negative value and use default value.

I guess man page was written as "less than or equal to zero" because of this confusing code(arg2<=0, but arg2
is an unsinged long value).
I think we can change this man page and also add bounds value description.

Also, I found a patch about arg2 is an unsigned long value

commit 7fe5e04292e71af34ae171b88caa2a139e0b6125
Author: Chen Gang<gang.chen@asianux.com>
Date:   Thu Feb 21 16:43:06 2013 -0800

     sys_prctl(): arg2 is unsigned long which is never<  0

     arg2 will never<  0, for its type is 'unsigned long'

     Also, use the provided macros.

What do you think about it ?

> Thoughts?
>
>
>

Re: [PATCH] sys_prctl(): simplify arg2 judgment when calling PR_SET_TIMERSLACK

[Cyrill Gorcunov]

On Tue, Jul 23, 2019 at 04:11:09PM +0800, Yang Xu wrote:
> > 2) according to man page passing negative value should be acceptable,
> >     though it never worked as expected. I've been grepping "git log"
> >     for this file and the former API is coming from
> > 
> > commit 6976675d94042fbd446231d1bd8b7de71a980ada
> > Author: Arjan van de Ven<arjan@linux.intel.com>
> > Date:   Mon Sep 1 15:52:40 2008 -0700
> > 
> >      hrtimer: create a "timer_slack" field in the task struct
> > 
> > which is 11 years old by now. Nobody complained so far even when man
> > page is saying pretty obviously
> > 
> >         PR_SET_TIMERSLACK (since Linux 2.6.28)
> >                Each thread has two associated timer slack values:  a  "default"
> >                value, and a "current" value.  This operation sets the "current"
> >                timer slack value for the calling  thread.   If  the  nanosecond
> >                value  supplied in arg2 is greater than zero, then the "current"
> >                value is set to this value.  If arg2 is less than  or  equal  to
> >                zero,  the  "current"  timer  slack  is  reset  to  the thread's
> >                "default" timer slack value.
> > 
> > So i think to match the man page (and assuming that accepting negative value
> > has been supposed) we should rather do
> > 
> > 	if ((long)arg2<  0)
> Looks correct. But if we set a ULONG_MAX(PR_GET_TIMERSLACK also limits ULONG_MAX)
> value(about 4s) on 32bit machine, this code will think this value is a negative value and use default value.
> 
> I guess man page was written as "less than or equal to zero" because of this confusing code(arg2<=0, but arg2
> is an unsinged long value).
> I think we can change this man page and also add bounds value description.

OK, seems reasonable. I think we should use comparision with zero
and simply update a man page.

[PATCH v2] sys_prctl(): remove unsigned comparision with less than zero

[Yang Xu]

Currently, when calling prctl(PR_SET_TIMERSLACK, arg2), arg2 is an
unsigned long value, arg2 will never < 0. Negative judgment is
meaningless, so remove it.

Fixes: 6976675d9404 ("hrtimer: create a "timer_slack" field in the task struct")
Signed-off-by: Yang Xu <xuyang2018.jy@cn.fujitsu.com>
Cc: Cyrill Gorcunov <gorcunov@gmail.com>
---
 kernel/sys.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/kernel/sys.c b/kernel/sys.c
index 2969304c29fe..701b5f00651d 100644
--- a/kernel/sys.c
+++ b/kernel/sys.c
@@ -2372,7 +2372,7 @@ SYSCALL_DEFINE5(prctl, int, option, unsigned long, arg2, unsigned long, arg3,
 			error = current->timer_slack_ns;
 		break;
 	case PR_SET_TIMERSLACK:
-		if (arg2 <= 0)
+		if (arg2 == 0)
 			current->timer_slack_ns =
 					current->default_timer_slack_ns;
 		else
-- 
2.18.1

Re: [PATCH v2] sys_prctl(): remove unsigned comparision with less than zero

[Cyrill Gorcunov]

On Wed, Jul 24, 2019 at 10:11:48AM +0800, Yang Xu wrote:
> Currently, when calling prctl(PR_SET_TIMERSLACK, arg2), arg2 is an
> unsigned long value, arg2 will never < 0. Negative judgment is
> meaningless, so remove it.
> 
> Fixes: 6976675d9404 ("hrtimer: create a "timer_slack" field in the task struct")
> Signed-off-by: Yang Xu <xuyang2018.jy@cn.fujitsu.com>
> Cc: Cyrill Gorcunov <gorcunov@gmail.com>
Acked-by: Cyrill Gorcunov <gorcunov@gmail.com>

Re: [PATCH v2] sys_prctl(): remove unsigned comparision with less than zero

[Andrew Morton]

On Wed, 24 Jul 2019 10:11:48 +0800 Yang Xu <xuyang2018.jy@cn.fujitsu.com> wrote:

> Currently, when calling prctl(PR_SET_TIMERSLACK, arg2), arg2 is an
> unsigned long value, arg2 will never < 0. Negative judgment is
> meaningless, so remove it.
> 
> ...
>
> --- a/kernel/sys.c
> +++ b/kernel/sys.c
> @@ -2372,7 +2372,7 @@ SYSCALL_DEFINE5(prctl, int, option, unsigned long, arg2, unsigned long, arg3,
>  			error = current->timer_slack_ns;
>  		break;
>  	case PR_SET_TIMERSLACK:
> -		if (arg2 <= 0)
> +		if (arg2 == 0)
>  			current->timer_slack_ns =
>  					current->default_timer_slack_ns;

A number of years ago Linus expressed approval of such comparisons with
unsigned quantities.  He felt that it improves readability a little -
the reader doesn't have to scroll back and check the type.

Re: [PATCH v2] sys_prctl(): remove unsigned comparision with less than zero

[Yang Xu]

on 2019/07/25 10:14, Andrew Morton wrote:

> On Wed, 24 Jul 2019 10:11:48 +0800 Yang Xu<xuyang2018.jy@cn.fujitsu.com>  wrote:
>
>> Currently, when calling prctl(PR_SET_TIMERSLACK, arg2), arg2 is an
>> unsigned long value, arg2 will never<  0. Negative judgment is
>> meaningless, so remove it.
>>
>> ...
>>
>> --- a/kernel/sys.c
>> +++ b/kernel/sys.c
>> @@ -2372,7 +2372,7 @@ SYSCALL_DEFINE5(prctl, int, option, unsigned long, arg2, unsigned long, arg3,
>>   			error = current->timer_slack_ns;
>>   		break;
>>   	case PR_SET_TIMERSLACK:
>> -		if (arg2<= 0)
>> +		if (arg2 == 0)
>>   			current->timer_slack_ns =
>>   					current->default_timer_slack_ns;
> A number of years ago Linus expressed approval of such comparisons with
> unsigned quantities.  He felt that it improves readability a little -
> the reader doesn't have to scroll back and check the type.
Hi Andrew

    It sounds good. ButWe still have to look at the actual situation. In here, this comparisons with unsigned
quantities doesn't improvereadability. In turn, the code give user a wrongdescription  as man page said "
If arg2 is less than or equal to zero, the "current" timer slack is reset to the thread's default" timer slack value."

If we set -1 in user space, we pass it into kernel as ULONG_MAX, it will not use default timer_slack value.

Also, I guess that if value has no actual sense we can use this comparisons. In here, arg2 represents slack time.
time will never less than 0.

ps: whether we change or not change this comparisons, it doesn't affect logic. So if you think this patch is meaningless,
I will accept it.

Thanks
Yang Xu

>
>
>
>

Re: [PATCH v2] sys_prctl(): remove unsigned comparision with less than zero

[Yang Xu]

on 2019/07/25 11:10, Yang Xu wrote:

>>> --- a/kernel/sys.c
>>> +++ b/kernel/sys.c
>>> @@ -2372,7 +2372,7 @@ SYSCALL_DEFINE5(prctl, int, option, unsigned long, arg2, unsigned long, arg3,
>>>                error = current->timer_slack_ns;
>>>            break;
>>>        case PR_SET_TIMERSLACK:
>>> -        if (arg2<= 0)
>>> +        if (arg2 == 0)
>>>                current->timer_slack_ns =
>>>                        current->default_timer_slack_ns;
>> A number of years ago Linus expressed approval of such comparisons with
>> unsigned quantities.  He felt that it improves readability a little -
>> the reader doesn't have to scroll back and check the type.
> Hi Andrew
>
>     It sounds good. ButWe still have to look at the actual situation. In here, this comparisons with unsigned
> quantities doesn't improvereadability. In turn, the code give user a wrongdescription  as man page said "
> If arg2 is less than or equal to zero, the "current" timer slack is reset to the thread's default" timer slack value."
> If we set -1 in user space, we pass it into kernel as ULONG_MAX, it will not use default timer_slack value.
> Also, I guess that if value has no actual sense we can use this comparisons. In here, arg2 represents slack time.
> time will never less than 0.
> ps: whether we change or not change this comparisons, it doesn't affect logic. So if you think this patch is meaningless,
> I will accept it.
Hi Andrew
   what do you think about it? update it or keep it.

Thanks
Yang Xu