[PATCH] Calling check_system_tsc_reliable before unsynchronized_tsc

[PATCH] Calling check_system_tsc_reliable before unsynchronized_tsc

[Zhenzhong Duan]

unsynchronized_tsc() checks value of tsc_clocksource_reliable which is set by
check_system_tsc_reliable(). It's better to move check_system_tsc_reliable() at
front, or else this check makes no sense.

Though X86_FEATURE_CONSTANT_TSC is usually set for TSC reliable system, just in
case.

Signed-off-by: Zhenzhong Duan <zhenzhong.duan@oracle.com>
---
  arch/x86/kernel/tsc.c |    4 ++--
  1 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/arch/x86/kernel/tsc.c b/arch/x86/kernel/tsc.c
index 714dfba..a316bdd 100644
--- a/arch/x86/kernel/tsc.c
+++ b/arch/x86/kernel/tsc.c
@@ -1412,11 +1412,11 @@ void __init tsc_init(void)
  
  	use_tsc_delay();
  
+	check_system_tsc_reliable();
+
  	if (unsynchronized_tsc())
  		mark_tsc_unstable("TSCs unsynchronized");
  
-	check_system_tsc_reliable();
-
  	detect_art();
  }
  
-- 
1.7.3

Re: [PATCH] Calling check_system_tsc_reliable before unsynchronized_tsc

[Thomas Gleixner]

On Thu, 8 Jun 2017, Zhenzhong Duan wrote:

> unsynchronized_tsc() checks value of tsc_clocksource_reliable which is set by
> check_system_tsc_reliable(). It's better to move check_system_tsc_reliable()
> at
> front, or else this check makes no sense.
> 
> Though X86_FEATURE_CONSTANT_TSC is usually set for TSC reliable system, just
> in
> case.
> 
> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@oracle.com>
> ---
>  arch/x86/kernel/tsc.c |    4 ++--
>  1 files changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/arch/x86/kernel/tsc.c b/arch/x86/kernel/tsc.c
> index 714dfba..a316bdd 100644
> --- a/arch/x86/kernel/tsc.c
> +++ b/arch/x86/kernel/tsc.c
> @@ -1412,11 +1412,11 @@ void __init tsc_init(void)
>   	use_tsc_delay();
>  +	check_system_tsc_reliable();
> +
>  	if (unsynchronized_tsc())
>  		mark_tsc_unstable("TSCs unsynchronized");
>  -	check_system_tsc_reliable();
> -

What kind of patch is that? Definitely nothing which can be applied.

Thanks,

	tglx

Re: [PATCH] Calling check_system_tsc_reliable before unsynchronized_tsc

[Zhenzhong Duan]

在 2017/6/13 0:03, Thomas Gleixner 写道:
> On Thu, 8 Jun 2017, Zhenzhong Duan wrote:
>
>> unsynchronized_tsc() checks value of tsc_clocksource_reliable which is set by
>> check_system_tsc_reliable(). It's better to move check_system_tsc_reliable()
>> at
>> front, or else this check makes no sense.
>>
>> Though X86_FEATURE_CONSTANT_TSC is usually set for TSC reliable system, just
>> in
>> case.
>>
>> Signed-off-by: Zhenzhong Duan <zhenzhong.duan@oracle.com>
>> ---
>>   arch/x86/kernel/tsc.c |    4 ++--
>>   1 files changed, 2 insertions(+), 2 deletions(-)
>>
>> diff --git a/arch/x86/kernel/tsc.c b/arch/x86/kernel/tsc.c
>> index 714dfba..a316bdd 100644
>> --- a/arch/x86/kernel/tsc.c
>> +++ b/arch/x86/kernel/tsc.c
>> @@ -1412,11 +1412,11 @@ void __init tsc_init(void)
>>    	use_tsc_delay();
>>   +	check_system_tsc_reliable();
>> +
>>   	if (unsynchronized_tsc())
>>   		mark_tsc_unstable("TSCs unsynchronized");
>>   -	check_system_tsc_reliable();
>> -
> What kind of patch is that? Definitely nothing which can be applied.
To avoid using unassigned variable tsc_clocksource_reliable.
Or what's the purpose of if (tsc_clocksource_reliable) check in 
unsynchronized_tsc()?

-- 
thanks
zduan

Re: [PATCH] Calling check_system_tsc_reliable before unsynchronized_tsc

[Thomas Gleixner]

[-- Attachment #1: Type: text/plain, Size: 1740 bytes --]

On Tue, 13 Jun 2017, Zhenzhong Duan wrote:
> 在 2017/6/13 0:03, Thomas Gleixner 写道:
> > On Thu, 8 Jun 2017, Zhenzhong Duan wrote:
> > 
> > > unsynchronized_tsc() checks value of tsc_clocksource_reliable which is set
> > > by
> > > check_system_tsc_reliable(). It's better to move
> > > check_system_tsc_reliable()
> > > at
> > > front, or else this check makes no sense.
> > > 
> > > Though X86_FEATURE_CONSTANT_TSC is usually set for TSC reliable system,
> > > just
> > > in
> > > case.
> > > 
> > > Signed-off-by: Zhenzhong Duan <zhenzhong.duan@oracle.com>
> > > ---
> > >   arch/x86/kernel/tsc.c |    4 ++--
> > >   1 files changed, 2 insertions(+), 2 deletions(-)
> > > 
> > > diff --git a/arch/x86/kernel/tsc.c b/arch/x86/kernel/tsc.c
> > > index 714dfba..a316bdd 100644
> > > --- a/arch/x86/kernel/tsc.c
> > > +++ b/arch/x86/kernel/tsc.c
> > > @@ -1412,11 +1412,11 @@ void __init tsc_init(void)
> > >    	use_tsc_delay();
> > >   +	check_system_tsc_reliable();
> > > +
> > >   	if (unsynchronized_tsc())
> > >   		mark_tsc_unstable("TSCs unsynchronized");
> > >   -	check_system_tsc_reliable();
> > > -
> > What kind of patch is that? Definitely nothing which can be applied.
> To avoid using unassigned variable tsc_clocksource_reliable.
> Or what's the purpose of if (tsc_clocksource_reliable) check in
> unsynchronized_tsc()?

That's not the question. The patch format itself is broken:

@@ -1412,11 +1412,11 @@ void __init tsc_init(void)
    	use_tsc_delay();
  +	check_system_tsc_reliable();
+
  	if (unsynchronized_tsc())
   		mark_tsc_unstable("TSCs unsynchronized");
  -	check_system_tsc_reliable();
-

That is not a patch created with diff -u or any other mechanism
which creates properly formatted patches.

Thanks,

	tglx