Re: [NEEDS-REVIEW] [PATCH V3 04/10] x86/pks: Preserve the PKRS MSR on context switch

[Dave Hansen]

On 12/17/20 8:10 PM, Ira Weiny wrote:
> On Thu, Dec 17, 2020 at 12:41:50PM -0800, Dave Hansen wrote:
>> On 11/6/20 3:29 PM, ira.weiny@xxxxxxxxx wrote:
>>>  void disable_TSC(void)
>>> @@ -644,6 +668,8 @@ void __switch_to_xtra(struct task_struct *prev_p, struct task_struct *next_p)
>>>  
>>>  	if ((tifp ^ tifn) & _TIF_SLD)
>>>  		switch_to_sld(tifn);
>>> +
>>> +	pks_sched_in();
>>>  }
>>
>> Does the selftest for this ever actually schedule()?
> 
> At this point I'm not sure.  This code has been in since the beginning.  So its
> seen a lot of soak time.

Think about it another way.  Let's say this didn't get called on the
first context switch away from the PKS-using task.  Would anyone notice?
 How likely is this to happen?

The function tracers or kprobes tend to be a great tool for this, at
least for testing whether the code path you expect to hit is getting hit.