Re: [RFC, v2] powerpc/powernv: Introduce kernel param to control fastsleep workaround behavior

[Shreyas B Prabhu <shreyas@linux.vnet.ibm.com>]

On Tuesday 17 March 2015 03:09 PM, Benjamin Herrenschmidt wrote:
> On Tue, 2015-03-17 at 19:57 +1100, Michael Ellerman wrote:
>>
>> So my first preference is that you just bite the bullet and decide to either
>> always apply the workaround, or just stick with the current behaviour. That's a
>> trade-off between (I think) better idle latency but a risk of checkstops, vs.
>> slower idle latency but less (how much less?) risk of checkstops.
>>
>> I think the reason you're proposing a kernel parameter is because we aren't
>> willing to make that decision, ie. we're saying that users should decide. Is
>> that right?
> 
> Correct. More specifically, a fairly high profile user that I will not
> name here has expressed interest in such a feature...
> 
>> I'm not a big fan of kernel parameters. They are a pain to use, and are often
>> just pushing a decision down one layer for no reason. What I mean is that
>> individual users are probably just going to accept whatever the default value
>> is from their distro.
> 
> Right. This is quite an obscure tunable.
> 
>> But anyway, that's a bit of a rant.
>>
>> As far as this patch is concerned, I don't think it actually needs to be a
>> kernel parameter.
>>
>> >From what I can see below, the decision as to whether you apply the workaround
>> or not doesn't affect the list of idle states. So this could just as well be a
>> runtime parameter, ie. a sysfs file, which can then be set by the user whenever
>> they like? They might do it in a boot script, but that's up to them.
> 
> Right, that would work too.

Okay. I'll send a patch with this design.
> 
>> For simplicity I think it would also be fine to make it a write-once parameter,
>> ie. you don't need to handle undoing it.
> 
> It would be easy enough to make it rw using stop machine I think... 
> 
>> I think the only complication that would add is that you'd need to be a little
>> careful about the order in which you nop out the calls vs applying the
>> workaround, in case some threads are idle when you're called.

Right, we should be safe with this sequence-
- NOP call to undo workaround
- Apply workaround on all cores.
- NOP call to apply workaround

> 
> I wouldn't bother with NOP'ing in that case, a runtime test will probably be noise
> in the measurement.
> 

Didn't get your point here. Do you mean, ignore the request if some
cores are in sleep or deeper state?
> Cheers,
> Ben.
> 
>