Re: [PATCH v3] ACPI / Sleep: Check low power idle constraints for debug only

[Rafael J. Wysocki]

On Sun, Aug 13, 2017 at 2:00 AM, Srinivas Pandruvada
<srinivas.pandruvada@xxxxxxxxxxxxxxx> wrote:
> On Sun, 2017-08-13 at 00:37 +0200, Rafael J. Wysocki wrote:
>> On Sat, Aug 12, 2017 at 5:59 PM, Srinivas Pandruvada
>> <srinivas.pandruvada@xxxxxxxxxxxxxxx> wrote:
>> >
>> > On Sat, 2017-08-12 at 16:27 +0200, Rafael J. Wysocki wrote:
>> > [...]
>> >
>> > >
>> > > >
>> > > > +
>> > > > +struct lpi_constraints {
>> > > > +   char *name;
>> > > > +   int min_dstate;
>> > > If you store the handle here as well, you won't need to
>> > > look it up every time _check_constraints() is called.
>> > The reason I didn't keep handle here, I thought handle can be stale
>> > or
>> > change for PnP device on plug in and out. Is this not true?
>> The handles don't go away on hot remove as a rule.  That may only
>> happen if tables get unloaded, but basically the constraints should
>> not point to anything in a table that may go away.
>
> So we don't need to worry about this case where tables gets unloaded
> and replaced?

I wouldn't. :-)

We generally don't support that case entirely, so why bother?

> This is in a debug path, so additional overhead of path
> to handle conversion may not be significant.

Well, it still is wasted cycles ...