On Sat, 5 Mar 2005, Pavel Machek wrote:
> > The issues in the previous email should be addressed before conversion can
> > be considered "complete".
>
> Umm, maybe, but we can't even switch pm_message_t to struct, yet,
> because it would break the build.
Obviously that's a necessary first step.
> > > We put all but one cpu in tight loop in smp_pause, so that it can't
> > > interfere. That's enough for STD, but not for STR.
> >
> > What goes wrong during STR?
>
> You'd have to boot secondary CPUs from real mode without touching
> memory. Not easy.
Don't you have that same problem with the primary CPU as well? Or to put
it another way, why should secondary CPUs be any harder to handle than the
primary? (Note: I know nothing about how SMP systems handle complex
things like booting...)
> > I don't know much about the scheduler. But if it can be made to ignore
> > processes that are in the refrigerator with no penalty, then it should be
> > able to ignore processes without PF_NOFREEZE also with no penalty.
>
> Well, I don't want to know about scheduler. Take a look if you can
> accomplish that. But having "if (in_suspend) { } " code in scheduler
> is going to be ugly.
>
> Ouch and it is not as easy as "ignore processes without
> PF_NOFREEZE". You need them to sleep at defined place where they do
> not hold any locks. What disadvantages do you see in refrigerator?
It's Ben who sees the disadvantages; I was merely echoing his comments.
Maybe I should bow out at this point and let him speak for himself. :-)
Alan Stern
