Re: [patch RFC 00/15] mm/highmem: Provide a preemptible variant of kmap_atomic & friends

From: Thomas Gleixner
Date: Thu Sep 24 2020 - 02:57:57 EST


On Wed, Sep 23 2020 at 17:12, Steven Rostedt wrote:
> On Wed, 23 Sep 2020 22:55:54 +0200
> Then scratch the idea of having anonymous local_lock() and just bring
> local_lock in directly? Then have a kmap local lock, which would only
> block those that need to do a kmap.

That's still going to end up in lock ordering nightmares and you lose
the ability to use kmap_local from arbitrary contexts which was again
one of the goals of this exercise.

Aside of that you're imposing reentrancy protections on something which
does not need it in the first place.

> Now as for migration disabled nesting, at least now we would have
> groupings of this, and perhaps the theorists can handle that. I mean,
> how is this much different that having a bunch of tasks blocked on a
> mutex with the owner is pinned on a CPU?
>
> migrate_disable() is a BKL of pinning affinity.

No. That's just wrong. preempt disable is a concurrency control,
i.e. protecting against reentrancy on a given CPU. But it's a cpu global
protection which means that it's not protecting a specific code path.

Contrary to preempt disable, migrate disable is not protecting against
reentrancy on a given CPU. It's a temporary restriction to the scheduler
on placement.

The fact that disabling preemption implicitely disables migration does
not make them semantically equivalent.

> If we only have local_lock() available (even on !RT), then it makes
> the blocking in groups. At least this way you could grep for all the
> different local_locks in the system and plug that into the algorithm
> for WCS, just like one would with a bunch of mutexes.

You cannot do that on RT at all where migrate disable is substituting
preempt disable in spin and rw locks. The result would be the same as
with a !RT kernel just with horribly bad performance.

That means the stacking problem has to be solved anyway.

So why on earth do you want to create yet another special duct tape case
for kamp_local() which proliferates inconsistency instead of aiming for
consistency accross all preemption models?

Thanks,

tglx