lkml.org 
[lkml]   [2020]   [May]   [25]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
Patch in this message
/
Date
From
SubjectRe: [PATCH v1 02/25] mm/swap: Don't abuse the seqcount latching API
Peter Zijlstra <peterz@infradead.org> wrote:
> On Tue, May 19, 2020 at 11:45:24PM +0200, Ahmed S. Darwish wrote:
> > @@ -713,10 +713,20 @@ static void lru_add_drain_per_cpu(struct work_struct *dummy)
> > */
> > void lru_add_drain_all(void)
> > {
>

Re-adding cut-out comment for context:

/*
* lru_drain_gen - Current generation of pages that could be in vectors
*
* (A) Definition: lru_drain_gen = x implies that all generations
* 0 < n <= x are already scheduled for draining.
*
* This is an optimization for the highly-contended use case where a
* user space workload keeps constantly generating a flow of pages
* for each CPU.
*/
> > + static unsigned int lru_drain_gen;
> > static struct cpumask has_work;
> > + static DEFINE_MUTEX(lock);
> > + int cpu, this_gen;
> >
> > /*
> > * Make sure nobody triggers this path before mm_percpu_wq is fully
> > @@ -725,21 +735,48 @@ void lru_add_drain_all(void)
> > if (WARN_ON(!mm_percpu_wq))
> > return;
> >
>

Re-adding cut-out comment for context:

/*
* (B) Cache the LRU draining generation number
*
* smp_rmb() ensures that the counter is loaded before the mutex is
* taken. It pairs with the smp_wmb() inside the mutex critical section
* at (D).
*/
> > + this_gen = READ_ONCE(lru_drain_gen);
> > + smp_rmb();
>
> this_gen = smp_load_acquire(&lru_drain_gen);

ACK. will do.

> >
> > mutex_lock(&lock);
> >
> > /*
> > + * (C) Exit the draining operation if a newer generation, from another
> > + * lru_add_drain_all(), was already scheduled for draining. Check (A).
> > */
> > + if (unlikely(this_gen != lru_drain_gen))
> > goto done;
> >
>

Re-adding cut-out comment for context:

/*
* (D) Increment generation number
*
* Pairs with READ_ONCE() and smp_rmb() at (B), outside of the critical
* section.
*
* This pairing must be done here, before the for_each_online_cpu loop
* below which drains the page vectors.
*
* Let x, y, and z represent some system CPU numbers, where x < y < z.
* Assume CPU #z is is in the middle of the for_each_online_cpu loop
* below and has already reached CPU #y's per-cpu data. CPU #x comes
* along, adds some pages to its per-cpu vectors, then calls
* lru_add_drain_all().
*
* If the paired smp_wmb() below is done at any later step, e.g. after
* the loop, CPU #x will just exit at (C) and miss flushing out all of
* its added pages.
*/
> > + WRITE_ONCE(lru_drain_gen, lru_drain_gen + 1);
> > + smp_wmb();
>
> You can leave this smp_wmb() out and rely on the smp_mb() implied by
> queue_work_on()'s test_and_set_bit().
>

Won't this be too implicit?

Isn't it possible that, over the years, queue_work_on() impementation
changes and the test_and_set_bit()/smp_mb() gets removed?

If that happens, this commit will get *silently* broken and the local
CPU pages won't be drained.

> > cpumask_clear(&has_work);
> > -
> > for_each_online_cpu(cpu) {
> > struct work_struct *work = &per_cpu(lru_add_drain_work, cpu);
> >
>
> While you're here, do:
>
> s/cpumask_set_cpu/__&/
>

ACK.

Thanks,

--
Ahmed S. Darwish
Linutronix GmbH

\
 
 \ /
  Last update: 2020-05-25 17:24    [W:0.699 / U:2.296 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site