Re: setup_per_zone_pages_min(): zone->lock vs. zone->lru_lock

[KAMEZAWA Hiroyuki]

On Mon, 29 Sep 2008 23:20:05 +0200
Gerald Schaefer <gerald.schaefer@xxxxxxxxxx> wrote:

> On Mon, 2008-09-29 at 18:36 +0100, Andy Whitcroft wrote:
> > The allocator protects it freelists using zone->lock (as we can see in
> > rmqueue_bulk), so anything which manipulates those should also be using
> > that lock.  move_freepages() is scanning the cmap and picking up free
> > pages directly off the free lists, it is expecting those lists to be
> > stable; it would appear to need zone->lock.  It does look like
> > setup_per_zone_pages_min() is holding the wrong thing at first look.
> 
> I just noticed that the spin_lock in that function is much older than the
> call to setup_zone_migrate_reserve(), which then calls move_freepages().
> So it seems that the zone->lru_lock there does (did?) have another purpose,
> maybe protecting zone->present_pages/pages_min/etc.
> 
Maybe.

> Looks like the need for a zone->lock (if any) was added later, but I'm not
> sure if makes sense to take both locks together, or if the lru_lock is still
> needed at all.
>
At first look, replacing zone->lru_lock with zone->lock is enough...
This function is an only one function which use zone->lru_lock in page_alloc.c
And zone_watermark_ok() which access zone->pages_min/low/high is not under any
locks. So, taking zone->lru_lock here doesn't seem to be necessary...


Thanks,
-Kame

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/