Re: [PATCH 2/2] sysvshm: SHM_LOCK use lru_add_drain_all_async()

[KOSAKI Motohiro]

(1/3/12 8:51 PM), Hugh Dickins wrote:
> On Sun, 1 Jan 2012, kosaki.motohiro@xxxxxxxxx wrote:
> > From: KOSAKI Motohiro<kosaki.motohiro@xxxxxxxxxxxxxx>
> >
> > shmctl also don't need synchrounous pagevec drain. This patch replace it with
> > lru_add_drain_all_async().
> >
> > Signed-off-by: KOSAKI Motohiro<kosaki.motohiro@xxxxxxxxxxxxxx>
>
> Let me answer this 2/2 first since it's easier.
>
> I'm going to thank you for bringing this lru_add_drain_all()
> to my attention, I had not noticed it; but Nak the patch itself.
>
> The reason being, that particular lru_add_drain_all() serves no
> useful purpose, so let's delete it instead of replacing it.  I believe
> that it serves no purpose for SHM_LOCK and no purpose for SHM_UNLOCK.
>
> I'm dabbling in this area myself, since you so cogently pointed out that
> I'd tried to add a cond_resched() to scan_mapping_unevictable_pages()
> (which is a helper for SHM_UNLOCK here) while it's under spinlock.
>
> In testing my fix for that, I find that there has been no attempt to
> keep the Unevictable count accurate on SysVShm: SHM_LOCK pages get
> marked unevictable lazily later as memory pressure discovers them -
> which perhaps mirrors the way in which SHM_LOCK makes no attempt to
> instantiate pages, unlike mlock.

Ugh, you are right. I'm recovering my remember gradually. Lee 
implemented immediate lru off logic at first and I killed it
to close a race. I completely forgot. So, yes, now SHM_LOCK has no 
attempt to instantiate pages. I'm ashamed.


> Since nobody has complained about that in the two years since we've
> had an Unevictable count in /proc/meminfo, I don't see any need to
> add code (it would need more than just your change here; would need
> more even than calling scan_mapping_unevictable_pages() at SHM_LOCK
> time - though perhaps along with your 1/2 that could handle it) and
> overhead to satisfy a need that nobody has.
>
> I'll delete that lru_add_drain_all() in my patch, okay?

Sure thing. :)


> (But in writing this, realize I still don't quite understand why
> the Unevictable count takes a second or two to get back to 0 after
> SHM_UNLOCK: perhaps I've more to discover.)

Interesting. I'm looking at this too.

--
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/