From: Izik Eidus <izik.eidus@ravellosystems.com>
To: Avi Kivity <avi@redhat.com>
Cc: nai.xia@gmail.com, Andrew Morton <akpm@linux-foundation.org>,
Andrea Arcangeli <aarcange@redhat.com>,
Hugh Dickins <hughd@google.com>,
Chris Wright <chrisw@sous-sol.org>,
Rik van Riel <riel@redhat.com>, linux-mm <linux-mm@kvack.org>,
Johannes Weiner <hannes@cmpxchg.org>,
linux-kernel <linux-kernel@vger.kernel.org>,
kvm <kvm@vger.kernel.org>
Subject: Re: [PATCH] mmu_notifier, kvm: Introduce dirty bit tracking in spte and mmu notifier to help KSM dirty bit tracking
Date: Wed, 22 Jun 2011 14:19:46 +0300 [thread overview]
Message-ID: <4E01CFD2.6000404@ravellosystems.com> (raw)
In-Reply-To: <4E01CDAD.3070202@redhat.com>
On 6/22/2011 2:10 PM, Avi Kivity wrote:
> On 06/22/2011 02:05 PM, Izik Eidus wrote:
>>>> + spte = rmap_next(kvm, rmapp, NULL);
>>>> + while (spte) {
>>>> + int _dirty;
>>>> + u64 _spte = *spte;
>>>> + BUG_ON(!(_spte& PT_PRESENT_MASK));
>>>> + _dirty = _spte& PT_DIRTY_MASK;
>>>> + if (_dirty) {
>>>> + dirty = 1;
>>>> + clear_bit(PT_DIRTY_SHIFT, (unsigned long *)spte);
>>>> + }
>>>
>>> Racy. Also, needs a tlb flush eventually.
>> +
>>
>> Hi, one of the issues is that the whole point of this patch is not do
>> tlb flush eventually,
>> But I see your point, because other users will not expect such
>> behavior, so maybe there is need into a parameter
>> flush_tlb=?, or add another mmu notifier call?
>>
>
> If you don't flush the tlb, a subsequent write will not see that
> spte.d is clear and the write will happen. So you'll see the page as
> clean even though it's dirty. That's not acceptable.
>
Yes, but this is exactly what we want from this use case:
Right now ksm calculate the page hash to see if it was changed, the idea
behind this patch is to use the dirty bit instead,
however the guest might not really like the fact that we will flush its
tlb over and over again, specially in periodically scan like ksm does.
So what we say here is: it is better to have little junk in the unstable
tree that get flushed eventualy anyway, instead of make the guest slower....
this race is something that does not reflect accurate of ksm anyway due
to the full memcmp that we will eventualy perform...
Ofcurse we trust that in most cases, beacuse it take ksm to get into a
random virtual address in real systems few minutes, there will be
already tlb flush performed.
What you think about having 2 calls: one that does the expected behivor
and does flush the tlb, and one that clearly say it doesnt flush the tlb
and expline its use case for ksm?
WARNING: multiple messages have this Message-ID (diff)
From: Izik Eidus <izik.eidus@ravellosystems.com>
To: Avi Kivity <avi@redhat.com>
Cc: nai.xia@gmail.com, Andrew Morton <akpm@linux-foundation.org>,
Andrea Arcangeli <aarcange@redhat.com>,
Hugh Dickins <hughd@google.com>,
Chris Wright <chrisw@sous-sol.org>,
Rik van Riel <riel@redhat.com>, linux-mm <linux-mm@kvack.org>,
Johannes Weiner <hannes@cmpxchg.org>,
linux-kernel <linux-kernel@vger.kernel.org>,
kvm <kvm@vger.kernel.org>
Subject: Re: [PATCH] mmu_notifier, kvm: Introduce dirty bit tracking in spte and mmu notifier to help KSM dirty bit tracking
Date: Wed, 22 Jun 2011 14:19:46 +0300 [thread overview]
Message-ID: <4E01CFD2.6000404@ravellosystems.com> (raw)
In-Reply-To: <4E01CDAD.3070202@redhat.com>
On 6/22/2011 2:10 PM, Avi Kivity wrote:
> On 06/22/2011 02:05 PM, Izik Eidus wrote:
>>>> + spte = rmap_next(kvm, rmapp, NULL);
>>>> + while (spte) {
>>>> + int _dirty;
>>>> + u64 _spte = *spte;
>>>> + BUG_ON(!(_spte& PT_PRESENT_MASK));
>>>> + _dirty = _spte& PT_DIRTY_MASK;
>>>> + if (_dirty) {
>>>> + dirty = 1;
>>>> + clear_bit(PT_DIRTY_SHIFT, (unsigned long *)spte);
>>>> + }
>>>
>>> Racy. Also, needs a tlb flush eventually.
>> +
>>
>> Hi, one of the issues is that the whole point of this patch is not do
>> tlb flush eventually,
>> But I see your point, because other users will not expect such
>> behavior, so maybe there is need into a parameter
>> flush_tlb=?, or add another mmu notifier call?
>>
>
> If you don't flush the tlb, a subsequent write will not see that
> spte.d is clear and the write will happen. So you'll see the page as
> clean even though it's dirty. That's not acceptable.
>
Yes, but this is exactly what we want from this use case:
Right now ksm calculate the page hash to see if it was changed, the idea
behind this patch is to use the dirty bit instead,
however the guest might not really like the fact that we will flush its
tlb over and over again, specially in periodically scan like ksm does.
So what we say here is: it is better to have little junk in the unstable
tree that get flushed eventualy anyway, instead of make the guest slower....
this race is something that does not reflect accurate of ksm anyway due
to the full memcmp that we will eventualy perform...
Ofcurse we trust that in most cases, beacuse it take ksm to get into a
random virtual address in real systems few minutes, there will be
already tlb flush performed.
What you think about having 2 calls: one that does the expected behivor
and does flush the tlb, and one that clearly say it doesnt flush the tlb
and expline its use case for ksm?
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
next prev parent reply other threads:[~2011-06-22 11:20 UTC|newest]
Thread overview: 96+ messages / expand[flat|nested] mbox.gz Atom feed top
2011-06-21 12:55 [PATCH 0/2 V2] ksm: take dirty bit as reference to avoid volatile pages scanning Nai Xia
2011-06-21 12:55 ` Nai Xia
2011-06-21 13:26 ` [PATCH 1/2 " Nai Xia
2011-06-21 13:26 ` Nai Xia
2011-06-21 21:42 ` Chris Wright
2011-06-21 21:42 ` Chris Wright
2011-06-22 0:02 ` Nai Xia
2011-06-22 0:02 ` Nai Xia
2011-06-22 0:42 ` Chris Wright
2011-06-22 0:42 ` Chris Wright
2011-06-21 13:32 ` [PATCH] mmu_notifier, kvm: Introduce dirty bit tracking in spte and mmu notifier to help KSM dirty bit tracking Nai Xia
2011-06-21 13:32 ` Nai Xia
2011-06-22 0:21 ` Chris Wright
2011-06-22 0:21 ` Chris Wright
2011-06-22 4:43 ` Nai Xia
2011-06-22 4:43 ` Nai Xia
2011-06-22 6:15 ` Izik Eidus
2011-06-22 6:15 ` Izik Eidus
2011-06-22 6:38 ` Nai Xia
2011-06-22 6:38 ` Nai Xia
2011-06-22 15:46 ` Chris Wright
2011-06-22 15:46 ` Chris Wright
2011-06-22 10:43 ` Avi Kivity
2011-06-22 10:43 ` Avi Kivity
2011-06-22 11:05 ` Izik Eidus
2011-06-22 11:05 ` Izik Eidus
2011-06-22 11:10 ` Avi Kivity
2011-06-22 11:10 ` Avi Kivity
2011-06-22 11:19 ` Izik Eidus [this message]
2011-06-22 11:19 ` Izik Eidus
2011-06-22 11:24 ` Avi Kivity
2011-06-22 11:24 ` Avi Kivity
2011-06-22 11:28 ` Avi Kivity
2011-06-22 11:28 ` Avi Kivity
2011-06-22 11:31 ` Avi Kivity
2011-06-22 11:31 ` Avi Kivity
2011-06-22 11:33 ` Nai Xia
2011-06-22 11:33 ` Nai Xia
2011-06-22 11:39 ` Izik Eidus
2011-06-22 11:39 ` Izik Eidus
2011-06-22 15:39 ` Rik van Riel
2011-06-22 15:39 ` Rik van Riel
2011-06-22 16:55 ` Andrea Arcangeli
2011-06-22 16:55 ` Andrea Arcangeli
2011-06-22 23:37 ` Nai Xia
2011-06-22 23:37 ` Nai Xia
2011-06-22 23:59 ` Andrea Arcangeli
2011-06-22 23:59 ` Andrea Arcangeli
2011-06-23 0:31 ` Nai Xia
2011-06-23 0:31 ` Nai Xia
2011-06-23 0:44 ` Andrea Arcangeli
2011-06-23 0:44 ` Andrea Arcangeli
2011-06-23 1:36 ` Nai Xia
2011-06-23 1:36 ` Nai Xia
2011-06-23 0:00 ` Rik van Riel
2011-06-23 0:00 ` Rik van Riel
2011-06-23 0:42 ` Nai Xia
2011-06-23 0:42 ` Nai Xia
2011-06-22 23:13 ` Nai Xia
2011-06-22 23:13 ` Nai Xia
2011-06-22 23:25 ` Andrea Arcangeli
2011-06-22 23:25 ` Andrea Arcangeli
2011-06-23 1:30 ` Nai Xia
2011-06-23 1:30 ` Nai Xia
2011-06-22 23:28 ` Rik van Riel
2011-06-22 23:28 ` Rik van Riel
2011-06-23 0:52 ` Nai Xia
2011-06-23 0:52 ` Nai Xia
2011-06-22 11:24 ` Nai Xia
2011-06-22 15:03 ` Andrea Arcangeli
2011-06-22 15:03 ` Andrea Arcangeli
2011-06-22 15:19 ` Izik Eidus
2011-06-22 15:19 ` Izik Eidus
2011-06-22 23:19 ` Nai Xia
2011-06-22 23:19 ` Nai Xia
2011-06-22 23:44 ` Andrea Arcangeli
2011-06-22 23:44 ` Andrea Arcangeli
2011-06-23 0:14 ` Nai Xia
2011-06-23 0:14 ` Nai Xia
2011-06-22 23:42 ` Nai Xia
2011-06-22 23:42 ` Nai Xia
2011-06-21 13:36 ` [PATCH 2/2 V2] ksm: take dirty bit as reference to avoid volatile pages scanning Nai Xia
2011-06-21 13:36 ` Nai Xia
2011-06-21 22:38 ` Chris Wright
2011-06-21 22:38 ` Chris Wright
2011-06-22 0:04 ` Nai Xia
2011-06-22 0:04 ` Nai Xia
2011-06-22 0:35 ` Chris Wright
2011-06-22 0:35 ` Chris Wright
2011-06-22 4:47 ` Nai Xia
2011-06-22 4:47 ` Nai Xia
2011-06-22 10:55 ` Nai Xia
2011-06-22 10:55 ` Nai Xia
2011-06-22 0:46 ` [PATCH 0/2 " Chris Wright
2011-06-22 0:46 ` Chris Wright
2011-06-22 4:15 ` Nai Xia
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=4E01CFD2.6000404@ravellosystems.com \
--to=izik.eidus@ravellosystems.com \
--cc=aarcange@redhat.com \
--cc=akpm@linux-foundation.org \
--cc=avi@redhat.com \
--cc=chrisw@sous-sol.org \
--cc=hannes@cmpxchg.org \
--cc=hughd@google.com \
--cc=kvm@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=nai.xia@gmail.com \
--cc=riel@redhat.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.