From: "Michael S. Tsirkin" <mst@redhat.com>
To: James Bottomley <James.Bottomley@hansenpartnership.com>
Cc: Christoph Hellwig <hch@infradead.org>,
Andrea Arcangeli <aarcange@redhat.com>,
Jason Wang <jasowang@redhat.com>,
David Miller <davem@davemloft.net>,
kvm@vger.kernel.org, virtualization@lists.linux-foundation.org,
netdev@vger.kernel.org, linux-kernel@vger.kernel.org,
peterx@redhat.com, linux-mm@kvack.org,
linux-arm-kernel@lists.infradead.org,
linux-parisc@vger.kernel.org
Subject: Re: [RFC PATCH V2 0/5] vhost: accelerate metadata access through vmap()
Date: Thu, 14 Mar 2019 06:42:21 -0400 [thread overview]
Message-ID: <20190314064004-mutt-send-email-mst@kernel.org> (raw)
In-Reply-To: <1552495028.3022.37.camel@HansenPartnership.com>
On Wed, Mar 13, 2019 at 09:37:08AM -0700, James Bottomley wrote:
> On Wed, 2019-03-13 at 09:05 -0700, Christoph Hellwig wrote:
> > On Tue, Mar 12, 2019 at 01:53:37PM -0700, James Bottomley wrote:
> > > I've got to say: optimize what? What code do we ever have in the
> > > kernel that kmap's a page and then doesn't do anything with it? You
> > > can
> > > guarantee that on kunmap the page is either referenced (needs
> > > invalidating) or updated (needs flushing). The in-kernel use of
> > > kmap is
> > > always
> > >
> > > kmap
> > > do something with the mapped page
> > > kunmap
> > >
> > > In a very short interval. It seems just a simplification to make
> > > kunmap do the flush if needed rather than try to have the users
> > > remember. The thing which makes this really simple is that on most
> > > architectures flush and invalidate is the same operation. If you
> > > really want to optimize you can use the referenced and dirty bits
> > > on the kmapped pte to tell you what operation to do, but if your
> > > flush is your invalidate, you simply assume the data needs flushing
> > > on kunmap without checking anything.
> >
> > I agree that this would be a good way to simplify the API. Now
> > we'd just need volunteers to implement this for all architectures
> > that need cache flushing and then remove the explicit flushing in
> > the callers..
>
> Well, it's already done on parisc ... I can help with this if we agree
> it's the best way forward. It's really only architectures that
> implement flush_dcache_page that would need modifying.
>
> It may also improve performance because some kmap/use/flush/kunmap
> sequences have flush_dcache_page() instead of
> flush_kernel_dcache_page() and the former is hugely expensive and
> usually unnecessary because GUP already flushed all the user aliases.
>
> In the interests of full disclosure the reason we do it for parisc is
> because our later machines have problems even with clean aliases. So
> on most VIPT systems, doing kmap/read/kunmap creates a fairly harmless
> clean alias. Technically it should be invalidated, because if you
> remap the same page to the same colour you get cached stale data, but
> in practice the data is expired from the cache long before that
> happens, so the problem is almost never seen if the flush is forgotten.
> Our problem is on the P9xxx processor: they have a L1/L2 VIPT L3 PIPT
> cache. As the L1/L2 caches expire clean data, they place the expiring
> contents into L3, but because L3 is PIPT, the stale alias suddenly
> becomes the default for any read of they physical page because any
> update which dirtied the cache line often gets written to main memory
> and placed into the L3 as clean *before* the clean alias in L1/L2 gets
> expired, so the older clean alias replaces it.
>
> Our only recourse is to kill all aliases with prejudice before the
> kernel loses ownership.
>
> > > > Which means after we fix vhost to add the flush_dcache_page after
> > > > kunmap, Parisc will get a double hit (but it also means Parisc
> > > > was the only one of those archs needed explicit cache flushes,
> > > > where vhost worked correctly so far.. so it kinds of proofs your
> > > > point of giving up being the safe choice).
> > >
> > > What double hit? If there's no cache to flush then cache flush is
> > > a no-op. It's also a highly piplineable no-op because the CPU has
> > > the L1 cache within easy reach. The only event when flush takes a
> > > large amount time is if we actually have dirty data to write back
> > > to main memory.
> >
> > I've heard people complaining that on some microarchitectures even
> > no-op cache flushes are relatively expensive. Don't ask me why,
> > but if we can easily avoid double flushes we should do that.
>
> It's still not entirely free for us. Our internal cache line is around
> 32 bytes (some have 16 and some have 64) but that means we need 128
> flushes for a page ... we definitely can't pipeline them all. So I
> agree duplicate flush elimination would be a small improvement.
>
> James
I suspect we'll keep the copyXuser path around for 32 bit anyway -
right Jason?
So we can also keep using that on parisc...
--
MST
next prev parent reply other threads:[~2019-03-14 10:42 UTC|newest]
Thread overview: 78+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-03-06 7:18 [RFC PATCH V2 0/5] vhost: accelerate metadata access through vmap() Jason Wang
2019-03-06 7:18 ` [RFC PATCH V2 1/5] vhost: generalize adding used elem Jason Wang
2019-03-06 7:18 ` [RFC PATCH V2 2/5] vhost: fine grain userspace memory accessors Jason Wang
2019-03-06 10:45 ` Christophe de Dinechin
2019-03-07 2:38 ` Jason Wang
2019-03-06 7:18 ` [RFC PATCH V2 3/5] vhost: rename vq_iotlb_prefetch() to vq_meta_prefetch() Jason Wang
2019-03-06 7:18 ` [RFC PATCH V2 4/5] vhost: introduce helpers to get the size of metadata area Jason Wang
2019-03-06 10:56 ` Christophe de Dinechin
2019-03-07 2:40 ` Jason Wang
2019-03-06 18:43 ` Souptick Joarder
2019-03-07 2:42 ` Jason Wang
2019-03-06 7:18 ` [RFC PATCH V2 5/5] vhost: access vq metadata through kernel virtual address Jason Wang
2019-03-06 16:31 ` Michael S. Tsirkin
2019-03-07 2:45 ` Jason Wang
2019-03-07 15:34 ` Michael S. Tsirkin
2019-03-07 19:09 ` Jerome Glisse
2019-03-07 19:38 ` Andrea Arcangeli
2019-03-07 20:17 ` Jerome Glisse
2019-03-07 21:27 ` Andrea Arcangeli
2019-03-08 9:13 ` Jason Wang
2019-03-08 19:11 ` Andrea Arcangeli
2019-03-11 7:21 ` Jason Wang
2019-03-11 14:45 ` Jan Kara
2019-03-08 8:31 ` Jason Wang
2019-03-07 15:47 ` Michael S. Tsirkin
2019-03-07 17:56 ` Michael S. Tsirkin
2019-03-07 19:16 ` Andrea Arcangeli
2019-03-08 8:50 ` Jason Wang
2019-03-08 14:58 ` Jerome Glisse
2019-03-11 7:18 ` Jason Wang
2019-03-08 19:48 ` Andrea Arcangeli
2019-03-08 20:06 ` Jerome Glisse
2019-03-11 7:40 ` Jason Wang
2019-03-11 12:48 ` Michael S. Tsirkin
2019-03-11 13:43 ` Andrea Arcangeli
2019-03-12 2:56 ` Jason Wang
2019-03-12 3:51 ` Michael S. Tsirkin
2019-03-12 2:52 ` Jason Wang
2019-03-12 3:50 ` Michael S. Tsirkin
2019-03-12 7:15 ` Jason Wang
2019-03-07 19:17 ` Jerome Glisse
2019-03-08 2:21 ` Michael S. Tsirkin
2019-03-08 2:55 ` Jerome Glisse
2019-03-08 3:16 ` Michael S. Tsirkin
2019-03-08 3:40 ` Jerome Glisse
2019-03-08 3:43 ` Michael S. Tsirkin
2019-03-08 3:45 ` Jerome Glisse
2019-03-08 9:15 ` Jason Wang
2019-03-08 8:58 ` Jason Wang
2019-03-08 12:56 ` Michael S. Tsirkin
2019-03-08 15:02 ` Jerome Glisse
2019-03-08 19:13 ` Andrea Arcangeli
2019-03-08 14:12 ` [RFC PATCH V2 0/5] vhost: accelerate metadata access through vmap() Christoph Hellwig
2019-03-11 7:13 ` Jason Wang
2019-03-11 13:59 ` Michael S. Tsirkin
2019-03-11 18:14 ` David Miller
2019-03-12 2:59 ` Jason Wang
2019-03-12 3:52 ` Michael S. Tsirkin
2019-03-12 7:17 ` Jason Wang
2019-03-12 11:54 ` Michael S. Tsirkin
2019-03-12 15:46 ` James Bottomley
2019-03-12 20:04 ` Andrea Arcangeli
2019-03-12 20:53 ` James Bottomley
2019-03-12 21:11 ` Andrea Arcangeli
2019-03-12 21:19 ` James Bottomley
2019-03-12 21:53 ` Andrea Arcangeli
2019-03-12 22:02 ` James Bottomley
2019-03-12 22:50 ` Andrea Arcangeli
2019-03-12 22:57 ` James Bottomley
2019-03-13 16:05 ` Christoph Hellwig
2019-03-13 16:37 ` James Bottomley
2019-03-14 10:42 ` Michael S. Tsirkin [this message]
2019-03-14 13:49 ` Jason Wang
2019-03-14 19:33 ` Andrea Arcangeli
2019-03-15 4:39 ` Jason Wang
2019-03-12 5:14 ` James Bottomley
2019-03-12 7:51 ` Jason Wang
2019-03-12 7:53 ` Jason Wang
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=20190314064004-mutt-send-email-mst@kernel.org \
--to=mst@redhat.com \
--cc=James.Bottomley@hansenpartnership.com \
--cc=aarcange@redhat.com \
--cc=davem@davemloft.net \
--cc=hch@infradead.org \
--cc=jasowang@redhat.com \
--cc=kvm@vger.kernel.org \
--cc=linux-arm-kernel@lists.infradead.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=linux-parisc@vger.kernel.org \
--cc=netdev@vger.kernel.org \
--cc=peterx@redhat.com \
--cc=virtualization@lists.linux-foundation.org \
/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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).