From: Vipin Sharma <vipinsh@google.com>
To: Ben Gardon <bgardon@google.com>
Cc: David Matlack <dmatlack@google.com>,
seanjc@google.com, pbonzini@redhat.com, kvm@vger.kernel.org,
linux-kernel@vger.kernel.org
Subject: Re: [Patch] KVM: x86/mmu: Make optimized __handle_changed_spte() for clear dirty log
Date: Wed, 25 Jan 2023 16:40:03 -0800 [thread overview]
Message-ID: <CAHVum0f=xKkE7YhPuQ-A9uOZMRMUSuiFcB3sxi0b_Bp837M5aQ@mail.gmail.com> (raw)
In-Reply-To: <CANgfPd9T7jdh1Cjjo4y6DcxC2poTaGhQ7wNLf6OgGtStg-yKJg@mail.gmail.com>
On Wed, Jan 25, 2023 at 2:25 PM Ben Gardon <bgardon@google.com> wrote:
>
> On Wed, Jan 25, 2023 at 2:00 PM David Matlack <dmatlack@google.com> wrote:
> >
> > On Wed, Jan 25, 2023 at 01:38:57PM -0800, Vipin Sharma wrote:
> > > @@ -736,7 +749,12 @@ static u64 __tdp_mmu_set_spte(struct kvm *kvm, int as_id, tdp_ptep_t sptep,
> > >
> > > old_spte = kvm_tdp_mmu_write_spte(sptep, old_spte, new_spte, level);
> > >
> > > - __handle_changed_spte(kvm, as_id, gfn, old_spte, new_spte, level, false);
> > > + if (record_dirty_log)
> > > + __handle_changed_spte(kvm, as_id, gfn, old_spte, new_spte,
> > > + level, false);
> > > + else
> > > + handle_changed_spte_clear_dirty_log(as_id, gfn, old_spte,
> > > + new_spte, level);
> >
> > I find it very non-intuitive to tie this behavior to !record_dirty_log,
> > even though it happens to work. It's also risky if any future calls are
> > added that pass record_dirty_log=false but change other bits in the
> > SPTE.
> >
> > I wonder if we could get the same effect by optimizing
> > __handle_changed_spte() to check for a cleared D-bit *first* and if
> > that's the only diff between the old and new SPTE, bail immediately
> > rather than doing all the other checks.
>
> I would also prefer that course. One big lesson I took when building
> the TDP MMU was the value of having all the changed PTE handling go
> through one function. That's not always the right choice but the
> Shadow MMU has a crazy spaghetti code of different functions which
> handle different parts of responding to a PTE change and it makes the
> code very hard to read. I agree this path is worth optimizing, but the
> more we can keep the code together, the better.
Let me see if I am able to get the same improvement in __handle_changed_spte().
next prev parent reply other threads:[~2023-01-26 0:40 UTC|newest]
Thread overview: 13+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-01-25 21:38 [Patch] KVM: x86/mmu: Make optimized __handle_changed_spte() for clear dirty log Vipin Sharma
2023-01-25 22:00 ` David Matlack
2023-01-25 22:25 ` Ben Gardon
2023-01-26 0:40 ` Vipin Sharma [this message]
2023-01-26 1:49 ` Sean Christopherson
2023-01-28 17:20 ` Vipin Sharma
2023-01-30 18:09 ` Sean Christopherson
2023-01-30 20:17 ` Vipin Sharma
2023-01-30 22:49 ` Sean Christopherson
2023-01-30 22:12 ` David Matlack
2023-01-30 23:49 ` Sean Christopherson
2023-01-31 17:41 ` David Matlack
2023-01-31 18:01 ` Vipin Sharma
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='CAHVum0f=xKkE7YhPuQ-A9uOZMRMUSuiFcB3sxi0b_Bp837M5aQ@mail.gmail.com' \
--to=vipinsh@google.com \
--cc=bgardon@google.com \
--cc=dmatlack@google.com \
--cc=kvm@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=pbonzini@redhat.com \
--cc=seanjc@google.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 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).