Received-SPF: pass (google.com: domain of linux-kernel+bounces-50648-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:45e3:2400::1 as permitted sender) client-ip=2604:1380:45e3:2400::1;
Precedence: bulk
MIME-Version: 1.0
References: <20240131155929.169961-1-alexghiti@rivosinc.com>
 <20240131155929.169961-5-alexghiti@rivosinc.com> <Zbuy1E7mz9Oui1Dl@andrea> <CAHVXubgw0PEZMhFmjA0cAFQ2+_JOYjVfk41qRC9TFdSJtej++w@mail.gmail.com>
In-Reply-To: <CAHVXubgw0PEZMhFmjA0cAFQ2+_JOYjVfk41qRC9TFdSJtej++w@mail.gmail.com>
From: Alexandre Ghiti <alexghiti@rivosinc.com>
Date: Fri, 2 Feb 2024 23:05:44 +0100
Message-ID: <CAHVXubjzpTru3Tn+nmv=GwG0YWxnkeYYT1KOJW5a27H2EDAZ0Q@mail.gmail.com>
Subject: Re: [PATCH RFC/RFT v2 4/4] riscv: Stop emitting preventive sfence.vma
 for new userspace mappings with Svvptc
To: Andrea Parri <parri.andrea@gmail.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>, Will Deacon <will@kernel.org>, 
	Thomas Bogendoerfer <tsbogend@alpha.franken.de>, Michael Ellerman <mpe@ellerman.id.au>, 
	Nicholas Piggin <npiggin@gmail.com>, Christophe Leroy <christophe.leroy@csgroup.eu>, 
	Paul Walmsley <paul.walmsley@sifive.com>, Palmer Dabbelt <palmer@dabbelt.com>, 
	Albert Ou <aou@eecs.berkeley.edu>, Andrew Morton <akpm@linux-foundation.org>, 
	Ved Shanbhogue <ved@rivosinc.com>, Matt Evans <mev@rivosinc.com>, Dylan Jhong <dylan@andestech.com>, 
	linux-arm-kernel@lists.infradead.org, linux-kernel@vger.kernel.org, 
	linux-mips@vger.kernel.org, linuxppc-dev@lists.ozlabs.org, 
	linux-riscv@lists.infradead.org, linux-mm@kvack.org
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

On Fri, Feb 2, 2024 at 4:42=E2=80=AFPM Alexandre Ghiti <alexghiti@rivosinc.=
com> wrote:
>
> Hi Andrea,
>
> On Thu, Feb 1, 2024 at 4:03=E2=80=AFPM Andrea Parri <parri.andrea@gmail.c=
om> wrote:
> >
> > On Wed, Jan 31, 2024 at 04:59:29PM +0100, Alexandre Ghiti wrote:
> > > The preventive sfence.vma were emitted because new mappings must be m=
ade
> > > visible to the page table walker but Svvptc guarantees that xRET act =
as
> > > a fence, so no need to sfence.vma for the uarchs that implement this
> > > extension.
> >
> > AFAIU, your first submission shows that you don't need that xRET proper=
ty.
> > Similarly for other archs.  What was rationale behind this Svvptc chang=
e?
>
> Actually, the ARC has just changed its mind and removed this new

The wording was incorrect here, the ARC did not state anything, the
author of Svvptc proposed an amended version of the spec that removes
this behaviour and that's under discussion.

> behaviour from the Svvptc extension, so we will take some gratuitous
> page faults (but that should be outliners), which makes riscv similar
> to x86 and arm64.
>
> >
> >
> > > This allows to drastically reduce the number of sfence.vma emitted:
> > >
> > > * Ubuntu boot to login:
> > > Before: ~630k sfence.vma
> > > After:  ~200k sfence.vma
> > >
> > > * ltp - mmapstress01
> > > Before: ~45k
> > > After:  ~6.3k
> > >
> > > * lmbench - lat_pagefault
> > > Before: ~665k
> > > After:   832 (!)
> > >
> > > * lmbench - lat_mmap
> > > Before: ~546k
> > > After:   718 (!)
> >
> > This Svvptc seems to move/add the "burden" of the synchronization to xR=
ET:
> > Perhaps integrate the above counts w/ the perf gains in the cover lette=
r?
>
> Yes, I'll copy that to the cover letter.
>
> Thanks for your interest!
>
> Alex
>
> >
> >   Andrea