From: Huacai Chen <chenhuacai@gmail.com>
To: Peter Zijlstra <peterz@infradead.org>
Cc: Boqun Feng <boqun.feng@gmail.com>, Guo Ren <guoren@kernel.org>,
Waiman Long <llong@redhat.com>,
Geert Uytterhoeven <geert@linux-m68k.org>,
Huacai Chen <chenhuacai@loongson.cn>,
Ingo Molnar <mingo@redhat.com>, Will Deacon <will@kernel.org>,
Arnd Bergmann <arnd@arndb.de>,
Linux-Arch <linux-arch@vger.kernel.org>,
Rui Wang <wangrui@loongson.cn>,
Xuefeng Li <lixuefeng@loongson.cn>,
Jiaxun Yang <jiaxun.yang@flygoat.com>
Subject: Re: [PATCH RFC 1/2] arch: Introduce ARCH_HAS_HW_XCHG_SMALL
Date: Fri, 30 Jul 2021 00:38:14 +0800 [thread overview]
Message-ID: <CAAhV-H6tQyb0o=ZBHp+LKi6LbOXwFD4JzBQjk1KkDhP2t5KmdA@mail.gmail.com> (raw)
In-Reply-To: <YQE5B6ecrnIoXhpI@hirez.programming.kicks-ass.net>
Hi Peter,
On Wed, Jul 28, 2021 at 7:02 PM Peter Zijlstra <peterz@infradead.org> wrote:
>
> On Wed, Jul 28, 2021 at 06:40:54PM +0800, Huacai Chen wrote:
> > Hi, Peter,
> >
> > On Tue, Jul 27, 2021 at 7:06 PM Peter Zijlstra <peterz@infradead.org> wrote:
> > >
> > > On Tue, Jul 27, 2021 at 10:29:59AM +0800, Boqun Feng wrote:
> > >
> > > > > "How to implement xchg_tail" shouldn't force with _Q_PENDING_BITS, but
> > > > > the arch could choose.
> > > >
> > > > I actually agree with this part, but this patchset failed to provide
> > > > enough evidences on why we should choose xchg_tail() implementation
> > > > based on whether hardware has xchg16, more precisely, for an archtecture
> > > > which doesn't have a hardware xchg16, why cmpxchg emulated xchg16() is
> > > > worse than a "load+cmpxchg) implemeneted xchg_tail()? If it's a
> > > > performance reason, please show some numbers.
> > >
> > > Right. Their problem is their broken xchg16() implementation.
> >
> > Please correct me if I'm wrong. Now my understanding is like this:
> > 1, _Q_PENDING_BITS=1 qspinlock can be used by all archs, though it may
> > be not optimized.
>
> Only if your arch has fwd progress guarantees for cmpxchg(). LL/SC based
> cmpxchg() is tricky, and typically needs software based backoff on
> failure.
>
> The qspinlock code is written in generic code, but it very much relies
> on an architecture to audit and vet the resulting code is sane for them.
> Clearly MIPS didn't do a good job of that.
>
> > 2, _Q_PENDING_BITS=8 qspinlock can be used if hardware supports
> > sub-word xchg/cmpxchg, or the software emulation is correctly
> > implemented. But the current MIPS emulation is not correct.
>
> Everything always relies on things being correctly implemented. 1)
> relies on cmpxchg() being correctly implemented. 2) relies on xchg16()
> being correctly implemented.
>
> Of these 2 is actually easier to implement correctly on LL/SC.
>
> > If so, I want to rename ARCH_HAS_HW_XCHG_SMALL to
> > ARCH_HAS_FAST_XCHG_SMALL, and let these archs select it:
> > 1, X86, ARM, ARM64, IA64, M68K, because they have hardware support.
> > 2, Other archs who select qspinlock currently (including MIPS),
> > because their current behavior is use _Q_PENDING_BITS=8 qspinlock and
> > we don't want to change anything in this patch. If their emulation is
> > broken or not as "fast" as expected, we can make new patches to
> > unselect the ARCH_HAS_FAST_XCHG_SMALL option.
>
> I utterly fail to see the point of any of this. If you use qspinlock,
> you had better have audited the whole thing for your architecture. And
> FAST_XCHG_SMALL is completely irrelevant for anything here.
>
Our original goal of this patch is let LoongArch use qspinlock, but
Arnd let us remove xchg_small, so we want to let LoongArch use
_Q_PENDING_BITS=1 lock . :)
Now Rui Wang has a better solution, and xchg_small will be
reimplemented for LoongArch, so I think this patch is no longer
needed.
Huacai
next prev parent reply other threads:[~2021-07-29 16:38 UTC|newest]
Thread overview: 26+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-07-24 12:36 [PATCH RFC 1/2] arch: Introduce ARCH_HAS_HW_XCHG_SMALL Huacai Chen
2021-07-24 12:36 ` [PATCH RFC 2/2] qspinlock: Use ARCH_HAS_HW_XCHG_SMALL to select _Q_PENDING_BITS definition Huacai Chen
2021-07-24 19:24 ` [PATCH RFC 1/2] arch: Introduce ARCH_HAS_HW_XCHG_SMALL Arnd Bergmann
2021-07-25 3:06 ` Jiaxun Yang
2021-07-25 10:08 ` Arnd Bergmann
2021-07-26 8:36 ` Geert Uytterhoeven
2021-07-26 8:56 ` Huacai Chen
2021-07-26 10:39 ` Boqun Feng
2021-07-26 16:41 ` Guo Ren
2021-07-26 17:03 ` Boqun Feng
2021-07-26 21:20 ` Waiman Long
2021-07-27 1:27 ` Guo Ren
2021-07-27 2:29 ` Boqun Feng
2021-07-27 2:46 ` Waiman Long
2021-07-27 11:05 ` Peter Zijlstra
2021-07-28 10:40 ` Huacai Chen
2021-07-28 11:01 ` Peter Zijlstra
2021-07-29 16:38 ` Huacai Chen [this message]
2021-07-27 10:17 ` Peter Zijlstra
2021-07-27 1:07 ` Guo Ren
2021-07-27 1:52 ` Wang Rui
2021-07-27 11:03 ` Peter Zijlstra
2021-07-27 2:00 ` Boqun Feng
2021-07-27 10:50 ` Peter Zijlstra
2021-07-27 10:12 ` Peter Zijlstra
2021-07-26 9:00 ` Arnd Bergmann
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='CAAhV-H6tQyb0o=ZBHp+LKi6LbOXwFD4JzBQjk1KkDhP2t5KmdA@mail.gmail.com' \
--to=chenhuacai@gmail.com \
--cc=arnd@arndb.de \
--cc=boqun.feng@gmail.com \
--cc=chenhuacai@loongson.cn \
--cc=geert@linux-m68k.org \
--cc=guoren@kernel.org \
--cc=jiaxun.yang@flygoat.com \
--cc=linux-arch@vger.kernel.org \
--cc=lixuefeng@loongson.cn \
--cc=llong@redhat.com \
--cc=mingo@redhat.com \
--cc=peterz@infradead.org \
--cc=wangrui@loongson.cn \
--cc=will@kernel.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).