Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable
From:   Andy Lutomirski <luto@amacapital.net>
Mime-Version: 1.0 (1.0)
Subject: Re: [RFC PATCH 7/7] lazy tlb: shoot lazies, a non-refcounting lazy tlb option
Date:   Tue, 14 Jul 2020 05:46:05 -0700
Message-Id: <6D3D1346-DB1E-43EB-812A-184918CCC16A@amacapital.net>
References: <1594708054.04iuyxuyb5.astroid@bobo.none>
Cc:     Anton Blanchard <anton@ozlabs.org>, Arnd Bergmann <arnd@arndb.de>,
        linux-arch <linux-arch@vger.kernel.org>,
        LKML <linux-kernel@vger.kernel.org>,
        Linux-MM <linux-mm@kvack.org>,
        linuxppc-dev <linuxppc-dev@lists.ozlabs.org>,
        Andy Lutomirski <luto@kernel.org>,
        Mathieu Desnoyers <mathieu.desnoyers@efficios.com>,
        Peter Zijlstra <peterz@infradead.org>, X86 ML <x86@kernel.org>
In-Reply-To: <1594708054.04iuyxuyb5.astroid@bobo.none>
To:     Nicholas Piggin <npiggin@gmail.com>
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk


> On Jul 13, 2020, at 11:31 PM, Nicholas Piggin <npiggin@gmail.com> wrote:
>=20
> =EF=BB=BFExcerpts from Nicholas Piggin's message of July 14, 2020 3:04 pm:=

>> Excerpts from Andy Lutomirski's message of July 14, 2020 4:18 am:
>>>=20
>>>> On Jul 13, 2020, at 9:48 AM, Nicholas Piggin <npiggin@gmail.com> wrote:=

>>>>=20
>>>> =EF=BB=BFExcerpts from Andy Lutomirski's message of July 14, 2020 1:59 a=
m:
>>>>>> On Thu, Jul 9, 2020 at 6:57 PM Nicholas Piggin <npiggin@gmail.com> wr=
ote:
>>>>>>=20
>>>>>> On big systems, the mm refcount can become highly contented when doin=
g
>>>>>> a lot of context switching with threaded applications (particularly
>>>>>> switching between the idle thread and an application thread).
>>>>>>=20
>>>>>> Abandoning lazy tlb slows switching down quite a bit in the important=

>>>>>> user->idle->user cases, so so instead implement a non-refcounted sche=
me
>>>>>> that causes __mmdrop() to IPI all CPUs in the mm_cpumask and shoot do=
wn
>>>>>> any remaining lazy ones.
>>>>>>=20
>>>>>> On a 16-socket 192-core POWER8 system, a context switching benchmark
>>>>>> with as many software threads as CPUs (so each switch will go in and
>>>>>> out of idle), upstream can achieve a rate of about 1 million context
>>>>>> switches per second. After this patch it goes up to 118 million.
>>>>>>=20
>>>>>=20
>>>>> I read the patch a couple of times, and I have a suggestion that could=

>>>>> be nonsense.  You are, effectively, using mm_cpumask() as a sort of
>>>>> refcount.  You're saying "hey, this mm has no more references, but it
>>>>> still has nonempty mm_cpumask(), so let's send an IPI and shoot down
>>>>> those references too."  I'm wondering whether you actually need the
>>>>> IPI.  What if, instead, you actually treated mm_cpumask as a refcount
>>>>> for real?  Roughly, in __mmdrop(), you would only free the page tables=

>>>>> if mm_cpumask() is empty.  And, in the code that removes a CPU from
>>>>> mm_cpumask(), you would check if mm_users =3D=3D 0 and, if so, check i=
f
>>>>> you just removed the last bit from mm_cpumask and potentially free the=

>>>>> mm.
>>>>>=20
>>>>> Getting the locking right here could be a bit tricky -- you need to
>>>>> avoid two CPUs simultaneously exiting lazy TLB and thinking they
>>>>> should free the mm, and you also need to avoid an mm with mm_users
>>>>> hitting zero concurrently with the last remote CPU using it lazily
>>>>> exiting lazy TLB.  Perhaps this could be resolved by having mm_count
>>>>> =3D=3D 1 mean "mm_cpumask() is might contain bits and, if so, it owns t=
he
>>>>> mm" and mm_count =3D=3D 0 meaning "now it's dead" and using some caref=
ul
>>>>> cmpxchg or dec_return to make sure that only one CPU frees it.
>>>>>=20
>>>>> Or maybe you'd need a lock or RCU for this, but the idea would be to
>>>>> only ever take the lock after mm_users goes to zero.
>>>>=20
>>>> I don't think it's nonsense, it could be a good way to avoid IPIs.
>>>>=20
>>>> I haven't seen much problem here that made me too concerned about IPIs=20=

>>>> yet, so I think the simple patch may be good enough to start with
>>>> for powerpc. I'm looking at avoiding/reducing the IPIs by combining the=

>>>> unlazying with the exit TLB flush without doing anything fancy with
>>>> ref counting, but we'll see.
>>>=20
>>> I would be cautious with benchmarking here. I would expect that the
>>> nasty cases may affect power consumption more than performance =E2=80=94=
 the=20
>>> specific issue is IPIs hitting idle cores, and the main effects are to=20=

>>> slow down exit() a bit but also to kick the idle core out of idle.=20
>>> Although, if the idle core is in a deep sleep, that IPI could be=20
>>> *very* slow.
>>=20
>> It will tend to be self-limiting to some degree (deeper idle cores
>> would tend to have less chance of IPI) but we have bigger issues on
>> powerpc with that, like broadcast IPIs to the mm cpumask for THP
>> management. Power hasn't really shown up as an issue but powerpc
>> CPUs may have their own requirements and issues there, shall we say.
>>=20
>>> So I think it=E2=80=99s worth at least giving this a try.
>>=20
>> To be clear it's not a complete solution itself. The problem is of=20
>> course that mm cpumask gives you false negatives, so the bits
>> won't always clean up after themselves as CPUs switch away from their
>> lazy tlb mms.
>=20
> ^^
>=20
> False positives: CPU is in the mm_cpumask, but is not using the mm
> as a lazy tlb. So there can be bits left and never freed.
>=20
> If you closed the false positives, you're back to a shared mm cache
> line on lazy mm context switches.

x86 has this exact problem. At least no more than 64*8 CPUs share the cache l=
ine :)

Can your share your benchmark?

>=20
> Thanks,
> Nick