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;
From:   Waiman Long <llong@redhat.com>
Subject: Re: [PATCH v15 3/6] locking/qspinlock: Introduce CNA into the slow
 path of qspinlock
To:     Davidlohr Bueso <dave@stgolabs.net>,
        Alex Kogan <alex.kogan@oracle.com>
Cc:     linux@armlinux.org.uk, peterz@infradead.org, mingo@redhat.com,
        will.deacon@arm.com, arnd@arndb.de, linux-arch@vger.kernel.org,
        linux-arm-kernel@lists.infradead.org, linux-kernel@vger.kernel.org,
        tglx@linutronix.de, bp@alien8.de, hpa@zytor.com, x86@kernel.org,
        guohanjun@huawei.com, jglauber@marvell.com,
        steven.sistare@oracle.com, daniel.m.jordan@oracle.com,
        dave.dice@oracle.com
References: <20210514200743.3026725-1-alex.kogan@oracle.com>
 <20210514200743.3026725-4-alex.kogan@oracle.com>
 <20210922192528.ob22pu54oeqsoeno@offworld>
Message-ID: <8e743acb-ec75-ea03-493a-d57154ab8fed@redhat.com>
Date:   Wed, 22 Sep 2021 15:52:18 -0400
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:78.0) Gecko/20100101
 Thunderbird/78.11.0
MIME-Version: 1.0
In-Reply-To: <20210922192528.ob22pu54oeqsoeno@offworld>
Content-Type: text/plain; charset=utf-8; format=flowed
Content-Transfer-Encoding: 8bit
Content-Language: en-US
Precedence: bulk

On 9/22/21 3:25 PM, Davidlohr Bueso wrote:
> On Fri, 14 May 2021, Alex Kogan wrote:
>
>> diff --git a/Documentation/admin-guide/kernel-parameters.txt 
>> b/Documentation/admin-guide/kernel-parameters.txt
>> index a816935d23d4..94d35507560c 100644
>> --- a/Documentation/admin-guide/kernel-parameters.txt
>> +++ b/Documentation/admin-guide/kernel-parameters.txt
>> @@ -3515,6 +3515,16 @@
>>             NUMA balancing.
>>             Allowed values are enable and disable
>>
>> +    numa_spinlock=    [NUMA, PV_OPS] Select the NUMA-aware variant
>> +            of spinlock. The options are:
>> +            auto - Enable this variant if running on a multi-node
>> +            machine in native environment.
>> +            on  - Unconditionally enable this variant.
>
> Is there any reason why the user would explicitly pass the on option
> when the auto thing already does the multi-node check? Perhaps strange
> numa topologies? Otherwise I would say it's not needed and the fewer
> options we give the user for low level locking the better.

I asked Alex to put in a command line option because we may want to 
disable it on a multi-socket server if we want to.


>
>> +            off - Unconditionally disable this variant.
>> +
>> +            Not specifying this option is equivalent to
>> +            numa_spinlock=auto.
>> +
>>     numa_zonelist_order= [KNL, BOOT] Select zonelist order for NUMA.
>>             'node', 'default' can be specified
>>             This can be set from sysctl after boot.
>> diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig
>> index 0045e1b44190..819c3dad8afc 100644
>> --- a/arch/x86/Kconfig
>> +++ b/arch/x86/Kconfig
>> @@ -1564,6 +1564,26 @@ config NUMA
>>
>>       Otherwise, you should say N.
>>
>> +config NUMA_AWARE_SPINLOCKS
>> +    bool "Numa-aware spinlocks"
>> +    depends on NUMA
>> +    depends on QUEUED_SPINLOCKS
>> +    depends on 64BIT
>> +    # For now, we depend on PARAVIRT_SPINLOCKS to make the patching 
>> work.
>> +    # This is awkward, but hopefully would be resolved once 
>> static_call()
>> +    # is available.
>> +    depends on PARAVIRT_SPINLOCKS
>
> We now have static_call() - see 9183c3f9ed7.
I agree that it is now time to look at using the static call for 
slowpath switching.
>
>
>> +    default y
>> +    help
>> +      Introduce NUMA (Non Uniform Memory Access) awareness into
>> +      the slow path of spinlocks.
>> +
>> +      In this variant of qspinlock, the kernel will try to keep the 
>> lock
>> +      on the same node, thus reducing the number of remote cache 
>> misses,
>> +      while trading some of the short term fairness for better 
>> performance.
>> +
>> +      Say N if you want absolute first come first serve fairness.
>
> This would also need a depends on !PREEMPT_RT, no? Raw spinlocks 
> really want
> the determinism. 

Agreed

Cheers,
Longman