linux-kernel - Re: [PATCH v7 3/5] locking/qspinlock: Introduce CNA into the slow path of qspinlock

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <20200122085816.GB15537@willie-the-truck>
Date:   Wed, 22 Jan 2020 08:58:17 +0000
From:   Will Deacon <will@...nel.org>
To:     Peter Zijlstra <peterz@...radead.org>
Cc:     Alex Kogan <alex.kogan@...cle.com>, linux-arch@...r.kernel.org,
        guohanjun@...wei.com, arnd@...db.de, dave.dice@...cle.com,
        jglauber@...vell.com, x86@...nel.org, will.deacon@....com,
        linux@...linux.org.uk, steven.sistare@...cle.com,
        linux-kernel@...r.kernel.org, rahul.x.yadav@...cle.com,
        mingo@...hat.com, bp@...en8.de, hpa@...or.com, longman@...hat.com,
        tglx@...utronix.de, daniel.m.jordan@...cle.com,
        linux-arm-kernel@...ts.infradead.org
Subject: Re: [PATCH v7 3/5] locking/qspinlock: Introduce CNA into the slow
 path of qspinlock

On Tue, Jan 21, 2020 at 09:29:19PM +0100, Peter Zijlstra wrote:
> 
> various notes and changes in the below.
> 
> ---
> 
> Index: linux-2.6/kernel/locking/qspinlock.c
> ===================================================================
> --- linux-2.6.orig/kernel/locking/qspinlock.c
> +++ linux-2.6/kernel/locking/qspinlock.c
> @@ -598,10 +598,10 @@ EXPORT_SYMBOL(queued_spin_lock_slowpath)
>  #define _GEN_CNA_LOCK_SLOWPATH
>  
>  #undef pv_wait_head_or_lock
> -#define pv_wait_head_or_lock		cna_pre_scan
> +#define pv_wait_head_or_lock		cna_wait_head_or_lock
>  
>  #undef try_clear_tail
> -#define try_clear_tail			cna_try_change_tail
> +#define try_clear_tail			cna_try_clear_tail
>  
>  #undef mcs_pass_lock
>  #define mcs_pass_lock			cna_pass_lock
> Index: linux-2.6/kernel/locking/qspinlock_cna.h
> ===================================================================
> --- linux-2.6.orig/kernel/locking/qspinlock_cna.h
> +++ linux-2.6/kernel/locking/qspinlock_cna.h
> @@ -8,37 +8,37 @@
>  /*
>   * Implement a NUMA-aware version of MCS (aka CNA, or compact NUMA-aware lock).
>   *
> - * In CNA, spinning threads are organized in two queues, a main queue for
> + * In CNA, spinning threads are organized in two queues, a primary queue for
>   * threads running on the same NUMA node as the current lock holder, and a
> - * secondary queue for threads running on other nodes. Schematically, it
> - * looks like this:
> + * secondary queue for threads running on other nodes. Schematically, it looks
> + * like this:
>   *
>   *    cna_node
> - *   +----------+    +--------+        +--------+
> - *   |mcs:next  | -> |mcs:next| -> ... |mcs:next| -> NULL      [Main queue]
> - *   |mcs:locked| -+ +--------+        +--------+
> + *   +----------+     +--------+         +--------+
> + *   |mcs:next  | --> |mcs:next| --> ... |mcs:next| --> NULL  [Primary queue]
> + *   |mcs:locked| -.  +--------+         +--------+
>   *   +----------+  |
> - *                 +----------------------+
> - *                                        \/
> + *                 `----------------------.
> + *                                        v
>   *                 +--------+         +--------+
> - *                 |mcs:next| -> ...  |mcs:next|          [Secondary queue]
> + *                 |mcs:next| --> ... |mcs:next|            [Secondary queue]
>   *                 +--------+         +--------+
>   *                     ^                    |
> - *                     +--------------------+
> + *                     `--------------------'
>   *
> - * N.B. locked = 1 if secondary queue is absent. Othewrise, it contains the
> + * N.B. locked := 1 if secondary queue is absent. Othewrise, it contains the

If we're redoing the comment, please can you s/Othewrise/Otherwise/ at the
same time? It catches me every time I read it!

Will