linux-kernel - Re: [PATCH v5 1/2] kernfs: use hashed mutex and spinlock in place of global ones.

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: <YgKm5aSCcCYWkck2@slm.duckdns.org>
Date:   Tue, 8 Feb 2022 07:22:45 -1000
From:   Tejun Heo <tj@...nel.org>
To:     Imran Khan <imran.f.khan@...cle.com>
Cc:     gregkh@...uxfoundation.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH v5 1/2] kernfs: use hashed mutex and spinlock in place of
 global ones.

On Sun, Feb 06, 2022 at 12:09:24PM +1100, Imran Khan wrote:
> +/*
> + * NR_KERNFS_LOCK_BITS determines size (NR_KERNFS_LOCKS) of hash
> + * table of locks.
> + * Having a small hash table would impact scalability, since
> + * more and more kernfs_node objects will end up using same lock
> + * and having a very large hash table would waste memory.
> + *
> + * At the moment size of hash table of locks is being set based on
> + * the number of CPUs as follows:
> + *
> + * NR_CPU      NR_KERNFS_LOCK_BITS      NR_KERNFS_LOCKS
> + *   1                  1                       2
> + *  2-3                 2                       4
> + *  4-7                 4                       16
> + *  8-15                6                       64
> + *  16-31               8                       256
> + *  32 and more         10                      1024
> + */
> +#ifdef CONFIG_SMP
> +#define NR_KERNFS_LOCK_BITS (2 * (ilog2(NR_CPUS < 32 ? NR_CPUS : 32)))
> +#else
> +#define NR_KERNFS_LOCK_BITS     1
> +#endif
> +
> +#define NR_KERNFS_LOCKS     (1 << NR_KERNFS_LOCK_BITS)

I have a couple questions:

* How did you come up with the above numbers? Are they based on some
  experimentation? It'd be nice if the comment explains why the numbers are
  like that.

* IIRC, we split these locks to per kernfs instance recently as a way to
  mitigate lock contention occurring across kernfs instances. I don't think
  it's beneficial to keep these hashed locks separate. It'd be both simpler
  and less contended to double one shared hashtable than splitting the table
  into two separate half sized ones. So, maybe switch to global hashtables
  and increase the size?

Thanks.

-- 
tejun