linux-kernel - Re: performance delta after VFS i_mutex=>i

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: <5756E6BF.9020908@hpe.com>
Date:	Tue, 7 Jun 2016 11:22:39 -0400
From:	Waiman Long <waiman.long@....com>
To:	<Valdis.Kletnieks@...edu>
CC:	Linus Torvalds <torvalds@...ux-foundation.org>,
	Dave Hansen <dave.hansen@...el.com>,
	"Chen, Tim C" <tim.c.chen@...el.com>,
	Ingo Molnar <mingo@...hat.com>,
	Davidlohr Bueso <dbueso@...e.de>,
	"Peter Zijlstra (Intel)" <peterz@...radead.org>,
	Jason Low <jason.low2@...com>,
	Michel Lespinasse <walken@...gle.com>,
	"Paul E. McKenney" <paulmck@...ux.vnet.ibm.com>,
	Waiman Long <waiman.long@...com>,
	Al Viro <viro@...iv.linux.org.uk>,
	LKML <linux-kernel@...r.kernel.org>
Subject: Re: performance delta after VFS i_mutex=>i_rwsem conversion

On 06/06/2016 11:22 PM, Valdis.Kletnieks@...edu wrote:
> On Mon, 06 Jun 2016 14:20:32 -0700, Linus Torvalds said:
>
>> I guess some "concurrent readdir with unlink" load would show that
>> behavior, but is it _realistic_? No idea. Let's not worry about it too
>> much until somebody shows a reason to worry.
> I've seen Makefiles where 'make clean' does a 'find . -name "*.o" | xargs rm'.
>
> But if somebody is doing that often enough against cache-cold directory trees
> to matter, they have bigger problems (like learning how to properly develop
> code using 'make').  So unless performance is *so* bad it triggers the lockup
> detector, it's probably OK....
>

The performance won't be very bad. It is just that with the right mix of 
readers and writers, the performance can be a bit worse  with i_rwsem 
than with i_mutex. But in other cases, the performance will be 
comparable or better with i_rwsem.

Regards,
Longman