From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mx2.suse.de ([195.135.220.15]:39292 "EHLO mx1.suse.de" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S1727575AbeLFK0A (ORCPT ); Thu, 6 Dec 2018 05:26:00 -0500 MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII; format=flowed Content-Transfer-Encoding: 7bit Date: Thu, 06 Dec 2018 11:25:57 +0100 From: Roman Penyaev To: Davidlohr Bueso Cc: Jason Baron , Alexander Viro , "Paul E. McKenney" , Linus Torvalds , linux-fsdevel@vger.kernel.org, linux-kernel@vger.kernel.org, akpm@linux-foundation.org Subject: Re: [RFC PATCH 1/1] epoll: use rwlock in order to reduce ep_poll_callback() contention In-Reply-To: <20181206040430.d7wjik6lt7kr75m6@linux-r8p5> References: <20181203110237.14787-1-rpenyaev@suse.de> <45bce871-edfd-c402-acde-2e57e80cc522@akamai.com> <20181206040430.d7wjik6lt7kr75m6@linux-r8p5> Message-ID: <98d200aa103fd6086c02dd620b65e961@suse.de> Sender: linux-fsdevel-owner@vger.kernel.org List-ID: On 2018-12-06 05:04, Davidlohr Bueso wrote: > On 12/3/18 6:02 AM, Roman Penyaev wrote: > >> The main change is in replacement of the spinlock with a rwlock, which >> is >> taken on read in ep_poll_callback(), and then by adding poll items to >> the >> tail of the list using xchg atomic instruction. Write lock is taken >> everywhere else in order to stop list modifications and guarantee that >> list >> updates are fully completed (I assume that write side of a rwlock does >> not >> starve, it seems qrwlock implementation has these guarantees). > > Its good then that Will recently ported qrwlocks to arm64, which iirc > had > a bad case of writer starvation. In general, qrwlock will maintain > reader > to writer ratios of acquisitions fairly well, but will favor readers > over > writers in scenarios where when too many tasks (more than ncpus). Thanks for noting that. Then that should not be a problem, since number of parallel ep_poll_callback() calls can't be greater then number of CPUs because of the wq.lock which is taken by the caller of ep_poll_callback(). BTW, did someone make any estimations how much does the latency on the write side increase if the number of readers is greater than CPUs? -- Roman