From mboxrd@z Thu Jan 1 00:00:00 1970 From: Evgeniy Polyakov Subject: Re: [take6 1/3] kevent: Core files. Date: Fri, 11 Aug 2006 11:27:49 +0400 Message-ID: <20060811072748.GA14398@2ka.mipt.ru> References: <20060810061433.GA4689@2ka.mipt.ru> <20060810001844.ff5e7429.akpm@osdl.org> <20060810075047.GB24370@2ka.mipt.ru> <20060810010254.3b52682f.akpm@osdl.org> <20060810082235.GA21025@2ka.mipt.ru> <20060810175639.b64faaa9.akpm@osdl.org> <20060811061535.GA11230@2ka.mipt.ru> <20060810232340.ab326d3f.akpm@osdl.org> <20060811063018.GB11230@2ka.mipt.ru> <20060811000454.d0345288.akpm@osdl.org> Mime-Version: 1.0 Content-Type: text/plain; charset=koi8-r Cc: lkml , David Miller , Ulrich Drepper , netdev , Zach Brown Return-path: Received: from relay.2ka.mipt.ru ([194.85.82.65]:50894 "EHLO 2ka.mipt.ru") by vger.kernel.org with ESMTP id S1161190AbWHKH2Z (ORCPT ); Fri, 11 Aug 2006 03:28:25 -0400 To: Andrew Morton Content-Disposition: inline In-Reply-To: <20060811000454.d0345288.akpm@osdl.org> Sender: netdev-owner@vger.kernel.org List-Id: netdev.vger.kernel.org On Fri, Aug 11, 2006 at 12:04:54AM -0700, Andrew Morton (akpm@osdl.org) wrote: > > This area can be decreased down to 70mb by reducing amount of > > information placed into the buffer (only user's data and flags) without > > additional hints. > > > > 70MB is still very bad, naturally. Actually I do not think that 4k events is a good choice - I expect people will scale it to tens of thousands at least, so we definitely want not to allow user to create way too many kevent fds. > There are other ways in which users can do this sort of thing - passing > fd's across sockets, allocating zillions of pagetables come to mind. But > we don't want to add more. > > Possible options: > > - Add a new rlimit for the number of kevent fd's > > - Add a new rlimit for the amount of kevent memory > > - Add a new rlimit for the total amount of pinned kernel memory. First > user is kevent. I think this rlimit and first one are the best choises. > - Account a kevent fd as being worth 100 regular fds, so the naughty user > hits EMFILE early (ug). > > A new rlimit is attractive, and they're easy to add. Problem is, userspace > support is hard (I think). afaik a standard Linux system doesn't have > global and per-user rlimit config files which are parsed and acted upon at > login. That would make rlimits more useful. As for now it is possible to use stack size rlimit for example. -- Evgeniy Polyakov