From: Petr Grillinger Subject: Re: procfs proc_write_file bug Date: Sat, 20 Nov 2004 10:05:22 +0100 Message-ID: <419F08D2.7080201@kiv.zcu.cz> References: Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Cc: linux-fsdevel@vger.kernel.org Return-path: Received: from s097-n054.tele2.cz ([213.246.97.54]:52743 "EHLO kapr.ath.cx") by vger.kernel.org with ESMTP id S261339AbUKTJFc (ORCPT ); Sat, 20 Nov 2004 04:05:32 -0500 To: Bryan Henderson In-Reply-To: Sender: linux-fsdevel-owner@vger.kernel.org List-Id: linux-fsdevel.vger.kernel.org Bryan Henderson wrote: >>But still thanks for the idea, it didn't occur to me that I could >>solve it by creating my own FS. Maybe next time :-) > > > I would resist the idea of creating a separate filesystem, because of the > added complexity for the user of having to mount it in addition to a the > extra (and duplicate) coding effort. > > Before sysfs, proc filled that role. > > What you're seeing in proc_file_write() isn't a bug but a limitation of > this simplified interface. You can do multipage writes by using the more > general interface where you supply your own file operation methods. (set > the proc_fops field in struct proc_dir_entry). > > -- > Bryan Henderson IBM Almaden Research Center > San Jose CA Filesystems That's it. Exactly what I needed - no need to change the kernel and very simple, thank you. Regarding the proc_file_write, it may not be an implementation bug but then, the FIXME comment, telling that the programmer wasn't sure at all when creating the interface, is confusing. I think, it should be replaced by your advice :-) Petr Grillinger