From mboxrd@z Thu Jan 1 00:00:00 1970 From: Cong Wang Subject: Re: [net-next PATCH v5 2/3] sysctl: add proc_do_large_bitmap Date: Mon, 22 Feb 2010 10:45:44 +0800 Message-ID: <4B81EFD8.8020202@redhat.com> References: <1266752533.3428.28.camel@Nokia-N900-42-11> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 7bit Cc: David Miller , Linux Kernel Network Developers , Linux Kernel Developers , "Eric W. Biederman" To: opurdila@ixiacom.com Return-path: Received: from mx1.redhat.com ([209.132.183.28]:58215 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754050Ab0BVCmQ (ORCPT ); Sun, 21 Feb 2010 21:42:16 -0500 In-Reply-To: <1266752533.3428.28.camel@Nokia-N900-42-11> Sender: netdev-owner@vger.kernel.org List-ID: Octavian Purdila wrote: > > But I think its worth to keep the whitespaces in beetween, e.g. allow > > $ echo '1, 2 ,3 ' > ip_local_reserved_ports. Sure. > >> Also, if I write an invalid value, it does reject this, but the previous >> value in that file is cleared, shouldn't we keep the previous one? >> >> > > The only way I see to fix this is to return EINVAL if we detect a write with offset. Yeah, we shouldn't continue once we find any invalid value. > > IMO we should do that for the other proc write routines as well, as otherwise ther result is confusing, e.g. > > write("1 2"); write(" 3"); > > will set first value in the vector to 1, than second value to 2 then *first* value to 3. > > I am all for it, but again, this changes userspace ABI. Sorry, is this related with the problem I mentioned above? Both "1 2" and " 3" are valid values. Thanks.