Return-path: Received: from he.sipsolutions.net ([78.46.109.217]:49940 "EHLO sipsolutions.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S934064Ab3DGRXA (ORCPT ); Sun, 7 Apr 2013 13:23:00 -0400 Message-ID: <1365355372.9172.1.camel@jlt4.sipsolutions.net> (sfid-20130407_192304_166156_39659473) Subject: Re: [PATCH 4/9] compat: backport EPROBE_DEFER value From: Johannes Berg To: "Luis R. Rodriguez" Cc: backports@vger.kernel.org Date: Sun, 07 Apr 2013 19:22:52 +0200 In-Reply-To: <1365301322-19327-4-git-send-email-mcgrof@do-not-panic.com> (sfid-20130407_042226_233475_AF88C7EE) References: <1365301322-19327-1-git-send-email-mcgrof@do-not-panic.com> <1365301322-19327-4-git-send-email-mcgrof@do-not-panic.com> (sfid-20130407_042226_233475_AF88C7EE) Content-Type: text/plain; charset="UTF-8" Mime-Version: 1.0 Sender: backports-owner@vger.kernel.org List-ID: On Sat, 2013-04-06 at 19:21 -0700, Luis R. Rodriguez wrote: > From: "Luis R. Rodriguez" > > Grant added the ability to the driver core to return > EPROBE_DEFER at device driver probe time and for that to > trigger a later deferred probe through a workqueue. It > seems we could in theory backport this functionality but > that requires quite a bit of work. For now just pass on > the EPROBE_DEFER definition safely and for older kernels > than 3.4 the EPROBE_DEFER would be treated as a regular > probe error. > +#define EPROBE_DEFER 517 /* Driver requests probe retry */ I'm not convinced this is useful, particularly not with that error number? If a driver returns -EPROBE_DEFER it will not work correctly, so why not just require it to be disabled (via the dependencies list) for kernels before 3.4? Also, if you return -517 userspace might get somewhat confused? johannes