From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-8.5 required=3.0 tests=HEADER_FROM_DIFFERENT_DOMAINS, INCLUDES_PATCH,MAILING_LIST_MULTI,SIGNED_OFF_BY,SPF_PASS,URIBL_BLOCKED, USER_AGENT_MUTT autolearn=ham autolearn_force=no version=3.4.0 Received: from mail.kernel.org (mail.kernel.org [198.145.29.99]) by smtp.lore.kernel.org (Postfix) with ESMTP id CD16BC43441 for ; Fri, 9 Nov 2018 13:34:48 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by mail.kernel.org (Postfix) with ESMTP id 83CF02081C for ; Fri, 9 Nov 2018 13:34:48 +0000 (UTC) DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org 83CF02081C Authentication-Results: mail.kernel.org; dmarc=fail (p=none dis=none) header.from=linux.intel.com Authentication-Results: mail.kernel.org; spf=none smtp.mailfrom=linux-kernel-owner@vger.kernel.org Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1727941AbeKIXP0 (ORCPT ); Fri, 9 Nov 2018 18:15:26 -0500 Received: from mga04.intel.com ([192.55.52.120]:47054 "EHLO mga04.intel.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1727667AbeKIXP0 (ORCPT ); Fri, 9 Nov 2018 18:15:26 -0500 X-Amp-Result: UNKNOWN X-Amp-Original-Verdict: FILE UNKNOWN X-Amp-File-Uploaded: False Received: from fmsmga001.fm.intel.com ([10.253.24.23]) by fmsmga104.fm.intel.com with ESMTP/TLS/DHE-RSA-AES256-GCM-SHA384; 09 Nov 2018 05:34:46 -0800 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.54,483,1534834800"; d="scan'208";a="106720099" Received: from kuha.fi.intel.com ([10.237.72.189]) by fmsmga001.fm.intel.com with SMTP; 09 Nov 2018 05:34:43 -0800 Received: by kuha.fi.intel.com (sSMTP sendmail emulation); Fri, 09 Nov 2018 15:34:43 +0200 Date: Fri, 9 Nov 2018 15:34:42 +0200 From: Heikki Krogerus To: Rob Herring Cc: Andy Shevchenko , "Rafael J. Wysocki" , Mika Westerberg , "linux-kernel@vger.kernel.org" , linux-acpi@vger.kernel.org, devicetree@vger.kernel.org Subject: Re: [PATCH v2 3/4] of/property: Add of_fwnode_name() Message-ID: <20181109133442.GA28183@kuha.fi.intel.com> References: <20181108165156.60073-1-heikki.krogerus@linux.intel.com> <20181108165156.60073-4-heikki.krogerus@linux.intel.com> <20181108192509.GW10650@smile.fi.intel.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: User-Agent: Mutt/1.10.1 (2018-07-13) Sender: linux-kernel-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Thu, Nov 08, 2018 at 03:18:21PM -0600, Rob Herring wrote: > On Thu, Nov 8, 2018 at 1:25 PM Andy Shevchenko > wrote: > > > > On Thu, Nov 08, 2018 at 12:23:51PM -0600, Rob Herring wrote: > > > On Thu, Nov 8, 2018 at 10:52 AM Heikki Krogerus > > > wrote: > > > > > > > > This implements get_name fwnode op for DT. > > > > > > > > Signed-off-by: Heikki Krogerus > > > > --- > > > > drivers/of/property.c | 11 +++++++++++ > > > > 1 file changed, 11 insertions(+) > > > > > > > > diff --git a/drivers/of/property.c b/drivers/of/property.c > > > > index f46828e3b082..9bc8fe136fa3 100644 > > > > --- a/drivers/of/property.c > > > > +++ b/drivers/of/property.c > > > > @@ -823,6 +823,16 @@ static void of_fwnode_put(struct fwnode_handle *fwnode) > > > > of_node_put(to_of_node(fwnode)); > > > > } > > > > > > > > +static int of_fwnode_get_name(const struct fwnode_handle *fwnode, char *buf) > > > > +{ > > > > + const char *name = kbasename(to_of_node(fwnode)->full_name); > > > > + size_t len = strchrnul(name, '@') - name; > > > > + > > > > + snprintf(buf, len + 1, "%s", name); > > > > > > This can be simplified to: > > > > > > snprintf(..., "%pOFn", to_of_node(fwnode)) > > > > > > But that presents a problem with knowing the length. Not passing in > > > the buf length is not good design because you can't tell if you > > > overflow the buffer. Either you can pass in the length of buf or do > > > the allocation here. In the latter case, then you can use kasnprintf. > > > The downside to doing the allocation here is then get_name() has side > > > effect of allocating memory that the caller needs to be aware of. > > > > Agree on matter of potential overflow. > > > > I wouldn't limit users with 32 characters for node name if it's not by both > > ACPI and DT specifications. > > While the DT spec says 31 characters, this has never been enforced. As > you might guess, we have node names longer than 31 characters. There's > been some discussion about what to do and the consensus seems to be > change the spec. > > > OTOH allocating and freeing memory in a loop each > > time when we would like to go through the child nodes sounds much worse > > scenario to me. > > Yes, I wrote that before looking at how you were using it... Of > course, if you want efficient, then you shouldn't use sprintf either > and use of_node_name_eq() as I've suggested. Since the fwnode API is just a wrapper layer (at least IMO), I don't think there should be any assumptions that it provides the optimal solution for anything. The low-level APIs should be the ones providing the optimal solutions. > > Thus, giving a length of the buffer is a good enough compromise. OK. That's what we'll do then. > > > However, I think the current API is better. It leaves low-level > > > details up to the firmware implementation. But as long as .get_name() > > > is not exposed to drivers I don't really care that much. Side note: I would prefer that we had something like of_node_name_get() function that of_fwnode_get_name() could simply call. I don't know how you want that implemented, but I'm expecting you will implement something like that in any case once you start removing that ->name member. I figured that at that point we can update of_fwnode_get_name() as well. > > I don't think this concept is changed by Heikki's patch series. It provides a > > common abstract function on top of more low-level firmware implementation which > > I consider a good approach. > > Generally, I would agree that's a worthwhile goal. However, in this > case you aren't saving anything. We still have at least a DT version > of the same thing (of_get_child_by_name). Maybe there's some dream > that the fwnode API will become the only one for both drivers and > non-drivers, but I really don't see that happening. As long as both DT > and fwnode APIs are in use, it is best to keep the APIs aligned. I don't think that anybody was planning to have the fwnode API as the only available API for the drivers. It's just a helper for the drivers on top of the low-level APIs, so the low-level APIs really need to always stay available for the drivers. There will always be corner cases. > There's another aspect that the node name comparison is case > insensitive on powerpc (and until 4.20, was for everything but Sparc). > I changed it in 4.20 and hopefully don't have to revert. Patch 4 does > a case sensitive compare. That probably is fine (as lots of powerpc > code already does case sensitive name compares), but no one really > knows until we break users. I actually used stccasecmp() in the first version. I don't know how, or why, I've changed it to strcmp(). I'll fix that. > Another issue is how are disabled nodes dealt with by different > firmwares? It's a frequent bug that we don't honor the 'status' > property (such as in the very code we're discussing). But then there > are some cases were want to ignore it so we can't just go add that > check in and we end up needing 2 flavors of everything. You're > probably okay though. Most devices with child nodes are > enabled/disabled only in the parent device node. thanks, -- heikki