From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1751422AbaEUCzt (ORCPT ); Tue, 20 May 2014 22:55:49 -0400 Received: from mail-pd0-f169.google.com ([209.85.192.169]:40733 "EHLO mail-pd0-f169.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750737AbaEUCzr (ORCPT ); Tue, 20 May 2014 22:55:47 -0400 Message-ID: <537C15B1.3020006@gmail.com> Date: Tue, 20 May 2014 19:55:45 -0700 From: Frank Rowand Reply-To: frowand.list@gmail.com User-Agent: Mozilla/5.0 (Windows NT 6.1; WOW64; rv:17.0) Gecko/20130801 Thunderbird/17.0.8 MIME-Version: 1.0 To: Grant Likely CC: linux-kernel@vger.kernel.org, devicetree@vger.kernel.org, David Daney , Pantelis Antoniou Subject: Re: [PATCH 2/3] of: Make of_find_node_by_path() handle /aliases References: <1399993115-21552-1-git-send-email-grant.likely@linaro.org> <1399993115-21552-3-git-send-email-grant.likely@linaro.org> In-Reply-To: <1399993115-21552-3-git-send-email-grant.likely@linaro.org> Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On 5/13/2014 7:58 AM, Grant Likely wrote: > Make of_find_node_by_path() handle aliases as prefixes. To make this > work the name search is refactored to search by path component instead > of by full string. This should be a more efficient search, and it makes > it possible to start a search at a subnode of a tree. > > Signed-off-by: David Daney > Signed-off-by: Pantelis Antoniou > [grant.likely: Rework to not require allocating at runtime] > Acked-by: Rob Herring > Signed-off-by: Grant Likely > --- > drivers/of/base.c | 60 +++++++++++++++++++++++++++++++++++++++++++++++++++---- > 1 file changed, 56 insertions(+), 4 deletions(-) Was this patch created against a tree that has modifications to device tree locking? I get a hang due to deadlock when I apply it. Patch to verify the cause is below. From: Frank Rowand Do not apply this patch -- it is not a valid fix. The point of this patch is to confirm the cause of a bug. Problem seen on 3.15-rc1 with dragonboard platform patches added. The dragonboard hangs on boot the very first time that of_find_node_by_path() is called: of_find_node_by_path() raw_spin_lock_irqsave(&devtree_lock, flags) np = __of_find_node_by_path(np, path) for_each_child_of_node(parent, child) { // This is a define: #define for_each_child_of_node(parent, child) \ for (child = of_get_next_child(parent, NULL); child != NULL; \ // Resulting in a deadlock when of_get_next_child() // attempts to lock the same lock: raw_spin_lock_irqsave(&devtree_lock, flags) When this patch is applied, the first lock is commented out so the deadlock does not occur and the boot completes. Not-signed-off-by: Frank Rowand --- drivers/of/base.c | 4 ++++ 1 file changed, 4 insertions(+) Index: b/drivers/of/base.c =================================================================== --- a/drivers/of/base.c +++ b/drivers/of/base.c @@ -835,13 +835,17 @@ struct device_node *of_find_node_by_path } /* Step down the tree matching path components */ +#if 0 raw_spin_lock_irqsave(&devtree_lock, flags); +#endif while (np && *path == '/') { path++; /* Increment past '/' delimiter */ np = __of_find_node_by_path(np, path); path = strchrnul(path, '/'); } +#if 0 raw_spin_unlock_irqrestore(&devtree_lock, flags); +#endif return np; } EXPORT_SYMBOL(of_find_node_by_path);