Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1756737AbdLPOOg (ORCPT ); Sat, 16 Dec 2017 09:14:36 -0500 Received: from bombadil.infradead.org ([65.50.211.133]:51612 "EHLO bombadil.infradead.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752724AbdLPOOe (ORCPT ); Sat, 16 Dec 2017 09:14:34 -0500 Date: Sat, 16 Dec 2017 12:14:27 -0200 From: Mauro Carvalho Chehab To: Jia-Ju Bai Cc: Fabien DESSENNE , "linux-media@vger.kernel.org" , "linux-kernel@vger.kernel.org" , Benjamin GAIGNARD , Hans Verkuil Subject: Re: [PATCH 1/2] bdisp: Fix a possible sleep-in-atomic bug in bdisp_hw_reset Message-ID: <20171216121427.6307c584@recife.lan> In-Reply-To: References: <1513086445-29265-1-git-send-email-baijiaju1990@gmail.com> <0370257c-ce0c-792f-6c85-50ebc18975f9@st.com> X-Mailer: Claws Mail 3.15.1-dirty (GTK+ 2.24.31; x86_64-redhat-linux-gnu) MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 1806 Lines: 55 Em Sat, 16 Dec 2017 19:53:55 +0800 Jia-Ju Bai escreveu: > Hi, > > On 2017/12/15 22:51, Fabien DESSENNE wrote: > > Hi > > > > On 12/12/17 14:47, Jia-Ju Bai wrote: > >> The driver may sleep under a spinlock. > >> The function call path is: > >> bdisp_device_run (acquire the spinlock) > >> bdisp_hw_reset > >> msleep --> may sleep > >> > >> To fix it, msleep is replaced with mdelay. > > May I suggest you to use readl_poll_timeout_atomic (instead of the whole > > "for" block): this fixes the problem and simplifies the code? > > Okay, I have submitted a patch according to your advice. > You can have a look :) This can still be usind mdelay() to wait for a long time. It doesn't seem wise to do that, as it could cause system contention. Couldn't this be reworked in a way to avoid having the spin locked while sleeping? Once we had a similar issue on Siano, and it was solved by this commit 3cdadc50bbe8f04c1231c8af614cafd7ddd622bf Author: Richard Zidlicky Date: Tue Aug 24 09:52:36 2010 -0300 V4L/DVB: dvb: fix smscore_getbuffer() logic Drivers shouldn't sleep while holding a spinlock. A previous workaround were to release the spinlock before callinc schedule(). This patch uses a different approach: it just waits for the siano hardware to answer. Signed-off-by: Richard Zidlicky Cc: stable@kernel.org Signed-off-by: Mauro Carvalho Chehab The code as changed to use wait_event() at the kthread that was waiting for data to arrive. Only when the data is ready, the code with the spin lock is called. It made the driver a way more stable, and didn't add any penalties of needing to do long delays on a non-interruptible code. Thanks, Mauro