From mboxrd@z Thu Jan 1 00:00:00 1970 From: Adam Wozniak Subject: spin_lock_irqsave and ISP interrupts on OMAP 3530 Date: Thu, 03 Feb 2011 08:31:51 -0800 Message-ID: <4D4AD877.8070702@irobot.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Return-path: Received: from smtp1.irobot.com ([206.83.81.187]:24300 "EHLO smtp1.irobot.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1756002Ab1BCQrL (ORCPT ); Thu, 3 Feb 2011 11:47:11 -0500 Sender: linux-omap-owner@vger.kernel.org List-Id: linux-omap@vger.kernel.org To: linux-omap@vger.kernel.org I'm hoping I'm just not understanding something and this isn't as broke as I fear it is. I'm using the ISP on an OMAP3530, linux 2.6.32. I've got some kernel code I want to lock in both the process context and the ISP interrupt context, so I'm using a spinlock. As I understand it, locking the spinlock should disable all interrupts. However, I'm still seeing the ISP interrupt happen when I've got the spinlock locked in the process context. i.e. somewhere I've got the code below, and my Warning is tripping. Is this broken, or am I not understanding something? int am_i_locked = 0; spinlock_t my_spinlock = SPIN_LOCK_UNLOCKED; void foo(void) { /* function could be called from process context, or could be called from omap34xx_isp_isr (interrupt context) */ unsigned long flags; spin_lock_irqsave(&my_spinlock, flags); WARN_ON(am_i_locked); am_i_locked++; /* do some stuff */ am_i_locked--; spin_unlock_irqrestore(&my_spinlock, flags); }