Re: [rtc-linux] Re: [PATCH] rtc: Make rtc_update_irq callable with irqs enabled

From: Atsushi Nemoto <anemo@mba.ocn.ne.jp>
To: alessandro.zummo@towertech.it
Cc: rtc-linux@googlegroups.com, david-b@pacbell.net,
	akpm@linux-foundation.org, linux-kernel@vger.kernel.org,
	hcegtvedt@atmel.com, vapier@gentoo.org,
	rongkai.zhan@windriver.com, balajirrao@openmoko.org,
	broonie@opensource.wolfsonmicro.com
Subject: Re: [rtc-linux] Re: [PATCH] rtc: Make rtc_update_irq callable with irqs enabled
Date: Sat, 25 Apr 2009 01:48:50 +0900 (JST)	[thread overview]
Message-ID: <20090425.014850.93020801.anemo@mba.ocn.ne.jp> (raw)
In-Reply-To: <20090424131334.10959f78@i1501.lan.towertech.it>

On Fri, 24 Apr 2009 13:13:34 +0200, Alessandro Zummo <alessandro.zummo@towertech.it> wrote:
> > >  use spin_lock() in the IRQ handler and spin_lock_irq/irq_save
> > >  in the setup functions.  
> > 
> > I think you're describing how the *current* scheme is supposed
> > to work ... except that some IRQ handlers aren't calling the
> > rtc_update_irq() routine with IRQs blocked.
> > 
> > Yes, that current scheme works ... modulo those buggy handlers.
> 
>  ok, but why it's necessary to disable the interrupts? Only because
>  the specs says so or because there's a locking issue I'm missing?

Here is a possible example:

1. RTC alarm interrupt handler takes rtc->irq_lock by spin_lock()
2. A timer interrupt handler calls rtc_uie_timer() for UIE emulation
3. rtc_uie_timer() waits on rtc->irq_lock .... deadlock!

This sort of deadlock can happen if a spin lock was used by multple
interrupt handlers.

---
Atsushi Nemoto