Re: OMAP3EVM not booting on l-o master

From: Tony Lindgren <tony@atomide.com>
To: Paul Walmsley <paul@pwsan.com>
Cc: "Mohammed, Afzal" <afzal@ti.com>,
	"R, Govindraj" <govindraj.raja@ti.com>,
	khilman@ti.com,
	"linux-omap@vger.kernel.org" <linux-omap@vger.kernel.org>
Subject: Re: OMAP3EVM not booting on l-o master
Date: Fri, 6 Apr 2012 11:24:36 -0700	[thread overview]
Message-ID: <20120406182435.GD15734@atomide.com> (raw)
In-Reply-To: <alpine.DEB.2.00.1204061159100.11100@utopia.booyaka.com>

* Paul Walmsley <paul@pwsan.com> [120406 11:11]:
> cc Kevin
> 
> On Fri, 6 Apr 2012, Tony Lindgren wrote:
> 
> > So something in Paul's series is changing the interrupt
> > behaviour somehow?
> 
> Am wondering if this might be due to the kernel not clearing some 
> bootloader mux settings?
> 
> Perhaps the bootloader is configuring the mux hardware for I/O wakeups on 
> a pad, but the kernel is missing an mux IRQ handler or mapping for it. Or 
> maybe one of the kernel mux entries is missing the OMAP_WAKEUP_EN or 
> OMAP_DEVICE_PAD_WAKEUP flags for a pad that the bootloader is configuring 
> - I seem to recall that Kevin found some problems here with the 
> mach-omap2/serial.c mux code.  Anyway, will create a patch to add some 
> debugging here for Afzal to try.

That's easy to check from dumping out the mux status from
/sys/kernel/debug/omap_mux/* entries after booting with a working
kernel.

> Also,it seems to me that when the mux layer starts, the code should 
> probably clear out all of the I/O wakeup configuration in the hardware 
> that the bootloader set up.

Yes that sounds right.

Regards,

Tony