[Gta04-owner] Now that 3.2-final is out...

[NeilBrown]

On Tue, 17 Jan 2012 20:29:34 +1100 NeilBrown <neilb at suse.de> wrote:

> On Tue, 17 Jan 2012 18:30:19 +1100 NeilBrown <neilb at suse.de> wrote:
> 
> > It should be possible to achieve the same thing by defining
> > CONFIG_EARLY_PRINTK and CONFIG_DEBUG_LL and including 'earlyprintk' in the
> > command-line options.
> > I've tried that and it didn't seem to meet the need I had at the time.  I
> > don't recall the details.
> > 
> > > 
> > > I have used this to debug NULL pointer dereferences in the backport
> > > of the omap-serial.c driver to 2.6.32.
> > > 
> > 
> > The fact that it worked for you gives me hope - I'll try again.
> 
> Isn't it amazing what a little hope can do!!
> 
> This patch:
> 
> diff --git a/arch/arm/plat-omap/include/plat/uncompress.h b/arch/arm/plat-omap/include/plat/uncompress.h
> index 2f472e9..de9bea7 100644
> --- a/arch/arm/plat-omap/include/plat/uncompress.h
> +++ b/arch/arm/plat-omap/include/plat/uncompress.h
> @@ -164,6 +164,7 @@ static inline void __arch_decomp_setup(unsigned long arch_id)
>  		DEBUG_LL_OMAP3(3, omap3517evm);
>  		DEBUG_LL_OMAP3(3, omap3_beagle);
>  		DEBUG_LL_OMAP3(3, omap3_pandora);
> +		DEBUG_LL_OMAP3(3, gta04);
>  		DEBUG_LL_OMAP3(3, omap_ldp);
>  		DEBUG_LL_OMAP3(3, overo);
>  		DEBUG_LL_OMAP3(3, touchbook);
> 
> on top of my current kernel will cause 'earlyprintk' to start working.
> 
> Now I can see where it freezes.
> Immediately after:
> 
> [    6.420043] PM: Adding info for i2c:1-004b
> [    6.429138] bus: 'i2c': driver_probe_device: matched device 1-004b with driver dummy
> [    6.445465] bus: 'i2c': really_probe: probing driver dummy with device 1-004b
> [    6.460540] driver: '1-004b': driver_bound: bound to device 'dummy'
> [    6.473815] bus: 'i2c': really_probe: bound device 1-004b to driver dummy
> 
> 
> which means immediately before:
> 
> [    6.513458] twl4030: PIH (irq 7) chaining IRQs 368..375
> [    6.525207] twl4030: power (irq 373) chaining IRQs 376..383
> [    6.537994] Registering platform device 'twl4030_gpio'. Parent at 1-0049
> [    6.552215] device: 'twl4030_gpio': device_add
> [    6.561767] bus: 'platform': add device twl4030_gpio
> [    6.572601] PM: Adding info for platform:twl4030_gpio
> 
> which makes it look a *lot* like interrupts from the twl4080 PMIC not being
> handled correctly.
> 
> I'll do some more digging in a little while.

Nope.  It wasn't interrupts.

It was power scripts.

I have about half an idea why this is a problem, but as the power scripts
aren't really used as we don't actually power-off the processor yet - just
put it in low-power 'retention' mode - we may as well just disable them for
now:

diff --git a/arch/arm/mach-omap2/board-omap3gta04.c b/arch/arm/mach-omap2/board-omap3gta04.c
index 1abeb17..809b1bb 100644
--- a/arch/arm/mach-omap2/board-omap3gta04.c
+++ b/arch/arm/mach-omap2/board-omap3gta04.c
@@ -753,8 +753,8 @@ static struct twl4030_resconfig twl4030_rconfig[] = {
 };
 
 struct twl4030_power_data gta04_power_scripts = {
-	.scripts	= twl4030_scripts,
-	.num		= ARRAY_SIZE(twl4030_scripts),
+//	.scripts	= twl4030_scripts,
+//	.num		= ARRAY_SIZE(twl4030_scripts),
 	.resource_config = twl4030_rconfig,
 	.use_poweroff	= 1,
 };


I'll push this into the git trees sometime soon.

Which just goes to show that it is even worth while reporting bugs that are
known!  It shows people care which is a great motivator to try harder.  It
also helps get ideas for more places.

Thanks,
NeilBrown
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 828 bytes
Desc: not available
URL: <http://lists.goldelico.com/pipermail/gta04-owner/attachments/20120117/50d06788/attachment.bin>