Barebox fails to boot 1/40 boots...

Barebox fails to boot 1/40 boots...

[Vanhauwaert Wouter]

Hello all,

I have this issue with our custom design (IMX53, NAND, DDR3)

Everything goes well most of the time, but now and then  the system does not boot and halts during initialization... 
To be more specific, the function on which the system blocks is 'setup_c()' in arch/arm/cpu/start-pbl.c

What can I do more to debug this? This function is completely in assembler...

Grtz,
Wouter



_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox

Re: Barebox fails to boot 1/40 boots...

[Lucas Stach]

Am Dienstag, den 18.06.2013, 14:59 +0200 schrieb Vanhauwaert Wouter:
> Hello all,
> 
> I have this issue with our custom design (IMX53, NAND, DDR3)
> 
> Everything goes well most of the time, but now and then  the system does not boot and halts during initialization... 
> To be more specific, the function on which the system blocks is 'setup_c()' in arch/arm/cpu/start-pbl.c
> 
> What can I do more to debug this? This function is completely in assembler...
> 
> Grtz,
> Wouter
> 

First of all: what barebox version are you running?

System crash in setup_c() smells suspiciously like the unclean cache
issue which was fixed with the 2013.05.1 release.

Regards,
Lucas
-- 
Pengutronix e.K.                           | Lucas Stach                 |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |
Peiner Str. 6-8, 31137 Hildesheim, Germany | Phone: +49-5121-206917-5076 |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |


_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox

RE: Barebox fails to boot 1/40 boots...

[Vanhauwaert Wouter]

>  -----Original Message-----
> From: Lucas Stach [mailto:l.stach@pengutronix.de]
> Sent: dinsdag 18 juni 2013 15:13
> To: Vanhauwaert Wouter
> Cc: barebox@lists.infradead.org
> Subject: Re: Barebox fails to boot 1/40 boots...
> 
> Am Dienstag, den 18.06.2013, 14:59 +0200 schrieb Vanhauwaert Wouter:
> > Hello all,
> >
> > I have this issue with our custom design (IMX53, NAND, DDR3)
> >
> > Everything goes well most of the time, but now and then  the system does not
> boot and halts during initialization...
> > To be more specific, the function on which the system blocks is
> > 'setup_c()' in arch/arm/cpu/start-pbl.c
> >
> > What can I do more to debug this? This function is completely in assembler...
> >
> > Grtz,
> > Wouter
> >
> 
> First of all: what barebox version are you running?
> 
> System crash in setup_c() smells suspiciously like the unclean cache issue which
> was fixed with the 2013.05.1 release.
> 

I'm using stable/v2013.05

_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox

Re: Barebox fails to boot 1/40 boots...

[Sascha Hauer]

On Tue, Jun 18, 2013 at 02:59:57PM +0200, Vanhauwaert Wouter wrote:
> Hello all,
> 
> I have this issue with our custom design (IMX53, NAND, DDR3)
> 
> Everything goes well most of the time, but now and then  the system does not boot and halts during initialization... 
> To be more specific, the function on which the system blocks is 'setup_c()' in arch/arm/cpu/start-pbl.c
> 
> What can I do more to debug this? This function is completely in assembler...

What makes you sure barebox stops there? Do you have DEBUG_LL enabled or
do you use a JTAG debugger?

Does this happen in first stage boot only or also when you chainload
barebox from an already running barebox?

You could try calling imx53_init_lowlevel_early from
barebox_arm_reset_vector(). This *could* make a difference.

Since you are using v2013.05-stable you should have the following
patches, but maybe it's worth rechecking.

76820df ARM: invalidate data caches during early init
f9d7582 ARM v7: added v7_mmu_cache_invalidate()
ee5f859 ARM v7: v7_mmu_cache_flush(): do not restore r0-r3 (minor optimization)
eb7d85c arm: properly init alignment trap bit
7904d90 ARM v7: fix mmu-off operation
9af1150 ARM: fix the memset fix

Sascha

-- 
Pengutronix e.K.                           |                             |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |
Peiner Str. 6-8, 31137 Hildesheim, Germany | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |

_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox

RE: Barebox fails to boot 1/40 boots...

[Vanhauwaert Wouter]

> -----Original Message-----
> From: Sascha Hauer [mailto:s.hauer@pengutronix.de]
> Sent: dinsdag 18 juni 2013 21:25
> To: Vanhauwaert Wouter
> Cc: barebox@lists.infradead.org
> Subject: Re: Barebox fails to boot 1/40 boots...
> 
> On Tue, Jun 18, 2013 at 02:59:57PM +0200, Vanhauwaert Wouter wrote:
> > Hello all,
> >
> > I have this issue with our custom design (IMX53, NAND, DDR3)
> >
> > Everything goes well most of the time, but now and then  the system does not
> boot and halts during initialization...
> > To be more specific, the function on which the system blocks is
> > 'setup_c()' in arch/arm/cpu/start-pbl.c
> >
> > What can I do more to debug this? This function is completely in assembler...
> 
> What makes you sure barebox stops there? Do you have DEBUG_LL enabled or
> do you use a JTAG debugger?
> 

Because these are real on/off tests (with which I mean with a relais which switches on/off every 10 seconds or so, a cold restart) I did not have the ability to test it with a debugger.
What I did was changing the LDO voltage from 1.2V to 1.3V and measure this with a scoop. When I put the adjustment right before this setup_c(), it goes well all the time (even when the board is blocked), 
when I put it after this setup_c(), it doesn't  get through when the board is blocked. So my assumption is it blocks in this function.

> Does this happen in first stage boot only or also when you chainload barebox
> from an already running barebox?

This is in a first stage boot...

> 
> You could try calling imx53_init_lowlevel_early from
> barebox_arm_reset_vector(). This *could* make a difference.
> 
> Since you are using v2013.05-stable you should have the following patches, but
> maybe it's worth rechecking.
> 
> 76820df ARM: invalidate data caches during early init
> f9d7582 ARM v7: added v7_mmu_cache_invalidate()
> ee5f859 ARM v7: v7_mmu_cache_flush(): do not restore r0-r3 (minor
> optimization) eb7d85c arm: properly init alignment trap bit
> 7904d90 ARM v7: fix mmu-off operation
> 9af1150 ARM: fix the memset fix
> 

I'll check

Grtz,
Wouter


_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox