OMAP 3530 arch_shutdown "undefined instruction"

OMAP 3530 arch_shutdown "undefined instruction"

[Orjan Friberg]

Hi,

I'm trying to use barebox as both the 2nd and 3rd stage bootloader (boot 
ROM being the 1st) on an OMAP 3530 (BeagleBoard).

Every go <address> command in the 2nd stage, regardless of whether I 
have uploaded anything to this address or not, is met with a

go 0x80e80000
## Starting application at 0x80E80000 ...
undefined instruction
pc : [<40208b00>]    lr : [<40203a34>]
sp : 87bff21c  ip : 00000028  fp : 87bffa4d
r10: 00000000  r9 : 00000000  r8 : 87bff64d
r7 : 87bffe44  r6 : 00000002  r5 : 4020b1e8  r4 : 80e80000
r3 : 00000000  r2 : 00000001  r1 : 0000843f  r0 : 4020ae50
Flags: nZCv  IRQs off  FIQs off  Mode SVC_32
Resetting CPU ...

which is in arch_shutdown:

40208af8 <arch_shutdown>:
40208af8:       e3a03000        mov     r3, #0  ; 0x0
40208afc:       ee073f17        mcr     15, 0, r3, cr7, cr7, {0}
40208b00:       e12fff1e        bx      lr

This is from the 2010.05.0 snapshot, though it seems that code hasn't 
changed in the latest trunk (arch/arm/cpu/cpu.c):

	/* flush I/D-cache */
	i = 0;
	asm ("mcr p15, 0, %0, c7, c7, 0": :"r" (i));


This is compiled with the CodeSourcery 2009q1 tool chain.

Commenting out the call to arch_shutdown makes the go command work 
without any obvious deficiencies, but that doesn't seem like the right 
way to do it.

(I haven't tried the latest trunk as it doesn't compile.)

Thanks for any suggestions,
Orjan

-- 
Orjan Friberg
FlatFrog Laboratories AB

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

Re: OMAP 3530 arch_shutdown "undefined instruction"

[Sascha Hauer]

Hi Orjan,

On Wed, May 26, 2010 at 04:48:09PM +0200, Orjan Friberg wrote:
> Hi,
>
> I'm trying to use barebox as both the 2nd and 3rd stage bootloader (boot  
> ROM being the 1st) on an OMAP 3530 (BeagleBoard).
>
> Every go <address> command in the 2nd stage, regardless of whether I  
> have uploaded anything to this address or not, is met with a
>
> go 0x80e80000
> ## Starting application at 0x80E80000 ...
> undefined instruction
> pc : [<40208b00>]    lr : [<40203a34>]
> sp : 87bff21c  ip : 00000028  fp : 87bffa4d
> r10: 00000000  r9 : 00000000  r8 : 87bff64d
> r7 : 87bffe44  r6 : 00000002  r5 : 4020b1e8  r4 : 80e80000
> r3 : 00000000  r2 : 00000001  r1 : 0000843f  r0 : 4020ae50
> Flags: nZCv  IRQs off  FIQs off  Mode SVC_32
> Resetting CPU ...
>
> which is in arch_shutdown:
>
> 40208af8 <arch_shutdown>:
> 40208af8:       e3a03000        mov     r3, #0  ; 0x0
> 40208afc:       ee073f17        mcr     15, 0, r3, cr7, cr7, {0}
> 40208b00:       e12fff1e        bx      lr
>
> This is from the 2010.05.0 snapshot, though it seems that code hasn't  
> changed in the latest trunk (arch/arm/cpu/cpu.c):
>
> 	/* flush I/D-cache */
> 	i = 0;
> 	asm ("mcr p15, 0, %0, c7, c7, 0": :"r" (i));
>
>

Seems this does not work on Cortex Processors. Can you try replacing
this with the following please:

	asm volatile (
		"bl __mmu_cache_flush;"
		"bl __mmu_cache_off;"
		:
		:
		: "r0", "r1", "r2", "r3", "r6", "r10", "r12", "cc", "memory"
	);

This should select the right cache flush functions.

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: OMAP 3530 arch_shutdown "undefined instruction"

[Orjan Friberg]

On 2010-05-27 11:31, Sascha Hauer wrote:
> Seems this does not work on Cortex Processors. Can you try replacing
> this with the following please:
>
> 	asm volatile (
> 		"bl __mmu_cache_flush;"
> 		"bl __mmu_cache_off;"
> 		:
> 		:
> 		: "r0", "r1", "r2", "r3", "r6", "r10", "r12", "cc", "memory"
> 	);

Thanks, this seems to work fine (in the sense that I can load and run a 
second stage bootloader; I haven't tried verifying what happens with the 
I and D cache).

-- 
Orjan Friberg
FlatFrog Laboratories AB

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

Re: OMAP 3530 arch_shutdown "undefined instruction"

[Sascha Hauer]

On Tue, Jun 08, 2010 at 10:02:44AM +0200, Luca Ceresoli wrote:
> Orjan Friberg wrote:> On 2010-05-27 11:31, Sascha Hauer wrote:
> > > Seems this does not work on Cortex Processors. Can you try replacing
> > > this with the following please:
> > >
> > > 	asm volatile (
> > > 		"bl __mmu_cache_flush;"
> > > 		"bl __mmu_cache_off;"
> > > 		:
> > > 		:
> > > 		: "r0", "r1", "r2", "r3", "r6", "r10", "r12", "cc", "memory"
> > > 	);
> > 
> > Thanks, this seems to work fine (in the sense that I can load and run a 
> > second stage bootloader; I haven't tried verifying what happens with the 
> > I and D cache).
> 
> Works for me as well.
> Is it going to be committed?

Ok, done.

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: OMAP 3530 arch_shutdown "undefined instruction"

[Luca Ceresoli]

Orjan Friberg wrote:> On 2010-05-27 11:31, Sascha Hauer wrote:
> > Seems this does not work on Cortex Processors. Can you try replacing
> > this with the following please:
> >
> > 	asm volatile (
> > 		"bl __mmu_cache_flush;"
> > 		"bl __mmu_cache_off;"
> > 		:
> > 		:
> > 		: "r0", "r1", "r2", "r3", "r6", "r10", "r12", "cc", "memory"
> > 	);
> 
> Thanks, this seems to work fine (in the sense that I can load and run a 
> second stage bootloader; I haven't tried verifying what happens with the 
> I and D cache).

Works for me as well.
Is it going to be committed?

Thanks,
Luca



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