Re: uboot & sheevaplug (dockstar)

[Jean-Yves Moulin <jym%baaz.fr@localhost>]

Hi Takashi,


On 24 Dec 2013, at 3:35 , KIYOHARA Takashi <kiyohara%kk.iij4u.or.jp@localhost> 
wrote:

> First of all, Has not your SheevaPlug broken?
> Is the OS of Original(Linux?) working correctly?
> Or Are you use u-boot original?

Yes, I'm able to boot the original linux. And I'm able to boot a netbsd
kernel with sources before 17-Jun-2013. 

I've tried to get back in source history:
* Source from 17-Jun-2013 04:00 UTC -> Kernel boot fine.
* Source from 17-Jun-2013 05:00 UTC -> I'm unable to boot.

After some search, this commit seems to breaks my kernel:

http://cvsweb.netbsd.org/bsdweb.cgi/src/sys/arch/arm/arm32/locore.S.diff?r1=1.31&r2=1.32&f=h

I'v two different issue with this diff:

* If I remove only the line 62 (sub r2,r2,r1) , the kernel panic:
panic: arm32_kernel_vm_init: Failed to allocate or align the kernel page 
directory

* If I change the other lines (lines 71 to 75), the kernel prints nothing.



> Can you execute the following command after tftp or fatload?
>  cmp.b 0x2000040 0x8000 <hex-ed size for your netbsd.bin>

(I always load the kernel with fatload)

I don't know if this is already needed, but just in case:

A good kernel:

u-boot>> cmp.b 0x2000040 0x8000 0x6beac4
byte at 0x02000106 (0x00) != byte at 0x000080c6 (0x01)
Total of 198 bytes were the same

A bad kernel:

u-boot>> cmp.b 0x2000040 0x8000 0x6beac4
byte at 0x020000fc (0x04) != byte at 0x000080bc (0xbc)
Total of 188 bytes were the same

 
> Can you moreover run a kernel with the diff which I prepare?
> Probably I will make the diff simply accessed to a COM port more
> early.  However, I cannot touch to sheevaplug or other Marvell SoC
> till a weekend.  It may contain a bug.


No problem. I've some dockstar with serial, ready for debug :-)


Thank you very much.

best,
jym