memory problems on ppc 8260 based hardware

memory problems on ppc 8260 based hardware

[13787783324]

I'm running into  memory problems on ppc 8260 based hardware. The kernel is based
on linux-2.4.4-2002-10-06.tar.bz2 from www.denx.de . (use uboot as bootloader.)

I use some debug tools and find when paging_init( ) function finished
(start_kernel --setup_arch--paging_init), some codes were destroyed such as
sort_exception_table( ) (the next line in setup_arch( ), so the kernel hang) ,
some codes were protected well such as memset( ).
(using LED in parallel port(adding some codes in kernel), I got the same
results,the kernel hang after paging_init( ))

what's the reason? Any advise will be appreciated.


best regards.


liu


** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

RE: memory problems on ppc 8260 based hardware

[Rune Torgersen]

Be aware that there is some timing problems on the SDRAM interface on
the 826x CPUs
(Look at motorolas site)
This sound very much like the problems I had before I fixed the hardware
on an inhouse board.


> -----Original Message-----
> From: owner-linuxppc-embedded@lists.linuxppc.org
> [mailto:owner-linuxppc-embedded@lists.linuxppc.org]
> Sent: Tuesday, May 04, 2004 10:11
> To: linuxppc-embedded@lists.linuxppc.org
> Subject: memory problems on ppc 8260 based hardware
>
>
>
> I'm running into  memory problems on ppc 8260 based hardware.
> The kernel is based on linux-2.4.4-2002-10-06.tar.bz2 from
www.denx.de . (use uboot as bootloader.)

I use some debug tools and find when paging_init( ) function finished
(start_kernel --setup_arch--paging_init), some codes were destroyed such
as sort_exception_table( ) (the next line in setup_arch( ), so the
kernel hang) , some codes were protected well such as memset( ). (using
LED in parallel port(adding some codes in kernel), I got the same
results,the kernel hang after paging_init( ))

what's the reason? Any advise will be appreciated.


best regards.


liu


** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/

RE: memory problems on ppc 8260 based hardware

[Rune Torgersen]

See
http://e-www.motorola.com/files/32bit/doc/white_paper/MPC826XSDRAMWP.pdf

Basically the clocks going to the SDRAM has to be leading the clock
going to the 8260 by about 1 ns.

This will only show up when doing burst accesses (like linux does all
the time) and no while doing standard memory tests.


> -----Original Message-----
> From: liuming [mailto:13787783324@hnmcc.com]
> Sent: Tuesday, May 04, 2004 13:38
> To: Rune Torgersen
> Subject: Re: memory problems on ppc 8260 based hardware
>
>
> Hi,
>       Mr Torgersen, thanks a lot .
>       I have read the Errata of 8260.  SDRAM is 32bit    (60x
> bus), is it the reason?  could you say something about you
> fixed the hardware on an inhouse board?
>
> Best regards!
>  liu

** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/