Problem: SDRAM configuration and refresh

Problem: SDRAM configuration and refresh

[adharankar]

Greetings,

I am trying to getup a Scout based board with
64MB SDRAM, and could use any recommendations
to fix this problem:

If I set the refresh-enable bit in the PSDMR,
the chipselect for flash and other devices
vanishes: consequently, this startup code
ends up executing junk. At this point if I try
to look at the flash from the BDI2000, the
flash cannot be seen.



On the other hand, if I initialize the SDRAM
without setting this refresh-enable bit, ppcboot
(the code using it) comes up and am able to run
commands etc. However, the problem now is that
without the SDRAM refresh we run into memory
corruption problem (mtest will fail) and a system
hang.

Its interesting to note that as long as memory
is accessed frequently (say via mtest), the
corruption is not see at that address. Also, an
idle system almost hangs in a very short time,
whereas an active one with mtest will continue on.



Best regards,
-Arun.

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

Re: Problem: SDRAM configuration and refresh

[Arun Dharankar]

Greetings!

Thanks to anyone who might have attempted to
help out! The problem turned out to by too low
SDRAM refresh interval.

Best regards,
-Arun.


On Tuesday 11 June 2002 03:22 pm, adharankar@attbi.com wrote:
> Greetings,
>
> I am trying to getup a Scout based board with
> 64MB SDRAM, and could use any recommendations
> to fix this problem:
>
> If I set the refresh-enable bit in the PSDMR,
> the chipselect for flash and other devices
> vanishes: consequently, this startup code
> ends up executing junk. At this point if I try
> to look at the flash from the BDI2000, the
> flash cannot be seen.
>
>
>
> On the other hand, if I initialize the SDRAM
> without setting this refresh-enable bit, ppcboot
> (the code using it) comes up and am able to run
> commands etc. However, the problem now is that
> without the SDRAM refresh we run into memory
> corruption problem (mtest will fail) and a system
> hang.
>
> Its interesting to note that as long as memory
> is accessed frequently (say via mtest), the
> corruption is not see at that address. Also, an
> idle system almost hangs in a very short time,
> whereas an active one with mtest will continue on.
>
>
>
> Best regards,
> -Arun.
>


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