MTD: linux-2.4.0-test9 on a PowerPC board.

MTD: linux-2.4.0-test9 on a PowerPC board.

[Brendan J Simon]

My company has a custom PowerPC board with a DOC2000 (8MB) on it.  I am 
trying to get linux-2.4.0-test9 running on it with the MTD drivers.  We 
think the DOC was detected initially but it seems that NFTL media header 
has been blown away.  I think this happened when we tried to write to it 
or format the device.

Trying to debug things we have noticed that some of the data returned 
has the bytes swapped.  It seems the NUM ERASE UNITS is returning 0xFA03 
where as it should be returning 0x03FA.  It looks like an endian 
problem.  Is the DOC code endian neutral ??  Will the endian problems be 
rampant throughout the code or will it be confined to a few places ??

Has anyone used a DOC on a PowerPC processor or other big endian 
architecture ??

Thanks for any help,
Brendan Simon.



To unsubscribe, send "unsubscribe mtd" to majordomo@infradead.org

Re: MTD: linux-2.4.0-test9 on a PowerPC board.

[David Woodhouse]

brendan.simon@bigpond.com said:
>  Trying to debug things we have noticed that some of the data returned
>  has the bytes swapped.  It seems the NUM ERASE UNITS is returning
> 0xFA03  where as it should be returning 0x03FA.  It looks like an
> endian  problem.  Is the DOC code endian neutral ??  Will the endian
> problems be  rampant throughout the code or will it be confined to a
> few places ??

We don't actually seem to use NumEraseUnits, although I have absolutely no 
recollection _why_ not. But that would explain why we never bother to 
byteswap it. The code is intended to be endian-aware, although 
I've personally only tested it on LE machines, so I wouldn't swear that 
I've caught all cases. 

> Has anyone used a DOC on a PowerPC processor or other big endian
> architecture ?? 

It's been known, but I can't remember whether the person who asked me about 
it ever admitted to it in public. 

--
dwmw2




To unsubscribe, send "unsubscribe mtd" to majordomo@infradead.org