[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: node types etc.




bjorn.wesen@xxxxxxx.com said:
>  How can something be cheaper than a single addition ? :) 

TCP checksums aren't just a simple addition. You have to fold back into 16 
bits. I'm told that CRC can be cheaper on 32-bit machines. A straight 
32-bit checksum, on the other hand, can quite possibly be cheaper than 
either.

>  Both checksum and checksum_and_copy are optimized (see csum_partial
> and csum_partial_copy). 

OK, cool. I have no preference here. I don't claim to know much about it. 

bjorn.wesen@xxxxxxx.com said:
>  Why do we need good checksumming, isn't the ECC stuff on the
> NAND-flashes supposed to take care of the most common bit-faults
> anyway ? 

ECC on NAND-flashes is currently a bit of a hack - it's done by the driver 
on a per-page basis, rather than per-node. I said at the time that it would 
be better to do it per-node, but promptly forgot about it. I think it would 
be useful to allow for proper ECC in the structure of JFFS itself, rather 
than expecting the hardware to bolt it on afterwards.

But I think I'll leave that for later - we don't want it on NOR flashes 
because it'll waste space, so we want different types of dirent and data 
node for it. I think.

--
dwmw2



To unsubscribe from this list: send the line "unsubscribe jffs-dev" in
the body of a message to majordomo@xxxxxxx.com