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

Re: node types etc.



On Tue, 16 Jan 2001, David Woodhouse wrote:
> >  TCP checksums avoid that [the problems, not the error checking],
> > reuse code, and are heavily optimized on most architectures.  I'd vote
> > for them. 
> 
> CRCs may well be cheaper on 32-bit machines than IP checksums, even if the 
> IP checksum routines are optimised.

How can something be cheaper than a single addition ? :) 

> I think the IP checksum routines are optimised for the copy_and_checksum 
> case, which I'm not sure is what we'll be doing in JFFS.

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

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 ?

The reason why we put the checksum there in the beginning, was to be able
to scan the flash and find complete nodes (i.e. the checksum should catch
the case where the system crashed during writing of the node). Not to find
actual errors in the flash. We figured a simple addition checksum would be
enough to catch the case of non-finished write, but we never really
thought a lot about that particular case :) 

-BW


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