phiggins at transzap.com
Fri Jun 2 18:47:12 EDT 2000
I'm going on vacation for a week. I'll be taking my books on
Reed-Solomon coding, but I won't be doing any hacking for a while. I'm
attaching the current state of my work (it's completely undocumented,
unfortunately) in case anyone wants to look.
Basically, I've collected data files by dumping out the buffer and
syndroms on every read and write. These are in ASCII form in the files
ecc.*. ecc.err contains a known failure, and the rest are ok. The ones
named ecc.ok* are from reads, and ecc.wr* are from writes. The files
named data.* are binary versions of the same files. The files syn.* are
the decoded syndromes from these data files. Nothing is special about the
decoding, it just turns them from their odd six-byte interleaved format
into something readable.
The files polys and polynomials contain the 60 primitive polynomials over
GF(2^10). They are in numeric format in polys, and polynomials contains
just the exponents of terms whose coefficient is 1. For example,
10 3 0 really means x^10 + x^3 + 1
Most of the C programs are just for converting data formats. The main one
of interest is gf.c, which actually attempts to compute the syndromes, but
I haven't been able to get a match with what the hardware returns yet.
I'm not sure if this will be of any use to anyone, but have a look if
you're interested. I wasn't able to find any good references on the web
for Reed-Solomon, but there were several good books at the library on
it. I really wouldn't recommend diving into this without a reference.
-------------- next part --------------
A non-text attachment was scrubbed...
Size: 4353 bytes
Url : http://lists.infradead.org/pipermail/linux-mtd/attachments/20000602/77ea93b8/attachment.gz
More information about the linux-mtd