[sebhc] list archive and a specific question

Jim Battle frustum at pacbell.net
Wed Jul 6 20:57:15 CDT 2005

Dwight K. Elvey wrote:

> Hi Dave
>  If you are seeing errors, you have some type of
> issue that needs to be fixed. Are you using the
> checksum directly from the sectors on the disk
> ( may be a CRC ) or are you generating it from
> a RAM buffer? Like I said, I've never seen a problem
> without a cause. I have designed embedded systems
> that do CRC checking on packets of data that
> run continuously, 24/7. Any logged errors required
> repairs. I've, also, done over a 100 disks on my Poly8813 without
> a single missmatch.
>  In any case, you should try using a CRC rather than
> a check sum. These can be used to find the error location
> for small burst of bit errors. I would suspect your RAM
> if the errors were always in the same location of your
> buffer. As a sanity check, you might try turning on
> the parity as well.
>  Of course, if you are using Windows, such errors are normal.
> Dwight

Dwight, I have seen errors that weren't fixable.  Say I'm transferring a lot of 
data from my Sol to my PC, and there is no flow control.  If I run at 9600 baud 
and windows hiccups for some reason (it decides to kick off a virusscan or 
something and that one task hogs resources for a couple of seconds), characters 
get dropped.  It is stupid that a 2 GHz machine can't keep up with a 9600 baud 
channel, but it happens.

(I know, don't run windows, but for all of its flaws, it has some nice 
properties too that, for me, make up for the heartburn)

Delivered by the SEBHC Mailing List

More information about the Sebhc mailing list