Yep. (Don't take offence).
> > > > showed that say only one in a million bits went wrong. In that
^^^
Yeah. Come to reread it the "measurements" part may make it look as
scientific data. But You didn't think that I could tell you the error
rate on YOUR SCSI cable in one simple number without sofisticated
equipment near the thing did you?
> > Incorrectly terminated SCSI busses or too long a SCSI bus lead to
> > erratic behaviour. Same (cable too long, or improper termination)
> > goes for IDE. (From the 16Mb/sec mode upwards, the motherboard side
> > of the cable has to be terminated.)
> >
> > > My opinion about IDE BUS is that it is not a suitable IO bus for mass
> > > storage devices, but looks like some extension of some system bus, since
> > it is.
> > > it is neither terminated, nor uses differential signals.
> > > If I enjoyed driving trabants with race car engine, I would probably
> > > use IDE Ultra 33 devices.
> >
> > Gerard, you do need to realize that a 32bit CRC detects all single
> > byte, double byte, and triple byte errors. It detects (2^-32)-1 out of
> > 2^32 of all quad byte and longer errors. Upto bursts of 64 bytes of
> > random data, this has a better chance of catching real errors than a
> > per-byte parity.
>
> A 32 bit value is only able to tag 2^32 differents bit strings.
> So, IMHO, the efficiency of a CRC depends on its algorithm with regard
> to the typology of error patterns that are probable to happen.
>
> One who wants to convince that this CRC is superior to something else
> must explained why and not just claim.
>
> What you write above (e.g. a 32bit CRC detects _all_ ...) can only be
> _false_ in my opinion.
> Note that I think that this CRC is very probably efficient since error
> patterns are not quite random.
> I just want to get explanations about this CRC capabilities and did'nt
> get any for the moment ...
Mark just "admitted" that it is only a 16 bit CRC. So lets start using
this number.
CRCs are based on a Linear Feedback Shift Regster (LFSR). With just a
few tricks you can design one that has a "cycle time" that achieves
the maximum theoretically possible (65536 for a 16 bit LFSR). If you
base a CRC on this, all perturbations shorter than the shift register
length have to end up with a different value in the register, as
otherwise the LFSR would be back to the beginning in less than the
theoretically max.
> > You're right that an unacceptable overhead would be incurred if
> > software would need to calculate the CRC. As to the speed of
> > calculating a CRC against that of parity, both can be implemented in
> > hardware with just a few xor gates.
>
> I donnot have any doc about this CRC.
> Could you let me know what algorithm or polynom is used ?
> > Gerard, may I ask you a question on YOUR field of expertise? I got a
>
> Are you aware that you are using a technology that may encounter a not
> detected error for _each_ byte transmitted with a probability of 1/2?
> :) :) :) :)
Yep, I'm using BOTH IDE and SCSI. I must be nuts :-)
> > With this as the hardware situtation, my machine once locked. I would
> > expect SERIOUS failures when my system would be using SCSI as the
> > root-device, but as it is, just a few "large" storage partitions were
> > mounted on the SCSI disks. With bad termination, I'd expect parity
> > errors, timeouts, but not a complete lockup.
>
> Agreed. Normally, even if the NCR chip is locked, the SCSI middle driver
> should ask the driver to reset the controller when a scsi abort request
> times out and all should restart. Something went wrong, perhaps at SCSI
> drivers level (including low-level one), perhaps in some other part of
> the kernel. Error recovery is very hard to test and it is IMO the
> weak point of the current Linux SCSI stack.
Well, :-) May I suggest you take off the termination of your bus :-)
you might get a nice chance to test error recovery within a few
minutes :-) (*)
Roger.
(*) Despite the smileys, I'm half serious.
-- ** R.E.Wolff@BitWizard.nl ** +31-15-2137555 ** http://www.BitWizard.nl/ ** Florida -- A 39 year old construction worker woke up this morning when a 109-car freight train drove over him. According to the police the man was drunk. The man himself claims he slipped while walking the dog. 080897