Hi Tim, > About your sector-dumping project: surely there must be a better way > than using the parallel port? ;-)
Well, there are worse ways. :-) The serial port is a lot slower, only 19,200 baud on some of these machines; their Standard Parallel Port is much quicker. Some of the drives are PATA IDE, so in theory I could move them to a modern PC and dd(1) or gddrescue, but given others are ST506 interface I'd have to find some other way for those anyway. And not disturbing a drive that's been in place for some decades has its appeal; I'm grateful if it spins up and accesses, let along survives a move to another machine. On these old machines, it doesn't matter that the parallel port is SPP and more or less uni-directional since under some of the operating systems, e.g. AIX 3.2.5 on a POWER CPU, it would be practically impossible to find out how to get enough access to the port to do anything more complex. The one thing I probably can do is get eight-bit bytes sent through unharmed. If I wrap the sectors with a frame, address info, CRC, possibly FEC too, depending on what I can cook up in BBC BASIC/ARM, then the Linux receiving end can work hard to come up with the image. If there are bits that need resending, Linux can tell me the addresses and I'll run the transfer again for just those areas, IOW I'm the other direction of the laggy protocol. There's even the option of having a microcontroller be the "printer" and just have it dump all data to a MultiMediaCard for when a PC can't be nearby. Since there's no protocol to speak of, other than normal SPP, the MCU doesn't have much to do, and wouldn't have knowledge of frames, CRC, etc., it just records the "print-out". Anyway, that's my idea so far. It seems the least intrusive to the old, fragile, systems. Cheers, Ralph. -- Next meeting: Blandford Forum, Tuesday 2010-09-07 20:00 http://dorset.lug.org.uk/ -- Meets, Mailing list, IRC, LinkedIn, ... How to Report Bugs Effectively: http://bit.ly/4sACa
