Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!rutgers!mit-eddie!uw-beaver!Teknowledge.COM!polya!Neon.Stanford.EDU!kaufman From: kaufman@Neon.Stanford.EDU (Marc T. Kaufman) Newsgroups: comp.sys.mac Subject: Re: 2 Macs on one HD? Message-ID: <12220@polya.Stanford.EDU> Date: 6 Oct 89 02:37:24 GMT References: <5513@merlin.usc.edu> <125867@sun.Eng.Sun.COM> Sender: news@polya.Stanford.EDU Reply-To: kaufman@Neon.Stanford.EDU (Marc T. Kaufman) Distribution: comp Organization: Stanford University, Computer Science Dept. Lines: 21 In article <125867@sun.Eng.Sun.COM> cramer@sun.com (Sam Cramer) writes: >Don't Macs reserve SCSI ID 7 for themselves? If so, it seems like having >two SCSI devices on the bus with the same ID is a bad idea. On the other >hand, maybe it all works out because Macs do polled I/O, so one Mac grabs >the bus until the SCSI command is completed. Polling would work if the SCSIGet/SCSISelect routines would perform a Bus Arbitration phase before blasting stuff out on the Bus. Unfortunately, under the current OS (6.x), they do not. Also, I would be VERY careful if a volume was simultaneously mounted on two systems. The Volume Control Blocks and allocation maps are almost certainly guaranteed to get out of Sync, since there is no way for one Mac to tell the other that they have been changed. One possible problem, EVEN WITH ARBITRATION, is that if both Macs tried to select simultaneously, they could not tell who won because they both would be setting the same selection bit (bit 7). In general, then, this is NOT a good idea. Keep your copy of the SUM utilities handy. Marc Kaufman (kaufman@Neon.stanford.edu)