Path: news.uiowa.edu!chi-news.cic.net!news.uoregon.edu!cie-2.uoregon.edu!nparker From: nparker@cie-2.uoregon.edu (Neil Parker) Newsgroups: comp.sys.apple2 Subject: Re: Quantum Fireball with Ramfast Date: 8 Nov 1995 06:30:23 GMT Organization: University of Oregon Campus Information Exchange Lines: 73 Message-ID: <47pipv$f0f@pith.uoregon.edu> References: <1995Nov5.201009.6826@freenet.victoria.bc.ca> NNTP-Posting-Host: cie-2.uoregon.edu In article dempson@atlantis.actrix.gen.nz (David Empson) writes: >In article <1995Nov5.201009.6826@freenet.victoria.bc.ca>, >Marc Sira wrote: >> In our last episode, dempson@atlantis.actrix.gen.nz (David Empson) said: >> >> >"Reselection" is a feature of both SCSI-1 and SCSI-2. It is not >> >supported by the RamFast, since the RamFast doesn't support the >> >arbitration phase, and has no SCSI ID of its own. >> > >> >If reselection is enabled, the drive will disconnect from the host >> >while a command is being actioned. When the command is complete (e.g. >> >a READ command is ready to transfer data) the drive will "reselect" >> >the host (i.e. wait for the bus to be free and try to connect to the >> >host, which involves sending the SCSI ID of the host) and transfer the >> >data. >> >> Interesting. Is it possible, then, to convince the drive not to do this? > >It should be. There doesn't appear to be a jumper to control this, so >I expect it would be necessary to send an appropriate SCSI "MODE >SELECT" command. I'd have to do some research into the relevant >standards to find out how this is supposed to be done. There is no jumper to control reselection, because the choice of whether or not to use it is always entirely the decision of the initiator (the device that issues the SCSI command, i.e. your SCSI card). The decision path works something like this: If the initiator doesn't do arbitration, the target (the receiver of the command, i.e. your disk drive) will not do reselection. If the initiator *does* use arbitration, then the target waits for an IDENTIFY message (the IDENTIFY message is a single byte that the intiator sends to the target to tell it what logical unit within the target is to be selected for I/O). Bit 6 of the IDENTIFY message is the reselection enable bit--if the initiator sends an IDENTIFY message with bit 6 set, and if target is capable of reselection, then reselection will occur. Since the RamFAST never does arbitration, no device connected to a RamFAST will ever try to do reselection. Note that the option to skip arbitration is not allowed in the SCSI-2 specs. All SCSI-2 initiators *must* do arbitration. However, a SCSI-2 *target* may optionally support non-arbitrating initiators, for compatibility with SCSI-1 initiators. The same thing is true of the option to support the SCSI-1 "single initiator option" (which, I suspect, is how the RamFAST manages not to have a SCSI ID)...a SCSI-2 initiator may never use the "single intiator option," but a SCSI-2 target may optionally support it for compatibility reasons. This could, in theory, cause compatibility problems between some newer SCSI devices and the RamFAST. If the device chooses not to support non-arbitrating hosts or the single initiator option, it will probably have trouble talking to the RamFAST. Disclaimer: The above discussion is my own interpretation of the SCSI draft docs. I don't have copies of the final SCSI standards. I doubt much changed between the last drafts and the published standards, but you never know... - Neil Parker P.S. The last drafts of both SCSI-1 and SCSI-2 can be FTPed from ftp.symbios.com. The last SCSI-2 draft is available on the Web at http://abekas.com:8080/SCSI2/. -- Neil Parker | No cute quote, no cute ASCII art, no cute nparker@cie-2.uoregon.edu | disclaimer, no deposit, no return.... nparker@cie.uoregon.edu | (This space intentionally left blank: http://cie-2.uoregon.edu/~nparker | )