ts@cup.portal.com (Tim W Smith) (02/24/89)
Here's another puzzle. Suppose you have a computer with a SCSI disk attached to that computer. You observe the following behaviour: You attempt to find the SCSI id of the disk by doing INQUIRY commands on id's 0 to 6. The disk responds to ALL of these. You reset the SCSI bus and do an inquiry on id 0. You get a unit attention condition. You do another inquiry and get the correct inquiry data. You now do an inquiry to id 1. You get a unit attention condition. What's going on? Tim Smith ps: anyone else have any weird things like this that would make interesting puzzles?
alexis@ccnysci.UUCP (Alexis Rosen) (02/26/89)
In article <15030@cup.portal.com> ts@cup.portal.com (Tim W Smith) writes: >[describes very weird disk behavior] >What's going on? Gee. Maybe your disk isn't working? :-) >ps: anyone else have any weird things like this that would make interesting >puzzles? Just out of curiosity, these aren't just hypothetical questions, are they? What are you doing that causes all this weirdness, anyway? This has got to be the most obscure question I've seen on this group in ages... Alexis Rosen alexis@ccnysci.uucp
ts@cup.portal.com (Tim W Smith) (02/27/89)
Here is the other puzzle I posted a few days ago: < Here's another puzzle. Suppose you have a computer with a SCSI disk < attached to that computer. You observe the following behaviour: < < You attempt to find the SCSI id of the disk by doing < INQUIRY commands on id's 0 to 6. The disk responds < to ALL of these. < < You reset the SCSI bus and do an inquiry on id 0. < You get a unit attention condition. You do another < inquiry and get the correct inquiry data. < < You now do an inquiry to id 1. You get a unit attention < condition. < < What's going on? The problem is that disk drive and the computer are both set to the same SCSI id. The way an initiator selects a target on SCSI goes something like this: Initiator grabs the bus Initiator turns on bits X and Y of the data bus, where X is the initiator id and Y is the target id ( this is why there is a limit of 8 devices: each one needs a bit on the data lines for selection ). Initiator also turns on a line that says "Hey you! I'm selecting you!" Target device sees that initiator is selecting someone. It looks at the data bus to see if it is the device being selected. If it is, it also looks to see what initiator is doing the selection. Well, suppose both devices are set to id 7. When the initiator tries to select device X, it will turn on bits 7 and X. The target device will look at the bus and see bit 7 on, so it thinks it is being selected. It will see bit X on and thing that the initiator has id X. Thus, from the point of view of the initiator, the target appears to respond to all id's that are not 7. This explains the first part of the puzzle. To understand the second part of the puzzle, you need to be aware that SCSI devices keep seperate status for each initiator. Now we can explain the second part. First, you do a reset of the SCSI bus. The target goes into a unit attention condition. The inquiry on id 0 gets a unit attention. The target thinks that the initiator had id 0, so it clears unit attention for initiator 0. The next inquiry on id 0 works fine, because the first cleared the unit attention condition. When you do the unit attention on id 1, the target thinks that the command is being issued by initiator 1, which has not yet done any command since the reset, so it gets a unit attention. Tim Smith
joseph@cooper.cooper.EDU (Joe Giannuzzi) (03/01/89)
in article <15030@cup.portal.com>, ts@cup.portal.com (Tim W Smith) says: > > You attempt to find the SCSI id of the disk by doing > INQUIRY commands on id's 0 to 6. The disk responds > to ALL of these. > I believe this will happen if the SCSI id of the disk is set to 7. (For instance, Jasmine drives have a switch for 0-7) I know that when the Jasmine drive id was accidentally set to 7, 7 identical and accessible copies (id's 0-6) of the drive appeared on the desktop. Joe Gunoz joseph@cooper.cooper.edu