home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!cs.utexas.edu!uwm.edu!linac!att!cbfsb!cbnewsc!att-out!rutgers!news.cs.indiana.edu!lynx!zia.aoc.nrao.edu!rmilner
- From: rmilner@zia.aoc.nrao.edu (Ruth Milner)
- Newsgroups: comp.sys.sun.hardware
- Subject: SCSI changes in 4.1.2
- Message-ID: <1992Aug31.221224.2005@zia.aoc.nrao.edu>
- Date: 31 Aug 92 22:12:24 GMT
- Organization: National Radio Astronomy Observatory, Socorro NM
- Lines: 46
-
- There's been a fair amount of discussion here lately about something in the
- SCSI code at SunOS 4.1.2 which results in tapes like Exabyte 8500's and DATs
- producing the "esp0: target n now synchronous ..." messages under some
- circumstances (renegotiating sync SCSI).
-
- I was wondering why these messages occur. In looking through first the list of
- SCSI bugs fixed in 4.1.2, and then the SunSolve bugslist, I find the following
- suspicious-looking fix:
-
- Bug Id: 1070535
- Category: scsa
- Subcategory: host_adapter
- Release summary: 4.1.2beta1
- Synopsis: esp: sync xfer rate is not renegotiated after a check condition
- Integrated in releases: 4.1.2beta1
- Patch id:
- Summary:
- after a check condition the sync xfer rate should be renegotiated.
- The esp driver does not correctly check for the status != 0.
- Consequently, the system may hang until a timeout resets the bus.
-
- Sounds to me like they put the renegotiation in regardless of what error the
- status actually indicates, i.e. after *every* check condition. But surely, in
- this particular case (where the "error" is that you didn't know how many bytes
- to expect in a block, so you ask for a whole bunch and just take what you get,
- and then bytes-requested > bytes-returned), you don't have a condition that
- would hang the bus, do you? So why do the renegotiation? Is it not possible to
- detect what the check condition really was?
-
- This is an especially interesting question since the problem does not occur
- with Exabyte 8200's, even though they also report they are using sync SCSI.
- Presumably they return a different status code, or the driver handles it
- differently, under these conditions. (?)
-
- Is there anyone out there who knows enough about the code to verify that this
- is what's happening?
-
- Also, I know this has been reported to Sun as a bug, but does anyone know what
- the status of getting it *fixed* is? Currently the only "patch" that I'm aware
- of is a new esp.o that does nothing but suppress the message that shows up -
- for every device on the bus. This is not a real fix, since the renegotiation
- is still happening, and now I can't tell if my disks might be doing something
- they shouldn't.
- --
- Ruth Milner NRAO/VLA Socorro NM
- Computing Division Head rmilner@zia.aoc.nrao.edu
-