home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!olivea!decwrl!deccrl!news.crl.dec.com!rdg.dec.com!decvax.dec.com!decvax.DEC.COM!jag
- From: jag@decvax.DEC.COM (John A. Gallant UEG)
- Newsgroups: comp.unix.ultrix
- Subject: Re: SCSI/CAM problems
- Message-ID: <1992Sep8.160410.17336@decvax.dec.com>
- Date: 8 Sep 92 16:04:10 GMT
- References: <ROBM.92Sep4150116@ataraxia.Berkeley.EDU>
- Sender: usenet@decvax.dec.com (Usenet News System)
- Reply-To: jag@zk3.dec.com
- Organization: OSF Engineering, Digital Equipment Corp.
- Lines: 188
- Nntp-Posting-Host: witsend.zk3.dec.com
-
- In article <ROBM.92Sep4150116@ataraxia.Berkeley.EDU>, robm@ataraxia.Berkeley.EDU (Rob McNicholas) writes:
-
- >I've recently installed the Open SCSI/CAM software on my DECstation,
- [lines deleted]
- >System: DEC5000/125, Ultrix 4.2C
-
- Is this the CAM kit out of the "product" box or a field test version ?
-
- > scsi bus#0: one RZ57
- > one RRD42
- > two EXB-8200 8mm drives Mountain Filesafe 2100D (Rev 252X)
- >
- > scsi bus#1: two RZ57s
- > one TZK50
- > one EXB-8200 (same as above)
-
- >This particular error given in the example was caused by using the
- >public domain "copytape" program, by David S. Hayes. The program
- >aborts in a read() with an ENOSPC error (the actual error output is:
- >"copytape: can't read input: I/O error"). This is consistenly
- >reproducible from any tape drive. Now we can't make duplicates of our
- >backups anymore :-( I saw a note in the CAM release notes that
- >mentioned a problem with model 100 machines returning the wrong number
- >of bytes on odd-byte tranfers. Could this be related? The code looks
- >like it's doing the right thing (checking the return value of the
- >read() call for the actual number of bytes tranferred).
-
- With the backup numbers 126b is an even value. The bug in the 100/25/240
- machine is that an extra byte is transfered in but the correct number of
- bytes is returned from the call.
-
- >I should point out that we can still read this tape "dd ibs=126b
- >if=/dev/nrmtNh | zcat | restore if -".
-
- Do you get similar CAM errors with the dd command ?
-
- >Anyone having similar experiences? Anyone have any suggestions?
- >
- >If I decide to back out on this "upgrade", will I have to reinstall
- >the 4.2A kernel config files? (I'd hate to have to do that since I've
- >since installed a bunch of patches.)
-
- During the installation of the CAMBIN42* subsets, the non-CAM only
- files are saved as *.nocam. You would only have to compare/diff these
- files. Off the top of my head I think that there are only about 6 or 7.
-
- >Thanks for any help, and apologies for the length of this message.
-
- I for one want to get this to work. We have spent a lot of engineering
- time and effort to make this new subsystem.
-
- How often do the error occur, on initial open ?, some where in the
- middle ?, or usually at the end ?
-
- What is the boot string text reported by the tape driver. If your
- tape drive is not one of the DEC supported devices, ie has the DEC name in
- it, there may not be an entry for it in the /usr/sys/data/cam_data.c file.
- The default TZxx entry is used, for devices that fall through the table
- lookup.
-
- What /dev/*rm* name did you use for the device. Is it possible that
- the tape may have been written in a block mode and the reads are trying
- variable ?
-
- You may want to make sure that you have an entry in the cam_data.c file
- for your device, you will want to make it look "a lot" like the TZK09 :-).
-
- >uerf reports these errors as:
-
- [Not worth looking at.]
-
- >cam_report gives this more verbose, but just as incomprehensible (to
- >me) output:
-
- But this is where all the "meat" is. :-)
-
- > Sequence number of error: 8
- > Time of error entry: Fri Sep 4 14:01:30 1992
-
- > SCSI device class: TAPE
- > Bus Number: 1
- > Target number: 1 Lun Number: 0
-
- This error is comming from the tape on the PMAZ-AA card, the second SCSI
- bus, with the lone Exabyte.
-
- > Routine Name: ctape_iodone
- > Routine Error Message: Hard Error Detected
- > Device Name: EXABYTE EXB-8200
- > Routine Message: Active CCB at time of error
- > Routine Message: CCB request completed with an error
-
- A "CCB" is the command/SCSI I/O Data structure used to communicate
- between the device drivers and the SCSI Interface Module, (SIM). The
- CAM tape driver has received an error from the tape drive that has
- terminated this I/O request.
-
- > CAM subsystem status ................. cam_status: 0xc4 (CCB request completed w
- > ith an error)
-
- The CCB has completed, and the error information has been returned as
- part of the CAM autosense functionality.
-
- > Data transfer length .............. cam_dxfer_len: 262144 (0x40000)
-
- This request has a transfer length of 256 Kbytes(!) this is a little more
- than the 126b (64512 bytes) indicated in your dump parameters.
-
- > SCSI device status code ......... cam_scsi_status: 0x2 (SCSI_STAT_CHECK_CONDITION)
-
- The SCSI status code, here signaling a check condition (a SCSI error),
- is returned from the device itself. The CAM subsystem will issue a SCSI
- REQUEST SENSE command to get the error information from the device.
-
- > Transfer residual length .............. cam_resid: 262144
-
- There was NO data transfer with this command.
-
- > Command descriptor block bytes ....... cam_cdb_io: 8 0 4 0 0 0
-
- Looking at the SCSI spec under the tape section, the command is a
- simple read (0x8) of 0x40000 bytes. The Fixed bit is not set, the tape
- driver is using the tape in variable length mode.
-
- The rest of the log report is an attempt to decode the error
- information that the device returned following it's report of an
- error. Only the standard error bytes are decoded.
-
- > ############### Entry End ###############
- >
- > Routine Message: Error, exception, or abnormal condition
- > Routine Message: Illegal request or CDB parameter
- >
- > Dumping Request Sense Data at 0x1000e278:
- > Error code ...................... : 0x0
- > Error class ..................... : 0x7
- > Information fields valid ........ : 0
- > Segment number .................. : 0x0
- > Sense Key ....................... : 0x5 (Illegal request or CDB parameter)
- > Illegal length indicator ........ : 0
- > End of medium ................... : 1
- > Tape file mark detected ......... : 0
- > Information byte 3 .............. : 0x0
- > Information byte 2 .............. : 0x0
- > Information byte 1 .............. : 0x0
- > Information byte 0 .............. : 0x0
- > Additional sense length ......... : 0x12
- > Command information byte 3 ...... : 0x0
- > Command information byte 2 ...... : 0x0
- > Command information byte 1 ...... : 0x0
- > Command information byte 0 ...... : 0x0
- > Additional sense code ........... : 0x0
- > Additional sense qualifier ...... : 0x0
- > Sense Code/Qualifier Message .... : No additional sense information
- > Field replaceable unit code ..... : 0x0
- > Additional sense bytes .......... : 0x0 0x0 0x0 0x0 0x0 0x1 0x20 0x0 0x0 0x22 0xfc
-
- I had to dig out our 8500 manual to try and understand the rest of
- these error bytes. Looking at the std error bytes, EOM - End of Media,
- has been reached. Sense Key 0x5 is for illegal request. From the
- manual - "an illegal parameter in the CDB" ... "or that the EXB-8500 is
- in the wrong mode to execute the command". According to the manual,
- Rev 001, there should be more error codes in the "Additional sense
- code" and "Additional sense qualifier" but there is none.
-
- The rest of the bytes are manually decoded:
- Additional sense bytes .......... :
- 15 16 17 18 19 20 21 22 23 24 25
- 0x0 0x0 0x0 0x0 0x0 0x1 0x20 0x0 0x0 0x22 0xfc
-
- Bytes 14:15 Are Reserved - We report 14 as the FRU code and 15 is 0x0
- Bytes 16:18 Are Read/Write data error counters : 0x0 0x0 0x0
- Bytes 19:21 Are Unit Sense
- 20:bit 0 : FE - Formatter Error
- 21:bit 5 : RSVD - Reserved
- Bytes 23:25 Is the Remaining tape in 1024 byte physical blocks
-
- Unfortunately there are no more bytes, I would to have liked to see
- byte 28, the fault symptom code, but the EXB-8500 did not return it.
-
- --
- John A. Gallant jag@zk3.dec.com
- Software Engineer - OSF Engineering Group
- Digital Equipment Corp. (603) 881-2472
-
- In the common people there is no wisdom, no penetration, no
- power of judgment.
- Marcus Cicero
-