stevens@hsi.UUCP (Richard Stevens) (07/23/86)
We just brought up 4.3 BSD on an 11-785 with a UDA50 with one RA80 (drive 0) and two RA81s (drives 1 & 2). When we try and make the ra0h partition on the RA80, newfs tries to mkfs with a size of 111,202 sectors. Mkfs very quickly bombs with the message endcd=200, stat=16001 Comments in the driver indicate that this is probably a response to an invalid command, with some mention of invalid block numbers. We backed off on the size by 1 cylinder and did the mkfs by hand, and the error still occured. (We've been using this complete partition under 4.2 for over 2 years without any problems.) We finally backed off to a size of 100,000 sectors and all is well. Anyone know whats going on ?? By the way, I looked at the driver's udamicro variable and it contained 0x4063 which indicates a microcode level of Version 3, model 6, if thats of any use. Richard Stevens Health Systems International, New Haven, CT ihnp4 ! hsi ! stevens
pep@princeton.UUCP (07/24/86)
Yes. The same thing that was going on in the 4.2 /etc/disktab. In the 4.3 disktab (version 4.11, date 5/31/86) the entry for the RA80 is incorrect (I just ran into the same problem). The RA80 has 546 cylinders, not 559. Make the following changes to /etc/disktab: :nc#546: :pc#236964: :pe#187640: :pg#187054: :ph#105560: I have made the h partition successfully; I haven't tried e or g. Pat Parseghian
chris@umcp-cs.UUCP (Chris Torek) (07/26/86)
In article <383@hsi.UUCP> stevens@hsi.UUCP (Richard Stevens) writes: >We just brought up 4.3 BSD on an 11-785 with a UDA50 with >one RA80 (drive 0) ... newfs tries to mkfs with a size of >111,202 sectors. Mkfs very quickly bombs with the message > > endcd=200, stat=16001 > >Comments in the driver indicate that this is probably a response >to an invalid command, with some mention of invalid block numbers. Code 0200 is M_OP_END. According to the MSCP documentation that I have never seen, this is given only in response to an invalid command. According to much more reliable legend, this is also given in response to an attempt to write too far past the end of the disk. According to established fact, an attempt to write a little past the end of an RA81 will write on the bad sector replacement information; the UDA50 driver is supposed to avert this kind of disaster, and I can only guess that the distributed 4.3 partition tables are wrong. >Anyone know whats going on ?? By the way, I looked at the driver's >udamicro variable and it contained 0x4063 which indicates a microcode >level of Version 3, model 6, if thats of any use. DEC is up to version 5 model 6, at least (we have two such). Version 3 still has the Get Unit Status microcode hang bug, I believe. Of course, it is practically impossible to get any reliable information on UDA50s and MSCP out of DEC. Rumour has it that this is because DEC `accidentally' sold such to various competitors, and decided this was a bad idea. If so, I cannot see why they do not simply require a nondisclosure/noncompetition agreement before selling the documentation. Until DEC starts selling MSCP documentation---under whatever strictures---I will continue to recommend against buying any of their products that use it. (I would recommend against RA81s anyway. Eagles are reliable, almost as large, less expensive, and faster; and DEC will service them.) -- In-Real-Life: Chris Torek, Univ of MD Comp Sci Dept (+1 301 454 1516) UUCP: seismo!umcp-cs!chris CSNet: chris@umcp-cs ARPA: chris@mimsy.umd.edu