klieb@tekigm2.MEN.TEK.COM (Kurt Liebezeit) (01/05/89)
I am currently running OS-9/68000 V2.1 (while I await the arrival of an upgrade to V2.2) on a Hazelwood Uniquad 2. I've been pretty busy at work until the Christmas shutdown occurred, and I have only had about a week's worth of intensive use on the 2.1 version. Even with this small amount of time on the system I keep running into a really nasty bug. The symptom of the bug is that at some point the hard disk controller seems to get lost. From the shell, it will fail to open a file for execution and return an error 021:247. All subsequent hard disk accesses fail with the same error. It also failed once when cc tried to load and execute the optimizer program. Once in this state, it won't execute anything that isn't already loaded into memory. The 021 is a Western Digital error code that means an invalid sector was requested, and the 247 is the code for a file seek error. There doesn't seem to be any consistency as to when it occurs; I've been using the YASE editor and doing C compiles with version 2.1 of the C compiler. I had to re-assemble YASE to be able to run it under OSK version 2.1, but it never gave me a problem when I ran an assembled version of the same source code under version 1.2. I have not even gotten to the point of trying to run the program that I'm developing, so I don't think that the problem is a stray pointer in the code that I'm writing. At first I thought that it might be related to having ghosting enabled, but it has occurred with ghosting off since then. Once it happens I have to power off to get the error to go away; a CPU reset won't do it. That makes me suspect that it is the controller being loaded with bad data. I have run dcheck many times to see if there is some corruption in the file system, but it always checks out intact. My hard disk was formatted under version 1.2; I ran the blk0up program as part of the upgrade to OSK 2.1, but my fear is that there is some hidden incompatibility that won't go away unless I re-format the hard disk under 2.1. Does this look like a known bug of V2.1? In my experience, Dave Bridger of Hazelwood writes very good drivers, but could the caching in the new drivers be getting corrupted somehow? Note that the system and its drivers are the same as the 68000 board sold by Frank Hogg, so there may be folks on the net with direct experience with this board. If you have any suggestions please e-mail, or post if you think that it is general enough. Thanks for your time and help. Kurt Liebezeit klieb@tekigm2.TEK.COM ...!tektronix!tekigm2!klieb