aps (07/21/82)
System III conspired against all of you! The recent problems with VAX-11/780's and System III is not the fault of the new 780 microcode (WCS124). In fact, it is System III that did not adhear to a limitation in the 780 set from the year of the flood. If one looks in the 1978-79 780 processor manual (I think that is the original) on page 78, under the description for SCBB (System Control Block Base register), one will note that bits 31 and 30 are MBS; this means MUST BE Zero, not, may be zero. System III, unlike Berkeley UNIX, loads the SCBB, it loads an address with bit 31 equal to a one. It turns out, that eariler versions of the microcode did not check for this bit and in fact did not care whether it was a one or not. WCS124, which is the new microcode for VAX-11/780 revision 7, does check it. If the SCBB is loaded with a 1 in bit 31, you will get a reserved operand fault (vector 18). In fact, if you use the wonderful standalone shell (SASH), you will get the cryptic message "HALT INSTRUCTION EXECUTED, HALTED AT 00000001". There is a way around this: The failed instruction is at AF8 and the offending data is in the longword at AFA. If you change the offending data (make the "8" a "0"), then restart the machine at the failed instruction (AF8), you can boot up System III. You better change the instruction in start.s mtpr $Scbbase,$SCBB # set SCBB to mtpr $Scbbase-0x80000000,$SCBB # set SCBB so this doesn't happen everytime you try to boot it. (NOTE: People trying to get wonderful System III onto 750's will run into the same problem.) AF8 = 02008FDA <--- Beginning of failing instruction AFA = DA118000 ^ +-- This "8" should be a "0". Change the "8" to a "0" and resume execution at AF8. Armando Stettner decvax!aps