rws%mit-bold@sri-unix.UUCP (12/13/83)
From: Robert W. Scheifler <rws@mit-bold> Description: For example, after a non-recoverable machine check, and panic: mchk, you get trap type 2, code = 0, pc = 80000d76 panic: Reserved operand a number of times and then panic: KSP not valid The problem is caused by REIing back to the interrupt stack at priority level 0, which causes a reserved operand fault on the VAX. Repeat-By: Cause a panic and have disk interrupts go off while waiting for bufs to go out in boot(). Fix: The fix is almost there. In 4.1, in panic(), there was an spl0() that caused this problem. In 4.2 the spl was moved into boot() and changed to an spl1(). However, asm.sed turns an spl1() into the same thing as an spl0(), and you still lose. So change s/calls $0,_spl1/mfpr $18,r0\ mtpr $0,$18/ to be s/calls $0,_spl1/mfpr $18,r0\ mtpr $1,$18/ (Actually, I think any value between 1 and 11 would work.)