Error Management; Traps - Prime Computer 50 Series Service Manual

Principles of Operation Prime Proprietary
e Calling procedure's keys and register values at point of call
@ Variables shared by the calling and called procedures
When the microcode executes the Procedure Call Instruction (PCL),
several things happen.
1. The microcode determines the called procedure's location in memory
and verifies that the calling procedure has access rights to it.
2. The microcode allocates memory space (stack frame) for the called
procedure.
3. The microcode stores the calling procedure''s point of call, keys
and register values, in the called procedure stack frame.
4. The microcode loads variables, keys and register values from the
ECB into its registers.
5. The microcode stores the memory addresses of the called
procedure's computation results in the called procedure stack
frame.
Once the called procedure is executed, the microcode restarts the
interrupted calling procedure. All keys, register values and
computational results are reloaded from the stack frame.
4.2.10 ERROR MANAGEMENT
Error management for the 50 Series is provided by traps, faults,
checks and interrupts. These means of error management are presented
in detail in the next subsections.
4.2.10.1 Traps
Traps are breaks in microcode execution. Some, but not all traps,
cause breaks in software execution (page not in memory, for example).
There are two basic types of traps:
@ Gl Traps - These traps are the direct result of a cache access,
'They are considered live, because the trap condition is
detected and handled in the same microcode step.
e G2 Traps - These traps are used for static conditions and are
generally caused by a preexisting machine condition.
When any trap condition is detected, the present microcode step is
aborted and the trap logic causes a branch to the appropriate trap
vector address. After the trap has been handled, the aborted step is
usually reexecuted.
4.2.10.2 Faults
Faults are breaks in currently executing software, which must be
serviced by the software. Faults are vectored through the software to
microcode fault handler routines. These routines suspend current
software operation, handle the fault and resume operation.
070-C 4- 14
"er