Future Commands - Motorola CPU32 Reference Manual

Hide thumbs Also See for CPU32:
Table of Contents

Advertisement

7.2.8.16 Future Commands

Unassigned command opcodes are reserved by Motorola for future expansion. All un-
used formats within any revision level will perform a NOP and return the ILLEGAL
command response.
7.3 Deterministic Opcode Tracking
The CPU32 utilizes deterministic opcode tracking to trace program execution. Two
signals, IPIPE and IFETCH, provide all the information required to analyze the opera-
tion of the instruction pipeline.
7.3.1 Instruction Fetch (IFETCH)
IFETCH indicates which bus cycles are accessing data to fill the instruction pipeline.
IFETCH is pulse-width modulated to multiplex two indications on a single pin. Asserted
for a single clock cycle, IFETCH indicates that the data from the current bus cycle is
to be routed to the instruction pipeline. IFETCH held low for two clock cycles indicates
that the instruction pipeline has been flushed. The data from the bus cycle is used to
begin filling the empty pipeline. Both user and supervisor mode fetches are signaled
by IFETCH.
Proper tracking of bus cycles via the IFETCH signal on a fast bus requires a simple
state machine. On a two-clock bus, IFETCH may signal a pipeline flush with associat-
ed prefetch followed immediately by a second prefetch. That is, IFETCH remains as-
serted for three clocks, two clocks indicating the flush/fetch and a third clock signaling
the second fetch. These two operations are easily discerned if the tracking logic sam-
ples IFETCH on the two rising edges of CLKOUT, which follow the address strobe (da-
ta strobe during show cycles) falling edge. Three-clock and slower bus cycles allow
time for negation of the signal between consecutive indications and do not experience
this operation.
7.3.2 Instruction Pipe (IPIPE)
The internal instruction pipeline can be modeled as a three-stage FIFO (see Figure 7-
11). Stage A is an input buffer — data can be used out of the stages B and C. IPIPE
signals advances of instructions in the pipeline.
Instruction register A (IRA) holds incoming words as they are prefetched. No decoding
takes place in the buffer. Instruction register B (IRB) provides initial decoding of the
opcode and decoding of extension words —it is a source of immediate data. Instruc-
tion register C (IRC) supplies residual opcode decoding during instruction execution.
CPU32
REFERENCE MANUAL
DEVELOPMENT SUPPORT
MOTOROLA
7-25

Advertisement

Table of Contents
loading

Table of Contents