Datasheet

PIC24FJXXXGA1/GB1
DS39907A-page 12 © 2007 Microchip Technology Inc.
3.2.1 SIX SERIAL INSTRUCTION
EXECUTION
The SIX control code allows execution of PIC24F family
assembly instructions. When the SIX code is received,
the CPU is suspended for 24 clock cycles, as the instruc-
tion is then clocked into the internal buffer. Once the
instruction is shifted in, the state machine allows it to be
executed over the next four PGC clock cycles. While the
received instruction is executed, the state machine
simultaneously shifts in the next 4-bit command (see
Figure 3-2).
Coming out of Reset, the first 4-bit control code is
always forced to SIX and a forced NOP instruction is
executed by the CPU. Five additional PGCx clocks are
needed on start-up, resulting in a 9-bit SIX command
instead of the normal 4-bit SIX command.
After the forced SIX is clocked in, ICSP operation
resumes as normal. That is, the next 24 clock cycles
load the first instruction word to the CPU.
FIGURE 3-2: SIX SERIAL EXECUTION
3.2.1.1 Differences Between Execution of
SIX and Normal Instructions
There are some differences between executing instruc-
tions normally and using the SIX ICSP command. As a
result, the code examples in this specification may not
match those for performing the same functions during
normal device operation.
The important differences are:
• Two-word instructions require two SIX operations
to clock in all the necessary data.
Examples of two-word instructions are GOTO and
CALL.
• Two-cycle instructions require two SIX operations.
The first SIX operation shifts in the instruction and
begins to execute it. A second SIX operation – which
should shift in a NOP to avoid losing data – provides
the CPU clocks required to finish executing the
instruction.
Examples of two-cycle instructions are table read
and table write instructions.
• The CPU does not automatically stall to account
for pipeline changes.
A CPU stall occurs when an instruction modifies a
register that is used for Indirect Addressing by the
following instruction.
During normal operation, the CPU automatically
will force a NOP while the new data is read. When
using ICSP, there is no automatic stall, so any
indirect references to a recently modified
register should be preceded by a NOP.
For example, the instructions, MOV #0x0,W0 and
MOV [W0],W1, must have a NOP inserted
between them.
If a two-cycle instruction modifies a register that is
used indirectly, it will require two following NOPs: one
to execute the second half of the instruction and a
second to stall the CPU to correct the pipeline.
Instructions such as TBLWTL [W0++],[W1]
should be followed by two NOPs.
• The device Program Counter (PC) continues to
automatically increment during ICSP instruction
execution, even though the Flash memory is not
being used.
As a result, the PC may be incremented to point to
invalid memory locations. Invalid memory spaces
include unimplemented Flash addresses and the
vector space (locations 0x0 to 0x1FF).
If the PC points to these locations, the device will
reset, possibly interrupting the ICSP operation. To
prevent this, instructions should be periodically
executed to reset the PC to a safe space. The opti-
mal method to accomplish this is to perform a
GOTO 0x200.
Note: To account for this forced NOP, all example
code in this specification begins with a
NOP to ensure that no data is lost.
P4
23 123 2324 1 2 3 4
P1
PGCx
P4A
PGDx
24-Bit Instruction Fetch
Execute PC – 1,
16
0000
Fetch SIX
456 78 181920212217
LSB X X X X X X X X X X X X X X MSB
PGDx = Input
P2
P3
P1B
P1A
78 9
0000000
Only for
Program
Memory Entry
Control Code
4 5
Execute 24-Bit
Instruction, Fetch
Next Control Code
00