
Real-Time Trace Support
MOTOROLA
Debug Support
19-5
appropriate marker and then the data transfer on the DDATA outputs. The length of the data transfer is
dependent on the operand size of the WDDATA instruction.
19.2.1.5
Begin Execution of Taken Branch (PST = $5)
The bytes are always displayed in a least-significant to most-significant order. The processor captures only
those target addresses associated with taken branches using a variant addressing mode. For example, all
JMP and JSR instructions using address register indirect or indexed addressing modes, all RTE and RTS
instructions as well as all exception vectors.
The simplest example of a branch instruction using a variant address is the compiled code for a C
language “case” statement. Typically, the evaluation of this statement uses the variable of an expression
as an index into a table of offsets, where each offset points to a unique case within the structure. For these
types of change-of-flow operations, the ColdFire processor uses the debug pins to output a sequence of
information on successive processor clock cycles:
1. Identify a taken branch has been executed using the PST pins ($5).
2. Using the PST pins, optionally signal the target address is to be displayed on the DDATA pins. The
encoding ($9, $A, $B) identifies the number of bytes that are displayed.
3. The new target address is optionally available on subsequent cycles using the nibble-wide DDATA
port. The number of bytes of the target address displayed on this port is a configurable parameter
(2, 3, or 4 bytes).
Another example of a variant branch instruction would be a JMP (A0) instruction.
Figure 19-2 shows the
Figure 19-2 Example PST/DDATA Diagram
PST is driven with a $5 indicating a taken branch. In the second cycle, PST is driven with a marker value of
$9 indicating a two-byte address that is displayed four bits at a time on the DDATA signals over the next
four clock cycles. The remaining four clock cycles display the lower two-bytes of the address (A0), least
significant nibble to most significant nibble. The output of the PST signals after the JMP instruction
completes is dependent on the target instruction. The PST can continue with the next instruction before the
address has completely displayed on the DDATA because of the DDATA FIFO. If the FIFO is full and the
next instruction needs to display captured values on DDATA, the pipeline stalls (PST = $0) until space is
available in the FIFO.
PSTCLK
PST
DData
$5
$0
$9
A[3:0]
A[7:4]
A[11:8]
A[15:12]
F
re
e
sc
a
le
S
e
m
ic
o
n
d
u
c
to
r,
I
Freescale Semiconductor, Inc.
For More Information On This Product,
Go to: www.freescale.com
n
c
..
.