Datasheet

Chapter 5. Debug Support 5-15
Programming Model
5.4.5 Data Breakpoint/Mask Registers (DBR/DBR1,
DBMR/DBMR1)
The data breakpoint registers (DBR/DBR1), Figure 5-10, specify data patterns used as part
of the trigger into debug mode. Only DBRn bits not masked with a corresponding zero in
DBMRn are compared with the data from the processor’s local bus, as defined in TDR.
12–11 DDC Debug data control. Controls operand data capture for PSTDDATA, which displays the number of
bytes defined by the operand reference size before the actual data; byte displays 8 bits, word
displays 16 bits, and long displays 32 bits (one nibble at a time across multiple clock cycles). See
Table 5-4.
00 No operand data is displayed.
01 Capture all write data.
10 Capture all read data.
11 Capture all read and write data.
10 UHE User halt enable. Selects the CPU privilege level required to execute the HALT instruction.
0 HALT is a supervisor-only instruction.
1 HALT is a supervisor/user instruction.
9–8 BTB Branch target bytes. Defines the number of bytes of branch target address PSTDDATA displays.
00 0 bytes
01 Lower 2 bytes of the target address
10 Lower 3 bytes of the target address
11 Entire 4-byte target address
See Section 5.3.1, “Begin Execution of Taken Branch (PST = 0x5).”
7 — Reserved, should be cleared.
6 NPL Non-pipelined mode. Determines whether the core operates in pipelined or mode.
0 Pipelined mode
1 Nonpipelined mode. The processor effectively executes one instruction at a time with no overlap.
This adds at least 5 cycles to the execution time of each instruction. Instruction folding is
disabled. Given an average execution latency of 1.6, throughput in non-pipeline mode would be
6.6, approximately 25% or less compared to pipelined performance.
Regardless of the NPL state, a triggered PC breakpoint is always reported before the triggering
instruction executes. In normal pipeline operation, the occurrence of an address and/or data
breakpoint trigger is imprecise. In non-pipeline mode, triggers are always reported before the next
instruction begins execution and trigger reporting can be considered precise.
An address or data breakpoint should always occur before the next instruction begins execution.
Therefore the occurrence of the address/data breakpoints should be guaranteed.
5 — Reserved, should be cleared.
4 SSM Single-step mode. Setting SSM puts the processor in single-step mode.
0 Normal mode.
1 Single-step mode. The processor halts after execution of each instruction. While halted, any
BDM command can be executed. On receipt of the
GO command, the processor executes the
next instruction and halts again. This process continues until SSM is cleared.
3–0 — Reserved, should be cleared.
Table 5-11. CSR Field Descriptions (Continued)
Bit Name Description