© 2004 Microchip Technology Inc. DS70053C-page 6-7
Section 6. Interrupts
Interrupts
6
6.2.1.2 Arithmetic Error Trap (Soft Trap, Level 11)
Any of the following events will cause an arithmetic error trap to be generated:
• Accumulator A Overflow
• Accumulator B Overflow
• Catastrophic Accumulator Overflow
• Divide by Zero
• Shift Accumulator (SFTAC) operation exceeding +/-16 bits
There are three enable bits in the INTCON1 register that enable the three types of accumulator
overflow traps. The OVATE control bit (INTCON1<10>) is used to enable traps for an Accumula-
tor A overflow event. The OVBTE control bit (INTCON1<9>) is used to enable traps for an
Accumulator B overflow event. The COVTE control bit (INTCON1<8>) is used to enable traps for
a catastrophic overflow of either accumulator.
An Accumulator A or Accumulator B overflow event is defined as a carry-out from bit 31. Note
that no accumulator overflow can occur if the 31-bit Saturation mode is enabled for the accumu-
lator. A catastrophic accumulator overflow is defined as a carry-out from bit 39 of either
accumulator. No catastrophic overflow can occur if accumulator saturation (31-bit or 39-bit) is
enabled.
Divide-by-zero traps cannot be disabled. The divide-by-zero check is performed during the first
iteration of the REPEAT loop that executes the divide instruction.
Accumulator shift traps cannot be disabled. The SFTAC instruction can be used to shift the
accumulator by a literal value or a value in one of the W registers. If the shift value exceeds
+/-16 bits, an arithmetic trap will be generated. The SFTAC instruction will execute, but the results
of the shift will not be written to the target accumulator.
An arithmetic error trap can be detected in software by polling the MATHERR status bit
(INTCON1<4>). To avoid re-entering the Trap Service Routine, the MATHERR status flag must
be cleared in software prior to returning from the trap with a RETFIE instruction. Before the
MATHERR status bit can be cleared, all conditions that caused the trap to occur must also be
cleared. If the trap was due to an accumulator overflow, the OA and OB status bits (SR<15:14>)
must be cleared. The OA and OB status bits are read only, so the user software must perform a
dummy operation on the overflowed accumulator (such as adding ‘0’) that will cause the
hardware to clear the OA or OB status bit.
6.2.2 Hard Traps
Hard traps include exceptions of priority level 13 through level 15, inclusive. The address error
(level 13) and oscillator error (level 14) traps fall into this category.
Like soft traps, hard traps can also be viewed as non-maskable sources of interrupt. The
difference between hard traps and soft traps is that hard traps force the CPU to stop code
execution after the instruction causing the trap has completed. Normal program execution flow
will not resume until after the trap has been Acknowledged and processed.
6.2.2.1 Trap Priority and Hard Trap Conflicts
If a higher priority trap occurs while any lower priority trap is in progress, processing of the lower
priority trap will be suspended and the higher priority trap will be Acknowledged and processed.
The lower priority trap will remain pending until processing of the higher priority trap completes.
Each hard trap that occurs must be Acknowledged before code execution of any type may
continue. If a lower priority hard trap occurs while a higher priority trap is pending, Acknowledged,
or is being processed, a hard trap conflict will occur. The conflict occurs because the lower
priority trap cannot be Acknowledged until processing for the higher priority trap completes.
The device is automatically reset in a hard trap conflict condition. The TRAPR status bit
(RCON<15> ) is set when the Reset occurs, so that the condition may be detected in software.
To Next Page
Loading...
