Cypress CY7C63310 ユーザーズマニュアル

ページ / 83
 
CY7C63310, CY7C638xx
Document 38-08035 Rev. *K
Page 51 of 83
17.2  Interrupt Processing
The sequence of events that occur during interrupt processing
follows:
1. An interrupt becomes active, because:
a. The interrupt condition occurs (for example, a timer expires).
b. A previously posted interrupt is enabled through an update 
of an interrupt mask register.
c. An interrupt is pending and GIE is set from 0 to 1 in the CPU 
Flag register.
2. The current executing instruction finishes.
3. The internal interrupt is dispatched, taking 13 cycles. During 
this time, the following actions occur: the MSB and LSB of 
Program Counter and Flag registers (CPU_PC and CPU_F) 
are stored onto the program stack by an automatic CALL 
instruction (13 cycles) generated during the interrupt 
acknowledge process.
a. The PCH, PCL, and Flag register (CPU_F) are stored onto 
the program stack (in that order) by an automatic CALL 
instruction (13 cycles) generated during the interrupt 
acknowledge process
b. The CPU_F register is then cleared. Because this clears the 
GIE bit to 0, additional interrupts are temporarily disabled.
c. The PCH (PC[15:8]) is cleared to zero.
d. The interrupt vector is read from the interrupt controller and 
its value placed into PCL (PC[7:0]). This sets the program 
counter to point to the appropriate address in the interrupt 
table (for example, 0004h for the POR/LVD interrupt).
4. Program execution vectors to the interrupt table. Typically, a 
LJMP instruction in the interrupt table sends execution to the 
user's Interrupt Service Routine (ISR) for this interrupt.
5. The ISR executes. Note that interrupts are disabled because
GIE = 0. In the ISR, interrupts are re-enabled by setting 
GIE = 1 (care must be taken to avoid stack overflow).
6. The ISR ends with a RETI instruction which restores the 
Program Counter and Flag registers (CPU_PC and CPU_F). 
The restored Flag register re-enables interrupts, because 
GIE = 1 again.
7. Execution resumes at the next instruction, after the one that 
occurred before the interrupt. However, if there are more 
pending interrupts, the subsequent interrupts are processed 
before the next normal program instruction.
17.3  Interrupt Trigger Conditions
Trigger conditions for most interrupts in 
have been explained in the relevant sections. However,
conditions under which the USB Active (interrupt address 0030h)
and PS2 Data Low (interrupt address 004Ch) interrupts are
triggered are explained follow.
1. USB Active Interrupt: Triggered when the D+/- lines are in a 
non-idle state, that is, K-state or SE0 state.
2. PS2 Data Low Interrupt: Triggered when SDATA becomes low 
when the SDATA pad is in the input mode for at least 6-7 
32 kHz cycles.
3. The GPIO interrupts are edge triggered.
17.4  Interrupt Latency
The time between the assertion of an enabled interrupt and the
start of its ISR is calculated from the following equation.
Latency = Time for current instruction to finish + Time for internal
interrupt routine to execute + Time for LJMP instruction in
interrupt table to execute.
For example, if the 5 cycle JMP instruction is executing when an
interrupt becomes active, the total number of CPU clock cycles
before the ISR begins is as follows:
(1 to 5 cycles for JMP to finish) + (13 cycles for interrupt routine)
+ (7 cycles for LJMP) = 21 to 25 cycles. 
In the previous example, at 24 MHz, 25 clock cycles take 
1.042 
μs. 
Figure 17-1.  Interrupt Controller Block Diagram
Interrupt
Source
(Timer,
GPIO, etc.)
Interrupt Taken
or
Posted
Interrupt
Pending
Interrupt
GIE
Interrupt Vector
Mask Bit Setting
D
R
Q
1
Priority
Encoder
M8C Core
Interrupt
Request
...
INT_MSKx
INT_CLRx Write
CPU_F[0]
...