![Infineon Technologies XC2200 User Manual Download Page 253](http://html1.mh-extra.com/html/infineon-technologies/xc2200/xc2200_user-manual_2055439253.webp)
XC2200 Derivatives
System Units (Vol. 1 of 2)
Interrupt and Trap Functions
User’s Manual
5-41
V2.1, 2008-08
ICU_X2K, V2.2
Sources for Additional Delays
Because the service requests are inserted into the current instruction stream, the
properties of this instruction stream can influence the request latency.
The actual response to an interrupt request may be delayed further depending on
programming techniques used by the application. The following factors can contribute:
•
Actual interrupt service routine is only reached via a JUMP from the interrupt vector
table.
Time-critical instructions can be placed directly into the interrupt vector table,
followed by a branch to the remaining part of the interrupt service routine. The space
between two adjacent vectors can be selected via bitfield VECSC in register
CPUCON1.
•
Context switching is executed before the intended action takes place (see
)
Time-critical instructions can be programmed “non-destructive” and can be executed
before switching context for the remaining part of the interrupt service routine.
Table 5-14
Additional Delays Caused by System Logic
Reason for Delay
Interrupt Response
PEC Response
Interrupt controller busy,
because the previous interrupt request
is still in process
max. 7 cycles
max. 7 cycles
Pipeline is stalled,
because instructions preceding the
injected instruction in the pipeline need
to write/read data to/from a peripheral
or memory
2
×
T
ACCmax
1)
1) This is the longest possible access time within the XC2200 system.
2
×
T
ACCmax
Pipeline cancelled,
because instructions preceding the
injected instruction in the pipeline
update core SFRs
4 cycles
4 cycles
Memory access for stack writes (if not
to DPRAM or DSRAM)
2/3
×
T
ACC
2)
2) Depending on segmentation off/on.
- - -
Memory access for vector table read
(except for intr. jump table cache)
2
×
T
ACC
- - -