fix SpinDelayUsPrecision() clock, no more wrapping 43ms> issues. Still capped at 1390ms. This affects a bunch of tearoff attacks 14b especially win 100ms actually being 16ms wrapped. Should be handled according to new HAL layer. Next things is rate renegotiated at reconnect for sim module. Makes things a more stable if something fails. Thanks Claude!
This commit is contained in:
1 parent
41c1809404
commit
9b61ef867c
5 files changed
+133
-37
No files matched your search
+9
-2
@@ -160,8 +160,15 @@ Current usages:
|
||||
|
||||
Busy wait based on 46.875 kHz PWM Channel 0
|
||||
|
||||
* 21.3 us precision and maximum 1.39 s
|
||||
* *Precision* variant: 0.7 us precision and maximum 43 ms
|
||||
* 21.3 us precision, no upper bound on the delay
|
||||
* *Precision* variant: 0.7 us precision, no upper bound on the delay
|
||||
|
||||
Both used to truncate their tick count into the 16 bit PWM counter and so
|
||||
returned early past those two limits without saying so: a requested 60 ms
|
||||
`SpinDelayUsPrecision()` measured 17.5 ms on an RDV4, and `hw tearoff --delay`
|
||||
accepts values right across that boundary. Longer waits are now walked in
|
||||
chunks of half a counter period. `SpinDelay()` still clamps its own argument at
|
||||
1390 ms and complains above that.
|
||||
|
||||
## Occasional TC0+TC1 / CountUS functions
|
||||
^[Top](#top)
|
||||
|
||||
Reference in new issue
Block a user