merged irq-refactor branch, which fixes some small but long-standing issues:

- occasional missed button events while asleep
- occasional short/aborted frames in aux LED sleep animation
- rare case of bogus voltage and/or temperature values
- fixed issue where nice_delay_ms() didn't work in setup()
- fixed theoretical possibility of extra-noisy buttons causing a hang

Also does some other things:
- cleans up the ADC code significantly
- adds a voltage stabilizer/lowpass option (enabled on t1634 builds)
- greatly reduces time spent per interrupt,
  which might make a future PWM-DSM technique possible
- moves most interrupt-handling logic to a non-critical code path,
  deferring that code until timing doesn't matter as much
- sped up button state measurements
- turned off muggle mode on a couple builds which were too big

1 files changed, 21 insertions, 1 deletions

diff --git a/spaghetti-monster/fsm-main.c b/spaghetti-monster/fsm-main.c
index e537a9e..c9ab69b 100644
--- a/spaghetti-monster/fsm-main.c
+++ b/spaghetti-monster/fsm-main.c
@@ -123,7 +123,6 @@ int main() {
     #else
     delay_4ms(1);
     #endif
-    empty_event_sequence();
 
     // fallback for handling a few things
     #ifndef DONT_USE_DEFAULT_STATE
@@ -160,6 +159,9 @@ int main() {
             standby_mode();
         }
 
+        // catch up on interrupts
+        handle_deferred_interrupts();
+
         // give the recipe some time slices
         loop();
 
@@ -168,4 +170,22 @@ int main() {
     }
 }
 
+
+void handle_deferred_interrupts() {
+    /*
+    if (irq_pcint) {  // button pressed or released
+        // nothing to do here
+        // (PCINT only matters during standby)
+    }
+    */
+    if (irq_adc) {  // ADC done measuring
+        ADC_inner();
+        // irq_adc = 0;  // takes care of itself
+    }
+    if (irq_wdt) {  // the clock ticked
+        WDT_inner();
+        // irq_wdt = 0;  // takes care of itself
+    }
+}
+
 #endif