feat: don't inline time-critical functions

Cargo.toml

@@ -58,7 +58,7 @@ codegen-units = 1
 debug = 2
 debug-assertions = true
 incremental = false
-opt-level = 's'
+opt-level = 3
 lto = "fat"
 overflow-checks = true
 
@@ -71,7 +71,7 @@ incremental = false
 lto = 'fat'
 # opt level needs to be benchmarked after every major feature
 # due to the changes in binary size and alignment
-opt-level = 's'
+opt-level = 3
 overflow-checks = false
 
 # do not optimize proc-macro crates = faster builds from scratch

src/filter.rs

@@ -179,7 +179,7 @@ pub struct KalmanState {
 }
 
 impl KalmanState {
-    // runs kalman filter
+    #[inline(never)]
     #[link_section = ".time_critical.run_kalman"]
     pub fn run_kalman(
         &mut self,
@@ -292,6 +292,7 @@ impl KalmanState {
 ///  output at the rim longer when released.
 ///
 /// Output is a tuple of the x and y positions.
+#[inline(never)]
 #[link_section = ".time_critical.run_waveshaping"]
 pub fn run_waveshaping(
     x_pos: f32,

src/input.rs

@@ -82,6 +82,7 @@ pub enum StickAxis {
     YAxis,
 }
 
+#[inline(never)]
 #[link_section = ".time_critical.read_ext_adc"]
 pub fn read_ext_adc<
     'a,
@@ -125,6 +126,7 @@ pub fn read_ext_adc<
 /// Gets the average stick state over a 1ms interval in a non-blocking fashion.
 /// Will wait until end_time is reached before continuing after reading the ADCs.
 #[allow(clippy::too_many_arguments)]
+#[inline(never)]
 #[link_section = ".time_critical.update_stick_states"]
 async fn update_stick_states(
     current_stick_state: &StickState,
@@ -540,6 +542,7 @@ pub async fn update_button_state_task(
 ///
 /// Has to run on core0 because it makes use of SPI0.
 #[embassy_executor::task]
+#[inline(never)]
 #[link_section = ".time_critical.update_stick_states_task"]
 pub async fn update_stick_states_task(
     spi: Spi<'static, SPI0, embassy_rp::spi::Blocking>,

src/stick.rs

@@ -882,6 +882,7 @@ pub fn calc_stick_values(angle: f32) -> (f32, f32) {
     (x, y)
 }
 
+#[inline(never)]
 #[link_section = ".time_critical.linearize"]
 pub fn linearize(point: f32, coefficients: &[f32; NUM_COEFFS]) -> f32 {
     coefficients[0] * (point * point * point)
@@ -890,6 +891,7 @@ pub fn linearize(point: f32, coefficients: &[f32; NUM_COEFFS]) -> f32 {
         + coefficients[3]
 }
 
+#[inline(never)]
 #[link_section = ".time_critical.notch_remap"]
 pub fn notch_remap(
     x_in: f32,