Tried cleaning things up and putting out some new assembly

didn't work

flake.nix

@@ -284,6 +284,28 @@
           ];
         };
 
+      packages."x86_64-linux"."ws2812_" = 
+      let
+        pkgs = import nixpkgs { system = "x86_64-linux"; };
+        avr = pkgs.pkgsCross.avr.buildPackages;
+      in packages."x86_64-linux"."ws2812" { gcc = "${avr.gcc}/bin/avr-gcc"; cflags = mcu_cflags attiny85; avr = true; };
+      packages."x86_64-linux"."ws2812" =
+        { gcc, cflags, avr }:
+        let
+          pkgs = import nixpkgs { system = "x86_64-linux"; };
+        in mkProgram {
+          pkgs = pkgs;
+          gcc = gcc;
+          cflags = cflags;
+          pname = "ws2812";
+          psrc = ./ws2812;
+          inherit avr;
+
+          pbuildInputs = [
+            (packages."x86_64-linux"."dio" { inherit gcc cflags; })
+          ];
+        };
+
       devShell."x86_64-linux" =
         let
           pkgs = import nixpkgs { system = "x86_64-linux"; };
@@ -297,9 +319,8 @@
           ];
         in
         pkgs.mkShell {
-          name = "avr-shell";
-          buildInputs = avr;
-
+          name = "Wearables-shell";
+          buildInputs = avr ++ [ pkgs.gnumake ];
           GCC = pkgs.pkgsCross.avr.buildPackages.gcc;
           SIMAVR = pkgs.pkgsCross.avr.buildPackages.simavr;
         };

flash-trinket

@@ -1 +1,2 @@
-avrdude -c usbtiny -p attiny85 -U flash:w:$1:i; sleep 5; avrdude -c usbtiny -p attiny85 -D -U flash:w:$1:i
+avrdude -c usbtiny -p attiny85 -U flash:w:$1:i; sleep 5; avrdude -c usbtiny -p attiny85 -D -U flash:w:$1:i
+# avrdude -c usbtiny -p attiny85 -D -U flash:w:$1:i

pwm/src/main.c

@@ -0,0 +1,74 @@
+#include <avr/io.h>
+#include <util/delay.h>
+#include <avr/interrupt.h>
+
+#include <base.h>
+
+#define FIVE_SECOND_TICK (8000000 * 10) / 256
+
+volatile uint32_t iteration = 0;
+
+void set_up_pwm(void);
+void start_pwm(void);
+
+ISR(TIMER0_OVF_vect) {
+    if (iteration >= FIVE_SECOND_TICK) {
+        GTCCR |= _BV(TSM);
+        /*
+        PORTB &= ~(_BV(2));
+        PORTB |= _BV(1);
+        */
+        PORTB = _BV(1);
+    } else {
+        iteration += 1;
+    }
+}
+
+void set_up_pwm(void) {
+    // Stop the clock
+    GTCCR = _BV(TSM) | _BV(PSR0);
+    
+    TCCR0B = _BV(CS00);
+
+    // Set normal mode
+    TCCR0A = _BV(COM0A1) | _BV(WGM01) | _BV(WGM00);
+
+    // Enable the overflow interrupt
+    TIMSK = _BV(TOIE0);
+}
+
+void start_pwm(void) {
+    GTCCR &= ~(_BV(TSM));
+}
+
+int main (void) {
+    int8_t step = 1;
+    uint8_t target = 0;
+
+    set_up_pwm();
+    OCR0A = 0;
+
+    PORTB = 0;
+    DDRB = _BV(2) | _BV(1) | _BV(0);
+    _delay_ms(50);
+    PORTB |= _BV(2);
+
+    sei();
+    start_pwm();
+
+    while (iteration != FIVE_SECOND_TICK) {
+        OCR0A = target;
+        target += step;
+
+        if (target == 255) {
+            step = -1;
+        } else if (target == 0) {
+            step = 1;
+        }
+
+        _delay_ms(10);
+    }
+
+    while (1) {}
+}
+

ws2812/direct_write_assembly.c

@@ -0,0 +1,52 @@
+// I have reduced noops at the end of each of these to take into account that there are several additional clock ticks of setup, after. However, I'm not totally sure that I get things right, seeing that there are four possible sequences, and I'm not really accounting for the timing of all four of them.
+#define write_zero(port, bit)           \
+    __asm__ __volatile__ (              \
+            "sbi %0, %1" "\n\t"         \
+            "nop" "\n\t"                \
+            "cbi %0, %1" "\n\t"         \
+            "nop" "\n\t"                \
+            : /* no outputs */          \
+            : "I" (_SFR_IO_ADDR(port)), \
+              "I" (bit)                 \
+    )
+
+#define write_one(port, bit)            \
+    __asm__ __volatile__ (              \
+            "sbi %0, %1" "\n\t"         \
+            "nop" "\n\t"                \
+            "nop" "\n\t"                \
+            "nop" "\n\t"                \
+            "nop" "\n\t"                \
+            "nop" "\n\t"                \
+            "cbi %0, %1" "\n\t"         \
+            : /* no outputs */          \
+            : "I" (_SFR_IO_ADDR(port)), \
+              "I" (bit)                 \
+    )
+
+#define write_byte(port, bit, byte)             \
+    __asm__ __volatile__ (                      \
+    ; count out eight bits                      \
+             "ldi %z, 8" "\n\t"                 \
+    ; load the current byte into a temporary register \
+             "ld __tmp_reg__, %[byte]" "\n\t"   \
+    ; shift the temporary register left, saving the msb in SREG (1 cycle) \
+    "L_%=: " "lsl __tmp_reg__" "\n\t"           \
+    ; if SREG is set, branch to I_%= (2 cycles if true, 1 cycle if false) \
+             "brbs I_%=" "\n\t"                 \
+    ; SREG was zero, so write a zero to the port \
+             write_zero(port, bit)              \
+    ; Jump to J_%=, the loop cleanup (2 cycles) \
+             "rjmp J_%=" "\n\t"                 \
+    ; SREG was one, so write a one to the port  \
+    "I_%=: " write_one(port, bit)               \
+    ; Decrement the bits counter (1 cycle)      \
+    "J_%=: " "dec %z" "\n\t"                    \
+    ; are there any bits left to send? (1 cycle) \
+             "cpi %z, 0" "\n\t"                 \
+    ; there are, so go back to L_%= (2 cycles)  \
+             "brne L_%=" "\n\t"                 \
+             : /* no outputs */                 \
+             : [byte] "I" (byte)                \
+    )
+

ws2812/main.c

@@ -0,0 +1,28 @@
+#include <avr/io.h>
+#include <util/delay.h>
+#include <dio.h>
+#include "ws2812.h"
+
+/*
+#include <simavr/avr/avr_mcu_section.h>
+AVR_MCU(F_CPU, "attiny85");
+
+const struct avr_mmcu_vcd_trace_t _mytrace[]  _MMCU_ = {
+   { AVR_MCU_VCD_SYMBOL("DDRB"), .what = (void*)&DDRB, },
+   { AVR_MCU_VCD_SYMBOL("PORTB"), .what = (void*)&PORTB, },
+};
+*/
+
+#define PIXEL_COUNT 1
+const rgb_t pixels[PIXEL_COUNT] = {
+    (rgb_t){ .r = 255,   .g = 0, .b = 255, },
+};
+
+int main (void) {
+    dio_t pin = (dio_t){ .ddr = &DDRB, .port = &PORTB, .pin = &PINB, .addr = 2 };
+    dio_set_direction(&pin, LINE_OUT);
+
+    np_write_rgb(&pin, pixels, PIXEL_COUNT);
+
+    return 0;
+}

ws2812/ws2812.c

@@ -0,0 +1,108 @@
+#include <avr/interrupt.h>
+# include "ws2812.h"
+
+// inline void write_zero(dio_t *pin, uint8_t hi, uint8_t low) {
+//     uint8_t *val;
+//     uint8_t *hi_reg;
+//     uint8_t *low_reg;
+//     __asm__ __volatile__ (
+//             "ldd %[hi_reg], %[hi]\n"
+//             "ldd %[low_reg], %[low]\n"
+//             "ldd %[val], %[pin]\n"
+//             "or %[val], %[hi_reg]\n"
+//             "std %[pin], %[val]\n"
+//             "nop\n"
+//             "nop\n"
+//             "and %[val], %[hi_reg]\n"
+//             "std %[pin], %[val]\n"
+//             "nop\n"
+//             "nop\n"
+//             "nop\n"
+//             /* Need another 5 cycles */
+//             : [val] "=r" (val), [hi_reg] "=r" (hi_reg), [low_reg] "=r" (low_reg)
+//             : [pin] "m" (pin->port), [hi] "m" (hi), [low] "m" (low)
+//         );
+// }
+
+
+// inline void write_one(dio_t *port, uint8_t hi, uint8_t low) {
+//     __asm__ __volatile__ (
+//             "or %0, %1\n"
+//             "nop\n"
+//             "nop\n"
+//             "nop\n"
+//             "nop\n"
+//             "nop\n"
+//             "nop\n"
+//             "and %0, %1\n"
+//             /* Need another 5 cycles */
+//             : /* no outputs */
+//             : "m" (port->port),
+//               "m" (hi),
+//               "m" (low)
+//         );
+// }
+
+inline void write_byte(dio_t *pin, uint8_t hi, uint8_t low, const uint8_t value) {
+    uint8_t *state_r;
+    uint8_t *hi_r;
+    uint8_t *low_r;
+    uint8_t *counter;
+
+    __asm__ __volatile__ (
+             "ldi r16, 8\n"
+             "ldi r17, %[value]\n"
+             "ldd %[hi_r], %[hi]\n"
+             "ldd %[low_r], %[low]\n"
+             "ldd %[state_r], %[pin]\n"
+
+    "L_%=: " "lsl __tmp_reg__\n"
+             "brbs I_%=\n"
+             "or %[state_r], %[hi_r]\n"
+             "nop\n"
+             "nop\n"
+             "and %[state_r], %[low_r]\n"
+             "nop\n"
+             "nop\n"
+             "nop\n"
+             "rjmp J_%=\n"
+
+    "I_%=: "
+             "brbs I_%=\n"
+             "or %[state_r], %[hi_r]\n"
+             "nop\n"
+             "nop\n"
+             "nop\n"
+             "nop\n"
+             "nop\n"
+             "and %[state_r], %[low_r]\n"
+
+    "J_%=: " "dec %[counter]\n"
+             "cpi %[counter], 0\n"
+             "brne L_%="
+
+            : [state_r] "=r" (state_r),
+              [hi_r] "=r" (hi_r),
+              [low_r] "=r" (low_r),
+              [counter] "=r" (counter)
+
+            : [pin] "m" (pin->port),
+              [hi] "m" (hi),
+              [low] "m" (low),
+              [value] "m" (value)
+            );
+}
+
+
+void np_write_rgb(dio_t *pin, const rgb_t *pixels, uint8_t length) {
+    cli();
+    uint8_t hi = 1 << pin->addr;
+    uint8_t low = ~hi;
+    for (int idx = 0; idx < length; idx++) {
+        write_byte(pin, hi, low, pixels[idx].r);
+        write_byte(pin, hi, low, pixels[idx].g);
+        write_byte(pin, hi, low, pixels[idx].b);
+    }
+    sei();
+}
+

ws2812/ws2812.h

@@ -0,0 +1,27 @@
+#ifndef __neopixels_h__
+#define __neopixels_h__
+
+#include <avr/io.h>
+#include <dio.h>
+
+typedef struct RGB_s {
+    uint8_t r;
+    uint8_t g;
+    uint8_t b;
+} rgb_t;
+
+typedef struct RGBW_s {
+    uint8_t r;
+    uint8_t g;
+    uint8_t b;
+    uint8_t w;
+} rgbw_t;
+
+// void np_initialize();
+void np_write_rgb(dio_t *, const rgb_t *, uint8_t);
+// void np_write_grb(io_pin_t *addr, rgb_t *values, uint8_t length);
+// void np_write_rgbw(io_pin_t *addr, rgbw_t *values, uint8_t length);
+// void np_write_grbw(io_pin_t *addr, rgbw_t *values, uint8_t length);
+
+#endif
+