9 files changed, 578 insertions, 171 deletions

diff --git a/common_features.mk b/common_features.mk
index 73aab5d84..883681324 100644
--- a/common_features.mk
+++ b/common_features.mk
@@ -263,11 +263,6 @@ ifneq ($(strip $(CUSTOM_MATRIX)), yes)
 endif
 
 ifeq ($(strip $(SPLIT_KEYBOARD)), yes)
-    SERIAL_BACKWARD_COMPAT := $(wildcard $(QUANTUM_DIR)/split_common/serial_backward_compatibility.h)
-    ifneq ($(SERIAL_BACKWARD_COMPAT),)
-      CONFIG_H += $(SERIAL_BACKWARD_COMPAT)
-      # $(info CONFIG_H=$(CONFIG_H))
-    endif
     OPT_DEFS += -DSPLIT_KEYBOARD
     QUANTUM_SRC += $(QUANTUM_DIR)/split_common/split_flags.c \
                 $(QUANTUM_DIR)/split_common/split_util.c
diff --git a/keyboards/miniaxe/matrix.c b/keyboards/miniaxe/matrix.c
index 55c458e1f..5fec1281d 100644
--- a/keyboards/miniaxe/matrix.c
+++ b/keyboards/miniaxe/matrix.c
@@ -32,9 +32,6 @@ along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #include "timer.h"
 #include "split_flags.h"
 
-#ifdef RGBLIGHT_ENABLE
-#   include "rgblight.h"
-#endif
 #ifdef BACKLIGHT_ENABLE
 #   include "backlight.h"
     extern backlight_config_t backlight_config;
@@ -55,6 +52,8 @@ along with this program.  If not, see <http://www.gnu.org/licenses/>.
     static bool debouncing = false;
 #endif
 
+#if defined(USE_I2C) || defined(EH)
+
 #if (MATRIX_COLS <= 8)
 #    define print_matrix_header()  print("\nr/c 01234567\n")
 #    define print_matrix_row(row)  print_bin_reverse8(matrix_get_row(row))
@@ -63,6 +62,27 @@ along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #else
 #    error "Currently only supports 8 COLS"
 #endif
+
+#else // USE_SERIAL
+
+#if (MATRIX_COLS <= 8)
+#    define print_matrix_header()  print("\nr/c 01234567\n")
+#    define print_matrix_row(row)  print_bin_reverse8(matrix_get_row(row))
+#    define matrix_bitpop(i)       bitpop(matrix[i])
+#    define ROW_SHIFTER ((uint8_t)1)
+#elif (MATRIX_COLS <= 16)
+#    define print_matrix_header()  print("\nr/c 0123456789ABCDEF\n")
+#    define print_matrix_row(row)  print_bin_reverse16(matrix_get_row(row))
+#    define matrix_bitpop(i)       bitpop16(matrix[i])
+#    define ROW_SHIFTER ((uint16_t)1)
+#elif (MATRIX_COLS <= 32)
+#    define print_matrix_header()  print("\nr/c 0123456789ABCDEF0123456789ABCDEF\n")
+#    define print_matrix_row(row)  print_bin_reverse32(matrix_get_row(row))
+#    define matrix_bitpop(i)       bitpop32(matrix[i])
+#    define ROW_SHIFTER  ((uint32_t)1)
+#endif
+
+#endif
 static matrix_row_t matrix_debouncing[MATRIX_ROWS];
 
 #define ERROR_DISCONNECT_COUNT 5
@@ -315,15 +335,39 @@ i2c_error: // the cable is disconnceted, or something else went wrong
 
 #else // USE_SERIAL
 
+
+typedef struct _Serial_s2m_buffer_t {
+    // TODO: if MATRIX_COLS > 8 change to uint8_t packed_matrix[] for pack/unpack
+    matrix_row_t smatrix[ROWS_PER_HAND];
+} Serial_s2m_buffer_t;
+
+volatile Serial_s2m_buffer_t serial_s2m_buffer = {};
+volatile Serial_m2s_buffer_t serial_m2s_buffer = {};
+uint8_t volatile status0 = 0;
+
+SSTD_t transactions[] = {
+    { (uint8_t *)&status0,
+      sizeof(serial_m2s_buffer), (uint8_t *)&serial_m2s_buffer,
+      sizeof(serial_s2m_buffer), (uint8_t *)&serial_s2m_buffer
+  }
+};
+
+void serial_master_init(void)
+{ soft_serial_initiator_init(transactions, TID_LIMIT(transactions)); }
+
+void serial_slave_init(void)
+{ soft_serial_target_init(transactions, TID_LIMIT(transactions)); }
+
 int serial_transaction(void) {
     int slaveOffset = (isLeftHand) ? (ROWS_PER_HAND) : 0;
 
-    if (serial_update_buffers()) {
+    if (soft_serial_transaction()) {
         return 1;
     }
 
+    // TODO:  if MATRIX_COLS > 8 change to unpack()
     for (int i = 0; i < ROWS_PER_HAND; ++i) {
-        matrix[slaveOffset+i] = serial_slave_buffer[i];
+        matrix[slaveOffset+i] = serial_s2m_buffer.smatrix[i];
     }
     
     #ifdef RGBLIGHT_ENABLE
@@ -332,7 +376,7 @@ int serial_transaction(void) {
     
     #ifdef BACKLIGHT_ENABLE
         // Write backlight level for slave to read
-        serial_master_buffer[SERIAL_BACKLIT_START] = backlight_config.enable ? backlight_config.level : 0;
+        serial_m2s_buffer.backlight_level = backlight_config.enable ? backlight_config.level : 0;
     #endif
 
     return 0;
@@ -375,8 +419,9 @@ void matrix_slave_scan(void) {
         i2c_slave_buffer[I2C_KEYMAP_START+i] = matrix[offset+i];
     }   
 #else // USE_SERIAL
+    // TODO: if MATRIX_COLS > 8 change to pack()
     for (int i = 0; i < ROWS_PER_HAND; ++i) {
-        serial_slave_buffer[i] = matrix[offset+i];
+        serial_s2m_buffer.smatrix[i] = matrix[offset+i];
     }
 #endif
     matrix_slave_scan_user();
diff --git a/quantum/split_common/matrix.c b/quantum/split_common/matrix.c
index d6359b51f..261137638 100644
--- a/quantum/split_common/matrix.c
+++ b/quantum/split_common/matrix.c
@@ -32,9 +32,6 @@ along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #include "timer.h"
 #include "split_flags.h"
 
-#ifdef RGBLIGHT_ENABLE
-#   include "rgblight.h"
-#endif
 #ifdef BACKLIGHT_ENABLE
 #   include "backlight.h"
     extern backlight_config_t backlight_config;
@@ -55,6 +52,8 @@ along with this program.  If not, see <http://www.gnu.org/licenses/>.
     static bool debouncing = false;
 #endif
 
+#if defined(USE_I2C) || defined(EH)
+
 #if (MATRIX_COLS <= 8)
 #    define print_matrix_header()  print("\nr/c 01234567\n")
 #    define print_matrix_row(row)  print_bin_reverse8(matrix_get_row(row))
@@ -63,6 +62,27 @@ along with this program.  If not, see <http://www.gnu.org/licenses/>.
 #else
 #    error "Currently only supports 8 COLS"
 #endif
+
+#else // USE_SERIAL
+
+#if (MATRIX_COLS <= 8)
+#    define print_matrix_header()  print("\nr/c 01234567\n")
+#    define print_matrix_row(row)  print_bin_reverse8(matrix_get_row(row))
+#    define matrix_bitpop(i)       bitpop(matrix[i])
+#    define ROW_SHIFTER ((uint8_t)1)
+#elif (MATRIX_COLS <= 16)
+#    define print_matrix_header()  print("\nr/c 0123456789ABCDEF\n")
+#    define print_matrix_row(row)  print_bin_reverse16(matrix_get_row(row))
+#    define matrix_bitpop(i)       bitpop16(matrix[i])
+#    define ROW_SHIFTER ((uint16_t)1)
+#elif (MATRIX_COLS <= 32)
+#    define print_matrix_header()  print("\nr/c 0123456789ABCDEF0123456789ABCDEF\n")
+#    define print_matrix_row(row)  print_bin_reverse32(matrix_get_row(row))
+#    define matrix_bitpop(i)       bitpop32(matrix[i])
+#    define ROW_SHIFTER  ((uint32_t)1)
+#endif
+
+#endif
 static matrix_row_t matrix_debouncing[MATRIX_ROWS];
 
 #define ERROR_DISCONNECT_COUNT 5
@@ -286,24 +306,48 @@ i2c_error: // the cable is disconnceted, or something else went wrong
 
 #else // USE_SERIAL
 
+
+typedef struct _Serial_s2m_buffer_t {
+    // TODO: if MATRIX_COLS > 8 change to uint8_t packed_matrix[] for pack/unpack
+    matrix_row_t smatrix[ROWS_PER_HAND];
+} Serial_s2m_buffer_t;
+
+volatile Serial_s2m_buffer_t serial_s2m_buffer = {};
+volatile Serial_m2s_buffer_t serial_m2s_buffer = {};
+uint8_t volatile status0 = 0;
+
+SSTD_t transactions[] = {
+    { (uint8_t *)&status0,
+      sizeof(serial_m2s_buffer), (uint8_t *)&serial_m2s_buffer,
+      sizeof(serial_s2m_buffer), (uint8_t *)&serial_s2m_buffer
+  }
+};
+
+void serial_master_init(void)
+{ soft_serial_initiator_init(transactions, TID_LIMIT(transactions)); }
+
+void serial_slave_init(void)
+{ soft_serial_target_init(transactions, TID_LIMIT(transactions)); }
+
 int serial_transaction(void) {
     int slaveOffset = (isLeftHand) ? (ROWS_PER_HAND) : 0;
 
-    if (serial_update_buffers()) {
+    if (soft_serial_transaction()) {
         return 1;
     }
 
+    // TODO:  if MATRIX_COLS > 8 change to unpack()
     for (int i = 0; i < ROWS_PER_HAND; ++i) {
-        matrix[slaveOffset+i] = serial_slave_buffer[i];
+        matrix[slaveOffset+i] = serial_s2m_buffer.smatrix[i];
     }
     
-    #ifdef RGBLIGHT_ENABLE
+    #if defined(RGBLIGHT_ENABLE) && defined(RGBLIGHT_SPLIT)
         // Code to send RGB over serial goes here (not implemented yet)
     #endif
     
     #ifdef BACKLIGHT_ENABLE
         // Write backlight level for slave to read
-        serial_master_buffer[SERIAL_BACKLIT_START] = backlight_config.enable ? backlight_config.level : 0;
+        serial_m2s_buffer.backlight_level = backlight_config.enable ? backlight_config.level : 0;
     #endif
 
     return 0;
@@ -346,8 +390,9 @@ void matrix_slave_scan(void) {
         i2c_slave_buffer[I2C_KEYMAP_START+i] = matrix[offset+i];
     }   
 #else // USE_SERIAL
+    // TODO: if MATRIX_COLS > 8 change to pack()
     for (int i = 0; i < ROWS_PER_HAND; ++i) {
-        serial_slave_buffer[i] = matrix[offset+i];
+        serial_s2m_buffer.smatrix[i] = matrix[offset+i];
     }
 #endif
     matrix_slave_scan_user();
diff --git a/quantum/split_common/matrix.h b/quantum/split_common/matrix.h
new file mode 100644
index 000000000..b5cb45bae
--- /dev/null
+++ b/quantum/split_common/matrix.h
@@ -0,0 +1,31 @@
+#ifndef SPLIT_COMMON_MATRIX_H
+#define SPLIT_COMMON_MATRIX_H
+
+#include <common/matrix.h>
+
+#ifdef RGBLIGHT_ENABLE
+#   include "rgblight.h"
+#endif
+
+typedef struct _Serial_m2s_buffer_t {
+#ifdef BACKLIGHT_ENABLE
+    uint8_t backlight_level;
+#endif
+#if defined(RGBLIGHT_ENABLE) && defined(RGBLIGHT_SPLIT)
+    rgblight_config_t rgblight_config; //not yet use
+    //
+    // When MCUs on both sides drive their respective RGB LED chains,
+    // it is necessary to synchronize, so it is necessary to communicate RGB information.
+    // In that case, define the RGBLIGHT_SPLIT macro.
+    //
+    // Otherwise, if the master side MCU drives both sides RGB LED chains,
+    // there is no need to communicate.
+#endif
+} Serial_m2s_buffer_t;
+
+extern volatile Serial_m2s_buffer_t serial_m2s_buffer;
+
+void serial_master_init(void);
+void serial_slave_init(void);
+
+#endif
diff --git a/quantum/split_common/serial.c b/quantum/split_common/serial.c
index 13e58d825..4df8d103b 100644
--- a/quantum/split_common/serial.c
+++ b/quantum/split_common/serial.c
@@ -1,5 +1,10 @@
 /*
  * WARNING: be careful changing this code, it is very timing dependent
+ *
+ * 2018-10-28 checked
+ *  avr-gcc 4.9.2
+ *  avr-gcc 5.4.0
+ *  avr-gcc 7.3.0
  */
 
 #ifndef F_CPU
@@ -9,17 +14,19 @@
 #include <avr/io.h>
 #include <avr/interrupt.h>
 #include <util/delay.h>
+#include <stddef.h>
 #include <stdbool.h>
 #include "serial.h"
+//#include <pro_micro.h>
 
 #ifdef SOFT_SERIAL_PIN
 
 #ifdef __AVR_ATmega32U4__
   // if using ATmega32U4 I2C, can not use PD0 and PD1 in soft serial.
   #ifdef USE_AVR_I2C
-     #if SOFT_SERIAL_PIN == D0 || SOFT_SERIAL_PIN == D1
-       #error Using ATmega32U4 I2C, so can not use PD0, PD1
-     #endif
+    #if SOFT_SERIAL_PIN == D0 || SOFT_SERIAL_PIN == D1
+      #error Using ATmega32U4 I2C, so can not use PD0, PD1
+    #endif
   #endif
 
   #if SOFT_SERIAL_PIN >= D0 && SOFT_SERIAL_PIN <= D3
@@ -63,215 +70,477 @@
  #error serial.c now support ATmega32U4 only
 #endif
 
-// Serial pulse period in microseconds. Its probably a bad idea to lower this
-// value.
-#define SERIAL_DELAY 24
+#define ALWAYS_INLINE __attribute__((always_inline))
+#define NO_INLINE __attribute__((noinline))
+#define _delay_sub_us(x)    __builtin_avr_delay_cycles(x)
+
+// parity check
+#define ODD_PARITY 1
+#define EVEN_PARITY 0
+#define PARITY EVEN_PARITY
+
+#ifdef SERIAL_DELAY
+  // custom setup in config.h
+  // #define TID_SEND_ADJUST 2
+  // #define SERIAL_DELAY 6             // micro sec
+  // #define READ_WRITE_START_ADJUST 30 // cycles
+  // #define READ_WRITE_WIDTH_ADJUST 8 // cycles
+#else
+// ============ Standard setups ============
+
+#ifndef SELECT_SOFT_SERIAL_SPEED
+#define SELECT_SOFT_SERIAL_SPEED 1
+//  0: about 189kbps (Experiment only)
+//  1: about 137kbps (default)
+//  2: about 75kbps
+//  3: about 39kbps
+//  4: about 26kbps
+//  5: about 20kbps
+#endif
 
-uint8_t volatile serial_slave_buffer[SERIAL_SLAVE_BUFFER_LENGTH] = {0};
-uint8_t volatile serial_master_buffer[SERIAL_MASTER_BUFFER_LENGTH] = {0};
+#if __GNUC__ < 6
+  #define TID_SEND_ADJUST 14
+#else
+  #define TID_SEND_ADJUST 2
+#endif
 
-#define SLAVE_DATA_CORRUPT (1<<0)
-volatile uint8_t status = 0;
+#if SELECT_SOFT_SERIAL_SPEED == 0
+  // Very High speed
+  #define SERIAL_DELAY 4             // micro sec
+  #if __GNUC__ < 6
+    #define READ_WRITE_START_ADJUST 33 // cycles
+    #define READ_WRITE_WIDTH_ADJUST 3 // cycles
+  #else
+    #define READ_WRITE_START_ADJUST 34 // cycles
+    #define READ_WRITE_WIDTH_ADJUST 7 // cycles
+  #endif
+#elif SELECT_SOFT_SERIAL_SPEED == 1
+  // High speed
+  #define SERIAL_DELAY 6             // micro sec
+  #if __GNUC__ < 6
+    #define READ_WRITE_START_ADJUST 30 // cycles
+    #define READ_WRITE_WIDTH_ADJUST 3 // cycles
+  #else
+    #define READ_WRITE_START_ADJUST 33 // cycles
+    #define READ_WRITE_WIDTH_ADJUST 7 // cycles
+  #endif
+#elif SELECT_SOFT_SERIAL_SPEED == 2
+  // Middle speed
+  #define SERIAL_DELAY 12            // micro sec
+  #define READ_WRITE_START_ADJUST 30 // cycles
+  #if __GNUC__ < 6
+    #define READ_WRITE_WIDTH_ADJUST 3 // cycles
+  #else
+    #define READ_WRITE_WIDTH_ADJUST 7 // cycles
+  #endif
+#elif SELECT_SOFT_SERIAL_SPEED == 3
+  // Low speed
+  #define SERIAL_DELAY 24            // micro sec
+  #define READ_WRITE_START_ADJUST 30 // cycles
+  #if __GNUC__ < 6
+    #define READ_WRITE_WIDTH_ADJUST 3 // cycles
+  #else
+    #define READ_WRITE_WIDTH_ADJUST 7 // cycles
+  #endif
+#elif SELECT_SOFT_SERIAL_SPEED == 4
+  // Very Low speed
+  #define SERIAL_DELAY 36            // micro sec
+  #define READ_WRITE_START_ADJUST 30 // cycles
+  #if __GNUC__ < 6
+    #define READ_WRITE_WIDTH_ADJUST 3 // cycles
+  #else
+    #define READ_WRITE_WIDTH_ADJUST 7 // cycles
+  #endif
+#elif SELECT_SOFT_SERIAL_SPEED == 5
+  // Ultra Low speed
+  #define SERIAL_DELAY 48            // micro sec
+  #define READ_WRITE_START_ADJUST 30 // cycles
+  #if __GNUC__ < 6
+    #define READ_WRITE_WIDTH_ADJUST 3 // cycles
+  #else
+    #define READ_WRITE_WIDTH_ADJUST 7 // cycles
+  #endif
+#else
+#error invalid SELECT_SOFT_SERIAL_SPEED value
+#endif /* SELECT_SOFT_SERIAL_SPEED */
+#endif /* SERIAL_DELAY */
+
+#define SERIAL_DELAY_HALF1 (SERIAL_DELAY/2)
+#define SERIAL_DELAY_HALF2 (SERIAL_DELAY - SERIAL_DELAY/2)
 
+#define SLAVE_INT_WIDTH_US 1
+#ifndef SERIAL_USE_MULTI_TRANSACTION
+  #define SLAVE_INT_RESPONSE_TIME SERIAL_DELAY
+#else
+  #define SLAVE_INT_ACK_WIDTH_UNIT 2
+  #define SLAVE_INT_ACK_WIDTH 4
+#endif
+
+static SSTD_t *Transaction_table = NULL;
+static uint8_t Transaction_table_size = 0;
+
+inline static void serial_delay(void) ALWAYS_INLINE;
 inline static
 void serial_delay(void) {
   _delay_us(SERIAL_DELAY);
 }
 
+inline static void serial_delay_half1(void) ALWAYS_INLINE;
+inline static
+void serial_delay_half1(void) {
+  _delay_us(SERIAL_DELAY_HALF1);
+}
+
+inline static void serial_delay_half2(void) ALWAYS_INLINE;
+inline static
+void serial_delay_half2(void) {
+  _delay_us(SERIAL_DELAY_HALF2);
+}
+
+inline static void serial_output(void) ALWAYS_INLINE;
 inline static
 void serial_output(void) {
   SERIAL_PIN_DDR |= SERIAL_PIN_MASK;
 }
 
 // make the serial pin an input with pull-up resistor
+inline static void serial_input_with_pullup(void) ALWAYS_INLINE;
 inline static
-void serial_input(void) {
+void serial_input_with_pullup(void) {
   SERIAL_PIN_DDR  &= ~SERIAL_PIN_MASK;
   SERIAL_PIN_PORT |= SERIAL_PIN_MASK;
 }
 
+inline static uint8_t serial_read_pin(void) ALWAYS_INLINE;
 inline static
 uint8_t serial_read_pin(void) {
   return !!(SERIAL_PIN_INPUT & SERIAL_PIN_MASK);
 }
 
+inline static void serial_low(void) ALWAYS_INLINE;
 inline static
 void serial_low(void) {
   SERIAL_PIN_PORT &= ~SERIAL_PIN_MASK;
 }
 
+inline static void serial_high(void) ALWAYS_INLINE;
 inline static
 void serial_high(void) {
   SERIAL_PIN_PORT |= SERIAL_PIN_MASK;
 }
 
-void serial_master_init(void) {
-  serial_output();
-  serial_high();
+void soft_serial_initiator_init(SSTD_t *sstd_table, int sstd_table_size)
+{
+    Transaction_table = sstd_table;
+    Transaction_table_size = (uint8_t)sstd_table_size;
+    serial_output();
+    serial_high();
 }
 
-void serial_slave_init(void) {
-  serial_input();
-
-  // Enable INT0
-  EIMSK |= _BV(INT0);
-  // Trigger on falling edge of INT0
-  EICRA &= ~(_BV(ISC00) | _BV(ISC01));
+void soft_serial_target_init(SSTD_t *sstd_table, int sstd_table_size)
+{
+    Transaction_table = sstd_table;
+    Transaction_table_size = (uint8_t)sstd_table_size;
+    serial_input_with_pullup();
+
+    // Enable INT0-INT3,INT6
+    EIMSK |= EIMSK_BIT;
+#if SERIAL_PIN_MASK == _BV(PE6)
+    // Trigger on falling edge of INT6
+    EICRB &= EICRx_BIT;
+#else
+    // Trigger on falling edge of INT0-INT3
+    EICRA &= EICRx_BIT;
+#endif
 }
 
-// Used by the master to synchronize timing with the slave.
+// Used by the sender to synchronize timing with the reciver.
+static void sync_recv(void) NO_INLINE;
 static
 void sync_recv(void) {
-  serial_input();
-  // This shouldn't hang if the slave disconnects because the
-  // serial line will float to high if the slave does disconnect.
+  for (uint8_t i = 0; i < SERIAL_DELAY*5 && serial_read_pin(); i++ ) {
+  }
+  // This shouldn't hang if the target disconnects because the
+  // serial line will float to high if the target does disconnect.
   while (!serial_read_pin());
-  serial_delay();
 }
 
-// Used by the slave to send a synchronization signal to the master.
+// Used by the reciver to send a synchronization signal to the sender.
+static void sync_send(void) NO_INLINE;
 static
 void sync_send(void) {
-  serial_output();
-
   serial_low();
   serial_delay();
-
   serial_high();
 }
 
 // Reads a byte from the serial line
-static
-uint8_t serial_read_byte(void) {
-  uint8_t byte = 0;
-  serial_input();
-  for ( uint8_t i = 0; i < 8; ++i) {
-    byte = (byte << 1) | serial_read_pin();
-    serial_delay();
-    _delay_us(1);
+static uint8_t serial_read_chunk(uint8_t *pterrcount, uint8_t bit) NO_INLINE;
+static uint8_t serial_read_chunk(uint8_t *pterrcount, uint8_t bit) {
+    uint8_t byte, i, p, pb;
+
+  _delay_sub_us(READ_WRITE_START_ADJUST);
+  for( i = 0, byte = 0, p = PARITY; i < bit; i++ ) {
+      serial_delay_half1();   // read the middle of pulses
+      if( serial_read_pin() ) {
+          byte = (byte << 1) | 1; p ^= 1;
+      } else {
+          byte = (byte << 1) | 0; p ^= 0;
+      }
+      _delay_sub_us(READ_WRITE_WIDTH_ADJUST);
+      serial_delay_half2();
   }
+  /* recive parity bit */
+  serial_delay_half1();   // read the middle of pulses
+  pb = serial_read_pin();
+  _delay_sub_us(READ_WRITE_WIDTH_ADJUST);
+  serial_delay_half2();
+
+  *pterrcount += (p != pb)? 1 : 0;
 
   return byte;
 }
 
 // Sends a byte with MSB ordering
-static
-void serial_write_byte(uint8_t data) {
-  uint8_t b = 8;
-  serial_output();
-  while( b-- ) {
-    if(data & (1 << b)) {
-      serial_high();
-    } else {
-      serial_low();
+void serial_write_chunk(uint8_t data, uint8_t bit) NO_INLINE;
+void serial_write_chunk(uint8_t data, uint8_t bit) {
+    uint8_t b, p;
+    for( p = PARITY, b = 1<<(bit-1); b ; b >>= 1) {
+        if(data & b) {
+            serial_high(); p ^= 1;
+        } else {
+            serial_low();  p ^= 0;
+        }
+        serial_delay();
     }
+    /* send parity bit */
+    if(p & 1) { serial_high(); }
+    else      { serial_low(); }
     serial_delay();
-  }
-}
 
-// interrupt handle to be used by the slave device
-ISR(SERIAL_PIN_INTERRUPT) {
-  sync_send();
+    serial_low(); // sync_send() / senc_recv() need raise edge
+}
 
-  uint8_t checksum = 0;
-  for (int i = 0; i < SERIAL_SLAVE_BUFFER_LENGTH; ++i) {
-    serial_write_byte(serial_slave_buffer[i]);
+static void serial_send_packet(uint8_t *buffer, uint8_t size) NO_INLINE;
+static
+void serial_send_packet(uint8_t *buffer, uint8_t size) {
+  for (uint8_t i = 0; i < size; ++i) {
+    uint8_t data;
+    data = buffer[i];
     sync_send();
-    checksum += serial_slave_buffer[i];
+    serial_write_chunk(data,8);
   }
-  serial_write_byte(checksum);
-  sync_send();
+}
 
-  // wait for the sync to finish sending
-  serial_delay();
+static uint8_t serial_recive_packet(uint8_t *buffer, uint8_t size) NO_INLINE;
+static
+uint8_t serial_recive_packet(uint8_t *buffer, uint8_t size) {
+  uint8_t pecount = 0;
+  for (uint8_t i = 0; i < size; ++i) {
+    uint8_t data;
+    sync_recv();
+    data = serial_read_chunk(&pecount, 8);
+    buffer[i] = data;
+  }
+  return pecount == 0;
+}
 
-  // read the middle of pulses
-  _delay_us(SERIAL_DELAY/2);
+inline static
+void change_sender2reciver(void) {
+    sync_send();          //0
+    serial_delay_half1(); //1
+    serial_low();         //2
+    serial_input_with_pullup(); //2
+    serial_delay_half1(); //3
+}
 
-  uint8_t checksum_computed = 0;
-  for (int i = 0; i < SERIAL_MASTER_BUFFER_LENGTH; ++i) {
-    serial_master_buffer[i] = serial_read_byte();
-    sync_send();
-    checksum_computed += serial_master_buffer[i];
+inline static
+void change_reciver2sender(void) {
+    sync_recv();     //0
+    serial_delay();  //1
+    serial_low();    //3
+    serial_output(); //3
+    serial_delay_half1(); //4
+}
+
+static inline uint8_t nibble_bits_count(uint8_t bits)
+{
+    bits = (bits & 0x5) + (bits >> 1 & 0x5);
+    bits = (bits & 0x3) + (bits >> 2 & 0x3);
+    return bits;
+}
+
+// interrupt handle to be used by the target device
+ISR(SERIAL_PIN_INTERRUPT) {
+
+#ifndef SERIAL_USE_MULTI_TRANSACTION
+  serial_low();
+  serial_output();
+  SSTD_t *trans = Transaction_table;
+#else
+  // recive transaction table index
+  uint8_t tid, bits;
+  uint8_t pecount = 0;
+  sync_recv();
+  bits = serial_read_chunk(&pecount,7);
+  tid = bits>>3;
+  bits = (bits&7) != nibble_bits_count(tid);
+  if( bits || pecount> 0 || tid > Transaction_table_size ) {
+      return;
   }
-  uint8_t checksum_received = serial_read_byte();
-  sync_send();
+  serial_delay_half1();
 
-  serial_input(); // end transaction
+  serial_high(); // response step1 low->high
+  serial_output();
+  _delay_sub_us(SLAVE_INT_ACK_WIDTH_UNIT*SLAVE_INT_ACK_WIDTH);
+  SSTD_t *trans = &Transaction_table[tid];
+  serial_low(); // response step2 ack high->low
+#endif
 
-  if ( checksum_computed != checksum_received ) {
-    status |= SLAVE_DATA_CORRUPT;
+  // target send phase
+  if( trans->target2initiator_buffer_size > 0 )
+      serial_send_packet((uint8_t *)trans->target2initiator_buffer,
+                         trans->target2initiator_buffer_size);
+  // target switch to input
+  change_sender2reciver();
+
+  // target recive phase
+  if( trans->initiator2target_buffer_size > 0 ) {
+      if (serial_recive_packet((uint8_t *)trans->initiator2target_buffer,
+                               trans->initiator2target_buffer_size) ) {
+          *trans->status = TRANSACTION_ACCEPTED;
+      } else {
+          *trans->status = TRANSACTION_DATA_ERROR;
+      }
   } else {
-    status &= ~SLAVE_DATA_CORRUPT;
+      *trans->status = TRANSACTION_ACCEPTED;
   }
-}
 
-inline
-bool serial_slave_DATA_CORRUPT(void) {
-  return status & SLAVE_DATA_CORRUPT;
+  sync_recv(); //weit initiator output to high
 }
 
-// Copies the serial_slave_buffer to the master and sends the
-// serial_master_buffer to the slave.
+/////////
+//  start transaction by initiator
+//
+// int  soft_serial_transaction(int sstd_index)
 //
 // Returns:
-// 0 => no error
-// 1 => slave did not respond
-int serial_update_buffers(void) {
-  // this code is very time dependent, so we need to disable interrupts
+//    TRANSACTION_END
+//    TRANSACTION_NO_RESPONSE
+//    TRANSACTION_DATA_ERROR
+// this code is very time dependent, so we need to disable interrupts
+#ifndef SERIAL_USE_MULTI_TRANSACTION
+int  soft_serial_transaction(void) {
+  SSTD_t *trans = Transaction_table;
+#else
+int  soft_serial_transaction(int sstd_index) {
+  if( sstd_index > Transaction_table_size )
+      return TRANSACTION_TYPE_ERROR;
+  SSTD_t *trans = &Transaction_table[sstd_index];
+#endif
   cli();
 
-  // signal to the slave that we want to start a transaction
+  // signal to the target that we want to start a transaction
   serial_output();
   serial_low();
-  _delay_us(1);
+  _delay_us(SLAVE_INT_WIDTH_US);
 
-  // wait for the slaves response
-  serial_input();
-  serial_high();
-  _delay_us(SERIAL_DELAY);
+#ifndef SERIAL_USE_MULTI_TRANSACTION
+  // wait for the target response
+  serial_input_with_pullup();
+  _delay_us(SLAVE_INT_RESPONSE_TIME);
 
-  // check if the slave is present
+  // check if the target is present
   if (serial_read_pin()) {
-    // slave failed to pull the line low, assume not present
+    // target failed to pull the line low, assume not present
+    serial_output();
+    serial_high();
+    *trans->status = TRANSACTION_NO_RESPONSE;
     sei();
-    return 1;
+    return TRANSACTION_NO_RESPONSE;
   }
 
-  // if the slave is present syncronize with it
-  sync_recv();
-
-  uint8_t checksum_computed = 0;
-  // receive data from the slave
-  for (int i = 0; i < SERIAL_SLAVE_BUFFER_LENGTH; ++i) {
-    serial_slave_buffer[i] = serial_read_byte();
-    sync_recv();
-    checksum_computed += serial_slave_buffer[i];
+#else
+  // send transaction table index
+  int tid = (sstd_index<<3) | (7 & nibble_bits_count(sstd_index));
+  sync_send();
+  _delay_sub_us(TID_SEND_ADJUST);
+  serial_write_chunk(tid, 7);
+  serial_delay_half1();
+
+  // wait for the target response (step1 low->high)
+  serial_input_with_pullup();
+  while( !serial_read_pin() ) {
+      _delay_sub_us(2);
   }
-  uint8_t checksum_received = serial_read_byte();
-  sync_recv();
 
-  if (checksum_computed != checksum_received) {
-    sei();
-    return 1;
+  // check if the target is present (step2 high->low)
+  for( int i = 0; serial_read_pin(); i++ ) {
+      if (i > SLAVE_INT_ACK_WIDTH + 1) {
+          // slave failed to pull the line low, assume not present
+          serial_output();
+          serial_high();
+          *trans->status = TRANSACTION_NO_RESPONSE;
+          sei();
+          return TRANSACTION_NO_RESPONSE;
+      }
+      _delay_sub_us(SLAVE_INT_ACK_WIDTH_UNIT);
   }
+#endif
 
-  uint8_t checksum = 0;
-  // send data to the slave
-  for (int i = 0; i < SERIAL_MASTER_BUFFER_LENGTH; ++i) {
-    serial_write_byte(serial_master_buffer[i]);
-    sync_recv();
-    checksum += serial_master_buffer[i];
+  // initiator recive phase
+  // if the target is present syncronize with it
+  if( trans->target2initiator_buffer_size > 0 ) {
+      if (!serial_recive_packet((uint8_t *)trans->target2initiator_buffer,
+                                trans->target2initiator_buffer_size) ) {
+          serial_output();
+          serial_high();
+          *trans->status = TRANSACTION_DATA_ERROR;
+          sei();
+          return TRANSACTION_DATA_ERROR;
+      }
+   }
+
+  // initiator switch to output
+  change_reciver2sender();
+
+  // initiator send phase
+  if( trans->initiator2target_buffer_size > 0 ) {
+      serial_send_packet((uint8_t *)trans->initiator2target_buffer,
+                         trans->initiator2target_buffer_size);
   }
-  serial_write_byte(checksum);
-  sync_recv();
 
   // always, release the line when not in use
-  serial_output();
-  serial_high();
+  sync_send();
 
+  *trans->status = TRANSACTION_END;
   sei();
-  return 0;
+  return TRANSACTION_END;
 }
 
-#endif /* SOFT_SERIAL_PIN */
+#ifdef SERIAL_USE_MULTI_TRANSACTION
+int soft_serial_get_and_clean_status(int sstd_index) {
+    SSTD_t *trans = &Transaction_table[sstd_index];
+    cli();
+    int retval = *trans->status;
+    *trans->status = 0;;
+    sei();
+    return retval;
+}
+#endif
+
+#endif
+
+// Helix serial.c history
+//   2018-1-29 fork from let's split and add PD2, modify sync_recv() (#2308, bceffdefc)
+//   2018-6-28 bug fix master to slave comm and speed up (#3255, 1038bbef4)
+//             (adjusted with avr-gcc 4.9.2)
+//   2018-7-13 remove USE_SERIAL_PD2 macro (#3374, f30d6dd78)
+//             (adjusted with avr-gcc 4.9.2)
+//   2018-8-11 add support multi-type transaction (#3608, feb5e4aae)
+//             (adjusted with avr-gcc 4.9.2)
+//   2018-10-21 fix serial and RGB animation conflict (#4191, 4665e4fff)
+//             (adjusted with avr-gcc 7.3.0)
+//   2018-10-28 re-adjust compiler depend value of delay (#4269, 8517f8a66)
+//             (adjusted with avr-gcc 5.4.0, 7.3.0)
+//   2018-12-17 copy to TOP/quantum/split_common/ and remove backward compatibility code (#4669)
diff --git a/quantum/split_common/serial.h b/quantum/split_common/serial.h
index 0b99f352d..b6638b3bd 100644
--- a/quantum/split_common/serial.h
+++ b/quantum/split_common/serial.h
@@ -1,5 +1,5 @@
-#ifndef MY_SERIAL_H
-#define MY_SERIAL_H
+#ifndef SOFT_SERIAL_H
+#define SOFT_SERIAL_H
 
 #include <stdbool.h>
 
@@ -7,20 +7,59 @@
 // Need Soft Serial defines in config.h
 // /////////////////////////////////////////////////////////////////
 // ex.
-//  /* Configuration of lower interface with the lower layer(hardware) of serial.c */
 //  #define SOFT_SERIAL_PIN ??   // ?? = D0,D1,D2,D3,E6
+//  OPTIONAL: #define SELECT_SOFT_SERIAL_SPEED ? // ? = 1,2,3,4,5
+//                                               //  1: about 137kbps (default)
+//                                               //  2: about 75kbps
+//                                               //  3: about 39kbps
+//                                               //  4: about 26kbps
+//                                               //  5: about 20kbps
 //
-//  /* Configuration of upper interface with the upper layer of serial.c */
-//  #define SERIAL_SLAVE_BUFFER_LENGTH MATRIX_ROWS/2
-//  #define SERIAL_MASTER_BUFFER_LENGTH 1
+// //// USE simple API (using signle-type transaction function)
+//   /* nothing */
+// //// USE flexible API (using multi-type transaction function)
+//   #define SERIAL_USE_MULTI_TRANSACTION
+//
+// /////////////////////////////////////////////////////////////////
 
-// Buffers for master - slave communication
-extern volatile uint8_t serial_slave_buffer[SERIAL_SLAVE_BUFFER_LENGTH];
-extern volatile uint8_t serial_master_buffer[SERIAL_MASTER_BUFFER_LENGTH];
+// Soft Serial Transaction Descriptor
+typedef struct _SSTD_t  {
+    uint8_t *status;
+    uint8_t initiator2target_buffer_size;
+    uint8_t *initiator2target_buffer;
+    uint8_t target2initiator_buffer_size;
+    uint8_t *target2initiator_buffer;
+} SSTD_t;
+#define TID_LIMIT( table ) (sizeof(table) / sizeof(SSTD_t))
 
-void serial_master_init(void);
-void serial_slave_init(void);
-int serial_update_buffers(void);
-bool serial_slave_data_corrupt(void);
+// initiator is transaction start side
+void soft_serial_initiator_init(SSTD_t *sstd_table, int sstd_table_size);
+// target is interrupt accept side
+void soft_serial_target_init(SSTD_t *sstd_table, int sstd_table_size);
 
+// initiator resullt
+#define TRANSACTION_END 0
+#define TRANSACTION_NO_RESPONSE 0x1
+#define TRANSACTION_DATA_ERROR  0x2
+#define TRANSACTION_TYPE_ERROR  0x4
+#ifndef SERIAL_USE_MULTI_TRANSACTION
+int  soft_serial_transaction(void);
+#else
+int  soft_serial_transaction(int sstd_index);
 #endif
+
+// target status
+// *SSTD_t.status has
+//   initiator:
+//       TRANSACTION_END
+//    or TRANSACTION_NO_RESPONSE
+//    or TRANSACTION_DATA_ERROR
+//   target:
+//       TRANSACTION_DATA_ERROR
+//    or TRANSACTION_ACCEPTED
+#define TRANSACTION_ACCEPTED 0x8
+#ifdef SERIAL_USE_MULTI_TRANSACTION
+int  soft_serial_get_and_clean_status(int sstd_index);
+#endif
+
+#endif /* SOFT_SERIAL_H */
diff --git a/quantum/split_common/serial_backward_compatibility.h b/quantum/split_common/serial_backward_compatibility.h
deleted file mode 100644
index 216b6c41f..000000000
--- a/quantum/split_common/serial_backward_compatibility.h
+++ /dev/null
@@ -1,11 +0,0 @@
-/* serial.h backward compatibility */
-
-// #ifndef SOFT_SERIAL_PIN
-//  #define SOFT_SERIAL_PIN D0
-// #endif
-
-#ifndef SERIAL_SLAVE_BUFFER_LENGTH
- #define SERIAL_SLAVE_BUFFER_LENGTH MATRIX_ROWS/2
- #define SERIAL_MASTER_BUFFER_LENGTH 1
-#endif
-
diff --git a/quantum/split_common/split_util.c b/quantum/split_common/split_util.c
index 56077e4f2..e4475c535 100644
--- a/quantum/split_common/split_util.c
+++ b/quantum/split_common/split_util.c
@@ -11,9 +11,6 @@
 #include "timer.h"
 #include "split_flags.h"
 
-#ifdef RGBLIGHT_ENABLE
-#   include "rgblight.h"
-#endif
 #ifdef BACKLIGHT_ENABLE
 #   include "backlight.h"
 #endif
@@ -24,8 +21,6 @@
 
 #if defined(USE_I2C) || defined(EH)
 #  include "i2c.h"
-#else
-#  include "serial.h"
 #endif
 
 volatile bool isLeftHand = true;
@@ -112,7 +107,7 @@ void keyboard_slave_loop(void) {
                 BACKLIT_DIRTY = false;
             }
         #else // USE_SERIAL
-            backlight_set(serial_master_buffer[SERIAL_BACKLIT_START]);
+            backlight_set(serial_m2s_buffer.backlight_level);
         #endif
     #endif
     // Read RGB Info
@@ -137,7 +132,9 @@ void keyboard_slave_loop(void) {
                 sei();
             }
         #else // USE_SERIAL
+          #ifdef RGBLIGHT_SPLIT
             // Add serial implementation for RGB here
+          #endif
         #endif
     #endif
    }
diff --git a/quantum/split_common/split_util.h b/quantum/split_common/split_util.h
index 45e8db0f2..d6cf3e72a 100644
--- a/quantum/split_common/split_util.h
+++ b/quantum/split_common/split_util.h
@@ -7,9 +7,6 @@
 #include <stdlib.h>
 #include "eeconfig.h"
 
-// backlight level store index in serial_master_buffer[] for slave to read
-#define SERIAL_BACKLIT_START   0x00
-
 #define SLAVE_I2C_ADDRESS           0x32
 
 extern volatile bool isLeftHand;