Add register parser and binary search for opcodes

12 files changed, 196 insertions, 33 deletions

diff --git a/newasm/Makefile b/newasm/Makefile
index 7b31eb4..fd057fe 100644
--- a/newasm/Makefile
+++ b/newasm/Makefile
@@ -11,7 +11,13 @@ parse.bin: parse.elf
 	arm-none-eabi-objcopy -O binary parse.elf parse.bin
 
 objects = main.o uart.o data.o string.o input.o \
-					statement.o whitespace.o label.o symbol.o opcode.o
+					statement.o optable.o \
+					parsers/whitespace.o \
+					parsers/label.o \
+					parsers/symbol.o \
+					parsers/opcode.o \
+					parsers/register.o \
+					parsers/decimal.o
 
 parse.elf: pico_bin.ld $(objects)
 	arm-none-eabi-ld -T pico_bin.ld -o parse.elf $(objects)
diff --git a/newasm/input.s b/newasm/input.s
index f3db1e9..7644a81 100644
--- a/newasm/input.s
+++ b/newasm/input.s
@@ -8,6 +8,7 @@
 getline:
         PUSH    {R4, R5, LR}    // save registers
         MOVS    R4, R0          // copy buffer start address
+        MOVS    R5, 0           // beginning offset
 next:   BL      uart_recv       // get a char
         CMP     R0, 0x03        // end of text (^C)
         BEQ     cancel          // don't submit, start on next line
diff --git a/newasm/main.s b/newasm/main.s
index 57a1b98..a27fec9 100644
--- a/newasm/main.s
+++ b/newasm/main.s
@@ -5,33 +5,34 @@
 .type main, %function
 .global main, strbuf
 
-main:     LDR     R4, =0x20002000
+main:     LDR     R5, =0x20002000
           BL      uart_recv
 loop:     BL      prompt
           LDR     R0, inpbuf
           BL      getline
           LDR     R0, inpbuf
           BL      putstrln
-          LDR     R1, inpbuf
+          LDR     R4, inpbuf
           LDR     R2, strbuf
           MOVS    R0, 0
           STRB    R0, [R2]
           BL      statement
           BNE     bad
+
 good:     ADR     R0, success
           PUSH    {R1}
           BL      putstrln
-          POP     {R0}
-          BL      putstrln
-          LDR     R0, strbuf
+          MOVS    R0, R4
           BL      putstrln
+          POP     {R0}
+          BL      send_hex
+          LDR     R0, =crlf
+          BL      putstr
           B       loop
+
 bad:      ADR     R0, fail
-          PUSH    {R1}
-          BL      putstrln
-          POP     {R0}
           BL      putstrln
-          LDR     R0, strbuf
+          MOVS    R0, R4
           BL      putstrln
           B       loop
 never:    BL      uart_recv
@@ -39,11 +40,11 @@ never:    BL      uart_recv
           BX      R0
 
 prompt:   PUSH    {LR}
-          MOVS    R0, R4
+          MOVS    R0, R5
           BL      send_hex
           MOVS    R0, ' 
           BL      uart_send
-          LDR     R0, [R4]
+          LDR     R0, [R5]
           BL      send_hex
           MOVS    R0, ' 
           BL      uart_send
diff --git a/newasm/optable.s b/newasm/optable.s
new file mode 100644
index 0000000..a063f02
--- /dev/null
+++ b/newasm/optable.s
@@ -0,0 +1,57 @@
+.syntax unified
+.cpu cortex-m0plus
+.thumb
+
+.type lookup, %function
+.global lookup
+
+// 1 unexpected char
+// 2 unexpected end of input
+
+// input:
+// R0 input string (unboxed)
+
+// output:
+// R0 return code
+// R1 instruction template
+// R2 instruction type
+// R3 special code
+
+lookup:   PUSH    {R4, R5, R6}
+          ADR     R4, op_table// op table base address
+          MOVS    R5, 0       // begin pointer
+          MOVS    R6, 7       // end pointer
+loop:     CMP     R5, R6      // if begin == end then range = 0
+          BEQ     error
+          ADDS    R2, R5, R6  // get sum
+          LSRS    R2, 1       // divide by two
+no_jiggle:LSLS    R3, R2, 3   // times 8 (index to offset)
+          LDR     R1, [R4, R3]// read opcode from optable
+          CMP     R0, R1      // compare opcode to input
+          BLO     lower       // if lower
+          BHI     higher      // if higher
+found:    LSLS    R2, 3       // times 8 (index to offset)
+          ADDS    R4, R2      // select row
+          MOVS    R0, 0       // success return code
+          LDRH    R1, [R4, 4] // get machine code template
+          LDRB    R2, [R4, 6] // get instruction type
+          LDRB    R3, [R4, 7] // get special code
+          POP     {R4, R5, R6}
+          BX      LR          // result in R2
+lower:    MOVS    R6, R2      // update end pointer
+          B       loop
+higher:   ADDS    R5, R2, 1   // update begin pointer
+          B     loop
+error:    MOVS    R0, 1       // return code 1 (not found)
+          POP     {R4, R5, R6}
+          BX      LR
+
+          .align  8
+op_table:
+          .ascii  "ADD3"; .hword 0x1C00;  .byte 0x01, 0x03  // instr type 1, special code 3 (imm width)
+          .ascii  "ADD8"; .hword 0x3000;  .byte 0x02, 0x08  // instr type 2, special code 8 (imm width)
+          .ascii  "ASPI"; .hword 0xB000;  .byte 0x04, 0x07  // instr type 4, special code 7
+          .ascii  "ASRI"; .hword 0x1000;  .byte 0x01, 0x05  // instr type 1, special code 5 (imm width)
+          .ascii  "ADCS"; .hword 0x4140;  .byte 0x00, 0x00  // instr type 0, special code 0
+          .ascii  "ADDS"; .hword 0x1400;  .byte 0x03, 0x00  // instr type 3, special code 0
+          .ascii  "BKPT"; .hword 0xBE00;  .byte 0x04, 0x08  // instr type 4, special code 8
diff --git a/newasm/parsers/decimal.s b/newasm/parsers/decimal.s
new file mode 100644
index 0000000..a1516d2
--- /dev/null
+++ b/newasm/parsers/decimal.s
@@ -0,0 +1,39 @@
+.syntax unified
+.cpu cortex-m0plus
+.thumb
+
+.type decimal, %function
+.global decimal
+
+// 1 not a digit
+
+// R4 input stream
+// R2 output value
+
+decimal:  LDRB    R0, [R4]    // get a char
+          CMP     R0, '0      // check if zero
+          BNE     notzero
+          ADDS    R4, 1       // consume the char
+          MOVS    R2, 0       // return value of zero
+success:  MOVS    R0, #0      // return code zero (success)
+          BX      LR
+notzero:  CMP     R0, '1      // if not [1-9] then error
+          BLO     bad
+          CMP     R0, '9
+          BHI     bad
+          ADDS    R4, 1       // consume the first digit
+          SUBS    R0, '0      // calculate the value
+          MOVS    R2, R0      // store it in R2
+loop:     LDRB    R0, [R4]    // get another char
+          CMP     R0, '0      // if not [0-9] then done
+          BLO     success
+          CMP     R0, '9
+          BHI     success
+          ADDS    R4, 1       // consume the additional digit
+          SUBS    R0, '0      // calculate the value
+          MOVS    R3, 10      // base 10
+          MULS    R2, R3      // shift result by one decimal place
+          ADDS    R2, R0      // accumulate into R2
+          B       loop        // keep getting digits
+bad:      MOVS    R0, #1      // return code 1 (not a digit)
+          BX      LR
diff --git a/newasm/label.s b/newasm/parsers/label.s
index af2f641..4770e6c 100644
--- a/newasm/label.s
+++ b/newasm/parsers/label.s
@@ -8,11 +8,11 @@
 // 1 unexpected begin char
 // 2 doesn't end with colon
 
-// R1 input buffer
+// R4 input buffer
 // R2 output buffer
 
 label:    PUSH    {LR}
-          LDRB    R0, [R1]    // get a char
+          LDRB    R0, [R4]    // get a char
           CMP     R0, 0x61    // a
           BLO     1f
           CMP     R0, 0x7A    // z
@@ -20,11 +20,11 @@ label:    PUSH    {LR}
 1:        MOVS    R0, #1      // return code 1 (expected lowercase)
           POP     {PC}
 2:        BL      symbol
-          LDRB    R0, [R1]    // get a char
+          LDRB    R0, [R4]    // get a char
           CMP     R0, ':      // colon
           BEQ     3f
           MOVS    R0, #2      // return code 2 (expected colon)
           POP     {PC}
-3:        ADDS    R1, 1       // consume the colon
+3:        ADDS    R4, 1       // consume the colon
           MOVS    R0, #0      // return code 0 (success)
           POP     {PC}
diff --git a/newasm/opcode.s b/newasm/parsers/opcode.s
index 71a5201..8ec327d 100644
--- a/newasm/opcode.s
+++ b/newasm/parsers/opcode.s
@@ -7,21 +7,21 @@
 
 // 1 unexpected first char
 
-// R1 input buffer
+// R4 input buffer
 // R2 output buffer
 
 opcode:   PUSH    {LR}
-          LDRB    R0, [R1]    // get a char
+          LDRB    R0, [R4]    // get a char
           CMP     R0, 0x41    // A
           BLO     1f
           CMP     R0, 0x5A    // Z
           BLS     2f
 1:        MOVS    R0, #1      // unexpected char
           POP     {PC}
-2:        ADDS    R1, 1       // consume the character
+2:        ADDS    R4, 1       // consume the character
           STRB    R0, [R2]    // store in temp buffer
           ADDS    R2, 1       // advance temp buffer pointer
-          LDRB    R0, [R1]    // get another character
+          LDRB    R0, [R4]    // get another character
           BL      goodchar    // check if valid symbol char
           BEQ     2b          // if so keep getting chars
           MOVS    R0, #0      // return code success
diff --git a/newasm/parsers/register.s b/newasm/parsers/register.s
new file mode 100644
index 0000000..48ebf7a
--- /dev/null
+++ b/newasm/parsers/register.s
@@ -0,0 +1,46 @@
+.syntax unified
+.cpu cortex-m0plus
+.thumb
+
+.type register, %function
+.global register
+
+// 1 unexpected char
+// 2 invalid general-purpose register number
+
+// R4 input buffer
+// R2 output buffer
+
+register: PUSH    {LR}
+          LDRB    R0, [R4]
+r_test:   CMP     R0, 'R
+          BNE     special
+          ADDS    R4, 1       // consume R
+          BL      decimal     // consume a decimal number (result in R2)
+          BEQ     validate
+          POP     {PC}        // error code already in R0
+validate: CMP     R2, #12     // general purpose registers 0-12
+          BLS     success
+          MOVS    R0, #2      // invalid register number error code
+          POP     {PC}
+special:  LDRH    R0, [R4]      // get two bytes from input stream
+          ADR     R2, table     // get address of table
+          MOVS    R3, 0         // set table offset to 0
+loop:     LDRH    R1, [R2, R3]  // get two bytes from table at current offset
+          CMP     R0, R1        // compare input to table row
+          BEQ     done          // if equal then done
+          ADDS    R3, 2         // increment table offset
+          CMP     R3, 6         // compare offset to table size
+          BLO     loop          // loop until end of table
+          MOVS    R0, #1        // return code 1 (unexpected char)
+          POP     {PC}
+done:     ADDS    R4, 2         // consume two chars
+          LSRS    R3, 1         // divide table offset by two to row
+          ADDS    R3, 13        // add 13 to get register number
+          MOVS    R2, R3
+success:  MOVS    R0, #0        // return code 0 (success)
+          POP     {PC}
+
+table:    .ascii  "SP"
+          .ascii  "LR"
+          .ascii  "PC"
diff --git a/newasm/symbol.s b/newasm/parsers/symbol.s
index 0677f6a..f57d493 100644
--- a/newasm/symbol.s
+++ b/newasm/parsers/symbol.s
@@ -7,19 +7,19 @@
 
 // 1 unexpected begin char
 
-// R1 input buffer
+// R4 input buffer
 // R2 output buffer
 
 symbol:   PUSH    {LR}
-          LDRB    R0, [R1]    // get a char
+          LDRB    R0, [R4]    // get a char
           BL      goodchar    // check if valid symbol char
           BEQ     loop
           MOVS    R0, #1      // unexpected begin char
           POP     {PC}
-loop:     ADDS    R1, 1       // consume the character
+loop:     ADDS    R4, 1       // consume the character
           STRB    R0, [R2]    // store in temp buffer
           ADDS    R2, 1       // advance temp buffer pointer
-          LDRB    R0, [R1]    // get another character
+          LDRB    R0, [R4]    // get another character
           BL      goodchar    // check if valid symbol char
           BEQ     loop        // if so keep getting chars
           MOVS    R0, #0      // return code success
diff --git a/newasm/whitespace.s b/newasm/parsers/whitespace.s
index f4d8d9f..708cab3 100644
--- a/newasm/whitespace.s
+++ b/newasm/parsers/whitespace.s
@@ -7,14 +7,16 @@
 
 // 1 unexpected char
 
+// R4 input buffer
+
 whitespace:
-          LDRB    R0, [R1]    // get a char
+          LDRB    R0, [R4]    // get a char
           CMP     R0, '       // space
           BEQ     1f
           MOVS    R0, #1      // return code 1 (unexpected char)
           BX      LR
-1:        ADDS    R1, 1       // consume the character
-          LDRB    R0, [R1]    // get another character
+1:        ADDS    R4, 1       // consume the character
+          LDRB    R0, [R4]    // get another character
           CMP     R0, '       // check if space
           BEQ     1b          // if so keep getting chars
           MOVS    R0, #0      // return code 0 (success)
diff --git a/newasm/statement.s b/newasm/statement.s
index 83f8d8d..4a199a6 100644
--- a/newasm/statement.s
+++ b/newasm/statement.s
@@ -7,24 +7,31 @@
 
 // 1 reported an error
 
-// R1 input buffer
+// R4 input buffer
 // R2 output buffer
 
 statement:PUSH    {LR}
           BL      whitespace
+          PUSH    {R2}
           BL      label
+          POP     {R2}
           CMP     R0, #2        // check for error code 2 (no colon)
           BNE     1f
           ADR     R0, no_colon
           B       err_exit
 1:        BL      whitespace
+          PUSH    {R2}
           BL      opcode
+          POP     {R2}
           BEQ     2f
           ADR     R0, no_opcode
           B       err_exit
-2:        MOVS    R0, #0
-          POP     {PC}
-
+2:        LDR     R0, [R2]
+          BL      lookup
+          BEQ     3f
+          ADR     R0, not_found
+          B       err_exit
+3:        POP     {PC}          // success code already in R0
 err_exit: BL      putstrln
           MOVS    R0, #1        // return code 1 (there was an error)
           POP     {PC}
@@ -33,3 +40,5 @@ err_exit: BL      putstrln
 no_colon:   .asciz  "Error: Expected colon at end of label"
             .align  4
 no_opcode:  .asciz  "Error: Expected an opcode"
+            .align  4
+not_found:  .asciz  "Error: Opcode not found"
diff --git a/newasm/string.s b/newasm/string.s
index c77958b..d897a98 100644
--- a/newasm/string.s
+++ b/newasm/string.s
@@ -28,7 +28,8 @@ putstrln:
         BL      putstr
         POP     {PC}
 
-cmpstr: MOVS    R4, 0
+cmpstr: PUSH    {R4}
+        MOVS    R4, 0
 1:      LDRB    R2, [R0, R4]
         LDRB    R3, [R1, R4]
         CMP     R2, R3
@@ -37,4 +38,5 @@ cmpstr: MOVS    R4, 0
         BEQ     2f
         ADDS    R4, 1
         B       1b
-2:      BX      LR
+2:      POP     {R4}
+        BX      LR