Add parsers for basic instruction typesHEADmaster

14 files changed, 309 insertions, 87 deletions

diff --git a/newasm/Makefile b/newasm/Makefile
index fd057fe..78a0828 100644
--- a/newasm/Makefile
+++ b/newasm/Makefile
@@ -2,40 +2,42 @@ PICO = /dev/disk/by-label/RPI-RP2
 DEVICE = /dev/ttyUSB0
 
 .PHONY: build
-build: parse.he
+build: assemble.he
 
-parse.he: parse.bin
-	od -An -tx2 -v parse.bin | sed "s/^ //" | tr " " "\n" | tr [:lower:] [:upper:] | sed "s/^/0x/" > parse.he
+assemble.he: assemble.bin
+	od -An -tx2 -v assemble.bin | sed "s/^ //" | tr " " "\n" | tr [:lower:] [:upper:] | sed "s/^/0x/" > assemble.he
 
-parse.bin: parse.elf
-	arm-none-eabi-objcopy -O binary parse.elf parse.bin
+assemble.bin: assemble.elf
+	arm-none-eabi-objcopy -O binary assemble.elf assemble.bin
 
 objects = main.o uart.o data.o string.o input.o \
 					statement.o optable.o \
+					regreg.o regregimm.o regimm.o regregreg.o imm.o \
 					parsers/whitespace.o \
 					parsers/label.o \
 					parsers/symbol.o \
 					parsers/opcode.o \
 					parsers/register.o \
-					parsers/decimal.o
+					parsers/decimal.o \
+					parsers/immediate.o
 
-parse.elf: pico_bin.ld $(objects)
-	arm-none-eabi-ld -T pico_bin.ld -o parse.elf $(objects)
+assemble.elf: pico_bin.ld $(objects)
+	arm-none-eabi-ld -T pico_bin.ld -o assemble.elf $(objects)
 
 $(objects): %.o: %.s
 	arm-none-eabi-as -o $@ $<
 
 .PHONY: clean
 clean:
-	rm -f parse.elf parse.bin parse.he slowcat $(objects)
+	rm -f assemble.elf assemble.bin assemble.he slowcat $(objects)
 
 .PHONY: dump
-dump: parse.bin
-	@od -Ax -tx2 -v parse.bin
+dump: assemble.bin
+	@od -Ax -tx2 -v assemble.bin
 
 .PHONY: serial
-serial: $(DEVICE) parse.he slowcat
-	cat parse.he | tr "\n" "\r" | ./slowcat | picocom -b 115200 -q $(DEVICE)
+serial: $(DEVICE) assemble.he slowcat
+	cat assemble.he | tr "\n" "\r" | ./slowcat | picocom -b 115200 -q $(DEVICE)
 	@echo
 	echo -n "G" | picocom -b 115200 -q $(DEVICE)
 	@echo
diff --git a/newasm/errors.s b/newasm/errors.s
new file mode 100644
index 0000000..3822851
--- /dev/null
+++ b/newasm/errors.s
@@ -0,0 +1,30 @@
+.syntax unified
+.cpu cortex-m0plus
+.thumb
+
+// .type print_error, %function
+// .global print_error
+
+// 00 success
+// 01 expected an opcode (or unexpected character)
+// 02 expected label or opcode
+// 03 expected colon at end of label
+// 04 opcode not found
+
+// 05 expected digit
+
+// 06 expected register
+// 07 invalid general-purpose register number
+// 08 invalid register for this register position
+// 09 invalid register combo for this instruction
+
+// 0A immediate value too large
+
+// 0B extra input at end of statement
+
+            .align  4
+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/imm.s b/newasm/imm.s
new file mode 100644
index 0000000..f6c7266
--- /dev/null
+++ b/newasm/imm.s
@@ -0,0 +1,27 @@
+.syntax unified
+.cpu cortex-m0plus
+.thumb
+
+.type imm, %function
+.global imm
+
+// R1 instruction under construction
+// R3 immediate width
+// R4 input buffer
+
+imm:      PUSH    {LR}
+          PUSH    {R3}
+          BL      immediate
+          POP     {R3}
+          BNE     exit
+          MOVS    R0, 1
+          LSLS    R0, R3
+          CMP     R2, R0
+          BLO     fine
+          MOVS    R0, 0x0A    // return code 0A (immediate value too large)
+          POP     {PC}
+fine:     ORRS    R1, R2      // fill in imm
+          MOVS    R0, 0       // return code 0 (success)
+exit:     POP     {PC}
+bad_reg:  MOVS    R0, 8       // return code 8 (invalid register for this register position)
+          POP     {PC}
diff --git a/newasm/main.s b/newasm/main.s
index a27fec9..f1d3c0d 100644
--- a/newasm/main.s
+++ b/newasm/main.s
@@ -6,38 +6,31 @@
 .global main, strbuf
 
 main:     LDR     R5, =0x20002000
-          BL      uart_recv
-loop:     BL      prompt
-          LDR     R0, inpbuf
+          BL      uart_recv           // wait for keypress
+loop:     BL      prompt              // display address and data
+          LDR     R0, inpbuf          // get a line of input
           BL      getline
-          LDR     R0, inpbuf
-          BL      putstrln
-          LDR     R4, inpbuf
-          LDR     R2, strbuf
-          MOVS    R0, 0
+          LDR     R4, inpbuf          // prepare input buffer
+          LDR     R2, strbuf          // prepare output buffer
+          MOVS    R0, 0               // clear output buffer
           STRB    R0, [R2]
-          BL      statement
-          BNE     bad
-
-good:     ADR     R0, success
-          PUSH    {R1}
-          BL      putstrln
-          MOVS    R0, R4
-          BL      putstrln
-          POP     {R0}
+          BL      statement           // call statement parser
+          BNE     bad                 // print message if failure
+          MOVS    R0, R1              // show assembled instruction
           BL      send_hex
           LDR     R0, =crlf
           BL      putstr
-          B       loop
-
-bad:      ADR     R0, fail
-          BL      putstrln
-          MOVS    R0, R4
+          B       loop                // repeat
+bad:      PUSH    {R0}
+          ADR     R0, fail
+          BL      putstr
+          POP    {R0}
+          BL      send_hex
+          LDR     R0, =crlf
+          BL      putstr
+          MOVS     R0, R4
           BL      putstrln
           B       loop
-never:    BL      uart_recv
-          LDR     R0, =0x20000001
-          BX      R0
 
 prompt:   PUSH    {LR}
           MOVS    R0, R5
@@ -57,4 +50,4 @@ strbuf:   .word   0x20001F80
           .align  4
 success:  .asciz  "The parser suceeded"
           .align  4
-fail:     .asciz  "The parser failed"
+fail:     .asciz  "The parser failed: "
diff --git a/newasm/optable.s b/newasm/optable.s
index a063f02..fb80ab4 100644
--- a/newasm/optable.s
+++ b/newasm/optable.s
@@ -5,9 +5,6 @@
 .type lookup, %function
 .global lookup
 
-// 1 unexpected char
-// 2 unexpected end of input
-
 // input:
 // R0 input string (unboxed)
 
@@ -25,7 +22,7 @@ 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)
+          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
@@ -42,7 +39,7 @@ 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)
+error:    MOVS    R0, 4       // return code 4 (opcode not found)
           POP     {R4, R5, R6}
           BX      LR
 
@@ -53,5 +50,5 @@ op_table:
           .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  "ADDS"; .hword 0x1800;  .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
index a1516d2..cc9dff4 100644
--- a/newasm/parsers/decimal.s
+++ b/newasm/parsers/decimal.s
@@ -5,8 +5,6 @@
 .type decimal, %function
 .global decimal
 
-// 1 not a digit
-
 // R4 input stream
 // R2 output value
 
@@ -35,5 +33,5 @@ loop:     LDRB    R0, [R4]    // get another char
           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)
+bad:      MOVS    R0, #5      // return code 5 (not a digit)
           BX      LR
diff --git a/newasm/parsers/immediate.s b/newasm/parsers/immediate.s
new file mode 100644
index 0000000..ac3e578
--- /dev/null
+++ b/newasm/parsers/immediate.s
@@ -0,0 +1,13 @@
+.syntax unified
+.cpu cortex-m0plus
+.thumb
+
+.type immediate, %function
+.global immediate
+
+// R4 input buffer
+// R2 output buffer
+
+immediate:PUSH    {LR}
+          BL      decimal
+          POP     {PC}
diff --git a/newasm/parsers/label.s b/newasm/parsers/label.s
index 4770e6c..45d39e5 100644
--- a/newasm/parsers/label.s
+++ b/newasm/parsers/label.s
@@ -5,9 +5,6 @@
 .type label, %function
 .global label
 
-// 1 unexpected begin char
-// 2 doesn't end with colon
-
 // R4 input buffer
 // R2 output buffer
 
@@ -23,7 +20,7 @@ label:    PUSH    {LR}
           LDRB    R0, [R4]    // get a char
           CMP     R0, ':      // colon
           BEQ     3f
-          MOVS    R0, #2      // return code 2 (expected colon)
+          MOVS    R0, #3      // return code 3 (expected colon)
           POP     {PC}
 3:        ADDS    R4, 1       // consume the colon
           MOVS    R0, #0      // return code 0 (success)
diff --git a/newasm/parsers/register.s b/newasm/parsers/register.s
index 48ebf7a..9334aa3 100644
--- a/newasm/parsers/register.s
+++ b/newasm/parsers/register.s
@@ -5,9 +5,6 @@
 .type register, %function
 .global register
 
-// 1 unexpected char
-// 2 invalid general-purpose register number
-
 // R4 input buffer
 // R2 output buffer
 
@@ -21,9 +18,12 @@ r_test:   CMP     R0, 'R
           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
+          MOVS    R0, #7      // invalid register number error code
           POP     {PC}
-special:  LDRH    R0, [R4]      // get two bytes from input stream
+special:  LDRB    R0, [R4]      // get two bytes from input stream
+          LDRB    R1, [R4, 1]
+          LSLS    R1, 8
+          ORRS    R0, R1
           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
@@ -32,7 +32,7 @@ loop:     LDRH    R1, [R2, R3]  // get two bytes from table at current offset
           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)
+          MOVS    R0, #6        // return code 6 (expected register)
           POP     {PC}
 done:     ADDS    R4, 2         // consume two chars
           LSRS    R3, 1         // divide table offset by two to row
@@ -41,6 +41,7 @@ done:     ADDS    R4, 2         // consume two chars
 success:  MOVS    R0, #0        // return code 0 (success)
           POP     {PC}
 
+          .align  4
 table:    .ascii  "SP"
           .ascii  "LR"
           .ascii  "PC"
diff --git a/newasm/regimm.s b/newasm/regimm.s
new file mode 100644
index 0000000..3e294c8
--- /dev/null
+++ b/newasm/regimm.s
@@ -0,0 +1,37 @@
+.syntax unified
+.cpu cortex-m0plus
+.thumb
+
+.type regimm, %function
+.global regimm
+
+// R1 instruction under construction
+// R3 immediate width
+// R4 input buffer
+
+regimm:   PUSH    {LR}
+          PUSH    {R3}
+          BL      register    // parse a register
+          POP     {R3}
+          BNE     exit        // exit if failure
+          CMP     R2, 7       // check that it's R0-R7
+          BHI     bad_reg
+          LSLS    R2, 8       // shift by 8
+          ORRS    R1, R2      // fill in Rdn
+          BL      whitespace  // mandatory whitespace
+          BNE     exit        // exit if failure
+          PUSH    {R3}
+          BL      immediate
+          POP     {R3}
+          BNE     exit
+          MOVS    R0, 1
+          LSLS    R0, R3
+          CMP     R2, R0
+          BLO     fine
+          MOVS    R0, 0x0A    // returun code 0A (immediate value too large)
+          POP     {PC}
+fine:     ORRS    R1, R2      // fill in imm
+          MOVS    R0, 0       // return code 0 (success)
+exit:     POP     {PC}
+bad_reg:  MOVS    R0, 8       // return code 8 (invalid register for this register position)
+          POP     {PC}
diff --git a/newasm/regreg.s b/newasm/regreg.s
new file mode 100644
index 0000000..6893dee
--- /dev/null
+++ b/newasm/regreg.s
@@ -0,0 +1,29 @@
+.syntax unified
+.cpu cortex-m0plus
+.thumb
+
+.type regreg, %function
+.global regreg
+
+// R1 instruction under construction
+// R2 output buffer
+// R4 input buffer
+
+regreg:   PUSH    {LR}
+          BL      register    // parse a register
+          BNE     exit        // exit if failure
+          CMP     R2, 7       // check that it's R0-R7
+          BHI     bad_reg
+          ORRS    R1, R2      // fill in Rdn
+          BL      whitespace  // mandatory whitespace
+          BNE     exit        // exit if failure
+          BL      register    // parse a register
+          BNE     exit
+          CMP     R2, 7       // check that it's R0-R7
+          BHI     bad_reg
+          LSLS    R2, 3       // shift by 3
+          ORRS    R1, R2      // fill in Rm
+          MOVS    R0, 0       // return code 0 (success)
+exit:     POP     {PC}
+bad_reg:  MOVS    R0, 8       // return code 8 (invalid register for this register position)
+          POP     {PC}
diff --git a/newasm/regregimm.s b/newasm/regregimm.s
new file mode 100644
index 0000000..6c649e2
--- /dev/null
+++ b/newasm/regregimm.s
@@ -0,0 +1,47 @@
+.syntax unified
+.cpu cortex-m0plus
+.thumb
+
+.type regregimm, %function
+.global regregimm
+
+// R1 instruction under construction
+// R3 immediate width
+// R4 input buffer
+
+regregimm:PUSH    {LR}
+          PUSH    {R3}
+          BL      register    // parse a register
+          POP    {R3}
+          BNE     exit        // exit if failure
+          CMP     R2, 7       // check that it's R0-R7
+          BHI     bad_reg
+          ORRS    R1, R2      // fill in Rd
+          BL      whitespace  // mandatory whitespace
+          BNE     exit        // exit if failure
+          PUSH    {R3}
+          BL      register    // parse a register
+          POP     {R3}
+          BNE     exit
+          CMP     R2, 7       // check that it's R0-R7
+          BHI     bad_reg
+          LSLS    R2, 3       // shift by 3
+          ORRS    R1, R2      // fill in Rn
+          BL      whitespace  // mandatory whitespace
+          BNE     exit        // exit if failure
+          PUSH    {R3}
+          BL      immediate
+          POP     {R3}
+          BNE     exit
+          MOVS    R0, 1
+          LSLS    R0, R3
+          CMP     R2, R0
+          BLO     fine
+          MOVS    R0, 0x0A    // returun code 0A (immediate value too large)
+          POP     {PC}
+fine:     LSLS    R2, 6       // shift by 6
+          ORRS    R1, R2      // fill in imm
+          MOVS    R0, 0       // return code 0 (success)
+exit:     POP     {PC}
+bad_reg:  MOVS    R0, 8       // return code 8 (invalid register for this register position)
+          POP     {PC}
diff --git a/newasm/regregreg.s b/newasm/regregreg.s
new file mode 100644
index 0000000..a8be6ee
--- /dev/null
+++ b/newasm/regregreg.s
@@ -0,0 +1,44 @@
+.syntax unified
+.cpu cortex-m0plus
+.thumb
+
+.type regregreg, %function
+.global regregreg
+
+// R1 instruction under construction
+// R2 output buffer
+// R4 input buffer
+
+regregreg:PUSH    {LR}
+
+          // TODO loop it
+          BL      register    // parse a register
+          BNE     exit        // exit if failure
+          CMP     R2, 7       // check that it's R0-R7
+          BHI     bad_reg
+          ORRS    R1, R2      // fill in Rd
+
+          BL      whitespace  // mandatory whitespace
+          BNE     exit        // exit if failure
+
+          BL      register    // parse a register
+          BNE     exit
+          CMP     R2, 7       // check that it's R0-R7
+          BHI     bad_reg
+          LSLS    R2, 3       // shift by 3
+          ORRS    R1, R2      // fill in Rn
+
+          BL      whitespace  // mandatory whitespace
+          BNE     exit        // exit if failure
+
+          BL      register    // parse a register
+          BNE     exit
+          CMP     R2, 7       // check that it's R0-R7
+          BHI     bad_reg
+          LSLS    R2, 6       // shift by 6
+          ORRS    R1, R2      // fill in Rm
+
+          MOVS    R0, 0       // return code 0 (success)
+exit:     POP     {PC}
+bad_reg:  MOVS    R0, 8       // return code 8 (invalid register for this register position)
+          POP     {PC}
diff --git a/newasm/statement.s b/newasm/statement.s
index 4a199a6..436ae03 100644
--- a/newasm/statement.s
+++ b/newasm/statement.s
@@ -5,40 +5,47 @@
 .type statement, %function
 .global statement
 
-// 1 reported an error
-
 // R4 input buffer
 // R2 output buffer
 
 statement:PUSH    {LR}
-          BL      whitespace
-          PUSH    {R2}
+          MOVS    R3, R2        // save output buffer
+          BL      whitespace    // skip leading whitespace
           BL      label
-          POP     {R2}
-          CMP     R0, #2        // check for error code 2 (no colon)
-          BNE     1f
-          ADR     R0, no_colon
-          B       err_exit
+          CMP     R0, 3         // only exit for code 3 (no colon at end of label)
+          BEQ     exit
+          MOVS    R1, R0
 1:        BL      whitespace
-          PUSH    {R2}
-          BL      opcode
-          POP     {R2}
-          BEQ     2f
-          ADR     R0, no_opcode
-          B       err_exit
-2:        LDR     R0, [R2]
+          MOVS    R2, 0         // reset output buffer
+          STRB    R2, [R3]
+          MOVS    R2, R3
+          BL      opcode        // TODO opcodes longer than 4
+          ADD     R0, R1
+          BNE     exit
+2:        MOVS    R0, R3
+          PUSH    {R3}
+          BL      putstrln
+          POP     {R3}
+          LDR     R0, [R3]
           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}
+          BNE     exit
+          BL      whitespace    // TODO new return code for expected whitespace
+          BNE     exit
+          ADR     R0, parsers
+          LSLS    R2, 2         // multiply by 4 to get byte offset
+          LDR     R2, [R0, R2]  // get address of parser
+          BLX     R2
+          BNE     exit
+          BL      whitespace
+          LDRB    R0, [R4]      // get a byte from the input stream
+          TST     R0, R0        // check if zero
+          BEQ     exit          // if it's zero then success
+          MOVS    R0, 0x0B      // return code 0B (extra input at end)
+exit:     POP     {PC}
 
-            .align  4
-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"
+          .align  4
+parsers:  .word   regreg
+          .word   regregimm
+          .word   regimm
+          .word   regregreg
+          .word   imm