7 Commits
Author SHA1 Message Date
PleegWat 304fe54833 Apply agreed naming conventions 2026-09-06 16:13:40 +02:00
PleegWat 6f7ad3aa4b Implement atoi
Parse strings to integer (binary, octal, decimal, or hexadecimal)
2026-09-06 16:12:50 +02:00
ShatteredMINT 9b7a8685e6 Merge pull request 'Add ISA definitions for symphony' (#18) from PleegWat/symphony_stdlib:add_isa into main
Reviewed-on: TCShenanigans/symphony_stdlib#18
2026-09-06 16:08:12 +02:00
PleegWat dd20822ee1 Use all unsigned compares 2026-09-05 16:41:52 +02:00
PleegWat 4c0cd79fdd Fix several bugs 2026-09-05 16:37:51 +02:00
PleegWat 2f09320fa5 Add optimization for arbitrary aligned values
Any value with 16 or less significant bits but larger than 64k is loaded
in one fewer cycle
2026-09-05 15:39:55 +02:00
PleegWat 8e7441fdb3 Add ISA definitions for symphony
Adds both a 'stock' ISA (copied from the game) and one modified for test
code
2026-09-05 15:36:54 +02:00
5 changed files with 262 additions and 13 deletions
+1 -1
View File
@@ -16,7 +16,7 @@ Performs a jump-and-link function call, with label size up to 32 bits.
_HARDWARE:_ Store PC+4 in the result register when in jump mode. _HARDWARE:_ Store PC+4 in the result register when in jump mode.
## `li %a(register) %b:U32(immediate | label) ## `li %a(register) %b:U32(immediate | label)`
Loads an immediate up to 32 bits long Loads an immediate up to 32 bits long
_HARDWARE:_ No requirements _HARDWARE:_ No requirements
+15 -12
View File
@@ -332,12 +332,13 @@ ret
li %r(register), %a:S33(immediate | label) li %r(register), %a:S33(immediate | label)
; assert(0 - 0x80000000 <= %a, "%a is too low") ; assert(0 - 0x80000000 <= %a, "%a is too low")
; assert(%a <= 0xffffffff, "%a is too high") ; assert(%a <= 0xffffffff, "%a is too high")
%nota = ~%a %s = trailing_zeros(%a)
when 0x0000 <= %a && %a <= 0xffff: 00110001 rrrr0000 %a[15:0] ; First 64k %sa = %a >> %s
when %nota <= 0xffff : 00110001 rrrr0000 %nota[15:0] 00100011 rrrr0000 0000rrrr 00000000 ; Last 64k %nota = %a ^ 0xffffffff
when !(%a & 0x0f) && %a < 0xffff0: 00110001 rrrr0000 %a[19:4] 00100111 rrrrrrrr 00000000 00000100 ; 16-byte aligned when %a <=u 0xffff : 00110001 %r[3:0]0000 %a[15:0] ; First 64k
when !(%a & 0x03) && %a < 0x3fffc: 00110001 rrrr0000 %a[17:2] 00100111 rrrrrrrr 00000000 00000010 ; 4-byte aligned when %nota <=u 0xffff: 00110001 %r[3:0]0000 %nota[15:0] 00100011 %r[3:0]0000 0000%r[3:0] 00000000 ; Last 64k
00110001 rrrr0000 %a[31:16] 00100111 rrrrrrrr 00000000 00010000 00110001 rrrrrrrr %a[15:0] ; Any 32-bit when %sa <=u 0xffff : 00110001 %r[3:0]0000 %sa[15:0] 00110111 %r[3:0]%r[3:0] 00000000 %s[7:0] ; Aligned
00110001 %r[3:0]0000 %a[31:16] 00110111 %r[3:0]%r[3:0] 00000000 00010000 00110001 %r[3:0]%r[3:0] %a[15:0] ; Any 32-bit
# Load 32-bit immediate %a into %r # Load 32-bit immediate %a into %r
qcall %a:U32(label) qcall %a:U32(label)
@@ -345,14 +346,16 @@ qcall %a:U32(label)
; assert(%a <= 0xffffffff, "%a is too high") ; assert(%a <= 0xffffffff, "%a is too high")
%r = 13 %r = 13
%f = 15 %f = 15
%nota = ~%a %s = trailing_zeros(%a)
when 0x0000 <= %a && %a <= 0xffff: 01011000 rrrrffff %a[15:0] ; First 64k %sa = %a >> %s
when %nota <= 0xffff : 00110001 rrrr0000 %nota[15:0] 00100011 rrrr0000 0000rrrr 00000000 01001000 rrrrffff 0000rrrr 00000000 ; Last 64k %nota = %a ^ 0xffffffff
when !(%a & 0x0f) && %a < 0xffff0: 00110001 rrrr0000 %a[19:4] 00100111 rrrrrrrr 00000000 00000100 01001000 rrrrffff 0000rrrr 00000000 ; 16-byte aligned when %a <=u 0xffff : 01011000 %r[3:0]ffff %a[15:0] ; First 64k
when !(%a & 0x03) && %a < 0x3fffc: 00110001 rrrr0000 %a[17:2] 00100111 rrrrrrrr 00000000 00000010 01001000 rrrrffff 0000rrrr 00000000 ; 4-byte aligned when %nota <=u 0xffff: 00110001 %r[3:0]0000 %nota[15:0] 00100011 %r[3:0]0000 0000%r[3:0] 00000000 01001000 %r[3:0]ffff 0000%r[3:0] 00000000 ; Last 64k
00110001 rrrr0000 %a[31:16] 00100111 rrrrrrrr 00000000 00010000 00110001 rrrrrrrr %a[15:0] 01001000 rrrrffff 0000rrrr 00000000 ; Any 32-bit when %sa <=u 0xffff : 00110001 %r[3:0]0000 %sa[15:0] 00110111 %r[3:0]%r[3:0] 00000000 %s[7:0] 01001000 %r[3:0]ffff 0000%r[3:0] 00000000 ; Aligned
00110001 %r[3:0]0000 %a[31:16] 00110111 %r[3:0]%r[3:0] 00000000 00010000 00110001 %r[3:0]%r[3:0] %a[15:0] 01001000 %r[3:0]ffff 0000%r[3:0] 00000000 ; Any 32-bit
# Jump and link to 32-bit label # Jump and link to 32-bit label
halt%c(condition) halt%c(condition)
1000cccc 00001111 00000000 00000000 1000cccc 00001111 00000000 00000000
# Halt on condition %c # Halt on condition %c
+1
View File
@@ -3,6 +3,7 @@ pub include imath
pub include array pub include array
pub include console pub include console
pub include mem pub include mem
pub include string_to_int
; Needs to be last! ; Needs to be last!
pub include LUTs pub include LUTs
+188
View File
@@ -0,0 +1,188 @@
; Internal register assignments:
; r1 - Partially parsed output integer
; r2 - Parsing position
; r3 - Character being parsed
; r4 - Set to -1 if the input is negative
; Convert string to integer (akin to libc atoi(), strotoi())
; Arguments:
; r1 - Pointer to string
; Result:
; r1 - Parsed integer
; r2 - Pointer to first rejected input byte
; Clobbers:
; flags
; r3 - last character read
; r4 - Negative marker
; Note: Unless the input is "0", tail-calls into a base-specific specialization.
pub auto:
mov r2, r1
mov r1, 0
mov r4, 0
load_8 r3, [r2]
cmp r3, 0x2D ; '-'
jne auto_positive
sub r4, zr, 1 ; Set r4 to -1
add r2, r2, 1
auto_positive:
load_16 r3, [r2] ; 2-byte prefix "0b", "0o", "0x", etc.
or r3, r3, 0x20 ; 2nd char to lower case
add r2, r2, 2
cmp r3, 0x3062 ; "0b"
je bin_loop
cmp r3, 0x306F ; "0o"
je oct_loop
cmp r3, 0x3078 ; "0x"
je hex_loop
sub r2, r2, 2 ; no matching prefix, move pointer back
jmp dec_loop
; Epilogue
done:
add r1, r1, r4 ; If r4 is -1, negate r1. Else it's 0 and no effect.
xor r1, r1, r4
mov flags, 0 ; No error
jmp r13
; Convert decimal string to integer
; Arguments:
; r1 - Pointer to string
; Result:
; r1 - Parsed integer
; r2 - Pointer to first rejected input byte
; Clobbers:
; flags
; r3 - last character read
; r4 - Negative marker
pub dec:
mov r2, r1
mov r1, 0
mov r4, 0
load_8 r3, [r2]
cmp r3, 0x2D ; '-'
jne dec_positive
sub r4, zr, 1 ; Set r4 to -1
add r2, r2, 1
dec_loop:
load_8 r3, [r2]
dec_positive:
sub r3, r3, 0x30 ; '0'
cmp r3, 9
ja done
add r2, r2, 1
lsl flags, r1, 2 ; Use flags to help multiply by 10
add r1, r1, flags
lsl r1, r1, 1
add r1, r1, r3
jmp dec_loop
; Convert binary string to integer
; Arguments:
; r1 - Pointer to string
; Result:
; r1 - Parsed integer
; r2 - Pointer to first rejected input byte
; Clobbers:
; flags
; r3 - last character read
; r4 - Negative marker
pub bin:
mov r2, r1
mov r1, 0
mov r4, 0
load_8 r3, [r2]
cmp r3, 0x2D ; '-'
jne bin_positive
sub r4, zr, 1 ; Set r4 to -1
add r2, r2, 1
bin_positive:
load_16 r3, [r2] ; check for prefix
or r3, r3, 0x20 ; 2nd char to lower case
cmp r3, 0x3062 ; "0b"
jne bin_loop
add r2, r2, 2
bin_loop:
load_8 r3, [r2]
sub r3, r3, 0x30 ; '0'
cmp r3, 1
ja done
add r2, r2, 1
lsl r1, r1, 1
add r1, r1, r3
jmp bin_loop
; Convert octal string to integer
; Arguments:
; r1 - Pointer to string
; Result:
; r1 - Parsed integer
; r2 - Pointer to first rejected input byte
; Clobbers:
; flags
; r3 - last character read
; r4 - Negative marker
pub oct:
mov r2, r1
mov r1, 0
mov r4, 0
load_8 r3, [r2]
cmp r3, 0x2D ; '-'
jne oct_positive
sub r4, zr, 1 ; Set r4 to -1
add r2, r2, 1
oct_positive:
load_16 r3, [r2] ; check for prefix
or r3, r3, 0x20 ; 2nd char to lower case
cmp r3, 0x306F ; "0o"
jne oct_loop
add r2, r2, 2
oct_loop:
load_8 r3, [r2]
sub r3, r3, 0x30 ; '0'
cmp r3, 7
ja done
add r2, r2, 1
lsl r1, r1, 3
add r1, r1, r3
jmp oct_loop
; Convert hexadecimal string to integer
; Arguments:
; r1 - Pointer to string
; Result:
; r1 - Parsed integer
; r2 - Pointer to first rejected input byte
; Clobbers:
; flags
; r3 - last character read
; r4 - Negative marker
pub hex:
mov r2, r1
mov r1, 0
mov r4, 0
load_8 r3, [r2]
cmp r3, 0x2D ; '-'
jne hex_positive
sub r4, zr, 1 ; Set r4 to -1
add r2, r2, 1
hex_positive:
load_16 r3, [r2] ; check for prefix
or r3, r3, 0x20 ; 2nd char to lower case
cmp r3, 0x3078 ; "0x"
jne hex_loop
add r2, r2, 2
hex_loop:
load_8 r3, [r2]
sub r3, r3, 0x30 ; '0'
cmp r3, 10
jb hex_add
sub r3, r3, 0x11 ; 'A' - '0'
and r3, r3, 0xdf ; to lower case
cmp r3, 5 ; 0-5: 6 letters
ja done
add r3, r3, 10 ; Adjust for digits below
hex_add:
add r2, r2, 1
lsl r1, r1, 4
add r1, r1, r3
jmp hex_loop
+57
View File
@@ -0,0 +1,57 @@
; test harness for string_to_int
; Registers:
; r8 - test address
; r9 - expected result
; r10 - actual length
; r11 - expected length
li r8, tests
li r12, end_of_tests
next_test:
; Set up arguments and run test
add r1, r8, 5 ; Start of test string
mov r13, atoi_test_ret
jmp stdlib.string_to_int.auto
atoi_test_ret:
; Load reference data
load_32 r9, [r8] ; Expected result
sub r10, r2, r8
sub r10, r10, 5 ; Actual string length
add r11, r8, 4 ; Address of expected length
load_8 r11, [r11] ; Expected length
; Verify results
cmp r1, r9
incorrect_result: jne incorrect_result
cmp r10, r11
incorrect_length: jne incorrect_length
; Next test
add r8, r8, 0x10
cmp r8, r12
jl next_test
; Done
success: jmp success
include ../src/stdlib
@0x20000
tests:
; inlen is the number of input bytes the function is expected to consume
; addr result inlen instr
@0x20000 U32 0 U8 0 "\0"
@0x20010 U32 0 U8 1 "0\0"
@0x20020 U32 1 U8 1 "1\0"
@0x20030 U32 2 U8 1 "2:\0"
@0x20040 U32 42 U8 2 "42\0"
@0x20050 U32 67 U8 2 "67lol\0"
@0x20060 U32 0x69a U8 5 "0x69a@\0"
@0x20070 U32 0x4B4 U8 5 "0x4B4g\0"
@0x20080 U32 0o23 U8 4 "0o239\0"
@0x20090 U32 0b1011 U8 6 "0b1011\0"
@0x200a0 U32 0b10001 U8 7 "0b100012\0"
@0x200b0 U32 0xFFFFFFEB U8 3 "-21\0" ; Yuk
@0x200c0
end_of_tests: