Irmen de Jong
|
2e37f5dee3
|
IR: support for multi-returnvalue function calls (asmsubs)
note: the VM can't execute these though as it has no CPU hardware registers
|
2024-03-23 00:30:17 +01:00 |
|
Irmen de Jong
|
66e7c51064
|
IR: fix some things related to asmsubs
|
2024-03-22 21:49:01 +01:00 |
|
Irmen de Jong
|
f5e332daf7
|
remove redundant IR instructions like SNZ
|
2024-03-02 23:19:39 +01:00 |
|
Irmen de Jong
|
040d75dafa
|
VM now supports indirect jump instruction
|
2024-02-22 16:19:26 +01:00 |
|
Irmen de Jong
|
39d2194d8f
|
IR: implemented inplace prefix op on split array
VM: NEG instructions also set N and Z flags
|
2024-01-28 22:33:13 +01:00 |
|
Irmen de Jong
|
504d1440cc
|
fixed rol(),rol2(),ror(),ror2()
|
2024-01-17 21:02:17 +01:00 |
|
Irmen de Jong
|
0cfcc5cd29
|
fix VM sgn() function for floats
|
2024-01-16 01:34:55 +01:00 |
|
Irmen de Jong
|
968609d06d
|
IR: fix problems with symbol offsets and unused subroutines/chunks
|
2024-01-13 16:43:41 +01:00 |
|
Irmen de Jong
|
38dc7fb7bd
|
IR: added SCC and SCS instructions
|
2024-01-09 01:41:37 +01:00 |
|
Irmen de Jong
|
37fa3b34a2
|
tweak IR
|
2024-01-07 22:12:09 +01:00 |
|
Irmen de Jong
|
6d9fccacb1
|
boolean not expression tweaks and optimizations
|
2024-01-05 13:32:16 +01:00 |
|
Irmen de Jong
|
f790182f0b
|
adding short-circuit boolean expression evaluation (in IR codegen) also -noshortcircuit cli option
|
2023-12-30 01:08:41 +01:00 |
|
Irmen de Jong
|
ae66fcac1e
|
added call builtin function for indirect JSR
|
2023-12-17 15:45:28 +01:00 |
|
Irmen de Jong
|
b24df31c2b
|
IR: fix codegen for routines returning in CPU Status register flag
|
2023-12-14 21:16:14 +01:00 |
|
Irmen de Jong
|
af5ca2d0b8
|
vm: treat floats as 64 bits doubles. 0.0 printed as "0".
|
2023-11-21 00:57:56 +01:00 |
|
Irmen de Jong
|
ff8de8e42d
|
removing redundant compares
|
2023-11-16 22:56:19 +01:00 |
|
Irmen de Jong
|
09d506194f
|
note
|
2023-11-15 22:27:16 +01:00 |
|
Irmen de Jong
|
92374e122b
|
IR: optimize concat with msb 0 into ext
|
2023-10-28 12:53:41 +02:00 |
|
Irmen de Jong
|
1da0c59182
|
vm: remove BNER opcode -> CMP + BSTNE
|
2023-09-23 11:47:24 +02:00 |
|
Irmen de Jong
|
36e8f10d2b
|
vm: remove BEQR opcode -> CMP + BSTEQ
|
2023-09-23 11:42:58 +02:00 |
|
Irmen de Jong
|
cdf5a8f20f
|
vm: remove BNE opcode -> CMPI + BSTNE
|
2023-09-23 11:22:33 +02:00 |
|
Irmen de Jong
|
eb64d92333
|
vm: remove BEQ opcode -> CMPI + BSTEQ
|
2023-09-23 11:21:43 +02:00 |
|
Irmen de Jong
|
c319233ddc
|
ir: added preparecall 'meta' instruction for functioncalls
|
2023-09-18 23:22:03 +02:00 |
|
Irmen de Jong
|
bde4be8231
|
fix VM indexed instructions to only use lsb part of the index
|
2023-09-06 02:44:04 +02:00 |
|
Irmen de Jong
|
2c9e50873c
|
use math.square for optimized X*X calculation (words only).
Added IR SQUARE instruction.
|
2023-08-14 01:05:17 +02:00 |
|
Irmen de Jong
|
c7d54570cc
|
IR: sXX, CONCAT instructions now use 3 register format
|
2023-07-21 00:07:56 +02:00 |
|
Irmen de Jong
|
6938c79f88
|
IR: added CMPI instruction
|
2023-07-14 23:17:29 +02:00 |
|
Irmen de Jong
|
a98cb50d55
|
Revert "ir: SCC now sets all bits to 1 (or 0)"
This reverts commit 7245aece4f .
|
2023-07-09 23:16:13 +02:00 |
|
Irmen de Jong
|
7245aece4f
|
ir: SCC now sets all bits to 1 (or 0)
|
2023-07-08 23:16:01 +02:00 |
|
Irmen de Jong
|
5037033fcf
|
ir: EXT and EXTS opcodes now have 2 registers to avoid type clash
|
2023-07-08 22:42:11 +02:00 |
|
Irmen de Jong
|
d26967a87d
|
ir doc
|
2023-07-07 22:35:05 +02:00 |
|
Irmen de Jong
|
334d382bfa
|
ir: JUMPI instruction added to support indirect jumps
|
2023-07-07 19:10:39 +02:00 |
|
Irmen de Jong
|
7ee162d98b
|
preparing version 9.0
|
2023-06-05 19:47:00 +02:00 |
|
Irmen de Jong
|
380f557c45
|
vm: implement split incr/decr
|
2023-06-03 22:22:13 +02:00 |
|
Irmen de Jong
|
bfaad1388c
|
IR: handle split arrays without new custom opcodes
|
2023-06-03 01:51:02 +02:00 |
|
Irmen de Jong
|
3ac9036c79
|
more split array stuff for 6502
|
2023-05-27 22:44:45 +02:00 |
|
Irmen de Jong
|
c94e292176
|
more split array stuff
|
2023-05-27 12:47:11 +02:00 |
|
Irmen de Jong
|
300e2fe9f8
|
IR: wrong attempt at optimizing register usage by reusing registers inside different code chunks
|
2023-05-18 21:57:21 +02:00 |
|
Irmen de Jong
|
b43223cb7a
|
added clamp() builtin function and floats.clampf()
|
2023-05-17 23:12:58 +02:00 |
|
Irmen de Jong
|
630c8a5faa
|
IR: fix romsub encoding
|
2023-05-14 18:08:06 +02:00 |
|
Irmen de Jong
|
905921a684
|
IR: new (sys)call instructions that encapsulate the full subroutine call
to fix the bugs resulting from nesting subroutine calls (as param to another call etc)
|
2023-05-14 15:20:25 +02:00 |
|
Irmen de Jong
|
bff3c4f95c
|
IR now converts IRInlineAsmChunk (of type IR) into regular code chunks directly.
.p8ir files usually won't contain <INLINEASM> nodes any longer
|
2023-05-09 21:04:31 +02:00 |
|
Irmen de Jong
|
97df33ab1a
|
IR: fix byte to word assignment not doing value extension
|
2023-05-08 22:47:00 +02:00 |
|
Irmen de Jong
|
add8a777d8
|
IR: binarydata fixes
|
2023-05-03 22:31:04 +02:00 |
|
Irmen de Jong
|
f465b2e2a0
|
some improvements to IR peephole optimizer
|
2023-05-02 00:29:04 +02:00 |
|
Irmen de Jong
|
0bea721c2e
|
docs
|
2023-04-27 01:26:25 +02:00 |
|
Irmen de Jong
|
efd7d6f0c0
|
tweak IR call args setting now via special SETPARAM instruction
|
2023-04-14 02:10:39 +02:00 |
|
Irmen de Jong
|
b55be093be
|
tweak IR
|
2023-04-11 22:48:20 +02:00 |
|
Irmen de Jong
|
9b37ac483f
|
vm fix str to word conversion
ir SYSCALL puts result(s) on value stack, instead of on hardcoded r0, r1
|
2023-04-10 16:26:42 +02:00 |
|
Irmen de Jong
|
090820958e
|
ir divmod returns its results on valuestack, to keep consistency with the rule that only 1 register can be a returnvalue
|
2023-04-10 15:26:30 +02:00 |
|