Commit Graph

546 Commits

Author SHA1 Message Date
Irmen de Jong
828d83dbef loadScaledArrayIndexIntoRegister(): useless type arg removed 2024-01-15 22:19:52 +01:00
Irmen de Jong
7de665d1e4 support for split wordarrays rol/ror and rol2/ror2
optimized rol and ror codegen

optimize/fix ror/ror2/rol/rol2 on word arrays
2024-01-15 03:22:37 +01:00
Irmen de Jong
0a356ba73a added containment check of float arrays 2024-01-14 14:14:09 +01:00
Irmen de Jong
0c1018ec61 dt error details 2024-01-12 17:34:19 +01:00
Irmen de Jong
d85c347a6c optimize /256 more, and fixed a unsigned byte word cast error 2024-01-07 02:34:05 +01:00
Irmen de Jong
8e6b91cb9e some optimizations 2024-01-06 00:44:00 +01:00
Irmen de Jong
334e6dca28 added string.contains().
fixed string and array containment check for length 1.
2024-01-05 20:46:26 +01:00
Irmen de Jong
bfd3edb617 fix expression evaluation bug where intermediate values were overwritten, yielding the wrong result 2024-01-04 21:04:11 +01:00
Irmen de Jong
19a2110ba2 fix exit() now actually correctly setting the return code in A
also, moved some cleanup stuff such as CLRCHN from exit() to the cleanup routine that is always called.
finally, also call the cleanup routine when  %option no_sysinit is used
2024-01-04 00:43:35 +01:00
Irmen de Jong
242a3eec63 fix data type difference error on range from and to values 2024-01-03 21:46:22 +01:00
Irmen de Jong
1e1f444cab cleanups.
also checked that value(12) < x < value(100) is indeed properly shortcircuited if x is 12 or less
2023-12-30 04:34:07 +01:00
Irmen de Jong
265e7aefbf clean up unused codegen for logical ops on words, also fix no-shortcircuit exception 2023-12-30 04:06:02 +01:00
Irmen de Jong
1c55a6c6dc shortcutting part one 2023-12-30 03:54:12 +01:00
Irmen de Jong
8f18b5b8a7 keep distinction between logical and bitwise boolean operators 2023-12-30 01:08:42 +01:00
Irmen de Jong
ccc11e49d2 fix asmgen for uword shift right 8 or more bits 2023-12-29 05:06:09 +01:00
Irmen de Jong
5d88717f32 fix non-existing instructions txy/tyx, oops 2023-12-29 03:27:35 +01:00
Irmen de Jong
9c1b11d605 some WARN messages are now INFO 2023-12-28 14:20:47 +01:00
Irmen de Jong
37fcde30d6 constants have p8c_ prefix instead of p8v_ in the asm 2023-12-28 05:28:32 +01:00
Irmen de Jong
2eb137618e refactor block options 2023-12-26 22:13:08 +01:00
Irmen de Jong
0d44492086 push,pushw,pop and popw are no longer built-in functions but regular routines in sys 2023-12-26 14:47:31 +01:00
Irmen de Jong
8ae435549d added -slabshigh N and -slabsgolden for memory() slabs 2023-12-23 20:45:30 +01:00
Irmen de Jong
9b113c0cbb added -varsgolden to put BSS into Golden Ram at $0400 2023-12-23 20:11:50 +01:00
Irmen de Jong
0e0fac8c4b BSSHIGHRAM_END more clearly defined (to be inclusive) 2023-12-23 19:05:06 +01:00
Irmen de Jong
ad9eaeafeb call now returns a word value 2023-12-22 22:24:11 +01:00
Irmen de Jong
66d939df0d docs about new asm symbol prefixes 2023-12-20 22:37:46 +01:00
Irmen de Jong
6bc079c7b7 more asm symbol prefixing: variables with p8v_, subroutines with p8s_ etc
labels with p8l_ . All this to avoid symbol clashes in the generated assembly code.
Everything got its own distinguishing prefix so we're done with it once and for all and have only 1 breaking change moment.
2023-12-20 22:20:59 +01:00
Irmen de Jong
69f6afe420 block names in asm now prefixed with p8b_ (instead of p8_)
as part of fixing var versus block symbol conflict handling
2023-12-19 23:00:20 +01:00
Irmen de Jong
ae66fcac1e added call builtin function for indirect JSR 2023-12-17 15:45:28 +01:00
Irmen de Jong
eba0bde6f3 Merge branch 'optimize-st'
# Conflicts:
#	examples/test.p8
2023-12-17 02:11:01 +01:00
Irmen de Jong
01c2112881 remove PtJump label, just use identifier with dummy 2023-12-13 04:16:49 +01:00
Irmen de Jong
a546c2247d optimize if-else handling of asmsub boolean result in status flags 2023-12-13 04:03:21 +01:00
Irmen de Jong
0da9142009 asm: also work for asmsub that return N or Z flag (Carry already worked) 2023-12-13 02:11:15 +01:00
Irmen de Jong
00b32f64e6 tweak for,sort,reverse st use 2023-12-12 20:26:45 +01:00
Irmen de Jong
f97b3f23e2 optimize symbol table for IR 2023-12-12 19:26:27 +01:00
Irmen de Jong
8e4319cd5a module directive %encoding to set the text encoding for that whole file (iso, petscii, etc.) 2023-12-06 23:54:08 +01:00
Irmen de Jong
28c1b208c1 optimized calls for float *10 and +0.5 2023-12-06 01:18:07 +01:00
Irmen de Jong
48f09f71ab fix TODO crash on uword[0] = uword[0] or 128 (byte register assign to word array) 2023-12-02 21:29:14 +01:00
Irmen de Jong
1e299bf360 better pokef() code 2023-11-28 23:04:27 +01:00
Irmen de Jong
ae2d96c455 added peekf and pokef builtin functions. Fixed sizeof() to allow number argument as well. 2023-11-27 23:36:02 +01:00
Irmen de Jong
6d8fbe0877 fixed float array indexing with an expression 2023-11-27 20:54:49 +01:00
Irmen de Jong
533090a68e fix expression result register of square, callfar, string compare functions in certain situations 2023-11-26 23:02:10 +01:00
Irmen de Jong
82a15b5a16 65C02 cpu: use TRB and TSB instructions for in-place AND/OR. 2023-11-24 00:50:52 +01:00
Irmen de Jong
935450a45f update kotest library 2023-11-22 18:40:07 +01:00
Irmen de Jong
0c9daf6eaf fix compiler crash on ptrvar[n+1] = ptrvar[2] 2023-11-14 21:46:11 +01:00
Irmen de Jong
76c09da961 make pokemon() be like poke, but also return the old value in the memory location. 2023-11-09 22:48:44 +01:00
Irmen de Jong
e7178ee496 optimized comparison with word variables 2023-11-05 00:20:12 +01:00
Irmen de Jong
3277544295 optimize assigning word array value to byte variable 2023-11-04 00:33:50 +01:00
Irmen de Jong
98d2c64d5d fix assembly error for uword[3] @zp @split word_addrs 2023-11-03 00:39:43 +01:00
Irmen de Jong
f68b46fc60 add a %zpallowed option to specify the range of zeropage register that can be used 2023-11-03 00:19:25 +01:00
Irmen de Jong
d54ab856e7 fix parameter passing bug introduced recently (byte not converted to word) 2023-11-02 00:31:35 +01:00
Irmen de Jong
16b24fadea gfx2 future mode, upgrate to Kotlin 1.9.20 2023-11-01 23:18:44 +01:00
Irmen de Jong
b3803cbdf1 more opportunities to use LDA(zp) instead of LDA(zp),Y on 65c02 2023-10-31 21:26:55 +01:00
Irmen de Jong
2ceaa25181 optimized code for (infrequently used) logical operations on word array 2023-10-29 23:41:34 +01:00
Irmen de Jong
7ec4ba40ad optimize asmsub arg evaluation order and stack usage 2023-10-28 17:29:00 +02:00
Irmen de Jong
0904712a00 remove last trace of getTempVar (arry index expression)
tiny optimization
2023-10-27 21:41:52 +02:00
Irmen de Jong
13ab4166c0 new kotest library version 2023-10-19 21:57:06 +02:00
Irmen de Jong
d88c09b098 fix signed byte to word casting issue uw = 8888 + (bb as ubyte) 2023-10-17 22:54:33 +02:00
Irmen de Jong
893b383bdf fix signed byte to word sign extension in assignment 2023-10-17 03:08:37 +02:00
Irmen de Jong
97c5c90eff fix codegen for var1>>=var2 and var1<<=var2 when var2 is zero 2023-10-16 00:04:21 +02:00
Irmen de Jong
daca87c6d0 added -breakinstr compiler option 2023-10-15 21:55:09 +02:00
Irmen de Jong
203ec5fa46 implement taking address of array var with variable index 2023-10-15 20:24:48 +02:00
Irmen de Jong
9ea69c07b8 optimize word array reads with indexvar 2023-10-14 07:30:54 +02:00
Irmen de Jong
68e62e4bd2 added cx16.MCIOUT() kernal call
correct case of several other cx16 kernal calls.

corrected to upper case: cx16 kernal calls CLOSE_ALL, LKUPLA, LKUPSA, JSRFAR, PRIMM, MACPTR.
2023-10-09 22:44:36 +02:00
Irmen de Jong
a48ce35f0b added %option verafxmuls 2023-10-05 22:06:33 +02:00
Irmen de Jong
a289b32053 Revert "added -verafxmul compiler option to use vera fx multiplication routine on cx16"
This reverts commit 690782bf.
It was too risky, using vera (especially fx) transparently in multiple places especially perhaps in IRQ handlers will create havoc unless much intricate care is taken to save/restore the vera state. Better to do vera fx explicitly where the programmer has full control.
2023-10-02 21:08:52 +02:00
Irmen de Jong
690782bf60 added -verafxmul compiler option to use vera fx multiplication routine on cx16 2023-10-01 22:44:45 +02:00
Irmen de Jong
8d177beb78 fix possible register corruption when calling asmsubs that require Carry flag as a parameter 2023-09-24 14:03:31 +02:00
Irmen de Jong
880c0a5da8 allow taking address of array element 2023-09-18 04:37:41 +02:00
Irmen de Jong
ccf6e32bf9 adding setlsb() and setmsb() builtin functions to 6502 codegen 2023-09-17 15:16:47 +02:00
Irmen de Jong
a1874f6f00 adding setlsb() and setmsb() builtin functions to 6502 codegen 2023-09-17 01:48:29 +02:00
Irmen de Jong
95e4490a8a adding setlsb() and setmsb() builtin functions 2023-09-15 02:39:16 +02:00
Irmen de Jong
31c132c2eb several optimizations and compiler error fix for @(&var) and @(&var+1) 2023-09-14 23:04:23 +02:00
Irmen de Jong
a1d0e5bb65 added list of software to docs 2023-09-13 21:51:48 +02:00
Irmen de Jong
03e0d4b2e8 reducing expression codegen complexity (no longer splitting conditional expressions, and using r9 as temp var) 2023-09-13 01:08:42 +02:00
Irmen de Jong
6afdd4e6fd preparing next version 2023-09-12 21:53:49 +02:00
Irmen de Jong
dd2463a440 proper fix for the previous commit. + fix for i/o channel reset in diskio.f_seek()
it wasn't the adressing mode, it was that it assumed the pointer variable was always in zeropage (which might not be)
2023-09-07 22:17:46 +02:00
Irmen de Jong
23a8bebd9e fix invalid addressing mode on 6502 cpu for bytevalue +/- bytearray[i] 2023-09-07 21:40:07 +02:00
Irmen de Jong
3caf9108ad finalizing 9.4.1 release 2023-09-06 21:18:01 +02:00
Irmen de Jong
b570bdaed7 fix codegen for array[i] += float expression 2023-09-05 22:38:52 +02:00
Irmen de Jong
8c0843cc87 fix an invalid 6502 instruction on c64 in certain float assignment 2023-09-05 21:54:52 +02:00
Irmen de Jong
eb018ae660 code optimization for bytearray[x] +/- bytearray[y]
use adc array,y or sbc array,y instead of tempvar
2023-09-04 23:01:53 +02:00
Irmen de Jong
c3fbdf34ca fixed c64 float problem 2023-09-03 16:40:10 +02:00
Irmen de Jong
48bd51e1a5 c64 float problem 2023-09-03 16:29:01 +02:00
Irmen de Jong
10d0b03a90 use less tempvars 2023-09-03 01:32:47 +02:00
Irmen de Jong
e1b3582f08 fix wordvar -= @(memory) 2023-09-03 01:12:26 +02:00
Irmen de Jong
95be1c9e22 fix optimized swapped in-place byte comparisons 2023-09-03 00:47:55 +02:00
Irmen de Jong
1ce8fe06d5 fix in-place <= for bytes 2023-09-03 00:01:11 +02:00
Irmen de Jong
e97303c226 fix word multiplication to not clobber r0 and r1 anymore
This was causing corruption in certain programs such as the cx16/amiga example.
The problem was introduced in 9.4 with the new multiply_words routine
2023-09-02 20:52:16 +02:00
Irmen de Jong
3b786c819d avoid using temp var even more 2023-09-01 23:47:01 +02:00
Irmen de Jong
04959dbd8b optimize asm: don't use temp var for some additions 2023-09-01 22:24:17 +02:00
Irmen de Jong
f14ea1b3de micro optimization to save 2 cycles: change some pha+pla into tax+txa 2023-09-01 20:37:24 +02:00
Irmen de Jong
09a7a4bbe5 optimize comparison against zero 2023-09-01 02:28:11 +02:00
Irmen de Jong
aa4d23a3d5 fix register stack saving on certain expression code that was broken on 6502 but not on 65c02 2023-08-29 11:50:35 +02:00
Irmen de Jong
bad9dd3b3b mention shadowing warnings from assembler 2023-08-28 16:55:28 +02:00
Irmen de Jong
ff35ba3696 added warnshadow cli option to enable assembler warnings about symbol shadowing 2023-08-28 16:41:46 +02:00
Irmen de Jong
75514fc7af fix some invalid instructions on 6502 (instead of 65c02) target for bit shifts 2023-08-14 21:58:26 +02:00
Irmen de Jong
bc8126eb16 2x faster word multiplication routine 2023-08-14 18:11:30 +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
151a206617 experimental Commodore PET target 2023-08-12 23:25:07 +02:00
Irmen de Jong
e3fbe37f9f fixed optimized code for >= and <= 2023-08-12 13:45:08 +02:00