AlanD Posted March 10, 2002 Share Posted March 10, 2002 Ok I need some help. I found a really good compression routine for the C64. The compressor is on the PC and the decompressor is in ASM for C64. It looks very feasible to use this but I have been trying to set it up and it'll only decompress about 4 bytes until it goes stupid and starts spewing garbage I'm not sure where I am going wrong here. Can anyone that is better at ASM than I am take a quick look at this program and see what they think? I talked to the original programmer so I know it does need to run out of RAM since it self-modifies. I would be glad to chat on the developers board about it and see what we can work out on it. The address for the program is Pucrunch. Also the designer said I was trying to do something similar to what he did with his Vimm demo in the Vicpic part. Address for info on this is Vimm. This is a major stumbling block for me and the further progress of Adventure 2. I have spent a week fighting with this. I would post the code I was using but it would be nasty spaghetti code because the original program is DASM and I use TASM and I put a lot of it in as .byte includes so would be useless for error detecting. At this point it would be better to get a fresh start and some help. Thanks again, AlanD Quote Link to comment Share on other sites More sharing options...
Heaven/TQA Posted March 11, 2002 Share Posted March 11, 2002 Alan, here am I... i give you all my help... i'll check it when i am at work... hve Quote Link to comment Share on other sites More sharing options...
Heaven/TQA Posted March 11, 2002 Share Posted March 11, 2002 maybe we can combine pucrunch and flashpack because both work with the same (?) pack method... but i am not sure and could be completly wrong... i have to check... hve Quote Link to comment Share on other sites More sharing options...
Thomas Jentzsch Posted March 11, 2002 Share Posted March 11, 2002 The assembler code looks very straight forward and there are only a few places where selfmodifying code is used. In most cases, changing the code into indirect access would solve the problem without to much performance loss. Example: code: INPOS = *+1 lda $aaaa change into code: INPOS .word ldy #0 lda (INPOS),y And perhaps you have to save and restore Y. [ 03-11-2002: Message edited by: Thomas Jentzsch ] Quote Link to comment Share on other sites More sharing options...
AlanD Posted March 12, 2002 Author Share Posted March 12, 2002 The code seemed to compile without a problem and the run from RAM was fine. Where I hit problems was when I tried to decompress a screen. There were 4 bytes that uncompressed fine then the last two should have been repeated say 10 times. At that point it just spit out garbage. I think there is an error somewhere in the LZW decompression routine but I can't find it unfortunately. Not sure where it went wrong. Also it seems the first two bytes of the header of the compressed file are never read off anywhere. Not sure why they are there either. I would really love to get this working since out of two screens I compressed I ended up averaging about 20% of original file size which is better than I hoped for. AlanD Quote Link to comment Share on other sites More sharing options...
Thomas Jentzsch Posted March 12, 2002 Share Posted March 12, 2002 I couldn't download the source codes (403, access denied). Could you mail them to me and perhaps your modifications too? Perhaps I find the time to have a closer look. Quote Link to comment Share on other sites More sharing options...
AlanD Posted March 13, 2002 Author Share Posted March 13, 2002 Ok I finally managed to find a few errors so it tries a little more but still only gets 4 bytes into the decompression and screws up. I can post more if needed. Just tell me what you need. I found a few errors such as every #$00 in the original file compiled as a #$20 which is simply an issue with me and DASM as I think that is where the error occured. The other error was in my copy routine when I move it to RAM at $1500. I was rolling my loop wrong and was missing $15FF. I fixed that. I did a byte for byte compare and now $1500 is storing exactly what is stored in the ROM at $b000 ($7000 in this disassemly as it doesn't accout for the beginning of the ROM being at $4000). Any help you guys can offer would be greatly appreciated, Alan code: ORG $1500 ; Call with X = HI of packed data, Y = LO of packed data ; Returns exec address in X = HI and Y = LO ; Carry will be set for error, cleared for OK decompress sty INPOS ; Setup read pointer stx INPOS+1 jsr getbyt ; 'p' jsr getbyt ; 'u' jsr getbyt ; skip endAddr jsr getbyt jsr getbyt sta esc+1 ; starting escape jsr getbyt ; read startAddr sta OUTPOS jsr getbyt sta OUTPOS+1 jsr getbyt ; read # of escape bits sta escB0+1 sta escB1+1 lda #8 sec sbc escB1+1 sta noesc+1 ; 8-escBits jsr getbyt sta mg+1 ; maxGamma + 1 lda #9 sec sbc mg+1 ; 8 - maxGamma == (8 + 1) - (maxGamma + 1) sta longrle+1 jsr getbyt sta mg1+1 ; (1< asl clc sbc #0 sta mg21+1 ; (2< jsr getbyt sta elzpb+1 jsr getbyt ; exec address sta lo+1 ; lo jsr getbyt sta hi+1 ; hi jsr getbyt ; rleUsed ldx #0 tay t1 beq t2 ; Y == 0 ? jsr getbyt sta rletabl,x inx dey bne t1 t2 lda #$80 ; setup bit store - $80 means empty sta bitstr+1 jmp main getbyt jsr getnew bitstr lda #0 ror rts newesc ldy esc+1 ; remember the old code (top bits for escaped byte) escB0 ldx #2 ; ** PARAMETER 0..8 jsr getchkf ; get & save the new escape code sta esc+1 tya ; pre-set the bits ; Fall through and get the rest of the bits. noesc ldx #6 ; ** PARAMETER 8..0 jsr getchkf jsr putch ; output the escaped/normal byte ; Fall through and check the escape bits again main ldy #0 ; Reset to a defined state tya ; A = 0 escB1 ldx #2 ; ** PARAMETER 0..8 jsr getchkf ; X = 0 esc cmp #0 bne noesc ; Fall through to packed code jsr getval ; X = 0 sta LZPOS+1 ; xstore - save the length for a later time lsr ; cmp #1 ; LEN == 2 ? (A is never 0) bne lz77 ; LEN != 2 -> LZ77 ;tya ; A = 0 jsr get1bit ; X = 0 lsr ; bit -> C, A = 0 bcc lz77_2 ; A=0 -> LZPOS+1 ;***FALL THRU*** ; e..e01 jsr get1bit ; X = 0 lsr ; bit -> C, A = 0 bcc newesc ; e..e010 ;***FALL THRU*** ; e..e011 srle iny ; Y is 1 bigger than MSB loops jsr getval ; Y is 1, get len, X = 0 sta LZPOS+1 ; xstore - Save length LSB mg1 cmp #64 ; ** PARAMETER 63-64 -> C clear, 64-64 -> C set.. bcc chrcode ; short RLE, get bytecode longrle ldx #2 ; ** PARAMETER 111111xxxxxx jsr getbits ; get 3/2/1 more bits to get a full byte, X = 0 sta LZPOS+1 ; xstore - Save length LSB jsr getval ; length MSB, X = 0 tay ; Y is 1 bigger than MSB loops chrcode jsr getval ; Byte Code, X = 0 tax ; this is executed most of the time anyway lda rletabl-1,x ; Saves one jump if done here (loses one txa) cpx #32 ; 31-32 -> C clear, 32-32 -> C set.. bcc t3 ; 1..31, we got the right byte from the table ; Ranks 32..64 (11111°xxxxx), get byte.. txa ; get back the value (5 valid bits) ldx #3 jsr getbits ; get 3 more bits to get a full byte, X = 0 t3 ldx LZPOS+1 ; xstore - get length LSB inx ; adjust for cpx#$ff;bne -> bne dorle jsr putch dex bne dorle ; xstore 0..255 -> 1..256 dey bne dorle ; Y was 1 bigger than wanted originally mainbeq beq main ; reverse condition -> jump always lz77 jsr getval ; X = 0 mg21 cmp #127 ; ** PARAMETER Clears carry (is maximum value) bne noeof ; EOF hi ldx #0 lo ldy #0 rts noeof sbc #0 ; C is clear -> subtract 1 (1..126 -> 0..125) elzpb ldx #0 ; ** PARAMETER (more bits to get) jsr getchkf ; clears Carry, X = 0 lz77_2 sta LZPOS+2 ; offset MSB jsr getbyte ; clears Carry, X = 0 ; Note: Already eor:ed in the compressor.. ;eor #255 ; offset LSB 2's complement -1 (i.e. -X = ~X+1) adc OUTPOS ; -offset -1 + curpos (C is clear) ldx LZPOS+1 ; xstore = LZLEN (read before it's overwritten) sta LZPOS+1 lda OUTPOS+1 sbc LZPOS+2 ; takes C into account sta LZPOS+2 ; copy X+1 number of chars from LZPOS to OUTPOS ;ldy #0 ; Y was 0 originally, we don't change it inx ; adjust for cpx#$ff;bne -> bne lzloop LZPOS lda $aaaa,y jsr putch ; Note: must be copied forwards! iny ; Y does not wrap because X=0..255 and Y initially 0 dex bne lzloop ; X loops, (256,1..255) beq mainbeq ; jump through another beq (-1 byte, +3 cycles) getnew pha ; 1 Byte/3 cycles INPOS = *+1 lda $aaaa ; ** PARAMETER inc INPOS bne t4 inc INPOS+1 t4 sec rol ; Shift out the next bit and ; shift in C=1 (last bit marker) sta bitstr+1 ; bitstr initial value = $80 == empty pla ; 1 Byte/4 cycles rts ; 25+12 = 37 ; getval : Gets a 'static huffman coded' value ; ** Scratches X, returns the value in A ** getval inx ; X <- 1 txa ; set the top bit (value is 1..255) gv0 asl bitstr+1 bne t5 jsr getnew t5 bcc getchk ; got 0-bit inx mg cpx #7 ; ** PARAMETER unary code maximum length + 1 bne gv0 beq getchk ; inverse condition -> jump always ; getval: 18 bytes ; 15 + 17*n + 6+15*n+12 + 36*n/8 = 33 + 32*n + 36*n/8 cycles ; getbits: Gets X bits from the stream ; ** Scratches X, returns the value in A ** getbyte ldx #7 get1bit inx ;2 getbits asl bitstr+1 bne t6 jsr getnew t6 rol ;2 getchk dex ;2 more bits to get ? getchkf bne getbits ;2/3 clc ;2 return carry cleared rts ;6+6 OUTPOS = *+1 ; ZP putch sta $aaaa ; ** parameter inc OUTPOS ; ZP bne t7 inc OUTPOS+1 ; ZP t7 rts rletabl .byte 0,0,0,0,0,0,0 .byte 0,0,0,0,0,0,0,0 .byte 0,0,0,0,0,0,0,0 .byte 0,0,0,0,0,0,0,0 here is the disassembly of what ended up in my 5200 bin. code: 00007000: 8C2B16 STY $162B 00007003: 8E2C16 STX $162C 00007006: 207E15 JSR $157E 00007009: 207E15 JSR $157E 0000700C: 207E15 JSR $157E 0000700F: 207E15 JSR $157E 00007012: 207E15 JSR $157E 00007015: 8DA215 STA $15A2 00007018: 207E15 JSR $157E 0000701B: 8D6116 STA $1661 0000701E: 207E15 JSR $157E 00007021: 8D6216 STA $1662 00007024: 207E15 JSR $157E 00007027: 8D8915 STA $1589 0000702A: 8D9D15 STA $159D 0000702D: A908 LDA #$08 0000702F: 38 SEC 00007030: ED9D15 SBC $159D 00007033: 8D9215 STA $1592 00007036: 207E15 JSR $157E 00007039: 8D4A16 STA $164A 0000703C: A909 LDA #$09 0000703E: 38 SEC 0000703F: ED4A16 SBC $164A 00007042: 8DC615 STA $15C6 00007045: 207E15 JSR $157E 00007048: 8DC215 STA $15C2 0000704B: 0A ASL A 0000704C: 18 CLC 0000704D: E900 SBC #$00 0000704F: 8DF515 STA $15F5 00007052: 207E15 JSR $157E 00007055: 8D2016 STA $1620 00007058: 207E15 JSR $157E 0000705B: 8DFB15 STA $15FB 0000705E: 207E15 JSR $157E 00007061: 8DF915 STA $15F9 00007064: 207E15 JSR $157E 00007067: A200 LDX #$00 00007069: A8 TAY 0000706A: F00A BEQ $0A 0000706C: 207E15 JSR $157E 0000706F: 9D6C16 STA $166C,X 00007072: E8 INX 00007073: 88 DEY 00007074: D0F4 BNE $F4 00007076: A980 LDA #$80 00007078: 8D8215 STA $1582 0000707B: 4C9915 JMP $1599 0000707E: 202916 JSR $1629 00007081: A900 LDA #$00 00007083: 6A ROR A 00007084: 60 RTS 00007085: ACA215 LDY $15A2 00007088: A202 LDX #$02 0000708A: 205C16 JSR $165C 0000708D: 8DA215 STA $15A2 00007090: 98 TYA 00007091: A206 LDX #$06 00007093: 205C16 JSR $165C 00007096: 206016 JSR $1660 00007099: A000 LDY #$00 0000709B: 98 TYA 0000709C: A202 LDX #$02 0000709E: 205C16 JSR $165C 000070A1: C900 CMP #$00 000070A3: D0EC BNE $EC 000070A5: 203C16 JSR $163C 000070A8: 8D1E16 STA $161E 000070AB: 4A LSR A 000070AC: D043 BNE $43 000070AE: 205116 JSR $1651 000070B1: 4A LSR A 000070B2: 9050 BCC $50 000070B4: 205116 JSR $1651 000070B7: 4A LSR A 000070B8: 90CB BCC $CB 000070BA: C8 INY 000070BB: 203C16 JSR $163C 000070BE: 8D1E16 STA $161E 000070C1: C940 CMP #$40 000070C3: 900C BCC $0C 000070C5: A202 LDX #$02 000070C7: 205216 JSR $1652 000070CA: 8D1E16 STA $161E 000070CD: 203C16 JSR $163C 000070D0: A8 TAY 000070D1: 203C16 JSR $163C 000070D4: AA TAX 000070D5: BD6B16 LDA $166B,X 000070D8: E020 CPX #$20 000070DA: 9006 BCC $06 000070DC: 8A TXA 000070DD: A203 LDX #$03 000070DF: 205216 JSR $1652 000070E2: AE1E16 LDX $161E 000070E5: E8 INX 000070E6: 206016 JSR $1660 000070E9: CA DEX 000070EA: D0FA BNE $FA 000070EC: 88 DEY 000070ED: D0F7 BNE $F7 000070EF: F0A8 BEQ $A8 000070F1: 203C16 JSR $163C 000070F4: C97F CMP #$7F 000070F6: D005 BNE $05 000070F8: A200 LDX #$00 000070FA: A000 LDY #$00 000070FC: 60 RTS 000070FD: E900 SBC #$00 000070FF: A200 LDX #$00 00007101: 205C16 JSR $165C 00007104: 8D1F16 STA $161F 00007107: 204F16 JSR $164F 0000710A: 6D6116 ADC $1661 0000710D: AE1E16 LDX $161E 00007110: 8D1E16 STA $161E 00007113: AD6216 LDA $1662 00007116: ED1F16 SBC $161F 00007119: 8D1F16 STA $161F 0000711C: E8 INX 0000711D: B9AAAA LDA $AAAA,Y 00007120: 206016 JSR $1660 00007123: C8 INY 00007124: CA DEX 00007125: D0F6 BNE $F6 00007127: F0C6 BEQ $C6 00007129: 48 PHA 0000712A: ADAAAA LDA $AAAA 0000712D: EE2B16 INC $162B 00007130: D003 BNE $03 00007132: EE2C16 INC $162C 00007135: 38 SEC 00007136: 2A ROL A 00007137: 8D8215 STA $1582 0000713A: 68 PLA 0000713B: 60 RTS 0000713C: E8 INX 0000713D: 8A TXA 0000713E: 0E8215 ASL $1582 00007141: D003 BNE $03 00007143: 202916 JSR $1629 00007146: 9013 BCC $13 00007148: E8 INX 00007149: E007 CPX #$07 0000714B: D0F1 BNE $F1 0000714D: F00C BEQ $0C 0000714F: A207 LDX #$07 00007151: E8 INX 00007152: 0E8215 ASL $1582 00007155: D003 BNE $03 00007157: 202916 JSR $1629 0000715A: 2A ROL A 0000715B: CA DEX 0000715C: D0F4 BNE $F4 0000715E: 18 CLC 0000715F: 60 RTS 00007160: 8DAAAA STA $AAAA 00007163: EE6116 INC $1661 00007166: D003 BNE $03 00007168: EE6216 INC $1662 0000716B: 60 RTS 0000716C: 00 BRK 0000716D: 00 BRK 0000716E: 00 BRK 0000716F: 00 BRK 00007170: 00 BRK 00's continue on for another 36 bytes or so Quote Link to comment Share on other sites More sharing options...
AlanD Posted March 13, 2002 Author Share Posted March 13, 2002 Grrrr Can anyone confirm that some of my jump addresses or self-modifying addresses are off? I'm going thru it line by line now code: 0000711D: B9AAAA LDA $AAAA,Y 00007120: 206016 JSR $1660 00007123: C8 INY 00007124: CA DEX That area when I disassemble memory in the 5200 after a partial run when it crashes is showing up as: code: B9 02 10 LDA $1002,Y ;looks ok as my screen is $1000 00 BRK ;huh? 60 RTS 16 C8 ASL $C8,X CA DEX well you can see that is quite a bit different. Seems something is modifying $1620 which is where that break appeared. Now just to figure out where and why. I think there were some mild problem when I did the assembly which are coming to haunt me Not sure where I went wrong tho. AlanD Quote Link to comment Share on other sites More sharing options...
AlanD Posted March 14, 2002 Author Share Posted March 14, 2002 Ok that $1620 being modified was the problem. All fixed now. It seems somehow that when my code was put into the source a bunch of the $00 changed to $20. Can you guess that it was supposed to be $1600 being modified and not $1620? Seems I missed one of the screwups. All fixed except it isn't loading in the colors and that is just a modification I need to do from when I began the changeover to compression. Thanks for everyone's help! Alan 1 Quote Link to comment Share on other sites More sharing options...
ndary Posted April 25, 2002 Share Posted April 25, 2002 what is the ratio?.. how much you can decompress compare to other packing utility for atari? Quote Link to comment Share on other sites More sharing options...
vesabios Posted December 2, 2021 Share Posted December 2, 2021 Did you ever get this to work (18 years ago!)? I tried the code you posted here and didn't get very far. It compiles but crashes. I'm including a binary file that I compressed with pucrunch (-d -c0) and trying to pass in the address in X/Y but it hangs. If you have working source you could share I'd much appreciate it --- just starting to learn my way around a8/6502 Quote Link to comment Share on other sites More sharing options...
Like us on Facebook
Follow us on Twitter
AtariAge on Instagram
Watch us on YouTube
Chat on Discord
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.