Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
Interrupt Mode 2
#1
So... if I was crazy and wanted to implement an IM2 routine.... how do you think I should do it?

Most of my attempts to put the 257 byte data block in memory have failed spectacularly. I think using ORG in an asm statement tends to upset the compiler; but the thing about setting up interrupts is we NEED to put a 257 byte vector table into a very very specific location in memory, and then need to ALSO have the actual interrupt service routine in the specific place that the vector table points to.

*ponder*

I suppose we could use DIM variable @address to force the issue with an array?

On the whole though, this fixed location code isn't really supported by the compiler at this time. How could we work around it?
Reply
#2
Thinking about it some more. If we assume the programmer is smart enough to NOT use the last three pages of memory, we can probably use them for IM2 purposes.

The program below is far from stable. It crashes eventually, depending what you try to do. It's probably because it tries to go back to basic, which we probably don't want to generally. That or I forgot some registers or something.

It's interesting to see it print in the middle of the screen and refuse to move it though!

Code:
sub im2Initialize() DIM addr as uInteger FOR addr=$FD00 TO $FE00 poke addr,$FE NEXT addr POKE $FEFE,243 '; di POKE $FEFF,205 '; CALL POKE Uinteger $FF00,@isr POKE $FF02,251 '; ei POKE $FF03,$ED POKE $FF04,77 '; ED4D reti asm LD A,$FD LD I,A IM 2 end asm return isr: asm ;This is the interrupt Service Rotuine ; First thing, save those registers. PUSH AF PUSH BC PUSH DE PUSH HL PUSH IX EXX PUSH AF PUSH BC PUSH DE PUSH HL end asm Print at 10,10;"Ha!" asm POP HL POP DE POP BC POP AF EXX POP IX POP HL POP DE POP BC POP AF ; This jumps to the ROM ISR. call $0038 ret end asm end sub PRINT "Hello world" im2Initialize()
Reply
#3
How adorable is this?

http://britlion.googlepages.com/Sprity.z80
Reply
#4
That's beyond my awasomeness already. I swear I knew enough about coding to understand what it does (and how!). So I need to study more Smile
Reply
#5
Ironically, once I'd worked out how to make the compiler do an interrupt mode 2 handler, getting your demo to run was scarily easy.

The IM2 code gets called once per interrupt, and hops over to do one run around what used to be the main loop.

So the main loop was renamed with a label instead of "DO" and has a return at the end of it. It's called with a gosub once per interrupt.

That's the whole trick.
Reply
#6
:o Wooow!

Don't know whether you use ZX BASIC to compile that. If you did, I guess -O3 won't work with this, as it won't probably understand IM2 routines, etc...
Again, astonishing!!!
Reply
#7
boriel Wrote::o Wooow!

Don't know whether you use ZX BASIC to compile that. If you did, I guess -O3 won't work with this, as it won't probably understand IM2 routines, etc...
Again, astonishing!!!

I was wondering when you'd notice that little gem of fun. Yes, it's ALL done with ZX Basic and as for -O3, I actually didn't, because I've been avoiding use of O3 etc - or was at that time; you want us to test it now :-)

It shouldn't make a difference, I think, because I sorta used a hammer to make it work. I used the routine listed above - which pokes the IM2 system into place the hard way. I couldn't find a way to let it allow me to put code where I wanted any other way.

Here's the actual whole program (and I know the IM2 bit isn't as efficient as it could be, it was a first pass, and hey, it worked):
Code:
#DEFINE BLACK 0 #DEFINE BLUE 1 #DEFINE RED 2 #DEFINE MAGENTA 3 #DEFINE GREEN 4 #DEFINE CYAN 5 #DEFINE YELLOW 6 #DEFINE WHITE 7 #DEFINE TRANSPARENT 8 #DEFINE CONTRAST 9 #DEFINE TRUE 1 #DEFINE FALSE 0 sub im2Initialize() DIM addr as uInteger FOR addr=$FD00 TO $FE00 poke addr,$FE NEXT addr POKE $FEFE,243 '; di POKE $FEFF,205 '; CALL POKE Uinteger $FF00,@isr POKE $FF02,251 '; ei POKE $FF03,$ED POKE $FF04,77 '; ED4D reti asm LD A,$FD LD I,A IM 2 end asm return isr: asm ;This is the interrupt Service Rotuine ; First thing, save those registers. PUSH AF PUSH BC PUSH DE PUSH HL PUSH IX EXX PUSH AF PUSH BC PUSH DE PUSH HL end asm gosub mainLoop asm POP HL POP DE POP BC POP AF EXX POP IX POP HL POP DE POP BC POP AF ; This jumps to the ROM ISR. call $0038 ret end asm end sub SUB fspInitialize (n as uByte, udgA as uByte, udgB as uByte, udgC as uByte, udgD as uByte,x as uByte,y as uByte) 'Initialize Sprite: n (sprite number 0-3);a,b,c,d (UDG number 0-20,99 means the sprite won't be printed) #ifndef FINAL_CODE IF n>3 OR (udgA>20 AND udgA<>99) or udgB>20 or udgC>20 or udgD>20 then Print "Out Of Bounds Call in ";"fspInitialize" : stop : END IF : REM Bounds checking - so a bad function call doesn't poke all over memory. #endif DIM targetAddr as uInteger targetAddr=@fspDataStart+(48*n) fspErase() POKE targetAddr,udgA targetAddr=targetAddr+1 POKE targetAddr,udgB targetAddr=targetAddr+1 POKE targetAddr,udgC targetAddr=targetAddr+1 POKE targetAddr,udgD targetAddr=targetAddr+1 POKE targetAddr,x targetAddr=targetAddr+1 POKE targetAddr,y targetAddr=targetAddr+1 POKE targetAddr,x targetAddr=targetAddr+1 POKE targetAddr,y fspBufferAndDraw() RETURN fspDataStart: REM This Section contains Fourspriter data and code that is called by other routines. ASM ;; 16x16 Sprites moving a character each frame ;; Copyleft 2009 The Mojon Twins. ;; Uses UDGs (it reads the address from the system variables). ;; It moves up to 4 sprites, preserving the backgrounds. ;; Sprite control is done writing a certain number ;; in the Sprite's first UDG ;; If the first UDG is 99, it won't be printed ; Note that the sprites are initialized to disabled. datap: ; each block is 48 bytes long. [This comment originally said 40 bytes, and it's clearly 48] ;; Sprite 1 udgs1: defb 99,0,0,0 ; Four UDGs of the first sprite. x_pos1: defb 0 ; X position in chars. y_pos1: defb 0 ; Y position in chars. cx_pos1: defb 0 ; Previous X position in chars. cy_pos1: defb 0 ; Previous Y position in chars. buffer1: defb 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 ; Background Gfx buffer. attrs1: defb 6,7,7,7 ; Sprite ATTRs buffatrs1: defb 0,0,0,0 ; Background Attr's buffer ;; Sprite 2 udgs2: defb 99,0,0,0 x_pos2: defb 0 y_pos2: defb 0 cx_pos2: defb 0 cy_pos2: defb 0 buffer2: defb 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 attrs2: defb 7,7,7,7 buffatrs2: defb 0,0,0,0 ;; Sprite 3 udgs3: defb 99,0,0,0 x_pos3: defb 0 y_pos3: defb 0 cx_pos3: defb 0 cy_pos3: defb 0 buffer3: defb 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 attrs3: defb 7,7,7,7 buffatrs3: defb 0,0,0,0 ;; Sprite 4 udgs4: defb 99,0,0,0 x_pos4: defb 0 y_pos4: defb 0 cx_pos4: defb 0 cy_pos4: defb 0 buffer4: defb 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0 attrs4: defb 7,7,7,7 buffatrs4: defb 0,0,0,0 ;; General use coordinates xpos: defb 0 ypos: defb 0 cxpos: defb 0 cypos: defb 0 fspCopyToScreen: ld b,(xpos) ld c,(ypos) call get_scr_address2 ; Returns (xpos, ypos) in HL ld a, (de) ld (hl), a inc h inc de ld a, (de) ld (hl), a inc h inc de ld a, (de) ld (hl), a inc h inc de ld a, (de) ld (hl), a inc h inc de ld a, (de) ld (hl), a inc h inc de ld a, (de) ld (hl), a inc h inc de ld a, (de) ld (hl), a inc h inc de ld a, (de) ld (hl), a inc h inc de ret fspCopyToScreen2: call get_scr_address2 ; Returns (xpos, ypos) in HL ld a, (de) ld (hl), a inc h inc de ld a, (de) ld (hl), a inc h inc de ld a, (de) ld (hl), a inc h inc de ld a, (de) ld (hl), a inc h inc de ld a, (de) ld (hl), a inc h inc de ld a, (de) ld (hl), a inc h inc de ld a, (de) ld (hl), a inc h inc de ld a, (de) ld (hl), a inc h inc de ret ; This function returns the address into HL of the screen address ; (xpos, ypos) in character grid notation. ; Original code was extracted by BloodBaz get_scr_address: push de ld a, (xpos) and 31 ld l, a ld a, (ypos) ld d, a and 24 add a, 64 ld h, a ld a, d and 7 rrca rrca rrca or l ld l, a pop de ret ; This function returns the address into HL of the screen address ; (xpos, ypos) in character grid notation. ; Original code was extracted by BloodBaz get_scr_address2: push de ld a,b ; b=x and 31 ld l, a ld a,c ; c=y ld d, a and 24 add a, 64 ld h, a ld a, d and 7 rrca rrca rrca or l ld l, a pop de ret ;; This function returns the memory address of the Character Position ;; (xpos, ypos) in the attribute screen memory. ;; Adapted from code by Jonathan Cauldwell. get_attr_address: ld a, (ypos) ; Cogemos y rrca rrca rrca ; Multiply by 32 ld l, a ; Pass to L and 3 ; Mask with 00000011 add a, 88 ; 88 * 256 = 22528 - start of attributes. ld h, a ; Put it in the High Byte ld a, l ; We get y value *32 and 224 ; Mask with 11100000 ld l, a ; Put it in L ld a, (xpos); Grab X add a, l ; Add it to the Low byte ld l, a ; Put it back in L, and we're done. HL=Address. ret get_attr_address2: ld a,c ;(ypos) ; Cogemos y rrca rrca rrca ; Multiply by 32 ld l, a ; Pass to L and 3 ; Mask with 00000011 add a, 88 ; 88 * 256 = 22528 - start of attributes. ld h, a ; Put it in the High Byte ld a, l ; We get y value *32 and 224 ; Mask with 11100000 ld l, a ; Put it in L ld a,b; (xpos); Grab X add a, l ; Add it to the Low byte ld l, a ; Put it back in L, and we're done. HL=Address. ret spare: defb 85,85,85,85,85,85,85,85 ;; Routine to save the background to the buffer fspEraseAsm: ld hl, datap ; ld b, 4 ; i4chars2: push bc ;; ld a, (hl) cp 99 jr z, nxt2 ;; A LD BC,6 ;EX DE,HL ; HL now points at the start sprite data instead ADD HL,BC ; HL moved on to point to old x-y positions. LD b,(HL) ; b=x INC HL LD c,(HL) ;c=y INC HL EX DE,HL ; Now it's DE that points into the data buffer. ; Draw first background character call fspCopyToScreen2 ;; B ; xpos++ inc b ; Draw second background Character call fspCopyToScreen2 ;; C ; xpos -- dec b ; ypos ++ inc c ; draw Third background Character call fspCopyToScreen2 ;; D ; xpos++ inc b ; Draw Fourth background Character call fspCopyToScreen2 ;; move to attr inc de inc de inc de inc de ;Set our position to top left corner again ; xpos -- dec b ; ypos -- dec c call get_attr_address2 call copyattrs EX DE,HL nxt2: pop bc djnz i4chars2 ret fspCopyFromScreen: call get_scr_address2 ld a, (hl) ld (de), a inc h inc de ld a, (hl) ld (de), a inc h inc de ld a, (hl) ld (de), a inc h inc de ld a, (hl) ld (de), a inc h inc de ld a, (hl) ld (de), a inc h inc de ld a, (hl) ld (de), a inc h inc de ld a, (hl) ld (de), a inc h inc de ld a, (hl) ld (de), a inc h inc de ret fspBufferAndDrawAsm: ld hl, datap ; ld b, 4 ; i4chars: push bc ;Active Sprite? ld a, (hl) cp 99 jr z, nxt1 ;; A LD BC,4 ADD HL,BC ; move past the udg part ;xpos ld b,(hl) inc hl ;ypos ld c,(hl) inc hl ld a,b ld (xpos),a ld a,c ld (xpos),a ; we won't need this soon. inc hl ; move past the old co-ordinates inc hl EX de,hl call fspCopyFromScreen ;; B ; xpos++ inc b call fspCopyFromScreen ;; C ; xpos -- dec b ; ypos ++ inc c call fspCopyFromScreen ;; D ; xpos++ inc b call fspCopyFromScreen ;; Now we point to the ATTR buffer,adding 4: inc de inc de inc de inc de ;Put ourselves back at the top left corner. ; xpos -- dec b ; ypos -- dec c call get_attr_address2 ld bc, 32 ;First Character LDI ; bc was loaded with 32 instead of 31 to account for the BC-- this instruction has. ;Second Character LDI add hl, bc ; we can get away with this because BC++ as well as hl-- ;Third Character LDI ;Fourth Character LDI EX DE,HL ; back to an HL pointer. nxt1: pop bc djnz i4chars ; Go straight into Painting new sprites for speed. ;; Print sprites routine ;; UDGs are labeled from 0 to 21. The addres of the UDG is: ;; *(23675) + 256 * *(23676) + 8 * N fspDrawAsm: draw_sprites: ld hl, datap ; ld b, 4 ; i4chars3: push bc ld a, (hl) cp 99 jr z, nxt3 ;; Copy pointer "HL" to alt reg push HL exx pop HL exx ld bc,4 add hl,bc ; Get xpos LD a,(hl) ld b,a ;x inc hl ; Get ypos LD a,(hl) ld c,a ;y inc hl push HL ; de Save pointer. EXX LD A,(HL) INC HL EXX call fspGetUdgAddr ;Draw the first character call fspCopyToScreen2 ; xpos++ inc b ; Address of the graphics EXX LD A,(HL) INC HL EXX call fspGetUdgAddr ; Draw the second Character call fspCopyToScreen2 ; xpos-- dec b ; ypos ++ inc c ; Address of Graphics EXX LD A,(HL) INC HL EXX call fspGetUdgAddr ; Draw the third character call fspCopyToScreen2 ; xpos ++ inc b ; Address of next Graphic EXX LD A,(HL) INC HL EXX call fspGetUdgAddr ; Draw the fourth Character call fspCopyToScreen2 pop hl ; Recover hl to point at the start of the pixel buffer ; hl = hl + 32 ld a,l add a,34 ld l,a jp nc, fspNoIncH inc h ; hl now points to the attr data. fspNoIncH: EX DE,HL ;Reset position to top left. ; xpos -- dec b ; ypos -- dec c ;; attr call get_attr_address2 call copyattrs EX DE,HL LD BC,4 ADD HL,BC ; Move pointer to start of next block nxt3: pop bc djnz i4chars3 ret fspGetUdgAddr: ; finds the address of the UDG # in A ld HL,(23675) rlca rlca rlca ; a = N * 8 ld d, 0 ld e, a add hl, de EX DE,HL ret copyattrs: ld bc, 32 EX DE,HL ;First LDI ;Second LDI ; DE=DE+BC EX DE,HL add hl,bc EX DE,HL ;Third LDI ;Fourth LDI EX DE,HL ret fspUpdateAsm: ; This name will be used from ASM update_coordinates: ; For each Sprite: ;; *(datap + 6) = *(datap + 4) ;; *(datap + 7) = *(datap + 5) ld hl, datap+4 ; Points to sprite 1 ld de, datap+6 ldi ldi ld hl, datap+4+48 ; Points to sprite 2 ld de, datap+6+48 ldi ldi ld hl, datap+4+48+48 ; Points to sprite 3 ld de, datap+6+48+48 ldi ldi ld hl, datap+4+48+48+48 ; Points to sprite 4 ld de, datap+6+48+48+48 ldi ldi ret END ASM END SUB SUB fspDisable (n as UByte) #ifndef FINAL_CODE IF n>3 then Print "Out Of Bounds Call in ";"fspDisable" : stop: END IF #endif POKE @fspDataStart+48*n,99 end SUB SUB fspCoord (n as uByte, x as uByte, y as uByte) 'Set sprite coords: n (sprite number);x,y (vertical,horizontal coords) #ifndef FINAL_CODE IF n>3 then Print "Out Of Bounds Call in ";"fspCoord" : stop: END IF #endif DIM targetAddr as uInteger targetAddr=@fspDataStart+4+48*n POKE targetAddr,x targetAddr=targetAddr+1 POKE targetAddr,y rem targetAddr=targetAddr+1 rem POKE targetAddr,x rem targetAddr=targetAddr+1 rem POKE targetAddr,y END SUB SUB fspAttrs (n as uByte, attra as uByte, attrb as uByte, attrc as uByte, attrd as uByte) 'Set sprite attrs: n (sprite number);a,b,c,d (UDG attrs) #ifndef FINAL_CODE IF n>3 then Print "Out Of Bounds Call in ";"fspAttrs" : stop: END IF #endif DIM targetAddr as uInteger targetAddr=@fspDataStart+40+48*n POKE targetAddr,attra targetAddr=targetAddr+1 POKE targetAddr,attrb targetAddr=targetAddr+1 POKE targetAddr,attrc targetAddr=targetAddr+1 POKE targetAddr,attrd END SUB FUNCTION fspAttrByte(fspInk as uByte,fspPaper as uByte,fspBright as uByte, fspFlash as uByte) as uByte #ifndef FINAL_CODE if fspInk > 7 OR fspPaper > 7 OR fspBright > 1 or fspFlash >1 then Print "Out Of Bounds Call in ";"fspAttrByte" : stop: END IF #endif return (fspFlash shl 7) + (fspBright shl 6) + (fspPaper shl 3) + fspInk END FUNCTION SUB fspAttr(n as uByte,fspInk as uByte,fspPaper as uByte,fspBright as uByte, fspFlash as uByte) #ifndef FINAL_CODE if fspInk > 7 OR fspPaper > 7 OR fspBright > 1 or fspFlash >1 then Print "Out Of Bounds Call in ";"fspAttr" : stop: END IF #endif DIM attrByte as uByte attrByte=fspAttrByte(fspInk,fspPaper,fspBright,fspFlash) fspAttrs(n,attrByte,attrByte,attrByte,attrByte) END SUB SUB FASTCALL fspErase() asm call fspEraseAsm end asm END SUB SUB FASTCALL fspBufferAndDraw() asm call fspBufferAndDrawAsm end asm END SUB SUB FASTCALL fspUpdate() asm call fspUpdateAsm END ASM END SUB SUB fspRedraw() asm ;halt we're in IM2, so no halting. NO! BAD! ; Wait for scan line LD BC,1930 ; Reduced this to the bare minimum to allow scanline to hit bottom of screen. Your program may need more or less. fspRedrawloop: DEC BC LD A,B OR C JR NZ,fspRedrawloop ;REM Erase the sprites call fspEraseAsm ;REM Save background and ;REM print sprites call fspBufferAndDrawAsm ;REM update coordinates call fspUpdateAsm end asm END SUB REM below this is Apenao's Demo Program: #include <sinclair.bas> #include <keys.bas> 100 DIM gentle (0 to 3,0 to 7) AS uByte => { { 15, 15, 15, 15, 15, 15, 13, 15} , _ { 240, 144, 208, 208, 240, 240, 176, 240} , _ { 15, 14, 63, 0, 0, 12, 26, 30} , _ { 176, 112, 252, 0, 0, 48, 104, 120}} 110 POKE Uinteger 23675, @gentle(0,0) CLS 'for n = 0 to 15 'print paper (n mod 7); ink ((n+3) mod 8); "ABCDEFGHIJKLMNOPQRSTUVWXYZ123456" 'next n LET gx=10:LET gy=10:let dx=3:let dy=3:let tx=6:let ty=6:let cx=18:let cy=18 let dmx=1:let dmy=1:let tmx=-1:let tmy=-1:let cmx=1:let cmy=-1 fspInitialize (0,0,1,2,3,gx,gy) : REM (sprite number, udgA, udgB,udgC,udgD, initial X co-ordinate, initial y-coordinate) fspInitialize (1,0,1,2,3,dx,dy) fspInitialize (2,0,1,2,3,tx,ty) fspInitialize (3,0,1,2,3,cx,cy) fspAttr(0,WHITE,MAGENTA,FALSE,FALSE) : REM (sprite number, ink, paper, bright, flash) fspAttr(1,YELLOW,BLACK,FALSE,FALSE) fspAttr(2,GREEN,BLUE,TRUE,FALSE) fspAttr(3,RED,WHITE,FALSE,FALSE) goto endofProgram mainLoop: IF MULTIKEYS(KEYO) and gx>0 THEN LET gx=gx-1: END IF IF MULTIKEYS(KEYP) and gx<30 THEN LET gx=gx+1:END IF IF MULTIKEYS(KEYQ) and gy>0 THEN LET gy=gy-1 :END IF IF MULTIKEYS(KEYA) and gy<22 THEN LET gy=gy+1: END IF let dx=dx+dmx:if dx=0 or dx=30 then let dmx=-dmx : END IF let dy=dy+dmy:if dy=0 or dy=22 then let dmy=-dmy: END IF let tx=tx+tmx:if tx=0 or tx=30 then let tmx=-tmx: END IF let ty=ty+tmy:if ty=0 or ty=22 then let tmy=-tmy: END IF let cx=cx+cmx:if cx=0 or cx=30 then let cmx=-cmx: END IF let cy=cy+cmy:if cy=0 or cy=22 then let cmy=-cmy: END IF fspCoord (0,gx,gy) fspCoord (1,dx,dy) fspCoord (2,tx,ty) fspCoord (3,cx,cy) fspRedraw() return endofProgram: fspErase() cls REM let's call a few things before we drop out to the IM2 mode so that the -O3 doesn't kill it REM Sadly, It does break it anyway! fspUpdate() gosub mainLoop im2Initialize()

O# Test: It does indeed compile without a hitch, and -O3 does indeed break it HARD!. At first it was because a lot of functions were not called; so I called them. Not quite sure what it's optimizing out that would break it now.

Anyway, that's how I did it; like I said sort of hammering it into place. But it definitely worked.
Reply
#8
britlion Wrote:
boriel Wrote::o Wooow!

Don't know whether you use ZX BASIC to compile that. If you did, I guess -O3 won't work with this, as it won't probably understand IM2 routines, etc...
Again, astonishing!!!

I was wondering when you'd notice that little gem of fun.
I was busy fixing -O3 up (I deeply rewrote some routines of the module). So I could only test this great program once I finished fixing it.

britlion Wrote:Yes, it's ALL done with ZX Basic and as for -O3, I actually didn't, because I've been avoiding use of O3 etc - or was at that time; you want us to test it now :-)

It shouldn't make a difference, I think, because I sorta used a hammer to make it work. I used the routine listed above - which pokes the IM2 system into place the hard way. I couldn't find a way to let it allow me to put code where I wanted any other way.
Ok, this is is something hard to do. I mean, if you use ORG XXXXX It will place the following code at that place, including the rest of the basic program.
Idea Idea: Why don't you put this in the wishlist (yes, it will take MUCH time, but not discarded)? For example:
  • Capability to declare functions or sub at a given memory address, like when declaring a variable with DIM .. AT @.... Example:
    Code:
    SUB MyInterrupt AT 0xFFE0h [...] END SUB
  • Capability to declare sub interrupts (this is already used in other compilers).
    Code:
    SUB INTERRUPT MyInterrupt : REM No parameters and no return. Must be a SUB [...] END SUB
    or more idiomatic
    Code:
    SUB Spritty AS INTERRUPT [...] END SUB
Some compilers allows to declare several interrupts and chain them, but I think one should suffice.
Britlion Wrote:Here's the actual whole program (and I know the IM2 bit isn't as efficient as it could be, it was a first pass, and hey, it worked):
I also thought it was. Again, thank you!

britlion Wrote:[... code ... ]

O# Test: It does indeed compile without a hitch, and -O3 does indeed break it HARD!. At first it was because a lot of functions were not called; so I called them. Not quite sure what it's optimizing out that would break it now.

Anyway, that's how I did it; like I said sort of hammering it into place. But it definitely worked.
Ok
Nice to hear that. Tongue Regarding to the routines being removed by -On, yes this is a problem I'm addressing. The compiler does not know anything about user's ASM block content (that's at the assembling stage). The reason is this is a "generic architecture" compiler (or a wanna be). So, for example, if I port the compiler to Commodore or PS3 (yes, it could be done, in fact), the assembler will be completely different. -O3 module has problem with this, because it's the nexus between BASIC layer (frontend) and asm/target machine layer (backend). In fact, such module must be completely rewritten for every architecture.

Having said that, I've thought some solutions:
  1. Put function calls at the end of the program, if you don't want them to be removed.
  2. The above does not prevent Variables to be removed, so if the ASM block write on basic variables,
    this could lead to another problem. So we can tell the compiler which identifiers (Variables, SUBs, Functions) our ASM block requires:
    Code:
    DIM MyVar1, MyVar2 as uInteger SUB MyFunction [...] END SUB ASM USES { MyFunction, MyVar1, MyVar2 } [...] call MyFunction ld (MyVar1), hl END ASM
    or alike
What do you think? If you like this, please, add it in the wishlist :wink:
Reply
#9
I love some of these ideas.

I think SUb <name> as interrupt has fascinating possibilities - not least of which that the compiler could assume this entry point will happen, and follow that thread to see what variables and functions are used, as well as main.

It probably means two passes, but it's a way of allowing -O3 to not exclude the interrupt path of the program.

Also, if it can work out where it has the memory to do the vector table and ISR (probably in 2 pages at the end of the program, if there's room), then that's awesome.

I hate to waste memory though. I think I'll wishlist a memory map option too.
Reply
#10
Shame this doesnt seem to work properly any more. Looks like an issue with fourspriter
Reply


Forum Jump:


Users browsing this thread: 1 Guest(s)