QLiberator decompiler

[mk79]

The 'dc.w $FF00 ; WINDS'. If I remember rightly, the QLIB source reads it as a byte, rather than a word, and treats it as a true/false flag.

Code: Select all

ROM:00010456 ; =============== S U B R O U T I N E =======================================
ROM:00010456
ROM:00010456
ROM:00010456 open_con_channels:                      ; CODE XREF: print_init_error:loc_11F86↓p
ROM:00010456                 cmpi.w  #4,obj_some_filetype(a5)
ROM:0001045C                 blt.s   check_mode
ROM:0001045E                 tst.w   obj_winds(a5)
ROM:00010462                 bmi.s   open_con_channels_exit
ROM:00010464
ROM:00010464 check_mode:                             ; CODE XREF: open_con_channels+6↑j
ROM:00010464                 moveq   #-1,d1
ROM:00010466                 moveq   #-1,d2
ROM:00010468                 moveq   #sms.dmod,d0    ; set the Display MODe
ROM:0001046A                 trap    #1
ROM:0001046C                 lea     con_m4_ch0(pc),a3 ; "CON_512x050a00x206"
ROM:00010470                 tst.b   d1
ROM:00010472                 beq.s   ch_start
ROM:00010474                 lea     con_m8_ch0(pc),a3 ; "CON_448x040a32x216"
ROM:00010478
ROM:00010478 ch_start:                               ; CODE XREF: open_con_channels+1C↑j
ROM:00010478                 moveq   #0,d4           ; Channel 0
ROM:0001047A
ROM:0001047A ch_loop:                                ; CODE XREF: open_con_channels+46↓j
ROM:0001047A                 moveq   #ch.len,d5
ROM:0001047C                 mulu.w  d4,d5
ROM:0001047E                 add.l   sb_chanb(a6),d5 ; long    channel table base
ROM:00010482                 move.w  ch_chid(a6,d5.l),d0
ROM:00010486                 bpl.s   ch_exists
ROM:00010488                 bsr.w   open_con
ROM:0001048C                 move.l  a0,ch_chid(a6,d5.l)
ROM:00010490                 bra.s   ch_next
ROM:00010492 ; ---------------------------------------------------------------------------
ROM:00010492
ROM:00010492 ch_exists:                              ; CODE XREF: open_con_channels+30↑j
ROM:00010492                 adda.w  (a3)+,a3        ; Skip CON string
ROM:00010494                 addq.l  #2,a3           ; Plus size bytes
ROM:00010496
ROM:00010496 ch_next:                                ; CODE XREF: open_con_channels+3A↑j
ROM:00010496                 addq.b  #1,d4           ; Next channel
ROM:00010498                 cmpi.b  #3,d4           ; 3 channels total
ROM:0001049C                 bne.s   ch_loop
ROM:0001049E
ROM:0001049E open_con_channels_exit:                 ; CODE XREF: open_con_channels+C↑j
ROM:0001049E                 rts
ROM:0001049E ; End of function open_con_channels

This is as far as I'm aware the only reference to obj_winds, so from a runtime standpoint this is a word.

You say you have done the private QLIB_obj extensions, Do you mean the embedded SuperBASIC extensions, with all the pointer interface and GENCODE type extensions?

Yes, not yet finished as some stuff is pretty difficult to figure out, but here's one completely finished example:

Code: Select all

		section code
		
		include 'dev8_keys_sbasic'
		include 'dev8_keys_qdos_sms'
		include 'dev8_keys_qdos_ioa'
		include 'dev8_keys_qdos_io'
		include 'dev8_keys_qlv'
		include 'dev8_keys_err'
		

; BASIC variables
sbq_qerr_old:    equ $C2		; WORD Error number as returned by old Q_ERR
sbq_qerr_wrappers: equ $DC
sbq_data:        equ $E0
sbq_runtime:     equ $EC		; Pointer to QLIB runtime

; GEN block
gb_chid:         equ 4
gb_pos:          equ 8
gb_align:        equ $B
gb_line_table:   equ $C                 ; Line table pointer

gen_start:                              
                lea     gen_procs(pc),a1
                movea.w sb.inipr,a2     ; INITialise PRocedure table
                jsr     (a2)
                moveq   #0,d0
                rts
; ---------------------------------------------------------------------------
gen_procs:      dc.w 12                 
                dc.w initgen-*
                dc.b 7,'INITgen'
                dc.w genop-*
                dc.b 5,'GENop'
                dc.w genb-*
                dc.b 4,'GENB',0
                dc.w genw-*
                dc.b 4,'GENW',0
                dc.w genl-*
                dc.b 4,'GENL',0
                dc.w genint-*
                dc.b 6,'GENint',0
                dc.w genfloat-*
                dc.b 8,'GENfloat',0
                dc.w gensfloat-*
                dc.b 9,'GENsfloat'
                dc.w genstring-*
                dc.b 9,'GENstring'
                dc.w gentext-*
                dc.b 7,'GENtext'
                dc.w genntd-*
                dc.b 6,'GENntd',0
                dc.w smem-*
                dc.b 4,'SMEM',0
                dc.w 0
                dc.w 3
                dc.w rp_addr-*
                dc.b 7,'RP_ADDR'
                dc.w odd-*
                dc.b 3,'ODD'
                dc.w shortf-*
                dc.b 6,'SHORTF',0
                dc.w 0

; =============== S U B R O U T I N E =======================================

; INITGEN #ch,pos
;
; pos = position, will be counted up

initgen:                                
                bsr     gen_get_chan_param ; Return channel ID in d7
                bne.s   initgen_rts
                movea.w sb.gtlin,a2     ; GeT Long INteger
                jsr     (a2)
                bne.s   initgen_rts
                movea.l sbq_data(a6),a2
                move.l  d7,gb_chid(a2)  ; Set channel
                move.l  (a6,a1.l),gb_pos(a2)

initgen_rts:                            
                rts

; =============== S U B R O U T I N E =======================================

; GENOP opcode
;
; Generate OPCODE byte with appropriate padding

genop:                                  
                movea.w sb.gtlin,a2     ; GeT Long INteger
                jsr     (a2)
                bne     genop_rts
                move.b  3(a6,a1.l),d5   ; Get LSB = opcode
                ext.w   d5
                move.b  genop_table(pc,d5.w),d0
                cmpi.b  #1,d0
                blt.s   genop_write_op
                bgt.s   genop_align_odd
                movea.l sbq_data(a6),a2
                bsr     gen_align
                bra.s   genop_write_op
; ---------------------------------------------------------------------------

genop_align_odd:                        
                movea.l sbq_data(a6),a2
                btst    #0,gb_align(a2)
                bne.s   genop_write_op
                bsr     gen_byte_0      ; Write 0

genop_write_op:                         
                rol.w   #1,d5           ; Opcodes are multiple of 2 as they index word table
                move.l  d5,d1
                bra     gen_byte_d1
; ---------------------------------------------------------------------------
; 0 = don't align stream
; 1 = insert alignment byte if position is odd
; 2 = insert alignment byte if position is even
genop_table:    dc.b 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0
                dc.b 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0
                dc.b 0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0
                dc.b 0,1,1,1,1,1,1,1,1,1,1,1,1,2,2,2,2
                dc.b 2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2,2
                dc.b 2,0,0,0,0,2,0,0,0,0,0,0,0,0,0,2,2
                dc.b 2,2,2,2,2,2,2,2,0,0,2,2,2,2,2,2,2
                dc.b 2,2,2,2,2,2,2,2,2,2,0

; =============== S U B R O U T I N E =======================================

; GENSTRING s
;
; Generate QDOS style string

genstring:                              
                movea.w sb.gtstr,a2     ; GeT STRing
                jsr     (a2)
                bne.s   genop_rts
                bsr     gen_align
                moveq   #0,d4
                move.w  (a6,a1.l),d4    ; String size
                addq.w  #2,d4
                moveq   #0,d6
                bra.s   gen_smul_loop   ; d4 = item size, d6 = item count - 1

; =============== S U B R O U T I N E =======================================

; GENTEXT s
;
; Generate text without size prefix

gentext:                                
                movea.w sb.gtstr,a2     ; GeT STRing
                jsr     (a2)
                bne.s   genop_rts
                moveq   #0,d4
                move.w  (a6,a1.l),d4
                addq.w  #2,a1
                moveq   #0,d6
                bra.s   gen_smul_loop   ; d4 = item size, d6 = item count - 1

; =============== S U B R O U T I N E =======================================

; GENB a,b,c,...
;
; Generate bytes

genb:                                   
                movea.w sb.gtlin,a2     ; GeT Long INteger
                jsr     (a2)
                bne.s   genop_rts
                moveq   #1,d4           ; Item size

gen_generic:                            
                move.w  d3,d6           ; Item count
                subq.w  #1,d6
                addq.w  #4,a1
                suba.w  d4,a1           ; Go to start of item

gen_smul_loop:                          
                movea.l sbq_data(a6),a2   ; d4 = item size, d6 = item count - 1
                movea.l gb_chid(a2),a0
                add.l   d4,gb_pos(a2)
                move.l  d4,d2
                bsr     gen_smul        ; Write multiple bytes. Ptr in a1, count in d2
                addq.l  #4,a1
                dbf     d6,gen_smul_loop ; d4 = item size, d6 = item count - 1

genop_rts:                              
                rts

; =============== S U B R O U T I N E =======================================

; GENW a,b,c,...
;
; Generate words (aligned)

genw:                                   
                bsr.s   gen_align

; =============== S U B R O U T I N E =======================================

; GENINT a,b,c,...
;
; Generate ints (not aligned)

genint:                                 
                movea.w sb.gtlin,a2     ; GeT Long INteger
                jsr     (a2)
                bne.s   genop_rts
                moveq   #2,d4           ; Item size
                bra.s   gen_generic

; =============== S U B R O U T I N E =======================================

; GENL a,b,c,...
;
; Generate longs (aligned)

genl:                                   
                bsr.s   gen_align
                movea.w sb.gtlin,a2     ; GeT Long INteger
                jsr     (a2)
                bne.s   genop_rts
                moveq   #4,d4           ; Item size
                bra.s   gen_generic

; =============== S U B R O U T I N E =======================================

; GENNTD x
;
; Generate word x * 8

genntd:                                 
                bsr.s   gen_align
                movea.w sb.gtlin,a2     ; GeT Long INteger
                jsr     (a2)
                bne.s   genop_rts
                rol     2(a6,a1.l)
                rol     2(a6,a1.l)
                rol     2(a6,a1.l)
                addq.w  #2,a1
                moveq   #2,d4           ; Item size
                moveq   #0,d6
                bra.s   gen_smul_loop   ; d4 = item size, d6 = item count - 1

; =============== S U B R O U T I N E =======================================


gen_align:                              
                movea.l sbq_data(a6),a2
                btst    #0,gb_align(a2)
                beq.s   gen_byte_rts

; =============== S U B R O U T I N E =======================================

; Write 0

gen_byte_0:                             
                moveq   #0,d1
; Write byte in d1

gen_byte_d1:                            
                movea.l sbq_data(a6),a2
                movea.l gb_chid(a2),a0
                moveq   #-1,d3
                moveq   #iob.sbyt,d0    ; Send BYTe to output
                trap    #3
                addq.l  #1,gb_pos(a2)

gen_byte_rts:                           
                rts

; =============== S U B R O U T I N E =======================================

; GENFLOAT f
;
; Generate 6-byte float

genfloat:                               
                moveq   #6,d4

genxfloat:                              
                movea.l nt_value(a6,a3.l),a1
                adda.l  sb_datab(a6),a1 ; long    data area base (first block)
                moveq   #0,d6           ; Only one time 6 bytes
                bra     gen_smul_loop   ; d4 = item size, d6 = item count - 1

; =============== S U B R O U T I N E =======================================

; GENSFLOAT
;
; Generate 4-byte float

gensfloat:                              
                moveq   #4,d4
                bra.s   genxfloat

; =============== S U B R O U T I N E =======================================

; SMEM addr,size
;
; Write memory contents

smem:                                   
                movea.w sb.gtlin,a2     ; GeT Long INteger
                jsr     (a2)
                bne.s   smem_rts
                cmpi.b  #2,d3
                bne.s   gen_err_ipar
                move.l  4(a6,a1.l),d2
                movea.l (a6,a1.l),a1
                movea.l sbq_data(a6),a2
                movea.l gb_chid(a2),a0
                add.l   d2,gb_pos(a2)
                moveq   #-1,d3
                moveq   #iob.smul,d0    ; Send MULtiple bytes
                trap    #3
                tst.l   d0

smem_rts:                               
                rts

; =============== S U B R O U T I N E =======================================

; Return channel ID in d7

gen_get_chan_param:                     
                cmpa.l  a3,a5
                ble.s   gen_err_ipar
                btst    #nt..hash,1(a6,a3.l) ; preceded by hash
                beq.s   gen_err_ipar
                movea.l sb_arthp(a6),a1 ; long    arithmetic stack pointer
                move.l  a5,-(sp)
                movea.l a3,a5
                addq.l  #8,a5
                movea.w sb.gtlin,a2     ; GeT Long INteger
                jsr     (a2)
                movea.l (sp)+,a5
                bne.s   gen_err_ipar
                addq.l  #8,a3
                bsr.s   gen_get_chan
                move.l  a0,d7
                moveq   #0,d0
                tst.l   d0
                rts
; ---------------------------------------------------------------------------

gen_err_ipar:                           
                moveq   #err.ipar,d0    ; Invalid PARameter (c.f. err.orng)
                tst.l   d0
                rts

; =============== S U B R O U T I N E =======================================


gen_get_chan:                           
                move.l  (a6,a1.l),d0
                moveq   #ch.len,d1
                mulu.w  d1,d0
                movea.l sb_chanb(a6),a2 ; long    channel table base
                adda.l  d0,a2
                movea.l (a6,a2.l),a0
                rts

; =============== S U B R O U T I N E =======================================

; Write multiple bytes. Ptr in a1, count in d2

gen_smul:                               
                moveq   #-1,d3
                moveq   #iob.smul,d0    ; Send MULtiple bytes
                trap    #4
                trap    #3
                suba.w  d1,a1
                tst.l   d0
                rts

; =============== S U B R O U T I N E =======================================

; addr = RP_ADDR
;
; Return current position in stream

rp_addr:                                
                movea.l sb_arthp(a6),a1 ; long    arithmetic stack pointer
                movea.l sbq_data(a6),a2
                move.l  gb_pos(a2),d0
                move.w  #$81F,d1
                tst.l   d0
                beq.s   rp_addr_0

rp_addr_normalize:                      
                asl.l   #1,d0
                bvs.s   rp_addr_backtrack
                subq.w  #1,d1
                bra.s   rp_addr_normalize
; ---------------------------------------------------------------------------

rp_addr_0:                              
                clr.w   d1
                bra.s   rp_addr_put
; ---------------------------------------------------------------------------

rp_addr_backtrack:                      
                roxr.l  #1,d0

rp_addr_put:                            
                move.l  d0,-4(a6,a1.l)
                move.w  d1,-6(a6,a1.l)
                subq.l  #6,a1
                moveq   #nt.fp,d4
                move.l  a1,sb_arthp(a6) ; long    arithmetic stack pointer
                moveq   #0,d0
                rts

; =============== S U B R O U T I N E =======================================

; is_odd = ODD(value)
;
; Return if the given value is odd (1) or not (0)

odd:                                    
                movea.w sb.gtlin,a2     ; GeT Long INteger
                jsr     (a2)
                bne     genop_rts
                moveq   #0,d1
                btst    #0,3(a6,a1.l)
                beq.s   loc_1A0D8
                moveq   #1,d1

loc_1A0D8:                              
                addq.l  #2,a1

gen_ret_int:                            
                move.w  d1,(a6,a1.l)
                move.l  a1,sb_arthp(a6) ; long    arithmetic stack pointer
                moveq   #nt.in,d4
                moveq   #0,d0
                rts

; =============== S U B R O U T I N E =======================================

; is_short = SHORTF(f)
;
; Check if float can be 32-bit

shortf:                                 
                movea.l sb_arthp(a6),a1 ; long    arithmetic stack pointer
                subq.l  #2,a1
                movea.l nt_value(a6,a3.l),a2
                adda.l  sb_datab(a6),a2 ; long    data area base (first block)
                moveq   #1,d1
                move.w  4(a6,a2.l),d0
                beq.s   gen_ret_int
                moveq   #0,d1
                bra.s   gen_ret_int

		end

As I was thinking about doing them. Because I thought you were just doing the stuff that came on the QLib disk.

Well, I wrote to you that I intend to do all the assembler stuff. I use a state-of-the-art PC based commercial disassembler which makes this stuff possible at all, I'd probably have given up long ago with the QL based ones.

Marcel

[mk79]

The dc.w 4 just after the line number table. I don't know exactly what it does, but in the QLIB source, if it's 0,1, or greater than 4, you get a BEEP, and maybe something else, otherwise it reads some value.

I'm not 100% sure yet there, either. I can tell that the OVERLAY command explicitly tests byte $2d for 4, otherwise the file is not recognised as a valid overlay. I've therefore classified it as a file-type, but in the mode-code it tests the whole word at $2c, so this is still a bit weird and unknown.

Marcel

[Martin_Head]

Code: Select all

base:           bra.w   qlib_start
; ---------------------------------------------------------------------------
                dc.w 0
                dc.w $4AFB
                dc.w 9
                dc.b 'Qlib_3.35'
                dc.b '             '
                dc.l token_base-base
                dc.l name_table-base
                dc.l 0                  ; Line table
                dc.w 4
                dc.l data_tokens-base   ; First DATA token
                dc.w 9                  ; Channel count
                

; =============== S U B R O U T I N E =======================================
write_rte_missing:
                lea     txt_runtime_missing(pc),a1 ; "\nRuntimes missing !\n"
                suba.l  a0,a0
                movea.w (ut.wtext).w,a2 ; Write TEXT
                jsr     (a2)
                rts
; End of function write_rte_missing

I have a couple of queries about the above.

'dc.l data_tokens-base ; First DATA token'
Looking at my disassembly of QLIB, should this be 'dc.l data_tokens-base+46', or maybe 'dc.l data_tokens-*'. Also is this just a pointer to the start of the DATA statements, or is it the DATA pointer that is updated every time a READ or RESTORE is done?

'movea.w (ut.wtext).w,a2 ; Write TEXT'
Is it just me, or has the disassembler got it's syntax wrong? And it should be move.w ut.wtext,a2

[mk79]

'dc.l data_tokens-base ; First DATA token'
Looking at my disassembly of QLIB, should this be 'dc.l data_tokens-base+46', or maybe 'dc.l data_tokens-*'.

Yes, it's data_tokens-*, sorry. I transfered this wrongly from my runtime analysis to the qlib_obj disassembly.

Also is this just a pointer to the start of the DATA statements, or is it the DATA pointer that is updated every time a READ or RESTORE is done?

Just the initial pointer, nothing in the OBJ area is ever modified. The runtime pointer is $C(a6).

'movea.w (ut.wtext).w,a2 ; Write TEXT'
Is it just me, or has the disassembler got it's syntax wrong? And it should be move.w ut.wtext,a2

It does use a different syntax, yes. I have created an automated post process to fix this, but for the post I just used copy/paste.

Marcel

[Martin_Head]

Yesterday I tried to get QLiberator to compile itself in Qemulator under QDOS. I thought I had already tried this, but I could not have.
It took 12 minutes to LOAD the Qlib basic program. And just under 2 hours to compile itself.

[RalfR]

Great! So perhaps you can change the place, where 512x256 is the maximum for dragging QLib's own window.

[duncan]

That is awsome work. Well done. I have noticed that the length of the compiled file is slightly longer than the QLIB_obj that I have which is 46158 bytes.

[Martin_Head]

Great! So perhaps you can change the place, where 512x256 is the maximum for dragging QLib's own window.

At the moment, my main concern is trying to make sure that the decompiled QLib is accurate to the original.
I would like to get a recompilation that is identical to the original. (If that's possible) At the moment the program size and dataspace values don't quite match. So I want to try doing byte by byte comparisons between the original and recompiled versions looking for differences. That's why I did the Qemulator compile yesterday, to see if using QDOS produced different results to SMSQ/E.

[RalfR]

At the moment, my main concern is trying to make sure that the decompiled QLib is accurate to the original.

I agree. Have you tried the recompiled version to compile a program? And to compare that with a one compiled with the original 3.36?

What, if you compile QLib with the recompiled Version? Same size and dataspace as if self-compiled?

[mk79]

Great! So perhaps you can change the place, where 512x256 is the maximum for dragging QLib's own window.

At the moment, my main concern is trying to make sure that the decompiled QLib is accurate to the original.
I would like to get a recompilation that is identical to the original. (If that's possible)

If you compile your QLIB source one time with the original qlib_obj and one time with your recompiled qlib_obj, do the results differ?

The 3.36 qlib_obj file claims that it was compiled with 3.36 itself (while 3.35 says it was compiled with 3.34), but we cannot be sure that it was the released version, so there might still be differences.

Marcel