27 June 2024, 21:25 | #1 |
German Translator
Join Date: Aug 2018
Location: Drübeck / Germany
Age: 50
Posts: 214
|
WinUAE Debugger - points for clarification or improvement
This is a summary of various points that I noticed during testing. Maybe some of them can be implemented in a future version?
1. Wc <addr> <endaddr> <destaddr>, copy memory. >Wc 30000 30030 30050 Copied from 00030000 - 0003002f to 00030050 - 0003007f Why not Copied from 00030000 - 00030030 to 00030050 - 00030080 ? include the last byte 2. helptext: e[x] wrong, better: e[a/x] 3. helptext: s "<string>"/<values> [<addr>] [<length>] is wrong, better s "<string>"/<values> [<addr>] [<end>] ; not length -> end >s 0839000600bfe001 20000 30000 Searching from 00020000 to 00030000.. 4. I <custom event> Is this command obsolet and free, othwise an example is needed. What is meaned with custom event? 5. S <file> <addr> <n> Save a block of Amiga memory. >S m31212 $31212 !400 ; Wrote 00031212 - 000313A2 (400 bytes) to 'm31212'. but better is ; Wrote 00031212 - 000313A1 (400 bytes) >S m1 $1 1 ; Wrote 00000001 - 00000002 (1 bytes) to 'm1'. --> better is: Wrote 00000001 - 00000001 (1 bytes) to 'm1'. >S m1 $1 0 ; Wrote 00000001 - 00000001 (0 bytes) to 'm1'. --> some error message like 'wrong argument' 6. S <file> <addr> <n> Save a block of Amiga memory. S C:\Amiga test $10 2 It is not possible to set a custom path for save. All files goes (in my case) to C:\Users\Public\Documents\Amiga Files\WinUAE 7. helptext wrong w <num> <address> <length> <R/W/I> <F/C/L/N> [V<value>[.x]] (read/write/opcode) (freeze/mustchange/logonly/nobreak). better is: w <num> [<address> [<length> [<R/W/I> <F/C/L/N> [V<value>[.x]] [<channel>]]]]] 8. Unknown command and function unknown >seg No executable loaded 9. Unknown command and function unknown >wl Memwatch breakpoints enabled Memwatch DMA validator enabled > 10. vm [<channel> <sub index> [<hex rgb>]] show DMA channel status / DMA channel enable/disable toggle/ change color of channel If sub index is zero: all sub index colors are changed. ok, if index 0 changed all, then it would be good to change the order from 7,0 -> 7,7 to 7,1 -> 7,8 >vm 7 1 00123456 7,0: 00123456 * Sprite ; the answer is 7,0 and not 7,1; better is: 7,0: --> 7,1: 11. >fl 0: PC == 0003173e [00000000 00000000] what is the meaning of [00000000 00000000] ? [00000000 00000000] Could be erased? I linked the other points here: >fo ; https://eab.abime.net/showpost.php?p...18&postcount=6 >il ; https://eab.abime.net/showpost.php?p...4&postcount=3; >smc ; https://eab.abime.net/showpost.php?p...7&postcount=21 >f ; https://eab.abime.net/showpost.php?p...4&postcount=22 >wd ; https://eab.abime.net/showpost.php?p...47&postcount=9 >vh ; https://eab.abime.net/showpost.php?p...6&postcount=28 >D ; https://eab.abime.net/showpost.php?p...65&postcount=8 >H ; https://eab.abime.net/showpost.php?p...30&postcount=3 >dj https://eab.abime.net/showthread.php?t=98511 ; relevant thread - no open points >did https://eab.abime.net/showthread.php?t=110586 ; relevant thread - no open points >wishes https://eab.abime.net/showthread.php?t=118334 Last edited by Rock'n Roll; Today at 09:00. |
28 June 2024, 15:07 | #2 |
German Translator
Join Date: Aug 2018
Location: Drübeck / Germany
Age: 50
Posts: 214
|
a slightly improved help with all the commands I know:
Code:
>h HELP for UAE Debugger ----------------------- g [<addr>] Start execution at the current address or <address>. c Dump state of the CIA, disk drives and custom registers. r Dump state of the CPU. r <reg> <value> Modify CPU registers (Dx,Ax,USP,ISP,VBR,...). rc[d] Show CPU instruction or data cache contents. m <addr> [<lines>] Memory dump starting at <address>. m r[b/w/l](addr/r(register)) [<lines>] Memory dump indirect a <addr> Assembler. d <addr> [<lines>] Disassembly starting at <address>. d r[b/w/l](address/r(register)) [<lines>] Disassembly starting indirect d[o/ppc] change disassembly to PPC or 'Original' 68k Assembly t [instructions] Step one or more instructions. tx Break when any exception. z Step through one instruction - useful for JSR, DBRA etc. f Step forward until PC in RAM ("boot block finder"). f <addr> [Nx] Add/remove breakpoint. fa <addr> [<start>] [<endaddr>] Find effective address <address>. fi Step forward until PC points to RTS, RTD or RTE. fi <opcode> [<w2>] [<w3>] Step forward until PC points to <opcode>. fp "<name>"/<addr> Step forward until process <name> or <addr> is active. fl List breakpoints. fd Remove all breakpoints. fs <lines to wait> | <vpos> <hpos> Wait n scanlines/position. fc <CCKs to wait> Wait n color clocks. fo <num> <reg> <oper> <val> [<mask> <val2>] [Nx] [Hx] Conditional register breakpoint. reg=Dx,Ax,PC,USP,ISP,VBR,SR. oper:!=,==,<,>,>=,<=,-,!- (-=val to val2 range). f <addr1> <addr2> Step forward until <addr1> <= PC <= <addr2>. e[a/x] Dump contents of all custom registers, ea = AGA colors. i Dump contents of interrupt and trap vectors. il [<mask>] Exception breakpoint. o <0-2|addr> [<lines>]View memory as Copper instructions. od Enable/disable Copper vpos/hpos tracing. ot Copper single step trace. ob <addr> Copper breakpoint. h Help for UAE Debugger H[H] <cnt> Show PC history (HH=full CPU info) <cnt> instructions. C [<value>] Search reset or search for values like energy or lifes in games. Cl List currently found trainer addresses. D[idxzs [<max diff>]] Deep trainer. i=new value must be larger, d=smaller, x = must be same, z = must be different, s = restart. W <addr> <values[.x] separated by space> Write into Amiga memory. W <addr> 'string' Write into Amiga memory. Wf <addr> <endaddr> <bytes or string like above>, fill memory. Wc <addr> <endaddr> <destaddr>, copy memory. w <num> <address> <length> [<R/W/I>] [<F/C/L/N>] [V<value>[.x]][<channel>] (read/write/opcode) (freeze/mustchange/logonly/nobreak). Add/remove memory watchpoints. wd [<0-1>] Enable illegal access logger. 1 = enable break. wl Memwatch DMA validator L <file> <addr> [<n>] Load a block of Amiga memory. S <file> <addr> <n> Save a block of Amiga memory. s "<string>"/<values> [<addr>] [<endaddr>] Search for string/bytes. T or Tt Show exec tasks and their PCs. Td,Tl,Tr,Tp,Ts,TS,Ti,TO,TM,Tf Show devs, libs, resources, ports, semaphores, residents, interrupts, doslist, memorylist, fsres. b Step to previous state capture position. M<a/b/s> <val> Enable or disable audio channels, bitplanes or sprites. sp <addr> [<addr2][<size>] Dump sprite information. di <mode> [<track>] Break on disk access. R=DMA read,W=write,RW=both,P=PIO. Also enables level 1 disk logging. did <log level> Enable disk logging. dj [<level bitmask>] Enable joystick/mouse input debugging. smc [<0-1>] Enable self-modifying code detector. 1 = enable break. dm Dump current address space map. v [<-1 to -6/o> | <vpos> [<hpos>] [<lines>] Show DMA data (accurate only in cycle-exact mode). v [-1 to -6] = enable visual DMA debugger. vh [ -1/?/c/d ] | [<channel> [<lines>]] "Heat map" vm [<channel> <sub index> [<hex rgb>]] show DMA channel status / DMA channel enable/ disable toggle/ change color of channel (v=V, vh=Vh, vm=Vm) commands like above I <custom event> Send custom event string reset[<h/k>] Reset (hardware, keyboard) ?<value> Hex ($ and 0x)/Bin (%)/Dec (!) converter and calculator. x Close debugger. xx Switch between console and GUI debugger. mg <address> Memory dump starting at <address> in GUI. dg <address> Disassembly starting at <address> in GUI. q Quit the emulator. You don't want to use this command. > Last edited by Rock'n Roll; Today at 09:00. Reason: update |
01 August 2024, 15:23 | #3 |
German Translator
Join Date: Aug 2018
Location: Drübeck / Germany
Age: 50
Posts: 214
|
further more points:
1. sp-command Is it possible to show also the sprite height? Sprite address 0002BD5C, width = 16, height = 13 ; EndY=109-StartY=96 OCS: StartX=416 StartY=96 EndY=109 ECS: StartX=1664 (416.0) StartY=96 EndY=109 Attach: 0. AGA SSCAN/SH10 bit: 1 >?!109-!96 $0000000D = %00000000`00000000`00000000`00001101 = 13 = 13 > 2. -- CYC ? CIA A: PRA C2 PRB 00 DRA 03 DRB 00 CRA 00 CRB 08 ICR 00 IM 0a TA ffff (ffff) TB 0863 (0863) TOD 00167f (0011aa) ALARM 000000 -- CYC=0000003427988E00 What is the normal user info about CYC=0000003427988E00? 3. >Wf 30050 30060 $F0 Wrote data to 00030050 - 00030060 >m 30050 2 00030050 F0F0 F0F0 F0F0 F0F0 F0F0 F0F0 F0F0 F0F0 ................ 00030060 0000 0000 0000 0000 0000 0000 0000 0000 ................ > it wrotes $F0 only from 00030050 to 0003005F, it doesn't include the last byte 30060 change answer: Wrote data (16 bytes) to 00030050 - 0003005f. 4. >s 3F 00 2000 Searching from 00000000 to 00002000.. Scanning.. 00000000 - 00010000 (Chip memory) 00000CF1 00000D78 00000D7E 00000D85 00000D89 many more unformated addresses.... Limit output to 8 entries per line, please formatting 5. fa <address> [<start>] [<end>] >fa c02878 ; normally Searching from 00000000 to 00C80000 Scanning.. 00000000 - 00200000 (Chip memory) Scanning.. 00c00000 - 00c80000 (Slow memory) [<start>] [<end>] doesn't work >fa c02878 20000 Searching from 00020000 to 00C80000 Scanning.. 00c00000 - 00c80000 (Slow memory) >fa c02878 $20000 $30000 Searching from 00020000 to 00030000 Scanning.. 00c00000 - 00c00000 (Slow memory) > 6. String returned: '' >w 0 1000 10 WR 0: 00001000 - 0000100F (16) RW CPU >w 0 1000 10 LR 0: 00001000 - 0000100F (16) R L CPU >w 0 1000 10 RW String returned: '' >w 0 1000 10 RL String returned: '' 7. works but it doesn't break the program w 0 22000 2 fw works but it comes no break or it works as w 0 22000 2 fwn, without f it triggers a break 8. >w ; if no memwatch breakpoint has been added, the answer to >w is nothing > >w No breakpoints. ; expected like >fl 9. The debugger always needs to wait until current CPU instruction has finished execution which means above breakpoints can't be 100% exact. >fc 1 Cycles: 1 Chip, 2 CPU. (V=105 H=166 -> V=105 H=167) ; change H=167 to 173 VPOS: 105 ($069) HPOS: 173 ($0ad) COP: $0001ede4 ... >fc 1 Cycles: 5 Chip, 10 CPU. (V=105 H=173 -> V=105 H=178) ; here is the correct elapsed cycles VPOS: 105 ($069) HPOS: 178 ($0b2) COP: $0001ede4 ... >fc 1 Cycles: 1 Chip, 2 CPU. (V=105 H=178 -> V=105 H=179) ; change H=179 to 188 VPOS: 105 ($069) HPOS: 188 ($0bc) COP: $0001ede4 better show always the needed cycles from the current instruction the same for fs >fs 115 20 Cycles: 68327 Chip, 136654 CPU. (V=127 H=20 -> V=115 H=20) ; better H=23 VPOS: 115 ($073) HPOS: 023 ($017) COP: $0001ede4 I prefer the correct state, like after a >t 10. if a program breakpoint interrupts a 'fc' and the breakpoint is deleted then the waiting CCKs are not correct anymore Code:
>f 23d60 Breakpoint added. >g Breakpoint 0 triggered. Cycles: 30 Chip, 60 CPU. (V=111 H=106 -> V=111 H=136) ; one loop is 30 CCK ... >fd All breakpoints removed. >fc 90 Cycles: 90 Chip, 180 CPU. (V=111 H=136 -> V=111 H=226) ; everything ok here ... >fc 90 Cycles: 90 Chip, 180 CPU. (V=111 H=226 -> V=112 H=89) ; everything ok here ... >f 23d60 Breakpoint added. >fc 90 Breakpoint 0 triggered. ; program breakpoint interrupts a 'fc' Cycles: 30 Chip, 60 CPU. (V=112 H=89 -> V=112 H=119) ; 30 CCK is ok ... >fd All breakpoints removed. >fc 90 Cycles: 60 Chip, 120 CPU. (V=112 H=119 -> V=112 H=179) ; 60 Chip and not 90, not ok ... >fc 90 Cycles: 30 Chip, 60 CPU. (V=112 H=179 -> V=112 H=209) ; 30 Chip and not 90, not ok ... 11. wish >d is >d pc , because I want see direct the next instruction Example: If I open the debugger with Shift+F12 or after a triggered breakpoint my interest is on >d pc Code:
Cycles: 4194303 Chip, 8388607 CPU. (V=122 H=131 -> V=210 H=0) VPOS: 210 ($0d2) HPOS: 004 ($004) COP: $0001ede4 ... 00022c10 0839 0006 00bf e001 btst.b #$0006,$00bfe001 Next PC: 00022c18 >d ; >d 'next pc' 00022c18 66f6 bne.b #$f6 == $00022c10 (T) ... >d pc ; I get this with a >d 00022c10 0839 0006 00bf e001 btst.b #$0006,$00bfe001 ... > Every next >d shows the next lines then. 12. >ex - reading, intepretation? >ex 096 DMACON 03A0 83A0 00022D10 CPU 18A COLOR05 0BB9 0000 00000000 CPU 098 CLXCON 0000 0000 00000000 CPU 18C COLOR06 0DE7 0000 00000000 CPU >d 22D10 1 00022d10 3b7c 83a0 0096 move.w #$83a0,(a5,$0096) == $00dff096 09A INTENA 0000 7FFF 00022B4E CPU 18E COLOR07 0E7D 0000 00000000 CPU 09C INTREQ 0028 7FFF 00022B4E CPU 190 COLOR08 07FF 0000 00000000 CPU >d 22B4E 1 00022b4e 2b7c 7fff 7fff 009a move.l #$7fff7fff,(a5,$009a) == $00dff09a This seems to be ok. But what about COP? It's in the next line? 0E0 BPL1PTH 0006 0006 00068764 COP 1C4 HBSTRT 0000 0000 00000000 CPU 0E2 BPL1PTL 8B48 8B48 00068768 COP 1C6 HBSTOP 0000 0000 00000000 CPU >o 68764 1 00068764: 00e2 8b48 ; BPL1PTL := 0x8b48 000 BLTDDAT 0000 0000 00000000 CPU 108 BPL1MOD FFF8 FFF8 00068754 COP 002 DMACONR 03A0 0000 00000000 CPU 10A BPL2MOD FFF8 FFF8 00068758 COP >o 68754 1 00068754: 010a fff8 ; BPL2MOD := 0xfff8 > 13. rc / rcd Whats the different between rc and rcd? How is the interpretation? >rc 00022AF0U:203900DF*00022AF4U:F0040280*00022AF8U:000FFF00*00022AFCU:0C800001* ... 00022af0 2039 00df f004 move.l $00dff004,d0 Last edited by Rock'n Roll; 02 August 2024 at 12:47. |
09 August 2024, 14:14 | #4 |
German Translator
Join Date: Aug 2018
Location: Drübeck / Germany
Age: 50
Posts: 214
|
ok, further more points
(Basically, however, the debugger works perfectly. It is nuances that are the issue here.) 1. >C4 Scanning.. 00000000 - 00200000 (Chip memory) Scanning.. 00c00000 - 00c80000 (Slow memory) 0000345B 00003473 000034AA 0008345B 00083473 000834AA 0010345B 00103473 001034AA 0018345B 00183473 001834AA Found 12 possible addresses with 0x4 (4) (1 bytes) Now continue with 'g' and use 'C' with a different value > Format output to 8 addresses per row, please 2. after a Cl the input prompt is on the far right >Cl 000034AA=0003 000834AA=0003 001034AA=0003 001834AA=0003 > 3. wish - Bitwise Not operation ~ >?$3 $00000003 = %00000000`00000000`00000000`00000011 = 3 = 3 >?~$3 ; ... 1111 1100 (for Blitter LF calculation: AB+!AC) 4. N counts one too much >f 2C1FC N3 Breakpoint added. >g Breakpoint 0 hit: PC=0002c1fc, count=3. Breakpoint 0 hit: PC=0002c1fc, count=2. Breakpoint 0 hit: PC=0002c1fc, count=1. Breakpoint 0 triggered. 5. visual DMA Debugger transparency mode is gone? v -5 : enable + visual mode + larger overlay with "transparency" was till 4.9 6. DMA Debugger - Horizontal diw ('(' and ')'), markers are on wrong positions - missed the L mark for long line (L) in second refresh slot 7. HDIW=0/1 in >c output are wrong 8. if an exception is debugged while the DMA debugger is active, the output does not stop after the number of lines e.g. H2 returns 500 lines >v-2 ; DMA Debugger on >il $08000000 ; Exception Brkpt. lvl-3 >g ; breaks the exception >H2 ; not 2, delivers 500 (maximum?) 9. wish - Hs command even if cycle exact or DMA debugger is active, a simplified output only CPU, without DMA channels or CPU cycles eg. H[s] <cnt> ; short >Hs1 0 209/199 0001fff6 0c39 00c8 00df f006 cmp.b #$c8,$00dff006 ; - 00 D1 BPL2 112 0000 0001CE08 ; without these lines 10. unification There are two differents output concepts for similar the same. The PC-History and DMA Debugger. Why not bring both in an unifcation output? It would be more readable. Code:
>H10 ; this are the last 10 executed instructions ... 0 209/199 0001fff6 0c39 00c8 00df f006 cmp.b #$c8,$00dff006 - 00 D1 BPL2 112 0000 0001CE08 - 00 D2 CPU-RWI 00DF 0001FFFA - 00 D3 BPL1 110 3C3C 00017E08 - 00 D4 CPU-RWI F006 0001FFFC - 00 D5 BPL2 112 0000 0001CE0A - 00 D6 CPU-RWI 66F6 0001FFFE - 00 D7 BPL1 110 3C00 00017E0A - 00 D8 CPU-RBD 00C7 00DFF006 - 00 D9 BPL2 112 0000 0001CE0C - 00 DA CPU-RWI 41F9 00020000 0 219/199 0001fffe 66f6 bne.b #$f6 == $0001fff6 (F) - 00 DB BPL1 110 0000 00017E0C - 00 DD BPL2 112 0000 0001CE0E - 00 DE CPU-RWI 0C39 0001FFF6 - 00 DF BPL1 110 0000 00017E0E - 00 E0 CPU-RWI 00C8 0001FFF8 0 225/199 0001fff6 0c39 00c8 00df f006 cmp.b #$c8,$00dff006 ;------------------------------------------------------------------------------ >v !199 !209 5 ; this are collected cycles from the previous frame Line: C7 199 HPOS D1 209: [D1 19A -] [D2 19C -] [D3 19E -] [D4 1A0 -] [D5 1A2 -] [D6 1A4 -] [D7 1A6 -] B [D8 1A8 -] BPL2 112 CPU-RWI BPL1 110 CPU-RBD BPL2 112 CPU-RWI BPL1 110 c 0000 66F6 3C3C 00C7 0000 41F9 3C00 c 0001CE08 0001FFFE 00017E08 00DFF006 0001CE0A 00020000 00017E0A 004 0E7 0FF 0FF 004 0BF 103 1F8 005 0E7 100 000 005 0BF [D9 1AA -] [DA 1AC -] [DB 1AE -] [DC 1B0 -] [DD 1B2 -] C [DE 1B4 -] [DF 1B6 -] [E0 1B8 -] BPL2 112 CPU-RWI BPL1 110 CPU-RWI BPL2 112 M CPU-RWI BPL1 110 CPU-RWI 0000 0C39 0000 00C8 0000 P 00DF 0000 F006 0001CE0C 0001FFF6 00017E0C 0001FFF8 0001CE0E 0001FFFA 00017E0E 0001FFFC 006 0E7 0FB 0FF 006 0BF 0FC 0FF 007 0E7 0FD 0FF 007 0BF 0FE 0FF ;------------------------------------------------------------------------------ 0 209/199 0001fff6 0c39 00c8 00df f006 cmp.b #$c8,$00dff006 ; one output-concept for both? - 00 D1 19A 0001CE08 BPL2 112 0000 xCIAx 004 0E7 - 00 D2 19C 0001FFFA CPU-RWI 00DF xCIAx 0FF 0FF |
Today, 09:02 | #5 |
German Translator
Join Date: Aug 2018
Location: Drübeck / Germany
Age: 50
Posts: 214
|
This is the final summary of open points related to the WinUAE debugger, as I have completed my general revision of my debugger lesson.
The lesson will be released later this year. First I will finish some other topics. 1. >od Copper debugger disabled. After disabling the output never changend from >o1 2 00000420: 0180 005a [000 008] ; COLOR00 := 0x005a 00000424: 00e2 0000 [000 00c] ; BPL1PTL := 0x0000 back to: (without cycle information) 00000420: 0180 005a ; COLOR00 := 0x005a 00000424: 00e2 0000 ; BPL1PTL := 0x0000 For this you have to close WinUAE and restart WinUAE again. (It should disabling collecting the data and don't show any state data.) 2. >ob 21590 Copper breakpoint @0x00021590 How can I list copper breakpoints? How can I delete copper breakpoints? 3. if a copper breakpoint triggers no info message is shown in first line about this >g ; no info message about copper breakpoint is reached Cycles: 8862 Chip, 17724 CPU. (V=105 H=4 -> V=144 H=13) VPOS: 144 ($090) HPOS: 023 ($017) COP: $0002159c ... Next PC: 00021558 4. copper breakpoint don't breaks again in the next frames >g ; breaks not again in next frame 5. >ot Sometimes it happens that the copper pointer is not aligned correctly VPOS: 300 ($12c) HPOS: 011 ($00b) COP: $0001edee ... 00fc0f90 4e72 2000 stop #$2000 Next PC: 00fc0f94 >o0 1 *0001edee: fffe 0000 ; NULL := 0x0000 >o 1edee-6 0001ede8: 0100 0200 ; BPLCON0 := 0x0200 *0001edec: ffff fffe ; Wait for vpos 6. xx change from Console Debugger to GUI Debugger works but backwards to Console Debugger the > prompt is missed empty debugger. it needs ShiFt+F12 again. 7. memory watchpoint with dma channel works only properly if cycle-exact is enabled, otherwise WinUAE Debugger hangs after >g w 1 $10508 10 BLT ; This command with 'BLT' requires 'cycle exact' otherwise it will not work! |
Currently Active Users Viewing This Thread: 1 (0 members and 1 guests) | |
Thread Tools | |
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
When closing the WinUAE debugger, WinUAE closes. | Photon | support.WinUAE | 2 | 21 September 2021 20:07 |
Debugger updates (was: WinUAE Debugger HH PC history) | selco | support.WinUAE | 8 | 14 March 2018 22:27 |
Disk-swapping improvement (WinUAE) | Maren | request.UAE Wishlist | 49 | 13 March 2010 22:34 |
Need help using the WinUAE Debugger | Steve | support.WinUAE | 9 | 10 January 2009 20:00 |
WinUAE 1.2 problem/improvement/request/proposal :) | jotd | support.WinUAE | 7 | 11 March 2006 18:54 |
|
|