Steem Boiler Requests

[Cyprian]

currently we we can set the "Monitor" for a particular address "B" or "W".
Would be cool to have possibility to set the Monitor to defined by user address range. E.g. I would like to break emulation when a program tries to read or write to the screen area (or any other memory area like chunky screen buffer) Therefore I define area of monitored memory: the beginning and the end memory (e.g. start: $03f8000, end: $03fffff)

[AtariZoll]

Huh ! This is something I want too for long time

[Cyrano Jones]

BP when [$Address]=[$Value] would also be nice (ie, halt when $ffff8254=$700 or similar)

Also, an option to select stack frame type ('00 or '30).

[Steven Seagal]

It's "request", right?
I will look at those. I already thought myself that a breakpoint condition like "write X to Y" would be cool.
But I don't understand "an option to select stack frame type ('00 or '30)".
Steem emulates the 68000, if that's what that means?

[AtariZoll]

As said I wanted it too. But truth is that I could do it myself, likely. Surely Steven Seagal will do it better and faster
Now, considering selecting of stack frame type, what CJ suggests - I guess that he does not want full 68030 emulation, just stack frame selectable.
This is what I did already, and is very useful for testing ST games - to see and fix stackframe related problems which would happen on Falcon, TT .
http://atari.8bitchip.info/stemt3.html
There is special patched TOS version too, and you will need it. TOS 2.06 has large stackframe support, but you need to patch it's CPU detection part.
I made another special Steem Debugger variant, to catch pipeline related errors in ST SW running on TT, Falcon - this helps in finding code parts which write in RAM location(s) only few bytes after writing instruction self - what causes errors, since it is already in CPU's pipeline, so it will execute old code/data. Such code is rarer than code with stackframe problems, and usage of my hacked version is not really elegant, but I guess that there is no more than 3 people on planet who needs such tool

[Steven Seagal]

@AtariZoll

http://atari.8bitchip.info/stemt3.html

Could you upload (or email me) the modified source, so that I integrate those fixes in Steem SSE?
68030 stack frame as Boiler option.
One bug you mention I recently saw:

first is bad date/timestamp shown (and used) by files in GEMDOS hard disk emulation. Instead time of file's last modification creation time is shown. It is incorrect, as TOS (and DOS) shows time of last modification on file. So, replaced CreationTime with LastWriteTime in stemdos.cpp .

[AtariZoll]

In attachment are all modded src files. Only cpu.h and cpu.cpp hold changes for stackframe hack. Others are with timestamp and floppy format changes. This should not go in normal Steem, only in Debugger (Boiler) . Not some partial 68030 emulation, just simple hack in purpose to track down possible errors when running ST(E) SW on TT, Falcon or on Mega STE with 68010 .

code.zip

I patched TOS 2.06 for this - just fooled CPU detection part. It will say that 68030 is in machine, although would work same if there would be 68010 or 20 . Removed cacr r/w at $E1A7AC too . Fixed TOS ROM CRC.

T2.06_30f.zip

[Steven Seagal]

Thx, I copied those fixes.
For the stack frame, I push 0 instead of SR, it seems clearer so.
All requests and more are in one way or another, with limitations, as will be apparent in tomorrow's beta.

[Cyprian]

Currently, Steem Boiler starts with emulation paused. Would be possible to add a checkbox (or a switch in the "Control mask browser") to start Steem Boiler with emulation started? And similar situation with Fast Forward, would be cool to have a check box.

Those changes would speed up a bit code debugging.

[AtariZoll]

It seems that Steven Seagal is absent from here last weeks. Perhaps filming new action movie ?

[Marakatti]

[Steven Seagal]

Currently, Steem Boiler starts with emulation paused. Would be possible to add a checkbox (or a switch in the "Control mask browser") to start Steem Boiler with emulation started? And similar situation with Fast Forward, would be cool to have a check box.

Those changes would speed up a bit code debugging.

I'll look if that's possible.
Meanwhile next beta should have PRG and TOS support.

It seems that Steven Seagal is absent from here last weeks. Perhaps filming new action movie ?

Secret mission!

[Cyprian]

I'll look if that's possible.

cool

Meanwhile next beta should have PRG and TOS support.

via drag'n'drop, command line or both?

It seems that Steven Seagal is absent from here last weeks. Perhaps filming new action movie ?

Secret mission!

hehe, will waiting for that movie

[Steven Seagal]

via drag'n'drop, command line or both?

Normally it should work all ways: drag on Steem icon, command-line, drag & drop in Steem window, and select in disk manager (zip/rar OK).
Caveat: only for 1 file.
It works well with all those single PRG demos lotek has been releasing.

[Cyprian]

One request regarding Logging.
I'm trying to recognize memory usage of a part of code. When I use logging "Log --> CPU" then I see following output:

Code: Select all

00F1EE: move.l -2(a1),d6
00F1F2: ror.l d7,d6
00F1F4: lea +$2000(a2),a5
00F1F8: lea +$60c(a5),a3
00F1FC: move.w d6,d1
00F1FE: subq.w #1,d1
00F200: movem.l D0-1/A0/A2,-(a7)
00F204: moveq #16,d3
00F206: bsr .l +$11a {$00F322}
00F322: sub.b d3,d7
00F324: bcc.s +12 {$00F332}
00F326: add.b #$10,d7
00F32A: move.l (a1),d6

Unfortunately, in case of my output (it has about 10000 lines) I'm not able to guess which memory address is read or written by the CPU instruction.

Would be possible to add calculated memory address in curly brackets as it is currently done in case of branches?

Code: Select all

00F1EE: move.l -2(a1),d6 {$00F920}
00F1F2: ror.l d7,d6
00F1F4: lea +$2000(a2),a5 {$00F900}

[Cyprian]

Another thing which I'd like to improve a bit is a pop-up dialog windows, like "New program executed", "Hit breakpoint at address" or "Wrote to address $XXXX, new value is XXXX". Generally that pop-up window is ok, until the moment when you have to press it fortieth once... (and also in case of my screen resolution 1600x900, every time you have to move mouse a lot because dialog box is in the center of the screen and the bolier in the top right corner)
Would be possible to move such a message to e.g. the status bar of the boiler window?

[Cyprian]

Hi Steven, would be possible to add some features to the History List?

- "Export" (as a text) - that would allow to analyze the list in a favorite text editor;

- "Clear" - that would allow to remove not needed code from the list;

- and the last one is to add Instruction Cycles, which would allow to analyze/improve a time-sensitive code.

[Steven Seagal]

One request regarding Logging.
I'm trying to recognize memory usage of a part of code. When I use logging "Log --> CPU" then I see following output:

Code: Select all

00F1EE: move.l -2(a1),d6
00F1F2: ror.l d7,d6
00F1F4: lea +$2000(a2),a5
00F1F8: lea +$60c(a5),a3
00F1FC: move.w d6,d1
00F1FE: subq.w #1,d1
00F200: movem.l D0-1/A0/A2,-(a7)
00F204: moveq #16,d3
00F206: bsr .l +$11a {$00F322}
00F322: sub.b d3,d7
00F324: bcc.s +12 {$00F332}
00F326: add.b #$10,d7
00F32A: move.l (a1),d6

Unfortunately, in case of my output (it has about 10000 lines) I'm not able to guess which memory address is read or written by the CPU instruction.

Would be possible to add calculated memory address in curly brackets as it is currently done in case of branches?

Code: Select all

00F1EE: move.l -2(a1),d6 {$00F920}
00F1F2: ror.l d7,d6
00F1F4: lea +$2000(a2),a5 {$00F900}

In current version 3.7, CPU log gives something like this:

Code: Select all

D0=3C0FFFF D1=0 D2=0 D3=3C00000 D4=0 D5=FFE0 D6=B D7=FF80
A0=174F0 A1=37F7A A2=444E2 A3=26108 A4=19074 A5=CC04 A6=1AFDC A7=1AFAC SR=2404
FC06EE 48E7 movem.l D0-7/A0-6,-(a7)

D0=3C0FFFF D1=0 D2=0 D3=3C00000 D4=0 D5=FFE0 D6=B D7=FF80
A0=174F0 A1=37F7A A2=444E2 A3=26108 A4=19074 A5=CC04 A6=1AFDC A7=1AF70 SR=2404
FC06F2 52B9 addq.l #1,$462

D0=3C0FFFF D1=0 D2=0 D3=3C00000 D4=0 D5=FFE0 D6=B D7=FF80
A0=174F0 A1=37F7A A2=444E2 A3=26108 A4=19074 A5=CC04 A6=1AFDC A7=1AF70 SR=2400
FC06F8 9BCD suba.l a5,a5

D0=3C0FFFF D1=0 D2=0 D3=3C00000 D4=0 D5=FFE0 D6=B D7=FF80
A0=174F0 A1=37F7A A2=444E2 A3=26108 A4=19074 A5=0 A6=1AFDC A7=1AF70 SR=2400
FC06FA 1039 move.b $ffff8260,d0

D0=3C0FFFC D1=0 D2=0 D3=3C00000 D4=0 D5=FFE0 D6=B D7=FF80
A0=174F0 A1=37F7A A2=444E2 A3=26108 A4=19074 A5=0 A6=1AFDC A7=1AF70 SR=2408
FC0700 C03C and.b #$3,d0

D0=3C0FF00 D1=0 D2=0 D3=3C00000 D4=0 D5=FFE0 D6=B D7=FF80
A0=174F0 A1=37F7A A2=444E2 A3=26108 A4=19074 A5=0 A6=1AFDC A7=1AF70 SR=2404
FC0704 B03C cmp.b #$2,d0



EDIT: is it possible to fix the title of this thread?

[Cyprian]

In current version 3.7, CPU log gives something like this:

Code: Select all

D0=3C0FFFF D1=0 D2=0 D3=3C00000 D4=0 D5=FFE0 D6=B D7=FF80
A0=174F0 A1=37F7A A2=444E2 A3=26108 A4=19074 A5=CC04 A6=1AFDC A7=1AFAC SR=2404
FC06EE 48E7 movem.l D0-7/A0-6,-(a7)

D0=3C0FFFF D1=0 D2=0 D3=3C00000 D4=0 D5=FFE0 D6=B D7=FF80
A0=174F0 A1=37F7A A2=444E2 A3=26108 A4=19074 A5=CC04 A6=1AFDC A7=1AF70 SR=2404
FC06F2 52B9 addq.l #1,$462

D0=3C0FFFF D1=0 D2=0 D3=3C00000 D4=0 D5=FFE0 D6=B D7=FF80
A0=174F0 A1=37F7A A2=444E2 A3=26108 A4=19074 A5=CC04 A6=1AFDC A7=1AF70 SR=2400
FC06F8 9BCD suba.l a5,a5

D0=3C0FFFF D1=0 D2=0 D3=3C00000 D4=0 D5=FFE0 D6=B D7=FF80
A0=174F0 A1=37F7A A2=444E2 A3=26108 A4=19074 A5=0 A6=1AFDC A7=1AF70 SR=2400
FC06FA 1039 move.b $ffff8260,d0

D0=3C0FFFC D1=0 D2=0 D3=3C00000 D4=0 D5=FFE0 D6=B D7=FF80
A0=174F0 A1=37F7A A2=444E2 A3=26108 A4=19074 A5=0 A6=1AFDC A7=1AF70 SR=2408
FC0700 C03C and.b #$3,d0

D0=3C0FF00 D1=0 D2=0 D3=3C00000 D4=0 D5=FFE0 D6=B D7=FF80
A0=174F0 A1=37F7A A2=444E2 A3=26108 A4=19074 A5=0 A6=1AFDC A7=1AF70 SR=2404
FC0704 B03C cmp.b #$2,d0



Really nice log, would be possible to add Instruction time in cycles?
How to activate it? I use Boiler 3.7 and mine is a bit different:

Code: Select all

>>> Start Emulation <<<
00F3B2: movem.l D0-7/A0-6,-(a7)
00F3B6: addq.w #1,$10a1c
00F3BC: cmpi.b #$2a,$fc02.W
00F3C2: beq.s +18 {$00F3D6}
00F3C4: move.l $10a10,d0
00F3CA: lsr.w #8,d0
00F3CC: move.l d0,$8200.w
00F3D0: jsr $f400
00F400: rts

BTW,
There is one a bit uncomfortable Boiler's behaviour. FlagLog --> CPU is always cleared after pressing "GO" button.

EDIT: is it possible to fix the title of this thread?

yes it is. What title do you propose?

[Steven Seagal]

Really nice log, would be possible to add Instruction time in cycles?
How to activate it? I use Boiler 3.7 and mine is a bit different:

Code: Select all

>>> Start Emulation <<<
00F3B2: movem.l D0-7/A0-6,-(a7)
00F3B6: addq.w #1,$10a1c
00F3BC: cmpi.b #$2a,$fc02.W
00F3C2: beq.s +18 {$00F3D6}
00F3C4: move.l $10a10,d0
00F3CA: lsr.w #8,d0
00F3CC: move.l d0,$8200.w
00F3D0: jsr $f400
00F400: rts

Check the powerful Control mask browser hack, line TRACE CPU:



See the potential for adding even more info, at user's choice?

Sorry, competition

BTW,
There is one a bit uncomfortable Boiler's behaviour. FlagLog --> CPU is always cleared after pressing "GO" button.

I already fixed that but only if you check 'suspend logging'.

EDIT: is it possible to fix the title of this thread?

yes it is. What title do you propose?

'Requests' instead of 'requerst' would be nice.

[Cyprian]

Check the powerful Control mask browser hack, line TRACE CPU:[/img]

See the potential for adding even more info, at user's choice?

need more options
As I mentioned would be cool to have "Instruction time in cycles",
Maybe also counting cycles since last VBL and HBL?

I already fixed that but only if you check 'suspend logging'.

good to know, actually I was wondering what does 'suspend logging' stand for.
Logging works fine now and I see that full output in file TRACE.txt

'Requests' instead of 'requerst' would be nice.

ops I missed it. Corrected. It seems its time to buy a stronger glasses

[Steven Seagal]

good to know, actually I was wondering what does 'suspend logging' stand for.
Logging works fine now and I see that full output in file TRACE.txt

'suspend logging' suppresses writing on original 'steem.log', if you don't use it check the option.
TRACE.txt is commanded by option 'SSE/Ouput TRACE to file'.
Silly but the TRACE system also works inside the Visual Studio interface, Boiler or not.

[Cyprian]

I observed that Boiler flag "Stop on Breakpoints" has no impact on any option in "Break on Interrupt".
E.g. when "Stop on Breakpoints" is unchecked and e.g "Timer A" is set then Boiler will stop on each Timer A occurrence.
From my point of view this is inconsistent behavior, because when I uncheck "Stop on Breakpoints" flag I expect that all breakpoints will be omitted (also Interrupts)

[Steven Seagal]

Yes, it's annoying, I'll change that the way you suggest.
It made sense in the way that breakpoints and IRQ are different.
We could add an option 'suspend break on interrupt', but there are many options already.