Gecko dotNET Bugs and Requests

[dcx2]

When Gecko.NET experiences a problem it writes to a GDNDebug.log file in the Logs folder.  It's timestamped to correspond to when the problem occurred.  If you can post or PM me the contents of the log file corresponding to a problem that you had, it will help me out.

I can't believe that the Wii resets itself, though.  I've never had that happen.  It's really weird.

[Arudo]

Interesting I'll have to check when I get home.

[Bully@Wiiplaza]

Aww... I was trying to debug one of my C2 codes, since it froze after the developers´ logo if the code was enabled right from the beginning.

--

So I set a conditional breakpoint after the bootstrap screen and geckodotnet skipped lots of instructions because the condition wasn´t true yet, but the instruction was executed there. Then, the game randomly froze quite fast and geckodotnet didn´t react anymore (I couldn´t cancel the breakpoint nor could I successfully press the "run" button)

Damn it, this appears to only happen with the newer version(s) of geckodotnet (probably because of the C2 codes fix codeshandler?) and only when the game is *before* the main screen therefore still *booting up*

Why I don´t use WiiRd then?
It´s somehow blocked by my firewall/anti virus program and I don´t know how to reenable it -.-

That bug makes it therefore almost impossible to debug this C2 code. :(

If someone cares which code I was trying to debug...
[spoiler]it´s the "offline only" inf. health code for Conduit2 that´s posted on the database.
The freeze can be bypassed by using button activators but that´s not what I want though!
Conduit2 and The Conduit are pretty much the biggest bitches with assembly of all games out there anyways... (you can never be sure that your assembly patch doesn´t randomly freeze at a certain point, instant freeze or your game gets glitchy)[/spoiler]

[dcx2]

0.65.2 can disable the RX/TX patches on the About tab.  I would restart the game after disabling the patches, though.

[Bully@Wiiplaza]

0.65.2 can disable the RX/TX patches on the About tab.  I would restart the game after disabling the patches, though.

Haha, nevermind.
Didn´t notice that it´s the feature I need to disable here.

Disabling it in-game could have possible crashes etc.?
If yes, I wouldn´t bother about it, though. (restarting doesn´t hurt since I crash often enough aswell xD)

EDIT:
I could possibly disable it before connection to a game to prevent the game from a restart?
It would also be nice if the change is saved on geckodotnet each time it is booted up, does it do it yet?

[dcx2]

The RX/TX patch suffer from the same problem as any C2 code.  When you apply a hook, and then try to disable it, the original instruction is not restored.

The game won't crash immediately, but the next time you send codes, the code handler will probably fail when it tries to execute the missing patches.

EDIT:

Yes, the RX/TX enable is a persistent setting.  So you won't need to disable it every time.

[Bully@Wiiplaza]

And you can´t possibly unhook this patch with the disable option on geckodotnet?

*got fixed*

[Bully@Wiiplaza]

i´d like to have the ghost add bug fixed. If you press "add gct code" and cancel it afterwards, it STILL adds the code... ???
Also, a little codeslines counter would be cool on the gct tab to see the amount of codeslines that are enabled (like the one that WiiRdGUI has!)
I discovered on geckodotnet 0.63.3 that some breakpoints don´t hit and just skip over (the counter tells me that it happened) even if no conditions were set by the user...! That makes it hard to create the specific code. Even though, it somtimes works afterwards, but not always.

[dcx2]

i´d like to have the ghost add bug fixed. If you press "add gct code" and cancel it afterwards, it STILL adds the code... ???
Also, a little codeslines counter would be cool on the gct tab to see the amount of codeslines that are enabled (like the one that WiiRdGUI has!)

I will look into these.

I discovered on geckodotnet 0.63.3 that some breakpoints don´t hit and just skip over (the counter tells me that it happened) even if no conditions were set by the user...! That makes it hard to create the specific code. Even though, it somtimes works afterwards, but not always.

What you are seeing is called Skip Unaligned Data Breakpoints.  It skips false breakpoints.  You can disable this on the (EDIT BP) About tab if you really want to.  However, I strongly recommend leaving it enabled unless you really understand what the rest of this post means.

---

The technical explanation is that a data breakpoint, such as read or write or read/write, actually has double-word (8-byte) alignment.  Let us say that *something* is writing the value 00000000 to the address 80123444.  Let us say you set a write breakpoint on 80123440.  Your write breakpoint will generate a false hit when *something* writes to 80123444.  This is because 80123444 is double-word aligned with 80123440, even though it is not word aligned.

You can verify this for yourself.  Look at DAR.  You will see that DAR does not match your breakpoint sometimes.  These are the false breakpoints; they are not word-aligned, but they are double-word aligned.

In binary terms, a non-exact breakpoint asks the following question.

(memory_access_address & 0xFFFFFFFE) == (breakpoint_address & 0xFFFFFFFE)

That is, the addresses are masked with a series of 1's ending with three 0's.  This enforces double-word alignment.  Skip Unaligned Data Breakpoint does a second check to ensure that the following is true.

(memory_access_address & 0xFFFFFFFC) == (breakpoint_address & 0xFFFFFFFC)

That is, the addresses are masked with a series of 1's ending with two 0's.  This will prevent the false breakpoint write to 80123444 from happening when we were actually interested in 80123440.

[Bully@Wiiplaza]

Learned something new again :smileyface:
I just need to get sure that my BP address is the EXACT same address than the one found in the Memory Viewer/Search.
If it still does not trigger I can also disable this option, ok (I didn´t like the unexact BP´s though, but that´s not a problem).

[dcx2]

It's not quite that simple.

An exact BP must match the address exactly.  Let's say you find a value in memory, like Mario's HP which is 3.  It might look like 00000003.  This is at address 80123444.

Is this a word access on 80123444?  a half-word access on 80123446?  a byte access on 80123447?  If you use exact, you must know exactly.  an lhz on 80123446 will not hit an exact BP on 80123444.

If you use non-exact, then any access to that word will hit.

Trust me.  Unless you fully understand the meaning of double-word breakpoint alignment, just leave the checkbox enabled.

[Bully@Wiiplaza]

great work with the update!

Sry to say that now, but I noticed that the "paste" option on the Breakpoint Condition Field gives a crash message.
But if Ctrl + v is pressed pasting works fine.
It says something like the command wasn´t assigned yet on the crash log.
And there´s still a ghost add on the memory viewer ("add gct code").

[Bully@Wiiplaza]

Apply this code via send cheats to activate the extended codes feature

how many codeslines become useable then?
It was like 220 before, wasn´t it?
One mostly doesn´t need that many codeslines but sometimes... it´s useful :P
e.g. on brawl or mkwii since there are lots of long and good codes.

I get a "too many codes found" message on brawl if I want to use sd cheats plus debugging.
SD Cheats alone does work since it doesn´t break the limit yet.

[dcx2]

Not sure how many code lines it was before, but if you use the Code Extender set to 32k, that would give you room for about 4,000 additional code lines.  Note that any non-extended codes after "Activate extended codes" will not be executed, so make sure that it is the last non-extended code in the GCT tab listbox.  Extended codes can be anywhere in the listbox, before or after Activate Extended Codes.

[Bully@Wiiplaza]

Not sure how many code lines it was before, but if you use the Code Extender set to 32k, that would give you room for about 4,000 additional code lines.  Note that any non-extended codes after "Activate extended codes" will not be executed, so make sure that it is the last non-extended code in the GCT tab listbox.  Extended codes can be anywhere in the listbox, before or after Activate Extended Codes.

Do you mean that the extender code must be the last code on my "non extended" codeslist or otherwise codes below it won´t be executed/applied?
After this was enabled, I can send more codes?
___

There something else I want to mention.
If I close geckodot.net after e.g. my game crashed, it sometimes remains active on the system, even though, the gui is gone.
It plays this (windows) notification sound from time to time. Probably because it couldn´t connect or something.
On task manager I can still see it that geckodot.net is there. I need to close it with task manager to get rid of the notification sound.
It´s a little annoying glitch that exists since geckodot.net was released. It still happens quite often. ???

But no worries, any time in the future there won´t be any glitches anymore that I can discover