Genesis GX Plus 1.4.0... and it's initial quirks

For anyone out there that doesn't have HBB, or who's just passing by and didn't notice it... a new version of the acclaimed emulator Genesis Plus for the Wii and Gamecube has come out. But my reason here isn't solely for stating the obvious.

I was quite thrilled to find out this emulator had finally received an update after 2 years of lacking updates. Moreso, because the previous version stopped finding and loading my roms when I moved all my stuff from a smaller USB to my currently used HDD. And thankfully, this version actually did find my roms.

But as soon as I fired it up, some weird little hanging instances ocurred. Among the first things I noticed that after pressing A on the Wiimote at the disclaimer, the Wiimote would not work with the emulator's menu. I had to switch to a Gamecube controller I had connected to the console to get the ball rolling.
The first hanging came from trying to see my current configuration, and then attempting to exit. It hanged, and required manually powering off the Wii. I tried turning it off with my Wiimote, and it surprised me that, unlike other apps the utterly failed to shutdown the Wii and instead only made the Wiimote turn off, this time the Wiimote itself didn't even bother turning off. Weird.

So anyways, several seconds later, after firing up the emulator again, I went straight to my roms. Since I had not played anything yet, I decided to open up my device and load a rom. I selected a random rom, pressed A to load it... hanged.
After reloading the emulator again, I remembered I didn't delete my history log from the previous emulator, so I decided to give it a shot and tried loading something from there. Success! Sonic the Hedgehog was up and running... but the controller did not respond at all. I could pop up the emulator menu, but I couldn't start the game at all. I thought that maybe the settings were rigged up to use my Wiimote by default, but even pressing all the buttons did nothing. After several times of summoning the menu, it stopped coming up and, while the game's demo was shown playing perfectly on-screen, there was no way to input anything on the Wii anymore. I decided to reset the Wii in hopes of it kicking me to HBC, but instead THE GAME ITSELF RESET. Nice feature, but most certainly not something I was expecting to see at a moment I wished to go back and try loading the emulator again.

Before I attempted messing with it again, I thought for a moment about the possibility of both the settings and history files from the previous version possibly conflicting with the emulator, so I hooked it up to my PC, deleted them, and tried again.
After I did this, everything seemed to work alot better now. I loaded Toejam & Earl from my device instead of my Recently Played list, and played for a few minutes. But I noticed that upon pressing some buttons, the game thought I had two controllers connected, and I accidentally summoned Player 2, which mimicked my every move. Thinking it was a simple bad setting, I exited to the emulator's menu... and it hanged again.

So now it's 4 am where I live, and I'd like to keep trying several ideas I had to make it work, but I still have to reassemble a USB Keyboard I had to fix and have no time right now. I took a few minutes to write this report, because otherwise I'd forget about it when I'd wake up.

Looking back...

+ Old Settings and (possibly) History files from previous Genesis Plus versions conflict with this new version
+ Emulator thinks there Player 1 is also simultaneously Player 2 (???)
+ Numerous random hangs

Now, I wasn't meaning to insult ekeeke in any way with this post. I love this emulator, as well as any other emulator in the GX line of apps, and the Wii64 Team releases too. But I just wanted to report these issues only to make this emulator achieve perfection faster.

Thank you man for your hard work.

A very concise review and expanded on some of the thoughts running through my head. Excited that it's received a quite a lengthy update following a couple of years in the wildnerness, but a little four steps forward - a couple back.

It's mainly the hanging for me, and a few issues that lead me to also suspect a conflict between past data held from the earlier version and this revised one. Not experienced the one-two player controller issues.

Full props to ekeeke for the update - I hope these minor niggles could be looked at in the next month or two for a fix to what is otherwise a decent overhaul of this project.

Thanks for the feedback. I have tried to test all this quirks but apparently it's always when it's released to the mass that they start appearing :)

Quote

But as soon as I fired it up, some weird little hanging instances ocurred. Among the first things I noticed that after pressing A on the Wiimote at the disclaimer, the Wiimote would not work with the emulator's menu.

Huh ?
Never seen this before, there are actually no reason wiimote button work on the disclaimer but not in the menu
Or do you mean you can not aim with the wiimote and select buttons ? Be more precise when writing "not working" because it can be many different things.

Quote

The first hanging came from trying to see my current configuration, and then attempting to exit. It hanged, and required manually powering off the Wii. I tried turning it off with my Wiimote, and it surprised me that, unlike other apps the utterly failed to shutdown the Wii and instead only made the Wiimote turn off, this time the Wiimote itself didn't even bother turning off. Weird.

how did you attempt to exit the emulator ? what option did you select ? The wiimote power off button work like in any other app but off course, only if the program is not crashed and it seems it did here, maybe while trying to update the config file.
Is your SD write-protected ?

Quote

I decided to reset the Wii in hopes of it kicking me to HBC, but instead THE GAME ITSELF RESET. Nice feature, but most certainly not something I was expecting to see at a moment I wished to go back and try loading the emulator again.

that's perfectly normal, make sure you have read the user manual

Quote

After I did this, everything seemed to work alot better now. I loaded Toejam & Earl from my device instead of my Recently Played list, and played for a few minutes. But I noticed that upon pressing some buttons, the game thought I had two controllers connected, and I accidentally summoned Player 2, which mimicked my every move. Thinking it was a simple bad setting, I exited to the emulator's menu... and it hanged again.

This is strange, works perfectly fine for me. It seems like a faulty controller configuration (same controller affected to both ports) Can you describe what controller configuration you are using ? Again, make sure you have carefully read the user manual regarding this menu also, I spent quite a lot of time typing it and would hope users to understand the features and settings correctly.

Quote

+ Old Settings and (possibly) History files from previous Genesis Plus versions conflict with this new version

for history, you need your old ROM files to be still at the previous location obviously, if they have been moved else where, then nothing will be loaded and eventually the emulator will hang looking for the file so better delete it if you are not sure.
I don't think there is conflict version for that file, it's basically a list with file name and directory and did not change since 1.3.1

the config.ini file, however, should have been automatically reseted to default .

that's said, I spotted an issue with the confi.ini being reseted everytime you start the emulator, this has been now fixed and you should download the archive again if you want your settings to be saved.

Quote

+ Numerous random hangs

Quote

It's mainly the hanging for me, and a few issues that lead me to also suspect a conflict between past data held from the earlier version and this revised one.

What kind of hangs ? When exactly ? Do hangs still happen after you deleted your old config files ? This is not very clear and I didn't experiment a single hang during the last weeks of testing. Hangs can off course occur if the config file got corrupted or if you modified the controller settings or key mapping (which would explain you failed at returning to the menu or start the game since the emulator did not recognize MENU & GAMEPAD keys anymore).


I would suggest the following:

1) start a fresh install, no configuration file, no history file, no saved states (there are not compatible with the new version) and verify the SD card is not write-protected.

2) download the updated version that fixes settings being reseted when the emulator is started.

3) before loading a game, go configure your controllers in the controllers settings menu as you want to use them (player 1, player 2, etc). Read the user manual if you need explanations. Also, before changing anything, please report the default configuration that was set by the emulator and report your current setup (number of gamecube controllers, number of registered wiimotes, etc...) so I can see if there is a problem here (unfortunately I can only test my own setup which is a single wiimote, a classic controller, one gamecube controller and ... that's all)



Edited 3 time(s). Last edit at 11/02/2010 03:41PM by ekeeke.

On a side note, I just noticed that some weird DSI exception could occur when the network is not initialized properly (i.e your Wii not connected to Wifi) after you launch the application through the HBC.

The crash is within libogc and is always the same, crashing during an IRQ handler (making the RESET button reload feature ineffective)

lwp_heap_fre->iosFree->c_irq_dispatcher->irq_exception_handler


It is quite immediate (~ one minute) after the application starts and does not seem to happen anymore when Wifi is running and network initialized. Having USB drive connected or not does not change anything so, contrary to what I initially thought, the instability does not seem to be in libogc USB storage driver but in the way HBC leaves the network when it starts an application.

Maybe the Wifi device is not shutdown properly and continue to generate IRQ to the CPU, which confuse libogc in the loaded application (what seems to happen is that libogc tries to free a resource that was not allocated, resulting in a unrecoverable exception)



Edited 2 time(s). Last edit at 11/03/2010 09:37AM by ekeeke.

Ok. Sorry for not reporting back sooner. I had to prep a PPT presentation for tomorrow, and had to catch up on some missed work. Haven't yet thoroughly tested, do I did download the supposed corrected version, which I only tested for around 5 minutes or so.

I don't think I'll be adding any new info, but rather clearing up some of my most vague parts of my report. I'll get to it as soon as I can.

Quote

But as soon as I fired it up, some weird little hanging instances ocurred. Among the first things I noticed that after pressing A on the Wiimote at the disclaimer, the Wiimote would not work with the emulator's menu.

Well, I didn't know this version worked with a pointer now, heh. Though I did wave the Wiimote around to see if I get some response. Also tried moving the cursor with the D-Pad with the Wiimote, but it didn't work. However, this only occurred to me the first time I ran it, with the Settings file from the previous version present. Wiimote has worked fine since it's deletion.

Quote

The first hanging came from trying to see my current configuration, and then attempting to exit. It hanged, and required manually powering off the Wii. I tried turning it off with my Wiimote, and it surprised me that, unlike other apps the utterly failed to shutdown the Wii and instead only made the Wiimote turn off, this time the Wiimote itself didn't even bother turning off. Weird.

What I meant to say that the emulator hanged the following matter: emulator loaded, then I accessed Settings, then without moving anything, I tried to go back to the main menu, but it froze.
Regarding the Wiimote part, I meant to say that while in the Wii was in it's frozen state, the Wiimote would not turn the Wii off. This is a typical expected thing to find when an app has crashed. And when pressing down the Power Button on the Wiimote while the console is in this state, the Wiimote would only turn itself off. What I said that surprised me was that the Wiimote didn't turn off as I'd thought it would while the Wii was frozen, but instead remained on and seemingly equally as unresponsive as the console.
Also, I'm using an HDD, not an SD card. And I don't know how to write protect my HDD anyways, so...

Quote

After I did this, everything seemed to work alot better now. I loaded Toejam & Earl from my device instead of my Recently Played list, and played for a few minutes. But I noticed that upon pressing some buttons, the game thought I had two controllers connected, and I accidentally summoned Player 2, which mimicked my every move. Thinking it was a simple bad setting, I exited to the emulator's menu... and it hanged again.

After deleting the Settings from the previous version, I accessed the Controller configuration before I did anything else to see the default controller settings as instructed. What I found was that there were two controllers setup. Player 1 had Wiimote enabled, and Player 2 had the GameCube controller. No wonder I couldn't play Sonic last time! Even while disabling Player 2 and leaving it with no physical control installed, somehow I could keep summoning Player 2 in Toejam & Earl (I'm not entirely sure about this right now... mind is kinda fuzzy 'cuz I need to sleep). I also noticed that if you change the controller settings while a game is loaded, the controller would stop responding in game. This might have only occurred to me in that one instance, but I'll try to see if I can reproduce it as soon as possible.
Ummm... I didn't mean NOT to read the manual... just as I said, I haven't had that much free time right now. For the record, I did read the one included in the previous version... if that counts for anything.

Quote

+ Old Settings and (possibly) History files from previous Genesis Plus versions conflict with this new version

While I went on the safe side and just deleted the old Settings File, before I used the revised version you mentioned that fixed the settings file not being saved, it wouldn't even create one. Now with the revised version, it does create one, but it doesn't seem to be of any use. After changing my controller settings, exiting to loader and then loading the emulator again, it still doesn't seem to be saving anything.
A "config.ini" file was created this time, but all it contains is a single line that says "GENPLUS-GX 1.4.0", and nothing else. I'd expect to see some strings with a yes/no activation kind of thing, or the gibberish kind of settings file, but nope. This one only had that one line, and that's it.

Quote

+ Numerous random hangs

These I cannot determine exactly. They just happen all of a sudden and without warning, and they seem to only happen when trying to load a ROM, or when trying to go back one screen in any given menu (such as going back from Settings menu to the Main Menu). Also seems to hang when trying to load another ROM after already having loaded one.

Regarding the controller setup: when I first tried it, I was using a Wiimote and had 2 GameCube controllers connected to the console (ports 1 and 2), as well as a Memory Card in slot A. I disconnected my Memory Card and one GameCube controller (port 2) since. I only have one Wiimote, and that sole is the only one registered on my console.
As I mentioned above, the initial controller settings after erasing the previous config.ini file revealed that Player 1 had the Wiimote and Player 2 had the GameCube controller plugged in port 1.

I'm quite lightheaded right now, and I know I'm missing a few more details, but I hope that what I clarified for now has given you some hints. I'll try to break the emulator down in as many crazy ways possible tomorrow so I can give you a more informative report (can't guarantee that I will, though).

Thanks again man, and keep the work comin'.

Quote

Even while disabling Player 2 and leaving it with no physical control installed, somehow I could keep summoning Player 2 in Toejam & Earl (I'm not entirely sure about this right now... mind is kinda fuzzy 'cuz I need to sleep).

I tried what you did with only wiimote connected to player 1 and player 2 is not summoned, you must have messed something with the emulated devices

Quote

I also noticed that if you change the controller settings while a game is loaded, the controller would stop responding in game.

it can stop responding if you accidentally switch to another type of device (for example, emulating a mouse instead of gamepad, or a multi-tap in a game that does not support them, etc...) without reseting the game ... affecting another input device definitively does not stop it from responding

I tested this part a lot and I can assure you there are no such bugs there, you must have done something wrong while playing blindly with controller options.

Quote

For the record, I did read the one included in the previous version... if that counts for anything.

the emulator interface was completely changed so the old manual does not really apply anymore. You should definitively first read the Manual before attempting to play with options like you did.

Quote

before I used the revised version you mentioned that fixed the settings file not being saved, it wouldn't even create one.

hum, are you sure about that ? because a config file is always created on your default FAT device (sd:/genplus/ if it is found, usb:/genplus/ otherwise), every time you go back from the option menu or when you exit the emulator normally (that means, NOT if you unplugged the power cable or hold the Power button on the console).

Quote

A "config.ini" file was created this time, but all it contains is a single line that says "GENPLUS-GX 1.4.0", and nothing else. I'd expect to see some strings with a yes/no activation kind of thing, or the gibberish kind of settings file, but nope.

it's not a text file so you won't be able to read it normally (apart from the version string); Please upload the generated file here or somewhere else

Quote

These I cannot determine exactly. They just happen all of a sudden and without warning, and they seem to only happen when trying to load a ROM, or when trying to go back one screen in any given menu (such as going back from Settings menu to the Main Menu). Also seems to hang when trying to load another ROM after already having loaded one.

It hangs just like that ? No exception screen with a lot of number, no reboot, just you can't control the menu anymore ?
that's really weird because it doesn't happen to me at all and I would have received many issues report if it was happening to others. I'm wondering, do you have Wifi on when you boot the HBC ? This is the only case (network configured in System menu but Wifi switched OFF ) where I have seen such crash, as explained in my previous post. Otherwwise, I can spend hours in the menu without a hang.

Quote

Thanks again man, and keep the work comin'.

Unfortunately, I don't see what I can fix on my side. Everything work perfectly here (well, as long as I have Wifi ON but that's definitively not an issue from the emulator itself) and as expected. I don't know, maybe it's because you are only using an USB drive, try using SD and see if that changes anything.



Edited 2 time(s). Last edit at 11/03/2010 04:12PM by ekeeke.

Another bug (?) I spotted with HBC (IOS58 + no_ios_reload). This only happen if no SD is inserted and you are loading apps from the USB drive.

Now, if you wait for network to initialize (bottom icon stops blinking) then load the application, the USB drive cannot be accessed by the application after calling fatInitDefault. If you don't wait for network and load the application immediately when HBC starts, everything is fine.

I don't know if it is a bug with IOS58, HBC with no_ios_reload or libogc, but it would definitively explain why settings are sometime not saved or restored when using only an USB drive.

Until this is clarified, I'd suggest removing the no_ios_reload flag from the meta.xml file if you have stability issues when loading from HBC.

It's a bug with HBC, where HBC tells the drive to go to sleep before launching your app. As a result, an immediate call to init it may fail. To work around this in WiiMC (if I don't find an SD card) I attempt to mount a usb drive for 5 seconds before giving up.

See here: [code.google.com]

For me it doesn't matter if the network is initialized or not.

Basically what's happening is ever since HBC added the ability to avoid the IOS reload, some of these bugs are cropping up. The bugs were always there, it's just no-one noticed (or cared) before because of the IOS reloading.

Ok thanks a lot, this was so much unstable I thought I was going nuts :-)
I'm doing something similar now and it indeed seems to work better.

  int retry = 0;
  int fatMounted = fatInitDefault();
  while (!fatMounted && (retry < 12))
  {
    usleep(250000);
    fatMounted = fatInitDefault();
    retry++;
  }

Any insight about the freezes and DSI exception that occurs when Wifi router is switched off ? Again, this does not happen when no_ios_reload is trashed.
I think I somehow fixed it by removing the redundant call to _ES_Init in _ahb_check_aprot (di.c) when DI_Init is called at the start of the application but I'm not sure, it seems it's still quite unstable and subject to crash sometime but not as often as before (it usually crashed after less than one minute)



Edited 3 time(s). Last edit at 11/04/2010 10:25AM by ekeeke.

Quote

It hangs just like that ? No exception screen with a lot of number, no reboot, just you can't control the menu anymore ?
that's really weird because it doesn't happen to me at all and I would have received many issues report if it was happening to others. I'm wondering, do you have Wifi on when you boot the HBC ?

Nope. That's why I said "hang" and not "crash with code dump". I was about to suggest you to make it crash and spit out a dump so I could paste the issue much easier, but now I know it's something wrong on my side.

I'm at school right now, and I just came by to drop a line. I'll edit this post later tonight to reflect upon the possibility that maybe my HDD is the one at fault, and test the emulator on an SD card.

I can't make the program crash as I want or display stuff before it hangs: indeed, it hangs because something got wrong in libogc interrupt callback processing... and never return to the application. There is no way to predict this.

As I said, also try to start the HBC with your Wifi router ON (if you ever have one) as freeze only happened to me when Wifi is shutdown.

On a last solution, edit the meta.xml file to remove the no_ios_reload line which seem the source of all these problems. You would not be able to load ROM from DVD anymore though.

Sadly, there isn't much I can do on my side, all this instability is caused by bugs in either libogc and/or HBC, under specific conditions; Network and USB devices, as well as IOS Startup/Shutdown seem to be very sensitive to handle.

About your other issues:

- Controller issues you were having are most likely due to a wrong configuration setup or you not using the setup controllers properly, I now have tested many more test case and I can ensure there are no bugs here; I might add some additional checks when returning to the game to ensure the user did not disconnected a configured controller, did not connected the same controller to 2 players (should not be possible through the menu but you never know), etc...

- I don't explain why your wiimote was not detected in the first place but it could be as well a consequence of the above instability or maybe you were not handling the wiimote properly ("not working" is generally a vague description).

- I am pretty confident that your old configuration file cannot have any incidence on any of the issues you had as an outdated config file cannot conflict with a new version or cause dysfunction (the default configuration is always restored if the file version mismatch, which should have been the case). Most of the hangs or crash seem to be caused by the instability of libogc when launched from HBC with no_ios_reload.

Quote
ekeeke
Any insight about the freezes and DSI exception that occurs when Wifi router is switched off ? Again, this does not happen when no_ios_reload is trashed.
I think I somehow fixed it by removing the redundant call to _ES_Init in _ahb_check_aprot (di.c) when DI_Init is called at the start of the application but I'm not sure, it seems it's still quite unstable and subject to crash sometime but not as often as before (it usually crashed after less than one minute)

Yes, this has been bothering me a lot too and all apps using no_ios_reload suffer from it. It's a bug in HBC - it makes an async net call that doesn't finish before it loads your app. The solution is for HBC to properly call net_deinit before loading the app. We'll just have to wait for an HBC update to do this.

Alright, thanks for confirming the bug, it's very annoying indeed.

So basically this means a net device is still opened on the IOS side and off course we can not close it anymore .
Does this also mean that you cannot use the network when using no_ios_reload, since net_init() would probably fail when trying to opening IOS devices ?

Though, I don't understand what is causing the DSI exception ? Is it the old async callback being called by IPC interrupt which now execute invalid code ? Could it be possible to prevent this by always clearing old callbacks in memory ? I'm not sure if I understand the IPC mechanism very well so I might be totally thinking this wrong . Shouldn't the random IPC_REQ_MAGIC prevent any IPC reponse generated by another compiled application from being valid ?



Edited 3 time(s). Last edit at 11/04/2010 06:25PM by ekeeke.

It's only a problem when an async IPC call is made and IOS doesn't respond before HBC exits. It's not a problem if the call returns to HBC.

There's a loop that's called when your app starts that looks something like this:

for(i=0; i < MAX_HANDLE; i++) IOS_Close(i);

So the problem isn't really that a handle can't be reopened.

The problem is actually two-fold - HBC tries to init the net even though there's no net available (and this takes while) AND HBC tried to init WC24 (due to a libogc bug that tueidj fixed), even though the net isn't available.

What's worse is that HBC does stuff like calling net_deinit from within a thread during shutdown, but doesn't wait for this thread to finish. So maybe net_deinit isn't reached, or maybe it is executed but doesn't finish in time before loading the app. Either way it has a bad result.

If you sit in HBC and wait for awhile (for HBC to finish trying to init the net), then load your app, you won't have a problem. But if you load your app right away, the app loads, and then a stray IPC response from IOS arrives to *your* app, initiated by HBC. Your app calls the callback, which points to a PPC address from HBC, and *CRASH*

An alternate possibility is that IOS panics because HBC doesn't leave it in a sane state.

Doing an IOS reload in HBC right before loading the app fixed any problems like this in the past, since IOS is now fresh and the PPC state is also fresh (since it's of course lost when the app loads).

This is the way I understand the situation, but I'm no expert...

Also having a look at __ES_Init(), I don't see a reason why calling it multiple times would hurt anything.

Quote

and then a stray IPC response from IOS arrives to *your* app, initiated by HBC. Your app calls the callback, which points to a PPC address from HBC, and *CRASH*

It seems to me it crashes when calling _ipc_freereq after the cb, this is the only iosFree occurence I see in the IPC interrupt callback (the DSI stack dump indeed indicates the crash occurs during execution of iosFree, in an IRQ handler). The pointer returned by IPC off course points to unallocated memory, hence the DSI exception.

However, this is that part I don't understand when reading libogc code. If you look at the __ipc_replyhandler function which is called by IPC interrupt handler, you see that the response first parameter is checked against IPC_REQ_MAGIC. This value is randomly generated when the application starts so normally, any response initiated by another application (i.e HBC) would be rejected and the callback should not be executed.

Quote

An alternate possibility is that IOS panics because HBC doesn't leave it in a sane state.

Indeed it's possible IOS is using some area in main memory that has been reinitialized when the application launches. But this does not explain why the PPC code would crash during the the IPC interrupt callback.

Quote

Also having a look at __ES_Init(), I don't see a reason why calling it multiple times would hurt anything.

well, the only thing I noticed is that it calls __ES_InitFS which is reentrant so AddDevice function was called each time you call __ES_Init but I don't know what it could break

but you are right, this didn't fix anything. I also tried to add a call to _IPC_Reinitialize on startup (this was usually done in __IOS_LaunchNewIOS) and expected there was a way to reset IOS without reloading it but did not change anything either. It indeed depends on the time you wait in the HBC before you load the application, and depending on where it was in the network init chain, it later crashes or not.

I guess I will wait patiently for a fixed HBC now, this thread is already far off topic :-)

This new version looks the nuts, but I'm having an issue.
How do you get it to save all initial settings ie pad type,button config, screen position etc as I have to alter these every time I load up the emulator?
Edit: I just read that this issue was fixed, but the file I downloaded from homebrew browser still resets the settings.
Another Edit: just downloaded file from wiki which must be newer than homebrew browser version as file is bigger and all works now.



Edited 2 time(s). Last edit at 11/07/2010 10:25PM by bluelion.

Homebrew Browser might not have updated the binary (i didn't changed the version number for the fixed one). Better download directly from genplus googlecode and update the boot.dol file.



Edited 1 time(s). Last edit at 11/07/2010 10:01PM by ekeeke.

Another problem is that arcade sticks won't work with the new version of the emulator. I haven't tested my new arcade stick with the old version, but did with Hugo GX and SMS emulators and it worked fine with those. The joystick just doesn't work with emulators that have the more advanced menus instead of the simple ones like the older version of Genesis Plus GX and the others I mentioned have. Can we try to fix this please?