4.6 iOS 64-bit beta

Testing my armv7 build with il2cpp on both 32 and 64 bit devices, if the player tries to enter their name using the keyboard on an iPad mini gen1 the game crashes. On iPad air2 no crash.

Edit: Report from beta tester on iPad mini 32 bit--


"On my iPad mini I have not updated to 8.1 I have 1.6gb available after the game is loaded. Is this enough? Should I make more space or upgrade the mini?
The game keeps crashing in random places, seems very unstable."

Should I try to get the full file to you? Submit a bug report with stripped file and then you could tell me how to get you the full file? Thanks for your help.

 

Any little news about the Case 670422, about ExternalCall missing method ?
Sorry for the inconvenience. Just looking to being able to compile on iOS in a near future.

 

Just noticed this:

Is there any plans to support Thread.Abort? "using safer mechanisms" is a nice advice, but it's not always possible, for example, when thread is inside a blocking call and I still have to abort it somehow.

 

Mono is greyed out in 4.6.3 for me. I don't have stack traces. That's not something I know how to do or how to ask my testers to do. Have never had to think about the stack before. Is it an easy thing to explain to me what I have to do?

But the game according to my testers now is also unstable on iPad air2 which is 64 bit. Much worse on 32 bit.

I built using armv7 only with il2cpp per JoshPeterson's suggestion. It was the first time I was able to get my game running very far without folding early on. So this is progress. At least I can test gameplay although rather lamely.

Are things more stable using Unity 5 beta?

 

@Lucas Meijer
Since I sent a bug report(671702) a week ago, I have not yet received any feedback from FogBugz and this thread.
while I'm waiting I have tried to build with unity 4.6.2p2 and 4.6.3 but it still crashes same as before.

I am looking forward to hear from you soon.

 

Sorry if already reported, I built a game with https://github.com/playgameservices/play-games-plugin-for-unity for iOS on Unity 4.6.3 f1 (IL2CPP, Graphics: Automatic), and got the following error from Unity. The error occurs even when no Google Play Games API is called in my scripts. Please take a look at it.

This file says our plugin version is 0.9.11.
https://github.com/playgameservices...inDev/Assets/GooglePlayGames/PluginVersion.cs

The offending method is
https://github.com/playgameservices...tforms/Native/PInvoke/PInvokeUtilities.cs#L52

 

Hi.

We've updated to 4.6.3 and now we can't create a new xCode project (I can confirm that the build is created with 4.6.2f1 both with Mono and IL2CPP backend). Error message:

Code (csharp):

1.  
2. stdout;
3. Fatal error in Mono CIL Linker
4. System.Exception: Error processing method: 'System.Void Ionic.Zlib.ParallelDeflateOutputStream::TraceOutput(Ionic.Zlib.ParallelDeflateOutputStream/TraceBits,System.String,System.Object[])' in assembly: 'Ionic.Zlib.dll' ---> Mono.Cecil.ResolutionException: Failed to resolve System.Void System.Console::set_ForegroundColor(System.ConsoleColor)
5.   at Mono.Linker.Steps.MarkStep.MarkMethod (Mono.Cecil.MethodReference reference) [0x00000] in <filename unknown>:0
6.   at Mono.Linker.Steps.MarkStep.MarkInstruction (Mono.Cecil.Cil.Instruction instruction) [0x00000] in <filename unknown>:0
7.   at Mono.Linker.Steps.MarkStep.MarkMethodBody (Mono.Cecil.Cil.MethodBody body) [0x00000] in <filename unknown>:0
8.   at Mono.Linker.Steps.MarkStep.ProcessMethod (Mono.Cecil.MethodDefinition method) [0x00000] in <filename unknown>:0
9.   at Mono.Linker.Steps.MarkStep.ProcessQueue () [0x00000] in <filename unknown>:0
10.   --- End of inner exception stack trace ---
11.   at Mono.Linker.Steps.MarkStep.ProcessQueue () [0x00000] in <filename unknown>:0
12.   at Mono.Linker.Steps.MarkStep.Process () [0x00000] in <filename unknown>:0
13.   at Mono.Linker.Steps.MarkStep.Process (Mono.Linker.LinkContext context) [0x00000] in <filename unknown>:0
14.   at Mono.Linker.Pipeline.Process (Mono.Linker.LinkContext context) [0x00000] in <filename unknown>:0
15.   at Mono.Linker.Driver.Run () [0x00000] in <filename unknown>:0
16.   at Mono.Linker.Driver.RunDriver (Mono.Linker.Driver driver) [0x00000] in <filename unknown>:0
17. stderr:
18.  
19. UnityEngine.Debug:LogError(Object)
20. UnityEditorInternal.Runner:RunManagedProgram(String, String, String, CompilerOutputParserBase)
21. UnityEditorInternal.AssemblyStripper:RunAssemblyLinker(IEnumerable`1, String&, String&, String, String)
22. UnityEditorInternal.AssemblyStripper:StripAssembliesTo(String, String, String&, String&, String, String, IEnumerable`1)
23. UnityEditorInternal.AssemblyStripper:Strip(String, String, String&, String&, String, String, IEnumerable`1)
24. UnityEditorInternal.AssemblyStripper:Strip(String[], String[], String, String, String&, String&, String, String, IEnumerable`1)
25. UnityEditorInternal.IL2CPPBuilder:RunAssemblyStripper(IEnumerable, String, String[], String[], String)
26. UnityEditorInternal.IL2CPPBuilder:StripAssemblies(String[], String)
27. UnityEditorInternal.IL2CPPBuilder:Run()
28. UnityEditorInternal.IL2CPPUtils:RunIl2Cpp(String, String, IIl2CppPlatformProvider, Action`1, RuntimeClassRegistry)
29. UnityEditor.HostView:OnGUI()
30.  

It's related to the Ionic Zlib dll we're using in our project. How can we exclude the dll from stripping? I tried some stuff in the link.xml, but nothing worked so far.

 

Thanks for explanation. I do understand that proper support for Thread.Abort is not that easy, and that Thread.Interrupt is a better option. It's just I assumed Thread.Abort will be supported as Mono supports it.
"blocking call" could be something as simple as Stream.Read or Socket.Read. Once a thread goes inside these methods, there is no common way out (unless you set a timeout, which is not always possible). Native call case is even worse.
That said, I'm okay with rewriting my own code, but sometimes all we have is a closed-source assembly, and there is absolutely nothing that can be done in this case.

 

MD5 Hashing isn't working on 4.6.3 and is breaking my highscores and shop. Bug reported with project repro, but here is the code just incase anyone can see the reason it's not working with IL2CPP. (this is stopping me ship my game)

(Case 674680)

Code (CSharp):

1.   public string Md5Sum(string input)
2.   {
3.     System.Security.Cryptography.MD5 md5 = System.Security.Cryptography.MD5.Create();
4.     byte[] inputBytes = System.Text.Encoding.ASCII.GetBytes(input);
5.     byte[] hash = md5.ComputeHash(inputBytes);
6.  
7.     StringBuilder sb = new StringBuilder();
8.     for(int i = 0; i < hash.Length; i++)
9.     {
10.       sb.Append(hash[i].ToString("X2"));
11.     }
12.     return sb.ToString();
13.   }
14.  

 

Hi Josh, I've just replied to the email.

 

Unsure on both. I only have Android 4.2 and cycles to try and push thru this il2cpp conversion.
The everyplay QA folks can prolly regress faster than I can
For the moment I used compiler directives around everyplay on Android. Android version can simply not have the feature. It isn't a blocking issue. I wanted you guys to be aware of it.
But I would like to have it at some point. Everyplay seems to suggest android 4.2 is enough to show it.

thanks

 

Thanks for the feedback Josh Peterson, and sorry for the inconvenience again, really hoping to test our apps soon !

 

@Lucas Meijer
Done. Case 674720

Thanks for the help.

 

Sorry if this has been asked already (this thread is getting quite big) but is there any word on when 4.6.3p1 will be released? As of now, I have a finished project that I can't release do to IL2CPP issues. These issues do not exist when I build the game for 32bit but unfortunately Apple is now requiring developers to support 64 bit.

Also, as a side note, it would be nice to have an option within the player settings to unlock the frame rate (or at least bump it up to 60fps on newer devices instead of the currently locked 30fps).

 

I don't have any insider knowledge about 4.6.3p1 specifically, but the patch releases often come out weekly on Thursday or Friday.

 

Hi,

Today we found out that loading embedded resources from a dll is not supported by IL2CPP. Do you plan to support it?

I've also created a bug report just in case: 674763.

Thanks!

 

Tried the 4.6.3 release this morning and I'm happy to say I don't get the "Monitor" crash during the splash screen (670035)!

This means that I can actually get into the game and find more bugs though

Trying to get the current process' information throws a SystemException

Code (CSharp):

1. System.Diagnostics.Process.GetCurrentProcess()

This works fine with Mono, but throws a SystemException with Il2Cpp. This is used by MongoDb's ObjectId class to generate a random hash for each machine that will be generating Ids. We already had a workaround on our side for this since it didn't work in WebPlayer anyway, but the exception that is thrown is different (SystemException vs MethodAccessException). Case #674706.

Padding exception when decrypting with AesManaged/CryptoStream
This also works fine with Mono and fails with Il2Cpp. It might have already been mentioned earlier in the thread but since there's so many different use-cases with encryption I decided to report it again. Case #674773.
Also, regarding encryption, it seems that you are using BouncyCastle behind the scenes to do the work. Is there any way to use it from inside Unity too or is this limited only to the Il2Cpp backend? (I'm asking because we also use BouncyCastle on our side, this might prevent having the library linked twice).

 

It's worth noting that there are some endian differences between Java and .NET. I've seen this many times with RSA Encryption (I actually wrote an RSA Encryption library for Silverlight and Windows Phone 7 a few years ago). You would run into issues with the data not decrypting properly. An example to fix it was that if you were encrypting with Bouncy Castle on the Java side, once you got your binary data in .NET it was necessary to reverse the byte array before decrypting. You could see a similar issue with other encryption / hashing schemes as well.

 

This is strictly client side (C#) and works fine in Mono, but thanks for the suggestion!

 

So what shall I do about my game that continues to crash on both 32 and 64 bit devices?

 

@catacomber: You need to look at the debug output in Xcode so that we can figure out why it's crashing. Run the build on your device in Xcode, and then go to the 'Debug navigator' tab. Then do whatever it is you need to do to make the game crash and then look at the most recent messages in the debug output. If those don't give you any clues then post them here.

 

i can confirm that 4.6.3 ThreadPool is working, (and performance is nice !) but seems like there is something wrong binary formatter

 

Guys has anyone else noticed their iPads running a lot hotter when using Metal? Mine is noticeably hotter.

 

Ok thanks. Will try that. Thanks for taking the time to do that pic.

 

Hi, there.

I found a problem about IL2CPP(Unity ver.4.6.3f1).

this code (in Unity Editor).

Code (CSharp):

1. public static T Instance
2.     {
3.         get
4.         {
5.             if(mInstance == null)
6.                 mInstance = new T();
7.            
8.             return mInstance;
9.         }
10.     }

Changed that code (in Xcode).

Code (C++):

1. extern MethodInfo Singleton_1_get_Instance_m37704_MethodInfo;
2. // T Singleton`1<System.Object>::get_Instance()
3.  
4. externMethodInfo Singleton_1_get_Instance_m37704_MethodInfo;
5.  
6. Object_t * Singleton_1_get_Instance_m37704_gshared (Object_t * __this/* static, unused */, MethodInfo* method)
7.  
8. {
9.  
10. Object_t * V_0 = {0};
11.  
12. Object_t * G_B4_0 = {0};
13.  
14. {
15.  
16. Object_t * L_0 = (((Singleton_1_t7660_StaticFields*)IL2CPP_RGCTX_DATA(method->declaring_type->rgctx_data, 0)->static_fields)->___mInstance);
17.  
18. if (((Object_t *)L_0))
19.  
20. {
21.  
22. goto IL_0037;
23.  
24. }
25.  
26. }
27.  
28. {
29.  
30. Initobj (&Object_t_il2cpp_TypeInfo, (&V_0));
31.  
32. Object_t * L_1 = V_0;
33.  
34. if (!((Object_t *)L_1))
35.  
36. {
37.  
38. goto IL_002d;
39.  
40. }
41.  
42. }
43.  
44. {
45.  
46. Initobj (&Object_t_il2cpp_TypeInfo, (&V_0));
47.  
48. G_B4_0 = V_0;
49.  
50. goto IL_0032;
51.  
52. }
53.  
54.  
55. IL_002d:
56.  
57. {
58.  
59. Object_t * L_2 = (( Object_t * (*) (Object_t * __this/* static, unused */, MethodInfo* method))IL2CPP_RGCTX_METHOD_INFO(method->declaring_type->rgctx_data, 2)->method)(NULL/*static, unused*/, /*hidden argument*/IL2CPP_RGCTX_METHOD_INFO(method->declaring_type->rgctx_data, 2));
60.  
61. G_B4_0 = L_2;
62.  
63. }
64.  
65.  
66. IL_0032:
67.  
68. {
69.  
70. ((Singleton_1_t7660_StaticFields*)IL2CPP_RGCTX_DATA(method->declaring_type->rgctx_data, 0)->static_fields)->___mInstance = G_B4_0;
71.  
72. }
73.  
74.  
75. IL_0037:
76.  
77. {
78.  
79. return (((Singleton_1_t7660_StaticFields*)IL2CPP_RGCTX_DATA(method->declaring_type->rgctx_data, 0)->static_fields)->___mInstance);
80.  
81. }
82.  
83. }
84.  

there is a problem.

Code (C++):

1. Object_t * L_0 = (((Singleton_1_t7660_StaticFields*)IL2CPP_RGCTX_DATA(method->declaring_type->rgctx_data, 0)->static_fields)->___mInstance);
2.         if (((Object_t *)L_0))
3.         {
4.             goto IL_0037;
5.         }

there is not checked 'NULL Exception'.

Code (C++):

1.  (Singleton_1_t7660_StaticFields*)IL2CPP_RGCTX_DATA(method->declaring_type->rgctx_data, 0)

this code has to change like that.

Code (C++):

1.  
2. if((Singleton_1_t7660_StaticFields*)IL2CPP_RGCTX_DATA(method->declaring_type->rgctx_data, 0) == NULL || (Singleton_1_t7660_StaticFields*)IL2CPP_RGCTX_DATA(method->declaring_type->rgctx_data, 0)->static_fields == NULL)
3.  
4.         {
5.  
6.             goto IL_002d;
7.  
8.         }
9.  
10.    
11.  
12. Object_t * L_0 = (((Singleton_1_t7660_StaticFields*)IL2CPP_RGCTX_DATA(method->declaring_type->rgctx_data, 0)->static_fields)->___mInstance);
13.  
14. if (((Object_t *)L_0))
15.  
16. {
17.  
18. goto IL_0037;
19.  
20. }
21.  

I wanna be updated that this problem solved ASAP.

 

Hello Joncham and Lucas, so this issue has not been fixed in Unity 4.6.3, correct? If so, has this been issue solved in future versions of Unity?

 

Does this not work for you:

Code (csharp):

1. Application.targetFrameRate = 60;

I've been using this for a long time - I get 60fps using this.

 

Sorry for going somewhat off topic - but does this suggest that IL2CPP is available for Android now? (I sure wish).

 

@JJC1138--trying to get to the point where I can play my game while attached to my computer while in Xcode so I can post a crash report in Debug View. I built it for archiving and adhoc distribution so had to rebuild it to run on my device.

The rest of this does not apply to you @JJC1138--thanks for all your help. : )

I really do appreciate that you (Unity other guys) responded to my post and tried to help me but I don't think posting stack traces is going to help me. If I'm wrong, please let me know.

Right now, having had to rebuild to run on my device, using il2cpp and armv7, I can't even get to playing on my device without rebooting so far.

I cannot build my game using mono and armv7---armv7 is greyed out. So please don't ask me how things run on mono. Am sick of that. I used to be fine building on mono. I think at this point we all know mono was and probably is ok---but for iOS you now can't use mono to build to submit to the asset store. My game used to run fine on mono as I've stated and restated many times.

I'm trying to reproduce the problems my testers are having on their devices by running the game on my iPad mini 1st gen while in Xcode. I currently keep having the game fold and getting the message lost connection reattach----i reattach and still fold. Only if I reboot my iPad mini can I keep the connection.

This is troubling because I know from past experiences that if this happens for me it will happen for my players--they'll have to reboot to be able even to run the game.

I don't get any messages in Xcode except to reattach to process.

Trying to play the game after reattaching and rebooting now to see what happens if it crashes.

But I re-ask my question that was never answered---would I be better off trying to build my game using the Unity 5 beta. Someone told me all focus is on that now and 4.6 is getting little attention. Seeing all the attention here to 4.6, I'm not sure they are correct but would like to know if 5.0 beta is more stable if you are trying to run a large game on a 32 bit device on ios.

Also have you, Unity, tried to run large---424 mb apps-- on il2cpp on a 32 bit device?

Does it run smoothly for you?

Although I don't mind suffering through trying to do this--to get my game running smoothly--keep in mind I'm paying $150 a month to have the ability to build a game on iOS that runs smoothly--that is $75 of that is for iOS but all I care about is iOS--that's my market-- so that amount is really for the privilege of building games that run smoothly on ios.

Not complaining about that right now but complaining that stack traces are going to help me when the real problem seems to be the bloated file size.

Still trying to get things running without problems -- will play the game after rebooting and see what happens and post any Xcode messages I get using the Debug viewer.

I think the fact that my testers find the game unstable on both their 32 and 64 bit devices is a flag that something isn't right with large files.

I never had this problem before il2cpp. I let my partners know we may not be able to release this game until April or May while you work your magic. They are hoping you will work that magic soon.

I'd really like to go back to making my game better and finding my own bugs rather than making it just run. And I love Unity. I have confidence things will be better. It's just right now a big time sponge for me.

I don't think my problem is one that posting stack traces is going to fix.

I think it's because il2cpp bloats the file size. And a 32 bit device just can't handle it.

The only time I've had this type of problem is if I've tried to play big, big files in the app store. They run fine on 64 bit devices but you have to do just what I've been doing--rebooting--to get them even to start on a 32 bit device.

So I don't think you can help me unless you try to run my full game built with il2cpp.

And to do that I have to have a way to get you my full game--not via the bug uploader because it won't take the size.

I know this was dumped on you by Apple requirements and maybe il2cpp down the road will be good for all of us but right now I just want to make my game as good as it can be and worry about my bugs---

I'm getting fretful but I think it's with good reason.

Can't you try to run a big file of your own (one that would have been 120 mb built with mono but is 424 MB or so built with il2cpp) with il2cpp on a 32 bit device? Not to mention on a 64 bit device? And see how it performs?

I really think that's the problem here and the bottom line.

Anything you ask me to do---other than being able to upload my app to you as a full app and not cut down---is imho not going to get me anywhere. And again--don't you have some app of your own somewhere that is a big file --rpg file size-- you can test?

Somehow I've lost my confidence that you really hear what my problem is. It's not about coding problems or bugs or stack traces or anything--

EXCEPT--the gargantuan, bloated file size trying to run on a 32 bit machine.

If you can't handle this right now--OK--I and my testers have to wait until you can.

Just don't ask me to put up stack traces unless you can tell me how that will help me show you that the problem is the large file size.