Test failure in TestBundleIntegerArrayList2 when the shared runtime is enabled #4596
Assignees
Labels
Milestone
Comments
|
I've got a better reproduction for this and it's also affecting d16-6. You can hit this by installing and running https://github.com/xamarin/monodroid-samples/tree/master/android5.0/DrawableTinting on device in Debug mode. The crash happens here - https://github.com/xamarin/monodroid-samples/blob/master/android5.0/DrawableTinting/DrawableTinting/DrawableTintingFragment.cs#L118 I've linked logcat output for the latest d16-6 results runtime test failure and the DrawableTinting sample crash. |
grendello
added a commit
to grendello/xamarin-android
that referenced
this issue
Apr 30, 2020
Fixes: dotnet#4596 Context: 7117414 7117414 made string lookup return into our type map support, for Debug builds only, and with them the old problem of duplicate type names returned as well - it is possible that a single Java type name will map to a number of managed types and vice versa. This is not a problem in most cases, but whenever a Java type name maps to a generic managed type (for instance "Android.Runtime.JavaList`1, Mono.Android"), the code added in 7117414 runs into trouble: System.MemberAccessException : Cannot create an instance of Android.Runtime.JavaList`1[T] because Type.ContainsGenericParameters is true. at System.Reflection.RuntimeConstructorInfo.DoInvoke (System.Object obj, System.Reflection.BindingFlags invokeAttr, System.Reflection.Binder binder, System.Object[] parameters, System.Globalization.CultureInfo culture) at System.Reflection.RuntimeConstructorInfo.Invoke (System.Reflection.BindingFlags invokeAttr, System.Reflection.Binder binder, System.Object[] parameters, System.Globalization.CultureInfo culture) at System.Reflection.ConstructorInfo.Invoke (System.Object[] parameters) at Java.Interop.TypeManager.CreateProxy (System.Type type, System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Java.Interop.TypeManager.CreateInstance (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer, System.Type targetType) at Java.Lang.Object.GetObject (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer, System.Type type) at Java.Lang.Object._GetObject[T] (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Java.Lang.Object.GetObject[T] (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Android.OS.Bundle.Get (System.String key) at Xamarin.Android.RuntimeTests.BundleTest.TestBundleIntegerArrayList2 () at (wrapper managed-to-native) System.Reflection.RuntimeMethodInfo.InternalInvoke(System.Reflection.RuntimeMethodInfo,object,object[],System.Exception&) at System.Reflection.RuntimeMethodInfo.Invoke (System.Object obj, System.Reflection.BindingFlags invokeAttr, System.Reflection.Binder binder, System.Object[] parameters, System.Globalization.CultureInfo culture) The reason this happens is due to the way the native code processes the managed type name it obtained from the `Java -> managed` map. Namely, it uses Mono embedding API to return a `MonoType` instance corresponding to the managed type name, then that `MonoType` instance is converted into `MonoReflectionType*` which is returned to the managed code (it corresponds to the `System.Type` instance there) and then an attempt to instantiate this type is made in the usual way. It works for non-generic types, but whenever a generic type is processed in this way it will result in us returning a `MonoReflectionType*` corresponding to an **open** generic type which then results in the exception shown above. Given just the type name, the native runtime is unable to "close" the generic type without information regarding the types of the generic parameters needed to do it. The solution, a workaround really, for now is to look at the type name and return `null` to the managed code whenever we see that the type name contains the "`" character, marking it as a generic type. The managed code will then proceed to look up the type and instantiate it in the correct way. In the future we need to find a better solution for this issue, but doing so would require quite extensive changes to our current code. The solution implemented here is sufficient for now.
jonpryor
added a commit
that referenced
this issue
May 1, 2020
Fixes: #4596 Context: xamarin/monodroid@192b1f1 Context: https://github.com/xamarin/java.interop/blob/fc18c54b2ccf14f444fadac0a1e7e81f837cc470/src/Java.Interop.Tools.JavaCallableWrappers/Java.Interop.Tools.JavaCallableWrappers/TypeNameMapGenerator.cs#L149-L186 In some environments, `BundleTest.TestBundleIntegerArrayList2()` fails when using the commercial shared runtime: Test 'Xamarin.Android.RuntimeTests.BundleTest.TestBundleIntegerArrayList2' failed: System.MemberAccessException : Cannot create an instance of Android.Runtime.JavaList`1[T] because Type.ContainsGenericParameters is true. at System.Reflection.RuntimeConstructorInfo.DoInvoke (System.Object obj, System.Reflection.BindingFlags invokeAttr, System.Reflection.Binder binder, System.Object[] parameters, System.Globalization.CultureInfo culture) at System.Reflection.RuntimeConstructorInfo.Invoke (System.Reflection.BindingFlags invokeAttr, System.Reflection.Binder binder, System.Object[] parameters, System.Globalization.CultureInfo culture) at System.Reflection.ConstructorInfo.Invoke (System.Object[] parameters) at Java.Interop.TypeManager.CreateProxy (System.Type type, System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Java.Interop.TypeManager.CreateInstance (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer, System.Type targetType) at Java.Lang.Object.GetObject (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer, System.Type type) at Java.Lang.Object._GetObject[T] (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Java.Lang.Object.GetObject[T] (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Android.OS.Bundle.Get (System.String key) at Xamarin.Android.RuntimeTests.BundleTest.TestBundleIntegerArrayList2 () at (wrapper managed-to-native) System.Reflection.RuntimeMethodInfo.InternalInvoke(System.Reflection.RuntimeMethodInfo,object,object[],System.Exception&) at System.Reflection.RuntimeMethodInfo.Invoke (System.Object obj, System.Reflection.BindingFlags invokeAttr, System.Reflection.Binder binder, System.Object[] parameters, System.Globalization.CultureInfo culture) The problem appears to be rooted in ce2bc68. In a pre-ce2bc689 world, there were two separate sets of "typemap" files that the application could use: * Java-to-managed mappings, in `typemap.jm` * Managed-to-Java mappings, in `typemap.mj`. These files did *not* need to have the same number of entries! % unzip /Library/Frameworks/Xamarin.Android.framework/Versions/10.2.0.100/lib/xamarin.android/xbuild/Xamarin/Android/Mono.Android.DebugRuntime-debug.apk typemap.jm % unzip /Library/Frameworks/Xamarin.Android.framework/Versions/10.2.0.100/lib/xamarin.android/xbuild/Xamarin/Android/Mono.Android.DebugRuntime-debug.apk typemap.mj % strings typemap.jm | wc -l 10318 % strings typemap.mj | wc -l 11956 The reason why they have a different number of entries is *aliasing*: there can be *multiple* bindings for a given Java type. The managed-to-Java mappings can contain *all* of them; the Java-to-managed mapping *must* pick Only One. For example, [`java.util.ArrayList`][0] contains (at least?) *three* bindings: * [`Android.Runtime.JavaList`][1] * [`Android.Runtime.JavaList<T>`][2] * `Java.Util.ArrayList` (generated at build time) In a pre-ce2bc689 world, this was "fine": we'd binary search each file *separately* to find type mappings. This changed in ce2bc68, as the typemap information was merged into a *single* array in order to de-duplicate information. This reduced `.apk` size. An unanticipated result of this is that the Java-to-managed mappings can now contain *duplicate* keys, "as if" the original `typemap.jm` had the contents: java/util/ArrayList Android.Runtime.JavaList, Mono.Android java/util/ArrayList Android.Runtime.JavaList`1, Mono.Android java/util/ArrayList Java.Util.ArrayLlist, Mono.Android Whereas pre-ce2bc689 there would have only been the first entry. "Sometimes" this duplication is "fine": the typemap search "Just happens" to return the first entry, or the 3rd entry, and apps continue to work as intended. "Sometimes" it isn't: the typemap binary search finds the 2nd entry, which is a generic type. This results in attempting to instantiate a generic type *without appropriate type parameters*, which results in the aforementioned `MemberAccessException`. Update `EmbeddedAssemblies::binary_search()` so that instead of returning the "first" entry that the binary search happens to land on, probe *previous* entries in the type mappings to see if they *also* contain the same key. `binary_search()` can instead return the *first entry* within the mapping, not just the first it happens to land upon while binary searching. This should allow for more *consistent* behaviors, ensuring that Java-to-managed typemap lookup obtains a *non*-generic type, thus avoiding the `MemberAccessException`. Additionally, update `TestBundleIntegerArrayList2()` so that it asserts the runtime *type* of `obj` returned, asserting that it is in fact a `JavaList` and not e.g. `Java.Util.ArrayList` or something. (The linker could otherwise "change" things, and this assertion will ensure that `JavaList` won't be linked away…) [0]: https://developer.android.com/reference/java/util/ArrayList [1]: https://github.com/xamarin/xamarin-android/blob/61f09bf2ef0c8f09550e2d77eb97f417432b315b/src/Mono.Android/Android.Runtime/JavaList.cs#L11-L13 [2]: https://github.com/xamarin/xamarin-android/blob/61f09bf2ef0c8f09550e2d77eb97f417432b315b/src/Mono.Android/Android.Runtime/JavaList.cs#L667-L668
jonpryor
added a commit
that referenced
this issue
May 1, 2020
Fixes: #4596 Context: xamarin/monodroid@192b1f1 Context: https://github.com/xamarin/java.interop/blob/fc18c54b2ccf14f444fadac0a1e7e81f837cc470/src/Java.Interop.Tools.JavaCallableWrappers/Java.Interop.Tools.JavaCallableWrappers/TypeNameMapGenerator.cs#L149-L186 In some environments, `BundleTest.TestBundleIntegerArrayList2()` fails when using the commercial shared runtime: Test 'Xamarin.Android.RuntimeTests.BundleTest.TestBundleIntegerArrayList2' failed: System.MemberAccessException : Cannot create an instance of Android.Runtime.JavaList`1[T] because Type.ContainsGenericParameters is true. at System.Reflection.RuntimeConstructorInfo.DoInvoke (System.Object obj, System.Reflection.BindingFlags invokeAttr, System.Reflection.Binder binder, System.Object[] parameters, System.Globalization.CultureInfo culture) at System.Reflection.RuntimeConstructorInfo.Invoke (System.Reflection.BindingFlags invokeAttr, System.Reflection.Binder binder, System.Object[] parameters, System.Globalization.CultureInfo culture) at System.Reflection.ConstructorInfo.Invoke (System.Object[] parameters) at Java.Interop.TypeManager.CreateProxy (System.Type type, System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Java.Interop.TypeManager.CreateInstance (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer, System.Type targetType) at Java.Lang.Object.GetObject (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer, System.Type type) at Java.Lang.Object._GetObject[T] (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Java.Lang.Object.GetObject[T] (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Android.OS.Bundle.Get (System.String key) at Xamarin.Android.RuntimeTests.BundleTest.TestBundleIntegerArrayList2 () at (wrapper managed-to-native) System.Reflection.RuntimeMethodInfo.InternalInvoke(System.Reflection.RuntimeMethodInfo,object,object[],System.Exception&) at System.Reflection.RuntimeMethodInfo.Invoke (System.Object obj, System.Reflection.BindingFlags invokeAttr, System.Reflection.Binder binder, System.Object[] parameters, System.Globalization.CultureInfo culture) The problem appears to be rooted in ce2bc68. In a pre-ce2bc689 world, there were two separate sets of "typemap" files that the application could use: * Java-to-managed mappings, in `typemap.jm` * Managed-to-Java mappings, in `typemap.mj`. These files did *not* need to have the same number of entries! % unzip /Library/Frameworks/Xamarin.Android.framework/Versions/10.2.0.100/lib/xamarin.android/xbuild/Xamarin/Android/Mono.Android.DebugRuntime-debug.apk typemap.jm % unzip /Library/Frameworks/Xamarin.Android.framework/Versions/10.2.0.100/lib/xamarin.android/xbuild/Xamarin/Android/Mono.Android.DebugRuntime-debug.apk typemap.mj % strings typemap.jm | wc -l 10318 % strings typemap.mj | wc -l 11956 The reason why they have a different number of entries is *aliasing*: there can be *multiple* bindings for a given Java type. The managed-to-Java mappings can contain *all* of them; the Java-to-managed mapping *must* pick Only One. For example, [`java.util.ArrayList`][0] contains (at least?) *three* bindings: * [`Android.Runtime.JavaList`][1] * [`Android.Runtime.JavaList<T>`][2] * `Java.Util.ArrayList` (generated at build time) In a pre-ce2bc689 world, this was "fine": we'd binary search each file *separately* to find type mappings. This changed in ce2bc68, as the typemap information was merged into a *single* array in order to de-duplicate information. This reduced `.apk` size. An unanticipated result of this is that the Java-to-managed mappings can now contain *duplicate* keys, "as if" the original `typemap.jm` had the contents: java/util/ArrayList Android.Runtime.JavaList, Mono.Android java/util/ArrayList Android.Runtime.JavaList`1, Mono.Android java/util/ArrayList Java.Util.ArrayLlist, Mono.Android Whereas pre-ce2bc689 there would have only been the first entry. "Sometimes" this duplication is "fine": the typemap search "Just happens" to return the first entry, or the 3rd entry, and apps continue to work as intended. "Sometimes" it isn't: the typemap binary search finds the 2nd entry, which is a generic type. This results in attempting to instantiate a generic type *without appropriate type parameters*, which results in the aforementioned `MemberAccessException`. Update `EmbeddedAssemblies::binary_search()` so that instead of returning the "first" entry that the binary search happens to land on, probe *previous* entries in the type mappings to see if they *also* contain the same key. `binary_search()` can instead return the *first entry* within the mapping, not just the first it happens to land upon while binary searching. This should allow for more *consistent* behaviors, ensuring that Java-to-managed typemap lookup obtains a *non*-generic type, thus avoiding the `MemberAccessException`. Additionally, update `TestBundleIntegerArrayList2()` so that it asserts the runtime *type* of `obj` returned, asserting that it is in fact a `JavaList` and not e.g. `Java.Util.ArrayList` or something. (The linker could otherwise "change" things, and this assertion will ensure that `JavaList` won't be linked away…) [0]: https://developer.android.com/reference/java/util/ArrayList [1]: https://github.com/xamarin/xamarin-android/blob/61f09bf2ef0c8f09550e2d77eb97f417432b315b/src/Mono.Android/Android.Runtime/JavaList.cs#L11-L13 [2]: https://github.com/xamarin/xamarin-android/blob/61f09bf2ef0c8f09550e2d77eb97f417432b315b/src/Mono.Android/Android.Runtime/JavaList.cs#L667-L668
Partly just for my own future reference, I'll record that setting |
|
The |
grendello
added a commit
to grendello/xamarin-android
that referenced
this issue
May 6, 2020
Fixes: dotnet#4596 Context: 7117414 Do the heavy lifting on the build time instead of the run time. Prefer base types and always map all the duplicate entries to the **first** entry encountered.
grendello
added a commit
to grendello/xamarin-android
that referenced
this issue
May 6, 2020
Fixes: dotnet#4596 Context: xamarin/monodroid@192b1f1 Context: https://github.com/xamarin/java.interop/blob/fc18c54b2ccf14f444fadac0a1e7e81f837cc470/src/Java.Interop.Tools.JavaCallableWrappers/Java.Interop.Tools.JavaCallableWrappers/TypeNameMapGenerator.cs#L149-L186 In some environments, `BundleTest.TestBundleIntegerArrayList2()` fails when using the commercial shared runtime: Test 'Xamarin.Android.RuntimeTests.BundleTest.TestBundleIntegerArrayList2' failed: System.MemberAccessException : Cannot create an instance of Android.Runtime.JavaList`1[T] because Type.ContainsGenericParameters is true. at System.Reflection.RuntimeConstructorInfo.DoInvoke (System.Object obj, System.Reflection.BindingFlags invokeAttr, System.Reflection.Binder binder, System.Object[] parameters, System.Globalization.CultureInfo culture) at System.Reflection.RuntimeConstructorInfo.Invoke (System.Reflection.BindingFlags invokeAttr, System.Reflection.Binder binder, System.Object[] parameters, System.Globalization.CultureInfo culture) at System.Reflection.ConstructorInfo.Invoke (System.Object[] parameters) at Java.Interop.TypeManager.CreateProxy (System.Type type, System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Java.Interop.TypeManager.CreateInstance (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer, System.Type targetType) at Java.Lang.Object.GetObject (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer, System.Type type) at Java.Lang.Object._GetObject[T] (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Java.Lang.Object.GetObject[T] (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Android.OS.Bundle.Get (System.String key) at Xamarin.Android.RuntimeTests.BundleTest.TestBundleIntegerArrayList2 () at (wrapper managed-to-native) System.Reflection.RuntimeMethodInfo.InternalInvoke(System.Reflection.RuntimeMethodInfo,object,object[],System.Exception&) at System.Reflection.RuntimeMethodInfo.Invoke (System.Object obj, System.Reflection.BindingFlags invokeAttr, System.Reflection.Binder binder, System.Object[] parameters, System.Globalization.CultureInfo culture) The problem appears to be rooted in ce2bc68. In a pre-ce2bc689 world, there were two separate sets of "typemap" files that the application could use: * Java-to-managed mappings, in `typemap.jm` * Managed-to-Java mappings, in `typemap.mj`. These files did *not* need to have the same number of entries! % unzip /Library/Frameworks/Xamarin.Android.framework/Versions/10.2.0.100/lib/xamarin.android/xbuild/Xamarin/Android/Mono.Android.DebugRuntime-debug.apk typemap.jm % unzip /Library/Frameworks/Xamarin.Android.framework/Versions/10.2.0.100/lib/xamarin.android/xbuild/Xamarin/Android/Mono.Android.DebugRuntime-debug.apk typemap.mj % strings typemap.jm | wc -l 10318 % strings typemap.mj | wc -l 11956 The reason why they have a different number of entries is *aliasing*: there can be *multiple* bindings for a given Java type. The managed-to-Java mappings can contain *all* of them; the Java-to-managed mapping *must* pick Only One. For example, [`java.util.ArrayList`][0] contains (at least?) *three* bindings: * [`Android.Runtime.JavaList`][1] * [`Android.Runtime.JavaList<T>`][2] * `Java.Util.ArrayList` (generated at build time) In a pre-ce2bc689 world, this was "fine": we'd binary search each file *separately* to find type mappings. This changed in ce2bc68, as the typemap information was merged into a *single* array in order to de-duplicate information. This reduced `.apk` size. An unanticipated result of this is that the Java-to-managed mappings can now contain *duplicate* keys, "as if" the original `typemap.jm` had the contents: java/util/ArrayList Android.Runtime.JavaList, Mono.Android java/util/ArrayList Android.Runtime.JavaList`1, Mono.Android java/util/ArrayList Java.Util.ArrayLlist, Mono.Android Whereas pre-ce2bc689 there would have only been the first entry. "Sometimes" this duplication is "fine": the typemap search "Just happens" to return the first entry, or the 3rd entry, and apps continue to work as intended. "Sometimes" it isn't: the typemap binary search finds the 2nd entry, which is a generic type. This results in attempting to instantiate a generic type *without appropriate type parameters*, which results in the aforementioned `MemberAccessException`. Update typemap generation code in `Xamarin.Android.Build.Tasks` so that all the duplicate Java type names will point to the same managed type name. Additionally, make sure we select the managed type in the same fashion the old typemap generator in `Java.Interop` worked - prefer base types to the derived ones if the type is an interface or an abstract class. The effect of this change is that no matter which entry `binary_search` ends up selecting it will always return the same managed type name for all the aliased Java types. Additionally, update `TestBundleIntegerArrayList2()` so that it asserts the runtime *type* of `obj` returned, asserting that it is in fact a `JavaList` and not e.g. `Java.Util.ArrayList` or something. (The linker could otherwise "change" things, and this assertion will ensure that `JavaList` won't be linked away…) [0]: https://developer.android.com/reference/java/util/ArrayList [1]: https://github.com/xamarin/xamarin-android/blob/61f09bf2ef0c8f09550e2d77eb97f417432b315b/src/Mono.Android/Android.Runtime/JavaList.cs#L11-L13 [2]: https://github.com/xamarin/xamarin-android/blob/61f09bf2ef0c8f09550e2d77eb97f417432b315b/src/Mono.Android/Android.Runtime/JavaList.cs#L667-L668 Do the heavy lifting on the build time instead of the run time. Prefer base types and always map all the duplicate entries to the **first** entry encountered.
grendello
added a commit
to grendello/xamarin-android
that referenced
this issue
May 6, 2020
Fixes: dotnet#4596 Context: xamarin/monodroid@192b1f1 Context: https://github.com/xamarin/java.interop/blob/fc18c54b2ccf14f444fadac0a1e7e81f837cc470/src/Java.Interop.Tools.JavaCallableWrappers/Java.Interop.Tools.JavaCallableWrappers/TypeNameMapGenerator.cs#L149-L186 In some environments, `BundleTest.TestBundleIntegerArrayList2()` fails when using the commercial shared runtime: Test 'Xamarin.Android.RuntimeTests.BundleTest.TestBundleIntegerArrayList2' failed: System.MemberAccessException : Cannot create an instance of Android.Runtime.JavaList`1[T] because Type.ContainsGenericParameters is true. at System.Reflection.RuntimeConstructorInfo.DoInvoke (System.Object obj, System.Reflection.BindingFlags invokeAttr, System.Reflection.Binder binder, System.Object[] parameters, System.Globalization.CultureInfo culture) at System.Reflection.RuntimeConstructorInfo.Invoke (System.Reflection.BindingFlags invokeAttr, System.Reflection.Binder binder, System.Object[] parameters, System.Globalization.CultureInfo culture) at System.Reflection.ConstructorInfo.Invoke (System.Object[] parameters) at Java.Interop.TypeManager.CreateProxy (System.Type type, System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Java.Interop.TypeManager.CreateInstance (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer, System.Type targetType) at Java.Lang.Object.GetObject (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer, System.Type type) at Java.Lang.Object._GetObject[T] (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Java.Lang.Object.GetObject[T] (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Android.OS.Bundle.Get (System.String key) at Xamarin.Android.RuntimeTests.BundleTest.TestBundleIntegerArrayList2 () at (wrapper managed-to-native) System.Reflection.RuntimeMethodInfo.InternalInvoke(System.Reflection.RuntimeMethodInfo,object,object[],System.Exception&) at System.Reflection.RuntimeMethodInfo.Invoke (System.Object obj, System.Reflection.BindingFlags invokeAttr, System.Reflection.Binder binder, System.Object[] parameters, System.Globalization.CultureInfo culture) The problem appears to be rooted in ce2bc68. In a pre-ce2bc689 world, there were two separate sets of "typemap" files that the application could use: * Java-to-managed mappings, in `typemap.jm` * Managed-to-Java mappings, in `typemap.mj`. These files did *not* need to have the same number of entries! % unzip /Library/Frameworks/Xamarin.Android.framework/Versions/10.2.0.100/lib/xamarin.android/xbuild/Xamarin/Android/Mono.Android.DebugRuntime-debug.apk typemap.jm % unzip /Library/Frameworks/Xamarin.Android.framework/Versions/10.2.0.100/lib/xamarin.android/xbuild/Xamarin/Android/Mono.Android.DebugRuntime-debug.apk typemap.mj % strings typemap.jm | wc -l 10318 % strings typemap.mj | wc -l 11956 The reason why they have a different number of entries is *aliasing*: there can be *multiple* bindings for a given Java type. The managed-to-Java mappings can contain *all* of them; the Java-to-managed mapping *must* pick Only One. For example, [`java.util.ArrayList`][0] contains (at least?) *three* bindings: * [`Android.Runtime.JavaList`][1] * [`Android.Runtime.JavaList<T>`][2] * `Java.Util.ArrayList` (generated at build time) In a pre-ce2bc689 world, this was "fine": we'd binary search each file *separately* to find type mappings. This changed in ce2bc68, as the typemap information was merged into a *single* array in order to de-duplicate information. This reduced `.apk` size. An unanticipated result of this is that the Java-to-managed mappings can now contain *duplicate* keys, "as if" the original `typemap.jm` had the contents: java/util/ArrayList Android.Runtime.JavaList, Mono.Android java/util/ArrayList Android.Runtime.JavaList`1, Mono.Android java/util/ArrayList Java.Util.ArrayLlist, Mono.Android Whereas pre-ce2bc689 there would have only been the first entry. "Sometimes" this duplication is "fine": the typemap search "Just happens" to return the first entry, or the 3rd entry, and apps continue to work as intended. "Sometimes" it isn't: the typemap binary search finds the 2nd entry, which is a generic type. This results in attempting to instantiate a generic type *without appropriate type parameters*, which results in the aforementioned `MemberAccessException`. Update typemap generation code in `Xamarin.Android.Build.Tasks` so that all the duplicate Java type names will point to the same managed type name. Additionally, make sure we select the managed type in the same fashion the old typemap generator in `Java.Interop` worked - prefer base types to the derived ones if the type is an interface or an abstract class. The effect of this change is that no matter which entry `binary_search` ends up selecting it will always return the same managed type name for all the aliased Java types. Additionally, update `TestBundleIntegerArrayList2()` so that it asserts the runtime *type* of `obj` returned, asserting that it is in fact a `JavaList` and not e.g. `Java.Util.ArrayList` or something. (The linker could otherwise "change" things, and this assertion will ensure that `JavaList` won't be linked away…) [0]: https://developer.android.com/reference/java/util/ArrayList [1]: https://github.com/xamarin/xamarin-android/blob/61f09bf2ef0c8f09550e2d77eb97f417432b315b/src/Mono.Android/Android.Runtime/JavaList.cs#L11-L13 [2]: https://github.com/xamarin/xamarin-android/blob/61f09bf2ef0c8f09550e2d77eb97f417432b315b/src/Mono.Android/Android.Runtime/JavaList.cs#L667-L668
grendello
added a commit
to grendello/xamarin-android
that referenced
this issue
May 7, 2020
Fixes: dotnet#4596 Fixes: dotnet#4660 Context: xamarin/monodroid@192b1f1 Context: https://github.com/xamarin/java.interop/blob/fc18c54b2ccf14f444fadac0a1e7e81f837cc470/src/Java.Interop.Tools.JavaCallableWrappers/Java.Interop.Tools.JavaCallableWrappers/TypeNameMapGenerator.cs#L149-L186 Context: https://github.com/xamarin/java.interop/blob/010161d1ef6625b762429334f02e7b15e1f7caae/src/Java.Interop.Tools.JavaCallableWrappers/Java.Interop.Tools.JavaCallableWrappers/TypeNameMapGenerator.cs#L155 ~~ Debug Mode ~~ In some environments, `BundleTest.TestBundleIntegerArrayList2()` fails when using the commercial shared runtime: Test 'Xamarin.Android.RuntimeTests.BundleTest.TestBundleIntegerArrayList2' failed: System.MemberAccessException : Cannot create an instance of Android.Runtime.JavaList`1[T] because Type.ContainsGenericParameters is true. at System.Reflection.RuntimeConstructorInfo.DoInvoke (System.Object obj, System.Reflection.BindingFlags invokeAttr, System.Reflection.Binder binder, System.Object[] parameters, System.Globalization.CultureInfo culture) at System.Reflection.RuntimeConstructorInfo.Invoke (System.Reflection.BindingFlags invokeAttr, System.Reflection.Binder binder, System.Object[] parameters, System.Globalization.CultureInfo culture) at System.Reflection.ConstructorInfo.Invoke (System.Object[] parameters) at Java.Interop.TypeManager.CreateProxy (System.Type type, System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Java.Interop.TypeManager.CreateInstance (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer, System.Type targetType) at Java.Lang.Object.GetObject (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer, System.Type type) at Java.Lang.Object._GetObject[T] (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Java.Lang.Object.GetObject[T] (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Android.OS.Bundle.Get (System.String key) at Xamarin.Android.RuntimeTests.BundleTest.TestBundleIntegerArrayList2 () at (wrapper managed-to-native) System.Reflection.RuntimeMethodInfo.InternalInvoke(System.Reflection.RuntimeMethodInfo,object,object[],System.Exception&) at System.Reflection.RuntimeMethodInfo.Invoke (System.Object obj, System.Reflection.BindingFlags invokeAttr, System.Reflection.Binder binder, System.Object[] parameters, System.Globalization.CultureInfo culture) The problem appears to be rooted in ce2bc68. In a pre-ce2bc689 world, there were two separate sets of "typemap" files that the application could use: * Java-to-managed mappings, in `typemap.jm` * Managed-to-Java mappings, in `typemap.mj`. These files did *not* need to have the same number of entries! % unzip /Library/Frameworks/Xamarin.Android.framework/Versions/10.2.0.100/lib/xamarin.android/xbuild/Xamarin/Android/Mono.Android.DebugRuntime-debug.apk typemap.jm % unzip /Library/Frameworks/Xamarin.Android.framework/Versions/10.2.0.100/lib/xamarin.android/xbuild/Xamarin/Android/Mono.Android.DebugRuntime-debug.apk typemap.mj % strings typemap.jm | wc -l 10318 % strings typemap.mj | wc -l 11956 The reason why they have a different number of entries is *aliasing*: there can be *multiple* bindings for a given Java type. The managed-to-Java mappings can contain *all* of them; the Java-to-managed mapping *must* pick Only One. For example, [`java.util.ArrayList`][0] contains (at least?) *three* bindings: * [`Android.Runtime.JavaList`][1] * [`Android.Runtime.JavaList<T>`][2] * `Java.Util.ArrayList` (generated at build time) In a pre-ce2bc689 world, this was "fine": we'd binary search each file *separately* to find type mappings. This changed in ce2bc68, as the typemap information was merged into a *single* array in order to de-duplicate information. This reduced `.apk` size. An unanticipated result of this is that the Java-to-managed mappings can now contain *duplicate* keys, "as if" the original `typemap.jm` had the contents: java/util/ArrayList Android.Runtime.JavaList, Mono.Android java/util/ArrayList Android.Runtime.JavaList`1, Mono.Android java/util/ArrayList Java.Util.ArrayLlist, Mono.Android Whereas pre-ce2bc689 there would have only been the first entry. "Sometimes" this duplication is "fine": the typemap search "Just happens" to return the first entry, or the 3rd entry, and apps continue to work as intended. "Sometimes" it isn't: the typemap binary search finds the 2nd entry, which is a generic type. This results in attempting to instantiate a generic type *without appropriate type parameters*, which results in the aforementioned `MemberAccessException`. Update typemap generation code in `Xamarin.Android.Build.Tasks` so that all the duplicate Java type names will point to the same managed type name. Additionally, make sure we select the managed type in the same fashion the old typemap generator in `Java.Interop` worked - prefer base types to the derived ones if the type is an interface or an abstract class. The effect of this change is that no matter which entry `binary_search` ends up selecting it will always return the same managed type name for all the aliased Java types. ~~ Release mode ~~ Another issue introduced in ce2bc68 is that we no longer ignore interface and generic types when generating the Java-to-managed maps. This adversely affects the `Release` mode in which it is possible that an attempt to instantiate an open generic type (or an interface) will happen, leading to error similar to: FATAL EXCEPTION: main Process: com.companyname.androidsingleviewapp1, PID: 24068 android.runtime.JavaProxyThrowable: System.MemberAccessException: Cannot create an instance of Com.Contoso.Javalibrary1.Class0`1[T] because Type.ContainsGenericParameters is true. at System.Reflection.RuntimeConstructorInfo.DoInvoke (System.Object obj, System.Reflection.BindingFlags invokeAttr, System.Reflection.Binder binder, System.Object[] parameters, System.Globalization.CultureInfo culture) at System.Reflection.RuntimeConstructorInfo.Invoke (System.Reflection.BindingFlags invokeAttr, System.Reflection.Binder binder, System.Object[] parameters, System.Globalization.CultureInfo culture) at System.Reflection.ConstructorInfo.Invoke (System.Object[] parameters) at Java.Interop.TypeManager.CreateProxy (System.Type type, System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Java.Interop.TypeManager.CreateInstance (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer, System.Type targetType) at Java.Lang.Object.GetObject (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer, System.Type type) at Java.Lang.Object._GetObject[T] (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Java.Lang.Object.GetObject[T] (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Com.Contoso.Javalibrary1.Class1.Create () at AndroidSingleViewApp1.MainActivity.OnCreate (Android.OS.Bundle savedInstanceState) at Android.App.Activity.n_OnCreate_Landroid_os_Bundle_ (System.IntPtr jnienv, System.IntPtr native__this, System.IntPtr native_savedInstanceState) at (wrapper dynamic-method) Android.Runtime.DynamicMethodNameCounter.1(intptr,intptr,intptr) at crc647d7dcab8da8ee782.MainActivity.n_onCreate(Native Method) at crc647d7dcab8da8ee782.MainActivity.onCreate(MainActivity.java:29) at android.app.Activity.performCreate(Activity.java:7144) at android.app.Activity.performCreate(Activity.java:7135) at android.app.Instrumentation.callActivityOnCreate(Instrumentation.java:1271) at android.app.ActivityThread.performLaunchActivity(ActivityThread.java:2931) at android.app.ActivityThread.handleLaunchActivity(ActivityThread.java:3086) at android.app.servertransaction.LaunchActivityItem.execute(LaunchActivityItem.java:78) at android.app.servertransaction.TransactionExecutor.executeCallbacks(TransactionExecutor.java:108) at android.app.servertransaction.TransactionExecutor.execute(TransactionExecutor.java:68) at android.app.ActivityThread$H.handleMessage(ActivityThread.java:1816) at android.os.Handler.dispatchMessage(Handler.java:106) at android.os.Looper.loop(Looper.java:193) at android.app.ActivityThread.main(ActivityThread.java:6718) at java.lang.reflect.Method.invoke(Native Method) at com.android.internal.os.RuntimeInit$MethodAndArgsCaller.run(RuntimeInit.java:493) at com.android.internal.os.ZygoteInit.main(ZygoteInit.java:858) Update typemap generation code to also ignore generic types and interfaces when generating the Java to Managed map in `Release` mode. We must not simply skip outputting entries for such types because we need to maintain element count parity between java-to-managed and managed-to-java lookup tables. Therefore, the way we skip such types is to output a type token ID of `0` instead of the actual one. This will cause the runtime code to fail to find a mapping, thus allowing the managed code to deal with such a type properly. Additionally, update `TestBundleIntegerArrayList2()` so that it asserts the runtime *type* of `obj` returned, asserting that it is in fact a `JavaList` and not e.g. `Java.Util.ArrayList` or something. (The linker could otherwise "change" things, and this assertion will ensure that `JavaList` won't be linked away…) [0]: https://developer.android.com/reference/java/util/ArrayList [1]: https://github.com/xamarin/xamarin-android/blob/61f09bf2ef0c8f09550e2d77eb97f417432b315b/src/Mono.Android/Android.Runtime/JavaList.cs#L11-L13 [2]: https://github.com/xamarin/xamarin-android/blob/61f09bf2ef0c8f09550e2d77eb97f417432b315b/src/Mono.Android/Android.Runtime/JavaList.cs#L667-L668
grendello
added a commit
to grendello/xamarin-android
that referenced
this issue
May 7, 2020
Fixes: dotnet#4596 Fixes: dotnet#4660 Context: xamarin/monodroid@192b1f1 Context: https://github.com/xamarin/java.interop/blob/fc18c54b2ccf14f444fadac0a1e7e81f837cc470/src/Java.Interop.Tools.JavaCallableWrappers/Java.Interop.Tools.JavaCallableWrappers/TypeNameMapGenerator.cs#L149-L186 Context: https://github.com/xamarin/java.interop/blob/010161d1ef6625b762429334f02e7b15e1f7caae/src/Java.Interop.Tools.JavaCallableWrappers/Java.Interop.Tools.JavaCallableWrappers/TypeNameMapGenerator.cs#L155 ~~ Debug Mode ~~ In some environments, `BundleTest.TestBundleIntegerArrayList2()` fails when using the commercial shared runtime: Test 'Xamarin.Android.RuntimeTests.BundleTest.TestBundleIntegerArrayList2' failed: System.MemberAccessException : Cannot create an instance of Android.Runtime.JavaList`1[T] because Type.ContainsGenericParameters is true. at System.Reflection.RuntimeConstructorInfo.DoInvoke (System.Object obj, System.Reflection.BindingFlags invokeAttr, System.Reflection.Binder binder, System.Object[] parameters, System.Globalization.CultureInfo culture) at System.Reflection.RuntimeConstructorInfo.Invoke (System.Reflection.BindingFlags invokeAttr, System.Reflection.Binder binder, System.Object[] parameters, System.Globalization.CultureInfo culture) at System.Reflection.ConstructorInfo.Invoke (System.Object[] parameters) at Java.Interop.TypeManager.CreateProxy (System.Type type, System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Java.Interop.TypeManager.CreateInstance (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer, System.Type targetType) at Java.Lang.Object.GetObject (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer, System.Type type) at Java.Lang.Object._GetObject[T] (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Java.Lang.Object.GetObject[T] (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Android.OS.Bundle.Get (System.String key) at Xamarin.Android.RuntimeTests.BundleTest.TestBundleIntegerArrayList2 () at (wrapper managed-to-native) System.Reflection.RuntimeMethodInfo.InternalInvoke(System.Reflection.RuntimeMethodInfo,object,object[],System.Exception&) at System.Reflection.RuntimeMethodInfo.Invoke (System.Object obj, System.Reflection.BindingFlags invokeAttr, System.Reflection.Binder binder, System.Object[] parameters, System.Globalization.CultureInfo culture) The problem appears to be rooted in ce2bc68. In a pre-ce2bc689 world, there were two separate sets of "typemap" files that the application could use: * Java-to-managed mappings, in `typemap.jm` * Managed-to-Java mappings, in `typemap.mj`. These files did *not* need to have the same number of entries! % unzip /Library/Frameworks/Xamarin.Android.framework/Versions/10.2.0.100/lib/xamarin.android/xbuild/Xamarin/Android/Mono.Android.DebugRuntime-debug.apk typemap.jm % unzip /Library/Frameworks/Xamarin.Android.framework/Versions/10.2.0.100/lib/xamarin.android/xbuild/Xamarin/Android/Mono.Android.DebugRuntime-debug.apk typemap.mj % strings typemap.jm | wc -l 10318 % strings typemap.mj | wc -l 11956 The reason why they have a different number of entries is *aliasing*: there can be *multiple* bindings for a given Java type. The managed-to-Java mappings can contain *all* of them; the Java-to-managed mapping *must* pick Only One. For example, [`java.util.ArrayList`][0] contains (at least?) *three* bindings: * [`Android.Runtime.JavaList`][1] * [`Android.Runtime.JavaList<T>`][2] * `Java.Util.ArrayList` (generated at build time) In a pre-ce2bc689 world, this was "fine": we'd binary search each file *separately* to find type mappings. This changed in ce2bc68, as the typemap information was merged into a *single* array in order to de-duplicate information. This reduced `.apk` size. An unanticipated result of this is that the Java-to-managed mappings can now contain *duplicate* keys, "as if" the original `typemap.jm` had the contents: java/util/ArrayList Android.Runtime.JavaList, Mono.Android java/util/ArrayList Android.Runtime.JavaList`1, Mono.Android java/util/ArrayList Java.Util.ArrayLlist, Mono.Android Whereas pre-ce2bc689 there would have only been the first entry. "Sometimes" this duplication is "fine": the typemap search "Just happens" to return the first entry, or the 3rd entry, and apps continue to work as intended. "Sometimes" it isn't: the typemap binary search finds the 2nd entry, which is a generic type. This results in attempting to instantiate a generic type *without appropriate type parameters*, which results in the aforementioned `MemberAccessException`. Update typemap generation code in `Xamarin.Android.Build.Tasks` so that all the duplicate Java type names will point to the same managed type name. Additionally, make sure we select the managed type in the same fashion the old typemap generator in `Java.Interop` worked - prefer base types to the derived ones if the type is an interface or an abstract class. The effect of this change is that no matter which entry `binary_search` ends up selecting it will always return the same managed type name for all the aliased Java types. ~~ Release mode ~~ Another issue introduced in ce2bc68 is that we no longer ignore interface and generic types when generating the Java-to-managed maps. This adversely affects the `Release` mode in which it is possible that an attempt to instantiate an open generic type (or an interface) will happen, leading to error similar to: FATAL EXCEPTION: main Process: com.companyname.androidsingleviewapp1, PID: 24068 android.runtime.JavaProxyThrowable: System.MemberAccessException: Cannot create an instance of Com.Contoso.Javalibrary1.Class0`1[T] because Type.ContainsGenericParameters is true. at System.Reflection.RuntimeConstructorInfo.DoInvoke (System.Object obj, System.Reflection.BindingFlags invokeAttr, System.Reflection.Binder binder, System.Object[] parameters, System.Globalization.CultureInfo culture) at System.Reflection.RuntimeConstructorInfo.Invoke (System.Reflection.BindingFlags invokeAttr, System.Reflection.Binder binder, System.Object[] parameters, System.Globalization.CultureInfo culture) at System.Reflection.ConstructorInfo.Invoke (System.Object[] parameters) at Java.Interop.TypeManager.CreateProxy (System.Type type, System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Java.Interop.TypeManager.CreateInstance (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer, System.Type targetType) at Java.Lang.Object.GetObject (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer, System.Type type) at Java.Lang.Object._GetObject[T] (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Java.Lang.Object.GetObject[T] (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Com.Contoso.Javalibrary1.Class1.Create () at AndroidSingleViewApp1.MainActivity.OnCreate (Android.OS.Bundle savedInstanceState) at Android.App.Activity.n_OnCreate_Landroid_os_Bundle_ (System.IntPtr jnienv, System.IntPtr native__this, System.IntPtr native_savedInstanceState) at (wrapper dynamic-method) Android.Runtime.DynamicMethodNameCounter.1(intptr,intptr,intptr) at crc647d7dcab8da8ee782.MainActivity.n_onCreate(Native Method) at crc647d7dcab8da8ee782.MainActivity.onCreate(MainActivity.java:29) at android.app.Activity.performCreate(Activity.java:7144) at android.app.Activity.performCreate(Activity.java:7135) at android.app.Instrumentation.callActivityOnCreate(Instrumentation.java:1271) at android.app.ActivityThread.performLaunchActivity(ActivityThread.java:2931) at android.app.ActivityThread.handleLaunchActivity(ActivityThread.java:3086) at android.app.servertransaction.LaunchActivityItem.execute(LaunchActivityItem.java:78) at android.app.servertransaction.TransactionExecutor.executeCallbacks(TransactionExecutor.java:108) at android.app.servertransaction.TransactionExecutor.execute(TransactionExecutor.java:68) at android.app.ActivityThread$H.handleMessage(ActivityThread.java:1816) at android.os.Handler.dispatchMessage(Handler.java:106) at android.os.Looper.loop(Looper.java:193) at android.app.ActivityThread.main(ActivityThread.java:6718) at java.lang.reflect.Method.invoke(Native Method) at com.android.internal.os.RuntimeInit$MethodAndArgsCaller.run(RuntimeInit.java:493) at com.android.internal.os.ZygoteInit.main(ZygoteInit.java:858) Update typemap generation code to also ignore generic types and interfaces when generating the Java to Managed map in `Release` mode. We must not simply skip outputting entries for such types because we need to maintain element count parity between java-to-managed and managed-to-java lookup tables. Therefore, the way we skip such types is to output a type token ID of `0` instead of the actual one. This will cause the runtime code to fail to find a mapping, thus allowing the managed code to deal with such a type properly. Additionally, update `TestBundleIntegerArrayList2()` so that it asserts the runtime *type* of `obj` returned, asserting that it is in fact a `JavaList` and not e.g. `Java.Util.ArrayList` or something. (The linker could otherwise "change" things, and this assertion will ensure that `JavaList` won't be linked away…) [0]: https://developer.android.com/reference/java/util/ArrayList [1]: https://github.com/xamarin/xamarin-android/blob/61f09bf2ef0c8f09550e2d77eb97f417432b315b/src/Mono.Android/Android.Runtime/JavaList.cs#L11-L13 [2]: https://github.com/xamarin/xamarin-android/blob/61f09bf2ef0c8f09550e2d77eb97f417432b315b/src/Mono.Android/Android.Runtime/JavaList.cs#L667-L668
grendello
added a commit
to grendello/xamarin-android
that referenced
this issue
May 7, 2020
Fixes: dotnet#4596 Fixes: dotnet#4660 Context: xamarin/monodroid@192b1f1 Context: https://github.com/xamarin/java.interop/blob/fc18c54b2ccf14f444fadac0a1e7e81f837cc470/src/Java.Interop.Tools.JavaCallableWrappers/Java.Interop.Tools.JavaCallableWrappers/TypeNameMapGenerator.cs#L149-L186 Context: https://github.com/xamarin/java.interop/blob/010161d1ef6625b762429334f02e7b15e1f7caae/src/Java.Interop.Tools.JavaCallableWrappers/Java.Interop.Tools.JavaCallableWrappers/TypeNameMapGenerator.cs#L155 ~~ Debug Mode ~~ In some environments, `BundleTest.TestBundleIntegerArrayList2()` fails when using the commercial shared runtime: Test 'Xamarin.Android.RuntimeTests.BundleTest.TestBundleIntegerArrayList2' failed: System.MemberAccessException : Cannot create an instance of Android.Runtime.JavaList`1[T] because Type.ContainsGenericParameters is true. at System.Reflection.RuntimeConstructorInfo.DoInvoke (System.Object obj, System.Reflection.BindingFlags invokeAttr, System.Reflection.Binder binder, System.Object[] parameters, System.Globalization.CultureInfo culture) at System.Reflection.RuntimeConstructorInfo.Invoke (System.Reflection.BindingFlags invokeAttr, System.Reflection.Binder binder, System.Object[] parameters, System.Globalization.CultureInfo culture) at System.Reflection.ConstructorInfo.Invoke (System.Object[] parameters) at Java.Interop.TypeManager.CreateProxy (System.Type type, System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Java.Interop.TypeManager.CreateInstance (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer, System.Type targetType) at Java.Lang.Object.GetObject (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer, System.Type type) at Java.Lang.Object._GetObject[T] (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Java.Lang.Object.GetObject[T] (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Android.OS.Bundle.Get (System.String key) at Xamarin.Android.RuntimeTests.BundleTest.TestBundleIntegerArrayList2 () at (wrapper managed-to-native) System.Reflection.RuntimeMethodInfo.InternalInvoke(System.Reflection.RuntimeMethodInfo,object,object[],System.Exception&) at System.Reflection.RuntimeMethodInfo.Invoke (System.Object obj, System.Reflection.BindingFlags invokeAttr, System.Reflection.Binder binder, System.Object[] parameters, System.Globalization.CultureInfo culture) The problem appears to be rooted in ce2bc68. In a pre-ce2bc689 world, there were two separate sets of "typemap" files that the application could use: * Java-to-managed mappings, in `typemap.jm` * Managed-to-Java mappings, in `typemap.mj`. These files did *not* need to have the same number of entries! % unzip /Library/Frameworks/Xamarin.Android.framework/Versions/10.2.0.100/lib/xamarin.android/xbuild/Xamarin/Android/Mono.Android.DebugRuntime-debug.apk typemap.jm % unzip /Library/Frameworks/Xamarin.Android.framework/Versions/10.2.0.100/lib/xamarin.android/xbuild/Xamarin/Android/Mono.Android.DebugRuntime-debug.apk typemap.mj % strings typemap.jm | wc -l 10318 % strings typemap.mj | wc -l 11956 The reason why they have a different number of entries is *aliasing*: there can be *multiple* bindings for a given Java type. The managed-to-Java mappings can contain *all* of them; the Java-to-managed mapping *must* pick Only One. For example, [`java.util.ArrayList`][0] contains (at least?) *three* bindings: * [`Android.Runtime.JavaList`][1] * [`Android.Runtime.JavaList<T>`][2] * `Java.Util.ArrayList` (generated at build time) In a pre-ce2bc689 world, this was "fine": we'd binary search each file *separately* to find type mappings. This changed in ce2bc68, as the typemap information was merged into a *single* array in order to de-duplicate information. This reduced `.apk` size. An unanticipated result of this is that the Java-to-managed mappings can now contain *duplicate* keys, "as if" the original `typemap.jm` had the contents: java/util/ArrayList Android.Runtime.JavaList, Mono.Android java/util/ArrayList Android.Runtime.JavaList`1, Mono.Android java/util/ArrayList Java.Util.ArrayLlist, Mono.Android Whereas pre-ce2bc689 there would have only been the first entry. "Sometimes" this duplication is "fine": the typemap search "Just happens" to return the first entry, or the 3rd entry, and apps continue to work as intended. "Sometimes" it isn't: the typemap binary search finds the 2nd entry, which is a generic type. This results in attempting to instantiate a generic type *without appropriate type parameters*, which results in the aforementioned `MemberAccessException`. Update typemap generation code in `Xamarin.Android.Build.Tasks` so that all the duplicate Java type names will point to the same managed type name. Additionally, make sure we select the managed type in the same fashion the old typemap generator in `Java.Interop` worked - prefer base types to the derived ones if the type is an interface or an abstract class. The effect of this change is that no matter which entry `binary_search` ends up selecting it will always return the same managed type name for all the aliased Java types. ~~ Release mode ~~ Another issue introduced in ce2bc68 is that we no longer ignore interface and generic types when generating the Java-to-managed maps. This adversely affects the `Release` mode in which it is possible that an attempt to instantiate an open generic type (or an interface) will happen, leading to error similar to: FATAL EXCEPTION: main Process: com.companyname.androidsingleviewapp1, PID: 24068 android.runtime.JavaProxyThrowable: System.MemberAccessException: Cannot create an instance of Com.Contoso.Javalibrary1.Class0`1[T] because Type.ContainsGenericParameters is true. at System.Reflection.RuntimeConstructorInfo.DoInvoke (System.Object obj, System.Reflection.BindingFlags invokeAttr, System.Reflection.Binder binder, System.Object[] parameters, System.Globalization.CultureInfo culture) at System.Reflection.RuntimeConstructorInfo.Invoke (System.Reflection.BindingFlags invokeAttr, System.Reflection.Binder binder, System.Object[] parameters, System.Globalization.CultureInfo culture) at System.Reflection.ConstructorInfo.Invoke (System.Object[] parameters) at Java.Interop.TypeManager.CreateProxy (System.Type type, System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Java.Interop.TypeManager.CreateInstance (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer, System.Type targetType) at Java.Lang.Object.GetObject (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer, System.Type type) at Java.Lang.Object._GetObject[T] (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Java.Lang.Object.GetObject[T] (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Com.Contoso.Javalibrary1.Class1.Create () at AndroidSingleViewApp1.MainActivity.OnCreate (Android.OS.Bundle savedInstanceState) at Android.App.Activity.n_OnCreate_Landroid_os_Bundle_ (System.IntPtr jnienv, System.IntPtr native__this, System.IntPtr native_savedInstanceState) at (wrapper dynamic-method) Android.Runtime.DynamicMethodNameCounter.1(intptr,intptr,intptr) at crc647d7dcab8da8ee782.MainActivity.n_onCreate(Native Method) at crc647d7dcab8da8ee782.MainActivity.onCreate(MainActivity.java:29) at android.app.Activity.performCreate(Activity.java:7144) at android.app.Activity.performCreate(Activity.java:7135) at android.app.Instrumentation.callActivityOnCreate(Instrumentation.java:1271) at android.app.ActivityThread.performLaunchActivity(ActivityThread.java:2931) at android.app.ActivityThread.handleLaunchActivity(ActivityThread.java:3086) at android.app.servertransaction.LaunchActivityItem.execute(LaunchActivityItem.java:78) at android.app.servertransaction.TransactionExecutor.executeCallbacks(TransactionExecutor.java:108) at android.app.servertransaction.TransactionExecutor.execute(TransactionExecutor.java:68) at android.app.ActivityThread$H.handleMessage(ActivityThread.java:1816) at android.os.Handler.dispatchMessage(Handler.java:106) at android.os.Looper.loop(Looper.java:193) at android.app.ActivityThread.main(ActivityThread.java:6718) at java.lang.reflect.Method.invoke(Native Method) at com.android.internal.os.RuntimeInit$MethodAndArgsCaller.run(RuntimeInit.java:493) at com.android.internal.os.ZygoteInit.main(ZygoteInit.java:858) Update typemap generation code to also ignore generic types and interfaces when generating the Java to Managed map in `Release` mode. We must not simply skip outputting entries for such types because we need to maintain element count parity between java-to-managed and managed-to-java lookup tables. Therefore, the way we skip such types is to output a type token ID of `0` instead of the actual one. This will cause the runtime code to fail to find a mapping, thus allowing the managed code to deal with such a type properly. ~~ Test Updates ~~ Update `TestBundleIntegerArrayList2()` so that it asserts the runtime *type* of `obj` returned, asserting that it is in fact a `JavaList` and not e.g. `Java.Util.ArrayList` or something. (The linker could otherwise "change" things, and this assertion will ensure that `JavaList` won't be linked away…) Add a unit test for issue dotnet#4660 (by @brendanzagaeski) It's very cute. :-) We sort the Java-to-managed type name table, so it can be binary searched to find the Managed type. Additionally, the "preferred" managed type is the *first* type in `monodis --typedef` output; the first type that we encounter when going through all types in an assembly. Brendan's cute way of provoking the bug is to add a new *generic* type which will come *before* the "normally preferred" (and correct!) type. Thus, even if/when binary search returns the first entry, that'll still be the *wrong entry*, because that'll be a generic type! Very cute. [0]: https://developer.android.com/reference/java/util/ArrayList [1]: https://github.com/xamarin/xamarin-android/blob/61f09bf2ef0c8f09550e2d77eb97f417432b315b/src/Mono.Android/Android.Runtime/JavaList.cs#L11-L13 [2]: https://github.com/xamarin/xamarin-android/blob/61f09bf2ef0c8f09550e2d77eb97f417432b315b/src/Mono.Android/Android.Runtime/JavaList.cs#L667-L668
grendello
added a commit
to grendello/xamarin-android
that referenced
this issue
May 8, 2020
Fixes: dotnet#4596 Fixes: dotnet#4660 Context: xamarin/monodroid@192b1f1 Context: https://github.com/xamarin/java.interop/blob/fc18c54b2ccf14f444fadac0a1e7e81f837cc470/src/Java.Interop.Tools.JavaCallableWrappers/Java.Interop.Tools.JavaCallableWrappers/TypeNameMapGenerator.cs#L149-L186 Context: https://github.com/xamarin/java.interop/blob/010161d1ef6625b762429334f02e7b15e1f7caae/src/Java.Interop.Tools.JavaCallableWrappers/Java.Interop.Tools.JavaCallableWrappers/TypeNameMapGenerator.cs#L155 ~~ Debug Mode ~~ In some environments, `BundleTest.TestBundleIntegerArrayList2()` fails when using the commercial shared runtime: Test 'Xamarin.Android.RuntimeTests.BundleTest.TestBundleIntegerArrayList2' failed: System.MemberAccessException : Cannot create an instance of Android.Runtime.JavaList`1[T] because Type.ContainsGenericParameters is true. at System.Reflection.RuntimeConstructorInfo.DoInvoke (System.Object obj, System.Reflection.BindingFlags invokeAttr, System.Reflection.Binder binder, System.Object[] parameters, System.Globalization.CultureInfo culture) at System.Reflection.RuntimeConstructorInfo.Invoke (System.Reflection.BindingFlags invokeAttr, System.Reflection.Binder binder, System.Object[] parameters, System.Globalization.CultureInfo culture) at System.Reflection.ConstructorInfo.Invoke (System.Object[] parameters) at Java.Interop.TypeManager.CreateProxy (System.Type type, System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Java.Interop.TypeManager.CreateInstance (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer, System.Type targetType) at Java.Lang.Object.GetObject (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer, System.Type type) at Java.Lang.Object._GetObject[T] (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Java.Lang.Object.GetObject[T] (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Android.OS.Bundle.Get (System.String key) at Xamarin.Android.RuntimeTests.BundleTest.TestBundleIntegerArrayList2 () at (wrapper managed-to-native) System.Reflection.RuntimeMethodInfo.InternalInvoke(System.Reflection.RuntimeMethodInfo,object,object[],System.Exception&) at System.Reflection.RuntimeMethodInfo.Invoke (System.Object obj, System.Reflection.BindingFlags invokeAttr, System.Reflection.Binder binder, System.Object[] parameters, System.Globalization.CultureInfo culture) The problem appears to be rooted in ce2bc68. In a pre-ce2bc689 world, there were two separate sets of "typemap" files that the application could use: * Java-to-managed mappings, in `typemap.jm` * Managed-to-Java mappings, in `typemap.mj`. These files did *not* need to have the same number of entries! % unzip /Library/Frameworks/Xamarin.Android.framework/Versions/10.2.0.100/lib/xamarin.android/xbuild/Xamarin/Android/Mono.Android.DebugRuntime-debug.apk typemap.jm % unzip /Library/Frameworks/Xamarin.Android.framework/Versions/10.2.0.100/lib/xamarin.android/xbuild/Xamarin/Android/Mono.Android.DebugRuntime-debug.apk typemap.mj % strings typemap.jm | wc -l 10318 % strings typemap.mj | wc -l 11956 The reason why they have a different number of entries is *aliasing*: there can be *multiple* bindings for a given Java type. The managed-to-Java mappings can contain *all* of them; the Java-to-managed mapping *must* pick Only One. For example, [`java.util.ArrayList`][0] contains (at least?) *three* bindings: * [`Android.Runtime.JavaList`][1] * [`Android.Runtime.JavaList<T>`][2] * `Java.Util.ArrayList` (generated at build time) In a pre-ce2bc689 world, this was "fine": we'd binary search each file *separately* to find type mappings. This changed in ce2bc68, as the typemap information was merged into a *single* array in order to de-duplicate information. This reduced `.apk` size. An unanticipated result of this is that the Java-to-managed mappings can now contain *duplicate* keys, "as if" the original `typemap.jm` had the contents: java/util/ArrayList Android.Runtime.JavaList, Mono.Android java/util/ArrayList Android.Runtime.JavaList`1, Mono.Android java/util/ArrayList Java.Util.ArrayLlist, Mono.Android Whereas pre-ce2bc689 there would have only been the first entry. "Sometimes" this duplication is "fine": the typemap search "Just happens" to return the first entry, or the 3rd entry, and apps continue to work as intended. "Sometimes" it isn't: the typemap binary search finds the 2nd entry, which is a generic type. This results in attempting to instantiate a generic type *without appropriate type parameters*, which results in the aforementioned `MemberAccessException`. Update typemap generation code in `Xamarin.Android.Build.Tasks` so that all the duplicate Java type names will point to the same managed type name. Additionally, make sure we select the managed type in the same fashion the old typemap generator in `Java.Interop` worked - prefer base types to the derived ones if the type is an interface or an abstract class. The effect of this change is that no matter which entry `binary_search` ends up selecting it will always return the same managed type name for all the aliased Java types. ~~ Release mode ~~ Another issue introduced in ce2bc68 is that we no longer ignore interface and generic types when generating the Java-to-managed maps. This adversely affects the `Release` mode in which it is possible that an attempt to instantiate an open generic type (or an interface) will happen, leading to error similar to: FATAL EXCEPTION: main Process: com.companyname.androidsingleviewapp1, PID: 24068 android.runtime.JavaProxyThrowable: System.MemberAccessException: Cannot create an instance of Com.Contoso.Javalibrary1.Class0`1[T] because Type.ContainsGenericParameters is true. at System.Reflection.RuntimeConstructorInfo.DoInvoke (System.Object obj, System.Reflection.BindingFlags invokeAttr, System.Reflection.Binder binder, System.Object[] parameters, System.Globalization.CultureInfo culture) at System.Reflection.RuntimeConstructorInfo.Invoke (System.Reflection.BindingFlags invokeAttr, System.Reflection.Binder binder, System.Object[] parameters, System.Globalization.CultureInfo culture) at System.Reflection.ConstructorInfo.Invoke (System.Object[] parameters) at Java.Interop.TypeManager.CreateProxy (System.Type type, System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Java.Interop.TypeManager.CreateInstance (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer, System.Type targetType) at Java.Lang.Object.GetObject (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer, System.Type type) at Java.Lang.Object._GetObject[T] (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Java.Lang.Object.GetObject[T] (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Com.Contoso.Javalibrary1.Class1.Create () at AndroidSingleViewApp1.MainActivity.OnCreate (Android.OS.Bundle savedInstanceState) at Android.App.Activity.n_OnCreate_Landroid_os_Bundle_ (System.IntPtr jnienv, System.IntPtr native__this, System.IntPtr native_savedInstanceState) at (wrapper dynamic-method) Android.Runtime.DynamicMethodNameCounter.1(intptr,intptr,intptr) at crc647d7dcab8da8ee782.MainActivity.n_onCreate(Native Method) at crc647d7dcab8da8ee782.MainActivity.onCreate(MainActivity.java:29) at android.app.Activity.performCreate(Activity.java:7144) at android.app.Activity.performCreate(Activity.java:7135) at android.app.Instrumentation.callActivityOnCreate(Instrumentation.java:1271) at android.app.ActivityThread.performLaunchActivity(ActivityThread.java:2931) at android.app.ActivityThread.handleLaunchActivity(ActivityThread.java:3086) at android.app.servertransaction.LaunchActivityItem.execute(LaunchActivityItem.java:78) at android.app.servertransaction.TransactionExecutor.executeCallbacks(TransactionExecutor.java:108) at android.app.servertransaction.TransactionExecutor.execute(TransactionExecutor.java:68) at android.app.ActivityThread$H.handleMessage(ActivityThread.java:1816) at android.os.Handler.dispatchMessage(Handler.java:106) at android.os.Looper.loop(Looper.java:193) at android.app.ActivityThread.main(ActivityThread.java:6718) at java.lang.reflect.Method.invoke(Native Method) at com.android.internal.os.RuntimeInit$MethodAndArgsCaller.run(RuntimeInit.java:493) at com.android.internal.os.ZygoteInit.main(ZygoteInit.java:858) Update typemap generation code to also ignore generic types and interfaces when generating the Java to Managed map in `Release` mode. We must not simply skip outputting entries for such types because we need to maintain element count parity between java-to-managed and managed-to-java lookup tables. Therefore, the way we skip such types is to output a type token ID of `0` instead of the actual one. This will cause the runtime code to fail to find a mapping, thus allowing the managed code to deal with such a type properly. ~~ Test Updates ~~ Update `TestBundleIntegerArrayList2()` so that it asserts the runtime *type* of `obj` returned, asserting that it is in fact a `JavaList` and not e.g. `Java.Util.ArrayList` or something. (The linker could otherwise "change" things, and this assertion will ensure that `JavaList` won't be linked away…) Add a unit test for issue dotnet#4660 (by @brendanzagaeski) It's very cute. :-) We sort the Java-to-managed type name table, so it can be binary searched to find the Managed type. Additionally, the "preferred" managed type is the *first* type in `monodis --typedef` output; the first type that we encounter when going through all types in an assembly. Brendan's cute way of provoking the bug is to add a new *generic* type which will come *before* the "normally preferred" (and correct!) type. Thus, even if/when binary search returns the first entry, that'll still be the *wrong entry*, because that'll be a generic type! Very cute. [0]: https://developer.android.com/reference/java/util/ArrayList [1]: https://github.com/xamarin/xamarin-android/blob/61f09bf2ef0c8f09550e2d77eb97f417432b315b/src/Mono.Android/Android.Runtime/JavaList.cs#L11-L13 [2]: https://github.com/xamarin/xamarin-android/blob/61f09bf2ef0c8f09550e2d77eb97f417432b315b/src/Mono.Android/Android.Runtime/JavaList.cs#L667-L668
jonpryor
pushed a commit
that referenced
this issue
May 8, 2020
Fixes: #4596 Fixes: #4660 Context: xamarin/monodroid@192b1f1 Context: https://github.com/xamarin/java.interop/blob/fc18c54b2ccf14f444fadac0a1e7e81f837cc470/src/Java.Interop.Tools.JavaCallableWrappers/Java.Interop.Tools.JavaCallableWrappers/TypeNameMapGenerator.cs#L149-L186 Context: https://github.com/xamarin/java.interop/blob/010161d1ef6625b762429334f02e7b15e1f7caae/src/Java.Interop.Tools.JavaCallableWrappers/Java.Interop.Tools.JavaCallableWrappers/TypeNameMapGenerator.cs#L155 ~~ Debug Mode ~~ In some environments, `BundleTest.TestBundleIntegerArrayList2()` fails when using the commercial shared runtime: Test 'Xamarin.Android.RuntimeTests.BundleTest.TestBundleIntegerArrayList2' failed: System.MemberAccessException : Cannot create an instance of Android.Runtime.JavaList`1[T] because Type.ContainsGenericParameters is true. at System.Reflection.RuntimeConstructorInfo.DoInvoke (System.Object obj, System.Reflection.BindingFlags invokeAttr, System.Reflection.Binder binder, System.Object[] parameters, System.Globalization.CultureInfo culture) at System.Reflection.RuntimeConstructorInfo.Invoke (System.Reflection.BindingFlags invokeAttr, System.Reflection.Binder binder, System.Object[] parameters, System.Globalization.CultureInfo culture) at System.Reflection.ConstructorInfo.Invoke (System.Object[] parameters) at Java.Interop.TypeManager.CreateProxy (System.Type type, System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Java.Interop.TypeManager.CreateInstance (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer, System.Type targetType) at Java.Lang.Object.GetObject (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer, System.Type type) at Java.Lang.Object._GetObject[T] (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Java.Lang.Object.GetObject[T] (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Android.OS.Bundle.Get (System.String key) at Xamarin.Android.RuntimeTests.BundleTest.TestBundleIntegerArrayList2 () at (wrapper managed-to-native) System.Reflection.RuntimeMethodInfo.InternalInvoke(System.Reflection.RuntimeMethodInfo,object,object[],System.Exception&) at System.Reflection.RuntimeMethodInfo.Invoke (System.Object obj, System.Reflection.BindingFlags invokeAttr, System.Reflection.Binder binder, System.Object[] parameters, System.Globalization.CultureInfo culture) The problem appears to be rooted in ce2bc68. In a pre-ce2bc689 world, there were two separate sets of "typemap" files that the application could use: * Java-to-managed mappings, in `typemap.jm` * Managed-to-Java mappings, in `typemap.mj`. These files did *not* need to have the same number of entries! % unzip /Library/Frameworks/Xamarin.Android.framework/Versions/10.2.0.100/lib/xamarin.android/xbuild/Xamarin/Android/Mono.Android.DebugRuntime-debug.apk typemap.jm % unzip /Library/Frameworks/Xamarin.Android.framework/Versions/10.2.0.100/lib/xamarin.android/xbuild/Xamarin/Android/Mono.Android.DebugRuntime-debug.apk typemap.mj % strings typemap.jm | wc -l 10318 % strings typemap.mj | wc -l 11956 The reason why they have a different number of entries is *aliasing*: there can be *multiple* bindings for a given Java type. The managed-to-Java mappings can contain *all* of them; the Java-to-managed mapping *must* pick Only One. For example, [`java.util.ArrayList`][0] contains (at least?) *three* bindings: * [`Android.Runtime.JavaList`][1] * [`Android.Runtime.JavaList<T>`][2] * `Java.Util.ArrayList` (generated at build time) In a pre-ce2bc689 world, this was "fine": we'd binary search each file *separately* to find type mappings. This changed in ce2bc68, as the typemap information was merged into a *single* array in order to de-duplicate information. This reduced `.apk` size. An unanticipated result of this is that the Java-to-managed mappings can now contain *duplicate* keys, "as if" the original `typemap.jm` had the contents: java/util/ArrayList Android.Runtime.JavaList, Mono.Android java/util/ArrayList Android.Runtime.JavaList`1, Mono.Android java/util/ArrayList Java.Util.ArrayLlist, Mono.Android Whereas pre-ce2bc689 there would have only been the first entry. "Sometimes" this duplication is "fine": the typemap search "Just happens" to return the first entry, or the 3rd entry, and apps continue to work as intended. "Sometimes" it isn't: the typemap binary search finds the 2nd entry, which is a generic type. This results in attempting to instantiate a generic type *without appropriate type parameters*, which results in the aforementioned `MemberAccessException`. Update typemap generation code in `Xamarin.Android.Build.Tasks.dll` so that all the duplicate Java type names will point to the same managed type name. Additionally, make sure we select the managed type in the same fashion the old typemap generator in `Java.Interop` worked: prefer base types to the derived ones if the type is an interface or an abstract class. The effect of this change is that no matter which entry `EmbeddedAssemblies::binary_search()` ends up selecting it will always return the same managed type name for all aliased Java types. ~~ Release config apps ~~ Another issue introduced in ce2bc68 is that we no longer ignore interface and generic types when generating the Java-to-managed maps. This adversely affects the `Release` mode in which it is possible that an attempt to instantiate an open generic type (or an interface) will happen, leading to error similar to: FATAL EXCEPTION: main Process: com.companyname.androidsingleviewapp1, PID: 24068 android.runtime.JavaProxyThrowable: System.MemberAccessException: Cannot create an instance of Com.Contoso.Javalibrary1.Class0`1[T] because Type.ContainsGenericParameters is true. at System.Reflection.RuntimeConstructorInfo.DoInvoke (System.Object obj, System.Reflection.BindingFlags invokeAttr, System.Reflection.Binder binder, System.Object[] parameters, System.Globalization.CultureInfo culture) at System.Reflection.RuntimeConstructorInfo.Invoke (System.Reflection.BindingFlags invokeAttr, System.Reflection.Binder binder, System.Object[] parameters, System.Globalization.CultureInfo culture) at System.Reflection.ConstructorInfo.Invoke (System.Object[] parameters) at Java.Interop.TypeManager.CreateProxy (System.Type type, System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Java.Interop.TypeManager.CreateInstance (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer, System.Type targetType) at Java.Lang.Object.GetObject (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer, System.Type type) at Java.Lang.Object._GetObject[T] (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Java.Lang.Object.GetObject[T] (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Com.Contoso.Javalibrary1.Class1.Create () at AndroidSingleViewApp1.MainActivity.OnCreate (Android.OS.Bundle savedInstanceState) at Android.App.Activity.n_OnCreate_Landroid_os_Bundle_ (System.IntPtr jnienv, System.IntPtr native__this, System.IntPtr native_savedInstanceState) at (wrapper dynamic-method) Android.Runtime.DynamicMethodNameCounter.1(intptr,intptr,intptr) at crc647d7dcab8da8ee782.MainActivity.n_onCreate(Native Method) at crc647d7dcab8da8ee782.MainActivity.onCreate(MainActivity.java:29) at android.app.Activity.performCreate(Activity.java:7144) at android.app.Activity.performCreate(Activity.java:7135) at android.app.Instrumentation.callActivityOnCreate(Instrumentation.java:1271) at android.app.ActivityThread.performLaunchActivity(ActivityThread.java:2931) at android.app.ActivityThread.handleLaunchActivity(ActivityThread.java:3086) at android.app.servertransaction.LaunchActivityItem.execute(LaunchActivityItem.java:78) at android.app.servertransaction.TransactionExecutor.executeCallbacks(TransactionExecutor.java:108) at android.app.servertransaction.TransactionExecutor.execute(TransactionExecutor.java:68) at android.app.ActivityThread$H.handleMessage(ActivityThread.java:1816) at android.os.Handler.dispatchMessage(Handler.java:106) at android.os.Looper.loop(Looper.java:193) at android.app.ActivityThread.main(ActivityThread.java:6718) at java.lang.reflect.Method.invoke(Native Method) at com.android.internal.os.RuntimeInit$MethodAndArgsCaller.run(RuntimeInit.java:493) at com.android.internal.os.ZygoteInit.main(ZygoteInit.java:858) Update typemap generation code to also ignore generic types and interfaces when generating the Java to Managed map in `Release` mode. We must not simply skip outputting entries for such types because we need to maintain element count parity between java-to-managed and managed-to-java lookup tables. Therefore, the way we skip such types is to output a type token ID of `0` instead of the actual one. This will cause the runtime code to fail to find a mapping, thus allowing the managed code to deal with such a type properly. ~~ Test Updates ~~ Update `BundleTest.TestBundleIntegerArrayList2()` so that it asserts the runtime *type* of `obj` returned, asserting that it is in fact a `JavaList` and not e.g. `Java.Util.ArrayList` or something. (The linker could otherwise "change" things, and this assertion will ensure that `JavaList` won't be linked away…) Add a unit test for the Release config app bug in Issue #4660; thanks to @brendanzagaeski. It works by provoking the binary search bug by "ensuring" that a *generic* type comes *before* the "normally preferred" *non*-generic type. Thus, even if/when binary search returns the first entry, that would still be the *wrong entry*, because that'll be a generic type! [0]: https://developer.android.com/reference/java/util/ArrayList [1]: https://github.com/xamarin/xamarin-android/blob/61f09bf2ef0c8f09550e2d77eb97f417432b315b/src/Mono.Android/Android.Runtime/JavaList.cs#L11-L13 [2]: https://github.com/xamarin/xamarin-android/blob/61f09bf2ef0c8f09550e2d77eb97f417432b315b/src/Mono.Android/Android.Runtime/JavaList.cs#L667-L668
jonpryor
pushed a commit
that referenced
this issue
May 8, 2020
Fixes: #4596 Fixes: #4660 Context: xamarin/monodroid@192b1f1 Context: https://github.com/xamarin/java.interop/blob/fc18c54b2ccf14f444fadac0a1e7e81f837cc470/src/Java.Interop.Tools.JavaCallableWrappers/Java.Interop.Tools.JavaCallableWrappers/TypeNameMapGenerator.cs#L149-L186 Context: https://github.com/xamarin/java.interop/blob/010161d1ef6625b762429334f02e7b15e1f7caae/src/Java.Interop.Tools.JavaCallableWrappers/Java.Interop.Tools.JavaCallableWrappers/TypeNameMapGenerator.cs#L155 ~~ Debug Mode ~~ In some environments, `BundleTest.TestBundleIntegerArrayList2()` fails when using the commercial shared runtime: Test 'Xamarin.Android.RuntimeTests.BundleTest.TestBundleIntegerArrayList2' failed: System.MemberAccessException : Cannot create an instance of Android.Runtime.JavaList`1[T] because Type.ContainsGenericParameters is true. at System.Reflection.RuntimeConstructorInfo.DoInvoke (System.Object obj, System.Reflection.BindingFlags invokeAttr, System.Reflection.Binder binder, System.Object[] parameters, System.Globalization.CultureInfo culture) at System.Reflection.RuntimeConstructorInfo.Invoke (System.Reflection.BindingFlags invokeAttr, System.Reflection.Binder binder, System.Object[] parameters, System.Globalization.CultureInfo culture) at System.Reflection.ConstructorInfo.Invoke (System.Object[] parameters) at Java.Interop.TypeManager.CreateProxy (System.Type type, System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Java.Interop.TypeManager.CreateInstance (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer, System.Type targetType) at Java.Lang.Object.GetObject (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer, System.Type type) at Java.Lang.Object._GetObject[T] (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Java.Lang.Object.GetObject[T] (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Android.OS.Bundle.Get (System.String key) at Xamarin.Android.RuntimeTests.BundleTest.TestBundleIntegerArrayList2 () at (wrapper managed-to-native) System.Reflection.RuntimeMethodInfo.InternalInvoke(System.Reflection.RuntimeMethodInfo,object,object[],System.Exception&) at System.Reflection.RuntimeMethodInfo.Invoke (System.Object obj, System.Reflection.BindingFlags invokeAttr, System.Reflection.Binder binder, System.Object[] parameters, System.Globalization.CultureInfo culture) The problem appears to be rooted in ce2bc68. In a pre-ce2bc689 world, there were two separate sets of "typemap" files that the application could use: * Java-to-managed mappings, in `typemap.jm` * Managed-to-Java mappings, in `typemap.mj`. These files did *not* need to have the same number of entries! % unzip /Library/Frameworks/Xamarin.Android.framework/Versions/10.2.0.100/lib/xamarin.android/xbuild/Xamarin/Android/Mono.Android.DebugRuntime-debug.apk typemap.jm % unzip /Library/Frameworks/Xamarin.Android.framework/Versions/10.2.0.100/lib/xamarin.android/xbuild/Xamarin/Android/Mono.Android.DebugRuntime-debug.apk typemap.mj % strings typemap.jm | wc -l 10318 % strings typemap.mj | wc -l 11956 The reason why they have a different number of entries is *aliasing*: there can be *multiple* bindings for a given Java type. The managed-to-Java mappings can contain *all* of them; the Java-to-managed mapping *must* pick Only One. For example, [`java.util.ArrayList`][0] contains (at least?) *three* bindings: * [`Android.Runtime.JavaList`][1] * [`Android.Runtime.JavaList<T>`][2] * `Java.Util.ArrayList` (generated at build time) In a pre-ce2bc689 world, this was "fine": we'd binary search each file *separately* to find type mappings. This changed in ce2bc68, as the typemap information was merged into a *single* array in order to de-duplicate information. This reduced `.apk` size. An unanticipated result of this is that the Java-to-managed mappings can now contain *duplicate* keys, "as if" the original `typemap.jm` had the contents: java/util/ArrayList Android.Runtime.JavaList, Mono.Android java/util/ArrayList Android.Runtime.JavaList`1, Mono.Android java/util/ArrayList Java.Util.ArrayLlist, Mono.Android Whereas pre-ce2bc689 there would have only been the first entry. "Sometimes" this duplication is "fine": the typemap search "Just happens" to return the first entry, or the 3rd entry, and apps continue to work as intended. "Sometimes" it isn't: the typemap binary search finds the 2nd entry, which is a generic type. This results in attempting to instantiate a generic type *without appropriate type parameters*, which results in the aforementioned `MemberAccessException`. Update typemap generation code in `Xamarin.Android.Build.Tasks.dll` so that all the duplicate Java type names will point to the same managed type name. Additionally, make sure we select the managed type in the same fashion the old typemap generator in `Java.Interop` worked: prefer base types to the derived ones if the type is an interface or an abstract class. The effect of this change is that no matter which entry `EmbeddedAssemblies::binary_search()` ends up selecting it will always return the same managed type name for all aliased Java types. ~~ Release config apps ~~ Another issue introduced in ce2bc68 is that we no longer ignore interface and generic types when generating the Java-to-managed maps. This adversely affects the `Release` mode in which it is possible that an attempt to instantiate an open generic type (or an interface) will happen, leading to error similar to: FATAL EXCEPTION: main Process: com.companyname.androidsingleviewapp1, PID: 24068 android.runtime.JavaProxyThrowable: System.MemberAccessException: Cannot create an instance of Com.Contoso.Javalibrary1.Class0`1[T] because Type.ContainsGenericParameters is true. at System.Reflection.RuntimeConstructorInfo.DoInvoke (System.Object obj, System.Reflection.BindingFlags invokeAttr, System.Reflection.Binder binder, System.Object[] parameters, System.Globalization.CultureInfo culture) at System.Reflection.RuntimeConstructorInfo.Invoke (System.Reflection.BindingFlags invokeAttr, System.Reflection.Binder binder, System.Object[] parameters, System.Globalization.CultureInfo culture) at System.Reflection.ConstructorInfo.Invoke (System.Object[] parameters) at Java.Interop.TypeManager.CreateProxy (System.Type type, System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Java.Interop.TypeManager.CreateInstance (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer, System.Type targetType) at Java.Lang.Object.GetObject (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer, System.Type type) at Java.Lang.Object._GetObject[T] (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Java.Lang.Object.GetObject[T] (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Com.Contoso.Javalibrary1.Class1.Create () at AndroidSingleViewApp1.MainActivity.OnCreate (Android.OS.Bundle savedInstanceState) at Android.App.Activity.n_OnCreate_Landroid_os_Bundle_ (System.IntPtr jnienv, System.IntPtr native__this, System.IntPtr native_savedInstanceState) at (wrapper dynamic-method) Android.Runtime.DynamicMethodNameCounter.1(intptr,intptr,intptr) at crc647d7dcab8da8ee782.MainActivity.n_onCreate(Native Method) at crc647d7dcab8da8ee782.MainActivity.onCreate(MainActivity.java:29) at android.app.Activity.performCreate(Activity.java:7144) at android.app.Activity.performCreate(Activity.java:7135) at android.app.Instrumentation.callActivityOnCreate(Instrumentation.java:1271) at android.app.ActivityThread.performLaunchActivity(ActivityThread.java:2931) at android.app.ActivityThread.handleLaunchActivity(ActivityThread.java:3086) at android.app.servertransaction.LaunchActivityItem.execute(LaunchActivityItem.java:78) at android.app.servertransaction.TransactionExecutor.executeCallbacks(TransactionExecutor.java:108) at android.app.servertransaction.TransactionExecutor.execute(TransactionExecutor.java:68) at android.app.ActivityThread$H.handleMessage(ActivityThread.java:1816) at android.os.Handler.dispatchMessage(Handler.java:106) at android.os.Looper.loop(Looper.java:193) at android.app.ActivityThread.main(ActivityThread.java:6718) at java.lang.reflect.Method.invoke(Native Method) at com.android.internal.os.RuntimeInit$MethodAndArgsCaller.run(RuntimeInit.java:493) at com.android.internal.os.ZygoteInit.main(ZygoteInit.java:858) Update typemap generation code to also ignore generic types and interfaces when generating the Java to Managed map in `Release` mode. We must not simply skip outputting entries for such types because we need to maintain element count parity between java-to-managed and managed-to-java lookup tables. Therefore, the way we skip such types is to output a type token ID of `0` instead of the actual one. This will cause the runtime code to fail to find a mapping, thus allowing the managed code to deal with such a type properly. ~~ Test Updates ~~ Update `BundleTest.TestBundleIntegerArrayList2()` so that it asserts the runtime *type* of `obj` returned, asserting that it is in fact a `JavaList` and not e.g. `Java.Util.ArrayList` or something. (The linker could otherwise "change" things, and this assertion will ensure that `JavaList` won't be linked away…) Add a unit test for the Release config app bug in Issue #4660; thanks to @brendanzagaeski. It works by provoking the binary search bug by "ensuring" that a *generic* type comes *before* the "normally preferred" *non*-generic type. Thus, even if/when binary search returns the first entry, that would still be the *wrong entry*, because that'll be a generic type! [0]: https://developer.android.com/reference/java/util/ArrayList [1]: https://github.com/xamarin/xamarin-android/blob/61f09bf2ef0c8f09550e2d77eb97f417432b315b/src/Mono.Android/Android.Runtime/JavaList.cs#L11-L13 [2]: https://github.com/xamarin/xamarin-android/blob/61f09bf2ef0c8f09550e2d77eb97f417432b315b/src/Mono.Android/Android.Runtime/JavaList.cs#L667-L668
jonpryor
pushed a commit
that referenced
this issue
May 8, 2020
Fixes: #4596 Fixes: #4660 Context: xamarin/monodroid@192b1f1 Context: https://github.com/xamarin/java.interop/blob/fc18c54b2ccf14f444fadac0a1e7e81f837cc470/src/Java.Interop.Tools.JavaCallableWrappers/Java.Interop.Tools.JavaCallableWrappers/TypeNameMapGenerator.cs#L149-L186 Context: https://github.com/xamarin/java.interop/blob/010161d1ef6625b762429334f02e7b15e1f7caae/src/Java.Interop.Tools.JavaCallableWrappers/Java.Interop.Tools.JavaCallableWrappers/TypeNameMapGenerator.cs#L155 ~~ Debug Mode ~~ In some environments, `BundleTest.TestBundleIntegerArrayList2()` fails when using the commercial shared runtime: Test 'Xamarin.Android.RuntimeTests.BundleTest.TestBundleIntegerArrayList2' failed: System.MemberAccessException : Cannot create an instance of Android.Runtime.JavaList`1[T] because Type.ContainsGenericParameters is true. at System.Reflection.RuntimeConstructorInfo.DoInvoke (System.Object obj, System.Reflection.BindingFlags invokeAttr, System.Reflection.Binder binder, System.Object[] parameters, System.Globalization.CultureInfo culture) at System.Reflection.RuntimeConstructorInfo.Invoke (System.Reflection.BindingFlags invokeAttr, System.Reflection.Binder binder, System.Object[] parameters, System.Globalization.CultureInfo culture) at System.Reflection.ConstructorInfo.Invoke (System.Object[] parameters) at Java.Interop.TypeManager.CreateProxy (System.Type type, System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Java.Interop.TypeManager.CreateInstance (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer, System.Type targetType) at Java.Lang.Object.GetObject (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer, System.Type type) at Java.Lang.Object._GetObject[T] (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Java.Lang.Object.GetObject[T] (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Android.OS.Bundle.Get (System.String key) at Xamarin.Android.RuntimeTests.BundleTest.TestBundleIntegerArrayList2 () at (wrapper managed-to-native) System.Reflection.RuntimeMethodInfo.InternalInvoke(System.Reflection.RuntimeMethodInfo,object,object[],System.Exception&) at System.Reflection.RuntimeMethodInfo.Invoke (System.Object obj, System.Reflection.BindingFlags invokeAttr, System.Reflection.Binder binder, System.Object[] parameters, System.Globalization.CultureInfo culture) The problem appears to be rooted in ce2bc68. In a pre-ce2bc689 world, there were two separate sets of "typemap" files that the application could use: * Java-to-managed mappings, in `typemap.jm` * Managed-to-Java mappings, in `typemap.mj`. These files did *not* need to have the same number of entries! % unzip /Library/Frameworks/Xamarin.Android.framework/Versions/10.2.0.100/lib/xamarin.android/xbuild/Xamarin/Android/Mono.Android.DebugRuntime-debug.apk typemap.jm % unzip /Library/Frameworks/Xamarin.Android.framework/Versions/10.2.0.100/lib/xamarin.android/xbuild/Xamarin/Android/Mono.Android.DebugRuntime-debug.apk typemap.mj % strings typemap.jm | wc -l 10318 % strings typemap.mj | wc -l 11956 The reason why they have a different number of entries is *aliasing*: there can be *multiple* bindings for a given Java type. The managed-to-Java mappings can contain *all* of them; the Java-to-managed mapping *must* pick Only One. For example, [`java.util.ArrayList`][0] contains (at least?) *three* bindings: * [`Android.Runtime.JavaList`][1] * [`Android.Runtime.JavaList<T>`][2] * `Java.Util.ArrayList` (generated at build time) In a pre-ce2bc689 world, this was "fine": we'd binary search each file *separately* to find type mappings. This changed in ce2bc68, as the typemap information was merged into a *single* array in order to de-duplicate information. This reduced `.apk` size. An unanticipated result of this is that the Java-to-managed mappings can now contain *duplicate* keys, "as if" the original `typemap.jm` had the contents: java/util/ArrayList Android.Runtime.JavaList, Mono.Android java/util/ArrayList Android.Runtime.JavaList`1, Mono.Android java/util/ArrayList Java.Util.ArrayLlist, Mono.Android Whereas pre-ce2bc689 there would have only been the first entry. "Sometimes" this duplication is "fine": the typemap search "Just happens" to return the first entry, or the 3rd entry, and apps continue to work as intended. "Sometimes" it isn't: the typemap binary search finds the 2nd entry, which is a generic type. This results in attempting to instantiate a generic type *without appropriate type parameters*, which results in the aforementioned `MemberAccessException`. Update typemap generation code in `Xamarin.Android.Build.Tasks.dll` so that all the duplicate Java type names will point to the same managed type name. Additionally, make sure we select the managed type in the same fashion the old typemap generator in `Java.Interop` worked: prefer base types to the derived ones if the type is an interface or an abstract class. The effect of this change is that no matter which entry `EmbeddedAssemblies::binary_search()` ends up selecting it will always return the same managed type name for all aliased Java types. ~~ Release config apps ~~ Another issue introduced in ce2bc68 is that we no longer ignore interface and generic types when generating the Java-to-managed maps. This adversely affects the `Release` mode in which it is possible that an attempt to instantiate an open generic type (or an interface) will happen, leading to error similar to: FATAL EXCEPTION: main Process: com.companyname.androidsingleviewapp1, PID: 24068 android.runtime.JavaProxyThrowable: System.MemberAccessException: Cannot create an instance of Com.Contoso.Javalibrary1.Class0`1[T] because Type.ContainsGenericParameters is true. at System.Reflection.RuntimeConstructorInfo.DoInvoke (System.Object obj, System.Reflection.BindingFlags invokeAttr, System.Reflection.Binder binder, System.Object[] parameters, System.Globalization.CultureInfo culture) at System.Reflection.RuntimeConstructorInfo.Invoke (System.Reflection.BindingFlags invokeAttr, System.Reflection.Binder binder, System.Object[] parameters, System.Globalization.CultureInfo culture) at System.Reflection.ConstructorInfo.Invoke (System.Object[] parameters) at Java.Interop.TypeManager.CreateProxy (System.Type type, System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Java.Interop.TypeManager.CreateInstance (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer, System.Type targetType) at Java.Lang.Object.GetObject (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer, System.Type type) at Java.Lang.Object._GetObject[T] (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Java.Lang.Object.GetObject[T] (System.IntPtr handle, Android.Runtime.JniHandleOwnership transfer) at Com.Contoso.Javalibrary1.Class1.Create () at AndroidSingleViewApp1.MainActivity.OnCreate (Android.OS.Bundle savedInstanceState) at Android.App.Activity.n_OnCreate_Landroid_os_Bundle_ (System.IntPtr jnienv, System.IntPtr native__this, System.IntPtr native_savedInstanceState) at (wrapper dynamic-method) Android.Runtime.DynamicMethodNameCounter.1(intptr,intptr,intptr) at crc647d7dcab8da8ee782.MainActivity.n_onCreate(Native Method) at crc647d7dcab8da8ee782.MainActivity.onCreate(MainActivity.java:29) at android.app.Activity.performCreate(Activity.java:7144) at android.app.Activity.performCreate(Activity.java:7135) at android.app.Instrumentation.callActivityOnCreate(Instrumentation.java:1271) at android.app.ActivityThread.performLaunchActivity(ActivityThread.java:2931) at android.app.ActivityThread.handleLaunchActivity(ActivityThread.java:3086) at android.app.servertransaction.LaunchActivityItem.execute(LaunchActivityItem.java:78) at android.app.servertransaction.TransactionExecutor.executeCallbacks(TransactionExecutor.java:108) at android.app.servertransaction.TransactionExecutor.execute(TransactionExecutor.java:68) at android.app.ActivityThread$H.handleMessage(ActivityThread.java:1816) at android.os.Handler.dispatchMessage(Handler.java:106) at android.os.Looper.loop(Looper.java:193) at android.app.ActivityThread.main(ActivityThread.java:6718) at java.lang.reflect.Method.invoke(Native Method) at com.android.internal.os.RuntimeInit$MethodAndArgsCaller.run(RuntimeInit.java:493) at com.android.internal.os.ZygoteInit.main(ZygoteInit.java:858) Update typemap generation code to also ignore generic types and interfaces when generating the Java to Managed map in `Release` mode. We must not simply skip outputting entries for such types because we need to maintain element count parity between java-to-managed and managed-to-java lookup tables. Therefore, the way we skip such types is to output a type token ID of `0` instead of the actual one. This will cause the runtime code to fail to find a mapping, thus allowing the managed code to deal with such a type properly. ~~ Test Updates ~~ Update `BundleTest.TestBundleIntegerArrayList2()` so that it asserts the runtime *type* of `obj` returned, asserting that it is in fact a `JavaList` and not e.g. `Java.Util.ArrayList` or something. (The linker could otherwise "change" things, and this assertion will ensure that `JavaList` won't be linked away…) Add a unit test for the Release config app bug in Issue #4660; thanks to @brendanzagaeski. It works by provoking the binary search bug by "ensuring" that a *generic* type comes *before* the "normally preferred" *non*-generic type. Thus, even if/when binary search returns the first entry, that would still be the *wrong entry*, because that'll be a generic type! [0]: https://developer.android.com/reference/java/util/ArrayList [1]: https://github.com/xamarin/xamarin-android/blob/61f09bf2ef0c8f09550e2d77eb97f417432b315b/src/Mono.Android/Android.Runtime/JavaList.cs#L11-L13 [2]: https://github.com/xamarin/xamarin-android/blob/61f09bf2ef0c8f09550e2d77eb97f417432b315b/src/Mono.Android/Android.Runtime/JavaList.cs#L667-L668
|
@grendello I'm seeing a new startup crash with the DrawableTinting test case with this patch applied to d16-6: https://gist.github.com/pjcollins/87762e81f1f3c7e8b821356e4612eecf I'm worried about other potential fallout/risk on d16-6 with changes coming this late (unfortunately in part due to detection of this issue coming late in the cycle). Do we have a sane way to revert to d16-5 behavior here for the shorter time frame and look to address this more properly in d16-7 or d16-8? I understand this would carry a performance hit but I think it would be preferable to limit the scope of change in d16-6 at this point if possible. |
grendello
added a commit
to grendello/xamarin-android
that referenced
this issue
May 11, 2020
Fixes: dotnet#4596 Context: a017561 Context: https://gist.github.com/pjcollins/87762e81f1f3c7e8b821356e4612eecf A missing parameter in a call to `log_debug` added in a017561 may lead to a segfault when `assembly` log category and `debug` log level are enabled: F libc : Fatal signal 11 (SIGSEGV), code 1 (SEGV_MAPERR), fault addr 0x3 in tid 922 (DrawableTinting), pid 922 (DrawableTinting) I crash_dump64: obtaining output fd from tombstoned, type: kDebuggerdTombstone I /system/bin/tombstoned: received crash request for pid 922 I crash_dump64: performing dump of process 922 (target tid = 922) F DEBUG : *** *** *** *** *** *** *** *** *** *** *** *** *** *** *** *** F DEBUG : Build fingerprint: 'Android/sdk_phone_x86_64/generic_x86_64:9/PSR1.180720.012/4923214:userdebug/test-keys' F DEBUG : Revision: '0' F DEBUG : ABI: 'x86_64' F DEBUG : pid: 922, tid: 922, name: DrawableTinting >>> com.xamarin.DrawableTinting <<< F DEBUG : signal 11 (SIGSEGV), code 1 (SEGV_MAPERR), fault addr 0x3 F DEBUG : Cause: null pointer dereference F DEBUG : rax 0000000000000000 rbx 00007ffed3c283c0 rcx 0000000000000003 rdx 0000000000000002 F DEBUG : r8 00007ffed3c283c0 r9 00000000ffffffff r10 00007ffed3c283d0 r11 00007ffed3c28824 F DEBUG : r12 00007c88774a2f17 r13 00000000ffffffff r14 0000000000000000 r15 00007ffed3c283d0 F DEBUG : rdi 0000000000000003 rsi 00007ffed3c283bb F DEBUG : rbp 00007ffed3c28f18 rsp 00007ffed3c28288 rip 00007c890f860761 F DEBUG : F DEBUG : backtrace: F DEBUG : #00 pc 0000000000020761 /system/lib64/libc.so (strlen+17) F DEBUG : #1 pc 000000000006e761 /system/lib64/libc.so (__vfprintf+5953) F DEBUG : #2 pc 000000000008df5d /system/lib64/libc.so (vsnprintf+189) F DEBUG : #3 pc 0000000000007b60 /system/lib64/liblog.so (__android_log_vprint+64) F DEBUG : dotnet#4 pc 000000000001350c /data/app/com.xamarin.DrawableTinting-zvchh4ya_DW11GfpEPFICw==/lib/x86_64/libmonodroid.so (log_debug_nocheck(_LogCategories, char const*, ...)+204) F DEBUG : dotnet#5 pc 000000000000de6a /data/app/com.xamarin.DrawableTinting-zvchh4ya_DW11GfpEPFICw==/lib/x86_64/libmonodroid.so (xamarin::android::internal::EmbeddedAssemblies::typemap_java_to_managed(char const*)+538) F DEBUG : dotnet#6 pc 000000000000df13 /data/app/com.xamarin.DrawableTinting-zvchh4ya_DW11GfpEPFICw==/lib/x86_64/libmonodroid.so (xamarin::android::internal::EmbeddedAssemblies::typemap_java_to_managed(_MonoString*)+99) F DEBUG : dotnet#7 pc 00000000000d57f8 <anonymous:0000000042d04000> Add the missing parameter to prevent the `SIGSEGV` from happening.
jonpryor
pushed a commit
that referenced
this issue
May 11, 2020
…4673) Fixes: #4596 Context: a017561 Context: https://gist.github.com/pjcollins/87762e81f1f3c7e8b821356e4612eecf A missing parameter in a call to `log_debug` added iFixes: #4596 Context: a017561 Context: https://gist.github.com/pjcollins/87762e81f1f3c7e8b821356e4612eecf A missing parameter in a call to `log_debug()` added in a017561 may lead to a segfault when `assembly` log category and `debug` log level are enabled: F libc : Fatal signal 11 (SIGSEGV), code 1 (SEGV_MAPERR), fault addr 0x3 in tid 922 (DrawableTinting), pid 922 (DrawableTinting) I crash_dump64: obtaining output fd from tombstoned, type: kDebuggerdTombstone I /system/bin/tombstoned: received crash request for pid 922 I crash_dump64: performing dump of process 922 (target tid = 922) F DEBUG : *** *** *** *** *** *** *** *** *** *** *** *** *** *** *** *** F DEBUG : Build fingerprint: 'Android/sdk_phone_x86_64/generic_x86_64:9/PSR1.180720.012/4923214:userdebug/test-keys' F DEBUG : Revision: '0' F DEBUG : ABI: 'x86_64' F DEBUG : pid: 922, tid: 922, name: DrawableTinting >>> com.xamarin.DrawableTinting <<< F DEBUG : signal 11 (SIGSEGV), code 1 (SEGV_MAPERR), fault addr 0x3 F DEBUG : Cause: null pointer dereference F DEBUG : rax 0000000000000000 rbx 00007ffed3c283c0 rcx 0000000000000003 rdx 0000000000000002 F DEBUG : r8 00007ffed3c283c0 r9 00000000ffffffff r10 00007ffed3c283d0 r11 00007ffed3c28824 F DEBUG : r12 00007c88774a2f17 r13 00000000ffffffff r14 0000000000000000 r15 00007ffed3c283d0 F DEBUG : rdi 0000000000000003 rsi 00007ffed3c283bb F DEBUG : rbp 00007ffed3c28f18 rsp 00007ffed3c28288 rip 00007c890f860761 F DEBUG : F DEBUG : backtrace: F DEBUG : #00 pc 0000000000020761 /system/lib64/libc.so (strlen+17) F DEBUG : #1 pc 000000000006e761 /system/lib64/libc.so (__vfprintf+5953) F DEBUG : #2 pc 000000000008df5d /system/lib64/libc.so (vsnprintf+189) F DEBUG : #3 pc 0000000000007b60 /system/lib64/liblog.so (__android_log_vprint+64) F DEBUG : #4 pc 000000000001350c /data/app/com.xamarin.DrawableTinting-zvchh4ya_DW11GfpEPFICw==/lib/x86_64/libmonodroid.so (log_debug_nocheck(_LogCategories, char const*, ...)+204) F DEBUG : #5 pc 000000000000de6a /data/app/com.xamarin.DrawableTinting-zvchh4ya_DW11GfpEPFICw==/lib/x86_64/libmonodroid.so (xamarin::android::internal::EmbeddedAssemblies::typemap_java_to_managed(char const*)+538) F DEBUG : #6 pc 000000000000df13 /data/app/com.xamarin.DrawableTinting-zvchh4ya_DW11GfpEPFICw==/lib/x86_64/libmonodroid.so (xamarin::android::internal::EmbeddedAssemblies::typemap_java_to_managed(_MonoString*)+99) F DEBUG : #7 pc 00000000000d57f8 <anonymous:0000000042d04000> Add the missing parameter to prevent the `SIGSEGV` from happening.n a017561 may lead to a segfault when `assembly` log category and `debug` log level are enabled: F libc : Fatal signal 11 (SIGSEGV), code 1 (SEGV_MAPERR), fault addr 0x3 in tid 922 (DrawableTinting), pid 922 (DrawableTinting) I crash_dump64: obtaining output fd from tombstoned, type: kDebuggerdTombstone I /system/bin/tombstoned: received crash request for pid 922 I crash_dump64: performing dump of process 922 (target tid = 922) F DEBUG : *** *** *** *** *** *** *** *** *** *** *** *** *** *** *** *** F DEBUG : Build fingerprint: 'Android/sdk_phone_x86_64/generic_x86_64:9/PSR1.180720.012/4923214:userdebug/test-keys' F DEBUG : Revision: '0' F DEBUG : ABI: 'x86_64' F DEBUG : pid: 922, tid: 922, name: DrawableTinting >>> com.xamarin.DrawableTinting <<< F DEBUG : signal 11 (SIGSEGV), code 1 (SEGV_MAPERR), fault addr 0x3 F DEBUG : Cause: null pointer dereference F DEBUG : rax 0000000000000000 rbx 00007ffed3c283c0 rcx 0000000000000003 rdx 0000000000000002 F DEBUG : r8 00007ffed3c283c0 r9 00000000ffffffff r10 00007ffed3c283d0 r11 00007ffed3c28824 F DEBUG : r12 00007c88774a2f17 r13 00000000ffffffff r14 0000000000000000 r15 00007ffed3c283d0 F DEBUG : rdi 0000000000000003 rsi 00007ffed3c283bb F DEBUG : rbp 00007ffed3c28f18 rsp 00007ffed3c28288 rip 00007c890f860761 F DEBUG : F DEBUG : backtrace: F DEBUG : #00 pc 0000000000020761 /system/lib64/libc.so (strlen+17) F DEBUG : #1 pc 000000000006e761 /system/lib64/libc.so (__vfprintf+5953) F DEBUG : #2 pc 000000000008df5d /system/lib64/libc.so (vsnprintf+189) F DEBUG : #3 pc 0000000000007b60 /system/lib64/liblog.so (__android_log_vprint+64) F DEBUG : #4 pc 000000000001350c /data/app/com.xamarin.DrawableTinting-zvchh4ya_DW11GfpEPFICw==/lib/x86_64/libmonodroid.so (log_debug_nocheck(_LogCategories, char const*, ...)+204) F DEBUG : #5 pc 000000000000de6a /data/app/com.xamarin.DrawableTinting-zvchh4ya_DW11GfpEPFICw==/lib/x86_64/libmonodroid.so (xamarin::android::internal::EmbeddedAssemblies::typemap_java_to_managed(char const*)+538) F DEBUG : #6 pc 000000000000df13 /data/app/com.xamarin.DrawableTinting-zvchh4ya_DW11GfpEPFICw==/lib/x86_64/libmonodroid.so (xamarin::android::internal::EmbeddedAssemblies::typemap_java_to_managed(_MonoString*)+99) F DEBUG : #7 pc 00000000000d57f8 <anonymous:0000000042d04000> Add the missing parameter to prevent the `SIGSEGV` from happening.
jonpryor
pushed a commit
that referenced
this issue
May 12, 2020
…4673) Fixes: #4596 Context: a017561 Context: https://gist.github.com/pjcollins/87762e81f1f3c7e8b821356e4612eecf A missing parameter in a call to `log_debug` added iFixes: #4596 Context: a017561 Context: https://gist.github.com/pjcollins/87762e81f1f3c7e8b821356e4612eecf A missing parameter in a call to `log_debug()` added in a017561 may lead to a segfault when `assembly` log category and `debug` log level are enabled: F libc : Fatal signal 11 (SIGSEGV), code 1 (SEGV_MAPERR), fault addr 0x3 in tid 922 (DrawableTinting), pid 922 (DrawableTinting) I crash_dump64: obtaining output fd from tombstoned, type: kDebuggerdTombstone I /system/bin/tombstoned: received crash request for pid 922 I crash_dump64: performing dump of process 922 (target tid = 922) F DEBUG : *** *** *** *** *** *** *** *** *** *** *** *** *** *** *** *** F DEBUG : Build fingerprint: 'Android/sdk_phone_x86_64/generic_x86_64:9/PSR1.180720.012/4923214:userdebug/test-keys' F DEBUG : Revision: '0' F DEBUG : ABI: 'x86_64' F DEBUG : pid: 922, tid: 922, name: DrawableTinting >>> com.xamarin.DrawableTinting <<< F DEBUG : signal 11 (SIGSEGV), code 1 (SEGV_MAPERR), fault addr 0x3 F DEBUG : Cause: null pointer dereference F DEBUG : rax 0000000000000000 rbx 00007ffed3c283c0 rcx 0000000000000003 rdx 0000000000000002 F DEBUG : r8 00007ffed3c283c0 r9 00000000ffffffff r10 00007ffed3c283d0 r11 00007ffed3c28824 F DEBUG : r12 00007c88774a2f17 r13 00000000ffffffff r14 0000000000000000 r15 00007ffed3c283d0 F DEBUG : rdi 0000000000000003 rsi 00007ffed3c283bb F DEBUG : rbp 00007ffed3c28f18 rsp 00007ffed3c28288 rip 00007c890f860761 F DEBUG : F DEBUG : backtrace: F DEBUG : #00 pc 0000000000020761 /system/lib64/libc.so (strlen+17) F DEBUG : #1 pc 000000000006e761 /system/lib64/libc.so (__vfprintf+5953) F DEBUG : #2 pc 000000000008df5d /system/lib64/libc.so (vsnprintf+189) F DEBUG : #3 pc 0000000000007b60 /system/lib64/liblog.so (__android_log_vprint+64) F DEBUG : #4 pc 000000000001350c /data/app/com.xamarin.DrawableTinting-zvchh4ya_DW11GfpEPFICw==/lib/x86_64/libmonodroid.so (log_debug_nocheck(_LogCategories, char const*, ...)+204) F DEBUG : #5 pc 000000000000de6a /data/app/com.xamarin.DrawableTinting-zvchh4ya_DW11GfpEPFICw==/lib/x86_64/libmonodroid.so (xamarin::android::internal::EmbeddedAssemblies::typemap_java_to_managed(char const*)+538) F DEBUG : #6 pc 000000000000df13 /data/app/com.xamarin.DrawableTinting-zvchh4ya_DW11GfpEPFICw==/lib/x86_64/libmonodroid.so (xamarin::android::internal::EmbeddedAssemblies::typemap_java_to_managed(_MonoString*)+99) F DEBUG : #7 pc 00000000000d57f8 <anonymous:0000000042d04000> Add the missing parameter to prevent the `SIGSEGV` from happening.n a017561 may lead to a segfault when `assembly` log category and `debug` log level are enabled: F libc : Fatal signal 11 (SIGSEGV), code 1 (SEGV_MAPERR), fault addr 0x3 in tid 922 (DrawableTinting), pid 922 (DrawableTinting) I crash_dump64: obtaining output fd from tombstoned, type: kDebuggerdTombstone I /system/bin/tombstoned: received crash request for pid 922 I crash_dump64: performing dump of process 922 (target tid = 922) F DEBUG : *** *** *** *** *** *** *** *** *** *** *** *** *** *** *** *** F DEBUG : Build fingerprint: 'Android/sdk_phone_x86_64/generic_x86_64:9/PSR1.180720.012/4923214:userdebug/test-keys' F DEBUG : Revision: '0' F DEBUG : ABI: 'x86_64' F DEBUG : pid: 922, tid: 922, name: DrawableTinting >>> com.xamarin.DrawableTinting <<< F DEBUG : signal 11 (SIGSEGV), code 1 (SEGV_MAPERR), fault addr 0x3 F DEBUG : Cause: null pointer dereference F DEBUG : rax 0000000000000000 rbx 00007ffed3c283c0 rcx 0000000000000003 rdx 0000000000000002 F DEBUG : r8 00007ffed3c283c0 r9 00000000ffffffff r10 00007ffed3c283d0 r11 00007ffed3c28824 F DEBUG : r12 00007c88774a2f17 r13 00000000ffffffff r14 0000000000000000 r15 00007ffed3c283d0 F DEBUG : rdi 0000000000000003 rsi 00007ffed3c283bb F DEBUG : rbp 00007ffed3c28f18 rsp 00007ffed3c28288 rip 00007c890f860761 F DEBUG : F DEBUG : backtrace: F DEBUG : #00 pc 0000000000020761 /system/lib64/libc.so (strlen+17) F DEBUG : #1 pc 000000000006e761 /system/lib64/libc.so (__vfprintf+5953) F DEBUG : #2 pc 000000000008df5d /system/lib64/libc.so (vsnprintf+189) F DEBUG : #3 pc 0000000000007b60 /system/lib64/liblog.so (__android_log_vprint+64) F DEBUG : #4 pc 000000000001350c /data/app/com.xamarin.DrawableTinting-zvchh4ya_DW11GfpEPFICw==/lib/x86_64/libmonodroid.so (log_debug_nocheck(_LogCategories, char const*, ...)+204) F DEBUG : #5 pc 000000000000de6a /data/app/com.xamarin.DrawableTinting-zvchh4ya_DW11GfpEPFICw==/lib/x86_64/libmonodroid.so (xamarin::android::internal::EmbeddedAssemblies::typemap_java_to_managed(char const*)+538) F DEBUG : #6 pc 000000000000df13 /data/app/com.xamarin.DrawableTinting-zvchh4ya_DW11GfpEPFICw==/lib/x86_64/libmonodroid.so (xamarin::android::internal::EmbeddedAssemblies::typemap_java_to_managed(_MonoString*)+99) F DEBUG : #7 pc 00000000000d57f8 <anonymous:0000000042d04000> Add the missing parameter to prevent the `SIGSEGV` from happening.
jonpryor
pushed a commit
that referenced
this issue
May 12, 2020
…4673) Fixes: #4596 Context: a017561 Context: https://gist.github.com/pjcollins/87762e81f1f3c7e8b821356e4612eecf A missing parameter in a call to `log_debug` added iFixes: #4596 Context: a017561 Context: https://gist.github.com/pjcollins/87762e81f1f3c7e8b821356e4612eecf A missing parameter in a call to `log_debug()` added in a017561 may lead to a segfault when `assembly` log category and `debug` log level are enabled: F libc : Fatal signal 11 (SIGSEGV), code 1 (SEGV_MAPERR), fault addr 0x3 in tid 922 (DrawableTinting), pid 922 (DrawableTinting) I crash_dump64: obtaining output fd from tombstoned, type: kDebuggerdTombstone I /system/bin/tombstoned: received crash request for pid 922 I crash_dump64: performing dump of process 922 (target tid = 922) F DEBUG : *** *** *** *** *** *** *** *** *** *** *** *** *** *** *** *** F DEBUG : Build fingerprint: 'Android/sdk_phone_x86_64/generic_x86_64:9/PSR1.180720.012/4923214:userdebug/test-keys' F DEBUG : Revision: '0' F DEBUG : ABI: 'x86_64' F DEBUG : pid: 922, tid: 922, name: DrawableTinting >>> com.xamarin.DrawableTinting <<< F DEBUG : signal 11 (SIGSEGV), code 1 (SEGV_MAPERR), fault addr 0x3 F DEBUG : Cause: null pointer dereference F DEBUG : rax 0000000000000000 rbx 00007ffed3c283c0 rcx 0000000000000003 rdx 0000000000000002 F DEBUG : r8 00007ffed3c283c0 r9 00000000ffffffff r10 00007ffed3c283d0 r11 00007ffed3c28824 F DEBUG : r12 00007c88774a2f17 r13 00000000ffffffff r14 0000000000000000 r15 00007ffed3c283d0 F DEBUG : rdi 0000000000000003 rsi 00007ffed3c283bb F DEBUG : rbp 00007ffed3c28f18 rsp 00007ffed3c28288 rip 00007c890f860761 F DEBUG : F DEBUG : backtrace: F DEBUG : #00 pc 0000000000020761 /system/lib64/libc.so (strlen+17) F DEBUG : #1 pc 000000000006e761 /system/lib64/libc.so (__vfprintf+5953) F DEBUG : #2 pc 000000000008df5d /system/lib64/libc.so (vsnprintf+189) F DEBUG : #3 pc 0000000000007b60 /system/lib64/liblog.so (__android_log_vprint+64) F DEBUG : #4 pc 000000000001350c /data/app/com.xamarin.DrawableTinting-zvchh4ya_DW11GfpEPFICw==/lib/x86_64/libmonodroid.so (log_debug_nocheck(_LogCategories, char const*, ...)+204) F DEBUG : #5 pc 000000000000de6a /data/app/com.xamarin.DrawableTinting-zvchh4ya_DW11GfpEPFICw==/lib/x86_64/libmonodroid.so (xamarin::android::internal::EmbeddedAssemblies::typemap_java_to_managed(char const*)+538) F DEBUG : #6 pc 000000000000df13 /data/app/com.xamarin.DrawableTinting-zvchh4ya_DW11GfpEPFICw==/lib/x86_64/libmonodroid.so (xamarin::android::internal::EmbeddedAssemblies::typemap_java_to_managed(_MonoString*)+99) F DEBUG : #7 pc 00000000000d57f8 <anonymous:0000000042d04000> Add the missing parameter to prevent the `SIGSEGV` from happening.n a017561 may lead to a segfault when `assembly` log category and `debug` log level are enabled: F libc : Fatal signal 11 (SIGSEGV), code 1 (SEGV_MAPERR), fault addr 0x3 in tid 922 (DrawableTinting), pid 922 (DrawableTinting) I crash_dump64: obtaining output fd from tombstoned, type: kDebuggerdTombstone I /system/bin/tombstoned: received crash request for pid 922 I crash_dump64: performing dump of process 922 (target tid = 922) F DEBUG : *** *** *** *** *** *** *** *** *** *** *** *** *** *** *** *** F DEBUG : Build fingerprint: 'Android/sdk_phone_x86_64/generic_x86_64:9/PSR1.180720.012/4923214:userdebug/test-keys' F DEBUG : Revision: '0' F DEBUG : ABI: 'x86_64' F DEBUG : pid: 922, tid: 922, name: DrawableTinting >>> com.xamarin.DrawableTinting <<< F DEBUG : signal 11 (SIGSEGV), code 1 (SEGV_MAPERR), fault addr 0x3 F DEBUG : Cause: null pointer dereference F DEBUG : rax 0000000000000000 rbx 00007ffed3c283c0 rcx 0000000000000003 rdx 0000000000000002 F DEBUG : r8 00007ffed3c283c0 r9 00000000ffffffff r10 00007ffed3c283d0 r11 00007ffed3c28824 F DEBUG : r12 00007c88774a2f17 r13 00000000ffffffff r14 0000000000000000 r15 00007ffed3c283d0 F DEBUG : rdi 0000000000000003 rsi 00007ffed3c283bb F DEBUG : rbp 00007ffed3c28f18 rsp 00007ffed3c28288 rip 00007c890f860761 F DEBUG : F DEBUG : backtrace: F DEBUG : #00 pc 0000000000020761 /system/lib64/libc.so (strlen+17) F DEBUG : #1 pc 000000000006e761 /system/lib64/libc.so (__vfprintf+5953) F DEBUG : #2 pc 000000000008df5d /system/lib64/libc.so (vsnprintf+189) F DEBUG : #3 pc 0000000000007b60 /system/lib64/liblog.so (__android_log_vprint+64) F DEBUG : #4 pc 000000000001350c /data/app/com.xamarin.DrawableTinting-zvchh4ya_DW11GfpEPFICw==/lib/x86_64/libmonodroid.so (log_debug_nocheck(_LogCategories, char const*, ...)+204) F DEBUG : #5 pc 000000000000de6a /data/app/com.xamarin.DrawableTinting-zvchh4ya_DW11GfpEPFICw==/lib/x86_64/libmonodroid.so (xamarin::android::internal::EmbeddedAssemblies::typemap_java_to_managed(char const*)+538) F DEBUG : #6 pc 000000000000df13 /data/app/com.xamarin.DrawableTinting-zvchh4ya_DW11GfpEPFICw==/lib/x86_64/libmonodroid.so (xamarin::android::internal::EmbeddedAssemblies::typemap_java_to_managed(_MonoString*)+99) F DEBUG : #7 pc 00000000000d57f8 <anonymous:0000000042d04000> Add the missing parameter to prevent the `SIGSEGV` from happening.
|
Release status update A Preview version of Xamarin.Android has now been published on macOS that includes the fix for this item. The fix is not yet available on Windows. I will update this item again when a version with the fix is available on Windows. Fix included in Xamarin.Android 10.3.1.0. Fix included on macOS in Visual Studio 2019 for Mac version 8.6 Preview 6. To try the Preview version that includes the fix, check for the latest updates on the Preview updater channel. Fix not yet available on Windows. |
|
Release status update A new Release version of Xamarin.Android has now been published on both Windows and macOS that includes the fix for this item. Fix included in Xamarin.Android 10.3.1.0. Fix included on Windows in Visual Studio 2019 version 16.6. To get the new version that includes the fix, check for the latest updates or install the latest version from https://visualstudio.microsoft.com/downloads/. Fix included on macOS in Visual Studio 2019 for Mac version 8.6. To get the new version that includes the fix, check for the latest updates on the Stable updater channel. |
ghost
locked as resolved and limited conversation to collaborators
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
Steps to Reproduce
I tried dropping the test source into a regular dummy app and did not see the same crash however...
The text was updated successfully, but these errors were encountered: