Does Android abstract the device architecture? - android

I'm new to android programming, I was reading the answers on this question Why shouldn't an Android app be written in C/C++ because you "simply prefer to program in C/C++"?
in the first answer by Devunwired. he mentioned that "your native code must be built into .so files (one for armv5, armv7 and x86) all packaged up into the same APK. This duplication of executable code makes your app 3x the size (i.e. a "fat binary") unless you take on the task of building separate APKs for each architecture when you distribute the application"
My question is, I always thought that Operating Systems in general provide an abstraction over the underlying architecture. So if I want to deploy a c/c++ program on different environments, I need to re-compile the source code using a compiler written for different operating systems, this operating system however might have different versions that support different architecture.
If I'm right then why the case is different when it comes to Android?

At some point the operating system hands the code over to the CPU. The binary must have instructions for that particular CPU. With Java each OS and CPU makes the same virtual machine available so you don't have the same considerations.

Related

Can a shared library (.so) file built for linux be included/linked and used in an Android application?

I am working on a project in which I need to include NGSpice simulation library in an Android application and of course be able to use it.
I tried including the NGSpice windows DLL in my android application using SWIG and Android NDK, but it turned out that it is not even possible, so now I started to think about building NGSpice as a shared library for linux.
And now my question is, can I use the linux shared library for NGSpice as it is in my Android application, or does it need to built differently somehow to work on my Android application.
Thanks.
No. Android typically has two key differences from a traditional linux:
1) It uses the Bionic C library and dynamic linker, instead of a more traditional set such as glibc.
2) Android is typically run on an ARM or 32-bit x86 processor (or in rare cases MIPS), while your desktop Linux library might be either 64-bit or 32-bit x86 code.
If you build a .so compatible with the architecture of the machine and with bionic's system library functions and dynamic linker, then it should be workable.
Alternatively, if you have something for a compatible architecture but the wrong libc, it may be possible to write your own loader to get it into memory in working form on a secured device, or on a rooted device it is possible to run a more traditional linux userspace (typically Debian derived) in a chroot. But neither of these would be easy to integrate into an Android application - for the latter you'd almost definitely have to pass work over via interprocess communication, and that might prove easier in the former case as well.
Your only really endorsed solution is to rebuild the library from source, using either the ndk build system, or an ndk-generated "stand alone toolchain" and the library's current build system.

Developing native C applications android without the NDK?

OK this is a strange one:
Is there a way someone can develop native C applications or libraries for Android without using the Android NDK?
What was happening before the NDK was released?
(It's not there for too long, I think it was released only one or two years ago).
Apparently, you can -- a friend of mine is a real Android guru and he managed to build a GCC-based native toolchain entirely by hand. He also fixed some missing parts in Android's libc. The main idea is the following: GCC has builtin support for the arm-elf-linux target, so with an appropriate build script, you can configure it to build for Android. However, you have to root the phone to run the resulting binaries. One even cooler thing is that because GCC is a self-hosting compiler, with the arm-linux-elf toolchain, you can recompile GCC once again and have the toolchain on the phone itself.
Before the NDK was released, the only officially supported way of developing Android applications was to use the Android SDK and writing your applications in Java.
As others have mentioned, it's possible to cross-compile some applications as completely stand-alone and run them on a rooted phone. The downsides of this should be obvious: very few people will be able to run your application (they also need to be root, plus you won't be able to get your application up on the Play store); and you might even run into compatibility problems yourself across different devices e.g. if you rely on dynamic linking against various libraries (which you might need to do in order to keep the size of the binary down).
TL;DR: it's possible, but severly limited, and not recommended.
You can compile your C code with an Android-compatible toolchain (such as CodeSourcery) and run it on a non-rooted phone, from its command line (for example through an SSH connection like SSHDroid).

Why porting Android to x86 is difficult?

I'd like to know the technical difficulty in porting Android to x86 architecture.
Since its source is open, what is the main difficulty in getting to run on x86?
Or is it correct to ask why it cannot be compiled to a Java bytecode to run on a JVM on a PC? Is it because of the Dalvik VM?
I'd appreciate if somebody could explain this.
Thank you.
Actually, it is already ported to x86: http://www.android-x86.org/
Also, Android does not run in a JVM. The Android Kernel is a modified Linux Kernel and written in C. You can't compile it to run in a JVM.
The DVM (used by Android) is a modified JVM based on the OpenSource JVM-implementation Apache Harmony. All Java-Applications on Android run in a DVM.
There isn't a problem. VirtualBox can run an android OS guest machine just fine.
The main difficulty is that x86 and ARM are totally different architectures. They have a totally different mentality about them, different instructions, different registers, different behaviours, different memory architecture, etc. Even the way they interface with other hardware is different.
In terms of technical difficulty, the architectures don't have instructions or mechanisms that are compatible with each other, so the behaviour that is relied upon by a compiler for ARM does not exist on x86, and vice versa.
Knowing enough about one of them to implement anything major usually means you've not spent much time dealing with the other, which just adds to the difficulty.

What is the Android Native Development Kit (NDK)?

What is the Android NDK (native development kit) ? How can one use it? Why should one use it?
The NDK (Native Development Kit) is a tool that allows you to program in C/C++ for Android devices. It's intended to integrate with the SDK (it's described as a "companion tool") and used only for performance-critical portions of a project. See here for more information.
NDK may improve application performance. This is usually true for many
processor-bound applications. Many multimedia applications and video games use
native code for processor-intensive tasks.
The performance improvements can come from three sources. Firstly, the native code is compiled to a binary code and run directly on OS, while Java code is translated into Java
byte-code and interpreted by Dalvik Virtual Machine (VM). At Android 2.2 or higher,
a Just-In-Time (JIT) compiler is added to Dalvik VM to analyze and optimize the Java
byte-code while the program is running (for example, JIT can compile a part of the
byte-code to binary code before its execution). But in many cases, native code still
runs faster than Java code.
Java code is run by Dalvik VM on Android. Dalvik VM is specially designed
for systems with constrained hardware resources (memory space, processor
speed, and so on).
The second source for performance improvements at NDK is that native code allows
developers to make use of some processor features that are not accessible at Android SDK,
such as NEON, a Single Instruction Multiple Data (SIMD) technology, allowing multiple
data elements to be processed in parallel. One particular coding task example is the color
conversion for a video frame or a photo. Suppose we are to convert a photo of 1920x1280
pixels from the RGB color space to the YCbCr color space. The naive approach is to apply a
conversion formula to every pixel (that is, over two million pixels). With NEON, we can process multiple pixels at one time to reduce the processing time.
The third aspect is that we can optimize the critical code at an assembly level, which is a
common practice in desktop software development.
Disadvantage
NDK cannot access lots of APIs available in the Android SDK directly, and developing in NDK will always introduce extra complexity
into your application.
The Android NDK is a companion tool used only in conjunction with Android SDK which allows application developers to build performance-critical portions of their apps by use of native (C/C++) code.
This provide benefits in form of reuse of existing code and increased speed.
Please go through below links.
Link-1
Link-2
Link-3
The Android NDK is a companion tool to the Android SDK that lets you build performance-critical portions of your apps in native code. It provides headers and libraries that allow you to build activities, handle user input, use hardware sensors, access application resources, and more, when programming in C or C++. If you write native code, your applications are still packaged into an .apk file and they still run inside of a virtual machine on the device. The fundamental Android application model does not change.
The following links also answers your question:
What is NDK?
When to Develop in Native Code
NDK Download
How to build NDK app
how to work with NDK
10 tips for Android NDK
The Android NDK is a toolset that lets you embed components that make
use of native code in your Android applications.
Android applications run in the Dalvik virtual machine. The NDK allows
you to implement parts of your applications using native-code
languages such as C and C++. This can provide benefits to certain
classes of applications, in the form of reuse of existing code and in
some cases increased speed.
Source: http://developer.android.com/sdk/ndk/overview.html
The Android NDK is a companion tool to the Android SDK that lets you
build performance-critical portions of your apps in native code. It
provides headers and libraries that allow you to build activities,
handle user input, use hardware sensors, access application resources,
and more, when programming in C or C++. If you write native code, your
applications are still packaged into an .apk file and they still run
inside of a virtual machine on the device. The fundamental Android
application model does not change.
Source: http://developer.android.com/sdk/ndk/index.html
NDK is just a set of tools which lets you to write C/C++ codes for your application.For example suppose you want to add a critical function/performance to your app and you want to write it in C/C++ then eclipse or any other IDE will not allow you to write your C/C++ and in that case you have to use NDK and integrate it in your app.
NDK is a toolset that allows you to implement parts of your app using native-code languages such as C and C++....Checkout this https://developer.android.com/tools/sdk/ndk/index.html
Android NDK (native development kit)
Android Native Development Kit (NDK) is developers to write code in C/C++ that compiles to native code
Why should one use it?
The source code is compiled directly into machine code for the CPU (and not into an intermediate language, as with Java) then developers are able to get the best performance
How can one use it?
Here best tutorials
https://developer.android.com/ndk/index.html
https://www.androidauthority.com/android-ndk-everything-need-know-677642/
https://www.ntu.edu.sg/home/ehchua/programming/android/Android_NDK.html

can native only code run on android as standalone application

We want to migrate a huge complex native program to Android system ,running it as a background service accepting command sent from Java Program using JNI along with IPC. However, the Android NDK state following words:
Please note that the NDK does not enable you to develop native-only applications. Android's primary runtime remains the Dalvik virtual machine.
Does that mean we have no way to run an standalone native-only application on Android as a background service? The native code can only exist in the form of library that will be loaded to the virtual machine through JNI?
The NDK itself is only for creating libraries, though if you do some web searching you will find that there are at least two sets of wrapper scripts or instructions for (ab)using its toolchain to make standalone executables linked against android's bionic libc (something you would not get from a non-android arm toolchain).
The google folks do not encourage people to do this. Unfortunately, their vision of android only includes java applications, with optional native libraries in support - it does not include any "stable" means of installing or launching a native executable, in the sense that they warn the methods you might be able to use today may not continue to work in new versions. This is really too bad, as it means giving up a lot of the general-purpose-computer potential of the device.
Well, it can be done. But to be honest i've never tried it using the NDK, but i've managed to create native applications using the toolchains provided with the android source code.
Your phone (incase your talking about phones) should be rooted.

Categories

Resources