unicorn/docs/COMPILE-NIX.md
2016-04-30 13:05:35 +08:00

170 lines
4.6 KiB
Markdown

This documentation explains how to compile, install & run Unicorn on MacOSX,
Linux, BSD, Solaris, Android & iOS.
To compile for Microsoft Windows, see [COMPILE-WINDOWS.md](COMPILE-WINDOWS.md)
----
[0] Dependencies
Unicorn requires few dependent packages as follows.
- For Mac OS X, "pkg-config" and "glib" are needed.
Brew users can install "pkg-config" and "glib" with:
$ brew install pkg-config glib
- For Linux, "glib2-dev" is needed.
Ubuntu/Debian users can install this with:
$ sudo apt-get install libglib2.0-dev
[1] Tailor Unicorn to your need.
Out of 6 archtitectures supported by Unicorn (Arm, Arm64, M68K, Mips, Sparc,
& X86), if you just need several selected archs, choose which ones you want
to compile in by editing "config.mk" before going to next steps.
By default, all 6 architectures are compiled.
The other way of customize Unicorn without having to edit config.mk is to
pass the desired options on the commandline to ./make.sh. Currently,
Unicorn supports 4 options, as follows.
- UNICORN_ARCHS: specify list of architectures to compiled in.
- UNICORN_STATIC: build static library.
- UNICORN_SHARED: build dynamic (shared) library.
- UNICORN_QEMU_FLAGS: specify extra flags for qemu's configure script
To avoid editing config.mk for these customization, we can pass their values to
make.sh, as follows.
$ UNICORN_ARCHS="arm aarch64 x86" ./make.sh
NOTE: on commandline, put these values in front of ./make.sh, not after it.
For each option, refer to docs/README for more details.
[2] Compile and install from source on *nix
To build Unicorn on *nix (such as MacOSX, Linux, *BSD, Solaris):
- To compile for current platform, run:
$ ./make.sh
- Unicorn requires Python 2.x to compile. If Python 2.x is not the default
Python interpreter, ensure that the appropriate option is set:
$ UNICORN_QEMU_FLAGS="--python=/path/to/python2" ./make.sh
- To cross-compile Unicorn on 64-bit OS to target 32-bit binary, run:
$ ./make.sh nix32
After compiling, install Unicorn with:
$ sudo ./make.sh install
For FreeBSD/OpenBSD, where sudo is unavailable, run:
$ su; ./make.sh install
Users are then required to enter root password to copy Unicorn into machine
system directories.
Afterwards, run ./samples/sample_all.sh to test the sample emulations.
NOTE: The core framework installed by "./make.sh install" consist of
following files:
/usr/include/unicorn/unicorn.h
/usr/include/unicorn/x86.h
/usr/include/unicorn/arm.h
/usr/include/unicorn/arm64.h
/usr/include/unicorn/mips.h
/usr/include/unicorn/ppc.h
/usr/include/unicorn/sparc.h
/usr/include/unicorn/m68k.h
/usr/lib/libunicorn.so (for Linux/*nix), or /usr/lib/libunicorn.dylib (OSX)
/usr/lib/libunicorn.a
[3] Cross-compile for iOS from Mac OSX.
To cross-compile for iOS (iPhone/iPad/iPod), Mac OSX with XCode installed is required.
- To cross-compile for ArmV7 (iPod 4, iPad 1/2/3, iPhone4, iPhone4S), run:
$ ./make.sh ios_armv7
- To cross-compile for ArmV7s (iPad 4, iPhone 5C, iPad mini), run:
$ ./make.sh ios_armv7s
- To cross-compile for Arm64 (iPhone 5S, iPad mini Retina, iPad Air), run:
$ ./make.sh ios_arm64
- To cross-compile for all iDevices (armv7 + armv7s + arm64), run:
$ ./make.sh ios
Resulted files libunicorn.dylib, libunicorn.a & tests/test* can then
be used on iOS devices.
[4] Cross-compile for Android
To cross-compile for Android (smartphone/tablet), Android NDK is required.
NOTE: Only ARM and ARM64 are currently supported.
$ NDK=/android/android-ndk-r10e ./make.sh cross-android arm
or
$ NDK=/android/android-ndk-r10e ./make.sh cross-android arm64
Resulted files libunicorn.so, libunicorn.a & tests/test* can then
be used on Android devices.
[5] By default, "cc" (default C compiler on the system) is used as compiler.
- To use "clang" compiler instead, run the command below:
$ ./make.sh clang
- To use "gcc" compiler instead, run:
$ ./make.sh gcc
[6] To uninstall Unicorn, run the command below:
$ sudo ./make.sh uninstall
[7] Language bindings
Look for the bindings under directory bindings/, and refer to README file
of corresponding languages.
[8] Unit tests
Automated unit tests use the cmocka unit testing framework (https://cmocka.org/).
It can be installed in most Linux distros using the package manager, e.g.
`sudo yum install libcmocka libcmocka-devel`, or you can easily build and install it from source.
You can run the tests by running `make test` in the project directory.