The project is initialy motivated by EPITA to understand how allocators work and what types of algorithms we can use for certain type of allocations. In order to implement the library in a kernel, I decided to implement the standard malloc(3) with a fast, secure and reliable algorithm which is really used in modern systems libraries (like the glibc) : the SLAB allocator method.
Currently, the project support the following features :
- Reliable single threaded allocations.
- Faster allocation using cached metadata.
- User-space metadata corruption safety.
In order to include this malloc(3) library in a micro-kernel, we are working on the following features :
- Multi-threaded allocations.
-
aligned_alloc(),malloc_usable_size(),memalign()andposix_memalign()support. -
long doubleaddress alignment for all objects (currently supporting objects upper or equal thansizeof(long double)bytes.
To compile the dynamic library, you need to use the following command :
make library
Note that you can also use the
make.
To compile the library in debug mode (debug symbols + assert(3) enabled), you need to use the following command :
make library_debug
Use make clean command to clean the compiled objects.
Once compiled as libmalloc.so, you can use this malloc implementation on your programs using the LD_PRELOAD bash environement variable :
LD_PRELOAD=<path/to>/libmalloc.so ./<your program>Where <path/to> is the path where libmalloc.so is and where <your program> is the name (path included) of the program you want to execute.
This way, <your program> will preload the libmalloc.so functions before loading the system's one. Thus libmalloc.so functions will be used instead of the system's ones.
To understand the implementation, you should read the IMPLEMENTATION.md file. If you want to contribute to the project and want to have a better understanding of the architecture, you should read the ARCHITECTURE.md file.