linear-malloc

DISCLAIMER! This is a quick hack I did to test things. Don't use!

An ultra simple single-threaded linear allocator.

Useful to make the output of applications running under Cachegrind/Callgrind more stable (since it doesn't try to do the clever optimizations that the system allocator does).

Usage

Linux:

cargo build
LD_PRELOAD=./target/debug/liblinear_malloc.so your-binary

macOS:

cargo build
DYLD_INSERT_LIBRARIES=./target/debug/liblinear_malloc.dylib DYLD_FORCE_FLAT_NAMESPACE=1 your-binary

Acknowledgements

Got some useful info about allocation strategies from @mtrebi's memory-allocators, and @ezrosent's "Allocators in Rust" was a good source to help set up the basic framework.