Kernel Memory: Slabs and the SLUB Allocator

• User space has malloc; the kernel has kmalloc, backed by the SLUB slab allocator (the default modern allocator).
• Objects are grouped by size into caches: general-purpose kmalloc-32, kmalloc-64, kmalloc-256, ... plus dedicated caches for hot structs (e.g. cred, files_struct).
• Freed objects are pushed on a per-cache freelist and handed back on the next same-size allocation — often in LIFO order, which attackers exploit for predictable reuse.
• Heap grooming / feng shui: allocate and free in a chosen pattern so a freed victim object is replaced by an attacker-shaped object of the same size class.