This article provides a comprehensive overview of Google’s TCMalloc memory allocator, detailing its three‑level cache architecture, allocation and reclamation strategies, installation methods, configuration options, and performance advantages over other allocators in C++ backend, game, and database applications.
The article examines how Android’s Scudo memory allocator has been refined over the past three years, detailing its four design goals, the cache and thread‑specific data mechanisms for fast allocation, strategies to balance safe memory release with performance, and the group‑based approach that reduces fragmentation while preserving security.
This article explores the inner workings of Linux's glibc malloc, detailing how heap memory is requested from the kernel, how multithreaded support is implemented with per‑thread arenas, and the structures such as arenas, bins, and chunks that together determine allocation efficiency.
The Android Scudo hardened allocator, introduced in Android R to replace jemalloc, uses a primary region‑based allocator, a secondary mmap‑backed allocator, thread‑specific caches, and a quarantine system with 64‑bit chunk headers and extensive safety checks, offering a balanced security‑performance trade‑off configurable via compile‑time, environment, and mallopt options.
This article investigates why MySQL often consumes far more physical memory than the configured InnoDB buffer pool size, explains the high‑water‑mark effect, and explores Linux process memory layout, memory allocators, and MySQL’s internal memory management mechanisms.
