Overview
Explore memory management in Linux architecture-specific code through this Linux Plumbers Conference talk. Dive into the similarities and duplications across different architectures, including page table allocation, early memory allocation, and initialization routines. Learn about the challenges of factoring out common code and refining the architecture-memory management interface. Discover potential improvements in areas such as per-CPU initialization, memory topology handling, crash kernel reservation, and vDSO mapping. Gain insights into cleaning up redundant code, streamlining page management, and addressing architecture-specific memory models and callbacks. Examine topics like flat memory handling, reservations, and page table management, with a focus on x86 architecture.
Syllabus
Introduction
How much do we really need
Cleanups
Page management
Memory models
Callback
Memory
Set apart
Memory initialization
Memory needs
Assumptions
Memory detection
Memory conversion
Page locator
Hotplug
Flat Memory
Reservations
Architecture specific
Page tables
Pigeon
x86
Taught by
Linux Plumbers Conference