P3L2

Question 1

Principles of Memory Management Systems

Answer 1

1. ) Use intelligently sized containers
2. ) Nota ll memory is needed at once
3. ) Optimized for performance

Question 2

Memory Management Goals

Answer 2

Allocate and Arbitrate

Question 3

Allocate

Answer 3

map pages from virt mem into page frames from physical

Question 4

Arbitrate

Answer 4

page tables

Question 5

MMU

Answer 5

• translate virtual to physical accesses

- report faults

Question 6

What causes faults

Answer 6

illegal access
permission
not present in mm

Question 7

HW Support Registers

Answer 7

Pointers to page table

base and limit size, number of segments

Question 8

Cache - TLB

Answer 8

Translation Lookaside Buffer

Question 9

VPN

Answer 9

Virtual Page Number

Question 10

PFN

Answer 10

Page Frame Number

Question 11

Page tables are per

Answer 11

process

Question 12

Page Table Flags

Answer 12

• Present (valid/invalid)
• Dirty
• Accessed
• R/W (permission bit)
• U/S (permission bit - user / supervisor)

Question 13

Page Fault when

Answer 13

H/W determines a physical mem access cannot be performed

Question 14

Page Fault Handler

Answer 14

determines action based on error code and faulting addr

Question 15

Hierarchical Page Tables

Answer 15

Outer page table == Page Table Directory

Internal page table == only for valid virtual memory regions

Question 16

Multi-level PT Tradeoffs

Answer 16

Smaller internal page tables / directories

but more memory accesses required for translation

Question 17

Page Table Cache

Answer 17

Translation Lookaside Buffer

MMU Level address translation cache

Question 18

Why small number of cached entry = high TLB hit rate

Answer 18

temporal and spacial locality

Question 19

Example of inverted page tables

Answer 19

Hashing Page Tables

Question 20

Memory allocator

Answer 20

determines VA to PA mapping

Question 21

Memory Allocation Challenges

Answer 21

permit coalescing / aggregation of free areas

avoid or limit extend of fragmentation

Question 22

Allocators in Linux

Answer 22

Buddy

Slab

Question 23

Buddy

Answer 23

Start with 2^x, split into powers of two until can satisfy request
-Aggregation works well and fast

Question 24

Slab allocator

Answer 24

-caches for common object type/sizes on top of contiguous memory

Question 25

Demand paging

Answer 25

virtual memory is much greater than physical memory | virtual memory page is not always in physical memory (swapping)

Question 26

Freeing up Physical Memory

Answer 26

when memory usage is above threshold (high watermark) | when CPU usage is below threshold (low watermark)

Question 27

LRU Policy

Answer 27

Least Recently Used

Question 28

What pages to swap out

Answer 28

- Least recently used (access-bit to track) - Pages that don't need to be written out (dirty bit) - Avoid non-swappable pages

Question 29

COW

Answer 29

Copy-on-write

Question 30

Checkpointing

Answer 30

1. write-protect and copy everything once | 2. copy diffs of "dirtied" pages for incremental checkpoints

Question 31

Debugging

Answer 31

Rewind-Replay | 1. Gradually go back to older checkpoints until problem found

Question 32

MIgration

Answer 32

Repeated checkpoints in a fast-loop until pause and copy becomes acceptable