The document discusses memory hierarchy and management. It describes how memory is organized in a hierarchy from fastest and smallest registers to slower but larger magnetic disks and tapes. It also covers concepts like cache hits and misses, main memory, multiprogramming and partitioning memory between multiple processes, and memory protection through relocation and segmentation.

1. Gerenciamento de Mem´ria
o

2. Sonho dos usu´rios
a
Idealmente, a mem´ria deveria ser
o
• r´pida,
a
• de custo baixo,
• imensa e
• n˜o vol´til.
a a

3. Realidade para os usu´rios
a
Hierarquia de Mem´ria
o
Typical access time Typical capacity
1 nsec Registers <1 KB
2 nsec Cache 1 MB
10 nsec Main memory 64-512 MB
10 msec Magnetic disk 5-50 GB
100 sec Magnetic tape 20-100 GB

4. Hierarquia de Mem´ria
o
Registradores
• Internos ` CPU
a
• Extremamente r´pidos
a
• Otimiza¸oes de c´digo podem mover temporariamente
c˜ o
vari´veis para registradores.
a
• Programas podem dar palpites sobre o que deve ficar
armazenado nos registradores
register int r;

5. Hierarquia de Mem´ria
o
Cache
• Internos ou muito pr´ximos ` CPU
o a
• Divididos em linhas de cache
• Controlados por hardware
• Cache hit
• Cache miss

6. Hierarquia de Mem´ria
o
Mem´ria Principal
o
• Random Access Memory (RAM)
• Compromisso entre pre¸o e desempenho
c
• Armazenamento vol´til
a

7. Hierarquia de Mem´ria
o
Disco
Read/write head (1 per surface)
Surface 7
Surface 6
Surface 5
Surface 4
Surface 3
Direction of arm motion
Surface 2
Surface 1
Surface 0

8. Hierarquia de Mem´ria
o
Fitas magn´ticas
e
• Utilizadas para c´pias de seguran¸a (backups)
o c
• Armazenamento de grandes quantidades de dados
• Acesso seq¨encial
u

9. Hierarquia de Mem´ria
o
Outros tipos de mem´ria
o
• ROM (Read Only Memory)
– r´pida e barata
a
– bootstrap loader est´ gravado em ROM
a
• EEPROM (Electrically Erasable ROM)
– podem ser apagadas (erros podem ser corrigidos)
• CMOS
– dependem de uma bateria
– armazenam rel´gio e configura¸oes
o c˜

10. Monoprograma¸˜o
ca
0xFFF …
Operating Device
system in drivers in ROM
ROM
User
program User
program
User
program
Operating Operating
system in system in
RAM RAM
0 0 0
(a) (b) (c)
mainframes palmtops MS-DOS

11. Multiprograma¸˜o e parti¸oes fixas
ca c˜
Multiple
input queues
800K
Partition 4 Partition 4
700K
Partition 3 Single Partition 3
input queue
400K
Partition 2 Partition 2
200K
Partition 1 Partition 1
100K
Operating Operating
system system
0
(a) (b)

12. Modelagem de multiprograma¸˜o
ca
• Quantos processos colocar na mem´ria?
o
• Estimativa de uso da CPU
20% I/O Wait
50% I/O Wait
80% I/O Wait

13. Modelagem de multiprograma¸˜o
ca
• Quantos processos colocar na mem´ria?
o
• Aposta otimista: 5 processos 80

14. Modelagem de multiprograma¸˜o
ca
• Vis˜o probabil´
a ıstica
• Taxa de utiliza¸˜o da CPU = 1 - (0.8)n
ca
...
...
...
...
...

15. Modelagem de multiprograma¸˜o
ca
20% I/O wait
CPU utilization (in percent)
100
50% I/O wait
80
60 80% I/O wait
40
20
0 1 2 3 4 5 6 7 8 9 10
Degree of multiprogramming
Utiliza¸˜o da CPU = 1 - p
ca n

16. An´lise de Desempenho
a
CPU
Arrival minutes # Processes
Job time needed 1 2 3 4
1 10:00 4 CPU idle .80 .64 .51 .41
2 10:10 3 CPU busy .20 .36 .49 .59
3 10:15 2 CPU/process .20 .18 .16 .15
4 10:20 2
(a) (b)
2.0 .9 .8 .3 Job 1 finishes
1
Job 2 starts .9 .8 .3 .9 .1
2
.8 .3 .9
3
.3 .9 .1 .7
4
0
0 10 15 20 22 27.6 28.2 31.7
Time (relative to job 1's arrival)
(c)

17. Swapping
CPU
CPU scheduler
Arriving
job
Input
queue
Main
Memory
Memory Disk
Admission
scheduler scheduler

18. Swapping
Time
C C C C C
B B B B
A
A A A
D D D
Operating Operating Operating Operating Operating Operating Operating
system system system system system system system
(a) (b) (c) (d) (e) (f) (g)

19. Reloca¸˜o e Prote¸˜o
ca ca
• Reloca¸˜o: um programa deve poder rodar em endere¸os
ca c
f´
ısicos distintos.
• Prote¸˜o: um programa n˜o pode fazer acesso ` ´rea de
ca a aa
mem´ria reservada a outro programa.
o

20. Reloca¸˜o durante a carga
ca
• Todos os endere¸os precisam ser identificados e alterados
c
• N˜o resolve o problema da prote¸˜o
a ca

21. Bits de prote¸˜o
ca
1010
1010
1010
0011
0011
O PSW de cada processo deve conter os bits de prote¸˜o
ca

22. Registradores base e limite
Address
Registers
when
0xFFFFFFFF program 2
is running
User program Registers
Limit-2
and data when
program 1 User-2 data
is running Base-2
Limit Limit-2
User program Base-2 User-1 data
and data Limit-1
Limit-1 User program
Base Base-1 Base-1
Operating Operating
System System
0
(a) (b)

23. Espa¸o para crescimento
c
B-Stack
Room for growth
Room for growth
B-Data
B Actually in use
B-Program
A-Stack
Room for growth
Room for growth Address (hex)
A-Data FFFF
A Actually in use Stack
A-Program
Gap
Operating Operating
system system
Data
(a) (b) Text
0000

24. Compacta¸˜o de mem´ria
ca o
A A
— B
B C
— —
C —

25. Bitmaps e lista de livres
A B C D E
8 16 24
(a)
11111000 P 0 5 H 5 3 P 8 6 P 14 4
11111111
11001111
H 18 2 P 20 6 P 26 3 H 29 3 X
11111000
Hole Starts Length Process
at 18 2
(b) (c)

26. Atualiza¸˜o da lista
ca
Before X terminates After X terminates
(a) A X B becomes A B
(b) A X becomes A
(c) X B becomes B
(d) X becomes

27. Algoritmos para aloca¸˜o de mem´ria
ca o
• First fit
• Next fit
• Best fit
• Worst fit

28. Problema relacionado
Malloc, free e realloc
void *malloc(size_t size);
void free(void *ptr);
void *realloc(void *ptr, size_t size);
A aloca¸˜o de registros dentro de um processo tem
ca
problemas semelhantes:
• administra¸˜o de uma lista de livres
ca
• fragmenta¸˜o
ca

29. Programas muito grandes
O que fazer se um programa for muito grande para caber na
mem´ria?
o

30. Overlays
dados d1, d2, d3, d4, d5;
f1(); g1(); h1();
f2(); g2(); h2();
f3(); g3(); h3();
main() {
fase_1(); /* fun¸oes f e dados d1, d2, d3 */
c
fase_2(); /* fun¸oes g e dados d1, d2, d4 */
c
fase_3(); /* fun¸oes h e dados d1, d2, d5 */
c
}

31. Mem´ria Virtual
o
The CPU sends virtual
CPU addresses to the MMU
package
CPU
Memory Disk
management Memory controller
unit
Bus
The MMU sends physical
addresses to the memory
Memory Management Unit (MMU)

32. Pagina¸˜o
ca
Virtual
address
space
60K-64K X
56K-60K X Virtual page
52K-56K X
48K-52K X
44K-48K 7
40K-44K X
Physical
36K-40K 5 memory
32K-36K X address
28K-32K X 28K-32K
24K-28K X 24K-28K
20K-24K 3 20K-24K
16K-20K 4 16K-20K
12K-16K 0 12K-16K
8K-12K 6 8K-12K
4K-8K 1 4K-8K
0K-4K 2 0K-4K
Page frame

33. Mapeamento dos endere¸os
c
Outgoing
1 1 0 0 0 0 0 0 0 0 0 0 1 0 0 physical
address
(24580)
15 000 0
14 000 0
13 000 0
12 000 0
11 111 1
10 000 0
9 101 1
12-bit offset
Page 8 000 0
copied directly
table 7 000 0 from input
6 000 0 to output
5 011 1
4 100 1
3 000 1
2 110 1 110
1 001 1
Present/
0 010 1
absent bit
Virtual page = 2 is used
as an index into the
page table Incoming
virtual
address
0 0 1 0 0 0 0 0 0 0 0 0 0 1 0 0
(8196)

34. Espa¸o de endere¸amento
c c
• Apenas as p´ginas ocupadas precisam ser mapeadas
a
Address (hex)
FFFF
Stack
Gap
Data
Text
0000

35. Mem´ria compartilhada
o
Processo A Processo B
Pilha Pilha
Shmem
Shmem
Data Data
Texto Texto

36. Mem´ria compartilhada
o
int shmget(key_t key, size_t size, int shmflg);
void *shmat(int shmid,
const void *shmaddr, int shmflg);
• Veja os exemplos: sh1.c sh2,c sh fork.c