CS 152 Computer Architecture and
Engineering
Lecture 10 - Virtual Memory
Krste Asanovic
Electrical Engineering and Computer Sciences
University of California at Berkeley
http://www.eecs.berkeley.edu/~krste
http://inst.eecs.berkeley.edu/~cs152
Last time in Lecture 9
• Protection and translation required for
multiprogramming
– Base and bounds, early simple scheme
• Page-based translation and protection avoids need
for memory compaction, easy allocation by OS
– But need to indirect in large page table on every access
• Address spaces accessed sparsely
– Can use multi-level page table to hold translation/protection
information
• Address space access with locality
– Can use “translation lookaside buffer” (TLB) to cache address
translations (sometimes known as address translation cache)
– Still have to walk page tables on TLB miss, can be hardware or
software talk
• Virtual memory uses DRAM as a “cache” of disk
memory, allows very cheap main memory
2/28/2008
CS152-Spring’08
2
Modern Virtual Memory Systems
Illusion of a large, private, uniform store
Protection & Privacy
OS
several users, each with their private
address space and one or more
shared address spaces
page table  name space
Demand Paging
Provides the ability to run programs
larger than the primary memory
useri
Primary
Memory
Swapping
Store
Hides differences in machine
configurations
The price is address translation on
each memory reference
2/28/2008
CS152-Spring’08
VA
mapping
TLB
PA
3
Hierarchical Page Table
Virtual Address
31
22 21
p1
0
12 11
p2
offset
10-bit 10-bit
L1 index L2 index
offset
Root of the Current
Page Table
p2
p1
(Processor
Register)
Level 1
Page Table
page in primary memory
page in secondary memory
Level 2
Page Tables
PTE of a nonexistent page
2/28/2008
Data Pages
CS152-Spring’08
4
Address Translation & Protection
Virtual Address
Virtual Page No. (VPN)
offset
Kernel/User Mode
Read/Write
Protection
Check
Address
Translation
Exception?
Physical Address
Physical Page No. (PPN)
offset
• Every instruction and data access needs address
translation and protection checks
A good VM design needs to be fast (~ one cycle) and
space efficient
2/28/2008
CS152-Spring’08
5
Translation Lookaside Buffers
Address translation is very expensive!
In a two-level page table, each reference
becomes several memory accesses
Solution: Cache translations in TLB
TLB hit
TLB miss
 Single Cycle Translation
 Page Table Walk to refill
virtual address
VRWD
tag
PPN
VPN
offset
(VPN = virtual page number)
(PPN = physical page number)
hit?
2/28/2008
physical address
CS152-Spring’08
PPN
offset
6
Handling a TLB Miss
Software (MIPS, Alpha)
TLB miss causes an exception and the operating system
walks the page tables and reloads TLB. A privileged
“untranslated” addressing mode used for walk
Hardware (SPARC v8, x86, PowerPC)
A memory management unit (MMU) walks the page
tables and reloads the TLB
If a missing (data or PT) page is encountered during the
TLB reloading, MMU gives up and signals a Page-Fault
exception for the original instruction
2/28/2008
CS152-Spring’08
7
Translation for Page Tables
• Can references to page tables cause TLB misses?
• Can this go on forever?
User PTE Base
System PTE Base
User Page Table
(in virtual space)
System Page Table
(in physical space)
Data Pages
2/28/2008
CS152-Spring’08
8
Variable-Sized Page Support
Virtual Address
31
22 21
p1
12 11
p2
0
offset
10-bit 10-bit
L1 index L2 index
offset
Root of the Current
Page Table
p2
p1
(Processor
Register)
Level 1
Page Table
page in primary memory
large page in primary memory
page in secondary memory
PTE of a nonexistent page
Level 2
Page Tables
Data Pages
2/28/2008
CS152-Spring’08
9
Variable-Size Page TLB
Some systems support multiple page sizes.
virtual address
V R WD
Tag
PPN
VPN
offset
PPN
offset
L
hit?
physical address
2/28/2008
CS152-Spring’08
10
Address Translation:
putting it all together
Virtual Address
Restart instruction
hardware
hardware or software
software
TLB
Lookup
miss
hit
Protection
Check
Page Table
Walk
 memory
the page is
Page Fault
(OS loads page)
 memory
denied
Protection
Fault
Update TLB
permitted
Physical
Address
(to cache)
SEGFAULT
2/28/2008
CS152-Spring’08
11
Address Translation in CPU Pipeline
PC
Inst
TLB
Inst.
Cache
D
Decode
E
TLB miss? Page Fault?
Protection violation?
+
M
Data
TLB
Data
Cache
W
TLB miss? Page Fault?
Protection violation?
• Software handlers need restartable exception on page fault or
protection violation
• Handling a TLB miss needs a hardware or software mechanism to
refill TLB
• Need mechanisms to cope with the additional latency of a TLB:
– slow down the clock
– pipeline the TLB and cache access
– virtual address caches
– parallel TLB/cache access
2/28/2008
CS152-Spring’08
12
Virtual Address Caches
CPU
VA
PA
TLB
Physical
Cache
Primary
Memory
Alternative: place the cache before the TLB
VA
CPU
Virtual
Cache
TLB
PA
Primary
Memory (StrongARM)
• one-step process in case of a hit (+)
• cache needs to be flushed on a context switch unless address
space identifiers (ASIDs) included in tags (-)
• aliasing problems due to the sharing of pages (-)
• maintaining cache coherence (-) (see later in course)
2/28/2008
CS152-Spring’08
13
Aliasing in Virtual-Address Caches
VA1
Page Table
Data Pages
PA
VA2
Two virtual pages share
one physical page
Tag
Data
VA1
1st Copy of Data at PA
VA2
2nd Copy of Data at PA
Virtual cache can have two
copies of same physical data.
Writes to one copy not visible
to reads of other!
General Solution: Disallow aliases to coexist in cache
Software (i.e., OS) solution for direct-mapped cache
VAs of shared pages must agree in cache index bits; this
ensures all VAs accessing same PA will conflict in directmapped cache (early SPARCs)
2/28/2008
CS152-Spring’08
14
CS152 Administrivia
• Tuesday Mar 4, Quiz 2
– Memory hierarchy lectures L6-L8, PS 2, Lab 2
– In class, closed book
2/28/2008
CS152-Spring’08
15
Concurrent Access to TLB & Cache
VA
VPN
L
TLB
PA
PPN
b
k
Page Offset
Tag
Virtual
Index
=
hit?
Direct-map Cache
2L blocks
2b-byte block
Physical Tag
Data
Index L is available without consulting the TLB
cache and TLB accesses can begin simultaneously
Tag comparison is made after both accesses are completed
Cases: L + b = k
2/28/2008
L+b<k
CS152-Spring’08
L+b>k
16
Virtual-Index Physical-Tag Caches:
Associative Organization
VA
VPN
a
L = k-b
TLB
PA
k
PPN
Virtual
Index
2a
b
Direct-map
2L blocks
Direct-map
2L blocks
Phy.
Tag
Page Offset
=
Tag
hit?
a
After the PPN is known, 2 physical tags are compared
=
2a
Data
Is this scheme realistic?
2/28/2008
CS152-Spring’08
17
Concurrent Access to TLB & Large L1
The problem with L1 > Page size
Virtual Index
VA
VPN
a
Page Offset
b
TLB
PPN
PA
Page Offset
L1 PA cache
Direct-map
VA1 PPNa
Data
VA2 PPNa
Data
b
=
Tag
hit?
Can VA1 and VA2 both map to PA ?
2/28/2008
CS152-Spring’08
18
A solution via
CPU
RF
Second Level Cache
L1
Instruction
Cache
Memory
Unified L2
Cache
L1 Data
Cache
Memory
Memory
Memory
Usually a common L2 cache backs up both
Instruction and Data L1 caches
L2 is “inclusive” of both Instruction and Data caches
2/28/2008
CS152-Spring’08
19
Anti-Aliasing Using L2: MIPS R10000
Virtual Index
VA
VPN
TLB
PPN
PA
a
Page Offset
b
into L2 tag
Page Offset
VA1 PPNa
Data
VA2 PPNa
Data
b
PPN
Tag
•
•
•
Suppose VA1 and VA2 both map to PA and VA1 is
already in L1, L2 (VA1  VA2)
After VA2 is resolved to PA, a collision will be
detected in L2.
VA1 will be purged from L1 and L2, and VA2 will be
loaded  no aliasing !
2/28/2008
CS152-Spring’08
L1 PA cache
Direct-map
PA
=
a1
hit?
Data
Direct-Mapped L2
20
Virtually-Addressed L1:
Anti-Aliasing using L2
VA
VPN
Page Offset
Virtual
Index & Tag
b
TLB
PA
PPN
Tag
Page Offset
2/28/2008
VA2
Data
“Virtual
Tag”
Physical
Index & Tag
CS152-Spring’08
Data
L1 VA Cache
b
Physically-addressed L2 can also be
used to avoid aliases in virtuallyaddressed L1
VA1
PA
VA1
Data
L2 PA Cache
L2 “contains” L1
21
Page Fault Handler
• When the referenced page is not in DRAM:
– The missing page is located (or created)
– It is brought in from disk, and page table is updated
Another job may be run on the CPU while the first job waits
for the requested page to be read from disk
– If no free pages are left, a page is swapped out
Pseudo-LRU replacement policy
• Since it takes a long time to transfer a page
(msecs), page faults are handled completely in
software by the OS
– Untranslated addressing mode is essential to allow kernel
to access page tables
2/28/2008
CS152-Spring’08
22
Hierarchical Page Table
Virtual Address
31
22 21
p1
0
12 11
p2
offset
10-bit 10-bit
L1 index L2 index
offset
Root of the Current
Page Table
p2
p1
(Processor
Register)
Level 1
Page Table
page in primary memory
page in secondary memory
Level 2
Page Tables
PTE of a nonexistent page
2/28/2008
Data Pages
CS152-Spring’08
23
Swapping a Page of a Page Table
A PTE in primary memory contains
primary or secondary memory addresses
A PTE in secondary memory contains
only secondary memory addresses
 a page of a PT can be swapped out only
if none its PTE’s point to pages in the
primary memory
Why?__________________________________
2/28/2008
CS152-Spring’08
24
Atlas Revisited
• One PAR for each physical page
PAR’s
• PAR’s contain the VPN’s of the pages
resident in primary memory
PPN
• Advantage: The size is proportional to
the size of the primary memory
VPN
• What is the disadvantage ?
2/28/2008
CS152-Spring’08
25
Hashed Page Table:
Approximating Associative Addressing
VPN
d
Virtual Address
Page Table
PID
hash
Offset
+
PA of PTE
Base of Table
VPN PID PPN
• Hashed Page Table is typically 2 to 3 times
larger than the number of PPN’s to reduce
collision probability
• It can also contain DPN’s for some nonresident pages (not common)
• If a translation cannot be resolved in this table
then the software consults a data structure
that has an entry for every existing page
2/28/2008
CS152-Spring’08
VPN PID DPN
VPN PID
Primary
Memory
26
Global System Address Space
User
map
Level A
User
•
•
•
2/28/2008
Global
System
Address
Space
map
Physical
Memory
Level B
map
Level A maps users’ address spaces into the global space
providing privacy, protection, sharing etc.
Level B provides demand-paging for the large global system
address space
Level A and Level B translations may be kept in separate TLB’s
CS152-Spring’08
27
Hashed Page Table Walk:
PowerPC Two-level, Segmented Addressing
Seg ID
64-bit user VA
0
hashS
PA of Seg Table
per process
+
35
Global Seg ID
+
Page
51
hashP
2/28/2008
51
63
PA
0
[ IBM numbers bits
with MSB=0 ]
Offset
Hashed Segment Table
80-bit System VA
PA of Page Table
system-wide
Page
67
Offset
79
Hashed Page Table
PA
0
40-bit PA
CS152-Spring’08
27
PPN
39
Offset
28
Power PC: Hashed Page Table
VPN
hash
d
Offset
80-bit VA
+
PA of Slot
Page Table
VPN
VPN
PPN
Base of Table
•
•
•
•
Each hash table slot has 8 PTE's <VPN,PPN> that are
searched sequentially
If the first hash slot fails, an alternate hash function is
used to look in another slot
All these steps are done in hardware!
Hashed Table is typically 2 to 3 times larger than the
number of physical pages
The full backup Page Table is a software data structure
2/28/2008
CS152-Spring’08
Primary
Memory
29
Virtual Memory Use Today - 1
• Desktops/servers have full demand-paged virtual
memory
–
–
–
–
Portability between machines with different memory sizes
Protection between multiple users or multiple tasks
Share small physical memory among active tasks
Simplifies implementation of some OS features
• Vector supercomputers have translation and protection
but not demand-paging
• (Older Crays: base&bound, Japanese & Cray X1/X2: pages)
– Don’t waste expensive CPU time thrashing to disk (make jobs fit in
memory)
– Mostly run in batch mode (run set of jobs that fits in memory)
– Difficult to implement restartable vector instructions
2/28/2008
CS152-Spring’08
30
Virtual Memory Use Today - 2
• Most embedded processors and DSPs provide physical
addressing only
– Can’t afford area/speed/power budget for virtual memory support
– Often there is no secondary storage to swap to!
– Programs custom written for particular memory configuration in
product
– Difficult to implement restartable instructions for exposed architectures
2/28/2008
CS152-Spring’08
31
Acknowledgements
• These slides contain material developed and
copyright by:
–
–
–
–
–
–
Arvind (MIT)
Krste Asanovic (MIT/UCB)
Joel Emer (Intel/MIT)
James Hoe (CMU)
John Kubiatowicz (UCB)
David Patterson (UCB)
• MIT material derived from course 6.823
• UCB material derived from course CS252
2/28/2008
CS152-Spring’08
32