+ All Categories
Home > Documents > Coverage Metrics Lecture - University of California, Berkeley · coverage metrics in formal...

Coverage Metrics Lecture - University of California, Berkeley · coverage metrics in formal...

Date post: 17-Oct-2019
Category:
Upload: others
View: 5 times
Download: 0 times
Share this document with a friend
45
1 Coverage Metrics Wenchao Li EECS 219C UC Berkeley
Transcript

1

Co

ve

rag

e M

etr

ics

We

nch

ao

Li

EE

CS

21

9C

UC

Be

rke

ley

2

Outlin

e o

f th

e lectu

re

•W

hy d

o w

e n

eed c

overa

ge m

etr

ics?

•C

rite

ria for

a g

ood c

overa

ge m

etr

ic.

•D

iffe

rent appro

aches to d

efine c

overa

ge

metr

ics.

•D

iffe

rent ty

pes o

f covera

ge m

etr

ics.

3

A d

iffe

rent

kin

d o

f covera

ge

4

Why d

o w

e n

eed C

overa

ge

Metr

ics?

~re

q

ack

LT

L s

pecific

ation:

Assert

G (

req→

F a

ck);

Extr

em

e C

ase:

0specific

ation

!

Ok,

so s

uppose n

ow

we k

no

w

we n

ee

d m

ore

specific

ations,

but

do w

e k

no

w w

hat

specific

ations t

o w

rite

?

Ante

cede

nt

Failu

re

The o

ther

extr

em

e?

1 s

pec f

or

every

sta

te t

ransitio

n

A s

imp

le E

xam

ple

:V

acuity

Dete

ction

5

Why d

o w

e n

eed C

overa

ge

Metr

ics?

In g

enera

l:•

specs a

re n

ot

necessari

ly c

om

ple

te;

•nee

d t

o p

rom

pt/

assis

t hard

ware

/soft

ware

teste

rs;

•tr

adeoff b

etw

een c

ost of

pro

vid

ing c

overa

ge

and

perf

orm

ance/r

elia

bili

ty;

•shou

ld w

e h

ave a

sin

gle

covera

ge m

etr

ics o

r m

any

app

lication

-depe

nd

ent

covera

ge m

etr

ics?

•covera

ge m

etr

ics in s

imu

lation-b

ased v

erification →

covera

ge m

etr

ics in form

al verification.

6

What

are

th

e c

rite

ria f

or

a g

ood

covera

ge m

etr

ics?

•D

irect

co

rresp

on

den

ce w

ith

bu

gs.

Question 1

: checkin

g incom

ple

teness o

f specific

ations ≠

findin

g

redundancie

s in the s

yste

m?

•R

easo

nab

le c

om

pu

tati

on

al an

d h

um

an

eff

ort

to

:(a

) com

pute

the m

etr

ics;

(b)

inte

rpre

t covera

ge d

ata

and g

enera

te s

tim

uli

to e

xerc

ise

uncovere

d a

spects

;(c

) achie

ve h

igh c

overa

ge;

Question 2

: 100%

covera

ge =

com

ple

te d

esig

n?

(d)

min

imal m

odific

ation to v

alid

ation fra

mew

ork

.•

Kn

ow

led

ge o

f th

e d

esig

n r

eq

uir

ed

?–

Covera

ge M

etr

ics for

Bla

ckbox

Testing.

[M. W

. W

hale

n e

t. a

l. 2

006] [A

jitha

Raja

n2006]

•“O

bserv

ab

ilit

y”

–com

ple

te s

pecific

ations v

s. abstr

act

specific

ations.

7

Definin

g C

overa

ge

(1)

“Sim

ula

tion A

ppro

ach”:

[S. K

atz

et.

al. 1

999]

•A

CT

L S

afe

ty p

ropert

ies.

•A

well-

covere

d im

ple

menta

tion s

hould

clo

sely

resem

ble

the r

educed table

au

of its

specific

ation.

•H

ence, a fully

covere

d im

ple

menta

tion is

bis

imila

rto

the r

educed table

au

of its

specific

ation, i.e. has the s

am

e s

et of

behavio

rs a

s the s

pecific

ation.

Bis

imu

lati

on

is

very

str

ict!

Th

e r

ed

uced

tab

leau

can

b

e h

ug

e!

8

Definin

g C

overa

ge

(1)

“Sim

ula

tion A

ppro

ach”

cont.:

•F

our

crite

ria in c

om

paring a

n im

ple

menta

tion I

with the r

educed

table

au S

of th

e s

pecific

ation.

(a)

UnIm

ple

mente

dS

tart

Sta

te, w

hic

h c

onta

ins the s

et of sta

tes w

0’

in W

0’fo

r w

hic

h a

ll w

0W

0have w

0’

sim

(w0).

(b)

UnIm

ple

mente

dS

tate

, w

hic

h c

onta

ins the s

et of sta

tes w

’W

’fo

r w

hic

h a

ll w

W

have w

’sim

(w).

(c)

UnIm

ple

mente

dT

ransitio

ns, w

hic

h c

onta

ins the s

et of tr

ansitio

ns

<w

’, u

’> R

’fo

r w

hic

h S

sim

ula

tes I

even w

ithout th

e tra

nsitio

n

<w

’, u

’>.

(d)

ManyT

oO

ne, w

hic

h c

onta

ins the s

et of sta

tes w

’W

’fo

r w

hic

h

sim

-1(w

’) is n

ot a s

ingle

ton.

∈∉

∈∉

Four

crite

ria a

re e

mpty

iff

the im

ple

menta

tion a

nd the

reduced table

au

of th

e s

pecific

ation a

re b

isim

ilar.

9

Definin

g C

overa

ge

(2)

“Mu

tan

t-b

ase

d A

pp

roa

ch

”: [Y

. H

oskote

et.

al. 1

999]

•In

sp

ire

d b

y m

uta

tio

n c

ove

rag

e in

sim

ula

tio

n-

ba

se

d v

eri

fica

tio

n.

[D.

L.

Dill

1998

]

•F

orm

ally

, fo

r a

n im

ple

me

nta

tio

n I

(mo

de

led

as a

lab

ele

d s

tate

-tra

nsitio

n g

rap

h),

a s

tate

win

I,

an

d

an

ob

se

rva

ble

sig

na

l q

, w

e s

ay t

ha

t w

is q

-

co

ve

red

by a

sp

ecific

atio

n S

if I

w,q

’(m

uta

nt

imp

lem

en

tatio

n b

y flip

pin

g t

he

va

lue

of

qin

w)

do

es n

ot sa

tisfy

S.

10

Ho

w a

re t

hese 2

ap

pro

ach

es

diffe

rent?

qA

GA

Gq

¬∨

q~

qq

~q

Syste

m I

1:

Sp

ecif

icati

on

:

Red

uced

tab

leau

S1:

Sim

ula

tio

n a

pp

roach

:all 4

cri

teri

a a

re e

mp

ty →

full c

overa

ge.

Mu

tan

t-b

ased

ap

pro

ach

:b

oth

sta

tes o

f I 1

are

no

t q

-co

vere

d.

11

Ho

w a

re t

hese 2

ap

pro

ach

es

diffe

rent?

Sp

ecif

icati

on

:A

Gq

qq

Syste

m I

2:

Red

uced

tab

leau

S2:

q

Sim

ula

tio

n A

pp

roach

:

Cri

teri

a 4

is n

ot

em

pty

: b

oth

sta

tes o

f I 2

are

sim

ula

ted

by t

he s

tate

t0.

Mu

tan

t-b

ased

Ap

pro

ach

:

I 2is

q-c

overe

d b

y S

2.

u0

u1

t 0

12

Muta

nt-

based A

ppro

ach

Tw

o c

overa

ge c

hecks:

(1)

Fals

ity c

overa

ge: does the m

uta

nt F

SM

still

satisfy

the s

pecific

ation?

(2)

Vacuity c

overa

ge: if the m

uta

nt F

SM

still

satisfies the s

pecific

ation, does it satisfy

it v

acuously

?

13

w0

w4

gra

nt 2

w3

w1

gra

nt 1

w2

gra

nt 2

LT

L s

pecific

ation S

: assert

G (

gra

nt 1→

F g

rant 2

);

I w

ant an e

xam

ple

!

w4

is fals

ity-

covere

d b

y S

w.r

.t.

muta

ton

on g

ran

t2

w1

is v

acuity-

covere

d b

y S

w.r

.t.

om

issio

n o

f w

1o

r

muta

tion o

n g

ran

t 1

w0

is v

acuity-

covere

d b

y S

w.r

.t.

muta

tion o

n g

ran

t 2

?

(1)

G (

gra

nt 1

→X

gra

nt 2

)?

(2)

Num

ber

of g

ran

t 2=

2?

(3)

Redundancy,

i.e.

w2

can

be o

mitte

d?

str

uctu

re-c

overe

d

(flip

ped a

lwa

ys)

vs.

node-c

overe

d

(flip

ped o

nly

once)

Qu

esti

on

: Is

w4

str

uctu

re-

co

ve

red

or

no

de-c

ove

red

by S

w.r

.t.

the m

uta

tio

n o

n g

ran

t 2?

14

Types o

f covera

ge m

etr

ics

A.S

ynta

ctic-c

overa

ge m

etr

ics

•C

od

e-b

ased

co

ve

rag

e

•C

ircu

it c

overa

ge

•H

it c

ou

nt

B.S

em

antic-c

overa

ge m

etr

ics

•F

SM

co

vera

ge

•A

ssert

ion

co

ve

rag

e

•M

uta

tio

n c

overa

ge

15

Types o

f covera

ge m

etr

ics

A.S

ynta

ctic-c

overa

ge m

etr

ics

•C

od

e-b

as

ed

co

ve

rag

e

•C

ircu

it c

overa

ge

•H

it c

ou

nt

B.S

em

antic-c

overa

ge m

etr

ics

•F

SM

co

vera

ge

•A

ssert

ion

co

ve

rag

e

•M

uta

tio

n c

overa

ge

16

Code C

overa

ge (

sim

ula

tion)

•sta

tem

en

t co

ve

rag

e

•b

ran

ch

co

ve

rag

e

•e

xp

ressio

n c

ove

rag

e

•G

ive

n a

CF

G c

alle

d G

, fo

r a

n in

pu

t se

qu

en

ce

t(2

I )*

su

ch t

ha

t th

e e

xe

cu

tio

n o

f G

on

t,

pro

jecte

d

on

th

e s

eq

ue

nce

of

loca

tio

ns,

is l

0,…

,lm

, w

e s

ay

tha

t a

sta

tem

en

t s

is c

ove

red

by t

if t

here

is 0

≤j

≤m

s.t.

l jco

rre

sp

on

ds t

o s

; a

bra

nch

<l, l’>

is

co

ve

red

by t

if th

ere

is 0

≤j ≤

m-1

su

ch

th

at

l j=

la

nd

lj+

1=

l’.

Wh

at

if t

here

is

co

ncu

rren

cy?

17

Code C

overa

ge (

F.V

.)

•G

iven a

CF

G c

alle

d G

and ξ

a

specific

ation s

atisfied in G

, w

e s

ay a

sta

tem

ent s

of G

is c

overe

d b

y ξ

if o

mitting

sfr

om

Gcauses v

acuous s

atisfa

ction o

f ξ

in the m

uta

nt C

FG

. S

imila

rly, a b

ranch

<l,l’>

of G

is c

overe

d if om

itting it

causes

vacuous s

atisfa

ction o

f ξ.

•W

hy v

acuous s

atisfa

ction

only

?

18

Types o

f covera

ge m

etr

ics

A.S

ynta

ctic-c

overa

ge m

etr

ics

•C

od

e-b

ased

co

ve

rag

e

•C

irc

uit

co

ve

rag

e

•H

it c

ou

nt

B.S

em

antic-c

overa

ge m

etr

ics

•F

SM

co

vera

ge

•A

ssert

ion

co

ve

rag

e

•M

uta

tio

n c

overa

ge

19

Circuit C

overa

ge

(sim

ula

tion)

•la

tch c

overa

ge

•to

ggle

covera

ge

•A

latc

h is c

overe

d if it c

hanges its

valu

e a

t le

ast once d

uring the e

xecution o

f th

e

input sequence.

•A

n o

utp

ut variable

is c

overe

d if its v

alu

e

has b

een toggle

d (

requires the v

alu

e to b

e

changed a

t le

ast tw

ice).

20

Circuit C

overa

ge

(F

.V.)

•R

ep

lace

th

e q

ue

stio

n b

y t

he q

ue

stio

n o

f w

he

the

r

dis

ab

ling

th

e c

ha

ng

e c

au

se

s t

he

sp

ecific

atio

n to

be

sa

tisfie

d v

acu

ou

sly

.

•A

la

tch

lis

co

ve

red

if th

e s

pecific

atio

n is

va

cu

ou

sly

sa

tisfied

in

th

e c

ircu

it o

bta

ine

d b

y

fixin

g t

he

va

lue

of

lto

its

in

itia

l va

lue

.

•A

n o

utp

ut

ois

co

vere

d if

the

sp

ecific

atio

n is

va

cu

ou

sly

sa

tisfied

in

th

e c

ircu

it o

bta

ine

d b

y

allo

win

g o

to c

ha

ng

e its

va

lue

on

ly o

nce

.

21

Types o

f covera

ge m

etr

ics

A.S

ynta

ctic-c

overa

ge m

etr

ics

•C

od

e-b

ased

co

ve

rag

e

•C

ircu

it c

overa

ge

•H

it c

ou

nt

B.S

em

antic-c

overa

ge m

etr

ics

•F

SM

co

vera

ge

•A

ssert

ion

co

ve

rag

e

•M

uta

tio

n c

overa

ge

22

Hit C

ount (s

imula

tion)

•R

epla

ce b

inary

covera

ge q

ueries w

ith

quantita

tive m

easure

ments

–th

e n

um

ber

of tim

es a

n o

bje

ct has b

een v

isited.

•V

isited o

ften →

functionalit

y b

etter

covere

d.

23

Hit C

ount (F

.V.)

•T

he m

inim

alnum

ber

of vis

its in w

hic

h w

e

have to p

erf

orm

the m

uta

tion (

or

om

issio

n

of th

e e

lem

ent)

in o

rder

to fals

ify the

specific

ation in the d

esig

n o

r to

make it

vacuously

satisfied.

24

Types o

f covera

ge m

etr

ics

A.S

ynta

ctic-c

overa

ge m

etr

ics

•C

od

e-b

ased

co

ve

rag

e

•C

ircu

it c

overa

ge

•H

it c

ou

nt

B.S

em

antic-c

overa

ge m

etr

ics

•F

SM

co

ve

rag

e

•A

ssert

ion

co

ve

rag

e

•M

uta

tio

n c

overa

ge

25

FS

M C

overa

ge (

sim

ula

tion)

•A

sta

te o

r a tra

nsitio

n o

f th

e F

SM

is

covere

d if it is v

isited d

uring the e

xecution

of th

e input sequence.

•T

ransitio

n c

overa

ge c

an b

e e

xte

nded to

path

covera

ge.

•P

roble

m?

lin

kin

g t

he u

nco

ve

red

part

s o

f th

e F

SM

to

u

nco

vere

d p

art

s o

f th

e

HD

L p

rog

ram

is n

ot

triv

ial!

co

mp

uti

ng

th

e p

ath

co

ve

rag

e i

s

exp

en

siv

e!

26

FS

M c

overa

ge (

F.V

.)

•In

sta

te c

overa

ge, w

e c

heck the influence

of om

issio

n o

f a s

tate

wor

changin

g the

valu

es o

f th

e o

utp

ut variable

s in w

on the

(nonvacuous)

satisfa

ction o

f th

e

specific

ation.

•In

path

covera

ge, w

e c

heck the influence

of om

itting o

r m

uta

ting a

fin

ite p

ath

on the

(nonvacuous)

satisfa

ction o

f th

e

specific

ation.

27

I w

ant an e

xam

ple

again

!

w0

w4

gra

nt 2

w3

w1

gra

nt 1

w2

gra

nt 2

LT

L s

pecific

ation S

: assert

G (

gra

nt 1→

F g

rant 2

);

28

I w

ant an e

xam

ple

again

!

w0

w4

gra

nt 2

w3

w1

gra

nt 1

w2

gra

nt 2

LT

L s

pecific

ation S

: assert

G (

gra

nt 1→

F g

rant 2

);

om

itti

ng

path

w0,

w2, w

3.

≠re

mo

vin

g

tran

sit

ion

s

<w

0,w

2>

an

d

<w

2,w

3>

Th

e s

pecif

icati

on

is s

ati

sfi

ed

n

on

vacu

ou

sly

.

29

Types o

f covera

ge m

etr

ics

A.S

ynta

ctic-c

overa

ge m

etr

ics

•C

od

e-b

ased

co

ve

rag

e

•C

ircu

it c

overa

ge

•H

it c

ou

nt

B.S

em

antic-c

overa

ge m

etr

ics

•F

SM

co

vera

ge

•A

ss

ert

ion

co

ve

rag

e

•M

uta

tio

n c

overa

ge

30

Assert

ion C

overa

ge (

sim

ula

tion)

•A

lso c

alle

d “

functional covera

ge”.

•A

ssert

ions c

an b

e p

ropositio

nalor

tem

pora

l.

•A

test tcovers

an a

ssert

ion a

if the

execution o

f th

e d

esig

n o

n t

satisfies a

.

•M

easure

s %

assert

ions c

overe

d for

a

giv

en s

et of in

put sequences.

31

Assert

ion C

overa

ge (

F.V

.)

•A

n a

ssert

ion a

is c

overe

d b

y a

specific

ation ξ

in a

FS

M F

if the m

uta

nt

FS

M F

’obta

ined fro

m F

by o

mitting a

ll behavio

rs that satisfy

asatisfies ξ

nonvacuously

.

•W

hat covera

ge m

etr

ic that w

e h

ave talk

ed

about is

sim

ilar

to this

one?

FS

M P

ath

Covera

ge!

32

Types o

f covera

ge m

etr

ics

A.S

ynta

ctic-c

overa

ge m

etr

ics

•C

od

e-b

ased

co

ve

rag

e

•C

ircu

it c

overa

ge

•H

it c

ou

nt

B.S

em

antic-c

overa

ge m

etr

ics

•F

SM

co

vera

ge

•A

ssert

ion

co

ve

rag

e

•M

uta

tio

n c

ov

era

ge

33

Muta

tion C

overa

ge (

sim

ula

tion)

•U

ser

intr

oduces a

sm

all

change to the

desig

n a

nd c

heck for

err

oneous b

ehavio

r.

•T

he c

overa

ge o

f a test tis

measure

d a

s

the %

muta

nt desig

ns that fa

il on t.

•T

he g

oal is

to fin

d a

set of in

put

sequences s

.t. fo

r each m

uta

nt desig

n

there

exis

ts a

t le

ast one test th

at fa

ils it.

34

Muta

tion C

overa

ge (

F.V

.)

•B

y m

uta

tion, w

e a

ctu

ally

mean local

muta

tion.

•W

e h

ave talk

ed a

bout th

is.

35

Muta

nt-

based A

ppro

ach

Tw

o c

overa

ge c

hecks:

[H.

Chockle

ret. a

l. 2

006]

(1)

Fals

ity c

overa

ge: does the m

uta

nt F

SM

still

satisfy

the s

pecific

ation?

(2)

Vacuity c

overa

ge: if the m

uta

nt F

SM

still

satisfies the s

pecific

ation, does it satisfy

it v

acuously

?

36

Com

puting C

overa

ge (

1)

•M

uta

nt-

based

Appro

ach:

[Y. H

osko

teet. a

l.

199

9]

•T

he g

oal is

to fin

d the s

et of covere

d

sta

tes.

•C

overa

ge =

num

ber

of covere

d s

tate

s /

num

ber

of re

achable

sta

tes

•R

ecurs

ive

alg

orith

m for

a g

iven A

CT

L

form

ula

and a

giv

en o

bserv

ed s

ignal.

37

Outlin

e o

f th

e a

lgorith

m

•S

ay,

we

wa

nt

to c

om

pu

te c

ove

rag

e f

or

A(f

1U

f 2).

)(

)(

''

21

00

fT

fT

SS

II

=f 1

S0

f 2

f 1

f 1

f 1

f 1

f 2

f 1f 1

f 2

C(S

0, A

(f1

Uf 2

)) =

C(t

ravers

e(S

0,f

1,f

2)

U C

(fir

str

each

ed

(S0,f

2),

f2)

)(

)(

''

21

00

fT

fT

SS

¬=

II

forw

ard

(S0)

giv

es s

tate

s

reach

ab

le i

n e

xactl

y o

ne s

tep

fr

om

th

e s

tart

sta

tes i

n S

0.

)))

,(

((

))(

(

),

(

22

0

20

20

ff

TS

forw

ard

edfi

rstr

eachf

TS

fS

edfi

rstr

each

¬

=

I

UI

)2

,1),'

0(

('

0

)2

,1,

0(

ff

Sfo

rwa

rdtr

avers

eS

ff

Str

avers

e

U

=

whe

re,

T(b

) re

pre

sen

t th

e

set

of

sta

tes w

hic

h

sati

sfy

b.

38

Com

puting C

overa

ge (

2)

modu

le e

xam

ple

(o1,o

2,o

3);

reg

o1,o

2,o

3;

initia

l beg

in

o1=

o2=

o3=

0;

end

alw

ays @

(posedg

eclk

) beg

in

assig

n o

1=

o1;

assig

n o

2=

o2|o

3;

assig

n o

3=

~o

3;

end

endm

odu

le

[H.

Ch

ockle

ret.

al.

2006]

S:

assert

G(o

2→

F o

3);

Co

de C

ov

era

ge:

This

sta

tem

ent

is

uncovere

d b

y t

he

specific

atio

n w

.r.t. to

both

om

issio

n a

nd m

uta

tions.

39

Com

puting C

overa

ge (

2)

modu

le e

xam

ple

(o1,o

2,o

3);

reg

o1,o

2,o

3;

initia

l beg

in

o1=

o2=

o3=

0;

end

alw

ays @

(posedg

eclk

) beg

in

assig

n o

1=

o1;

assig

n o

2=

o2|o

3;

assig

n o

3=

~o

3;

end

endm

odu

le

[H.

Ch

ockle

ret.

al.

2006]

S:

assert

G(o

2→

F o

3);

Cir

cu

it C

overa

ge:

Latc

h o

1is

uncovere

d

–fixin

g o

1to

0 f

or

the

wh

ole

execution d

oes

not

affect th

e

satisfa

ction

of S

.

40

Com

puting C

overa

ge (

2)

[H.

Ch

ockle

ret.

al.

2006]

S:

assert

G(o

2→

F o

3);

000

100

101

010

011

111

110

001

o1o

2o

3

FS

M C

ove

rag

e:

All

sta

tes in w

hic

h o

1 =

1 a

re u

nre

acha

ble

, and

thus u

ncovere

d w

.r.t. all

specific

atio

ns.

?

41

Com

puting C

overa

ge (

2)

[H.

Ch

ockle

ret.

al.

2006]

(1)

Naiv

e w

ay:

en

um

era

teth

rou

gh

all

mu

tan

t

FS

M t

o c

heck b

oth

fals

ity a

nd

vacu

ity c

overa

ge

for

a s

pecif

icati

on

ξ.

(2)

A B

ett

er

way:

co

mp

ute

co

ve

rag

e

sym

bo

licall

y.

oT

he id

ea i

s t

o l

oo

k f

or

a f

air

path

in

th

e

pro

du

ct

of

the m

uta

nt

FS

M F

’an

d a

n

au

tom

ato

n A

for

the n

eg

ati

on

of ξ.

oA

dd

a n

ew

va

riab

le x

th

at

en

co

des a

su

bfo

rmu

lain

ξth

at

is b

ein

g r

ep

laced

by

tru

e/f

als

e.

Th

e v

alu

e 0

fo

r x s

tan

ds f

or

“n

o

rep

lacem

en

t”.

Th

en

we c

heck t

he

sati

sfa

cti

on

of ξ

in t

he s

yste

m.

oC

on

sid

er

an

au

gm

en

ted

pro

du

ct

wit

h s

tate

sp

ace 2

XX

2X

X S

.

A c

yc

le i

s r

ea

ch

ab

le in

th

e

au

gm

en

ted

au

tom

ato

n.

<w

,u0,s

0>

<w

,u,s

>

P

<w

,w’,s

><

w,w

,s>

fair

pa

th

flip

pin

g

q

42

Co

mple

xity o

f C

om

puting C

ove

rage

3n

+ 2

m+

log

lH

it C

ount

3n

+ 4

m+

logk

Assert

ion C

overa

ge

3n

+ 4

m+

logp

FS

M P

ath

Covera

ge

2n

+ 3

m+

log

lC

ircuit C

overa

ge

2n

+ 2

m+

logk

Co

de C

ove

rage

3n

+ 3

mV

acu

ity C

overa

ge

3n

+ 2

mM

uta

tion C

overa

ge

Co

mp

lexit

yM

etr

ic

Co

mp

lexit

y i

n t

erm

s o

f R

OB

DD

vari

ab

les r

eq

uir

ed

.

nand m

are

the

num

ber

of

variable

s r

equired fo

r e

ncodin

g the s

tate

space o

f F

and A

~ξ;

ldenote

s the n

um

ber

of

latc

hes in the c

ircuit, k

de

note

s the n

um

be

r o

f

assert

ions/n

um

ber

of lin

es o

f nodes; p

denote

s th

e length

of

the p

ath

; tdeno

tes th

e

thre

shold

of hit c

ount.

43

Conclu

sio

n

•T

wo p

roble

ms inhere

nt w

ith a

ny c

overa

ge

metr

ics.

•T

wo a

ppro

aches to d

efine c

overa

ge

metr

ics for

form

al verification.

•D

iffe

rent sem

antic a

nd s

ynta

ctic c

overa

ge

metr

ics –

how

are

they s

imila

r to

/diffe

rent

from

the o

nes u

sed in s

imula

tion.

•S

till

an o

pen p

roble

m.

44

Refe

rence L

ist

•M

ichae

l W

. W

hale

n,

Ajit

ha

Raja

n,

Mats

P.E

. H

eim

dahl, S

teven P

. M

iller.

C

overa

ge m

etr

ics f

or

require

ments

-base

d t

esting.

Pro

c.

of

Inte

rnationa

l S

ym

posiu

m o

n S

oft

ware

Testing a

nd A

naly

sis

, page 2

5-3

6,

2006.

•A

jitha

Ra

jan.

Covera

ge M

etr

ics t

o M

easure

Adequ

acy o

f B

lack-B

ox T

est

Suites.

Pro

c.

of

Inte

rnationa

l C

onfe

rence o

n A

uto

mate

d S

oft

wa

re

Engin

eeri

ng,

page 3

35-3

38,

2006.

•S

. K

atz

, O

. G

rum

berg

, D

. G

eis

t. “

Have I

wri

tten e

nou

gh p

rop

ert

ies? -

A

meth

od o

f com

pariso

n b

etw

ee

n s

pecific

ation a

nd im

ple

me

nta

tion”.

Pro

c.

of

10th

AC

M A

dvanced R

esearc

h W

ork

ing C

on

fere

nce o

n C

orr

ect

Hard

ware

D

esig

n a

nd V

erification M

eth

ods (

CH

AR

M),

page 2

80-2

97,

1999.

•D

.L.

Dill

. W

hat’s b

etw

een s

imula

tio

n a

nd f

orm

al verification

?P

roc.

of

35th

D

esig

n A

uto

mation C

onfe

ren

ce,

page 3

28–

329,

199

8.

•Y

. H

oskote

, T

. K

am

, P

.-H

Ho, X

. Z

hao.

Covera

ge e

stim

ation f

or

sym

bo

lic

model ch

eckin

g.

Pro

c.

of

36th

Desig

n A

uto

mation C

onfe

ren

ce,

page 3

00–

305,

199

9.

•H

ana

Chockle

r, O

rna

Kupfe

rman,

Moshe Y

. V

ard

i. C

overa

ge m

etr

ics f

or

form

al verification.

Inte

rnal Jo

urn

al o

n S

oft

ware

Tools

for

Techno

log

y

Tra

nsfe

r (S

TT

T),

Vol. 8

, Is

sue 4

, P

age 3

73-3

86,

August

20

06.

45

Refe

rence L

ist

•H

ana

Chockle

r, O

rna

Kupfe

rman,

Moshe Y

. V

ard

i. C

overa

ge m

etr

ics f

or

form

al verification.

Pro

c.

of

12th

Advance

d R

esearc

h W

ork

ing C

onfe

rence

on C

orr

ect

Hard

ware

Desig

n a

nd V

erificatio

n M

eth

ods (

CH

AR

ME

), 2

003.

•H

ana

Chockle

r, O

rna

Kupfe

rman.

Covera

ge o

f Im

ple

menta

tions b

y

Sim

ula

ting S

pe

cific

ations.

Pro

c.

Of 2nd I

FIP

Inte

rnatio

nal C

Onfe

rence

on

Theore

tical C

om

pute

r S

cie

nce:

Foundatio

ns o

f In

form

ation T

echnolo

gy in

the E

ra o

f N

etw

ork

ing a

nd M

obile

Com

putin

g,

page 4

09-4

21,

2002.

•H

ana

Chockle

r, O

rna

Kupfe

rman,

Robert

P.

Kurs

han,

Moshe Y

. V

ard

i. A

P

ractical A

ppro

ach t

o C

overa

ge in M

ode

l C

he

ckin

g.

In P

roc.

of

13th

In

tern

ation

al C

onfe

rence o

n C

om

pute

r A

ide

d V

erificatio

n,

page 6

6-7

8,

2001.

•H

ana

Chockle

r, O

rna

kupfe

rman,

Moshe Y

. V

ard

i. C

overa

ge M

etr

ics f

or

Tem

pora

l Logic

Model C

heckin

g.

In P

roc.

of 7th

Inte

rnation

al C

onfe

rence

on T

ools

and A

lgorith

ms f

or

the C

onstr

uctio

n a

nd A

na

lysis

of

Syste

ms,

page 5

28-5

42,

2001.

•O

rna

Kupfe

rman,

Moshe Y

. V

ard

i. V

acuity D

ete

ction in T

em

pora

l M

odel

Checkin

g.

In P

roc.

of

10th

IF

IP W

G 1

0.5

Advanced R

esearc

h W

ork

ing

Confe

rence o

n C

orr

ect

Hard

ware

Desig

n a

nd

Verification M

eth

ods,

pag

e

82-9

6,

19

99.


Recommended