►
From YouTube: Smart Contract Security
Description
Visit the https://archive.devcon.org/ to gain access to the entire library of Devcon talks with the ease of filtering, playlists, personalized suggestions, decentralized access on IPFS and more.
https://archive.devcon.org/archive/watch/2/smart-contract-security
After a quick overview of smart contract failures in the past, a list of important takeaways will be covered. Some coding techniques to prevent unexpected behaviour in smart contracts will be covered as well as some remarks about governance in decentralized systems.
Speaker(s): Christoph Jentzsch
Skill level: Intermediate
Track: Security
Keywords: DAO, hack, exploit, cap, solidity, formal, proof, verification, invariance, centralization, fork, failsafe, governance, dapps, community, multisig, updates, delays, libraries, developers, tools, compiler, IDEs
Follow us: https://twitter.com/efdevcon, https://twitter.com/ethereum
Learn more about devcon: https://www.devcon.org/
Learn more about ethereum: https://ethereum.org/
Devcon is the Ethereum conference for developers, researchers, thinkers, and makers.
Devcon 2 was held in Shanghai, China on Sep 19 - 21, 2016.
Devcon is organized and presented by the Ethereum Foundation, with the support of our sponsors. To find out more, please visit https://ethereum.foundation/
A
A
You
know
about
the
Dow
and
I
was
mainly
responsible
for
writing
down
smart
contract.
So
on
7th
of
June
there
was
an
attack
robbing
3.5
million
ISA
out
of
this
smart
contract
by
using
the
reentry
exploit.
We
have
heard
about
this
before
this
conference.
I
will
not
go
into
details
how
this
works,
but
those
were
the
lines
of
code
in
the
dow
itself.
Basically,
there
was
this.
One
function
was
12
yacht
with
tois
we
wore
4,
which
was
vulnerable
to
this
exploit.
A
So
what
can
we
learn
from
all
of
this?
There's?
A
lot
of
things
have
been
said
yesterday.
I
mean
I
will
repeat
some
of
them,
of
course,
but
give
my
personal
perspective
on
some
of
those
things.
First
of
all,
cap,
smart
contracts,
it's
very
early
days
and
B
lack
experience,
meaning
we
just
need
to
learn
some
lessons
before
to
know
which
kind
of
bucks
we
can
do.
If
you
just
look
at
version
numbers
of
the
current
software,
you
can
get
a
feeling
of
how
early
it
is.
A
A
Realloc
suggested
in
a
blog
post
that,
for
example,
right
now,
a
cap
of
about
ten
million
dollars
seems
about
right.
It's
an
it.
It's
an
individual
decision
but,
of
course,
would
have
been
nice
if
the
Dow
would
have
had
such
a
cap.
But
of
course
nobody
didn't
know
that
it
would
rise
to
such
an
ass
eyes.
So
next
one
formal
proof,
verification
I,
will
not
go
into
this.
We
have
heard
about
a
lot
of
this
yesterday,
but,
yes,
a
very
important
topic.
I
hope
we
make
two
could
make
continuous
progress
on
this
topic,
invariant
checks.
A
So
basically
you
can
write
see
that
your
smart
contact
has
certain
invariants
such
as
for
a
doubt,
as
would
have
been
that
a
total
supply
is
smaller
than
the
balance
plus
reward
Hung's
and
then,
after
every
function,
you
can
check
if
the
invariant
still
hold
its
reduces,
of
course,
the
risk.
Then
this
is
an
interesting
topic
centralization.
A
Of
course,
we
want
to
build
decentralized
application,
that's
all
main
goal
and
one
of
the
weaknesses
so
to
say
of
the
dowel
boss,
that
it
was
we
D
decentralized,
is
that
we
had
no
control,
nobody
had
any
control
over
it,
and
this
was
the
problem
and
in
the
end,
because
nobody
could
save
the
doubt
only
the
community
by
doing
a
hard
fall,
which
of
course,
is
a
horrible
option
to
do
so.
We
need
to
go
stepwise
from
centralization
to
decentralization.
If
you
think
about
a
theorem,
we
had
Olympic
tests.
A
In
a
time
you
had
the
franchise
net
that
we
had.
Those
canneries
I
had
one
of
those
keys
where
basically,
two
out
of
four
key
holders
could
more
or
less
switch
of
the.
If
they
would
listen
to
the
canneries.
We
had
have
homestead
that
we
have
this
difficulty
increase,
which
means
we
need
to
have
a
hard
fork
in
about
a
year
or
so,
and
that's
only
a
couple
of
people
able
to
code
up
the
next
version
of
etherium.
A
So
it's
also
kind
of
centralized
so
and
we
will
go
step
to
step
to
more
decentralization
for
a
doubt.
I
have
been
curators,
they
have
been
given
a
lot
of
power
except
of
the
function
split
doubt
because
this
was
a
function
meant
to
as
a
fail-safe
or
malicious
creators.
So
that's
why
I
didn't
have
control
over
this
function.
Couldn't
do
anything
there's
all
the
other
Taoist.
A
It
should
I
Omega
down
after
thousand
workers,
which
start
centralized
and
we'll
go
into
decentralization,
then
equations,
because
who
can
control
for
a
doubt
what
some
token
holders
it
could
be
central
transit
authorities
and
there's
more,
but
a
slight
stone
showed
up
here
right
now.
It
could
be
assumption
area,
community
multi-sig.
A
It
could
be
a
something
like
a
stake
vote
that
you
build
in
some
senior
smart
contract
when
a
certain
amount
of
ether
holders
vote
for
its
top
something
so
there's
certain
ways
of
controlling
this,
but
I
think
we
need
to
go
step
by
step,
but
also
I
think
it's
important
that
we
really
want
to
build
decentralized
application
and
do
not
use
the
towers
and
excuse
to
only
build
centralized
applications.
Although
it's
good
to
start
like
this
I,
don't
think
it
should
end
like
this.
We
should
move
forward
to
fool
decentralization
one
day.
A
Establish
security
patterns,
meaning
learning
we
didn't
know
about
the
call
stack
that
to
that
call
stack
dams
attack.
We
know
about
the
plot
gas
limit.
Don't
have
a
battery
links
loops,
we
know
about,
we
only
exploit
now
we
know
about
that.
Either
sent
the
contractors
out
it's
possible
to
send
either
to
a
contract
without
any
contract
invitation.
So
even
if
you
use
this
modifier
payable
or
you
don't
use
it
on
any
function,
you
try
to
avoid
getting
either
as
a
contract.
A
You
cannot
avoid
it
because
you
can
use
the
suicide
opcode
to
set
transfer
either
to
a
contract
without
executing
any
code.
For
example,
specify
the
right
amount
of
gas
send
roses
call
depending
on
what
your
exact
you
want
to
do.
You
have
to
be
careful
with
the
block
time
stamp
because
it
can
be
manipulated.
A
Transaction
versus
message,
sender,
for
example,
can
be
used
for
phishing
attacks
and
much
more
she's
better.
The
slides
are
not
really
working
Bruce,
there
are
links
coming
up.
It's
just.
There
are
some
very
good
resources.
One
is
from
the
consensus
website,
best
smart
contractors,
let's
correct
best
practice
of
small
contracts
or
something
like
this.
You
find
out
a
consensus
website,
it's
a
really
nice
overview
of
all
the
things
we
can
get
me
to
learn
and
what
we
have
learned
up
to
now.
A
A
Yes,
updatable
contracts,
so
the
dow
had
a
possibility
of
updating
the
contract
through
a
vote.
It
did
take
two
weeks
debating
time
and
now
one
point
wanted
to
say
was
walking
progress,
but
it
was
too
late,
so
it
is
I
would
advise
you
in
your
smile
when
you
write
smart
contracts
to
have
an
option
to
update
them.
The
question
is
only
who
can
update
them
in
the
beginning.
It
can
be
you
of
course,
centralized.
A
Time
delays
also
something
important
when
you
whispering
someone
can
restore
ether
or
to
take
ether
out
of
a
small
contract.
If
you
have
a
time
delay,
this
gives
an
option
if
you
implement
at
some
authority
to
do
something
to
act
in
case
of
the
doubt,
there
was
no
such
authority
in
the
least
resort.
Was
this
misters
awful
hard
fork?
So,
therefore,
if
you
have
time
delays
and
this
Authority
implemented,
then
this
can
work
nicely
together
to
reduce
the
risk
of
a
smart
contract,
of
course,
minimal
complexity.
A
There
are
statistics
out
there
which
saying
there
are
15
up
to
50
bucks
per
thousand
lines
of
code
for
website.
It's
okay,
you
can
fix
it,
but
for
smart
contract,
that's
really
bad.
So,
of
course
we
need
a
much
much
more
security
for
smart
contracts
and
we
need
for
normal
stuff.
So
this
also
means
not
everything,
needs
to
centralization
and
needs
to
be
in
a
smart
contract
only
put
in
what
you
really
need
the
core
elements
of
a
decent
last
application.
Everything
else
can
be
maybe
in
swarm
whisper
or
even
other
techniques.
A
Yeah
and
the
other
thing
is
to
reuse
trusted
proofing
code.
Can
you
also
be
dangerous
at,
for
example,
the
standard
token
contract,
which
was
also
used
in
the
Dow
which
worked
just
nice?
You
have
the
foundation.
Moody
SiC,
which
seems
to
hold
up
to
now,
so
means
also
looks
safe,
and
maybe
it
will
be
even
a
Taoist
and
a
framework
one
day
just
to
reuse
a
little
safe
code.
This
also
come
as
a
danger,
because
if
then
you,
someone
finds
it
back
in
this
kind
of
code
if
it
affect
a
lot
of
applications.
A
A
A
So
as
a
conclusion,
I
would
say
it
is
very
early
days
and
we
all
need
to
be
very,
very
careful,
but
I
think
there
is,
of
course,
a
lot
of
lessons
to
be
learned
from
this,
but
being
here,
the
last
couple
of
days
got
me
really
excited
to
see
what
came
out
of
this.
We
have
a
lot
of
security
experts
from
academics
to
have
a
lot
of
media
attention
lot
of
people.
A
You
can
only
really
say
thank
you.
The
last
couple
of
months
have
been
as
opposed
as
a
person
for
me,
my
personal
life,
also
as
a
company
horrible,
of
course,
but
it's
only
because
of
the
people
sitting
here
because
of
the
lot
of
the
ìiî
and
community.
That
I
can
stand
here
today
and
give
this
talk
and
I
just
want
to
say.
Thank
you
very
much.