en/glep/glep-0059.txt

GLEP: 59
Title: Manifest2 hash policies and security implications
Version: $Revision: 1.6 $
Last-Modified: $Date: 2010/01/31 09:55:43 $
Author: Robin Hugh Johnson <robbat2@gentoo.org>, 
Status: Draft
Type: Standards Track
Content-Type: text/x-rst
Requires: 44
Created: October 2006
Updated: November 2007, June 2008, July 2008, October 2008, January 2010
Updates: 44
Post-History: December 2009, January 2010

Abstract
========
While Manifest2 format allows multiple hashes, the question of which
checksums should be present, why, and the security implications of such
have never been resolved. This GLEP covers all of these issues, and
makes recommendations as to how to handle checksums both now, and in
future.

Motivation
==========
This GLEP is being written as part of the work on signing the Portage
tree, but is only tangentially related to the actual signing of
Manifests. Checksums present one possible weak point in the overall
security of the tree - and a comprehensive security plan is needed.

This GLEP is not mandatory for the tree-signing specification, but
instead aims to improve the security of the hashes used in Manifest2.
As such, it is also able to stand on it's own.

Specification
=============
The bad news
------------
First of all, I'd like to cover the bad news in checksum security.
A much discussed point, as been the simple question: What is the
security of multiple independent checksums on the same data?
The most common position (and indeed the one previously held by myself),
is that multiple checksums would be an increase in security, but we
could not provably quantify the amount of security this added.
The really bad news, is that this position is completely and utterly
wrong. Many of you will be aghast at this. There is extremely little
added security in multiple checksums as noted by Joux [J04]. For any set
of checksums, the actual strength lies in that of the strongest
checksum.

Wang et al [W04] extended Joux's [J04] work on SHA-0 to cover MD4, MD5,
HAVAL-128 and RIPEMD families of hashes.

How fast can MD5 be broken?
---------------------------
For a general collision, not a pre-image attack, since the announcement
by Wang et al [W04], the time required to break MD5 has been massively
reduced. Originally at 1 hour on a near-supercomputer (IBM P690) and
estimated at 64 hours with a Pentium-3 1.7Ghz. This has gone down to
less than in two years, to 17 seconds [K06a].

08/2004 - 1 hour, IBM pSeries 690 (32x 1.7Ghz POWER4+) = 54.4 GHz-Hours
03/2005 - 8 hours, Pentium-M 1.6Ghz = 12.8 Ghz-Hours
11/2005 - 5 hours, Pentium-4 1.7Ghz = 8.5 Ghz-Hours
03/2006 - 1 minute, Pentium-4 3.2Ghz = .05 Ghz-Hours
04/2006 - 17 seconds, Pentium-4 3.2Ghz = .01 Ghz-Hours

If we accept a factor of 800x as a sample of how much faster a checksum
may be broken over the course of 2 years (MD5 using the above data is
>2000x), then existing checksums do not stand a significant chance of
survival in the future. We should thus accept that whatever checksums we
are using today, will be broken in the near future, and plan as best as
possible. (A brief review [H04] of the SHA1 attacks indicates an
improvement of ~600x in the same timespan).

And for those that claim implementation of these procedures is not yet
feasible, see [K06b] for an application that can produce two
self-extracting EXE files, with identical MD5s, and whatever payload you
want.

The good news
-------------
Of the checksums presently used by Manifest2 (SHA1, SHA256, RIPEMD160),
one stands close to being completely broken: SHA1; and another is
significantly weakened: RIPEMD160. The SHA2 series has suffered some
attacks, but still remains reasonably solid [G07],[K08]. 

To reduce the potential for future problems and any single checksum
break leading to a rapid decrease in security, we should incorporate the
strongest hash available from each family of checksums, and be prepared
to retire old checksums actively, unless there is a overriding reason to
keep a specific checksum, such as part of a migration plan.

What should be done
-------------------
Portage should always try to verify all supported hashes that are
available in a Manifest2, starting with the strongest ones as maintained
by a preference list. Over time, the weaker checksums should be removed
from Manifest2 files, once all old Portage installations have had
sufficient time to upgrade. We should be prepared to add stronger
checksums wherever possible, and to remove those that have been
defeated.

As soon as feasible, we should add the SHA512 and WHIRLPOOL algorithms.
In future, as stream-based checksums are developed (in response to the
development by NIST [AHS]), they should be considered and used.

The SHA512 algorithm is available in Python 2.5, which has been a
dependency of Portage since approximately Portage 2.1.6.13.

The WHIRLPOOL checksum is not available within the PyCrypto library or
hashlib that is part of Python 2.5, but there are multiple alternative
Python implementations available, ranging from pure Python to C-based
(python-mhash).

The existence unsupported hash is not considered to be a failure unless
no supported hashes are available for a given Manifest entry.

Checksum depreciation timing
----------------------------
For the current Portage, both SHA1 and RIPEMD160 should be immediately
removed, as they present no advantages over the already present SHA256.
SHA256 cannot be replaced immediately with SHA512, as existing Portage
versions need at least one supported algorithm present (SHA256 support
was added in June 2006), so it must be retained for some while.

Immediately:
- Add WHIRLPOOL and SHA512.
- Remove SHA1 and RIPEMD160.

After the majority of Portage installations include SHA512 support:
- Remove SHA256.

Backwards Compatibility
=======================
Old versions of Portage may support and expect only specific checksums.
This is accounted for in the checksum depreciation discussion.

For maximum compatiability, we should only have to include each of the
old algorithms that we are officially still supporting, as well as the
new ones that we prefer.

References
==========

[AHS] NIST (2007). "NIST's Plan for New Cryptographic Hash Functions",
  (Advanced Hash Standard). http://csrc.nist.gov/pki/HashWorkshop/

[BOBO06] Boneh, D. and Boyen, X. (2006). "On the Impossibility of
  Efficiently Combining Collision Resistant Hash Functions"; Proceedings
  of CRYPTO 2006, Dwork, C. (Ed.); Lecture Notes in Computer Science
  4117, pp. 570-583. Available online from:
  http://crypto.stanford.edu/~dabo/abstracts/hashing.html

[H04] Hawkes, P. and Paddon, M. and Rose, G. (2004). "On Corrective
  Patterns for the SHA-2 Family". CRYPTO 2004 Cryptology ePrint Archive,
  Report 2004/204. Available online from:
  http://eprint.iacr.org/2004/207.pdf

[J04] Joux, Antoie. (2004).  "Multicollisions in Iterated Hash 
  Functions - Application to Cascaded Constructions;" Proceedings of
  CRYPTO 2004, Franklin, M. (Ed); Lecture Notes in Computer Science
  3152, pp.  306-316. Available online from:
  http://web.cecs.pdx.edu/~teshrim/spring06/papers/general-attacks/multi-joux.pdf

[K06a] Klima, V. (2006). "Tunnels in Hash Functions: MD5 Collisions
  Within a Minute". Cryptology ePrint Archive, Report 2006/105.
  Available online from: http://eprint.iacr.org/2006/105.pdf

[K06b] Klima, V. (2006). "Note and links to high-speed MD5 collision
  proof of concept tools". Available online from:
  http://cryptography.hyperlink.cz/2006/trick.txt

[K08] Klima, V. (2008). "On Collisions of Hash Functions Turbo SHA-2".
  Cryptology ePrint Archive, Report 2008/003. Available online from:
  http://eprint.iacr.org/2008/003.pdf

[G07] Gligoroski, D. and Knapskog, S.J. (2007). "Turbo SHA-2".
  Cryptology ePrint Archive, Report 2007/403. Available online from:
  http://eprint.iacr.org/2007/403.pdf

[W04] Wang, X. et al: "Collisions for Hash Functions MD4, MD5,
  HAVAL-128 and RIPEMD", rump session, CRYPTO 2004, Cryptology ePrint
  Archive, Report 2004/199, first version (August 16, 2004), second
  version (August 17, 2004). Available online from:
  http://eprint.iacr.org/2004/199.pdf

Thanks to
=========
I'd like to thank the following folks, in no specific order:
 - Ciaran McCreesh (ciaranm) - for pointing out the Joux (2004) paper,
   and also being stubborn enough in not accepting a partial solution.
 - Marius Mauch (genone), Zac Medico (zmedico) and Brian Harring
   (ferringb): for being knowledgeable about the Portage Manifest2
   codebase.

Copyright
=========
Copyright (c) 2006-2010 by Robin Hugh Johnson. This material may be
distributed only subject to the terms and conditions set forth in the
Open Publication License, v1.0.

vim: tw=72 ts=2 expandtab:
1	GLEP: 59
2	Title: Manifest2 hash policies and security implications
3	Version: $Revision: 1.6 $
4	Last-Modified: $Date: 2010/01/31 09:55:43 $
5	Author: Robin Hugh Johnson <robbat2@gentoo.org>,
6	Status: Draft
7	Type: Standards Track
8	Content-Type: text/x-rst
9	Requires: 44
10	Created: October 2006
11	Updated: November 2007, June 2008, July 2008, October 2008, January 2010
12	Updates: 44
13	Post-History: December 2009, January 2010
14
15	Abstract
16	========
17	While Manifest2 format allows multiple hashes, the question of which
18	checksums should be present, why, and the security implications of such
19	have never been resolved. This GLEP covers all of these issues, and
20	makes recommendations as to how to handle checksums both now, and in
21	future.
22
23	Motivation
24	==========
25	This GLEP is being written as part of the work on signing the Portage
26	tree, but is only tangentially related to the actual signing of
27	Manifests. Checksums present one possible weak point in the overall
28	security of the tree - and a comprehensive security plan is needed.
29
30	This GLEP is not mandatory for the tree-signing specification, but
31	instead aims to improve the security of the hashes used in Manifest2.
32	As such, it is also able to stand on it's own.
33
34	Specification
35	=============
36	The bad news
37	------------
38	First of all, I'd like to cover the bad news in checksum security.
39	A much discussed point, as been the simple question: What is the
40	security of multiple independent checksums on the same data?
41	The most common position (and indeed the one previously held by myself),
42	is that multiple checksums would be an increase in security, but we
43	could not provably quantify the amount of security this added.
44	The really bad news, is that this position is completely and utterly
45	wrong. Many of you will be aghast at this. There is extremely little
46	added security in multiple checksums as noted by Joux [J04]. For any set
47	of checksums, the actual strength lies in that of the strongest
48	checksum.
49
50	Wang et al [W04] extended Joux's [J04] work on SHA-0 to cover MD4, MD5,
51	HAVAL-128 and RIPEMD families of hashes.
52
53	How fast can MD5 be broken?
54	---------------------------
55	For a general collision, not a pre-image attack, since the announcement
56	by Wang et al [W04], the time required to break MD5 has been massively
57	reduced. Originally at 1 hour on a near-supercomputer (IBM P690) and
58	estimated at 64 hours with a Pentium-3 1.7Ghz. This has gone down to
59	less than in two years, to 17 seconds [K06a].
60
61	08/2004 - 1 hour, IBM pSeries 690 (32x 1.7Ghz POWER4+) = 54.4 GHz-Hours
62	03/2005 - 8 hours, Pentium-M 1.6Ghz = 12.8 Ghz-Hours
63	11/2005 - 5 hours, Pentium-4 1.7Ghz = 8.5 Ghz-Hours
64	03/2006 - 1 minute, Pentium-4 3.2Ghz = .05 Ghz-Hours
65	04/2006 - 17 seconds, Pentium-4 3.2Ghz = .01 Ghz-Hours
66
67	If we accept a factor of 800x as a sample of how much faster a checksum
68	may be broken over the course of 2 years (MD5 using the above data is
69	>2000x), then existing checksums do not stand a significant chance of
70	survival in the future. We should thus accept that whatever checksums we
71	are using today, will be broken in the near future, and plan as best as
72	possible. (A brief review [H04] of the SHA1 attacks indicates an
73	improvement of ~600x in the same timespan).
74
75	And for those that claim implementation of these procedures is not yet
76	feasible, see [K06b] for an application that can produce two
77	self-extracting EXE files, with identical MD5s, and whatever payload you
78	want.
79
80	The good news
81	-------------
82	Of the checksums presently used by Manifest2 (SHA1, SHA256, RIPEMD160),
83	one stands close to being completely broken: SHA1; and another is
84	significantly weakened: RIPEMD160. The SHA2 series has suffered some
85	attacks, but still remains reasonably solid [G07],[K08].
86
87	To reduce the potential for future problems and any single checksum
88	break leading to a rapid decrease in security, we should incorporate the
89	strongest hash available from each family of checksums, and be prepared
90	to retire old checksums actively, unless there is a overriding reason to
91	keep a specific checksum, such as part of a migration plan.
92
93	What should be done
94	-------------------
95	Portage should always try to verify all supported hashes that are
96	available in a Manifest2, starting with the strongest ones as maintained
97	by a preference list. Over time, the weaker checksums should be removed
98	from Manifest2 files, once all old Portage installations have had
99	sufficient time to upgrade. We should be prepared to add stronger
100	checksums wherever possible, and to remove those that have been
101	defeated.
102
103	As soon as feasible, we should add the SHA512 and WHIRLPOOL algorithms.
104	In future, as stream-based checksums are developed (in response to the
105	development by NIST [AHS]), they should be considered and used.
106
107	The SHA512 algorithm is available in Python 2.5, which has been a
108	dependency of Portage since approximately Portage 2.1.6.13.
109
110	The WHIRLPOOL checksum is not available within the PyCrypto library or
111	hashlib that is part of Python 2.5, but there are multiple alternative
112	Python implementations available, ranging from pure Python to C-based
113	(python-mhash).
114
115	The existence unsupported hash is not considered to be a failure unless
116	no supported hashes are available for a given Manifest entry.
117
118	Checksum depreciation timing
119	----------------------------
120	For the current Portage, both SHA1 and RIPEMD160 should be immediately
121	removed, as they present no advantages over the already present SHA256.
122	SHA256 cannot be replaced immediately with SHA512, as existing Portage
123	versions need at least one supported algorithm present (SHA256 support
124	was added in June 2006), so it must be retained for some while.
125
126	Immediately:
127	- Add WHIRLPOOL and SHA512.
128	- Remove SHA1 and RIPEMD160.
129
130	After the majority of Portage installations include SHA512 support:
131	- Remove SHA256.
132
133	Backwards Compatibility
134	=======================
135	Old versions of Portage may support and expect only specific checksums.
136	This is accounted for in the checksum depreciation discussion.
137
138	For maximum compatiability, we should only have to include each of the
139	old algorithms that we are officially still supporting, as well as the
140	new ones that we prefer.
141
142	References
143	==========
144
145	[AHS] NIST (2007). "NIST's Plan for New Cryptographic Hash Functions",
146	(Advanced Hash Standard). http://csrc.nist.gov/pki/HashWorkshop/
147
148	[BOBO06] Boneh, D. and Boyen, X. (2006). "On the Impossibility of
149	Efficiently Combining Collision Resistant Hash Functions"; Proceedings
150	of CRYPTO 2006, Dwork, C. (Ed.); Lecture Notes in Computer Science
151	4117, pp. 570-583. Available online from:
152	http://crypto.stanford.edu/~dabo/abstracts/hashing.html
153
154	[H04] Hawkes, P. and Paddon, M. and Rose, G. (2004). "On Corrective
155	Patterns for the SHA-2 Family". CRYPTO 2004 Cryptology ePrint Archive,
156	Report 2004/204. Available online from:
157	http://eprint.iacr.org/2004/207.pdf
158
159	[J04] Joux, Antoie. (2004). "Multicollisions in Iterated Hash
160	Functions - Application to Cascaded Constructions;" Proceedings of
161	CRYPTO 2004, Franklin, M. (Ed); Lecture Notes in Computer Science
162	3152, pp. 306-316. Available online from:
163	http://web.cecs.pdx.edu/~teshrim/spring06/papers/general-attacks/multi-joux.pdf
164
165	[K06a] Klima, V. (2006). "Tunnels in Hash Functions: MD5 Collisions
166	Within a Minute". Cryptology ePrint Archive, Report 2006/105.
167	Available online from: http://eprint.iacr.org/2006/105.pdf
168
169	[K06b] Klima, V. (2006). "Note and links to high-speed MD5 collision
170	proof of concept tools". Available online from:
171	http://cryptography.hyperlink.cz/2006/trick.txt
172
173	[K08] Klima, V. (2008). "On Collisions of Hash Functions Turbo SHA-2".
174	Cryptology ePrint Archive, Report 2008/003. Available online from:
175	http://eprint.iacr.org/2008/003.pdf
176
177	[G07] Gligoroski, D. and Knapskog, S.J. (2007). "Turbo SHA-2".
178	Cryptology ePrint Archive, Report 2007/403. Available online from:
179	http://eprint.iacr.org/2007/403.pdf
180
181	[W04] Wang, X. et al: "Collisions for Hash Functions MD4, MD5,
182	HAVAL-128 and RIPEMD", rump session, CRYPTO 2004, Cryptology ePrint
183	Archive, Report 2004/199, first version (August 16, 2004), second
184	version (August 17, 2004). Available online from:
185	http://eprint.iacr.org/2004/199.pdf
186
187	Thanks to
188	=========
189	I'd like to thank the following folks, in no specific order:
190	- Ciaran McCreesh (ciaranm) - for pointing out the Joux (2004) paper,
191	and also being stubborn enough in not accepting a partial solution.
192	- Marius Mauch (genone), Zac Medico (zmedico) and Brian Harring
193	(ferringb): for being knowledgeable about the Portage Manifest2
194	codebase.
195
196	Copyright
197	=========
198	Copyright (c) 2006-2010 by Robin Hugh Johnson. This material may be
199	distributed only subject to the terms and conditions set forth in the
200	Open Publication License, v1.0.
201
202	vim: tw=72 ts=2 expandtab: