PowerPoint - Common Vulnerabilities and Exposures

advertisement
Common Vulnerabilities
and Exposures
Steve Christey
Margie Zuk
June 22, 2000
Context: The Vulnerability Life Cycle
•Mailing lists, Newsgroups, Hacker sites
Start Here
Discovery
•Incident
Response Incident
Teams
Handling
•Incident
Reports
•Academic
Analysis
Study
•Advisories
CVE
•Intrusion
Detection Detection
Systems
Collection
Protection
•Vulnerability Assessment Tools
•Databases
•Newsletters
A Roadblock to Information Sharing:
Same Problem, Different Names
Organization Name
CERT
CyberSafe
ISS
AXENT
Bugtraq
BindView
Cisco
IBM ERS
CERIAS
L-3
CA-96.06.cgi_example_code
Network: HTTP ‘phf’ Attack
http-cgi-phf
phf CGI allows remote command execution
PHF Attacks – Fun and games for the whole family
#107 – cgi-phf
#3200 – WWW phf attack
Vulnerability in NCSA/Apache Example Code
http_escshellcmd
#180 HTTP Server CGI example code compromises http server
The Implications
 Difficult to correlate data across multiple organizations
and tools
- E.g. IDS and assessment tools
- E.g. security tools and fix information
- Incident information
 Difficult to conduct a detailed comparison of tools or
databases
- Vulnerabilities are counted differently
- Which is more comprehensive?
Common Vulnerabilities and Exposures (CVE):
One Common Language
Name
Description
CVE-1999-0003
ToolTalk (rpc.ttdbserverd) buffer
overflow
CVE-1999-0006
Buffer overflow in qpopper
CVE-1999-0067
Shell metacharacters in phf
CVE-1999-0344
Windows NT debug-level access
bug (a.k.a. Sechole)
 Lists all publicly known security problems
 Assigns unique identifier to each problem
 Remains independent of multiple perspectives
 Is publicly open and shareable
 Community-wide effort via the CVE Editorial Board
Addressing Common Misconceptions of CVE
 Not a full-fledged vulnerability database
- Simplicity avoids competition, limits debate
- Intended for use by vulnerability database maintainers
 Not a taxonomy or classification scheme
 Focuses on vulnerabilities instead of attacks
- Does not cover activities such as port mapping
 Not just “vulnerabilities” in the classical sense
- Definitions of “vulnerability” vary greatly
- “Exposure” covers a broader notion of “vulnerability”
 Competing vendors are working together to adopt CVE
CVE Editorial Board
 Members from 25 different organizations including
researchers, tool vendors, response teams, and end
users
 Mostly technical representatives
 Review and approve CVE entries
 Discuss issues related to CVE maintenance
 Monthly meetings (face-to-face or phone)
 Publicly viewable mailing list archives
Active Editorial Board Members
(as of June 19, 2000)
Tool Vendors
David Balenson - NAI
Andy Balinsky - Cisco
Scott Blake - BindView
Andre Frech - ISS
Kent Landfield - info-ops.com
Jim Magdych - NAI
David Mann - BindView
Craig Ozancin - AXENT
Paul E. Proctor - CyberSafe
Mike Prosser - Symantec
Marcus Ranum - NFR
Steve Schall - Intrusion.com
Tom Stracener - Hiverworld
Bill Wall - Harris
Kevin Ziese - Cisco
OS Vendors
David LeBlanc - Microsoft
Casper Dik - Sun
MITRE
Dave Baker, Steve Christey,
Bill Hill
Response Teams
Ken Armstrong - CanCERT
Bill Fithen - CERT Coordination Center
Scott Lawler - DOD-CERT
Academic/Educational
Matt Bishop - UC Davis Computer Security Lab
Pascal Meunier - Purdue University CERIAS
Alan Paller - SANS Institute
Gene Spafford - Purdue University CERIAS
Network Security
Eric Cole - Vista IT
Kelly Cooper - GTE Internetworking
Information Providers
Russ Cooper - NTBugtraq
Elias Levy - Bugtraq, Security Focus
Ron Nguyen - Ernst and Young
Ken Williams - eSecurityOnline.com
Other Security Analysts
Steve Northcutt - SANS
Adam Shostack - Zero-Knowledge Systems
Stuart Staniford - Silicon Defense
CVE Enables Detailed Product Comparisons
CVE
Name
CVE-XXXX-0001
Tool
A
X
Tool
B
X
DB
1
CVE-XXXX-0002
X
X
X
CVE-XXXX-0003
CVE-XXXX-0004
X
X
DB
2
X
Hacker
Site
X
X
X
X
Using CVE from Advisories to IDSes
Do my systems
have these
problems?
Popular
Attacks
CVE-1
CVE-2
CVE-3
CVE-4
Which tools
test for these
problems?
Tool 1
CVE-1
CVE-2
CVE-3
Does my IDS
have the
signatures?
IDS
CVE-1
CVE-3
CVE-4
Tool 2
CVE-3
CVE-4
I can’t detect exploits
of CVE-2 - how well
does Tool 1 check for it?
Using CVE from Attacks to Incident Recovery
I detected
an attack on CVE-3.
Did my assessment
say my system
has the problem?
Tool 2
CVE-3 Tool 1
CVE-4CVE-1
CVE-2
CVE-3
YES
Public
Databases
Clean up
Close the hole CVE-2
CVE-3
Advisories
Report the
CVE-1
incident
CVE-2
NO
CVE-3
Don’t send an alarm
But the attack succeeded!
Tell your assessment vendor
Go to YES
CVE Compatibility
 Ensures that a tool or database can “speak CVE” and
correlate data with other CVE-compatible products
 Requirements
- Find items by CVE name (CVE searchable)
- Include CVE name in output for each item (CVE output)
- Provide MITRE with database items that are not in CVE yet
Good faith effort to keep mappings accurate
 25 organizations have declared their intentions to make their
products CVE compatible
-
Organizations Working Toward CVE
Compatibility
- Advanced Research Corp.
*- Alliance Qualité Logiciel
*- AXENT
*- BindView
- CERIAS/Purdue University
- CERT
- Cisco
*-- CyberSafe
Cyrano
Ernst and Young
*- Harris Corp.
*- Hiverworld, Inc.
- Intrusion.com
*
=
- Internet Security Systems
*- Nessus Project
*- Network Associates
- Network Security Wizards
*- NIST
*- NTBugtraq
Institute
*-- SANS
Security Focus - Bugtraq
*- Symantec (L-3)
- UC Davis
White Hats
*- World Wide Digital Security
*
CVE already being used in a product
Adding New Entries to CVE
 Board member submits raw information to MITRE
 Submissions are grouped, refined, and proposed back to
the Board as candidates
Form: CAN-YYYY-NNNN
- Strong likelihood of becoming CVE-YYYY-NNNN
 Not a guarantee
- Delicate balance between timeliness and accuracy
 Board reviews and votes on candidates
- Accept, modify, recast, reject, reviewing
 If approved, the candidate becomes a CVE entry
 Entry is included in a subsequent CVE version
Published on CVE web site
 Entries may later be modified or deprecated
-
-
Stages of Security Information in CVE
Submissions
Candidates
•Raw information
•Obtained from MITRE,
Board members, and
other data feeds
•Combined and refined
…..
….. …..
…..
…..
…..
…..
…..
•Placed in clusters
•Proposed to Editorial Board
•Accepted or rejected
•Backmap tells submitters what
candidates were assigned to
their submissions
Entries
•Added to CVE list
•Submissions, candidates
removed from the “pool”
•Published in an official CVE
version
CAN-2000-0001
CVE-2000-0001
CAN-2000-0002
<REJECTED>
CAN-2000-0003
CVE-2000-0003
Back-map
Content Decisions
 Explicit guidelines for content of CVE entries
- Ensure consistency within CVE
- Provide “lessons learned” for researchers
 Three basic types
- Inclusion
 What
goes into CVE? What doesn’t, and why?
 Example: weak encryption, bugs in beta code
- Level of Abstraction
 Example: default passwords, or multiple bugs in
the same application - one or many entries?
Format
 Example: what goes into a CVE description?
 Challenge: what to do with incomplete information?
-
The CVE Strategy
Unreviewed
Bugtraqs, Mailing lists,
Hacker sites
Discovery
Products
Policy
time
Reviewed Advisories
CERT, CIAC,
Vendor advisories
Scanners,
Intrusion Detection,
Vulnerability Databases
Methodologies
Purchasing
Requirements
Education
1. Inject Candidate
numbers into advisories
2. Establish CVE at product
level in order to...
3. … enable CVE to permeate
the policy level.
CVE Status as of June 2000
 Latest CVE version: 20000602
- 700 entries
 722 additional candidates being reviewed by Editorial Board
 Received vulnerability databases from 10 organizations
- Will help create more legacy candidates
- Processing 10,000 submissions (database items)
 May
produce over 1000 additional candidates?
 Candidate numbers used in 5 security advisories
 CVE names included in Top Internet Security Threats list
 Editorial Board discussing content decisions
- Affects ~300 candidates
Future Directions
 Add more entries to CVE
- Goal: 1000 entries by September 2000
- Add entries for all 1999, 2000 advisories
 Add more candidates
- Goals: 1000 new candidates by September 2000
 Cover 80% of items in each participating tool or database
 Use CVE identifiers in advisories, newsletters, etc.
 Use CVE in deeper product analysis
 Work with users and vendors to establish CVE as a de
facto standard
For More Information
http://cve.mitre.org
Download