[bitcoin-dev] CVE-2017-18350 disclosure

LORD HIS EXCELLENCY JAMES HRMH willtech at live.com.au
Fri Nov 8 17:03:02 UTC 2019


It goes without saying in that all privately known CVE should be handled so professionally but, that is, well done team.

Regards,
LORD HIS EXCELLENCY JAMES HRMH


________________________________
From: bitcoin-dev-bounces at lists.linuxfoundation.org <bitcoin-dev-bounces at lists.linuxfoundation.org> on behalf of Luke Dashjr via bitcoin-dev <bitcoin-dev at lists.linuxfoundation.org>
Sent: Saturday, 9 November 2019 2:07 AM
To: bitcoin-dev at lists.linuxfoundation.org <bitcoin-dev at lists.linuxfoundation.org>
Cc: security at bitcoincore.org <security at bitcoincore.org>
Subject: [bitcoin-dev] CVE-2017-18350 disclosure

CVE-2017-18350 is a buffer overflow vulnerability which allows a malicious
SOCKS proxy server to overwrite the program stack on systems with a signed
`char` type (including common 32-bit and 64-bit x86 PCs).

The vulnerability was introduced in 60a87bce873ce1f76a80b7b8546e83a0cd4e07a5
(SOCKS5 support) and first released in Bitcoin Core v0.7.0rc1 in 2012 Aug 27.
A fix was hidden in d90a00eabed0f3f1acea4834ad489484d0012372 ("Improve and
document SOCKS code") released in v0.15.1, 2017 Nov 6.

To be vulnerable, the node must be configured to use such a malicious proxy in
the first place. Note that using *any* proxy over an insecure network (such
as the Internet) is potentially a vulnerability since the connection could be
intercepted for such a purpose.

Upon a connection request from the node, the malicious proxy would respond
with an acknowledgement of a different target domain name than the one
requested. Normally this acknowledgement is entirely ignored, but if the
length uses the high bit (ie, a length 128-255 inclusive), it will be
interpreted by vulnerable versions as a negative number instead. When the
negative number is passed to the recv() system call to read the domain name,
it is converted back to an unsigned/positive number, but at a much wider size
(typically 32-bit), resulting in an effectively infinite read into and beyond
the 256-byte dummy stack buffer.

To fix this vulnerability, the dummy buffer was changed to an explicitly
unsigned data type, avoiding the conversion to/from a negative number.

Credit goes to practicalswift (https://twitter.com/practicalswift) for
discovering and providing the initial fix for the vulnerability, and Wladimir
J. van der Laan for a disguised version of the fix as well as general cleanup
to the at-risk code.

Timeline:
- 2012-04-01: Vulnerability introduced in PR #1141.
- 2012-05-08: Vulnerability merged to master git repository.
- 2012-08-27: Vulnerability published in v0.7.0rc1.
- 2012-09-17: Vulnerability released in v0.7.0.
...
- 2017-09-21: practicalswift discloses vulnerability to security team.
- 2017-09-23: Wladimir opens PR #11397 to quietly fix vulernability.
- 2017-09-27: Fix merged to master git repository.
- 2017-10-18: Fix merged to 0.15 git repository.
- 2017-11-04: Fix published in v0.15.1rc1.
- 2017-11-09: Fix released in v0.15.1.
...
- 2019-06-22: Vulnerability existence disclosed to bitcoin-dev ML.
- 2019-11-08: Vulnerability details disclosure to bitcoin-dev ML.
_______________________________________________
bitcoin-dev mailing list
bitcoin-dev at lists.linuxfoundation.org
https://lists.linuxfoundation.org/mailman/listinfo/bitcoin-dev
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.linuxfoundation.org/pipermail/bitcoin-dev/attachments/20191108/aace00bf/attachment-0001.html>


More information about the bitcoin-dev mailing list