LG webOS
TV
| name | value | 
|---|---|
| model | 43UH6100 | 
| product | 3.0 | 
| firmware | 4.30.40 | 
| features | app marketplace, live TV listings | 
| vulnerabilities | all phone-home calls are done over HTTP | 
the 43UH6100 is a 'smart' TV, running LG's webOS
since it is a fair assumption it is running OpenWrt underneath, the original goal was rooting the device, but initial investigations showed some other interesting vectors
digging
nmap
from nmap -PN -sV <device>, we get:
PORT     STATE SERVICE  VERSION
1175/tcp open  upnp
3000/tcp open  http     LG smart TV http service
3001/tcp open  ssl/http LG smart TV http service
9998/tcp open  http     Google Chromecast httpd
aside from the obvious flag running of both HTTP and HTTPS versions of (likely) the same service, interested to see that the Chromecast plugged in to the TV is also being exposed on the same IP as the TV
since there is an LG smart TV app available for Android/iOS, assuming that there is an API of some sort running on 3000 or 3001, so:
$ curl http://<device>:3000
Hello world
we see the same response on 3001, but have to use -k as the device uses a self-signed certificate
so, something is there, we just don't know how to talk to it yet
sniffing
switching tactics and connected the TV to a wireless network that has a tap, and we start to see some interesting things:
on boot
every time the TV starts up, within 30 seconds, it calls home:
POST /CheckSWAutoUpdate.laf HTTP/1.1
Accept: */*
User-Agent: User-Agent: Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1; SV1)
Host: snu.lge.com:80
Connection: Keep-Alive
Content-type: application/x-www-form-urlencoded
Content-Length: 572
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
HTTP/1.1 200 OK
Date: Wed, 16 Nov 2016 08:23:56 GMT
Content-length: 508
Content-type: application/octet-stream;charset=UTF-8
Pragma: no-cache;
Expires: -1;
Content-Transfer-Encoding: binary;
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
that looks a lot like base64 encoded data, and when decoded, yields
request:
<REQUEST>
  <PRODUCT_NM>webOSTV 3.0</PRODUCT_NM>
  <MODEL_NM>HE_DTV_W16P_AFADATAA</MODEL_NM>
  <SW_TYPE>FIRMWARE</SW_TYPE>
  <MAJOR_VER>04</MAJOR_VER>
  <MINOR_VER>30.40</MINOR_VER>
  <COUNTRY>US2</COUNTRY>
  <COUNTRY_GROUP>US</COUNTRY_GROUP>
  <DEVICE_ID>de:ad:be:ef:ca:fe</DEVICE_ID>
  <AUTH_FLAG>N</AUTH_FLAG>
  <IGNORE_DISABLE>N</IGNORE_DISABLE>
  <ECO_INFO>01</ECO_INFO>
  <CONFIG_KEY>00</CONFIG_KEY>
  <LANGUAGE_CODE>en-US</LANGUAGE_CODE>
</REQUEST>
pretty standard, but the auth_flag, ignore_disable and config_key values are potentially interesting
response:
<RESPONSE>
  <RESULT_CD>900</RESULT_CD>
  <MSG>Success</MSG>
  <REQ_ID>00000000000000000001</REQ_ID>
  <IMAGE_URL></IMAGE_URL>
  <IMAGE_SIZE></IMAGE_SIZE>
  <IMAGE_NAME></IMAGE_NAME>
  <UPDATE_MAJOR_VER></UPDATE_MAJOR_VER>
  <UPDATE_MINOR_VER></UPDATE_MINOR_VER>
  <FORCE_FLAG></FORCE_FLAG>
  <KE></KE>
  <GMT>16 Nov 2016 08:23:56 GMT</GMT>
  <ECO_INFO>01</ECO_INFO>
  <CDN_URL></CDN_URL>
  <CONTENTS></CONTENTS>
</RESPONSE>
much more interesting than the request:
| key | assumption | 
|---|---|
IMAGE_URL | 
the URL of a firmware update | 
IMAGE_SIZE | 
the size of the firmware update - are they doing this instead of checksum? | 
IMAGE_NAME | 
the name of the firmware update - not sure why this is necessary | 
UPDATE_MAJOR_VER | 
the major version of the firmware update | 
UPDATE_MINOR_VER | 
the minor version of the firmware update | 
FORCE_FLAG | 
whether or not to force the update - unclear if true | 
CDN_URL | 
URL that the firmware update is available at | 
CONTENTS | 
none | 
channel search
when configuring the cable connections, the TV makes a number of calls:
request:
GET /fts/gftsDownload.lge?biz_code=IBS&func_code=ONLINE_EPG_FILE&file_path=/ibs/online/epg_file/20161116/f_1479280636996tmsepgcrawler_merged000004417_201611160600_06_20161116070000.zip HTTP/1.1
Host: aic-ngfts.lge.com
Accept: */*
response:
HTTP/1.1 200 OK
Server: Apache
Content-Disposition: attachment; filename="f_1479280636996tmsepgcrawler_merged000004417_201611160600_06_20161116070000.zip"
Content-Transfer-Encoding: binary;
Last-Modified: Wed, 16 Nov 2016 07:25:17 GMT
Content-Length: 135700
Content-Type: application/octet-stream;charset=UTF-8
Date: Wed, 16 Nov 2016 08:24:01 GMT
Connection: keep-alive
parameters in request:
| parameter | assumption | 
|---|---|
biz_code | 
none | 
func_code | 
none | 
file_path | 
none | 
looking at the file path, if not in a chroot'd environment, potential for ~LFI - attempts thus far have shown nothing but 404
looking at the file itself:
$ curl -o foo "http://aic-ngfts.lge.com/fts/<path>"
  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                 Dload  Upload   Total   Spent    Left  Speed
100  132k  100  132k    0     0   230k      0 --:--:-- --:--:-- --:--:--  230k
$ file foo
foo: Zip archive data, at least v2.0 to extract
$ unzip foo
Archive:  foo
  inflating: schedule.json
  inflating: program.json
schedule.json
sample entry:
{
  "dbAction": "I",
  "schdId": "100006/EP010865380045/2016-11-11-10:00",
  "contentId": "EP010865380045",
  "seqNo": "0",
  "chanCode": "100006",
  "strtTime": "2016,11,11,10,00,00",
  "strtTimeLong": 1478858400,
  "endTime": "2016,11,11,12,00,00",
  "endTimeLong": 1478865600,
  "schdSummary": "",
  "timeType": "",
  "schdPgmTtl": "Late Night Gifts",
  "schdSubTtl": "Lisa Rinna",
  "rebrdcstFlag": "Y",
  "capFlag": "",
  "liveFlag": "",
  "dataBrdcstFlag": "",
  "scExplnBrdcstFlag": "",
  "scQualityGbn": "",
  "signBrdcstFlag": "",
  "voiceMultiBrdcstCount": "",
  "threeDFlag": "",
  "schdAdultClassCode": "-1",
  "schdAgeGrdCode": "TVG",
  "pgmGrId": "SH010865380000",
  "genreCode": "61",
  "realEpsdNo": "0"
}
program.json
{
  "dbAction": "I",
  "contentId": "EP000000510045",
  "seqNo": "0",
  "pgmGrId": "SH000000510000",
  "connectorId": "1013932",
  "serId": "184628",
  "serNo": "",
  "seasonId": "7895341",
  "seasonNo": "3",
  "pgmType": "Series",
  "realEpsdNo": "1",
  "summary": "Whitley encounters a new Dwayne on the plane ride back to school.",
  "pgmImgUrlName": "http://ngfts.lge.com/fts/gftsDownload.lge?biz_code=IBS&func_code=TMS_PROGRAM_IMG&file_path=/ibs/tms/program_img/p184628_b_v7_ab.jpg",
  "orgGenreType": "",
  "orgGenreCode": "188",
  "oGenreCode": "2",
  "oGenreType": "",
  "subGenreType": "",
  "subGenreCode": "",
  "makeCom": "",
  "makeCntry": "",
  "makeYear": "1989-09-28",
  "usrPplrSt": "",
  "pplrSt": "",
  "audLang": "en",
  "dataLang": "ENG",
  "audQlty": "",
  "genreImgUrl": "http://aic-ngfts.lge.com/fts/gftsDownload.lge?biz_code=IBS&func_code=GENRE_IMG&file_path=/ibs/genre_img_v/2_36_V_Sitcom.png",
  "vodFlag": "N",
  "pgmImgSize": "V480X720",
  "genreImgSize": "V480X704",
  "lgGenreCode2": "36",
  "lgGenreName2": "Sitcom",
  "programLock": "",
  "castingFlag": "Y"
}
generate_slimmed-aic-json.rb can be used to create a small schedule starting at the current time.
application marketplace
bar
license manager
after an update of an application (and potentially other times), the device calls a different home:
request:
POST /license_manager.asp HTTP/1.1
User-Agent: Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1; SV1;)
Host: us.security.lgtvsdp.com
Content-Length: 210
Content-Type:application/x-www-form-urlencoded
X-Device-Product:webOSTV 3.0
X-Device-Platform:W16P
X-Device-Model:HE_DTV_W16P_AFADATAA
X-Device-Netcast-Platform-Version:3.3.1
X-Device-Eco-Info:1
X-Device-Country-Group:US
X-Device-Publish-Flag:Y
X-Device-ContentsQA-Flag:Y
X-Device-FW-Version:04.30.40
X-Device-SDK-VERSION:3.3.1
X-Device-ID:<redacted>
X-Device-Type:T01
X-Device-Language:en-US
X-Device-Country:US
X-Device-Remote-Flag:N
X-Authentication:<redacted>
mode=issuelicense4pre&sid=1827712162&deviceid=<redacted>>&devicemodel=webostv&p=D1609DEB7189B744D4BC272550CBF5BF&g=5&A=52013FFC91EA5A6F41BE025B5E4461FB&hmac=OnSGJj7D3yth5HPdafdtnArDKYc%3D
nothing in the request really jumps out
response:
HTTP/1.1 200 OK
Cache-Control: private
Content-Length: 935
Content-Type: text/html
Server: Microsoft-IIS/7.5
Set-Cookie: ASPSESSIONIDAARCQAST=DOLHEFIBIEAONHPFCIFPECDL; path=/
X-Powered-By: ASP.NET
Date: Tue, 15 Nov 2016 19:00:39 GMT
<?xml version='1.0' encoding='utf-8'?><response result='0' message=''><responsedata>B=957172C7AF8EFA66326A7639D1C5301B;license=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;hmac=l0kBybteRX6bdSGjD/w0LV86MVU=</responsedata></response>
breaking down the XML response:
<?xml version='1.0' encoding='utf-8'?>
<response result='0' message=''>
  <responsedata>
    B=957172C7AF8EFA66326A7639D1C5301B;
    license=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;
    hmac=l0kBybteRX6bdSGjD/w0LV86MVU=
  </responsedata>
</response>
both license and hmac values are obviously hashes, but have been unable to determine what kind
impersonating
most (all?) of this data is based on impersonate-lge.rb interactions
OS update
impersonate-lge.rb catches the POST to /CheckSWAutoUpdate.laf, changes:
| key | value | 
|---|---|
image_url | 
http://snu.lge.com/fizbuzz | 
image_size | 
400 | 
image_name | 
fizzbuzz | 
update_major_ver | 
04 | 
update_minor_ver | 
30.50 | 
force_flag | 
Y | 
cdn_url | 
http://snu.lge.com/fizzbuzz | 
contents | 
'' | 
since the update_minor_ver specified is greater than the existing value (30.40), the TV prompts the user that an upgrade is available.
the traffic after the user chooses to upgrade starts with a GET of the image_url:
GET /fizzbuzz HTTP/1.1
Accept: */*
Host: snu.lge.com
Range: bytes=0-1715
Connection: Closed
followed by 5 retries, since they all received 404 as we're not sure what the format of the update actually is (yet), but assume it will be an .ipk as well.
then some base64 encoded data with a log :
POST /SWDownloadStartLog.laf HTTP/1.1
Accept: */*
User-Agent: User-Agent: Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1; SV1)
Host: snu.lge.com:80
Connection: Keep-Alive
Content-type: application/x-www-form-urlencoded
Content-Length: 268
PFJFUVVFU1Q+CjxSRVFfSUQ+MDAwMDAwMDAwMDg2MTMyNDQ2NjA8L1JFUV9JRD4KPFBST0RVQ1RfTk0+d2ViT1NUViAzLjA8L1BST0RVQ1RfTk0+CjxNT0RFTF9OTT5IRV9EVFZfVzE2UF9BRkFEQVRBQTwvTU9ERUxfTk0+CjxTV19UWVBFPkZJUk1XQVJFPC9TV19UWVBFPgo8SU1BR0VfTkFNRT5maXp6YnV6ejwvSU1BR0VfTkFNRT4KPC9SRVFVRVNUPgo=
decoded:
<REQUEST>
  <REQ_ID>00000000008613244660</REQ_ID>
  <PRODUCT_NM>webOSTV 3.0</PRODUCT_NM>
  <MODEL_NM>HE_DTV_W16P_AFADATAA</MODEL_NM>
  <SW_TYPE>FIRMWARE</SW_TYPE>
  <IMAGE_NAME>fizzbuzz</IMAGE_NAME>
</REQUEST>
and then some similar data to a different endpoint:
POST /DownloadResult.laf HTTP/1.1
Accept: */*
User-Agent: User-Agent: Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1; SV1)
Host: snu.lge.com:80
Connection: Keep-Alive
Content-type: application/x-www-form-urlencoded
Content-Length: 308
PFJFUVVFU1Q+CjxSRVFfSUQ+MDAwMDAwMDAwMDg2MTMyNDQ2NjA8L1JFUV9JRD4KPFBST0RVQ1RfTk0+d2ViT1NUViAzLjA8L1BST0RVQ1RfTk0+CjxNT0RFTF9OTT5IRV9EVFZfVzE2UF9BRkFEQVRBQTwvTU9ERUxfTk0+CjxTV19UWVBFPkZJUk1XQVJFPC9TV19UWVBFPgo8VVBEQVRFX1JFU1VMVD43MjI8L1VQREFURV9SRVNVTFQ+CjxSRVRSWV9DT1VOVD4wPC9SRVRSWV9DT1VOVD4KPC9SRVFVRVNUPgo=
decoded:
<REQUEST>
  <REQ_ID>00000000008613244660</REQ_ID>
  <PRODUCT_NM>webOSTV 3.0</PRODUCT_NM>
  <MODEL_NM>HE_DTV_W16P_AFADATAA</MODEL_NM>
  <SW_TYPE>FIRMWARE</SW_TYPE>
  <UPDATE_RESULT>722</UPDATE_RESULT>
  <RETRY_COUNT>0</RETRY_COUNT>
</REQUEST>
so, now we know what the process is, just need to determine what the format/contents of the OS update is.
after shutting down impersonate-lge.com.rb, the real snu.lge.com responds to /CheckSWAutoUpdate.laf with:
GET /GlobalSWDownloadCdn.laf?IMG=/<redacted>-prodkey_nsu_V3_SECURED.epk HTTP/1.1
Accept: */*
Host: su.lge.com:80
Range: bytes=0-1715
Connection: Closed
taking a look at the (850mb) file:
$ binwalk -v --dd='.*' <redacted>-prodkey_nsu_V3_SECURED.epk
Scan Time:     2016-12-28 22:41:37
Target File:   <redacted>-prodkey_nsu_V3_SECURED.epk
MD5 Checksum:  eadf4625c8033f286f7459766558d43b
Signatures:    344
DECIMAL       HEXADECIMAL     DESCRIPTION
--------------------------------------------------------------------------------
1437257       0x15EE49        HPACK archive data
88501492      0x5466CF4       StuffIt Deluxe Segment (data): f
116751487     0x6F57C7F       VMware4 disk image
151796947     0x90C3CD3       LANCOM OEM file
184522619     0xAFF977B       MySQL ISAM compressed data file Version 4
188949815     0xB432537       QEMU QCOW Image
202964337     0xC18FD71       MySQL ISAM compressed data file Version 8
360991579     0x15844B5B      MySQL ISAM compressed data file Version 9
403720767     0x18104A3F      MySQL ISAM compressed data file Version 5
438498638     0x1A22F54E      Cisco IOS experimental microcode, for ""
558916980     0x21506574      QEMU QCOW Image
652690023     0x26E74267      COBALT boot rom data (Flat boot rom or file system)
673373671     0x2822DDE7      StuffIt Deluxe Segment (data): f
752461107     0x2CD9A533      MySQL ISAM index file Version 11
798709823     0x2F9B583F      LANCOM OEM file
828143551     0x315C77BF      MySQL ISAM index file Version 11
828353910     0x315FAD76      MySQL ISAM compressed data file Version 4
however, given the 'encrypted' portion of the filename and the fact that none of the files are actually usable as the type indicated here
- the encryption is throwing off 
binwalkfile type detection 
attempting to find an unencrypted version of the file by fuzzing the original URL has, so far, proved unsuccessful.
TODO how would we determine the type of encryption in order to start attacking it?
channel guide
in _public/aic/_source/slimmed/schedule.json, changed:
| key | value | 
|---|---|
schdSummary | 
h4ck the planet | 
schdPgmTtl | 
h4ck the planet | 
schdSubTtl | 
h4ck the planet | 
in _public/aic/_source/slimmed/program.json, changed:
| key | value | 
|---|---|
contentId | 
EP022959710001 | 
genreImgUrl | 
http://aic-gfts.lge.com/aic/hacktheplanet.jpg | 
pgmGrId | 
SH022959710000 | 
pgmImgUrlName | 
http://aic-gfts.lge.com/aic/hacktheplanet.jpg | 
summary | 
h4ck the planet | 
contentId and pgmGrId were changed to make them line up with changes made to schedule.json
application update
fizzbang