Similarities with newly found Linux malware utilized in Operation DreamJob corroborate the idea that the notorious North Korea-aligned group is behind the 3CX supply-chain assault
ESET researchers have found a brand new Lazarus Operation DreamJob marketing campaign focusing on Linux customers. Operation DreamJob is the title for a collection of campaigns the place the group makes use of social engineering methods to compromise its targets, with pretend job affords because the lure. On this case, we had been capable of reconstruct the total chain, from the ZIP file that delivers a pretend HSBC job provide as a decoy, up till the ultimate payload: the SimplexTea Linux backdoor distributed by way of an OpenDrive cloud storage account. To our data, that is the primary public point out of this main North Korea-aligned risk actor utilizing Linux malware as a part of this operation.
Moreover, this discovery helped us affirm with a excessive stage of confidence that the current 3CX supply-chain assault was the truth is performed by Lazarus – a hyperlink that was suspected from the very starting and demonstrated by a number of safety researchers since. On this blogpost, we corroborate these findings and supply further proof concerning the connection between Lazarus and the 3CX supply-chain assault.
The 3CX supply-chain assault
3CX is a global VoIP software program developer and distributor that gives telephone system companies to many organizations. In line with its web site, 3CX has greater than 600,000 prospects and 12,000,000 customers in varied sectors together with aerospace, healthcare, and hospitality. It supplies consumer software program to make use of its programs through an internet browser, cell app, or a desktop utility. Late in March 2023, it was found that the desktop utility for each Home windows and macOS contained malicious code that enabled a gaggle of attackers to obtain and run arbitrary code on all machines the place the applying was put in. Quickly, it was decided that this malicious code was not one thing that 3CX added themselves, however that 3CX was compromised and that its software program was utilized in a supply-chain assault pushed by exterior risk actors to distribute further malware to particular 3CX prospects.
This cyber-incident has made headlines in current days. Initially reported on March 29th, 2023 in a Reddit thread by a CrowdStrike engineer, adopted by an official report by CrowdStrike, stating with excessive confidence that LABIRINTH CHOLLIMA, the corporate’s codename for Lazarus, was behind the assault (however omitting any proof backing up the declare). Due to the seriousness of the incident, a number of safety firms began to contribute their summaries of the occasions, specifically Sophos, Check Point, Broadcom, Trend Micro, and extra.
Additional, the a part of the assault affecting programs operating macOS was coated intimately in a Twitter thread and a blogpost by Patrick Wardle.
Timeline of occasions

Determine 1. Timeline of occasions associated to the preparation and distribution of 3CX trojanized functions
The timeline exhibits that the perpetrators had deliberate the assaults lengthy earlier than execution; as early as December 2022. This implies they already had a foothold inside 3CX’s community late final 12 months.
Whereas the trojanized 3CX macOS utility exhibits it was signed in late January, we didn’t see the unhealthy utility in our telemetry till February 14th, 2023. It’s unclear whether or not the malicious replace for macOS was distributed previous to that date.
Though ESET telemetry exhibits the existence of the macOS second-stage payload as early as February, we didn’t have the pattern itself, nor metadata to tip us off about its maliciousness. We embrace this data to assist defenders decide how far again programs might need been compromised.
A number of days earlier than the assault was publicly revealed, a mysterious Linux downloader was submitted to VirusTotal. It downloads a brand new Lazarus malicious payload for Linux and we clarify its relationship to the assault later within the textual content.
Attribution of the 3CX supply-chain assault to Lazarus
What’s already revealed
There may be one area that performs a major function in our attribution reasoning: journalide[.]org. It’s talked about in a few of the vendor studies linked above, however its presence isn’t defined. Apparently, articles by SentinelOne and ObjectiveSee don’t point out this area. Neither does a blogpost by Volexity, which even avoided offering attribution, stating “Volexity can’t at the moment map the disclosed exercise to any risk actor”. Its analysts had been among the many first to research the assault in depth and so they created a instrument to extract an inventory of C&C servers from encrypted icons on GitHub. This instrument is helpful, because the attackers didn’t embed the C&C servers instantly within the intermediate phases, however reasonably used GitHub as a lifeless drop resolver. The intermediate phases are downloaders for Home windows and macOS that we denote as IconicLoaders, and the payloads they get as IconicStealer and UpdateAgent, respectively.
On March 30th, Joe Desimone, a safety researcher from Elastic Security, was among the many first to supply, in a Twitter thread, substantial clues that the 3CX-driven compromises are most likely linked to Lazarus. He noticed {that a} shellcode stub prepended to the payload from d3dcompiler_47.dll is just like AppleJeus loader stubs attributed to Lazarus by CISA again in April 2021.
On March 31st it was being reported that 3CX had retained Mandiant to supply incident response companies regarding the supply-chain assault.
On April 3rd, Kaspersky, by way of its telemetry, confirmed a direct relationship between the 3CX supply-chain victims and the deployment of a backdoor dubbed Gopuram, each involving payloads with a standard title, guard64.dll. Kaspersky information exhibits that Gopuram is linked to Lazarus as a result of it coexisted on sufferer machines alongside AppleJeus, malware that was already attributed to Lazarus. Each Gopuram and AppleJeus had been noticed in assaults towards a cryptocurrency firm.
Then, on April 11th, the CISO of 3CX summarized Mandiant’s interim findings in a blogpost. In line with that report, two Home windows malware samples, a shellcode loader known as TAXHAUL and a posh downloader named COLDCAT, had been concerned within the compromise of 3CX. No hashes had been offered, however Mandiant’s YARA rule, named TAXHAUL, additionally triggers on different samples already on VirusTotal:
- SHA-1: 2ACC6F1D4656978F4D503929B8C804530D7E7CF6 (ualapi.dll),
- SHA-1: DCEF83D8EE080B54DC54759C59F955E73D67AA65 (wlbsctrl.dll)
The filenames, however not MD5s, of those samples coincide with these from Kaspersky’s blogpost. Nevertheless, 3CX explicitly states that COLDCAT differs from Gopuram.
The subsequent part accommodates a technical description of the brand new Lazarus malicious Linux payload we lately analyzed, in addition to the way it helped us strengthen the present hyperlink between Lazarus and the 3CX compromise.
Operation DreamJob with a Linux payload
The Lazarus group’s Operation DreamJob includes approaching targets by way of LinkedIn and tempting them with job affords from business leaders. The title was coined by ClearSky in a paper revealed in August 2020. That paper describes a Lazarus cyberespionage marketing campaign focusing on protection and aerospace firms. The exercise has overlap with what we name Operation In(ter)ception, a collection of cyberespionage assaults which were ongoing since no less than September 2019. It targets aerospace, army, and protection firms and makes use of particular malicious, initially Home windows-only, instruments. Throughout July and August 2022, we discovered two situations of Operation In(ter)ception focusing on macOS. One malware pattern was submitted to VirusTotal from Brazil, and one other assault focused an ESET person in Argentina. A number of weeks in the past, a local Linux payload was discovered on VirusTotal with an HSBC-themed PDF lure. This completes Lazarus’s capacity to focus on all main desktop working programs.
On March 20th, a person within the nation of Georgia submitted to VirusTotal a ZIP archive known as HSBC job provide.pdf.zip. Given different DreamJob campaigns by Lazarus, this payload was most likely distributed by way of spearphishing or direct messages on LinkedIn. The archive accommodates a single file: a local 64-bit Intel Linux binary written in Go and named HSBC job provide․pdf.
Apparently, the file extension just isn’t .pdf. It’s because the obvious dot character within the filename is a leader dot represented by the U+2024 Unicode character. The usage of the chief dot within the filename was most likely an try to trick the file supervisor into treating the file as an executable as a substitute of a PDF. This might trigger the file to run when double-clicked as a substitute of opening it with a PDF viewer. On execution, a decoy PDF is exhibited to the person utilizing xdg-open, which can open the doc utilizing the person’s most popular PDF viewer (see Determine 3). We determined to name this ELF downloader OdicLoader, because it has an identical function because the IconicLoaders on different platforms and the payload is fetched from OpenDrive.
OdicLoader drops a decoy PDF doc, shows it utilizing the system’s default PDF viewer (see Determine 2), after which downloads a second-stage backdoor from the OpenDrive cloud service. The downloaded file is saved in ~/.config/guiconfigd (SHA-1: 0CA1723AFE261CD85B05C9EF424FC50290DCE7DF). We name this second-stage backdoor SimplexTea.
Because the final step of its execution, the OdicLoader modifies ~/.bash_profile, so SimplexTea is launched with Bash and its output is muted (~/.config/guiconfigd >/dev/null 2>&1).
SimplexTea is a Linux backdoor written in C++. As highlighted in Desk 1, its class names are similar to perform names present in a pattern, with filename sysnetd, submitted to VirusTotal from Romania (SHA-1: F6760FB1F8B019AF2304EA6410001B63A1809F1D). Due to the similarities at school names and performance names between SimplexTea and sysnetd, we imagine SimplexTea is an up to date model, rewritten from C to C++.
Desk 1. Comparability of the unique image names from two Linux backdoors submitted to VirusTotal
guiconfigd |
sysnetd |
CMsgCmd::Begin(void) | MSG_Cmd |
CMsgSafeDel::Begin(void) | MSG_Del |
CMsgDir::Begin(void) | MSG_Dir |
CMsgDown::Begin(void) | MSG_Down |
CMsgExit::Begin(void) | MSG_Exit |
CMsgReadConfig::Begin(void) | MSG_ReadConfig |
CMsgRun::Begin(void) | MSG_Run |
CMsgSetPath::Begin(void) | MSG_SetPath |
CMsgSleep::Begin(void) | MSG_Sleep |
CMsgTest::Begin(void) | MSG_Test |
CMsgUp::Begin(void) | MSG_Up |
CMsgWriteConfig::Begin(void) | MSG_WriteConfig |
MSG_GetComInfo | |
CMsgHibernate::Begin(void) | |
CMsgKeepCon::Begin(void) | |
CMsgZipDown::Begin(void) | |
CMsgZip::StartZip(void *) | |
CMsgZip::Begin(void) | |
CHttpWrapper::RecvData(uchar *&,uint *,uint,signed char) | |
RecvMsg | |
CHttpWrapper::SendMsg(_MSG_STRUCT *) | SendMsg |
CHttpWrapper::SendData(uchar *,uint,uint) | |
CHttpWrapper::SendMsg(uint,uint,uchar *,uint,uint) | |
CHttpWrapper::SendLoginData(uchar *,uint,uchar *&,uint *) |
How is sysnetd associated to Lazarus? The next part exhibits similarities with Lazarus’s Home windows backdoor known as BADCALL.
BADCALL for Linux
We attribute sysnetd to Lazarus due to its similarities with the next two recordsdata (and we imagine that sysnetd is a Linux variant of the group’s backdoor for Home windows known as BADCALL):
- P2P_DLL.dll (SHA-1: 65122E5129FC74D6B5EBAFCC3376ABAE0145BC14), which exhibits code similarities to sysnetd within the type of domains used as a entrance for pretend TLS connection (see Determine 4). It was attributed to Lazarus by CISA in December 2017. From September 2019, CISA began to name newer variations of this malware BADCALL (SHA-1: D288766FA268BC2534F85FD06A5D52264E646C47).

Determine 4. Similarities between a Home windows and a Linux variant of BADCALL (an inventory of domains used as a entrance for a pretend TLS connection)
- prtspool (SHA-1: 58B0516D28BD7218B1908FB266B8FE7582E22A5F), which exhibits code similarities to sysnetd (see Determine 5). It was attributed to Lazarus by CISA in February 2021. Observe as effectively that SIMPLESEA, a macOS backdoor discovered in the course of the 3CX incident response, implements the A5/1 stream cipher.

Determine 5. Similarities between AppleJeus for macOS and the Linux variant of BADCALL (the important thing for the A5/1 stream cipher)
This Linux model of the BADCALL backdoor, sysnetd, hundreds its configuration from a file named /tmp/vgauthsvclog. Since Lazarus operators have beforehand disguised their payloads, the usage of this title, which is utilized by the VMware Visitor Authentication service, means that the focused system could also be a Linux VMware digital machine. Apparently, the XOR key on this case is similar as one utilized in SIMPLESEA from the 3CX investigation.
Having a look on the three 32-bit integers, 0xC2B45678, 0x90ABCDEF, and 0xFE268455 from Determine 5, which signify a key for a customized implementation of the A5/1 cipher, we realized that the identical algorithm and the an identical keys had been utilized in Home windows malware that dates again to the tip of 2014 and was concerned in some of the infamous Lazarus circumstances: the cybersabotage of Sony Photos Leisure (SHA-1: 1C66E67A8531E3FF1C64AE57E6EDFDE7BEF2352D).

Determine 7. The decryption routine shared between the BADCALL for Linux and focused harmful malware for Home windows from 2014
Further attribution information factors
To recap what we’ve coated up to now, we attribute the 3CX supply-chain assault to the Lazarus group with a excessive stage of confidence. That is primarily based on the next components:
- Malware (the intrusion set):
- The IconicLoader (samcli.dll) makes use of the identical sort of sturdy encryption – AES-GCM – as SimplexTea (whose attribution to Lazarus was established through the similarity with BALLCALL for Linux); solely the keys and initialization vectors differ.
- Based mostly on the PE Wealthy Headers, each IconicLoader (samcli.dll) and IconicStealer (sechost.dll) are tasks of an identical dimension and compiled in the identical Visible Studio surroundings because the executables iertutil.dll (SHA-1: 5B03294B72C0CAA5FB20E7817002C600645EB475) and iertutil.dll (SHA-1: 7491BD61ED15298CE5EE5FFD01C8C82A2CDB40EC) reported within the Lazarus cryptocurrency campaigns by Volexity and Microsoft. We embrace under the YARA rule RichHeaders_Lazarus_NukeSped_IconicPayloads_3CX_Q12023, which flags all these samples, and no unrelated malicious or clear recordsdata, as examined on the present ESET databases and up to date VirusTotal submissions.
- SimplexTea payload hundreds its configuration in a really comparable option to the SIMPLESEA malware from the 3CX official incident response. The XOR key differs (0x5E vs. 0x7E), however the configuration bears the identical title: apdl.cf (see Determine 8).
- Infrastructure:
- There may be shared community infrastructure with SimplexTea, because it makes use of https://journalide[.]org/djour.php because it C&C, whose area is reported within the official results of the incident response of the 3CX compromise by Mandiant.
Conclusion
The 3CX compromise has gained a variety of consideration from the safety neighborhood since its disclosure on March 29th. This compromised software program, deployed on varied IT infrastructures, which permits the obtain and execution of any form of payload, can have devastating impacts. Sadly, no software program writer is resistant to being compromised and inadvertently distributing trojanized variations of their functions.
The stealthiness of a supply-chain assault makes this technique of distributing malware very interesting from an attacker’s perspective. Lazarus has already used this technique up to now, focusing on South Korean customers of WIZVERA VeraPort software program in 2020. Similarities with present malware from the Lazarus toolset and with the group’s typical methods strongly recommend the current 3CX compromise is the work of Lazarus as effectively.
It’s also attention-grabbing to notice that Lazarus can produce and use malware for all main desktop working programs: Home windows, macOS, and Linux. Each Home windows and macOS programs had been focused in the course of the 3CX incident, with 3CX’s VoIP software program for each working programs being trojanized to incorporate malicious code to fetch arbitrary payloads. Within the case of 3CX, each Home windows and macOS second-stage malware variations exist. This text demonstrates the existence of a Linux backdoor that most likely corresponds to the SIMPLESEA macOS malware seen within the 3CX incident. We named this Linux element SimplexTea and confirmed that it’s a part of Operation DreamJob, Lazarus’s flagship marketing campaign utilizing job affords to lure and compromise unsuspecting victims.
ESET Analysis affords non-public APT intelligence studies and information feeds. For any inquiries about this service, go to the ESET Threat Intelligence web page.
IoCs
Information
SHA-1 | Filename | ESET detection title | Description |
---|---|---|---|
0CA1723AFE261CD85B05C9EF424FC50290DCE7DF | guiconfigd | Linux/NukeSped.E | |
3A63477A078CE10E53DFB5639E35D74F93CEFA81 | HSBC_job_offer․pdf | Linux/NukeSped.E | OdicLoader, a 64-bit downloader for Linux, written in Go. |
9D8BADE2030C93D0A010AA57B90915EB7D99EC82 | HSBC_job_offer.pdf.zip | Linux/NukeSped.E | A ZIP archive with a Linux payload, from VirusTotal. |
F6760FB1F8B019AF2304EA6410001B63A1809F1D | sysnetd | Linux/NukeSped.G | BADCALL for Linux. |
Community
IP handle | Area | Internet hosting supplier | First seen | Particulars |
---|---|---|---|---|
23.254.211[.]230 | N/A | Hostwinds LLC. | N/A | C&C server for BADCALL for Linux |
38.108.185[.]79 38.108.185[.]115 |
od[.]lk | Cogent Communications | Distant OpenDrive storage containing SimplexTea (/d/NTJfMzg4MDE1NzJf/vxmedia) | |
172.93.201[.]88 | journalide[.]org | Nexeon Applied sciences, Inc. | C&C server for SimplexTea (/djour.php) |
MITRE ATT&CK methods
Tactic | ID | Identify | Description |
---|---|---|---|
Reconnaissance | T1593.001 | Search Open Web sites/Domains: Social Media | Lazarus attackers most likely approached a goal with a pretend HSBC-themed job provide that might match the goal’s curiosity. This has been performed principally through LinkedIn up to now. |
Useful resource Growth | T1584.001 | Purchase Infrastructure: Domains | In contrast to many earlier circumstances of compromised C&Cs utilized in Operation DreamJob, Lazarus operators registered their very own area for the Linux goal. |
T1587.001 | Develop Capabilities: Malware | Customized instruments from the assault are very seemingly developed by the attackers. | |
T1585.003 | Set up Accounts: Cloud Accounts | The attackers hosted the ultimate stage on the cloud service OpenDrive. | |
T1608.001 | Stage Capabilities: Add Malware | The attackers hosted the ultimate stage on the cloud service OpenDrive. | |
Execution | T1204.002 | Person Execution: Malicious File | OdicLoader masquerades as a PDF file with a view to idiot the goal. |
Preliminary Entry | T1566.002 | Phishing: Spearphishing Hyperlink | The goal seemingly obtained a hyperlink to third-party distant storage with a malicious ZIP archive, which was later submitted to VirusTotal. |
Persistence | T1546.004 | Occasion Triggered Execution: Unix Shell Configuration Modification | OdicLoader modifies the sufferer’s Bash profile, so SimplexTea is launched every time Bash is stared and its output is muted. |
Protection Evasion | T1134.002 | Entry Token Manipulation: Create Course of with Token | SimplexTea can create a brand new course of, if instructed by its C&C server. |
T1140 | Deobfuscate/Decode Information or Info | SimplexTea shops its configuration in an encrypted apdl.cf. | |
T1027.009 | Obfuscated Information or Info: Embedded Payloads | The droppers of all malicious chains include an embedded information array with a further stage. | |
T1562.003 | Impair Defenses: Impair Command Historical past Logging | OdicLoader modifies the sufferer’s Bash profile, so the output and error messages from SimplexTea are muted. SimplexTea executes new processes with the identical method. | |
T1070.004 | Indicator Elimination: File Deletion | SimplexTea has the flexibility to delete recordsdata securely. | |
T1497.003 | Virtualization/Sandbox Evasion: Time Based mostly Evasion | SimplexTea implements a number of customized sleep delays in its execution. | |
Discovery | T1083 | File and Listing Discovery | SimplexTea can record the listing content material along with their names, sizes, and timestamps (mimicking the ls -la command). |
Command and Management | T1071.001 | Software Layer Protocol: Net Protocols | SimplexTea can use HTTP and HTTPS for communication with its C&C server, utilizing a statically linked Curl library. |
T1573.001 | Encrypted Channel: Symmetric Cryptography | SimplexTea encrypts C&C site visitors utilizing the AES-GCM algorithm. | |
T1132.001 | Information Encoding: Customary Encoding | SimplexTea encodes C&C site visitors utilizing base64. | |
T1090 | Proxy | SimplexTea can make the most of a proxy for communications. | |
Exfiltration | T1041 | Exfiltration Over C2 Channel | SimplexTea can exfiltrate information as ZIP archives to its C&C server. |
Appendix
This YARA rule flags the cluster containing each IconicLoader and IconicStealer, in addition to the payloads deployed within the cryptocurrency campaigns from December 2022.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 |
/* The next rule will solely work with YARA model >= 3.11.0 */ import „pe“ rule RichHeaders_Lazarus_NukeSped_IconicPayloads_3CX_Q12023 { meta: description = “ Wealthy Headers-based rule masking the IconicLoader and IconicStealer from the 3CX provide chain incident, and likewise payloads from the cryptocurrency campaigns from 2022-12″ writer = „ESET Analysis“ date = „2023-03-31“ hash = „3B88CDA62CDD918B62EF5AA8C5A73A46F176D18B“ hash = „CAD1120D91B812ACAFEF7175F949DD1B09C6C21A“ hash = „5B03294B72C0CAA5FB20E7817002C600645EB475“ hash = „7491BD61ED15298CE5EE5FFD01C8C82A2CDB40EC“ situation: pe.rich_signature.toolid(259, 30818) == 9 and pe.rich_signature.toolid(256, 31329) == 1 and pe.rich_signature.toolid(261, 30818) >= 30 and pe.rich_signature.toolid(261, 30818) <= 38 and pe.rich_signature.toolid(261, 29395) >= 134 and pe.rich_signature.toolid(261, 29395) <= 164 and pe.rich_signature.toolid(257, 29395) >= 6 and pe.rich_signature.toolid(257, 29395) <= 14 } |