Eran Tromer Slides credit: Dan Boneh, Stanford Introduction to Information Security 0368-3065, Spring 2014 Lecture 9: Virtual machine confinement, trusted computing architecture Eran Tromer Slides credit: Dan Boneh, Stanford

Confinement using Virtual Machines

Virtual machines Security benefits: Key Untrusted Code US patent 6,922,774 (NSA NetTop) Security benefits: Confinement (isolation, sandboxing) Management Monitoring Recovery Forensics (replay) Process Process Process Process Process Process OS services OS services OS kernel OS kernel Hypervisor Host OS (if Type 2) CPU Memory Devices

VMM security assumption Malware may infect guest OS and guest apps But malware cannot escape from the infected VM Cannot infect host OS Cannot infect other VMs on the same hardware Requires that VMM protect itself and is not buggy VMM is much simpler than full OS VMM API is much simpler than OS API (but host OS still has device drivers)

Example of VM security application: VMM Introspection Example of VM security application: VMM Introspection protecting the anti-virus system

Example: intrusion Detection / anti-virus Runs as part of OS kernel and user space process Kernel root kit can shutdown protection system Common practice for modern malware Standard solution: run IDS system in the network Problem: insufficient visibility into user’s machine Better: run IDS as part of VMM (protected from malware) VMM can monitor virtual hardware for anomalies VMI: Virtual Machine Introspection Allows VMM to check Guest OS internals IDS = Intrusion Detection System

Sample checks Stealth malware: Creates processes that are invisible to “ps” Opens sockets that are invisible to “netstat” 1. Lie detector check Goal: detect stealth malware that hides processes and network activity Method: VMM lists processes running in GuestOS VMM requests GuestOS to list processes (e.g. ps) If mismatch, kill VM

Sample checks (cont.) 2. Application code integrity detector VMM computes hash of user app-code running in VM Compare to whitelist of hashes Kills VM if unknown program appears 3. Ensure GuestOS kernel integrity example: detect changes to sys_call_table 4. Virus signature detector Run virus signature detector on GuestOS memory 5. Detect if GuestOS puts NIC in promiscuous mode

Virtualization – covert channels and side channels Covert channel: unintended communication channel between isolated but cooperating components (sender and receiver) Can be used to leak classified data from secure component to public component Side channel: unintended channel that lets an attacker component retrieve information from an victim component without the latter’s cooperation Often induced by low-level resource contention Process Process Process Process Process Process Key Untrusted Code OS services OS services OS kernel OS kernel Hypervisor CPU Memory Devices

An example covert channel Both VMs use the same underlying hardware To send a bit b  {0,1} malware does: b= 1: at midnight do CPU intensive calculation b= 0: at midnight do nothing At midnight, listener does a CPU intensive calculation and measures completion time Now b = 1  completion-time > threshold Many covert channel exist in running system: File lock status, cache contents, interrupts, … Very difficult to eliminate

Cache-based side-channels in cloud computing Demonstrated, using Amazon EC2 as a study case: Cloud cartography Mapping the structure of the “cloud” and locating a target on the map. Placement vulnerabilities An attacker can place his VM on the same physical machine as a target VM (40% success for a few dollars) Cross-VM exfiltration Once VMs are co-resident, secret information can be exfiltrated across VM boundary. (Simulated: theft of decryption keys!)

Virtual machine confinement: a blessing or a curse? Motivation Virtual machine confinement: a blessing or a curse?

Subvirt [King Chen Wang Verbowski Wang Lortch 2006] Virus idea: Once on the victim machine, install a malicious VMM Virus hides in VMM Invisible to virus detector running inside VM Anti-virus  Anti-virus OS VMM and virus OS HW HW

The MATRIX

Blue=virtual, red=physical

VM Based Malware (blue pill virus) [Rutkowska 2006] A virus that installs a malicious VMM (hypervisor) on-the- fly under running OS Use SVM/VT-x to create VM Microsoft Security Bulletin: (Oct, 2006) : Suggests disabling hardware virtualization features by default for client-side systems http://www.microsoft.com/whdc/system/platform/virtual/CPUVirtExt.mspx VMBRs are easy to defeat A guest OS can detect that it is running on top of VMM

VMM Detection Can an OS detect it is running on top of a VMM? Applications: Virus detector can detect VMBR Normal virus (non-VMBR) can detect VMM refuse to run to avoid reverse engineering Software that binds to hardware (e.g. MS Windows) can refuse to run on top of VMM DRM systems may refuse to run on top of VMM

VMM detection (red pill techniques) VM platforms often emulate simple hardware VMWare emulates an ancient i440bx chipset … but report 8GB RAM, dual Opteron CPUs, etc. 2. VMM introduces time latency variances Memory cache behavior differs in presence of VMM Results in relative latency in time variations for any two operations 3. VMM shares the TLB with GuestOS GuestOS can detect reduced TLB size 4. Deduplication (VMM saves single copies of identical pages) … and many more methods [GAWF’07]

VMM Detection Compatibility: ensure off the shelf software works Bottom line: The perfect VMM does not exist VMMs today (e.g. VMWare) focus on: Compatibility: ensure off the shelf software works Performance: minimize virtualization overhead VMMs do not provide transparency Anomalies reveal existence of VMM

Trusted Computing Architecture

Background TCG consortium. Founded in 1999 as TCPA. Main players (promoters): (>200 members) AMD, HP, IBM, Infineon, Intel, Lenovo, Microsoft, Sun Goals: Hardware protected (encrypted) storage: Only “authorized” software can decrypt data e.g.: protecting key for decrypting file system Secure boot: method to “authorize” software Attestation: Prove to remote server what software is running on my machine.

Secure boot History of BIOS/EFI malware: CIH (1998): CIH virus corrupts system BIOS Heasman (2007): System Management Mode (SMM) “rootkit” via EFI Sacco, Ortega (2009): infect BIOS LZH decompressor CoreBOOT: generic BIOS flashing tool Main point: BIOS runs before any defenses (e.g. antivirus) Proposed defense: lock system configuration (BIOS + OS) Today: TCG approach SMM: runs privileged code with no interrupts, not even from OS

TCG: changes to PC Extra hardware: TPM Trusted Platform Module (TPM) chip Single 33MhZ clock. TPM Chip vendors: (~.3$) Atmel, Infineon, National, STMicro Intel D875GRH motherboard Software changes: BIOS, EFI (UEFI) OS and Apps 7$ for TPM from National Semiconductors (per 1000 units)

TPMs in the real world TPMs widely available on laptops, desktops and some servers Software using TPMs: File/disk encryption: BitLocker, IBM, HP, Softex Attestation for enterprise login: Cognizance, Wave Client-side single sign on: IBM, Utimaco, Wave

What the TPM does How to use it TPM Basics What the TPM does How to use it

Non Volatile Storage (> 1280 bytes) Components on TPM chip Non Volatile Storage (> 1280 bytes) Other Junk PCR Registers (16 registers) LPC bus I/O API calls Crypto Engine: RSA, SHA-1, HMAC, RNG RSA: 1024, 2048 bit modulus SHA-1: Outputs 20 byte digest

Non-volatile storage 1. Endorsement Key (EK) (2048-bit RSA) Created at manufacturing time. Cannot be changed. Used for “attestation” (described later) 2. Storage Root Key (SRK) (2048-bit RSA) Used for implementing encrypted storage Created after running TPM_TakeOwnership( OwnerPassword, … ) Can be cleared later with TPM_ForceClear from BIOS 3. OwnerPassword (160 bits) and persistent flags Private EK, SRK, and OwnerPwd never leave the TPM TPM_ForceClear requires “proof of presence,” namely holding down the Fn key

PCR: the heart of the matter PCR: Platform Configuration Registers Lots of PCR registers on chip (at least 16) Register contents: 20-byte SHA-1 digest (+junk) Updating PCR #n : TPM_Extend(n,D): PCR[n]  SHA-1 ( PCR[n] || D ) TPM_PcrRead(n): returns value(PCR(n)) PCRs initialized to default value (e.g. 0) at boot time TPM can be told to restore PCR values in NVRAM via TPM_SaveState and TPM_Startup(ST_STATE) for system suspend/resume TPM_SaveState stores PCR values in NV-RAM

Using PCRs: the TCG boot process BIOS boot block executes Calls TPM_Startup (ST_CLEAR) to initialize PCRs to 0 Calls PCR_Extend( n, <BIOS code> ) Then loads and runs BIOS post boot code BIOS executes: Calls PCR_Extend( n, <MBR code> ) Then runs MBR (master boot record), e.g. GRUB. MBR executes: Calls PCR_Extend( n, <OS loader code, config> ) Then runs OS loader … and so on 0. During boot TPM receives a TPM_Init signal from LPC bus 1. TMP_Startup: does one of three things: (1) deactivates TPM, (2) resets PCRs, or (3) restores PCR values from data created with TPM_SaveState --- used for resume 2. TPM_Startup can only be called once after power-up (sets postInitialise flag to True) 3. Note: MBR is GRUB Stage 1

In a diagram After boot, PCRs contain hash chain of booted software Hardware BIOS boot block OS loader BIOS MBR Application OS Root of trust in integrity measurement measuring Single PCR can identify entire platform TPM Extend PCR Root of trust in integrity reporting After boot, PCRs contain hash chain of booted software Collision resistance of SHA-1 ensures commitment

Example: Trusted GRUB (IBM’05) What PCR # to use and what to measure specified in GRUB config file

Using PCR values after boot Application 1: encrypted (a.k.a sealed) storage. Step 1: TPM_TakeOwnership( OwnerPassword, … ) Creates 2048-bit RSA Storage Root Key (SRK) on TPM Cannot run TPM_TakeOwnership again without OwnerPwd: Ownership Enabled Flag  False Done once by IT department or laptop owner. (optional) Step 2: TPM_CreateWrapKey / TPM_LoadKey Create more RSA keys on TPM protected by SRK Each key identified by 32-bit keyhandle OwnerPassword can later be used to change owner and remove SRK SRK key handle ID is 0x40000000

Protected Storage Main Step: Encrypt data using RSA key on TPM TPM_Seal (some) Arguments: keyhandle: which TPM key to encrypt with KeyAuth: Password for using key `keyhandle’ PcrValues: PCRs to embed in encrypted blob data block: at most 256 bytes (2048 bits) Used to encrypt symmetric key (e.g. AES) Returns encrypted blob. Main point: blob can only be decrypted with TPM_Unseal when PCR-reg-vals = PCR-vals in blob. TPM_Unseal will fail othrewise TPM_Seal: allows to specify arbitrary PCR values for unseal.

Protected Storage Embedding PCR values in blob ensures that only certain apps can decrypt data. e.g.: Messing with MBR or OS kernel will change PCR values.

Sealed storage: applications Lock software on machine: OS and apps sealed with MBR’s PCR. Any changes to MBR (to load other OS) will prevent locked software from loading. Prevents tampering and reverse engineering Web server: seal server’s SSL private key Goal: only unmodified Apache can access SSL key Problem: updates to Apache or Apache config General problem with software upgrades/patches: Upgrade process must re-seal all blobs with new PCRs

Security? Resetting TPM after boot Attacker can disable TPM until after boot, then extend PCRs arbitrarily (one-byte change to boot block) [Kauer 07] Software attack: send TPM_Init on LPC bus allows calling TPM_Startup again (to reset PCRs) Simple hardware attack: use a wire to connect TPM reset pin to ground Once PCRs are reset, they can be extended to reflect a fake configuration. Rollback attack on encrypted blobs e.g. undo security patches without being noticed. Can be mitigated using Data Integrity Regs (DIR) Need OwnerPassword to write DIR K’07: Kauer, Usenix Security 2007. Need owner password to write to DIR. Anyone can read DIR. Stored in NV RAM.