1Intel(R) TXT Overview:
   4Intel's technology for safer computing, Intel(R) Trusted Execution
   5Technology (Intel(R) TXT), defines platform-level enhancements that
   6provide the building blocks for creating trusted platforms.
   8Intel TXT was formerly known by the code name LaGrande Technology (LT).
  10Intel TXT in Brief:
  11o  Provides dynamic root of trust for measurement (DRTM)
  12o  Data protection in case of improper shutdown
  13o  Measurement and verification of launched environment
  15Intel TXT is part of the vPro(TM) brand and is also available some
  16non-vPro systems.  It is currently available on desktop systems
  17based on the Q35, X38, Q45, and Q43 Express chipsets (e.g. Dell
  18Optiplex 755, HP dc7800, etc.) and mobile systems based on the GM45,
  19PM45, and GS45 Express chipsets.
  21For more information, see
  22This site also has a link to the Intel TXT MLE Developers Manual,
  23which has been updated for the new released platforms.
  25Intel TXT has been presented at various events over the past few
  26years, some of which are:
  27      LinuxTAG 2008:
  29      TRUST2008:
  31          3_David-Grawrock_The-Front-Door-of-Trusted-Computing.pdf
  32      IDF, Shanghai:
  34      IDFs 2006, 2007 (I'm not sure if/where they are online)
  36Trusted Boot Project Overview:
  39Trusted Boot (tboot) is an open source, pre-kernel/VMM module that
  40uses Intel TXT to perform a measured and verified launch of an OS
  43It is hosted on SourceForge at
  44The mercurial source repo is available at
  47Tboot currently supports launching Xen (open source VMM/hypervisor
  48w/ TXT support since v3.2), and now Linux kernels.
  51Value Proposition for Linux or "Why should you care?"
  54While there are many products and technologies that attempt to
  55measure or protect the integrity of a running kernel, they all
  56assume the kernel is "good" to begin with.  The Integrity
  57Measurement Architecture (IMA) and Linux Integrity Module interface
  58are examples of such solutions.
  60To get trust in the initial kernel without using Intel TXT, a
  61static root of trust must be used.  This bases trust in BIOS
  62starting at system reset and requires measurement of all code
  63executed between system reset through the completion of the kernel
  64boot as well as data objects used by that code.  In the case of a
  65Linux kernel, this means all of BIOS, any option ROMs, the
  66bootloader and the boot config.  In practice, this is a lot of
  67code/data, much of which is subject to change from boot to boot
  68(e.g. changing NICs may change option ROMs).  Without reference
  69hashes, these measurement changes are difficult to assess or
  70confirm as benign.  This process also does not provide DMA
  71protection, memory configuration/alias checks and locks, crash
  72protection, or policy support.
  74By using the hardware-based root of trust that Intel TXT provides,
  75many of these issues can be mitigated.  Specifically: many
  76pre-launch components can be removed from the trust chain, DMA
  77protection is provided to all launched components, a large number
  78of platform configuration checks are performed and values locked,
  79protection is provided for any data in the event of an improper
  80shutdown, and there is support for policy-based execution/verification.
  81This provides a more stable measurement and a higher assurance of
  82system configuration and initial state than would be otherwise
  83possible.  Since the tboot project is open source, source code for
  84almost all parts of the trust chain is available (excepting SMM and
  85Intel-provided firmware).
  87How Does it Work?
  90o  Tboot is an executable that is launched by the bootloader as
  91   the "kernel" (the binary the bootloader executes).
  92o  It performs all of the work necessary to determine if the
  93   platform supports Intel TXT and, if so, executes the GETSEC[SENTER]
  94   processor instruction that initiates the dynamic root of trust.
  95   -  If tboot determines that the system does not support Intel TXT
  96      or is not configured correctly (e.g. the SINIT AC Module was
  97      incorrect), it will directly launch the kernel with no changes
  98      to any state.
  99   -  Tboot will output various information about its progress to the
 100      terminal, serial port, and/or an in-memory log; the output
 101      locations can be configured with a command line switch.
 102o  The GETSEC[SENTER] instruction will return control to tboot and
 103   tboot then verifies certain aspects of the environment (e.g. TPM NV
 104   lock, e820 table does not have invalid entries, etc.).
 105o  It will wake the APs from the special sleep state the GETSEC[SENTER]
 106   instruction had put them in and place them into a wait-for-SIPI
 107   state.
 108   -  Because the processors will not respond to an INIT or SIPI when
 109      in the TXT environment, it is necessary to create a small VT-x
 110      guest for the APs.  When they run in this guest, they will
 111      simply wait for the INIT-SIPI-SIPI sequence, which will cause
 112      VMEXITs, and then disable VT and jump to the SIPI vector.  This
 113      approach seemed like a better choice than having to insert
 114      special code into the kernel's MP wakeup sequence.
 115o  Tboot then applies an (optional) user-defined launch policy to
 116   verify the kernel and initrd.
 117   -  This policy is rooted in TPM NV and is described in the tboot
 118      project.  The tboot project also contains code for tools to
 119      create and provision the policy.
 120   -  Policies are completely under user control and if not present
 121      then any kernel will be launched.
 122   -  Policy action is flexible and can include halting on failures
 123      or simply logging them and continuing.
 124o  Tboot adjusts the e820 table provided by the bootloader to reserve
 125   its own location in memory as well as to reserve certain other
 126   TXT-related regions.
 127o  As part of its launch, tboot DMA protects all of RAM (using the
 128   VT-d PMRs).  Thus, the kernel must be booted with 'intel_iommu=on'
 129   in order to remove this blanket protection and use VT-d's
 130   page-level protection.
 131o  Tboot will populate a shared page with some data about itself and
 132   pass this to the Linux kernel as it transfers control.
 133   -  The location of the shared page is passed via the boot_params
 134      struct as a physical address.
 135o  The kernel will look for the tboot shared page address and, if it
 136   exists, map it.
 137o  As one of the checks/protections provided by TXT, it makes a copy
 138   of the VT-d DMARs in a DMA-protected region of memory and verifies
 139   them for correctness.  The VT-d code will detect if the kernel was
 140   launched with tboot and use this copy instead of the one in the
 141   ACPI table.
 142o  At this point, tboot and TXT are out of the picture until a
 143   shutdown (S<n>)
 144o  In order to put a system into any of the sleep states after a TXT
 145   launch, TXT must first be exited.  This is to prevent attacks that
 146   attempt to crash the system to gain control on reboot and steal
 147   data left in memory.
 148   -  The kernel will perform all of its sleep preparation and
 149      populate the shared page with the ACPI data needed to put the
 150      platform in the desired sleep state.
 151   -  Then the kernel jumps into tboot via the vector specified in the
 152      shared page.
 153   -  Tboot will clean up the environment and disable TXT, then use the
 154      kernel-provided ACPI information to actually place the platform
 155      into the desired sleep state.
 156   -  In the case of S3, tboot will also register itself as the resume
 157      vector.  This is necessary because it must re-establish the
 158      measured environment upon resume.  Once the TXT environment
 159      has been restored, it will restore the TPM PCRs and then
 160      transfer control back to the kernel's S3 resume vector.
 161      In order to preserve system integrity across S3, the kernel
 162      provides tboot with a set of memory ranges (RAM and RESERVED_KERN
 163      in the e820 table, but not any memory that BIOS might alter over
 164      the S3 transition) that tboot will calculate a MAC (message
 165      authentication code) over and then seal with the TPM. On resume
 166      and once the measured environment has been re-established, tboot
 167      will re-calculate the MAC and verify it against the sealed value.
 168      Tboot's policy determines what happens if the verification fails.
 169      Note that the c/s 194 of tboot which has the new MAC code supports
 170      this.
 172That's pretty much it for TXT support.
 175Configuring the System:
 178This code works with 32bit, 32bit PAE, and 64bit (x86_64) kernels.
 180In BIOS, the user must enable:  TPM, TXT, VT-x, VT-d.  Not all BIOSes
 181allow these to be individually enabled/disabled and the screens in
 182which to find them are BIOS-specific.
 184grub.conf needs to be modified as follows:
 185        title Linux 2.6.29-tip w/ tboot
 186          root (hd0,0)
 187                kernel /tboot.gz logging=serial,vga,memory
 188                module /vmlinuz-2.6.29-tip intel_iommu=on ro
 189                       root=LABEL=/ rhgb console=ttyS0,115200 3
 190                module /initrd-2.6.29-tip.img
 191                module /Q35_SINIT_17.BIN
 193The kernel option for enabling Intel TXT support is found under the
 194Security top-level menu and is called "Enable Intel(R) Trusted
 195Execution Technology (TXT)".  It is marked as EXPERIMENTAL and
 196depends on the generic x86 support (to allow maximum flexibility in
 197kernel build options), since the tboot code will detect whether the
 198platform actually supports Intel TXT and thus whether any of the
 199kernel code is executed.
 201The Q35_SINIT_17.BIN file is what Intel TXT refers to as an
 202Authenticated Code Module.  It is specific to the chipset in the
 203system and can also be found on the Trusted Boot site.  It is an
 204(unencrypted) module signed by Intel that is used as part of the
 205DRTM process to verify and configure the system.  It is signed
 206because it operates at a higher privilege level in the system than
 207any other macrocode and its correct operation is critical to the
 208establishment of the DRTM.  The process for determining the correct
 209SINIT ACM for a system is documented in the SINIT-guide.txt file
 210that is on the tboot SourceForge site under the SINIT ACM downloads.