Skip to contentSkip to conversation

x86/txt for v2.6.32

GIT PULL H. Peter Anvin 8 participants 23 messages Sep 14, 2009 – Oct 17, 2009

lore
Share & tools
lore
#0 lore raw

The following changes since commit 326ba5010a5429a5a528b268b36a5900d4ab0eba:
Linus Torvalds (1):

        Linux 2.6.31-rc8

are available in the git repository at:

git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip.git x86-txt-for-linus

Arnaldo Carvalho de Melo (1):
x86, intel_txt: Fix typos in Kconfig help

H. Peter Anvin (3):
x86, intel_txt: tboot.c needs <asm/fixmap.h> x86, intel_txt: Factor out the code for S3 setup x86, intel_txt: Handle ACPI_SLEEP without X86_TRAMPOLINE

Ingo Molnar (1):

      Merge commit 'v2.6.31-rc8' into x86/txt

Joseph Cihula (4):
x86, intel_txt: Intel TXT boot support
x86, intel_txt: Intel TXT reboot/halt shutdown support x86, intel_txt: Intel TXT Sx shutdown support
intel_txt: Force IOMMU on for Intel TXT launch

Shane Wang (1):
x86, intel_txt: clean up the impact on generic code, unbreak non-x86

 Documentation/intel_txt.txt      |  210 ++++++++++++++++++
 Documentation/x86/zero-page.txt  |    1 +
 arch/x86/Kconfig                 |    4 +
 arch/x86/include/asm/bootparam.h |    3 +-
 arch/x86/include/asm/fixmap.h    |    3 +
 arch/x86/kernel/Makefile         |    1 +
 arch/x86/kernel/reboot.c         |    7 +
 arch/x86/kernel/setup.c          |    3 +
 arch/x86/kernel/smpboot.c        |    2 +
 arch/x86/kernel/tboot.c          |  447 ++++++++++++++++++++++++++++++++++++++
 drivers/acpi/acpica/hwsleep.c    |    3 +
 drivers/pci/dmar.c               |    7 +
 drivers/pci/intel-iommu.c        |   17 ++-
 include/linux/tboot.h            |  162 ++++++++++++++
 kernel/cpu.c                     |    1 +
 security/Kconfig                 |   30 +++
 16 files changed, 898 insertions(+), 3 deletions(-)
 create mode 100644 Documentation/intel_txt.txt
 create mode 100644 arch/x86/kernel/tboot.c
 create mode 100644 include/linux/tboot.h
#1 lore raw

Hi!

Quoted 8 lines

git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip.git x86-txt-for-linus

Arnaldo Carvalho de Melo (1):
x86, intel_txt: Fix typos in Kconfig help

H. Peter Anvin (3):
x86, intel_txt: Factor out the code for S3 setup x86, intel_txt: Handle ACPI_SLEEP without X86_TRAMPOLINE

This went in despite my NAK/request for more info. It seems you are simply ignoring feedback you don't like :-(

Signature · 2 lines
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
#2 lore raw
Quoted 7 lines · Pavel Machek

Hi!

Quoted 5 lines

git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip.git x86-txt-for-linus

Arnaldo Carvalho de Melo (1):
x86, intel_txt: Fix typos in Kconfig help

Quoted 6 lines
Quoted 3 lines

H. Peter Anvin (3):
x86, intel_txt: Factor out the code for S3 setup x86, intel_txt: Handle ACPI_SLEEP without X86_TRAMPOLINE

This went in despite my NAK/request for more info. It seems you are simply ignoring feedback you don't like :-(

The "x86, intel_txt: Handle ACPI_SLEEP without X86_TRAMPOLINE" commit looks entirely correct to me.

I have no comments on "x86, intel_txt: Factor out the code for S3 setup" as well, maybe except that the BUG() in tboot_setup_sleep() in the !CONFIG_ACPI_SLEEP case is a bit unfortunate.

Thanks,
Rafael

#3 lore raw
Quoted 15 lines · Rafael J. Wysocki
Quoted 7 lines · Pavel Machek

Hi!

Quoted 5 lines

git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip.git x86-txt-for-linus

Arnaldo Carvalho de Melo (1):
x86, intel_txt: Fix typos in Kconfig help

Quoted 6 lines
Quoted 3 lines

H. Peter Anvin (3):
x86, intel_txt: Factor out the code for S3 setup x86, intel_txt: Handle ACPI_SLEEP without X86_TRAMPOLINE

This went in despite my NAK/request for more info. It seems you are simply ignoring feedback you don't like :-(

Quoted 6 lines

The "x86, intel_txt: Handle ACPI_SLEEP without X86_TRAMPOLINE" commit looks entirely correct to me.

I have no comments on "x86, intel_txt: Factor out the code for S3 setup" as well, maybe except that the BUG() in tboot_setup_sleep() in the !CONFIG_ACPI_SLEEP case is a bit unfortunate.

Well, I worry that S3 support for TXT makes TXT completely useless. A little liquid nitrogen, remove RAM, place it in another machine, modify it in any way you want, more liquid nitrogen, place it back.

Oops, protection provided by TXT is lost.

Pavel

Signature · 2 lines
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
#4 lore raw
Quoted 23 lines · Pavel Machek
Quoted 15 lines · Rafael J. Wysocki
Quoted 7 lines · Pavel Machek

Hi!

Quoted 5 lines

git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip.git x86-txt-for-linus

Arnaldo Carvalho de Melo (1):
x86, intel_txt: Fix typos in Kconfig help

Quoted 6 lines
Quoted 3 lines

H. Peter Anvin (3):
x86, intel_txt: Factor out the code for S3 setup x86, intel_txt: Handle ACPI_SLEEP without X86_TRAMPOLINE

This went in despite my NAK/request for more info. It seems you are simply ignoring feedback you don't like :-(

Quoted 6 lines

The "x86, intel_txt: Handle ACPI_SLEEP without X86_TRAMPOLINE" commit looks entirely correct to me.

I have no comments on "x86, intel_txt: Factor out the code for S3 setup" as well, maybe except that the BUG() in tboot_setup_sleep() in the !CONFIG_ACPI_SLEEP case is a bit unfortunate.

Quoted 5 lines

Well, I worry that S3 support for TXT makes TXT completely useless. A little liquid nitrogen, remove RAM, place it in another machine, modify it in any way you want, more liquid nitrogen, place it back.

Oops, protection provided by TXT is lost.

Ah, I see your point now.

Thanks,
Rafael

#5 lore raw
Quoted 9 lines · Rafael J. Wysocki
Quoted 6 lines

Well, I worry that S3 support for TXT makes TXT completely useless. A little liquid nitrogen, remove RAM, place it in another machine, modify it in any way you want, more liquid nitrogen, place it back.

Oops, protection provided by TXT is lost.

Ah, I see your point now.

Shane Wang sent me a patch for S3 support, but it missed the merge window:

	http://marc.info/[email protected]

*As far as I understand* -- and I haven't looked into it in detail yet, having just come back from Plumber's -- this provides integrity protection, not content extraction protection.

Shane, could you please comment?

Thanks,

-hpa

#6 lore raw
Quoted 11 lines · H. Peter Anvin
Quoted 9 lines · Rafael J. Wysocki
Quoted 6 lines

Well, I worry that S3 support for TXT makes TXT completely useless. A little liquid nitrogen, remove RAM, place it in another machine, modify it in any way you want, more liquid nitrogen, place it back.

Oops, protection provided by TXT is lost.

Ah, I see your point now.

Quoted 7 lines

Shane Wang sent me a patch for S3 support, but it missed the merge window:

	http://marc.info/[email protected]

*As far as I understand* -- and I haven't looked into it in detail yet, having just come back from Plumber's -- this provides integrity protection, not content extraction protection.

How does it provide integrity protection? I'm free to modify RAM content in the other machine....
Pavel

Signature · 2 lines
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
#7 lore raw
Quoted 13 lines · Pavel Machek
Quoted 12 lines · H. Peter Anvin
Quoted 7 lines · Rafael J. Wysocki
Quoted 5 lines

Well, I worry that S3 support for TXT makes TXT completely useless. A little liquid nitrogen, remove RAM, place it in another machine, modify it in any way you want, more liquid nitrogen, place it back.

Oops, protection provided by TXT is lost.

Ah, I see your point now.

Shane Wang sent me a patch for S3 support, but it missed the merge window:

	http://marc.info/[email protected]
Quoted 7 lines
Quoted 3 lines

*As far as I understand* -- and I haven't looked into it in detail yet, having just come back from Plumber's -- this provides integrity protection, not content extraction protection.

How does it provide integrity protection? I'm free to modify RAM content in the other machine....
Pavel

Hi Pavel,

Before S3 sleep, tboot patch will MAC the memory, and after S3 resume, the memory integrity will be verified according to the MAC value. So, you can't modify RAM, or else you will fail on S3 resume.

The current patch hpa mentioned is for userspace memory integrity. For kernel memory integrity, the code is already in with the previous txt patch.

Thanks.
Shane

#8 lore raw
Quoted 10 lines · Shane Wang
Quoted 9 lines · Pavel Machek
Quoted 8 lines · H. Peter Anvin
Quoted 7 lines · Rafael J. Wysocki
Quoted 5 lines

Well, I worry that S3 support for TXT makes TXT completely useless. A little liquid nitrogen, remove RAM, place it in another machine, modify it in any way you want, more liquid nitrogen, place it back.

Oops, protection provided by TXT is lost.

Ah, I see your point now.

Quoted 7 lines
Quoted 7 lines
Quoted 7 lines

Shane Wang sent me a patch for S3 support, but it missed the merge window:

	http://marc.info/[email protected]

*As far as I understand* -- and I haven't looked into it in detail yet, having just come back from Plumber's -- this provides integrity protection, not content extraction protection.

Well, documentation seems to suggest it provides content protection, too. If not, should that be clearly documented in Doc*/intel_txt? [Also, I'd expect threat model aka "what does it protect against there"].

Quoted 11 lines
Quoted 2 lines

How does it provide integrity protection? I'm free to modify RAM content in the other machine....

Before S3 sleep, tboot patch will MAC the memory, and after S3 resume, the memory integrity will be verified according to the MAC value. So, you can't modify RAM, or else you will fail on S3 resume.

The current patch hpa mentioned is for userspace memory integrity. For kernel memory integrity, the code is already in with the previous txt patch.

Ok, and what prevents me from commenting out the MAC checking code?

Pavel

Signature · 2 lines
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
#9 lore raw
Quoted 2 lines · On Tue, 29 Sep 2009 19:13:18 +0200 Pavel Machek <[email protected]> wrote:

Ok, and what prevents me from commenting out the MAC checking code?

because the bios verified some code that verified the kernel which includes the MAC checking code .. as part of returning from S3 ?

Quoted 1 line
Signature · 3 lines
Arjan van de Ven 	Intel Open Source Technology Centre
For development, discussion and tips for power savings, 
visit http://www.lesswatts.org
#10 lore raw
Quoted 8 lines · Arjan van de Ven
Quoted 2 lines · On Tue, 29 Sep 2009 19:13:18 +0200 Pavel Machek <[email protected]> wrote:

Ok, and what prevents me from commenting out the MAC checking code?

because the bios verified some code that verified the kernel which includes the MAC checking code .. as part of returning from S3 ?

Yes, S3 sleep/resume cause another cycle to build the measured environment.
i.e. SINIT will verify tboot, tboot will verify kernel mem, kernel will verify userspace mem. If you comment out the MAC checking code in any party, the chain will lost and S3 resume will fail.

Thanks.
Shane

#11 lore raw
Quoted 7 lines · Shane
Quoted 6 lines · Arjan van de Ven
Quoted 2 lines · On Tue, 29 Sep 2009 19:13:18 +0200 Pavel Machek <[email protected]> wrote:

Ok, and what prevents me from commenting out the MAC checking code?

Quoted 7 lines
Quoted 2 lines

because the bios verified some code that verified the kernel which includes the MAC checking code .. as part of returning from S3 ?

Yes, S3 sleep/resume cause another cycle to build the measured environment.
i.e. SINIT will verify tboot, tboot will verify kernel mem, kernel will verify userspace mem. If you comment out the MAC checking code in any party, the chain will lost and S3 resume will fail.

Ok, that means that you are protecting integrity but not secrecy? Should that be written down in documentation, along with threat model?

So I modify the RAM content so that BIOS does not think measured environment existed before suspend?
Pavel

Signature · 2 lines
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
#12 lore raw
Quoted 3 lines

So I modify the RAM content so that BIOS does not think measured environment existed before suspend?
Pavel

Hi Pavel, what do you mean on this question?
When do you modify the RAM? before S3 sleep?

Thanks.
Shane

#13 lore raw
Quoted 7 lines · Shane
Quoted 3 lines

So I modify the RAM content so that BIOS does not think measured environment existed before suspend?
Pavel

Hi Pavel, what do you mean on this question?
When do you modify the RAM? before S3 sleep?

He was talking earlier about a liquid nitrogen freeze attack.

-hpa

Signature · 2 lines
H. Peter Anvin, Intel Open Source Technology Center
I work for Intel.  I don't speak on their behalf.
#14 lore raw
Quoted 9 lines · H. Peter Anvin
Quoted 6 lines · Shane
Quoted 2 lines

So I modify the RAM content so that BIOS does not think measured environment existed before suspend?

Hi Pavel, what do you mean on this question?
When do you modify the RAM? before S3 sleep?

He was talking earlier about a liquid nitrogen freeze attack.

You wish one would need liquid nitrogen for it... frozen CO2 (dry ice) is more than cold enough for such an attack, and _MUCH_ easier to both acquire and handle than liquid nitrogen.

Signature · 4 lines
  "One disk to rule them all, One disk to find them. One disk to bring
  them all and in the darkness grind them. In the Land of Redmond
  where the shadows lie." -- The Silicon Valley Tarot
  Henrique Holschuh
#15 lore raw
In reply to Wang, Shane
Quoted 6 lines · Shane
Quoted 3 lines

So I modify the RAM content so that BIOS does not think measured environment existed before suspend?
Pavel

Hi Pavel, what do you mean on this question?
When do you modify the RAM? before S3 sleep?

During the sleep, using something cold and second machine.

Signature · 2 lines
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
#16 lore raw
Quoted 9 lines · Pavel Machek
Quoted 6 lines · Shane
Quoted 3 lines

So I modify the RAM content so that BIOS does not think measured environment existed before suspend?
Pavel

Hi Pavel, what do you mean on this question?
When do you modify the RAM? before S3 sleep?

During the sleep, using something cold and second machine.

And it is ridiculously easy to pull off, too:
http://www.engadget.com/2008/02/21/cold-boot-disk-encryption-attack-is-shockingly-effective/

Shows the attack being used to read sensitive keys, but you can use it also to *modify* system running state (it will be more difficult, as you need to remove and replace the RAM while on S3 instead of S5, but it should be doable by someone who knows what he is doing).

Signature · 4 lines
  "One disk to rule them all, One disk to find them. One disk to bring
  them all and in the darkness grind them. In the Land of Redmond
  where the shadows lie." -- The Silicon Valley Tarot
  Henrique Holschuh
#17 lore raw
Quoted 11 lines
Quoted 3 lines
Quoted 3 lines

So I modify the RAM content so that BIOS does not think measured environment existed before suspend?

And it is ridiculously easy to pull off, too:
http://www.engadget.com/2008/02/21/cold-boot-disk-encryption-attack-is-shockingly-effective/

Shows the attack being used to read sensitive keys, but you can use it also to *modify* system running state (it will be more difficult, as you need to remove and replace the RAM while on S3 instead of S5, but it should be doable by someone who knows what he is doing).

I believe the whole point of this TXT / S3 handling is that the resume from S3 will then be able to detect that the contents of RAM have been modified while the system was asleep.

TXT simply produces a reasonably trustworthy measurement of system state. If you modify RAM while the system is asleep, then you will not be able to produce a measurement showing an unmodified system state.

- R.

#18 lore raw
Quoted 7 lines · Roland Dreier
Quoted 7 lines
Quoted 3 lines
Quoted 3 lines

So I modify the RAM content so that BIOS does not think measured environment existed before suspend?

And it is ridiculously easy to pull off, too:
http://www.engadget.com/2008/02/21/cold-boot-disk-encryption-attack-is-shockingly-effective/

Quoted 8 lines
Quoted 4 lines

Shows the attack being used to read sensitive keys, but you can use it also to *modify* system running state (it will be more difficult, as you need to remove and replace the RAM while on S3 instead of S5, but it should be doable by someone who knows what he is doing).

I believe the whole point of this TXT / S3 handling is that the resume from S3 will then be able to detect that the contents of RAM have been modified while the system was asleep.

...and you are able to read out any keys, etc. Maybe that's expected & ok, but Doc*/intel_txt.txt does not actually tell me what it protects against and is pretty much useless... making patches impossible to review.

So... what does txt protect?

Data integrity only?

Data privacy, too?

Who is it designed to protect against?

Remote attacker?

Local user trying to subvert it?

...and has soldering iron he's willing to use?

...and has soldering iron, osciloscope and PCI analyzer he's willing to use?

Quoted 3 lines

TXT simply produces a reasonably trustworthy measurement of system state. If you modify RAM while the system is asleep, then you will not be able to produce a measurement showing an unmodified system state.

Well, actually I see some auditing to be done in proposed patches.

Pavel

Signature · 2 lines
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
#19 lore raw
Quoted 21 lines

From: Pavel Machek [mailto:[email protected]]
Sent: Tuesday, October 06, 2009 1:13 AM

Quoted 6 lines · Roland Dreier
Quoted 6 lines
Quoted 3 lines
Quoted 3 lines

So I modify the RAM content so that BIOS does not think measured environment existed before suspend?

And it is ridiculously easy to pull off, too:
http://www.engadget.com/2008/02/21/cold-boot-disk-encryption-attack-is-shockingly-

effective/

Quoted 9 lines
Quoted 5 lines

Shows the attack being used to read sensitive keys, but you can use it also to *modify* system running state (it will be more difficult, as you need to remove and replace the RAM while on S3 instead of S5, but it should be doable by someone who knows what he is doing).

I believe the whole point of this TXT / S3 handling is that the resume from S3 will then be able to detect that the contents of RAM have been modified while the system was asleep.

Quoted 8 lines

...and you are able to read out any keys, etc. Maybe that's expected & ok, but Doc*/intel_txt.txt does not actually tell me what it protects against and is pretty much useless... making patches impossible to review.

So... what does txt protect?

From Documentation/intel_txt.txt:

Intel TXT in Brief:
o Provides dynamic root of trust for measurement (DRTM)
o Data protection in case of improper shutdown
o Measurement and verification of launched environment

Intel TXT doesn't protect anything itself--it provides a foundation for software to provide protections and security. tboot and the associated Linux patches do this. The section of intel_txt.txt titled "Value Proposition for Linux or "Why should you care?"" tries to describe what is provided.

Quoted 1 line

Data integrity only?

Data integrity, yes, but not only. The code also provides for DRTM-based measurements, data protection in case of improper shutdown, etc.

Quoted 1 line

Data privacy, too?

No.

Quoted 3 lines

Who is it designed to protect against?

Remote attacker?

Yes.

Quoted 1 line

Local user trying to subvert it?

No.

Quoted 3 lines

...and has soldering iron he's willing to use?

...and has soldering iron, osciloscope and PCI analyzer he's willing to use?

No and no.

Quoted 5 lines
Quoted 3 lines

TXT simply produces a reasonably trustworthy measurement of system state. If you modify RAM while the system is asleep, then you will not be able to produce a measurement showing an unmodified system state.

Well, actually I see some auditing to be done in proposed patches.

All comments are welcome.

Joe

#20 lore raw

Hi!

Quoted 24 lines
Quoted 9 lines
Quoted 8 lines
Quoted 4 lines

Shows the attack being used to read sensitive keys, but you can use it also to *modify* system running state (it will be more difficult, as you need to remove and replace the RAM while on S3 instead of S5, but it should be doable by someone who knows what he is doing).

I believe the whole point of this TXT / S3 handling is that the resume from S3 will then be able to detect that the contents of RAM have been modified while the system was asleep.

Quoted 8 lines

...and you are able to read out any keys, etc. Maybe that's expected & ok, but Doc*/intel_txt.txt does not actually tell me what it protects against and is pretty much useless... making patches impossible to review.

So... what does txt protect?

From Documentation/intel_txt.txt:

Intel TXT in Brief:
o Provides dynamic root of trust for measurement (DRTM)
o Data protection in case of improper shutdown
o Measurement and verification of launched environment

Intel TXT doesn't protect anything itself--it provides a foundation for software to provide protections and security. tboot and the associated Linux patches do this. The section of intel_txt.txt titled "Value Proposition for Linux or "Why should you care?"" tries to describe what is provided.

Quoted 7 lines
Quoted 1 line

Data integrity only?

Data integrity, yes, but not only. The code also provides for DRTM-based measurements, data protection in case of improper shutdown, etc.

Quoted 1 line

Data privacy, too?

No.

So why does it protect data "in case of improper shutdown"?

Quoted 5 lines
Quoted 3 lines

Who is it designed to protect against?

Remote attacker?

Yes.

Existing mechanisms should be adequate to protect against then.

Quoted 3 lines
Quoted 1 line

Local user trying to subvert it?

No.

Then again, why does it protect data "in case of improper shutdown"?

Quoted 7 lines
Quoted 5 lines
Quoted 3 lines

TXT simply produces a reasonably trustworthy measurement of system state. If you modify RAM while the system is asleep, then you will not be able to produce a measurement showing an unmodified system state.

Well, actually I see some auditing to be done in proposed patches.

All comments are welcome.

Well, without detailed design goals, comments are pretty much impossible. Please improve Documentation/intel_txt.txt to explain what it protects, and against who.
Pavel

Signature · 2 lines
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
#21 lore raw
In reply to Henrique de Moraes Holschuh
Quoted 11 lines · On Sat, 3 Oct 2009 17:36:20 -0300 Henrique de Moraes Holschuh <[email protected]>…
Quoted 9 lines · Pavel Machek
Quoted 6 lines · Shane
Quoted 3 lines

So I modify the RAM content so that BIOS does not think measured environment existed before suspend?
Pavel

Hi Pavel, what do you mean on this question?
When do you modify the RAM? before S3 sleep?

During the sleep, using something cold and second machine.

Quoted 8 lines

And it is ridiculously easy to pull off, too:
http://www.engadget.com/2008/02/21/cold-boot-disk-encryption-attack-is-shockingly-effective/

Shows the attack being used to read sensitive keys, but you can use it also to *modify* system running state (it will be more difficult, as you need to remove and replace the RAM while on S3 instead of S5, but it should be doable by someone who knows what he is doing).

that assumes all state is in ram, and not some bit in the TPM..

(and as for that "we can steal the content"... that's confusing DRM with TXT. TXT is integrity, DRM is. well evil ;-)

Signature · 3 lines
Arjan van de Ven 	Intel Open Source Technology Centre
For development, discussion and tips for power savings, 
visit http://www.lesswatts.org
#22 lore raw
Quoted 16 lines · Arjan van de Ven
Quoted 14 lines · On Sat, 3 Oct 2009 17:36:20 -0300 Henrique de Moraes Holschuh <[email protected]>…
Quoted 9 lines · Pavel Machek
Quoted 6 lines · Shane
Quoted 3 lines

So I modify the RAM content so that BIOS does not think measured environment existed before suspend?
Pavel

Hi Pavel, what do you mean on this question?
When do you modify the RAM? before S3 sleep?

During the sleep, using something cold and second machine.

And it is ridiculously easy to pull off, too:
http://www.engadget.com/2008/02/21/cold-boot-disk-encryption-attack-is-shockingly-effective/

Quoted 6 lines
Quoted 4 lines

Shows the attack being used to read sensitive keys, but you can use it also to *modify* system running state (it will be more difficult, as you need to remove and replace the RAM while on S3 instead of S5, but it should be doable by someone who knows what he is doing).

that assumes all state is in ram, and not some bit in the TPM..

Indeed, it does.

Quoted 2 lines

(and as for that "we can steal the content"... that's confusing DRM with TXT. TXT is integrity, DRM is. well evil ;-)

Agreed :) But TXT itself (or the BIOS, but not the kernel) needs to recheck RAM content signatures on resume. Does it? If it doesn't, it is easy to "fix" too, just drop all previous atestation before suspend/restart/shutdown :P

Signature · 4 lines
  "One disk to rule them all, One disk to find them. One disk to bring
  them all and in the darkness grind them. In the Land of Redmond
  where the shadows lie." -- The Silicon Valley Tarot
  Henrique Holschuh