Just a user interface issue?
Just a user interface issue?
Posted Feb 28, 2013 6:20 UTC (Thu) by jreiser (subscriber, #11027)Parent article: Loading keys from Microsoft PE binaries
According to UEFI Secure Boot, Microsoft says "systems certified for Windows 8 [on non-ARM machines] must allow secure boot to enter custom mode or be disabled." Then in theory the user can enter any key in custom mode, and secure boot will honor that key. Apparently current UEFI instances require the user to type the long custom key, but instead UEFI could read the key from a USB device or filesystem and display it to the user for approval. Is there more to the brouhaha than just a user interface issue?
Posted Feb 28, 2013 12:22 UTC (Thu)
by k3ninho (subscriber, #50375)
[Link] (18 responses)
In the past, there's been a trickle-down solution where the more-technical people build their own platform and community and it slowly becomes something that less-technical people can do. In this case, that doesn't apply: security is intended to be a blocker to access, and being on the wrong side of that divide must be enforced or it's not security.
For this problem, the Linux Foundation should be paying for a technical advisor to go round the mainboard manufacturers enrolling them in the Linux UEFI Certification Program and supplying them with a signing key which is used in addition to the MS one. The program is created to allow for corporations to build system images with their own platform keys under which they control the entire secure environment. That should be the real motivation for this, with desktop users being secondary beneficiaries. Thus is no extra burden on the non-technical user and such a setup allows UEFI Secure Boot to work as intended.
K3n.
Posted Feb 28, 2013 17:26 UTC (Thu)
by raven667 (subscriber, #5198)
[Link] (17 responses)
It's a kind of logical insanity, since logic seems to get one to a place where MS/Verisign is the only signing authority.
Posted Feb 28, 2013 21:39 UTC (Thu)
by Lennie (subscriber, #49641)
[Link] (6 responses)
So how much money does it take to maintain a selfsigned CA ?
You don't need an organisation like Verisign to sign the CA.
Posted Feb 28, 2013 21:45 UTC (Thu)
by dlang (guest, #313)
[Link] (1 responses)
you need to have processes in place to keep the bad guys out, this probably means that it takes more work to do the signing
you need redundancy
you need to spend time figuring out if you should sign things (unless you are a commercial CA, in which case you just need to see if the credit card accepts the charge)
That being said, the cost of running the CA itself is trivial compared to the cost of getting your cert accepted and in the various places it needs to be to do any good.
Posted Feb 28, 2013 22:33 UTC (Thu)
by Lennie (subscriber, #49641)
[Link]
Posted Feb 28, 2013 22:14 UTC (Thu)
by mjg59 (subscriber, #23239)
[Link] (3 responses)
Posted Feb 28, 2013 22:59 UTC (Thu)
by Lennie (subscriber, #49641)
[Link] (2 responses)
Doing a secure custom CA needs these things, I guess ?
If you get yourself some cheap netbooks with a builtin TPM and install Linux on it you can then store two copies of your keys in two seperate safes possibly in different buildings. Then you have 3 things solved.
An east european TLD does this for their DNSSEC keysigning keys if I remember correctly. The zone singing keys are on a machine behind a firewall which is used to push updates to the publicly visible servers.
In DNSSEC the zone signing keys are used to sign the DNS data and key signing keys are used to sign the zone signing keys every couple of months.
Posted Feb 28, 2013 23:36 UTC (Thu)
by raven667 (subscriber, #5198)
[Link] (1 responses)
http://www.webtrust.org/homepage-documents/item27839.aspx
Posted Feb 28, 2013 23:45 UTC (Thu)
by Lennie (subscriber, #49641)
[Link]
I know if that is what you want, you need a lot of stuff done because I've been following what CAcert is doing.
Posted Mar 4, 2013 11:51 UTC (Mon)
by Max.Hyre (subscriber, #1054)
[Link] (9 responses)
Posted Mar 4, 2013 15:16 UTC (Mon)
by raven667 (subscriber, #5198)
[Link] (8 responses)
Posted Mar 5, 2013 17:07 UTC (Tue)
by ortalo (guest, #4654)
[Link] (7 responses)
The only thing that made me doubt of the inutiliy of UEFI "security" up to now and eventually consider *not* deactivating secure boot is the involvement of people like Matthew Garret (and co.)...
Maybe I am concluding to fast and getting old. Or maybe not...
Posted Mar 5, 2013 17:20 UTC (Tue)
by raven667 (subscriber, #5198)
[Link] (4 responses)
I'm not sure how that's supposed to work, uEFI isn't TPM and provides no way to validate a license key or enforce DRM, it's really only useful for preventing malware from modifying the boot process.
It would have been nice if mjg59's proposal for automatic key enrollment would have gotten some traction because that could have been the mechanism to become independent of the MS signing infrastructure. IIUC the problem here is that the key signing through MS is made to fit MS existing tooling for signing PE binaries, which is understandable, but it seems to me the solution is to build an alternate signing infrastructure that works the way we want, within the constraints of the uEFI standard. Maybe it would be even better to modify the uEFI SecureBoot standard to do more exactly what we want but it might be too late for that as the current standard is now widely shipping and can't be changed. Maybe can't be changed for 20+ years.
Posted Mar 6, 2013 8:47 UTC (Wed)
by ortalo (guest, #4654)
[Link] (3 responses)
Linux users in this game are just the troublemakers who made apparent that M$ was grabbing control of OS instances. I am speculating yes, but I suspect this is more due to licensing reasons and monetary interest rather than security reasons and moral issues...
The actual full details may not be perfect, but that would not be the first a commercial security mechanism has design vulnerabilities... Furthermore, the objective is to increase the bill, not block the customer system.
Posted Mar 6, 2013 16:42 UTC (Wed)
by mjg59 (subscriber, #23239)
[Link] (2 responses)
No, it doesn't. You're able to disable the signature checking and you're able to install your own keys. Having done that, you're then free to lie to the OS about whether or not it booted a signed binary.
Posted Mar 6, 2013 18:42 UTC (Wed)
by raven667 (subscriber, #5198)
[Link] (1 responses)
Do I have that right, mjg59? 8-)
Posted Mar 6, 2013 18:50 UTC (Wed)
by mjg59 (subscriber, #23239)
[Link]
Posted Mar 5, 2013 17:21 UTC (Tue)
by mjg59 (subscriber, #23239)
[Link] (1 responses)
Posted Mar 6, 2013 8:57 UTC (Wed)
by anselm (subscriber, #2796)
[Link]
Also it's not as if piracy was keeping Microsoft up at night. People have been stealing Microsoft software for nearly 40 years and the company is still doing great. Bill Gates is still #2 on the Forbes list, so there doesn't seem to be an obvious problem.
Microsoft is doing little things here and there to make life harder for pirates (think »activation«), but actually getting rid of piracy altogether isn't that high up on their agenda. If they wanted to, they could certainly stamp out piracy nearly completely but (a) this would also inconvenience legitimate users, in particular »enterprise« users, which would be counterproductive – these users might get silly ideas like looking at operating systems that make for less hassle, such as Linux –, and (b) a certain level of piracy ensures that anybody who wants Windows badly enough will be able to get it, which is way better than encouraging them to look at icky free alternatives like Linux.
Posted Feb 28, 2013 12:24 UTC (Thu)
by keeperofdakeys (guest, #82635)
[Link]
You also can view a talk with James explaining it in detail. http://mirror.linux.org.au/linux.conf.au/2013/mp4/Making_...
Just a user interface issue?
Just a user interface issue?
Just a user interface issue?
Just a user interface issue?
Just a user interface issue?
Just a user interface issue?
Just a user interface issue?
- processs/time/people
- physical security
- key security
- redundancy of the physical security - and key security solution
Just a user interface issue?
Just a user interface issue?
So why not get them the money?
The problem is that the Linux Foundation doesn't have the money to be a CA [....]
Sounds like a perfect case for a Kickstarter project. I'd certainly pay a significant (to me) amount of cash to get a native Linux UEFI key built into most machines. I suspect I'm not alone.
So why not get them the money?
So why not get them the money?
Personnally, I would prefer to pay *him* directly to design an alternative security BIOS for Linux rather than reuse the one from M$...
So why not get them the money?
So why not get them the money?
What I am questioning is the true objective of this thing. What I am questioning too is whether bills will increase or not btw... At least, we have an occasion to demonstrate that servers running linux do not pay... anything!
So why not get them the money?
So why not get them the money?
So why not get them the money?
So why not get them the money?
So why not get them the money?
Just a user interface issue?