Linux 4.10 UEFI changes

Matt Fleming posted UEFI changes for Linux 4.10 kernel.

Folks, please pull the following v4.10 material. There isn’t a huge amount of stuff here. The biggest change is the EFI dev path parser code from Lukas to get thunderbolt working on his macbook.
 * Fix an allocation bug in the generic EFI libstub where alignment and adjusted size isn’t taken into account – Roy Franz
 * Update the EFI MAINTAINERS entry to include ARM and arm64 files and directories – Ard Biesheuvel
 * Add new feature to seed the RNG from the stashed value returned by EFI_RNG_PROTOCOL in EFI stub and wire up for ARM/arm64 – Ard Biesheuvel
 * Retrieve Apple device properties from within the EFI stub to fully support thunderbolt devices on Apple Macbooks – Lukas Wunner

More details on the Thunderbolt patch:

thunderbolt: Use Device ROM retrieved from EFI:
Macs with Thunderbolt 1 do not have a unit-specific DROM: The DROM is empty with uid 0x1000000000000. (Apple started factory-burning a unit- specific DROM with Thunderbolt 2.) Instead, the NHI EFI driver supplies a DROM in a device property. Use it if available. It’s only available when booting with the efistub.  If it’s not available, silently fall back to our hardcoded DROM.  The size of the DROM is always 256 bytes. The number is hardcoded into the NHI EFI driver. This commit can deal with an arbitrary size however, just in case they ever change that.  Background information: The EFI firmware volume contains ROM files for the NHI, GMUX and several other chips as well as key material. This strategy allows Apple to deploy ROM or key updates by simply publishing an EFI firmware update on their website. Drivers do not access those files directly but rather through a file server via EFI protocol AC5E4829-A8FD-440B-AF33-9FFE013B12D8. Files are identified by GUID, the NHI DROM has 339370BD-CFC6-4454-8EF7-704653120818.  The NHI EFI driver amends that file with a unit-specific uid. The uid has 64 bit but its entropy is much lower: 24 bit represent the model, 24 bit are taken from a serial number, 16 bit are fixed. The NHI EFI driver obtains the serial number via the DataHub protocol, copies it into the DROM, calculates the CRC and submits the result as a device property.  A modification is needed in the resume code where we currently read the uid of all switches in the hierarchy to detect plug events that occurred during sleep. On Thunderbolt 1 root switches this will now lead to a mismatch between the uid of the empty DROM and the EFI DROM. Exempt the root switch from this check: It’s built in, so the uid should never change. However we continue to *read* the uid of the root switch, this seems like a good way to test its reachability after resume.

git://git.kernel.org/pub/scm/linux/kernel/git/efi/efi.git tags/efi-next


more info on PCI Leech

 DMA attacking over USB-C and Thunderbolt 3
I just got an Intel NUC Skull Canyon that has an USB-C port capable of Thunderbolt 3. Thunderbolt is interesting since it’s able to carry PCI Express which is Direct Memory Access (DMA) capable. I have previously demonstrated how it is possible to DMA-attack macs over Thunderbolt 2 in my DEF CON talk “Direct Memory Attack the Kernel”. To attack my MacBook Air in the DEF CON demo I used a Sonnet Echo ExpressCard Thunderbolt 2 to ExpressCard adapter together with a PCILeech ExpressCard. I also got a Thunderbolt 3 to Thunderbolt 2 adapter from Startech and I wanted to try it on the NUC to see if it’s possible to use it for DMA attacks, or if Thunderbolt has been secured. […]




additional Apple device property support for Linux efistub

Lukas Wunner submitted a 6-part patch to the Linux-(EFI,Kernel) lists with additional Apple EFI firmware support.

Apple device properties
Apple EFI drivers supply device properties which are needed to support Macs optimally. This series extends the efistub to retrieve the device properties before ExitBootServices is called (patch [1/6]). They are assigned to devices in an fs_initcall (patch [5/6]). As a first use case, the Thunderbolt driver is amended to take advantage of the Device ROM supplied by EFI (patch [6/6]). A by-product is a parser for EFI Device Paths which finds the struct device corresponding to a given path. This is needed to assign properties to their devices (patch [3/6]). […]

For more info:


Intel 01.org mailing lists

It is sometimes funny to watch a company do open source. Intel’s 01.org, for Open Source projects, has a mailing list server with multiple lists:

There are lists for LUV and CHIPSEC. These work fine!

There is a list for Thunderbolt Software. …but it is a closed list, with no public archives. 😦

The text that it is a closed list:
“This is a hidden list, which means that the list of members is available only to the list administrator.”

There’s a list for Intel Kernel Guard Technology (KGT). It also is a closed list, with the same text as the Thunderbolt list. BUT, their archives are publicly-available.

There’s a list for BIOS Implementation Test Suite (BITS)!
But there are no archives, perhaps a closed list, or just broken archives?

I rather wish Intel used intel.com or 01.com for closed lists, and kept the Open Source-centric 01.0rg’s list all public, with working archives. 😦


Intel 01.org Thunderbolt project

I just noticed a Thunderbolt software project on 01.org, the Intel Open Source project site:


The last release was a few months ago. However, this is not a normal open source project. It is “restricted open source”, only only for Thunderbolt vendors, not the public.

The code can be accessed from the restricted Git Hub at: https://github.com/01org/thunderbolt-software
An access can requested at thunderbolt-linux@intel.com only for manufacturers having a valid Thunderbolt technology license.




Interview with LegbaCore (and: their OpROM checker ships!)

A while ago, I emailed Corey and Xeno of LegbaCore, and sent them a few questions for an ‘interview’ for this blog. Well, here’s the results (also see EOM for URLs):

Q: This October in Singapore you’re giving 3-days of training at HITB. Besides new Apple EFI skills, can you give us some other new things that’ll be in this training?
A: The course introduces the basics of evaluating firmware and SMM on modern platforms for security vulnerabilities, as well as for potential compromises. Specifically we work through methodologies for identifying whether or not your system contains publically known vulnerabilities (which a great majority of them do). We also introduce a firmware forensic and reverse engineering methodology for identifying potential firmware compromises… so say if you got comprised by something like the hacking team UEFI rootkit, we’d talk about the tools and procedures that would be useful for identifying and analyzing this threat both on one particular system and at scale across your enterprise.
    The most important point of the training is it will be focused on real hardware that is deployed in real environments. A large portion of the course will involve evaluating the hardware that students bring with them. This way if you’re in charge of malware detection/response in an enterprise, you’ll walk away with actionable information related to the hardware you are deploying on your network.

Q: You had a Twitter post a few weeks (months?) back, saying that you were going to start releasing information about OEM systems’s vulnerabilities. What’s up with that project, I’m eager to see this data, as Consumer Reports and other computer review sources are useless for this most crucial pre-sales information. Any chance you could give FirmwareSecurity.com a teaser of this information, perhaps one new OEM model released in the last 6 months that’s insecure? 🙂
A: We anticipate that the project to start making some vendors’ firmware security failings more apparent (via a public website) will probably kick off in early 2016. We want to give all vendors that we think may have an interest in improving their security a chance to either talk with us about working with them, or show that they can make measurable security improvements on their own within this timeframe.

Q: You had a Twitter post a few days ago, pointing to a new LegbaCore Github repository for a new Option ROM checking tool. This sounds very interesting! What kinds of Option ROM(s) will it support? What platform(s) will it run on? When can we expect initial release?
A: It will only integrity check the Apple Thunderbolt to Ethernet adapter’s option ROM. It will work in conjunction with a port of the linux tg3 kernel driver to run on Macs to dump the OROM. The “ethtool” command can also be run to dump the OROM on linux systems that have a thunderbolt port and the tg3 driver available. The tool will be released to coincide with the talk at BlackHat, August 6th.

Q: Beyond this new Option ROM checker, does LegbaCore have any additional tool plans in the works? If so, any details to reveal?
A: Our current thinking on tools is that we expect that we will make clean-slate “best effort” tools freely available at some point in the future. These tools would be like all typical security tools, being not particularly trustworthy, and thus vulnerable to attacker subversion. They would mostly be useful for catching attackers as they first enter into the domain, and are not particularly sophisticated. However we will only make those tools available once we have commercial-grade tools available, that have a trust argument for why these paid tools have the ability to stand up to sophisticated and well-resourced adversaries. And in the background we will work with vendors to add capabilities such as SMM Dual Monitor mode, which significantly strengthen the trust arguments

Q: The Copernicus release is mostly Windows-centric, but also includes a cross-platform, bios_diff.py tool in the release. Will the new LegbaCore github tree include a open source bios_diff project, perhaps to get open source patches for improved BIOS parsing beyond EFIPWN, perhaps like that in UEFITool?
A: While bios_diff.py continues to provide the only simple, semantically aware integrity checking capability for BIOS, it has a number of issues. E.g. in the context of our latest work, it simply doesn’t work to integrity check Apple firmware, because Apple firmware update structure does not look the same as the structure on-disk.

Q: Post-October/HITB, what’s the next LegbaCore BIOS/UEFI security training event you’ll be giving?
A: Later in October I’ll be offering a similar training at Ruxcon breakpoint. Beyond that we are fairly busy with private training engagements.

Q: Any hints what kind of new firmware vulnerability research you’re working on, and when we might see some of the results?
A: We are branching out to the security of peripheral devices such as network cards, HDs/SSDs, embedded controllers, GPUs, the Intel Management Engine, etc. We are shooting to have our first talk on one of these topics around December.

Q: Recently Stephan of coreboot mentioned that in the past MITRE said that Chrome OS firmware was more secure than UEFI. Both coreboot+depthcharge’s Verified Boot and UEFI’s Secure Boot both seem pretty similar, in terms of PKI usage. Have you an opinion on the security strength of either of these firmware solutions?
A: We have never evaluated CoreBoot to any level of depth. Hardware-wise, we like the Chromebook requirement to provide physical write protection for the flash chip via a screw. The way that Chromebooks supposedly root their boot trust in the embedded controller hardware, as described to us, sounded like a good idea. But without having actually spent time looking at the platforms, we cannot say much in terms of the security (or lack thereof) relative to UEFI-based systems.

—-[End of interview.]

Thanks Corey and Xeno!

The code was added 5 days ago, and I missed it!

Upcoming training:

Stephan’s comment on coreboot security: