Researchers Find Critical Vulnerabilities in AMD's Ryzen and EPYC Processors, But They Gave the Chipmaker Only 24 Hours Before Making the Findings Public (cnet.com)

Alfred Ng, reporting for CNET: Researchers have discovered critical security flaws in AMD chips that could allow attackers to access sensitive data from highly guarded processors across millions of devices. Particularly worrisome is the fact that the vulnerabilities lie in the so-called secure part of the processors -- typically where your device stores sensitive data like passwords and encryption keys. It's also where your processor makes sure nothing malicious is running when you start your computer. CTS-Labs, a security company based in Israel, announced Tuesday that its researchers had found 13 critical security vulnerabilities that would let attackers access data stored on AMD's Ryzen and EPYC processors, as well as install malware on them. Ryzen chips power desktop and laptop computers, while EPYC processors are found in servers. The researchers gave AMD less than 24 hours to look at the vulnerabilities and respond before publishing the report. Standard vulnerability disclosure calls for 90 days' notice so that companies have time to address flaws properly. An AMD spokesperson said, "At AMD, security is a top priority and we are continually working to ensure the safety of our users as new risks arise. We are investigating this report, which we just received, to understand the methodology and merit of the findings," an AMD spokesman said. Zack Whittaker, a security reporter at CBS, said: Here's the catch: AMD had less than a day to look at the research. No wonder why its response is so vague.

Comments

Sponsored by, Intel! (R)

[Anonymous Coward]

... someone needs to dig (deep) into who registered the amdflaw domain and who is funding this.

Re:Sponsored by, Intel! (R)

[sinij]

Yes, couple days to respond is a hit job and not a responsible disclosure. However, if AMD and Intel get into "flaw disclosure" wars, the only winner will be consumers. This is not a bad thing.

Re:Sponsored by, Intel! (R)

[Penguinisto]

Devil's Advocate: the disclosure(s) is (are) vague as hell on exploit details, let alone demonstrations or proof-of-concepts, so there is that.

All said though, still a dick move by CTS-Labs.

Re:Sponsored by, Intel! (R)

[Opportunist]

Care to inform me how I would be the winner if flaws in hardware become published with ZERO chance for their makers to deliver any kind of patch before malware creators get a chance to exploit them?

Re:Sponsored by, Intel! (R)

[Lonewolf666]

Yes, the combination of publication within a day and registering an AMD-denigrating domain for the purpose stinks. As others have written already, it looks like a PR hit job.

With a quick Google search (5 minutes) I could also find nothing substantial about CTS Labs. They have a professional looking website with quite a bit of Bullshit Bingo appeal, and a contact e-mail address on it.
Otherwise not much:
      -no postal address
      -no references from past projects
One might wonder if this is more than a shell company ;-)

Re:Sponsored by, Intel! (R)

[Baloroth]

As opposed to Intel, whose chips are perfectly secure. Except Intel had ~5 months to fix the problem before public disclosure (longer than responsible disclosure standards required). AMD is somehow only given 24 hours? That's not just irresponsible disclosure, that's an indirect attack.

Re:Sponsored by, Intel! (R)

[Carewolf]

Care to inform me how I would be the winner if flaws in hardware become published with ZERO chance for their makers to deliver any kind of patch before malware creators get a chance to exploit them?

The place this hole is, is the AMD version of IME, a useless piece of malware designed to remote-controlled your computer, which Intel and AMD puts there for enterprise purposes. Get rid of it from or make it default off and these issues goes away...

I have no fucking clue why they installed those crappy Internet-of-shit operating systems in there by default in the first place.

Re:Sponsored by, Intel! (R)

[sinij]

You stole my comment!

It isn't my fault that your speculative execution and prediction thinking leaked your post idea for everyone to see.

Re:Sponsored by, Intel! (R)

[DRJlaw]

Pretty clearly Intel-funded, yes.

Pretty clearly? Based on what evidence? All you've done is speculated as to motive.

Re:Sponsored by, Intel! (R)

[DarkOx]

We could be using single cycle machines with no pipe-lining, in order execution and several megabytes of SRAM.

They would be slow but they could be they could be secure by now. We chose fast and cheap over reliable.

Re:Sponsored by, Intel! (R)

[Burz]

Have to agree that the intent behind this super-fast disclosure looks malicious. It follows that the research was probably undertaken with malicious intent as well.

A very large chunk of Intel's operations are based in Israel, so that is one possible motivation for Israelis to go after AMD, which is based in the EU. Its widely known that the EU fined Intel over a $billion for threatening PC makers to avoid using too many AMD chips in PC products. There is revanchism and monopolist warfare going on here.

Not quite comparable to Intel's snafu

[erapert]

These vulnerabilities look like they are almost all problems with the chipset or AMD's equivalent to Intel's Management Engine.
So these aren't quite on par with Spectre and Meltdown.

Some firmware updates should fix almost all of this.
Still, it was sort of an asshole move to only give AMD 24 hours' notice just so they could get their 15 minutes of fame.
And, yes, it's disgusting to see AMD put out products with lots of weaknesses like this.

Re:Not quite comparable to Intel's snafu

[sl3xd]

Saying they aren't on par with Spectre or Meltdown is missing the point - it's an apples to oranges comparison, just like IME's many problems aren't comparable to Spectre or Meltdown.

It's not clear that firmware updates can fix it -- it depends on whether it's something that can be updated in firmware. Many security-critical hardware designs doesn't allow firmware updates, because at that stage modifiable firmware is a security hole in and of itself.

At the end of the day, it sounds like AMD's Secure Processor has similar problems as Intel's Management Engine. It's not exactly unexpected, as every remote management 'feature' of the type has historically been riddled with security holes, regardless of vendor.

I can't help but wonder, though, what the source of "24 hours notice" is; the articles I saw don't explain. I recall in years past, there are cases where researchers tried for months to get Microsoft to take their claims seriously. Microsoft wouldn't even acknowledge them, and when the researchers released it as a zero-day, and Microsoft shrieked they weren't given any notice...

If AMD really was only given 24 hours notice, it was outrageously unprofessional and unethical behavior by the research company.

Honestly, I'm more willing to believe corporate America would lie in an attempt to CYA than researchers would act in a way so unethical that nobody will work with them in the future.

pretty lame summary

[nimbius]

https://amdflaws.com/ for the actual exploits detailed. the "whitepaper" is mostly fluff, unless you enjoy pretty icons and charts..completely remiss of any technical implementation details outside of how vulnerable Windows is to this flaw. Idiotic green screen video confirms this exploit appears to have more studio production value than actual security value. https://www.youtube.com/watch?...

Re:pretty lame summary

[Luthair]

I feel like in the aftermath of Heartbleed it started to become common for researchers to try to brand their discoveries and are increasingly hyping hoping the mainstream press covers which works as an advertisement for the researcher and their org.

Intel gets 6 months and AMD gets a day?

[Anonymous Coward]

This all smells fishy. Hand me the tin-foil. I need a hat.

Follow the money

[spaceman375]

In collusion with intel or not, I'd bet these "researchers" have bought a bunch of intel stock over the last few months.

Re:Follow the money

[slack_justyb]

They literally spell it out on their disclaimer page.

Although we have a good faith belief in our analysis and believe it to be objective and unbiased, you are advised that we may have, either directly or indirectly, an economic interest in the performance of the securities of the companies whose products are the subject of our reports.

So while these exploits might be real, they just straight up fess to being shady as shit. This is some blackballing level of unethical behavior. They literally hit and run AMD for profit. Whoever these engineers are, this whole episode should be the end of any future career they might have had and it just stops short of what I would think would constitute an outright FTC investigation.

Twenty-four hour notice and then posting publicly the exploits isn't research, that's a willful attack.

trying to make a name for themselves...

[jmdevince]

CTS Labs only registered their domain (cts-labs.com) 6 months ago. They registered amdflaws.com 2018-02-22. So they spent time tweaking the marketing material. This is nothing but a new company trying to make a name for themselves and have instead pissed off true security researchers by not following responsible disclosure. From CTS' own site: "Due to the sensitive nature of security vulnerabilities, we usually work under strict mutual NDAs with our customers to ensure maximum safety and privacy". ... Horseshit.

Re:trying to make a name for themselves...

[bluefoxlucid]

I used to be a full disclosure guy.

I grew up.

Re:trying to make a name for themselves...

[MachineShedFred]

The sentence on the web site was probably edited from:

"Due to the sensitive nature of security vulnerabilities, we usually work under strict mutual NDAs with our customers to ensure maximum safety and privacy. If you would like to become one of our customers by handing over a signed NDA and a fat bag of money, you can contact us at the following email address. Should we find a flaw in a product that is not produced by one of our NDA partners, we'll first ask them for a fat bag of money, and if they don't immediately capitulate, we'll be publishing their dirty laundry as "full disclosure with previous notification".

Somehow I have a feeling that the "disclosure" to AMD included the offer of a mutual NDA and business-to-business financial arrangement, with AMD telling them to pound it.

They all have insane requirements

[Anonymous Coward]

All of those "vulnerabilities" have insane requirements like being able to defeat OEM BIOS flash protections or Windows' driver signing...

MASTERKEY:

        Exploiting MASTERKEY requires an attacker to be able to re-flash the BIOS with a specially crafted BIOS update. This update would contain Secure Processor metadata that exploits one of the vulnerabilities, as well as malware code compiled for ARM Cortex A5 – the processor inside the AMD Secure Processor.

RYZENFALL:

        Exploitation requires that an attacker be able to run a program with local-machine elevated administrator privileges. Accessing the Secure Processor is done through a vendor supplied driver that is digitally signed.

FALLOUT:

        Exploitation requires that an attacker be able to run a program with local-machine elevated administrator privileges. Accessing the Secure Processor is done through a vendor supplied driver that is digitally signed.

CHIMERA:

        Prerequisites for Exploitation: A program running with local-machine elevated administrator privileges. Access to the device is provided by a driver that is digitally signed by the vendor.

Re:They all have insane requirements

[93 Escort Wagon]

You're missing the point.

The point is - they came up with really cool names for each exploit.

They shortened AMD stocks

[xxxLCxxx]

Looks like somebody has shortened AMD stocks. This should be under investigation soon.

From reddit.com:

FRANKFURT, March 12 (Reuters) - German financial watchdog Bafin said on Monday that short-seller Viceroy Research breached German securities law with a research report on ProSiebenSat.1 as it did not notify the regulator of its activities.

Under German law, any entity that is not a securities firm, a fund manager, an EU administrative firm or an investment company that intends to publish recommendations on investments in assets must notify Bafin ahead of time, it said.

It also said Viceroy’s website did not contain information on where the company was based.

ProSieben last week rejected a critical report by Viceroy that led to a drop in its share price by as much as 9 percent, saying the allegations of questionable accounting contained in it were“unfounded and distorting reality”. (Reporting by Maria Sheahan Editing by Arno Schuetze)

Not a vulnerability

[FeelGood314]

This is both an attack on AMD (and possibly their stock price) and a way for the researchers to get publicity. This happens way to often, just this time it got more publicity than usual. What happens is researchers looking to make a name for themselves finds what they think could sound like exploit, the fact that it might already be public knowledge or hell even the way a device is supposed to work (e.g. exploit needs signed drivers and physical access) doesn't matter. Usually the "researchers" aren't very good. They use automated tools to scan for a vulnerability that they don't really understand and when you respond that "yeah, that 32 bit signed/unsign error might be exploitable if you send me a buffer with 2^31 + 7 bytes of data to a processes on an old 32 bit server but since the process only has 2GB of memory good luck.* The researches intentionally published right away so that the organization they are attacking doesn't have time to respond. The researchers didn't want a response because they knew the response would be "fuck off, this isn't a vulnerability!"

*yes, I had this conversation.

Now this is suspicious

[Megol]

Look at how the information is delivered. "This site is to inform the public about the vulnerabilities and call upon AMD and the security community to fix the vulnerable products." - but doesn't actually give AMD the time to fix the problem(s).

Look at the website: amdflaws.com
Nice name.

"MASTERKEY requires an attacker to be able to re-flash the BIOS with a specially crafted BIOS update"
So this is a low impact problem. Yes they try to hype it but the fact is if anyone have access to a computer one should always assume they can gain control.
For just a few years ago people wouldn't even try to portrait it as a problem.

The rest are similar things - bypassing security while still needing physical and/or elevated privileges. Yes there may be problems caused by this, no the problems aren't really bad.

I wouldn't be surprised if Intel spent some $$$ to encourage the group behind this to select the website name, the naming of the exploits (or "exploits" in some cases), how they are presented on the website and the white paper, and lastly to not giving AMD any chance to patch the problems. Add to this the quote above that show an exceptional level of dishonesty.

And if Intel didn't give them anything the group missed out - Intel have dedicated resources for these kind of operations as anyone that have been into computers for a while should know.

Disgusting.

From their own Disclaimer

[iCEBaLM]

https://amdflaws.com/disclaime...

"Although we have a good faith belief in our analysis and believe it to be objective and unbiased, you are advised that we may have, either directly or indirectly, an economic interest in the performance of the securities of the companies whose products are the subject of our reports."

24 hours notice. "Researchers" who seem to spring up out of nowhere. Creating a website and videos for maximum publicity. All the security flaws seem overblown (require actual flashing of firmware or bypassing driver signing), and.. wait, what's this?

https://www.reddit.com/r/AMD_S...

A huge number of put option (a bet that share price will fall dramatically) volume 5 days ago?

Nah, this is totally legit!