AMD's turn for some side channel fun

[moproblems99]

So, I haven't had a chance to read the whole thing but it looks possibly a variant of Spectre/meltdown.
It also looks like at least Zen and Zen+ are affected.

https://www.google.com/amp/s/www.to...l-attacks-discovered-impacts-zen-architecture

Like we've seen with many of the recent attacks against modern processors, the two AMD vulnerabilities center on side-channel approaches, in this case a Spectre-based attack, that enable researchers to tease out what would normally be protected information.

 

[xkm1948]

Zen is still very young, plenty of time to find new security vulnerabilities. Nothing designed by human is perfect. Key is how fast it will take for AMD to patch the holes.

Also i hope amd fans take this one well. Just remember these are billion dollar for profit companies, not your bff

 

[eidairaman1]

So, I haven't had a chance to read the whole thing but it looks possibly a variant of Spectre/meltdown.
It also looks like at least Zen and Zen+ are affected.

https://www.google.com/amp/s/www.to...l-attacks-discovered-impacts-zen-architecture

So are you goin back to intel then?

 

[biffzinker]

 

[mtcn77]

So, I haven't had a chance to read the whole thing but it looks possibly a variant of Spectre/meltdown.

I have just 1 problem with that statement. Spectre is not a side channel attack, Meltdown is. Spectre is a branch prediction attack.

 

[hat]

Disappointing, yet not particularly surprising. On that note, I've said many times before that I'll be interested to see Intel come up with a new architecture that's not full of holes like Skylake is (apparently that's supposed to be Rocket Lake, in.... uh, 2021?) but it's only a matter of time until they find the holes in that too.

 

[Chomiq]

View attachment 147407

View attachment 147408

HU is really stretching this. Both companies fund research into vulnerabilities, I recall recent Intel one where researchers thanked both Intel and AMD for their donations/gifts (it might have been the same group that discovered this one). It's a standard practice and as a researcher you're obligated to disclose this in your paper, otherwise your funding may be revoked, not to mention the impact on your credibility.
The "gifts" were probably hardware samples for testing, not yachts and speedboats.

 

[steen]

I have just 1 problem with that statement. Spectre is not a side channel attack, Meltdown is. Spectre is a branch prediction attack.

Yep. A followup tweet from the author: "The attacks leak a few bit of meta-data. Meltdown and Zombieload leak tons of actual data." As others have said, it's a matter of time & resources before vulnerabilities are discovered in any piece of hw/sw.

HU is really stretching this.

Agreed. The paper's authors are academics & have published their findings & detailed COI disclosure. Many IHVs fund genuine research. Intel is not the bad actor here. Pity they weren't as concerned over the DLSS 2.0=good/competitor products=bad story shopped by Nv PR since the beginning of the year.

 

[Bill_Bright]

Key is how fast it will take for AMD to patch the holes.

Not just how fast, but IF patching is even possible, and IF it is, IF the patch does not impact performance.

Also i hope amd fans take this one well.

And, if applicable, how they treated and commented on Intel when the shoe was on the other foot!

 

[john_]

5 March. AMD Financial Day. Great success.
8 March. Tom's Hardware(who else?) comes out with this article. Coincidence?
Nothing is full proof, but the time of publication, considering that the whole research had Intel behind it, it's not a coincidence, in my opinion.
I do expect extra drama from certain tech press, about how AMD CPUs are as vulnerable as Intel CPUs (ignoring the fact that for every one AMD vulnerability discovered we hear about a dozen for Intel CPUs).

 

[mtcn77]

Spectre is more like the original hyperthreading logical core physical access isolation issue that got canned. I cannot quote on that, forgot the quote.

 

[R-T-B]

So are you goin back to intel then?

Depends on who is better bang for the buck at the time of purchase. All you can really do.

I got a superduper deal on my present chip or I'd have went AMD again. Both sides can and will have issues, decide on what you know at time of purchase.

Spectre is more like the original hyperthreading logical core physical access isolation issue that got canned. I cannot quote on that, forgot the quote.

Spectre is a general speculative execution bug, always affected AMD but not to this degree. I think you are thinking of MDS bug family.

 

[mtcn77]

Spectre is a general speculative execution bug, always affected AMD but not to this degree. I think you are thinking of MDS bug family.

Do you think that is how hyperthreading thread snooping worked in the previous instant?
I don't have the words to describe it.

 

[phill]

I've personally made my bed and I'm enjoying it (well, so to speak )

AMD have been an utter winner for me, so that's now where my money will go.. Until such time things change or I feel like a change

 

[R-T-B]

Do you think that is how hyperthreading thread snooping worked in the previous instant?
I don't have the words to describe it.

I'm saying hyperthreading-thread snooping is not a spectre class vulnerability and thus offtopic. It is in the MDS-category, which is in itself a speculative execution bug but generally considered seperately for it's different attack vector.

This is literlly what I do for a living, FYI. I'm a full-time employed security consultant.

 

[Vya Domus]

5 March. AMD Financial Day. Great success.
8 March. Tom's Hardware(who else?) comes out with this article. Coincidence?
Nothing is full proof, but the time of publication, considering that the whole research had Intel behind it, it's not a coincidence, in my opinion.
I do expect extra drama from certain tech press, about how AMD CPUs are as vulnerable as Intel CPUs (ignoring the fact that for every one AMD vulnerability discovered we hear about a dozen for Intel CPUs).

I noticed they went as far as testing this on ancient K8 processors, that's pretty ridiculous, we're talking about CPUs from 2003. Clearly they were digging deep everywhere they could to find something.

You know what's funny though, because of the colossal amount of vulnerabilities found in Intel's architecture everyone has become numb to these things, they could uncover a million more and no one would really care.

 

[R-T-B]

I noticed they went as far as testing this on ancient K8 processors, that's pretty ridiculous, we're talking about CPUs from 2003. Clearly they were digging deep everywhere they could to find something.

No, that's standard practice when hardware vulnerabity testing. It helps you understand future architectures to understand the ancestry, is how I'd describe it to the layman.

I believe meltdown was tested down to Pentium II class chips, as an example.

You know what's funny though, because of the colossal amount of vulnerabilities found in Intel's architecture everyone has become numb to these things, they could uncover a million more and no one would really care.

The layman may not care. But some of us in the know watch this stuff with a career on the line.

 

[Vya Domus]

The layman may not care. But some of us in the know watch this stuff with a career on the line.

Well tell me how many careers have been on the line because of these things then. Come on, this is overly dramatic to say the the least, I haven't heard about a single business or service that has been endangered due to any of these vulnerabilities. I'd go as far as to bet most aren't even aware of them and they're probably never going to be.

 

[R-T-B]

Well tell me how many careers have been on the line because of these things then. Come on, this is overly dramatic to say the the least, I haven't heard about a single business or service that has been endangered due to any of these vulnerabilities.

*I don't know why everyone is worried about getting sick. Every sickness I have ever had, I survived!

Survivership bias. It only takes one bad one.

At any rate, cloud hosting has been impacted by this in a big way. Not much else yet, but if you're the one who'll be fired when they find "the one" you do subscribe to the issue in general. Believe me it's not just me watching.

Nothing dramatic about that, just my reality.

 

[Vya Domus]

At any rate, cloud hosting has been impacted by this in a big way.

In what way ?

 

[R-T-B]

In what way ?

Performance or cross-vm exploitation. There were articles at some point on the performance impact. The other? You probably won't find any.

You don't hear about it because these exploitation issues are usually settled privately in the cloud. No one likes hurting their parent companies share price when they can just bribe a customer and retroactively patch/fire somebody.

Heck, to be a host for certain cloud services, they literally audit your microcode version and bios settings now. So yeah, you run into it.

 

[Vya Domus]

You don't hear about it because these exploitation issues are usually settled privately in the cloud.

And/or nothing major occurred, you're not really going to tell me everything is in shambles but somehow they're keeping this frail facade up.

 

[R-T-B]

And/or nothing major occurred, you're not really going to tell me everything is in shambles but somehow they're keeping this frail facade up.

It's not. Most providers patched microcode and/or settled it within weeks with some retroactive firings and patchings. There was some bios-update downtime but nothing unmanageable. Maybe some money lost for the careless, but few if anything significant enough to make news. That wasn't really my point.

My point was more personal: If you were fired, it was probably dramatic and life altering and you probably sucked at your job by not keeping your ears to the ground.

I don't mean to come across as rude, but what's so hard to believe about that? If you work this field, you care?

 

[thesmokingman]

Yea, paid for generously at that by Intel.

This is some ridiculous shit. They even had to resort to using a modified kernel.

https://twitter.com/i/web/status/1236333792358531074

Then AMD's take... lolol.

3/7/20

We are aware of a new white paper that claims potential security exploits in AMD CPUs, whereby a malicious actor could manipulate a cache-related feature to potentially transmit user data in an unintended way. The researchers then pair this data path with known and mitigated software or speculative execution side channel vulnerabilities. AMD believes these are not new speculation-based attacks.


AMD continues to recommend the following best practices to help mitigate against side-channel issues:

 

[R-T-B]

This is some ridiculous shit. They even had to resort to using a modified kernel.

I mean, if you're going to attack a VM hypervisor, a modified kernel is a given. You don't exactly expect the bad guys to apply their own mitigations, do you?