News [Ars] Silent Windows update patched side channel that leaked data from Intel CPUs

[Hitman928]

https://arstechnica.com/information-technology/2019/08/silent-windows-update-patched-side-channel-that-leaked-data-from-intel-cpus/

 

[NTMBK]

Yet another speculative execution security hole? Good grief.

 

[DrMrLordX]

Might want to merge this into the previous Xeon vulnerability thread.

https://forums.anandtech.com/threads/massive-security-hole-in-xeons-incoming-official-meltdown-spectre-discussion-thread.2532563/

Note that this drama has been going on for over 18 months. Also note that Intel's latest offerings that are touted as having mitigations for Spectre/Meltdown in hardware (well, mostly) have no hardware mitigations for Zombieload or this new sidechannel attack. Each new patch is more performance that is potentially lost.

 

[ehume]

. . . Each new patch is more performance that is potentially lost.

Lost from crippling a CPU or lost from accelerants forgone? E.g. -- speculative execution was an accelerant.

 

[NeoLuxembourg]

[phoronix.com] Initial Benchmarks Of The Spectre "SWAPGS" Mitigation Performance Impact

In the other common benchmarks tested, the SWAPGS performance impact tended to be 1% or less at least for this latest generation Intel Core i9 9900K desktop. Additional tests are currently being conducted on more hardware and more benchmarks. At least as it stands now it seems to be largely a 1% or less performance hit (with select exceptions at ~5%), so not nearly as bad as the other Spectre/Meltdown/Foreshadow/Zombieload mitigations, but this is on top of all that we've seen since January 2018.

Sauce: https://www.phoronix.com/scan.php?page=article&item=swapgs-spectre-impact&num=1

 

[VirtualLarry]

Another day, another Intel exploit, another Microsoft patch.

Thank God I'm not running Intel.

 

[lopri]

If I am understanding correctly, affected Intel CPUs are all HT-enabled ones, yes?

 

[DrMrLordX]

Lost from crippling a CPU or lost from accelerants forgone? E.g. -- speculative execution was an accelerant.

Accelerant. But you know the drill by now.

[phoronix.com] Initial Benchmarks Of The Spectre "SWAPGS" Mitigation Performance Impact

Less than 1% isn't so bad. Better than the other patches. Cascade Lake and Cooper Lake users won't get hurt too badly.

 

[Phynaz]

Ugh...Did anyone actually read the article? AMD is affected too, as is ARM. You fell for the clickbait headline. This nothing more than another Specter V1 variant.

As far as performance, here's Redhat:

What is the performance impact from this update?

The fix for this CVE has shown to cause a minimal performance impact. The impact will be felt more in applications with high rates of user-kernel-user space transitions. For example, in system calls, NMIs, and kernel interrupts.

Although there is no way to say what the impact will be for any given workload, in our testing we determined that the performance changes are not measurable in current benchmarks.

 

[Hitman928]

Ugh...Did anyone actually read the article? AMD is affected too, as is ARM. You fell for the clickbait headline. This nothing more than another Specter V1 variant.

As far as performance, here's Redhat:

There was some confusion on that at first. The original researchers said they tried it on 2 AMD chips but they weren't vulnerable. AMD says they are not affected. Microsoft says AMD is not affected. The Linux kernel patch only applies to Intel CPUs. Redhat is the only one saying AMD is vulnerable despite having no patch that applies to AMD systems.

I think the confusion comes in because there were multiple attack vectors for this exploit to work. Intel CPUs, even with the most current mitigations were vulnerable to all of the listed attack vectors. AMD was immune to all of them except one on a hardware level. For the one vector that could work on AMD, the Spectre variant 1 mitigation that was applied way back when this all started closes that vulnerability. So as long as you are patched for AMD with the initial Spectre mitigation, you're good to go, which is probably why the attacks failed on AMD CPUs for the original researchers.

tl;dr: AMD is not vulnerable to the new attack, Intel CPUs are. Performance hit doesn't look like hardly anything except a ~5% hit in some throughput tests.

 

[Phynaz]

What are you talking about? RHEL has been patched for AMD.

Don't be a Larry, facts are your friend.

You just do not learn. User insults
are not allowed in the tech areas.

AT Mod Usandthem

 

[Hitman928]

What are you talking about? RHEL has been patched for AMD.

Don't be a Larry, facts are your friend.

For someone who complains that people don't read the article, maybe you should look a little closer yourself:

Contrary to Red Hat's report initially saying AMD CPUs are affected, the Linux kernel is not applying this SWAPGS mitigation to AMD hardware

 

[VirtualLarry]

To be fair, Phynaz, both the thread title, and the article link title in the OP, specifically said Intel.

Edit: When I posted that.

 

[darkswordsman17]

Still not even an article reporting the mere existence of the ZombieLoad exploit from AT. Forget who, but one of them commented on another article that they would once they got an official response from Intel. Guess we can figure out what Intel's plan for handling this stuff is, just ignore it! As long as places like AT let them get away with it, no reason why they'd change. But we can certainly get every PR release reported on.

 

[DrMrLordX]

What are you talking about? RHEL has been patched for AMD.

Does anyone still do "self-pnwage of the year" in ATOT anymore?

Don't be a Larry, facts are your friend.

I, for one, would be honored. You rock, VirtualLarry! Mostly!

Still not even an article reporting the mere existence of the ZombieLoad exploit from AT.

Probably one of the reasons why I hit these forums 99x more often than the main page.