Catch up on stories from the past week (and beyond) at the Slashdot story archive

 


Forgot your password?
Close
typodupeerror
×
Intel Security Hardware

Intel's Cascade Lake CPUs Impacted By New Zombieload v2 Attack (zdnet.com) 43

Posted by msmash from the security-woes dept.
The Zombieload vulnerability disclosed earlier this year in May has a second variant that also works against more recent Intel processors, not just older ones, including Cascade Lake, Intel's latest line of high-end CPUs -- initially thought to have been unaffected. From a report: Intel is releasing microcode (CPU firmware) updates today to address this new Zombieload attack variant, as part of its monthly Patch Tuesday -- known as the Intel Platform Update (IPU) process. Back in May, two teams of academics disclosed a new batch of vulnerabilities that impacted Intel CPUs. Collectively known as MDS attacks, these are security flaws in the same class as Meltdown, Spectre, and Foreshadow. The attacks rely on taking advantage of the speculative execution process, which is an optimization technique that Intel added to its CPUs to improve data processing speeds and performance. Vulnerabilities like Meltdown, Spectre, and Foreshadow, showed that the speculative execution process was riddled with security holes. Disclosed in May, MDS attacks were just the latest line of vulnerabilities impacting speculative execution. They were different from the original Meltdown, Spectre, and Foreshadow bugs disclosed in 2018 because they attacked different areas of a CPU's speculative execution process. Further reading: Flaw in Intel PMx driver gives 'near-omnipotent control over a victim device'.
This discussion has been archived. No new comments can be posted.

Intel's Cascade Lake CPUs Impacted By New Zombieload v2 Attack More

Intel's Cascade Lake CPUs Impacted By New Zombieload v2 Attack

Comments Filter:

  • intel servers should be in the dumps by now go AMD (Score:4, Insightful)

    by Joe_Dragon ( 2206452 ) on Tuesday November 12, 2019 @03:31PM (#59407636)

    intel servers should be in the dumps by now go AMD

    • Re:intel servers should be in the dumps by now go (Score:4, Insightful)

      by SirAstral ( 1349985 ) on Tuesday November 12, 2019 @03:44PM (#59407694)

      Yea, AMD should nearly be an only choice about now but that is not how things work. Couple that with the fact that these vulnerabilities are not exactly easy to pull off and have low value in comparison to all of the other vulnerabilities out there. There is a never ending deluge of poor industry choices where security is concerned. It costs a lot of money to righteously dump Intel systems for AMD and it costs a lot of money to change standards even when it should be far less than it does cost.

      The corporate space is in a nutshell... designed to be as lumbering, painful, and idiotic as possible to make changes... from the absolute bottom to the top. In many respects, "The Cloud" is a natural result of this bullshit mentality that pervades the IT/IS industry. And the crazy thing is that all of these "change controls" got put in to stop the crazy and all it did was make it worse and make it cost more without seeing a single benefit.

      • AMD should nearly be an only choice about now but that is not how things work.

        Of course, Intel's bribes took care of that.

        • There is definitely some truth to that. There is a lot of "insider trading" in the server rack space. Lots of executive have preferred vendors for a reason. Sure some businesses will institute a "procurement" department to head this behavior off, but it just changes who gets enriched and sometimes does nothing to change anything. I remember trying to handle a procurement call one time in my career and all I was able to be was a little yes bitch. I hated it and wound up leaving the place. Not exclusive

          • There was a funny story on Reddit that either a university or another similar institution ordered an newest Ryzen OEM model in quite a significant quantity...and ooops, the vendor delivered Intel boxes, their bad. It was "a mistake" but it already got paid for by another department. The IT people were absolutely fuming (this was at the height of the security vulnerabilities) but they couldn't ultimately do anything about it. I vaguely recall that it turned out to be shadier than "a mistake" but I can't find

            • Yeah I really think Intel sends goons over to vendors to make sure only Intel boxes leave the warehouse.

            • Best Buy had a bunch of items in their PC section titled "Third-generation Ryzen" that magically change to Ryzen 2xxx CPUs when you pull up the full spec sheet. They're probably still there if you want to look for them.

              I ended up going to Newegg, and got delivered a Ryzen 3800x CPU - and got dicked by Gigabyte on the motherboard instead. It seems they're currently busy dumping broken Aorus X570 (AM4/Ryzen 3xxx) motherboards on the market - no returns, please. Windows won't read any SATA or NVME disks in an

              • Lets take a look at how many people gave them 5 eggs, 4 eggs, etc...

                x570 Aorus Master - 56% - 9% - 13% - 8% - 14% - 64 reviews - DO NOT BUY
                x570 Aorus Elite - 49% - 15% - 5% - 7% - 24% - 64 reviews - DO NOT BUY
                x570 Aorus Xtreme - 79% - 0% - 4% - 4% - 13% - 23 reviews - MAYBE
                x570 Aurus Ultra - 50% - 8% - 8% - 17% - 17% - 12 reviews - DO NOT BUY
                x570 Aurus Pro Wifi - 37% - 17% - 9% - 9% - 28% - 47 reviews - DO NOT BUY
                x570 Aurus Gaming X - 27% - 17% - 3% - 13% - 40% - 30 reviews - DO NOT BUY
                x570 Aurus Pr

                • I most definitely bought the motherboard from Newegg, and that was the shock. I've never seen this before in my 15 years of dealing with them. I know how to use the site, and I have used their RMA process before as well.

                  Under Warranty & Returns on the product listing it states:

                  "Return for refund within: non-refundable

                  Return for replacement until: January 31, 2020"

                  However, when I go to my orders page, where my other orders have two buttons for "Refund" and "Replace" - this order has only "Ineligible for

              • Well an mb thst cant read either nvme or sata is not fit for purpos, do demand a refund, law in hand, I would love to see them refuse that. Or am I missing something here?

        • Re: (Score:2)

          by Holi ( 250190 )
          Or the fact that until Zen AMD did not have a competitive product in the server space.

    • intel servers should be in the dumps by now go AMD

      Unfortunately nobody's perfect in this... AMD had the rdrand issue this year.

  • That's one argument settled (Score:5, Insightful)

    by raymorris ( 2726007 ) on Tuesday November 12, 2019 @03:42PM (#59407686) Journal

    We have a lot of debates on Slashdot, rarely resolving any. Back when Meltdown was news I said that this general class of attacks wouldn't be solved in the foreseeable future. Others argued that there was a fix and Meltdown would be a one-off event, solved with a patch and done with.

    > these are security flaws in the same class as Meltdown, Spectre, and Foreshadow.

    Can we now agree this class of attacks is not readily fixable, not fixed and done kind of thing? If so we could finally have a Slashdot debate resolved.

    • Re: (Score:2)

      by geek ( 5680 )

      Back when Meltdown was news I said that this general class of attacks wouldn't be solved in the foreseeable future. Others argued that there was a fix and Meltdown would be a one-off event, solved with a patch and done with

      So do you want a medal or something? Grats man, first guy to ever win an argument on the internet! What did you win?

      • Thinking more about it, I MIGHT have said on Slashdot that Trump was un-electable, that he could never win. In which case I would have decidely lost an argument on Slashdot. Anyone who said otherwise would have won a Slashdot argument prior to this one. :)

        I'm not sure if I said that or not.

    • Aside from Spectre v1 and a variant of Spectre-NG almost all other hardware vulnerabilities [wikipedia.org] have been fixed in Intel Ice Lake in hardware and AMD/Zen didn't have most of them in the first place, so you're simultaneously right and wrong.

      From what I've heard Spectre class vulnerabilities are indeed an implication of OoOE [wikipedia.org], so either you have both of them, or none of them.

      • so you're simultaneously right and wrong.

        Schroedinger's slashdotter?

      • > Aside from Spectre v1 and a variant of Spectre-NG almost all other hardware vulnerabilities have been

        The you linked says they've patched 18 versions over the the last 18 months. My prediction is that they'll continue to play whack-a-mole. That'll continue until they make a major change, then we'll have another, related series of issues.

    • Re: (Score:1)

      by Luthair ( 847766 )
      As someone who isn't an Intel fan (and just bought a 3700X), is it really surprising that a targeted attack against an Intel chip doesn't work on an AMD chip? I imagine AMD processors aren't undergoing the level of scrutiny that Intel CPUs currently are due to the much smaller marketshare over the past decade.
      • Then again Intel has been on the same 14nm Monolithic die process for what 5 years now? The first was on the Core M chips back in 2014. After 2021 AMD should be changing it's architecture, and socket in 2021 for consumer grade CPU's and the Threadripper line is already going to the new sTR4 socket. This is typically what happens when architectural innovation becomes stale, and gives people time to chip away at it over a period of time.

        • Re: (Score:2)

          by Luthair ( 847766 )
          What in your inane ramblings do you think applies to speculative execution vulnerabilities?
  • Here's one more for today The Jump Conditional Code (JCC) erratum [phoronix.com] - affects all SkyLake derived CPUs (Kaby, Coffee, Comet, Cascade), causes a performance loss up to 4%.

    • Re:Add one more (Score:4, Informative)

      by Artem S. Tashkinov ( 764309 ) on Tuesday November 12, 2019 @04:02PM (#59407772) Homepage

      Actually it's a whole lot more [intel.com] - 18 security advisories have been published today.

      And speaking of transient execution CPU vulnerabilities. Intel now has eighteen of them [wikipedia.org]. Looks like they cut a lot of corners when designing their CPUs to run faster.

      This makes this page [intel.com] look quite silly: Enabling Innovation with Security at the Core. Hopefully they will make the Willow Cove uArch a lot more secure.

      • Ironically many tried to play this off as if it wasn't really that big of a deal until it became a serious issue. I am curious as to when these vulnerabilities will become active hot targets for Intel consumers.

      • Looks like they cut a lot of corners when designing their CPUs to run faster.

        Actually it looks like they made a decision to promote speed over security. And you can't argue it didn't work.

        Still worth remembering is the insane difficulty of doing anything with speculative execution. This kind of attack is not just limited to the CPU, and yet here we are over a year later and a widely unpatchable critically classed exploit remains unused by malware.

        Now excuse me while I save myself some time and just phish someone's paypal password.

    • Re: (Score:2)

      by ameline ( 771895 )

      I can't find the actual erratum from Intel for that one. (JCC borkage)

      These JCC instructions are two bytes long (for the most part) -- so there is a roughly 1 in 16 chance that any given JCC instruction is a danger. It's far from clear what the actual failure mode is or how likely it is aside from that 1 in 16 chance of landing on a 32 byte boundary. It will be fairly easy for compilers and assemblers to avoid it by tossing in a NOP at the right point or scheduling instructions slightly differently. But if

      • It will be fairly easy for compilers and assemblers to avoid it by tossing in a NOP at the right point

        ...slowing down code on everything, to fix Intels bullshit, and only fixed if you recompile/reassemble with an updated compiler/assembler, and only if the libraries you use are also recompiled/reassembled.

        What a stupid idea, pretending to be simple and cost-free, because.... well... why?

        Do explain to use why you are covering for Intel right now.

  • Convenient Marketing!! (Score:3)

    by Matheus ( 586080 ) on Tuesday November 12, 2019 @04:23PM (#59407832) Homepage

    Co-branding with the recent release of Zombieland 2!!

  • walk away from a brand and find another better brand.
    As a consumer you have that freedom.
    You also have the freedom to review, test and tell the world why one CPU brand is better than others.. Keep testing and support brands that support your security needs.

    • Keep testing and support brands that support your security needs.

      You should buy AMD for many reasons. Price vs performance alone should be enough. Not skimping on PCIe lanes at a time where they are becoming increasingly relevant. Having better integrated graphics. Not hiding functionality such as hardware RAID behind plug in dongles.

      But no consumers should be worried about these classes of security exploits unless they also have Volcano insurance, run around afraid of getting struck by meteors, and have security detail guarding the front door of their house.

      • Price vs performance alone should be enough.

        Dominating! [cpubenchmark.net] as a matter of fact. Its not even close. Intel owned the high end segment entirely such that AMD only dominated the low end, and then chiplettes solving the yield problem. Intel is fucked.

        Not skimping on PCIe lanes at a time where they are becoming increasingly relevant.

        Intel isnt so bad on the server side...

        Not hiding functionality such as hardware RAID behind plug in dongles.

        hardware raid is a disaster waiting to happen

  • Comment removed based on user account deletion

Slashdot Top Deals

Stellar rays prove fibbing never pays. Embezzlement is another matter.

Close