Kernel memory leaking' Intel processor design flaw forces Linux, Windows redesign

Discussion in 'Frontpage news' started by zero_cool, Jan 2, 2018.

  1. scatman839

    scatman839 Ancient Guru

    Messages:
    14,121
    Likes Received:
    538
    GPU:
    3080, KD55XD800
  2. pegasus1

    pegasus1 Ancient Guru

    Messages:
    5,120
    Likes Received:
    3,508
    GPU:
    TUF 4090
    They will find a way, they always do, remember the line "Its impossible to code games to use more than a single core".
     
  3. __hollywood|meo

    __hollywood|meo Ancient Guru

    Messages:
    2,991
    Likes Received:
    139
    GPU:
    6700xt @2.7ghz
    not too surprising. the patch impacts I/O heavily.
     
  4. Turanis

    Turanis Guest

    Messages:
    1,779
    Likes Received:
    489
    GPU:
    Gigabyte RX500
    Jagman and Silva like this.

  5. Kaarme

    Kaarme Ancient Guru

    Messages:
    3,513
    Likes Received:
    2,355
    GPU:
    Nvidia 4070 FE
    Somebody should benchmark that latest Assassin's Creed after the patch. Doesn't it use a VM for its paranoid schizophrenic copy protection?
     
  6. Turanis

    Turanis Guest

    Messages:
    1,779
    Likes Received:
    489
    GPU:
    Gigabyte RX500
    Yes,2 in 1 DRM protection.
     
  7. H83

    H83 Ancient Guru

    Messages:
    5,465
    Likes Received:
    3,003
    GPU:
    XFX Black 6950XT
    OK i understand that his is really bad but i have a question: this flaw/bug only affects people running virtual machines or affects anyone that´s using an Intel CPU??? If it´s the latter than that´s a problem for me because i´m running W7 and i´m not expecting MS to patch it...

    It seems Intel want´s me to go with AMD next time i build a new rig...
     
  8. mbk1969

    mbk1969 Ancient Guru

    Messages:
    15,541
    Likes Received:
    13,561
    GPU:
    GF RTX 4070
    I think you do interpret "virtual" wrong. All processes run in virtual machine, every process "thinks" that it runs exclusively on computer, having all CPU, RAM, keyboard and mouse - all hardware for itself only.

    And to me when OS kernel should get security patch for CPU flaw means bad design of OS kernel.
     
  9. Evildead666

    Evildead666 Guest

    Messages:
    1,309
    Likes Received:
    277
    GPU:
    Vega64/EKWB/Noctua
    The more i'm reading about, the more I see mentioned Virtual Memory, not just Virtual Machines.

    So basically, every Intel CPU would be affected, and need to be patched, but not everyone would get the slowdown due to it being only in specific cases.
     
  10. k3vst3r

    k3vst3r Ancient Guru

    Messages:
    3,702
    Likes Received:
    177
    GPU:
    KP3090
    Performance impact is big on NVME SSD's. 50%+ performance reduction.

    "In some of Phoronix's I/O based benchmarks we can see that system performance is impacted on both their i7 6800K and i7 8700K based systems. It must be noted that the 6800K-based system is using a SATA SSD while the 8700K-based system is using a Samsung 950 Pro M.2 NVMe SSD, explaining why its performance is hit hardest.

    The performance degradation here is astounding, showcasing exactly how much these changes can affect the performance of certain applications. Given the fact that these are synthetic benchmarks, the real-world performance impact of this kernel patch is expected to be lower for consumers, though it does highlight how bad the problem can be."

    https://overclock3d.net/news/softwa...mpacting_security_update_have_been_released/1
     

  11. BlackScout

    BlackScout Active Member

    Messages:
    83
    Likes Received:
    30
    GPU:
    RTX 3060 12GB
  12. D3M1G0D

    D3M1G0D Guest

    Messages:
    2,068
    Likes Received:
    1,341
    GPU:
    2 x GeForce 1080 Ti
    Wow, that performance hit is insane! If I was a data center operator, I would be extremely worried about the impact this could have on my business. To say that Intel massively screwed up would be a severe understatement. Little wonder why AMD stock is shooting up.

    Also, since this is a hardware bug in their current design, it's likely that upcoming CPUs would also be affected (e.g., Ice Lake). Otherwise, they'll need to quickly do a redesign to avoid this issue, which will cause delays. No real positives here.
     
  13. Ryu5uzaku

    Ryu5uzaku Ancient Guru

    Messages:
    7,547
    Likes Received:
    608
    GPU:
    6800 XT
    Don't worry it's all the core cpus. Intel might fix it hardware wise in some future cpu.
     
  14. DeskStar

    DeskStar Guest

    Messages:
    1,307
    Likes Received:
    229
    GPU:
    EVGA 3080Ti/3090FTW
    Sooooooooo going GPU accelerated in your servers has seems/sounds like a better approach as more and more time goes on. Not only are they faster, but could they be more secure!?!
     
  15. AsiJu

    AsiJu Ancient Guru

    Messages:
    8,811
    Likes Received:
    3,369
    GPU:
    KFA2 4070Ti EXG.v2
    Yep, article refers to virtual memory address space. Actually I don't think it was stated anywhere that this would affect VMs only or in particular.

    Problem being that this flaw could allow exploiting speculative execution to access kernel mode address space from user mode processes.
    So the actual issue is that (based on current belief) Intel CPUs speculative execution does not check for such violations (as user mode process trying to access kernel mode addresses) before execution.

    Ofc one cannot directly run such code but if Intel CPU "guesses" such code would be run then it will execute such code.

    Again, that's what I got from the article and the article is based on speculation (based on the fact Linux patch separates user mode and kernel mode address spaces).

    edit: typos
     
    Last edited: Jan 4, 2018

  16. Alessio1989

    Alessio1989 Ancient Guru

    Messages:
    2,941
    Likes Received:
    1,239
    GPU:
    .
    This is not going to be fixed with microcode in a easy way. Disabling TLB would be worst.
    GPU GUI acceleration has nothing to do with this. Actually accelerating the GUI with the GPU will cause more context switch to kernel than a software rendering.
     
  17. Aura89

    Aura89 Ancient Guru

    Messages:
    8,413
    Likes Received:
    1,483
    GPU:
    -
    ....lol?

    That was meant as a joke, right?

    I mean, it must have been. There's literally no way possible someone could have that mindset....
     
  18. mbk1969

    mbk1969 Ancient Guru

    Messages:
    15,541
    Likes Received:
    13,561
    GPU:
    GF RTX 4070
    Why? If this flaw was with us for years and our rigs were not suffered from any sort of breach then problem with this flaw is not worth a loss in performance for home users. I would prefer to decide whether to install that patch with fix for myself (as well as other things in OS).
     
    jaggerwild likes this.
  19. Evildead666

    Evildead666 Guest

    Messages:
    1,309
    Likes Received:
    277
    GPU:
    Vega64/EKWB/Noctua
    I think the whole point is that the fix, won't impact most users either. ;)
     
  20. mbk1969

    mbk1969 Ancient Guru

    Messages:
    15,541
    Likes Received:
    13,561
    GPU:
    GF RTX 4070
    Regardless of that I prefer to decide for my home rig.

    If performance loss is originated namely in changes in transitions from user code to kernel code (and back) then the impact should be bigger for apps which do huge amount of kernel code calls. For example SSD/HDD benchmark apps - essentially all they do is calling kernel code for read-write operations.
     

Share This Page