Announcement

Collapse
No announcement yet.

Intel Releases New Microcode For Skylake CPUs (20200616)

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • archkde
    replied
    Originally posted by zerothruster View Post

    Yes, except that not all machines hung while booting. Mine is one, but now there is a report of a hang on a high-end (Xeon) 5e machine. So I'll keep using the 20200609 firmware.
    I've only seen reports of Skylake-U/Y machines hanging (and my Skylake-U also hung immediately when I tried to apply the dc microcode). Your experience is consistent with that (high-end Xeon does not sound like Skylake-U). I wonder why Intel reverted the update for more processor families.

    Leave a comment:


  • starshipeleven
    replied
    Originally posted by zerothruster View Post
    now there is a report of a hang on a high-end (Xeon) 5e machine.
    <loud angry top manager noises>

    Leave a comment:


  • zerothruster
    replied
    Originally posted by archkde View Post
    Judging by the revision number, this "update" actually reverts the microcode for the affected models to the older version that does not hang the processor as soon as trying to apply it.
    Yes, except that not all machines hung while booting. Mine is one, but now there is a report of a hang on a high-end (Xeon) 5e machine. So I'll keep using the 20200609 firmware.

    Leave a comment:


  • archkde
    replied
    Judging by the revision number, this "update" actually reverts the microcode for the affected models to the older version that does not hang the processor as soon as trying to apply it.

    Leave a comment:


  • zerothruster
    replied
    The problem with unbootable systems seems to have been confined to the -4e- versions, at least from looking at the issue on Monday. My own -5e- i3 was booting with the 20200609 microcode and with linux-5.7.2 it reported it had Microcode mitigation for SRBDS. So, if I "update" to the latest release (which reverts all the skylake updates to what was available at the back end of last year) I'll lose that.

    Of course, for my use-case SRBDS doesn't seem to be a major threat, but this is still disappointing.

    Leave a comment:


  • angrypie
    replied
    Originally posted by milkylainen View Post
    Seems that "Patch-Tuesday" is a contagious disease after all...
    More like "Patch Tuesday, Unbootable Wednesday."

    Leave a comment:


  • starshipeleven
    replied
    Originally posted by aphysically View Post
    A lot of distributions were having to nuke one of the last microcode updates because it was making certain systems unbootable.
    Going for the maximum security option.

    Leave a comment:


  • mroche
    replied
    Originally posted by scratchi View Post

    I assume this might be impacting RHEL too in this case.
    No need to assume, it is:

    - 6.10 https://bugzilla.redhat.com/show_bug.cgi?id=1846134
    - 7.9 https://bugzilla.redhat.com/show_bug.cgi?id=1846133
    - 8.3 https://bugzilla.redhat.com/show_bug.cgi?id=1846119

    Turns out I was mistaken in my prior posting, our Skylake (Xeon Gold 6148) servers were running off of the base 2.1-61microcode (C7.8), and not the problematic 2.1-61.6. Going to hold off until this 'update' gets pushed downstream to us.

    Cheers,
    Mike
    Last edited by mroche; 06-16-2020, 04:42 PM.

    Leave a comment:


  • scratchi
    replied
    Originally posted by aphysically View Post
    A lot of distributions were having to nuke one of the last microcode updates because it was making certain systems unbootable. Could the update be related?

    https://github.com/intel/Intel-Linux...iles/issues/31
    Yea, like this bug report from CentOS that got posted in yesterday's CentOS 8.2 release article: https://bugs.centos.org//view.php?id=17452

    I assume this might be impacting RHEL too in this case.

    Leave a comment:


  • aphysically
    replied
    A lot of distributions were having to nuke one of the last microcode updates because it was making certain systems unbootable. Could the update be related?

    https://github.com/intel/Intel-Linux...iles/issues/31

    Leave a comment:

Working...
X