Changeset 3179c69


Ignore:
Timestamp:
06/23/2020 06:39:31 PM (15 months ago)
Author:
Ken Moffat <ken@…>
Branches:
10.0, 10.1, 11.0, qt5new, trunk, xry111/git-date, xry111/git-date-for-trunk, xry111/git-date-test
Children:
6df48a7
Parents:
dc96204
Message:

Firmware: update some details:

  1. Intel (CPU) microcode - update the details for Skylake in the light

of the later release.

  1. Video firmware - I had overlooked that Nvidia provided signed

firmware to linux-firmware in January and February for Turing GPUs
and that it is only needed for hardware acceleration so not
essential for working KMS. For completeness, also mention the
problematic Intel iGPU firmware for Skylake and later.

git-svn-id: svn://svn.linuxfromscratch.org/BLFS/trunk/BOOK@23317 af4574ff-66df-0310-9fd7-8a98e5e911e0

File:
1 edited

Legend:

Unmodified
Added
Removed
  • postlfs/config/firmware.xml

    rdc96204 r3179c69  
    6969    <listitem>
    7070      <para>
    71         Firmware for video controllers. On x86 machines this seems to mostly
    72         apply to ATI devices (Radeon and AMDGPU chips) and Nvidia Maxwell
    73         and Pascal cards which all require firmware to be able to use KMS
     71        Firmware for video controllers. On x86 machines this is required for
     72        ATI devices (Radeon and AMDGPU chips) and may be useful for Intel (Skylake
     73        and later) and Nvidia (Kepler and later) GPUs.
     74      </para>
     75
     76      <para>
     77        ATI Radeon and AMGPU devices all require firmware to be able to use KMS
    7478        (kernel modesetting - the preferred option) as well as for Xorg. For
    75         earlier radeon chips (before the R600), the firmware is still in the
    76         kernel.
     79        old radeon chips (before the R600), the firmware is still in the
     80        kernel source.
     81      </para>
     82
     83      <para>
     84        Intel integrated GPUs from Skylake onwards can use firmware for GuC
     85        (the Graphics microcontroller), and also for the HuC (HEVC/H265
     86        microcontroller which offloads to the GPU) and the DMC (Display
     87        Microcontroller) to provide additional low-power states. The GuC and
     88        HuC have had a chequered history in the kernel and updated firmware
     89        may be disabled by default, depending on your kernel version. Further
     90        details may be found at <ulink
     91        url="https://01.org/linuxgraphics/downloads/firmware/">01.org</ulink>
     92        and <ulink
     93        url="https://wiki.archlinux.org/index.php/intel_graphics/">Arch
     94        linux</ulink>.
     95      </para>
     96
     97      <para>
     98        Nvidia GPUs from Kepler onwards require signed firmware, otherwise the
     99        nouveau driver is unable to provide hardware acceleration. Nvidia has
     100        now released firmware up to Turing (most, maybe all, GTX16xx and RTX20xx
     101        GPUs) to linux-firmware, and kernels from linux-5.6 should support it,
     102        although Mesa support may require a development version until Mesa-20.2
     103        is released. Note that faster clocks than the default are not enabled
     104        by the released firmware.
    77105      </para>
    78106    </listitem>
     
    178206        The first step is to get the most recent version of the Intel
    179207        microcode.  This must be done by navigating to <ulink url=
    180          'https://github.com/intel/Intel-Linux-Processor-Microcode-Data-Files/releases/'/>
     208        'https://github.com/intel/Intel-Linux-Processor-Microcode-Data-Files/releases/'/>
    181209        and downloading the latest file there.  As of this writing the most
    182         recent version of the microcode is microcode-20200609.  Extract this
     210        secure version of the microcode, for those machines which can boot it,
     211        is microcode-20200609. If you have a Skylake machine, please read the
     212        Caution in the 'Early loading of microcode' section below.  Extract this
    183213        file in the normal way, the microcode is in the <filename>intel-ucode
    184214        </filename> directory, containing various blobs with names in the form
     
    383413      <caution>
    384414        <para>
    385           On some Skylake machines with hex Model Number '4e' (78 decimal) the upgrade to
    386           microcode version '0xdc' is reported to cause the machine to hang in early boot,
    387           and the fix is to revert to version 0xd6 which was shipped in the 20191115
    388           microcode release.
     415          On some Skylake machines with hex Model Number '4e' (78 decimal) the
     416          upgrade to microcode version '0xdc' is reported to cause the machine to
     417          hang in early boot, and the fix is to revert to version 0xd6 which was
     418          first shipped in the 20191115 microcode release.
    389419        </para>
    390420
    391421        <para>
    392           Doing that, while making the machine usable, will mean that the SRBDS
    393           mitigations are not present.
     422          At least one model '5e' Skylake does boot successfully with version
     423          0xdc, but Intel has now shipped a 20200616 release which is intended for
     424          distros which need an initrd that will boot on everyone's machine: it
     425          reverts both Skylake variants ('4e' and '5e') to the old 0xd6.
    394426        </para>
    395427
    396428        <para>
    397           The Skylake machines with hex Model Number '5e' (94 decimal) use similar
    398           firmware versions, but boot correctly with the mitigation applied.
     429          For a Skylake which does not boot with 0xdc, reverting to 0xd6 will make
     430          the machine usable, but without the SRBDS mitigations.
    399431        </para>
    400432      </caution>
Note: See TracChangeset for help on using the changeset viewer.