[GIT PULL] x86/microcode for v6.6-rc1

[GIT PULL] x86/microcode for v6.6-rc1

[Borislav Petkov]

Hi Linus,

please pull the first part of Thomas' microcode loader work, for 6.6.
The other part wasn't fully ready in time so it will follow on later.

Thx.

---

The following changes since commit 6eaae198076080886b9e7d57f4ae06fa782f90ef:

  Linux 6.5-rc3 (2023-07-23 15:24:10 -0700)

are available in the Git repository at:

  git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip.git tags/x86_microcode_for_v6.6_rc1

for you to fetch changes up to 4d2b748305e96fb76202a0d1072a285b1500bff3:

  x86/microcode: Remove remaining references to CONFIG_MICROCODE_AMD (2023-08-26 13:37:55 +0200)

----------------------------------------------------------------
- The first, cleanup part of the microcode loader reorg tglx has been
  working on. This part makes the loader core code as it is practically
  enabled on pretty much every baremetal machine so there's no need to
  have the Kconfig items. In addition, there are cleanups which prepare
  for future feature enablement.

----------------------------------------------------------------
Ashok Raj (2):
      x86/microcode: Include vendor headers into microcode.h
      x86/microcode/intel: Rename get_datasize() since its used externally

Borislav Petkov (AMD) (1):
      x86/microcode/AMD: Rip out static buffers

Lukas Bulwahn (1):
      x86/microcode: Remove remaining references to CONFIG_MICROCODE_AMD

Sebastian Andrzej Siewior (1):
      x86/microcode: Remove microcode_mutex

Thomas Gleixner (7):
      x86/mm: Remove unused microcode.h include
      x86/microcode: Hide the config knob
      x86/microcode/intel: Move microcode functions out of cpu/intel.c
      x86/microcode: Make reload_early_microcode() static
      x86/microcode: Move core specific defines to local header
      x86/microcode/intel: Remove debug code
      x86/microcode/intel: Remove pointless mutex

 arch/x86/Kconfig                         |  38 +---
 arch/x86/configs/i386_defconfig          |   1 -
 arch/x86/configs/x86_64_defconfig        |   1 -
 arch/x86/include/asm/microcode.h         | 154 +++++-----------
 arch/x86/include/asm/microcode_amd.h     |  58 ------
 arch/x86/include/asm/microcode_intel.h   |  88 ---------
 arch/x86/kernel/cpu/common.c             |   4 +-
 arch/x86/kernel/cpu/intel.c              | 176 +-----------------
 arch/x86/kernel/cpu/microcode/Makefile   |   4 +-
 arch/x86/kernel/cpu/microcode/amd.c      | 133 +++++++-------
 arch/x86/kernel/cpu/microcode/core.c     |  17 +-
 arch/x86/kernel/cpu/microcode/intel.c    | 304 +++++++++++++++++++++----------
 arch/x86/kernel/cpu/microcode/internal.h | 131 +++++++++++++
 arch/x86/mm/init.c                       |   1 -
 drivers/platform/x86/intel/ifs/load.c    |   7 +-
 15 files changed, 466 insertions(+), 651 deletions(-)
 delete mode 100644 arch/x86/include/asm/microcode_amd.h
 delete mode 100644 arch/x86/include/asm/microcode_intel.h
 create mode 100644 arch/x86/kernel/cpu/microcode/internal.h


-- 
Regards/Gruss,
    Boris.

https://people.kernel.org/tglx/notes-about-netiquette

Re: [GIT PULL] x86/microcode for v6.6-rc1

[Linus Torvalds]

On Mon, 28 Aug 2023 at 05:25, Borislav Petkov <bp@alien8.de> wrote:
>
> please pull the first part of Thomas' microcode loader work, for 6.6.

Bah.

The Zenbleed fix conflicted with this because of how it declared the
amd_check_microcode() function.

The natural way to do it would have been in the new
microcode/internal.h file, but that isn't included by the caller, and
while I could have just done something like

    #include "microcode/internal.h"

in the common.c file, that seemed to be against the spirit here.

So I hacked something up in <asm/microcode.h> instead.

That is probably not what either you or Thomas wanted, but the Intel
code has some similar hacks in that header, so it's not _entirely_ out
of line.

Please just let me know how much I messed up.

I apologize in advance.

                 Linus

Re: [GIT PULL] x86/microcode for v6.6-rc1

[Linus Torvalds]

On Mon, 28 Aug 2023 at 16:09, Linus Torvalds
<torvalds@linux-foundation.org> wrote:
>
> Please just let me know how much I messed up.
>
> I apologize in advance.

Ahh, I see that Ingo put it in <asm/processor.h> in the linux-next
tree. I'm not convinced it's any better than my <asm/microcode.h>
choice, but it does result in not needing another

  #ifdef CONFIG_CPU_SUP_AMD
 ..

thing, so...

Feel free to send me a follow-up patch to move that
amd_check_microcode() declaration to whatever the appropriate place
is.

           Linus

Re: [GIT PULL] x86/microcode for v6.6-rc1

[Linus Torvalds]

On Mon, 28 Aug 2023 at 16:25, Linus Torvalds
<torvalds@linux-foundation.org> wrote:
>
> Feel free to send me a follow-up patch to move that
> amd_check_microcode() declaration to whatever the appropriate place
> is.

.. actually, I have now gotten around to Ingo's "Misc x86 cleanups"
tree, and there I find the declaration movement to <asm/processor.h>
by Arnd.

So I ended up just making my tree match at that point.

                Linus

Re: [GIT PULL] x86/microcode for v6.6-rc1

[Borislav Petkov]

On Mon, Aug 28, 2023 at 05:11:09PM -0700, Linus Torvalds wrote:
> .. actually, I have now gotten around to Ingo's "Misc x86 cleanups"
> tree, and there I find the declaration movement to <asm/processor.h>
> by Arnd.
> 
> So I ended up just making my tree match at that point.

Yap, looks good, thanks.

The logic was to have the declarations in the processor.h header as the
definition is in the processor-specific file cpu/amd.c.

-- 
Regards/Gruss,
    Boris.

https://people.kernel.org/tglx/notes-about-netiquette

Re: [GIT PULL] x86/microcode for v6.6-rc1

[pr-tracker-bot]

The pull request you sent on Mon, 28 Aug 2023 14:25:33 +0200:

> git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip.git tags/x86_microcode_for_v6.6_rc1

has been merged into torvalds/linux.git:
https://git.kernel.org/torvalds/c/42a7f6e3ffe06308c1ec43a7dac39a27de101574

Thank you!

-- 
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/prtracker.html

Re: [GIT PULL] x86/microcode for v6.6-rc1

[Ed Spiridonov]

From the commit message:
> This part makes the loader core code as it is practically enabled on pretty much every baremetal machine

Hmm... I guess there are a lot more virtualized Linux instances than
baremetal ones, why should the microcode boot code be mandatory?