Re: [PATCH] x86: Unbreak early processor microcode loading

[Daniel J Blueman <daniel@numascale.com>]

On Wed, Mar 4, 2015 at 7:45 PM, Borislav Petkov <bp@alien8.de> wrote:
> On Wed, Mar 04, 2015 at 10:18:47AM +0100, Borislav Petkov wrote:
>>  Let me try to reproduce that here.
> 
> Well, it works fine on a single-socket box here:
> 
> [    1.045426] microcode: updated early to new patch_level=0x010000dc
> [    1.060957] microcode: CPU0: patch_level=0x010000dc
> [    1.061143] microcode: CPU1: patch_level=0x010000dc
> [    1.061310] microcode: CPU2: patch_level=0x010000dc
> [    1.061495] microcode: CPU3: patch_level=0x010000dc
> [    1.061677] microcode: CPU4: patch_level=0x010000dc
> [    1.061863] microcode: CPU5: patch_level=0x010000dc
> 
> That's early patching.
> 
> $ grep -E "(FIRMWARE|MICROCODE)" .config
> CONFIG_MICROCODE=y
> CONFIG_MICROCODE_INTEL=y
> CONFIG_MICROCODE_AMD=y
> CONFIG_MICROCODE_OLD_INTERFACE=y
> CONFIG_MICROCODE_INTEL_EARLY=y
> CONFIG_MICROCODE_AMD_EARLY=y
> CONFIG_MICROCODE_EARLY=y
> CONFIG_PREVENT_FIRMWARE_BUILD=y
> CONFIG_FIRMWARE_IN_KERNEL=y
> CONFIG_EXTRA_FIRMWARE="microcode_amd.bin microcode_amd_fam15h.bin 
> microcode_amd_fam16h.bin"
> CONFIG_EXTRA_FIRMWARE_DIR="/lib/firmware/amd-ucode"
> # CONFIG_DRM_LOAD_EDID_FIRMWARE is not set
> CONFIG_FIRMWARE_EDID=y
> CONFIG_FIRMWARE_MEMMAP=y
> # CONFIG_GOOGLE_FIRMWARE is not set
> # CONFIG_TEST_FIRMWARE is not set

Thanks for checking Boris!

I can't reproduce your early-loading case with this same config; I 
suspect your udev is loading the firmware which is available in the 
initramfs, or it's being loaded later from the rootfs non-early.

In my case, the microcode is in neither, hence we build it into the 
kernel image.

The full console log, config and bzImage are at 
https://resources.numascale.com/telemetry/

I find the changes I posted earlier fix this. Maybe I'm missing 
something still?

Daniel