Revision 0x00000122 for 06-b7-01 (i9-14900HX) breaks initialisation of mei_me

[wmkoolen]

Dear all,

I'm seeing a difference before and after updating the microcode for 06-b7-01. I don't quite know what it means, but I do see a multi-second stall on boot when the MEI initialisation fails. This happens even on the same kernel with and without the dis_ucode_ldr kernel option.

Old behaviour:
[ 0.674435] microcode: Current revision: 0x0000011f
[ 4.934601] mei_me 0000:00:16.0: enabling device (0000 -> 0002)

New behaviour:
[ 0.712298] microcode: Current revision: 0x00000122
[ 0.712375] microcode: Updated early from: 0x0000011f
[ 0.915088] mei_me 0000:00:16.0: enabling device (0000 -> 0002)
[ 0.915704] mei_me 0000:00:14.0: H_RST is not set = 0x1A000840
[ 2.943674] mei_me 0000:00:14.0: wait hw ready failed
[ 2.943822] mei_me 0000:00:14.0: hw_start failed ret = -62 fw status = 803401FD 00000000 00006032 00000000 00000000 00000000
[ 2.944064] mei_me 0000:00:14.0: H_RST is not set = 0x1A000840
[ 4.959607] mei_me 0000:00:14.0: wait hw ready failed
[ 4.959746] mei_me 0000:00:14.0: hw_start failed ret = -62 fw status = 803401FD 00000000 00006032 00000000 00000000 00000000
[ 4.959988] mei_me 0000:00:14.0: H_RST is not set = 0x1A000840
[ 6.975616] mei_me 0000:00:14.0: wait hw ready failed
[ 6.975769] mei_me 0000:00:14.0: hw_start failed ret = -62 fw status = 803401FD 00000000 00006032 00000000 00000000 00000000
[ 6.981211] mei_me 0000:00:14.0: reset: reached maximal consecutive resets: disabling the device
[ 6.986604] mei_me 0000:00:14.0: reset failed ret = -19
[ 6.991920] mei_me 0000:00:14.0: link layer initialization failed.
[ 6.997228] mei_me 0000:00:14.0: init hw failure.
[ 7.002843] mei_me 0000:00:14.0: initialization failed.

[wmkoolen]

Dear all,

In searching for related thins online, I found this dmesg log: https://intel-gfx-ci.01.org/tree/drm-tip/CI_DRM_14187/bat-rpls-2/boot.html That log is also from a system with intel 06-b7-01 cpu. Theirs is an engineering sample cpu that boots with microcode revision 0x00000122. Mine instead is being updated to it from 0x0000011f. The log shows that their system does not suffer from the mei_me initialisation failure. Curious.