Skip to content

# 06-5e-03 (i7-6700HQ) rev 0xf0: hang when secondary CPUs start after early load from 0x88, at boot and on S3 resume (Clevo P640RE) #114

Open

Description

@imaginitivename

Summary

On a Clevo P640RE (i7-6700HQ, CPUID 0x506e3), Linux hangs with a black screen when microcode is loaded early from the BIOS revision 0x88 to 0xf0 and the kernel then starts the secondary CPUs. The same hang occurs on resume from S3. With dis_ucode_ldr the machine boots and all 8 CPUs start normally on 0x88.

This looks like the 06-5e-03 hang reported against revision 0xdc in 2020, which was described as resolved in 0xea. On this machine it still occurs with 0xf0.

System

CPU Intel Core i7-6700HQ, family 6, model 94 (0x5e), stepping 3, CPUID 0x506e3
Platform ID flags (processor_flags) 0x20
Machine Clevo P640RE (sold as Sager P640RE / Metabox P640RE), HM170 chipset
BIOS American Megatrends 1.05.08, dated 04/08/2016, UEFI boot
Microcode in BIOS 0x88
Microcode loaded by OS 0xf0 (Arch intel-ucode 20261110-1)
Kernels tested 7.3.3, 7.3.4 and 6.18.60 LTS (CachyOS builds), same result on all
x2APIC disabled (BIOS sets the x2APIC opt-out bit)

Results

Each line is one boot. Only the listed parameter was changed.

Kernel parameters Microcode Result
(default) early load 0x88 -> 0xf0 Hang, black screen
dis_ucode_ldr stays 0x88 Boots, 8 CPUs online
maxcpus=1 early load 0x88 -> 0xf0 on CPU0 Boots with 1 CPU
maxcpus=2 early load Hang
nosmt early load Hang
cpuhp.parallel=0 early load Hang
idle=poll early load Hang
intel_idle.max_cstate=1 early load Hang
intremap=off early load Hang
pcie_port_pm=off early load Hang

earlycon=efifb keep_bootcon ignore_loglevel produced no output before the hang, and nothing reaches the journal, so I do not have a log of the exact point of failure.

Starting the secondary CPUs later works

Booting with maxcpus=1 and then onlining the other CPUs from userspace about 15 seconds into boot works every time:

smp: Bringing up secondary CPUs ...
smp: Brought up 1 node, 1 CPU
microcode: Current revision: 0x000000f0
microcode: Updated early from: 0x00000088
...
[   17.978319] smpboot: Booting Node 0 Processor 1 APIC 0x2
[   17.989855] smpboot: Booting Node 0 Processor 2 APIC 0x4
[   17.997450] smpboot: Booting Node 0 Processor 3 APIC 0x6
[   18.002009] smpboot: Booting Node 0 Processor 4 APIC 0x1
[   18.008064] smpboot: Booting Node 0 Processor 5 APIC 0x3
[   18.012018] smpboot: Booting Node 0 Processor 6 APIC 0x5
[   18.014152] smpboot: Booting Node 0 Processor 7 APIC 0x7

Afterwards /proc/cpuinfo shows revision 0xf0 on all 8 logical CPUs and the system is stable.

So the update itself applies on every CPU. The hang only occurs when the secondary CPUs are started early, straight after CPU0 has been updated.

S3 resume shows the same pattern

  • Resume from S3 with CPUs 1-7 online at suspend time: hang. The kernel restarts them during early resume, after the firmware has reset them.
  • Resume from S3 with CPUs 1-7 offlined before suspend and onlined from userspace after resume: works.
  • pm_test levels processors and core (CPUs offlined and re-onlined without a real firmware S3 transition): pass.
  • s2idle: works.

Current workaround

maxcpus=1 plus a service that onlines CPUs 1-7 after boot, and a sleep hook that offlines them before S3 and onlines them after resume. This keeps all CPUs on 0xf0.

Not tested

  • Intermediate revisions (for example 0xd6, 0xdc, 0xea), so I cannot say which revision first shows the hang on this machine.
  • A vanilla mainline kernel. The kernels above are CachyOS builds; the proprietary NVIDIA module is installed but is not loaded at the point of the hang.
  • A newer BIOS. 1.05.08 is the version installed; I have not found a later one for this model.

Activity

  1. klynastor commented on Nov 14, 2026

    @klynastor

    I think this issue slightly predates what was reported in 2020 (#31). I believe the fix is that the motherboard BIOS must have a microcode revision of likely 0xc0 or newer installed, for Linux to successfully apply any early microcode update at or above that revision. I have a system that successfully boots from 0xc2 this way (it's a 06-4e-03, but it uses a revision scheme similar to 06-5e-03):

    microcode: Current revision: 0x000000f0
    microcode: Updated early from: 0x000000c2
    

    My system vendor was kind enough to supply a new BIOS for me when I asked, but that was many years ago. Unfortunately, I can't find a newer BIOS version for your laptop.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions