Tiger Lake laptop fails to resume from S3 suspend (black screen, keyboard dead, requires hard reboot)

tahaspc

tahaspc

Hi everyone,

I'm running into what appears to be a suspend/resume issue on my Tiger Lake laptop. I've spent quite a while debugging it, but I'm running out of ideas.

System
  • FreeBSD 15.1-RELEASE-p3 amd64
  • XFCE + LightDM
  • Intel Core i5 (11th Gen, Tiger Lake)
  • Intel Iris Xe graphics
  • drm-66-kmod
  • gpu-firmware-kmod installed
  • Insyde BIOS (reported as INSYDE TGL-ULT)

Symptoms

Whenever the machine enters S3 sleep (either using
Code: 
acpiconf -s 3
or by closing the lid), it never resumes correctly.

The power LED indicates that the machine wakes up, but:

  • The internal display remains completely off (not even the backlight turns on)
  • The keyboard is completely dead
  • Caps Lock LED never toggles
  • No TTY appears
  • No graphical session returns
  • Only holding the power button performs a shutdown

This happens even if X isn't running.

What I tested

  • Suspend from XFCE
  • Suspend from a text console
  • Suspend via lid close
  • Suspend via
    Code: 
    acpiconf -s 3

All produce the exact same result.

Interesting observations

The lid switch itself works correctly.

Polling

Code: 
sysctl dev.acpi_lid.0.state

shows

Code: 
1 -> 0 -> 1

when opening and closing the lid.

The ACPI lid device exists:

Code: 
acpi_lid0 <Control Method Lid Switch>

and

Code: 
hw.acpi.lid_switch_state

changes appropriately.

Supported sleep states:

Code: 
hw.acpi.supported_sleep_state:
S3 S4 S5

Current suspend state:

Code: 
hw.acpi.suspend_state:
S3

Graphics

Using drm-kmod 6.6:

Code: 
drm-66-kmod-6.6.25.1501000_8

Firmware loads correctly:

Code: 
[drm] Finished loading DMC firmware i915/tgl_dmc_ver2_12.bin

No obvious GPU hangs appear in dmesg.

Loader.conf

I've already tried disabling Intel power-saving features:

Code: 
hw.drm.i915.enable_dc="0"
hw.drm.i915.enable_fbc="0"
hw.drm.i915.enable_psr="0"

None changed the behavior.

Other hardware

Touchpad uses:

Code: 
iichid0: Using sampling mode

Keyboard is still detected as:

Code: 
atkbd0

so the keyboard should not depend on iichid.

dmesg

The only recurring errors I see are HD Audio timeouts:

Code: 
hdac0: Command ... timeout on address 2

which appear unrelated.

No GPU reset or i915 hang messages appear.

Power manager

XFCE power manager is running.

Lid actions are configured, but even manually calling

Code: 
acpiconf -s 3

outside of X exhibits the same failure.

Conclusion

Since the issue occurs:

  • outside X
  • outside XFCE
  • using acpiconf directly
  • with a dead keyboard after resume

I'm starting to suspect either:

  • an ACPI firmware issue (Insyde BIOS)
  • a Tiger Lake suspend/resume regression
  • or an interaction between Intel Tiger Lake and drm-kmod on resume.

Additional related issue

I also experience a seemingly related problem even when the machine does not enter suspend.

If I configure XFCE's power manager so that closing the lid does one of the following:

  • Do nothing
  • Turn off display
  • Lock screen

and then leave the laptop closed for around 10 minutes, opening the lid often results in:

  • An active, but completely unresponsive screen
  • The cursor may still be visible and working perfectly
  • The desktop never redraws
  • Keyboard input appears to stop working
  • The only recovery is a hard reboot

This happens despite the machine never actually entering S3 sleep.

Conclusion

There appear to be two possibly related issues:

  • S3 suspend never resumes correctly.
  • The system can also become unusable after the display has been powered off for an extended period without entering suspend.

Since the problem occurs:

  • outside X
  • outside XFCE
  • using acpiconf directly
  • with a dead keyboard after resume
  • and also during display wake without suspend

I'm beginning to suspect a firmware, ACPI, or Intel Tiger Lake graphics/power-management issue rather than a desktop-environment problem.

Because of this, I'm wondering whether the suspend issue and this display wake issue are symptoms of the same underlying problem.

Has anyone with Tiger Lake hardware seen similar behavior?

Are there any known loader tunables, ACPI quirks, or debugging methods that could help narrow this down?

Thanks!
 
tahaspc I think tiger lake does not support S3 only S0ix (i.e. modern sleep). The BIOS may report that it does (ACPI table) but that does not make it true. Can you check on Linux to see if it's actually enabled there or not (tip: if you put it to sleep at 100% and after half a day it's mostly down to fumes, it's not S3).
 
Can anyone explain to me why suspending your computer is so critical?

I've said this somewhere else:

1) Set it up so closing the lid keeps it running.
2) If you are going to the bathroom in Starbucks and you take your laptop with you because you are a careful person in any big city (except the Netherlands, which is the last safe country on Earth): lock the session, close the lid, put the laptop into your backpack, go to the bathroom, come back, keep working.
3) If you are going home: turn the thing off. FreeBSD boots very fast.
 
AlfredoLlaquet said:
Can anyone explain to me why suspending your computer is so critical?

I've said this somewhere else:

1) Set it up so closing the lid keeps it running.
2) If you are going to the bathroom in Starbucks and you take your laptop with you because you are a careful person in any big city (except the Netherlands, which is the last safe country on Earth): lock the session, close the lid, put the laptop into your backpack, go to the bathroom, come back, keep working.
3) If you are going home: turn the thing off. FreeBSD boots very fast.
Click to expand...
for me freebsd takes 30 secs to a minute though
 
tahaspc said:
Hi everyone,

I'm running into what appears to be a suspend/resume issue on my Tiger Lake laptop. I've spent quite a while debugging it, but I'm running out of ideas.

System
  • FreeBSD 15.1-RELEASE-p3 amd64
  • XFCE + LightDM
  • Intel Core i5 (11th Gen, Tiger Lake)
  • Intel Iris Xe graphics
  • drm-66-kmod
  • gpu-firmware-kmod installed
  • Insyde BIOS (reported as INSYDE TGL-ULT)

Symptoms

Whenever the machine enters S3 sleep (either using
Code: 
acpiconf -s 3
or by closing the lid), it never resumes correctly.

The power LED indicates that the machine wakes up, but:

  • The internal display remains completely off (not even the backlight turns on)
  • The keyboard is completely dead
  • Caps Lock LED never toggles
  • No TTY appears
  • No graphical session returns
  • Only holding the power button performs a shutdown

This happens even if X isn't running.

What I tested

  • Suspend from XFCE
  • Suspend from a text console
  • Suspend via lid close
  • Suspend via
    Code: 
    acpiconf -s 3

All produce the exact same result.

Interesting observations

The lid switch itself works correctly.

Polling

Code: 
sysctl dev.acpi_lid.0.state

shows

Code: 
1 -> 0 -> 1

when opening and closing the lid.

The ACPI lid device exists:

Code: 
acpi_lid0 <Control Method Lid Switch>

and

Code: 
hw.acpi.lid_switch_state

changes appropriately.

Supported sleep states:

Code: 
hw.acpi.supported_sleep_state:
S3 S4 S5

Current suspend state:

Code: 
hw.acpi.suspend_state:
S3

Graphics

Using drm-kmod 6.6:

Code: 
drm-66-kmod-6.6.25.1501000_8

Firmware loads correctly:

Code: 
[drm] Finished loading DMC firmware i915/tgl_dmc_ver2_12.bin

No obvious GPU hangs appear in dmesg.

Loader.conf

I've already tried disabling Intel power-saving features:

Code: 
hw.drm.i915.enable_dc="0"
hw.drm.i915.enable_fbc="0"
hw.drm.i915.enable_psr="0"

None changed the behavior.

Other hardware

Touchpad uses:

Code: 
iichid0: Using sampling mode

Keyboard is still detected as:

Code: 
atkbd0

so the keyboard should not depend on iichid.

dmesg

The only recurring errors I see are HD Audio timeouts:

Code: 
hdac0: Command ... timeout on address 2

which appear unrelated.

No GPU reset or i915 hang messages appear.

Power manager

XFCE power manager is running.

Lid actions are configured, but even manually calling

Code: 
acpiconf -s 3

outside of X exhibits the same failure.

Conclusion

Since the issue occurs:

  • outside X
  • outside XFCE
  • using acpiconf directly
  • with a dead keyboard after resume

I'm starting to suspect either:

  • an ACPI firmware issue (Insyde BIOS)
  • a Tiger Lake suspend/resume regression
  • or an interaction between Intel Tiger Lake and drm-kmod on resume.

Additional related issue

I also experience a seemingly related problem even when the machine does not enter suspend.

If I configure XFCE's power manager so that closing the lid does one of the following:

  • Do nothing
  • Turn off display
  • Lock screen

and then leave the laptop closed for around 10 minutes, opening the lid often results in:

  • An active, but completely unresponsive screen
  • The cursor may still be visible and working perfectly
  • The desktop never redraws
  • Keyboard input appears to stop working
  • The only recovery is a hard reboot

This happens despite the machine never actually entering S3 sleep.

Conclusion

There appear to be two possibly related issues:

  • S3 suspend never resumes correctly.
  • The system can also become unusable after the display has been powered off for an extended period without entering suspend.

Since the problem occurs:

  • outside X
  • outside XFCE
  • using acpiconf directly
  • with a dead keyboard after resume
  • and also during display wake without suspend

I'm beginning to suspect a firmware, ACPI, or Intel Tiger Lake graphics/power-management issue rather than a desktop-environment problem.

Because of this, I'm wondering whether the suspend issue and this display wake issue are symptoms of the same underlying problem.

Has anyone with Tiger Lake hardware seen similar behavior?

Are there any known loader tunables, ACPI quirks, or debugging methods that could help narrow this down?

Thanks!
Click to expand...
I am using a comet lake Thinkpad on 16-CURRENT. I had this issue for about a week-+. On the latest update, FreeBSD 16.0-CURRENT #0 main-0380d016d9ba: Thu Sep 17 07:16:27 +08 2026, sleep n resume work. It has been working for the past 2/3 days.
 
rshad muhammad said:
I am using a comet lake Thinkpad on 16-CURRENT. I had this issue for about a week-+. On the latest update, FreeBSD 16.0-CURRENT #0 main-0380d016d9ba: Thu Sep 17 07:16:27 +08 2026, sleep n resume work. It has been working for the past 2/3 days.
Click to expand...
Wait, version 16?! Thought the latest is 15.1
 
Which laptop model is this? Is there something in the BIOS related to sleep states . I had to change that setting for the thinkpad.
On 15.1 try with the 6.1 version ( drm-61-kmod-6.1.128.1501000_8 and gpu-firmware-intel-kmod-tigerlake-20260519.1501000 ) . It's been super stable on my T14 Gen2.
Comment out all the #hw.i915kms stuff in your loader.conf before reboot.
And then do the testing scenarios... ( tty login + startx and then zzz , ly/sddm etc...)
 
Suspend works for me with
hw.acpi.power_button_state=S3
hw.acpi.lid_switch_state=S3

FreeBSD-14.4
Chip i3-1115G4 (that would be tigerlake)

It did not work right from the start, and IIRC there was an issue with the TPM, which had to be disabled in BIOS. But this seems fixed for quite a while now, and sadly I do not find any notes about what I did.

And yes, suspend is critical. Not because of the startup time, but before cutting power one needs to contemplate about a place where to save all the open things.
 
You must log in or register to reply here.
Back
Top