I had a simple task: Install and configure my mothers new laptop.
To my surprise she wanted to to try Linux, because Windows 11 again has a new UI and her reasoning was that if she needs to get used to something new it might as well be Linux - and I guess the fact that Win11 only got buggier over time has reached normies now.

Now this should be no problem.

Unfortunately that laptop, a Lenovo IdeaPad Slim 3 16ABR8, turned out to have firmware bugs related to suspend (s2idle).

I debugged the issues and eventually found ways to fix/work around them - this blogpost shows how.

The Problems

After suspend+resume (s2idle standby) the keyboard stopped working. Several sources suggested setting a kernel parameter (i8042.nopnp), which made the keyboard mostly work after resume, except the Fn keys/multimedia keys didn’t.
For example, pressing the F1/Mute key only registered as F1, even when pressed together with Fn, same for all other Fn combinations.

The lid switch also stopped working after the first suspend+resume, so when closing the laptop it didn’t go to sleep or at lock the screen or whatever was configured - here I didn’t find any kernel parameter that helped.

At least those problems were undone by rebooting… until the next suspend.

These problems seem to be common on several IdeaPads (and possibly other notebooks) with Zen3 or Zen3+ CPUs (“Barcelo” or “Rembrandt”)¹.

A minor issue was an ACPI error that happened on resume - though of course at the time I didn’t know whether it’s minor or the root of the problems.

ACPI: \_SB_.PEP_: Successfully transitioned to state lps0 exit
ACPI: \_SB_.PEP_: Successfully transitioned to state lps0 ms exit
ACPI BIOS Error (bug): Could not resolve symbol [\_SB.PC00.LPCB.EC0.SNTM],
	AE_NOT_FOUND (20251212/psargs-332)

No Local Variables are initialized for Method [_DSM]

Initialized Arguments for Method [_DSM]:  (4 arguments defined for method
	invocation)
  Arg0:   000000002ff511a6 <Obj>           Buffer(16) 56 0D E0 11 64 CE CE 47
  Arg1:   000000005176ebd5 <Obj>           Integer 0000000000000000
  Arg2:   00000000e877e9f8 <Obj>           Integer 0000000000000004
  Arg3:   0000000024f78b95 <Obj>           Package 0000000024f78b95

ACPI Error: Aborting method \_SB.PEP._DSM due to previous error (AE_NOT_FOUND)
	(20251212/psparse-529)
ACPI: \_SB_.PEP_: Failed to transitioned to state screen on

Solutions

Avoid breaking the keyboard and lid switch with suspend

It turned out that the laptops EC (embedded controller) needs a little me-time before suspend, or it becomes uncooperative after waking up² and doesn’t deliver keyboard and lid switch events anymore - this most probably is a firmware bug³.

I have a patch that adds a 2.5 second sleep right before suspend at the appropriate place in the amd_pmc kernel module, which will hopefully be merged in the Linux kernel.

Until it has been merged (and that kernel version ends up in your Linux distribution), you can use my out-of-tree version of the module that contains the patches and should work with Kernel 6.5 and newer. You can find it and installation instructions at https://github.com/DanielGibson/amd_pmc-ideapad, if it works for you I recommend installing it with DKMS so it gets automatically built and installed when your distro ships a kernel update.

If your device isn’t already detected automatically as affected by this issue, the updated amd_pmc kernel module has a parameter delay_suspend that allows explicitly enabling (or disabling) it; the aforementioned Github project has more information on this.

Possibly remaining problems (with wakealarm)

Unfortunately, this does not completely fix the issues on all devices.

Normal suspend and resume (with keyboard or lid switch or touchpad, …) works fine, but timer-base suspend+resume⁴ still has the old problems on some machines (like my 16ABR8), but apparently works fine on others (like 14ARP10).

I think this shouldn’t be a big problem in practice, but you should probably be aware of this.

At least on 16ABR8 you can still use the i8042.nopnp kernel commandline parameter⁵ to keep most of the keyboard functional after a timed suspend. That will allow you to unlock your screensaver after resume, save your open files and then reboot to restore full keyboard and lid switch functionality.

Workaround for the ACPI AE_NOT_FOUND problem

The aforementioned ACPI error (Could not resolve symbol [\_SB.PC00.LPCB.EC0.SNTM], AE_NOT_FOUND) turned out to be unrelated to the keyboard/lid problems and I haven’t found any other issues caused by it.

It’s still ugly and there’s a reasonably easy way to fix it which I’d like to describe, if only to allow others having similar issues to find it.

NOTE: While the keyboard issue seems to affect lots of IdeaPads with Zen3 (“Barcelo”) and Zen3+ (“Rembrandt”) CPUs, this ACPI error seems to be more specific to the IdeaPad Slim 3 16ABR8 and probably other *ABR8 devices⁶, at least with the current BIOS version (KYCN41WW).
Don’t blindly apply my fix, check first if you have this exact error (about \_SB.PC00.LPCB.EC0.SNTM) in dmesg after suspend!

Maye you have similar AE_NOT_FOUND errors for other symbols, in which case my fix may at least show a way to work around them, but you’ll have to adapt it.

So the complaint is that the method \_SB.PC00.LPCB.EC0.SNTM can’t be found, because it doesn’t exist. Most probably it should’ve been \_SB.PCI0.LPC0.EC0.SNTM, a method from dsdt.dsl that is otherwise unused. PCI0, LPC0 and EC0 are all (nested) devices.

Thankfully it’s possible to create ACPI fake devices (or ACPI dummy devices or ACPI stub devices⁷), they’re called “Module Device” there and at least when they do not contain a _CRS object or method they’re “simple container objects”⁸.

So one can just create fake devices PC00 and LPCB and EC0 and a SNTM method in them which calls the actual method.
Alternatively I guess the method could remain empty to do nothing, which would still make the ACPI error go away so the ACPI method calling the nonexisting method can finish.

It looks like this:

// create stub/fake/dummy devices and a stub method for the
// method missing in the BIOS ACPI tables (\_SB.PC00.LPCB.EC0.SNTM)
// and make it call the proper method (\_SB.PCI0.LPC0.EC0.SNTM)
DefinitionBlock ("", "SSDT", 2, "LENOVO", "SNTMFIX", 0x00000001)
{
	// the actually existing method from dsdt.dsl
	External (_SB.PCI0.LPC0.EC0.SNTM, MethodObj) // 0 Arguments

	Scope (\_SB)
	{
		Device (PC00)
		{
			// https://uefi.org/specs/ACPI/6.6/09_ACPI_Defined_Devices_and_Device_Specific_Objects.html#module-device
			// says that devices with HID "ACPI0004" are "module devices"
			// that act as simple containers if they contain no _CRS object
			Name (_HID, "ACPI0004" /* Module Device */)  // _HID: Hardware ID
			Device (LPCB)
			{
				Name (_HID, "ACPI0004" /* Module Device */)
				Device (EC0)
				{
					Name (_HID, "ACPI0004" /* Module Device */)
					Method (SNTM, 0, NotSerialized)
					{
						// call real method from dsdt.dsl
						\_SB.PCI0.LPC0.EC0.SNTM ()
					}
				}
			}
		}
	}
}

Save this as sntmfix.dsl and use iasl⁹ to compile it (iasl sntmfix.dsl).

If you’re using the grub bootloader, copy the resulting sntmfix.aml to /boot/ and tell grub to load it on boot by editing /etc/grub.d/40_custom (or another file that ends up in the generated /boot/grub/grub.cfg, or edit that file directly if you don’t generate it) and adding the following line:
acpi /boot/sntmfix.aml
(if /boot/ has its own partition on your system it should be just acpi /sntmfix.aml)

Afterwards run # update-grub or, if that doesn’t exist, # grub-mkconfig -o /boot/grub/grub.cfg.

If you’re not using grub, see the Arch Wiki for instructions (yes, they should also work on other distros).