Hi Alessandro, hi.

Alessandro Vesely via Dng - 09.06.25, 19:23:03 CEST:
[…]

I did not answer so far.

Cause I do not even know where to start.

Let me put it this way: On my main laptop I have even given up on it. I
shutdown the machine and boot it. Over the day I use standby.

My conclusion from various other laptops, all ThinkPads, is this:

Hibernation is broken.

It is a feature I cannot rely on anymore.

It works with a kernel and is broken with the next kernel. It may work
after some time with a newer kernel – or not. On one laptop a X260 I kept
an older kernel for that reason. And once I tried a newer one after quite
a while, it worked. But then after some other kernel upgrade it broke
again. So I keep the last known working kernel for now. I had the habit to
"apt download" the kernel Debian package so I have a copy of a known
working one.

On another, a T14 AMD Gen 2, I am lucky. There was a bug with some
6.12.something Debian Trixie kernel in Devuan Ceres but apparently it got
fixed.

On this T14 AMD Gen 5 no luck. Even after disabling Qualcomm WLAN and
Bluetooth drivers which are known not to work with hibernation at the
moment.

I feel there are at least nine reasons for the unreliability of that
feature I observed over the years:

1. Linux kernel is not a micro kernel. It cannot forcefully stop, restart
and reinitialize faulty drivers in a safe way. Ideally the Linux kernel
would have full authority over drivers, but that is not possible within a
monolithic kernel.

2. The development speed of the Linux kernel is insane. If you ask me: It
is way beyond ridiculous. That speed does not allow much time for
regression fixing.

3. The Linux kernel has become much, much, much more complex.

4. Hardware has become more and more complex.

5. There are much more different hardware devices around than 20 years
ago.

6. Firmware is usually closed source. If it has bugs, bad luck. Unless the
hardware manufacturer or firmware provider actually cares. But even then
it may take months for a fix.

7. Sometimes fixing something for Linux breaks Windows. Windows has
priority over fixes for Linux. So bad luck again.

8. In my point of view the Linux implementation of hibernate is subpar. I
fondly remember the times of Software Suspend 2 aka TuxOnIce by Nigel
Cunningham¹. These were the times where I could type "who" inside a KDE
session in Juni and see a date from April. I.e. use the same KDE sessions
for months.

9. Kernel developers usually ask for a bisect. But doing one often
requires hours or even half a day or more to get to the last working
commit and determine the commit that breaks it. How many users are
actually willing to put up with that?

I have been told Lenovo does not support hibernation with Linux on devices
they do provide Linux support for. I meanwhile understand: For a reason.

But anyway… best of luck with finding a way to hibernate again. Using an
older kernel may be the best option. And then eventually test again after
a while with the then current one.

And if you have a working setup: Do not change it.

But on the other hand people want to use up to date software.

I do have some hope: Currently there are developers working on making
hibernation more reliable, including someone from AMD.

But still I think it will be challenging for the reasons I outlined above.

For my main private laptop, that T14 AMD Gen 5, my current approach is:
Shutdown and reboot, standby over the day. And for my music laptop, that
X260, it is: Use that old kernel I know works okay.

I had it that some kernels could not even power off a machine anymore. I
saw this with a T14 AMD Gen 1 and that X260. It is ridiculous. It was a
regular shutdown and the machine would not power off. Are you even kidding
me?

It is a sad story, unfortunately.


[1] https://gitlab.com/nigelcunningham/tuxonice-kernel

The original upstream website appears to be gone.

Best,
--
Martin - please no carbon copy to me