an acpi menace to linux in some uefi-ed computers

[TJ Hoye]

demo@mx1:~
$ dmx | grep --count ACPI
32
demo@mx1:~
$ acpi -V
Battery 0: Full, 100%
Battery 0: design capacity 3684 mAh, last full capacity 3500 mAh = 95%
Adapter 0: on-line
Thermal 0: ok, 25.0 degrees C
Thermal 0: trip point 0 switches to mode critical at temperature 107.0 degrees C
Cooling 0: Processor 0 of 3
Cooling 1: Processor 0 of 3
Cooling 2: iwlwifi no state information available
Cooling 3: Processor 0 of 3
Cooling 4: Processor 0 of 3
Cooling 5: INT3400 Thermal no state information available
Cooling 6: Processor 0 of 3
Cooling 7: SEN1 no state information available
Cooling 8: Processor 0 of 3
Cooling 9: x86_pkg_temp no state information available
Cooling 10: Processor 0 of 3
Cooling 11: intel_powerclamp no state information available
Cooling 12: pch_skylake no state information available
Cooling 13: Processor 0 of 3
Cooling 14: TMEM no state information available
demo@mx1:~
$ lsmod | grep -i acpi
snd_soc_acpi_intel_match 28672 1 snd_soc_skl
snd_soc_acpi 16384 2 snd_soc_acpi_intel_match,snd_soc_skl
acpi_thermal_rel 16384 1 int3400_thermal
acpi_pad 20480 0
demo@mx1:~

Here's a few more clues:

1. I had a dream that a later kernel might know how to talk to uefi better.
ND. Still getting 32 dmesg gripes. Remastered, brought in 5.0.1 amd64 antiX kernel,
and used MX-Tools to do LiveUSB Kernel Install. Except(!) for acpi, 5.0.1 seems happy on Inspiron 5570 uefi boot.

2. Using acpi -V, some acpi functions are actually working; they were working for kernels 4.19 and 4.20 as well.

3. Grepping the modules loaded, four of them seem appropriate, just a minority of those required? Was same
for kernels 4.19 and 4.20.

4. I have the suspicion that there might be a missing cache of names which include catchwords like:
Thermal, Battery, Cooling, Adapter, possibly eight in all, that uefi wants to communicate-with concerning ACPI.
In this phantasy, it would be necessary to provide all of these names and the parameters they in turn require,
for this Inspiron 5570 to perform the necessary ACPI functionality. There a some tantalizing clues in acpi -V.

TJH

[Stevo]

If the laptop works better even with the acpi warnings at boot, why turn acpi off?

[TJ Hoye]

If the laptop works better even with the acpi warnings at boot, why turn acpi off?

Hello, Stevo.

It 'works' well enough either way, except that acpi features chores are short-changed.
I think uefi is telling us the linux kernels I've tried need to tell us more about this fabulous 5570 computer.
I wish I knew someone that could look at the 5.0 kernel source and see if there are some acpi hooks
that might be used, that normally wouldn't be of any use on a lot of 'older' computers.
Not being a programmer, that's something I could not probably pull off successfully myself.

TJH

[j2mcgreg]

Have you tried the Liquorix 5 series kernel that is available via MXPI? I installed it into my test machine and then made a snapshot which I used to successfully install MX18.2 on my troubled Acer Aspire 3 a315. I seem to remember some reference here that the newer Dells use the same Insydeh20 Set-up Utility as the Acer does.

[TJ Hoye]

Have you tried the Liquorix 5 series kernel that is available via MXPI? I installed it into my test machine and then made a snapshot which I used to successfully install MX18.2 on my troubled Acer Aspire 3 a315. I seem to remember some reference here that the newer Dells use the same Insydeh20 Set-up Utility as the Acer does.

Hello, j2ncgreg, and thanks for your question.

No, I have not tried a Liquorix 5 kernel. I've only added a repo to my Synaptic and brought in one of antiX's kernels.
It's just a hunch that antiX couldn't care less about acpi, since his genre seems to be more comfortable with 'older' hardware.
I'm probably not qualified myself to do much more than complain, but how might I go about getting some other kernel than
what I'm immediately able to find in MX and antiX territory?

Also, what do you get for dmesg comments re acpi, and what do you get for acpi -V? n.b. that's cap V.

TJH

[Stevo]

The latest Liquorix kernel is dead easy to install in the MX Package installer...

Also, if you do web searches for the ACPI warnings you're getting along with "Dell", I'm sure you will find other posts from concerned Dell owners in other distros' forums that should be informative.

[TJ Hoye]

The latest Liquorix kernel is dead easy to install in the MX Package installer...

Also, if you do web searches for the ACPI warnings you're getting along with "Dell", I'm sure you will find other posts from concerned Dell owners in other distros' forums that should be informative.

Hello again, Stevo.

I tried using MX-Tools to install a Liquorix kernel.
This process failed, but returned me to a significantly different situation with the antiX 5.0 kernel.
1. There is now only one dmesg entry about ACPI instead of 32.
2. There are two less modules loaded which address ACPI.
3. acpi -V still says it isnt getting the stuff it needs to set up ACPI.

I'm using a LiveUSB, and I think using MX-Tool pkg installer is not meant to work for LIveUSB kernel installs.
That's why MX-Tools has a LiveUSB kernel installer. That installer uses Synaptic for its kernel material,
and I don't know where to point it to get Liquorix material.

So, without my participation, my antiX kernel is a lot quieter, but still not helping on the ACPI front.

Evidence is as follows, for comparison with previous similar commands.
BitJam can probably explain why the MX package installer fails for LiveUSB kernels.

demo@mx1:~
$ acpi -V
No support for device type: power_supply
No support for device type: power_supply
Cooling 0: intel_powerclamp no state information available
Cooling 1: x86_pkg_temp no state information available
Cooling 2: pch_skylake no state information available
Cooling 3: iwlwifi no state information available

demo@mx1:~
$ alias dmx; dmx | grep -c ACPI
alias dmx='dmesg | grep -i --color " bug\|warn\|fail\|error\|corrupt"'
1

demo@mx1:~
$ dmx | grep -i acpi
[ 1.736482] ACPI: <n/a>: failed to evaluate _DSM (0x1001)

demo@mx1:~
$ lsmod | grep -i acpi
snd_soc_acpi_intel_match 28672 1 snd_soc_skl
snd_soc_acpi 16384 2 snd_soc_acpi_intel_match,snd_soc_skl
demo@mx1:~

TJH

[TJ Hoye]

I have tentatively concluded the following about acpi in MX-Linux.

1. Acpi is an enormous black hole to avoid as much as we can.
A more experienced person than I has expressed his feeling this way:
https://www.markshuttleworth.com/archives/1332
2. You can't ignore it entirely without hurting yourself; for example,
acpi=off is not a good policy because it may disrupt other things.
3. All those features that help your battery last longer don't matter
much if you always power up with an ac adapter.
4. Acpi features which tell you about the thermal situation, and other
states of your computer may be available to you at the command line.
Try out acpi -V sometime at the command line; this begs for a little GUI
to present these a little more easily.
5. Some linuxes, MX-for example, don't fully implement all the features
of acpi, but acpi interfaces and unused hooks abound. And these are
both plentiful and well advertised. This would seem to me to be possibly
exploitable by someone smarter than me.
6. I have two laptop pcs: a Dell 1545 legacy-boot 10-year-old and
a brand-new, almost, Dell 5570 with uefi. These comments represent
my total personal hardware experience with acpi.

I'd be glad to hear other opinions than my own on this topic.

[Stevo]

How did the Package Installer fail to install the latest Liquorix kernel? That should have "just worked". Could you provide some more details?

[TJ Hoye]

How did the Package Installer fail to install the latest Liquorix kernel? That should have "just worked". Could you provide some more details?

Hello again, Stevo.

You will have to ask BitJam to get a good technical answer here.

Back in MX-14 days I was as I still am working exclusively LiveUSB. Back then, both AntiX and BitJam were working on how to do a kernel upgrade on MX-LiveUSBs.
Not knowing why I had then tried to use the MX-Tools pkg installer to do this and failed. I repeated my mistake recently with your suggestion.
I never was a party to AntiX's and/or BitJam's solution except to cheer them on. BitJam gives me underserved credit other than being just a test case
for their invention.

All I know is for a LIveUSB the pkg installer fails somewhat softly, but may somewhat corrupt the initial LiveUSB kernel's install, IIRC.
The LiveUSB Kernel Installer works just fine for LiveUSBs, but it looks for kernel material in Synaptic.

TJH