307 Commits

Author SHA1 Message Date
aebc3fdeab
pygoscelis: Disable CONFIG_USB_NET_RNDIS_WLAN 2022-01-08 12:06:21 +01:00
ae89d527a4
pygoscelis: Disable CONFIG_NVME_MULTIPATH 2022-01-08 12:06:21 +01:00
6cf3c6477c
pygoscelis: Degrade CONFIG_X86_PMEM_LEGACY to module
This module is most-probably not used nor necessary on my machine and
might be removed in the future. For details about the technique see e.g.
https://github.com/pmem/ndctl/issues/32
https://www.intel.com/content/www/us/en/developer/articles/training/how-to-emulate-persistent-memory-on-an-intel-architecture-server.html
2022-01-08 12:06:21 +01:00
df0e62b1b1
pygoscelis: Bump to 5.15.13-T14s 2022-01-08 12:06:20 +01:00
f4b17afad6
pygoscelis: Bump to 5.15.12-T14s 2021-12-29 13:06:20 +01:00
d905acd2c8
Reverting the systemd migration
This is even worse:
- pcscd keeps being restarted, even with explictly disabling it,
  blocking my GPG hw token
- bridge interface setup would require either full systemd networkd
  usage (yet I need NetworkManager for VPNs) or disabling most parts
  of systemd-networkd (hoping that would work better than with pcscd)

Revert "pygoscelis: Give up, migrate to systemd"

This reverts commit aab957d951f56ff0905b4d0501caba5139aa03ef.
2021-12-28 13:51:32 +01:00
aab957d951
pygoscelis: Give up, migrate to systemd
Desktops without systemd became a PTIA to use. Examples:
- KDE is fucked up w/o systemd: https://bugs.kde.org/show_bug.cgi?id=432975
- new project expect systemd everywhere and the cool elogind project
  isn't enough anymore:
  e.g. https://invent.kde.org/plasma/plasma-firewall
  or pipewire, which expects `systemctl --user restart ...` to restart properly
- https://wiki.gentoo.org/wiki/Hard_dependencies_on_systemd
2021-12-26 21:11:13 +01:00
86d770a9f6
pygoscelis: Bump to 5.15.11-T14s 2021-12-22 22:45:39 +01:00
0335173dec
pygoscelis: Bump to 5.15.10-T14s 2021-12-18 15:29:36 +01:00
1d6d1fcd18
pygoscelis: Switch (back) to a higher timer frequency
In commit 23eecad7b48913b12bd83917647de99e66d69b49 (2021-10-13) I
lowered the timer frequency from 1 000 Hz to 100 Hz, but while I wasn't
able to recognize a reproducible drop in energy consumption there were
some heavy-load situations when the system started to stutter.

Setting it to 300 Hz as some binary distributions do sounds therfor like
a sane decision.
2021-12-15 22:51:33 +01:00
6330bf7d68
pygoscelis: Bump to 5.15.8-T14s 2021-12-15 22:46:03 +01:00
09625a9717
pygoscelis: Bump to 5.15.7-T14s 2021-12-08 21:35:25 +01:00
bf5acb0514
pygoscelis: Bump to 5.15.5-T14s 2021-11-27 17:21:26 +01:00
dbc7e9e357
pygoscelis: Bump to 5.15.3-T14s 2021-11-19 09:44:39 +01:00
531a637741
pygoscelis: Disable CONFIG_CRYPTO_MD5 2021-11-04 21:46:53 +01:00
61cbddb4fa
pygoscelis: Enable CONFIG_ZERO_CALL_USED_REGS 2021-11-04 21:46:52 +01:00
2bff3ce69f
pygoscelis: Switch to CONFIG_IOMMU_DEFAULT_DMA_STRICT 2021-11-04 21:46:52 +01:00
a585380847
pygoscelis: Disable CONFIG_X86_PLATFORM_DRIVERS_INTEL 2021-11-04 21:46:52 +01:00
90e6b8c0c1
pygoscelis: Enable some nice sounding conntrack stuff. 2021-11-04 21:46:52 +01:00
8f078787b0
pygoscelis: Disable some netfilter assets I never used. 2021-11-04 21:46:51 +01:00
5b516ef701
pygoscelis: Enable CONFIG_TCP_MD5SIG
I don't need it as I don't do anything with BGP.
2021-11-04 21:46:51 +01:00
994ec4bcab
pygoscelis: Enable CONFIG_WERROR 2021-11-04 21:46:51 +01:00
704829168a
pygoscelis: Bump to 5.15.0-T14s 2021-11-04 21:46:51 +01:00
5cb9736ab1
pygoscelis: Bump to 5.14.16-T14s 2021-11-03 19:52:55 +01:00
3b2a1f38f5
pygoscelis: Bump to 5.14.15-T14s 2021-10-27 12:17:44 +02:00
23eecad7b4
pygoscelis: Set the timer frequency to 100 Hz
Trying out a lower time frequency and check if it has any influence at all
after a short discussion in IRC about it.
To me it looks as there is no real impact on my hardware with this
After a debate in IRC and some reading it looks to me as there is no general
rule but many people with opinions. Therefore I don’t expect a difference in
overall power consumption or responsiveness, but will just try it out.

I set up a quick and dirty battery test:
- display brightness at 35 %
- wifi on
- bluetooth off
- some default background deamons running (cron, chrony/NTP, syncthing)
- one termina (konsole) open
- that energy monitor from KDE open
- CPU boost disabled with 'echo 0 > /sys/devices/system/cpu/cpufreq/boost'
- verified that temperatures are low enough that the fan is off

Both setups (5.14.11-T14s with CONFIG_HZ_1000=y vs. 5.14.12-T14s
with CONFIG_HZ_100=y) showed values around 5.6 W, down to 5.2 W, up to > 30W
as I called emerge to check if it's not just a display glitch.

As this is a Notebook I don't have to care at all about some video cards and
TV frequencies that are mentioned in various threads.

To end this commit message I’d like to share a quote that amused me during my
mini-research:

„864Hz is indeed the win - I won't use a kernel wihout that option anymore.”
 - User “creidiki”, forums.gentoo.org
2021-10-13 16:45:08 +02:00
aeb29dd984
pygoscelis: Bump to 5.14.12-T14s 2021-10-13 16:45:08 +02:00
140a0f3d3f
pygoscelis: Bump to 5.14.11-T14s 2021-10-10 16:55:49 +02:00
860eb36800
pygoscelis: Bump to 5.14.10-T14s 2021-10-07 09:07:52 +02:00
2323fe4222
pygoscelis: Bump to 5.14.9-T14s 2021-09-30 14:35:30 +02:00
1f2087ac7b
pygoscelis: Bump to 5.14.8-T14s 2021-09-28 09:53:08 +02:00
6c17796453
pygoscelis: Enable support for my external USB NIC 2021-09-17 11:53:01 +02:00
469a7d5985
pygoscelis: Bump to 5.14.5-T14s 2021-09-17 11:45:30 +02:00
c81cc0c5d3
pygoscelis: Bump to 5.14.4-T14s 2021-09-15 11:01:03 +02:00
21a735a8da
pygoscelis: Bump to 5.14.3-T14s 2021-09-12 18:27:29 +02:00
18a94ce728
pygoscelis: Expose thermal information to userspace
Enable CONFIG_INT340X_THERMAL=m
2021-09-08 10:36:42 +02:00
8be72aa078
Revert "pygoscelis: Enable support for another sensor found on some AMD SoCs"
This reverts commit 4b4b7789969d54e3ecf90dcc3f0cff1806cedd1c as I didn't
see any further sensors or sensor data.
2021-09-08 10:36:42 +02:00
bff4fd3cc3
pygoscelis: Enable ACPI NVDIMM Firmware Interface Table (NFIT) 2021-09-08 10:36:42 +02:00
8c16c564a1
pygoscelis: Disable CONFIG_X86_MSR
I _think_ it was only set for powertop on my legacy intel machine.
2021-09-08 10:36:42 +02:00
62587243e3
pygoscelis: Bump to 5.14.2-T14s 2021-09-08 10:36:41 +02:00
36a52721e8
pygoscelis: Bump to 5.14.1-T14s 2021-09-03 21:59:34 +02:00
3c2797ea2c
pygoscelis: Enable LMI (expose UEFI/BIOS in running system)
Knobs to read -- and write -- should be visible in
/sys/class/firmware-attributes/*

For more details see Documentation/ABI/testing/sysfs-class-firmware-attributes.
2021-09-02 19:13:28 +02:00
4b4b778996
pygoscelis: Enable support for another sensor found on some AMD SoCs 2021-09-02 19:13:28 +02:00
6c21f8de7e
pygoscelis: Enable the sysfs entry for GPIO 2021-09-02 19:13:28 +02:00
7661231c48
pygoscelis: Disable SLAB debugging
I don't use slab_debug=P anymore as it disables certain other functionalities.
2021-09-02 19:13:27 +02:00
028ddfbf77
pygoscelis: Disable CONFIG_CGROUP_PERF 2021-09-02 19:13:27 +02:00
894e771bb2
pygoscelis: Enable core scheduling 2021-09-02 19:13:27 +02:00
14ffb8edc0
pygoscelis: Default to CONFIG_BPF_UNPRIV_DEFAULT_OFF 2021-09-02 19:13:27 +02:00
5077c82174
pygoscelis: Bump to 5.14.0-T14s 2021-09-02 19:13:27 +02:00
f632d4bca5
pygoscelis: Bump to 5.13.13-T14s 2021-08-26 23:41:05 +02:00