Arch Linux

Description:
My laptop always boots into 'powersave'-mode, although I have specifically installed and enabled laptop-mode and cupower and have configured both to use the 'performance'-governor, whatever the powersource this 5y-old laptop is running on.
It fails every time: I always have to wait for my scaled-to-798Mhz processor to be able to load a terminal and then manually run a command like
# systemctl restart cpupower, or
# cupower frequency-set -g performance
... which works every time.

My logs tell me this:
# journalctl --boot | grep cpu
jan 08 08:49:00 mbp-vvh kernel: Initializing cgroup subsys cpuset
jan 08 08:49:00 mbp-vvh kernel: Initializing cgroup subsys cpu
jan 08 08:49:00 mbp-vvh kernel: Initializing cgroup subsys cpuacct
jan 08 08:49:00 mbp-vvh kernel: setup_percpu: NR_CPUS:128 nr_cpumask_bits:128 nr_cpu_ids:2 nr_node_ids:1
jan 08 08:49:00 mbp-vvh kernel: PERCPU: Embedded 32 pages/cpu @ffff88013fc00000 s92824 r8192 d30056 u1048576
jan 08 08:49:00 mbp-vvh kernel: pcpu-alloc: s92824 r8192 d30056 u1048576 alloc=1*2097152
jan 08 08:49:00 mbp-vvh kernel: pcpu-alloc: [0] 0 1
jan 08 08:49:00 mbp-vvh kernel: RCU restricting CPUs from NR_CPUS=128 to nr_cpu_ids=2.
jan 08 08:49:00 mbp-vvh kernel: RCU: Adjusting geometry for rcu_fanout_leaf=64, nr_cpu_ids=2
jan 08 08:49:00 mbp-vvh kernel: cpuidle: using governor ladder
jan 08 08:49:00 mbp-vvh kernel: cpuidle: using governor menu
jan 08 08:49:00 mbp-vvh kernel: HPET: 4 timers in total, 0 timers will be used for per-cpu timer
jan 08 08:49:00 mbp-vvh kernel: ledtrig-cpu: registered to indicate activity on CPUs
jan 08 08:49:01 mbp-vvh kernel: ACPI: acpi_idle registered with cpuidle
jan 08 08:49:01 mbp-vvh systemd[1]: Starting Apply cpupower configuration...
jan 08 08:49:01 mbp-vvh cpupower[431]: Setting cpu: 0
jan 08 08:49:01 mbp-vvh cpupower[431]: Setting cpu: 1
jan 08 08:49:01 mbp-vvh systemd[1]: Started Apply cpupower configuration.
jan 08 08:49:03 mbp-vvh kernel: nvidia_bl(O+) nvidia(PO) input_leds joydev mousedev hid_apple bcm5974 hid_appleir bnep usbhid hid msr mac_hid evdev uvcvideo nls_iso8859_1 snd_hda_codec_cirrus snd_hda_codec_generic videobuf2_vmalloc nls_cp437 videobuf2_memops vfat fat videobuf2_core v4l2_common applesmc led_class input_polldev videodev nls_utf8 media coretemp kvm_intel kvm wl(PO) snd_hda_intel cfg80211 pcspkr snd_hda_codec snd_hda_core snd_hwdep shpchp drm snd_pcm uas snd_timer snd forcedeth soundcore nv_tco usb_storage i2c_nforce2 acpi_als kfifo_buf industrialio sbs battery sbshc fjes acpi_cpufreq button processor ac sch_fq_codel vboxdrv(O) btusb btrtl btbcm btintel bluetooth rfkill ip_tables x_tables firewire_ohci firewire_core crc_itu_t ohci_pci ohci_hcd ehci_pci ehci_hcd usbcore usb_common sr_mod pata_acpi cdrom

... which leads me to suspect that the acpi_cpufreq-module (which is the driver used by my laptop) is not loaded at the time cpupower is setting the governor. Whatever is causing this, it is not causing it when I manually restart cpupower or manually set the governor from the Gnome terminal. This looks like a race condition to me?

Additional info:
* package version(s)
Arch 64bit fully up to date
* config and/or log files etc.

I have tried to disable laptop-mode.service. I have now also disabled intel_pstate (which I think is irrelevant). I also manually edited the cpupower-script and added the following lines at the bottom:
modprobe acpi_cpufreq
cpupower frequency-set -g performance

The second line is executed (this clearly follows from the above logs at 8:38:21). The first one, I do not know.

Steps to reproduce:
Boot.