This was meant to be removed in bbcd9a58 already, but the specified
version was too low which resulted in the file not being removed on
package upgrades.

Signed-off-by: Martin Kepplinger <martin.kepplinger@puri.sm>

The /usr/bin/power_key.py script basically does 2 things:

* set a devfreq governor once (and handle frequencies but only on old kernels)
* shut down the system when the power button is being pressed for ca. 2 seconds

The first task is to be replaced via a simple udev rule. The second one
is being tracked by Librem5/OS-issues#146


and will be replaced by a GUI anyways.

This solves problems with the power_key script:
* it is not syncronised with the display system.
* it is a nightmare to maintain and keep working

In my tests, the long-press shutdown mechanism didn't work anymore, but if
it does for you, this will remove it (temporarily, see the above issue).

Signed-off-by: Martin Kepplinger <martin.kepplinger@puri.sm>

Angus Ainslie authored 4 years ago and Angus Ainslie committed 4 years ago

By moving the AUX BM818 antenna connector to the unmarked center position
the modem can get a GPS signal. This NOT the GNSS controlled by HKS but is
part of the WWAN modem so is bound to the modem firmware and the WWAN HKS.

Signed-off-by: Angus Ainslie <angus@akkea.ca>

Guido Gunther authored 4 years ago and Angus Ainslie committed 4 years ago

We want to use the devkit's uboot, not the phone one.

Guido Gunther authored 4 years ago and Angus Ainslie committed 4 years ago

This makes `head -1 *.lst` misleading

Guido Gunther authored 4 years ago and Angus Ainslie committed 4 years ago

This makes `head -1 *.lst` misleading

Sebastian Krzyszkowiak authored 4 years ago and Angus Ainslie committed 4 years ago

This allows easy switching between installed kernel packages.

Phone images are more likely to build so use that as the
librem5-devkit-flash-image test image.

Signed-off-by: Angus Ainslie <angus@akkea.ca>

/boot is '1' on the phone

It got dropped last release but not removed on existing
installations.

Signed-off-by: Martin Kepplinger <martin.kepplinger@puri.sm>

devfreq is being handled differently on our kernels. With 5.6+ frequency
is switched automatically by the kernel for devfreq, when the powersave
governor is set. So in that case we only to that (done by udev in the future).

This also makes devfreq settings compatible with the devkit again because
it reads the minimum frequency which is different than on the phone,
from sysfs.

Also, we particularly handle 5.6-rc based kernel versions that
include a broken devfreq implementation / setup. starting with version
5.6.0 we need to increment the ABI version to 2.

Use gbp dch --multimaint-merge to generate the changelog and improve
bad commit messages manually.

Signed-off-by: Martin Kepplinger <martin.kepplinger@puri.sm>

Guido Gunther authored 4 years ago and Martin Kepplinger committed 4 years ago

This makes sure all drivers got their dependencies probed as well.

Guido Gunther authored 4 years ago and Sebastian Krzyszkowiak committed 4 years ago

We have a fix in wlroots now.

Martin Kepplinger authored 4 years ago and Guido Gunther committed 4 years ago

At least on recent kernels when using backlight_led the path
does not differ on the phone.

Signed-off-by: Martin Kepplinger <martin.kepplinger@puri.sm>

Martin Kepplinger authored 4 years ago and Guido Gunther committed 4 years ago

I get the following error on the L5:

TypeError: argument of type 'CompletedProcess' is not iterable

and using out.stdout solves the problem.

Signed-off-by: Martin Kepplinger <martin.kepplinger@puri.sm>

It's currently blacklisted, so in order to test it anyway, do modprobe.

Signed-off-by: Martin Kepplinger <martin.kepplinger@puri.sm>

Signed-off-by: Martin Kepplinger <martin.kepplinger@puri.sm>

Signed-off-by: Martin Kepplinger <martin.kepplinger@puri.sm>

Guido Gunther authored 5 years ago and Guido Gunther committed 5 years ago

This also adds helper that makes testing other regulators simple.

Suggested by Guido. Thanks.

Signed-off-by: Martin Kepplinger <martin.kepplinger@puri.sm>

Linux changes the sysfs naming of devfreq from what we currently use, see
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=66d0e797bf095d407479c89952d42b1d96ef0a7f



Signed-off-by: Martin Kepplinger <martin.kepplinger@puri.sm>

This reverts commit b8314d28.

ETNA_MESA_DEBUG introduced in 2700da0c is required
by phoc to function (see: mesa#4), but it breaks other GL apps causing blinking
artifacts.

Closes OS-issues#101

As on Linux 5.6 this is totally stable for me on Birch, I'd like to use 25Mhz
instead of 100 when the screen is off.

I ran "stress -c 2 -vm 2" while using the phone on the console and
turning on the screen, browsing the web and the system kept usable
the whole time.

With this, the device cools down more quickly when the screen is
off and improves battery time.

For Linux 5.3 (only), we keep the current behaviour as 25Mhz doesn't seem
to be stable there.

Signed-off-by: Martin Kepplinger <martin.kepplinger@puri.sm>

Using the dat format uses the correct number of channels and clock for
the sai interface.

Signed-off-by: Angus Ainslie <angus@akkea.ca>

Signed-off-by: Angus Ainslie <angus@akkea.ca>

Signed-off-by: Angus Ainslie <angus@akkea.ca>

Signed-off-by: Angus Ainslie <angus@akkea.ca>

Guido Gunther authored 5 years ago and Angus Ainslie committed 5 years ago

Used in the tests

Martin Kepplinger authored 5 years ago and Guido Gunther committed 5 years ago

blacklist the magnetometer driver as we don't use it (yet).

Signed-off-by: Martin Kepplinger <martin.kepplinger@puri.sm>

Signed-off-by: Angus Ainslie <angus@akkea.ca>