If an external keyboard with a backlight gets unplugged then
up_kbd_backlight_event_io would constantly get called, burning 100% CPU.
To make things worse, up_kbd_backlight_event_io would also constantly post
DBUS events, causing gnome-shell to also become very unresponsive.
This commit fixes this by returning FALSE from up_kbd_backlight_event_io
on unplug.
While at it also fix calling up_kbd_backlight_emit_change with a negative
brightness value in other error scenarios. Specifically this fixes
calling up_kbd_backlight_emit_change with -1 on the initial
up_kbd_backlight_event_io call in which case up_kbd_backlight_brightness_read
will typically fail with ENODATA.
Those devices have been supported in the Linux kernel since at least
3.19, so there's no need to try and support those with our user-space
handling.
Closes: #3
Check that external devices which show up as "battery" types have their
capacity_level correctly read, and that the device's type will be
correctly read after an input node is attached.
This check is supposed to start a poll for 1) batteries embedded in
external devices 2) batteries when battery polling is not disabled.
Make sure to start the poll for all batteries embedded in devices even
the ones that are just detected as "battery".
On some devices the power_supply node will show up before a related node
such as "input", making it impossible to know which type of device we're
attached to without waiting. Try to detect the device type again if the
device type is still "battery".
This fixes some Logitech devices appearing as "Battery" instead of their
respective device type, such as "mouse".
Use the correct refresh function for device batteries which aren't of a
known type. This fixes battery information not getting updated on many
Logitech wireless devices.
Reproducer:
- Unplug Logitech receiver
- Replug Logitech receiver
- Press button on Logitech keyboard
- Move mouse/touchpad
The touchpad battery would be stuck in "unknown" and with a "type" of
battery. This commit forces the correct refresh function to be used, one
that reads the capacity_level on those devices, and will update the
overall battery level.
Closes: #72
This directory may not exist when cleaning out /var (e. g. on stateless
systems). As upowerd cannot mkdir it by itself due to `ProtectSystem`,
let systemd create it via `StateDirectory`.
Keep `ReadWritePaths` and the `mkdir` in the Makefile to continue to
support a non-default `--historydir`.
To better support stateless systems with an empty /etc, the old location
in /etc/dbus-1/system.d/ should only be used for local admin changes.
Package provided D-Bus policy files are supposed to be installed in
/usr/share/dbus-1/system.d/.
This is supported since dbus 1.9.18.
https://lists.freedesktop.org/archives/dbus/2015-July/016746.html
G_TYPE_INSTANCE_GET_PRIVATE has been deprecated since glib version
2.58 and should be replaced with the xxx_get_instance_private (obj)
which is generated by G_ADD_PRIVATE.
Use G_DEFINE_TYPE_WITH_CODE (..., G_PRIVATE_ADD (...)) instead of
the (deprecated since glib 2.58) function g_type_class_add_private
to add a private structure for a type.
Bump the minimal required version of glib to 2.38.0, the version
where G_PRIVATE_ADD was added.
Some devices report "Not charging" when the battery is full and AC power
is connected. In this situation we should report fully-charged instead
of pending-charge.
Closes: #86.
Currently up_device_supply_refresh_battery sets out_state before the
state value is definitive, so the wrong state value is returned to the
caller. Luckily the only caller does not make use of this value at the
moment, so there are no user-visible consequences. Nonetheless this is a
bug, so this commit fixes it.
Test relaying the pending-charge state to the DisplayDevice. This commit
adds three tests: only one battery pending-charge, one battery
pending-charge and another one discharging, and one battery
pending-charge and another one charging.
Without this change if all batteries in the system are in the
pending-charge state, the display device state is set to unknown, and
its icon to battery-missing-symbolic.
This change makes the pending-charge state be considered when
calculating the DisplayDevice state, setting it to pending-charge if at
least one battery in the system is pending-charge and no other is
charging or discharging.
Closes: #81Closes: #19
This will help make it more clear when adding an extra state on the
following commit. It also makes the language consistent between the
different lines. There are no changes on the meaning of these lines nor
any functional changes on this commit.
automake's documentation says that it works depth-first, but it doesn't
actually go depth first when dist'ing a directory. So our .ref.xml files
aren't generated when we try to get them added to the dist.
Fix this with a big hammer by moving the generation to the current
directory, rather than spending any more time working with automake.
Can't wait for the meson port.
