Pull the 2 previous changes together to refresh the battery status for a
couple of seconds after receiving an event on the power line. This fixes
charge status being inaccurate for up to 120 seconds (the normal poll
timeout on MacBooks).
https://gitlab.freedesktop.org/upower/upower/issues/56
When an event happens on the power line, and we are using a particular
device (in this case a MacBook or MacBookAir), we might need to poll
more aggressively after the event.
This adds a function to detect and export this.
Fix double-close on exit. If we created a GIOChannel from the file
descriptor, then g_io_channel_shutdown() will close that file descriptor
as well. Close the channel first, so that the file descriptor is only
closed manually if we didn't manage to create a GIOChannel from it.
https://gitlab.freedesktop.org/upower/upower/issues/60
Don't leave the history files installed when running "make uninstall".
Those files will likely have been created by the test suite when running
"make distcheck"
sysfs_get_double_with_error() used to return -1.0 for errors, but in
some cases, we might want -1.0 to be a valid value, such as for negative
discharge rates. Return FALSE on error instead.
The existing function didn't set the free function on the GPtrArray.
This means a libupower user can easily make a mistake in the reference
counting of the array contents.
The current power_supply support in the kernel will advertise an unknown
capacity when a device is being charged. This tests this state as well
as the expected behaviour when the device is fully charged.
https://bugs.freedesktop.org/show_bug.cgi?id=97484
Rather than leaving the percentage untouched, set the battery level at
50.0 as a "middle-ground".
This percentage is supposed to be ignored if a BatteryLevel is set so
this is better than showing "0%" which makes it look like a bug.
https://bugs.freedesktop.org/show_bug.cgi?id=97484
Newer kernels emit bind/unbind uevents that are not of interest to
powerd. To avoid littering logs with scary messages, let's lower their
severity to "debug".
https://bugs.freedesktop.org/show_bug.cgi?id=106018
This adds an extra check to query if the iOS device has a
battery at all before actually trying to retrieve the battery
status. Should still work for older iOS versions which do not
provide this key.
https://bugzilla.freedesktop.org/show_bug.cgi?id=105625
As Apple TVs do not have a battery and do not provide the
"BatteryIsCharging" lockdown key, upower would crash due to a
missing check whether the retrieved node is actually valid.
https://bugzilla.freedesktop.org/show_bug.cgi?id=105625
When UPower would appear, or disappear, we'd get warnings like:
g_object_notify: object class 'UpDevice' has no property named 'g-name-owner'
or:
g_object_notify: object class 'UpClient' has no property named 'g-name-owner'
This was caused by the property proxying added in 7531dbd and 28438a7
being too lax about which properties it tried to proxy, which included
ones that didn't apply to the object types in question.
See https://bugs.freedesktop.org/show_bug.cgi?id=43001https://bugs.freedesktop.org/show_bug.cgi?id=102350
As exported through BlueZ's org.bluez.Battery1 D-Bus interface. This
interface is only used for device where the battery information cannot
be processed in the kernel.
This is the first UpDevice type that doesn't use UdevDevice for the
Linux backend, and it is also the first that does not poll() status at
all.
https://bugs.freedesktop.org/show_bug.cgi?id=92370
Since libimobiledevice 5a8449a, LOCKDOWN_E_NOT_ENOUGH_DATA was renamed
to LOCKDOWN_E_RECEIVE_TIMEOUT. We'll use "-7" to avoid having to use
build glue.
The take_action_timeout_cb() function returns G_SOURCE_REMOVE which
makes GLib destroy the timeout. However the action_timeout_id
stayed != 0 so when warning level turned to "action" again the daemon
assumed that the timeout is already set and did nothing.
(It only logged: "Not taking action, timeout id already set")
https://bugs.freedesktop.org/show_bug.cgi?id=87791
Remove pointless upperbound, SENSOR_MAX_TYPES has no meaning here
and we break from the loop anyway when needed (i.e. when ENOENT
is returned). This aligns the code with other sensor device
walkers in OpenBSD.
https://bugs.freedesktop.org/show_bug.cgi?id=95257
If the poll timeout is actually changed because the warning level
has changed, only reset the actual poll code, not the warning level
notifications and the booking structures.
https://bugs.freedesktop.org/show_bug.cgi?id=99862
When a device is refreshed because the poll timeout has been reached,
the warning-level change notification can also be fired, which then
will reset (i.e. disable, re-enable) polling. For batteries this can
happen three times in a row.
Now we reset polling only if the calculated timeout actually differs
from the current one.
https://bugs.freedesktop.org/show_bug.cgi?id=99862
