GitLab

Start work on new ofono interface. So far, this detects ofono on dbus
and complains if both mm and ofono are running.

See the ofono issue in the eg25-manager repo for more information.

There's no reason we should wait 3s before retrying a failed AT command. 
This can even cause issues with suspend inhibition: 3 retries of a 
failed command would take 9s, while logind ignores the inhibitor after 
only 5s.

Reducing the delay to 500ms should be enough and wouldn't interfere with 
suspend inhibition.

Fixes #3

This makes its role more explicit.

Most of the modem issues follow a (incomplete) USB device reset. Instead 
of relying solely on the existing timer, this patch adds a udev monitor 
which resets the modem as soon as its associated USB device is reset, 
which greatly improves recovery time.

This patch adds a 3s delay when resetting the modem during which we 
avoid triggering a new reset. This makes sure we don't trigger a reset 
twice in a row.

It also disables any related running timer to avoid being re-triggered 
unnecessarily.

Sometimes it takes just a little bit longer than usual, so this avoids 
unnecessary modem recovery.

mm-iface: fix GError pointer

See merge request mobian1/devices/eg25-manager!2

use g_autoptr for GError

See merge request mobian1/devices/eg25-manager!1

This means the system is in a very bad shape as no other software should 
make use of those, so exit the daemon (will be restarted by systemd).

This makes sure the resume commands (disable URC cache and enable GPS)
are executed.

As ModemManager releases the modem while going to sleep, we clear the
USB ID too, which causes a segfault when trying to recover the modem.

This reverts commit ff9b26b8.

The response can include the command and an error number, so we want to 
only check it contains ERROR, even if it's replying more than just the 
'ERROR' string.

These are set commands, no need to verify the current value.

This way we can make sure the AT commands are executed only once 
ModemManager has released the modem, preventing any race condition.

For some reason, during system the main loop exits before we call 
g_main_loop_quit(), so don't rely on it and use a simple polling loop to 
check for modem poweroff.

`Developer edition` phones have a different compatible string but should 
be considered identical to BraveHeart edition here. Therefore BH is the 
default phone, unless the compatible is that of a CE phone.

When powering off the modem, we must assert the RESET line only when the 
modem has shut down (otherwise it'll cause a hard reset and won't allow 
the modem to shut down properly).

This commit also polls the STATUS pin during poweroff and quits 
immediately once this pin goes high.

RI won't be accessible due to it being used by the kernel as an 
interrupt source, so we can check the STATUS pin (CE only) which will 
allow us to know precisely when the modem is shut down.