FP4 idle battery use varies a lot depending on 2G, 3G or 4G setting. Why?

janmv

One of the topics on my UT-on-Fairphone4-wishlist is to understand what I can do to use the phone's battery optimally. In the past half year I have seen large differences in battery use when the phone is off, so screen off and no active apps (nothing shows if you swipe from the right). The phone used to need a recharge every 3 or 4 days, but currently it is just 1 day, while my daily usage pattern is roughly the same.

I probably changed settings, not knowing this would be the effect. I have now been checking more structurally. What I find is that setting Mobile Connection type to "2G only" or "2G/3G" instead of "2G/3G/4G" causes the full difference. I could have known, as the first options says 'saves battery'. The effect is like what I read in an old post (topic 6211), but that is a different phone (Xiaomi mi A2), and in my case it does not matter whether I turn data on or off. (I have a phone-only SIM-card, no reason to turn data on. Still I checked.)

From the system settings sources (Sim.qml) I see that the options are used to set a property 'technologyPreference' of an OfonoRadioSettings QML component, where '2G only' is 'gsm', '2G/3G' is 'umts', and '2G/3G/4G' is 'lte. In the ofono documentation I also find a Dbus radio-settings API (radio-settings-api.txt), where there is a FastDormancy property which is described as 'a major power-saving feature for mobile devices'. It seems as if this parameter is never explicitly being set in UT: searching the UBports core respositories does not result in hits. But this looks as a possible source of the differences.

I checked if the the modem might be limited, but from what I find on the internet, the FP4 is built on the Qualcomm Snapdragon 750G processor, which includes a modem called X52. This modem does not seem to pose a hardware limitation.

I also find a modem-manager package which might allow me to experiment with the modem. I tried to install it and run it in Libertine container, but from within the container, I do not have access to the Dbus.

So there are my questions: * What is the phone doing in the "2G/3G/4G"-setting that it does not do in the other two settings? * If it is related to the FastDormancy-setting, where is this setting use? Where should I look? Is there a way to experiment with that? * How to get modem-manager (mmcli) to run?

(I'm sorry for not posting links, apparently not enough reputation to do that.)

Thanks!

lduboeuf

@janmv There has been an attempt to set FastDormancy, but land to nowhere, maybe you could ask there ?.
@flohack can maybe tell what is missing ?

https://gitlab.com/ubports/development/core/repowerd/-/merge_requests/58

I agree LTE even without data On drains a lot battery ( Pixel3a), i feel lucky to still have 2G available here, otherwise the phone would be unusable as a daily driver.

lduboeuf

hmm by looking quickly ibto it, it appears modem or maybe the component in between modem and ofono doesn't provide the "FastDormancy" feature. Repowerd seems to call properly the method so not an issue tn that repo.

janmv

Thanks for this reaction. Looking at the repowerd source explains why I did not find more hits for "FastDormancy": most of the time, it is "fast_dormancy", with an underscore . I'll check if this brings me any further.

janmv

Some progress to mention:

As far as I can tell from the sources, setting FastDormancy should work, but I'm not sure I am looking at the right sources (gitlab ubports development core ofono-ubports)
For a check:

phablet@ubuntu-phablet:/usr/sbin$ ./ofonod --version
1.29

and

phablet@ubuntu-phablet:/usr/lib/systemd/system$ dpkg -l | grep ofono
...
ii  ofono-sailfish 1.29+git8-0ubports1~20240527210340.1~66073d0+ubports20.04                        arm64        Mobile telephony stack (daemon)
...

If I run testscript set-fast-dormancy (location mentioned above, from subdirectory test), it responds

phablet@ubuntu-phablet:~/.local/bin$ set-fast-dormancy 1
Setting fast dormancy for modem /ril_0...
Traceback (most recent call last):
  File "/home/phablet/.local/bin/set-fast-dormancy", line 26, in <module>
    radiosettings.SetProperty("FastDormancy", dbus.Boolean(enable));
  File "/usr/lib/python3/dist-packages/dbus/proxies.py", line 72, in __call__
    return self._proxy_method(*args, **keywords)
  File "/usr/lib/python3/dist-packages/dbus/proxies.py", line 141, in __call__
    return self._connection.call_blocking(self._named_service,
  File "/usr/lib/python3/dist-packages/dbus/connection.py", line 652, in call_blocking
    reply_message = self.send_message_with_reply_and_block(
dbus.exceptions.DBusException: org.ofono.Error.NotImplemented: Implementation not provided

Testscript get-tech-preference returns:

phablet@ubuntu-phablet:~/.local/bin$ get-tech-preference
Technology preference: umts

and this value changes correctly if I switch to 2G or to 2G/3G/4G

I added two lines to this script slightly to return all radiosettings:

for p in properties:
    print(p)

It shows:

phablet@ubuntu-phablet:~/.local/bin$ get-radiosettings
TechnologyPreference
AvailableTechnologies

So it seems that my current ofono implementation does not have a FastDormancy setting. @lduboeuf You suggest the reason might be the component in between the modem and ofono. What component is that?

lduboeuf

@janmv i don't know well that part, but that is the layer between Android and Ofono.

lduboeuf

@janmv i've ask the team. Apparently FastDormancy is handled natively under ofono

lduboeuf

said in FP4 idle battery use varies a lot depending on 2G, 3G or 4G setting. Why?:

@janmv i've ask the team. Apparently FastDormancy is handled natively under ofono

the componebt itself is called ofono-binder-plugin

janmv

Thanks. I'll check there if I can find why FastDormancy does not seem to be implemented.

janmv

It appears that the RIL-modem-software has no implementation of setting fast dormancy.

The FP4 has five plugins installed in folder /usr/lib/aarch64-linux-gnu/ofono-sailfish/plugins:

-rw-r--r-- 1 root root  19176 Aug 23  2023 apndbplugin.so
-rw-r--r-- 1 root root 367264 Oct  7  2024 binderplugin.so
-rw-r--r-- 1 root root  70592 Sep 25  2024 mtkbinderpluginext.so
-rw-r--r-- 1 root root   6096 Nov 19  2023 rilbinderplugin.so
-rw-r--r-- 1 root root 320424 Aug 23  2023 rilplugin.so

The packaging information in gitlab/ubports/development/core/packaging says that sources of binderplugin, rilbinderplugin and rilplugin are available at mer-hybris at github. (I suppose the other two plugins are not related to setting fast dormancy.)

An ofono plugin initializes driver methods when it is registered. Radio settings are initialized in this struct (defined in sailfishos/ofono/.../ofono/include/radio-settings.h):

struct ofono_radio_settings_driver {
   (...)
   void (*query_fast_dormancy)(...)
   void (*set_fast_dormancy)(...)
   (...)
}

In the mer-hybris binderplugin this structure is initialized (ofono-binder-plugin/src/binder_radio_settings.c line 234):

binder_radio_settings_init() {
   ofono_radio_settings_driver_register(&binder_radio_settings_driver);
}

but binder_radio_settings_driver does not provide fast dormancy methods (line 224). It only sets:

static const struct ofono_radio_settings_driver binder_radio_settings_driver = {
	.name                 = BINDER_DRIVER,
	.probe                = binder_radio_settings_probe,
	.remove               = binder_radio_settings_remove,
	.query_rat_mode       = binder_radio_settings_query_rat_mode,
	.set_rat_mode         = binder_radio_settings_set_rat_mode,
	.query_available_rats = binder_radio_settings_query_available_rats
};

As a result, radio_set_property_handler() in ofono/src/radio-settings.c (line 658) returns "Not implemented":

if (rs->driver->set_fast_dormancy == NULL)
	return __ofono_error_not_implemented(msg);

There are other repositories on github with the same code, but none of them provide fast dormancy methods.

Now what could be next? Of course I could be overlooking things. Is my conclusion right (RIL modem software does not implement fast dormancy)? If so, is it worth fixing? I'd say yes, as the difference in battery usage is large and 2G/3G are being phased out. And if it is worth fixing, then where? The ofono-binder-plugin repository is active. Would this be the place to ask? @lduboeuf can I have your advice once more?

lduboeuf

@janmv
sorry, i don't know the full story.
Some say that that modern drivers implement it directly, so not needed on upper stack.

Maybe could be something to ask for next Q&A or directly in repository has you said.

janmv

@lduboeuf I'll formulate a question for next Q&A-session. There is probably more to know and to consider than just a driver which could be implemented more completely.
I can still bring up the issue in the ofono-binder-plugin repository later.
Thanks!

lduboeuf

@janmv said in FP4 idle battery use varies a lot depending on 2G, 3G or 4G setting. Why?:

@lduboeuf I'll formulate a question for next Q&A-session. There is probably more to know and to consider than just a driver which could be implemented more completely.
I can still bring up the issue in the ofono-binder-plugin repository later.
Thanks!

thanks for working on it

lduboeuf

@janmv

on core dev Telegram channel,
There is ongoing discussion about the FastDormancy thing. It appears something is not correctly working, you can follow here:

https://t.me/ubports_devel/94429

janmv

@lduboeuf Thank you Lionel. I was already following the discussion in the UT-FP4 Telegram group on the same topic, but I was not aware of this one in Core Dev. Same contributors