Installing microG on Sailfish OS

[attah]

I will now stop feeding the troll.

[robang74]

@emva

The Jolla phone supported Android 4.1.2 way back in 2015

Great, because I read that “The First Tablet with Sailfish OS” has been released on 2014. Thus, Jolla started quite immediately supporting Android, but not SailFish OS in free (as gratis) version because licensing issue as far as I read:

[Q] The 'Alien Dalvik' licensing issue

Please, feel free to correct me if I am wrong.

[robang74]

@robang74

I would happy to change my mind because as tiner the microG role into the device functioning and thus the Android system below, as better.

The last time I switch on the GPS was to try the compass apps, some of them, at least. It was 18h ago, more or less:

@robang74

COMPASS CALIBRATION

Open StoreMan market and search for “ compass ”, choose:

• OrienteeringCompass

add the repository and install the app. Launch the app and swipe down to calibrate the compass. It claims that 30% is enough but with a doing the oo-loop slow and long enouhg can easily reach the 100%

Two things happen to my smartphone in this time:

• GPSinfo has been updated from v0.8 to v0.15.1-1 and now is able to calibrate the compass
• I have installed Google app and I granted to Google app the full priviledges, then I restarted the Android Support, then I removed the Google app (which failed and the icon remained) then I restarted the Android Support again (and the icon svanished).

After this two actions - which was completely indipendent because the first was an automatic suggestion due to the new native apps Chum market and the 2nd because I was giving a try to Google Lens app - my GPS started to work fine and also in the middle of my room it fix 3 or 4 satellites on 33 or 34 available indoor.

For sure the magic happes becuase GPSinfo calibrated the compass but compass calibration has been done also under another native apps, before.

I have a 6th sense for data leaking and about this GPS, it is triggered!

I might be wrong, I might be wrong, I might be wrong, I might be wrong, but…

[adekker]

I think you would do a lot of people a favor if you would ease down on all this incoherent reports/conclusions/speculations until you have more insight on how these components interact.

[robang74]

@adekker

I think you would do a lot of people a favor if you would ease down on all this incoherent reports/conclusions/speculations until you have more insight on how these components interact.

I agree.

[lkraav]

@robang74

This seems coherent with the solved issue reported by @lkraav refering to an Android app bolt after upgrading microG:

I never reported this as a “fix”, merely that for me, new MicroG didn’t seem to break location working by moving away from UnifiedNLP. Outdoors, Bolt was working fine with earlier versions, too, but indoors performance is much worse.

With new MicroG, I am mainly interested if indoors fix could also become high-performing. Testing right now, it seems indoors fix is obtained in a few seconds, but something like 100m away down the street.

[robang74]

@lkraav

I am mainly interested if indoors fix could also become high-performing. Testing right now, it seems indoors fix is obtained in a few seconds, but something like 100m away down the street.

What’s surprising me is that we have the same hardware (X10 II) with the same SFOS version (4.5.0.19) both we use the same version of microG (last available) with UnfiedNLP services active but our experience about GPS functioning is quite different. This is the reason for which I am still doing experiment on it.

Blackbox approch is a way but on the long run, I will go for the ADB console and trying to check the system log and everything else that can put me on the right way. I will keep you informed.

By the way I am using GPSinfo v0.15 and I have noticed that it is draining the battery at a rate I never experienced before. This is another novelty in the scene.

[lkraav]

@robang74

What’s surprising me is that we have the same hardware (X10 II) with the same SFOS version (4.5.19) both we use the same version of microG (last available) with UnfiedNLP services active but our experience about GPS functioning is quite different. This is the reason for which I am still doing experiment on it.

Maybe because I also do other modifications, like suplpatcher GPS stopped working - #425 by nekron and related gps.conf changes.

This usually makes all the difference.

[miau]

If doing GPS tests you need to reboot the phone after each fix to get comparable results.

[robang74]

@lkraav

Maybe because I also do other modifications, like suplpatcher GPS stopped working - #425 by nekron and related gps.conf changes.

This usually makes all the difference.

Yes, in particular this one.

A Word On 4G SUPL Support

It is mandatory to use SUPLv2 on 4G connected devices. To change the SUPL version setting I found out that patching the modems SUPL register is not needed because Sony vendor blob is initializing the setting for you based on gps.conf entry.

For SUPLv2 support please add the following setting into the gps.conf file:

SUPL_VER=0x20000

In fact, using the fiber connection available on my parents home, I got the GPS fixed indoor at the 4th floor of six. While using my 4G connection with UDP+VPN, I cannot manage to fix my GPS at 6th floor of six. Moreover, also the time skew is a disaster - with time.is - I got something between 1.1s and 0.4s. This because the auto-update time service works under a stressed connection which is shared with my laptop as well by the WiFi tethering.

This also explain why Android system apps are able to influence the GPSinfo results. Because they introduce more suitable parameters and keep the time skew much lesser impactful. In fact with Android Support activated the time skew rarely reach 0.1s with my 4G connection used as above described.

Considering that autoupdate time service could be the source of time jittering and /vendor/etc/gps.conf is misconfigured for my 4G connection. I am testing the “device only” GPS option (factory setting) and I have disabled the autoupdate time/date.

The result seems promising expecially because as soon I visit the time.is website I got the 4/38 fix but soon I loose it again unless I reload that page. Yes, keep the system hardware clock stable is a master key for indoor GPS while in outdoor the signal streght is enough to support this variances.

[robang74]

@attah

@robang74

native and Android apps, both have problem with the GPS because the release version of microG, then microG should be considered part of the hardware support under the point of functionality,

This is untrue. The “because of” specifically.

And i quote an actual Jolla-employee (emphasis mine):

@vige

Note that this only helps with android apps, as sailfish apps naturally don’t have access to location provided by microG.

Well, as far as I seen - microG is a jailed component that can be disabled and even when it runs cannot influence others subsystem unless it allows Android applications with system priviledges like Google app to interact with the network and stabilise the system clock.

Installing microG on Sailfish OS

Yes, in particular this one. A Word On 4G SUPL Support It is mandatory to use SUPLv2 on 4G connected devices. To change the SUPL version setting I found out that patching the modems SUPL register is not needed because Sony vendor blob is initializing the setting for you based on gps.conf entry. For SUPLv2 support please add the following setting into the gps.conf file: SUPL_VER=0x20000 In fact, using the fiber connection available on my parents home, I got the GPS fixed indoor at the 4th …

The system clock stabilisation, is enough to do the magic to fix the GPS quickly also indoor. Under this point of view the green area is - even - much bigger!

In fact, a malicius Android application that would messup the system clock would be able to create a GPS service DOS. Please, keep in consideration that I verified the clock skew with the default native browser and also with DuckDuckGo for Android to check if the two source of time were different.

Again, this test has been done in a blackbox mode with appaling rude instruments. Thus a more specific and detailed investigation could arrive to another conclusion. However, about GPS there are 676 post in a single thread (and there are few many thread) - thus, it is time to fix such issue.

---

GENERAL SOLUTION PROPOSAL

@vige: I think that it is time to give a end to GPS stops working which is a thread with 676 posts about 3 years long. The following UI change will make a lot people happy expecially using the proper values as default.

Parameter to set are the following in /etc/gps.conf or /vendor/etc/gps.conf:

NTP_SERVER (selection dropmenu)

• time.google.com (default)
• asia.pool.ntp.org
• europe.pool.ntp.org
  …
• oceania.pool.ntp.org
• north-america.pool.ntp.org
• south-america.pool.ntp.org
• [custom to enter by user]

SUPL_HOST:PORT

• supl.google.com:7276
  …
• supl.grapheneos.org:7275
• [custom to enter by user]

in order to support 4G mobile connection the version 2 is mandatory:

• SUPL_VER=0x20000

plus optional but suggested these other parameters:

• XTRA_SERVER_QUERY=1
• INTERMEDIATE_POS=1
• ACCURACY_THRES=5000

Which can set by a menu or simply by three check boxes [0/1].

[throwaway69]

@robang74

I will go for the ADB console and trying to check the system log and everything else that can put me on the right way.

Did you notice this is an actual linux and not yet another android ROM pretending to be an OS? Just enable developer mode and set up a password then you can ‘root’ your phone with ‘devel-su’ and password in the provided terminal. What exactly gave you the idea that visiting a website like time.is will cause the system to change its clock? Looks like you’re chasing a crazy placebo effect (sorry don’t want to feed but wtf)

[miau]

Have you rebooted the phone between the two tests?

[robang74]

@throwaway69

Just enable developer mode and set up a password then you can ‘root’ your phone with ‘devel-su’ and password in the provided terminal.

Because if I got a root shell under my fingers, I solve the problem but few people are able to digest a thread of 676 posts like GPS stops wroking started on July 2020 and with the last post 9 day ago Jun 2023.

Three years for a patch of 3 lines changed and not more than 6 changed!

[> trolling mode on <]
You all are the worst customer service I have ever had to deal with in my life and you all should pay a very expensive monthly fee to write on this forum!
[> trolling mode off <]

THE PATCH

NTP_SERVER=europe.pool.ntp.org # just to replace google
SUPL_HOST=supl.grapheneos.org  # just to replace google
SUPL_PORT=7275
SUPL_VER=0x20000  # v2 mandatory for the 4G connections

XTRA_SERVER_QUERY=1 # why not?
INTERMEDIATE_POS=1  # why not?
ACCURACY_THRES=5000 # uncertanty needs a limit

@miao: yes, solved. See my post before this.

[miau]

I don’t see anything new here

The GPS stopped working thread is initally about the XA2, which cannot connect to a SUPL server without suplpatch, because it has no certificate for the secure connection. So your settings were already tested and do not work on the XA2.

And many people claimed that they don’t wan’t to use A-GPS, because this does not fix the initial problem.

Also SUPL only works with mobile data not when connected to a wifi network.

All other settings you posted are already known.

[robang74]

@miau

I don’t see anything new here

A simple summary, that’s the novelty.

@miau

The GPS stopped working thread is initally about the XA2, which cannot connect to a SUPL server without suplpatch, because it has no certificate for the secure connection. So your settings were already tested and do not work on the XA2.

The patch exists so the XA2 users should be happy.

@miau

And many people claimed that they don’t wan’t to use A-GPS, because this does not fix the initial problem.

This part is not clear for me

I can understand that they wish to use ‘device only’ mode because privacy concerns but I did not get that the GPS was previously working indoor without A-GPS. Is it a specific problem of XA2 or a general problem? Considering that I have a X10II, it might be a general problem.

@miau

Also SUPL only works with mobile data not when connected to a wifi network.

Without changes was working well with a WiFi connection over fiber probably because such a connection was able to sync the system clock precisely enough and obviously this require the autoupdate of time/date set.

With these changes is working with a mobile 4G connection + UDP VPN, also. So, from my PoV is a great advantage because now I can have indoor GPS quickly working with my mobile connection. Which was my goal.

[babo]

@flypig

@vige

It is possible to install MozillaNlpBackend.

Are you able to activate it in the Settings? When I try, it immediately deactivates itself, whereas at least Apple UnifiedNlp Backend stays activated.

@vige

Yes I am able to activate it. It first did that with me as well, but after I did something it stopped doing that. I wish I knew what that “something” was.

Quite a late reply, but maybe could still be relevant. In my experience, that “something” is giving the module the permission to allow it to access location at all times.

[CaMpEoN]

I can´t find to tick…" grant it signature spoofing using the " enable system package replacement " permission
For the PlayStore advaced tips…
Can anyone give me a clue… Thanks

[demislition]

Is that the one in Settings - System - Android AppSupport, which you check to allow the replacing of system packages with their unofficial versions?

[CaMpEoN]

I thin its the same… but with different name…
Thanks