AGPS does not work

[GoogleCodeExporter]

What steps will reproduce the problem?
1. Start Google maps
2. Select My Location
3.

What is the expected output? What do you see instead?
The GPS icon should show up. It does not. Its blank. Nor does it show location 
by GPS.

What version of the boot.img and system.img are you using?
System.img V4.1 and boot.img V1.2

Can you provide logcat output during that error?

Please provide any additional information below.

Original issue reported on code.google.com by cosmicbu...@gmail.com on 31 Jul 2010 at 1:07

[GoogleCodeExporter]

Maybe you removed the driver from the qcom makefile?

Original comment by caf...@gmail.com on 7 Nov 2010 at 3:46

[GoogleCodeExporter]

Build fix pushed. Indeed there was a change in my source tree, but in a part 
not version controlled so it wouldn't show up with all the git commands in the 
world. Anyway, duplicated code now removed and should build for you all again. 
Enjoy your gps...

Richard.

Original comment by richarda...@gmail.com on 7 Nov 2010 at 8:51

[GoogleCodeExporter]

thanks richardaburton, my gps starts locating me...

Original comment by iareh...@gmail.com on 7 Nov 2010 at 10:33

[GoogleCodeExporter]

perfect :)

Original comment by saxand...@gmail.com on 7 Nov 2010 at 11:28

• Changed state: Fixed

[GoogleCodeExporter]

After installing the 20101106 nightly build, the gps working fine, not 
necessary to disable/enable procedure, the fix was less then 5 sec.

Original comment by peter.z...@gmail.com on 7 Nov 2010 at 11:38

[GoogleCodeExporter]

GPS starts working but phone started to restart after richardaburton's build. 
reflashed nightly build 20101106.

Original comment by iareh...@gmail.com on 7 Nov 2010 at 12:33

[GoogleCodeExporter]

GPS still not working after flashing nightly build 20101106 ;(. GAPPS gets 
installed.

Original comment by iareh...@gmail.com on 7 Nov 2010 at 1:21

[GoogleCodeExporter]

I am having the same problem still. I have flashed nightly build 20101107-1300. 
Although the GPS no longer requires that I turn it off and then on. it still 
does not aquire a fix. If I disable and then reenable the gps via the widget, 
then the phone reboots when I try to use it.

Original comment by jonathan...@gmail.com on 7 Nov 2010 at 10:35

[GoogleCodeExporter]

I have extracted some logs one of me attempting to use google maps right after 
reboot and one of me disabling and re-enabling the GPS then opening google 
maps. The second log clearly shows a bad file number error as soon as the gps 
thread starts. The end of the log is the last data I receive before the phone 
rebooted.

Original comment by jonathan...@gmail.com on 7 Nov 2010 at 10:57

Attachments:

• googlemaps.log.diff
• disable_reenable.log.diff

[GoogleCodeExporter]

1.5 hours after the satellites does not hook up mobilemaps, while google 
navigator is written with "gps waiting ..." and does not work

Original comment by fabiosol...@gmail.com on 7 Nov 2010 at 11:12

[GoogleCodeExporter]

Works really well for me with the 07112010 1300 release.
Does not take more than 30 seconds to fix in my house !!!
It s as fast as the 1.6 korean rom i had tested
Good job !!

Original comment by aurelien...@gmail.com on 8 Nov 2010 at 12:44

[GoogleCodeExporter]

build 07112010 1300 does not work for me. GPS unable to get lat long...

Original comment by iareh...@gmail.com on 8 Nov 2010 at 1:04

[GoogleCodeExporter]

same problem as in the comments 58, 59 and 62.

GPS netither works both on 20101106 and 20101107 nightly builds. Phone just 
reboots. 

Original comment by emiron...@gmail.com on 8 Nov 2010 at 3:55

[GoogleCodeExporter]

I confirm rebooting as mentioned in 63

Original comment by iareh...@gmail.com on 8 Nov 2010 at 4:11

[GoogleCodeExporter]

Run mine for 2 days without reboot, gps active for several hours total, 
near-instant fix everytime.

When are you getting reboot? Random without gps active? Only when gps active? 
Only after turn off & on again with widget (like 59)?

Do you have Settings -> Location & Security -> Use wireless networks ticked? 
(You must agree to the popup too), this is needed to use agps.

Original comment by richarda...@gmail.com on 8 Nov 2010 at 7:17

[GoogleCodeExporter]

Hey, good news indeed!
Actually, I was experimenting with the open-source drivers, too, but you were 
faster. I think it's time for a new release soon.

richardaburton: Please try to keep the repository clean by not pushing those
"Merge branch 'froyo' of git://github.com/OpenEtna/android_hardware_msm7k into 
froyo".
with a workflow of

git checkout -b froyo github/froyo
[write code]
git commit
git pull --rebase
git push

you should be able to prevent it. Note the git pull --rebase. This updates your 
local code by fetching the remote repository and applies your local commits on 
top of them instead of merging them.

Nevertheless, great work!

Original comment by polyth...@gmail.com on 8 Nov 2010 at 9:12

[GoogleCodeExporter]

Hmmm in my opinion we shouldn't mark things as Fixed until the real release is 
done.

Otherwise people not using the nightly builds will come here and report again...

Like Issue 190 for instance...

Original comment by Tetsuo6...@gmail.com on 8 Nov 2010 at 9:51

[GoogleCodeExporter]

Issue 290 has been merged into this issue.

Original comment by Tetsuo6...@gmail.com on 8 Nov 2010 at 9:54

[GoogleCodeExporter]

@richardaburton: Reboot was random... It restarted on starting a call... I can 
retry your build if you say so

Original comment by iareh...@gmail.com on 8 Nov 2010 at 9:54

[GoogleCodeExporter]

I'd like it at ConfirmFix until people confirm the fix for an official release, 
too.

Original comment by polyth...@gmail.com on 8 Nov 2010 at 11:53

• Changed state: ConfirmFix

[GoogleCodeExporter]

@richardaburton

"Use wireless networks" in the "Location and Security" never was related with 
agps. It always worked without this option enabled.

My phone reboots the same as described in comment 59. logcat contents are the 
same.

Original comment by emiron...@gmail.com on 8 Nov 2010 at 1:16

[GoogleCodeExporter]

I tried with Nightly Build from 2010-11-06. I´m inside a building and open 
GMaps with GPS and 3G enable and even inside the building I get a position by 
AGPS (GPS itself didn´t get a fix, but maps showed correctly my position).

So I think that this could confirm that AGPS is working, since it get my 
position without GPS itself (only using 3G). I will try later get a position 
outside.

Original comment by trysta...@gmail.com on 8 Nov 2010 at 2:26

[GoogleCodeExporter]

@trystan2k are you talking about determining positions by wireless networks? 

AGPS is not a replacement for GPS, it's technology that makes GPS cold start 
faster by receiveing data about orbital positions and visible sattelites via 
internet.

In that fact, we can't tell that AGPS is working without GPS (only using 3G), 
because it will be incorrect, i think.

Correct me, if i wrong.

Original comment by emiron...@gmail.com on 8 Nov 2010 at 2:33

[GoogleCodeExporter]

@emironoff,

Yeah. I think so. I know that if you disable 3G and go on Maps, it will give 
your position only using 3G. 

But if you enable GPS and open Maps, it will only give you position when the 
GPS fix, won´t it ? If it should give the position using 3G and give it more 
accurate using GPS after getting a fix then the other versions are right and 
this one no.

But I think you are right. I just realize that now I get a position without GPS 
get a fix first and before this not happens... I actually don´t know what is 
the correct and incorrect way...

Anyway, later I will try outside and see how much time it takes to get a fix 
(in previous version I tested and it was more than 1 minute)...

Original comment by trysta...@gmail.com on 8 Nov 2010 at 2:46

[GoogleCodeExporter]

I mean that "Wireless networks positioning" includes Wi-Fi Positioning system 
and cell-site triangulation. They have nothing to do with gps, because they 
don't receive coordinates from sattelites.

And there are no problems with that way of positioning.

But we are talking about AGPS, which doesn't work. In fact, in latest releases 
(including 6.1-rc4) GPS status utility show no found sattelites at all. That is 
the problem, i think.

Original comment by emiron...@gmail.com on 8 Nov 2010 at 3:06

[GoogleCodeExporter]

oh, i made a mistake. I wanted to tell that "Wireless network positioning" 
doesn't calculate coordinates by sattelites, not receiveng them.

coordinates, of course, being calculated by gps receiver (or, maybe, with A-GPS 
assistance service, depending of A-GPS type)

Original comment by emiron...@gmail.com on 8 Nov 2010 at 3:09

[GoogleCodeExporter]

i never got a fix after richardaburton's changes and i've experienced the 
reboot issue after enabling/disabling gps, however i didn't wipe the phone 
before upgrading the rom.
gps status shows me there are up to 8 satellites available (yellow dots).

Original comment by skun...@gmail.com on 8 Nov 2010 at 3:39

[GoogleCodeExporter]

i confirm with @skunk73, but gps status shows me no satellites at all. 
But reboot issue is the same, also wipe before flashing doesn't affect this 
issue. I've tried both with and without wiping, it didn't change result.

Original comment by emiron...@gmail.com on 8 Nov 2010 at 4:16

[GoogleCodeExporter]

emironoff: quite correct (about agps and the setting I mentioned) I was writing 
about 2 different things in a hurry and ended up editing them together! Btw, 
there is a setting for agps that has been removed from the settings app 
(upstream), but it appears that it does default to on without the user being 
able to enable it. As for the suggestion to enable the network based location 
option that will get you a course grained location fix very rapidly, so in 
google maps (the application the original bug report mentions), with this 
enabled you should get an almost immediate location (in my case accurate to 
about 2.5km) as a starting point while waiting for a gps fix. It's also all 
part of the location api so it is important to know this setting because it 
could be this that is causing the reboot issue mentioned, not actual gps.

Anyway, I can confirm a reboot issue after disabling and re-enabling the gps 
with the widget, then opening gps status. I'll have a look at that as soon as I 
get chance (I've got a busy week in theatre this week!), luckily in the mean 
time there is a simple workaround - don't disable and re-enable it. I have been 
running the updated version since I made the changes and that is the only issue 
I've had with it. I get a fix in a matter of seconds and I've had no reboots 
other than the specific scenario I just mentioned. Also, I did not do a wipe 
when moving to this version and I can't see any reason that would help so no 
need for anyone to be doing that.

Now, the interesting question is why has moving to the open source driver 
totally fixed gps for some people, like me, and apparently made no difference 
to others?

Original comment by richarda...@gmail.com on 8 Nov 2010 at 5:17

[GoogleCodeExporter]

It's worth noting, I was in the group of people who could get a gps fix, but it 
took forever, with the old driver. Hence my assumption that near instant fixes 
with the new driver implied agps was now working. Are the people who cannot get 
a gps fix now the same ones who couldn't get one before with the old driver? 
And are the people who posted positive things about the fix the people who 
could previously get a very slow fix with the old driver? Oh, and I'm in the 
UK, just in case there is a geographical element to any pattern.

Original comment by richarda...@gmail.com on 8 Nov 2010 at 5:37

[GoogleCodeExporter]

i belong to the people living in spain whom did wait a long time before getting 
a satellite fix before the changes (from 5 to 10 minutes) and doesn't get a fix 
at all after (was waiting over 30 min.).
as i wrote above gps status can see up to 8 satellites it should suffice to get 
a fix, does it?

Original comment by skun...@gmail.com on 8 Nov 2010 at 6:29

[GoogleCodeExporter]

the minimum required for a positional fix is 3 (4 for altitude). Depending on 
the SNR, 8 should be way more than enough.

Original comment by jonathan...@gmail.com on 8 Nov 2010 at 6:33

[GoogleCodeExporter]

i'm from Russia, Moscow and problems with gps fix started with 6.x OpenEtna 
branch.
A couple of times i used to get a fix, with 6.0 firmware.

I tried to do a rollback on Eclair (System 4.5-rc1, Boot 1.7) and with that 
image GPS worked perfectly. Very fast fix and accurate positioning. 

Original comment by emiron...@gmail.com on 8 Nov 2010 at 7:33

[GoogleCodeExporter]

I live in France, and when I open Map, the gps icon appears and blink and I've 
got a aproximate position of my phone. 3 min. later, the gps icon stop 
blinking, and I've got a great phone position. I think the fix work :-) thank 
you !

Original comment by benoit.f...@gmail.com on 8 Nov 2010 at 7:33

[GoogleCodeExporter]

rc4 in Canada still does not get a GPS fix  (GMaps and 20 minutes of 
attempting).  The GPS icon flashes like it's trying to get a fix, but never 
does.

Original comment by mark.mob...@gmail.com on 8 Nov 2010 at 8:23

[GoogleCodeExporter]

@richardaburton i've just tried to turn on both satellite and wireless networks 
positioning.

When i open maps, i receive inaccurate position data (with about 2km range) and 
that's all. GPS icon doesn't stop flashing and coordinates not become accurate.

Tell me if you need any additional information for debugging. I'm very 
interested in fix of this problem and i want to help with any testing/debugging.

Original comment by emiron...@gmail.com on 8 Nov 2010 at 10:05

[GoogleCodeExporter]

I've what appears to be the source of the reboot error after toggling gps off 
and back on again. There appears to be a bug in the qcom gps code. It appears 
to be missing a bit of de-init code, which should (presumably) get rid of a 
particular object. Some cleanup is done, but not all, so on a second attempt to 
init the module we don't then do a full init. The bit of init we miss is not 
the same bit as wasn't previosuly cleaned up (otherwise we'd probably get away 
with it). Does that make sense?

I think I've fixed it, but I'm going to test. I've disabled and re-enabled my 
gps and no reboot so far (always seemed to happen within a minute previously, 
now been about 5), plus no errors in logcat. I'm going to bed now, if the phone 
is still running happily in the morning I'll be happy with it. Although since 
we haven't forked the project I'm not sure what to do with it at the moment. 
Ideally it'd go upstream, after a bit more testing.

Richard.

p.s. emironoff: not sure at the moment, will have a think, also have you tried 
clearning agps data and redownloading with gps status?

Original comment by richarda...@gmail.com on 8 Nov 2010 at 11:21

[GoogleCodeExporter]

Hi.

I just did the test outside and for me gps and agps is working ok. In first 
time that I open Maps, it take 4 minutes to fix a positin (but with 3g I 
already have my position on map) but I do not know why I lost 3g connectio. 
Then i reboot the phone and start maps again and it takes less than 1 minute to 
fix gps position.

In previous versio it was working for me (take long time to. Fix but works.) 
Just like with richard...

Original comment by trysta...@gmail.com on 8 Nov 2010 at 11:21

[GoogleCodeExporter]

@richardaburton yes, i tried. But it says "A-GPS data will be downloaded as 
soon as you have data connection". But data connection is established, i've 
tried to use 3G/2G and Wi-Fi.

While trying to download data via wi-fi, connection being lost for a couple of 
seconds, then re-enabled.

But i think more interesing will be results of try to clean/download gps data 
via mobile network. I have attached logcat contents.

Original comment by emiron...@gmail.com on 9 Nov 2010 at 1:30

Attachments:

• [gps logcat.txt](https://storage.googleapis.com/google-code-attachments/openetna/issue-49/comment-89/gps logcat.txt)

[GoogleCodeExporter]

Well the phone survived the night. I disabled and re-enabled the gps again with 
the widget just for fun, then went outside and started gps status. Fix in 51 
seconds, no reboots, no errors in logcat. I think the reboot issue is fixed, 
just need advice on what to do with the patch now...

Original comment by richarda...@gmail.com on 9 Nov 2010 at 7:51

[GoogleCodeExporter]

Can you post/attach the patch?

Original comment by polyth...@gmail.com on 9 Nov 2010 at 11:29

• Changed state: Started

[GoogleCodeExporter]

Reboot fix patch attached...

polytheus: Can we just drop the patched files in our overlay or does this only 
work for resource files (I can't find much documentation). Or will we need to 
fork it? And sorry about those dodgy merge commits, still getting the hang of 
git!

Original comment by richarda...@gmail.com on 9 Nov 2010 at 6:35

Attachments:

• qcom-gps.diff

[GoogleCodeExporter]

I decided to open a bug upstream and attach the patch, but I wouldn't expect 
that to move at our desire pace (if at all). 
http://code.google.com/p/android/issues/detail?id=12463

Original comment by richarda...@gmail.com on 9 Nov 2010 at 7:17

[GoogleCodeExporter]

@richard we can fork until the patch is adopted upstream, and then go back to 
their repo...

Original comment by saxand...@gmail.com on 9 Nov 2010 at 8:07

[GoogleCodeExporter]

Overlays only work for resources. Forking seems to be the best options. I have 
some scripts that make it easy to merge changes from upstream (if there are 
any), so this should not be a big issue. I'll fork it tomorrow.

Original comment by polyth...@gmail.com on 9 Nov 2010 at 9:40

[GoogleCodeExporter]

polytheus: thanks

emironoff: yes that is interesting, but I can see this happening as this is a 
callback and it could interrupt a request already in progress, in which case 
it'll simply be abandoned, i imagine it's not that important about the clock 
because you are only 15 seconds off (is that close enough? maybe not...)

what's more interesting is that i don't think the deletion of the agps data 
should actually get as far as it does if the gps is enabled, the data can only 
be deleted while gps is disabled (according to comments in the code), however 
i'm not sure it checks that correctly so could be getting called at an 
inappropriate point, i'll need to look more closely when i'm less tired, of 
course not being able to recreate the issue means it going to be difficult to 
really see what's going on

Original comment by richarda...@gmail.com on 9 Nov 2010 at 9:56

[GoogleCodeExporter]

I have forked it. Do a "repo sync -j 10", so the right remote gets added.

You may also want to reenable the hidden AGPS Setting in the settings app 
packages/apps/Settings/res/xml/security_settings.xml
to help debugging. I'm not sure if we can safely assume which value this 
setting has on the openenta phones out there.

And there is also a setting config_gps_xtra_download_on_boot in
device/lg/eve/overlay/frameworks/base/core/res/res/values/config.xml which you 
may want to experiment with.

Original comment by polyth...@gmail.com on 10 Nov 2010 at 8:30

[GoogleCodeExporter]

There is a little bug : when I use GPS Essentials, it see 0 satellite when my 
position is fixed.

Original comment by maxgenil...@gmail.com on 10 Nov 2010 at 4:18

Attachments:

• snap20101110_070910.png

[GoogleCodeExporter]

I don't now if it is related but i open a 
issue(http://code.google.com/p/openetna/issues/detail?id=295) regarding a 
Exception that i sow in the logcat when i use GPS

Original comment by ivolu...@gmail.com on 10 Nov 2010 at 6:19

[GoogleCodeExporter]

Reboot patch now committed. Also option to enable/disable agps has been 
reinstated to the settings app. Please ensure it is enabled in settings after 
updating to a build containing it.

Original comment by richarda...@gmail.com on 10 Nov 2010 at 9:05

• Changed state: ConfirmFix