We use the routing rule to control the flow of data connection when wifi and mobile data connection are both enabled. The default route will be set to Wifi interface when we connect to Wifi network. Not sure if we should disconnect the mobile data connection also. Android seems disconnect the mobile data connection when it is connected to Wifi network.

I think we should. UX comment?

My understanding from reading the description is that the data connection remains active but all network traffic is directed to wifi. (ie. The data connection is not in use while connected to wifi.) I don't think we provide any visual indication to the user that the data connection is still active. As such, basecamp-. Please renom with additional justification if I'm missing something here.

IMHO we should indeed disconnect the mobile data connection. When it is not in use it will use a fair amount of battery power to keep the connection alive. While this is not that bad in 2G it's more worse with 3G or even 4G. Checking that on my Android device shows me that the mobile connection is always disconnected too when WIFI is active.

Good point, Henrik. For the sake of battery life this seems like something we should seriously consider for v1, if it's feasible. Daniel and Chris, flagging for you, since this is more of a product call than UX.

How risky would this change be?

I saw Android implementing this in Connectivity Manager. It is not trivial job to implement it in gecko. Not sure if we can have quick solution from gaia.

Based on the risk, I don't think we can justify bb+ at this point.

3G active connections are a battery drainer and it is also pretty bad for operator networks as keeping a PDP context open is pretty expensive in terms of network resources. In 2G the problem is not the battery or the network, but that in most of the countries DTM (Dual Transfer Mode) is not supported, which means that having an active cellular connection, impedes the device to receive incoming calls. I am afraid this will be problematic for certification, but David will be able to confirm it.

Hi, Yes, I also agree on the need to PS detach when registered in a WiFi network. There is no need to maintain the connectivity up when using WiFi, and this has an impact on battery consumption and network resources. Thanks! David

Battery consumption is unfortunate but don't have a clear solution to this for basecamp. We'll have to figure this out after basecamp.

OK, bear in mind that battery is not the most important consequence, doing this in 2G will make the user miss calls without knowing why. This will be very common in LatAm.

Deassigning: it seems a bit too much work for the WW and I wouldn't be able to test it in Canada.

Can you help us understand why having a 2G connection prevents making or receiving phone calls? My phone indicates a 2G connection is present much of the time, but I can still use the phone so long as I'm not actively using data. Is the wifi connection causing the phone to assume its actively using 2G data when its not? Otherwise it seems like this bug would waste resources but not affect telephony.

(In reply to Lucas Adamski from comment #13) > Can you help us understand why having a 2G connection prevents making or > receiving phone calls? My phone indicates a 2G connection is present much > of the time, but I can still use the phone so long as I'm not actively using > data. > This is very true. If the 2G connection is not actively used you should be able to make/receive voice calls. > Is the wifi connection causing the phone to assume its actively using 2G > data when its not? > > Otherwise it seems like this bug would waste resources but not affect > telephony. Yes, my bad. More details about resources consumption: - Network: Leaving a PDP context open without needing it is problematic because most of GGSN licenses take into account the simultaneous open contexts in order to charge operator. - Battery: This depends pretty much on the support (both on the network and device side) of Fast Dormancy Version 8, which makes the device go to Idle faster and reduces battery consumption a lot even if the device is connected. You can find an interesting paper about these issues at [1]. David (dpv) can provide you more details about it. [1] www.nokiasiemensnetworks.com/sites/default/files/document/Smart_Lab_WhitePaper_27012011_low-res.pdf.

Taipei triaged.

WIP V1 to address this bug.

Comment on attachment 703771 [details] [diff] [review] WIP V1 Review of attachment 703771 [details] [diff] [review]: ----------------------------------------------------------------- Looks good.

If AGPS or MMS activated the data call which shares with default APN, we should not turn off data call when wifi enabled. This patch takes this into consideration.

Need to make sure that APGS or MMS deactivates data call when not been used.

Can we fix this for v1 without involving AGPS or MMS?

(In reply to Andrew Overholt [:overholt] from comment #20) > Can we fix this for v1 without involving AGPS or MMS? We can ignore MMS for v1. But we have to consider AGPS case, because most operators have AGPS sharing same APN with default APN. The two AGPS bugs depended by this bug should be easy to fix.

This is a clean up from WIP V2. If Wifi is connected, the data connection for Internet access will be deactivated, unless there are any activated AGPS/MMS data calls that shares same APN as default APN.

To be landed after bug 832925 land.

Landed on mozilla-b2g18/gaia master prior to the 1/25 branching to mozilla-b2g18_v1_0_0/v1.0.0, updating status-b2g-v1.0.0 to fixed.

Works great with todays build on the Unagi device.

Unagi Build ID: 20130313070202 Gecko: http://hg.mozilla.org/releases/mozilla-b2g18_v1_0_1/rev/e74dafa6b2d9 Gaia: b34e726147f8e671ad8c538b50900ccfbffcb084 Kernel: Dec 5th On the build above, I cannot get the bug to occur. When I launch adb and type adb shell netcfg, I'm unable to get the data connection to be active if Wifi is active. Only by turning off Wifi or being out of range can I see data working. Marking as Verified Fixed.