1149829 - nsINetworkInterface is broken on b2g emulator

Status: RESOLVED FIXED

(Firefox OS Graveyard :: RIL, defect)

Description

[Byron Campen [:bwc]]

After bypassing bug 1149792, we still permafail because nICEr is unable to get any local interfaces. Maybe only localhost is available now?

Comment 1

[Byron Campen [:bwc]]

Nils, do you have a patch laying around to enable localhost that I can try?

Comment 2

[Nils Ohlmeier [:drno]]

Bug 1128130 has a patch on it which should accomplish that.

Comment 3

[Byron Campen [:bwc]]

I've verified in another try push that |isNetworkReady| in network.js still fails once the network-ready event fires:

https://treeherder.mozilla.org/#/jobs?repo=try&revision=60e4840433c5

https://dxr.mozilla.org/mozilla-central/source/dom/media/tests/mochitest/network.js#13
https://dxr.mozilla.org/mozilla-central/source/dom/media/tests/mochitest/network.js#63

Additionally, it appears that the GetAddresses call is always returning 0 addresses here, even though we wait for a network-ready event before starting the test (the logging bears this out):

https://dxr.mozilla.org/mozilla-central/source/media/mtransport/gonk_addrs.cpp?from=gonk_addrs.cpp&case=true#73

The try run:

https://treeherder.mozilla.org/#/jobs?repo=try&revision=34498d08886c

Comment 4

[Byron Campen [:bwc]]

Not sure what the component is here, exactly. The logging shows that there are two network interfaces, but neither has any addresses. I am not sure what type these two interfaces are (maybe one wifi and one radio; I don't see anything that would register a loopback interface, which is unfortunate).

Comment 5

[Byron Campen [:bwc]]

I doubt this is a problem in isNetworkReady, since even once the network-ready event fires we still don't actually have any addresses, but maybe you know who to ask about this?

Comment 6

[Shih-Chiang Chien [:schien] (UTC+8) (use ni? plz)]

Let me explain how this worked before.

The network interface for b2g emulator is in gonk layer and there is no default network interface on b2g. B2G will only get the local address after turning on either 3G data or wifi and that's the entire code of isNetworkReady for.

@hsinyi, do you know if anything changed about setting up 3G data on emulator? The code we used in WebRTC test cases [1] is enabling radio network via mozMobileConnections and waiting for |datachange| event. I'm wondering if that trick still works on current codebase.

[1] https://dxr.mozilla.org/mozilla-central/source/dom/media/tests/mochitest/NetworkPreparationChromeScript.js

Comment 7

[Hsin-Yi Tsai (she/her) [:hsinyi]]

I believe Edgar can answer better than I do ;)

Comment 8

[Edgar Chen [:edgar]]

(In reply to Shih-Chiang Chien [:schien] (UTC+8) (use ni? plz) from comment #6)
> Let me explain how this worked before.
> 
> The network interface for b2g emulator is in gonk layer and there is no
> default network interface on b2g. B2G will only get the local address after
> turning on either 3G data or wifi and that's the entire code of
> isNetworkReady for.
> 
> @hsinyi, do you know if anything changed about setting up 3G data on
> emulator? The code we used in WebRTC test cases [1] is enabling radio
> network via mozMobileConnections and waiting for |datachange| event. I'm
> wondering if that trick still works on current codebase.

It should work on current codebase.

> 
> [1]
> https://dxr.mozilla.org/mozilla-central/source/dom/media/tests/mochitest/
> NetworkPreparationChromeScript.js

Comment 9

[Edgar Chen [:edgar]]

(In reply to Byron Campen [:bwc] from comment #4)
> Not sure what the component is here, exactly. The logging shows that there
> are two network interfaces, but neither has any addresses. I am not sure
> what type these two interfaces are (maybe one wifi and one radio; I don't
> see anything that would register a loopback interface, which is unfortunate).

Hi Byron, could you help to push try again with enabling ril log by changing |DEBUG_ALL| to true? Maybe we can get some clues from the log. Thank you.

[1] https://dxr.mozilla.org/mozilla-central/source/dom/system/gonk/ril_consts.js#17

Comment 10

[Byron Campen [:bwc]]

Try run with DEBUG_ALL=true

https://treeherder.mozilla.org/#/jobs?repo=try&revision=ac2dc88a6722

Comment 11

[Edgar Chen [:edgar]]

From the log, the cellular data is established and the ip address is "10.0.2.100/24".

11:10:49     INFO -  -*- RadioInterface[0]: Received message from worker: {"radioTech":3,"apn":"epc.tmobile.com","chappap":3,"pdptype":"IP","rilMessageClientId":0,"rilMessageToken":7,"rilMessageType":"setupDataCall","rilRequestType":27,"rilRequestError":0,"status":0,"suggestedRetryTime":54864,"cid":"1","active":2,"type":"IP","ifname":"rmnet0","addresses":["10.0.2.100/24"],"dnses":["10.26.75.40","10.26.75.41"],"gateways":["10.0.2.2"],"state":1}

I am going to check dose the ip address of nsINetworkInterface be updated correctly. ni? me for tracking.

Comment 12

[Edgar Chen [:edgar]]

Attachment: Test patch, v1

Hi Byron, I found there is something wrong in RIL's nsINetworkInterface.getAddresses(). Could you try this patch to see if it helps? Thank you.

Comment 13

[Byron Campen [:bwc]]

Let's see what happens!

https://treeherder.mozilla.org/#/jobs?repo=try&revision=1a58fa017990

Comment 14

[Byron Campen [:bwc]]

Ok, it looks like we're getting network interfaces now.

Comment 15

[Shih-Chiang Chien [:schien] (UTC+8) (use ni? plz)]

Gecko 39 also have this bug, remember to uplift to aurora. :)

Comment 16

[Edgar Chen [:edgar]]

When I tried to write a test case for nsINetworkInterfaceListService, I found another issue: addresses/gateways/dnses in RIL's nsINetworkInterface won't be clean after disconnect, I will provide a fix here, too.

Comment 17

[Edgar Chen [:edgar]]

Attachment: Patch, v2

Comment 18

[Edgar Chen [:edgar]]

Comment on attachment 8590128 [details] [diff] [review]
Patch, v2

Review of attachment 8590128 [details] [diff] [review]:
-----------------------------------------------------------------

- Fix the getAddresses issue.
- Clean addresses/gateways/dnses when disconnect.
- Add test case for NetworkInterfaceListService.

Hi Hsinyi, would you mind reviewing this? Thank you.

Comment 19

[Hsin-Yi Tsai (she/her) [:hsinyi]]

Comment on attachment 8590128 [details] [diff] [review]
Patch, v2

Review of attachment 8590128 [details] [diff] [review]:
-----------------------------------------------------------------

Looks good to me.

Comment 20

[Edgar Chen [:edgar]]

Try result: https://treeherder.mozilla.org/#/jobs?repo=try&revision=f316d82ab6c1

Comment 21

[Edgar Chen [:edgar]]

https://treeherder.mozilla.org/#/jobs?repo=b2g-inbound&revision=76007d3799af

Comment 22

[Wes Kocher (:KWierso) (Not reading bugmail; email directly if needed)]

https://hg.mozilla.org/mozilla-central/rev/76007d3799af