1094572 - B2G RIL: handle data call fail cause accordingly

Status: RESOLVED FIXED

(Firefox OS Graveyard :: RIL, defect)

Description

[Jessica Jong [:jessica] (inactive)]

We should check the status returned by RIL_REQUEST_SETUP_DATA_CALL and act accordingly, such as to retry or not, see [1].

[1] https://github.com/mozilla-b2g/android-hardware-ril/blob/master/include/telephony/ril.h#l366-406

Comment 1

[Jessica Jong [:jessica] (inactive)]

Attachment: patch, v1.

Comment 2

[Jessica Jong [:jessica] (inactive)]

Comment on attachment 8521268 [details] [diff] [review]
patch, v1.

Review of attachment 8521268 [details] [diff] [review]:
-----------------------------------------------------------------

::: dom/system/gonk/RadioInterfaceLayer.js
@@ +3825,4 @@
>      this.state = RIL.GECKO_NETWORK_STATE_DISCONNECTED;
> +
> +    if (this.requestedNetworkIfaces.length === 0) {
> +      if (DEBUG) this.debug("This DataCall is not requested anymore.")

missed semicolon, will fix it in the upcoming patch.

Comment 3

[Jessica Jong [:jessica] (inactive)]

Comment on attachment 8521268 [details] [diff] [review]
patch, v1.

I wonder if we need to detect and block the case where retry is not needed, but data call setup is still triggered by dataregistration/networkinfo change. Actually, updateRILNetworkInterface() should only be called when there is a change, but it is always called now due to design limitation.

Edgar, may I have your review? Thanks.

Comment 4

[Edgar Chen [:edgar]]

Comment on attachment 8521268 [details] [diff] [review]
patch, v1.

Review of attachment 8521268 [details] [diff] [review]:
-----------------------------------------------------------------

Thank you.

::: dom/system/gonk/RadioInterfaceLayer.js
@@ -775,5 @@
>          oldSettings.oldEnabled = oldSettings.enabled;
>          oldSettings.enabled = false;
>        }
>  
> -      if (oldConnHandler.anyDataConnected()) {

It seems we could remove |anyDataConnected()| as well.

::: dom/system/gonk/ril_worker.js
@@ +4142,5 @@
>        if (newDataCall.status != DATACALL_FAIL_NONE) {
> +        if (newDataCallOptions) {
> +          newDataCallOptions.status = newDataCall.status;
> +          newDataCallOptions.suggestedRetryTime = newDataCall.suggestedRetryTime;
> +        }

Hmm, the code here isn't straightforward to understand to me. But I haven't better idea to handle this, because the |_processDataCallList()| is used for both REQUEST_SETUP_DATA_CALL and REQUEST_DATA_CALL_LIST now. Let's see if we need to do some refactoring.

Comment 5

[Jessica Jong [:jessica] (inactive)]

Thanks for the review, Edgar.

(In reply to Edgar Chen [:edgar][:echen] from comment #4)
> Comment on attachment 8521268 [details] [diff] [review]
> patch, v1.
> 
> Review of attachment 8521268 [details] [diff] [review]:
> -----------------------------------------------------------------
> 
> Thank you.
> 
> ::: dom/system/gonk/RadioInterfaceLayer.js
> @@ -775,5 @@
> >          oldSettings.oldEnabled = oldSettings.enabled;
> >          oldSettings.enabled = false;
> >        }
> >  
> > -      if (oldConnHandler.anyDataConnected()) {
> 
> It seems we could remove |anyDataConnected()| as well.

Yes, will do.

> 
> ::: dom/system/gonk/ril_worker.js
> @@ +4142,5 @@
> >        if (newDataCall.status != DATACALL_FAIL_NONE) {
> > +        if (newDataCallOptions) {
> > +          newDataCallOptions.status = newDataCall.status;
> > +          newDataCallOptions.suggestedRetryTime = newDataCall.suggestedRetryTime;
> > +        }
> 
> Hmm, the code here isn't straightforward to understand to me. But I haven't
> better idea to handle this, because the |_processDataCallList()| is used for
> both REQUEST_SETUP_DATA_CALL and REQUEST_DATA_CALL_LIST now. Let's see if we
> need to do some refactoring.

I agree that this is kind of confusing, especially the part that we pass completely different things to |_sendDataCallError()| on data call setup failure and on data call drop. In data call drop case, we are not copying the corresponding apn, username, password, etc, so it relies mostly on 'datacallstatechange' that is sent later. This needs some refactoring indeed.

Comment 6

[Jessica Jong [:jessica] (inactive)]

Attachment: patch (final).

Address review comment 4:
- remove |anyDataConnected()|

Comment 7

[Jessica Jong [:jessica] (inactive)]

Attachment: patch (final).

Sorry, add missing semicolon in line 3816.

Comment 8

[Jessica Jong [:jessica] (inactive)]

try looks good!

https://treeherder.mozilla.org/ui/#/jobs?repo=try&revision=c3b34a7a64e5

Comment 9

[Jessica Jong [:jessica] (inactive)]

https://hg.mozilla.org/integration/b2g-inbound/rev/8af17af9fc2e

Comment 10

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/mozilla-central/rev/8af17af9fc2e

Comment 11

[Jessica Jong [:jessica] (inactive)]

Attachment: patch (v2.0m)

Since I am going to be OOO tomorrow, I am attaching the patch for v2.0m, just in case.