1018051 - Add a slight delay in test_outgoing_emergency_in_airplane_mode.js

Status: RESOLVED FIXED

(Firefox OS Graveyard :: RIL, defect)

Description

[Szu-Yu Chen [:aknow] (inactive)]

When running the test case (test_outgoing_emergency_in_airplane_mode.js), we found that sometimes the CALL STATE CHANGE from modem is missing after the console command "gsm cancel". It results in timeout and fails the test.

We suspected that this is caused by multiple entries happened in emulator. The emulator were processing (1) console command "gsm cancel", and (2) SIM IO request at the same time. Both of these two actions have to write their response to a buffer and then transmit them through AT channel. However, there is only one such buffer. Action 2 might interrupt action 1 and overwrite the output buffer, so the original data on buffer, e.g., "CALL STATE CHANGE" is erased.

The patch provide a workaround, adding a slight delay, to postpone action (1) so these two actions will not collide together.

Comment 1

[Szu-Yu Chen [:aknow] (inactive)]

Attachment: Delay 1s in test_outgoing_emergency_in_airplane_mode.js

Edgar,
Hsinyi is OOO now. Could you review it for me?
I'd like to provide a workaround first to solve the issue on try test.

Comment 2

[Edgar Chen [:edgar]]

(In reply to Szu-Yu Chen [:aknow] from comment #0)
> When running the test case (test_outgoing_emergency_in_airplane_mode.js), we
> found that sometimes the CALL STATE CHANGE from modem is missing after the
> console command "gsm cancel". It results in timeout and fails the test.
> 
> We suspected that this is caused by multiple entries happened in emulator.
> The emulator were processing (1) console command "gsm cancel", and (2) SIM
> IO request at the same time. Both of these two actions have to write their
> response to a buffer and then transmit them through AT channel. However,
> there is only one such buffer. Action 2 might interrupt action 1 and
> overwrite the output buffer, so the original data on buffer, e.g., "CALL
> STATE CHANGE" is erased.

Please help to file a bug for this.

> 
> The patch provide a workaround, adding a slight delay, to postpone action
> (1) so these two actions will not collide together.

Comment 3

[Edgar Chen [:edgar]]

Comment on attachment 8431376 [details] [diff] [review]
Delay 1s in test_outgoing_emergency_in_airplane_mode.js

Review of attachment 8431376 [details] [diff] [review]:
-----------------------------------------------------------------

r=me with the comment addressed.

::: dom/telephony/test/marionette/test_outgoing_emergency_in_airplane_mode.js
@@ +33,5 @@
>    setRadioEnabled(false)
>      .then(() => gDial("112"))
>      .then(call => { outCall = call; })
>      .then(() => gRemoteAnswer(outCall))
> +    .then(() => gDelay(1000))

Please help to add some comments about why we need a delay here. Thank you.

Comment 4

[Szu-Yu Chen [:aknow] (inactive)]

Attachment: [final] Delay 1s in test_outgoing_emergency_in_airplane_mode.js. r=echen

Comment 5

[Szu-Yu Chen [:aknow] (inactive)]

https://tbpl.mozilla.org/?tree=Try&rev=a1045dc22e5c

No failures are relate to test_outgoing_emergency_in_airplane_mode.js.

Comment 6

[Edgar Chen [:edgar]]

https://hg.mozilla.org/integration/b2g-inbound/rev/16b0ccfa474d

Comment 7

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/mozilla-central/rev/16b0ccfa474d