1020060 - REQUEST_GET_CURRENT_CALLS fails on emulator

Status: RESOLVED FIXED

(Firefox OS Graveyard :: RIL, defect)

Description

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

https://tbpl.mozilla.org/php/getParsedLog.php?id=40998657

18:44:18     INFO -  06-03 21:09:38.992    44   148 I Gecko   : RIL Worker: Parcel (size 12): 0,0,0,0,179,1,0,0,2,0,0,0
18:44:18     INFO -  06-03 21:09:38.992    44   148 I Gecko   : RIL Worker: We have at least one complete parcel.
18:44:18     INFO -  06-03 21:09:38.992    44   148 I Gecko   : RIL Worker: [0] Solicited response for request type 9, token 435, error 2
18:44:18     INFO -  06-03 21:09:38.992    44   148 I Gecko   : RIL Worker: [0] Handling parcel as REQUEST_GET_CURRENT_CALLS

I found that we got GENERIC_FAILURE from ril on emulator. There are something we have to check:
- The request on emulator should not fail. So why it fails?
- Maybe we should add some error handling in gecko

Comment 1

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

Attachment: Retry getCurrentCalls

Add a retry mechanism for getCurrentCalls().

Comment 2

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

(In reply to Szu-Yu Chen [:aknow] from comment #1)
> Created attachment 8438209 [details] [diff] [review]
> Retry getCurrentCalls
> 
> Add a retry mechanism for getCurrentCalls().

May I know with this retry mechanism, is the try failure obviously improved?
I am asking because it seems to me that the root cause should be something wrong on emulator. And I am inclined to avoiding some workarounds for emulator on gecko side as much as we can, especially if the improvement isn't as much as we expect.

Comment 3

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

(In reply to Hsin-Yi Tsai [:hsinyi] from comment #2)
> (In reply to Szu-Yu Chen [:aknow] from comment #1)
> > Created attachment 8438209 [details] [diff] [review]
> > Retry getCurrentCalls
> > 
> > Add a retry mechanism for getCurrentCalls().
> 
> May I know with this retry mechanism, is the try failure obviously improved?
> I am asking because it seems to me that the root cause should be something
> wrong on emulator. And I am inclined to avoiding some workarounds for
> emulator on gecko side as much as we can, especially if the improvement
> isn't as much as we expect.

I don't know. It's a really seldom issue happened about twice a week. We should land it first, then we could know whether the patch is useful.

The problem happened on the emulator means that it could also happened on a real modem. In my opinion, we have to work on both solutions including gecko side and emulator side.

Comment 4

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

(In reply to Szu-Yu Chen [:aknow] from comment #3)
> (In reply to Hsin-Yi Tsai [:hsinyi] from comment #2)
> > (In reply to Szu-Yu Chen [:aknow] from comment #1)
> > > Created attachment 8438209 [details] [diff] [review]
> > > Retry getCurrentCalls
> > > 
> > > Add a retry mechanism for getCurrentCalls().
> > 
> > May I know with this retry mechanism, is the try failure obviously improved?
> > I am asking because it seems to me that the root cause should be something
> > wrong on emulator. And I am inclined to avoiding some workarounds for
> > emulator on gecko side as much as we can, especially if the improvement
> > isn't as much as we expect.
> 
> I don't know. It's a really seldom issue happened about twice a week. We
> should land it first, then we could know whether the patch is useful.

Hmm, I thought we should land it after we know it's helpful.
I am just concerned if we re-send the request in a very short time, emulator state likely remains the same as that when the 1st failed request. In this way, does resending benefit? But we all understand the tricky situation on try... If we don't land it first, we never know the answer. This is like chickens and eggs. :(

So, let's give it a try and evaluate the performance, and remove it if it's not helpful at all.

> 
> The problem happened on the emulator means that it could also happened on a
> real modem. 

Indeed, then we need to know what goes wrong on modem.

> In my opinion, we have to work on both solutions including gecko
> side and emulator side.

No objection to this opinion. :) Please file a emulator bug in case we forget!

Comment 5

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

Attachment: [final] Retry getCurrentCalls. r=hsinyi

Comment 6

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

https://tbpl.mozilla.org/?tree=Try&rev=a52b53d1d30d

Comment 7

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/integration/b2g-inbound/rev/2d95cd8f3e0e

Comment 8

[Ryan VanderMeulen [:RyanVM]]

https://hg.mozilla.org/mozilla-central/rev/2d95cd8f3e0e