928643 - Disable MOZ_B2G_RIL on Flatfish

Status: RESOLVED FIXED

(Firefox OS Graveyard :: RIL, defect, P1)

Description

[Vicamo Yang [:vicamo][:vyang]]

+++ This bug was initially created as a clone of Bug #920551 +++

Comment 1

[Vicamo Yang [:vicamo][:vyang]]

nominate for koi+ for it blocks bug 911684 and bug 921912, which are both koi+.

Comment 2

[Vicamo Yang [:vicamo][:vyang]]

I think one can simply set `GECKO_CONFIGURE_ARGS` to '--disable-b2g-ril' in per device BoardConfig.mk (or any Android.mk) to disable RIL.  See https://github.com/mozilla-b2g/gonk-misc/blob/master/Android.mk#L273

Comment 3

[Francis Lee [:frlee]]

according to triage result, change to Koi+

Comment 4

[[e-mail]]

Askening, could you help add the '--disable-b2g-ril' ARG addressed in Comment 2 to TWCI flatfish build, then we could evaluate it quickly. 

thanks

Comment 5

[[e-mail]]

Askening, Sorry its build param in BoardConfig.mk, not for CI params

Comment 6

[Askeing Yen[:askeing][:fyen]]

(In reply to Fred Lin [:gasolin] from comment #4)
> Askening, could you help add the '--disable-b2g-ril' ARG addressed in
> Comment 2 to TWCI flatfish build, then we could evaluate it quickly. 
> 
> thanks

After added GECKO_CONFIGURE_ARGS="--disable-b2g-ril" into environment variable, the build failed.

Comment 7

[Vicamo Yang [:vicamo][:vyang]]

(In reply to Askeing Yen[:askeing] from comment #6)
> After added GECKO_CONFIGURE_ARGS="--disable-b2g-ril" into environment
> variable, the build failed.

Have you applied patches from bug 814556 and bug 915604?  Could you detail the error messages?

Comment 8

[Askeing Yen[:askeing][:fyen]]

No, TWCI doesn't applied patches.

error message:
http://release1-qa.corp.tpe1.mozilla.com:8080/view/All.B2G.Build/job/B2G.v1.2.0.flatfish/25/consoleFull

Comment 9

[Askeing Yen[:askeing][:fyen]]

There is a issue Bug 929854 for creating the mater flatfish config profile.
After we have the config profile, then TWCI can build master flatfish.

Comment 10

[Askeing Yen[:askeing][:fyen]]

depend on Bug 929854

Comment 11

[[e-mail]]

Thomas, could you help add environment variable described in Comment 2 into flatfish config profile?

Comment 12

[thomas tsai]

It will lead to build-break that nsIWifi.idl won't be parsed, so there will no nsIWifi.h be found.

Comment 13

[[e-mail]]

Vicamo, do we still miss some configurations after bug 920551 landed?

Comment 14

[thomas tsai]

It works in master, so v1.2 is waiting for the patch.

Comment 15

[shawn ku [:sku]]

Due ot Thomas already replied, clear ni?sku.

Comment 16

[Vicamo Yang [:vicamo][:vyang]]

Attachment: default-gecko-config-without-ril.global.example

This patch illustrates how to disable RIL on all b2g builds.  DO NOT COMMIT IT.

Comment 17

[Vicamo Yang [:vicamo][:vyang]]

Attachment: Android.mk.device.example

This patch demos how to disable RIL in ordinary Android.mk.  DO NOT COMMIT IT.

Comment 18

[Vicamo Yang [:vicamo][:vyang]]

So, to realize what I had in comment 2, here are examples for disabling RIL.  No additional patch is required (in master branch for now.  I'm going to uplift bug 920551 to aurora).

Comment 19

[Vicamo Yang [:vicamo][:vyang]]

(In reply to Vicamo Yang [:vicamo][:vyang] from comment #18)
> I'm going to uplift bug 920551 to aurora

Done.

Comment 20

[[e-mail]]

check with 11/1 flatfish master build, `window.navigator.mozMobileConnection` API still exist

Does there any settings we can do from gaia or gecko side to disable these API?

Comment 21

[[:fabrice] Fabrice Desré]

I think that we don't want to ship different gecko builds to different devices. So removing RIL apis from flatfish should rather be done via a pref and the appropriate webidl functionality that let you hide apis based on a pref.

Comment 22

[Vicamo Yang [:vicamo][:vyang]]

(In reply to Fabrice Desré [:fabrice] from comment #21)
> I think that we don't want to ship different gecko builds to different
> devices. So removing RIL apis from flatfish should rather be done via a pref
> and the appropriate webidl functionality that let you hide apis based on a
> pref.

WebIDL is awesome.  WebIDL is the final solution.  But it just can't be NOW.

And please note Flatfish doesn't even build Gecko from Mozilla's hg/git tree.  There is another git tree hosted on bitbucket where not everyone has access.  That's an even more serious problem.

Comment 23

[[:fabrice] Fabrice Desré]

(In reply to Vicamo Yang [:vicamo][:vyang] from comment #22)
> (In reply to Fabrice Desré [:fabrice] from comment #21)
> > I think that we don't want to ship different gecko builds to different
> > devices. So removing RIL apis from flatfish should rather be done via a pref
> > and the appropriate webidl functionality that let you hide apis based on a
> > pref.
> 
> WebIDL is awesome.  WebIDL is the final solution.  But it just can't be NOW.

Why? we use it with many apis already as you know.

> And please note Flatfish doesn't even build Gecko from Mozilla's hg/git
> tree.  There is another git tree hosted on bitbucket where not everyone has
> access.  That's an even more serious problem.

Terrible indeed. I don't see how we can provide any support in these conditions.

Comment 24

[Vicamo Yang [:vicamo][:vyang]]

(In reply to Fabrice Desré [:fabrice] from comment #23)
> (In reply to Vicamo Yang [:vicamo][:vyang] from comment #22)
> > (In reply to Fabrice Desré [:fabrice] from comment #21)
> > WebIDL is awesome.  WebIDL is the final solution.  But it just can't be NOW.
> 
> Why? we use it with many apis already as you know.

Just not in CellBroadcast/Icc/MobileConnection/MobileMessage/Voicemail yet.  It's true some event targets had been converted, but not all.  Besides, I don't want to interfere DSDS work, either.

Comment 25

[[e-mail]]

Checked b2g-manifest and I think it use standard gaia/gecko from https://git.mozilla.org/releases

https://bitbucket.org/thomastsai/a31-b2g-manifest/src/d8899d7620670bb7ebc7316640978fc4f9530b78/base-jb.xml?at=master

They did build Gecko/gaia from Mozilla's hg/git tree

Comment 26

[Vicamo Yang [:vicamo][:vyang]]

(In reply to Fred Lin [:gasolin] from comment #25)
> They did build Gecko/gaia from Mozilla's hg/git tree

Only after bug 929854.  And there is still https://bitbucket.org/thomastsai/a31-b2g-manifest/src/d8899d762067/flatfish.xml#cl-23 , a hidden patch pool to Gecko that require some membership to access.

Besides, the way Gaia determines whether a API set is supported is by checking the validity of |navigator.foo|.  You can do this on any platform with a customized Gecko, patched with 5 or 6 new lines that always returns null on attributes you're interested in.  You're not blocked by anything.  Never has been.

Comment 27

[Ken Chang[:kenkai]]

Hi Vicamo, this bug is yours...:)

Comment 28

[[e-mail]]

Hi vicamo, we do have some patches that are prepared and tested via your suggestion, but some reviewer ask us test on real tablet device.
That's why we're care of these platform/gecko issues :)

Comment 29

[Ken Chang[:kenkai]]

Viral, please help this bugs. Thanks.

Comment 30

[thomas tsai]

gecko and gaia of flatfish are all from mozilla hg/git.
Only few vendor specific folder like device, vendors and are from bitbucket.
Now gecko can be built from master or gaia.
Please refer to https://bugzilla.mozilla.org/show_bug.cgi?id=904377.

Comment 31

[thomas tsai]

how to build v1.2 or master for flatfish:
https://bugzilla.mozilla.org/show_bug.cgi?id=904377#c16

Comment 32

[Masatoshi Kimura [:emk]]

(In reply to thomas tsai from comment #30)
> Please refer to https://bugzilla.mozilla.org/show_bug.cgi?id=904377.

Only Mozilla employees?

Comment 33

[viral [:viralwang]]

Hi Vicamo,

we have plan to remove all gecko/gaia patches and merge it into mozilla hg/git in bug 934831.
I'm not sure we should provide a temporary gecko build for gaia testing only or consider merge it into our hg/git tree.

Any suggestion will be help, thanks.

Comment 34

[Vicamo Yang [:vicamo][:vyang]]

(In reply to thomas tsai from comment #31)
> how to build v1.2 or master for flatfish:
> https://bugzilla.mozilla.org/show_bug.cgi?id=904377#c16

[ -n "`cat config.sh |grep flatfish`" ] || echo "not available"

And I got "not available".

It's really irrelevant for me actually.  I know it might be still in development and unavailable for the public.  The reason I had this bug open is I want to remind us that we need somewhere to send the pull request.  And as far as I know now, we're not ready yet.

Comment 35

[viral [:viralwang]]

This bug should be fixed now, we already add "--disable-b2g-ril" for flatfish.
I think we can have build in TWCI server tomorrow. (both master/v1.2 should be fine in daily build today)

Comment 36

[[:fabrice] Fabrice Desré]

(In reply to viral [:viralwang] from comment #35)
> This bug should be fixed now, we already add "--disable-b2g-ril" for
> flatfish.
> I think we can have build in TWCI server tomorrow. (both master/v1.2 should
> be fine in daily build today)

I'll bring this up in the appropriate mailing lists, but I still thinks that having a custom gecko build is really not what we want. We have time to fix CellBroadcast/Icc/MobileConnection/MobileMessage/Voicemail before we ship.

Comment 37

[steve]

compile error after add ac_add_options --disable-b2g-ril in  default-gecko-config

gecko/dom/bluetooth/bluedroid/BluetoothHfpManager.cpp:20:27: fatal error: nsIDOMIccInfo.h: No such file or directory

master branch

Comment 38

[Vicamo Yang [:vicamo][:vyang]]

(In reply to steve from comment #37)
> compile error after add ac_add_options --disable-b2g-ril in 
> default-gecko-config
> 
> gecko/dom/bluetooth/bluedroid/BluetoothHfpManager.cpp:20:27: fatal error:
> nsIDOMIccInfo.h: No such file or directory
> 
> master branch

Filed new bug 943758.