994463 - [Flame][Build][RIL] After flashing gecko, RIL part doesn't work.

Status: VERIFIED FIXED

(Firefox OS Graveyard :: Vendcom, defect)

Description

[Al Tsai [:atsai]]

After flashing gecko, RIL part doesn't work.

If we use the image provided by partner, RIL work correctly. There should be something changed in partner build and didn't feedback to Mozilla branch.

Comment 1

[Al Tsai [:atsai]]

Hi Francis, could you please help on it? Thanks in advance :)

Comment 2

[Francis Lee [:frlee]]

hi Yanan,

we need your help to sort out if there's any patches you added in gecko. the problem is, once we replace our own gaia/gecko, gsm is not working at all.

please work with Viral to resolve this issue.

Comment 3

[shawn ku [:sku]]

What does "RIL part doesn't work" mean?

Could you please share with us radio/device log to check the issue?

Comment 4

[Ken Chang[:kenkai]]

Currently, the commercial RIL is used in flame release. Do you want to enable Moz RIL in flame?

Comment 5

[shawn ku [:sku]]

Attachment: 994463 radio/device log

The behavior looks weird after flash Moz Gecko.
There is no RADIO_STATE change, nor RADIO_POWER request,
All we can see is RILD get connected.

Comment 6

[viral [:viralwang]]

it looks like gecko/gaia 1.3 works but master don't work :(
It may not need partner support in this case.

Comment 7

[Jason Smith [:jsmith]]

Clearing nom - vendor issues won't block.

Comment 8

[Naoki Hirata :nhirata (please use needinfo instead of cc)]

I think there's a change in bug 994246 that might affect this?

Comment 9

[Naoki Hirata :nhirata (please use needinfo instead of cc)]

oorrrr... maybe not. sorry.  please ignore comment 8.

Comment 10

[Tony Zhao]

Attachment: Manifest file for get code

Hi All,

T2M didn’t change Gecko on our side,  these codes are from git://codeaurora.org/, please refer to attached Manifest file.

Comment 11

[:gerard-majax]

Not sure if it's the bug I experienced, but building and flashing for Flame, I got only one SIM to be recognized.

Comment 12

[Ken Chang[:kenkai]]

(In reply to Alexandre LISSY :gerard-majax from comment #11)
> Not sure if it's the bug I experienced, but building and flashing for Flame,
> I got only one SIM to be recognized.
DSDS properties for Moz RIL are not added in flame. Please add following properties in release image of flame.
[ro.moz.ril.0.network_types]: [gsm,wcdma]
[ro.moz.ril.1.network_types]: [gsm]
[ro.moz.ril.numclients]: [2]

Comment 13

[Ken Chang[:kenkai]]

(In reply to Tony Zhao from comment #10)
> Created attachment 8405201 [details]
> Manifest file for get code
> 
> Hi All,
> 
> T2M didn’t change Gecko on our side,  these codes are from
> git://codeaurora.org/, please refer to attached Manifest file.
I think this problem comes from the interface changes between 1.3 and master. If we want to make flame work with current m-c, T2M needs to have a new commercial RIL including the interface changes of m-c or we need to enable mozRIL in flame.

Comment 14

[:gerard-majax]

(In reply to Ken Chang[:ken] from comment #12)
> (In reply to Alexandre LISSY :gerard-majax from comment #11)
> > Not sure if it's the bug I experienced, but building and flashing for Flame,
> > I got only one SIM to be recognized.
> DSDS properties for Moz RIL are not added in flame. Please add following
> properties in release image of flame.
> [ro.moz.ril.0.network_types]: [gsm,wcdma]
> [ro.moz.ril.1.network_types]: [gsm]
> [ro.moz.ril.numclients]: [2]

Okay, with those properties set, I do get both slots visibles, but then no SIM card is detected at all.

Comment 15

[Ken Chang[:kenkai]]

(In reply to Alexandre LISSY :gerard-majax from comment #14)
> Okay, with those properties set, I do get both slots visibles, but then no
> SIM card is detected at all.
What gecko version did you used? I guess that bug 988979 occurred.

Comment 16

[:gerard-majax]

(In reply to Ken Chang[:ken] from comment #15)
> (In reply to Alexandre LISSY :gerard-majax from comment #14)
> > Okay, with those properties set, I do get both slots visibles, but then no
> > SIM card is detected at all.
> What gecko version did you used? I guess that bug 988979 occurred.

Master, built from scratch reusing blobs from image version 10D. I don't think bug 988979 plays a role a all: airplane mode has no impact, and I don't even see the two SIM slots.

If bug 988979 is at play, then it would be bug 976497 as airplane mode does nothing.

Comment 17

[Ken Chang[:kenkai]]

I find that the rilproxy isn't in system image. Please add rilproxy in system image.
And add following commands in init.rc
#begin
service rilproxy /system/bin/rilproxy
    class main
    socket rilproxy stream 660 root system
    user root
    group radio
#end

Comment 18

[:gerard-majax]

(In reply to Ken Chang[:ken] from comment #17)
> I find that the rilproxy isn't in system image. Please add rilproxy in
> system image.
> And add following commands in init.rc
> #begin
> service rilproxy /system/bin/rilproxy
>     class main
>     socket rilproxy stream 660 root system
>     user root
>     group radio
> #end

I'll try, but it means our buildsystem is severely broken then.

Comment 19

[Ken Chang[:kenkai]]

(In reply to Alexandre LISSY :gerard-majax from comment #18)
> > And add following commands in init.rc
> > #begin
> > service rilproxy /system/bin/rilproxy
> >     class main
> >     socket rilproxy stream 660 root system
> >     user root
> >     group radio
> > #end
> 
> I'll try, but it means our buildsystem is severely broken then.
This message is for T2M. Since this modification need to be done in boot image.

Comment 20

[Tony Zhao]

Hi Ken& Francis, 

As internal estimate of adding this modification, may cause some side effect like not able to detect SIM cards. so that we not suggest to adding this on official version.
If you really need to verify it, we can make a temp version with it.
Thanks

Comment 21

[Ken Chang[:kenkai]]

(In reply to Tony Zhao from comment #20)
> Hi Ken& Francis, 
> 
> As internal estimate of adding this modification, may cause some side effect
> like not able to detect SIM cards. so that we not suggest to adding this on
> official version.
> If you really need to verify it, we can make a temp version with it.
> Thanks
You don't need to include ril_proxy in your office image. To add ril_proxy is only for test. But you have to add these shell commands in the init file in official release image.

Comment 22

[Michael Wu [:mwu]]

(In reply to Ken Chang[:ken] from comment #21)
> You don't need to include ril_proxy in your office image. To add ril_proxy
> is only for test. But you have to add these shell commands in the init file
> in official release image.

Ken, rilproxy is already in the init configuration. See init.b2g.rc. We control the contents of that file.

What we need here is to figure out what quirks in RIL are different this time and make sure mozril accounts for those differences.

Comment 23

[Ken Chang[:kenkai]]

(In reply to Michael Wu [:mwu] from comment #22)
> (In reply to Ken Chang[:ken] from comment #21)
> > You don't need to include ril_proxy in your office image. To add ril_proxy
> > is only for test. But you have to add these shell commands in the init file
> > in official release image.
> 
> Ken, rilproxy is already in the init configuration. See init.b2g.rc. We
> control the contents of that file.
I had tested current flame image. After pushing rilproxy and reboot flame, rilproxy wasn't executed. It seems flame doesn't use the init.b2g.rc. Moreover, the socket connecting to RILD of flame need to have root privilege, which also causes mozril being not able to run on flame.

Comment 24

[Tony Chung [:tchung]]

can we get this prioritized and fixed asap?   We have received our first shipment of Flame devices in MTN View, and plan to distribute these devices next week.   Without the proper gecko patch to detect the RIL, many engineers will be stuck without a working RIL after flashing mozilla builds.

Thanks!

ni? mwu since he seems to have a solution on how to proceed.

Comment 25

[Francis Lee [:frlee]]

T2M has provided a test build which includes properties/shell commands requested by Ken.
https://mozilla.app.box.com/files/0/f/1708478428/Foxfone-One
Foxfone-One_V10E-rilproxy-black.zip

Comment 26

[Tony Chung [:tchung]]

(In reply to Francis Lee [:frlee] from comment #25)
> T2M has provided a test build which includes properties/shell commands
> requested by Ken.
> https://mozilla.app.box.com/files/0/f/1708478428/Foxfone-One
> Foxfone-One_V10E-rilproxy-black.zip

Was this test build supposed to work out of box, by just applying gecko over it?  if so, i'm still getting no SIM detected after applying gecko over the v10E-rilproxy-black base image.

Environment
Gaia      bbc76bcdc937bdbaf3935351ba2d532449c63bf8
Gecko     https://hg.mozilla.org/mozilla-central/rev/33ca5e321046
BuildID   20140423160202
Version   31.0a1
ro.build.version.incremental=81
ro.build.date=Wed Apr 23 16:48:30 CST 2014



Logcat
*****
Tonys-MacBook-Pro:B2G-flash-tool tchung$ adb logcat -b radio
D/RILD    (  332): **RIL Daemon Started**
D/RILD    (  332): **RILd param count=1**
D/RILD    (  332): RIL Client Id:=0
D/RILD    (  335): **RIL Daemon Started**
D/RILD    (  335): **RILd param count=3**
D/RILD    (  335): RIL Client Id:=1
D/RILD    (  332): RIL_Init argc = 5 clientId = 0
D/RILD    (  335): RIL_Init argc = 5 clientId = 1
E/RILC    (  332): RIL_register: RIL version 8
W/RILC    (  335): RIL_onUnsolicitedResponse called before RIL_register
E/RILC    (  335): RIL_register: RIL version 8
E/RILQ    (  332): (0/332): RIL[0][main] qcril_qmi_init_core_client_handles: qcril_qmi_presence_init returned failure(31)
E/RILQ    (  335): (1/335): RIL[1][main] qcril_qmi_init_core_client_handles: qcril_qmi_presence_init returned failure(31)
E/RILQ    (  332): (0/332): RIL[0][main] qcril_qmi_nas_init: QCRIL_OPERATOR_NAME is persist.env.spec,  property_name is persist.env.spec. Failed to get operator name
E/RILQ    (  335): (1/335): RIL[1][main] qcril_qmi_nas_init: QCRIL_OPERATOR_NAME is persist.env.spec,  property_name is persist.env.spec. Failed to get operator name
E/RILQ    (  332): (0/332): RIL[0] get_ecc_property_name: ecc list name : ril.ecclist.
E/RILQ    (  332): (0/332): RIL[0] get_ecc_property_name: ecc list name : ril.ecclist.
E/RILQ    (  335): (1/335): RIL[1] get_ecc_property_name: ecc list name : ril.ecclist1.
E/RILQ    (  332): (0/332): RIL[0][event] qcril_uim_update_pbm_card_event: PBM_HANDLE_CARD 397313 processing Failed!
E/RILQ    (  335): (1/335): RIL[1][event] qcril_uim_update_pbm_card_event: PBM_HANDLE_CARD 397313 processing Failed!
E/RILQ    (  335): (1/335): RIL[1][qmi_cb] get_ecc_property_name: ecc list name : ril.ecclist1.
E/RILQ    (  332): (0/332): RIL[0][qmi_cb] get_ecc_property_name: ecc list name : ril.ecclist.
E/RILQ    (  335): (1/335): RIL[1][event] qcril_uim_update_pbm_card_event: PBM_HANDLE_CARD 397313 processing Failed!
E/RILQ    (  332): (0/332): RIL[0][event] qcril_uim_update_pbm_card_event: PBM_HANDLE_CARD 397313 processing Failed!
I/RILC    (  332): libril: new connection
I/RILC    (  332): RIL Daemon version: Qualcomm RIL 1.0
E/RILQ    (  332): (0/332): RIL[0][qmi_cb] get_ecc_property_name: ecc list name : ril.ecclist.
E/RILQ    (  335): (1/335): RIL[1][qmi_cb] get_ecc_property_name: ecc list name : ril.ecclist1.
E/RILQ    (  332): (0/332): RIL[0][qmi_cb] get_ecc_property_name: ecc list name : ril.ecclist.
E/RILQ    (  332): (0/332): RIL[0][event] qcril_qmi_nas_report_restricted_details: .. reporting 18
E/RILQ    (  332): (0/332): RIL[0][main] qcril_qmi_nas_report_restricted_details: .. reporting 18

Comment 27

[Francis Lee [:frlee]]

> Was this test build supposed to work out of box, by just applying gecko over
> it?  if so, i'm still getting no SIM detected after applying gecko over the
> v10E-rilproxy-black base image.
> 

No, this test build is for Ken's team to enable/debug Moz' RIL.

Comment 28

[Tony Chung [:tchung]]

(In reply to Francis Lee [:frlee] from comment #27)
> > Was this test build supposed to work out of box, by just applying gecko over
> > it?  if so, i'm still getting no SIM detected after applying gecko over the
> > v10E-rilproxy-black base image.
> > 
> 
> No, this test build is for Ken's team to enable/debug Moz' RIL.

okay sorry for the noise.  i got a bit overexcited.

Comment 29

[Ken Chang[:kenkai]]

I had enabled MozRIL in flame and showed it to AI and Flame. Here are summary what we need to do for flame.
A. T2M need to add following things in boot image of flame.
1) Add following properties.
  [ro.moz.ril.0.network_types]: [gsm,wcdma]
  [ro.moz.ril.1.network_types]: [gsm]
  [ro.moz.ril.numclients]: [2]
2) Add one shell command in init.rc.
  symlink /dev/socket/rild1 /dev/socket/rilproxy1  
=> We need T2M to provide a official image for verification.

B. Moz need to add some interfaces for supporting flame device in MozRIL. Here are the bugs, bug 936325 and bug 1001352.

Comment 30

[:gerard-majax]

(In reply to Ken Chang[:ken] from comment #29)
> I had enabled MozRIL in flame and showed it to AI and Flame. Here are
> summary what we need to do for flame.
> A. T2M need to add following things in boot image of flame.
> 1) Add following properties.
>   [ro.moz.ril.0.network_types]: [gsm,wcdma]
>   [ro.moz.ril.1.network_types]: [gsm]
>   [ro.moz.ril.numclients]: [2]
> 2) Add one shell command in init.rc.
>   symlink /dev/socket/rild1 /dev/socket/rilproxy1  
> => We need T2M to provide a official image for verification.

I took a v10E base image, ontop of which I flashed a freshly baked master gecko and gaia. Hacked the b2g.sh script to make sure both the properties and the symlink are there, and my device still sees no SIM at all.

Not talking about the state of the device after flash all partitions from a build I did myself.

Comment 31

[:gerard-majax]

(In reply to Alexandre LISSY :gerard-majax from comment #30)
> (In reply to Ken Chang[:ken] from comment #29)
> > I had enabled MozRIL in flame and showed it to AI and Flame. Here are
> > summary what we need to do for flame.
> > A. T2M need to add following things in boot image of flame.
> > 1) Add following properties.
> >   [ro.moz.ril.0.network_types]: [gsm,wcdma]
> >   [ro.moz.ril.1.network_types]: [gsm]
> >   [ro.moz.ril.numclients]: [2]
> > 2) Add one shell command in init.rc.
> >   symlink /dev/socket/rild1 /dev/socket/rilproxy1  
> > => We need T2M to provide a official image for verification.
> 
> I took a v10E base image, ontop of which I flashed a freshly baked master
> gecko and gaia. Hacked the b2g.sh script to make sure both the properties
> and the symlink are there, and my device still sees no SIM at all.

As far as I could debug, the RIL on the Gecko side was seeing SIM cards and being able to request their ICC ID. But no iccdetected event ever triggered back to Gaia.

And the hack from bug 976497 does not help.

Comment 32

[Ken Chang[:kenkai]]

(In reply to Alexandre LISSY :gerard-majax from comment #31)
> 
> As far as I could debug, the RIL on the Gecko side was seeing SIM cards and
> being able to request their ICC ID. But no iccdetected event ever triggered
> back to Gaia.
> 
> And the hack from bug 976497 does not help.
Unfortunately, there are some proprietary dsds functions in flame. We need to add the functions of activating SIM, bug 936325, and enabling data service, bug 1001352 for flame DSDS. Then, and only then, the Moz RIL is able to be enabled in flame.

Comment 33

[Ken Chang[:kenkai]]

We will use bug 1003011 instead of bug 936325 for tracking the implementing status of activating SIM function.

Comment 34

[Chris Lord [:cwiiis]]

I don't know if it's tracked, so sorry if this is just noise - applying the patches in the various blocking bugs, I get a working RIL for the Flame, but text messaging doesn't work, or works extremely intermittently. Are we aware of this/tracking this? QC-RIL on 1.3 works fine in this regard. Calls (both calling and receiving) and data work fine on MozRIL.

Comment 35

[:gerard-majax]

Chris, can you try something?

In your B2G build tree, edit hardware/ril/libril/ril.cpp and switch the RILC_LOG define to 1. Then rebuild and push libril.so:
$ ./build.sh out/target/product/flame/system/lib/libril.so
$ adb root && adb wait-for-device && adb remount
$ adb push out/target/product/flame/system/lib/libril.so /system/lib/libril.so 

Then reboot the device. You will get more logs in |adb logcat -b radio| output. Please look at them when you reproduce the issue :)

Comment 36

[:gerard-majax]

In my case, when this is happening, I see:
> D/RILC    (  345): [0246]> SEND_SMS ((null),01000A816071894055000003477C1D)
> D/RILC    (  345): [0246]< SEND_SMS fails by E_GENERIC_FAILURE

Comment 37

[Jason Smith [:jsmith]]

Can't be a QA blocker because it's not a regression.

Comment 38

[Tony Chung [:tchung]]

After speaking with ken chang, we still need vendor to push a fix on the Flame kernel side to allow the RIL bits to be seen from mozilla's build process.  Just enabling bug 1003011 is not enough if we want to build daily mozilla builds for testing.

ni? ken and francis to communicate this to vendor

Comment 39

[Ken Chang[:kenkai]]

Yestoday, vendor had provided a test image and it worked. Francis is talking with T2M to get the official release. Maybe, Francis can provide more information for that.

Comment 40

[Francis Lee [:frlee]]

https://mozilla.app.box.com/files/0/f/1708478428/Foxfone-One
Foxfone-One_V10F-3_BlackforMozRIL.zip

this is the test image T2M provided. now there's one SMS issue under clarification. once its done, official image will be provided by T2M

Comment 41

[Tony Chung [:tchung]]

(In reply to Francis Lee [:frlee] from comment #40)
> https://mozilla.app.box.com/files/0/f/1708478428/Foxfone-One
> Foxfone-One_V10F-3_BlackforMozRIL.zip
> 
> this is the test image T2M provided. now there's one SMS issue under
> clarification. once its done, official image will be provided by T2M

Thanks Francis.  please keep us updated.   this is urgently needed for any flame user that requires telephony.

Comment 42

[Tony Chung [:tchung]]

(In reply to Tony Chung [:tchung] from comment #41)
> (In reply to Francis Lee [:frlee] from comment #40)
> > https://mozilla.app.box.com/files/0/f/1708478428/Foxfone-One
> > Foxfone-One_V10F-3_BlackforMozRIL.zip
> > 
> > this is the test image T2M provided. now there's one SMS issue under
> > clarification. once its done, official image will be provided by T2M
> 
> Thanks Francis.  please keep us updated.   this is urgently needed for any
> flame user that requires telephony.

Hi Francis, what's the ETA on T2M delivering an updated image to us with this fix?    Can we get this tomorrow?

Comment 43

[Tony Chung [:tchung]]

ni? vance since francis is on PTO

Comment 44

[Vance Chen [:vchen][skype:vance.lucida][[e-mail]]]

Later today(5/22), T2M will provide a new build that fix the wrong properties for MOZ RIL problem. Will update the image link here later.

Comment 45

[Tony Chung [:tchung]]

Hi Vance,

Doing a gecko/ gaia flash over the latest T2M build (v10G-2), is now detecting the RIL and DSDS again.   

I can make a phone call, and connect to the web via Data.   however, sending SMS is not working yet.   Need to investigate further why.

Thanks,
Tony

Comment 46

[Anthony Ricaud (:rik)]

(In reply to Tony Chung [:tchung] from comment #45)
> however, sending SMS is not working yet.
Alexandre Lissy told me that turning airplane mode on and off will let you send SMS after.

Comment 47

[Ken Chang[:kenkai]]

This bug is only to enable mozRil in flame. The original purpose of this bug is fixed by T2M. I'd like to close this bug. 

Because Flame is the first DSDS device with QC platform, we may still face a lot of bugs in main trunk after enabling mozRil in flame. Therefore, if you find any bug, please file a bug and make it a blocker of bug 1009139. Afterwords, lets use bug 1009139 to track all bugs related to flame.

Comment 48

[Tony Chung [:tchung]]

Verified fix on v10G-2 image.   DSDS and RIL is working for 1.4 and 2.0 gecko builds.  Thanks!