1173295 - [Flame L] merge patches for those fork from t2m branch

Status: RESOLVED FIXED

(Firefox OS Graveyard :: GonkIntegration, defect)

Description

[viral [:viralwang]]

There are some repositories fork from t2m/caf need some patches for b2g:

external/bluetooth/bluedroid
frameworks/av
frameworks/base
hardware/libhardware
kernel
system/core

Since most of patches already land in other branches, I think I can collect all of them in this bug.

Comment 1

[viral [:viralwang]]

Attachment: PR for kernel

as we did in bug 1086987, we need to enable seccomp and fanotify for flame-l

Comment 2

[viral [:viralwang]]

Attachment: PR for system/core

we need to make sure b2g binaries executable in system/core.

Comment 3

[viral [:viralwang]]

Attachment: PR for hardware/libhardware

refer to this commit from CAF to make it build pass
https://www.codeaurora.org/cgit/external/gigabyte/platform/hardware/libhardware/commit/?h=caf/LF.BR.1.2.3&id=5fd7bf8d840afb3f0e4f38b8b9879427af30922f

Comment 4

[viral [:viralwang]]

Attachment: PR for external/bluetooth/bluedroid

refer the commit from CAF:
https://www.codeaurora.org/cgit/external/gigabyte/platform/external/bluetooth/bluedroid/commit/?h=caf/LF.BR.1.2.3&id=1bd2b0420e78060c3f8fd236179bb9391de34f96

Comment 5

[viral [:viralwang]]

Attachment: PR for frameworks/base

Same as bug 1104239, we need to remove fonts makefile.

Comment 6

[Michael Wu [:mwu]]

Comment on attachment 8620872 [details] [review]
PR for kernel

This doesn't look quite right.

The branch name doesn't match the existing naming scheme.

The commits that are being merged shouldn't be there, because they should have already existed when you branched. This probably means you're branching from an incorrect point. Generally we should start from a commit tagged on CAF and add any changes on top of that. All the seccomp backports are provided by CAF, not T2M.

Comment 7

[Michael Wu [:mwu]]

This port looks like it started from the wrong point. You need to start from the appropriate CAF tag as a base for any additional changes. Our b2g-* ports are only used as a reference for SoC vendors and for AOSP ports. CAF has their own set of B2G port changes. The only thing that we override is platform_build.

Comment 8

[Michael Wu [:mwu]]

If there are no additional changes you need beyond CAF's branch, point the manifest to the appropriate tag. We only need to make forks when we need to make changes that upstream won't take or can't take in time.

Comment 9

[viral [:viralwang]]

Hi Michael,

It will be best if we can have CAF tag for L. However, CAF won't provide L version with b2g for msm8x10 chipset. (or we can wait/ask for b2g_l_3.5?)
All QCT works on L are base on msm8909 with 5.1, and QCT has officially stopped L on msm8x10 and we can only have flame with 5.0. That's why we can't use the same tag as msm8909 b2g since there are something kernel not support like ion.

The only choice we have is to use the AOSP tag provided by t2m for kernel and backport those necessary patches.
For other repositories, we also need to use the AOSP tag (it could be t2m/CAF) and merge patches again like we did for Nexus-5 L.

Comment 10

[Michael Wu [:mwu]]

Got it, makes sense. Then Flame L doesn't necessarily seem like something that people should upgrade to. It sounds like more of an option for people who need to test L and have a Flame.

Comment 11

[Francis Lee [:frlee]]

the main reason to get Flame L on PVT is due to DSDS and NFC testing. QA actually raised the need to have Flame L PVT. On top of that, so far, Flame is still the only device which supports both DSDS and NFC functions + people can easily get a Flame. i think it is worthy to put it on PVT, since we have almost prepared all it requires (config, manifest, etc).

Comment 12

[viral [:viralwang]]

Attachment: PR for frameworks/av

Hi Micheal,

I refer the code in https://www.codeaurora.org/cgit/quic/la/platform/frameworks/av/commit/?h=LF.BR.1.2.3&id=a4b9ff7252272cedff593937e8e6dab842bd4209 , we have to revert the commit to make it build pass.

Comment 13

[viral [:viralwang]]

Comment on attachment 8620872 [details] [review]
PR for kernel

Hi Michael,

As our discussion in comment 9, this PR still need your help to review.
Thank you.

Comment 14

[Michael Wu [:mwu]]

(In reply to Francis Lee [:frlee] from comment #11)
> the main reason to get Flame L on PVT is due to DSDS and NFC testing. QA
> actually raised the need to have Flame L PVT. On top of that, so far, Flame
> is still the only device which supports both DSDS and NFC functions + people
> can easily get a Flame. i think it is worthy to put it on PVT, since we have
> almost prepared all it requires (config, manifest, etc).

Yeah that's fine. It's just that there's not a whole lot of value here compared to our other ports, but the work is done so I'll stamp it. It's just a really odd port (and I didn't realize when reviewing). It supports 5.0, when we want to focus on 5.1, and it does it without official chipset support and releases from QC.

Comment 15

[viral [:viralwang]]

https://github.com/mozilla-b2g/codeaurora_kernel_msm/commits/foxfone-one-lollipop
https://github.com/mozilla-b2g/platform_system_core/commits/foxfone-one-lollipop
https://github.com/mozilla-b2g/platform_hardware_libhardware/commits/foxfone-one-lollipop
https://github.com/mozilla-b2g/platform_external_bluetooth_bluedroid/commits/foxfone-one-lollipop
https://github.com/mozilla-b2g/platform_frameworks_base/commits/foxfone-one-lollipop
https://github.com/mozilla-b2g/platform_frameworks_av/tree/LA.BF.1.1.2_rb1.12