1176285 - [aries-l] Update script to download vendor blobs automatically during build

Status: RESOLVED FIXED

(Firefox OS Graveyard :: GonkIntegration, defect)

Description

[Kai-Zhen Li [:seinlin][:kli]]

Accordingly to bug 1163550 comment 91, developers will be noticed to download the blobs from vendor site manually. If the download can be done automatically, it would be perfect.

Comment 1

[Kai-Zhen Li [:seinlin][:kli]]

Alexandre, could you have a look to this bug? Thanks!

Comment 2

[:gerard-majax]

Attachment: Shinano platform PR

Comment 3

[:gerard-majax]

Attachment: Leo device PR

Comment 4

[:gerard-majax]

Attachment: Aries device PR

Comment 5

[:gerard-majax]

Adam, I need your help to cross check this is working well. Thanks :)

Comment 6

[Michael Wu [:mwu]]

Hmm, can you add a check to make sure w3m is installed? curl/wget can usually be assumed, but I don't think w3m can be assumed.

Comment 7

[:gerard-majax]

(In reply to Michael Wu [:mwu] from comment #6)
> Hmm, can you add a check to make sure w3m is installed? curl/wget can
> usually be assumed, but I don't think w3m can be assumed.

Done. I have also changed the triggering a logic a little bit since a clean tree was failing completely because of make-dependency .mk files not here (thanks adam). We should now be safe.

Comment 8

[Michael Wu [:mwu]]

Is there a specific reason you're using blobs.mk to trigger the running of the blobs downloader? The blob downloader is usually automatically triggered by build.sh as long as it's named download-blobs.sh and placed in the device directory.

Comment 9

[:gerard-majax]

(In reply to Michael Wu [:mwu] from comment #8)
> Is there a specific reason you're using blobs.mk to trigger the running of
> the blobs downloader? The blob downloader is usually automatically triggered
> by build.sh as long as it's named download-blobs.sh and placed in the device
> directory.

Yeah. For now, build.sh will execute either download-blobs.sh or extract-files.sh to my understanding. But given Sony has not been granted the right to redistribute modem firmware in the ZIP file, we still need extract-files to grab those.

I'm not very happy of the blobs.mk hacks, neither is Adam or you, so I'll look forward to a better fix. That being said, configure_device() relies on load-config.sh which is ran by setup.sh to read .config for having the value of DEVICE variable.

Given the way blobs are packaged, there is a $(call ...) dependency on device's .mk ( $(PRODUCT_NAME)-vendor.mk ) which will make the lunch step fail. So we won't even execute the configure_device() function in build.sh.

That's why early port was checking the .mk in each device mk file, and why I have mutualized it at blobs.mk level.

That being said, I have just spoken with Alin who agrees that we fix upstream replacing the call inherit-product with a inherit-product-if-exists like those at https://github.com/mozilla-b2g/device-sony-aries/blob/master/aosp_d5803.mk#L23

This way we should have lunch not failing on missing blobs and we can move the download-sony-blobs.sh back into extract-files.sh.

Michael, is there any reason for exclusively running either download-blobs.sh OR extract-files.sh ? If no, we could run both by making them just symlink on the device/sony/shinano/ files on each tree.

Comment 10

[:gerard-majax]

I've updated the PRs with:
 - moving back to calling download-sony-blobs.sh from extract-files.sh
 - changing inherit-product to inherit-product-if-exists on aosp_*.mk files: I'll do PR upstream for this before landing my changes

Comment 11

[Michael Wu [:mwu]]

We've never had a device where running both would make sense till now. Doing a download in extract-files.sh seems like a reasonable compromise though.

Comment 12

[Michael Wu [:mwu]]

Comment on attachment 8635249 [details] [review]
Shinano platform PR

Looks good. There's just one nit that I put on the PR.

Comment 13

[:gerard-majax]

(In reply to Michael Wu [:mwu] from comment #12)
> Comment on attachment 8635249 [details] [review]
> Shinano platform PR
> 
> Looks good. There's just one nit that I put on the PR.

That's fixed! Thanks!

Comment 14

[:gerard-majax]

https://github.com/mozilla-b2g/device-sony-shinano/commit/8ea26484d91181dfbe08c65d6c4abdad5dc79555
https://github.com/mozilla-b2g/device-sony-leo/commit/1587a6b073c2c75f2211780f066b20a859089da5
https://github.com/mozilla-b2g/device-sony-aries/commit/67e79ae3eff129a44bb5a7d451746d8811e054c2