1579834 - [Webrtc] Add mips64el build support

Status: RESOLVED FIXED

(Core :: WebRTC, defect, P5)

Description

[Wang Qing]

Attachment: build-mips64el.log

User Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10_14_0) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/62.0.3189.0 Safari/537.36

Steps to reproduce:

1. get inbound code
2. ./mach configure
3. ./mach build

Actual results:

/home/loongson/gecko-dev/media/webrtc/trunk/webrtc/api/audio_codecs/isac/audio_decoder_isac.h:19:2: error: "Must choose either fix or float"
#error "Must choose either fix or float"

Expected results:

build success

Comment 1

[Wang Qing]

Attachment: Bug-1579834-WebRTC-Add-mips64-support-with-defined-WEBRTC_CODEC_.patch

I have a patch needed to review, thanks!!!

Comment 2

[Wang Qing]

Attachment: Bug-1579834-WebRTC-Add-mips64-support-with-defined-W.patch

Sorry, this patch is ok!

Comment 3

[Mu Tao]

Attachment: Bug 1579834 - [WebRTC] Add mips64 support with defined WEBRTC_CODEC_ISAC; r=dminor

Comment 4

[Mu Tao]

Attachment: Bug 1579834 - [WebRTC] Add mips64 support; r=dminor

Comment 5

[Dan Minor [:dminor]]

Hey Chris, I think there might be something strange with the moz.build code generation with the json files attached to this bug. On my x86_64 Linux system, I end up with sandbox validation errors due to files being added to UNIFIED_SOURCES twice. The first instance is:

Source file should only be added to UNIFIED_SOURCES once: /media/webrtc/trunk/webrtc/common_audio/signal_processing/complex_bit_reverse.c

Looking at media/webrtc/trunk/webrtc/common_audio/common_audio_c_gn/moz.build, the generated file is adding it for:

if CONFIG["MOZ_DEBUG"] == "1":
...

and then again for:

if CONFIG["CPU_ARCH"] == "x86_64" and CONFIG["OS_TARGET"] == "Linux":
...

If I fix that instance of the problem I hit something similar with media/webrtc/trunk/webrtc/modules/audio_coding/isac_fix_c_gn/moz.build and media/webrtc/trunk/webrtc/modules/audio_processing/audio_processing_c_gn/moz.build, and maybe more, that was the point where I gave up.

I was wondering if you could have a look?

Comment 6

[Chris Manchester (limited bugmail, email directly)]

Of course. The issue appears to be "MOZ_DEBUG": 1, (1 as an int, not a string as in the others) in the mips64_True_mips64_linux.json file. If I change that to be a string things seem to be generated correctly.

Sorry this is so brittle, I briefly looked into how to make this raise a useful error in cases like these but it seems a little awkward. Please file another bug if you'd like me to look into that at some point. Thanks!

Comment 7

[Mu Tao]

@Chris Thanks for your advice!!!

Comment 8

[Dan Minor [:dminor]]

That fixed my local build. I've started a try job here: https://treeherder.mozilla.org/#/jobs?repo=try&revision=95b5085b651f26a77a9d568b24f29faf13401b47. As long as that looks good, this should be fine to land. Thanks for your patience.

Comment 9

[Dan Minor [:dminor]]

Would you like me to land this for you or are you waiting on the additional reviewers?

Comment 10

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/1387b2c480b5
[WebRTC] Add mips64 support; r=dminor

Comment 11

[Dorel Luca [:dluca]]

https://hg.mozilla.org/mozilla-central/rev/1387b2c480b5