1506644 - -msse2 flag slips in again in non-x86 builds

Status: RESOLVED FIXED

(Core :: WebRTC, defect, P5)

Description

[Dan Horák]

After a change on Friday (Nov 09) builds fail on ppc64/ppc64le due unrecognized compiler flag

...
 0:10.09 gmake[4]: Entering directory '/mnt/dan/firefox.git/obj-powerpc64le-unknown-linux-gnu/media/webrtc/trunk/webrtc/common_audio/common_audio_sse2_gn'
 0:10.09 media/webrtc/trunk/webrtc/common_audio/common_audio_sse2_gn
 0:10.10 /usr/bin/g++ -o Unified_cpp_common_audio_sse2_gn0.o -c  -I/mnt/dan/firefox.git/obj-powerpc64le-unknown-linux-gnu/dist/stl_wrappers -I/mnt/dan/firefox.git/obj-powerpc64le-unknown-linux-gnu/dist/system_wrappers -include /mnt/dan/firefox.git/config/gcc_hidden.h -DDEBUG=1 -DCHROMIUM_BUILD -DV8_DEPRECATION_WARNINGS -DWEBRTC_ENABLE_PROTOBUF=0 -DWEBRTC_MOZILLA_BUILD -DWEBRTC_NON_STATIC_TRACE_EVENT_HANDLERS=0 -DWEBRTC_RESTRICT_LOGGING -DDYNAMIC_ANNOTATIONS_ENABLED=1 -DWTF_USE_DYNAMIC_ANNOTATIONS=1 -DUSE_NSS_CERTS=1 -DUSE_X11=1 -DWEBRTC_LINUX -DWEBRTC_POSIX -D_FILE_OFFSET_BITS=64 -DSTATIC_EXPORTABLE_JS_API -DMOZ_HAS_MOZGLUE -DMOZILLA_INTERNAL_API -DIMPL_LIBXUL -I/mnt/dan/firefox.git/media/webrtc/trunk/webrtc/common_audio/common_audio_sse2_gn -I/mnt/dan/firefox.git/obj-powerpc64le-unknown-linux-gnu/media/webrtc/trunk/webrtc/common_audio/common_audio_sse2_gn -I/mnt/dan/firefox.git/obj-powerpc64le-unknown-linux-gnu/ipc/ipdl/_ipdlheaders -I/mnt/dan/firefox.git/ipc/chromium/src -I/mnt/dan/firefox.git/ipc/glue -I/mnt/dan/firefox.git/media/webrtc/trunk/webrtc -I/mnt/dan/firefox.git/obj-powerpc64le-unknown-linux-gnu/dist/include -fPIC -DMOZILLA_CLIENT -include /mnt/dan/firefox.git/obj-powerpc64le-unknown-linux-gnu/mozilla-config.h -U_FORTIFY_SOURCE -D_FORTIFY_SOURCE=2 -Wall -Wempty-body -Wignored-qualifiers -Woverloaded-virtual -Wpointer-arith -Wsign-compare -Wtype-limits -Wunreachable-code -Wwrite-strings -Wno-invalid-offsetof -Wc++1z-compat -Wduplicated-cond -Wimplicit-fallthrough -Wno-error=maybe-uninitialized -Wno-error=deprecated-declarations -Wno-error=array-bounds -Wno-error=free-nonheap-object -Wno-error=multistatement-macros -Wno-error=class-memaccess -Wformat -Wformat-overflow=2 -fno-sized-deallocation -U_FORTIFY_SOURCE -D_FORTIFY_SOURCE=2 -fno-exceptions -fno-strict-aliasing -fno-rtti -fno-exceptions -fno-math-errno -pthread -pipe -g -Og -fno-omit-frame-pointer -funwind-tables -msse2  -MD -MP -MF .deps/Unified_cpp_common_audio_sse2_gn0.o.pp  -fdiagnostics-color  /mnt/dan/firefox.git/obj-powerpc64le-unknown-linux-gnu/media/webrtc/trunk/webrtc/common_audio/common_audio_sse2_gn/Unified_cpp_common_audio_sse2_gn0.cpp
 0:10.10 g++: error: unrecognized command line option '-msse2'; did you mean '-misel'?
 0:10.10 gmake[4]: *** [/mnt/dan/firefox.git/config/rules.mk:1134: Unified_cpp_common_audio_sse2_gn0.o] Error 1

Comment 1

[Makoto Kato [:m_kato]]

bug 1434589 and bug 1465274's JSON files aren't merged by bug 1376873. Why?

Comment 2

[Dan Minor [:dminor]]

(In reply to Makoto Kato [:m_kato] from comment #1)
> bug 1434589 and bug 1465274's JSON files aren't merged by bug 1376873. Why?

Because of changes in the webrtc.org gn files upstream, it is necessary to regenerate json files for every platform after the branch update. If you can attach updated json files for your platforms here, I'll be happy to review them. I'm sorry, I know this process is painful.

Comment 3

[Dan Horák]

Cameron, could you take a look, please? You have fixed that last time, if I see right.

Comment 4

[Dan Horák]

Or maybe I'll have a patch, seems I got the process for update working.

Comment 5

[Dan Horák]

Attachment: 0001-add-ppc64-to-webrtc-regenerate-moz.build-to-remove-S.patch

Comment 6

[Dan Minor [:dminor]]

Comment on attachment 9024755 [details] [diff] [review]
0001-add-ppc64-to-webrtc-regenerate-moz.build-to-remove-S.patch

Review of attachment 9024755 [details] [diff] [review]:
-----------------------------------------------------------------

This looks good to me. Bug 1503363 is also trying to add updated json and moz.build files. I'll land this for you once that lands to avoid conflicts with the generated moz.build files.

Comment 7

[Dan Horák]

OK, let me know if you need a refresh of the moz.build file(s).

Comment 8

[Dan Minor [:dminor]]

(In reply to Dan Horák from comment #7)
> OK, let me know if you need a refresh of the moz.build file(s).

Thank you for the offer, but I can take care of that prior to landing.

By the way, Bug 818631 tracks future webrtc.org updates. We don't have one planned at the moment (maybe Summer 2019?) but if you cc yourself there, you can get advance warning when the bug for a new update is filed and then you can get your updated json files included as part of the update itself instead of having to deal with a broken build after the fact. I'm sorry I missed cc'ing you on Bug 1376873.

Comment 9

[Dan Horák]

No problem, I've subscribed to the rebase bug. I wonder if bug 1506571 (webrtc on Linux/aarch64) needs also the JSON file + moz.build update or if it is a deeper issue.

Comment 10

[Cameron Kaiser [:spectre]]

(In reply to Dan Horák from comment #3)
> Cameron, could you take a look, please? You have fixed that last time, if I
> see right.

Sorry, I just saw this. I do have a gn tool compiled for POWER9 (at great personal exasperation), so let me know if you want a copy (E-mail me or something and I'll send you the binary).

Comment 11

[Dan Horák]

(In reply to Cameron Kaiser [:spectre] from comment #10)
> (In reply to Dan Horák from comment #3)
> > Cameron, could you take a look, please? You have fixed that last time, if I
> > see right.
> 
> Sorry, I just saw this. I do have a gn tool compiled for POWER9 (at great
> personal exasperation), so let me know if you want a copy (E-mail me or
> something and I'll send you the binary).

Thanks, but I have got gn running, learned the process and I've already produced the necessary JSON files :-)

Comment 12

[Dan Horák]

Attachment: arm64_False_arm64_linux.json

Comment 13

[Dan Horák]

Attachment: arm64_True_arm64_linux.json

Comment 14

[Dan Horák]

Attachment: arm_False_arm_linux.json

Comment 15

[Dan Horák]

Attachment: arm_True_arm_linux.json

Comment 16

[Dan Horák]

Dan, as you can see, I've uploaded the re-generated json files for Linux/aarch64 (for bug 1506571) and Linux/arm (not yet reported by me) here. Do I understand it right, that the moz.build file takes all the json files into account, so it can be generated by "GnMozbuildWriter" ex-post?

Comment 17

[Dan Minor [:dminor]]

(In reply to Dan Horák from comment #16)
> Dan, as you can see, I've uploaded the re-generated json files for
> Linux/aarch64 (for bug 1506571) and Linux/arm (not yet reported by me) here.
> Do I understand it right, that the moz.build file takes all the json files
> into account, so it can be generated by "GnMozbuildWriter" ex-post?

Yes, that is correct. If this is the complete set of json files, I can review them, generate moz.build files and check them in.

Comment 18

[Dan Horák]

yes, it is a complete set from me

Comment 19

[Dan Minor [:dminor]]

Try run here: https://treeherder.mozilla.org/#/jobs?repo=try&revision=1db044cd29604921e777c13e1af24bb1427c1cc6

Looks like the new configs caused no problems with the existing set.

Comment 20

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/mozilla-inbound/rev/94adc4dd3151
Add webrtc.org build configs for PPC and Arm Linux platforms; r=dminor

Comment 22

[Natalia Csoregi [:nataliaCs]]

https://hg.mozilla.org/mozilla-central/rev/94adc4dd3151