tracking: canary doesn't compile on s390 and ppcle

[refack]

• Version: canary
• Platform: s390x and ubuntu@ppcle
• Subsystem: V8

s390x:

15:48:02 In file included from ../../src/compiler/backend/s390/code-generator-s390.cc:14:0:
15:48:02 ../../src/s390/macro-assembler-s390.h:6:2: error: #error This header must be included via macro-assembler.h
15:48:02  #error This header must be included via macro-assembler.h
15:48:02   ^

ubuntu@ppcle

15:48:41 FAILED: obj/v8_base/code-generator-ppc.o 
15:48:41 g++ -MMD -MF obj/v8_base/code-generator-ppc.o.d -DV8_DEPRECATION_WARNINGS -DUSE_UDEV -DUSE_AURA=1 -DUSE_GLIB=1 -DUSE_NSS_CERTS=1 -DUSE_X11=1 -DFULL_SAFE_BROWSING -DSAFE_BROWSING_CSD -DSAFE_BROWSING_DB_LOCAL -DCHROMIUM_BUILD -DFIELDTRIAL_TESTING_ENABLED -D_FILE_OFFSET_BITS=64 -D_LARGEFILE_SOURCE -D_LARGEFILE64_SOURCE -D__STDC_CONSTANT_MACROS -D__STDC_FORMAT_MACROS -DNDEBUG -DNVALGRIND -DDYNAMIC_ANNOTATIONS_ENABLED=0 -DV8_TYPED_ARRAY_MAX_SIZE_IN_HEAP=64 -DENABLE_GDB_JIT_INTERFACE -DENABLE_MINOR_MC -DV8_DEPRECATION_WARNINGS -DV8_IMMINENT_DEPRECATION_WARNINGS -DV8_INTL_SUPPORT -DENABLE_HANDLE_ZAPPING -DV8_USE_SNAPSHOT -DV8_USE_EXTERNAL_STARTUP_DATA -DV8_CONCURRENT_MARKING -DV8_EMBEDDED_BUILTINS -DV8_EMBEDDED_BYTECODE_HANDLERS -DV8_TARGET_ARCH_PPC -DV8_TARGET_ARCH_PPC64 -DV8_TARGET_ARCH_PPC_LE -DDISABLE_UNTRUSTED_CODE_MITIGATIONS -DU_USING_ICU_NAMESPACE=0 -DU_ENABLE_DYLOAD=0 -DUSE_CHROMIUM_ICU=1 -DU_STATIC_IMPLEMENTATION -DICU_UTIL_DATA_IMPL=ICU_UTIL_DATA_FILE -DUCHAR_TYPE=uint16_t -I../.. -Igen -I../.. -Igen -I../../third_party/icu/source/common -I../../third_party/icu/source/i18n -I../../include -fno-strict-aliasing --param=ssp-buffer-size=4 -fstack-protector -Wno-builtin-macro-redefined -D__DATE__= -D__TIME__= -D__TIMESTAMP__= -funwind-tables -fPIC -pipe -pthread -m64 -Wall -Werror -Wno-unused-local-typedefs -Wno-maybe-uninitialized -Wno-deprecated-declarations -Wno-comments -Wno-missing-field-initializers -Wno-unused-parameter -fno-omit-frame-pointer -g0 -fvisibility=hidden -Wno-strict-overflow -Wno-return-type -O3 -fno-ident -fdata-sections -ffunction-sections -std=gnu++14 -Wno-narrowing -fno-exceptions -fno-rtti -fvisibility-inlines-hidden -c ../../src/compiler/backend/ppc/code-generator-ppc.cc -o obj/v8_base/code-generator-ppc.o
15:48:41 In file included from ../../src/compiler/backend/ppc/code-generator-ppc.cc:15:0:
15:48:41 ../../src/ppc/macro-assembler-ppc.h:6:2: error: #error This header must be included via macro-assembler.h
15:48:41  #error This header must be included via macro-assembler.h
15:48:41   ^

[mhdawson]

@john-yan can you take a look.

[john-yan]

Please wait until these 4 changes get in:
https://chromium-review.googlesource.com/c/v8/v8/+/1342738
https://chromium-review.googlesource.com/c/v8/v8/+/1342977
https://chromium-review.googlesource.com/c/v8/v8/+/1342661
https://chromium-review.googlesource.com/c/v8/v8/+/1342978

remaining ppc issue has to be waited until:
https://chromium-review.googlesource.com/c/v8/v8/+/1336110

[refack]

To reiterate this is just a tracking issue, and is related to our canary build. If you are already dealing with it, it would be enough if you just keep up in the loop.
FTR: Still relevant as of last night with V8 7.2.378

P.S. I just saw your reply @john-yan , thank you very much for the context.

[mhdawson]

@refack is this still an issue?

[mhdawson]

Actually, I'll answer that myself. I think the answer is yes. The originally reported issues were fixed but there have been new issues since them.

I'm wondering if this should be closed or if we should keep it. I think a lot of the time canary will not be working for those platforms as canary is the bleeding edge and we are always catching up.

[refack]

I'm wondering if this should be closed or if we should keep it. I think a lot of the time canary will not be working for those platforms as canary is the bleeding edge and we are always catching up.

I'd say it's your (s390 & linux@ppc team's) call.

[refack]

Wrong button (the time off made me rusty)
I'm thinking maybe there's a wider conversation to be had. Maybe we want to have a different CI/CR flow for these platforms....

[mhdawson]

I think everything is fine except for Canary. We might want to exclude them from the Canary builds but they should be good for everything else.

[targos]

Do we care to know if/when they fail on canary? If not, I can remove them from the test job.

[mhdawson]

@john-yan what's your take. If they are not going to be passing more than 50% of the time it might be better to just remove them.

@targos since this would just be for canary but its the same job for release we'd have to be careful that they are only removed for the canary builds.

[john-yan]

I don't think it's necessary to track master branch either.

[targos]

@mhdawson It's a different job: https://ci.nodejs.org/view/MyJobs/job/node-test-commit-node-v8/

[mhdawson]

Ok I was thinking of the nightly canary builds. I'm fine removing it from the job testing canary v8 with Node. While I'd like our platforms to be there, at this point I think it adds more noise than benefit.

[targos]

Ok, I removed aix, plinux and linuxone from the job. I think we can keep the release job as it is, so we get a chance to have canary builds on those platforms when compilation works.

[mhdawson]

@targos + 1

[targos]

@refack Is this resolved from your POV?

[refack]

this cycle was resolved (and another one since then)