CPP/HOST_CPP were probably not necessary already, but now that we leave
it to configure to figure out the appropriate compiler flags, we don't
need to set HOST_CC/HOST_CXX to remove the flags from CC/CXX.
Differential Revision: https://phabricator.services.mozilla.com/D14382
Rather than manually passing -isysroot to clang. Ideally, we shouldn't
need to fill BINDGEN_CFLAGS from the mozconfig, but we're not quite
there yet.
Differential Revision: https://phabricator.services.mozilla.com/D14381
Because we now set the sysroot include flags early in python configure,
we don't need to set CPP/CXXCPP. We also skip the explicit compiler test
because more complete tests follow anyways.
Differential Revision: https://phabricator.services.mozilla.com/D14380
This matches more closely cross toolchains prefixes (as can be seen in
e.g. media/libvpx/libvpx/README for x86_64-darwin*-gcc), and leaves it
to the build system to figure out the right --target to pass to clang on
its own.
Differential Revision: https://phabricator.services.mozilla.com/D14376
We're always setting -dead_strip on mac builds, per
cross-mozconfig.common, we might as well not do that and revert bug
638149, which disabled adding -dead_strip with LTO: that is apparently
not a problem anymore.
Differential Revision: https://phabricator.services.mozilla.com/D14373
Currently, when Firefox crashes when running the PGO profile, the error
is ignored, and the result is a build that is not optimized as expected.
So instead of such things going unnoticed, we make profileserver.py
return an error code, further failing the build.
Unfortunately, this does not make the output useful to know what went
wrong exactly, but the most important thing is that the build fails
instead of going through and leading to bad results.
This removes several patches that have been upstreamed.
--HG--
extra : rebase_source : 8462b42b8342935cfb5ecb912fac06c67b9165b9
extra : amend_source : 271b5d4a5b37fa74555ef87cb0c4444f8f7468f0
Summary:
I've chosen linux64-debug since it's the most visible build I usually do, but I
could do another build task or something, or use the static analysis builds, or
what not. Just let me know if there's a better way to do this.
Caveat: This might make updating Rust toolchains a bit more painful. I think
this is better and we should just deal with warnings before updating toolchains,
but I don't know if there'd be strong opposition to that.
Note that this does _not_ affect third-party code since Cargo passes
`--cap-lint warn` automatically for those.
Proof that it works:
* https://treeherder.mozilla.org/#/jobs?repo=try&revision=4ad1e4e1392f71b574cff683e90c7b13bf8781d1
* https://treeherder.mozilla.org/#/jobs?repo=try&revision=57604f92624bbe49037eee87c56fdb6bf2b5017d
Reviewers: #firefox-build-system-reviewers, ted
Reviewed By: #firefox-build-system-reviewers, ted
Subscribers: reviewbot, glandium, ted
Bug #: 1513009
Differential Revision: https://phabricator.services.mozilla.com/D14083
But keep the crashreporter disabled at runtime because it doesn't work
yet.
This has the side effect of creating the artifacts with the
crashreporter symbols and pdbs.
Differential Revision: https://phabricator.services.mozilla.com/D14550
--HG--
extra : moz-landing-system : lando
This patch is based on the cmake cache files for Android checked in to the
clang repo.
Differential Revision: https://phabricator.services.mozilla.com/D14004
--HG--
extra : moz-landing-system : lando
This is an egregious hack. The web-platform-tests were not meant to run in the shell.
The eight tests that are included are the ones that we pass with flying colors.
In most of the others, we still have a failure or two.
Differential Revision: https://phabricator.services.mozilla.com/D11533
--HG--
rename : dom/imptests/testharness.js => js/src/jit-test/lib/w3c-testharness.js
extra : moz-landing-system : lando
This is an egregious hack. The web-platform-tests were not meant to run in the shell.
The eight tests that are included are the ones that we pass with flying colors.
In most of the others, we still have a failure or two.
Differential Revision: https://phabricator.services.mozilla.com/D11533
--HG--
rename : dom/imptests/testharness.js => js/src/jit-test/lib/w3c-testharness.js
extra : moz-landing-system : lando
This pulls a newer version of the nasm package, and patches
out doc building in order to make it compile on debian 7.
Differential Revision: https://phabricator.services.mozilla.com/D13510
--HG--
extra : moz-landing-system : lando
https://clang.llvm.org/docs/DiagnosticsReference.html#wbitfield-enum-conversion
This clang warning caught a real layout bug related to implicitly truncated enums in the table border style code (bug 1485179):
layout/tables/nsTableFrame.cpp:5318:14 [-Wbitfield-enum-conversion] bit-field 'ownerElem' is not wide enough to store all enumerators of 'BCBorderOwner'
layout/tables/nsTableFrame.cpp:5358:16 [-Wbitfield-enum-conversion] bit-field 'ownerElem' is not wide enough to store all enumerators of 'BCBorderOwner'
layout/tables/nsTableFrame.cpp:5374:18 [-Wbitfield-enum-conversion] bit-field 'subElem' is not wide enough to store all enumerators of 'BCBorderOwner'
layout/tables/nsTableFrame.cpp:5385:18 [-Wbitfield-enum-conversion] bit-field 'subElem' is not wide enough to store all enumerators of 'BCBorderOwner'
Also update the comment linking to clang and gcc's warning documentation.
Differential Revision: https://phabricator.services.mozilla.com/D12513
--HG--
extra : moz-landing-system : lando
This is required on Mac, where there is a dummy nasm executable
that prints an error if the real nasm isn't found.
Differential Revision: https://phabricator.services.mozilla.com/D12371
--HG--
extra : moz-landing-system : lando
Bug 1492663 upgraded those builds to clang 7, and bug 1503330 brought
them back to clang 6 because of speedometer regressions.
With the previous change, the regression doesn't happen any more,
allowing to upgrade again.
Depends on D12394
Differential Revision: https://phabricator.services.mozilla.com/D12395
--HG--
extra : moz-landing-system : lando