On one-go MOZ_PGO builds, it's generally not wanted to do LTO during the
profile-generate phase. And the build system doesn't really support
different build options between both phases in this case, so we relied
on MOZ_PROFILE_GENERATE to disable the LTO flags.
However, in standalone profile-generate builds, if --enable-lto is
passed explicitly, the build should respect that choice.
So instead of checking MOZ_PROFILE_GENERATE to disable the LTO flags,
we disable them when MOZ_LTO is not set, and we force it to be disabled
during the profile-generate phase of one-go MOZ_PGO builds.
Differential Revision: https://phabricator.services.mozilla.com/D21659
--HG--
extra : moz-landing-system : lando
Consequently, this removes:
- MOZ_LIBPRIO, which is now always enabled.
- non_msvc_compiler, which is now always true.
- The cl.py wrapper, since it's not used anymore.
- CL_INCLUDES_PREFIX, which was only used for the cl.py wrapper.
- NONASCII, which was only there to ensure CL_INCLUDES_PREFIX still
worked in non-ASCII cases.
This however keeps a large part of detecting and configuring for MSVC,
because we still do need it for at least headers, libraries, and midl.
Depends on D19614
Differential Revision: https://phabricator.services.mozilla.com/D19615
--HG--
extra : moz-landing-system : lando
Support for running profileserver.py multiple times was originally added
in bug 663251 and subsequently removed in bug 672563. Since then we've
been copying around the argument to profileserver.py from mozconfigs in
buildbot-configs, to mozconfigs in mozilla-central, to client.mk, and
ultimately to Makefile.in. It no longer does anything and can be
removed.
Differential Revision: https://phabricator.services.mozilla.com/D15748
--HG--
extra : moz-landing-system : lando
The test is actually redundant with the TEXTREL check in
check_binary.py (R_386_PC32 relocations will only happen as text
relocations)
Differential Revision: https://phabricator.services.mozilla.com/D16002
--HG--
extra : moz-landing-system : lando
This addresses a related issue along the way: a build that results in running
configure would not update the value of self.config_environment (and therefore
self.substs) as seen from the build driver, so out of date values would have
been used.
The changes to Makefile.in and client.mk made exploit the assumption that by
he time anything in the Make build is running, config.status is up to date.
Users running "make" without the benefit of "mach" will need to manually run
configure when necessary in order to take this into account.
Differential Revision: https://phabricator.services.mozilla.com/D8473
--HG--
extra : moz-landing-system : lando
This patch makes it so we write the sccache log directly to the artifact
directory, so that it will be uploaded even if the build fails. It also makes
the log slightly more verbose. Both of these should help with diagnosing
sccache failures in CI.
The sccache log will no longer be explicitly gzip compressed, but some
Taskcluster client implementations will store it as gzip compressed.
Differential Revision: https://phabricator.services.mozilla.com/D6187
--HG--
extra : moz-landing-system : lando
This patch moves the logic for include/exclude/compress from the makefile into
symbols_archive.py, and adds an additional check to the full-archive case to
allow the inclusion of test symbols in try builds.
--HG--
extra : amend_source : 306419aae90a1935ced49fec89112df05142501c
Now that we've generated an order file of the first N functions invoked
during startup, let's tell the linker about said functions so it can
cluster them appropriately.
test_build.py fails on local builds with messages like:
make[4]: Entering directory '/tmp/tmp65BjCH'
make[4]: *** No rule to make target 'buildid.h'. Stop.
make[4]: Leaving directory '/tmp/tmp65BjCH'
/home/froydnj/src/gecko-dev.git/config/faster/rules.mk:76: recipe for target '/tmp/tmp65BjCH/buildid.h' failed
make[3]: *** [/tmp/tmp65BjCH/buildid.h] Error 2
make[3]: *** Waiting for unfinished jobs....
make[4]: Entering directory '/tmp/tmp65BjCH'
make[4]: *** No rule to make target 'source-repo.h'. Stop.
make[4]: Leaving directory '/tmp/tmp65BjCH'
/home/froydnj/src/gecko-dev.git/config/faster/rules.mk:76: recipe for target '/tmp/tmp65BjCH/source-repo.h' failed
make[3]: *** [/tmp/tmp65BjCH/source-repo.h] Error 2
make[3]: Leaving directory '/tmp/tmp65BjCH/faster'
Makefile:155: recipe for target 'faster' failed
The tests pass in automation, however, because automation always defines
the rules for buildid.h and source-repo.h in the toplevel Makefile.in.
For local builds, however, those rules are not defined to avoid build
churn. Let's ensure that the necessary rules are defined during testing
as well.
Tup can have a lot of unnecessary log lines, particularly on a fresh
build when it is scanning directories for the first time. Turning on
--quiet when MOZ_AUTOMATION is set will help trim down the log size
significantly.
MozReview-Commit-ID: AaRppurkXsA
--HG--
extra : rebase_source : dca18a0e3657853eacfc6cdbb935998cadb339ce
For automation builds, force buildid.h to be regenerated, and always
#include buildid.h in nsAppRunner.cpp. This will eliminate the mismatch at the
cost of re-linking libxul on every build.
For developer builds, always #include buildid.h in nsAppRunner.cpp, but do not
force buildid.h to be re-generated for incremental builds. This means we will
never have a mismatch between platform and application buildids, although
subsequent builds are no longer guaranteed to have a new buildid.
MozReview-Commit-ID: 19hvu8AQXgN
--HG--
extra : rebase_source : 9f4fd063f89b9b2165d5889fdaa8f83104c4411c
For automation builds, force buildid.h to be regenerated, and always
#include buildid.h in nsAppRunner.cpp. This will eliminate the mismatch at the
cost of re-linking libxul on every build.
For developer builds, always #include buildid.h in nsAppRunner.cpp, but do not
force buildid.h to be re-generated for incremental builds. This means we will
never have a mismatch between platform and application buildids, although
subsequent builds are no longer guaranteed to have a new buildid.
MozReview-Commit-ID: 19hvu8AQXgN
--HG--
extra : rebase_source : 181b53c2cc47526bd632a0931ba6e65fc74179ef
It is a one-liner and doesn't need to exist in a separate make file.
MozReview-Commit-ID: 5w3eLWD0CuB
--HG--
extra : rebase_source : 15ebb63156f8b4e8d3288dfaf0d9ec2c328b5914
Previously, client.mk made the decision of whether to perform a PGO
build. This required passing around MOZ_PGO and invoking a separate
make target if this variable was set.
In this commit, we move this logic to Makefile.in. We employ a special
mechanism in rules.mk to override the default make target so
`make` evaluates "profiledbuild" if MOZ_PGO is set. This also
required using an explicit target for $(MAKE) invocations inside
the "profiledbuild" rule to avoid infinite recursion.
MozReview-Commit-ID: 8sHiVspMisM
--HG--
extra : rebase_source : 76c8a0c592015802f2b9ad52fe2001012a4611f6
extra : source : 24e28b37fbc0ad55ee449699e09c62042af41b55
client.mk is going away. This commit moves the PGO build target out of
it.
We move the code to Makefile.in. This likely isn't its final
home, as we'll want to eliminate Makefile.in at some point. However,
it is easier to move things to the root Makefile.in than other
alternatives. Keep in mind that the logic needs access to primitives
like BUILDSTATUS and variables from mk_add_options, which the
Makefile.in mechanism has easy access to.
MozReview-Commit-ID: 4C43lBhZz4l
--HG--
extra : rebase_source : 0ff1e95dcf8cb0082b019713db0cb7f47f0bccb8
extra : source : 0d2083b8edc43fc13c77f71199e0fcaddb32fcfb
client.mk hasn't been well-supported for years. Stop recommending it.
MozReview-Commit-ID: KWFhAQPIXWu
--HG--
extra : rebase_source : dafb307d82f1f0a3141d077102fd2fe21add65b4
The --track flag provides a more accurate accounting of what files were
installed by the manifest, so they can be appropriately removed. For
example, test files are now removed from _tests if an entry in a test
file is deleted.
The --no-remove flag is removed as an alternative, and the --track flag
is now mandatory.
MozReview-Commit-ID: Wiup4Gzwkb
--HG--
extra : rebase_source : 4a44c7fe066ba9b5f1e37ec682464f7f4f6cb2cf
The SDK installation code was removed in bug 1333826.
MozReview-Commit-ID: 3h63nFSKOfi
--HG--
extra : rebase_source : 4a99f56e62c749e8991831f1a86cb4d98a58d31e
This change makes us upload an `$(PKG_BASENAME).generated-files.tar.gz` archive
alongside other build artifacts which contains all the generated source files
from the build. A change after this will introduce an `upload-generated-sources`
task to take this artifact and upload the individual files to an S3 bucket.
This will be used to provide links to generated source files when they appear
in stack traces in crash reports.
MozReview-Commit-ID: 6yQAdlZ5q3O
--HG--
extra : rebase_source : d92fb17ae737d1360e9724997f6688e29bedef12
extra : source : 14d18d7cf454c4c3d0f6d49d1d01660e06e4be4b
This change makes us upload an `$(PKG_BASENAME).generated-files.tar.gz` archive
alongside other build artifacts which contains all the generated source files
from the build. A change after this will introduce an `upload-generated-sources`
task to take this artifact and upload the individual files to an S3 bucket.
This will be used to provide links to generated source files when they appear
in stack traces in crash reports.
MozReview-Commit-ID: 6yQAdlZ5q3O
--HG--
extra : rebase_source : 3f6ef734c062e0f5e9c2ca433ffad51fdf14b1ad
Note this will only enable it on try, autoland and mozilla-inbound,
which are the only branches where sccache are enabled at the moment.
Enabling on more builds (or not) is the subject of bug 1373460.
Also note that bug 1181040, that ensured PGO builds weren't using
sccache mentions that back then, link times went up when using sccache
(with -Z7) vs. without, but that was presumably with MSVC 2013. Try
suggests link times are the same using sccache now (still using -Z7,
pending bug 1318370).
--HG--
extra : rebase_source : 9f9d87714f3c841b790eb7d692ea0968b1755b70
When investigating why a particular build got a particularly high number
of cache misses, it is useful to know exactly *what* had cache misses,
and you can't get the missing information after the fact.
We used to have some stats stored in a log file uploaded as artifact
with the old python sccache per bug 1005495 (and before that inline in
the build log), and we actually still have remains of that in the form
of a EXTRA_UPLOAD_FILES in build/mozconfig.cache.
Let's reuse that, and get some useful logs from sccache. Ideally sccache
would give better logs, but those are sufficient for now. We'll figure
things out eventually. https://github.com/mozilla/sccache/issues/151
--HG--
extra : rebase_source : a2e7093697a5b720ab28f155560211d6c0a4347e
Thunderbird is still using MOZ_PKG_PRETTYNAMES which has spaces in the package
name, which gets mangled by abspath.
MozReview-Commit-ID: 23ngx6gIESI
--HG--
extra : rebase_source : 9038f295e582bec5872898a5e8f052bfacf8f85a
This commit makes sccache dump JSON stats at the end of the build, and then
reads them in `BuildScript.generate_build_stats` and adds them to the
build_metrics we submit to Perfherder. The stats dumping is done in
Makefile.in where we currently dump verbose sccache stats because sccache
doesn't persist stats to disk right now and it will also shut down its server
process after 5 minutes, so when the post-build automation steps take more
than 5 minutes the server shuts down and the stats are lost.
Currently it's collecting:
* Cache hit rate
* Cache write errors
* Non-cacheable requests (compiler invocations that sccache can't cache)
We can always grow this list later.
MozReview-Commit-ID: J9CwU7XB05I
--HG--
extra : rebase_source : 084b09c3b0621330ac331a99b1bca9a15cf833b7
This commit moves symbol dumping to the compile tier, to be run via "syms"
targets. Tracking files are used for the sake of incremental builds, because
dump_syms may genearate multiple outputs whose paths are not known ahead of
time.
Minimal changes to symbolstore.py are made here. More extensive
simplifications will be made in a future commit on the basis of symbolstore.py
handling one file at a time.
MozReview-Commit-ID: 3mOP8A6Y7iM
--HG--
extra : rebase_source : bfe97afcdfc05b9e79f01577701c83e8b00eb4e9
This removes the UNIFY_DIST and UNIFIED_BUILD variables, as well as the
--unify flag from the packager and UnifiedBuildFinder from mozpack. As a
result the STAGEPATH variable is never defined anymore, so its uses can
be removed as well.
test_unify.py is currently the only mozbuild/mozpack test that fails
without running configure first, and there isn't much point in fixing
tests for things that we don't actually use anymore.
MozReview-Commit-ID: F5q1FPW3Did
--HG--
extra : rebase_source : cadbd237f51c23ea1983135294521d628d16f0df
This replaces the 'run-tests-deps' make target with a python function that will directly
read moz.build files, emit them with TestManifestEmitter, then consume them with
TestManifestBackend. Because the TestResolver is the only place that actually reads the
test metadata files, we can remove this logic from the CommonBackend as well.
MozReview-Commit-ID: DXgMoeH5dKf
MozReview-Commit-ID: HstZ57qkqf2
--HG--
extra : rebase_source : f377fa6863ef66d3adb86ed64f844e346686862f
These automation steps are no longer relevant, since the last use of
MOZ_AUTOMATION_PRETTY was removed along with non-unified builds in bug
1121000.
MozReview-Commit-ID: 11VDu8pIs1p
--HG--
extra : rebase_source : a82b52400b73224e06e124dd10d6646d4b21a7fc
This updates the client artifact code to locate test support files in the
common test archive and populate the objdir with these files appropriately.
MozReview-Commit-ID: GuXjwUtsl
--HG--
extra : rebase_source : 3560efee22533f60be1e7394fa0841005991cca6
This updates the client artifact code to locate test support files in the
common test archive and populate the objdir with these files appropriately.
MozReview-Commit-ID: GuXjwUtsl
--HG--
extra : rebase_source : 3560efee22533f60be1e7394fa0841005991cca6
In bug 1320194 we want to generate a TestManifest backend without first
invoking configure. However, we still need a way to update the backend,
which is normally handled by the top-level Makefile. We can split this
logic out into a separate file, rebuild-backend.mk, which can be invoked
directly as long as the appropriate environment variables are set
(BUILD_BACKEND_FILES, BACKEND_GENERATION_SCRIPT, and PYTHON).
--HG--
extra : rebase_source : b6a486db3e3c97b406b11074fda052bd63fe1b8f
This patch does a few things:
1) Change all the in-tree tooltool manifests to contain sccache2 instead of the existing Python sccache
2) Change mozconfig.cache to point at sccache.
3) Lightly tweak the --with-cccache configure option to support sccache, and detect whether we're using ccache or sccache and set an option appropriately.
4) Add a MOZ_SCCACHE_VERBOSE_STATS option, and add a target in the top-level Makefile to make sccache spit out its stats at the end of the build. This is useful to see the cache hits/errors until we get something better.
5) Add MOZ_USING_SCCACHE to the build telemetry. Not that I think it will be immediately useful, but for future use.
MozReview-Commit-ID: 9lrdLwNj5Bm
--HG--
extra : rebase_source : d323457df10d0ee0ac5811940e518d9422a7e070
Each archive takes 10+ seconds to generate. Split the archiving
part into separate make targets so this completes faster.
I'm not super thrilled about adding yet more make targets to the
root Makefile.in. We can refactor this later. For now, let's
take the quick win.
MozReview-Commit-ID: 2Oeq1rX6yLw
--HG--
extra : rebase_source : 55282a1dacf65adcaf559c178c61ae42591b29ea
This also fixes the issue of processing the artifacts twice in some
situations (bug 1275673). Note that the artifact download no longer
happens when a specific target is passed to 'mach build'.
MozReview-Commit-ID: Ktys6u3r1kG
On my i7-6700K time and size for the full zip archive for a Windows
debug build is as follows:
Time Size
Before: 25.4s 278,708,278
After: 13.0s 280,623,521
Anything lower than level 5 and the archive size starts increasing by
several megabytes. Level 5 is definitely the sweet spot.
MozReview-Commit-ID: 9lW2W1htq93
--HG--
extra : rebase_source : 26458386e4032dfd85eddd937bf6e6bff057d99b
This moves test installation for test files out of the monolithic install
manifest for $objdir/_tests, and determines the test and support files
to install based on the object derived from all-tests.json. Additionally,
the files resulting from TEST_HARNESS_FILES are installed, as some tests
will depend on them.
As a result, the time to install tests when invoking the test runner will
scale with the number of tests requested to run rather than the entire set
of tests in the tree, resulting in significantly less overhead.
MozReview-Commit-ID: LeIrUVh1yD4
The behavior is not entirely idempotent (most notably for
buildconfig.html), but this can be improved later if necessary.
It is idempotent where it matters.
This allows to get rid of config/makefiles/rcs.mk and its uses.
FasterMake needs some RecursiveMake install manifests to have been
processed before doing its work, so we can actually end up processing
them twice because of the going back and forth from FasterMake in the
hybrid build system.
Set the dependency at the RecursiveMake level when doing an hybrid
build.
--HG--
extra : commitid : 7nD60DTsoHz
extra : rebase_source : 61b5803732b0ecdff421e4e15a2086d4eae7a937
extra : amend_source : c0ae62708b2019888ea320c3793d4ea3f6d6d460
For some reason, recent changes with the install-dist/% rules broke the
dummy wrapper used by `mach build faster`, leading to `Nothing to be done
for install-dist_idl.`, subsequently failing to build xpt files because
of the missing idls in dist/idl.
--HG--
extra : commitid : 7SoYbKdAeSQ
extra : rebase_source : 8fee8fdae0d04aed3906d00d33dd7b830f82d75f
extra : amend_source : 2a9a1086d409fbcd4386b7d589c8df50dcdf4b26
The current rule is only for "backend.RecursiveMakeBackend", but, with
the current default of generating both the RecursiveMake and FasterMake
backends, the command creates/refreshes both backends. This is, in fact,
how the FasterMake backend is refreshed in most cases.
Moreover, with an hybrid backends, the generated file is not
"backend.RecursiveMakeBackend" anymore, so we need a more generic way to
handle this.
Furthermore, it's not necessarily desirable for all backends to have a
dependency file to handle the dependencies to refresh the backend, so
generate a plain list instead. This has the side effect of making `mach
build-backend --diff` more readable for changes to that file.
Finally, make the backend.* files created like any other backend file,
such that its diff appears in the `mach build-backend --diff` output.
Install manifests are not empty in normal conditions, apart a few
exceptions where they are only used for a "magic" `rm -rf`.
However, we're going to introduce changes that will empty some of
the install manifests and make their work happen from a different
backend, in which case we don't want them to correspond to a `rm -rf`.
While it would be possible to move those things in the export tier, it
is still interesting to have reporting for them separately, especially
considering I expect things to gradually move from the other tiers to
this new one.
While not entirely obvious, the recurse_pre-export target doesn't imply
actual recursion as long as the RecursiveMake backend doesn't emit
traversal information for it, so nothing will actually happen on this
target, but the interesting part is that it runs, per the generic
config/recurse.mk rules for tiers, between BUILDSTATUS TIER_START and
BUILDSTATUS TIER_FINISH, so that all its dependencies are accounted
as being part of the pre-export tier.
This change allows specifying objdir-relative paths in EXPORTS to enable
exporting entries from GENERATED_FILES. Objdir paths in EXPORTS that are
not in GENERATED_FILES will raise an exception.
Example:
```
EXPORTS += ['!g.h', 'f.h']
GENERATED_FILES += ['g.h']
```
Given the implementation, this should also work for FINAL_TARGET_FILES,
FINAL_TARGET_PP_FILES, and TESTING_FILES, but those are not well-tested.
This patch also renames the install manifest for '_tests' to match the
directory name for convenience in some code I refactored.
--HG--
extra : commitid : CwayzXtxv1O
extra : rebase_source : 5fb6f461fc740da9bce14bbdbfabdfe618af8803
We have many unit tests in the tree for some small parts of the build system
pipeline, but we don't have anything that resembles an end to end test, and we
kind of rely on the resulting Firefox not being broken by our changes.
With the Faster make backend growing, I want to ensure it produces the same
thing as the recursive make backend, at least for the parts it supports.
This adds some kind of test that allows to check that.
The configure option has explicitly thrown an error for more than a year now,
and it happens that the remaining way to still forcefully use it has been
broken for more than 8 months.
