With clang 12, win32 debug tests run out of stack in the 1000-deep layout crashtests.
Usually when this type of thing happens, it's because the compiler made a drastically heavier inlining decision along the recursive part of the stack, with a difference in the hundreds of bytes, and we pay that price of that space 1000 times. In those cases we normally try to snip out the heavy piece with a MOZ_NEVER_INLINE.
However, in this case, the difference between old and new compilers was only something like 16 bytes of the cyclical piece. This suggests that we're already up pretty close against the limit, and we're only a bad day away from failing tests even with existing clang. This only happens in debug builds since they have lots of logging helpers on the stack. Release builds are still well under the limit.
So I propose to increase the stack limit in debug. While we normally have to be pretty careful about address space on win32, it should be less of a concern with these non-shipping builds.
Differential Revision: https://phabricator.services.mozilla.com/D101196
The `clobber` targets are superseded by `mach clobber`, so we don't need them for any reason. The `clean` target is meant to get you to a post-`configure` state, but it doesn't really work, and if it's necessary for you to be in that state for some reason you can just clobber and re-`configure`, so it doesn't seem worth it to get it working again. Instead, delete all of them. Also delete `everything` which is not useful when `clobber` doesn't exist.
Differential Revision: https://phabricator.services.mozilla.com/D93514
Bug 1645986 solved the problem for most generated files by moving their
rules to the top-level, but we're going to add rules that will end up in
subdirectories, so we have to solve the same problem again, in the
subdirectories.
Differential Revision: https://phabricator.services.mozilla.com/D88389
This makes us use one less tool from MSVC, and removes one more use of wine
in cross builds.
We replace the call to either rc/llvm-rc or windres with a wrapper script.
While the script is not strictly needed for the latter, we use a wrapper
in that case anyway because it's one step towards fixing bug 1498414.
For llvm-rc, however, we need a wrapper because llvm-rc doesn't preprocess
on its own, so the wrapper does that too.
The wrapper script also allows to deal with the default flags passed to
llvm-rc or windres, rather than inherit them from old-configure.
We also need to explicitly pass the codepage to llvm-rc, which was not
necessary with rc (presumably, llvm-rc has a different default).
While here, remove the unused WINDRES subst from js/src/old-configure.in.
Also, while here, we remove --use-temp-file, because as described in the
linked bug and in the windres manual page, it was used to work around bugs
on Windows 98 and earlier.
Differential Revision: https://phabricator.services.mozilla.com/D86312
l10n-merge creates a full merge dir for a while now, let's
simplify the build logic to only read from that directory
during repacks and langpacks.
Differential Revision: https://phabricator.services.mozilla.com/D77023
`ply`, [by design](https://github.com/dabeaz/ply/issues/79), does not produce reproducible table files; hence bug 1633156. (Note that this was *always* true, but only became a problem once we switched to Python 3, which has more unpredictable dict iteration order than Python 2.7, at least prior to [3.7](https://docs.python.org/3/whatsnew/3.7.html#summary-release-highlights).)
In any other circumstance I would consider submitting a patch to `ply` to fix this, but as of the [in-progress version 4.0 of the library](https://github.com/dabeaz/ply/blob/master/CHANGES), it doesn't even emit this cached data any more, and indeed the [latest version of the code](1fac9fed64/ply) doesn't even call `open()` at all except to do logging or to read the text data to be parsed from `stdin`. So if we were going to pin our future on `ply` and upgrade to later versions of the library in the future, we would have to live in a world where `ply` doesn't generate cached table files for us anyway.
Emitting the cached table files so later build steps can consume them is an "optimization", but it's not clear exactly how much actual value that optimization provides overall. Quoth the `CHANGES` file from that repository:
```
PLY no longer writes cached table files. Honestly, the use of
the cached files made more sense when I was developing PLY on
my 200Mhz PC in 2001. It's not as much as an issue now. For small
to medium sized grammars, PLY should be almost instantaneous.
```
In practice, I have found this to be true; namely, `./mach build pre-export export` takes just about as long on my machine after this patch as it did before, and in a try push I performed, there's no noticeable performance regression from applying this patch. In local testing I also found that generating the LALR tables in calls to `yacc()` takes about 0.01s on my machine generally, and we generate these tables a couple dozen times total over the course of the `export` tier now. This isn't *nothing*, but in my opinion it's also not nearly long enough where it would be a concern given how long `export` already takes.
That `CHANGES` file also stresses that if caching this data is important, we have the option of doing so via `pickle`. If and when we decide that re-enabling this optimization is valuable for us, we should take control of this process and perform the generation in such a way that we can guarantee reproducibility.
Differential Revision: https://phabricator.services.mozilla.com/D73484
Also enable all config/tests with python3. unit-nsinstall.py was the
last one that didn't pass with python 3.
Switch the test to using @unittest.skipIf and disable the subprocess
test because we purposely broke running nsinstall.py independently with
python 2.
Differential Revision: https://phabricator.services.mozilla.com/D70160
--HG--
extra : moz-landing-system : lando
At the moment, everything we're putting in a wasm sandbox is not
performance-critical, so we don't need PGO. It's also not clear that
PGO would actually work properly with code that's been run through
wasm. Let's leave figuring that out until, at the very least, the wasm
toolchain is a little more mature.
Differential Revision: https://phabricator.services.mozilla.com/D58513
--HG--
extra : moz-landing-system : lando
In 1-tier PGO builds that shared the objdir between the instrumented and
profile-use builds, the instrumented build objects used a different
suffix (.i_o) to separate them from the profile-use build (which uses
the default .o suffix). These builds are now always in separate objdirs,
and don't need special suffix rules anymore.
As a bonus, this helps fix an issue with buildid.cpp continually
rebuilding because libxul_so.list always lists the inputs as *.o, which
don't exist if we're using a .i_o suffix. Make would always re-create
buildid.cpp and therefore libxul.so in the instrumented build even when
nothing has changed.
Differential Revision: https://phabricator.services.mozilla.com/D56115
--HG--
extra : moz-landing-system : lando
Add backend stuff to build sandboxed wasm libraries. (Don't actually update any moz.build files to consume this yet.)
Differential Revision: https://phabricator.services.mozilla.com/D54152
--HG--
extra : moz-landing-system : lando
Actually we set _DEPEND_CFLAGS to both host and target compiler. But if host and target are different compiler type, we may pass invalid option.
Differential Revision: https://phabricator.services.mozilla.com/D38457
--HG--
extra : moz-landing-system : lando
Bug 1560527 was not supposed to change verbosity for mach build, but it
turns out it did, because the ifeq it copied from one place to another
was wrong in the first place.
While here, replace a ifeq that did work with the now equivalent
BUILD_VERBOSE_LOG.
Differential Revision: https://phabricator.services.mozilla.com/D35966
--HG--
extra : moz-landing-system : lando
This makes running without mach more consistent. e.g. running
`make -C $objdir/toolkit/library/rust target` makes the cargo log
verbose, and adding `-s` makes it less verbose.
Differential Revision: https://phabricator.services.mozilla.com/D35521
--HG--
extra : moz-landing-system : lando
We're moving to IR-level PGO instrumentation for clang-cl. We've also
moved to using static linker ordering files, which was the primary
application of the previous style of PGO instrumentation. We therefore
we no longer need this code.
Differential Revision: https://phabricator.services.mozilla.com/D31134
--HG--
extra : moz-landing-system : lando
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
This patch adds detection for when icecream is in use to build telemetry.
icecream is commonly enabled in two ways: by either setting CC/CXX to point
to icecream's cc/c++ symlinks, or by setting adding
mk_add_options 'export CCACHE_PREFIX=icecc' to a mozconfig when also using
ccache. For the former, this patch adds a simple configure check to see
if CXX is a symlink to a file named 'icecc'. For the latter this patch adds
CCACHE_PREFIX as a configure subst to capture the value.
We don't currently have a facility for writing telemetry tests that depend on
configure values. Local manual testing shows that it does work as expected.
Differential Revision: https://phabricator.services.mozilla.com/D18138
--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
This was added back in bug 1414287, as a sort of quick hack, but since
then, bug 1515528 fixed things such that the hack is not necessary
anymore, and bug 1523201 allows for things to work on automation
(HOST_LINKER ended up being wrong because of the lack of "proper" VC
configuration on automation)
Differential Revision: https://phabricator.services.mozilla.com/D17790
--HG--
extra : moz-landing-system : lando
The only use in configure itself is for a MSVC version check that is now
always true (we don't accept versions < 19.15 anymore).
The only uses in the build system are in code that could just use
CC_TYPE instead.
Differential Revision: https://phabricator.services.mozilla.com/D17207
--HG--
extra : moz-landing-system : lando
At some point we made L10NBASEDIR required. That means that
env L10NBASEDIR=... make chrome-AB_CD
takes the value set by configure. That is different than
make chrome-AB_CD L10NBASEDIR=...
which uses the value passed on the command line. Rather than making
the latter style work with `mach build`, we instead set the "correct"
value for L10NBASEDIR in automation.
We could remove the --with-l10n-base stanzas from many automation
mozconfigs, but there's some small advantage to keeping them explicit.
Perhaps eventually we will remove them -- hopefully after
standardizing l10n vs l10n-central!
Differential Revision: https://phabricator.services.mozilla.com/D15776
--HG--
extra : moz-landing-system : lando
This also requires the 64-bits rust compiler and some build system
tweaks.
And since we make the 32-bits builds cross-compiles on CI, we also need
to adjust the MSVC build mozconfigs such that the host compiler points
to the right MSVC cl. Likewise, the DIA SDK is used for host things, so
use the 64-bits version or it.
Differential Revision: https://phabricator.services.mozilla.com/D7845
--HG--
extra : moz-landing-system : lando
This also requires the 64-bits rust compiler and some build system
tweaks.
Differential Revision: https://phabricator.services.mozilla.com/D7845
--HG--
extra : moz-landing-system : lando
This merely centralizes logic that was formerly used at multiple sites
into the single remaining use site.
None of the JAVA* flags have been used for a long time.
Differential Revision: https://phabricator.services.mozilla.com/D7313
--HG--
extra : moz-landing-system : lando
BUILD_TOOLS was only ever used for things that another variable provides
equally well. Removing BUILD_TOOLS means that we can remove win_srcdir
and WIN_TOP_SRC as well.
Currently mozconfig.cache overrides a few build options for sccache.
This patch moves them into toolchain.configure so that the build system
will set them properly when sccache is in use. Additionally,
{CC,CXX}_WRAPPER are set in config.mk, so just avoid setting them when
sccache is in use.
David Major
Ricky Stewart
Mike Hommey
Axel Hecht
Bogdan Tara
Nathan Froyd
Mike Shal
Makoto Kato
Tom Ritter
Gijs Kruitbosch
Andreea Pavel
Henri Sivonen
Ted Mielczarek
Nick Alexander
Noemi Erli
Ciure Andrei
Nathan Froyd