The original commit message for the order files says "It's OK if these files are slightly out of date", and that's true. However, in the 11 months since then, our builds have undergone major changes that affect large numbers of final symbol names, like switching from Frontend to IR PGO, and enabling the new pass manager. I think at this point a refresh wouldn't hurt.
I generated these files by locally creating a "stage 2.5" PGO build that's both `-fprofile-use` _and_ `-finstrument-functions-after-inlining`, and otherwise following the instructions from bug 1444171. Doing this at the `use` stage rather than the `gen` stage lets us get around the problem of mixing IR and Frontend instrumentation. It also gets us a more accurate picture of the final optimizations.
I haven't exhaustively run this through Talos. I did some spot checks, most were within noise, Speedometer maaybe gets 0.6%. I don't think this is worth hogging the talos machines for a full run, we can get the results after landing.
If for no other reason, this is worth doing to reduce 8000 lines of warning LNK4037 (order file refers to a nonexistent symbol) in build logs down to about 15.
Differential Revision: https://phabricator.services.mozilla.com/D72171
Specifically, this renames
* nsTArray_CopyChooser to nsTArray_RelocationStrategy
* the Copy template argument of nsTArray_base to RelocationStrategy
* nsTArray_CopyWithConstructors to nsTArray_RelocateUsingMoveConstructor
* nsTArray_CopyWithMemutils to nsTArray_RelocateUsingMemutils
* DECLARE_USE_COPY_CONSTRUCTORS to MOZ_DECLARE_RELOCATE_USING_MOVE_CONSTRUCTOR
Differential Revision: https://phabricator.services.mozilla.com/D66243
--HG--
extra : moz-landing-system : lando
The remaining uses all need adjustements to in-tree mozconfigs, so they
all need to be done at once.
However, to make things slightly more intelligible, we do this in two
steps. This is step 1: we modify the use_toolchain transform to take care of
the transformation, while keeping the task definitions intact, so that
we only deal with mozconfig and build script adjustements here.
Differential Revision: https://phabricator.services.mozilla.com/D41890
We're planning on switching to IR-based profiling, so we can't use the
frontend-based instrumentation to collect the order in which functions
are executed...at least not during the build itself. Performance tests
indicate that not having the order information decreases performance
significantly. So we're going to check in static files for Win32 and
Win64 and use those to perform the ordering. It's OK if these files are
slightly out of date; as of this writing, builds that generate and then
use these files complain that ~1/3 of the functions can't be found (!).
We're just trying to do something slightly smarter than whatever the
linker default is.
Differential Revision: https://phabricator.services.mozilla.com/D31132
--HG--
extra : moz-landing-system : lando
The current setup uses different ways for different platforms, with
different workarounds, even using extra configuration items for Windows.
Now that there can't be a difference between the host per the build
system and the host per rust, we can get rid of those configuration
items, and use a more common infrastructure.
We cannot, however, avoid using wrapper scripts, because per-target rust
link-arg flags don't work up great.
The downside is that multiplies the number of wrappers, as we now have
to have a different one for host and target, and then we have .bat files
and shell scripts for, respectively, Windows hosts, and other hosts.
Depends on D24321
Differential Revision: https://phabricator.services.mozilla.com/D24322
--HG--
extra : moz-landing-system : lando
The current setup uses different ways for different platforms, with
different workarounds, even using extra configuration items for Windows.
Now that there can't be a difference between the host per the build
system and the host per rust, we can get rid of those configuration
items, and use a more common infrastructure.
We cannot, however, avoid using wrapper scripts, because per-target rust
link-arg flags don't work up great.
The downside is that multiplies the number of wrappers, as we now have
to have a different one for host and target, and then we have .bat files
and shell scripts for, respectively, Windows hosts, and other hosts.
Depends on D24321
Differential Revision: https://phabricator.services.mozilla.com/D24322
--HG--
extra : moz-landing-system : lando
Now that everything is lined up nicely, we don't need those anymore.
We can fill cargo-linker.bat with HOST_LINKER and HOST_LINKER_LIBPATHS,
which, unfortunately, can't be used as is, so we add a variant for
substitution in .bat files.
Differential Revision: https://phabricator.services.mozilla.com/D18034
--HG--
extra : moz-landing-system : lando
Now that it's automatically derived and that in-tree mozconfigs are
setting DIA_SDK_PATH, we don't need to set WIN_DIA_SDK_BIN_DIR on
automation, or elsewhere, for that matter.
Differential Revision: https://phabricator.services.mozilla.com/D17915
--HG--
extra : moz-landing-system : lando
And remove all the variables that configure will figure out for us as a
consequence. This has the side effect of making the automation builds a
little more like local builds, in that they don't rely on preset PATH,
LIB, etc. for the build to work.
Differential Revision: https://phabricator.services.mozilla.com/D17789
--HG--
extra : moz-landing-system : lando
It turns out, we don't need to `mk_add_options export` the variables
from the in-tree mozconfigs. If anything, that causes problems when
trying to simplify the mozconfigs, because it makes the variables
exported from .mozconfig.mk, overriding what configure may change and
store in autoconf.mk.
All the variables are handled by configure in a way that makes them
available in autoconf.mk, so there's no loss there, and with the
python/shell-based mozconfig loader, it turns out we don't need to go
through extra normalization via cmd.
autospider.py, being its own pseudo-mozconfig parser, still does need
it, though, but it was hooking into it already, so just inline that.
Differential Revision: https://phabricator.services.mozilla.com/D17769
--HG--
extra : moz-landing-system : lando
This patch also changes how pdbs for the ASAN job are copied:
we relax restrictions so that pdbs if present) are always copied out
and add an environment variable MOZ_COPY_PDBS to indicate when we
want to produce pdbs for copying.
We currently use a 32-bit Rust toolchain for win32 builds, but this can lead
to OOM situations. This patch makes win32 builds use a 64-bit Rust toolchain,
which requires a little bit of extra configuration because rustc needs to
be able to find a link.exe that produces 64-bit binaries for building
things like build scripts, which are host binaries.
We will now generate a batch file that sets LIB to the paths to 64-bit
libraries and invokes the x64-targeting link.exe, and add a section to the
.cargo/config file to instruct cargo to use that batch file as the linker
when producing 64-bit binaries.
MozReview-Commit-ID: 9vKBbm7Gvra
--HG--
extra : rebase_source : 599b3b661c7a8a5db1f32a2a9732fc202fb55e1e
The new VS package is now based on update 15.4.2, although that release didn't affect any files in our package. I'm picking up the update mostly just to make filename unique.
The `cd $PATH && pwd` pattern no longer works because tooltool packages haven't
been downloaded yet when mach evaluates mozconfigs, check the directory
existence beforehand.
MozReview-Commit-ID: 68Axh1RWYUr
--HG--
extra : rebase_source : d851e4ae7f465bced1dce2774f871dce9313c8e3
David Major
Stefan Hindli
Jesse Schwartzentruber
Tetsuharu OHZEKI
Oana Pop Rus
Simon Giesecke
Marco Bonardo
Tom Schuster
Ehsan Akhgari
Narcis Beleuzu
Noemi Erli
Emma Malysz
Mike Hommey
Razvan Maries
Masatoshi Kimura
Boris Zbarsky
Jonathan Kingston
Nathan Froyd
Tom Ritter
Margareta Eliza Balazs
Ryan VanderMeulen
Ted Mielczarek
Ting-Yu Chou
Wes Kocher