Get a list of leaf nodes in the dependency graph
from ninja, and make sure none of them are in the
output directory. This ensures that there are no
rules that depend on a file in the output directory
that doesn't have rule to generate it. The check
will catch a common set of build failures where
a rule to generate a file is deleted (either by
deleting a module in an Android.mk file, or by
modifying the build system incorrectly). These
failures are often not caught by a local incremental
build because the previously built files are still
present in the output directory.
Bug: 36843214
Bug: 68062417
Test: manual
Change-Id: I4933187e8b72f2ef0c32d18ffea756e2c6fa417c
This way config.mk no longer needs to check which java is in PATH and
fix it. It'll be consistent for all build steps under soong_ui.
Also unify handling of ANDROID_JAVA_HOME / JAVA_HOME with
OVERRIDE_ANDROID_JAVA_HOME / EXPERIMENTAL_USE_OPENJDK9.
Test: m nothing
Test: build/soong/soong_ui.bash --make-mode nothing (w/o envsetup.sh)
Test: aosp_arm ninja files are the same before/after
Test: before/after ninja files match with OVERRIDE_ANDROID_JAVA_HOME
Test: before/after ninja files match with EXPERIMENTAL_USE_OPENJDK9
Change-Id: Icdb65093d9c346524074de239a4f895e4230a24d
This may happen if there's been another call to microfactory before
we've read the log file. Just ignore the error and continue.
Test: Run `while true; do get_build_var TARGET_PRODUCT; done` in parallel
Change-Id: I670144f4637281ebf1299dbd9151dd9819c363e3
Create a lock file during log rotation so that multiple processes won't
step on each other.
Test: Run `while true; do get_build_var TARGET_PRODUCT; done` in parallel
Test: m blueprint_tools
Change-Id: I7144cd42aca47c694487ddae44713f82665ed81e
build/soong/soong_ui.bash --dumpvars-mode \
--vars="..." \
--abs-vars="..." \
--var-prefix="..." \
--abs-var-prefix="..."
is similar to the previous:
CALLED_FROM_SETUP=true BUILD_SYSTEM=build/core \
make -f build/core/config.mk dump-many-vars \
DUMP_MANY_VARS="..." \
DUMP_MANY_ABS_VARS="..." \
DUMP_VAR_PREFIX="..." \
DUMP_ABS_VAR_PREFIX="..."
and
build/soong/soong_ui.bash --dumpvar-mode [--abs] VAR
is similar to the previous:
CALLED_FROM_SETUP=true BUILD_SYSTEM=build/core \
make -f build/core/config.mk dumpvar-[abs-]-VAR
But uses soong_ui and ckati, so that we use a consistent make parser and
sandboxing configurations.
One major output difference between the pure make implementation and
this one is that report_config in Go is implemented using embedded
newlines in single quotes, while the make implementation uses `` with
embedded echo commands. This seems to work fine for both bash and zsh,
and report_config isn't meant to be machine-parsed anyways.
Test: build/soong/soong_ui.bash --dumpvar-mode report_config
Test: build/soong/soong_ui.bash --dumpvar-mode TARGET_DEVICE
Test: build/soong/soong_ui.bash --dumpvar-mode --abs PRODUCT_OUT
Test: build/soong/soong_ui.bash --dumpvar-mode --abs ALL_PRODUCTS
Test: build/soong/soong_ui.bash --dumpvars-mode --vars="report_config TARGET_DEVICE" --abs-vars="ALL_PRODUCTS"
Test: build/soong/soong_ui.bash --dumpvars-mode --vars=TARGET_DEVICE --abs-vars=PRODUCT_OUT --var-prefix=v_ --abs-var-prefix=a_
Change-Id: I0fbd0732bbf6fcfcd24084cf3c830a91a4b6bfc2
This reverts commit a3e6c520ac. I fixed
a few heavy stack users, and we've got better debugging.
Bug: 36182021
Test: m dist
Change-Id: Iefd8589185720e44c40a2e0ceff5fd1438211b86
Pass --dump-stats to ckati, but filter out the '*kati*' lines so that
they only end up in our verbose output. That way we've always got access
to the statistics.
Bug: 36182021
Test: m nothing; cat out/soong.log
Change-Id: Iaf7a814fc67f3e475c913faf69924a7f4e2ae3b3
This speeds up dumping make variables from ~380ms using make to ~220ms
using ckati. It also means that we're consistently using the same parser
for builds (with the same .KATI_READONLY/etc extensions).
envsetup.sh (lunch) / other scripts still use make, changing those to go
through soong_ui will be a future change.
Test: m clean; m nothing
Test: USE_GOMA=true m nothing
Test: m PRODUCT-aosp_x86-sdk
Test: m APP-Calculator
Test: build/soong/build_test.bash -only-config (on AOSP and internal master)
Change-Id: I6ca554de8de4955fb869001d06d29969b75751cc
They don't really affect anything general in android except for the
output path.
Bug: 65453318
Test: TARGET_BUILD_TYPE=debug m
Change-Id: I98c17fefae61f9f7936e184bd1f8441086bb7d23
Instead of pairing the reading of CleanSpec.mk files with the reading of
Android.mk files, split them into separate, individually cachable steps.
This way we only read Android.mk files once after a clean/sync. We'll
still read the CleanSpec.mk files multiple times, but that's
significantly faster than reading all the Android.mk files.
This adds about 50ms if kati doesn't need to reread the CleanSpec.mk
files. Reading all the CleanSpec.mk files takes about a second.
Bug: 35970961
Test: m clean; m nothing; m nothing
Test: Add CleanSpec.mk line, see it executed.
Change-Id: I83bad15c50709510959d5b8b673a907b8aa7de82
Kati sets detect_leaks=0 by default now.
Test: unset ASAN_OPTIONS; SANITIZE_HOST=address m nothing
Change-Id: I0cbf04ae9ed40de520abca3fb35c65e7f543682d
We can call directly into the blueprint bootstrap.bash using values that
soong_ui has already calculated.
Instead of calling into blueprint.bash, build minibp with microfactory,
and directly run ninja. This allows us to get individual tracing data
from each component.
Test: m -j blueprint_tools
Test: m clean; m -j blueprint_tools
Change-Id: I2239943c9a8a3ad6e1a40fa0dc914421f4b5202c
The Finder runs roughly 200ms faster than findleaves.py in aosp,
and runs roughly 400ms faster in internal master.
Bug: 64363847
Test: m -j
Change-Id: I62db8dacc90871e913576fe2443021fb1749a483
This reverts commit b6d161bf16.
Reason for revert: New Build Breakage: aosp-master/sdk_mac @ 4260825
Change-Id: I8bda8c50c5e5c9f84621d11a4c15b168833bcd21
This way we only have one way to start a build, which always has logging
/ tracing / etc, even if we don't need Kati.
There's two ways to use this:
As a direct replacement for mkdir out; cd out; ../bootstrap.bash;
./soong -- as long as --skip-make is always passed, we'll never run
Kati, and Soong will run outside of it's "make" mode. This preserves
most of the speed, and allows full user control over the Soong
configuration.
A (experimental, dangerous) way to temporarily bypass the product
variable and kati steps of a build. As long as a user is sure that
nothing has changed from the last build, and they know exactly which
Ninja targets they want to build (which may not be the same as the
arguments normally passed to 'm'), this can lead to shorter build
startup times.
Test: rm -rf out; m --skip-make libc
Test: rm -rf out; m libc; m --skip-make libc
Test: rm -rf out; mkdir out; cd out; ../bootstrap.bash; ./soong libc
Test: build/soong/scripts/build-ndk-prebuilts.sh
Change-Id: Ic0f91167b5779dba3f248a379fbaac67a75a946e
It can find every Android.bp in internal master in
about 2.5 sec the first time and 0.3 sec subsequent times
Bug: 62455338
Test: m -j blueprint_tools # which runs the unit tests
Test: m -j blueprint_tools && \
out/soong/host/linux-x86/bin/finder \
-v
--db /tmp/mydb \
--names Android.mk \
--prune-files .android-out-dir \
--exclude-dirs .git,.repo \
. \
>/tmp/finder-log 2>&1
Change-Id: I5ab2650459a1dae0d5d076faf411ec2d053c743d
Instead of calling SetNinjaBuildDir, pass it to bootstrap.bash, so that
the bootstrap package can set it consistently during bootstrapping and
normal execution.
Bug: 63720725
Test: m -j nothing
Test: mkdir o; ../bootstrap.bash; ./soong
Change-Id: Ica88d2d5f1461b5be49bfe6316c6ec4ef4d89d49
The build now uses a prebuilt javac binary distributed with the
source, checking the host java version is no longer required.
Reapplies Ieef83481d0e6d68371bbd8a422f870ffe2bbb428 after updating
some more manifests to include the prebuilts.
Test: m -j checkbuild
Bug: 62956999
Change-Id: I774dbb75c8ae2ff1a990423a3827c77886e44779
(cherry picked from commit c6f07f00da)
The build now uses a prebuilt javac binary distributed with the
source, checking the host java version is no longer required.
Test: m -j checkbuild
Bug: 62956999
Change-Id: Ieef83481d0e6d68371bbd8a422f870ffe2bbb428
MAKELEVEL and MAKEFLAGS were are both unset by makeparallel when invoked
from Make. In preparation for removing the Make wrapper (and
makeparallel), strip them in soong_ui as well.
Add MFLAGS for completeness.
Test: m -j
Change-Id: I029f9464d11b432e979b32ede360450f4facb038
Make supports specifying all types of variables on the command line
(using =, :=, +=, and other variable references. When running soong_ui
through make/makeparallel these all effectively become environment
variables.
So in preparation to remove the Make wrapper, support a simplified form
of this syntax, roughly equivalent to what the shell supports if
specified before the command (<NAME>=<VALUE>).
Test: m -j blueprint_tools
Change-Id: I08fa2b86710f282e619b0cc324a3e5bbaf62d26e
In preparation to remove Make/makeparallel from soong_ui startup, we
need to preserve compatibility with the different ways that make
supports the -j option.
Nothing changes unless Make/makeparallel is removed from the startup.
Once that is removed, not specifying a -j value will be equivalent to
'-j' instead of '-j1', like Ninja. A value will also be supported when
specifying -k, like Ninja (though specifying it alone will be equivalent
to '-k 0').
Test: m -j blueprint_tools
Change-Id: I9d5d59bedd4f6e5cca76bdb4cd47e0b5b7d523f0
This causes Soong to put the outputs of each genrule into a temporary
location and copy the declared outputs back to the output directory.
This gets the process closer to having an actual sandbox.
Bug: 35562758
Test: make
Change-Id: I8048fbf1a3899a86fb99d71b60669b6633b07b3e
This doesn't catch all the possible causes of timeouts,
(like if Ninja is only partially stuck or if Kati is stuck)
but it should clarify some causes of stuckness
Bug: 62065855
Test: m -j showcommands NINJA_HEARTBEAT_INTERVAL=500ms
Change-Id: I73a792ae91873b19d7b336166a2d47f37c549906
I removed *.txt from the installclean list yesterday, but it turns out
that android-info.txt being removed was the only thing triggering
system.prop to be rebuilt on the incremental build servers. So continue
android-info.txt so that the version information in system.prop is
correct.
I'll follow this up with a more complicated change to make the
system.prop generation depend on the build number/versions more
directly, but that's a more complicated change since we need to change
how the build number file is written.
Bug: 62252890
Test: m ...intermediates/system.prop; m installclean; m ...intermediates/system.prop
Change-Id: I25dd8ddd99f7b2c927a8cbfb7f1aca1dd3e82a82
For *.ini/*.txt files we should be recreating them as the instructions
change, so we shouldn't need to remove them. The biggest reason to
remove them is that the auto-installclean was removing
$PRODUCT_OUT/build_fingerprint.txt just after Kati wrote it out, causing
the fingerprint to be blank on builds where the auto installclean code
ran.
There's no reference to xlb in our tree, and it's been around since the
beginning of git history, so remove it.
Bug: 62224537
Test: lunch aosp_arm-eng; m nothing;
lunch aosp_arm-userdebug; m -j nothing;
check out/target/product/generic/build_fingerprint.txt
Change-Id: Id391af5c7d95ea78e4e68fae294d8295fccb1964
By default, the Android build enforces an OpenJDK 8 toolchain,
whose name contains the strings "openjdk" and "1.8".
After this CL, the check can be changed to enforce a toolchain
name starting with "9" and without the need for "openjdk" having
to occur in the name.
This experimental new check can be enabled by running:
export EXPERIMENTAL_USE_OPENJDK9=true
To switch back to the standard check, run:
unset EXPERIMENTAL_USE_OPENJDK9
Test: make ANDROID_COMPILE_WITH_JACK=false checkbuild tests \
&& make checkbuild tests
(with OpenJDK 8u45 toolchain on the PATH)
Test: make EXPERIMENTAL_USE_OPENJDK9=true \
ANDROID_COMPILE_WITH_JACK=false checkbuild
(with jdk 9-ea+170 toolchain on the PATH)
Bug: 38177295
Change-Id: I75de3e23fe0b7f41eb6dd3f55dadd3fa3c3383bd
This flag allowed for building with an OpenJDK 7 toolchain. It was
used for build bot builds that now work with an OpenJDK 8 toolchain.
Hence, this feature is no longer required. This CL drops it.
Bug: 27583810
Test: Treehugger build succeeds.
Change-Id: Iba9cf12cff2597fa70aa9999e8a6bda190f29f3c
This speeds things up a little bit, but the major win is that we don't
need to recursively call into the build during the auto installclean.
Test: m -j installclean
Test: m -j dataclean
Change-Id: I09ded8dbd8a2015c7848bc0042dfd3adad1b31c9
This way kati won't need to be run as often (either initially, or when
switching products with the same device).
Bug: 35970961
Test: m clean; m -j blueprint_tools; m -j blueprint_tools; m -j blueprint_tools
Test: lunch aosp_arm-eng; m -j blueprint_tools; lunch full-eng; m -j blueprint_tools; <repeat>
Change-Id: Ie9fca3c8f1dd412459ea47c7090c7c5fdb0bcf6e
When kati keeps state around, it has to regenerate the ninja file every
time the state is changed. So move the java version checking into
soong_ui, where we can parallelize it with other operations instead of
only checking it occasionally.
Bug: 35970961
Test: Put java7 in PATH, m -j
Test: Put java8-google in PATH, m -j
Test: Put a space in TOP, m -j
Test: OUT_DIR=<case-preserving fs> m -j
Test: OUT_DIR=<path with space> m -j
Test: DIST_DIR=<path with sapce> m -j
Change-Id: I3245c8dd6d856240d17d54cb05d593dc9df71a27
It now uses the same output style as ninja, overwriting status lines in
smart terminals.
Test: multiproduct_kati
Test: multiproduct_kati | cat
Change-Id: I8db5198ffdc5ebc5503241ac492379753d92978e
So that we don't have to load up all the makefile state just to remove
the output directory.
Starting from a completely empty out directory:
kati: 16s
soong_ui: 2.0s
From a minimal out directory (m -j blueprint_tools):
kati: 3.8s
soong_ui: 0.4s
Test: m -j clean
Test: m -j clobber
Change-Id: Ibeec6fbfa29387750342a752a55336caca4b3992
Otherwise Make and soong_ui disagree about where the soong.variables
file should be. For example, when "OUT_DIR=out/":
Error dumping make vars: Failed to parse make line: "make: Nothing to be done for `out/soong/soong.variables'."
We would need to pass out//soong/soong.variables into make, since it
doesn't attempt to match clean/unclean paths.
Bug: 37553659
Test: OUT_DIR=out/ m -j
Change-Id: Iebeea4868b96b11e8b08735f272e22be5db1d701
We're seeing some rare crashes that appear to be stack overflows. To get
better debugging, use the asan version of ckati when 'dist' is
specified.
Bug: 36182021
Test: m -j
Test: m -j dist
Test: make -j dist (w/o lunch)
Change-Id: Ic3a5590974bfd718bf1929355d344b5933ac1d4f
Wrap os/exec.Cmd to use our Context and Config interfaces for automatic
logging and error handling. It also simplifies environment modification
based on the Config's environment.
This also adds sandboxing on Macs using sandbox-exec. A simple profile
is provided that only logs on violations, though multiproduct_kati on
AOSP has no violations. This isn't applied to ninja, only make / soong /
kati to start with. I measured <5% time increase in reading all
makefiles, and no noticable difference when kati doesn't regenerate.
I'd like to spin up a process to dump violation logs into our log file,
but the log reporting changed over the range of Mac versions that we
support, so that's going to be more complicated. Opening Console.app
works in all cases if you're local -- just search/filter for sandbox.
Linux sandboxing will be implemented later -- the sandbox definition is
opaque enough to support a different implementation.
Test: multiproduct_kati on AOSP master on Mac
Change-Id: I7046229333d0dcc8f426a493e0f7380828879f17
I missed this when converting to soong_ui.
Test: m -j blueprint_tools (check soong.log)
Test: SANITIZE_HOST=address m -j blueprint_tools
Change-Id: I01eb567db6848dc36dd679557291a4e600a63bba
We keep having makefiles reading $DISPLAY, and GREP_OPTIONS with
--color=always has caused problems with shell commands in the past.
Bug: 37333696
Test: GREP_OPTIONS="--color=always" m -j
Test: DISPLAY=test m -j (modify an Android.mk to check DISPLAY)
Change-Id: Iafab37d61ecde5c6fdb35a05f18709abc17e8554
To allow to inject a wrapper program for compiling java code, let
me allow soong to export JAVAC_WRAPPER.
Test: Set JAVAC_WRAPPER and ANDROID_COMPILE_WITH_JACK=false envs,
Test: and make -j 32
Change-Id: Ib77fe00bd781f8f4ddf965c5e7905eab466a057a
Currently, Kati is the only one that is respecting OUT_DIR_COMMON_BASE,
causing the build to fail when ninja tries to match up the Soong and
Kati ninja files.
The soong_ui.bash script change is necessary to move the microfactory
built outputs to the correct location.
The config.go change takes care of setting OUT_DIR in the environment of
every subprocess, so that we could remove the OUT_DIR_COMMON_BASE
handling from Make once the USE_SOONG_UI=false path goes away.
Bug: 35929763
Test: m -j blueprint_tools
Test: OUT_DIR=... m -j blueprint_tools
Test: OUT_DIR_COMMON_BASE=... m -j blueprint_tools
Change-Id: I4184f2b4392d32c5bea51e51162a48df0e5c90d5
We can start removing out directories again in multiproduct_kati, since
the opendir bug has been fixed.
Add --color_warnings to the Kati command line. Since this is different
from Make, take this opportunity to reorder the command line to make
more sense. This wasn't done before because kati forces a regen whenever
the command line changes.
Test: USE_SOONG_UI=true m -j blueprint_tools
Change-Id: I5ad03359fbc16db482722946202297c1ae0f2b90
Bug: 35214134
Test: ctrl-C during build
Test: add for{} to hang soong_ui in multiple places, ensure it exits
Change-Id: Ic71eedd4b1814ab2f3c441ae61a97570eda4fe16
1. Collapses the "including .../Android.mk ..." lines like ninja does,
so that we overwrite the last line if there wasn't anything else to
print.
2. Strips ansi control codes so that log files don't include unreadable
characters.
Test: m -j
Test: m -j | tee output.log (with colored output)
Change-Id: Ib18437f6f9d37084360097a9d586800c833072c5
For proper ninja smart terminal support, we need to pass stdin to
./soong. Otherwise it starts a new line if the terminal isn't wide
enough.
Test: `rm -rf out/soong/.bootstrap; m -j` in narrow terminal
Change-Id: I643a526001adc2323a420a03fa1df282554c7886
This creates a rotating build.trace.gz in the out directory that can be
loaded with chrome://tracing. It'll include start and end timings for
make/soong/kati/ninja, and it will import and time-correct the ninja log
files.
Test: m -j; load out/build.trace.gz in chrome://tracing
Test: multiproduct_kati -keep; load out/multiproduct*/build.trace.gz
Change-Id: Ic060fa9515eb88d95dbe16712479dae9dffcf626
This is a replacement for build/tools/kati_all_products.sh using the new
Soong ui/build package. It doesn't even attempt to run ninja, and it can
be configured to run only the product config, or only the product config
and Soong.
For AOSP on my machine:
-only-config 1.4s
-only-soong 1m20s
<none> 13m
Test: multiproduct_kati
Change-Id: Ie3e6e7bdf692e46a8b8eb828f437190f8003500b
Right now this mostly just copies what Make is doing in
build/core/ninja.mk and build/core/soong.mk. The only major feature it
adds is a rotating log file with some verbose logging.
There is one major functional difference -- you cannot override random
Make variables during the Make phase anymore. The environment variable
is set, and if Make uses ?= or the equivalent, it can still use those
variables. We already made this change for Kati, which also loads all of
the same code and actually does the build, so it has been half-removed
for a while.
The only "UI" this implements is what I'll call "Make Emulation" mode --
it's expected that current command lines will continue working, and
we'll explore alternate user interfaces later.
We're still using Make as a wrapper, but all it does is call into this
single Go program, it won't even load the product configuration. Once
this is default, we can start moving individual users over to using this
directly (still in Make emulation mode), skipping the Make wrapper.
Ideas for the future:
* Generating trace files showing time spent in Make/Kati/Soong/Ninja
(also importing ninja traces into the same stream). I had this working
in a previous version of this patch, but removed it to keep the size
down and focus on the current features.
* More intelligent SIGALRM handling, once we fully remove the Make
wrapper (which hides the SIGALRM)
* Reading the experimental binary output stream from Ninja, so that we
can always save the verbose log even if we're not printing it out to
the console
Test: USE_SOONG_UI=true m -j blueprint_tools
Change-Id: I884327b9a8ae24499eb6c56f6e1ad26df1cfa4e4
