Currently, installation of a module is defined as an action of copying
the built artifact of the module to an install path like out/soong/host
(for host modules) and out/target/product/<device>/<partition> (for
device modules). After the modules are installed, the installed files
are further processed to create packages like system.img, vendor.img,
cvd-host-package.tar.gz, etc.
This notion of installation seems to have originated from the old time
when system.img is the primary product of the entire build process
(modulo a few more like root.img). Packaging the installed files as the
filesystem image was considered as a post-build step then.
However, this model doesn't seem to fit well to the current and future
environment where we have a lot more filesystem images (system, vendor,
system_ext, product, ...). The filesystem images themselves are even
grouped together to form a higher-level filesystem image like super.img.
Furthermore, things like cvd-host-package.tar.gz requires us to be able
to group some of the host tools in a format that isn't filesystem image.
Lastly, we are expected to have more filesystem images that are subsets
of system.img (and their friends) for the Android-like mini OS that will
be running on on-device virtual machines. These all imply that the
packaging (which we call installation today) is not a global post-build
step, but a part of the build rules for creating the package-like
modules.
A model better fits to the new sitatuation might be this; a module
specifies its built artifact and the path where it should be placed. The
latter path is not rooted at out/. It's a relative path to the root
directory which will be determined by another module that implements the
packaging. For example, cc_library will have ./lib (or ./lib64), not
out/target/product/<device>/<partition>/lib as the path. Then packages
like system.img, cvd-host-package.tar.gz, etc. are explicitly modeled as
modules and they have deps to other modules. Then the modules are placed
at the relative path under the package root, and the entire root
directory finally is packaged as the output file (be it img, tar.gz, or
whatever).
PackagingSpec is the first step to implement the new model. It abstracts
a request to place a built artifact at a certain path in a package. It
has extra information about whether the path should be a symlink or not,
and whether the path is for an executable. It currently is created when
InstallFiles (and its friends) are called, and can be retrieved via
the new method PackagingSpecs().
In this CL, no one is using PackagingSpec. The installation is still
done by the existing rules created in InstallFiles, etc. and the
structs are not used for the filesystem images like system.img.
Bug: 159685774
Bug: 172414391
Test: m
Change-Id: Ie1dec72d1ac14382fc3b74e5c850472e9320d6a3
Soong currently assumes that installed files should depend on
installed files of all transitive dependencies, which results
in extra installed file dependencies through genrules, static
libs, etc.
Annotate dependency tags for dependencies for which the
installed files are necessary such as shared libraries
and JNI libraries.
This avoids extra installed files, and is also a first step
towards genrules using their own copy of tools instead of
the installed copy.
Bug: 124313442
Test: m checkbuild
Test: java.TestBinary
Test: cc.TestInstallSharedLibs
Test: deptag_test.go
Change-Id: Ic22603a5c0718b5a21686672a7471f952b4d1017
Add vendor_ramdisk_available and vendor_ramdisk attribute to
various rules. When a vendor_ramdisk variant of a module is
generated, it is installed to $OUT/vendor-ramdisk.
It is similar to a ramdisk image.
Test: m nothing -j
Change-Id: Ib2d16459f3094dbe21c3bdb7c016cb4b2bf62765
This CL adds symlink_outputs to various locations in Soong that creates
actions that creates symlink outputs, and explicitly mark them as such.
Test: m
Bug: 160568334
Change-Id: I322751bada52a9f49011c74731d84761586e03e7
A global variant was used to store the global mapping between
modules and APEXes. Replace it with storing pointers to APEX
contents inside each module so that they can query the contents
of any APEXes they belong to.
Bug: 146393795
Test: all Soong tests
Test: single line change to build.ninja host install dependency ordering
Test: no Android-${TARGET_PRODUCT}.mk, make_vars-${TARGET_PRODUCT}.mk or late-${TARGET_PRODUCT}.mk
Change-Id: Id2d7b73ea27f8c3b41d30820bdd86b65c539bfa4
It will be used by the cvd_host_package module type to gather installed
files and package them in a distributable format.
Bug: 168086242
Test: m
Change-Id: I3af6dcda0bce1f298b1500fbac311c9593ddf375
Mulitple init_rc files may be necessary to support multilib binaries
with different installed names.
Test: m checkbuild
Change-Id: I17d3a3b2346a25b230e5a325deb0d78f0ee045f0
A host target is considered as being cross-compiled when the target
can't run natively on the build machine. For example, linux_glibc/x86_64
is a non-cross target on a standard x86/Linux machine, but is a cross
host on Mac. Previously, whether cross or not was a static attribute of
an OsType. For example, Windows was always considered as cross host,
while linux_bionic was not. This becomes a problem when we support more
host targets like linux_bionic/arm64 which should be cross-host on
standard x86/Linux machines.
This change removes HostCross from the OsClass type and instead adds a
property HostCross to the Target type. When a target is being added, it
is initialized to true when the target can't run natively on the current
build machine.
Bug: 168086242
Test: m
Change-Id: Ic37c8db918873ddf324c86b12b5412952b0f2be2
Move PrimaryModule, FinalModule, and VisitAllModuleVariants to
BaseModuleContext so they can be used by mutators.
Test: m checkbuild
Change-Id: I4280333f19283edc8f65fc4af6f49b654af31782
Copy documentation from Blueprint for interface methods that are
wrapped into Soong interfaces.
Test: m checkbuild
Change-Id: Ib58c9201b8e7dfbf900a252e709f0a3cfed73286
This change fixes some places where host arch is assumed to be x86 to
form a path element like linux-x86. In preparation for non-x86 host
targets, the host arch part is derived from the context.
In addition, InstallForceOS() is refactored so that it can override not
only OsType, but also ArchType. Without this, the paths for the
robolectic modules will be changed from linux-x86 to linux-common, which
breaks several other places where the old paths are expected.
Bug: 134795810
Test: m
Change-Id: Ib38c715948ae546e55021ece82bac1d82e9e5da0
Common properties are automatically inherited from a parent module
(i.e. one that calls CreateModule()) to the child module that it
creates. This makes no sense for dist/dists properties so this change
extracts them into their own structure separate to the
commonProperties.
Test: m checkbuild and TreeHugger
Bug: 160448975
Change-Id: Icceb20455e13394dd3b3bce464fb9bb34882d6c3
APEX variants that share the same SDK version and updatability
almost always use identical command line arguments to build but
with different intermediates directories. This causes unnecessary
build time and disk space for duplicated work.
Deduplicate APEX variants that would build identically. Create
aliases from the per-APEX variations to the new shared variations
so that the APEX modules can continue to depend on them via the
APEX name as the variation.
This has one significant change in behavior. Before this change,
if an APEX had two libraries in its direct dependencies and one
of those libraries depended on the other, and the second library
had stubs, then the first library would depend on the implementation
of the second library and not the stubs. After this change, if
the first library is also present in a second APEX but the second
library is not, then the common variant shared between the two
APEXes would use the stubs, not the implementation.
In a correctly configured set of build rules this change will
be irrelevant, because if the compilation worked for the second
APEX using stubs then it will work for the common variant using
stubs. However, if an incorrect change to the build rules is
made this could lead to confusing errors, as a previously-working
common variant could suddenly stop building when a module is added
to a new APEX without its dependencies that require implementation
APIs to compile.
This change reduces the number of modules in an AOSP arm64-userdebug
build by 3% (52242 to 50586), reduces the number of variants of the
libcutils module from 74 to 53, and reduces the number of variants
of the massive libart[d] modules from 44 to 32.
Bug: 164216768
Test: go test ./build/soong/apex/...
Change-Id: I0529837476a253c32b3dfb98dcccf107427c742c
In preparation for reusing the same variation for multiple apexes,
rename ApexName to ApexVariationName.
Bug: 164216768
Test: all soong tests
Change-Id: I88f2c5b192ffa27acd38e01952d0cefd413222a0
unavailable-to-platform case.
This removes the special case added in https://r.android.com/1274763
from SkipInstall(), so that it doesn't cause conflicting AndroidMk
entries when a cc_prebuilt_library_static module has prefer:true and
the corresponding source module exists.
Test: `m` in a tree with a snapshot created from art-module-sdk where
the libartimagevalues module has prefer:true
Bug: 151303681
Change-Id: I651ae325753b707296892adb4cae80daaddb6af2
When a module is disabled by calling Disable(), it could be
unexpectedly enabled by the archMutator. For example, a module has
```
enabled : false,
arch : {
arm64 : {
enabled : true,
},
}
```
When this modules is disabled by calling Disable() before the
archMutator, it became enabled again ignoring Disable() command.
By defining the `ForcedDisabled` property, we can disable a module
regardless of `Enabled` property.
Bug: 161565086
Test: m
Change-Id: I443720715edbac25cb6d7c3d9eb1bc9806719161
dependencyTag uses a set of predefined tags to identify different types
of dependencies. There are already multiple bits of metadata stored
in the dependency tag (Library, Shared, ReexportFlags), and supporting
them all requires a combinatorial explosion of predefined tags and
causes issues when using equality comparisons if a new bit of metadata
is added.
Add a new libraryDependencyTag type that will contain the metadata
bits, and replace the quality comparisons with checks on the metadata
bits.
There are 5 TODOs where modifying the checks identified problems with
the existing checks. These were left in place to produce identical
build output and will be fixed separately.
Bug: 162437057
Test: no change to build.ninja or {Android,make_vars,late}-${TARGET_PRODUCT}.mk
Change-Id: I72d4207dcf381c07c92e00e5a03968ebb5ed8d30
When WITHOUT_CHECK_API=true, the droidstubs apiFile can be nil. Fix this
by defensively checking for nil paths when making the TaggedDistFiles,
and also in android.MakeDefaultDistFiles.
The error message reported in r.android.com/1335521 was:
"internal error: panic in GenerateBuildActions for singleton androidmk
Dist file should not be nil for the default tag in
android.net.ipsec.ike.stubs.source in translateAndroidMkModule for
module android.net.ipsec.ike.stubs.source variant android_common"
Test: WITHOUT_CHECK_API=true m droid dist
Bug: 152834186
Signed-off-by: Jingwen Chen <jingwen@google.com>
Change-Id: I1b1f7c0b7a0e1c0ed5e15957d0162c47fd3ec197
Previously, java_sdk_library_import added the dependencies on its child
components in the deps mutator after prebuilts without a matching
source module are renamed to the source module name. That meant that
the java_sdk_library_import has to use the source module name and ended
up depending on the source module unless it was preferred.
This change adds a new component deps mutator that runs before the
PrebuiltRenameMutator so that the java_sdk_library_import can add
dependencies onto the prebuilt modules. It also updates an affected
test.
Bug: 159902351
Test: m nothing
Change-Id: I3576c4873302743e51aff547ea1497bef6d748ac
"excludes" should be applied both SourceFileProducer and
OutputFileProducer. But we've been missing the latter.
(see android/paths_test.go)
There's one more to note. Previously, the exclusion was done by
modifying slice directly, which caused some problem(b/76179848).
The fix was returning a copy from Srcs() method. But there're still many
implementors which just return internal slices directly.
This change creates a new array when it needs to exclude something from
returned slice.
Bug: n/a
Test: m (soong test added)
Change-Id: I1ba91e490cb55fc243020681667278cdc75f59df
This CL adds "dists" to the base property struct to support multiple
dist file configurations, and generic tag support to dist tagged outputs
of modules.
Fixes: b/152834186
Test: soong tests and `m sdk dist`
Change-Id: I80c86bc9b7b09e671f640a4480c45d438bdd9a2a
Signed-off-by: Jingwen Chen <jingwen@google.com>
Fix a bug where host-only modules were incorrectly listed as VNDK.
Also refactor VndkMutator() / apexVndkDepsMutator() module skipping
logic.
Bug: 158543482
Test: Add unit test to cc/cc_test.go
Change-Id: I50b09f526cbc081149d8241c2a091e3ee48ef4d7
Merged-In: I50b09f526cbc081149d8241c2a091e3ee48ef4d7
(cherry picked from commit bba545e039)
Jiyong Park