Re: [android-building] libncurses5 package missing from Ubuntu 23.10+

Try https://android-review.googlesource.com/c/platform/prebuilts/clang/host/linux-x86/+/2760210 instead On Fri, Sep 29, 2023 at 1:37 PM xstefen wrote: > > Would love to but one does not simply walk into *.corp.google.com > On Friday, September 22, 2023 at 6:59:09 PM UTC-5 Stephen Hines wrote:

Re: [android-building] how the dist target is built?

I've described the `dist` command line argument as a modifier, as it's not really a target itself. In general, it's changing the behavior of other targets so that they also perform copies into $DIST_DIR. 1. there seem no Makefile in the AOSP root now, how does the make command > find the make

Re: [android-building] How to do Incremental build with updated compiler flags

We tried to make reading files from Soong difficult to do, since this is very error prone, as shown by this thread. It looks like you may have bypassed these protections by running very early during initialization. > > Also I think that go files are executed only in the clean build not in >

Re: [android-building] How to do Incremental build with updated compiler flags

Any changes in .mk / .bp files should be producing correct incremental builds -- in general since the ninja migration in Android N, though there have been some bug fixes over the years since for some specific use cases. Bug reports would be useful for any cases that you can reproduce on AOSP.

Re: [android-building] building lunch #48

FYI: The lunch number is only meaningful within a single lunch invocation -- it can change every time you sync (or make edits to the COMMON_LUNCH_CHOICES) If I remember correctly, `arm_v7_v8` is a product that was set up for some art testing, and isn't expected to build a full android image. The

Re: [android-building] Re: How to compile the user version for Android 12？

> > I think by "user", you mean "release". > Just to clarify, it is "user". There used to be a "release" vs "debug" choice too, but that hasn't existed or done anything for many years. And yes, the list you see in the `lunch` menu is just a small selection of targets someone thought would be

Re: [android-building] Compile zipalign.exe as 64-bit

There is build system support for 64-bit windows binaries, but it's not very well tested. A change like https://android-review.git.corp.google.com/c/platform/build/+/2645079 will produce a 64-bit zipalign.exe. On Wed, Jun 28, 2023 at 3:25 PM 'enh' via Android Building <

Re: [android-building] Re: How to skip robolectric host tests

Just m Settings would work as well, mma builds and installs everything under the directory, which would include building the robotests (even if you disable running them with DISABLE_ROBO_RUN_TESTS) - Dan On Fri, May 5, 2023 at 1:27 AM satur...@gmail.com wrote: > I figured out how to do it.

Re: [android-building] execute envsetup.sh

The AOSP master source tree is >100GB on disk (though likely less due to compression when downloading via git). You do need the full tree in order to run a build. If you aren't using it, the `--partial-clone` option to `repo init` can help reduce the amount you need to download at first, though

Re: [android-building] Encountered PHONY error on owners.mk when first full-building of aosp

Don't clone the manifest first, just run: $ mkdir $ cd $ repo init -u https://android.googlesource.com/platform/manifest -b master $ repo sync $ source build/envsetup.sh $ lunch aosp_redfin-userdebug (Even with aosp_arm64-eng, result is same) $ make -j4 What's happening is that by cloning the

Re: [android-building] How to build aapt2 for macOS?

Building device images is not supported on MacOS as of then, but we still support building the SDK tools. You will need to checkout the source tree on a case-sensitive filesystem (apfs works, it just needs a case-sensitive volume set up). Other than repo, the only thing you should need to install

Re: [android-building] Re: missing and no known rule to make it

The code that prints that out is in ninja , but it really won't help to debug it from there -- it has printed everything it knows. ninja: error: >

Re: [android-building] How to include external header files in cc_binary build

A common way for this to work would be to use a static library: Common/Android.bp: cc_library_static { name: "libCommon", srcs: ["Common.cpp"], export_include_dirs: ["."], } Foo/Android.bp: cc_library_shared { name: "libFoo", static_libs: ["libCommon"],

Re: [android-building] Re: dex2oatd fails due to unexpected mmap address

om> wrote: > >> >> >> On Thu, Jul 28, 2022 at 8:58 PM 'Dan Willemsen' via Android Building < >> android-building@googlegroups.com> wrote: >> >>> Have you tried to reproduce with a newer version of Android? Android 7 >>> is about 6 years old, so I

Re: [android-building] Cannot build Android 13 with audio.a2dp.default module

audio.a2dp.default is now part of the bluetooth mainline module , and cannot be installed independently. It will be embedded in the mainline module instead. - Dan On Tue, Sep 13, 2022 at 6:11 PM Michael Goffioul wrote: > I'm

Re: [android-building] Mirror manifest contains unavailable repo

This repo has been removed from the manifest in https://android-review.googlesource.com/2176262 - Dan On Thu, Aug 4, 2022 at 7:33 PM Al Sutton wrote: > Hey folks, > > The mirror manifest has a change[1] which adds a repo which isn't > available from the AOSP servers. > > Does anyone know if

Re: [android-building] platform/system/bpf is not reachable

There were some problems in our git infra, I believe they've mostly been resolved now, so please try again. Thanks, Dan On Mon, Aug 1, 2022 at 12:09 PM Amit Pundir wrote: > Hi Team, > > I'm not able to sync AOSP lately because platform/system/bpf is not > reachable. I see: > >

Re: [android-building] Re: dex2oatd fails due to unexpected mmap address

Have you tried to reproduce with a newer version of Android? Android 7 is about 6 years old, so I wouldn't be expecting any changes to it. You may want to talk to your distribution, this could be a configuration change during the kernel upgrade. Breaking existing binaries is something the kernel

Re: [android-building] Re: Build AOSP 11 with Google RBE service

> > In our company the AOSP 11 building time is *1.5hr* which is still slow, > as daily build and dev build are heavily depending on it. I have seen the > large potential of RBE to improve the build efficiency, > That's your clean build time, right? Are you using incremental builds when possible?

Re: [android-building] Re: Lunch combo options not shown

There are also a variety of log files under out/dumpvars-* that may help find an error during the dumping of the list (it may show up better if you try to use an explicit target instead of the menu fallback: `lunch aosp_arm-eng`) - Dan On Wed, Jun 15, 2022 at 11:57 AM Thilo Jäggi wrote: > Is

Re: [android-building] AOSP build time vs CPU core count (up to 56)

This is probably one of two things: 1. Disk Speed. Once you saturate your CPUs (and have the associated memory required -- somewhere around 2GB/core is usually decent), the next thing you'll often hit is disk bandwidth. This is where ccache tends to hurt more than it helps, unless the ccache is

Re: [android-building] Addr2line support in Android-13 release

We've been moving away from GCC and GNU Binutils over to Clang+llvm tools. It looks like there's a version of llvm-addr2line in prebuilts/clang/host/linux-x86/llvm-binutils-stable/llvm-addr2line - Dan On Fri, Jun 3, 2022 at 4:21 PM Bala Ramasamy wrote: > Hi All, > We are trying to use

Re: [android-building] Missing blueprints

Did your `repo sync` finish successfully? It sounds like the build/blueprint project is missing or corrupted. - Dan On Tue, May 17, 2022 at 9:08 PM Red wrote: > red@Red-LINUX:~/source$ lunch > > You're building on Linux > > Warning: Cannot display lunch menu. > > Note: You can invoke lunch

Re: [android-building] Compilation error while following codelab

I'd suggest syncing your source tree and trying to build again -- it appears that you have this frameworks/base change , but not this telephony change

Re: [android-building] Android 9 compilation error in Ubuntu 18.04.

The error isn't in this log snippet, look for a section starting with "FAILED:". Newer versions have an out/error.log file with the errors, but Android 9 is too old for that. - Dan On Thu, Mar 10, 2022 at 2:43 AM hani wrote: > I compiled it into a VMWare environment on two computers, but the

Re: [android-building] Building ADB and Fastboot results in huge binaries

You can build the entire platform-tools package for linux and windows with something like m dist sdk-repo-platform-tools on the master branch (and an upcoming platform-tools tag, but I don't think that one -- while there is an equivalent command, it's a much much much larger build, so I wouldn't

Re: [android-building] error.GitError: Cannot checkout platform/build/blueprint: Cannot initialize work tree for platform/build/blueprint

It sounds like you're having some issues with git, or your network/filesystem. There's not really enough information here to understand what issue you're hitting. Some things that may help: - Which repo command you're running - What git version you're using (git --version) - More

Re: [android-building] Android envsetup.sh overrides pre-set TARGET_BUILD_TYPE env variable!

We never really use anything other than TARGET_BUILD_TYPE=release -- in AOSP changing that flag is effectively a no-op. TARGET_BUILD_VARIANT=eng or runtime flags tend to control more debug-like features, but in general it's problematic to turn on any sort of "debug" for everything in a build

Re: [android-building] Raw ninja build log (not hashes)

The ninja log file, which is an implementation detail, does not have that information. The verbose.log.gz does have the command, and it's also possible to query the commands from ninja. In any case, remember that the command is executed within the build sandbox environment, so it may not act the

Re: [android-building] product_config.mk syntax error ?

It appears that make & ckati both treat that as "" == "," (when $(TARGET_BUILD_VARIANT) is `user`), so that case never triggers. I've posted a removal of the second comma here: https://android-review.googlesource.com/c/platform/build/+/1931200/ Thanks, Dan On Tue, Dec 21, 2021 at 1:10 PM

Re: [android-building] Finding target based on source files

This isn't a simple problem, even just in Soong (what do you do when an Android.bp file changes? what about a product config file that changes how a module is built, but is nowhere near it?). At Google we're largely just building a representative sample, but using incremental builds -- so we're

Re: [android-building] Re: Run shell script before include $(BUILD_PREBUILT)

Embedding build systems into one another is a very complicated issue, and requires deep knowledge of both build systems in order to keep all the functionality of both systems. Given that the Android build attempts to provide safe, fast incremental builds that work on any machine, it gets very

Re: [android-building] Re: building cts/vts

The build variant (eng/userdebug/user) generally should only affect the device builds, not the test builds, but there may be some differences. The instructions at https://source.android.com/compatibility/cts/development don't set anything, so you'll be using the default of `eng` if you build cts

[android-building] Re: building cts/vts

Generally, use the prebuilts -- you want to make sure that you're actually compatible with the same tests everyone else is running. The major exception is if you're developing a change to CTS/VTS itself (to upstream into AOSP), then you'll want to build it to test your change. In that case,

Re: [android-building] AOSP Compile Time

Most of the time you probably should be doing incremental builds instead of clean builds (it's very rare to see incremental build issues anymore), so clean build numbers should be somewhat less important than they used to be. If you're interested in diving into your build's performance, we've got

Re: [android-building] Error while building Android 12 for a signed image

The *-sign-targetfile-* rule doesn't appear to be in AOSP, so this looks like an issue in your custom make code. Switching to the built sign_target_files_apks is definitely a good start. out/.path should not have a working soong_zip, though it may be present -- the intention is to limit the

Re: [android-building] Android bootstrap fails with Golang panic

It looks like this was an issue in "VSDK"-based trees that are using a vendor snapshot, and was fixed in master (but not in Android 12): https://android-review.googlesource.com/q/topic:vendor_snapshot-192647618 You can either try to cherry-pick that topic, or disable the VSDK. - Dan On Wed, Oct

Re: [android-building] post install step for cc_binary

No, the source tree should be considered read-only from within the build. I'd suggest doing this as a wrapper around the build, rather than inside the build. You can set up `dist` options so that `m ... dist` builds will copy artifacts (and logs -- it's meant to be used on build servers) to

Re: [android-building] cc_binary module name and binary name

No, that's not possible, as the installation system still sets up the install rules for every module, so that the user can request a module to be installed even if it isn't in PRODUCT_PACKAGES. Either have them named differently on the device, or use soong namespaces

Re: [android-building] ninja error while upgrading external/v8 in Nougat due to Android.bp

Android.bp (and the Soong build system in general) wasn't really supported until Android-O, so an Android.bp will be ignored in Nougat. (There was some prototype support in Nougat under a USE_SOONG flag, but I really doubt that'll work for you) - Dan On Tue, Sep 14, 2021 at 4:19 PM Saurabh

Re: [android-building] Creating Symbolic Links in the /vendor partition

Ah, now I understand why I'm confused -- Android 12 (and main for the last year+) doesn't allow you to use PRODUCT_COPY_FILES to install binaries and libraries: docs . This lets us

Re: [android-building] build android q version on ubuntu 18.04 all cpu at low priority

Is there a specific use case where you need the nice value increased? Realize that we're also running in SCHED_BATCH mode, which also affects how the processes are scheduled. We didn't really see any build time changes when we switched this, only an increase in interactivity when using the desktop

Re: [android-building] Build faild randomly caused by resouce copy race condition during apk buiding

One fix for this is to move over to using an Android.bp to build your app rather than an Android.mk  But I also submitted a fix for this into aosp/master last week: https://android-review.googlesource.com/c/platform/build/+/1764131 - Dan On Tue, Jul 20, 2021 at 9:42 AM raymond@gmail.com <

Re: [android-building] Re: How to build AOSP with 8GB RAM

> > I hope AOSP will provide a way to disable the documentation build part of > it. Can someone from Google help me out here..? > Unfortunately, the documentation part of the build also generates code stubs and other information extracted from the code/docs (@hidden is a javadoc tag, not an

[android-building] Re: Platform development on MacOS isn't supported as of June 22, 2021.

> -g default,tools The "tools" group in the manifest is unrelated to building these SDK tools, so that shouldn't be necessary (though all of the "tools" grouped projects are also in "default" currently, so it really shouldn't change anything either). > LOCAL_MULTILIB=64 USE_NINJA=false

Re: [android-building] ABI update Failed

I'd recommend contacting your source code provider about that build issue, as that doesn't appear to be AOSP, and the AOSP-based targets don't seem to work in that source tree. In general, the idea behind API/ABI stability is that it's stable across Android releases, so it's not something that

Re: [android-building] Re: Including Library in AOSP/frameworks/base/core

> > *AOSP/build/make/core/tasks/check_boot_jars/package_allowed_list.txt* Be careful if you're planning on being compatible with the CDD , particularly section 3.6 . Adding

Re: [android-building] How to fix or avoid “Android.mk: error: ”libxyz (native:vendor) can not link against libgui.vendor (native:vndk_private)"?

vndk_private libraries are not part of the stable ABI provided to /vendor code, so they're not safe to link against, as it's possible for a VNDK update to be installed without recompiling your vendor code. They still have ".vendor" versions because they're used in the implementations of other VNDK

Re: [android-building] Build error for target db845c-userdebug

That doesn't actually look like the error -- there should be a line starting with `FAILED:` earlier in your console. You can always look at the dedicated error log after the build too, in out/error.log, which should only contain errors (not other build output that finished after the error). - Dan

Re: [android-building] clang++.real stack dump llvm when building nougat-86 android-x86

The common reason that clang crashes come up on this mailing list is that your system has run out of memory during the build -- you might want to watch `top` during a build, or look at the kernel logs afterwards and see if it was OOM-killed or crashed due to some other reason. Sometimes it's a bug

Re: [android-building] Re: Build problem

If that doesn't work, try rebuilding the blob files, my understanding is that they've been fixed. Thanks, Dan On Tue, Jun 15, 2021 at 3:34 PM cptl...@gmail.com wrote: > In vendor/qcom/redfin/proprietary/Android.mk: > > There are several lines that say: > *LOCAL_NOTICE_FILE :=

Re: [android-building] AOSP on aws a1 instance (arm)

Building Android requires a x86-64 machine, it is not supported on arm. As you've found, all of our prebuilt tools are x86-64, and I suspect there are further issues even if you manage to replace them with arm equivalents, as that hasn't been tested. There's limited support in Soong for

Re: [android-building] how to distribute android 10/11 ?

> > While I understand goma support is being deprecated/removed on AOSP > master, is there a way to still be able to use Goma on Android 12 since I > already have a goma client and server setup to work with Android 10. > No, it was removed before Android 12 was branched, so it's not in Android 12

Re: [android-building] Missing "deployagent.inc" in standalone ADB build

We've got a public code search site that can help find these: https://cs.android.com/search?q=deployagent.inc==android%2Fplatform%2Fsuperproject -- in this case, it's defined in packages/modules/adb/Android.bp. - Dan On Mon, Apr 26, 2021 at 9:07 AM Daniel Caruso wrote: > Hi all, > > This

Re: [android-building] How to avoid soong_build run in every build

> > Clang SA is not enabled > This isn't an AOSP tree. My guess is that either an Android.bp, environment variable, or other file/directory soong_build reads is changing every build, but since you have some custom changes to the build system, I can't say for certain. You can check out/soong.log

Re: [android-building] How to run AOSP module in a verbose mode?

You can see a list of all the commands run in the out/verbose.log.gz file after a build completes (at least on modern builds -- for older builds specify `showcommands` on the command line) - Dan On Wed, Mar 3, 2021 at 3:06 PM anands...@gmail.com wrote: > Hi All, > > I am debugging one AOSP

Re: [android-building] Re: Soong build system install & use

Soong is intended to build the Android Platform, so while it can build "unbundled" apps in some contexts, that's only by downloading a large portion of the android platform in order to do so. If you're just building an Android application, it's more than likely not the right tool to use. If you

Re: [android-building] Is there tool like androidmk to convert Android.bp to Android.mk?

It would be simple enough to write one that handles a few use cases, but no, I don't know of one. The Android.mk that ndk-build uses was never the same format as the Android.mk used by the platform builds -- they just happened to share some names (and both used make). - Dan On Wed, Feb 3, 2021

Re: [android-building] how to distribute android 10/11 ?

GOMA support has been removed on the AOSP master branch, and has been replaced with RBE. - Dan On Wed, Feb 3, 2021 at 9:30 AM 唐木村 wrote: > @Dan Willemsen > > >>> The build team does not recommend using Goma -- it does not work > particularly well with the Android build, and we expect to be

Re: [android-building] [Help] clang error while building on Mac OS X El Capitan

Oh, I forgot El Capitan is MacOSX 10.11, which is probably old enough to be supported on that branch. But the MacOS 11.1 SDK shouldn't work on that, and the screenshot claims that you're running 10.15.7 (Darwin 19.6.0). So something doesn't line up here. Regardless, the Linux VM is probably your

Re: [android-building] [Help] clang error while building on Mac OS X El Capitan

Older versions of Android don't build with newer versions of MacOS (the inverse works, you can build newer android versions with older MacOS versions, though that's not usually interesting except for CI servers) If you really need to build a 4+ year old android version on your Mac, then your best

Re: [android-building] android-9.0.0_r8 build error

Did you sync to that from some other revision? The `repo` tool doesn't go back and clean up "removed" / files when changing manifests, so there's some number of dangling symlinks in your worktree. If you remove the five files that it complains about, that's probably enough to get it working. -

Re: [android-building] Not able to build with OUT_DIR?

This is definitely a bug, I uploaded a fix here: https://android-review.googlesource.com/c/platform/build/soong/+/1489596 - Dan On Thu, Nov 5, 2020 at 8:30 AM Yongqin Liu wrote: > Hi, All > > I want to build hikey960 with the default aosp tree with OUT_DIR specified, > but I got following

Re: [android-building] Establishing a build environment

In general, at least with modern versions of Android, you shouldn't run into any issues with any fairly modern linux distribution. And with each release, we've been using fewer and fewer tools from your host distribution to help with this. Please report any problems you run into. Our build farm

Re: [android-building] Replacement for BUILD_BROKEN_PHONY and phony-rebuild in android R

gt; Please sugget any alternative for this so that the kernel intermidiates > are actually generated in the real *out/target/product/**/obj/KERNEL_OBJ* > directory > . > > Regards, > Sumit Kumar > > On Tuesday, August 11, 2020 at 8:43:40 PM UTC+5:30 rr44r...@gmail.com > wrote: > &

Re: [android-building] Replacement for BUILD_BROKEN_PHONY and phony-rebuild in android R

> > but in my case this rebuilding is required . > I think this is where we're disconnected. Forcing rebuilds means that incremental builds will nearly never be as fast as they should be, as the entire tree starting from that phony would need to be rebuilt even if you weren't touching anything

Re: [android-building] Add build tag from a blueprint

Huh, I hadn't even realized that was customizable. In any case, we'd expect global configuration like this to happen in the global/device configuration part of the build (build/make/core/config.mk / product makefiles / BoardConfig.mk, and things they include), not happen in Android.mk / Android.bp

Re: [android-building] aosp master code build fail with "FAILED: out/soong/.primary/soong-ui-build/test/test.passed"

I believe this was fixed with https://android-review.googlesource.com/c/platform/build/soong/+/1384901, so sync your tree and that should go away. - Dan On Fri, Jul 31, 2020 at 11:11 AM Suhang Wang wrote: > I sync the latest code, but get fallowing build fail. > > Mac OS Catalina 10.15.4

Re: [android-building] repo sync fails on the latest Ubuntu

Can you file an issue at https://bugs.chromium.org/p/gerrit/issues/entry?template=GoogleSource+Issue ? Thanks On Wed, Jul 1, 2020 at 2:38 PM Sasha Sirotkin wrote: > Tried 10 times already > > On Wed, Jul 1, 2020, 22:25 Ljube Babunski wrote: > >> Try again to sync. >>> >>> -- >> -- >> You

Re: [android-building] Controlling clang-tidy during the builds

`tidy: true` always runs clang-tidy. If you only want to run it on specific builds, you can omit that property and instead set WITH_TIDY=true in your environment. Unfortunately, that doesn't let you select which code to run tidy on, it'll be run on anything that is built whenever that environment

Re: [android-building] Um........Please help me!!!

Does prebuilts/go/linux-x86/bin/go exist in your source tree? It should exist: https://android.googlesource.com/platform/manifest/+/refs/heads/android-8.1.0_r64/default.xml#552 https://android.googlesource.com/platform/prebuilts/go/linux-x86/+/refs/tags/android-8.1.0_r64/bin/ If it does exist,

Re: [android-building] Re: android.bp privileged prop not work

The Android.bp next to that go file is going to have a dep entry of "soong-cc" (which gives you access to the "android/soong/cc" package). "soong-java" will allow you to reference the "android/soong/java" package. - Dan On Tue, May 19, 2020 at 12:00 AM fanx lin wrote: > Hi Dan, I have one more

Re: [android-building] android.bp privileged prop not work

I believe the problem is that you need to use `Privileged` instead of `privileged`. (In Go, uppercase leading characters in field names are exported and can be used by others, while lowercase ones are considered private) if ctx.AConfig().IsEnvTrue("PAX_PRI"){ > Unless this is for temporary

Re: [android-building] Issue building AOSP with docker running on Mojave

How much RAM have you given docker? `free -h` shows how much is available. IIRC, unlike on linux, because docker on Mac is a VM, it can't grow to use as much as necessary. Our docs list 16GB as the minimum available to the build. - Dan On Thu, May 14, 2020 at 9:30 AM 6020peaks /\/\/\

Re: [android-building] Re: shared library can't be copied into OTATOOLS

Ah, yes, the (internal, unstable) INTERNAL_OTATOOLS_FILES variable has changed meaning. Your local change to copy the files into OTA/tools wouldn't be able to use that variable anymore. I think you'd need the contents of INTERNAL_OTATOOLS_PACKAGE_FILES before the rest of the keys/etc get added to

Re: [android-building] Re: shared library can't be copied into OTATOOLS

No, libc++.so is already included in the otatools zip file that I linked earlier (missing that file would cause issues in our infrastructure using otatools as well). I've confirmed it's there in our internal R trees as well. If it's not showing up in your builds, are you sure you don't have a

Re: [android-building] Re: shared library can't be copied into OTATOOLS

Just adding "libc++" (etc) to that list would trigger the shared library to be installed, but in general (just like PRODUCT_PACKAGES), you shouldn't need to refer to shared libraries because they should be included as a byproduct of being used by one of the binaries in that list (unless they're

Re: [android-building] Re: android build take eons

Yep, this greatly depends on your machine and which target you're building -- there are builds and configurations that can do a full clean build in ~30 minutes, but that's a simpler config and a high-end machine. For the workflow: if i want to include a new library, or write a new driver, or

Re: [android-building] shared library can't be copied into OTATOOLS

What version of AOSP are you using? The current aosp/master builds of otatools.zip definitely have lib64/libbase.so: https://ci.android.com/builds/submitted/6459548/aosp_cf_x86_phone-userdebug/latest And printing out INTERNAL_OTATOOLS_MODULES_WITH_DEPS, it's getting libbase added from

Re: [android-building] "Build sandboxing disabled due to nsjail error. This may become fatal in the future."

Yes, we're familiar with the docker problems, and we'll likely have to live with it for the time being: https://issuetracker.google.com/123210688 . I've removed the above message from master and Android 11 (the other common case was distributions that turned off user namespaces). > I suppose this

Re: [android-building] Memory leak?

The minimum requirements for the build is 16GB of RAM available (and even with that it can be difficult later in the build). The virtualization mention was meant to say that a machine with 16GB of RAM wasn't enough when virtualizing -- the virtual machine would need 16GB allocated to it. `m -j1`

Re: [android-building] equivalent flag "optimize" from Android.bp in Android.mk

`LOCAL_PROGUARD_ENABLED := disabled` would be the equivalent of `optimize: { enabled: false }` - Dan On Tue, Apr 28, 2020 at 11:51 AM Kbc K wrote: > Hi Everyone, > > Soong build system "Android.bp" has a useful flag called "optimize" which > allow to disable compiler optimization. > > I'm

Re: [android-building] Android.mk deprecation status and plan

The AOSP master branch is already reflecting the 2021 work in progress. For the Android 11 (2020) release, https://cs.android.com/android/_/android/platform/build/+/be01cf841e95b8ae396625486b8fb746e5e71492:Deprecation.md;drc=97ec8127081ce0e4b372349caa513ccde9f0fcf7;bpv=0;bpt=0 will be the state:

Re: [android-building] repo init python2.7 or python3.6 which one to use?

When using `--repo-url=git://codeaurora.org/tools/repo.git --repo-branch=caf-stable`, you likely need to use a repo launcher downloaded from codeaurora as well, as it seems the upstream repo launcher is not compatible with it. I'm not familiar with codeaurora though -- there's a chance you could

Re: [android-building] Replacing acloud prebuilt binary

I believe the prebuilt comes from the python support in the platform build system, not setup.py: source build/envsetup.sh m acloud cp out/host/linux-x86/bin/acloud-dev .../acloud - Dan On Tue, Apr 21, 2020 at 11:16 AM Andre D wrote: > Hello, > > Is it possible to replace the prebuilt

Re: [android-building] Classes.jar and classes-header.jar with same module name

That generally should be an error, as we normally require modules to be unique based on the combination of LOCAL_MODULE / LOCAL_IS_HOST_MODULE / LOCAL_MODULE_CLASS (this is in build/make/core/base_rules.mk

Re: [android-building] Android 11` Preview building failed initially

Sorry, apparently that link didn't work. https://groups.google.com/d/msg/android-building/t-70r0IdkZY/sY_NhNd6AgAJ should work better. - Dan On Tuesday, April 7, 2020 at 9:45:15 AM UTC-7 Dan Willemsen wrote: > Yep, that's not intended to be a buildable tag. See my previous messages: >

Re: [android-building] Android 11` Preview building failed initially

Yep, that's not intended to be a buildable tag. See my previous messages: https://groups.google.com/g/android-building/c/t-70r0IdkZY/m/sY_NhNd6AgAJ - Dan On Tue, Apr 7, 2020 at 9:41 AM Varun Kumar wrote: > Hi Dan Willemsen, > > I have used below branch/tag > > *repo init -u

Re: [android-building] Re: Kotlin illegal reflection when building AOSP

Yeah, our minimum requirement is 16GB (ignore the comment about virtual machines there -- the point is that we expect 16GB available for the OS+Build). Even on 16GB we've been seeing some number of issues (hence the workarounds I mentioned

Re: [android-building] Kotlin illegal reflection when building AOSP

The problem isn't the kolin warning, but java.lang.OutOfMemoryError -- your machine ran out of ram. How much ram is available/free before the build starts? Are you using AOSP master, or some other branch/tag? We've made some attempts to help with this on master (running fewer "highmem" actions

Re: [android-building] Android 11` Preview building failed initially

What branch/tag did you use to initialize this source tree? (`repo init ...`) - Dan On Thu, Apr 2, 2020 at 8:57 PM Varun Kumar wrote: > > PLATFORM_VERSION_CODENAME=R > PLATFORM_VERSION=R > TARGET_PRODUCT=aosp_arm64 > TARGET_BUILD_VARIANT=eng >

Re: [android-building] Android Build Teststub -- Linker failed "undefined symbol" Android.bp // maybe missing an include path?

The `libsonic` in Android only includes `sonic.c`: https://cs.android.com/android/platform/superproject/+/master:external/sonic/Android.bp openInputWaveFile and readFromWaveFile are in `wave.c` - Dan On Tue, Mar 31, 2020 at 2:56 PM jrw wrote: > Hello, > > i'm facing an issue with the linker,

Re: [android-building] Re: Dex2oat failed to compile a boot image

As posted with the announcement: This preview is not a full platform update and only for reference. I would not expect this to build or necessarily get any further (eg boot/etc). Only some components have been updated, and that's done on a git-project boundary, so the entire build graph gets

Re: [android-building] verbose: *kati*: slow included makefiles (1037.829698): art/Android.mk

Interesting, 20 minutes still seems super excessive for that  You'll probably save some amount of time sharing a single out directory between devices, but not necessarily if you switch between them frequently, as some of the device-specific files will clobber each other (this was Soong trying to

Re: [android-building] Disallowed PATH tool "m4", but used by flex

As a reminder: This preview is not a full platform update and only for reference. It is not expected that everything will build, as it only contains updates to some projects. For this breakage in particular, you'll need a change like this one, which fixes up a bunch of things around how mesa

Re: [android-building] verbose: *kati*: slow included makefiles (1037.829698): art/Android.mk

So that time was mostly spent initializing the find emulator: verbose: *kati*: init find emulator time: 1311.781312 What that's doing is walking through your source tree and building up a data structure to more efficiently (normally!) run `find` commands. This can become very slow if you've

Re: [android-building] [Android 10] AOSP LibFuzzer example (soong) build problems.

That page expects that you're using AOSP master, not 10.0.0_r30, which did not yet have support for cc_fuzz (it had some support in Make instead, but I know the fuzzing tooling as been changing over time, so I don't know if there are other things on that page that may work differently as well.

[android-building] Re: how to integrate ninja and RBE to support remote build AOSP so fast?

Someone should be reaching out to you directly about RBE, but I'd like to clarify that the remote execution API has more clients than just Bazel (even though it is in the bazelbuild github organization). The rewrapper support in AOSP is unrelated to

Re: [android-building] How to integrate individual packages in cuttlefish

Cuttlefish has opted into the mainline restrictions about using a common system image. If this is meant to be generic, then maybe modify the mainline_system.mk file (or there are variables to provide temporary exemptions). Otherwise, configure your MyService / MyFramework modules to be on the

Re: [android-building] fatal error happens when I'm trying to include header files

libupdater_core (which builds blockimg.cpp) does not depend on libinstall. It gets edify/include and otautil/include through the libedify and libotautil static libraries listed in libupdater_static_libs:

  1   2   3   >