-
Notifications
You must be signed in to change notification settings - Fork 4.7k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Add NativeAOT build and runtime pack for linux-bionic #86781
Conversation
Build and runtime pack for linux-bionic (Android without the Java part). I tried not to regress the existing CoreCLR Android build that is based on a crossrootfs (dotnet#56622) - the if's basically deal with that. Note that it's likely broken anyway (dotnet#66562).
Tagging subscribers to this area: @agocke, @MichalStrehovsky, @jkotas Issue DetailsBuild and runtime pack for linux-bionic (Android without the Java part). I tried not to regress the existing CoreCLR Android build that is based on a crossrootfs (#56622) - the if's basically deal with that. Note that it's likely broken anyway (#66562). Cc @dotnet/ilc-contrib
|
@MichalStrehovsky can you kick off an official build of this branch to make sure it's ok? |
Triggered this build: https://dnceng.visualstudio.com/internal/_build/results?buildId=2188707&view=results I think it looks good? It failed at signing, but I assume it's because I triggered it and it's not the main branch. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
One question but looks good to me otherwise.
@@ -84,6 +84,8 @@ build_native() | |||
exit 1 | |||
fi | |||
|
|||
cmakeArgs="-C $__RepoRootDir/eng/native/tryrun.cmake $cmakeArgs" |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
why is this necessary?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
The same reason we had to add a tryrun for iDevices - cmake cannot execute these detectoids because it's a crossbuild. Normally the infra passes tryrun to crossbuild but this is not considered a crossbuild by the infra (see the first issue I linked in top post - we don't pass -cross
to build.sh to build Bionic)
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Yes, what I meant was why is this not necessary for the mono-based Android build today?
I looked a bit deeper now and I think it is because we're not using any of these cmake checks in the System.Native build so we don't hit this, e.g. we explicitly skip the checks that would cause binaries to be run during configure on mobile and manually set the results:
runtime/src/native/libs/configure.cmake
Lines 572 to 605 in c525645
if(CLR_CMAKE_TARGET_IOS) | |
# Manually set results from check_c_source_runs() since it's not possible to actually run it during CMake configure checking | |
unset(HAVE_SHM_OPEN_THAT_WORKS_WELL_ENOUGH_WITH_MMAP) | |
unset(HAVE_ALIGNED_ALLOC) # only exists on iOS 13+ | |
unset(HAVE_CLOCK_MONOTONIC) # only exists on iOS 10+ | |
unset(HAVE_CLOCK_REALTIME) # only exists on iOS 10+ | |
unset(HAVE_FORK) # exists but blocked by kernel | |
elseif(CLR_CMAKE_TARGET_MACCATALYST) | |
# Manually set results from check_c_source_runs() since it's not possible to actually run it during CMake configure checking | |
# TODO: test to see if these all actually hold true on Mac Catalyst | |
unset(HAVE_SHM_OPEN_THAT_WORKS_WELL_ENOUGH_WITH_MMAP) | |
unset(HAVE_ALIGNED_ALLOC) # only exists on iOS 13+ | |
unset(HAVE_CLOCK_MONOTONIC) # only exists on iOS 10+ | |
unset(HAVE_CLOCK_REALTIME) # only exists on iOS 10+ | |
unset(HAVE_FORK) # exists but blocked by kernel | |
elseif(CLR_CMAKE_TARGET_TVOS) | |
# Manually set results from check_c_source_runs() since it's not possible to actually run it during CMake configure checking | |
unset(HAVE_SHM_OPEN_THAT_WORKS_WELL_ENOUGH_WITH_MMAP) | |
unset(HAVE_ALIGNED_ALLOC) # only exists on iOS 13+ | |
unset(HAVE_CLOCK_MONOTONIC) # only exists on iOS 10+ | |
unset(HAVE_CLOCK_REALTIME) # only exists on iOS 10+ | |
unset(HAVE_FORK) # exists but blocked by kernel | |
elseif(CLR_CMAKE_TARGET_ANDROID) | |
# Manually set results from check_c_source_runs() since it's not possible to actually run it during CMake configure checking | |
unset(HAVE_SHM_OPEN_THAT_WORKS_WELL_ENOUGH_WITH_MMAP) | |
unset(HAVE_ALIGNED_ALLOC) # only exists on newer Android | |
set(HAVE_CLOCK_MONOTONIC 1) | |
set(HAVE_CLOCK_REALTIME 1) | |
elseif(CLR_CMAKE_TARGET_BROWSER OR CLR_CMAKE_TARGET_WASI) | |
set(HAVE_FORK 0) | |
else() | |
if(CLR_CMAKE_TARGET_OSX) | |
unset(HAVE_ALIGNED_ALLOC) # only exists on OSX 10.15+ | |
else() |
So this is fine since we'll override the values for e.g. HAVE_SHM_OPEN_THAT_WORKS_WELL_ENOUGH_WITH_MMAP
but we should probably clean this up and move all of the detection into tryrun.cmake. Would you mind filing an issue for that?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Filed #86829.
It should work for non-main builds as well but I checked and I've seen the same error on some 8.0-preview5 branch builds as well so I think it is unrelated. |
Build and runtime pack for linux-bionic (Android without the Java part).
I tried not to regress the existing CoreCLR Android build that is based on a crossrootfs (#56622) - the if's basically deal with that. Note that it's likely broken anyway (#66562).
Cc @dotnet/ilc-contrib