* Share the stopwatch code between `mtouch` and `mmp`;
* Add more details on linker steps. Sadly substeps are executed on the
metadata so they can't be reported individually;
* Add a few places where timestamps where missing to get better precision
on the linking part;
On my test project we were calling `IsArchEnabled` 334,713 times which
made the enumerator called enough to show on the memory usage (13MB),
even if each instance is only 40 bytes.
Caching the result makes the method called only 5 times.
* [runtime] build interp-, icalltable- and ilgen libraries so they can be consumed in interpreter configuration
* [mtouch] add --interpreter configuration support
* [xharness] Add support for removing defines from cloned project files.
* [xharness] Run monotouch-test, mscorlib and mini tests with the interpreter too.
They're ignored for now, which means we won't run them on the bots.
* [mtouch] We're using dlsym when using the interpreter.
* [xharness] enable mini regression tests with interpreter on CI
Fix https://github.com/xamarin/xamarin-macios/issues/3653 differently as
`UseShellExecute` cannot be used when redirecting output so the original
fix [1] caused an exception which affected it from both macOS and windows
(thru XMA) instead of being an issue only with the later.
It's not clear how the original [1] fix was validated successfully, it's
possible than an older version of mono did not throw (since that
limitation seems windows specific).
[1] https://github.com/xamarin/xamarin-macios/pull/3781
Fixes this test failure:
1) Failed : Xamarin.MTouch.MT0113_linker
The error 'MT0113: Native code sharing has been disabled for the extension 'testServiceExtension' because the managed linker settings are different between the container app (None) and the extension (All).' was not found in the output:
Message #1 did not match:
actual: 'Native code sharing has been disabled for the extension 'testServiceExtension' because the remove-dynamic-registrar optimization differ between the container app (default) and the extension (false).'
expected: 'Native code sharing has been disabled for the extension 'testServiceExtension' because the managed linker settings are different between the container app (None) and the extension (All).'
which happens because:
* Removing the dynamic registrar requires the linker, so removal of the dynamic registrar is disabled if the linker is not disabled
* This results in the app and appex having different values for the remove-dynamic-registrar option
* Thus the error message.
Technically either error is correct, but I prefer the previous one (about the
linker), because it directly assigns blame (the linker setting). Figuring out
what has to change (the linker setting) when the error message complains about
an optimization is not so straight forward for users.
Assemblies that satisfy all of these conditions:
* Are not in the container project.
* Are in multiple app extensions.
can (and should) be bundled in the container app.
Fixes https://bugzilla.xamarin.com/show_bug.cgi?id=58873
- https://github.com/xamarin/xamarin-macios/issues/3367
- App Store will now fail builds if you add in a 32-bit dylib
- If you are a 32-bit app you don't need the 64-bit part of your fat
dylib anyway
- Add --optimize=-trim-architectures to allow customization of behavior, as not everyone
uses app store
In addition, while writing tests for this is was noticed that mmp tests did not "really" run Release configuration correctly in most cases. Fixing this turned out to be a bit of a pain, but necessary to correctly test this (and other things).
- Turns out that /p:configuration:debug is not sufficient to tell mmp to
do the right thing
- That, in most projects, sets the DebugSymbols property, which really
is what is checked.
- However, two of our projects did not have that, so we always did
release mmp work.
- Removed configuration property for tests and added real "Release"
configuration option
This fixes a startup crash in code shared app extensions due to having the
wrong value set in the runtime (the dynamic registrar was removed, but the
executable didn't know it).
When code sharing is enabled, only the container app/target will have a
LinkContext.
So make sure the static registrar finds that link context when it needs
information from the linker.
https://github.com/xamarin/xamarin-macios/issues/3514
There's only one static registrar now, so there's no need for code to select
which one, just create the one and only.
Also unify this code between mtouch and mmp.
* [mtouch/mmp] Give users more control over optimizations, and share more code between mtouch and mmp.
1. Add an --optimize flag to mtouch/mmp that allows users to select which
optimizations to apply (or not). This makes it easier to add future
optimizations, and allow users to disable any optimization that causes
problems without having to disable many other features.
2. Share as much optimization code as possible between mtouch and mmp. This
immediately gives a benefit to mmp, which has three new optimizations only
mtouch had: NSObject.IsDirectBinding inlining, IntPtr.Size inlining and
dead code elimination.
This results in ~6kb of disk space saved for a linked Xamarin.Mac app:
* link sdk: [Debug][1], [Release][2]
* link all: [Debug][3], [Release][4]
Testing also verifies that monotouchtest ([Debug][5], [Release][6]) has not
changed size at all, which means that no default optimizations have changed
inadvertedly.
[1]: https://gist.github.com/rolfbjarne/6b731e3b5ca6170355662e6505c3d492#link-sdk-mac--debug
[2]: https://gist.github.com/rolfbjarne/6b731e3b5ca6170355662e6505c3d492#link-sdk-mac--release
[3]: https://gist.github.com/rolfbjarne/6b731e3b5ca6170355662e6505c3d492#link-all-mac--debug
[4]: https://gist.github.com/rolfbjarne/6b731e3b5ca6170355662e6505c3d492#link-all-mac--release
[5]: https://gist.github.com/rolfbjarne/6b731e3b5ca6170355662e6505c3d492#monotouchtest-iphonedebug64
[6]: https://gist.github.com/rolfbjarne/6b731e3b5ca6170355662e6505c3d492#monotouchtest-iphonerelease64
* [tools] Don't enable the IsDirectBinding optimization by default for Xamarin.Mac apps, it's not safe.
* Fix whitespace issues.
* [doc] Document optimizations.
* Officially support optimizations by adding them to the Versions.plist.
* [linker] Improve IntPtr.Size inliner + dead code eliminatior and add tests.
* Properly handle operands for the ldc_i4_s instruction (they're sbyte).
* Fix less-than condition to actually do a less-than comparison.
* Make sure to look up the bitness in the Target, not the Application, since
the Application's value will be incorrect when building fat apps (both
Is32Build and Is64Build will be true).
* Remove unnecessary checks for the IntPtr.Size inliner: this optimization
does not depend on other instructions than the IntPtr.get_Size call, so
remove the checks that verify surrounding instructions. This makes the
IntPtr.Size inliner kick in in more scenarios (such as the new tests).
* Add tests.
* [tests] Add mmp tests for optimizations.
* [tests] Fix XM optimization tests.
* [tests] Fix test build error.
* [mtouch] Don't allow building for 32-bit when deployment target is >= 11. Fixes#57966.
Also bump maccore to get an mlaunch error for launching a 32-bit app in a 64-bit-only simulator.
https://bugzilla.xamarin.com/show_bug.cgi?id=57966
* [tests][msbuild] Make sure all Info.plists have deployment targets.
Otherwise we get different behavior (32-bit allowed or not) depending on which
Xcode is used to build.
* [mtouch] Default to 64-bit arch if not specified and targeting iOS 11+.
* [tests] Tweak tests to either specify a deployment target < 11 or not build a 32-bit arch.
* [mtouch] Improve how we make sure native symbols aren't stripped away. Fixes#51710 and #54417.
* Refactor required symbol collection to store more information about each
symbol (field, function, Objective-C class), and in general make the code
more straight forward.
* Implement support for generating source code that references these symbols,
and do this whenever we can't ask the native linker to keep these symbols
(when using bitcode). Additionally make it possible to do this manually, so
that the source code can be generated for non-bitcode platforms too (which
is useful if the number of symbols is enormous, in which case we might
surpass the maximum command-line length).
* Also make it possible to completely ignore native symbols, or ignore them on
a per-symbol basis. This provides a fallback for users if we get something
right and we try to preserve something that shouldn't be preserved (for
instance if it doesn't exist), and the user ends up with unfixable linker
errors.
* Don't collect Objective-C classes unless they're in an assembly with
LinkWith attributes. We don't need to preserve Objective-C classes in any
other circumstances.
* Implement everything for both Xamarin.iOS and Xamarin.Mac, and share the
code between them.
* Remove previous workaround for bug #51710, since it's no longer needed.
* Add tests.
https://bugzilla.xamarin.com/show_bug.cgi?id=54417https://bugzilla.xamarin.com/show_bug.cgi?id=51710
* [mtouch] Make sure to only keep symbols from the current app when code sharing.
This fixes a build problem with the interdependent-binding-projects test when
testing in Today Extension mode.
In 11390f119c we stopped setting the force flag
when the cache was invalid, because we'd delete the cache anyway, and it was
determined that deleting the cache was enough.
Unfortunately it's not, because some output is not in the cache, and might not
get correctly updated.
Scenario:
* User builds app.
* User changes some build option (for instance switching off incremental
builds).
* User does an insignificant change in a source file for the executable
process.
* User builds app again (without cleaning). This will rebuild the exe, but
since the change was insignificant, all the IL, except the MVID, would
remain identical.
* mtouch would see that the command-line options changed, and invalidate the
cache. This would delete the cache, and everything would be rebuilt,
including AOT-compiling the assemblies again.
* When the time came for mtouch to copy assemblies to the app directory,
mtouch would realize that the existing .exe in the app (which was not
deleted because it's not in the cache, but the actual output directory) was
only insignificantly different (only the MVID was different, which our cache
logic knows to ignore when comparing assemblies), so it wouldn't copy the
.exe to the .app.
* At runtime we'd assert, because the MVID in the aot-compiled code was
different from the MVID in the assembly:
error: Failed to load AOT module '(null)' while running in aot-only mode: doesn't match assembly.
* The exact assert varies depending on which build option changed. Other
variations:
Failed to load AOT module '(null)' while running in aot-only mode: compiled against GC (4, while the current runtime uses GC sgen)
* Assertion at /Users/builder/data/lanes/4691/0719ced1/source/xamarin-macios/external/mono/mono/metadata/metadata.c:1118, condition `idx < t->rows' not met
Because of this I'm reverting 11390f119c, and
once again setting the force flag when the cache is invalid. It might be
overkill, but it's the safest option (cache invalidation is after all the only
hard problem in computer science), and bugs are very annoying and
timeconsuming to track down.
https://bugzilla.xamarin.com/show_bug.cgi?id=54973
* [mtouch] Allow code sharing assemblies from multiple locations if they're identical. Fixes#56498.
We disallow code sharing when the same assembly (based on name) is referenced
from multiple paths, but poke a hole in this logic by allowing the same
assembly from multiple paths when those assemblies are 100% identical, since
that should be 100% safe.
https://bugzilla.xamarin.com/show_bug.cgi?id=56498
* [tests] Comment out assert that asserts due to another bug.
If both an extension and the container app (or multiple extensions) reference
the same binding assembly for a framework, then we'd error out with an
internal error when trying to copy the framework from both locations to the
container app.
So instead detect when we're trying to copy multiple identical (by comparing
the on-disk contents) frameworks, and only copy one of them.
We'll still show an error if the frameworks are different, but now a nice
MT1035 error with a proper error description instead of an internal error.
https://bugzilla.xamarin.com/show_bug.cgi?id=56635
It does not make sense to support incremental builds for the simulator (since
no AOT compilation is done), it just makes the test matrix more complicated.
So simplify things by removing support for incremental builds.
We also ignore any (other) --assembly-build-target arguments, because building
to frameworks doesn't make sense either in the simulator.
https://bugzilla.xamarin.com/show_bug.cgi?id=55712
This dramatically decreases the size of watchOS apps built for debug when
using frameworks, because it ends up removing all the bitcode from
Mono.framework.
https://bugzilla.xamarin.com/show_bug.cgi?id=55256
Of particular importance is if we're building for LLVM or not: this fixes a
bug where we wouldn't pass --llvm to the AOT compiler when compiling
assemblies to frameworks (which we do when sharing code).
https://bugzilla.xamarin.com/show_bug.cgi?id=55555
* Remove the MT0008 test, since the error will never be shown again.
* Check non-existent root assemblies and report MT0018 instead of MT0007 if
they look like command-line arguments.
* Collect all MT0018/MT0007 errors before reporting any of them.
Put simulator assemblies in MonoBundle/simulator for frameworks, so that we
can have a single framework that contains both device and simulator
assemblies without assemblies conflicting between device and simulator.
The device assemblies continue in the same place, in the MonoBundle directory,
so no additional checks are needed on device.
Also stop using `mdb` as the name for debug symbols and remove
> static MdbReader mdb_reader;
since we're not mkbundl'ing mtouch anymore.
Related to https://github.com/xamarin/xamarin-macios/pull/2002 for mmp
* [mtouch] Only iOS has version 8 (and earlier OS versions that don't support frameworks).
* [mtouch] Update logging to be less confusing when logging about WatchKit extensions.
* [mtouch] Don't compile P/Invoke wrappers for extensions that are sharing code.
The container app already has the P/Invoke wrappers.
We want to copy the aot data for both the 32-bit and the 64-bit versions of an
assembly even if the 32-bit and 64-bit versions of the assembly are identical.
https://bugzilla.xamarin.com/show_bug.cgi?id=54499
No frameworks should be bundled in WatchKit 1 extensions, they should be
bundled in the container (iOS) app.
This broke when merging the equivalent fix for master into the framework-sdk
branch (i.e. a broken merge).