Egor Bogatov [Fri, 10 Feb 2023 16:58:51 +0000 (17:58 +0100)]
Apply jkotas's patch to fix race condition in AllocateManagedClassObject (#81947)
Jan Kotas [Fri, 10 Feb 2023 16:57:30 +0000 (08:57 -0800)]
Fix major performance regression in legacy struct marshalling APIs (#81941)
* Fix major performance regression in legacy struct marshalling APIs
This change adds a simple cache for the marshalling stubs. This cache improves performance of the test mentioned in the issue by 20+x.
Fixes #45100
Co-authored-by: Aaron Robinson <arobins@microsoft.com>
Marek Safar [Fri, 10 Feb 2023 16:27:31 +0000 (17:27 +0100)]
Update area-owners.md (#81958)
Co-authored-by: Theodore Tsirpanis <teo@tsirpanis.gr>
Katelyn Gadd [Fri, 10 Feb 2023 16:24:45 +0000 (08:24 -0800)]
[wasm] Reimplement jiterpreter float-to-int conversions (#81923)
Reimplement float-to-int conversions in the jiterpreter by generating a series of conditional checks that match what clang does, in order to work around the wasm conversion opcode being unspecified for nan/inf.
Fix bug in jiterpreter instrumentation that would suppress the dumped module blob and error message if an unhandled error occurred while generating a block.
Fixes #81900
Michal Strehovský [Fri, 10 Feb 2023 14:43:15 +0000 (23:43 +0900)]
Fix `getMethodSig` for array methods (#81942)
Found by JIT tests. This also affects crossgen2, wonder why we haven't seen it.
Steve Pfister [Fri, 10 Feb 2023 11:47:22 +0000 (06:47 -0500)]
[mono] Move mono.mscordbi subset off the offical build (#81917)
Co-authored-by: Steve Pfister <steve.pfister@microsoft.com>
dotnet-maestro[bot] [Fri, 10 Feb 2023 11:45:39 +0000 (12:45 +0100)]
[main] Update dependencies from dotnet/roslyn-analyzers (#81885)
Co-authored-by: dotnet-maestro[bot] <dotnet-maestro[bot]@users.noreply.github.com>
Egor Bogatov [Fri, 10 Feb 2023 11:28:13 +0000 (12:28 +0100)]
Remove contention from HandleHistogramProfileRand (#81932)
Jakob Botsch Nielsen [Fri, 10 Feb 2023 10:14:34 +0000 (11:14 +0100)]
JIT: Pick some low-hanging fruit in LSRA throughput (#81916)
* Remove LinearScan::RegisterSelection::score
* Remove currentLocation from state
* Remove nextRefPos from state
* Remove rangeEndRefPosition
* Remove lastLocation
* Remove prevRegRec from state
* Remove unused field
* Reorder zero initialization
* Stop resetting foundRegBit
Tom Deseyn [Fri, 10 Feb 2023 10:03:07 +0000 (11:03 +0100)]
Make DotNetBuildFromSource work in more cases. (#81480)
* Make DotNetBuildFromSource work in more cases.
DotNetBuildFromSource is mostly used when this repository is built as part of source-build.
With these changes the flag works in more cases, making it easier to reproduce the
source-build behavior when building the runtime repository directly.
* Microsoft.NETCore.Platforms.Tests: use for .NET Framwork.
* Remove redundant PortableBuild property.
* eng/SourceBuild.props: remove DotNetBuildFromSource defaults.
* PR feedback.
* Use a proper default RuntimeOS and OutputRid when DotNetBuildFromSource is true.
* SourceBuild.props: move KeepNativeSymbols to repo, remove BuildDepPackage.
* SourceBuild.props: add comment describing what properties go where.
Šimon Rozsíval [Fri, 10 Feb 2023 09:24:50 +0000 (10:24 +0100)]
[Android] Enable two networking tests (#81895)
* Avoid crashing when AndroidCryptoNative_RegisterRemoteCertificateValidationCallback is called more than once
* Enable GetAsync_IPv6LinkLocalAddressUri_Success test on Android
* Enable supported cases of SslStream_ServerCallbackNotSet_UsesLocalCertificateSelection on Android
Ilona Tomkowicz [Fri, 10 Feb 2023 08:30:00 +0000 (09:30 +0100)]
Fix. (#81879)
Vitek Karas [Fri, 10 Feb 2023 08:29:04 +0000 (00:29 -0800)]
Validate the Kept attribute against the presence of EEType in NativeAOT (#81830)
Previously we validated the Kept on types against the presence of constructed EEType. But the compiler avoids constructed types as much as possible and the tests do use cases where the type in question is never instantiated (or fully static type). Validating against the presence of EEType (basically a method table) is much closer to the ILLink behavior which validates the presence of the type in IL metadata.
Fixes #80408
Michal Strehovský [Fri, 10 Feb 2023 08:15:47 +0000 (17:15 +0900)]
Fix compiler crashes for invalid IL sequences (#81940)
The JIT test tree has many tests that test invalid IL. We were crashing on some of them in optimized builds.
The problem is that optimized builds look at the program twice - first time when building whole program view, and second time for codegen. The whole program view building doesn't look at the IL in detail and doesn't need to detect invalid IL like "filter block that doesn't end with endfilter". At the same time, scanning phase must precompute the whole world, including situations when invalid code is going to be replaced by a call to a throw helper.
We were ending up in situations where the throw helper was not predicted and we didn't even know how the vtable of `InvalidProgramException` should look like (because it wasn't scanned). A simple fix is to always generate the helper. It's a bit of garbage 99% of time, but probably fine.
Mike McLaughlin [Fri, 10 Feb 2023 05:54:38 +0000 (21:54 -0800)]
Change the default createdump memory enumeration to the fast path (#81922)
* Change the default createdump memory enumeration to the fast path
Also reduces the memory createdump uses generating a dump.
Setting DOTNET_EnableFastHeapDumps=0 opts out going back to the
slow/high memory usage generation.
Issue: https://github.com/dotnet/runtime/issues/72148
* Update src/coreclr/debug/createdump/crashinfo.cpp
Co-authored-by: Noah Falk <noahfalk@users.noreply.github.com>
---------
Co-authored-by: Noah Falk <noahfalk@users.noreply.github.com>
Jeremy Koritzinsky [Fri, 10 Feb 2023 04:01:16 +0000 (20:01 -0800)]
Remove dead code (#81927)
Jeremy Koritzinsky [Fri, 10 Feb 2023 04:01:02 +0000 (20:01 -0800)]
Set the last error in MAPMapPEFile when it fails (#81929)
Adeel Mujahid [Thu, 9 Feb 2023 23:32:54 +0000 (01:32 +0200)]
Declare CoreClrPgoDataArg variable in CI script (#80324)
Co-authored-by: Jeremy Koritzinsky <jkoritzinsky@gmail.com>
Co-authored-by: Jeremy Koritzinsky <jekoritz@microsoft.com>
Eirik Tsarpalis [Thu, 9 Feb 2023 23:23:42 +0000 (23:23 +0000)]
Remove a number of unnecessary intermediate allocations in JsonMetadataServices. (#81903)
dotnet-maestro[bot] [Thu, 9 Feb 2023 22:33:30 +0000 (16:33 -0600)]
[main] Update dependencies from dotnet/icu dotnet/xharness dotnet/emsdk dotnet/msquic (#81827)
* Update dependencies from https://github.com/dotnet/xharness build
20230208.1
Microsoft.DotNet.XHarness.CLI , Microsoft.DotNet.XHarness.TestRunners.Common , Microsoft.DotNet.XHarness.TestRunners.Xunit
From Version 1.0.0-prerelease.23106.1 -> To Version 1.0.0-prerelease.23108.1
* Update dependencies from https://github.com/dotnet/emsdk build
20230207.2
Microsoft.NET.Workload.Emscripten.Current.Manifest-8.0.100-preview.2
From Version 8.0.0-preview.2.23107.1 -> To Version 8.0.0-preview.2.23107.2
* Update dependencies from https://github.com/dotnet/msquic build
20230207.1
System.Net.MsQuic.Transport
From Version 8.0.0-alpha.1.23106.6 -> To Version 8.0.0-alpha.1.23107.1
* Update dependencies from https://github.com/dotnet/emsdk build
20230208.1
Microsoft.NET.Workload.Emscripten.Current.Manifest-8.0.100-preview.2
From Version 8.0.0-preview.2.23107.1 -> To Version 8.0.0-preview.2.23108.1
* Update dependencies from https://github.com/dotnet/icu build
20230208.2
Microsoft.NETCore.Runtime.ICU.Transport
From Version 8.0.0-preview.2.23106.1 -> To Version 8.0.0-preview.2.23108.2
---------
Co-authored-by: dotnet-maestro[bot] <dotnet-maestro[bot]@users.noreply.github.com>
Co-authored-by: Ankit Jain <radical@gmail.com>
Katelyn Gadd [Thu, 9 Feb 2023 22:24:28 +0000 (14:24 -0800)]
[wasm] Optimize out redundant null checks in the jiterpreter (#81811)
* Optimize out duplicate null checks in the jiterpreter
* Fix the last opcode of a jiterpreter trace sometimes executing again after the end of the trace
Katelyn Gadd [Thu, 9 Feb 2023 19:28:55 +0000 (11:28 -0800)]
Disable EnsureCleanEnvironment on wasm since it can't possibly work (#81893)
The related test is also skipped on CI thanks to https://github.com/dotnet/runtime/pull/81860
Eric Erhardt [Thu, 9 Feb 2023 18:59:14 +0000 (12:59 -0600)]
Add Host.CreateEmptyApplicationBuilder (#81728)
* Add Host.CreateEmptyApplicationBuilder
Introduce a way to create an "empty" HostApplicationBuilder that doesn't bring any optional/unnecessary dependencies into the app's closure. This has the same functional behavior as DisableDefaults=true while allowing for the unused code to be trimmed.
Also add CreateApplicationBuilder(HostApplicationBuilderSettings settings) overload to make the Host factory methods complete.
Fix #81280
Radek Doulik [Thu, 9 Feb 2023 18:56:20 +0000 (19:56 +0100)]
Fix vector set_Item (#81898)
Store the xinsert result when needed.
Fixes `System.Numerics.Tests.Matrix4x4Tests.Matrix4x4IndexerSetTest`
test.
Bruce Forstall [Thu, 9 Feb 2023 18:43:05 +0000 (11:43 -0700)]
Cache copies of register sets and data that varies with AVX512 (#81868)
* Cache copies of register sets and data that varies with AVX512
1. Fix availableRegCount to only be stored in LSRA, where it is used.
2. Make copies of rbmAllFloat/rbmFltCalleeTrash to CodeGen and LSRA
classes during construction, so access is "closer", in case TP is
impacted. (This also makes it possible to define inline accessors in
the `emitter` class.)
* Fix initialization ordering
Pavel Savara [Thu, 9 Feb 2023 18:39:48 +0000 (19:39 +0100)]
[browser] improve asset loading (#81886)
* new `enableDownloadRetry` configuration
* fix for actual_downloaded_assets_count assert in use case with `asset.pendingDownload`
* make `AssetEntry` mutable and pass it back to `downloadResource` API callback
Co-authored-by: Marek Fišera <mara@neptuo.com>
Egor Bogatov [Thu, 9 Feb 2023 18:34:05 +0000 (19:34 +0100)]
Fix AllocateRegularStaticBox for FOH (#81857)
Co-authored-by: SingleAccretion <62474226+SingleAccretion@users.noreply.github.com>
Bruce Forstall [Thu, 9 Feb 2023 17:53:37 +0000 (10:53 -0700)]
Introduce insGroup/instrDesc backwards navigation (#80840)
* Introduce insGroup/instrDesc backwards navigation
Currently, it is only possible to walk forwards in insGroup/instrDesc
data: insGroup are in a single-linked list, and variable-sized
instrDesc are packed into a memory array where function `emitSizeOfInsDsc()`
can tell you the size of an instrDesc, and `emitAdvanceInstrDesc()`
can advance an `instrDesc*` forward, but there is no facility for
walking to the previous instrDesc.
This change introduces a `igPrev` pointer so the insGroup list is
doubly-linked. Also, each instrDesc is annotated with 4 or 5 bits of
data indicating how large the previous instruction in the buffer is,
so it is possible to find it by subtracting that size from a `instrDesc*`.
Also, each insGroup maintains an `igLastIns` pointer to the last instruction
in the group, to allow seeding the backwards walk.
The instrDesc backwards walk data is taken from the "small constant" area,
so some small constants get pushed to using larger instrDesc sizes.
Walking backwards is done using `emitGetLastIns()` to find the last
instruction in the function, and `emitPrevID()` to walk to the previous one.
This information is currently only used in the JitDump.
Basic measurement of the benchmarks SPMI collection showed about a 2.5% increase
in emitter memory usage.
* Fix one type
* Put code under EMIT_BACKWARDS_NAVIGATION define
* Disable backwards navigation
Radek Doulik [Thu, 9 Feb 2023 17:49:10 +0000 (18:49 +0100)]
[mono] Fix SIMD multiplication by scalar (#81888)
OP_XBINOP_BYSCALAR expects scalar as second argument.
Fixes `System.Numerics.Tests.GenericVectorTests.TestMultiplicationWithScalar<T>()`
tests.
Will Smith [Thu, 9 Feb 2023 17:45:54 +0000 (09:45 -0800)]
[JIT] X64 - Peephole optimization to skip emitting `movsxd` (#81721)
* Allow GT_CAST for long to small types on 64bit
* Check unsigned
* Minor cleanup
* Added AreUpper32BitsSigned for peephole optimization
* Renamed to AreUpper32BitsFFFF. Simplified check.
* comments
* Formatting::
* Feedback. Added disasm test.
devsko [Thu, 9 Feb 2023 17:45:39 +0000 (18:45 +0100)]
XmlWriter respects NewLineOnAttributes when writing xmlns (#81822)
* XmlWriter respects NewLineOnAttributes when writing xmlns
* formatting
Jakob Botsch Nielsen [Thu, 9 Feb 2023 17:30:00 +0000 (18:30 +0100)]
JIT: Transform SELECT(relop, 1/0, 0/1) to relop (#81880)
Fix #81479
Stephen Toub [Thu, 9 Feb 2023 16:10:34 +0000 (11:10 -0500)]
Improve ConcurrentDictionary performance, in particular for strings (#81557)
* Improve ConcurrentDictionary performance, in particular for strings
- By default, string hash codes are randomized. This is an important defense-in-depth security measure, but it also adds some overhead when strings are used as keys in dictionaries. `Dictionary<>` addresses that overhead by starting out with using non-randomized comparers, and then upgrades to randomized comparers only once enough collisions have been detected. This PR updates `ConcurrentDictionary<>` with similar tricks. The comparer is moved from being stored on the dictionary itself to instead be stored on the Tables object that's atomically swapped when the table grows; that way, the comparer always remains in sync with the hashcodes stored in the nodes in that table. When we enumerate a bucket looking for an existing key as part of an add, we count how many items we traverse, and if that resulting number is larger than the allowed threshold and we're currently using a non-randomized comparer, we force a rehash; that rehash will replace the non-randomized comparer with the equivalent randomized one.
- The `ConcurrentDictionary<>` ctor is improved to presize based on the size of a collection being passed in; otherwise, it might resize multiple times as it's populating the dictionary. The sizing logic is also changed to use the same prime bucket selection size as does `Dictionary<>`.
- The method we were using to compute the bucket for a key wasn't being inlined for reference type keys due to the generic context; that method has been moved to the outer type as a static to avoid the non-inlined call and extra generic dictionary lookup.
- For all key types, we were also paying for a non-inlined ldelema helper call when reading the head node of a bucket; that's been addressed via a struct wrapper with a volatile node field, rather than using Volatile.Read to access the array element.
- We were inconsistent in whether checked math was used in computing the size of the table. In some situations it would be possible to overflow without it being detected, or for it to be detected and manifest in various ways. This simplifies to just always use checked for computing the counts.
- Remove unnecessary try/finally blocks that are leftover from CERs and thread abort protection.
- Deduped some code with calls to helper functions.
* Address PR feedback
* Address PR feedback
* Address PR feedback
Marek Safar [Thu, 9 Feb 2023 16:07:14 +0000 (17:07 +0100)]
Enable same set of analyzers for C# tools that are used by the repo (#81073)
Co-authored-by: Tlakaelel Axayakatl Ceja <tlakaelel.ceja@microsoft.com>
Eirik Tsarpalis [Thu, 9 Feb 2023 15:37:47 +0000 (15:37 +0000)]
Fix a couple of JsonConstructor bugs & clean up JsonParameterInfo implementation. (#81842)
* Fix a couple of JsonConstructor bugs & clean up JsonParameterInfo implementation.
* add test coverage for ignored null values
* fix whitespace
Jo Shields [Thu, 9 Feb 2023 13:41:04 +0000 (08:41 -0500)]
Force building of Mono manifest packages when building via source-build (#81790)
* We can't pass source-build properties externally, they need to come via locally defined properties.
* Indicate that we can consume emsdk manifest via source-build.
Radek Doulik [Thu, 9 Feb 2023 12:40:40 +0000 (13:40 +0100)]
Fix vector's Narrow intrinsics (#81843)
The BCL Vector classes have non-saturating Narrow methods, while wasm instructions are saturating. AFAIK wasm does not have non-saturating narrow instructions. So instead of
i8x16.narrow_i16x8_s
i8x16.narrow_i16x8_u
i16x8.narrow_i32x4_s
i16x8.narrow_i32x4_u
use `v8x16.shuffle` instruction to implement the extract narrow operation.
This fixes `System.Numerics.Tests.GenericVectorTests.Narrow[U]Int*` tests.
Pavel Savara [Thu, 9 Feb 2023 11:41:07 +0000 (12:41 +0100)]
handle PDB files as asset type (#81828)
Jakob Botsch Nielsen [Thu, 9 Feb 2023 10:16:05 +0000 (11:16 +0100)]
JIT: Support containing compares in GT_SELECT for xarch (i.e. start emitting cmov instructions) (#81267)
This adds support for contained compares in GT_SELECT nodes for xarch.
As part of this, also enables if-conversion on xarch.
Michal Strehovský [Thu, 9 Feb 2023 07:49:19 +0000 (16:49 +0900)]
Fix range check in LdelemaRef (#81748)
Also made `LdelemaRef` look closer to how it looks like on the CoreCLR-JIT side.
Found by Pri0 tests.
Bruce Forstall [Thu, 9 Feb 2023 04:27:10 +0000 (21:27 -0700)]
Remove unused PHASE_COMPUTE_PREDS (#81863)
Ankit Jain [Thu, 9 Feb 2023 02:59:35 +0000 (21:59 -0500)]
Skip `tracing/eventpipe/simpleprovidervalidation/` on wasm (#81860)
Fixes https://github.com/dotnet/runtime/issues/81859
Andy Ayers [Thu, 9 Feb 2023 02:41:40 +0000 (18:41 -0800)]
JIT: add concept of edge likelihood (#81738)
Now that FlowEdges are created early and persist, decorate them with likelihood
information early, if we have edge-based PGO data.
We don't use the likelihood for anything yet, but I need to get it in circulation
so I can start working on refining both initial and subsequent consistency of
the data.
Also add a diagnostic checker for likelhood, and a way to enable it. All of
this is off by default.
Ankit Jain [Thu, 9 Feb 2023 01:20:08 +0000 (20:20 -0500)]
evaluate-default-paths.yml: Add paths for wasi (#81817)
* evaluate-default-paths.yml: Add paths for wasi
.. and exclude wasm specific paths from coreclr.
* disable runtime-wasm-optional automatic trigger for PRs
* CI: skip debugger tests on runtime-wasm if they run on runtime pipeline
.. too.
* fix condition
* address review feedback from @kg
* Fix yml issue
Using `Was?Runner*` failed to validate on azdo with:
`/eng/pipelines/common/evaluate-default-paths.yml (Line: 10, Col: 24): While parsing a flow sequence, did not find expected ',' or ']'.`
* fix syntax error in yml
Sven Boemer [Thu, 9 Feb 2023 00:12:10 +0000 (16:12 -0800)]
Fix src/tests/build.sh argument processing (#81836)
The logic to process test filter args isn't working correctly
because the `shift` happens inside a sourced script, and doesn't
actually shift the outer arguments. The right way to do this is
via `__ShiftArgs`.
Without this fix, `tree nativeaot` was working because
`nativeaot` happened to be a supported argument as well. With the
fix, other test subtrees work as expected.
anthonycanino [Wed, 8 Feb 2023 23:10:39 +0000 (15:10 -0800)]
Enable AVX512 Additional 16 SIMD Registers (#79544)
* Change regMask_enum and regMaskTP to unsigned __int64_t on AMD64.
This allows for more registers to be encoded in the register allocator.
* Add upper 16 SIMD registers to allocator.
Commit includes refactoring code to use `const instrDesc *` instead of `instruction`
so information about when EVEX is needed (due to high SIMD registers) is
available to the emitter.
* Limit high SIMD reg to compatible intrinsics lsra build.
* Limit high SIMD reg to compatible intrinsics lsra build.
* Limit high SIMD reg to compatible intrinsics and gentree nodes.
Commit constrains certain hw intrinsics and gentree nodes to use
lower SIMD registers even if upper SIMD registers are available due
to limitations of EVEX encoding for certain instructions.
For example, SSE `Reciprocal` lowers to `rcpps` which does not have an
EVEX encoding form, hence, we cannot allow that hw intrincis node to use
a high SIMD register.
These intrinsics are marked with `HW_Flag_NoEvexSemantics`. Other such
intructions related to masking (typically marked with
`HW_Flag_ReturnsPerElementMask`) also have similar issues (though they
can be replaced with the EVEX k registers and associated masking when
implemented).
In addition, the callee/calleer save registers have also been adjusted
to properly handle the presence and absence of AVX512 upper simd
registers at runtime.
* Fix for X86 throughput.
* Add upper simd stress test to the AVX512 testing pipeline.
* Formatting.
* Fix wrong-sized attr for simd mov instruction.
* Fix non-AMD64 LSRA stress mask.
* Update src/coreclr/jit/compiler.h
Co-authored-by: Bruce Forstall <brucefo@microsoft.com>
* Update src/coreclr/jit/compiler.cpp
Co-authored-by: Bruce Forstall <brucefo@microsoft.com>
* Update src/coreclr/jit/gentree.cpp
Co-authored-by: Bruce Forstall <brucefo@microsoft.com>
* Update src/coreclr/jit/hwintrinsic.h
Co-authored-by: Bruce Forstall <brucefo@microsoft.com>
* Update src/coreclr/jit/target.h
Co-authored-by: Bruce Forstall <brucefo@microsoft.com>
* Update src/coreclr/jit/emitxarch.cpp
Co-authored-by: Bruce Forstall <brucefo@microsoft.com>
* Remove unneeded vars
* Address PR comments.
* Allow `emitinl.h` access to the `rbm` variables.
* Replace RBM_LOWSIMD with `BuildEvexIncompatibleMask`.
* Move AVX512 dependent `targetamd.h` vars into compiler object.
* Fixing some edge cases for `targetamd.h` variables.
* Fix a merge/rebase bug.
* Update src/coreclr/jit/compiler.h
Co-authored-by: Bruce Forstall <brucefo@microsoft.com>
* Update src/coreclr/jit/lsra.cpp
Co-authored-by: Bruce Forstall <brucefo@microsoft.com>
* Update src/coreclr/jit/compiler.h
Co-authored-by: Bruce Forstall <brucefo@microsoft.com>
* Fix nits.
* Trying VM changes.
* VM hack.
* VM hack.
* Revert "VM hack."
This reverts commit
91cf3db9115e94ca1d759045fe4da6a496228cfc.
* Adjust ACTUAL_REG_COUNT based on availability of AVX512.
* Use inline accessor functions instead of macros
Convert from macros to accessor functions for
RBM_ALLFLOAT, RBM_FLT_CALLEE_TRASH, CNT_CALLEE_TRASH_FLOAT.
Convert LSRA use of ACTUAL_REG_COUNT to AVAILABLE_REG_COUNT,
and create an accessor for that value for AMD64 as well.
* Clearifying comments.
---------
Co-authored-by: Bruce Forstall <brucefo@microsoft.com>
Co-authored-by: Kunal Pathak <Kunal.Pathak@microsoft.com>
Jeff Handley [Wed, 8 Feb 2023 21:41:21 +0000 (13:41 -0800)]
Refine corerun instructions. Improve argument parsing in src/tests/build. (#81733)
* Refine corerun instructions. Improve arg parsing in src/tests/build.
* Revert the libraries configuration default. Refine usage and docs.
* Exit argument processing at first unrecognized arg per existing pipeline usage
Nikola Milosavljevic [Wed, 8 Feb 2023 21:08:50 +0000 (13:08 -0800)]
Allow source-build to set UsingToolMicrosoftNetCompilers property (#81180)
* Allow source-build to set UsingToolMicrosoftNetCompilers property
* Update condition to detect full source-build
Fan Yang [Wed, 8 Feb 2023 20:58:22 +0000 (15:58 -0500)]
[Mono] Fix typo in arm64-codegen.h (#81845)
Remove extra `)`
Jakob Botsch Nielsen [Wed, 8 Feb 2023 20:54:47 +0000 (21:54 +0100)]
JIT: Handle promoted struct liveness updates in forward sub (#81789)
This logic needed to be updated together with 7131ef9, without the
update we have a correctness bug.
The new correct and more precise unmarking also allows us to remove a
case from fgMakeOutgoingStructArgCopy. The case is covered by the
last-use optimization one now (specifically it was only happening for
some undone promotion cases).
dotnet-maestro[bot] [Wed, 8 Feb 2023 20:20:01 +0000 (15:20 -0500)]
[main] Update dependencies from dotnet/arcade (#81634)
* Update dependencies from https://github.com/dotnet/arcade build
20230203.1
Microsoft.DotNet.Arcade.Sdk , Microsoft.DotNet.Build.Tasks.Archives , Microsoft.DotNet.Build.Tasks.Feed , Microsoft.DotNet.Build.Tasks.Installers , Microsoft.DotNet.Build.Tasks.Packaging , Microsoft.DotNet.Build.Tasks.TargetFramework , Microsoft.DotNet.Build.Tasks.Templating , Microsoft.DotNet.Build.Tasks.Workloads , Microsoft.DotNet.CodeAnalysis , Microsoft.DotNet.GenAPI , Microsoft.DotNet.GenFacades , Microsoft.DotNet.Helix.Sdk , Microsoft.DotNet.PackageTesting , Microsoft.DotNet.RemoteExecutor , Microsoft.DotNet.SharedFramework.Sdk , Microsoft.DotNet.VersionTools.Tasks , Microsoft.DotNet.XUnitConsoleRunner , Microsoft.DotNet.XUnitExtensions
From Version 8.0.0-beta.23102.4 -> To Version 8.0.0-beta.23103.1
* Update dependencies from https://github.com/dotnet/arcade build
20230206.4
Microsoft.DotNet.Arcade.Sdk , Microsoft.DotNet.Build.Tasks.Archives , Microsoft.DotNet.Build.Tasks.Feed , Microsoft.DotNet.Build.Tasks.Installers , Microsoft.DotNet.Build.Tasks.Packaging , Microsoft.DotNet.Build.Tasks.TargetFramework , Microsoft.DotNet.Build.Tasks.Templating , Microsoft.DotNet.Build.Tasks.Workloads , Microsoft.DotNet.CodeAnalysis , Microsoft.DotNet.GenAPI , Microsoft.DotNet.GenFacades , Microsoft.DotNet.Helix.Sdk , Microsoft.DotNet.PackageTesting , Microsoft.DotNet.RemoteExecutor , Microsoft.DotNet.SharedFramework.Sdk , Microsoft.DotNet.VersionTools.Tasks , Microsoft.DotNet.XUnitConsoleRunner , Microsoft.DotNet.XUnitExtensions
From Version 8.0.0-beta.23102.4 -> To Version 8.0.0-beta.23106.4
* Update dependencies from https://github.com/dotnet/arcade build
20230207.1
Microsoft.DotNet.Arcade.Sdk , Microsoft.DotNet.Build.Tasks.Archives , Microsoft.DotNet.Build.Tasks.Feed , Microsoft.DotNet.Build.Tasks.Installers , Microsoft.DotNet.Build.Tasks.Packaging , Microsoft.DotNet.Build.Tasks.TargetFramework , Microsoft.DotNet.Build.Tasks.Templating , Microsoft.DotNet.Build.Tasks.Workloads , Microsoft.DotNet.CodeAnalysis , Microsoft.DotNet.GenAPI , Microsoft.DotNet.GenFacades , Microsoft.DotNet.Helix.Sdk , Microsoft.DotNet.PackageTesting , Microsoft.DotNet.RemoteExecutor , Microsoft.DotNet.SharedFramework.Sdk , Microsoft.DotNet.VersionTools.Tasks , Microsoft.DotNet.XUnitConsoleRunner , Microsoft.DotNet.XUnitExtensions
From Version 8.0.0-beta.23102.4 -> To Version 8.0.0-beta.23107.1
---------
Co-authored-by: dotnet-maestro[bot] <dotnet-maestro[bot]@users.noreply.github.com>
Tomas Weinfurt [Wed, 8 Feb 2023 20:14:31 +0000 (12:14 -0800)]
don't send server_name when literal IP (#81631)
* don't send server_name when literal IP
* fix unit test
Eirik Tsarpalis [Wed, 8 Feb 2023 19:44:45 +0000 (19:44 +0000)]
Support use of fast-path serialization in combined JsonSerializerContexts. (#80741)
* Support use of fast-path serialization in combined JsonSerializerContexts.
* Strengthen assertions in the nested JsonTypeInfo accessors.
Tanner Gooding [Wed, 8 Feb 2023 19:37:50 +0000 (11:37 -0800)]
Minimally light up Quaternion and Plane for Mono SIMD (#81541)
* Add support for Quaternion
* Add support for Plane
Tanner Gooding [Wed, 8 Feb 2023 19:09:09 +0000 (11:09 -0800)]
Ensure CNS_VEC and BLK are handled for `setLclRelatedToSIMDIntrinsic` (#81760)
Giridhar Trivedi [Wed, 8 Feb 2023 16:45:44 +0000 (22:15 +0530)]
Handle crash when [out] parameter of type int is passed to delegate (#81276)
When a delegate is passed an [out] parameter of type "int" it will crash
since this data type is not handled in emit_managed_wrapper_ilgen.
This will lead to a crash. With this fix [out] parameter of type "int"
is handled after adding an addition case statement to handle MONO_TYPE_I
in emit_managed_wrapper_ilgen. Supporting UT is added.
Co-authored-by: Giridhar Trivedi <giridhar.trivedi@ibm.com>
Lakshan Fernando [Wed, 8 Feb 2023 15:55:57 +0000 (07:55 -0800)]
Initial work to port EventPipe library to NativeAOT (#80382)
* CoreCLR shim files renamed to AOT
* change coreclr references to aot
* Native AOT runtime implementation
* hooks to EP library from AOT
* Container code
* Maanged to Native hooks
* build artifacts
* required changes to NativeAOT to support EP
* changes to common EP source
* sample EventSource test app
* Add EventPipe lib to the linker
* Additional conditions to include the EventPipe library
* Fixing the Checked and Release and Linux builds
* separating maanged and native EventSource switches
* Update src/coreclr/nativeaot/BuildIntegration/Microsoft.NETCore.Native.Unix.targets
Co-authored-by: Michal Strehovský <MichalStrehovsky@users.noreply.github.com>
* FB, build break and test fixes
* fix non-windows build breaks and using paldebugbreak for TODOs
* trying another option to get arounf inttypes.h and PRIu64
* reverting the changes in ep-json file in common code
* Add an NativeAOT EventSource test
* FB and only enabling FEATURE_PERFTRACING in Windows
* fix DiagnosticEventSource test
* FB
* Update src/coreclr/nativeaot/BuildIntegration/Microsoft.NETCore.Native.Windows.targets
Co-authored-by: Michal Strehovský <MichalStrehovsky@users.noreply.github.com>
* Added GuardCF version and a standalone test
* Moving NativeAOT headers away from EventPipe common code
* fix linux build break
* missed a needed definition in Linux
* inline eventpipe common source code
* Adding Evenpipe lib to a lib test
* Small cleanups
* Undo unnecessary changes
* Make sure runtime can build without FEATURE_PERFTRACING
* Fix linux build
---------
Co-authored-by: Michal Strehovský <MichalStrehovsky@users.noreply.github.com>
Jakob Botsch Nielsen [Wed, 8 Feb 2023 15:38:24 +0000 (16:38 +0100)]
JIT: Avoid omitting copies for struct fields passed implicitly byrefs (#81787)
The code does not properly mark the parent struct as address exposed,
but also doing so would completely disable any tracking/optimization for
these locals which we do not want to do.
Fix #81739
Eirik Tsarpalis [Wed, 8 Feb 2023 14:10:17 +0000 (14:10 +0000)]
Fix race condition when accessing nested JsonTypeInfo metadata (#81821)
* Fix race condition when accessing nested JsonTypeInfo metadata
* Improve wording in comments.
* Remove DEBUG conditional
Radek Doulik [Wed, 8 Feb 2023 12:20:29 +0000 (13:20 +0100)]
[json] Use S.R.I vectors in JsonReaderHelper (#81758)
* [json] Use S.R.I vectors in JsonReaderHelper
wasm performance improvement
| measurement | new simd | old simd | nosimd |
|-:|-:|-:|-:|
| Json, non-ASCII text deserialize | 0.4234ms | 0.4280ms | 0.4234ms |
| Json, small deserialize | 0.0376ms | 0.0402ms | 0.0397ms |
| Json, large deserialize | 10.4231ms | 11.1614ms | 11.0021ms |
microbenchmarks perf
--filter *Text.Json*Read*
summary:
better: 170, geomean: 1.120
worse: 15, geomean: 1.080
total diff: 185
* Use SRI vectors only on net7 and later
Viktor Hofer [Wed, 8 Feb 2023 10:48:55 +0000 (11:48 +0100)]
Update version of Microsoft.DotNet.ApiCompat.Task (#81772)
* Update version of Microsoft.DotNet.ApiCompat.Task
* Update Version.Details.xml
Jakob Botsch Nielsen [Wed, 8 Feb 2023 10:39:03 +0000 (11:39 +0100)]
SPMI: Warn when comparing throughput of JITs compiled with native PGO (#81716)
Fix #81690
dotnet bot [Wed, 8 Feb 2023 08:07:32 +0000 (00:07 -0800)]
Localized file check-in by OneLocBuild Task: Build definition ID 679: Build ID 2107735 (#81700)
* Localized file check-in by OneLocBuild Task: Build definition ID 679: Build ID 2105824
* Localized file check-in by OneLocBuild Task: Build definition ID 679: Build ID 2107383
* Localized file check-in by OneLocBuild Task: Build definition ID 679: Build ID 2107383
Michal Strehovský [Wed, 8 Feb 2023 06:34:39 +0000 (15:34 +0900)]
Fix handling of the generic byreflike constraint (#81684)
Some support for NativeAOT was written in #67783 but the test didn't actually pass because of these missing pieces.
Ankit Jain [Wed, 8 Feb 2023 06:12:15 +0000 (01:12 -0500)]
[wasm] Pin the version of chrome used for tests - `109.0.5414.119` (#81814)
* [wasm] Pin the version of chrome used for tests
Related: https://github.com/dotnet/runtime/issues/81792
The debugger tests are breaking with the latest stable chrome version
`110.0.5481.77`. As a temporary workaround to keep the CI green, pin the
version to the previous working one.
* fix windows url
Zoltan Varga [Wed, 8 Feb 2023 04:01:29 +0000 (23:01 -0500)]
[mono][aot] Optimize constrained calls made from gsharedvt methods. (#79339)
The calls are of the form:
.constrained T_GSHAREDVT
callvirt <method>
Whenever T_GSHAREDVT is a reference or value type is only known at runtime.
Previously these were handled by passing the arguments to a JIT icall which
computed the target method and did a runtime invoke.
Added 2 optimizations:
* Precompute the data which depends only on the type and the method,
store it in an rgctx slot and pass it to the JIT icall.
* Add a fastpath for simpler cases which makes an indirect call
from generated code.
Katelyn Gadd [Wed, 8 Feb 2023 03:54:38 +0000 (19:54 -0800)]
Add an optimized C implementation of STFLD_O that uses the correct kind of write barrier (#81806)
This PR moves most of the jiterpreter's STFLD_O implementation into a C function that is responsible for also doing the null check. As a bonus that function is able to use the correct kind of write barrier (though it's not clear to me whether the previous one was broken in any way).
dotnet-maestro[bot] [Wed, 8 Feb 2023 01:43:43 +0000 (17:43 -0800)]
[main] Update dependencies from dotnet/cecil (#81756)
* Update dependencies from https://github.com/dotnet/cecil build
20230206.4
Microsoft.DotNet.Cecil
From Version 0.11.4-alpha.23080.1 -> To Version 0.11.4-alpha.23106.4
* Remove Microsoft.DotNet.Cecil.Pdb from being referenced since it now is part of Microsoft.DotNet.Cecil package
---------
Co-authored-by: dotnet-maestro[bot] <dotnet-maestro[bot]@users.noreply.github.com>
Co-authored-by: Tlakollo <tlcejava@microsoft.com>
Jan Vorlicek [Wed, 8 Feb 2023 00:47:32 +0000 (01:47 +0100)]
Prevent unwinding through stack bottom (#81770)
When processing unhandled exception on the most recent Alpine 3.17,
the libunwind doesn't stop at the bottom
frame of the main thread (the caller of `main`) and tries to unwind
further. The reason is that the method is missing dwarf unwind
information, so the libunwind falls back to using RBP chain, but the RBP
points to a garbage and so it ends up crashing with SIGSEGV.
While the missing DWARF unwind info seems to be a bug in the Alpine 3.17
(older ones work fine), we can prevent issues like this by stopping at
the hosting API boundary and not trying to unwind past that. This is
what this PR does.
Austin Wise [Tue, 7 Feb 2023 23:50:01 +0000 (15:50 -0800)]
[Jiterpreter] Add support for TryGetHashCode intrinsic (#81644)
#80520 added support in the Jiterpreter for the intrinsic that underlies RuntimeHelpers.GetHashCode. This PR adds similar support for the intrinsic that underlies RuntimeHelpers.TryGetHashCode.
For reference RuntimeHelpers.TryGetHashCode was added in #80059. It is currently only used in ConditionalWeakTable.TryGetValue.
Michal Strehovský [Tue, 7 Feb 2023 23:48:09 +0000 (08:48 +0900)]
Baseline most of the Pri0 test suite (#81685)
I made a pass over all Pri0 tests except the JIT directory.
Vitek Karas [Tue, 7 Feb 2023 23:23:05 +0000 (15:23 -0800)]
Revert "Fix generic parameter data flow validation in NativeAOT (#81532)" (#81783)
This reverts commit
e71a4fb10d7ea6b502dd5efe7a8fcefa2b9c1550.
Adeel Mujahid [Tue, 7 Feb 2023 23:14:12 +0000 (01:14 +0200)]
Fix condition for module registration export (#81765)
* FIx condition for module registration export
* Typo
Stephen Toub [Tue, 7 Feb 2023 22:47:28 +0000 (17:47 -0500)]
Fix two compilation constants for .NET 8 (#81775)
Eric Erhardt [Tue, 7 Feb 2023 22:28:15 +0000 (16:28 -0600)]
EnableAotAnalyzer for System.Diagnostics.EventLog (#81724)
* EnableAotAnalyzer for System.Diagnostics.EventLog
Contributes to #75480
* EnableAotAnalyzer for System.Diagnostics.PerformanceCounters
* Avoid boxing T.
devsko [Tue, 7 Feb 2023 21:39:47 +0000 (22:39 +0100)]
Fix EncodingExtensions.Convert for sliced single segment ROSequence (#81665)
Fixes https://github.com/dotnet/runtime/issues/45346
Jeremy Koritzinsky [Tue, 7 Feb 2023 21:19:13 +0000 (13:19 -0800)]
Use MiniDumpWriteDump to create dumps on Windows (#81509)
Ankit Jain [Tue, 7 Feb 2023 20:49:22 +0000 (15:49 -0500)]
Improve CI path-changes-based-triggers to work better (#79127)
## Background:
`eng/pipelines/common/evaluate-default-paths.yml` is used to specify named subsets of paths, and `evaluate-changed-paths.sh` decides which subsets have "changed files". And these subsets are used in conditions for the various jobs to determine when they should be run.
---
`evaluate-changed-paths.sh: Add support for include+exclude combined`
## Problem
This script currently supports `--include`, and `--exclude`
parameters to determine if we have any changed paths.
```
Scenarios:
1. exclude paths are specified
Will include all paths except the ones in the exclude list.
2. include paths are specified
Will only include paths specified in the list.
1st we evaluate changes for all paths except ones in excluded list. If we can not find
any applicable changes like that, then we evaluate changes for included paths
if any of these two finds changes, then a variable will be set to true.
```
As described above, these two are evaluated exclusively.
For example:
```
include: [ '/a/b/*' ], exclude: [ '/a/b/d.txt' ]
```
- This would return true if there are changes:
- `[ '/a/b/c.txt' ]` - caught by `include`
- or `[ '/r/s.txt' ]` - caught by `exclude`
- but it would also return true for `[ '/a/b/d.txt' ]` because `include`
does not consider the `exclude` list.
## Solution:
- This commit adds a new `--combined` parameter which essentially supports
`$include - $exclude`. Thus allowing exact more constrained path
specification.
- It is passed in addition to `include`, and `exclude`.
Given:
```
include: [ '/a/b/*' ], exclude: [ '/a/b/d.txt' ]
```
- For the following path changes:
- `[ '/a/b/c.txt' ]` - `true`
- `[ '/r/s.txt' ]` - `false` because this does not fall in
`$include - $exclude`
- `[ '/a/b/d.txt' ]` - `false` - excluded in `$include - $exclude`
The implementation is trivially implemented by passing both the lists to
`git diff` which supports this already.
---
- Track subset name changes in the ymls
- Update `wasm` jobs to have tighter trigger conditions
- This results in wasm specific changes only triggering relevant wasm jobs. For example, if there are wasm debugger changes then only the wasm debugger jobs will be triggered. Or if there are only workload manifest[1] changes then Wasm.Build.Tests will be triggered only.
Thays Grazia [Tue, 7 Feb 2023 20:17:02 +0000 (17:17 -0300)]
[wasm][debug] Avoid pausing on breakpoint while invoking methods. (#81162)
* Trying to avoid pausing on breakpoint while invoking methods.
* Adding test case for it as discussed with @radical offline.
* Creating new tests and fixing its behavior.
* Addressing @radical comments offline.
* same behavior for stepping, disable global_stepping while evaluating expression.
* Renaming function as suggested by @radical
Marie Píchová [Tue, 7 Feb 2023 19:52:36 +0000 (20:52 +0100)]
libnuma-dev (#81764)
Fix #81706
Thays Grazia [Tue, 7 Feb 2023 19:48:36 +0000 (16:48 -0300)]
[wasm][debugger] Skip generated wrappers when trying to step into/out with JustMyCode enabled (#81715)
* Skip generated wrappers when trying to step.
* Changing to use System.Diagnostics.DebuggerNonUserCode
dotnet-maestro[bot] [Tue, 7 Feb 2023 19:10:39 +0000 (13:10 -0600)]
[main] Update dependencies from dotnet/icu dotnet/xharness dotnet/runtime-assets dotnet/emsdk dotnet/msquic dotnet/hotreload-utils (#81755)
* Update dependencies from https://github.com/dotnet/icu build
20230206.1
Microsoft.NETCore.Runtime.ICU.Transport
From Version 8.0.0-preview.2.23103.1 -> To Version 8.0.0-preview.2.23106.1
* Update dependencies from https://github.com/dotnet/xharness build
20230206.1
Microsoft.DotNet.XHarness.CLI , Microsoft.DotNet.XHarness.TestRunners.Common , Microsoft.DotNet.XHarness.TestRunners.Xunit
From Version 1.0.0-prerelease.23102.1 -> To Version 1.0.0-prerelease.23106.1
* Update dependencies from https://github.com/dotnet/runtime-assets build
20230207.1
Microsoft.DotNet.CilStrip.Sources , System.ComponentModel.TypeConverter.TestData , System.Drawing.Common.TestData , System.Formats.Tar.TestData , System.IO.Compression.TestData , System.IO.Packaging.TestData , System.Net.TestData , System.Private.Runtime.UnicodeData , System.Runtime.Numerics.TestData , System.Runtime.TimeZoneData , System.Security.Cryptography.X509Certificates.TestData , System.Text.RegularExpressions.TestData , System.Windows.Extensions.TestData
From Version 8.0.0-beta.23103.1 -> To Version 8.0.0-beta.23107.1
* Update dependencies from https://github.com/dotnet/emsdk build
20230207.1
Microsoft.NET.Workload.Emscripten.Current.Manifest-8.0.100-preview.2
From Version 8.0.0-preview.2.23081.3 -> To Version 8.0.0-preview.2.23107.1
* Update dependencies from https://github.com/dotnet/msquic build
20230206.6
System.Net.MsQuic.Transport
From Version 7.0.0-alpha.1.22406.1 -> To Version 8.0.0-alpha.1.23106.6
* Update dependencies from https://github.com/dotnet/hotreload-utils build
20230207.1
Microsoft.DotNet.HotReload.Utils.Generator.BuildTool
From Version 1.1.0-alpha.0.23080.1 -> To Version 1.1.0-alpha.0.23107.1
---------
Co-authored-by: dotnet-maestro[bot] <dotnet-maestro[bot]@users.noreply.github.com>
Egor Bogatov [Tue, 7 Feb 2023 18:36:14 +0000 (19:36 +0100)]
PGO: Reduce number of block counters (#81731)
Ivan Povazan [Tue, 7 Feb 2023 18:14:50 +0000 (19:14 +0100)]
Enable AppleAppBuilder to bundle iOS applications built for NativeAOT runtime (#81711)
Fixes: https://github.com/dotnet/runtime/issues/80910
Hong Li [Tue, 7 Feb 2023 17:41:35 +0000 (09:41 -0800)]
update owner for are:System.Transactions (#81640)
Anton Firszov [Tue, 7 Feb 2023 17:28:13 +0000 (18:28 +0100)]
Harden and re-enable DualMode socket tests (#80715)
- UDP tests are moved to a non-parallel collection, since there is no way to prevent parallel UDP tests from intercepting each other's packets. This only increases the execution time by ~2 seconds.
- TCP interferences are prevented by a utility PortBlocker, which creates and binds a "shadow" socket of the opposite address family to prevent sockets in parallel tests to bind to the same port.
- Accept and SendTo tests are refactored to use SocketTestHelperBase<T>, reducing duplicate code.
Kevin Jones [Tue, 7 Feb 2023 17:25:31 +0000 (12:25 -0500)]
Add support for SP800-108 CTR Key Derivation Function
This introduces a class for producing values per the "KDF in Counter Mode" section in NIST Special Publication 800-108r1 (Recommendation for Key Derivation Using Pseudorandom Functions).
The `SP800108HmacCounterKdf` class is part of the inbox cryptography library on .NET 8. Based on demonstrated need for older TFMs, the type is also being exposed via a NuGet package: Microsoft.Bcl.Cryptography. This package may, in the future, contain other types that belong as part of inbox cryptography but have a demonstrated need to be available to older TFMs.
Tanner Gooding [Tue, 7 Feb 2023 15:56:56 +0000 (07:56 -0800)]
Fix a bug around Vector4.Distance and Sse41.Insert lowering (#81725)
* Adding a regression test for dotnet/runtime#81585
* Fixing a bug around Sse41.Insert lowering
* Ensure that Distance/DistanceSquared are correctly imported
* Account for another case around Sse41.Insert chain folding
Jeremy Koritzinsky [Tue, 7 Feb 2023 13:12:06 +0000 (05:12 -0800)]
Move all of the ref-pack generators to use the "LatestVS" Roslyn build so they don't pull in prebuilts in source-build (#81561)
Buyaa Namnan [Tue, 7 Feb 2023 12:59:20 +0000 (04:59 -0800)]
Fix warnings found with new analyzer CA1860: Prefer Length/Count/IsEmpty property check over Any() (#81583)
Co-authored-by: Marek Safar <marek.safar@gmail.com>
Vitek Karas [Tue, 7 Feb 2023 09:07:33 +0000 (01:07 -0800)]
Fix generic parameter data flow validation in NativeAOT (#81532)
[This is a revert of a revert of #80956 with additional fixes for #81358)
This reworks how generic parameter data flow validation is done in the NativeAOT compiler.
Previously generic data flow was done from generic dictionary nodes. Problem with that approach is that there's no origin information at that point. The warnings can't point to the place where the problematic instantiation is in the code - we only know that it exists.
Aside from it being unfriendly for the users, it means any RUC or suppressions don't work on these warnings the same way they do in linker/analyzer.
This change modifies the logic to tag the method as "needs data flow" whenever we spot an instantiation of an annotated generic in it somewhere.
Then the actual validation/marking is done from data flow using the trim analysis patterns.
The only exception to this is generic data flow for base types and interface implementations, that one is done on the EEType nodes.
Note that AOT implements a much more precise version of the generic data flow validation as compared to linker/analyzer. See the big comment at the beginning of `GenericParameterWarningLocation.cs` for how that works.
Due to an issue with DependencyInjection, this change also implements a behavior where if a method or field is reflection accessible, the compiler will perform generic argument data flow on all types in the signature of the method/field (which it normally wouldn't do). See #81358 for details about the issue and discussions on the fix approach.
Test changes:
Adds the two tests from linker which cover this functionality.
Change the test infra to use token to compare message origins for expected warnings. Consistently converting generic types/methods into strings across two type systems is just very difficult - the tokens are simple and reliable.
Changes the tests to avoid expecting specific generic types/methods formatting in the messages - again, it's too hard to make this consistent without lot of effort. And the tests don't really need it.
Adds a smoke test which has a simplified version of the DI problem from https://github.com/dotnet/runtime/issues/81358.
dotnet-maestro[bot] [Tue, 7 Feb 2023 08:22:47 +0000 (09:22 +0100)]
[main] Update dependencies from 7 repositories (#81536)
Co-authored-by: dotnet-maestro[bot] <dotnet-maestro[bot]@users.noreply.github.com>
Co-authored-by: Marek Safar <marek.safar@gmail.com>
Co-authored-by: tlakollo <tlakaelel_axayakatl@outlook.com>
Co-authored-by: Buyaa Namnan <bunamnan@microsoft.com>
Michal Strehovský [Tue, 7 Feb 2023 02:48:42 +0000 (11:48 +0900)]
Add a simple size test (#81517)
The size of Hello World is currently 1.8 MB. This is achieved by avoiding any non-field reflection in the codepaths that are part of a hello world. If reflection comes into picture, the size jumps by several 100 kB because suddenly we need a lot of code to support that. This is a smoke test to detect those situations. We're getting proper size testing in the dotnet/performance repo with trend histories, etc., but the invariant for Hello World is easy to check for and nice to gate commits on.
Stephen Toub [Tue, 7 Feb 2023 01:28:31 +0000 (20:28 -0500)]
Avoid using `file` in generated regexes pre-C# 11 (#81726)
Katelyn Gadd [Tue, 7 Feb 2023 01:09:47 +0000 (17:09 -0800)]
[wasm] Migrate jiterpreter hit counting to C (#81729)
* Move jiterpreter trace hit counting to C for better performance
Aaron Robinson [Tue, 7 Feb 2023 00:22:33 +0000 (16:22 -0800)]
Fix contract violations. (#81727)
Removes GC_TRIGGERS contract for functions that
do not trigger a GC. The removed contract definition
appears to be out of date.
Fallout from #81571
Michal Strehovský [Tue, 7 Feb 2023 00:03:30 +0000 (09:03 +0900)]
Match ILScanner with CorInfoImpl (#81683)
ILScanner and the JitInterface have to be in sync wrt exceptions throw or there are problems. Fixes one of the Pri0 tests.
Andy Ayers [Tue, 7 Feb 2023 00:00:07 +0000 (16:00 -0800)]
JIT: rename flowList to FlowEdge and encapsulate all fields (#81621)
Restructure this type in anticipation of modifying how it represents data.
Rename some of the fields (and now related accessor methods) because subsequent
changes may do things like add the target block or successor edge links.
Also, remove some stray references to fgComputePreds
Michal Strehovský [Mon, 6 Feb 2023 23:54:55 +0000 (08:54 +0900)]
Adjust workaround for event source resource stripping (#81528)
The used approach only worked in debug builds. `InlineableStringResource` is not one of the nodes that survives from scanning phase to compilation phase. Make the compiler think `GetResourceString` was reflected on instead.
Sven Boemer [Mon, 6 Feb 2023 22:49:12 +0000 (14:49 -0800)]
Run nativeaot tests in helix (#81439)
This changes the existing NativeAot smoke test runs to run in
helix. The multi-module test run (only supported on windows)
now runs only in the checked configuration, to avoid
having to add a separate windows build script invocation inside
one of the yaml templates.