The nullability of parameter 'destinationType' in
TypeConverter.CanConvertTo(...) was changed from
not-nullable to nullable during the development of .NET 6.

Since a destination type supported by this TypeConverter
can never be null, a NotNullWhenAttribute is added to the
'destinationType' parameter when the result value of
TypeConverter.CanConvertTo(...) is 'true'.

Fix #63186

* Remove reference to deleted setup-stress-dependencies.cmd/sh

Automatically download and copy coredistools.dll to the libraries testhost.

* Remove reference to setup-stress-dependencies.cmd/sh scripts

This is in the libraries gcstress pipeline.

It is no longer necessary.

* Set a huge timeout for GCStress

* Add nativeaot trim annotations experiment

* Fix mono libraries

* Update src/libraries/System.Reflection.Emit.Lightweight/ref/System.Reflection.Emit.Lightweight.cs
Co-authored-by: NEric Erhardt <eric.erhardt@microsoft.com>

* FB on attribute message

* FB

* missed nativeaot message change
Co-authored-by: NEric Erhardt <eric.erhardt@microsoft.com>

* [wasm] Fix a race condition in adding payloads for helix

The issue shows up only in Wasm.Build.Tests for the
`TestUsingWorkloads!=true` case, as `clang --version` failing with:

```
/datadisks/disk1/work/A0100914/p/build/emsdk/upstream/bin/clang: error while loading shared libraries: libLLVM-13git.so: cannot open shared object file: No such file or directory (TaskId:231)
```

For this non-workloads testing case, we copy `emscripten` from the
system copy to the git checkout. And then use that directory for the
payload.

The build logs show that the missing file (`libLLVM-13git.so`) does get
copied as part of copying `/usr/local/emscripten/emsdk` to
`$(RepoRoot)/src/mono/wasm/emsdk`. But the file, and a few others seem
to be missing in the final helix payload.

We add `emsdk` to the payload for target path `build/emsdk`.

But a recent change also added `node` for this case with a target path
`build/emsdk/node`, with an overlapping path with `build/emsdk`. I
believe this is causing an issue where these directories are being
processed in parallel, and cause some files get missed.

This commit:

1. Add `node` only when needed (skip WBT for example);
2. Use a non-overlapping path for `node`, `build/emsdk-node`.

Fixes https://github.com/dotnet/runtime/issues/65956

* Change `_HelixLocalNodePath` evaluation order
Co-authored-by: NRadek Doulik <radekdoulik@gmail.com>

I observed that the inlined call to clear_brick_table in clear_region_info took more CPU samples than necessary - it's about 7x faster to call memset than it is to code a straighforward loop.

* add missing GC.SuppressFinalize(this) to FileStream.DisposeAsync

* DerivedFileStreamStrategy.DisposeAsync does not need to call _fileStream.BaseDisposeAsync, as FileStream.DisposeAsync calls Dispose(false) now

* Use ConvereterStrategy.Object in ObjectConverter

* address feedback

* [mono][wasm] Allow AOTing methods with catch clauses.

This works as follows:
* During EH, when a catch clause is found in an AOTed method, the
  EH state is saved into TLS and a c++ exception is thrown.
* The C++ exception is caught by the AOTed method, and the landing
  pad code calls mono_llvm_resume_exception_il_state ().
* That call will run the catch clause and the rest of the method
  code using the interpreter, storing the possible return value
  back into the AOTed method's stack frame.
* After the call, the method skips the rest of its code, and
  returns immediately to its caller.

* Fix console bench sample.

* Fix landing pads.

* Fix issues.

* Add support for filter clauses.

* Implement all wasm return conventions.

* Fix arg/local write back.

* Avoid throwing a c++ exception from do_jit_call () so the caller can clean up the interpreter stack.

* Disable AOTing some more assemblies on CI.

* Rename llvmonly EH functions to mini_llvmonly_ for clarity.

* Improve unwinding through interpreter frames.

Instead of throwing a c++ exception from mono_handle_exception ()
when an exception is caught in AOTed code, set context->has_resume_state,
so the intepreter will normally unwind until exiting interpreted code.
Then throw the c++ exception after the call to interp_exec_method ()
to unwind through the runtime code and the AOTed frames
which can't handle the exception.

* null-check the redirect context before using.

* tweak the comment

* do not allocate context if InitializeContext has unexpected results.

Fixes #65573

* Fix GC hole with multi-reg local var stores

Change #64857 exposed an existing problem where when generating code
for a multi-reg GT_STORE_LCL_VAR, if the first register slot was not
enregistered, but the second or subsequent slots was, and those non-first
slots contained GC pointers, we wouldn't properly add those GC pointers
to the GC tracking sets. This led to cases where the register lifetimes
would be killed in the GC info before the actual lifetime was complete.

The primary fix is to make `gtHasReg()` handle the `IsMultiRegLclVar()`
case. As a side-effect, this fixes some LSRA dumps that weren't displaying
multiple registers properly.

There are about 50 SPMI asm diffs on win-arm64 where register lifetimes
get extended, fixing GC holes.

I also made `GetMultiRegCount()` handle the `IsMultiRegLclVar()` case.

I made a number of cleanup changes along the way:
1. Fixed two cases of calling `gcInfo.gcMarkRegSetNpt` with regNumber, not regMaskTP
2. Marked some functions `const`
3. Improved some comments
4. Changed "ith" to "i'th" in comments which still doesn't read great,
but at least I'm not left trying to parse "ith" as an English word.
5. Use `OperIsScalarLocal()` more broadly
6. Renamed `gtDispRegCount` to `gtDispMultiRegCount` to make it clear
it only applies to the multi-reg case.

Fixes #65476.

* Update src/coreclr/jit/gentree.cpp
Co-authored-by: NKunal Pathak <Kunal.Pathak@microsoft.com>
Co-authored-by: NKunal Pathak <Kunal.Pathak@microsoft.com>

Move diagnostics for invalid GeneratedDllImportAttribute usage to generator instead of analyzer (#65915)

* Add regex "unit tests" test project

This follows the convention used by the networking tests, which typically have two distinct test projects per library: functional tests and unit tests.  The former are what we typically refer to as our tests for a library, whereas the latter build product source into the test project in order to directly validate internals (an alternative to this is to use InternalsVisibleTo, but that negatively impacts trimming and makes it more challenging to maintain a property boundary for the functional tests).  All the existing tests are moved unedited into the FunctionalTests, and a new UnitTests project is added with some initial tests for the RegexTreeAnalyzer code. The generator parser tests were also consolidated into the functional tests, as there's no longer a good reason for those few tests to be separate.

I fixed a few bugs in RegexTreeAnalyzer as a result.  In particular, we were over-annotating things as potentially containing captures or backtracking because in the implementation we were using the lookup APIs meant to be used only once all analysis was complete.  This doesn't have a negative functional impact, but it does negatively impact perf of compiled / source generator, which then generate unnecessary code.  We were also incorrectly conflating atomicity conferred by a grandparent with atomicity conferred by a parent; we need MayBacktracks to reflect only the atomicity directly contributed by a node, not by its parent's influence, as we need the parent to be able to understand whether the child might backtrack.

* Add some unit tests for RegexFindOptimizations

* Address PR feedback

This corrects two uses of RandomNumberGenerator that resulted in slight
bias in their results.

* Make Is*Project properties unambiguous

Currently these properties exist which categorize projects:
- IsReferenceAssembly: The project's parent directory is 'ref'
- IsGeneratorProject: The project's parent directory is 'gen'
- IsTestProject: The project is located somewhere under a '/tests/'
                 directory and the project name's suffix is either
                 '.Tests' or '.UnitTests'.
- IsTrimmingTestProject: Same as IsTestProject but the project name's
                         suffix is '.TrimmingTests'.
- IsTestSupportProject: The project is located somewhere under a
                        '/tests/' directory and the above IsTestProject
                        property is false.
- UsingMicrosoftNoTargetsSdk: The project uses the NoTargets msbuild SDK
- UsingMicrosoftTraversalSdk: The project uses the Traversal msbuild SDK
- IsRuntimeAssembly: True when all above is false
- IsSourceProject: True when the project's parent directory is 'src'

The IsRuntimeAssembly and IsSourceProject properties meanings are
ambiguous and the property names aren't consistent (IsReferenceAssembly
vs IsGeneratorProject).

I'm changing the above to the following:
- **IsReferenceSourceProject: The project's parent directory is 'ref'
                        or the parent of the parent directory is 'ref'**
- IsGeneratorProject: The project's parent directory is 'gen'
- IsTestProject: The project is located somewhere under a '/tests/'
                 directory and the project name's suffix is either
                 '.Tests' or '.UnitTests'.
- IsTrimmingTestProject: Same as IsTestProject but the project name's
                         suffix is '.TrimmingTests'.
- IsTestSupportProject: The project is located somewhere under a
                        '/tests/' directory and the above IsTestProject
                        **and IsTrimmingTestProject** props are false.
- UsingMicrosoftNoTargetsSdk: The project uses the NoTargets msbuild SDK
- UsingMicrosoftTraversalSdk: The project uses the Traversal msbuild SDK
- IsSourceProject: **True when all above is false.**

* React to PR feedback

* Update GenAPI usage and code clean-up

* Setup pipeline for testing.

* First test with the Maui Version.

* Remove perf-job parameter version typo.

* Try another way of doing the output variable.

* Third times the charm?

* Add a second test for two to be done in one run.

* Fix variable problem with xplat-setup.yml

* Third time was not the charm...

* Try another setup.

* Only include one of the tests as the perf setup is failing potentially due to this.

* Try 2 variable attempts and write out potential get methods.

* Remove failed thing and split them to all be seperate and always run.

* Move variables into the second job stuff.

* Won't run when specified under the job parameters.

* Try hardcoded values to see if it is the output or input that seems to not be connected properly.

* Try new variable format.

* Missed bracket that wasn't found until after most of the way through.

* Maui installs are failing, work around this for variable testing.

* Need to keep mono build enabled.

* Revert "Need to keep mono build enabled."

This reverts commit ff32e46c080efb8a103907de6dd1a13cd0d1be04.

* Revert "Maui installs are failing, work around this for variable testing."

This reverts commit 6dbf6b0ee26d8047d14be86e12e56872c7dab7cc.

* Temp fix for workload install.

* Removed breaking variable test as the new format works along with changing to only having one variable for testing. This should allow us to have access to the maui version in the perf setup for testing.

* Full run test. Still using perf repo fork.

* Add comment explaining RetrievedMauiVersion.

* Add maui version to the iOS runs.

* Undo testing changes to prepare for merge once install is working again.

* Update comment in BinaryFormatterWriter to indicate a NRT bug won't be fixed

* Update comment to reflect known issue in .NET Framework

* Update comment to indicate a known issue

It is the case that the IR supports the "zero" node for structs
in two positions: on the RHS of an assignment (InitBlk form) and
under a return, in case the ABI return type is scalar.

Meanwhile, assertion propagation "blindly" replaced structs with
zeroes, and so workarounds had to be applied in order for the IR
to remain valid, in the form of the NO_CSE flag, applied either
explicitly (multi-reg returns) or implicitly (ADDR(LCL) created by
"impNormSturctVal" for call args).

This was:

a) A CQ problem in cases where we forgot to clear the NO_CSE flag
   when the node's parent was updated, say after inlining.
b) A burden for enabling struct LCL_VAR arguments, as one had to
   remembered to mark them NO_CSE in all situation.

This change fixes the problem by deleting propagation of zeroes for
local uses, instead propagating them as part of their parents (ASGs
and RETURNs).

This has the CQ benefits of not being affected by stale NO_CSEs and
the drawback of not participating in the "chained" propagation, where
we first copy-propagated something, and then zero-propagated into the
new local as well. These cases seem rather rare, so I decided not to
spend TP on fixing them by looking at the copy assertions in the new
code.

This change also deletes the zero propagation code for SIMDs. It was
only useful in cases we had a promoted SIMD field that was zero-inited
via an InitBlk on the parent struct. That promotion code was (and is)
creating nodes that look like integral constants, except they are of
TYP_SIMD. We should delete that form and use proper SIMD zero nodes
instead, and design the propagation story for them separately.

* Adding Regex.IsMatch(Span) and RegexRunner.Scan() methods

* Addressing some PR Feedback

* Fixing case where span-based APIs are called from a precompiled regex and adding a test for it.

* Adding Tests for IsMatch span overload

* Refactor Run<TState> and Run to share centralized logic

* Emit IsBoundary and IsECMABoundary instead of exposing them.

* Address Emitter changes feedback.

* Addressing PR Feedback.

* Addressing most of the PR comments

* Fix issue where runtextbeg and length do not match the size of the input

* Add Precompiled tests for all overloads of methods.

* Fix precompiled test failing in non-Windows due to new lines.

* Apply suggestions from code review
Co-authored-by: NStephen Toub <stoub@microsoft.com>

* Addressing more PR Feedback

* Fix protected Scan method

* Addressing feedback and adding more tests

* Fix few failing tests in Mono runtime
Co-authored-by: NStephen Toub <stoub@microsoft.com>

This allows us to remove a bunch of huge tables that weren't actually needed.
Removes support for setting MONO_EXTERNAL_ENCODINGS env var which is discouraged.
Also removes the code for .mdb debug symbol support which was already disabled everywhere.

Fix OSR jitstress failures by disabling STRESS_LCL_FLD for two cases:
* for Tier0 methods with patchpoints (because patchpoint info generation
does not produce the right offsets when locals are padded)
* for OSR locals in OSR methods (because they live on the Tier0 frame and
so can't have their storage altered.

Enable struct promotion for OSR locals. In most cases this ends up being
dependent promotion. Enabling independent promotion will take more work.

* GetTypeField should return ELEMENT_TYPE_BYREF.

* Remove setting of token2 in all cases. This value is not used in any
scenario and therefore is more confusing than helpful. Add a comment about
the expectations of the returned values.

* Setup pipeline and setup for testing.

* Updated scenario names for new scenario name setup and added no animation tests.

* Revert "Setup pipeline and setup for testing."

This does a few things:
- The previous changes for Antimirov changed the representation from being represented as a DfaMatchingState to instead being a list of states.  Unfortunately, this came at the expense of a significant increase in overhead for the DFA mode.  This commit melds the old and new design, by still using a CurrentState struct that can be either a DFA state or an NFA state, but with a simplified representation that lacks any processing.  The processing is left to two separate structs that implement static abstract interface methods, as was previously done, using that to provide streamlined, generic-specialized processing for the DFA path.
- Overhauled how we switch from DFA mode to NFA mode.  Previously, the builder would store whether we were in NFA mode, and any matching after it had been switched over immediately started in NFA mode.  We now always start in DFA, and only switch over to NFA if we try to create a new state but doing so would exceed the graph size constraint.
- Overhauled how we handle the outer/inner loop logic.  Previously we wanted to remain in a tight inner loop and so would use some "leeway" to allow us to go beyond the graph size constraint in order to avoid having to check too frequently whether we'd exceeded that size.  Now, the transition function itself returns whether we need to exit out to the outer loop in order to upgrade from DFA to NFA, and until/unless that happens, we can stay in the inner loop.
- Avoids inlining huge amounts of NFA code that can result in worse DFA code
- Renamed Brzozowski and Antimirov to DFA and NFA.
- Added a lot of comments to SymbolicRegexMatcher
- Fixes "watchdog" issue with captures
- A few other renames / cleanups.

* Set android to only build one RID (android-arm64).

* Update the mv file path for the android apk.

* Add --self-contained to get rid of android build warning.

* Add comment about dotnet restore being done before publish.

* Test a potential fix for the maui install failures.

* Only run the app build.

* Reenable rest of the pipeline.

* Remove usages of native bootstrapping
* Make sure cmake is in the path for mono wasm builds
Co-authored-by: NAdeel Mujahid <3840695+am11@users.noreply.github.com>

We currently either guess at some of this state based on the immediate surrounding nodes (e.g. whether the immediate child backtracks) or we do potentially-expensive walks each time we need to check (e.g. walking all ancestors until root to determine whether a given node is to be considered atomic).  This changes the code to do a pass over the graph to compute the relevant information, which can then be used by the code generators any time they need to access that information.  This provides the code with faster and more accurate answers.

As JanV noticed during an independent investigation,
TypeGeneratorTests historically used to produce an enormous amount
of stdout spew due to individually reporting the details of all
passing test cases; for the entire set of 1501 tests the total
size of the spew was about 150 MB. In combination with the change
to capture stdout spew in the xml test reports this effectively
canceled out all previous wins achieved via test merging - in fact,
the tests became twice as slow compared to before the merging.

This modification only reports mismatches i.e. failures where we
don't receive the expected answer in one of the test cases. This
reduces the total spew from 150 MB to just over a megabyte and
restores the perf improvement achieved via test merging.

Thanks

Tomas

This change adds initial provisions for platform detection similar
to library logic to CoreCLRTestLibrary and marks a few
architecture-conditional tests with ConditionalFact attributes.

While I must admit I'm not happy about the IL representation of
the ConditionalFact attributes, it's technically using the same
representation as the library tests do and it only affects a handful
of tests. During consolidation of the remaining tests we'll
continue working on an easier solution even though it may require
diverging further away from Xunit.

I'm not yet removing the CLRTestTargetUnsupported clauses from
the corresponding project files, that can only be done after
switching these tests over to use the merged wrappers (my next
change) as the legacy XUnit wrapper generator (in particular the
test execution script generator) relies on those properties.

Thanks

Tomas

* Tmp change for debugging.

* Added chromium, needed for running debugger tests.

* Corrected line breaking, removed suso.

* Removed update duplicate.

* Prevent regex on null.

* Set special env variable only for containers.

* Locally docker build crashes without it.

* Applied @eerhardt suggestion.

* Moved env assignment to a place that is executed by onCreateCommand.sh

* Revert accidential new line removal.

* Upgrade is unnecessary, works without it.

* Moved all logic to C# code.

* Remove duplicated line.

* Applied @thays's and @lewing's comments.

* Enabling @radical approach for chromium installation.
Co-authored-by: NAnkit Jain <radical@gmail.com>