* Use OpenSSL 3 on macOS

* Update documentation

* Restart CI

The virtual vector instruction used internally by the JIT were leaking out and reported as unsupported

Fixes #68222

CallArgABIInfo::ArgNum is essentially unnecessary now that arguments are
stored in order so remove it.

Also address some feedback from the refactoring and fill out some
function headers.

* Exposing ExpM1, Exp2, Exp2M1, Exp10, Exp10M1, LogP1, Log2P1, and Log10P1

* Adding tests covering ExpM1, Exp2, Exp2M1, Exp10, Exp10M1, LogP1, Log2P1, and Log10P1

* Ensure Log10P1Test64 is 64-bit only

Fix #58874
Fix #64802
Fix #68157

Currently the JIT's constant folding (gtFoldExprConst and VNs
EvalOpSpecialized) assumes that BSWAP16 zero extends into the upper 16
bits. This was not the case, and in fact the behavior of BSWAP16
depended on platform.

Normally this would not be a problem since we always insert normalizing
casts when creating BSWAP16 nodes, however VN was smart enough to remove
this cast in some cases (see the test).

Change the semantics of BSWAP16 nodes to zero extend into the upper 16
bits to match constant folding, and add a small peephole to avoid
inserting this normalization in the common case where it is not
necessary.

Fixes #67723
Subsumes #67726

It allows them to be inlined by the R2R compiler.

We load the library dynamically after #55037 so it is not necessary to have the export lib available for the build.

Number of HRESULTs and error messages related to GAC, CAS and other concepts remove in .NET Core
existed just to provide mapping in Marshal.GetExceptionForHR API. They were otherwise unused.

Deleting these HRESULTs is observable behavior change in Marshal.GetExceptionForHR. It should be
acceptable change given that these HRESULTs are never returned by the runtime and very unlikely to
be used by user code.

* Avoid duplicate Strings.resx import & code-cleanup

Fixes https://github.com/dotnet/runtime/issues/68218

Makes assemblies with String.resx resources smaller by avoiding the duplicate resources.

* Overhaul when/where we check for timeouts

Regex timeouts have never been about guaranteeing exact timeout handling; rather, they're about avoiding catastrophic backtracking.  As such, we already allow an O(n) amount of work in many cases between timeout checks.  This change formalizes that, such that we now check for a timeout at every place where we could do at least an O(n) amount of work, which essentially means every time we match at a new index and every time we backtrack.  It also removes the counting logic that previously translated only 1 out of 1000 CheckTimeout calls into a timeout check; now every CheckTimeout will query the current tick count.

* Address PR feedback

* Enable System.Text.Json tests on netfx (#63803)

* Enable System.Text.Json tests on netfx

* use NETFRAMEWORK define

* disable another two test cases after rebase

* disable last test case which repros only on CI for me

* add p2p only on netfx

* Remove hardcoded assembly qualified names

* Address feedback
Co-authored-by: NKrzysztof Wicher <kwicher@microsoft.com>

Remove one of the quirks added in the call args refactoring.

* Remove all Unix-specific files.

* Throw PNSE on non-Windows.

* Merge the Windows-specific files into their formerly cross-platform counterparts.

* Remove all mentions of Unix in the tests.

* Remove two always-on defines.

* Merge two item groups in the project file.
No reason to sort them; the list is already unsorted.
And rename a remaining formerly Windows-specific file.

* Remove the NoCOMWrappers files.

* Fail on unsupported platforms when a library is trying to be loaded.

* Fix compile errors.

* Small changes in the project file.

* Remove two meaningless asserts.

* Run BinaryFormatter tests on SDC types only on Windows.

* Use `[ThreadStatic]` in Gdip.ThreadData.

* Remove `TargetsAnyOS`.
Co-authored-by: NViktor Hofer <viktor.hofer@microsoft.com>

* SDC project file clean-up

* Remove `!!` from System.Drawing.Common.
Co-authored-by: NViktor Hofer <viktor.hofer@microsoft.com>

* Exclude all markdown files from triggering CI

Since 08/09/2021 it's now possible to use wild cards in exclusion patterns: https://docs.microsoft.com/en-us/azure/devops/release-notes/2021/sprint-192-update#support-for-wild-cards-in-path-filters.

Fixing our exclude yml statements to ignore all markdown files in the repo and not just the ones under docs/.

See https://github.com/dotnet/runtime/pull/68221

* Use OperatingSystem.IsWindows() in Uri.

* Remove platform specific files.

* Remove platform specific compilation for System.Private.Uri.

* Exposing MaxMagnitudeNumber, MaxNumber, MinMagnitudeNumber, and MinNumber

* Ensure Max, MaxMagnitude, Min, and MinMagnitude tests are covering the relevant scenarios

* Moving the floating-point AssertEqual helper from the Math/MathF tests into AssertExtensions

* Adding tests covering MaxMagnitudeNumber, MaxNumber, MinMagnitudeNumber, and MinNumber

* Disabling a few Half tests due to an issue around -0

* Disabling a few Half tests due to an issue around +0

* Fixing 67993 as the test helper had a bug where it used the wrong target type

* Fixing up the doc comment for Min/MaxMagnitudeNumber and Min/MaxNumber to clarify it doesn't propagate NaN

Now, any SuperPMI failure in the replay, asmdiffs, and checked/release
asmdiffs pipeplines is ignored, and doesn't cause AzDO to treat the entire
job as failed. The job does get a "!" icon, but that's only if you visit the
AzDO UI: the GitHub page on a PR doesn't see that failure.

Now that we have more experience with these pipelines, and we are using
replay and asmdiffs with JIT PRs, we want the failures to be more visible,
and treated as failures.

Note that asm diffs should not be treated as failures.

* Delete AssemblySpec::m_wszCodeBase

This was only used for assembly loading using hosting APIs that can be short-circuited to load directly without going through AssemblySpec

* Delete unnecessary ownedFlags argument

* Delete AssemblyBuilder.CheckContext checks

This check is trying to ensure that the Types passed in are somewhat well-behaving by verifying that
Module and Assembly properties are non-null. This validation is not done consistently and it has very
value. There are many other ways custom types can misbehave.

Also, this saves allocations of the object[] arrays for the params argument.

* Simplify Exception.Source

Simplify Exception.Source to avoid Reflection.Emit dependencies in Exception implementation

* Delete unnecessary clone of CustomAttributeBuilder enumeration

* Factor out simple ModuleBuilder and TypeBuilder forwardering methods

* Delete ComVisible annotations in Mono

* Delete redundant FCall

* Add initial version of {Last}IndexOfAnyExcept

These are functional but not vectorized.  At least some of these should be vectorized for at least some data types subsequently, but that's a more intensive change.  Once that's in, we can update a few places to use these, e.g. Regex should end up using any of the overloads that are vectorized.

* Fix comments

* Fix a race condition in the thread pool

There is a case where on a work-stealing queue, both `LocalPop()` and `TrySteal()` may fail when running concurrently, and lead to a case where there is a work item but no threads are released to process it. Fixed to always ensure that there's a thread request when there was a missed steal. Also when `LocalPop()` fails, the thread does not attempt to pop anymore and that can be an issue if that thread is the last thread to look for work items. Fixed to always check the local queue.

Fixes https://github.com/dotnet/runtime/issues/67545

* avoid allocating collection for intermediate certificates

* fix NRE for custom trust

* feedback from review

We can now have Tier0 locals at byte offsets, so rework how the offset
information is incoded in patchpoints to make this possible.

Closes #68194.

* Switch to 2-phase matching in NonBacktracking

First phase now finds the true match end position.
The implicit .* is now a lazy .*? to prioritize the earliest match.
Third phase is now only run for subcaptures, which no longer needs to
find match end position.
Remove counter optimization that no longer applies with OrderedOr.
Fix a problem in SymbolicRegexInfo where begin/end anchors were
marked as line anchors.
Also remove dead fields from SymbolicRegexInfo.
Fix captures not being handled for empty matches at start of input.

* Improve comments for NonBacktracking

Especially fix comments for the new 2-phase match generation algorithm.

* Add a failing test for the earlier NonBacktracking

* Avoid transitions to deadends for capuring NFA

Need to use `emitActualTypeSize` for the operand size.

Fixes #68170.

* Move MD4 implementation into Common/src/System/Net/Security

* Add minimal RC4 implementation

* WIP: Integrate managed Ntlm implementation into SocketsHttpHandler
Co-authored-by: NTomas Weinfurt <tweinfurt@yahoo.com>

* WIP: Makeshift tests for NTLM/Negotiate authentication

* Fix compilation, clean up some of the hashing

* Avoid using a temporary buffer

* Add computation of signing keys, sealing keys and mechListMIC

* Various cleanups

* Send SPN in target information

* Add some validation, mark spots with missing validation

* Clean up some of the memory offset manipulation

* Move NTLM version into static variable

* Add support for channel bindings, clean up

* Fix hash calculation in makeNtlm2Hash accidentally broken with last commit.
Read NegTokenResp explicitly.
Add mechListMIC reading and verification.

* Verify last authentication token in HTTP Negotiate authentication

* Address feedback

* Fix tvOS builds by making few methods static

* Enable System.Net.Security tests on Android and iOS
Co-authored-by: NTomas Weinfurt <tweinfurt@yahoo.com>
Co-authored-by: NSteve Pfister <steve.pfister@microsoft.com>

This reverts commit 20563211.

* Improving STJ source generator support for record types

Improving the analyzer to not flag a warning about init-only
properties that are actually constructor parameters. In particular,
record types with positional parameters desugar to classes with both
a constructor and init-only properties (eg, to support "with"
expressions). Since these properties are deserialized via the
constructor and not the property setter, they support the fast path and
round trip just fine, so the diagonstic analyzer should not raise the
SYSLIB1037 warning.

Issue 58770

* Improving STJ source generator support for record types

Edit to fix missing semi colon in unit test's CreateCompiliation string

Issue 58770

* Case insensitive matching of props with ctor params

* Additional unit tests

* Test for source gen positional records roundtrip
Co-authored-by: NJoe Harjung <jharjung@integratedlabor.com>

* Fix #65633, multiplication overflow causing incorrect right shift results

* Add unit test reproducing incorrect right shift results for BigInteger (#65633)

* Fix inconsistent formatting

* Expand LargeNegativeBigIntegerShiftTest to validate values, expose private BigInteger members to tests

* Apply suggestions from code review
Co-authored-by: NTanner Gooding <tagoo@outlook.com>

* Restrict unit test to 64-bit processors

* Document reasoning for restricting the test to 64-bit
Co-authored-by: NDan Moseley <danmose@microsoft.com>
Co-authored-by: NTanner Gooding <tagoo@outlook.com>
Co-authored-by: NDan Moseley <danmose@microsoft.com>