This reverts commit a0f7927c.

It broke wasm builds: https://github.com/dotnet/runtime/pull/67381

* Concatenate cookies with semicolon

* Restore tests that run on .NET Framework

* Change Cookie header to Custom

* PR feedback

* Call `EnsureLoadableMethod` from scanner, same as we do in JitInterface.
* Make scan/compile diffing less sensitive to differences caused by invalid IL.

* Fix handling of makeref/refvalue in generic code

Found in Pri0 tests. We didn't have any coverage for TypedReference in smoke tests.

* Update Generics.cs

* Avoid pre-allocating large arrays in some scenarios for Enumerable.Chunk().

Fix #67132

* Switch Chunk() to use List-based implementation.

See discussion here:
https://github.com/dotnet/runtime/pull/67210#discussion_r836561697

* Address feedback from https://github.com/dotnet/runtime/pull/67210/files

Co-authored-by: NJan Kotas <jkotas@microsoft.com>
Co-authored-by: NJakob Botsch Nielsen <Jakob.botsch.nielsen@gmail.com>

Before this change the execution script generator arbitrarily
referred to "InputAssemblyName", "AssemblyName" and
"MSBuildProjectName" w.r.t. names of the scripts and the
executables. This is unfortunate because for merged
ILPROJ projects we need to adjust the assembly name during build.
This simple change makes the scripts consistently refer to
AssemblyName everywhere.

Thanks

Tomas

- Fix Null ref on non-existing reference. This is in line with how
  the old code worked, it returned false on CheckSignature but didn't
  throw on load.

Replace Marshal.PtrToStructure and similar methods with unmanaged pointers. Fixes a few AOT compatibility warnings.

* Update dependencies from https://github.com/dotnet/linker build 20220330.1

Microsoft.NET.ILLink.Tasks
 From Version 7.0.100-1.22178.1 -> To Version 7.0.100-1.22180.1

* Update dependencies from https://github.com/dotnet/linker build 20220331.1

Microsoft.NET.ILLink.Tasks
 From Version 7.0.100-1.22178.1 -> To Version 7.0.100-1.22181.1

* Update dependencies from https://github.com/dotnet/linker build 20220401.3

Microsoft.NET.ILLink.Tasks
 From Version 7.0.100-1.22178.1 -> To Version 7.0.100-1.22201.3
Co-authored-by: Ndotnet-maestro[bot] <dotnet-maestro[bot]@users.noreply.github.com>

Loop cloning needs to parse what morph creates from GT_INDEX nodes
to determine if there are array accesses with bounds checks that
could potentially be optimized. For jagged array access, this can
be a "comma chain" of bounds checks and array element address expressions.
For a case where an array of structs had a struct field, such as
`ValueTuple<int[], int>[]`, cloning was confusing the expression
`a[i].Item1[j]` for the jagged array access `a[i][j]`.

The fix here is to keep track of the type of the `GT_INDEX` node that is
being morphed, in the `GT_BOUNDS_CHECK` node that is created for it.
(This is the only thing cloning parses, to avoid the need to parse
the very complex trees morph can create.) This type is then checked
when parsing the "comma chain" trees. If a non-`TYP_REF` is found (such
as a `TYP_STRUCT` in the above example), no more levels of array indexing
are considered. (`TYP_REF` is what an array object would have, for a jagged
array.)

Fixes #66254.

We can sometimes see trees like "ARR_ADDR(long 0)" that are the
result of morph's folding logic.

In general it does not seem great to allow folding of BYREFs
into LONGs like that, but it is also not incorrect per-se.

* Add {ReadOnly}Span ctor (internal for now) that takes an in and no length

* Address PR feedback

* Fix native references to _pointer

Co-authored-by: NStephen Halter <halter73@gmail.com>

* add quotes around dotnet path

* address feedback

Fixes #65612. This regression was exposed in #62891 - we had a mismatch between what methods are considered reflectable during scanning vs during optimized compilation due to extra array being injected for reflectable `params` methods on instantiated generic types.

There are two potential fixes - either expand the set of methods considered reflectable during scanning, or limit the number of methods considered reflectable in optimized compilation. It doesn't look like we need the expanded set, so going with a restriction instead. The CI may prove me otherwise, so keeping an open mind about this.

* Move copying of merged wrapper native components to copynative step

* Move export of test exclusion file before ILLink changing CORE_ROOT

* Fix file mode for test execution bash scripts

* Add out-of-proc test folders to Helix work item payloads

* Fix Helix publishing in the presence of out-of-process tests

* Use a marker file triggered off of a project property to determine which files to skip running the Mono AOT compiler on (some tests fail at AOT time, so we need this to work around that limitation)

* Pass runtimeVariant to the Core_Root construction for the test exclusion list creation

* Don't write the output element when there's no test output. This causes issues with Helix test uploading.

* Write xunit results the way xharness expects to read them for wasm so we correctly report failures and don't report a test harness failure.

* Fix GeneratedTestRunner to build.

* Fix copying native wrappers for merged test runners.

Pre-emptively mark another test as RequiresProcessIsolation as I stumbled upon it while fixing this

* Fix copying of native test components to merged wrapper outputs

Due to Pri0 / Pri1 test grouping we may end up building the
wrapper in a different group than its components. Make sure that
we populate all native components of merged wrapper dependencies
before we copy them over to the merged wrapper output folder.

* Exclude wasm support files.
Co-authored-by: NJeremy Koritzinsky <jekoritz@microsoft.com>

Removed the hard heap limit heap balancing logic for regions + UOH and fixed infinite retries in case at the LOH boundary (#67406)

* Removed the hard heap limit heap balancing logic for uoh and fixed inifite retries

* Addressed feedback - refactored region checking directives

* Removed all instances of hard_limit_exceeded_p as it is no longer needed

Tracking: https://github.com/dotnet/runtime/issues/66921

Tracking: https://github.com/dotnet/runtime/issues/58699

Tracking: https://github.com/dotnet/runtime/issues/66279

Remove `ObsoleteAttribute` from COM interop values.

* Detect systemd when inside a container.

Modify IsSystemdService to function correctly
for containerised services.

Fix #67203

Adds conditional compilation to fall back
to Process.GetCurrentProcess().Id when
Environment.ProcessId is not available.

* Apply suggestions from code review

The minimum supported .NETCoreApp version in the repository is 6.0, so just using NETCOREAPP.

Co-authored-by: james <judilsteve>
Co-authored-by: NEric Erhardt <eric.erhardt@microsoft.com>

* Dissolve structs into primitives for PUTARG_STK

The primary benefit of this change is "free" support for
complex addressing modes, which is always desirable when
we are loading a primitive (as opposed to the large struct
case, where we would not want to use the 3-operand LEA
in a loop, but instead cache the address in a register).

The additional (future) benefit is that we will no longer
need to mark the source local as DNER, once LCL_VAR sources
for struct PUTARG_STKs are supported.

* Contain PUTARG_STK sources for "push [mem]"

Today logic for emitting the legacy XUnit wrappers is based on
scanning the test output tree for execution scripts (sh / cmd);
we however need to emit execution scripts even in the presence of
merged wrappers in case of tests requiring project isolation.
This change filters these tests out so that they don't end up
creating a spurious legacy XUnit test wrapper.

Thanks

Tomas

Based on (non-exhaustive) testing, chain builds from a Let's Encrypt issued certificate have
the following characteristics:

* Live OCSP request required (uncached/unstapled): 577ms
* OCSP response retrieved from cache (unstapled): 183ms
* OCSP response utilized from TLS stapling (bypasses cache): 182ms

In both cached and stapled the revocation portion was about 39ms.
(The revocation mode was ExcludeRoot, the CRL pertaining to the intermediate was cached for all three measurements.)

If the OCSP response was stapled (and the math worked out OK on it) then we completely ignore the OCSP cache.
While it could potentially be useful to update the cache if the stapled response was newer, the extra I/O of doing
the "newer" test didn't feel justified at this time.

This should improve the experience when working with complex SAML tokens.