…#103363)

Fixes #103276

Added an automated test that the testcase failed in the above issue. It now passes with this change.

Ideally, Mono should share the same assembly name parsing logic as CoreCLR, which is `AssemblyNameParser.TryParse`.

* Base64Url encoding, validation impelementation

* Validation related updates

* Try fix perf regression in vectorized methods

* Add decoder implementation and unit tests

* Share code in place decoding code

* Add span<char> oveloads with vectorization

* Generalize Span char/byte implementations

* Move ref update to runtime and other cleanup

* Generalize the AdvSimd.Arm64 vectorization added recently

* Apply suggestions from code review

Co-authored-by: Günther Foidl <gue@korporal.at>

* Apply some feedback

* Try fix ARM failure

* Use array pool whenever applicable

* Handle '%' as url padding, add more tests and fix findings

* Fix assertion failure, apply some feedback, try fix ARM failure

* Update docs, small clean ups

* Try fix ARM failure

* Update src/libraries/System.Private.CoreLib/src/System/Buffers/Text/Base64Url/Base64UrlEncoder.cs

Co-authored-by: Günther Foidl <gue@korporal.at>

* Rename StoreToDetionation overloads, reuse some duplicate code

* Improve perf for Base.IsValid() oveerloads, exclude ARM vectorization for char(ushort) overload

* Apply feedbacks

* Revert Assert

* Fix ARM vectorization failure for char overload

* Apply suggestions from code review

Co-authored-by: Jeremy Barton <jbarton@microsoft.com>

* Apply suggestions from code review

Co-authored-by: Günther Foidl <gue@korporal.at>

* Apply more feedback

* Apply suggestions from code review

Co-authored-by: Günther Foidl <gue@korporal.at>

* Apply review comment left overs

* Apply suggestions from code review

Co-authored-by: Miha Zupan <mihazupan.zupan1@gmail.com>

* Apply remaining feedback

* Apply suggestions from code review

Co-authored-by: Jeremy Barton <jbarton@microsoft.com>

* Apply the doc feedback for other API docs

* Fix Base64Url fuzzer findings

* Apply suggestions from code review

Co-authored-by: Jeremy Barton <jbarton@microsoft.com>

* Rename utf8 -> source/destintion

* Apply feedbacks

* Apply feedback

* Apply left out feedbacks

---------

Co-authored-by: Günther Foidl <gue@korporal.at>
Co-authored-by: Jeremy Barton <jbarton@microsoft.com>
Co-authored-by: Miha Zupan <mihazupan.zupan1@gmail.com>

* Add OrderedDictionary

* Open-code collection data structure rather than using Dictionary+List

This rewrites the core of the type to be based on a custom data structure rather than wrapping Dictionary and List. The core structure is based on both Dictionary in corelib and the OrderedDictionary prototype in corefxlab.

This also adds missing TryAdd, Capacity, EnsureCapacity, and TrimExcess members for parity with Dictionary, and fixes debugger views for the Key/ValueCollections.

* Address PR feedback

* Add more tests based on code coverage gaps

* Address PR feedback

* Try to fix NativeAOT tests in CI

…ctor` APIs (#103392)

* Added Sve LoadVectorNonTemporal/NonFaulting/128AndReplicateToVector APIs

* Feedback

Saves up to 5.2% in file size per rt-sz measurements.

Adds a phase before object writing that looks for identical method bodies and deduplicates those that are same.

To keep delegate equivalence working, the compiler also distinguishes between references to symbols and references to symbols that have an observable address identity. When a method is folded that has an observable address identity, the references that require observable address identity go through a unique jump thunk. This means that delegates point to jump thunks and reflection mapping tables point to jump thunks (whenever a method body got folded into a different method body). We do not need the jump thunks for references that are not address exposed (so a `call` in a method body will no go through a jump thunk).

Since method body folding is still observable with stack trace APIs or debuggers, this is opt in. The user gets opted in by setting `StackTraceSupport=false` (or using an undocumented switch).

I took a shortcut in a couple places where references that may or may not be address exposed get treated as address exposed. There are TODO comments around those. We may want to fix tracking within the compiler to tighten this. It may not matter much. I also took a shortcut in deduplication - we currently only look at leaf identical method bodies. The method bodies that become identical after first level of folding currently don't get folded. This leaves a bit size on the table still. There's a TODO comment as well. We also don't consider function pointers address exposed since there's no API to compare these. That's also a TODO for whenever we add such API.

Should help discoverability.

This was enabled in #103273 but needs more work to actually build. Official builds are failing.

* Use ThrowHelper in Queue<T>

* Don't use ThrowHelper in ThrowHelper

* Remove empty line

Not clear why we can't find path to ILC, but this is failing official builds and we need to unblock.

```
error MSB3073: The command ""\tools\\ilc" @"/__w/1/s/artifacts/obj/cdacreader/Release/net9.0/linux-musl-arm/native/libcdacreader.ilc.rsp"" exited with code 127.
```

Co-authored-by: Jan Kotas <jkotas@microsoft.com>
Co-authored-by: Eric StJohn <ericstj@microsoft.com>

…tue (#103408)

* Disable EnableWriteXorExecute by default on riscv64 architectue

On riscv64 architectue any dotnet process that loads corossgened assembly
without DOTNET_EnableWriteXorExecute=0 environment variable set
crashes with "Segmentation fault" error.

For example System.Private.CoreLib.dll is compiled with crossgen2 by default
during build phase so this crash applies to any processes that loads it.

Setting EnableWriteXorExecute=0 disables this options on riscv64 architecure
and prevents these crashes.

Co-authored-by: Dong-Heon Jung <clamp03@gmail.com>

* Add a link to an issue about re-enabling WriteXorExecute on riscv64 arch.

Co-authored-by: Jan Kotas <jkotas@microsoft.com>

---------

Co-authored-by: Dong-Heon Jung <clamp03@gmail.com>
Co-authored-by: Jan Kotas <jkotas@microsoft.com>

…uild 20240613.1 (#103473)

Microsoft.CodeAnalysis.Analyzers , Microsoft.CodeAnalysis.NetAnalyzers
 From Version 3.11.0-beta1.24306.2 -> To Version 3.11.0-beta1.24313.1

Co-authored-by: dotnet-maestro[bot] <dotnet-maestro[bot]@users.noreply.github.com>

* ARM64-SVE: gathervector extends

* Comment out 32bit address APIs

* Replace triple quotes with doubles

* restore commented API

* Set HW_Category_MemoryLoad for all gatherloads

* Fix cast checking for all load types

* Remove GT_CAST changes

* Add predicate registers

* Increase REGNUM_BITS to 7

* Assign float registers if node is mask

* Remove some TODO-SVE present for predicate register implementation

* Make sure to use vector registers if there is no mask

* handle some more printing of predicate registers

* jit format

* try to fix gcc failure

* Revert "try to fix gcc failure"

This reverts commit 5452f6a.

* proper gcc-14 build error fix

Fix #103486

…103308)

* Add File.Append/WriteAllText/Bytes{Async} overloads for span/memory

* Fix faulty assert

* Expose the BigMul APIs publicly

* Ensure that inputs for BigMuil32x32Signed are all naturally typed as int

* Mark StringContent mediaType parameters as nullable

* Update tests

* Fix linux arm64 release build
- Workaround issue where the link register isn't being reported properly in the unwind data

Works around #103489

…esting. (#103483)

…oldBlocks (#103492)

Based on feedback in #102763 (comment), define "cold" blocks based on whether their weights are below a certain threshold, rather than only considering blocks marked with BBF_RUN_RARELY, in fgMoveColdBlocks. I added a BasicBlock method for doing this weight check rather than localizing it to fgMoveColdBlocks, as I plan to use it elsewhere in the layout phase.

* Add a line stating that debugger scenario is not yet supported

* add the message in other 2 files as well

For stdcpp conformance, remove the remaining special extensions.
* clean up unnecessary __llvm  special handling.
* Add class, enum, struct prefixes to cases where the variable name is clashing with type name and it changes meaning  (-Werror=changes-meaning)
  * In some trivial cases, I just renamed the variables..
* On x86, replace `__asm {}` syntax with `__asm ("")`
* Two small illumos related build fixes under `src/native` which helped validating the rest of the changes on the platform.

…n', 'JIT_GetDynamicGCStaticBase_SingleAppDomain' after PR#99183. (#103467)

* Implement `JIT_GetDynamicNonGCStaticBase_SingleAppDomain`, `JIT_GetDynamicGCStaticBase_SingleAppDomain` for LoongArch64 after PR #99183.
* Also fix the build error for LA64.

Change-Id: Ifb9ee599bca3971270a5bc654770dfcc7e931955

* Revert "Do not set UseNativeAotForComponents for arm32 MUSL (#103469)"

This reverts commit f1f0750.

* Revert "Set UseNativeAotForComponents to false on bionic (#103454)"

This reverts commit d901213.

* Revert "Try fixing x86 Windows legs (#103411)"

This reverts commit 6927fea.

* Revert "Consolidate <NativeAotSupported definitions (#103273)"

This reverts commit b86c463.