* Enable Regex to use SearchValues<string> in compiled / source generator TryFindNextStartingPosition
The analyzer determines a set of prefixes that can start any match, and then uses SearchValues with IndexOfAny to find the next one from that set. It's currently only enabled for case-insensitive; we need to do some more perf validation before enabling for case-sensitive.
* Address PR feedback
* Fix unit test
Add some minimal CI testing for this new CSE heuristic.
Remove testing of cross-block assertion prop, since it's now on by default.
Contributes to #92915.
Fix doc issues raised in https://github.com/dotnet/runtime/issues/98499
DiagnosticListener can send events that don't satisfy the filter. This is non-obvious behavior so the docs should call it out.
There was some funky guidance about naming of payload types that conflicts with our standard namespace naming guidelines so I removed it.
Previously, we would cache the InterfaceImpl that pointed to the interface that provided the DIM. It should be the original type that created the need for the DIM.
Part of #94549. Implements the following encodings:
IF_SVE_FL_3A
IF_SVE_FM_3A
IF_SVE_FN_3A
IF_SVE_FN_3B
IF_SVE_FO_3A
IF_SVE_FP_3A
IF_SVE_FQ_3A
IF_SVE_FS_3A
IF_SVE_FW_3A
IF_SVE_FX_3A
When compiling with `--instruction-set:native` detect the CPUs that implemented AVX-512 in a way where it's better not to use it much.
This replicates the check in codeman.cpp and passes the same flag to JIT that the VM would pass.
I noticed this because passing `--instruction-set:native` to our ASP.NET benchmark machines murders the performance (minus 10+% in RPS) instead of helping it in any way. Passing `--application.environmentVariables DOTNET_PreferredVectorBitWidth=512` kills it in similar way with CoreCLR-JIT, so this detection will likely fix it.
* Add generic support for calli
* Consider classInst as well
* Block runtime marshaling for calli
* Fix the error mode
* Adding tests
* Compare instantiations as well when lookup
* Adding more test cases
* Avoid getting export multiple times
* Improve tests
* Use standard boolean
* Handle collectible assemblies
* Handle inlined P/Invoke transitions
* Reimplement unloadability handling
* Fix native library resolving in tests
* Use AllocMemTracker properly
* Skip tests on windows x86
* Skip negative tests on Windows x86 and NativeAOT
* Use better exception message
* Exclude tests for mono
* Use correct loader elsewhere
* Put VASigCookieBlock on loader module
* Move VASigCookie creation to a new method
* Update issues.targets
* Allocate StubMD on loader module
* Add an error mode for generic varargs
* Apply suggestions from code review
Co-authored-by: Jan Kotas <jkotas@microsoft.com>
* Tweak the tests
* Add link to an GH issue
---------
Co-authored-by: Jan Kotas <jkotas@microsoft.com>
Fixes#98772. Recently, we started checking successor edge likelihoods in Compiler::fgDebugCheckProfileWeights by default; this means we call Compiler::fgDebugCheckOutgoingProfileData in Debug/Checked builds to verify successor edges' likelihoods, and thus call BasicBlock::GetSucc(unsigned, Compiler*) to iterate the successor edges. For switch blocks, GetSucc(unsigned, Compiler*) calls GetSwitchDescMap, and builds m_switchDescMap if it doesn't exist yet. Upon finishing edge likelihood verification, we don't reset m_switchDescMap, so it is possible for this map to be created earlier in Debug/Checked builds versus Release builds.
This doesn't result in behavioral diffs if the map contains the same state between Debug/Release builds by the time it is read. However, if the map is null by the time it is needed, it is created on-demand, ensuring it is completely up-to-date. If the map is not null, the correctness of its current state depends on how judiciously it was maintained with UpdateSwitchTableTarget, creating potential for diffs in the map's state. By invalidating the map instead of updating it as state changes, we can force it to be rebuilt with the latest state when it's needed.
Microsoft.CodeAnalysis.Analyzers , Microsoft.CodeAnalysis.NetAnalyzers
From Version 3.11.0-beta1.24117.1 -> To Version 3.11.0-beta1.24121.1
Co-authored-by: dotnet-maestro[bot] <dotnet-maestro[bot]@users.noreply.github.com>
Microsoft.SourceBuild.Intermediate.emsdk , Microsoft.NET.Runtime.Emscripten.3.1.34.Python.win-x64 , Microsoft.NET.Workload.Emscripten.Current.Manifest-9.0.100.Transport
From Version 9.0.0-preview.2.24120.1 -> To Version 9.0.0-preview.2.24121.1
Co-authored-by: dotnet-maestro[bot] <dotnet-maestro[bot]@users.noreply.github.com>
* Fix VS debugger tests issues
This change fixes three issues I have found in the debugger related code
after I was finally able to run the VS debugger tests with .NET 9.
These are the fixed issues:
* The old EH for hardware exceptions puts incorrectly the exception
location adjusted by -1 to the exception stack trace. The new EH
didn't have this problem, but VS expects that to be the case, so I
have added a compensation for it until the old EH is gone or until VS
can be changed to support both cases.
* Exception interception was not working correctly for cases when the
exception needed to be propagated over native runtime frames. After
catching the rethrown SEH / PAL_SEHException and also in
ProcessCLRException for Windows, the interception needs to be detected
and instead of calling the DispatchManagedException, we need to
call the Ex.RhUnwindAndIntercept again.
* The first chance exception debugger notification was sometimes sent
before even the first stack frame was added to the exception stack
trace. That has broken quite a number of VS tests.
* Fix missing ifdef to fix x86 build
* Fix crossdac build
* Generate check for unsigned divide by zero only when ExceptionSetFlags::DivideByZeroException is present
Not doing so caused JITting on FullOpts fail on assert(add->acdUsed) in genJumpToThrowHlpBlk_la when the DivByZero check was optimized out as a result of #98113
* Change check for GT_DIV or GT_MOD in LSRA to look the same as CodeGen::genCodeForDivMod
Egor BogatovandSingleAccretion
Stephen Toub
Milena Hristova
Kunal Pathak
Andy Ayers
Michal Strehovský
Noah Falk
Jackson Schuster
Aman Khalid
Will Smith
SteveandJan Kotas
Katelyn Gadd
dotnet-maestro[bot]anddotnet-maestro[bot]
Jan Vorlicek
Radek Zikmund
Tomasz Sowiński
Meri Khamoyan