- 20 1月, 2023 8 次提交
-
-
由 Julien Couvreur 提交于
-
由 Juan Hoyos 提交于
Since https://github.com/dotnet/runtime/commit/66af85735dc40dc7aa257329c91e790b7418b429 it looks like the runtime is not creating the clr <-> dac correlation index in symbol servers. `publishInstallersAndChecksums` is no longer needed, so skip passing that to the in-stage publish.
-
由 Aaron Robinson 提交于
* Fix nested non-byref like VC with no pointer fields
-
由 Steve Pfister 提交于
For whatever reason, the path was wrong and a bunch of tests are being run on CI that shouldn't.
-
由 Radek Doulik 提交于
-
由 Zoltan Varga 提交于
-
由 Pavel Savara 提交于
- return bytes of streaming response as soon as available - fix unhandled error in reader.cancel() promise - return cancelable promise from http_wasm_get_streamed_response_bytes - unit test for slowly streamed chunks - unit test for streaming and default cancellation
-
由 Thefrank 提交于
* add inotify for FreeBSD during linking * remove FreeBSD for now * netstandard does not understand FreeBSD * add FreeBSD platform to HostModel Tests * add inotify for FreeBSD during linking * remove FreeBSD for now * netstandard does not understand FreeBSD * add FreeBSD platform to HostModel Tests * Use Clang12, update LinkerArg * WIP FreeBSD NativeAOT * add inotify for FreeBSD during linking * remove FreeBSD for now * netstandard does not understand FreeBSD * add FreeBSD platform to HostModel Tests * WIP FreeBSD NativeAOT * remove FreeBSD for now * add FreeBSD platform to HostModel Tests * Use Clang12, update LinkerArg * Update JITTools and ObjWriter for FreeBSD-x64 * fixes for tests * fixes for tests * add TargetTriple for FreeBSD * Some changes from feedback, remove out of scope * HostModel back to netstandard2.0 * Update src/coreclr/CMakeLists.txt * Publish crossgen2 as single-file on cross-os * pass NativeAotSupported to MSBuild crossgen2 tasks * Can't use NativeAOT in source build yet Co-authored-by: NAdeel Mujahid <3840695+am11@users.noreply.github.com> * Update private string HostFxr and HostPolicy * Fix private string HostPolicy * Apply suggestions from code review Co-authored-by: NAdeel Mujahid <3840695+am11@users.noreply.github.com> Co-authored-by: NMichal Strehovský <MichalStrehovsky@users.noreply.github.com> Co-authored-by: NJan Kotas <jkotas@microsoft.com> Co-authored-by: NRobin Sue <robinsue@live.de>
-
- 19 1月, 2023 27 次提交
-
-
由 Andy Ayers 提交于
We use the bbNum to decide how to manipulate "in loop" edges when inverting loops, and this was sometimes misleading as the bbNums did not reflect block order. Per the code this is not a correctness issue, but needless redirections can confuse loop recognition. Fixes #80809.
-
由 Vitek Karas 提交于
Removes the `Origin` class as it doesn't seem to serve any purpose anymore. The code used it to basically just only call `GetDisplayName` on members and pass that as the `reason` to dependency analysis. Replaced with passing just string directly as the `reason`. Fixes bugs in warning subsystem when the warning is generated from an origin which is a module or assembly. This would previously crash the compiler. There are basically no tests for this right now - some of the tests in linker which hit this can't be easily ported yet due to NativeAOT not producing "reflection access to DAM" warnings. I have these mostly working locally but enabling the warnings fails some tests outside of the linker ones, so it needs more work. Fixes analysis bugs when methods or fields are accessed via `ldtoken` or `ldftn` instructions. In such cases we still need to process them for RUC and DAM based warnings and so on. There are only a few tests for this in this change - more will come when we port the `AnnotatedMembersAccessedViaReflection` from linker, but those all require to enable the "reflection access to DAM" warnings as mentioned above. * This required addition of a new TrimAnalysisPattern since the processing happens in reflection scanning and all warnings/marking hapening from there must go through the patterns. Changes `DynamicDependencyAttribute` processing to happen in a separate dependency node. Previously if there were warnings generated due to the `DynamicDependencyAttribute` the compiler may report them multiple times (if it hit the same member twice during the rest of the analysis). To avoid this, the processing is postponed by adding a new dependency node - which then acts as the unification mechanism and thus warnings are produced just once.
-
由 Anton Firszov 提交于
-
由 Eric Erhardt 提交于
* LambdaExpression.CanCompileToIL should respect IsDynamicCodeSupported With the new feature switch added in https://github.com/dotnet/runtime/issues/39806, calling LambdaExpression.Compile is throwing a PlatformNotSupportedException. Instead, LambdaExpression should respect IsDynamicCodeSupported and switch to using the interpreter when IsDynamicCodeSupported is false. * Add tests
-
由 Jan Vorlicek 提交于
Recent change in the `ThreadLocalModule::AllocateDynamicClass` has accidentally removed zeroing for new CollectibleDynamicEntry instances. That leads to crashes when running code in an unloadable context. This change fixes that.
-
由 Egor Bogatov 提交于
-
-
由 Vlad Brezae 提交于
We were storing uninitialized values. Bug caught with valgrind, hopefully fixes some real crashes.
-
由 Michal Strehovský 提交于
Grabbing custom attribute data brings the entire reflection stack (we need properties, constructors, fields, methods). We ask `IsByRefLike` in type loader constraint validation. Ideally we should be able to get rid of the type loader from a hello world, but that's for later. This is a more efficient implementation either way.
-
由 Michal Strehovský 提交于
Contributes to #80165. Not aware of a good reason to keep using it. Removes `TypeDelegator` from apps that don't use it. Saves 7 kB (but also stepping stone for the larger savings later).
-
由 Jeremy Koritzinsky 提交于
Remove the TypeKey concept as the primary user for the concept is unable to use it effectively. (#79418)
-
由 Michal Strehovský 提交于
Contributes to #80165. Dispensing of reflection member infos is done through a member policies class. This is a generic class that has specialized implementations for each kind of member info. It used a clever trick to get to the specific implementations. Just access `MemberPolicies<EventInfo>.Default` to get one for events or `MemberPolicies<PropertyInfo>.Default` to get one for properties. It was also absolutely not trimming friendly. This change removes the clever trick and replaces it with good old fashioned parameter passing.
-
由 LateApexEarlySpeed 提交于
Squashed commit of the following: commit e89bcfe2d8c3f45232f3d0b861be34ccede948f1 Author: lateapexearlyspeed <lateapexearlyspeed@163.com> Date: Wed Nov 9 11:56:41 2022 +0800 Change comment behavior: add newline between comments. commit 60d730ae151c9c2bffdaf7b6ca3cfa68180ba8b8 Author: lateapexearlyspeed <lateapexearlyspeed@163.com> Date: Thu Oct 27 15:36:43 2022 +0800 Fix comment: test multiple comments in same line. commit 977e76a298e52ad184adf78ab7172be99b97b9e2 Merge: 64b6be97fd4 9b9b0c95 Author: LateApexEarlySpeed <72254037+lateapexearlyspeed@users.noreply.github.com> Date: Fri Sep 23 12:56:01 2022 +0800 Merge branch 'main' into lateapexearlyspeed-MakeWriteCommentValueOverloadSameBehavior commit 64b6be97fd4c9b59fc601db8a0e6ddd8a8b97a7f Author: lateapexearlyspeed <lateapexearlyspeed@163.com> Date: Thu Sep 22 10:39:32 2022 +0800 Fix comment: keep newtonsoft.json asserting tests; add non-indented tests; use pattern match. commit e3a772ac985801eb89abc028f5a71f2ad502cd55 Author: lateapexearlyspeed <lateapexearlyspeed@163.com> Date: Tue Sep 20 16:20:59 2022 +0800 Inline expectedIndentedJson to local var. commit 9e16463cc8826890e6fbeb115ba9a05867ee5017 Author: lateapexearlyspeed <lateapexearlyspeed@163.com> Date: Tue Sep 20 11:46:44 2022 +0800 Fix comment: update test cases to use string literal to assert. commit bebb09c0c9bc7b7b1c5076ab6dbcf1f5ec1d884a Author: lateapexearlyspeed <lateapexearlyspeed@163.com> Date: Thu Sep 15 17:56:44 2022 +0800 Add tests. commit 2f9dbff3629f0b9b8b24f2bb2c32ec9c620a6fcd Author: lateapexearlyspeed <lateapexearlyspeed@163.com> Date: Wed Sep 7 11:01:11 2022 +0800 Add NewLine between comment and start of (object and array). commit 9c1bb280f59e3755227747251a1a51fdca9eeab1 Author: lateapexearlyspeed <lateapexearlyspeed@163.com> Date: Mon Sep 5 15:11:57 2022 +0800 Make WriteCommentValue() overloads have same behavior.
-
由 Bruce Forstall 提交于
-
由 Andy Ayers 提交于
Loop canonicalization now maintains pred edges. GC poll insertion was already maintaining the edges but was rebuilding them anyways. Now pred lists are never rebuilt. Also revise `fgUpdateChangedFlowGraph` so that it no longer has the ability to remove or rebuild. Fixes #49030. Also fixes #80772.
-
由 Thays Grazia 提交于
* press alt-shift-d and open tab to debug blazor app on firefox. * Update src/mono/wasm/debugger/BrowserDebugProxy/Firefox/FirefoxDebuggerProxy.cs Co-authored-by: NAnkit Jain <radical@gmail.com> * addressing @radical comments Co-authored-by: NAnkit Jain <radical@gmail.com>
-
由 Jeremy Koritzinsky 提交于
Switch to using the .NET SDK properties that are defined to provide the TFM instead of using the assembly that defines System.Object (#80646) Fixes https://github.com/dotnet/runtime/issues/80621
-
由 dotnet-maestro[bot] 提交于
* Update dependencies from https://github.com/dotnet/roslyn build 20230113.14 Microsoft.CodeAnalysis , Microsoft.CodeAnalysis.CSharp , Microsoft.Net.Compilers.Toolset From Version 4.6.0-1.23058.1 -> To Version 4.6.0-1.23063.14 * Update dependencies from https://github.com/dotnet/roslyn build 20230116.1 Microsoft.CodeAnalysis , Microsoft.CodeAnalysis.CSharp , Microsoft.Net.Compilers.Toolset From Version 4.6.0-1.23058.1 -> To Version 4.6.0-1.23066.1 * Update dependencies from https://github.com/dotnet/roslyn build 20230117.16 Microsoft.CodeAnalysis , Microsoft.CodeAnalysis.CSharp , Microsoft.Net.Compilers.Toolset From Version 4.6.0-1.23058.1 -> To Version 4.6.0-1.23067.16 Co-authored-by: Ndotnet-maestro[bot] <dotnet-maestro[bot]@users.noreply.github.com>
-
由 Aleksey Kliger (λgeek) 提交于
Fixes #47462 **CoreCLR** This also makes some changes to CoreCLR to decouple EventPipe and startup hooks. Presently if startup hooks are disabled, `RuntimeEventSource.Initialize` is never called. The PR makes the two features independent by moving runtime event source initialization out of the startup hook feature check. * Implement startup hooks support in Mono * Keep StartupHookProvider.ProcessStartupHooks under feature flag * Don't catch/cleanup the exceptions from startup hooks. * Add an ios simulator startup hook functional test * Implement Android functional test * Add WASM functional test * Make a single managed startup method for CoreCLR A common configuration for coreclr is event source enabled, startup hooks disabled, so at least one managed call is inevitable. Since we have to call into managed no matter what, let the trimmer determine what happens once we get there. This is different from mono where published trimmed apps may have both startup hooks and event source disabled. In that case we would rather avoid a managed call to an empty method early in startup. * fix build and line damage
-
由 Jakob Botsch Nielsen 提交于
TreeLifeUpdater at most needs to flip 4 bits in two liveness sets when it is called, however before this change it uses several full width bitset operations to do this. This changes TreeLifeUpdater to do its job in a much more direct way by updating the liveness sets directly.
-
由 Jakob Botsch Nielsen 提交于
jit-analyze no longer accepts comma-separated metrics after https://github.com/dotnet/jitutils/pull/362, instead requiring each metric to be specified separately on the command line.
-
由 Miha Zupan 提交于
-
由 Miha Zupan 提交于
* Improve IndexOf codegen for non-char types * Call directly into NonPackedIndexOf where it makes sense
-
由 Tomas Weinfurt 提交于
* add unsupported attributes to System.Net.Security * using * CA1416
-
由 Jan Kotas 提交于
-
https://github.com/dotnet/linker由 dotnet-maestro[bot] 提交于
Microsoft.NET.ILLink.Tasks From Version 8.0.100-1.23064.1 -> To Version 8.0.100-1.23067.1 Co-authored-by: Ndotnet-maestro[bot] <dotnet-maestro[bot]@users.noreply.github.com>
-
https://github.com/dotnet/arcade由 dotnet-maestro[bot] 提交于
Microsoft.DotNet.Arcade.Sdk , Microsoft.DotNet.Build.Tasks.Archives , Microsoft.DotNet.Build.Tasks.Feed , Microsoft.DotNet.Build.Tasks.Installers , Microsoft.DotNet.Build.Tasks.Packaging , Microsoft.DotNet.Build.Tasks.TargetFramework , Microsoft.DotNet.Build.Tasks.Templating , Microsoft.DotNet.Build.Tasks.Workloads , Microsoft.DotNet.CodeAnalysis , Microsoft.DotNet.GenAPI , Microsoft.DotNet.GenFacades , Microsoft.DotNet.Helix.Sdk , Microsoft.DotNet.PackageTesting , Microsoft.DotNet.RemoteExecutor , Microsoft.DotNet.SharedFramework.Sdk , Microsoft.DotNet.VersionTools.Tasks , Microsoft.DotNet.XUnitConsoleRunner , Microsoft.DotNet.XUnitExtensions From Version 8.0.0-beta.23066.1 -> To Version 8.0.0-beta.23067.5 Co-authored-by: Ndotnet-maestro[bot] <dotnet-maestro[bot]@users.noreply.github.com>
-
- 18 1月, 2023 5 次提交
-
-
由 Elinor Fung 提交于
-
由 Elinor Fung 提交于
-
由 Vlad Brezae 提交于
* [mono][interp] Remove MINT_VT_ALIGNMENT It serves no purpose nowadays. All vars are aligned to MINT_STACK_SLOT_SIZE. * [mono][interp] Add 16 byte default alignment to optimized code This will enable us to control alignment of Vector128 vars at compile time. * [mono][interp] Add stack alignment for unoptimized code For normal calls, we introduce a new opcode before the call that will move all the arguments to aligned stack location. When emitting the code for the call, we emit directly the aligned call args offset instead. Unoptimized code has its own opcodes that do moving of param, we tweak them to copy them into aligned location.
-
由 Zoltan Varga 提交于
-
由 Stephen Toub 提交于
-