- 08 7月, 2022 40 次提交
-
-
由 Tom Deseyn 提交于
* Tar: for directories, use a default mask that has the 'x' bit. Without the 'x'-bit, the directories that get created are not accessible. * Apply suggestions from code review Co-authored-by: NStephen Toub <stoub@microsoft.com> * Update src/libraries/System.Formats.Tar/tests/TarTestsBase.cs Co-authored-by: NStephen Toub <stoub@microsoft.com> Co-authored-by: NStephen Toub <stoub@microsoft.com>
-
由 Daniel Svensson 提交于
* Dont create HashSet unless needed in IsTypeSerializable * Apply suggestions from code review
-
由 Kunal Pathak 提交于
-
由 Tom Deseyn 提交于
* Tar, Zip: respect umask when creating files. * Update UnixExtractFilePermissionsCompat * Replace const int with const UnixFileMode. * Also run tests with zero umask
-
由 Stephen Toub 提交于
-
由 Stephen Toub 提交于
* Use file-scoped types in RegexGenerator output * Update System.Text.RegularExpressions.Tests.csproj
-
由 SingleAccretion 提交于
-
由 SingleAccretion 提交于
* Delete the unused GTF_ICON_FIELD_OFF * Do not try to find "IND(struct ARR_ADDR)" handle Leftover from GT_INDEX. * Delete the zero-offset field sequence map For statics at a zero offset, preserve the "CNS_INT 0" node instead. * Make all field sequences singletons * Print struct field offsets in the sequence * Improve address mode marking to handle COMMAs This avoids regressing cases where we now have new CSE opportunities for statics with static init helper calls. * Delete NotAField Replace with using "nullptr" directly. Note that "nullptr" can mean both an absense of a field sequence and an "implicit" sequence of struct fields. * Refactor the field sequence store Now that we don't need the map to be a set, we can apply some optimizations. * FieldSeqNode -> FieldSeq * Turn asserts into "noway_assert"s We don't expect empty sequences: they are either coming from the "IsFieldAddr" code path, which disallows these, or from the boxed static code path, where we should always have them. * Fix formatting * Take down SPMI * Resurrect SPMI to work around a CG2 bug
-
由 Ilona Tomkowicz 提交于
[wasm][debugger] Enable onCallFrame evaluation of static fields in classes/interfaces from a static method. (#70560) * Test for DIM - expected behavior. * Static tests: renaming variables and small refactoring. * Added new testcases and merged existing tests into Theories. * Fix for resolving the name with namespace prefix matching the current frame location. * Fix for tests failing on InvokeMethod in runtime. * Applied smallest suggestions of @radical's review. * Reverted deleting and renaming tests - only adding new. * Corrected the logic + added new tests. * Removed unnecessary logging.
-
由 Jakob Botsch Nielsen 提交于
Convert them to GenTreeVisitor.
-
由 Jakob Botsch Nielsen 提交于
We may see this in stress modes, and in general the JIT should be able to handle the flag.
-
由 Michal Strehovský 提交于
System.Linq.Parallel has multiple generic cycles in it. We probably hit them all in the System.Linq.Parallel.Tests assembly. We would be compiling the test forever (or until running out of memory) because we didn't detect one of them. Add detection for the case where the recursion is entered through a generic interface virtual method call. The compiler terminates the infinite expansion after the user-specifiable cutoff point. The test shows the pattern and the expected behavior.
-
由 Michal Strehovský 提交于
There was an optimization that was trying to prevent the compiler from doing unnecessary work when analyzing generic virtual methods. The optimization is invalid.
-
由 Elinor Fung 提交于
- Find shape data for stateful collection marshallers - Add marshalling strategy implementations for stateful collection marshalling of blittable elements and non-blittable elements - Refactor non-blittable element handling into a base class for shared functionality between stateless and stateful - Probably can share more for blittable element handling too, but I'll plan to do that in a follow-up after getting general support in. - Tests for stateful marshallers for `List<T>` and arrays
-
由 Olli Saarikivi 提交于
* Concurrency fixes and refactoring for clarity Removed builder reference from SymbolicRegexNode instances; builder now has to be passed in. Since the builder is not thread safe this clarifies the locking required in the matcher when using it. Moved matching specific state from the builder to the matcher. This includes state and transition arrays. Simplify character kind code by eliminating duplication of logic. * Changes from review and cleanup DfaMatchingState is now just MatchingState * Comment on NFA mode IDs
-
由 Parker Bibus 提交于
-
由 Bruce Forstall 提交于
Fixes #61482
-
由 Jeremy Koritzinsky 提交于
-
由 Eirik Tsarpalis 提交于
* Remove implicit fallback to reflection-based serialization. Fix #71714 Include JsonSerializerContext in JsonSerializerOptions copy constructor. Fix #71716 Move reflection-based converter resolution out of JsonSerializerOptions. Fix #68878 * Address feedback & add one more test * Update src/libraries/System.Text.Json/src/System/Text/Json/Serialization/JsonSerializerContext.cs Co-authored-by: NStephen Toub <stoub@microsoft.com> * fix build * Bring back throwing behavior in JsonSerializerContext and add tests * Only create caching contexts if a resolver is populated * Add null test for JsonSerializerContext interface implementation. * skip RemoteExecutor test in netfx targets * Add DefaultJsonTypeInfoResolver test for types with JsonConverterAttribute * remove nullability annotation * Update src/libraries/System.Text.Json/src/System/Text/Json/Serialization/JsonSerializerOptions.Converters.cs Co-authored-by: NKrzysztof Wicher <mordotymoja@gmail.com> Co-authored-by: NStephen Toub <stoub@microsoft.com> Co-authored-by: NKrzysztof Wicher <mordotymoja@gmail.com>
-
由 Jakob Botsch Nielsen 提交于
This adds some richer debug information in the form of inline trees and IL<->native mappings with inlinee information, plus an ETW event in the private runtime provider to report it back. A COMPlus variable is used to opt-in to generating and storing this larger debug information. On the JIT side, we now store a few more fields of data for each inline context in release builds: * Its ordinal which is required to be incrementally assigned * The actual IL offset of the IL instruction leading to the creation of the inline context * The method handle of the inlinee On the EE side we store the new debug information together with the normal debug info, reusing the flag byte used for patchpoint information. The hope is that the new richer format of debug info can eventually replace the old one, but this won't happen in .NET 7 as I expect it will take some time to come up with the right format for this data and the right set of fields to expose. In the meantime this environment variable and event/debugging APIs are mainly to be used as an opt-in way to start incrementally prototyping on the tooling side without needing special checked builds of the runtime/JIT.
-
由 Theodore Tsirpanis 提交于
-
由 Kevin Jones 提交于
* Obsolete Create(string) cryptographic factory methods * Add to list of diagnostics
-
由 tfenise 提交于
-
由 Noah Falk 提交于
* Fix EventSource deadlock Initializing NativeRuntimeEventSource inside the EventListener static constructor could lead to a deadlock where a 2nd thread holding the ETW lock calls back though ETWCallback and tries to get EventListener lock which blocks on the static constructor. Fixes #70870
-
由 Anton Firszov 提交于
* ReturnsSocketFlags test * fix * ReceiveTruncated_TruncatedFlagIsSetOnReceive
-
由 Kevin Jones 提交于
* Fix some null reference exceptions and improve nullable annotations. * Remove new line that was introduced
-
由 Jakob Botsch Nielsen 提交于
While experimenting with doing less spilling in #70932 I have realized that the change I made in #68736 was not correct. There I commented that "we now only check interference with the stack entries that are not arguments since we now know that arguments will be normalized and kept in order" -- this is not correct since the normalization may spill args to new statements. It is possible that a later argument that we are normalizing interferes with an earlier argument that was not normalized. In that case the logic could reorder these two arguments by spilling only the later argument to a new statement. This changes the logic back to be more similar to what it was before -- we iterate the arguments in reverse order, popping them from the top as we go along with normalization now checking interference with previous args as well. Since we cannot walk args backwards we also need an extra loop (as before) to walk the signature information in order.
-
由 SwapnilGaikwad 提交于
* Implement GetIndexOfFirstNonAsciiChar intrinsic on AArch64 * Fix typo in a method name
-
由 Tanner Gooding 提交于
* Ensure that CreateChecked, CreateSaturating, and CreateTruncating are publicly exposed * Fixing the constraint on the private TryConvertFrom helpers
-
由 David Mason 提交于
-
由 SingleAccretion 提交于
Namely, cases where we cannot transform local indirections into local nodes: 1. Unused indirections - we simply remove them. 2. Volatile FIELDs - we mark the underlying local address-exposed. 3. Large offsets - likewise. Overall this (as well as most other ADDR-related changes) works towards enabling the invariant that after local morph, every local access is either address-exposed, or transformed into LCL_VAR/LCL_FLD.
-
由 Tanner Gooding 提交于
* Updating a few BitConverter APIs to be intrinsic * Reacting to PR feedback * Update src/coreclr/jit/valuenum.cpp Co-authored-by: NSingleAccretion <62474226+SingleAccretion@users.noreply.github.com> * Update src/coreclr/jit/gentree.h Co-authored-by: NSingleAccretion <62474226+SingleAccretion@users.noreply.github.com> * Apply formatting patch * Ensure DoubleToInt64Bits and Int64BitsToDouble intrinsics only pop when appropriate * Fix VNForBitCast to not handle exception set logic since its always a normal value now * Applying formatting patch Co-authored-by: NSingleAccretion <62474226+SingleAccretion@users.noreply.github.com>
-
由 SingleAccretion 提交于
Preparatory changes for enabling direct decomposition of promoted locals into `FIELD_LIST`s on ARM (#71726) * Unify the "3-byte struct" ARM case with other targets The only case when "structSize != passingSize" could be true is when we "upcast" a "mis-sized" struct to a primitive. For ARM, we never did that -- the relevant "#ifdef"s didn't include it, thus "OperIs(GT_OBJ) && (structSize != passingSize)" was dead code. Instead, there was bespoke handling for such structs: "if (structSize < TARGET_POINTER_SIZE)". Delete the bespoke handling and the "#ifdef", unifying the handling. * Update a comment * Fix PUTARG_SPLIT codegen We must use the "field type" for stores. Example "struct { int a; byte b; byte c; }" is split between one register and one stack slot, and both "b" and "c" become propagated constants, typed INT. * Enable OBJ spilling for ARM * Fold the last ARM-specific condition * Do not convert large split args into FIELD_LISTs We do not check the register requirements on the fields or their sizes (extents). * Restructure support for "mis-sized" structs Add a comment, simplify the code.
-
由 David Cantú 提交于
* Make full assembly unsupported on browser * Add new MultiSign APIs and address API review feedback on existing ones * Add SignaturePadding tests and address #70189 * Address feedback (typos and leftover code) * Add tests for CoseHeaderValue and wrap CborReader/Writer errors * nits and line chopping * Move alg. validations to CoseSigner and add remaining AddSignatureFor* APIs
-
由 Carlos Sanchez 提交于
-
由 Carlos Sanchez 提交于
-
由 Eric Erhardt 提交于
* Better error handling in SubtleCrypto workers Handle exceptions from SubtleCrypto by catching and logging exceptions coming from the crypto stack. Reset web worker when a request fails. Also, fix race conditions where the web worker can read its own response as part of the next request. Contributes to #69740
-
由 CyrusNajmabadi 提交于
Polyfill the incremental generator ForAttributeWithMetadataName from roslyn (for LoggingGenerator). (#71651) * Polyfill the incremental generator ForAttributeWithMetadataName from roslyn (for LoggingGenerator). * Simplify * Fix project file * Remove * revert
-
由 Koundinya Veluri 提交于
* Fix long latencies in `ReaderWriterLockSlim` lock operations due to `Sleep(1)` - Removed usage of `Sleep(1)` - Modified the spin-waiting scheme beyond the `Sleep(0)` threshold to avoid potentially long sequences of ineffective `Sleep(0)`s - Retuned the threshold to reprioritize deprioritized spin-lock requests Fixes https://github.com/dotnet/runtime/issues/69206
-
由 Jan Kotas 提交于
Fixes #71733
-