- 12 10月, 2022 10 次提交
-
-
由 Carlos Sanchez 提交于
Co-authored-by: Ncarlossanlop <carlossanlop@users.noreply.github.com>
-
由 Egor Bogatov 提交于
Co-authored-by: NJan Kotas <jkotas@microsoft.com>
-
由 Jeremy Koritzinsky 提交于
-
由 Drew Scoggins 提交于
* Update Arm64 cross build container * Reomve suffix
-
由 Egor Bogatov 提交于
-
由 Jan Kotas 提交于
- Use proper macOS capitalization - Delete superfluous details Co-authored-by: NTheodore Tsirpanis <teo@tsirpanis.gr>
-
由 Filip Navara 提交于
-
由 Mark Plesko 提交于
Inspired by comments ([here](https://github.com/dotnet/runtime/pull/73424#discussion_r939068635) and [here]( https://github.com/dotnet/runtime/pull/73424#discussion_r939090204)) - Remove `RefPosition* currentRefPosition = &reverseIterator` as it creates two variables that are essentially the same, especially since the iterator has operators like `->`. In fact, I had to spend a bit of time verifying that there weren't any updates to one but not the other in the code. (And actually #73424 did slightly impact this relationship since it moved the assignment from the loop iteration step to the loop body, but I don't believe that this mattered.) I renamed the "iterator" variable to the "position" name to reduce textual churn in the code. This didn't work as well for the range-based loops since they yield -references- to the underlying object so a bunch of C++ punctuation changes. - Break apart two complicated `for` conditions and remove duplication between the condition and the loop body. Search for `continueLoop` to see them. - Delete list's `operator&` on each iterator since they are no longer used and not part of normal iterators. Manually verified no diffs (other than memory addresses) on a jitdump for a test case with minopts on and off to exercise the dump code in lsra. Manually verified asm diffs were spurious (differences in runtime addresses, sometimes also causing changes in the compiler internal representation size and then locations of labels). This regains the unexplained small throughput loss in #73424.
-
由 Heath Baron-Morgan 提交于
Co-authored-by: NBuyaa Namnan <bunamnan@microsoft.com>
-
由 Mitchell Hwang 提交于
This test has flakily caused app crashes on tvOS arm64 lanes. dotnet/arcade#11167 When the app crashes, it was observed that of the last tests ran, there commonly was only 2 InvalidIVSizes see https://gist.github.com/mdh1418/563ce4066e16dfee055e0903e2c70a1e https://gist.github.com/mdh1418/ed11c7ba361c3fdb8906e57034c58f90 https://gist.github.com/mdh1418/c831b807dc0d949bc01cdeee61a1d795 Whereas on a successful test suite run, there are 3. https://gist.github.com/mdh1418/912914871d580475751d460719624224 Its suspected that the last test case is problematic, so disabling to make CI cleaner
-
- 11 10月, 2022 30 次提交
-
-
由 Matous Kozak 提交于
* Abs intrinsics * rename OP_SSSE3_ABS
-
由 Thays Grazia 提交于
-
由 Thays Grazia 提交于
* The description of an object that has ToString implemented should be the result of the ToString and not the type name. * Fix if cannot find ToString method. * removing my name of the code * Implementing support for inherited classes with ToString overriden. Implement support for value types with ToString overriden. Fix Debugger* is more relevant than ToString * Fix test cases. * Apply suggestions from code review Co-authored-by: NAnkit Jain <radical@gmail.com> * Adding more tests and addressing @radical comments * Adding tests for record * Moving implementation of InvokeToStringAsync to sdbhelper and reuse it on objects and valuetypes. * Apply suggestions from code review Co-authored-by: NAnkit Jain <radical@gmail.com> * Addressing @radical comments. Co-authored-by: NAnkit Jain <radical@gmail.com>
-
-
由 Eirik Tsarpalis 提交于
* Ensure reflection property metadata is instantiated lazily. * Address feedback
-
由 Eirik Tsarpalis 提交于
Co-authored-by: stoub@microsoft.com
-
由 Vlad Brezae 提交于
* [mono][interp] Remove redundant opcode Also allowing general optimizations that are being applied to ldfld opcodes. * [mono][interp] Remove ldloca when used with ldfld/stfld Generate direct moves instead to the valuetype interior offset * [mono][interp] Replace ldloca + ldind/stdind with simple var moves
-
由 Shay Rojansky 提交于
Closes #76469 (cherry picked from commit 2070def2)
-
由 Bruce Forstall 提交于
* Fix loop canonicalization for call-finally case If a call-finally is followed by a loop body that requires canonicalization, we hit an assert that the `head` block preceding the loop must be a fall-through block. This was true in all cases except if the preceding block was the BBJ_ALWAYS of a BBJ_CALLFINALLY/BBJ_ALWAYS pair. To fix this, add a new canonicalization. If this case occurs, simply insert a new fall-through block above the loop top and redirect the BBJ_ALWAYS to that new block. Fixes #76346 * Add unit test
-
由 Stephen Toub 提交于
* Improve performance of integer formatting 1. The DivRem(..., 10) for each digit in the number ends up being the most expensive aspect of formatting. This employs a trick other formatting libraries use, which is to have a table for all the formatted values between 00 and 99, and to then DivRem(..., 100) to cut the number of operations in half, which for longer values is meaningful. 2. Avoids going through the digit counting path when we know at the call site it won't be needed. 3. Employs a branch-free, table-based lookup for CountDigits(ulong) to go with a similar approach added for uint. * Address PR feedback (move license notice)
-
由 Austin Wise 提交于
The SDK also set the RuntimeHostConfigurationOption item. So setting in this project results in multiple copies of the same item. I'm not sure if this causes a problem with any currently supported configuration of .NET, but it is detectable when compiling this test with NativeAOT. ILC fails because multiple copies of the `--appcontextswitch` flag are passed with the same key.
-
由 Jan Vorlicek 提交于
Remove unused functionality from the CoreCLR PAL. Also replace the `DeleteFile` by standard `remove` function.
-
由 Tom Deseyn 提交于
* ZipArchive.CreateEntry: include file type in default ExternalAttributes. For Unix-type zip files, ExternalAttributes value matches 'struct stat' st_mode. Besides permissions it should also include the file type. * Use 'int' everywhere for ExternalAttributes. * Change back to use uint for external file attributes field.
-
由 Katelyn Gadd 提交于
If a method has the Intrinsic attribute, the interpreter should treat it as if it had AggressiveInlining set. (#76475) This significantly improves performance for System.Numerics vectors
-
由 Brian Bohe 提交于
* Fixing repro command line args * Update src/coreclr/tools/aot/crossgen2/Program.cs Co-authored-by: NBrian Bohe <brianbohe@microsoft.com> Co-authored-by: NJan Kotas <jkotas@microsoft.com>
-
由 Jakob Botsch Nielsen 提交于
JIT: Move impMathIntrinsic, lookupNamedIntrinsic, impUnsupportedNamedIntrinsic, impArrayAccessIntrinsic, impKeepAliveIntrinsic
-
由 Jakob Botsch Nielsen 提交于
-
由 Jakob Botsch Nielsen 提交于
JIT: Move impMarkInlineCandidate, impMarkInlineCandidateHelper, IsTargetIntrinsic, IsIntrinsicImplementedByUserCall, IsMathIntrinsic, impDevirtualizeCall, impConsiderCallProbe, compClassifyGDVProbeType, impGetSpecialIntrinsicExactReturnType
-
由 Jakob Botsch Nielsen 提交于
JIT: Move SpillRetExprHelper, addFatPointerCandidate, pickGDV, isCompatibleMethodGDV, considerGuardedDevirtualization, addGuardedDevirtualizationCandidate
-
由 Jakob Botsch Nielsen 提交于
JIT: Move impTransformThis, impCanPInvokeInline, impCanPInvokeInlineCallSite, impCheckForPInvokeCall
-
由 Jakob Botsch Nielsen 提交于
-
由 Jakob Botsch Nielsen 提交于
-
由 Jakob Botsch Nielsen 提交于
-
由 Jakob Botsch Nielsen 提交于
-
由 Jakob Botsch Nielsen 提交于
Also avoid marking some functions as 'inline' since then they cannot be used from other TUs (in general, we can just leave the DCE up to the compiler).
-
由 MSDN.WhiteKnight 提交于
-
由 Elinor Fung 提交于
-
由 Joni Aromaa 提交于
* Fix converting old SIMD ctors to HWI ones * Address PR feedback * Fix Linux build
-
由 Ivan Diaz Sanchez 提交于
* Quality Week 2022: The First Step to a Better Documentation. * Removed the Markdown Extension's additional development tags. * Removed redundant stuff. * Removed redundant stuff. * Removed guide testing versions to avoid confusion that they would be the only supported versions. * Addressed some Powershell comments, added a pointer to a yaml with links to other .NET sibling repos, and did a couple mini-corrections. Still got a few comments to address... * Fixed a ';:' with Powershell's syntax. * Addressed review comments: Redaction improvements, localization url's, improved code snippets, added info on individual subsets building configurations, changed COMPlus for DOTNET, and other misc improvements. * Addressed remaining comments. * Addressed some last comments regarding corerun, and the subset flags in the build scripts. Co-authored-by: NIvan Diaz <bluehorizon186@gmail.com>
-