- 16 6月, 2022 14 次提交
-
-
由 Thays Grazia 提交于
-
由 Andrew Au 提交于
Co-authored-by: NJuan Hoyos <juan.hoyos@microsoft.com> Co-authored-by: NNoah Falk <noahfalk@users.noreply.github.com>
-
由 Andy Ayers 提交于
Rework loop canonicalization in anticipation of extending it to handle non-loop edges like those seen in #70802 (details in #70569). The code to fix up flow out of `h` was not needed and has been removed. We now assert upstream that no such fix up will be needed.
-
由 Jan Kotas 提交于
FastInterlock* operations were #defined as aliases to regular Interlock* operations. Deleted the FastInterlock* aliases and replaced usage with Interlocked* operations.
-
由 Andy Gocke 提交于
Upgrade to released Preview 5 build
-
由 Jeremy Barton 提交于
-
由 Filip Navara 提交于
The support for building with GSS_KRB5_CRED_NO_CI_FLAGS_X was broken for quite some time. Attempts to reenable it failed due to bug in the krb5 GSSAPI implementation resulting in invalid memory accesses.
-
由 Michal Strehovský 提交于
Enables more aggressive metadata stripping by default. In this mode, the AOT compiler only generates reflection metadata for artifacts that were deemed reflection-visible by either dataflow analysis, or user inputs. Previously we would consider everything that got generated reflection-visible. * Change the defaults within the compiler (making the conservative mode opt-in). * Add tests for scenarios in https://github.com/dotnet/runtimelab/issues/656 (resolves dotnet/runtimelab#656). * Update tests * Do more marking where it was necessary
-
由 Andy Ayers 提交于
-
由 Egor Bogatov 提交于
-
由 Egor Bogatov 提交于
-
由 Aaron Robinson 提交于
* Predicate for valid Tracker Reference is inverted * Fix test to not inline wrapper creation function.
-
由 Jan Vorlicek 提交于
This change modifies the way hardware exceptions are propagated from the vectored exception handler. Until now, runtime was returning from the vectored exception handler with instruction pointer in the context set to an asm helper. That redirected the thread to that helper and we have raised an exception from its call chain. While working on CET support, it was found that propagating exceptions from the vectored exception handler directly works just fine. So this change makes it work that way for all Windows targets.
-
由 Vladimir Sadov 提交于
-
- 15 6月, 2022 26 次提交
-
-
由 Jan Kotas 提交于
The method was not actually using interlocked operation. It made it possible for multiple different values to win. Fixes #70742
-
由 Fan Yang 提交于
* First round of change of fixing C4018 * Address part of review feedback * Change the type of idx to unsigned for `effective_table_slow` * Add idx range check after the type was changed * Address review feedback for `class-init.c` * Change the return type of `*table_num_rows*` to `guint32`. Deal with the consequence of return type change of `table_info_get_rows`. Correct the type of a few local variables which store the return of `mono_metadata_token_index`. * Update return type * Address review feedbacks of metadata.c * Fix native crash * Make counter private to for-loop * Address review feedbacks * Address review feedbacks
-
由 Andrii Kurdiumov 提交于
* NativeAOT Unlock assignability comparison in reflection-free Closes #69960 * Document support for `IsAssignableFrom` * Document support for `IsInstanceOfType` * Document `IsAssignableTo` * Add suggestion from Michal * Update src/coreclr/nativeaot/System.Private.DisabledReflection/src/Internal/Reflection/RuntimeTypeInfo.cs Co-authored-by: NMichal Strehovský <MichalStrehovsky@users.noreply.github.com> * Remove whitespaces Co-authored-by: NMichal Strehovský <MichalStrehovsky@users.noreply.github.com>
-
由 Radek Zikmund 提交于
-
由 Tanner Gooding 提交于
Updating HWIntrinsicInfo::lookupId to not accelerate Vector256 APIs when AVX2 is unsupported (#70686) * Updating HWIntrinsicINfo::lookupId to not accelerate Vector256 APIs when AVX2 is unsupported * Fixing a check in lookupId to properly negate the condition * Ensure the special-cased EA_32BYTE constants only happen when AVX/AVX2 are supported * Fixing a bad assert in morph
-
由 Radek Zikmund 提交于
* Inline state transition helpers Fixes #55437 * Add high level comments for HandleEventReceive and ReadAsync
-
由 SingleAccretion 提交于
* Fix numbering of HWI loads Take into account all operands and describe the exception sets precisely. * Add a test * Recast the check
-
由 Vladimir Sadov 提交于
* Enabling return address hijacking on ARM64 Windows * pass right flags to the GcInfoDecoder
-
由 Jan Kotas 提交于
-
由 Tomas Weinfurt 提交于
This reverts commit 84f7cad0.
-
由 SingleAccretion 提交于
* Delete the unused "GTF_USE_FLAGS" It used to indicate that a branch operation didn't need to materialize its operand and could just "use flags" instead, but that purpose has long been lost now that we have explicit SETCC nodes in lowering. * Make GTF_DONT_EXTEND a shared flag So that it can be used for "LCL_FLD" as well as "GT_IND". No diffs. * Enable TYP_STRUCT on the RHS * fgMorphBlockOperand * Tweak TYP_STRUCT LCL_FLD costs Model it as two load, like OBJ. Note we could be more precise here, by using the register type of the layout. For now, we defer. * Block CSE Preserve previous behavior to avoid diffs.
-
由 SingleAccretion 提交于
* Add GenTree::GetLayout * x86/Unix x64: lowering * x86/Unix x64: "genConsumePutStructArgStk" * x86/Unix x64: "genStructPutArgUnroll" * x86/Unix x64: codegen asserts * x86: "genStructPutArgPush" * Unix x64: "genStructPutArgPartialRepMovs"
-
由 SingleAccretion 提交于
* Separate out LowerPutArgStk Call it from "LowerArg" instead of "ContainCheckCallOperands", as it does more than just containment analysis. * Support "PUTARG_STK(STRUCT LCL_VAR/LCL_FLD)" on ARM/64 To test this, transform "OBJ(LCL_VAR|FLD_ADDR)" to "LCL_FLD" on lowering. This additionally simplified the codegen code as it doesn't have to support two representations of the same thing.
-
由 SingleAccretion 提交于
* Do not change "gtRetClsHdl" in "impNormStructVal" Doing so leads breaking the proper ABI handling for the call. * Add a test
-
由 SingleAccretion 提交于
* Refactor implicit by-refs morphing Move it to "fgMorphLocal". Enabled folding in "LocalAddressVisitor". * Add FEATURE_IMPLICIT_BYREFS And use it in all the necessary places. Resolves the TP regression on Unix x64, turning it into a TP improvement. * Add a zero-diff quirk
-
由 SingleAccretion 提交于
* Add a test * Fix out-of-bounds loads in ARM/64's "genPutArgSplit" * Split the test out
-
由 Tomáš Rylek 提交于
According to my comparative perf measurements composite framework spends about 5M more instructions in the method coreclr.dll!CMiniMdTemplate<CMiniMd>::SearchTableForMultipleRows(CMiniColDef sColumn) I tracked this down to the R2R table AttributePresence which is used to speed up attribute queries against the ECMA metadata model. In composite mode we were putting the table in the image header, not component header, and so the runtime was unable to locate it. This change fixes generation of the table in Crossgen2; I have verified that with this change I no longer see the 5M outlier. Thanks Tomas
-
由 Egor Bogatov 提交于
-
由 Buyaa Namnan 提交于
-
由 Badre BSAILA 提交于
* ConfigurationBinder.Bind on virtual properties duplicates elements * simplify GetAllProperties * return array instead of list
-
由 Andy Gocke 提交于
* Add doc on Unix temporary file security practice * Update unix-tmp.md * Update unix-tmp.md * Add example permissions encoding * Update docs/design/security/unix-tmp.md Co-authored-by: NDan Moseley <danmose@microsoft.com> * Update unix-tmp.md Co-authored-by: NDan Moseley <danmose@microsoft.com>
-
由 Thays Grazia 提交于
* Implement get bytes from loaded_files using debugger protocol. * fix pdb size == nul * Adressing @radical comments. * Fix build. * fix compilation * Addressing @radical comments. Co-authored-by: NAnkit Jain <radical@gmail.com>
-
由 Gleb Balykov 提交于
Do not use ExecutableMemoryAllocator on 64 bit platforms if default base address usage is requested (#70563)
-
由 Aleksey Kliger (λgeek) 提交于
* [wasm] Make runtime_is_initialized promise callbacks one-shot Throw if runtime_is_initialized_resolve or runtime_is_initialized_reject is called more than once * Add a slightly generalized GuardedPromise<T> object Protects against multiple-resolve, multiple-reject, reject after resolve and resolve after reject. Does not protect against the executor throwing.
-
由 Aleksey Kliger (λgeek) 提交于
* Fix cmake error ``` Manually-specified variables were not used by the project: CONFIGURATION_WASM_OPT_FLAGS ``` * Build the interpreter with -O1 on Wasm in Debug configs Otherwise `interp_exec_method` and `generate_code` can easily overflow the stack in some browsers with even a few recursive calls (for example during .cctor initializaiton)
-
由 Dennis Yemelyanov 提交于
* align memory load calculation in GC with Docker and Kubernetes * pass inactive file field name as a parameter
-