Age | Commit message (Collapse) | Author | Files | Lines |
|
As part of my work on CPAOT implementation of GC ref map info
I have implemented decoder of the info so that it can be displayed
next to the import cell signatures.
This also uncovered one possible cause of R2RDump GcInfo-related
crashes that were observed by Andon and myself. It looks like Amy
in her initial implementation confused the various GC encodings
and used GC info to parse the import section auxiliary data which
actually contains the GC ref map info.
Thanks
Tomas
|
|
Update var life for multireg local
|
|
|
|
preview-27213-01, respectively (master) (#21503)
* Update CoreClr, CoreFx, CoreSetup to preview-27213-02, preview.18613.4, preview-27213-01, respectively
* Add System.Threading.AccessControl to CoreFX test deps
|
|
Fixes #20452.
|
|
will block to avoid the async break deadlock (#21319)
Setting the m_willBlockOnGarbageCollectionEvent earlier and only when we will block to avoid the async break deadlock
|
|
Signed-off-by: dotnet-bot <dotnet-bot@microsoft.com>
|
|
* Enable Azure Pipelines reporter so the test results can be seen at one place (i.e. "Tests" tab) - TestRunNamePrefix should encode all the needed information for a test run to be identifiable.
* Scenarios property should use comma (not semi-colon) to work correctly on non-Windows.
* Specify timeout to test jobs (the numbers are taken from netci.groovy) - we would adjust them as soon as we get this running in each PR.
* Specify timeout for Helix work items (the number is taken from buildpipeline) - same as above.
* Specify helixType and helixRepo properties to the job and use _HelixType and _HelixSource variables in MSBuild project instead of defining them manually.
* Limit Pri0 Checked test jobs to running against pull requests. This also should be limited in the future to "default" build definition name - so when we have PR workload selection feature working we can distinguish between default PR validation and user triggered jobs.
* Remove "Prepare tests for helix" step - instead of archiving Core_Root and tests folders ourselves - directly specify them as correlation payload and work item payload directories and let the archiving be handled by Helix Sdk.
|
|
When a multi-reg var is defined by a call, but doesn't currently reside in a register,
we must still update liveness.
Fix #21500
|
|
Issue #21485: fix EnumProcessModules hPseudoCurrentProcess bug.
Added handle reference.
Issue #21484: createdump segfaults with ASP.NET app
The problem is the ClrDataModule Request faulted on a dynamic module
getting the file layout flag.
Fixed the Request code not get the file layout and in the crash dump
code skip any dynamic modules.
|
|
In general, don't fold operations on relocatable immediates. Only allow EQ/NE folding, since relocation should preserve identity but not bit values or relative comparisons.
Closes #21483.
|
|
|
|
|
|
Fix HWIntrinsics ifdef
|
|
|
|
* Fix AVX2 and SSE2 XML comments
* Fix ARM intrinsic XML comments
|
|
with EEJitManager::SetCpuInfo in how ISA support checks are done (#21499)
|
|
Don't require BLK nodes for SIMD
|
|
|
|
Fix for PMI assert: 'vnWx != NoVN'
|
|
Optimize away the unbox type test when the jit knows the type of object
being unboxed and can resolve the type comparison at jit time.
Closes #14473.
|
|
We may now retype some immediates (for instance RVA statics feeding into Spans)
as byrefs. These need to be tracked as such during encoding so that the encoder
GC liveness tracking stays current.
Fixes the assert seen in #21466.
|
|
|
|
|
|
|
|
This change updates the following dependencies
- Microsoft.DotNet.Arcade.Sdk - 1.0.0-beta.18610.4
- Microsoft.DotNet.Helix.Sdk - 1.0.0-beta.18610.4
|
|
* Fixed bug in StackString where the size is not initialized correctly to STACK_COUNT
* Added CharString and WCharString template classes and a generic CharStringFromLPCWSTR method
* Added support for debugging a sandboxed app on Mac
This change fixes the usage of IPC while debugging while running in a sandbox. When running in a sandbox, the temporary folder for each process will be different. Thus the pipes being created in TwoWayPipe right now would be created in different directories in the debugger process and the process being debugged.
This change configures the folder to be used based on the application group ID that the sandboxed app belongs to.
For the same reasons, the names sempahores being used to synchronize the debugger attach need to be prefixed with the application group ID. This change was abit more involved since the name of the semaphore is limited to 31 characters, so we had to encode the semaphore names differently to make them shorter.
Last, new APIs to the debugger shim were added to support this new feature. This change only handles the runtime side and the dbgshim. An additional change to vsdbg needs to be done to use the new APIs.
fixes #21066
* Fixed build breaks on non Mac Unix platforms
* Fixed usage of gApplicationGroupId in non apple environments
* Fixed bug in semaphore names
* Got rid of usage of StackString
* Made PAL_GetApplicationGroupId Apple specific
* Added comment about pragma pack
* Fixed comment
* Added exported symbols
* Duplicated applicationGroupId so it can be used from another thread during register complete callback
* Renamed BitNum2ByteNum to GetExtraEncodedAreaSize to make intent clearer
* Fixed nit comments
* Removed redundant changes in StackString
* Fixed windows build break
* Fixed compilation switch from __APPLE to __APPLE__
* Added missing exports
|
|
We were not capturing failing inline observations for failed inlines
that got reported by the runtime directly. Move the point of capture
up in the jitted code so it always happens regardless of reporting
obligations.
|
|
other intrinsics (#21451)
* Updating Vector256.Create(V128, V128) to be implemented using other intrinsics
* Updating Vector128.Create(T, ...) and Vector256.Create(T, ...) to be implemented using other intrinsics
* Renaming CreateSoftware to SoftwareFallback and adding clarifying comments to the helper method code
* Fixing Vector128.Create(int, int, int, int) to properly consume the elements for the Sse2 codepath
|
|
|
|
|
|
|
|
|
|
GC (#21478)
|
|
Issue #18201 / Hackathon
|
|
* Adding support for the 2-byte VEX encoding to the emitter
* Relocating and expanding the comment which explains the two-byte VEX encoding check
|
|
Preserve the exception sets from child nodes when we create a new unique value number
|
|
* Parametrize test-job with helixQueues(Public|Internal) in eng/test-job.yml and specify them in eng/platform-matrix.yml
* Remove HelixTargetQueues property in tests/helixpublishwitharcade.proj
* Fix the error "unknown os: linux2" with Python script
|
|
Eliminate most cases where an OBJ or BLK node is required for SIMD
values. The exception is the case where a value produced by an intrinsic
(SIMD or HWIntrinsic) is used as an argument but the argument is of a
different SIMD type (e.g. a different baseType),
|
|
Signed-off-by: dotnet-bot <dotnet-bot@microsoft.com>
|
|
Fixes #21028
|
|
|
|
|
|
versions of Windows. (#21469)
|
|
|
|
* Add support for RVA fields to crossgen/R2R
RVA fields are became more common with pre-inititialized ReadOnlySpan<byte>. Fix crossgen to deal with them for R2R.
* Fix tests, map new JIT helper for R2R
|
|
* Update dependencies from https://github.com/dotnet/arcade build 677
This change updates the following dependencies
- Microsoft.DotNet.Arcade.Sdk - 1.0.0-beta.18607.6
- Microsoft.DotNet.Helix.Sdk - 1.0.0-beta.18607.6
* Update dependencies from https://github.com/dotnet/arcade build 679
This change updates the following dependencies
- Microsoft.DotNet.Arcade.Sdk - 1.0.0-beta.18608.1
- Microsoft.DotNet.Helix.Sdk - 1.0.0-beta.18608.1
* Update dependencies from https://github.com/dotnet/arcade build 679
This change updates the following dependencies
- Microsoft.DotNet.Arcade.Sdk - 1.0.0-beta.18608.1
- Microsoft.DotNet.Helix.Sdk - 1.0.0-beta.18608.1
|
|
respectively (#21452)
|
|
attributes) (#21462)
* Fix for #21456 - restrict increased generic attribute allocations to only generic attributes
This is a trivial quick-fix for #21456 where regressions between 2.1 and 3.0 were discovered on most attibute pathways due to the allocation overhead in the generic-supporting pathways. The workaround is to simply not take that slow/expensive path for non-generics.
While I'd like to optimize `RuntimeModule.ResolveMethod` further, there's a public surface area in play there that makes the changes non-trivial. There, we'll have to choose overhead on the public path (which may still be a net win), or duplication in code for another path.
* Update comments
|
|
* Delete vm/context.*
Leftover from remoting
|