Age | Commit message (Collapse) | Author | Files | Lines |
|
Add symbol packages for coreclr runtime.
|
|
Fix error return in getPath
|
|
Fix brick table issue on Unix
|
|
Fix ByValArray support.
ByValArray marshalling is implemented actually in OleVariant.cpp, which implements both regular array marshalling and variant marshalling). Regular array marshalling code should be supported in x-plat but it was incorrectly #ifdef-ed out. Fix by only #ifdef the COM interop VARIANT part.
Fix #4262 and #4194
|
|
Depends on the new buildtools symbol support.
Update build tools version containing the new symbol file support.
|
|
Add support for x86 long subtract operator
|
|
Clean up JIT build rules
|
|
Cleanup: be consistent about compCalleeFPRegsSavedMask #ifdefs
|
|
Cleanup: remove getFramePointerReg()
|
|
Add genEstablishFramePointer() to enapsulate frame pointer creation
|
|
Minor JIT changes
|
|
With #3296, we have all the support for subtracting longs on x86. Subtract
can use the same logic as addition for longs.
Fixes #4189.
|
|
actually implements both regular array marshalling and variant marshalling) and is incorrectly #ifdef-ed out under x-plat. Fix by only #ifdef the COM interop VARIANT part.
Re-enable ByValArray test as well.
|
|
Inliner: initial version of a model-based policy
|
|
There is a problem with Clang compiling a loop that sets a range of brick
table entries that are "shorts" as a memset. That causes it to be written
by bytes in some cases on OSX (and it is a mere chance that the memset on
Linux is implemented so that it doesn't use byte access).
This was causing corruption of the brick table on OSX.
The fix is to mark the pointer that iterates over the brick table as volatile
to prevent the compiler from using the memset.
|
|
Fix Linux-ARM Broken Stack w/ ThePreStub
|
|
Fix a couple of erroneous assert(trues)
|
|
|
|
|
|
[Arm64]:Fixing the Logic of ResolveWorkerChainLookupAsmStub for VSD
|
|
|
|
Add a model-based size predictor to the `DiscretionaryPolicy` and
dump its data along with observations. This can be used to assess
the performance of the model. Currently using a linear model that
can explain about 55% of the variability in code size impact for
mscorlib. Whether this holds up for other situations is still to
be determined.
Create a `ModelPolicy` that relies on this code size model to make
inlining decisions. This is very much a work in progress as the code
size models are still not well vetted and there is no real code
quality model yet.
|
|
ARM64 Work Item 3817, 3524 - Struct16 decomposition
|
|
Add a useful assert about SP to FP offsets.
Add "const" to a few member functions.
|
|
|
|
|
|
|
|
Better sos message for runtime not initialized
|
|
Replace CQuickSort with qsort in GC info encoders.
|
|
Changes to support passing of MultiReg structs using GT_LISTs
Optional support for struct promotion for multireg structs:
To enable set FEATURE_MULTIREG_STRUCT_PROMOTE to 1
Morphs the 16-byte structs at the end of fgMorphArgs
Careful refactoring to avoid changes to UNIX_AMD64 code
Covers all of the 16-byte struct expansion cases in fgMorph
Added function header comments
Passing the tests for Arm64
No AsmDiffs for non-Arm64 targets.
Codegen uses Contained nodes for PUTARG_STK 16-byte stack args
Created a genPutArgStk method for Arm64
|
|
Merge changes from TFS
|
|
Also fixed the corerun sos command checked build problem.
|
|
Add SIMD files to x86 build (they aren't used, yet).
Separate out the LEGACY_BACKEND files, to be explicit about them.
Make protojit and standalone build rules more similar.
|
|
[tfs-changeset: 1594511]
|
|
This removes another utilcode dependency from the GC info encoders.
|
|
Remove Marshal.Read/Write (AsAny object) overload from CoreCLR.
|
|
expect anyone to use these overloads and it doesn't make sense to keep them. Their semantics are rather convoluted and most likely not what you might expect.
|
|
Add error logging to 4 PAL tests that fail periodically on OSX
|
|
The call to PathCharString.Set returns true on succes, so this code path
was returning false if it succeeded in setting lpPathFileName to
lpFileNameString. It looks like this is just a holdover from the way
this code used to be written.
|
|
Inliner: update DiscretionaryPolicy to look at opcodes and arg info
|
|
Remove non-portable dependencies from SimplerHashTable.
|
|
Have the `DiscretionaryPolicy` do some simple opcode histogramming
to try and provide better observations for code size estimation.
Likewise, look at the types and sizes of arguments and report those.
|
|
Add header file for the hosting API
|
|
[ARM64] Fixes Updating Registers for managed frames in GC stack walks
|
|
Add CoreDisTools header
|
|
|
|
Make current culture travel with async operations on Linux
|
|
|
|
Fix an issue where COREHOST_SERVER_GC and COREHOST_CONCURRENT_GC were…
|
|
|