| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
joperezr/DisableNativeResourceCompileForManagedCode
Disabling NativeResourceCompile Target for Managed libraries
|
|
'readytorun/mainv1' and 'readytorun/mainv2' tests on Linux wait for human
response asking permission to remove write-protected file:
rm: remove write-protected regular file 'test.dll'?
'rm' command shall be used with '-f' flag to make it non-interactive.
|
|
This change fixes annotations of AMD64 assembler helpers on Unix. The helpers that
don't use RBP as a stack frame were missing setting the RSP as a CFA register and
unwinding through them was failing.
This was hit in R2R scenarios, with the DelayLoad_Helper_Obj not being unwindable.
|
|
While fixing up a call from a Ready-to-Run method, we don't always
have the right implementation MethodDesc for the target of the call.
Thus it is potentially unsafe to DoPrestub using the MethodDesc we have,
causing a test failue (Loader.classloader_regressions_429802_CMain)
in issue #5366. This is fixed by not calling DoPrestub from
ExternalMethodFixupWorker.
|
|
Currently, we can build coreclr with -O1 flag of clang in release
build mode for Linux/ARM. This instruction is to enable -O3 optimization
level of clang on ARM/Linux by fixing the bug of llvm.
Signed-off-by: Geunsik Lim <geunsik.lim@samsung.com>
|
|
CoreDisTools Package: Make minor version generic
|
|
Inliner: enable inlining of methods with cpblk
|
|
Change the version requirement of CoreDisTools package
from 1.0.1-prerelease-00001 to 1.0.1-prerelease-*
so that the test infrastructure picks up the latest
available version.
If only the 1.0.1-prerelease-xxxxx number changes,
there is no change in the library API. There may be
a change in the implementation.
The current need for this change is that we need to pick up
a new version of the CoreDisTools package (with an
implementation fix) for X86 testing.
|
|
Disable perf_codequality_math_functions
|
|
|
|
abssingle relies on a certain level of floating point precision and fails if the diff is outside of that precision.
|
|
Disable RVA statics tests outside Windows.
|
|
Fix Details for ILAsm/ILDasm/ClrJit Binaries
|
|
|
|
GCStress: Fix a race-condition
|
|
Fix for methods that tail call via helpers
|
|
couple scenarios. (#5399)
* Fixing an error in PAL_pow that caused it to return an incorrect value for a couple scenarios.
* Adding additional PAL_pow tests.
|
|
Fixes https://github.com/dotnet/coreclr/issues/5408
|
|
Fix details for CoreCLR binary
|
|
Include license files into the packages
|
|
These tests are not supported in general and are already disabled
for Windows. Disable them for *nix as well.
|
|
|
|
Currently, FEATURE_HIJACK is disabled for ARM, and thus CoreCLR got
stuck if GC is invoked from JITed code.
This commit attempts to port FEATURE_HIJACK from ARM (windows) in order
to fix #5403.
|
|
Existing Ready to Run implementation doesn't fully validate requests
to create object instances, and allows some invalidate requests (e.g.,
to create an instance of an abstract class). This causes failure of
test case Loader.classloader_generics_Instantiation_Negative_abstract01
in issue #5366.
|
|
|
|
Fix for HFA args on ARM32
|
|
Recursive generic definitions can easily make CrossGen take very
long time to complete. This is the cause of a failure in issue #5366.
This is fixed by limiting the number of methods to compile by CrossGen.
|
|
|
|
* Adding some basic System.Math performance tests.
* Renaming 'floatnative' to 'floatdouble'.
* Removing outdated workarounds in the floatdouble interop code.
* Renaming 'finite.cpp' to 'math.cpp'
* Updating the double-precision math tests.
* Updating PAL_EPSILON to be more precise.
|
|
on the caller's frame. If an argument lives on the caller caller's frame, it may get
overwritten if that frame is reused for the tail call. Therefore, we should always copy
struct parameters if they are passed as arguments to a tail call.
The simple il regression test has a scenario similar to that of the F# repro in #5164.
Closes #5164.
|
|
* Factor out import of `new` multi-dimmensional array
* Implement import of `new` multi-dimmensional array using the non-vararg helper
* Delete RETURN_TYPE_IS_COMPOSITE inlining observation because of it is never hit.
* Add R2R id for non-vararg helper
* Reformat affected code to follow coding conventions
|
|
Remove some lexical-CSE-related code.
|
|
This change ensures that calls to CORINFO_HELP_STOP_FOR_GC() themselves
are not converted to GC-Stress traps -- thus preventing the race between
the GC thread, and a managed thread under GCStress.
Identification of calls to CORINFO_HELP_STOP_FOR_GC():
Since this is a GCStress only requirement, its not worth special identification in the GcInfo
Since CORINFO_HELP_STOP_FOR_GC() calls are realized as indirect calls by the JIT, we cannot identify
them by address at the time of SprinkleBreakpoints().
So, we actually let the SprinkleBreakpoints() replace the call to CORINFO_HELP_STOP_FOR_GC()
with a trap, and revert it back to the original instruction the first time we hit the trap in
OnGcCoverageInterrupt().
Upside: No change to GCInfo or JIT is necessary
Downside: Need to decode a few bytes on every GCStress interrupt.
Fixes #4794
|
|
Named mutexes: fix handle leak, add stress tests
|
|
Publish pkg/symbolpkg folders within Release/Debug folders
|
|
|
|
|
|
Lexical CSE is no more; remove some related code:
- Remove the `lexicalCSE` parameter and related checks from `optConfigDisableCSE`.
- Remove the `VALNUM_CSE_ENABLED` symbol: this symbol is unconditionally
set to 1 and doesn't control much other than the behavior of
`optConfigDisableCSE`
|
|
|
|
uninstantiated generic method." (#5347)
The problem was that the delegates code base was not correctly instantiating interface GVMs on target types
(after finding the target method on the target type using slot numbers), and ended up calling uninstantiated
generic method definitions on delegate invokes.
|
|
ARM64: eh11_large exclusion
|
|
Publish packages into folders named for their configuration
|
|
|
|
Add new document on how to setup/view JIT dumps
|
|
This test often fails in release run.
|
|
Shrink Task size by 10%
|
|
|
|
Remove double avoid in comments
|
|
Add extra options to 'runtest.sh' for effective test monitoring.
|
|
|
