Age | Commit message (Collapse) | Author | Files | Lines |
|
Extend PGO support from VC++ on WIN32 to Clang/LLVM on UNIX as well.
* Just like on Windows: if profile data is missing, skip enabling PGO
(allows non-PGO builds in branches where we don't publish PGO data).
* PGO with LTO requires additional dependencies (namely a discoverable
`ld.gold` and `LLVMgold.so`). To protect against broken support and
keep the build flexible across a wider array of distros, attempt to
detect whether PGO compilation would work (using cmake's
`try_compile()`), and fall back to a non-PGO/non-LTO build if the test
fails.
|
|
* Fix dependency to coreclr and mscorrc in Linux
Remove dependency with coreclr in some projects for Linux
Recover build mscorrc for cross architecture in Linux
* Fix CMakeLists.txt in src/dlls
minimize uses of CLR_CROSS_COMPONENTS_BUILD flag in CMakeLists.txt
Modify root/CMakeLists.txt to exclude corerun and coreconsole build
during cross component build
|
|
Enable Windows hosted, Linux target amd64 altjit
With this change, we build a JIT that runs on Windows amd64
and targets Linux amd64, as an altjit named linuxnonjit.dll.
This is useful for debugging, or generating asm code or diffs.
You can even easily create Windows/non-Windows asm diffs
(either to compare the asm, or compare the generated code size).
For this to work, the JIT-EE interface method
getSystemVAmd64PassStructInRegisterDescriptor() was changed
to always be built in, by defining `FEATURE_UNIX_AMD64_STRUCT_PASSING_ITF`
in all AMD64 builds. The `_ITF` suffix indicates that this is
functionality specific to implementing the JIT-EE interface
contract. There were many places in the VM that used this
interchangeably with `FEATURE_UNIX_AMD64_STRUCT_PASSING`. Now,
`FEATURE_UNIX_AMD64_STRUCT_PASSING` means code in the VM needed
to implement this feature, but not required to implement the
JIT-EE interface contract. In particular, MethodTables compute
and cache the "eightbyte" info of structs when loading a type.
This is not done when only `FEATURE_UNIX_AMD64_STRUCT_PASSING_ITF`
is set, to avoid altering MethodTable behavior on non-Unix
AMD64 builds. Instead, if `getSystemVAmd64PassStructInRegisterDescriptor()`
is called on a non-Unix build (by the altjit), the `ClassifyEightBytes()`
function is called, and nothing is cached. Hopefully (though it was
hard for me to guarantee by observation), calling `ClassifyEightBytes()`
does not have any side effects on MethodTables. It doesn't really matter,
since if called for altjit, we don't care too much about running.
The previously used `PLATFORM_UNIX` define is now insufficient.
I introduced the `#define` macros `_HOST_UNIX_` to indicate the
JIT being built will run on Unix, and `_TARGET_UNIX_` to indicate
the JIT is generating code targeting Unix. Some things were
converted to use the `UNIX_AMD64_ABI` define, which makes more
sense.
|
|
* Add scripts for building for arm Android
* Use arm soft float for bringup
* Use NDK r14
* Set HAVE_PT_REGS to false, as they are not available on Android
* _aligned_alloc is not available on clang
* Detect Visual C++ instead of clang
* Detect Android instead of clang
|
|
Signed-off-by: Jiyoung Yun <jy910.yun@samsung.com>
|
|
- Fix ABI flag based on target
- Exclude corefx for cross-architecture component
|
|
* Support cross-building CoreCLR for arm64 Android
* Link gcc_s before libunwind
|
|
platforms (#8976)
* Add way to build with FEATURE_STANDALONE_GC from build.sh
* Make CMake changes to build the GC 'PAL' as its own build target (to avoid -nostdinc).
In addition, introduce a "GC PAL" that provides an implementation of
GCToOSInterface on Unix-like platforms, for use with
FEATURE_STANDALONE_GC.
|
|
ARM32-target (#8687)
* fix CMakeLists.txt for palrt
* Modify CMakeLists.txt for cross-architecture component configuaration
Add coreclrpal project to fix a dependency issue when CMake configuration
* Allow only x64-host/arm-target for linux
allow x64-host/arm-target only, because pal is not stable for x86 now
* Update build.sh
delete NumProc=1 (inserted for debugging)
* Change x64-host to x86-host for arm-target
JIT correctness problem,
and fix CMake configuration for x86 in pal
* Fix type
* Update CMakeLists.txt
* fix style
* Set x86 when arm-target/x64-host
* fix typo
|
|
* Adding arm64 and updating default subsystems
|
|
* Fix buildsystem for linux cross-architecture component build
* refactoring build.sh, bug fix and typo fix
* Update build.sh
|
|
This patch will add x86 part for cross build
- add x86 in build-rootfs and related cross files
- add x86 parts in CMakeLists.txt to pass configuration
- add required source files to pass configuration
|
|
|
|
|
|
|
|
* Add PGO GENPROFILE support to coreclr and clrjit
Update the cmake build system to enable support for Profile Guided
Optimization (PGO) on Windows, and enable this feature for two target
binaries (coreclr and clrjit).
With this change, toggle between instrumented and profile-optimized
settings for target binaries by passing pgoinstrument argument to the build.cmd
Assume profile-optimized mode by default. Fall back to regular non-PGO
optimized builds if profile data is not available.
|
|
This change enables build of CoreCLR on Alpine Linux. Here is the list
of changes:
- Disable asserts checking RSP in arbitrary threads against cached stack limit
for the respective thread. The stack on Alpine obviously grows over the limit
reported by the pthread functions.
- Disable using XSTATE. This should be re-enabled after MUSL gets the _xstate,
_fpx_sw_bytes and related data structures added to the signal.h header.
- Disable setting rlimit of RLIMIT_NOFILE to the max value, since it breaks
debugging for some reason.
- Add skipping over the hardware signal trampoline in the PAL_VirtualUnwind.
While we were not trying to walk over it in a simple case, in a case where
an exception was thrown from a catch handler of a hardware exception, we
still attempted to walk over it and it fails on Alpine.
- Fix detection of Alpine Linux in the PAL's CMakeLists.txt so that it works
in Docker containers too.
- Modified PAL_VirtualUnwind to make the check for unwinding past the bottom
of the stack unconditional. We had a long list of platforms where we were
doing this check and it doesn't hurt to do it on platforms where it is not
needed. I have done that rather than adding a check for Alpine Linux as
another platform that needs it.
|
|
|
|
|
|
|
|
Fix platform logic in symbol stripping code
|
|
|
|
Pass --build-id=sha1 to linker explicitly
|
|
* Scripts: find out compiler definitions of CMake
In order to find mismatch between native and managed,
we need to know the list of definitions of native.
The copmiler definitions are stored at cmake.definitions
This addresses the complaints of #4674
Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
* Scripts: add check-definitions.py
scripts/check-definitions.py checks the consistency between
the native-build (CMake) compiler definitions and
the managed-build (MSBuild/mscorlib) compiler definitions
at build-time and prints out potentially dangerous
inconsistencies.
In order to get the proper results, managed build should
be executed after the native build (build.sh will do so
if no options such as skipnative or skipmanaged are given.)
Fix #4674
Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
* Scripts: allow check-definitions py to ignore the harmless
The third argument of check-definitions.py specifies harmless keywords
to be suppressed from emitting warning messages.
Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
* Scripts: add ignored cdefine keywords for warning
As an example of how to declare compiler definition keywords
that are harmless to be inconsistent between the native and the
managed, we have added FEATURE_IMPLICIT_TLS and FEATURE_HIJACK.
Developers may add more keywords in System.Private.CoreLib.csproj
if the keywords are verified to be harmless; i.e., although
the keywords exist in both cmake and clr.coreclr.props,
the keywords are NEVER used in either side of the sources
or the keywords only happen to have the same name while they
denote the completely different semantics and may be disjoint.
Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
|
|
On some platforms, a build-id was not being added to native artifacts
and we would like it to be present. Explicitly pass --build-id=sha1 to
the linker.
Fixes #5796
|
|
1. Update arm64 toolset to latest.
2. Link against ucrt.
3. Fix tls offsets which have changed due to updated toolset
4. Fix source code in decimal.cpp to avoid integer overflow. Result of signed integer overlfow is undefined in C++.
5. Enable build of sos which can be loaded in arm64 windbg
6. Add nop to empty assembly marker methods as new toolset generates invalid .pdata for them.
|
|
Adding Versioning to Native components in non-Windows builds.
|
|
Currently only enabled for arm64
|
|
Linux/ARM: Unifying the redundant usages of objcopy
|
|
Enable ReJIT on all platforms
|
|
Fix Arm64 break
|
|
|
|
Print message about detected NetBSD
|
|
|
|
|
|
DelayLoadAndFixWinRTInterop
|
|
|
|
This change adds a new argument to build.cmd, buildjit32, that configures
the build to build and link JIT32 instead of RyuJIT if the sources are
available in `src/jit32`.
|
|
|
|
Implement software write watch and make concurrent GC functional outs…
|
|
- Implemented software write watch using write barriers
- A new set of write barriers is introduced, each corresponding to an existing one, but which also updates the write watch table. The GC switches to a write watch barrier during concurrent GC, and switches back to a non write watch barrier after the final query for dirty pages.
- The write watch table is alloacted along with the card table
- Since the card table is used differently, different synchonization is used for the write watch table. The runtime is suspended during resize since that is the most infrequently occuring operation, of that, ResetWriteWatch, and GetWriteWatch.
- ResetWriteWatch() doesn't need a suspend, but since the software WW version is much faster than the Windows version, moved it into the suspended region to avoid some synchronization that would otherwise be required
- The background calls to GetWriteWatch() don't need or do a suspend. They only need to synchronize with the resize path, not for the purpose of correct functionality, but to not miss dirty pages such that concurrent GC is effective.
Miscellaneous:
- Fixed runtests.sh to copy mscorlib.dll and delete the Windows version of mscorlib.ni.dll
|
|
|
|
Let's remove the redundent file names using unified structure.
Signed-off-by: Geunsik Lim <geunsik.lim@samsung.com>
|
|
Add header file for the hosting API
|
|
Adding Common versioning to managed and Native components of coreclr
|
|
Arm softfp For Issue #3785
|
|
|
|
Add the header file and also modify coreruncommon.cpp to use it.
|
|
|
|
in integer registers for arm-softfp architecture.
|