Jan Vorlicek [Tue, 28 Feb 2017 19:19:24 +0000 (20:19 +0100)]
Fix mismatch in the cmake script endif condition (#9859)
This change fixes a mismatch in the cmake script endif condition introduced
by #9809.
Gaurav Khanna [Tue, 28 Feb 2017 19:02:07 +0000 (11:02 -0800)]
Merge pull request #9781 from dotnet-bot/master-UpdateDependencies
Update CoreClr, CoreFx to beta-25028-02, stable, respectively (master)
Bruce Forstall [Tue, 28 Feb 2017 17:57:34 +0000 (09:57 -0800)]
Merge pull request #9852 from mskvortsov/ryujit-arm32-label
[RyuJIT/ARM32] enable CG for GT_LABEL nodes
Andy Ayers [Tue, 28 Feb 2017 17:15:14 +0000 (09:15 -0800)]
Remove some unused benchmarks (#9843)
Pat Gavlin [Tue, 28 Feb 2017 16:30:46 +0000 (08:30 -0800)]
Merge pull request #9796 from pgavlin/FixOurGetThreadCycles
Fix _our_GetThreadCycles for clang/xarch.
Hyeongseok Oh [Tue, 28 Feb 2017 13:09:37 +0000 (22:09 +0900)]
Fix cross component builderror - CMakeLists.txt for PAL (#9819)
define & use PAL_ARCH_SOURCES_DIR in PAL CMakeLists.txt to build assembly based on platform architecture.
root/CMakeLists.txt defines ARCH_SOURCES_DIR based on target architecture.
Mikhail Skvortcov [Tue, 28 Feb 2017 13:01:20 +0000 (16:01 +0300)]
RyuJIT/ARM32: enable CG for GT_LABEL nodes
Frederik Carlier [Tue, 28 Feb 2017 12:55:59 +0000 (13:55 +0100)]
Disable DBGIPC_TRANSPORT_DI and DBGIPC_TRANSPORT_VM on Android (#9809)
* Disable DBGIPC_TRANSPORT_DI and DBGIPC_TRANSPORT_VM on Android
Jonghyun Park [Tue, 28 Feb 2017 10:36:55 +0000 (19:36 +0900)]
Use BIT64 instead of _WIN64 for LONG helpers (#9845)
dotnet-bot [Tue, 28 Feb 2017 09:28:23 +0000 (09:28 +0000)]
Update CoreClr, CoreFx to beta-25028-02, stable, respectively
Yongseop Kim [Tue, 28 Feb 2017 09:25:10 +0000 (18:25 +0900)]
[x86/linux] Implement BackPatchWorkerAsmStub (#9690)
* [x86/linux] Implement BackPatchWorkerAsmStub
This implementation is ported from BackPatchWorkerAsmStub in
vm/i386/virtualcallstubcpu.hpp
Sergey Andreenko [Tue, 28 Feb 2017 04:14:27 +0000 (20:14 -0800)]
Merge pull request #9774 from sandreenko/coreRT-fat-pointers-fix
coreRT fat pointer hidden argument fix for calls with retBufArg
Dan Moseley [Tue, 28 Feb 2017 03:39:04 +0000 (19:39 -0800)]
Replace INSIDE_CLR with CORECLR (#9841)
Alex Perovich [Tue, 28 Feb 2017 02:54:54 +0000 (20:54 -0600)]
Move Used interop sources into shared folder (#9840)
Sergey Andreenko [Tue, 28 Feb 2017 02:41:40 +0000 (18:41 -0800)]
CC fix
Andy Ayers [Tue, 28 Feb 2017 02:10:27 +0000 (18:10 -0800)]
Some tests for devirtualization (#9834)
Jonghyun Park [Tue, 28 Feb 2017 02:06:38 +0000 (11:06 +0900)]
Use flags in exceptionhandling.cpp (#9823)
* Add USE_FUNCLET_CALL_HELPER
* Use STACK_RANGE_BOUNDS_ARE_CALLER_SP
Jonghyun Park [Tue, 28 Feb 2017 01:26:51 +0000 (10:26 +0900)]
[x86/Linux] Fix GetCurrentSP (#9838)
Sergey Andreenko [Fri, 24 Feb 2017 21:39:36 +0000 (13:39 -0800)]
coreRT fat pointer hidden argument fix for calls with retBufArg
Aditya Mandaleeka [Mon, 27 Feb 2017 23:34:48 +0000 (15:34 -0800)]
Merge pull request #9837 from adityamandaleeka/remove_useless_message
Remove useless message from runtest output.
Aditya Mandaleeka [Mon, 27 Feb 2017 23:30:15 +0000 (15:30 -0800)]
Remove useless message from runtest output.
Michal Strehovský [Mon, 27 Feb 2017 23:21:57 +0000 (15:21 -0800)]
Don't generate a runtime lookup for static base if not needed (#9833)
`fgInitThisClass` makes sure that the type initializer for the type that
owns the method being compiled has run. On CoreCLR, the fallthrough code
somehow ends up not doing a runtime lookup if the owning type is not
canonical. We need to avoid it on CoreRT as well.
Pat Gavlin [Mon, 27 Feb 2017 21:05:59 +0000 (13:05 -0800)]
Merge pull request #9795 from pgavlin/RefactorIsContained
Refactor isContained().
Per Lundberg [Mon, 27 Feb 2017 20:57:48 +0000 (22:57 +0200)]
README.md: More idiomatic Markdown usage (#9799)
- In essence, use backticks where applicable.
- Also, changed some incorrect case (`Build.cmd` vs `build.cmd` which is the canonical name; if you have your git repo on a case-sensitive filesystem like I do, this makes a real difference - you can't run it with the incorrect case).
- Fixed naming of Git and CMake to use their official capitalization.
mikedn [Mon, 27 Feb 2017 20:44:27 +0000 (22:44 +0200)]
Improve BitConverter.ToIntNN methods (#9803)
* Improve BitConverter.ToIntNN methods
- Avoid duplicate argument validation
- Fix failed optimization attempt: (uint)i < a.Length needs to be (uint) < (uint)a.Length otherwise it gets promoted to long and that's bad for 32 bit architectures
- Workaround JIT limitation: fixed variables cannot be enregistered
- Use Unsafe.ReadUnaligned to avoid alignment handling complications
- Use Unsafe.As instead of fixed as it generates shorter code
- Don't forward unsigned method to signed methods. The JIT doesn't want to inline the signed methods even though the generated code isn't significantly larger
mikedn [Mon, 27 Feb 2017 20:05:13 +0000 (22:05 +0200)]
Use Unsafe.ReadUnaligned/WriteUnaligned in UMA (#9829)
mikedn [Mon, 27 Feb 2017 16:40:51 +0000 (18:40 +0200)]
Add Unsafe.ReadUnaligned/WriteUnaligned (#9808)
Jonghyun Park [Mon, 27 Feb 2017 15:00:42 +0000 (00:00 +0900)]
Use WIN64EXCEPTIONS to declare EE_PERSONALITY_ROUTINE helpers (#9821)
Jonghyun Park [Mon, 27 Feb 2017 14:57:20 +0000 (23:57 +0900)]
[x86/Linux] Clean up unused NakedThrowHelper stub (#9824)
Jonghyun Park [Mon, 27 Feb 2017 10:07:05 +0000 (19:07 +0900)]
[x86/Linux] 16-byte aligned Funclet Helpers (#9810)
Jonghyun Park [Mon, 27 Feb 2017 05:21:19 +0000 (14:21 +0900)]
Clean up PInvokeStubForHost and PInvokeStubForHostInner (#9815)
Jonghyun Park [Mon, 27 Feb 2017 05:20:43 +0000 (14:20 +0900)]
Clean up nested WIN64EXCEPTIONS-ifdef (#9817)
Jonghyun Park [Mon, 27 Feb 2017 04:09:10 +0000 (13:09 +0900)]
[x86/Linux] Use _X86_ instead of _TARGET_X86_ (#9816)
Jonghyun Park [Mon, 27 Feb 2017 02:46:19 +0000 (11:46 +0900)]
[x86/Linux] Clean up PopSEHRecords (#9812)
Jonghyun Park [Mon, 27 Feb 2017 02:45:50 +0000 (11:45 +0900)]
[x86/Linux] Clean up CallRtlUnwind (#9813)
Jonghyun Park [Mon, 27 Feb 2017 02:41:48 +0000 (11:41 +0900)]
[x86/Linux] Clean up ARRAY STUB(s) (#9811)
Jan Kotas [Sun, 26 Feb 2017 20:55:49 +0000 (12:55 -0800)]
Delete a bit of dead code (#9801)
Pat Gavlin [Sun, 26 Feb 2017 05:10:01 +0000 (21:10 -0800)]
Merge pull request #9789 from mikedn/le-intrinsic
Treat BitConverter.IsLittleEndian as an intrinsic
Eric Mellino [Sun, 26 Feb 2017 02:54:06 +0000 (18:54 -0800)]
Merge pull request #9782 from mellinoe/ilasm-pathlength-fix
Increase the length of allowed parameters in ilasm
Jan Kotas [Sun, 26 Feb 2017 02:27:32 +0000 (18:27 -0800)]
Update BitConverter from CoreFX (#9788)
* Update BitConverter from CoreFX
Related to #9701
* Use ThrowHelper
Ben Adams [Sun, 26 Feb 2017 00:02:14 +0000 (00:02 +0000)]
Better ArraySegment .ctor tests (#9800)
Hyung-Kyu Choi [Sat, 25 Feb 2017 19:24:16 +0000 (04:24 +0900)]
[ARM/CI] Fix arm32 CI git check failure (#9798)
Because layout in corefx binary has been changed, it causes arm CI failure.
And skip arm32 tests because test is not working recently.
Signed-off-by: Hyung-Kyu Choi <hk0110.choi@samsung.com>
Hyung-Kyu Choi [Sat, 25 Feb 2017 18:29:20 +0000 (03:29 +0900)]
[ARM/CI] Update arm CI to sync up with corefx change (#9797)
Signed-off-by: Hyung-Kyu Choi <hk0110.choi@samsung.com>
Pat Gavlin [Sat, 25 Feb 2017 18:09:00 +0000 (10:09 -0800)]
Fix _our_GetThreadCycles for clang/xarch.
The `"=A"` constraint in `asm volatile("rdtsc" : "=A"(cycles))` does not work as expected
on AMD64. Instead, the assembly must use an output constraint for both `eax` and `edx`
(i.e. `=a"(lo), "=d"(hi)`).
Pat Gavlin [Fri, 24 Feb 2017 19:19:27 +0000 (11:19 -0800)]
Refactor isContained().
This changes isContained to terminate early if it is not possible for the node in question
to ever be contained. There are many nodes that are statically non-containable: any node
that is not a value may not be contained along with a small number of value nodes. To avoid
the need for a `switch` to identify the latter, their node kinds have been marked with a
new flag, `GTK_NOCONTAIN`. The checks for identifying non-containable nodes are encapsulated
withing a new function, `canBeContained`.
Jan Kotas [Sat, 25 Feb 2017 16:09:54 +0000 (08:09 -0800)]
Merge pull request #9791 from dotnet/revert-9231-MinOptsSimplifiedGCtracking
Revert "Simplified MinOpts GC ref tracking."
Jan Kotas [Sat, 25 Feb 2017 14:33:11 +0000 (06:33 -0800)]
Revert "Simplified MinOpts GC ref tracking."
Jan Vorlicek [Sat, 25 Feb 2017 10:08:35 +0000 (11:08 +0100)]
Fix GC hole when exception filter throws unhandled exception (#9785)
The extra Unix specific piece of code in the StackFrameIterator::Filter that
handles the difference in the exception stack unwinding on Unix was not
skipping exception trackers belonging to filter clauses. But that was not
right, since filter funclet stack frames behave the same way on Windows and
Unix. They can be present on the stack when we reach their parent frame if
the filter hasn't finished running yet or they can be gone if the filter
completed running, either succesfully or with unhandled exception.
This change adds skipping of filter funclet related exception trackers at
that place so that the common code processes them.
This fixes the GC hole mentioned in the title that was discovered when
running some tests with GCStress mode 2.
Mike Danes [Sat, 25 Feb 2017 09:45:02 +0000 (11:45 +0200)]
Treat BitConverter.IsLittleEndian as an intrinsic
James Ko [Sat, 25 Feb 2017 06:48:36 +0000 (01:48 -0500)]
Remove `this == null` checks from string.Equals (#9768)
Alex Perovich [Sat, 25 Feb 2017 06:46:59 +0000 (00:46 -0600)]
Move S.Buffers and S.IO sources shared with corert into shared folder (#9778)
Andy Ayers [Sat, 25 Feb 2017 01:51:10 +0000 (17:51 -0800)]
Merge pull request #9756 from AndyAyersMS/RefCountGenericContext
JIT: keep ref count for runtime generic context lookups
Stephen Toub [Sat, 25 Feb 2017 01:39:44 +0000 (20:39 -0500)]
Use clock_gettime(CLOCK_REALTIME) in DateTime.UtcNow (#9772)
gettimeofday() is currently used in DateTime.UtcNow. Depending on the system, it may have accuracy and precision in line with either clock_gettime(CLOCK_REALTIME) or clock_gettime(CLOCK_REALTIME_COARSE), but in returning its data it's necessarily limited by struct timeval's tv_usec field (microseconds), whereas clock_gettime uses struct timespec which has tv_nsec (nanoseconds). This commit switches to use clock_gettime(CLOCK_REALTIME) to gain the additional precision (though the FILETIME through which the data is returned only supports 100-nsec precision). On my machine, there was no measurable impact to throughput.
Ahson Ahmed Khan [Sat, 25 Feb 2017 01:27:54 +0000 (17:27 -0800)]
Merge pull request #9598 from ahsonkhan/OptimizeSpanClear
Optimize span clear
Eric Mellino [Sat, 25 Feb 2017 00:48:50 +0000 (16:48 -0800)]
Increase the length of allowed parameters in ilasm.
The way this code is structured prevents parameters longer than 128 characters
from being passed. This change simply increases that to 1024 characters, although
a more robust change could be made in the future to account for platform differences.
Aditya Mandaleeka [Sat, 25 Feb 2017 00:37:26 +0000 (16:37 -0800)]
Remove optimizedtls.cpp (#9777)
Jan Vorlicek [Fri, 24 Feb 2017 21:47:31 +0000 (22:47 +0100)]
Remove NakedThrowHelper and ifdef-out its callers (#9771)
This change removes NakedThrowHelper function for Unix since it was not used.
It also ifdefs out its upstream callers.
Andy Ayers [Fri, 24 Feb 2017 21:35:37 +0000 (13:35 -0800)]
review feedback
Pat Gavlin [Fri, 24 Feb 2017 21:17:56 +0000 (13:17 -0800)]
Merge pull request #9336 from mikedn/const-fold
Fix incorrect cast in gtFoldExprConst
Ian Hays [Fri, 24 Feb 2017 21:08:17 +0000 (13:08 -0800)]
Merge pull request #9769 from ianhays/filelock_osx
Remove OSX FileStream Lock/Unlock
Jan Vorlicek [Fri, 24 Feb 2017 20:29:15 +0000 (21:29 +0100)]
Remove getcontext and setcontext usage (#9759)
It turns out that the getcontext and setcontext that I have used in my stack
oveflow reporting change are not present e.g. on Alpine Linux or on Android.
So I am replacing their usage with RtlCaptureContext and RtlRestoreContext
instead.
I have also found that the addition of the .cfi_adjust_cfa_offset to the
PROLOG_SAVE_REG_PAIR has broken unwinding of all helpers that use
PROLOG_WITH_TRANSITION_BLOCK, because the PROLOG_STACK_ALLOC macro
updates the CFA offset. So I am fixing that by removing the CFA offset
updating from the PROLOG_STACK_ALLOC and adding explicit one to the
ARM64 CallSignalHandlerWrapper.
Bruce Forstall [Fri, 24 Feb 2017 20:28:50 +0000 (12:28 -0800)]
Merge pull request #9732 from hseok-oh/fix_9442
Fix code generation for array length check in legacy JIT
Mike Danes [Sat, 4 Feb 2017 10:17:13 +0000 (12:17 +0200)]
Enable DevDiv_377155
Mike Danes [Sat, 4 Feb 2017 10:14:54 +0000 (12:14 +0200)]
Fix incorrect cast in gtFoldExprConst
The fact that an operation is unsigned affects the operation itself but that doesn't mean that the result of the operation is also unsigned. Constants are stored as ssize_t and the node type is TYP_INT so the result has to be sign extended, not zero extended.
Otherwise code that uses the return of IconValue() without first narrowing it to int will behave incorrectly. Such code does exists, even gtFoldExprConst does this when folding shift operations:
uint a = uint.MaxValue;
uint b = 0;
int r = (int)checked(a + b);
Console.WriteLine((r >> 2).ToString("X"));
The above code prints
3FFFFFFF instead of the expected
FFFFFFFF.
This also makes gtFoldExprConst consistent with ValueNumStore::EvalFuncForConstantArgs which evaluates TYP_INT nodes as int and then casts to ssize_t.
Ian Hays [Fri, 24 Feb 2017 19:46:25 +0000 (11:46 -0800)]
Update FileLock/Unlock PNSE message for OSX
Justin Van Patten [Fri, 24 Feb 2017 19:08:08 +0000 (11:08 -0800)]
Avoid unnecessary branch in CultureAwareComparer (#9767)
* Avoid unnecessary branch in CultureAwareComparer
Determine the CompareOptions up-front instead of during each operation.
William Godbe [Fri, 24 Feb 2017 18:54:40 +0000 (10:54 -0800)]
Merge pull request #9749 from wtgodbe/XunitCopyStuff
Use -f instead of -n when copying from Core_Root in runtest.sh
Ian Hays [Thu, 23 Feb 2017 22:46:15 +0000 (14:46 -0800)]
Remove OSX FileStream Lock/Unlock
OSX doesn't support usage of both fcntl and flock. Since we're already using one in FileShare for the entire file, we cannot enable partial file locking like we do on other Unix platforms. The alternative is to throw a PNSE and suggest using FileShare on the whole file instead.
Stephen Toub [Fri, 24 Feb 2017 18:35:48 +0000 (13:35 -0500)]
Merge pull request #9766 from discostu105/patch-1
fix typo in sos_md.h
Sean Gillespie [Fri, 24 Feb 2017 18:12:26 +0000 (10:12 -0800)]
[Local GC] Move workstation GC DAC globals to a struct shared between GC and DAC (#9255)
* [Local GC] Move workstation GC DAC globals to a struct shared between
the GC and the DAC
* (Some) code review feedback and bug fixes for issues found while debugging on OSX
* Address some code review feedback:
1. Make g_gcDacGlobals a pointer and dacvar on the VM side, so
that publishing the GC dac vars is done atomically (through
a pointer assignment). This fixes a race that Noah noticed.
2. Remove the requirement for the GC's generation class struct
to be known at compile-time, by using a dacvar as the size
of the generation class at run-time (for pointer arithmetic)
3. Move all DAC-interesting fields to be at the start of GC
internal classes, so that the DAC does not need to know the
size or exact layout of the class past the fields it cares
about.
* Split the definition of the size of several arrays across the SOS/DAC and GC/DAC interfaces, and add static asserts that they are the same
* Repair the Windows Release build
* Implement the GC DAC scheme for Server GC and eliminate the duplicate GC dac vars
* Some work
* Decouple use of the GC generation table from a write barrier by having the EE store a copy of the global during initialization
* Actually make it work with server GC
* Checkpoint
* Checkpoint where everything works
* Code cleanup
* Fix debugger test failures
* Additional code cleanup
* Address code review feedback by adding a static assert and standardizing the way that we iterate over the generation table
* Repair the Windows x86 build
* Revert "Decouple use of the GC generation table from a write barrier by having the EE store a copy of the global during initialization"
This reverts commit
573f61a16b4fa8c2fc4c568c0b968a921230f31c.
* Revert "Repair the Windows x86 build"
This reverts commit
188c22d87e1d65abf00ab8fa28f46ad607a9028f.
* Partial revert, move `generation_table` back the global namespace for a single-proc allocation helper
* Fix a debugger test failure
* Repair crash dump scenarios
Christoph Neumüller [Fri, 24 Feb 2017 14:52:11 +0000 (15:52 +0100)]
fix typo in sos_md.h
Jonghyun Park [Fri, 24 Feb 2017 14:48:58 +0000 (23:48 +0900)]
[Linux/ARM,ARM64] Fix undefined FixContextHandler issue (#9755)
Sujin Kim [Fri, 24 Feb 2017 14:43:15 +0000 (23:43 +0900)]
Fix segmentation fault when using ZapImage::PrintStats() (#9731)
Currently, ZapImage::PrintStats is not used anywhere. But I want to get detailed informations about Native Image(ni.dll) components.
So I just used ZapImage::PrintStats when called ZapImage::SaveImage. After that I met the segmentation fault likes #9680.
I found the reason for this issue is accessing the null pointer to get sizes.
John Gorter [Fri, 24 Feb 2017 14:41:42 +0000 (15:41 +0100)]
typo in CMakeLists.txt (#9763)
Stephen Toub [Fri, 24 Feb 2017 12:43:36 +0000 (07:43 -0500)]
Merge pull request #9738 from stephentoub/ordinalignorecase_hashing
Tweak OrdinalComparer for OrdinalIgnoreCase
Hanjoung Lee [Fri, 24 Feb 2017 07:02:21 +0000 (16:02 +0900)]
[x86/Linux] Initial patch for EH funclet (#9601)
- Generate a simple EH funclet frame and support SP-based stack unwinding for funclets.
- Introduce assembly helpers : CallEHFunclet and CallEHFilterFunclet
Bruce Forstall [Fri, 24 Feb 2017 06:34:58 +0000 (22:34 -0800)]
Merge pull request #9758 from seanshpark/fixx86break
Fix build break for !FEATURE_FIXED_OUT_ARGS
Sivarv [Fri, 24 Feb 2017 04:35:39 +0000 (20:35 -0800)]
Merge pull request #9723 from fiigii/is4byte
Simplify and streamline Is4ByteAVXInstruction() and Is4ByteSSE4Instruction() functions
Stephen Toub [Thu, 23 Feb 2017 18:15:27 +0000 (13:15 -0500)]
Address PR feedback
Stephen Toub [Thu, 23 Feb 2017 16:25:03 +0000 (11:25 -0500)]
Improve CompareOrdinalIgnoreCaseHelper by removing return from loop
Stephen Toub [Thu, 23 Feb 2017 16:37:23 +0000 (11:37 -0500)]
Strongly type StringComparer static comparer fields
Stephen Toub [Thu, 23 Feb 2017 14:28:30 +0000 (09:28 -0500)]
Split OrdinalIgnoreCaseComparer from OrdinalComparer
They're both used very frequently, and we can eliminate the branches on checking _ignoreCase from every operation.
Jan Kotas [Fri, 24 Feb 2017 02:51:03 +0000 (18:51 -0800)]
Reject byref-like object in FormatterServices.GetUninitializedObject (#9754)
Fixes #9739
Jonghyun Park [Fri, 24 Feb 2017 02:01:10 +0000 (11:01 +0900)]
[x86/Linux] Enable FEATURE_ARRAYSTUB_AS_IL (#9752)
* [x86/Linux] Enable FEATURE_ARRAYSTUB_AS_IL
* Move FeatureStubsAsIL into Windows-specific block
SaeHie Park [Fri, 24 Feb 2017 01:03:26 +0000 (10:03 +0900)]
Fix build break for !FEATURE_FIXED_OUT_ARGS
Bruce Forstall [Fri, 24 Feb 2017 01:03:58 +0000 (17:03 -0800)]
Merge pull request #9681 from mskvortsov/ryujit-arm32-reload
[RyuJIT/ARM32] Fix helper kill mask and call target consuming
Stephen Toub [Fri, 24 Feb 2017 00:43:58 +0000 (19:43 -0500)]
Use GetSystemTimePreciseAsFileTime if available in DateTime.UtcNow (#9736)
Tanner Gooding [Fri, 24 Feb 2017 00:36:20 +0000 (16:36 -0800)]
Merge pull request #9717 from tannergooding/math-intrinsics2
Updating codegenxarch to support single-precision CORINFO_INTRINSIC_Sqrt
Andy Ayers [Sat, 18 Feb 2017 04:51:59 +0000 (20:51 -0800)]
JIT: keep ref count for runtime generic context lookups
Keep a ref count for runtime generic context lookups, and track when
an inlinee requires a runtime lookup but then ignores the resulting
context. Only report/keep the context alive if there are known uses.
This pattern happens frequently in inlined generic code that uses
type tests to extract properties of types.
Addresses #1691.
Sivarv [Fri, 24 Feb 2017 00:08:41 +0000 (16:08 -0800)]
Merge pull request #9742 from sivarv/singleField
Enable promotion of structs containing fields of structs with a single pointer-sized scalar type field.
Andy Ayers [Fri, 24 Feb 2017 00:08:23 +0000 (16:08 -0800)]
Merge pull request #9748 from AndyAyersMS/FixJitTestDir
Move jit tests to directory with proper casing
wtgodbe [Thu, 23 Feb 2017 22:29:41 +0000 (14:29 -0800)]
Use -f instead of -n when copying from Core_Root in runtest.sh
Alex Perovich [Thu, 23 Feb 2017 23:20:04 +0000 (17:20 -0600)]
Merge pull request #9726 from alexperovich/codeFormat
Run dotnet/codeformatter on S.P.CoreLib sources.
Andy Ayers [Thu, 23 Feb 2017 22:53:00 +0000 (14:53 -0800)]
Add jit test project checker as jit test (#9721)
Add the tool used to fix/repair projects as a test case. It will
fail with an error code if any project file in the jit test subtree
needs fixing.
This test requires CORE_ROOT to point at the test overlay and
that directory be rooted under the repo (other test cases already
have similar requirements).
CodeGenBringUp tests are noted as needing fixes, but are accounted
for under the category of "deferred fixes".
Andy Ayers [Thu, 23 Feb 2017 22:52:10 +0000 (14:52 -0800)]
JIT: defer setting outgoing args size until after optimization (#9683)
For fixed out args ABIs, the jit currently computes the size of the
out arg area during morph, as it encounters calls that must pass arguments
via the stack.
Subsequently, the optimizer may delete calls, either because they are
pure and the results are unused, or because they are conditional and the
guarding predicates are resolved by the optimizer in such a way that
the call is no longer reachable.
In particular if all the calls seen by morph are subsequently removed by
the optimizer, the jit may generate a stack frame that it never uses and
doesn't need. If only some calls are removed then the stack frame may end
up larger than necessary.
One motivating example is the inlining of a shared generic method that
ignores its generic context parameter. The caller typically must invoke a pure
helper to determine the proper argument to pass. Post inline the call's result
is unused, and the helper call is deleted by the optimizer.
This change defers the outgoing arg size computation until fgSimpleLowering,
which runs after optimization. The code in morph now simply records the
needed size for each call in the associated fgArgInfo.
As before, the outgoing arg size computation ignores fast tail calls, since
they can use the caller-supplied scratch area for memory arguments.
The jit may introduce helper calls after optimization which could seemingly
invalidate the out args area size. The presumption here is that these calls
are carefully designed not to use the caller-supplied scratch area. The
current code makes the same assumption.
This change introduces some unanticipated diffs. Optcse estimates the
frame size to decide how aggressive it should be about cses of local
variable references. This size estimate included the out args area, which
appears to be something of an oversight; there are no cse opportunities
in this area and the offsets of the other local variables are not impacted
by the size of this area. With this change the out args area size is seen
as zero during optimization and so the cse strategy for a method may be
different.
Pat Gavlin [Thu, 23 Feb 2017 22:20:45 +0000 (14:20 -0800)]
Merge pull request #9582 from pgavlin/vzeroupper_x86_epilog
Do not report FP restores in x86 epilogs.
Brian Sullivan [Thu, 23 Feb 2017 21:56:09 +0000 (13:56 -0800)]
Merge pull request #9454 from mikedn/cgtun-to-cne
Morph (x GT_GT.unsigned 0) into (x GT_NE 0)
Andy Ayers [Thu, 23 Feb 2017 21:46:20 +0000 (13:46 -0800)]
Move jit tests to directory with proper casing
Closes #9747.
Alex Perovich [Wed, 22 Feb 2017 23:23:44 +0000 (15:23 -0800)]
Run codeformatter on S.P.CoreLib sources
Alex Perovich [Wed, 22 Feb 2017 23:23:25 +0000 (15:23 -0800)]
Manual change to make codeformatter work