platform/upstream/coreclr.git
7 years agoMerge pull request #11085 from wtgodbe/SkipTests
William Godbe [Wed, 19 Apr 2017 22:05:26 +0000 (15:05 -0700)]
Merge pull request #11085 from wtgodbe/SkipTests

Skip failing tests in Helix

7 years agoSkip failing tests in Helix
wtgodbe [Wed, 19 Apr 2017 22:04:46 +0000 (15:04 -0700)]
Skip failing tests in Helix

7 years agoUpdate CheckProjects to not fail for non-dev CORE_ROOT
Andy Ayers [Wed, 19 Apr 2017 21:07:24 +0000 (14:07 -0700)]
Update CheckProjects to not fail for non-dev CORE_ROOT

CheckProjects expects to be able to find the jit test project files
relative to the value of the `CORE_ROOT` environment variable. This
fails when we have test setups like cross-tests or helix where the
test binary tree has been copied away from the repo after building.

Tolerate these cases by returning a success code if the `CORE_ROOT`
path is not pointing at the expected place.

Fixes #11079.
Fixes #10635.

7 years agoConvert some noway_assert to static_assert
Mike Danes [Wed, 19 Apr 2017 21:06:40 +0000 (00:06 +0300)]
Convert some noway_assert to static_assert

7 years ago[x86/Linux] Corretly pop args after unmanaged function call (#11007)
Jonghyun Park [Wed, 19 Apr 2017 20:50:56 +0000 (05:50 +0900)]
[x86/Linux] Corretly pop args after unmanaged function call (#11007)

* [x86/Linux] Corretly pop args after unmanaged function call

* Fix x86/Windows build error

7 years agoAdd a regression test.
Pat Gavlin [Wed, 19 Apr 2017 20:38:05 +0000 (13:38 -0700)]
Add a regression test.

7 years agoMerge pull request #11049 from briansull/fixMaxOffset
Brian Sullivan [Wed, 19 Apr 2017 19:33:20 +0000 (12:33 -0700)]
Merge pull request #11049 from briansull/fixMaxOffset

Fix incorrect test of structSize during struct promotion

7 years agoAdd test coverage for Rank 1 multidimensional arrays (#11034)
Michal Strehovský [Wed, 19 Apr 2017 19:01:57 +0000 (12:01 -0700)]
Add test coverage for Rank 1 multidimensional arrays (#11034)

These have some weird behaviors worth covering.

7 years agoIBC: Disable `-partialNGEN -minify` for SPC.dll (#10950)
Daniel Podder [Wed, 19 Apr 2017 18:59:58 +0000 (13:59 -0500)]
IBC: Disable `-partialNGEN -minify` for SPC.dll (#10950)

Add some fixes around the use of IBCMerge in CoreCLR, along with a BuildTools patch (which the fixes are dependent upon):
* don't restore the default optdata package through BuildTools
* don't use partial ngen by default in CoreCLR for SPC.dll

7 years agoSpill side-effects in impAssignMultiRegTypeToVar.
Pat Gavlin [Wed, 19 Apr 2017 18:26:46 +0000 (11:26 -0700)]
Spill side-effects in impAssignMultiRegTypeToVar.

This importer function converts IR from the form

    tree

to
            /- tree
    STMT - =
            \- lclVar

in order to conform to the JIT's representation of multi-register values.
Today, it does not check whether or not the tree being spilled has any
side-effects that may interfere with the side-effects of trees that are
already on the evaluation stack, which can lead to bad codegen. This
change fixes this function to spill any side-effects on the stack
before generating the temp assign.

Fixes #10940.

7 years agoMerge pull request #11044 from alpencolt/ryu-arm-localloc-subs-fix
Bruce Forstall [Wed, 19 Apr 2017 18:25:35 +0000 (11:25 -0700)]
Merge pull request #11044 from alpencolt/ryu-arm-localloc-subs-fix

[RyuJIT/ARM32] Use 'subs' instead of 'sub' on localloc

7 years agoMerge pull request #11070 from stephentoub/add_stat_fields
Stephen Toub [Wed, 19 Apr 2017 17:44:12 +0000 (13:44 -0400)]
Merge pull request #11070 from stephentoub/add_stat_fields

Add Dev/Ino fields to Interop.Sys.FileStatus

7 years agoFor Reflection invoke, use ArgLocDesc
jashook [Wed, 19 Apr 2017 17:03:13 +0000 (10:03 -0700)]
For Reflection invoke, use ArgLocDesc

For arm64 if the struct needs to be passed in registers, then pass
that information along when we do the reflection invoke.

Note this adds more test cases for passing different struct types via reflection.
I am guessing similar work would need to be done for Arm32.

7 years agoChange ICLRRuntimeHost2 to ICLRRuntimeHost4 for Windows corerun. (#10967)
Jim Ma [Wed, 19 Apr 2017 16:39:23 +0000 (16:39 +0000)]
Change ICLRRuntimeHost2 to ICLRRuntimeHost4 for Windows corerun. (#10967)

* Change ICLRRuntimeHost2 to ICLRRuntimeHost4 for Windows corerun.

Since we've introduced ICLRRuntimeHost4 in PR #10842 for Unix, we apply it for Windows corerun as well.

Fix #10960

7 years agoStop running all tests on first failure, if the run is private or rolling.
José Rivero [Wed, 19 Apr 2017 15:49:19 +0000 (08:49 -0700)]
Stop running all tests on first failure, if the run is private or rolling.

7 years agoStop treating test files as executable and fix encoding (#11065)
Yi Zhang (CLR) [Wed, 19 Apr 2017 15:37:44 +0000 (08:37 -0700)]
Stop treating test files as executable and fix encoding (#11065)

* change encoding to UTF-8

* change line ending to UNIX

* remove executable attribute

7 years agoAdd Dev/Ino fields to Interop.Sys.FileStatus
Stephen Toub [Wed, 19 Apr 2017 15:20:42 +0000 (11:20 -0400)]
Add Dev/Ino fields to Interop.Sys.FileStatus

7 years agoMac OS X is now macOS (README.md) (#11068)
Roger [Wed, 19 Apr 2017 15:08:00 +0000 (11:08 -0400)]
Mac OS X is now macOS (README.md) (#11068)

7 years ago[x86/Linux] Do NOT run GC on Transition (#11056)
Jonghyun Park [Wed, 19 Apr 2017 08:55:18 +0000 (17:55 +0900)]
[x86/Linux] Do NOT run GC on Transition (#11056)

7 years agoRevert "Add SecurityZone stub" (#11053)
Viktor Hofer [Wed, 19 Apr 2017 02:36:14 +0000 (04:36 +0200)]
Revert "Add SecurityZone stub" (#11053)

7 years agoUpdating the measurement.py call to be setup like the submission.py call
José Rivero [Wed, 19 Apr 2017 01:31:16 +0000 (18:31 -0700)]
Updating the measurement.py call to be setup like the submission.py call

7 years agoSet default BenchView run type to local
José Rivero [Wed, 19 Apr 2017 00:59:00 +0000 (17:59 -0700)]
Set default BenchView run type to local
- This is the case when running on the dev boxes (outside automation)

7 years agoMerge pull request #10933 from hqueue/ryujit/enable_nodebash_stats
Bruce Forstall [Wed, 19 Apr 2017 00:14:13 +0000 (17:14 -0700)]
Merge pull request #10933 from hqueue/ryujit/enable_nodebash_stats

Fix clang build error when enabling NODEBASH_STATS

7 years agoMerge pull request #11021 from mskvortsov/array-offset-mla
Bruce Forstall [Wed, 19 Apr 2017 00:13:16 +0000 (17:13 -0700)]
Merge pull request #11021 from mskvortsov/array-offset-mla

[RyuJIT/ARM32] Use mla for array offset computation

7 years agoSome fixes for the GC Reliability Framework (#11029)
Sean Gillespie [Wed, 19 Apr 2017 00:12:09 +0000 (17:12 -0700)]
Some fixes for the GC Reliability Framework (#11029)

* Quality-of-life fixes for the GC reliability framework to make it run
a little better on CoreCLR

* Remove some p/invokes to kernel32 when building on Core

* Minor fixes to run on non-Windows platforms

* Re-enable ReliabilityFramework sanity run

7 years agoImproving robustness of the run-xunit-perf.cmd script.
José Rivero [Tue, 18 Apr 2017 23:51:13 +0000 (16:51 -0700)]
Improving robustness of the run-xunit-perf.cmd script.
- Added error checking to the script and make it fail fast
- Updated the xUnit Performance API version (build0002 -> build0003)

7 years agoImplement the createdump policy options and env variables (#11032)
Mike McLaughlin [Tue, 18 Apr 2017 23:54:53 +0000 (16:54 -0700)]
Implement the createdump policy options and env variables (#11032)

Added these command line options:

-f, --name - dump path and file name. The pid can be placed in the name with %d. The default is "/tmp/coredump.%d"
-n, --normal - create minidump (default).
-h, --withheap - create minidump with heap.
-m, --micro - create triage minidump.
-d, --diag - enable diagnostic messages.

Added these environment variables:

COMPlus_DbgMiniDumpType - if set to "1" generate MiniDumpNormal, "2" MiniDumpWithPrivateReadWriteMemory, "3" MiniDumpFilterTriage. Default is MiniDumpNormal.
COMPlus_DbgMiniDumpName - if set, use as the template to create the dump path and file name. The pid can be placed in the name with %d. The default is "/tmp/coredump.%d".

7 years agoMerge pull request #11054 from wtgodbe/spacing
William Godbe [Tue, 18 Apr 2017 23:43:11 +0000 (16:43 -0700)]
Merge pull request #11054 from wtgodbe/spacing

Fix spacing in build-test.cmd in Pipeline

7 years agoFix spacing in build-test.cmd in Pipeline
wtgodbe [Tue, 18 Apr 2017 23:42:00 +0000 (16:42 -0700)]
Fix spacing in build-test.cmd in Pipeline

7 years agoMerge pull request #11050 from ellismg/remove-duplicate-resources
Matt Ellis [Tue, 18 Apr 2017 23:36:55 +0000 (16:36 -0700)]
Merge pull request #11050 from ellismg/remove-duplicate-resources

Remove some duplicate resource strings

7 years agoMerge pull request #11052 from wtgodbe/emptyArgs
William Godbe [Tue, 18 Apr 2017 22:37:22 +0000 (15:37 -0700)]
Merge pull request #11052 from wtgodbe/emptyArgs

Create default variables for Crossgen, TargetsNonWindows in pipebuild

7 years agoCreate default variables for Crossgen, TargetsNonWindows in pipebuild
wtgodbe [Tue, 18 Apr 2017 22:36:14 +0000 (15:36 -0700)]
Create default variables for Crossgen, TargetsNonWindows in pipebuild

7 years agoConvert a few test files from UTF16 to ASCII and add license (#11051)
Jan Kotas [Tue, 18 Apr 2017 22:29:43 +0000 (17:29 -0500)]
Convert a few test files from UTF16 to ASCII and add license (#11051)

7 years agoMerge pull request #10979 from wtgodbe/R2RWindows
William Godbe [Tue, 18 Apr 2017 22:02:10 +0000 (15:02 -0700)]
Merge pull request #10979 from wtgodbe/R2RWindows

Enable running Crossgen tests on Windows in Helix

7 years agoRemove /RELEASE option required for device drivers (#11023)
Gaurav Khanna [Tue, 18 Apr 2017 21:02:14 +0000 (14:02 -0700)]
Remove /RELEASE option required for device drivers (#11023)

7 years agoMerge pull request #11047 from stephentoub/task_iscompletedsuccessfully
Stephen Toub [Tue, 18 Apr 2017 20:43:53 +0000 (16:43 -0400)]
Merge pull request #11047 from stephentoub/task_iscompletedsuccessfully

Expose Task.IsCompletedSuccessfully

7 years agoRemove some duplicate resource strings
Matt Ellis [Tue, 18 Apr 2017 20:26:01 +0000 (13:26 -0700)]
Remove some duplicate resource strings

7 years agoEnable running Crossgen tests on Windows in Helix
wtgodbe [Fri, 14 Apr 2017 21:31:34 +0000 (14:31 -0700)]
Enable running Crossgen tests on Windows in Helix

7 years agoMerge pull request #10948 from eerhardt/GetPreviousAdjRulePerf
Eric Erhardt [Tue, 18 Apr 2017 19:07:13 +0000 (14:07 -0500)]
Merge pull request #10948 from eerhardt/GetPreviousAdjRulePerf

Fix perf of DateTime.Now on Unix

7 years agoMerge pull request #11045 from JosephTremoulet/TraceEventVersion
Joseph Tremoulet [Tue, 18 Apr 2017 18:43:19 +0000 (14:43 -0400)]
Merge pull request #11045 from JosephTremoulet/TraceEventVersion

Run-xunit-perf: update TraceEvents version number

7 years agofail fast on exceptions that indicate corrupted process state (#10957)
Rahul Kumar [Tue, 18 Apr 2017 18:33:59 +0000 (11:33 -0700)]
fail fast on exceptions that indicate corrupted process state (#10957)

7 years agoFix incorrect test of structSize in struct promoted
Brian Sullivan [Tue, 18 Apr 2017 18:32:21 +0000 (11:32 -0700)]
Fix incorrect test of structSize in struct promoted

Fixes the desktop test Crossgen NI and Ngen NI validation

7 years ago[x86/Linux] EH Support for Per-Frame P/Invoke Init (#10966)
Jonghyun Park [Tue, 18 Apr 2017 18:27:29 +0000 (03:27 +0900)]
[x86/Linux] EH Support for Per-Frame P/Invoke Init (#10966)

7 years agoExpose Task.IsCompletedSuccessfully
Stephen Toub [Tue, 18 Apr 2017 17:57:19 +0000 (13:57 -0400)]
Expose Task.IsCompletedSuccessfully

7 years agosmall clean-up and bug fix in genIntToIntCast (#10989)
Sergey Andreenko [Tue, 18 Apr 2017 17:44:01 +0000 (10:44 -0700)]
small clean-up and bug fix in genIntToIntCast (#10989)

genintToIntCast was cleaned-up, the assert that was wrong for x86 was deleted. The repro was added.

7 years agoRun-xunit-perf: update TraceEvents version number
Joseph Tremoulet [Tue, 18 Apr 2017 16:47:25 +0000 (12:47 -0400)]
Run-xunit-perf: update TraceEvents version number

The perf harness needs 1.0.3-alpha-experimental now.

7 years agoAssert GetAdjustmentRuleForTime sets ruleIndex correctly.
Eric Erhardt [Tue, 18 Apr 2017 16:39:01 +0000 (11:39 -0500)]
Assert GetAdjustmentRuleForTime sets ruleIndex correctly.

7 years ago[RyuJIT/ARM32] Use 'subs' instead of 'sub' on localloc
Alexander Soldatov [Tue, 18 Apr 2017 14:51:20 +0000 (17:51 +0300)]
[RyuJIT/ARM32] Use 'subs' instead of 'sub' on localloc

Previous implementation used sub instruction for counter
conditional execution for iterating over loops which cause incorrect
behavior.

7 years agoMerge pull request #11035 from BruceForstall/NumSuccImprovement
Bruce Forstall [Tue, 18 Apr 2017 14:18:10 +0000 (07:18 -0700)]
Merge pull request #11035 from BruceForstall/NumSuccImprovement

Hoist calls to NumSucc() out of loops

7 years agoRyuJIT/ARM32: use mla for array offset computation
Mikhail Skvortcov [Tue, 18 Apr 2017 13:02:17 +0000 (16:02 +0300)]
RyuJIT/ARM32: use mla for array offset computation

7 years ago[x86/Linux] changed to the AT&T syntax to solve a problem related broken stack. ...
ragmani [Tue, 18 Apr 2017 10:01:55 +0000 (19:01 +0900)]
[x86/Linux] changed to the AT&T syntax to solve a problem related broken stack. (#10932)

This is a problem similar to #10383.

Signed-off-by: ragmani <ragmani0216@gmail.com>
7 years ago* fixed class size calculation (#10916)
Buyduck [Tue, 18 Apr 2017 08:02:14 +0000 (11:02 +0300)]
* fixed class size calculation (#10916)

7 years agoFix String.Replace implementation (#11039)
Tarek Mahmoud Sayed [Tue, 18 Apr 2017 05:02:08 +0000 (22:02 -0700)]
Fix String.Replace implementation (#11039)

* Fix String.Replace implementation

String.Replace was callin ga low level CompareInfo method IndexOfCore which cannot be called in some conditions like when using ordinal ignore case option or when running in invariant mode.
The fix is to call a higher level method in CompareInfo which can handle such cases becfore calling the lower level method

* fix typo

7 years agoMerge pull request #10995 from svick/patch-1
Andy Ayers [Tue, 18 Apr 2017 04:47:38 +0000 (21:47 -0700)]
Merge pull request #10995 from svick/patch-1

Update Viewing JIT dumps to 2.0 and csproj

7 years agoFix error caused by passing -TargetsWindows at end of Command Line (#11031)
William Godbe [Tue, 18 Apr 2017 03:16:22 +0000 (20:16 -0700)]
Fix error caused by passing -TargetsWindows at end of Command Line (#11031)

7 years agoDo not throw an assertion on no mscorlib
Michael Letterle [Fri, 31 Mar 2017 15:32:32 +0000 (11:32 -0400)]
Do not throw an assertion on no mscorlib

It's okay to use debuggable attributes from other BaseAsmRef's

Fixes #10608

7 years agoMerge pull request #11036 from brianrob/remove_eventcounter
Brian Robbins [Tue, 18 Apr 2017 01:35:37 +0000 (18:35 -0700)]
Merge pull request #11036 from brianrob/remove_eventcounter

Remove EventCounter from System.Private.CoreLib

7 years agoFix build break
danmosemsft [Mon, 17 Apr 2017 23:58:10 +0000 (16:58 -0700)]
Fix build break

7 years agoRename attribute to IsReadOnlyAttribute (#11026)
Omar Tawfik [Mon, 17 Apr 2017 23:33:09 +0000 (16:33 -0700)]
Rename attribute to IsReadOnlyAttribute (#11026)

7 years agoUse isInternalRegDelayFree to avoid temp reg overallocation (#11025)
Bruce Forstall [Mon, 17 Apr 2017 23:32:36 +0000 (16:32 -0700)]
Use isInternalRegDelayFree to avoid temp reg overallocation (#11025)

Use isInternalRegDelayFree to avoid temp reg overallocation

7 years agoRemove EventCounter from System.Private.CoreLib.
Brian Robbins [Mon, 17 Apr 2017 23:16:22 +0000 (16:16 -0700)]
Remove EventCounter from System.Private.CoreLib.

7 years agoHoist calls to NumSucc() out of loops
Bruce Forstall [Mon, 17 Apr 2017 23:00:51 +0000 (16:00 -0700)]
Hoist calls to NumSucc() out of loops

Should be a minor throughput improvement.

7 years agoRemove handle assignment validation from GC side.
Aditya Mandaleeka [Mon, 17 Apr 2017 22:53:38 +0000 (15:53 -0700)]
Remove handle assignment validation from GC side.

7 years agoAdd handle assignment validation to VM side.
Aditya Mandaleeka [Mon, 17 Apr 2017 22:51:58 +0000 (15:51 -0700)]
Add handle assignment validation to VM side.

7 years agoImprove performance of Object.GetType for arrays (#10992)
Jan Kotas [Mon, 17 Apr 2017 22:37:09 +0000 (17:37 -0500)]
Improve performance of Object.GetType for arrays (#10992)

* Improve performance of Object.GetType for arrays

Cache managed Type object on MethodTable for arrays. Makes Object.GetType for arrays about 5x faster.

7 years agoMerge pull request #11028 from danmosemsft/api.zone
Viktor Hofer [Mon, 17 Apr 2017 21:39:21 +0000 (23:39 +0200)]
Merge pull request #11028 from danmosemsft/api.zone

Add SecurityZone stub

7 years agoMove handle manipulation functions to interface.
Aditya Mandaleeka [Wed, 12 Apr 2017 02:11:32 +0000 (19:11 -0700)]
Move handle manipulation functions to interface.

7 years agoFormat spmi sources with clang-format (#11027)
Sergey Andreenko [Mon, 17 Apr 2017 21:05:11 +0000 (14:05 -0700)]
Format spmi sources with clang-format (#11027)

Apply jit style format on spmi sources.

7 years agoJit intrinsics for Span<T>.get_Item and ReadOnlySpan<T>.get_Item. (#10910)
Andy Ayers [Mon, 17 Apr 2017 20:04:23 +0000 (13:04 -0700)]
Jit intrinsics for Span<T>.get_Item and ReadOnlySpan<T>.get_Item. (#10910)

Implement these two methods as optional-expand jit intrinsics.
Uses `GT_ARR_BOUNDS_CHECK` for the bounds check so in some cases
downstream code is able to eliminate redundant checks. Fully general
support (on par with arrays in most cases) is still work in progress.

Update one bit of code in the optimizer that assumed it knew the
tree types that appeared in a `GT_ARR_BOUNDS_CHECK`.

Add benchmark tests for Span and ReadOnlySpan indexers.

Tests ability of jit to reason about indexer properties with respect
to loop bounds and related indexer uses. Some cases inspired by span
indexer usage in Kestrel.

Closes #10785.

Also addresses lack of indexer inlining noted in #10031. Span indexers
should now always be inlined, even when invoked from shared methods.

7 years agoAdd scenario-based tests
Swaroop Sridhar [Fri, 14 Apr 2017 03:07:41 +0000 (20:07 -0700)]
Add scenario-based tests

Modified Xunit runner to run Scenario Benchmarks
Updating xunit-performance packages

7 years agoAdd SecurityZone stub
danmosemsft [Mon, 17 Apr 2017 19:15:07 +0000 (12:15 -0700)]
Add SecurityZone stub

7 years agoIn function Assembly.Load ignore AssemblyName.CodeBase even if it is set (#11010)
Rahul Kumar [Mon, 17 Apr 2017 18:47:47 +0000 (11:47 -0700)]
In function Assembly.Load ignore AssemblyName.CodeBase even if it is set (#11010)

7 years agoRemove Alpine for pipe builds (#11022)
Gaurav Khanna [Mon, 17 Apr 2017 18:29:35 +0000 (11:29 -0700)]
Remove Alpine for pipe builds (#11022)

7 years agoSupport FEATURE_EVENT_TRACE override (#10961)
Jonghyun Park [Mon, 17 Apr 2017 18:21:45 +0000 (03:21 +0900)]
Support FEATURE_EVENT_TRACE override (#10961)

7 years agoAllow build.sh to have multiple cmakeargs options (#10962)
Jonghyun Park [Mon, 17 Apr 2017 18:20:21 +0000 (03:20 +0900)]
Allow build.sh to have multiple cmakeargs options (#10962)

7 years agoUpdate CoreClr, CoreFx to preview1-25215-03, preview1-25214-03, respectively (#10952)
dotnet bot [Mon, 17 Apr 2017 18:17:42 +0000 (11:17 -0700)]
Update CoreClr, CoreFx to preview1-25215-03, preview1-25214-03, respectively (#10952)

7 years ago[RyuJit/ARM32] Fix assertion of failure on "Instruction cannot be encoded" (#10789)
Sujin Kim [Mon, 17 Apr 2017 17:18:39 +0000 (02:18 +0900)]
[RyuJit/ARM32] Fix assertion of failure on "Instruction cannot be encoded" (#10789)

Fix assertion of failure on "Instruction cannot be encoded"

I figured out the bit set mask for '0xffffff00' which is the imm is not supportted on ARM.
So at first, It would be checked the validation of using instruction with immediate constants.
If not, it would be use the instruction with registers.

7 years agoMerge pull request #11013 from mskvortsov/nullcheck
Bruce Forstall [Mon, 17 Apr 2017 17:14:13 +0000 (10:14 -0700)]
Merge pull request #11013 from mskvortsov/nullcheck

[RyuJIT/ARM32] Update nullcheck codegen

7 years agoMerge pull request #11016 from alpencolt/ryu-arm-localloc-fix
Bruce Forstall [Mon, 17 Apr 2017 16:53:15 +0000 (09:53 -0700)]
Merge pull request #11016 from alpencolt/ryu-arm-localloc-fix

[RyuJIT/ARM32] Fix amount of allocated registers for Localloc.

7 years agoFix ARM32/Linux release CI test failure (#11008)
Hyeongseok Oh [Mon, 17 Apr 2017 15:56:19 +0000 (00:56 +0900)]
Fix ARM32/Linux release CI test failure (#11008)

* Fix ARM32/Linux release CI test failure

Fix ARM32/Linux release CI test failure by removing libcoreclrptprovider.so from CoreFX build results.

* Fix typo

Fix typo in netci.groovy

7 years agoAdd Span and ReadOnlySpan from CoreCLR
Jan Kotas [Tue, 14 Mar 2017 16:06:56 +0000 (09:06 -0700)]
Add Span and ReadOnlySpan from CoreCLR

Signed-off-by: dotnet-bot <dotnet-bot@microsoft.com>
7 years ago[RyuJIT/ARM32] Fix amount of allocated registers for Localloc.
Alexander Soldatov [Mon, 17 Apr 2017 13:10:36 +0000 (16:10 +0300)]
[RyuJIT/ARM32] Fix amount of allocated registers for Localloc.

Now for small allocation (up to 4 store instructions) we always have
1 temporary register which is different from register where result
of stackalloc is returned. Also one additional register will be allocated
for PSP if needed.

7 years agoRyuJIT/ARM32: update nullcheck codegen
Mikhail Skvortcov [Mon, 17 Apr 2017 09:48:09 +0000 (12:48 +0300)]
RyuJIT/ARM32: update nullcheck codegen

7 years agoFix comment about pinning bit in object.h (#11004)
303248153 [Mon, 17 Apr 2017 03:01:52 +0000 (11:01 +0800)]
Fix comment about pinning bit in object.h (#11004)

7 years agoFix warning in LPTSTRTestNative (#10973)
Yi Zhang (CLR) [Mon, 17 Apr 2017 02:21:55 +0000 (19:21 -0700)]
Fix warning in LPTSTRTestNative (#10973)

* Fix warning in LPTSTRTestNative and add support for wcsncmp in xplatform.h

7 years agoFix OOM exception thrown in case insensitive replace for an empty string (#11001)
Hugh Bellamy [Sun, 16 Apr 2017 18:36:47 +0000 (01:36 +0700)]
Fix OOM exception thrown in case insensitive replace for an empty string (#11001)

7 years agoFixed CI link (#11003)
Cyd Haselton [Sun, 16 Apr 2017 18:36:17 +0000 (13:36 -0500)]
Fixed CI link (#11003)

Updated link referenced in CI builds; old link was broken

7 years agoMove Span/ReadOnlySpan to shared CoreLib partition (#10988)
Jan Kotas [Sat, 15 Apr 2017 20:32:58 +0000 (13:32 -0700)]
Move Span/ReadOnlySpan to shared CoreLib partition (#10988)

Fix a few method names to better names used in CoreRT

Contributes to https://github.com/dotnet/corert/issues/2966

7 years agoMake some Dictionary code smaller (#10993)
mikedn [Sat, 15 Apr 2017 19:11:51 +0000 (22:11 +0300)]
Make some Dictionary code smaller (#10993)

Store a reference to the relevant entry at the start of the loop to avoid indexing `entries` multiple times.
This avoids some redundant range checks in `Remove` and allows the elimination of a duplicate `index++` in `Enumerator.MoveNext`.

Saves ~7KB of code.

7 years agoMerge pull request #10983 from danmosemsft/revert.jeremy
Michelle McDaniel [Sat, 15 Apr 2017 18:40:09 +0000 (11:40 -0700)]
Merge pull request #10983 from danmosemsft/revert.jeremy

Revert "Strip out unused reg code. (#10741)"

7 years agoUpdate Viewing JIT dumps to 2.0 and csproj
Petr Onderka [Sat, 15 Apr 2017 17:22:00 +0000 (19:22 +0200)]
Update Viewing JIT dumps to 2.0 and csproj

7 years agoMake createdump build and work on Centos (#10985)
Mike McLaughlin [Sat, 15 Apr 2017 18:09:03 +0000 (11:09 -0700)]
Make createdump build and work on Centos (#10985)

7 years agoRemember AdjustementRule index
Eric Erhardt [Sat, 15 Apr 2017 16:54:29 +0000 (11:54 -0500)]
Remember AdjustementRule index

Profiling TimeZoneInfo scenarios on Unix shows most of the time being spent in GetPreviousAdjustmentRule. This is because we need to scan all of the AdjustmentRules, looking for the specified rule, and then getting the one before it.

This change remembers the AdjustmentRule index when passing the AdjustmentRule into GetPreviousAdjustmentRule. This allows us to not scan all the rules, and increases the performance of DateTime.Now significantly in time zones that have daylight savings time.

Fix https://github.com/dotnet/corefx/issues/3571

7 years ago[Arm64] Fix assembly errors (#10977)
Steve MacLean [Sat, 15 Apr 2017 15:39:02 +0000 (11:39 -0400)]
[Arm64] Fix assembly errors (#10977)

* [Arm64] Fix assembly errors

* [Arm64] Static use ldrb w2, [x2, w1, UXTW]

* Revert changes to src/vm/arm64/asmhelpers.asm

7 years agoMerge pull request #10958 from pgavlin/VSO397793
Pat Gavlin [Sat, 15 Apr 2017 05:16:44 +0000 (22:16 -0700)]
Merge pull request #10958 from pgavlin/VSO397793

Preserve VNs in fgMorphIntoHelperCall.

7 years agoRevert "Strip out unused reg code. (#10741)"
danmosemsft [Fri, 14 Apr 2017 22:53:33 +0000 (15:53 -0700)]
Revert "Strip out unused reg code. (#10741)"

This reverts commit ed4f594abf41a71b126152bb8755051d0831e12d.

7 years agoMerge pull request #10936 from BruceForstall/TempRegApi
Bruce Forstall [Fri, 14 Apr 2017 22:47:58 +0000 (15:47 -0700)]
Merge pull request #10936 from BruceForstall/TempRegApi

Introduce API for codegen getting temp registers from gtRsvdRegs

7 years agoAdd a brief comment to the new test.
Pat Gavlin [Fri, 14 Apr 2017 22:18:12 +0000 (15:18 -0700)]
Add a brief comment to the new test.

7 years agoSmall clean up of stackEntry (#10820)
Sergey Andreenko [Fri, 14 Apr 2017 20:58:25 +0000 (13:58 -0700)]
Small clean up of stackEntry (#10820)

Make all stackDepth equal unsigned int.
Delete possible wrong function (the comment said : "// used in the inliner, where we can assume typesafe code. please don't use in the importer!!", but it was used).
Delete methods, that used arguments as output buffers.

* delete code without effects

7 years agoIntroduce API for codegen getting temp registers from gtRsvdRegs
Bruce Forstall [Thu, 13 Apr 2017 00:16:46 +0000 (17:16 -0700)]
Introduce API for codegen getting temp registers from gtRsvdRegs

LSRA puts a set of temporary registers needed by a node in the
gtRsvdRegs register mask. Currently, evey codegen function that
needs to use a temporary register manually manipulates this mask.

Introduce a few functions to make this simpler and more regular:

1. GetSingleTempReg() // Gets a temp; asserts there is exactly one temp reg.
2. ExtractTempReg() // Gets the lowest temp, removes it from gtRsvdRegs so subsequent calls don't see it.
3. AvailableTempRegCount() // Returns a count of available temp registers.

All take an optional register mask, so you can extract a register from
just the set specified by the mask (typically RBM_ALLINT or RBM_ALLFLOAT).