Swaroop Sridhar [Tue, 4 Apr 2017 00:23:11 +0000 (17:23 -0700)]
Fix invocation of ILLink in linux testing
Fix the input to illink -a argument to not include the extension.
The current command line caused silent wrong code gneration
because of https://github.com/mono/linker/issues/56
Jonghyun Park [Wed, 5 Apr 2017 00:29:50 +0000 (09:29 +0900)]
[x86/Linux] 16-byte aligned ResolveWorkerAsmStub (#10655)
William Godbe [Tue, 4 Apr 2017 23:42:20 +0000 (16:42 -0700)]
Merge pull request #10715 from wtgodbe/alpineFix
Fix Alpine build break in pipeline
Jan Vorlicek [Tue, 4 Apr 2017 23:41:25 +0000 (01:41 +0200)]
Update module not found error message (#10692)
This message can be misleading, since it says "The specified module
could not be found", but it would also report case when the module
was found, but one of the module's dependencies were not found.
Multiple people in the past were hit by that when their module was
not loaded, they were thinking that there is some problem with
the path coreclr is loading modules from or something like that and
spent non-trivial amount of time trying to figure that out.
The fact that the module could have missing dependencies is really
not obvious.
So I am updating the message to reflect that fact.
wtgodbe [Tue, 4 Apr 2017 23:24:51 +0000 (16:24 -0700)]
Fix Alpine build break in pipeline
Sergey Andreenko [Tue, 4 Apr 2017 23:20:29 +0000 (16:20 -0700)]
CoreRT: make CORINFO_HELP_READYTORUN_DELEGATE_CTOR optimization for verifiable sequences. (#10663)
delegate ctor optimization for CoreRT.
Rahul Kumar [Tue, 4 Apr 2017 22:33:36 +0000 (15:33 -0700)]
Merge pull request #10190 from sdmaclea/PR-ARM64-BACKPORT-9500
[Arm64/Windows] Backport #9500 changes
Jarret Shook [Tue, 4 Apr 2017 20:58:38 +0000 (13:58 -0700)]
Merge pull request #10614 from sdmaclea/PR-ARM64-GENRET-GCINFO
[Arm64] Fix GCStress hole in genReturn
Bruce Forstall [Tue, 4 Apr 2017 20:20:32 +0000 (13:20 -0700)]
Merge pull request #10609 from alpencolt/ryu-arm-locallock
[RyuJIT/ARM32] Localloc implementation in codegen
Aditya Mandaleeka [Tue, 4 Apr 2017 20:10:40 +0000 (13:10 -0700)]
Merge pull request #10699 from adityamandaleeka/fix_env_empty_var_handling
Fix handling of empty environment variables in PAL
Bruce Forstall [Tue, 4 Apr 2017 18:09:36 +0000 (11:09 -0700)]
Merge pull request #10681 from BruceForstall/LinuxArmAltjit
Add Linux host support for arm32/arm64 altjit
Aditya Mandaleeka [Tue, 4 Apr 2017 17:31:46 +0000 (10:31 -0700)]
Fix handling of empty environment variables in PAL.
Rion Williams [Tue, 4 Apr 2017 17:47:54 +0000 (12:47 -0500)]
Removed Unused Local Random Instances (#10696)
* Removed Unused Local Random Instances
Discovered two cases of a `Random` instance that weren't being used
within their respective methods and removed them.
* Retained Earlier Comment
Retained an earlier comment for consistency
Matt Mitchell [Tue, 4 Apr 2017 15:29:26 +0000 (08:29 -0700)]
Use setMachineAffinity for arm64 (#10558)
Hanjoung Lee [Tue, 4 Apr 2017 14:25:40 +0000 (23:25 +0900)]
[x86/Linux] Fix crash when GC triggered inside funclets (#10188)
* [x86/Linux] Fix crash when GC triggered inside funclets
* [x86/Linux] Make main function not to report
Hyeongseok Oh [Tue, 4 Apr 2017 12:07:09 +0000 (21:07 +0900)]
Pass crossbuild and linux-id to CMakeLists.txt using CMake arguments (#10686)
Remove in CMakeLists.txt
1) read cross-build from environment variable
2) read linux-id from file system
Instead, pass cross-build and linux-id from gen-buildsys-clang.sh using cmake arguments
Jonghyun Park [Tue, 4 Apr 2017 08:50:02 +0000 (15:50 +0700)]
[x86/Linux] Correctly unwind esp frames (#10685)
Rahul Kumar [Tue, 4 Apr 2017 08:02:19 +0000 (01:02 -0700)]
Merge pull request #10601 from rahku/c4996
1. Remove disable of C4996
Bruce Forstall [Tue, 4 Apr 2017 05:32:22 +0000 (22:32 -0700)]
Merge pull request #10682 from BruceForstall/CleanUpJitEEDefine
Remove unused code under typo define
Noah Falk [Tue, 4 Apr 2017 04:04:16 +0000 (21:04 -0700)]
Merge pull request #10580 from noahfalk/fitjit_tier0_flag
Add Tier0 jit flag
Nicolò Carandini [Tue, 4 Apr 2017 01:42:15 +0000 (03:42 +0200)]
Update README.md (#10641)
In this file I’ve found a “project.json” occurrence that I've changed to ".csproj".
Matt Warren [Tue, 4 Apr 2017 01:37:48 +0000 (02:37 +0100)]
Move '#pragma warning(disable:21000)' to correct place (#10455)
Whilst I was writing a blog post ['A Hitchhikers Guide to the CoreCLR Source Code'](http://mattwarren.org/2017/03/23/Hitchhikers-Guide-to-the-CoreCLR-Source-Code/#top-10-lists) I noticed that a `#pragma warning(disable:21000)` was mis-aligned.
I'm pretty sure that having it in the wrong place doesn't cause a problem (i.e. this doesn't actually *fix* anything), because the other methods are smaller, so I understand if you don't want to take the PR.
Ben Adams [Tue, 4 Apr 2017 01:20:54 +0000 (02:20 +0100)]
Improve Guid.ToString (#10669)
* Improve Guid.ToString
* revert HexToChar
Brian Sullivan [Tue, 4 Apr 2017 00:30:36 +0000 (17:30 -0700)]
Merge pull request #10678 from briansull/contract-fix
Remove the SO_TOLERANT contact from CEEInfo::resolveVirtualMethodHelper
Dan Moseley [Tue, 4 Apr 2017 00:15:24 +0000 (17:15 -0700)]
Debug.Assert FF (#10652)
Igor Kulaychuk [Tue, 4 Apr 2017 00:10:40 +0000 (03:10 +0300)]
Fix 'this' pointer in struct methods (#10668)
The 'this' method argument should always be a pointer.
Bruce Forstall [Mon, 3 Apr 2017 23:59:35 +0000 (16:59 -0700)]
Remove unused code under typo define
The code is under COR_JIT_EE_VER which never existed (it used to
be COR_JIT_EE_VERSION), thus the code has never been used.
Dan Moseley [Mon, 3 Apr 2017 23:43:43 +0000 (16:43 -0700)]
remove fedora 23 (#10667)
Russell C Hadley [Mon, 3 Apr 2017 23:37:00 +0000 (16:37 -0700)]
Merge pull request #10629 from russellhadley/EHWriteThru-design-doc
Add initial design document for EH WriteThru
Koundinya Veluri [Mon, 3 Apr 2017 23:35:47 +0000 (16:35 -0700)]
Fix InternalsVisibleTo when it references an assembly with some DebuggableAttribute flags (#10664)
Fixes #3541
- Mask out the DebuggableAttribute bits from when comparing assembly spec flags for matching an InternalsVisibleTo reference to an assembly
William Godbe [Mon, 3 Apr 2017 23:26:02 +0000 (16:26 -0700)]
Merge pull request #10671 from wtgodbe/Redist
Publish VC Redist DLLs as part of CoreCLR Package
Brian Sullivan [Mon, 3 Apr 2017 23:01:23 +0000 (16:01 -0700)]
Remove the SO_TOLERANT contact from CEEInfo::resolveVirtualMethodHelper
that contract should only apply to CEEInfo::resolveVirtualMethod which uses the JIT_TO_EE_TRANSITION() transition macros.
Pat Gavlin [Mon, 3 Apr 2017 22:34:34 +0000 (15:34 -0700)]
Merge pull request #10672 from pgavlin/FixExceptionReference
Fix a reference to System.Exception in a JIT test.
Bruce Forstall [Mon, 3 Apr 2017 21:44:27 +0000 (14:44 -0700)]
Add Linux host support for arm32/arm64 altjit
Now, on Linux, the x64 build will build an arm64-targeting
altjit. The x86 build will build an arm32-targeting JIT.
The altjit is named libprotononjit.so/dylib. It can be
used for debugging or for generating asm diffs using jit-diff.
wtgodbe [Mon, 3 Apr 2017 20:29:27 +0000 (13:29 -0700)]
Publish VC Redist DLLs as part of CoreCLR Package
Jarret Shook [Mon, 3 Apr 2017 20:56:40 +0000 (13:56 -0700)]
Merge pull request #10326 from jashook/arm64_test_update_20_march
Update the arm64 test drop - WIP
Pat Gavlin [Mon, 3 Apr 2017 20:53:04 +0000 (13:53 -0700)]
Fix a reference to System.Exception in a JIT test.
This bad reference was casuing a failure in the x64 pri1 R2R jobs.
ragmani [Mon, 3 Apr 2017 19:27:44 +0000 (04:27 +0900)]
[x86/Linux] add the process of handling when entry point is funclet. (#10602)
* [x86/Linux] add the process of handling the entry point when it is funclet in case x86.
* [x86/Linux] enable only for TARGET_X86 && FEATURE_PAL.
Signed-off-by: ragmani <ragmani0216@gmail.com>
jashook [Mon, 20 Mar 2017 21:14:31 +0000 (14:14 -0700)]
Update the arm64 test drop
Alexander Soldatov [Mon, 3 Apr 2017 14:43:30 +0000 (17:43 +0300)]
[RyuJIT/ARM32] Localloc implementation in codegen
Algorithm is almost the same as for other platforms but code
is little bit refactored to improve readability.
Added test to cover almost all possible branches of codegen.
Since arrays allocation of int type is needed multiplication and overflow
checks new tests cover only byte arrays.
Alexander Soldatov [Mon, 3 Apr 2017 12:01:56 +0000 (15:01 +0300)]
[RyuJIT/ARM32] GS Cookie check implementation.
Implementation of GS Cookie check for RyuJIT/ARM32 almost the same as
for ARM64. So this code was moved to codegencommon with.
Also added optimization to use tmp register for emitting helper call.
Dan Moseley [Mon, 3 Apr 2017 12:44:48 +0000 (05:44 -0700)]
Cut down FormatterServices (#10653)
Tarek Mahmoud Sayed [Sun, 2 Apr 2017 19:18:22 +0000 (12:18 -0700)]
Fix Double.ToString performance on Linux and OSX (#10572)
* Fix Double.ToString performance on Linux and OSX
* some feedback fixes
* Remove the implementation depending on ecvt_r
Hugh Bellamy [Sun, 2 Apr 2017 16:12:29 +0000 (23:12 +0700)]
Bring back kApplicationException (#10650)
Dan Moseley [Sun, 2 Apr 2017 10:16:23 +0000 (03:16 -0700)]
Remove securitystate (#10649)
Jan Kotas [Sat, 1 Apr 2017 19:25:28 +0000 (12:25 -0700)]
Move identical Globalization files to shared partition
Jan Kotas [Sat, 1 Apr 2017 18:40:23 +0000 (11:40 -0700)]
Move identical files to shared partition
Jan Kotas [Sat, 1 Apr 2017 18:26:05 +0000 (11:26 -0700)]
Move identical Interop files to shared partition
Sean Gillespie [Sat, 1 Apr 2017 21:25:46 +0000 (14:25 -0700)]
[Local GC] Move some EE-specific finalize-on-unload logic out of the GC (#10598)
* Move some EE-specific logic to a callback on GCToEEInterface
* Fix the sample
* Consistent style for pointers
* Code review feedback: Move app domain index check into EE callback and make ShouldFinalizeObjectForUnload always return true (due to CoreCLR not having app domains)
* Code review feedback: Fix a comment and add a TODO for bringing these changes to desktop
Sean Gillespie [Sat, 1 Apr 2017 21:25:07 +0000 (14:25 -0700)]
[Local GC] Remove static fields from GC interface (#10566)
* Remove max_generation from GC interface
* [Local GC] Clean up the GC interface by removing all static fields
from it:
1) gcHeapType is replaced by g_heap_type (EE side) and g_gc_heap_type
(GC side), each of which is set on initialization. g_heap_type is
read by the DAC to determine what kind of heap is being used.
2) maxGeneration is not necessary due to the GC DAC changes.
* Rebase against master
* Comments and cleanup
* Use a heap allocation instead of alloca
* Code review feedback: remove a redundant cast and allocate generation count table once, on first use
* Address code review feedback: cache some calls to GetMaxGeneration
Aditya Mandaleeka [Sat, 1 Apr 2017 19:25:08 +0000 (12:25 -0700)]
Merge pull request #10584 from adityamandaleeka/handle_table_local_gc_init
Change how the VM calls ObjectFromHandle.
Matt Ellis [Sat, 1 Apr 2017 18:13:03 +0000 (11:13 -0700)]
Add .gitmirror file
Brian Sullivan [Sat, 1 Apr 2017 17:46:58 +0000 (10:46 -0700)]
Merge pull request #10627 from briansull/profile-data
Added DisableInlining flag to the ProfileData.
Russell C Hadley [Mon, 27 Mar 2017 17:14:25 +0000 (10:14 -0700)]
Add initial design document for EH WriteThru
Document describes an initial approch to enregistering local variables
that live across EH flow. Today the JIT forces all local vars that are live
across EH flow that kills all registers to live on the stack. The transformation outlined ensures consistency on the stack across the EH kills but allows the enregistration with in either the normal flow of control or with in any EH handlers.
Luis G [Sat, 1 Apr 2017 09:00:12 +0000 (06:00 -0300)]
Refactor dotnet download code in init-tools.cmd (#10527)
* Refactor dotnet download code in init-tools.cmd
This addresses the improvements proposed in issue #10526.
Includes the init-tools.cmd script refactor and a new script called
dotnet-download.ps1 which includes the extracted code and logic
* Code review feedback changes
* Fix spacing
Andy Ayers [Sat, 1 Apr 2017 07:56:33 +0000 (00:56 -0700)]
Jit: fix a few issues with single def local tracking (#10633)
Fixes for three issues that came up in desktop testing.
1. Don't track class types for locals when in import only mode,
since the jit may be looking at an uninstantiated generic method.
In these cases the jit's use of TYP_REF to represent CORINFO_TYPE_VAR
confuses the tracking code. In import only mode the jit is not going
to use the information, so there is no need to track it at all.
Import only mode is tied to verification and CoreCLR runs in full
trust mode, so no test was added.
2. Allow `lvaUpdateClass` to be called more than once but assert
that the second and subsequent calls provide the exact same update
as the first call.
This can happen for single def ref class locals whose definition
block is reimported. Reimportation is triggered by some tolerable
non-ref type mismatches at block joins and so ref type information
should be consistent for each importation attempt.
Add a couple of IL test cases translated from destkop MC++ tests
that will trigger this kind of reimportation.
3. Bail out of devirtualization (for desktop only) if the base class
has precise initialization semantics, since desktop seems to trigger
class initialization for ref type callvirts (contrary to ECMA355 I.8.9.5).
See #4853 for some discussion. Since this is desktop only behavior
no test was added.
Andy Ayers [Sat, 1 Apr 2017 06:04:48 +0000 (23:04 -0700)]
Merge pull request #10638 from AndyAyersMS/EmptyTryUpdateFinallyNestLevel
Jit: decrement finally nesting level when removing empty trys
Jan Kotas [Fri, 31 Mar 2017 23:20:12 +0000 (16:20 -0700)]
Delete moved files
Michal Strehovský [Fri, 31 Mar 2017 22:08:40 +0000 (15:08 -0700)]
Merge pull request dotnet/corert#3175 from jkotas/nmirror-merge
Merge nmirror to master
Signed-off-by: dotnet-bot <dotnet-bot@microsoft.com>
Aditya Mandaleeka [Sat, 1 Apr 2017 00:12:03 +0000 (17:12 -0700)]
Change how VM calls ObjectFromHandle.
Aditya Mandaleeka [Sat, 1 Apr 2017 00:09:46 +0000 (17:09 -0700)]
Add way to get table for handle and ADIndex for handletable.
noahfalk [Thu, 30 Mar 2017 07:07:59 +0000 (00:07 -0700)]
Add Tier0/1 jit flags
These flags provides a hook to change the JIT policy in the future and diverge tier0/tier1 compilation from min_opt/speed_opt respectively.
Sergey Andreenko [Sat, 1 Apr 2017 01:36:06 +0000 (18:36 -0700)]
delete GetMethod2 (#10626)
* delete GetMethod2
* delete unused code
Andy Ayers [Sat, 1 Apr 2017 01:13:32 +0000 (18:13 -0700)]
Jit: decrement finally nesting level when removing empty trys
For non-funclet EH models, the GT_END_LFIN statement tracks the
nesting level of the associated finally. When removing a try-finally
with an empty try we need to decrement this level for any try-finallys
nested within the associated finally.
Added a test case with several levels of empty try and nesting.
Closes #10621.
Sergey Andreenko [Sat, 1 Apr 2017 01:12:53 +0000 (18:12 -0700)]
delete wrong and unused value (#10634)
Bruce Forstall [Sat, 1 Apr 2017 00:19:54 +0000 (17:19 -0700)]
Merge pull request #10467 from helloguo/VectorConversionTest
Add vector conversion tests
Bruce Forstall [Sat, 1 Apr 2017 00:14:19 +0000 (17:14 -0700)]
Merge pull request #10619 from BruceForstall/legacynonjit
Add ARM32 legacy altjit build
Ahson Ahmed Khan [Sat, 1 Apr 2017 00:10:50 +0000 (17:10 -0700)]
Removing Slice on string overloads and adding AsSpan (#10544)
Jan Kotas [Fri, 31 Mar 2017 23:15:56 +0000 (16:15 -0700)]
Merge pull request #10625 from jkotas/debug
Move System.Diagnostics.Debug from corefx to coreclr
Ahson Ahmed Khan [Fri, 31 Mar 2017 22:27:46 +0000 (15:27 -0700)]
Temporary removing string slice span bench test. (#10617)
* Temporary removing string slice bench test.
* Using if false directive to skip breaking tests
Brian Sullivan [Fri, 31 Mar 2017 01:16:58 +0000 (18:16 -0700)]
Added DisableInlining flag to the ProfileData.
Extended CompileStatus to have both COMPILE_HOT_EXCLUDED and COMPILE_COLD_EXCLUDED.
Fixes the IsNull implementation
Fixes getBBProfileData to handle the case where pos can now be zero
Rahul Kumar [Wed, 29 Mar 2017 22:31:29 +0000 (15:31 -0700)]
1. Remove disable of C4996
2. delete some dead code
3. Remove implementation of GetOSVersion and hardcode OS version to be 0.0.0
4. Add implementation of IsWindows8OrAbove & IsWinrtSupported (donno why this was deleted earlier)
5. remove GetVersionEx from PAL
Bruce Forstall [Fri, 31 Mar 2017 17:11:28 +0000 (10:11 -0700)]
Add altjit build for Windows x86-hosted, Windows ARM32 LEGACY_BACKEND target named legacynonjit.dll
Bruce Forstall [Fri, 31 Mar 2017 21:21:23 +0000 (14:21 -0700)]
Merge pull request #10193 from mskvortsov/ryujit-arm32-fix-struct
[RyuJIT/ARM32] Fix lvOnFrame for struct args
Jan Kotas [Fri, 31 Mar 2017 20:33:53 +0000 (13:33 -0700)]
Change the test hook from interface to delegate
Jan Kotas [Fri, 31 Mar 2017 18:45:17 +0000 (11:45 -0700)]
Fix build breaks
Atsushi Kanamori [Fri, 31 Mar 2017 18:41:39 +0000 (11:41 -0700)]
NS2.0 Inject more missing types into CoreRT by sharing them. (#10613)
* NS2.0 Inject more missing types into CoreRT by sharing them.
- ArgIterator was cleaned up for sharing but actually
sharing it today is too messy with TypedReference
being in the wrong namespace on half of CoreRT.
Some preparatory fixes discovered during the
attempt will be going in on CoreRT...
* Leave RuntimeArgumentHandle and ArgIterator unshared.
Jan Kotas [Fri, 31 Mar 2017 18:18:51 +0000 (11:18 -0700)]
Copy System.Diagnostics.Debug implementation from corefx
Ben Adams [Fri, 31 Mar 2017 17:50:36 +0000 (18:50 +0100)]
Fallback for enum.Format(G) when not valid (#10610)
[Arm64] Fix GCStress hole in genReturn
Atsushi Kanamori [Fri, 31 Mar 2017 14:20:30 +0000 (07:20 -0700)]
Prepare TimeZoneInfo.cs for move to shared partition. (#10606)
(This is the CoreRT side of https://github.com/dotnet/coreclr/pull/10606)
TimeZoneInfo.cs needs no changes but it brings in
another file CurrentTimeZone.cs.
Which drags in yet another file (Hashtable.cs)
which is widely used inside CoreCLR but aside
from this one file, it seems to be stuff that'll
will never be ported over or already has been
ported over sans Hashtable.
So we'll refactor the memoization logic of
CurrentTimeZone.cs into its own partial file
and share the rest. CoreCLR will use continue to use Hashtable.
CoreRT will use ConcurrentUnifier.
And while we're at it, will rename it to
CurrentSystemTimeZone.cs to match the actual class name.
Once we make a corresponding change on the CoreCLR side,
both TimeZone.cs and CurrentSystemTimeZone.cs will
be fully synced and can move to the shared partition.
Steve MacLean [Fri, 31 Mar 2017 02:06:23 +0000 (22:06 -0400)]
Add assert to GCHeap::ValidateObjectMember (#10591)
In presence of a corrupt heap, objects can contain
null method table. Add assertion to prevent segfault
in checked/debug builds.
Michal Strehovský [Fri, 31 Mar 2017 02:00:24 +0000 (19:00 -0700)]
Add test coverage for ldtoken of open generic methods (#10592)
Jeff Schwartz [Thu, 30 Mar 2017 23:37:42 +0000 (16:37 -0700)]
Merge pull request #10593 from dotnet/standaloneservergc
Enable StandAloneGC and Server GC
Hyung-Kyu Choi [Thu, 30 Mar 2017 22:40:19 +0000 (07:40 +0900)]
[Ryujit/ARM32][ReadyToRun] Fix invocation to Thunk (#10348)
Fix invocation to Thunk code from code generated from ReadyToRun.
Thunk code for ARM32 accept r12 as a parameter.
Signed-off-by: Hyung-Kyu Choi <hk0110.choi@samsung.com>
Matt Mitchell [Thu, 30 Mar 2017 22:05:38 +0000 (15:05 -0700)]
Enable CROSS for testing (#10594)
Brian Sullivan [Thu, 30 Mar 2017 22:03:49 +0000 (15:03 -0700)]
Merge pull request #10513 from briansull/profile-data
Crossgen support for ExcludeHotMethodCode and ExcludeColdMethodCode
Pat Gavlin [Thu, 30 Mar 2017 22:02:10 +0000 (15:02 -0700)]
Fix Issue #10022. (#10587)
This issue occurred because the JIT currently inserts the P/Invoke
method prolog into the first block of a function without ensuring that
this block will only execute once. This is not safe, as executing the
P/Invoke prolog multiple times can create cycles in the frame list and
casue the stack unwinder to hang.
This change ensures that lowering inserts a new, single-entry basic
block at the beginning of the function for functions that require P/Invoke prologs.
Jeff Schwartz [Thu, 30 Mar 2017 22:01:26 +0000 (15:01 -0700)]
Fix tabs and spaces
Jonghyun Park [Thu, 30 Mar 2017 21:20:01 +0000 (06:20 +0900)]
[x86/Linux] Use CDECL instead of STDCALL (#10410)
[x86/Linux] Use CDECL instead of STDCALL
Use STDCALL/THISCALL/FASTCALL if callconv is explicitly specified
Merge two adjustment (argument/alignment) into one
Bruce Forstall [Thu, 30 Mar 2017 21:16:47 +0000 (14:16 -0700)]
Merge pull request #10228 from mskvortsov/ryujit-arm32-copy-barrier
[RyuJIT/ARM32] Enable GT_COPY and GT_MEMORYBARRIER
Jarret Shook [Thu, 30 Mar 2017 20:46:59 +0000 (13:46 -0700)]
Merge pull request #10588 from jashook/add_correct_failure_metadata_for_b08046
Re-disable b08046 and add correct metadata.
Jeff Schwartz [Thu, 30 Mar 2017 20:41:39 +0000 (13:41 -0700)]
Change necessary to enable standalonegc on server gc. Now CreateThread in gcenv.windows.cpp matches gcenv.os.cpp
Sujin Kim [Thu, 30 Mar 2017 20:37:13 +0000 (05:37 +0900)]
[Ryujit/ARM32] Implement NYI related with overflow for ARM (#10491)
* Implement NYI(overflow checks) for ARM
On last comment of #8496, the NYI message of overflow checks is printed after running the CodeGenBringUpTests.
That was the message about temp register setup for overflow checks.
It was referenced
https://github.com/dotnet/coreclr/blob/master/src/jit/lsraarm64.cpp#L399
I think it doesn't make any problem even though writing it the same as arm64.
* modifiy for coding convention
* Implement NYI : Unimplmented GT_CAST:int <--> int with overflow
I think it doesn't make any problem even though writing it the same as arm64.
So I copied parts of CodeGen::genIntToIntCast() and modified some below codes.
```
if (emitter::emitIns_valid_imm_for_cmp(castInfo.typeMax, cmpSize))
```
-->
```
if (emitter::emitIns_valid_imm_for_cmp(castInfo.typeMax, INS_FLAGS_DONT_CARE))
```
* Implement NYI : genLongToIntCast: overflow check
I copied and pasted codes from codegenxarch.cpp.
But It seemed be necessary that conditional execution values are changed by each architectures.
So I used 'genJumpKindForOper' for getting the emitJumpKind value.
The sample app has been checked to work well.
* Modify the implementation of emitter::emitIns_valid_imm_for_cmp
According to reference manual, I figured out CMP and ADD have different mechanisms on ARM unlike ARM64.
So I defined "...for_cmp" function not just use "..for_add" in the function likes ARM64.
Joseph Tremoulet [Thu, 30 Mar 2017 18:57:06 +0000 (14:57 -0400)]
Merge pull request #10569 from sdmaclea/PR-FIX-10359
Handle null ref in Compiler::optCreateAssertion
jashook [Thu, 30 Mar 2017 18:07:13 +0000 (11:07 -0700)]
Re-disable b08046 and add correct metadata.
\JIT\Regression\CLR-x86-JIT\V1.2-M01\b08046\b08046\b08046.cmd was disabled because the issue #2414
was closed but the tests tagged with that issue were not removed. b08046's correct issue is #4849.
Sean Gillespie [Thu, 30 Mar 2017 17:11:28 +0000 (10:11 -0700)]
Fix an issue where the DAC and GC versions of the heap_segment data structure did not agree on the location of the heap field (#10576)
Fix formatting
Andrey Akinshin [Thu, 30 Mar 2017 13:25:42 +0000 (18:25 +0500)]
Fix magic number explanation in filetime.cpp (#10582)
89 * 366 + 280 * 365 equals to 134774 (not 134744)
Igor Kulaychuk [Thu, 30 Mar 2017 09:54:02 +0000 (12:54 +0300)]
[Linux][GDB-JIT] Remove (nothrow) when using new (#10457)
* Merge ByteTypeInfo into PrimitiveTypeInfo
* Remove ArrayTypeInfo ownership over its array element type
ArrayTypeInfo should not delete memory for its array element type
because element type is owned by NotifyGdb::PTK_TypeInfoMap.
* Make NamedRefTypeInfo to own its value type pointer
* Fix memory leak when creating ArrayTypeInfo
* Refactor creation of PrimitiveTypeInfo class
* Move code from GetTypeInfoFromTypeHandle to ClassTypeInfo constructor
* Move code from GetTypeInfoFromTypeHandle to RefTypeInfo constructor
* Remove (nothrow) usage from GetTypeInfoFromTypeHandle function
* Remove most usages of (nothrow) from gdbjit
* Remove all remaining (nothrow) usages from gdbjit
* Throw COR_E_NOTSUPPORTED from GetTypeInfoFromTypeHandle when dumping unsupported types
* Remove (nothrow) usage from MemBuf::Resize
* Use NewArrayHolder for vars in FunctionMember class
* Use NewArrayHolder for m_type_name in TypeInfoBase class
* Use NewArrayHolder for m_typedef_name in TypeDefInfo class
* Use NewArrayHolder for m_member_name in TypeMember class
* Use NewArrayHolder for m_var_name in VarDebugInfo class
* Use NewArrayHolder for members in ClassTypeInfo class
* Add m_symbol_name field to Elf_Symbol struct for managing symbol name memory
* Add m_value_type_storage field to NamedRefTypeInfo class for managing referenced type memory
* Remove CalledMethod list cleanup - avoid having dangling pointer in CodeHeader class
* Fix memory leak in GetMethodNativeMap function
* Simplify FunctionMemberPtrArrayHolder class
Since FunctionMemberPtrArrayHolder instance lifetime is limited to
NotifyGdb::MethodCompiled method, there is no need for complicated reallocation code.
* Wrap NotifyGdb::MethodCompiled endtry point in try/catch block
* Fix argument naming style - rename MethodDescPtr to methodDescPtr