Pat Gavlin [Mon, 22 Aug 2016 21:01:31 +0000 (14:01 -0700)]
Remove fgDebugCheckLinear{Tree,NodeLinks}. (#6842)
These functions are not necessary for LIR.
Hannah Zhang [Mon, 22 Aug 2016 20:12:48 +0000 (13:12 -0700)]
Merge pull request #6840 from hann013/t-hazhan/SuppressIldasmAttribute
Added SuppressIldasmAttribute
Eric Mellino [Mon, 22 Aug 2016 17:29:35 +0000 (10:29 -0700)]
Merge pull request #6821 from mellinoe/opensuse421
Add a CI job for openSUSE 42.1
Hannah Zhang [Mon, 22 Aug 2016 17:24:29 +0000 (10:24 -0700)]
Added SuppressIldasmAttribute to mscorlib and model.xml
Brian Sullivan [Mon, 22 Aug 2016 14:05:34 +0000 (07:05 -0700)]
Merge pull request #6806 from wateret/fix-hfa
[ARM/Linux] Fix HFA structs
James Ko [Mon, 22 Aug 2016 02:00:19 +0000 (22:00 -0400)]
Further optimizations for String.Join, String.Concat (#6800)
* Apply length-1 optimization to Join(string, object[])
* Apply length-1 optimization to Concat(IEnumerable<string>)
* Apply length-1 optimization to Concat(IEnumerable<T>)
* Apply lengths 0/1 optimizations to Concat(object[]), Concat(string[])
Hanjoung Lee [Fri, 19 Aug 2016 07:32:00 +0000 (16:32 +0900)]
[ARM/Linux] Fix HFA structs
Re-enable FEATURE_MULTIREG_RET for ARM32 which was disabled by #6467.
Fix #6677
Bruce Forstall [Sun, 21 Aug 2016 03:51:31 +0000 (20:51 -0700)]
Merge pull request #6832 from dotnet-bot/from-tfs
Merge changes from TFS
Matt Ellis [Sun, 21 Aug 2016 01:29:38 +0000 (18:29 -0700)]
Merge remote-tracking branch 'upstream/master' into from-tfs
Bruce Forstall [Sat, 20 Aug 2016 20:27:20 +0000 (13:27 -0700)]
Disable SuperPMI unit test for x86 legacy backend
[tfs-changeset: 1623421]
Bruce Forstall [Sat, 20 Aug 2016 20:15:09 +0000 (13:15 -0700)]
SuperPMI: fix ARM build break
[tfs-changeset: 1623419]
Pat Gavlin [Sat, 20 Aug 2016 17:41:28 +0000 (10:41 -0700)]
Assert that any node with a non-zero dstCount produces a value. (#6827)
* Assert that any node with a non-zero dstCount produces a value.
The register allocator will otherwise misallocate, as such nodes will
produce an unconsumed value.
* Fix the dstCount for GT_RETFILT.
Bruce Forstall [Sat, 20 Aug 2016 17:15:38 +0000 (10:15 -0700)]
Disable superpmi test for RyuJIT/x86
[tfs-changeset: 1623413]
Bruce Forstall [Sat, 20 Aug 2016 17:07:50 +0000 (10:07 -0700)]
Fix build break related to LogException requiring non-empty varargs arguments
[tfs-changeset: 1623412]
Peter Jas [Sat, 20 Aug 2016 07:05:50 +0000 (10:05 +0300)]
Fix ARM64 build breakage (#6819)
Jan Kotas [Sat, 20 Aug 2016 06:51:25 +0000 (23:51 -0700)]
Delete CanChangeType from model.xml
[tfs-changeset: 1623392]
Jan Kotas [Sat, 20 Aug 2016 03:28:13 +0000 (20:28 -0700)]
Fix build breaks and warnings in internal build
[tfs-changeset: 1623382]
Eugene Rozenfeld [Sat, 20 Aug 2016 01:47:13 +0000 (18:47 -0700)]
Merge pull request #6793 from erozenfeld/TailCallInCalleeInlining
Don't inline methods that have tail-prefixed calls.
Jose Perez Rodriguez [Sat, 20 Aug 2016 00:41:31 +0000 (17:41 -0700)]
Fix casing of DaylightTime.cs
[tfs-changeset: 1623368]
Aditya Mandaleeka [Sat, 20 Aug 2016 00:29:09 +0000 (17:29 -0700)]
Merge pull request #6799 from adityamandaleeka/osx_ci_dumps
Enable additional functionality for OS X CI dumps.
Carol Eidt [Fri, 19 Aug 2016 23:39:19 +0000 (16:39 -0700)]
Merge pull request #6816 from CarolEidt/FixArmCrossgen2
Fix dstCount of GT_RETURNTRAP for Arm64
Mike McLaughlin [Fri, 19 Aug 2016 23:06:56 +0000 (16:06 -0700)]
Create SOS SymbolReader managed project and change SOS to use it. (#6783)
* Change SOS to use wrapper class that is now in System.Diagnostics.StackTrace.
The portable PDB helper code for SOS source/line support has been moved from
System.Diagnostics.Debug.SymbolReader to a new managed SOS project in the coreclr
repo called SOS.NETCore.
The public APIs have now been made internal.
Plumb through the loaded PE address to the managed SymbolReader functions so it can be used as a key.
Fixed a stack trashing/overflow when a unresolved managed breakpoint is resolved because one of the
module name buffers was too small (MAX_PATH_FNAME). Changed it (and others) to MAXLONGPATH.
SOS now works with Portable PDBs on Windows.
New dac private get module data request. Used to get the necessary info for portable PDBs.
SOS now supports in-memory PE's on xplat and Windows. Needed to get and plumb though the in-memory
PE layout where it is file based or loaded.
Better Windows GetLineByILOffset support. Uses the SymbolReader and now works with in-memory PEs.
Misc code formatting and general cleanup.
* Code review feedback.
Bruce Forstall [Fri, 19 Aug 2016 22:56:04 +0000 (15:56 -0700)]
Open source SuperPMI
OVERVIEW
========
This directory contains the SuperPMI tool used for testing the .NET
just-in-time (JIT) compiler.
SuperPMI has two uses:
1. Verification that a JIT code change doesn't cause any asserts.
2. Finding test code where two JIT compilers generate different code, or verifying
that the two compilers generate the same code.
Case #1 is useful for doing quick regression checking when making a source
code change to the JIT compiler. The process is: (a) make a JIT source code
change, (b) run that newly built JIT through a SuperPMI run to verify no
asserts have been introduced.
Case #2 is useful for generating assembly language diffs, to help analyze the
impact of a JIT code change.
SuperPMI works in two phases: collection and playback. In the collection phase,
the system is configured to collect SuperPMI data. Then, run any set of .NET managed
programs. When these managed programs invoke the JIT compiler, SuperPMI gathers and
captures all information passed between the JIT and its .NET host. In the
playback phase, SuperPMI loads the JIT directly, and causes it to compile all
the functions that it previously compiled, but using the collected data to
provide answers to various questions that the JIT needs to ask. The .NET
execution engine (EE) is not invoked at all.
TOOLS
==========
There are two native executable tools: superpmi and mcs. There is a .NET Core
C# program that is built as part of the coreclr repo tests build called
superpmicollect.exe.
All will show a help screen if passed -?.
COLLECTION
==========
Set the following environment variables:
SuperPMIShimLogPath=<full path to an empty temporary directory>
SuperPMIShimPath=<full path to clrjit.dll, the "standalone" JIT>
COMPlus_AltJit=*
COMPlus_AltJitName=superpmi-shim-collector.dll
(On Linux, use libclrjit.so and libsuperpmi-shim-collector.so. On Mac,
use libclrjit.dylib and libsuperpmi-shim-collector.dylib.)
Then, run some managed programs. When done running programs, un-set these variables.
Now, you will have a large number of .mc files. Merge these using the mcs
tool:
mcs -merge base.mch *.mc
One benefit of SuperPMI is the ability to remove duplicated compilations, so
on replay only unique functions are compiled. Use the following to create a
"unique" set of functions:
mcs -removeDup -thin base.mch unique.mch
Note that -thin is not required. However, it will delete all the compilation
result collected during the collection phase, which makes the resulting MCH
file smaller. Those compilation results are not required for playback.
Use the superpmicollect.exe tool to automate and simplify this process.
PLAYBACK
========
Once you have a merged, de-duplicated MCH collection, you can play it back
using:
superpmi unique.mch clrjit.dll
You can do this much faster by utilizing all the processors on your machine,
and replaying in parallel, using:
superpmi -p unique.mch clrjit.dll
REMAINING WORK
=============
The basic of assembly diffing are there, using the "coredistools"
package. The open source build needs to be altered to use this package
to wire up the correct build steps.
[tfs-changeset: 1623347]
Eugene Rozenfeld [Thu, 18 Aug 2016 17:28:37 +0000 (10:28 -0700)]
Don't inline methods that have tail-prefixed calls.
This commit changes the inlining behavior when the callee has tail-prefixed calls.
The jit was turning tail-prefixed calls in inlinees into normal calls. That means that
in some cases tail prefix wasn't honored. I changed the code to not inline such callees.
This matches the behavior of the legacy x64 jit. A possible improvement would be to allow
inlining when the tail-prefixed calls in the inlinee could still be dispatched as tail calls from the caller.
I enabled TailcallVerifyWithPrefix set of tests. They were disabled because Condition8.Test1, Condition8.Test2,
and Condition8.Test3 used varargs calling convention. I commented out code that was calling those tests.
I didn't delete them in case CoreCLR will support varargs in the future.
I also turned on Condition21.Test1, Condition21.Test2, Condition21.Test3, Condition21.Test6, and Condition21.Test7.
Condition21.Test3 is the test that was failing because the jit was inlining calees with tail-prefixed calls.
The other Condition21 tests above were passing and just weren't called.
Eric Mellino [Fri, 19 Aug 2016 17:15:12 +0000 (10:15 -0700)]
Use getBootstrapPublishRid function to override __PUBLISH_RID while bringing up new platforms
Eric Mellino [Thu, 18 Aug 2016 17:53:01 +0000 (10:53 -0700)]
Support MSBuild on opensuse.42.1-x64
Eric Mellino [Wed, 17 Aug 2016 22:49:15 +0000 (15:49 -0700)]
Add a run for opensuse 42.1
Carol Eidt [Fri, 19 Aug 2016 22:28:23 +0000 (15:28 -0700)]
Merge pull request #6813 from CarolEidt/FixArmCrossgen
Fix One Arm64 crossgen failure
Eric Mellino [Fri, 19 Aug 2016 22:17:04 +0000 (15:17 -0700)]
Merge pull request #6811 from mellinoe/update-buildtools
Update buildtools to 1.0.26-prerelease-00718-04, remove duplicate props
Koundinya Veluri [Fri, 19 Aug 2016 21:07:55 +0000 (14:07 -0700)]
Merge pull request #6812 from kouvel/Api
Expose some types deriving from SafeHandle
Carol Eidt [Fri, 19 Aug 2016 20:51:05 +0000 (13:51 -0700)]
Fix dstCount of GT_RETURNTRAP
Carol Eidt [Fri, 19 Aug 2016 19:14:28 +0000 (12:14 -0700)]
Fix Arm64 crossgen
The type of the ADDR node needs to be set to TYP_BYREF for arm64.
For x64/ux, oddly, it requires the struct type. I added a TODO-Cleanu8p
comment because a GT_ADDR should never be TYP_STRUCT.
Jose Perez Rodriguez [Fri, 19 Aug 2016 20:30:25 +0000 (13:30 -0700)]
Using the public DaylightTime instead of the internal one in corefx dir for Linux/OSX builds.
[tfs-changeset: 1623319]
Hannah Zhang [Fri, 19 Aug 2016 20:27:20 +0000 (13:27 -0700)]
Merge pull request #6810 from hann013/t-hazhan/SystemVersion
Added constructor to System.Version
Eric Mellino [Fri, 19 Aug 2016 17:24:34 +0000 (10:24 -0700)]
Remove properties which are now defined in buildtools
These all have the same values in buildtools now
Aditya Mandaleeka [Fri, 19 Aug 2016 02:35:45 +0000 (19:35 -0700)]
Don't create dumpling list file unless there are actually dumps.
Koundinya Veluri [Fri, 19 Aug 2016 18:20:21 +0000 (11:20 -0700)]
Expose some types deriving from SafeHandle
- SafeHandleMinusOneIsInvalid
- SafeHandleZeroOrMinusOneIsInvalid
Part of dotnet/corefx#10933
Stephen Toub [Fri, 19 Aug 2016 17:59:55 +0000 (13:59 -0400)]
Merge pull request #6808 from stephentoub/fix_wr
Fix WeakReference in mscorlib ref assembly
Eric Mellino [Thu, 18 Aug 2016 18:23:31 +0000 (11:23 -0700)]
Update buildtools to 1.0.26-prerelease-00718-04
Pat Gavlin [Fri, 19 Aug 2016 17:44:46 +0000 (10:44 -0700)]
Implement the proposed design for RyuJIT's LIR. (#6689)
These changes implement the design for RyuJIT's LIR described in https://github.com/dotnet/coreclr/blob/master/Documentation/design-docs/removing-embedded-statements.md.
The following passes required changes:
Rationalize, which has been almost completely rewritten
Long decomposition
Target-independent lowering
Target-dependent lowering
LSRA
Liveness
Flowgraph optimization
Codegen
For the most part, these changes are confined to the backend. Common code that needed to be updated included liveness, flowgraph optimization, and a few miscellaneous utilities.
The utilities used to analyze and manipulate LIR live (almost) entirely in src/jit/lir.{cpp,h}. The core concepts that are unique to LIR are LIR::Use and LIR::Range. The latter is a tuple that captures an SDSU def (i.e. an LIR node) and its corresponding use->def edge and user. The former serves to abstract a self-contained sequence of LIR nodes that make up e.g. the contents of a basic block.
Testing indicates that neither JIT throughput nor code quality are significantly impacted by these changes.
Hannah Zhang [Fri, 19 Aug 2016 17:22:17 +0000 (10:22 -0700)]
Added constructor to System.Version
Sepideh Khoshnood [Fri, 19 Aug 2016 17:14:47 +0000 (10:14 -0700)]
Expose missing members of System.Reflection in contract to prep for dotnet/corefx#1053 (#6614)
* Add missing Reflection APIs, Misc
* Add missing Reflection APIs, System.Reflection.Assembly
* Add missing Reflection APIs, some AssemblyAttributes, AssemblyName, AssemblyNameFlags
* Add missing Reflection APIs, ConstructorInfo, CustomAttributeData, CustomAttributeNamedArgument, CustomAttributeTypedArgument
* Add missing Reflection APIs, EventInfo, FieldInfo
* Add missing Reflection APIs, IReflect, ImageFileMachine, ExceptionHandlingClause
* Add missing Reflection APIs, ParameterInfo, PropertyInfo, Pointer, ResourceAttributes
* Add missing Reflection APIs, MethodBase, MethodBody, MethodInfo
* Fix some issues and add few more missing APIs
* Add more missing APIs
* Add rest from System.Reflection.AssemblyName
* Add serialization features, AssemblyName, Pointer, Missing
* Add more from System.Reflection.Module and add System.ModuleHandle
* Add equality operators to System.Reflection.Module* Add equality operators to System.Reflection.MemberInfo and System.Type
* Add some of missing APIs from System.Type
* Add some missing changes for System.Type
Stephen Toub [Fri, 19 Aug 2016 12:01:38 +0000 (08:01 -0400)]
Fix WeakReference in mscorlib ref assembly
It's missing the ISerializable implementation.
Aditya Mandaleeka [Fri, 19 Aug 2016 01:13:52 +0000 (18:13 -0700)]
Print info from core dumps on OS X using LLDB.
Carol Eidt [Thu, 18 Aug 2016 23:50:29 +0000 (16:50 -0700)]
Merge pull request #6780 from CarolEidt/Pre1stClassStructBlkOps
Preparatory changes for Blk Ops IR
Pat Gavlin [Thu, 18 Aug 2016 23:09:29 +0000 (16:09 -0700)]
Merge pull request #6797 from dotnet-bot/from-tfs
Merge changes from TFS
Koundinya Veluri [Thu, 18 Aug 2016 22:06:27 +0000 (15:06 -0700)]
Merge pull request #6774 from kouvel/Apis
Expose some threading type members
Bryan P. Arant [Thu, 18 Aug 2016 22:03:52 +0000 (15:03 -0700)]
Merge pull request #6765 from bryanar/cat_call
printing dumpling url at dumpling upload time
Carol Eidt [Tue, 16 Aug 2016 20:42:50 +0000 (13:42 -0700)]
Preparatory changes for Blk Ops IR
These are mostly refactoring changes, in preparation for the change to the IR for block assignments.
Rama krishnan Raghupathy [Thu, 18 Aug 2016 20:39:11 +0000 (13:39 -0700)]
Merge pull request #6782 from ramarag/testjson
Restoring the Test Runtime libraries from a single project.json
Fixes #5174
Stephen Toub [Thu, 18 Aug 2016 20:17:49 +0000 (16:17 -0400)]
Merge pull request #6796 from stephentoub/fix_serialization_issues
Fix several serialization issues
Koundinya Veluri [Mon, 15 Aug 2016 18:17:48 +0000 (11:17 -0700)]
Expose some threading type members
- Exposed some public members of WaitHandle
- Fixed "path too long" error message to take a length parameter (as in mscorlib.txt) and uses of it to pass in the correct value for the length
- Made SignalAndWait implementation available on Windows, and throw PlatformNotSupportedException on others
Part of dotnet/corefx#10933
Aditya Mandaleeka [Thu, 18 Aug 2016 19:53:00 +0000 (12:53 -0700)]
Enable additional functionality for OS X CI dumps.
Stephen Toub [Thu, 18 Aug 2016 18:14:02 +0000 (14:14 -0400)]
Remove TypeForwardedFrom attributes for System and System.Core
These assemblies don't exist in .NET Core. Having these attributes on serializable types causes deserialization for such types to fail, as it attempts to load the assembly from which they were forwarded.
Stephen Toub [Thu, 18 Aug 2016 18:06:39 +0000 (14:06 -0400)]
Fix serialization of delegates without targets
A null check is missing when FEATURE_REMOTING is not defined, resulting in an invalid cast exception when deserializing a delegate without a target (e.g. a delegate for a static method).
Stephen Toub [Thu, 18 Aug 2016 18:05:35 +0000 (14:05 -0400)]
Fix serialization of several exception types
VerificationException's deserialization ctor was being tree-shaken away.
SecurityException's deserialization ctor was ifdef'd out.
Joseph Tremoulet [Thu, 18 Aug 2016 15:02:56 +0000 (11:02 -0400)]
Merge pull request #6747 from JosephTremoulet/VNLoops
Record expr location for opqaue value numbers
Viacheslav Nikolaev [Thu, 18 Aug 2016 12:12:09 +0000 (15:12 +0300)]
An easy fix for #6760 (#6770)
As mentioned in #6760 we can postpone the calculation of the NeedsTurkishCasing property.
Conflicts:
src/mscorlib/corefx/System/Globalization/TextInfo.Unix.cs
Rahul Kumar [Thu, 18 Aug 2016 02:37:23 +0000 (19:37 -0700)]
Due to comdat folding some fcalls are pointing to same code. This fix ensures that these fcall has a unique code address. (#6776)
Rama Krishnan Raghupathy [Thu, 18 Aug 2016 01:03:39 +0000 (18:03 -0700)]
Restoring the Test Runtime libraries from a single project.json
This will ensure the expected version to be present at the runtime
Koundinya Veluri [Wed, 17 Aug 2016 21:48:59 +0000 (14:48 -0700)]
Fix CRWLock ID generation overflow bug
There are several problems with the CRWLock constructor, where it tries to increment a 64-bit lock ID in two 32-bit pieces using lock-free code in a fast path:
- For the lower 32 bits, LLockID, zero is reserved for identifying a free lock entry. Upon incrementing LLockID from -1 to 0, the code was not skipping zero or incrementing the upper 32 bits, ULockID. Instead, it waited for one more increment to LLockID before increment ULockID. Assignment of the invalid LLockID caused a free lock entry to be reused for a new lock, and corruption in lock state thereafter. This is the issue identified by the bug.
- If LLockID != -1 in thread A, but the compare-exchange fails because thread B had updated the value before thread A, ULockID was not being read again in thread A. If thread B had also incremented ULockID, thread A would continue to use the old ULockID.
- In the locked slow path that handles incrementing ULockID, ULockID is updated before LLockID. The constructor though, was reading ULockID before LLockID. This allows a race where Thread A reads an old value for ULockID and a new value for LLockID, increments LLockID successfully, and continues to use the old ULockID.
Due to the availability of InterlockedCompareExchange64 on all supported platforms, we decided to use that instead to simplify the solution and fix all of the above issues at the same time.
Bug: 242568
Integrated from changes: 1621197, 1621810
[tfs-changeset: 1622814]
Jarret Shook [Wed, 17 Aug 2016 20:05:42 +0000 (13:05 -0700)]
Merge pull request #6773 from jashook/disable_jit32_timeouts_round_two
Disable tests that timeout for jit32
Rahul Kumar [Wed, 17 Aug 2016 18:00:15 +0000 (11:00 -0700)]
Merge pull request #6748 from rahku/arm64ShadowHeap
update GC shadow heap as part of writebarrier jit helper
jashoo [Wed, 17 Aug 2016 17:47:26 +0000 (10:47 -0700)]
Disable tests that timeout for jit32
See issues #6720, #6718, #6714 for more information.
Wes Haggard [Wed, 17 Aug 2016 15:29:26 +0000 (08:29 -0700)]
Merge pull request #6766 from dotnet-bot/UpdateDependencies20160817062015
Updating External dependencies to beta-24417-00
Jonghyun Park [Wed, 17 Aug 2016 06:32:41 +0000 (15:32 +0900)]
Makes FCallMemcpy unwindable (#6753)
FCallMemcpy function in vm/arm/memcpy.S creates an stack frame that
libunwind cannot recognize, which leads to the issue in #6752.
This commit makes the stack frame of FCallMemcpy unwindable to fix
issue #6752.
dotnet-bot [Wed, 17 Aug 2016 06:20:15 +0000 (06:20 +0000)]
Updating External dependencies to beta-24417-00
Hyung-Kyu Choi [Wed, 17 Aug 2016 04:03:30 +0000 (13:03 +0900)]
Apply correct symbolic name _ARM_ instead of _TARGET_ARM_ (#6756)
Signed-off-by: Hyung-Kyu Choi <hk0110.choi@samsung.com>
Bryan P. Arant [Wed, 17 Aug 2016 00:25:27 +0000 (17:25 -0700)]
printing dumpling url at upload time.
Rama krishnan Raghupathy [Wed, 17 Aug 2016 01:44:22 +0000 (18:44 -0700)]
Merge pull request #6749 from ramarag/sos_misc
Arm64 Sos fixes for
Jan Kotas [Wed, 17 Aug 2016 01:33:14 +0000 (18:33 -0700)]
Merge pull request #6763 from dotnet-bot/from-tfs
Merge changes from TFS
Tarek Mahmoud Sayed [Wed, 17 Aug 2016 00:36:06 +0000 (17:36 -0700)]
Merge pull request #6362 from svick/culturedata-regionnames
Move RegionNames out of #ifs
Koundinya Veluri [Tue, 16 Aug 2016 20:14:29 +0000 (13:14 -0700)]
Fix build break from new alignments
- MSVC seems to require alignment specification to be on the declaration as well as the definition
- Ignore warning about padding parent struct due to __declspec(align()), as that is intentional
Original change PR: https://github.com/dotnet/coreclr/pull/6516
[tfs-changeset: 1622589]
Stephen Toub [Tue, 16 Aug 2016 19:42:38 +0000 (15:42 -0400)]
Merge pull request #6761 from stephentoub/mscorlib_deserialization
Add deserialization ctors to mscorlib ref assembly
Fadi Hanna [Tue, 16 Aug 2016 17:58:50 +0000 (10:58 -0700)]
Adding new generics tests (#6709)
Stephen Toub [Tue, 16 Aug 2016 17:14:07 +0000 (13:14 -0400)]
Add deserialization ctors to mscorlib ref assembly
In a previous PR, I'd added these to model.xml, but we're still consuming the mscorlib ref assembly in corefx, so we need these deserialization ctors there, too.
Gaurav Khanna [Tue, 16 Aug 2016 16:56:29 +0000 (09:56 -0700)]
Merge pull request #6743 from dotnet-bot/UpdateDependencies20160815051908
Updating External dependencies to beta-24415-00
Joseph Tremoulet [Sat, 13 Aug 2016 18:17:09 +0000 (14:17 -0400)]
Record expr location for opqaue value numbers
When value numbering sees an expression that is too complex, runs out of
compile-time budget for deep field/phi sequences, etc., it assigns
`VNForExpr()` for the expression whose value number is being computed.
This produces a new, unique, opaque value number that will compare as
equal to itself but not to any other value number.
This change updates value numbering to record the location of the
expression whose value such a value numbers is created to stand in for.
While attaching location information to value numbers in general must be
done with care regarding subtleties of correlating value numbers for
expressions in different locations that compute the same value, those
subtleties are not relevant in the case of these unique/opaque value
numbers, whose point is to stand in for the value of one particular
expression and whose value may be propagated by copies but will never be
computed by other redundant expressions.
Loop-invariant code hoisting is updated to take these locations into
account when determining loop-invariance; the opaque value number is
loop-invariant if the expression whose value it represents is outside the
loop.
The `VNForExpr()` method is updated to take a `BasicBlock*` to identify
the location, but the backing store only records the block's loop number,
since the loop-invariance check is the only consumer of this information
and doing so allows a more compact representation (in particular, we avoid
allocating `m_defs` backing storage for these value numbers, since they are
used for the "give up and be conservative" cases).
The `BasicBlock*` passed to `VNForExpr()` can be `nullptr` to still have
the prior semantics of a fully opaque value number that can't be proved
loop-invariant; this is used for a handful of cases where the budgeting
results are memoized and so expressions at different locations could end
up sharing opaque value numbers, as well as a few cases where the
`VNForExpr()` call is covering a rare corner case in a utility for which
the corresponding `BasicBlock` may not be handy.
Fixes #6303.
Rama Krishnan Raghupathy [Tue, 16 Aug 2016 01:34:24 +0000 (18:34 -0700)]
Arm64 Sos fixes for
!verify heap
!u
!u -gcinfo
Rahul Kumar [Tue, 16 Aug 2016 01:32:59 +0000 (18:32 -0700)]
update GC shadow heap as part of writebarrier jit helper
Stephen Toub [Tue, 16 Aug 2016 00:33:00 +0000 (20:33 -0400)]
Merge pull request #6423 from stephentoub/restore_serialization
Restore most serialization implementations in coreclr
Koundinya Veluri [Mon, 15 Aug 2016 18:52:40 +0000 (11:52 -0700)]
Merge pull request #6516 from benaadams/WorkerThreadStart
WorkerThreadStart volatile read+cmpxchg loop
Atsushi Kanamori [Mon, 15 Aug 2016 17:00:09 +0000 (10:00 -0700)]
Merge pull request #6745 from AtsushiKan/rv
Fix issue https://github.com/dotnet/coreclr/issues/6600
Atsushi Kanamori [Mon, 15 Aug 2016 13:52:29 +0000 (06:52 -0700)]
Fix issue https://github.com/dotnet/coreclr/issues/6600
Return parameters not accomodated in
Attribute.GetParentDefinition(ParameterInfo)
Stephen Toub [Sun, 24 Jul 2016 23:57:33 +0000 (19:57 -0400)]
Add deserialization ctors to model.xml
dotnet-bot [Mon, 15 Aug 2016 05:19:08 +0000 (05:19 +0000)]
Updating External dependencies to beta-24415-00
Stephen Toub [Sat, 23 Jul 2016 17:49:15 +0000 (10:49 -0700)]
Port serialization implementation to Unix globalization types
These will need to be tested. They are a straight port from the Windows implementation, but the implementation has diverged, and it's possible (likely) there are some differences I didn't catch.
Stephen Toub [Sat, 23 Jul 2016 04:53:10 +0000 (21:53 -0700)]
Ensure On* methods are not removed by BclRewriter
Stephen Toub [Sat, 23 Jul 2016 03:26:01 +0000 (20:26 -0700)]
Restore most serialization implementations in coreclr
I audited all uses of FEATURE_SERIALIZATION in coreclr and removed most of the ifdef'ing. The remaining uses of FEATURE_SERIALIZATION are primary related to actually implementing formatters, CAS-related types, and a few types I've left non-serializable (e.g. ExecutionContext) but that we can revisit later if needed.
Jan Kotas [Sun, 14 Aug 2016 06:15:57 +0000 (23:15 -0700)]
Merge pull request #6740 from dotnet-bot/from-tfs
Merge changes from TFS
Jan Kotas [Sun, 14 Aug 2016 01:52:58 +0000 (18:52 -0700)]
Fixup a few contracts
[tfs-changeset: 1622296]
Wes Haggard [Sat, 13 Aug 2016 07:05:33 +0000 (00:05 -0700)]
Merge pull request #6731 from weshaggard/FixSerializationPackageVersion
Update version of S.R.Serialization.Primitives to version 4.1.1
Rahul Kumar [Sat, 13 Aug 2016 06:09:19 +0000 (23:09 -0700)]
Merge pull request #6728 from rahku/typeloaderHang
Issue: If two threads try to load the same type it can cause clr to h…
Rahul Kumar [Sat, 13 Aug 2016 06:08:18 +0000 (23:08 -0700)]
Merge pull request #6726 from rahku/fixes
In intGCShadow() first reserve memory before committing
Wes Haggard [Sat, 13 Aug 2016 04:51:07 +0000 (21:51 -0700)]
Update version of S.R.Serialization.Primitives to version 4.1.1
Pat Gavlin [Sat, 13 Aug 2016 02:11:09 +0000 (19:11 -0700)]
Merge pull request #6729 from pgavlin/Reformat
Re-reformat the JIT sources.
Rahul Kumar [Sat, 13 Aug 2016 00:41:03 +0000 (17:41 -0700)]
review fixes
Rahul Kumar [Sat, 13 Aug 2016 00:29:02 +0000 (17:29 -0700)]
In intGCShadow() first reserve memory before committing. VirtualCommit does not accept NULL as input address
Pat Gavlin [Fri, 12 Aug 2016 23:51:22 +0000 (16:51 -0700)]
Reformat the JIT sources.
The sources were formatted using jit-format on Windows with the
following command lines in the following order:
`jit-format.cmd -o Windows -b Checked -a x64 -f`
`jit-format.cmd -o Windows -b Debug -a x64 -f`
`jit-format.cmd -o Windows -b Release -a x64 -f`
The versions of clang-format and clang-tidy that were invoked by
jit-format were:
clang-format version 3.8.1 (branches/release_38)
LLVM (http://llvm.org/):
LLVM version 3.8.1
Optimized build with assertions.
Built Jun 22 2016 (12:19:56).
Default target: x86_64-pc-windows-msvc
Host CPU: sandybridge
Pat Gavlin [Fri, 12 Aug 2016 23:34:18 +0000 (16:34 -0700)]
Massage the code a bit for jit-format w/ Clang 3.8.
Manually insert a few braces that would otherwise have been inserted
in the wrong place.