Carol Eidt [Tue, 19 Jul 2016 05:31:02 +0000 (22:31 -0700)]
Merge pull request #6227 from parjong/revert-6021-LessConservativeGtObj
Revert "Less conservative gt obj"
Andy Ayers [Tue, 19 Jul 2016 03:13:21 +0000 (20:13 -0700)]
Merge pull request #6324 from AndyAyersMS/ReplayCapturesProfitModel
Inliner: capture profit model estimate during replay
William Godbe [Mon, 18 Jul 2016 22:40:16 +0000 (15:40 -0700)]
Merge pull request #6322 from wtgodbe/packageFixes
Fix package build in build-packages.sh to use dotnet instead of corerun
wtgodbe [Mon, 18 Jul 2016 21:49:30 +0000 (14:49 -0700)]
Fix package build in build-packages.sh to use dotnet instead of corerun
Peter Jas [Mon, 18 Jul 2016 21:13:24 +0000 (00:13 +0300)]
Fix libc path for macOS El Capitan (#6302)
Dan Moseley [Mon, 18 Jul 2016 20:42:18 +0000 (13:42 -0700)]
Add to model.xml several members we wish to expose in the FX (#6310)
* Add to model.xml several members we wish to expose in the FX.
Done by pulling lines selectively from a model.xml created from the full
s.p.corelib.dll.
* Add parameter to build.cmd to build both mscorlib and its native image, then stop. Also dump any errors to console for CI.
* Fix up to date check of BCLRewriter: should build when model.xml is changed
Peter Jas [Mon, 18 Jul 2016 19:03:29 +0000 (22:03 +0300)]
Add Lazy<T>(T value) constructor (#6304)
Fix dotnet/corefx#2578
Kyungwoo Lee [Mon, 18 Jul 2016 15:49:37 +0000 (08:49 -0700)]
Merge pull request #6315 from kyulee1/disemail
ARM64: Temporarrily disable email notification
Kyungwoo Lee [Mon, 18 Jul 2016 15:46:45 +0000 (08:46 -0700)]
ARM64: Temporarrily disable email notification
Currently all Arm64 boxes are dead.
They need to be upgraded.
Jostein Kjønigsen [Mon, 18 Jul 2016 06:23:25 +0000 (08:23 +0200)]
Provide guidance for common build-failure scenario. (#6164)
Stephen Toub [Mon, 18 Jul 2016 05:44:15 +0000 (01:44 -0400)]
Add ICloneable back to several globalization types on Unix (#6307)
The globalization implementation used for corefx on Unix came from corert, which didn't have ICloneable, and thus the globalization types that implement ICloneable in desktop and coreclr on Windows don't on Unix. Now that ICloneable is coming back, we need these types to implement ICloneable appropriately.
James Ko [Mon, 18 Jul 2016 05:18:26 +0000 (01:18 -0400)]
Minimize buffer allocations in Stream.CopyTo for seekable streams (#4540)
The current implementation of Stream.CopyTo allocates a giant, 81920-byte buffer if no bufferSize parameter is passed. This is incredibly wasteful if the stream we're copying from can seek, because then we can use the Length and Position properties to determine how many bytes are left and allocate a buffer of that size.
Jonghyun Park [Tue, 12 Jul 2016 02:03:35 +0000 (11:03 +0900)]
Revert "Less conservative gt obj"
Jan Kotas [Sun, 17 Jul 2016 07:24:32 +0000 (09:24 +0200)]
GC update from CoreRT (#6305)
https://github.com/dotnet/corert/tree/master/src/Native/gc
bc9671edca37502794c01647215519d4254d3ad4
Aditya Mandaleeka [Fri, 15 Jul 2016 22:16:31 +0000 (15:16 -0700)]
Merge pull request #6299 from mairaw/patch-1
fix small typos on contributing guide
Maira Wenzel [Fri, 15 Jul 2016 21:50:54 +0000 (14:50 -0700)]
fix small typos
Sean Gillespie [Fri, 15 Jul 2016 21:15:29 +0000 (14:15 -0700)]
Merge pull request #6296 from swgillespie/vscode-gitignore
Add VS Code local settings folder to gitignore
Andy Ayers [Fri, 15 Jul 2016 20:06:44 +0000 (13:06 -0700)]
Inliner: capture profit model estimate during replay
When using ReplayPolicy and dumping data, capture the profitability
estimate made by the ModelPolicy, so we can more easily cross-validate
it against externally measured profitability.
William Godbe [Fri, 15 Jul 2016 19:50:58 +0000 (12:50 -0700)]
Merge pull request #6295 from wtgodbe/initToolsFix
Stop chmoding nonexistant corerun file in Tools dir
Sean Gillespie [Fri, 15 Jul 2016 19:00:17 +0000 (12:00 -0700)]
Add VS Code local settings folder to gitignore
wtgodbe [Fri, 15 Jul 2016 18:10:49 +0000 (11:10 -0700)]
Stop chmoding nonexistant corerun file in Tools dir
Sejong Oh [Fri, 15 Jul 2016 17:46:42 +0000 (10:46 -0700)]
Merge pull request #6037 from sejongoh/linux_struct_arg
Unecessary Linux stack argument copy
Jarret Shook [Fri, 15 Jul 2016 16:58:33 +0000 (09:58 -0700)]
Merge pull request #6249 from prajwal-aithal/devel/arm-ci-documentation
Documentation: Add documentation for building coreclr for emulator
Michal Strehovský [Fri, 15 Jul 2016 06:11:27 +0000 (23:11 -0700)]
Merge pull request #6287 from MichalStrehovsky/removeAssertOnMustExpand
Remove assert validating intrinsic expansion on mustExpand
Jarret Shook [Thu, 14 Jul 2016 22:58:09 +0000 (15:58 -0700)]
Merge pull request #6050 from prajwal-aithal/devel/arm-ci-runtests
ARM-CI: Add tests to CI script
Michal Strehovský [Thu, 14 Jul 2016 22:26:20 +0000 (15:26 -0700)]
Remove assert validating intrinsic expansion on mustExpand
The assert is incorrect. Consider expansion of
CORINFO_INTRINSIC_RTH_GetValueInternal, where the RuntimeTypeHandle was
created through a generic lookup helper call. The intrinsic expansion of
this sequence becomes the generic lookup helper call on it's own.
Sejong OH [Mon, 11 Jul 2016 20:44:09 +0000 (13:44 -0700)]
Remove unnecessary copies for stack struct argument except:
- struct is promoted to registers
- fgMakeOutgoingStructArgCopy is called
- tree pattern of obj - addr - simd operator
* call void foo
...
+--* obj simd16
| \--* addr byref
| | /--* lclVar simd16 V05 loc4
| \--* simd simd16 int -
| \--* lclVar simd16 V08 tmp1
...
William Godbe [Thu, 14 Jul 2016 20:20:44 +0000 (13:20 -0700)]
Merge pull request #6256 from wtgodbe/buildToolsVersion
Update to latest version of buildtools
wtgodbe [Wed, 13 Jul 2016 17:29:43 +0000 (10:29 -0700)]
Update to latest version of buildtools
Bruce Forstall [Thu, 14 Jul 2016 18:39:48 +0000 (11:39 -0700)]
Merge pull request #6276 from BruceForstall/LocAllocCleanup
Clean up localloc implementation
Bruce Forstall [Wed, 13 Jul 2016 23:00:35 +0000 (16:00 -0700)]
Clean up localloc implementation; improve AMD64 codegen for initialized locallocs
Brian Sullivan [Thu, 14 Jul 2016 14:14:50 +0000 (07:14 -0700)]
Merge pull request #6274 from hqueue/arm/xor_swap
Introduce XOR swap to legacy codegen for ARM32
Hyung-Kyu Choi [Thu, 14 Jul 2016 07:23:57 +0000 (16:23 +0900)]
Introduce XOR swap to legacy codegen for ARM32
Signed-off-by: Hyung-Kyu Choi <hk0110.choi@samsung.com>
Prajwal A N [Wed, 13 Jul 2016 04:12:12 +0000 (13:12 +0900)]
docs: Add building coreclr for emulator
Added documentation for building coreclr for the Linux ARM Emulator.
Signed-off-by: Prajwal A N <an.prajwal@samsung.com>
Stephen Toub [Thu, 14 Jul 2016 02:50:23 +0000 (22:50 -0400)]
Merge pull request #6263 from stephentoub/remove_serializable
Add #if FEATURE_SERIALIZATION around customized [Serializable]s
Rahul Kumar [Thu, 14 Jul 2016 02:26:40 +0000 (19:26 -0700)]
Merge pull request #6257 from rahku/crossgen
ARM64:Similar to arm, in order to preserve volatile semantics MemoryB…
Jan Vorlicek [Wed, 13 Jul 2016 22:54:42 +0000 (00:54 +0200)]
Make StubHelpers::ProcessByrefValidationList NOTHROW (#6258)
This change fixes StubHelpers::ProcessByrefValidationList so that it doesn't throw.
While there was a EX_TRY / EX_CATCH, the catch handler calls FormatValidationMessage
to format the message before aborting the process and that function can throw.
The fix is to wrap the inside of the EX_CATCH in one more EX_TRY / EX_CATCH and
abort the process without message if there is an exception in the FormatValidationMessage.
Stephen Toub [Wed, 13 Jul 2016 22:51:38 +0000 (18:51 -0400)]
Merge pull request #6205 from stephentoub/environmentaugments
Add EnvironmentAugments to coreclr
Carol Eidt [Wed, 13 Jul 2016 22:44:44 +0000 (15:44 -0700)]
Merge pull request #6182 from wateret/fix-6051
[ARM-softfp/Linux] Get precise type for struct
Stephen Toub [Wed, 13 Jul 2016 22:43:45 +0000 (18:43 -0400)]
Add #if FEATURE_SERIALIZATION around non-functional [Serializable]
We're adding BinaryFormatter to corefx. At present, ISerializable and friends are declared in corefx rather than being exported from System.Private.Corelib. As a result, there are a fair number of types in mscorlib that try to customize their serialization, but because they're doing so using interfaces that aren't exported (and in some cases, aren't even currently compiled in), a formatter outside of Corelib sees a type as serializable but then tries to serialize it using the default rules and fails.
This commit ifdefs the [Serializable] attribute on types that try to customize their serialization. If in the future we decide we want to expose ISerializable and friends from Corelib in order to enable the serialization of these types (which we'll likely want to), e.g Delegate, Exception, etc. we'll need to turn on FEATURE_SERIALIZATION for coreclr, and these will all come back automatically.
William Godbe [Wed, 13 Jul 2016 22:14:26 +0000 (15:14 -0700)]
Merge pull request #6255 from wtgodbe/targetingpack
Build the TargetingPack package as part of build-packages scripts
Pat Gavlin [Wed, 13 Jul 2016 22:11:13 +0000 (15:11 -0700)]
Merge pull request #6254 from pgavlin/RemoveRenameUpdatedVars
Remove RenameUpdatedVars.
Stephen Toub [Wed, 13 Jul 2016 20:51:54 +0000 (16:51 -0400)]
Conditionally compile EnvironmentAugments.cs for coreclr
Stephen Toub [Sat, 9 Jul 2016 21:46:12 +0000 (17:46 -0400)]
Add EnvironmentAugments to coreclr
Bruce Forstall [Wed, 13 Jul 2016 20:30:33 +0000 (13:30 -0700)]
Merge pull request #6251 from BruceForstall/Localloc
Implement localloc for RyuJIT/x86
Bruce Forstall [Tue, 12 Jul 2016 23:38:17 +0000 (16:38 -0700)]
Implement localloc for RyuJIT/x86
Fixes #4182
Change RyuJIT to generate localloc code for x86 the same way legacy JIT does.
E.g., use a push/dec loop for initialized memory. And avoid using "sub esp"
to keep the emitter from tracking stack pointer adjustments.
Bruce Forstall [Wed, 13 Jul 2016 20:15:02 +0000 (13:15 -0700)]
Merge pull request #6224 from BruceForstall/Fix4177
Fix #4177: RyuJIT/x86 NYI related to lvaArgType()
Bruce Forstall [Mon, 11 Jul 2016 23:44:41 +0000 (16:44 -0700)]
Fix #4177: RyuJIT/x86 NYI related to lvaArgType()
For x86, this function can simply return the variable type. There is
no need to map struct types. The function actually will never be
called on x86 for struct types because it is only called for register
arguments.
This NYI was only hit in the case of JMP calls. I fixed two cases
in the JMP implementation where FEATURE_VARARGS code assumed AMD64, due
to the Windows AMD64 ABI convention to pass float varargs values in both
integer and floating point registers. I put that under `#ifdef _TARGET_AMD64_`.
This removes 426 NYI.
Stephen Toub [Wed, 13 Jul 2016 19:21:36 +0000 (15:21 -0400)]
Merge pull request #6253 from stephentoub/stop_stripping_serializable
Stop BclRewriter from trimming serializable metadata
Rahul Kumar [Wed, 13 Jul 2016 19:06:44 +0000 (12:06 -0700)]
ARM64:Similar to arm, in order to preserve volatile semantics MemoryBarrier instructions needs to be inserted. Otherwise writes to memory locations can be re-ordered. For eg. In function leave_spin_lock() we set holding_thread to -1 and then set pSpinLock->lock to -1. Without memorybarrier it is possible for different threads to see pSpinLock->lock new value of -1 and old value of holding_thread. This causes assertion pSpinLock->holding_thread == (Thread*)-1 in function enter_spin_lock() to fire
Matt Ellis [Wed, 13 Jul 2016 17:50:26 +0000 (10:50 -0700)]
Merge pull request #6247 from ellismg/add-back-exception-members
Add back some surface area to thinned corelib
wtgodbe [Wed, 13 Jul 2016 17:21:50 +0000 (10:21 -0700)]
build the TargetingPack package as part of build-packages scripts
Stephen Toub [Wed, 13 Jul 2016 15:36:34 +0000 (11:36 -0400)]
Stop BclRewriter from trimming serializable metadata
Pat Gavlin [Tue, 12 Jul 2016 21:48:52 +0000 (14:48 -0700)]
Remove RenameUpdatedVars.
The variable renaming performed by rationalize was originally intended
(according to its comments) to allow assignments that were embedded in
subtrees to be hoisted out into a preceding statement. It appears that
this logic has been removed, so the rename pass is no longer needed.
Hanjoung Lee [Wed, 13 Jul 2016 10:39:21 +0000 (19:39 +0900)]
[ARM-softfp/Linux] Get precise type for pointer-size struct
Compiler::getPrimitiveTypeForStruct may return TYP_FLOAT but is off on softfp.
So we need to check that in another way rather than `IsHfa()`.
Otherwise struct types with single float32 will make inconsistency in gentree.
Which ends up with assert failure "Bad IL: Illegal assignment of float into integer!".
Check is done in Compiler::getPrimitiveTypeForStruct, by isSingleFloat32Struct.
Fix #6051
Rahul Kumar [Wed, 13 Jul 2016 01:08:53 +0000 (18:08 -0700)]
Merge pull request #6246 from rahku/crossgen
Arm64: remove gcstress_fail tag from certain tests in tests.lst
Matt Ellis [Wed, 13 Jul 2016 01:08:37 +0000 (18:08 -0700)]
Add back some surface area to thinned corelib
We are working on expsosing some existing exception types from
System.Runtime in CoreFX. The implementations will facade over
System.Private.CoreLib, but we need to make sure all the members we expect
to expose from the contract are present in the implementation assembly.
Expose the missing members for ExecutionEngineException,
NotFiniteNumberException and ExternalException.
Andy Ayers [Wed, 13 Jul 2016 01:06:43 +0000 (18:06 -0700)]
Merge pull request #6244 from AndyAyersMS/FixDataCollection
Inliner: Update data collection by targeting a single inline
Rahul Kumar [Wed, 13 Jul 2016 01:05:12 +0000 (18:05 -0700)]
Arm64: remove gcstress_fail tag from certain tests in tests.lst
Rahul Kumar [Wed, 13 Jul 2016 01:01:22 +0000 (18:01 -0700)]
Merge pull request #6226 from rahku/crossgen
ARM64: In GetDelegateTarget a portion of the _WIN64 code was under _T…
Matt Ellis [Wed, 13 Jul 2016 00:58:02 +0000 (17:58 -0700)]
Suppress Interactive Shell when running PS
On recent builds of Windows/PowerShell it seems like PS wants to write a
prompt after invoking the script, which causes a problem because that puts
extra gunk into the generated source file from h2inc, which causes
downstream failures.
Pat Gavlin [Wed, 13 Jul 2016 00:10:37 +0000 (17:10 -0700)]
Merge pull request #6225 from pgavlin/RationalizeCleanup
Refactor top-level comma rewriting.
Pat Gavlin [Wed, 13 Jul 2016 00:09:58 +0000 (17:09 -0700)]
Add a comment at the use of RewriteTopLevelComma.
Andy Ayers [Tue, 12 Jul 2016 23:12:46 +0000 (16:12 -0700)]
Inliner: Update data collection by targeting a single inline
The inliner currently will record detailed data about the last successful
inline performed (given a build with DEBUG or INLINE_DATA defined).
However, for purposes of inline profitability analysis we might be more
interested in the data from an earlier inline.
This change creates a mechanism where the replay log can flag one inline
per method as the target of data collection. The inliner checks for this
attribute during replay and captures that inline's data.
Pat Gavlin [Mon, 11 Jul 2016 22:03:42 +0000 (15:03 -0700)]
Refactor top-level comma rewriting.
Top-level commas were being handled in three separate parts of
rationalize: TreeTransformRationalization, RewriteSimpleTransforms,
and DoPhase. Furthermore, the transform performed by the latter
was distinct from the transforms performed by the former two parts,
as those parts called out to RewriteTopLevelComma in order to do
the necessary rewrite. This change collects the two transforms
in RewriteTopLevelComma and centralizes the comma handling in
RewriteSimpleTransforms.
With the comma handling centralized in RewriteSimpleTransforms,
TreeTransformRationalization was not doing much of anything, so
it has been removed and RewriteSimpleTransforms renamed to
TreeTransformRationalization.
chunseoklee [Tue, 12 Jul 2016 21:28:26 +0000 (06:28 +0900)]
active exception regression fix (#6117)
This patch fixes regression caused by 597e160 commit.
Previous PAL_VirtualUnwind does not set CONTEXT_UNWOUND_TO_CALL
properly. In this patch, the flag is added for non-signaled exception.
Mike McLaughlin [Tue, 12 Jul 2016 21:12:45 +0000 (14:12 -0700)]
Fix !PrintException (pe) -lines bug where line number/source info is not being displayed. (#6241)
The argument bLineNumbers to FormatGeneratedException was being used a local flag.
Dan Moseley [Tue, 12 Jul 2016 20:26:53 +0000 (13:26 -0700)]
Expose missing members of Array to prep for dotnet/corefx#9998. (#6230)
* Expose missing members of Array to prep for dotnet/corefx#9998. modelgen.exe is helpful.
* Remove dead code
* Implement Array.LongLength.
* Revert "Implement Array.LongLength."
This reverts commit
fdf7e96172d3b816551b827b6a45fbc692b0551d.
* Revert "Remove dead code"
This reverts commit
432087c2d6f826db93fa30a07476dc2decb12633.
* Expose Array.GetLongLength from VM
Rahul Kumar [Tue, 12 Jul 2016 01:10:13 +0000 (18:10 -0700)]
ARM64: In GetDelegateTarget a portion of the _WIN64 code was under _TARGET_AMD64_ whereas at other places (like NDirectStubLinker::DoNDirect & InlinedCallFrame::HasFunction ) related code is under _WIN64. Make the code in GetDelegateTarget to be in sync with other functions and remove ifdef _TARGET_ADM64_. Without this change for arm64 InlinedCallFrame::m_datum stores unmanaged target pointer but StackWalker assumes it to be a NDirectMethodDesc and tries to dereference it and crashes.
Bruce Forstall [Tue, 12 Jul 2016 18:16:22 +0000 (11:16 -0700)]
Merge pull request #6137 from papaslavik/int_softfp_fix
Fix for #5877: pre-spill just the floating arguments
Bruce Forstall [Tue, 12 Jul 2016 18:14:56 +0000 (11:14 -0700)]
Merge pull request #5966 from mikedn/x86-neg-long
Implement GT_NEG decomposition
Brian Sullivan [Tue, 12 Jul 2016 17:06:13 +0000 (10:06 -0700)]
Merge pull request #6198 from briansull/multireg-refactor
Refactor and cleanup work for passing and returning struct types
Jarret Shook [Tue, 12 Jul 2016 14:31:50 +0000 (07:31 -0700)]
Merge pull request #5948 from jyoungyun/fix/issue_5947
Add testsRunningInsideARM.txt
Jan Vorlicek [Tue, 12 Jul 2016 07:44:15 +0000 (09:44 +0200)]
Fix exceptions in GC wave 2 (#6220)
I've removed blocks with PAL_TRY that were under CATCH_GC. I also had to
fix contract problem in GCToEEInterface::GcStartWork that was marked as
THROWS due to StubHelpers::ProcessByrefValidationList being marked as
THROWS, but the StubHelpers::ProcessByrefValidationList in fact doesn't
throw since it has body wrapped in EX_TRY / EX_CATCH.
This also fixes a problem that started to appear after my previous exceptions
fix change.
Mike Danes [Thu, 23 Jun 2016 19:02:50 +0000 (22:02 +0300)]
Implement GT_NEG decomposition
Brian Sullivan [Thu, 7 Jul 2016 20:23:20 +0000 (13:23 -0700)]
Refactor and cleanup work for passing and returning struct types
Replaces argOrReturnTypeForStruct with two new methods:
getArgTypeForStruct - Provides information on how to pass a struct type
getReturnTypeForStruct - Provides information on how to return a struct type
A struct can be passed or return in the following different ways:
1. A struct type may be passed/returned as a primitive type when its size is small
2. A struct type may be passed/returned by reference to a copy
3. A struct type may be passed/returned by value using multiple registers
4. A struct type may be passed by value using a copy on the stack
Incorporated code review feedback with expanded comments.
Aditya Mandaleeka [Mon, 11 Jul 2016 22:11:03 +0000 (15:11 -0700)]
Merge pull request #6215 from dotnet-bot/UpdateDependencies20160711061707
Updating External dependencies to beta-24311-00
Pat Gavlin [Mon, 11 Jul 2016 22:00:29 +0000 (15:00 -0700)]
Merge pull request #6221 from pgavlin/RemoveDeadCodeInRationalize
Remove dead code in Rationalize.
Pat Gavlin [Mon, 11 Jul 2016 20:38:30 +0000 (13:38 -0700)]
Remove dead code in Rationalize.
Most of the removed code deals with QMARK rewriting. These nodes
are now rewritten as part of fgMorph, so the rationalizer does not
need to deal with them (other than asserting that they do not
exist).
Other dead code:
- The `Rationalizations` type
- A few split-related declarations and helpers
Andy Ayers [Mon, 11 Jul 2016 18:10:56 +0000 (11:10 -0700)]
Merge pull request #6204 from AndyAyersMS/ModelPolicyDepthCheck
Inliner: add depth check to ModelPolicy
Aditya Mandaleeka [Mon, 11 Jul 2016 17:27:00 +0000 (10:27 -0700)]
Merge pull request #6216 from hqueue/fix/typo_20160710
Fix a misleading typo in comment
Bruce Forstall [Mon, 11 Jul 2016 17:20:03 +0000 (10:20 -0700)]
Merge pull request #6202 from BruceForstall/RefactorDecomp
Refactor RyuJIT/x86 long decomposition code
Hyung-Kyu Choi [Mon, 11 Jul 2016 14:40:38 +0000 (23:40 +0900)]
Fix a misleading typo in comment
Signed-off-by: Hyung-Kyu Choi <hk0110.choi@samsung.com>
James Ko [Mon, 11 Jul 2016 12:30:13 +0000 (08:30 -0400)]
Avoid repeated virtual method calls in List.Contains (#6212)
dotnet-bot [Mon, 11 Jul 2016 06:17:06 +0000 (06:17 +0000)]
Updating External dependencies to beta-24311-00
Bruce Forstall [Sat, 9 Jul 2016 01:38:59 +0000 (18:38 -0700)]
Refactor RyuJIT/x86 long decomposition code
All decomposition code is now in decompose.cpp, in a Decompose class.
Each node type that is decomposed has its own member function to do
the decomposition. Various helpers have been added to reduce code
duplication. In general, though, the code is as it was before.
Andy Ayers [Sat, 9 Jul 2016 17:58:02 +0000 (10:58 -0700)]
Inliner: add depth check to ModelPolicy
Give the ModelPolicy the standard depth bailout logic. This mainly
serves to prevent runaway inlining for recursive methods marked with
AggressiveInlining attributes.
Andy Ayers [Sat, 9 Jul 2016 15:47:14 +0000 (08:47 -0700)]
Merge pull request #6201 from AndyAyersMS/LowerCallsiteWeight
Inliner: updates to ModelPolicy
Brian Sullivan [Sat, 9 Jul 2016 02:14:10 +0000 (19:14 -0700)]
Merge pull request #6190 from dotnet-bot/from-tfs
Merge changes from TFS
Andy Ayers [Fri, 8 Jul 2016 00:53:43 +0000 (17:53 -0700)]
Inliner: updates to ModelPolicy
Updates to bring CS and TP impact of the ModelPolicy into more acceptable
ranges.
For CS, reduce the call site weights to values that are more in keeping
with the legacy policy weights. Local test runs show this does not
drastically alter CQ and brings CS down below LegacyPolicy levels, on
average.
Implement an early out rejection based solely on ILSize. The threshold
value is set by conservatively determining when ILSize alone indicates
the method in question will never be inlined (note the policy itself
does not have an explicit ILSize cutoff). See comments for
`ModelPolicy::NoteInt` for details. Note if we adjust the model's size
and profitability estimates, this threshould will also need updating.
CQ (as measured by the CoreCLR perf tests) continues to show about a
2.5% geomean improvement over LegacyPolicy.
Jan Vorlicek [Fri, 8 Jul 2016 23:56:00 +0000 (01:56 +0200)]
Fix exceptions in GC (#6192)
The GC code in general doesn't expect exceptions to be thrown from functions
it calls. However, the event creation functions were in fact throwing, so
in case of OOM, exception would be thrown from various places of the code.
This change creates non-throwing versions of the event creation functions
and replaces all usages of the throwing ones in the GC.
In addition, I've removed blocks with PAL_TRY that were under NO_CATCH_HANDLERS
define and also added Object::SetAppDomainNoThrow, which allowed me to removed
EX_TRY from GCHeap::StressHeap where the Object::SetAppDomain was called before.
Bruce Forstall [Fri, 8 Jul 2016 23:33:19 +0000 (16:33 -0700)]
Merge pull request #6193 from BruceForstall/FixTypos
Fix some typos in comments
Rahul Kumar [Fri, 8 Jul 2016 23:30:51 +0000 (16:30 -0700)]
Merge pull request #6195 from rahku/crossgen
Fix SOS bpmd for windows
William Godbe [Fri, 8 Jul 2016 23:26:18 +0000 (16:26 -0700)]
Merge pull request #6197 from wtgodbe/heapVerify
Add new scenario to code coverage job
wtgodbe [Fri, 8 Jul 2016 22:49:32 +0000 (15:49 -0700)]
Add new scenario to code coverage job
Sivarv [Fri, 8 Jul 2016 22:15:39 +0000 (15:15 -0700)]
Merge pull request #6176 from sivarv/lsraTodo
Optimize use of a reg-optional lclVar that is marked for both reload and spillAfter
Rahul Kumar [Fri, 8 Jul 2016 22:03:41 +0000 (15:03 -0700)]
Fix SOS bpmd for windows
Bruce Forstall [Fri, 8 Jul 2016 21:24:24 +0000 (14:24 -0700)]
Fix some typos in comments
Tarek Mahmoud Sayed [Fri, 8 Jul 2016 18:47:55 +0000 (11:47 -0700)]
Fixing the build break/unsigned to signed warning.
This checkin onbehalf of Tijoytk
[tfs-changeset: 1616784]
sivarv [Fri, 8 Jul 2016 00:19:28 +0000 (17:19 -0700)]
Optimize use of a reg-optional lclVar that is marked for both reload
and spillAfter.