Andy Ayers [Tue, 12 Jun 2018 16:49:36 +0000 (09:49 -0700)]
JIT: finally cloning -- handle fall through out of try (#18416)
When searching for a handler to clone, handle the case where
control can fall out of the try into a callfinally pair.
Closes #18353.
Carol Eidt [Tue, 12 Jun 2018 15:38:56 +0000 (08:38 -0700)]
Unify struct arg handling (#18358)
* Unify struct arg handling
Eliminate unnecessary struct copies, especially on Linux, and reduce code duplication.
Across all targets, use GT_FIELD_LIST to pass promoted structs on stack, and avoid
requiring a copy and/or marking `lvDoNotEnregister` for those cases.
Unify the specification of multi-reg args:
- numRegs now indicates the actual number of reg args (not the size in pointer-size units)
- regNums contains all the arg register numbers
Igor Kulaychuk [Tue, 12 Jun 2018 09:10:25 +0000 (12:10 +0300)]
Fix unwinding of funclet with no epilog on x86/Linux (#17458)
Tanner Gooding [Tue, 12 Jun 2018 02:58:34 +0000 (19:58 -0700)]
Adding containment support to more x86 HWIntrinsics (#18349)
* Adding containment support for the Sse42.Crc32 intrinsic
* Adding containment support for the x86 Compare*OrderedScalar and Compare*UnorderedScalar HWIntrinsics
* Adding containment support to several intrinsics that were marked NoContainment
Pent Ploompuu [Tue, 12 Jun 2018 01:01:14 +0000 (04:01 +0300)]
Fix Decimal.NearNegativeZero/NearPositiveZero (#18424)
Dan Moseley [Mon, 11 Jun 2018 21:42:35 +0000 (14:42 -0700)]
Reword invalid program exception message (#18409)
* Reword invalid program exception message
* Reword
* Update msg
Sergey Andreenko [Mon, 11 Jun 2018 21:29:17 +0000 (14:29 -0700)]
skip empty regMask in verifyRegistersUsed (#18415)
Stephen Toub [Mon, 11 Jun 2018 19:41:49 +0000 (15:41 -0400)]
Standardize on PLATFORM_WINDOWS instead of PLATFORM_UNIX (#18411)
There are a bunch of uses of PLATFORM_WINDOWS in System.Private.CoreLib, and just a few uses of PLATFORM_UNIX. Standardize on the former, removing all occurrences of the latter.
Tarek Mahmoud Sayed [Mon, 11 Jun 2018 19:23:26 +0000 (12:23 -0700)]
Fix UTF8 encoding fallback index (#18387)
* Fix UTF8 encoding fallback index
When falling back in UTF8 decoding, we’ll call the DecoderFallbackBuffer.InternalFallback method which takes a pointer to the source bytes pointer which point to where we have progressed inside the source bytes. The problem is before calling DecoderFallbackBuffer.InternalFallback we call GetBytesUnknown which can adjust the source byte pointer and DecoderFallbackBuffer.InternalFallback need to be called with the pointer before the adjustment. The fix is simply storing the original pointer value of the source byte pointer before calling GetBytesUnknown and then pass the stored original pointer to DecoderFallbackBuffer.InternalFallback instead of the adjusted source byte pointer
* Fix the comment
Luqun Lou [Mon, 11 Jun 2018 19:22:50 +0000 (12:22 -0700)]
WinRT attribute Cleanup (#18412)
Luqun Lou [Mon, 11 Jun 2018 18:29:46 +0000 (11:29 -0700)]
Remove WinRTSynchronizationContextFactoryBase (#18385)
Jarret Shook [Mon, 11 Jun 2018 18:28:26 +0000 (11:28 -0700)]
(Re)Enable NativeVarargs for CoreCLR (#18373)
* Enable NativeVarargs for CoreCLR
This will allow coreclr run programs which have native varargs.
It also re-enables the ArgIterator type for managed to managed native
vararg calls. This will allow the use of the __arglist keyword.
Limitations:
NYI statements have been added for all Unix Targets.
With this change __arglist (native varargs) will be supported, but not
yet tested on:
Amd64 Windows
x86 Windows
Arm32 Windows
Arm64 Windows
This change does not re-enable native vararg tests. This will be done
in a separate change.
Paulo Janotti [Mon, 11 Jun 2018 18:22:12 +0000 (11:22 -0700)]
Adds the CoreLib IL to packages (#18372)
* Adds the CoreLib IL to packages
The purpose of this is to enable code coverage runs to include CoreLib in CI - for that it is necessary to make the IL version available.
Chris Sienkiewicz [Mon, 11 Jun 2018 18:12:44 +0000 (11:12 -0700)]
Fix PCODE not having thumb bit set when comparing two addresses (#18247)
Anirudh Agnihotry [Mon, 11 Jun 2018 14:29:58 +0000 (07:29 -0700)]
Removed length restrictions on named synchronization primitives lengths on Windows (#18402)
* Removed Path Length check
* waitHandleNameMax removed
William Godbe [Sun, 10 Jun 2018 01:11:09 +0000 (18:11 -0700)]
Update Buildtools (#18380)
Jan Kotas [Sat, 9 Jun 2018 20:39:58 +0000 (13:39 -0700)]
Avoid NativeOverlapped pinning by allocating unmanaged memory for it (#18360)
It makes PinnableBufferCache unnecessary
Stephen Toub [Sat, 9 Jun 2018 20:34:46 +0000 (16:34 -0400)]
Use nameof of Encoding classes (#18395)
Bruce Forstall [Sat, 9 Jun 2018 19:12:37 +0000 (12:12 -0700)]
Merge pull request #18394 from BruceForstall/Fix10115
Re-enable tests for Windows arm64
Bruce Forstall [Sat, 9 Jun 2018 06:08:54 +0000 (23:08 -0700)]
Merge pull request #18363 from sdmaclea/PR-waitany-debug
Waitany test debug and cleanup
Maryam Ariyan [Thu, 7 Jun 2018 21:24:29 +0000 (14:24 -0700)]
Some extra changes needed in shared
Signed-off-by: dotnet-bot <dotnet-bot@microsoft.com>
Eric Erhardt [Sat, 9 Jun 2018 03:09:04 +0000 (22:09 -0500)]
Remove Substring in TimeZoneInfo parsing (#18378)
* Remove Substring in TimeZoneInfo parsing
Convert Substring to Slice to eliminate string allocations while parsing Posix daylight savings rules.
Stephen Toub [Sat, 9 Jun 2018 02:53:52 +0000 (22:53 -0400)]
Use cached strings for values 0 to 9 (#18383)
When doing ToString on single-digit integers, just return a cached string. This helps avoid string allocation/formatting costs for this very common case, while only costing one comparison for other cases.
Stephen Toub [Sat, 9 Jun 2018 02:41:27 +0000 (22:41 -0400)]
Remove ConcurrentDictionary from CoreLib (#18308)
Bruce Forstall [Sat, 9 Jun 2018 01:39:02 +0000 (18:39 -0700)]
Re-enable tests for Windows arm64
These tests don't fail anymore. Re-enable them.
Fixes #10115
Bruce Forstall [Sat, 9 Jun 2018 01:23:57 +0000 (18:23 -0700)]
Merge pull request #18382 from BruceForstall/MakeArmGcStress3Clean
Disable 3 tests incompatible with GCStress=3
Brian Robbins [Sat, 9 Jun 2018 00:44:11 +0000 (17:44 -0700)]
Fix EventPipe EventHandle Caching for TraceLogging (#18355)
Russ Keldorph [Fri, 8 Jun 2018 17:58:18 +0000 (10:58 -0700)]
Fix coredistools download script now that Packages layout has changed
Fixes #18321
Anirudh Agnihotry [Fri, 8 Jun 2018 21:51:10 +0000 (14:51 -0700)]
Moved Mutex.cs From src to shared (#18320)
* Moved Mutex.cs to shared
* Reduced #ifdef in mutex.cs
* Moved windows specific Implementation and renamed mincore to kernal32
* safeWaitHandle changed to _waitHandle
* moved using statement out of buffer namespace
Luqun Lou [Fri, 8 Jun 2018 21:41:24 +0000 (14:41 -0700)]
Expose public implementation AsyncCausalitySupport to S.R.WindowsRuntime (#18376)
Bruce Forstall [Fri, 8 Jun 2018 21:27:18 +0000 (14:27 -0700)]
Merge pull request #18287 from sdmaclea/TryClearGCStressInstructionUpdate
Use atomic ops in CommitGCStressInstructionUpdate()
Bruce Forstall [Fri, 8 Jun 2018 20:50:16 +0000 (13:50 -0700)]
Disable 3 tests incompatible with GCStress=3
The following fail in CI with arm Windows GCStress=3, and are now marked
`GCStressIncompatible`.
1. GC\Coverage\smalloom\smalloom.cmd
2. Issue #17480: tracing\tracevalidation\rundown\rundown\rundown.cmd, tracing\tracevalidation\jittingstarted\JittingStarted\JittingStarted.cmd
Steve MacLean [Fri, 8 Jun 2018 19:46:31 +0000 (15:46 -0400)]
Fix threading gcStress fails
Mark GCStressIncompatible ond/or fix synchronization logic
Amy [Fri, 8 Jun 2018 19:14:11 +0000 (12:14 -0700)]
Merge pull request #18326 from acmyu/gc
R2RDump - GcInfo and slot table
Sergey Andreenko [Fri, 8 Jun 2018 17:30:43 +0000 (10:30 -0700)]
Workaround for compiler.hpp (1848) - Assertion failed 'lvRefCnt' (#18292)
* add test
* clean fgRemoveStmt a bit
* fix the issue (more like a workaround).
Egor Chesakov [Fri, 8 Jun 2018 16:32:32 +0000 (09:32 -0700)]
Disable failing CoreFx tests on Windows/arm and Linux/arm (#18347)
* System.Dynamic.Runtime.Tests on Windows/arm and Linux/arm
* System.Linq.Expressions.Tests on Windows/arm (JitStress=2)
* System.Runtime.Numerics.Tests on Windows/arm (JitStress=1/JitStress=2)
Andy Ayers [Fri, 8 Jun 2018 06:16:56 +0000 (23:16 -0700)]
JIT: Fix bug in finally cloning caused by unsound callfinally reordering (#18348)
We need to make sure that if we reorder a callfinally during finally cloning
that the callfinally is actually the one being targeted by the last block in
the try range.
Closes #18332. Linked issue has some more detailed notes.
Pent Ploompuu [Fri, 8 Jun 2018 00:47:44 +0000 (03:47 +0300)]
Fixed and optimized Decimal.GetHashCode (#18288)
Brian Robbins [Thu, 7 Jun 2018 20:44:49 +0000 (13:44 -0700)]
Add missing #ifdef.
Signed-off-by: dotnet-bot-corefx-mirror <dotnet-bot@microsoft.com>
Aaron Robinson [Fri, 8 Jun 2018 00:23:08 +0000 (17:23 -0700)]
Interop Copy tests now all return 101 during fail rather than crash (#18310)
* Interop Copy tests now all return 101 during fail rather than crash
* Remove unused Assert APIs
Bruce Forstall [Fri, 8 Jun 2018 00:03:30 +0000 (17:03 -0700)]
Merge pull request #18354 from sdmaclea/PR-eventsourcetrace
Mark eventsourcetrace GCStressIncompatible
Bruce Forstall [Thu, 7 Jun 2018 23:58:03 +0000 (16:58 -0700)]
Merge pull request #18340 from mikedn/ins-desc
Increase instrDesc::_idIns width on XARCH
Carol Eidt [Thu, 7 Jun 2018 23:53:45 +0000 (16:53 -0700)]
Merge pull request #18356 from CarolEidt/Fix_GT_START_NONGC
GT_START_NONGC has no kill set
Amy Yu [Thu, 7 Jun 2018 23:38:10 +0000 (16:38 -0700)]
GcInfo belongs to method, not runtime function
Steve MacLean [Thu, 7 Jun 2018 22:59:31 +0000 (18:59 -0400)]
Waitany test debug and cleanup
Krzysztof Wicher [Thu, 7 Jun 2018 22:55:54 +0000 (15:55 -0700)]
Skip old time zones adjustments with offsets higher than 14h (#18305)
* Skip old time zones with offsets higher than 14h
* apply offline feedback
* fix braces
Carol Eidt [Thu, 7 Jun 2018 20:58:22 +0000 (13:58 -0700)]
GT_START_NONGC has no kill set
It was mistakenly added to the `GT_PROF_HOOK` case in the recent LSRA changes.
Steve MacLean [Thu, 7 Jun 2018 20:31:10 +0000 (16:31 -0400)]
Mark eventsourcetrace GCStressIncompatible
Noah Falk [Thu, 7 Jun 2018 19:53:08 +0000 (12:53 -0700)]
Allow ILCodeVersion to fallback to default IL (#18322)
For compat with profilers that used our APIs in unexpected ways we can allow
the ILCodeVersion to fallback to the default IL code when no IL was explicitly
given.
Maryam Ariyan [Thu, 7 Jun 2018 19:40:40 +0000 (12:40 -0700)]
Moves CurrentSystemTimeZone, ICastable, RegistryView and RegistryValueKind to shared (#18337)
* Moves CurrentSystemTimeZone to shared completely
* Moves ICastable to shared
* Moves RegistryView and RegistryValueKind to shared
* Moves ICastableHelpers to its own file
Tarek Mahmoud Sayed [Thu, 7 Jun 2018 18:01:26 +0000 (11:01 -0700)]
Force short date pattern to use yyyy on Linux (#18316)
* Force short date pattern to use yyyy on Linux
The default pattern we get is using yy which causes the years to be displayed as 2 digits. This is not acceptable for many users. The fix here is to force 4-digit year as a default and still keep the original pattern which has 2-digits year in the optional list
* Address the review feedback
Stephen Toub [Thu, 7 Jun 2018 17:24:11 +0000 (13:24 -0400)]
Remove dead case from string.Compare (#18345)
Marek Safar [Thu, 7 Jun 2018 17:07:47 +0000 (10:07 -0700)]
Guard Windows code under PLATFORM_WINDOWS (#18344)
Steve MacLean [Mon, 4 Jun 2018 17:55:34 +0000 (13:55 -0400)]
Use atomic ops in CommitGCStressInstructionUpdate()
Mike Danes [Thu, 7 Jun 2018 15:17:04 +0000 (18:17 +0300)]
Increase instrDesc::_idIns width on XARCH
XARCH exceeded the 512 instruction limit but it still has room for instruction formats (only 77 at the moment) so a bit was taken from _idInsFmt.
Also, cleaned up some tiny while instrDesc traces while trying to fix "Space taken up to here:" kind of comments.
Alfred Myers [Thu, 7 Jun 2018 07:06:45 +0000 (04:06 -0300)]
Fixed a couple of comments about ArgumentException on timeout parameters (#18335)
Stephen Toub [Thu, 7 Jun 2018 07:06:09 +0000 (03:06 -0400)]
Remove a dead branch from __DTString (#18334)
All call sites to the ctor validate that the DateTimeFormatInfo has been created.
Stephen Toub [Thu, 7 Jun 2018 07:04:30 +0000 (03:04 -0400)]
Delete FormatterConverter from CoreLib (#18336)
It's not used in the library. The type is implemented/exposed from corefx.
Egor Chesakov [Thu, 7 Jun 2018 06:49:48 +0000 (23:49 -0700)]
Cross-bitness support in JitInterface (#18329)
* Replace sizeof(tailCallType) / sizeof(tailCallType[0]) with _countof(tailCallType)
* Remove TransparentProxyObject and RealProxyObject
* InlinedCallFrame::GetEEInfo should not be used for R2R compilation
* Hardcode constants used in CEEInfo::getEEInfo
* Thread fields should not be used for R2R compilation
* Zero pEEInfoOut->inlinedCallFrameInfo when compiling R2R code
* Rename OFFSETOF__PtrArray__m_Array to avoid compilation conflicts with asmconstants
* Add assertions enabled in every non-crossbitness scenario
* Define and use CROSSBITNESS_COMPILE macro
mikedn [Thu, 7 Jun 2018 05:18:59 +0000 (08:18 +0300)]
Fix ARM cast codegen (#18063)
* Fix ARM cast codegen
ARM cast codegen is rather convoluted and sometimes does the wrong thing by applying GTF_UNSIGNED to the destination type even though this flag is only about the source type.
* Add more conversion tests
These tests are intended to support casts with contained operands. They're also useful to catch issues caused by load nodes having small types and generally improve test coverage for casts.
Tanner Gooding [Thu, 7 Jun 2018 01:24:04 +0000 (18:24 -0700)]
Adding containment support for more x86 hardware intrinsics (#18297)
* Adding containment support to one-operand scalar HWIntrinsics (x86)
* Adding containment support to two-operand imm HWIntrinsics (x86)
* Adding containment support to three-operand imm HWIntrinsics (x86)
* Updating hwintrinsiccodegenxarch to properly mask Sse41.Insert for TYP_FLOAT
* Updating the Sse41.Insert tests for TYP_FLOAT
* Adding containment support for Sse2.CompareLessThan and BlendVariable (Sse41/Avx/Avx2)
* Fixing `genHWIntrinsic_R_RM_I` to call `emitIns_SIMD_R_R_I`, rather than `emitIns_R_R_I`
* Updating emitOutputSV to not modify the code for IF_RWR_RRD_SRD_CNS
* Cleaning up some of the emitxarch code.
* Moving roundps and roundpd into the IsDstSrcImm check
Amy Yu [Wed, 6 Jun 2018 23:57:33 +0000 (16:57 -0700)]
Refactoring gcInfo, added verbose option to cli
Amy Yu [Wed, 6 Jun 2018 22:05:46 +0000 (15:05 -0700)]
Fix error cause by merge
Amy Yu [Wed, 6 Jun 2018 21:42:16 +0000 (14:42 -0700)]
Fix unwindInfo bug caused by missing alignmentPad, dump raw byes of unwindInfo and gcInfo
Amy Yu [Wed, 6 Jun 2018 17:56:06 +0000 (10:56 -0700)]
Use single variable for multiple tabs
Amy Yu [Wed, 6 Jun 2018 17:38:57 +0000 (10:38 -0700)]
Read slot table, dump GcInfo
Amy Yu [Tue, 5 Jun 2018 19:11:11 +0000 (12:11 -0700)]
Extract other fields from gcinfo
Amy Yu [Mon, 4 Jun 2018 23:01:13 +0000 (16:01 -0700)]
Extract gcinfo header, return kind and code length. Use code length as size of the runtime function
Amy Yu [Sat, 2 Jun 2018 00:04:48 +0000 (17:04 -0700)]
Extract gcinfo from image
Amy Yu [Mon, 4 Jun 2018 23:59:42 +0000 (16:59 -0700)]
Avoid passing R2RReader to R2RMethod and RuntimeFunction constructors
Amy Yu [Mon, 4 Jun 2018 20:49:13 +0000 (13:49 -0700)]
Use typeDef.Namespace to get full namespace, move to separate function
Amy [Wed, 6 Jun 2018 21:19:56 +0000 (14:19 -0700)]
R2RDump - Unwind Info (#18248)
* Extract unwind info from image
* Use typeDef.Namespace to get full namespace, move to separate function
* Avoid passing R2RReader to R2RMethod and RuntimeFunction constructors
* Dump unwind info
* Changes to unwindInfo output format
Jan Vorlicek [Wed, 6 Jun 2018 20:39:47 +0000 (22:39 +0200)]
Fix issue in Compiler::impImportStaticFieldAccess (#18328)
When CORINFO_FIELD_STATIC_SHARED_STATIC_HELPER accessor is being processed for
structs, the method creates a GT_IND node instead of GT_OBJ node. For structs
that can be normalized to a specific type (like TYP_SIMD8), it causes an assert
in Compiler::impNormStructVal in the switch case for GT_IND.
This change modifies the node creation to work the same way as in the default
case for accessors, creating GT_OBJ for structs and GT_IND otherwise.
Andy Ayers [Wed, 6 Jun 2018 19:37:09 +0000 (12:37 -0700)]
JIT: implement assembly name dissassembly filtering (#18306)
Add `COMPlus_JitDisasmAssemblies` to specify a whitelist of assemblies
whose methods will be disassembled (also filters out order, EH, GC, and
unwind info dumps too).
Intended use is for PMI asm diffing. Since PMI runs in process, current
disassembly will dissassemble various methods in PMI itself as well
as framework methods that did not get prejitted.
With this new config option we can restrict disassembly to just the assemblies
that we're processing via PMI.
Maryam Ariyan [Wed, 6 Jun 2018 17:15:19 +0000 (10:15 -0700)]
Moves FastResourceComparer and TaskAwaiter to shared (#18283)
* Reduces diff in BinaryReader
* Moves FastResourceComparer to shared
* Reduces diff in FileBasedResourceGroveler
* Minor comment diff in ManifestBasedResourceGroveler
* Reducing diff by removing extra blank lines in ResourceReader
* Reduces diff by renaming private fields in AssemblyName
* String to string and Object to object in AssemblyName
* Reduces diff in TaskAwaiter
* Moves TaskAwaiter to shared
* Apply code review feedback
* Fixes failure in corert
* Revert Renaming fields in AssemblyName
Koundinya Veluri [Wed, 6 Jun 2018 17:09:29 +0000 (10:09 -0700)]
Increase timeout and log the result for some WaitAny tests that expect AbandonedMutexException (#18312)
Follow-up to https://github.com/dotnet/coreclr/pull/13298, missed a few cases.
Noah Falk [Wed, 6 Jun 2018 03:40:12 +0000 (20:40 -0700)]
Add instructions for making idl file changes (#18188)
Brian Sullivan [Wed, 6 Jun 2018 00:58:21 +0000 (17:58 -0700)]
Merge pull request #18313 from dotnet-bot/from-tfs
Merge changes from TFS
Carol Eidt [Wed, 6 Jun 2018 00:31:31 +0000 (17:31 -0700)]
Merge pull request #18303 from mikedn/lockadd4
Use the actual type of LOCKADD's data operand
Brian Sullivan [Tue, 5 Jun 2018 22:11:08 +0000 (15:11 -0700)]
Fix a warning as error issue that causes a build break on the Desktop CLR
[tfs-changeset: 1702868]
Egor Chesakov [Tue, 5 Jun 2018 18:38:56 +0000 (11:38 -0700)]
Use sysconf(_SC_NPROCESSORS_CONF) instead of sysconf(_SC_NPROCESSORS_ONLN) in PAL and GC ONLY on ARM and ARM64 (#18289)
Alfred Myers [Tue, 5 Jun 2018 18:23:36 +0000 (15:23 -0300)]
Fixed typos on readme (#18304)
Mike Danes [Tue, 5 Jun 2018 16:29:22 +0000 (19:29 +0300)]
Use the actual type of LOCKADD's data operand
The data operand can theoretically be a small int so we need to use its actual type to determine instruction size.
In practice this issue does not seem to happen as the importer is pretty insistent in spilling interlocked arguments to lclvars that have the proper type. Even when the data argument is already a small int lclvar (e.g. short method argument) this still doesn't cause problems because there's a cast between the method argument and LOCKADD.
Andy Ayers [Tue, 5 Jun 2018 14:52:35 +0000 (07:52 -0700)]
Update SDK used to build jitutils to 2.1 RC (#18290)
In anticipation of taking a dependence on netcoreapp2.1 for the
forthcoming PMI work from jitutils.
Also make sure to shut down build servers before cleaning up the SDK.
Luqun Lou [Tue, 5 Jun 2018 13:40:34 +0000 (06:40 -0700)]
Use WChar instead of char (#18294)
Andy Ayers [Mon, 4 Jun 2018 23:02:01 +0000 (16:02 -0700)]
JIT: tolerate byref/nint return type mismatches during inlining (#18274)
Have the jit tolerate byref/nint return type mismatches for inlines
same way that it tolerates them for ret instructions.
Fixes #18270.
Tanner Gooding [Sun, 3 Jun 2018 03:53:03 +0000 (20:53 -0700)]
Adding function headers to the 'emitIns_SIMD_*' methods and clarifying comments on related code.
Tanner Gooding [Sun, 3 Jun 2018 03:41:06 +0000 (20:41 -0700)]
Updating the x86 HWIntrinsics to support containment for most one-operand intrinsics.
Tanner Gooding [Sun, 3 Jun 2018 01:55:51 +0000 (18:55 -0700)]
Fixing up ContainCheckHWIntrinsic to use intrinsicId and to switch on the category for 1/3 arg intrinsics
Tanner Gooding [Sun, 3 Jun 2018 01:53:20 +0000 (18:53 -0700)]
Fixing the name of the paramters in the emitIns_SIMD_* methods.
Carol Eidt [Mon, 4 Jun 2018 21:57:09 +0000 (14:57 -0700)]
Merge pull request #18267 from mikedn/lockadd3
Cleanup LOCKADD handling
Jan Kotas [Mon, 4 Jun 2018 20:09:29 +0000 (13:09 -0700)]
Make Encoding.Unicode.GetString faster on 64-bit platforms (#18263)
UnicodeEncoding was taking fast path only when both source and destination were 8-byte aligned. It was not
the case for Unicode.GetString.
The fix is to just check the source alignment as fast path precondition, and use unaligned write for destination.
Also removed alternative fast path that is not needed anymore, tweaked local variable assignments for better
register allocation, and fixed performance issue introduced by recent big endian portability fix.
Amy [Mon, 4 Jun 2018 18:40:58 +0000 (11:40 -0700)]
R2RDump - Dump AvailableTypes section contents (#18227)
* Move R2RReader constructor to separate functions, parse READYTORUN_SECTION_AVAILABLE_TYPES
* Fix bug from merging, output formatting changes
* Rename availableTypes variables, save availableTypes as string instead of tuple
Mike Danes [Sun, 3 Jun 2018 21:50:17 +0000 (00:50 +0300)]
Cleanup LOCKADD handling
LOCKADD nodes are generated rather early and there's no reason for that:
* The CORINFO_INTRINSIC_InterlockedAdd32/64 intrinsics are not actually used. Even if they would be used we can still import them as XADD nodes and rely on lowering to generate LOCKADD when needed.
* gtExtractSideEffList transforms XADD into LOCKADD but this can be done in lowering. LOCKADD is an XARCH specific optimization after all.
Additionally:
* Avoid the need for special handling in LSRA by making GT_LOCKADD a "no value" oper.
* Split LOCKADD codegen from XADD/XCHG codegen, attempting to use the same code for all 3 just makes things more complex.
* The address is always in a register so there's no real need to create an indir node on the fly, the relevant emitter functions can be called directly.
The last point above is actually a CQ issue - we always generate `add [reg], imm`, more complex address modes are not used. Unfortunately this problem starts early, when the importer spills the address to a local variable. If that ever gets fixed then we'll could probably generate a contained LEA in lowering.
Alfred Myers [Mon, 4 Jun 2018 18:01:50 +0000 (15:01 -0300)]
Fixed a couple of comments on TimeSpan.SecondsPerTick and Thread.SleepInternal (#18276)
* Fixed comment for TimeSpan.SecondsPerTick
* Fixed comment on Thread.SleepInternal
Benjamin Hodgson [Mon, 19 Feb 2018 14:20:05 +0000 (14:20 +0000)]
Optimise HtmlEncode to produce less garbage
do the same optimisation for HtmlDecode
remove now-unneeded reference
Add tests for new ValueStringBuilder methods
Signed-off-by: dotnet-bot-corefx-mirror <dotnet-bot@microsoft.com>
Brian Robbins [Mon, 4 Jun 2018 17:26:53 +0000 (10:26 -0700)]
Flow EventSources to EventPipe on Windows (#18217)
Egor Chesakov [Mon, 4 Jun 2018 17:08:22 +0000 (10:08 -0700)]
Work in VM towards cross-bitness crossgen (#18245)
* Replace sizeof expressions with target-specific constants:
* sizeof(ObjHeader) -> OBJHEADER_SIZE
* sizeof(Object) -> OBJECT_SIZE
* ObjSizeOf(Object) -> OBJECT_BASESIZE
* sizeof(ArrayBase) -> ARRAYBASE_SIZE
* ObjSizeOf(ArrayBase) -> ARRAYBASE_BASESIZE
* Remove ObjSizeOf macro
* Use OBJECT_SIZE in ArrayBase::GetBoundsOffset
* Cast ppObj to CORCOMPILE_GCREFMAP_TOKENS* before dereferencing in FakePromote
* LOG2_PTRSIZE should depend on _TARGET_64BIT_
Andrey Kurdyumov [Mon, 4 Jun 2018 16:24:37 +0000 (22:24 +0600)]
Fix error during compilation with /permissive- (#18272)
Michelle McDaniel [Mon, 4 Jun 2018 16:06:09 +0000 (09:06 -0700)]
Be more aggressive with Partial NGen (#18240)
* Be more aggressive with partial ngen
In partial ngen, we only compile those methods that both have profile
data and have the ReadMethodCode flag set in the profile flags. That
means that we miss some functions that do have counts, which affects the
performance of assemblies crossgened with partial ngen. This change
makes partial ngen more aggressive by, in the cold path, first checking
to see if we already considered this method in the hot section, which
means that it has profile data, and only rejecting those methods that
were not already considered.