Remove relocations for MethodTable's vtable-1st-level-indirection

Commit migrated from https://github.com/dotnet/coreclr/commit/c4c04ea45cda3194cdf2189d757b96602ea344ee

Remove relocations for MethodTable::m_pParentMethodTable for Linux ARM

Commit migrated from https://github.com/dotnet/coreclr/commit/cf1fb9e17fc8b6ee849edab5a696d0ec5c6eadd2

Remove relocations for MethodTable::m_pPerInstInfo for Linux ARM

Commit migrated from https://github.com/dotnet/coreclr/commit/7a107e9622ff5803987add587414610014eeca4e

Remove relocations for MethodTable::m_pWriteableData for Linux ARM

Commit migrated from https://github.com/dotnet/coreclr/commit/33e3886802d95e7d4b5bfde630c0067d07c6291a

Remove relocations for InterfaceInfo_t::m_pMethodTable for Linux ARM

Commit migrated from https://github.com/dotnet/coreclr/commit/e085f07a5fa0aff778ff7562a33a88513b79b3b7

Additional fixes for RelativePointer, FixupPointer, RelativeFixupPointer, PlainPointer

Commit migrated from https://github.com/dotnet/coreclr/commit/e8adb62978b887eda3617dff7eccec1fc7719d07

Add FixupPlainOrRelativePointerField for MethodDesc fields

Commit migrated from https://github.com/dotnet/coreclr/commit/fead79c7c612a7262621ec75a0a76a83f08903bf

Remove relocations from SECTION_Readonly for fields not accessed from jit code on ARM

Commit migrated from https://github.com/dotnet/coreclr/commit/b4786838985ca249cca04a7d0676ae57ee61ca2b

Merge pull request dotnet/coreclr#12572 from wateret/armel/dblarg_stack

[RyuJIT/armel] Passing stack double arguments

Commit migrated from https://github.com/dotnet/coreclr/commit/74a3f9691e490e9732da55c46b678159c64fae74

[RyuJIT/ARM32] Avoid promoting structures which can be handled by HFA (dotnet/coreclr#12604)

* [RyuJIT/ARM32] Avoid promoting structures which can be handled by HFA

Signed-off-by: Hyung-Kyu Choi <hk0110.choi@samsung.com>
* [RyuJIT/ARM32] Add comments for ARM32

Signed-off-by: Hyung-Kyu Choi <hk0110.choi@samsung.com>
Commit migrated from https://github.com/dotnet/coreclr/commit/07207c0402a0d822b910935e3054faaca293ddfd

Note that tests don't have to be built to view JIT dumps (dotnet/coreclr#12699)

Commit migrated from https://github.com/dotnet/coreclr/commit/1f57f54b9f79a356476df9ba7a043ea967980ca4

Unify LazyInitializers with CoreRT (dotnet/coreclr#12697)

CoreCLR implementation cached s_activatorFactorySelector delegate, but CoreRT did not cache it - it got commented out during earlier AOT binary size optimization efforts. I have unified it by avoiding s_activatorFactorySelector delegate altogether to keep the best of both.

Commit migrated from https://github.com/dotnet/coreclr/commit/19042a92989904d1502d1f14798ade735ee1836a

TypeForward cleanup for Culture=neutral (dotnet/coreclr#12698)

Commit migrated from https://github.com/dotnet/coreclr/commit/5041fa732b48728db01e9875358d6e564bd31548

Move files to shared CoreLib partition (dotnet/coreclr#12694)

Commit migrated from https://github.com/dotnet/coreclr/commit/fa5544c2b0589f1b1811dbfaf6f227327c4757fc

Use R2R helper for runtime lookups only when USEHELPER was specified (dotnet/coreclr#12669)

The CoreRT compiler would like to be able to do indirections-based
lookup even when ReadyToRun codegen is specified.

Commit migrated from https://github.com/dotnet/coreclr/commit/29de0696d2e870a891c7312316620083f953604e

Avoid a large local in a recursive visitor. (dotnet/coreclr#12693)

This local of type `CORINFO_CALL_INFO` bloats the frame size of its
caller when the function is inlined and ends up aggravating stack size
issues in deep recursion. This change refactors the offending method a
bit in order to avoid the local entirely.

Hopefully this fixes dotnet/coreclr#12691.

Commit migrated from https://github.com/dotnet/coreclr/commit/ce6c5640ff3e9443f1c58e1070d478f282ce90a6

Update CoreClr, CoreFx to preview2-25508-04, preview2-25508-04, respectively (dotnet/coreclr#12633)

Commit migrated from https://github.com/dotnet/coreclr/commit/ed91ce69cbc4e026f297075d123a28a5838f0580

Remove redundant call to TimeSpan.TotalMilliseconds (dotnet/coreclr#12692)

Commit migrated from https://github.com/dotnet/coreclr/commit/e616560d5de4ccb39d846860eee27ff8806f28b2

Ensure that ZeroMemoryInGCHeap writes in pointer-sized increments (dotnet/coreclr#12689)

by adding the volatile keyword (which disables memset optimization).
Fixes dotnet/coreclr#12207

Commit migrated from https://github.com/dotnet/coreclr/commit/c28b3c9f928ba0ff1be6727dcbdb9ba8d1522a82

Merge pull request dotnet/corertdotnet/coreclr#4110 from jkotas/nmirror-merge

Merge nmirror to master

Signed-off-by: dotnet-bot <dotnet-bot@microsoft.com>
Commit migrated from https://github.com/dotnet/coreclr/commit/5e2d5d8f2ff3e3bc317592419f11e5d6ba12dd0e

Merge pull request dotnet/corertdotnet/coreclr#4100 from dotnet/nmirror

Merge nmirror to master

Signed-off-by: dotnet-bot <dotnet-bot@microsoft.com>
Commit migrated from https://github.com/dotnet/coreclr/commit/d9ba7e32d5562ec1f26592d7718ac19f913129e6

Fix Unix debug assert to not use instance handle (dotnet/coreclr#12687)

Commit migrated from https://github.com/dotnet/coreclr/commit/cf4ede607d7cf506a1aa2f53905a6cda11b51db6

Fix GitHub_3449 test (dotnet/coreclr#12683)

* Fix GitHub_3449 test

This test was looking for one answer for ARM and ARM64 but another
for other architectures. This didn't work for ARM32 on Windows
(compared to ARM32 on Linux).

Instead, simply allow either answer. This doesn't seem like a
terrible loss of precision, since we already allowed different
answers based on the platform.

* Update comment

Commit migrated from https://github.com/dotnet/coreclr/commit/866fc9fc6ae6367a70e3c2dd4a819213bfaf214e

Minor fix added to mirrored changes

Commit migrated from https://github.com/dotnet/coreclr/commit/4cdcaa8708fe6d59a3f5a0be69b245eb47d735a8

Merge pull request dotnet/corertdotnet/coreclr#4086 from tarekgh/CleanupSerializationInEncoding

Clean up serialization from Encoding

Signed-off-by: dotnet-bot <dotnet-bot@microsoft.com>
Commit migrated from https://github.com/dotnet/coreclr/commit/78c661d7e6c79d6647c6c18bf5cab17aa11b4a3b

Merge pull request dotnet/coreclr#12682 from adityamandaleeka/arm64_dump_hang

Temporarily block core dump printing on arm64

Commit migrated from https://github.com/dotnet/coreclr/commit/e6727e656094143730c16a8288e0e6c4412a59aa

Merge pull request dotnet/coreclr#12675 from mikedn/range-disp

Dump tree IDs instead of pointers in RangeCheck

Commit migrated from https://github.com/dotnet/coreclr/commit/2b974e13e3f580c87e56dec5379ca1686121e269

Merge pull request dotnet/coreclr#12673 from ruben-ayrapetyan/fix-issue-12643

Fix copying of FieldMarshaler structures

Commit migrated from https://github.com/dotnet/coreclr/commit/d74ac9150e314aa70e4c6c3d4aaded1b1a4b8bc6

Merge pull request dotnet/coreclr#12681 from BruceForstall/RemoveNonexistentVarargsTests

Remove varargs test references

Commit migrated from https://github.com/dotnet/coreclr/commit/d87da4c581c7a77bc19a200f00d50a6cf61a2da2

Restore missing native *.ni.pdb file from the Microsoft.NETCore.Runtime.CoreCLR (dotnet/coreclr#12677)

* Restore missing native *.ni.pdb file from the Microsoft.NETCore.Runtime.CoreCLR package

DLLs that have native code, need both the *.pdb (for IL information) and the *.ni.pdb (for
native information).

When System.Private.Corlib was renamed from System.Private.Corlib.ni.pdb to System.Private.Corlib,
the logic that decides what goes into the  Microsoft.NETCore.Runtime.CoreCLR package lost
the *.ni.pdb for System.Private.Corlib.

This change basically causes this logic to search both for *.ni.pdb as well as the *.pdb file and
thus fixes this issue.

* Remove unnecessary condition (review feedback)

Commit migrated from https://github.com/dotnet/coreclr/commit/597a7996c5a7d2a0b52e0c9b404f6fbff0d34aad

Temporarily block core dump printing on arm64 to unblock CI.

Commit migrated from https://github.com/dotnet/coreclr/commit/f82c45150188a57c245f4a430592008c35ad1849

Revert change that always publish runtime layout (dotnet/coreclr#12679)

Commit migrated from https://github.com/dotnet/coreclr/commit/14908c750664a000c45936a7b42532a2ed2ad233

Merge pull request dotnet/coreclr#12500 from adiaaida/throughputExclusionList

Update throughput exclusion list

Commit migrated from https://github.com/dotnet/coreclr/commit/88311c57064bfe69dd60b85d16ee84c93a1d1598

Remove varargs test references

varargs tests were removed with dotnet/coreclr@0a427c6d6d87e69ce0f87dafd2f965ee989e6d29

Commit migrated from https://github.com/dotnet/coreclr/commit/0d3209f58b35ce8c6c70e339772482f28b5d9568

Dump tree IDs instead of pointers in RangeCheck

Commit migrated from https://github.com/dotnet/coreclr/commit/3ce09a193367cc6b361a2911b85b6264dca36c5d

Fix skipcrossgen option (dotnet/coreclr#12650)

* Fix skipcrossgen option

* Use return instead of exit

Commit migrated from https://github.com/dotnet/coreclr/commit/8b4d63bea132d4cb73933ca9e2411f55bca1f539

Merge pull request dotnet/coreclr#12429 from mikedn/dead-qmark

Remove QMARK related dead code

Commit migrated from https://github.com/dotnet/coreclr/commit/9f6c2f6d83d9367252ba14bc515a53c601074261

Merge pull request dotnet/coreclr#12623 from hqueue/arm/ryujit/issue_12615

[RyuJIT/ARM32] Unassign double register properly

Commit migrated from https://github.com/dotnet/coreclr/commit/8f8e2abe707754bc3eefce619403b5dc9199182b

Merge pull request dotnet/coreclr#12597 from hseok-oh/ryujit/struct_split/lsra

[RyuJIT/ARM32] Codegen: split struct argument less than 16 bytes

Commit migrated from https://github.com/dotnet/coreclr/commit/d83ddabc3fc2b3099cc2e10bf2626f1d1a9b8c7d

Fix copying of FieldMarshaler structures in EEClassLayoutInfo::CollectLayoutFieldMetadataThrowing.

Related issue: dotnet/coreclr#12643

Commit migrated from https://github.com/dotnet/coreclr/commit/b4131eab7371b3a34c76349771e8766037754bd9

[RyuJIT/ARM32] Codegen: split struct argument less than 16 bytes

Enable passing split struct (less than 16 bytes)
- LSRA phase
- Codegen phase

Commit migrated from https://github.com/dotnet/coreclr/commit/e62cb301df336a682c9e26b9876d5c7279d8e1e4

Merge pull request dotnet/coreclr#12653 from wateret/arm/lvadump

[RyuJIT/arm] Fix double type local variable dump

Commit migrated from https://github.com/dotnet/coreclr/commit/28e0a4f8fc21afcf0ca963ddfd3d8ac22d3f0998

[RyuJIT/ARM32] Apply review feedback

Rename vairable and rewrite code to use less #ifdef

Signed-off-by: Hyung-Kyu Choi <hk0110.choi@samsung.com>
Commit migrated from https://github.com/dotnet/coreclr/commit/0acd5abc5725353833f9d6fb14a6fd4290a6c1d4

[RyuJIT/armel] Rewrite changes for readibility

Commit migrated from https://github.com/dotnet/coreclr/commit/0c1b669e53f73e8c1e4967ff436eb25b0cb57155

Pin GC threads to corresponding processors on Linux (dotnet/coreclr#12486)

* fix SetThreadGroupAffinity, add SetThreadAffinityMask and pin GC threads to processors on Linux

* fix loop count and error handler

Commit migrated from https://github.com/dotnet/coreclr/commit/39ce84d04ee9a783a20b77b820d999271edbb31b

Merge pull request dotnet/coreclr#12667 from rartemev/issue_12465_1

Added explicit cast float -> double for arguments

Commit migrated from https://github.com/dotnet/coreclr/commit/a72a34db3f702abe54dc0512069abc90f2bcff37

Merge pull request dotnet/coreclr#12634 from rartemev/issue_12466_1

Fix misprint in complex condition

Commit migrated from https://github.com/dotnet/coreclr/commit/b7e10aa038dc67fabd18d9f71c83e3edc1a89e4d

Added explicit cast float -> double for arguments

Commit migrated from https://github.com/dotnet/coreclr/commit/fcf7571a0edc8ba028897bdb468d6b4fbd23c77c

Use SUB instead of XOR to clear fat pointer bit (dotnet/coreclr#12658)

This makes the code generated for fat calli smaller because of the clearing of the fat pointer bit gets fused into address mode.

Commit migrated from https://github.com/dotnet/coreclr/commit/8e3f231e944e4adb7b8e9e8c364e95ae0b8798da

Merge pull request dotnet/coreclr#12666 from mskvortsov/UnalignedLdSt

[RyuJIT/Arm32] Support unaligned floating point loads/stores

Commit migrated from https://github.com/dotnet/coreclr/commit/ce1fdc0e3d8509523a729026e2e6432f88bda6c2

Merge pull request dotnet/coreclr#12655 from YongseopKim/ryujit/arm32/add_nyi_promoted_struct

[RyuJIT/ARM32] Add NYI_ARM for handling promoted struct

Commit migrated from https://github.com/dotnet/coreclr/commit/f9f0ec4ec02b24210ea8fc7bb2acd2c759b6d57a

Update throughput exclusion list

There are many assemblies in our throughput benchmark package that are
reference or forwarding assemblies. There is no jit work in these
assemblies, so we want to exclude them.

This change also adds an iterations parameter so users can specify a
specific number of iterations for testing.

Commit migrated from https://github.com/dotnet/coreclr/commit/ddaa2cd5de83fa90012991a18bc8c4edd4e97755

Merge pull request dotnet/coreclr#12053 from mikedn/xcopy-bindir

Do not copy bin subdirectories to test core_root

Commit migrated from https://github.com/dotnet/coreclr/commit/ccc6f7213265d45d6d4cf0f4b7da3b18bbba8436

Merge pull request dotnet/coreclr#12652 from hseok-oh/ryujit/fix_split_lower

[RyuJIT/ARM32] Fix register type setting for split struct arg

Commit migrated from https://github.com/dotnet/coreclr/commit/7861f5b4ed8abcfb6b337212e1f9dd73fed2afa9

Support unaligned floating point loads/stores

Commit migrated from https://github.com/dotnet/coreclr/commit/6ad0c4d285ea681fec0b8bad2cf14d373ff3ee15

[RyuJIT/ARM32] Add NYI_ARM for handling promoted struct

- In RyuJIT/ARM32, promoted struct on stack can't be handled like
  splitted struct arg's one.
- Until implementing handling it, make it entering the NYI_ARM.

Commit migrated from https://github.com/dotnet/coreclr/commit/d78aa192cdd3bc95046b0d78ac1c74dba2ca6029

Fix DumpLog to be compatible with CoreRT (dotnet/coreclr#12636)

Commit migrated from https://github.com/dotnet/coreclr/commit/32b52269a270f9b7800da3ba119b92061f528789

build runtime layout when skipping tests (dotnet/coreclr#12587)

Commit migrated from https://github.com/dotnet/coreclr/commit/79ce844d7a68a62ad0a49a5db202dabf343ecad7

[x86/Linux] Enable thread-abort reraise loop prevention (dotnet/coreclr#12646)

Commit migrated from https://github.com/dotnet/coreclr/commit/e5557aea38564be9c2e1630ae7b075c3b76837b3

[RyuJIT/arm] Fix double type local variable dump

Commit migrated from https://github.com/dotnet/coreclr/commit/45cd0b74063ef74e584988639ff19df619796663

[RyuJIT/ARM32] Fix register type setting for split struct arg

Fix register type setting for split struct argument using GT_FIELD_LIST node

Commit migrated from https://github.com/dotnet/coreclr/commit/ccaf2e88b9661501ca8d7141bbc0cd4b64864f9b

[RyuJIT/ARM32] Fix assertion failed 'gcPtrCount == 0' (dotnet/coreclr#12621)

* [RyuJIT/ARM32] Fix assertion failed 'gcPtrCount == 0'

Change wrong gcPtrCount's value from gtNumSlots to gtNumberReferenceSlots of GenTreePutArgStk

* Fix a formatting error

Commit migrated from https://github.com/dotnet/coreclr/commit/7e15d8f121df2f51e1eff74e765e64109a9574c0

Merge pull request dotnet/coreclr#12647 from dotnet/mirror-merge-9210244

Mirror changes from dotnet/corert

Commit migrated from https://github.com/dotnet/coreclr/commit/474d9618e844e9a0fb3fd2b687d558d1239b021d

Fix the merge from the corert changes

Commit migrated from https://github.com/dotnet/coreclr/commit/d11d281a2141bb6880371d878d558859c56f4901

Merge pull request dotnet/corertdotnet/coreclr#4080 from dotnet/nmirror

Merge nmirror to master

Signed-off-by: dotnet-bot <dotnet-bot@microsoft.com>
Commit migrated from https://github.com/dotnet/coreclr/commit/9673552bb036470ebb57aa647d0995686cbd94c8

Merge pull request dotnet/coreclr#12606 from adityamandaleeka/mask_activation_on_altstack

Mask INJECT_ACTIVATION_SIGNAL while on alternate signal stack

Commit migrated from https://github.com/dotnet/coreclr/commit/334259414f8455b93989ee1ac01a7facbdff8950

Mask INJECT_ACTIVATION_SIGNAL while on alternate signal stack.

Commit migrated from https://github.com/dotnet/coreclr/commit/94f2d05cd8e2f0d0be0e264b808dccb58b5d8f33

Update gtDispRegVal() for multi register nodes (dotnet/coreclr#12573)

* Update gtDispRegVal() for multi register nodes

* [RyuJIT/armel] Fix

Commit migrated from https://github.com/dotnet/coreclr/commit/bcfb4b64f2f5186318410d5054486e8a948271ca

Cleanup officialbuild warnings (dotnet/coreclr#12578)

* Cleanup officialbuild warnings

* Re-enable "sync test native binaries" step

Commit migrated from https://github.com/dotnet/coreclr/commit/511099820df8a8ab750a436d5f78ebb9e8b43df1

Merge pull request dotnet/coreclr#12619 from parjong/fix/merge_jit_and_engine

Fix missing references in FEATURE_MERGE_JIT_AND_ENGINE

Commit migrated from https://github.com/dotnet/coreclr/commit/a211a077bd199cf5ea27f4983a13d77a9efba8f2

Merge pull request dotnet/coreclr#12637 from parjong/fix/skipcrossgen

Add skipcrossgen option

Commit migrated from https://github.com/dotnet/coreclr/commit/5ea45b71361607e99d752615c491600ac4fafeca

Merge pull request dotnet/coreclr#12575 from wateret/armel/floatret

[RyuJIT/armel] Move returned float to float reg

Commit migrated from https://github.com/dotnet/coreclr/commit/866d4189c77a703e6af805646e927266222d8876

Add skipcrossgen option

Commit migrated from https://github.com/dotnet/coreclr/commit/6193e8b18c29a62d12e36c6d2fbc86ca91f287d1

Support interprter build when INTERP_TRACING is not set (dotnet/coreclr#12625)

Commit migrated from https://github.com/dotnet/coreclr/commit/17b815123a609a4c326875ff845a8e70aaab278c

Remove unused global variable (dotnet/coreclr#12620)

Commit migrated from https://github.com/dotnet/coreclr/commit/b063eb64d5f7fa229677934b9a8f83ef46a7423c

Hide methods in IdDispenser (instead of using DacNotImpl) (dotnet/coreclr#12624)

Commit migrated from https://github.com/dotnet/coreclr/commit/85e8f7b2e1fca663e55e37f396c85f60e0375935

Fix misprint in complex condition

Commit migrated from https://github.com/dotnet/coreclr/commit/102b767e2ce82a4d380abb4b764997e579b6e3c8

Update to latest System.Memory

Commit migrated from https://github.com/dotnet/coreclr/commit/f738e8861a5705dbd74a3088c0f4a1ee4e15935d

Update CoreClr, CoreFx to preview2-25504-01, preview2-25504-01, respectively

Commit migrated from https://github.com/dotnet/coreclr/commit/bc8b81b75bf43ea707e0d47b00da34a68f795194

Delete unnecessary code (dotnet/coreclr#12607)

Commit migrated from https://github.com/dotnet/coreclr/commit/38a2a69c786e4273eb1339d7a75f939c410afd69

[RyuJIT/armel] Move returned float to float reg

Returned float values are in r0(float) or r0:r1(double).
These values must be moved to proper float register after call.

Fix dotnet/coreclr#12574

Commit migrated from https://github.com/dotnet/coreclr/commit/305ddc1380e3373e7b5422ada70f100b38ccedc1

[RyuJIT/ARM32] Unassign double register properly

In LinearScan::tryAllocateFreeReg(), unassign a double register properly
when a found float register is a part of a double register.

Signed-off-by: Hyung-Kyu Choi <hk0110.choi@samsung.com>
Commit migrated from https://github.com/dotnet/coreclr/commit/127cbf2a7ff14cb5281c22e5a9b10c47f6605a26

Fix missing refernces in FEATURE_MERGE_JIT_AND_ENGINT

Commit migrated from https://github.com/dotnet/coreclr/commit/06a7b5b7c6cbaf2390a3916fbfb1b3ea06a7e420

Merge pull request dotnet/coreclr#12608 from pgavlin/FixDSE

Fix a bug in LIR dead store removal.

Commit migrated from https://github.com/dotnet/coreclr/commit/77ac6921d8b6b23e682adc42440501e34a083720

create GetAssertionIndex to convert from unsigned to AssertionIndex (dotnet/coreclr#12554)

Commit migrated from https://github.com/dotnet/coreclr/commit/2b5da8429093ad757849686fd668330a3102a1a5

Fix a bug in LIR dead store removal.

This came in with the recent operand visitor changes: instead of marking
the operands of a dead indirect store unused, DSE would mark the operands
of its address unsued.

Commit migrated from https://github.com/dotnet/coreclr/commit/bee2ec645438be185bb77371b0eb5093573f7b9d

[x86/Linux] Restore PTR_MethodTable in FastPrimitiveArrayAllocator (dotnet/coreclr#12569)

Commit migrated from https://github.com/dotnet/coreclr/commit/9665cb5b9af94cccaa2e127337ff6d471ea18a0b

[RyuJIT/armel] Passing stack double arguments

Fix dotnet/coreclr#11927

Commit migrated from https://github.com/dotnet/coreclr/commit/65f83c9da4c954ab89e3565738bb6db835168a98

Move the bitset iterator for bitsetasuint64inclass inside the bitset. (dotnet/coreclr#12553)

It is how it is done for others bitsets.

Commit migrated from https://github.com/dotnet/coreclr/commit/01088c7b3e47e22c08bfcdc3a73dc5f7a7f1696e

Improve fgAddRefPred (dotnet/coreclr#12585)

* Improve fgAddRefPred

Only do one search through the predecessor list looking for an
existing predecessor edge, or looking for the proper location to
insert a new predecessor edge. Previously, we traversed the list
twice (or part of twice) for new edges. For new edges, stop the
search when we reach the correct location in sorted order.

This reduces x86 release minopts instruction count by 0.42% and
full opts instruction count by 0.33% (of a superpmi replay of
the tests).

Fixes dotnet/coreclr#12582

* Update for feedback

Commit migrated from https://github.com/dotnet/coreclr/commit/58d0ead6d34837fab60da1473090588db415fc1d

Merge pull request dotnet/corertdotnet/coreclr#4046 from dotnet/nmirror

Merge nmirror to master

Signed-off-by: dotnet-bot <dotnet-bot@microsoft.com>
Commit migrated from https://github.com/dotnet/coreclr/commit/bfc3141f97a9f9f0d095f3d2e99a7c02ce869ab0

Reducing diffs with CoreRT (dotnet/coreclr#12579)

* Rename m_firstChar to _firstChar to reduce CoreRT diffs
* Reducing diffs with CoreRT
* Delete dead code

Commit migrated from https://github.com/dotnet/coreclr/commit/57214f1948e56835cde1b6162acc58751f483e14

Merge pull request dotnet/coreclr#12568 from hseok-oh/ryujit/fix_11781

[RyuJIT/ARM32] Fix using fgMorphMultiregStructArg for mkrefany

Commit migrated from https://github.com/dotnet/coreclr/commit/3390498956eeccd6d4813fb65d534e6108c16ebb

Remove unnecessary update of RegRecord in LSRA (dotnet/coreclr#12474)

* Remove unnecessary update of RegRecord

We don't need to clear assignedInterval of regRec,
because it was already cleared by checkAndClearInterval() above.

Signed-off-by: Hyung-Kyu Choi <hk0110.choi@samsung.com>
* Apply review feedback

- Add a assertion and comment for the assertion

Signed-off-by: Hyung-Kyu Choi <hk0110.choi@samsung.com>
Commit migrated from https://github.com/dotnet/coreclr/commit/f0d3c439dc16a3218d53578f14c973696de04cb1

[RyuJIT/ARM32] Return HFA struct (dotnet/coreclr#12453)

* [RyuJIT/ARM32] Return HFA struct

- Enable return HFA struct
- Merge ARM32/ARM64 CodeGen::isStructReturn(GenTreePtr treeNode)
- Merge ARM32/ARM64 CodeGen::genStructReturn(GenTreePtr treeNode)

* Add comment

Add TODO comment to optimize two-float loading

Commit migrated from https://github.com/dotnet/coreclr/commit/7318ee5ce4306030908e6c88fd9b908914d4c685

Merge pull request dotnet/coreclr#12539 from ellismg/update-prerelease-tag

Update to master prerelease2 tag

Commit migrated from https://github.com/dotnet/coreclr/commit/abc11397af25d50159e35538569eae581d0dbdbf

Merge pull request dotnet/coreclr#12527 from gsfreema/relative-paths

Fix contributing workflow commit message link

Commit migrated from https://github.com/dotnet/coreclr/commit/eaab9f34b0f5e139af82ef9aea09ee2fdb8badcc

Fix an issue where g_ephemeral_low would never get bashed into the write (dotnet/coreclr#12563)

barrier when starting up with Server GC (x86 only).

Commit migrated from https://github.com/dotnet/coreclr/commit/7a283118b871458fd46d4593f2e6e3126701e4bc

Merge pull request dotnet/coreclr#12577 from dpodder/optdata-170630

Update optdata to version 20170630-0138

Commit migrated from https://github.com/dotnet/coreclr/commit/e95af8cc002e49592fe4d6b20b03330eb4ffd41a

Merge pull request dotnet/coreclr#12236 from rartemev/fix_legacyjit_tests

Fix legacyjit CI testing

Commit migrated from https://github.com/dotnet/coreclr/commit/2beb07a856087b65a4729ddfb20701e244039f59