Delete !FEATURE_IMPLICIT_TLS (#14398)

Linux and Windows arm64 are using the regular C/C++ thread local statics. This change unifies the remaining Windows architectures to be on the same plan.

44 files changed, 167 insertions, 2776 deletions

diff --git a/src/vm/CMakeLists.txt b/src/vm/CMakeLists.txt
index f8790cf85d..0aed676d94 100644
--- a/src/vm/CMakeLists.txt
+++ b/src/vm/CMakeLists.txt
@@ -351,7 +351,6 @@ if(CLR_CMAKE_TARGET_ARCH_AMD64)
         ${ARCH_SOURCES_DIR}/InstantiatingStub.asm
         ${ARCH_SOURCES_DIR}/JitHelpers_Fast.asm
         ${ARCH_SOURCES_DIR}/JitHelpers_FastWriteBarriers.asm
-        ${ARCH_SOURCES_DIR}/JitHelpers_InlineGetAppDomain.asm
         ${ARCH_SOURCES_DIR}/JitHelpers_InlineGetThread.asm
         ${ARCH_SOURCES_DIR}/JitHelpers_SingleAppDomain.asm
         ${ARCH_SOURCES_DIR}/JitHelpers_Slow.asm
@@ -359,7 +358,6 @@ if(CLR_CMAKE_TARGET_ARCH_AMD64)
         ${ARCH_SOURCES_DIR}/RedirectedHandledJITCase.asm
         ${ARCH_SOURCES_DIR}/ThePreStubAMD64.asm
         ${ARCH_SOURCES_DIR}/ExternalMethodFixupThunk.asm
-        ${ARCH_SOURCES_DIR}/TlsGetters.asm # Condition="'$(FeatureImplicitTls)' != 'true'
         ${ARCH_SOURCES_DIR}/UMThunkStub.asm
         ${ARCH_SOURCES_DIR}/VirtualCallStubAMD64.asm
     )
diff --git a/src/vm/amd64/AsmMacros.inc b/src/vm/amd64/AsmMacros.inc
index f95a291929..58fd8d130f 100644
--- a/src/vm/amd64/AsmMacros.inc
+++ b/src/vm/amd64/AsmMacros.inc
@@ -188,29 +188,25 @@ Section ends
 
 
 ;
-; Macro to Call GetThread() correctly whether it is indirect or direct
-;
-CALL_GETTHREAD macro
-ifndef GetThread
-extern GetThread:proc
-endif
-        call    GetThread
-        endm
+; Inlined version of GetThread
+; Trashes rax and r11
+; 
+INLINE_GETTHREAD macro Reg
+
+        EXTERN _tls_index : DWORD
+        EXTERN gCurrentThreadInfo:DWORD
+
+        mov     r11d, [_tls_index]
+        mov     rax, gs:[OFFSET__TEB__ThreadLocalStoragePointer]
+        mov     rax, [rax + r11 * 8]
+        mov     r11d, SECTIONREL gCurrentThreadInfo
+        mov     Reg, [rax + r11]
 
-CALL_GETAPPDOMAIN macro
-ifndef GetAppDomain
-extern GetAppDomain:proc
-endif
-        call    GetAppDomain
         endm
 
-;
 ; if you change this code there will be corresponding code in JITInterfaceGen.cpp which will need to be changed
 ;
 
-; DEFAULT_TARGET needs to always be futher away than the fixed up target will be
-
-
 JIT_HELPER_MONITOR_THUNK macro THUNK_NAME, Section
 Section segment para 'CODE'
         align 16
diff --git a/src/vm/amd64/InstantiatingStub.asm b/src/vm/amd64/InstantiatingStub.asm
index dff1b6f5a6..8601e4ae44 100644
--- a/src/vm/amd64/InstantiatingStub.asm
+++ b/src/vm/amd64/InstantiatingStub.asm
@@ -90,13 +90,11 @@ NESTED_ENTRY InstantiatingMethodStubWorker, _TEXT
         ;
         ; link the StubHelperFrame
         ;
-        CALL_GETTHREAD
-        mov     rdx, [rax + OFFSETOF__Thread__m_pFrame]
+        INLINE_GETTHREAD r12
+        mov     rdx, [r12 + OFFSETOF__Thread__m_pFrame]
         mov     [rbp + OFFSETOF_FRAME + OFFSETOF__Frame__m_Next], rdx
         lea     rcx, [rbp + OFFSETOF_FRAME]
-        mov     [rax + OFFSETOF__Thread__m_pFrame], rcx
-
-        mov     r12, rax                ; store the Thread pointer 
+        mov     [r12 + OFFSETOF__Thread__m_pFrame], rcx
 
         add     rsp, SIZEOF_MAX_OUTGOING_ARGUMENT_HOMES
 
diff --git a/src/vm/amd64/JitHelpers_Fast.asm b/src/vm/amd64/JitHelpers_Fast.asm
index f004be549e..5e0d79f74f 100644
--- a/src/vm/amd64/JitHelpers_Fast.asm
+++ b/src/vm/amd64/JitHelpers_Fast.asm
@@ -583,58 +583,6 @@ endif
         nop
 LEAF_END_MARKED JIT_WriteBarrier, _TEXT
 
-ifndef FEATURE_IMPLICIT_TLS
-LEAF_ENTRY GetThread, _TEXT
-        ; the default implementation will just jump to one that returns null until 
-        ; MakeOptimizedTlsGetter is run which will overwrite this with the actual 
-        ; implementation.
-        jmp short GetTLSDummy
-
-        ;
-        ; insert enough NOPS to be able to insert the largest optimized TLS getter 
-        ; that we might need, it is important that the TLS getter doesn't overwrite
-        ; into the dummy getter.
-        ;
-        db (TLS_GETTER_MAX_SIZE_ASM - 2) DUP (0CCh)
-
-LEAF_END GetThread, _TEXT
-
-LEAF_ENTRY GetAppDomain, _TEXT
-        ; the default implementation will just jump to one that returns null until 
-        ; MakeOptimizedTlsGetter is run which will overwrite this with the actual 
-        ; implementation.
-        jmp short GetTLSDummy
-
-        ;
-        ; insert enough NOPS to be able to insert the largest optimized TLS getter 
-        ; that we might need, it is important that the TLS getter doesn't overwrite
-        ; into the dummy getter.
-        ;
-        db (TLS_GETTER_MAX_SIZE_ASM - 2) DUP (0CCh)
-
-LEAF_END GetAppDomain, _TEXT
-
-LEAF_ENTRY GetTLSDummy, _TEXT
-        xor    rax, rax
-        ret
-LEAF_END GetTLSDummy, _TEXT
-
-LEAF_ENTRY ClrFlsGetBlock, _TEXT
-        ; the default implementation will just jump to one that returns null until 
-        ; MakeOptimizedTlsGetter is run which will overwrite this with the actual 
-        ; implementation.
-        jmp short GetTLSDummy
-
-        ;
-        ; insert enough NOPS to be able to insert the largest optimized TLS getter 
-        ; that we might need, it is important that the TLS getter doesn't overwrite
-        ; into the dummy getter.
-        ;
-        db (TLS_GETTER_MAX_SIZE_ASM - 2) DUP (0CCh)
-
-LEAF_END ClrFlsGetBlock, _TEXT
-endif
-
 ; Mark start of the code region that we patch at runtime
 LEAF_ENTRY JIT_PatchedCodeLast, _TEXT
         ret
@@ -986,12 +934,11 @@ if 0 ne 0
         ;
         ; link the TailCallFrame
         ;
-        CALL_GETTHREAD
-        mov     r14, rax
-        mov     r15, [rax + OFFSETOF__Thread__m_pFrame]        
+        INLINE_GETTHREAD r14
+        mov     r15, [r14 + OFFSETOF__Thread__m_pFrame]        
         mov     [r13 + OFFSETOF_FRAME + OFFSETOF__Frame__m_Next], r15
         lea     r10, [r13 + OFFSETOF_FRAME]
-        mov     [rax + OFFSETOF__Thread__m_pFrame], r10
+        mov     [r14 + OFFSETOF__Thread__m_pFrame], r10
 endif
 
         ; the pretend call would be here
diff --git a/src/vm/amd64/JitHelpers_InlineGetAppDomain.asm b/src/vm/amd64/JitHelpers_InlineGetAppDomain.asm
deleted file mode 100644
index 187decf14d..0000000000
--- a/src/vm/amd64/JitHelpers_InlineGetAppDomain.asm
+++ /dev/null
@@ -1,123 +0,0 @@
-; Licensed to the .NET Foundation under one or more agreements.
-; The .NET Foundation licenses this file to you under the MIT license.
-; See the LICENSE file in the project root for more information.
-
-; ==++==
-;
-
-;
-; ==--==
-; ***********************************************************************
-; File: JitHelpers_InlineGetAppDomain.asm, see history in jithelp.asm
-;
-; Notes: These routinues will be patched at runtime with the location in 
-;        the TLS to find the AppDomain* and are the fastest implementation 
-;        of their specific functionality.
-; ***********************************************************************
-
-include AsmMacros.inc
-include asmconstants.inc
-
-; Min amount of stack space that a nested function should allocate.
-MIN_SIZE equ 28h
-
-; Macro to create a patchable inline GetAppdomain, if we decide to create patchable
-; high TLS inline versions then just change this macro to make sure to create enough
-; space in the asm to patch the high TLS getter instructions.
-PATCHABLE_INLINE_GETAPPDOMAIN macro Reg, PatchLabel
-PATCH_LABEL PatchLabel
-        mov     Reg, gs:[OFFSET__TEB__TlsSlots]
-        endm
-
-extern JIT_GetSharedNonGCStaticBase_Helper:proc
-extern JIT_GetSharedGCStaticBase_Helper:proc
-
-LEAF_ENTRY JIT_GetSharedNonGCStaticBase_InlineGetAppDomain, _TEXT
-        ; Check if rcx (moduleDomainID) is not a moduleID
-        mov     rax, rcx
-        test    rax, 1
-        jz      HaveLocalModule
-
-        PATCHABLE_INLINE_GETAPPDOMAIN rax, JIT_GetSharedNonGCStaticBase__PatchTLSLabel
-
-        ; Get the LocalModule, rcx will always be odd, so: rcx * 4 - 4 <=> (rcx >> 1) * 8
-        mov     rax, [rax + OFFSETOF__AppDomain__m_sDomainLocalBlock + OFFSETOF__DomainLocalBlock__m_pModuleSlots]
-        mov     rax, [rax + rcx * 4 - 4]
-
-    HaveLocalModule:
-        ; If class is not initialized, bail to C++ helper
-        test    byte ptr [rax + OFFSETOF__DomainLocalModule__m_pDataBlob + rdx], 1
-        jz      CallHelper
-        REPRET
-
-    align 16
-    CallHelper:
-        ; Tail call JIT_GetSharedNonGCStaticBase_Helper
-        mov     rcx, rax
-        jmp     JIT_GetSharedNonGCStaticBase_Helper
-LEAF_END JIT_GetSharedNonGCStaticBase_InlineGetAppDomain, _TEXT
-
-LEAF_ENTRY JIT_GetSharedNonGCStaticBaseNoCtor_InlineGetAppDomain, _TEXT
-        ; Check if rcx (moduleDomainID) is not a moduleID
-        mov     rax, rcx
-        test    rax, 1
-        jz      HaveLocalModule
-
-        PATCHABLE_INLINE_GETAPPDOMAIN rax, JIT_GetSharedNonGCStaticBaseNoCtor__PatchTLSLabel
-
-        ; Get the LocalModule,  rcx will always be odd, so: rcx * 4 - 4 <=> (rcx >> 1) * 8
-        mov     rax, [rax + OFFSETOF__AppDomain__m_sDomainLocalBlock + OFFSETOF__DomainLocalBlock__m_pModuleSlots]
-        mov     rax, [rax + rcx * 4 - 4]
-        ret
-
-    align 16
-    HaveLocalModule:
-        REPRET
-LEAF_END JIT_GetSharedNonGCStaticBaseNoCtor_InlineGetAppDomain, _TEXT
-
-LEAF_ENTRY JIT_GetSharedGCStaticBase_InlineGetAppDomain, _TEXT
-        ; Check if rcx (moduleDomainID) is not a moduleID
-        mov     rax, rcx
-        test    rax, 1
-        jz      HaveLocalModule
-
-        PATCHABLE_INLINE_GETAPPDOMAIN rax, JIT_GetSharedGCStaticBase__PatchTLSLabel
-
-        ; Get the LocalModule, rcx will always be odd, so: rcx * 4 - 4 <=> (rcx >> 1) * 8
-        mov     rax, [rax + OFFSETOF__AppDomain__m_sDomainLocalBlock + OFFSETOF__DomainLocalBlock__m_pModuleSlots]
-        mov     rax, [rax + rcx * 4 - 4]
-
-    HaveLocalModule:
-        ; If class is not initialized, bail to C++ helper
-        test    byte ptr [rax + OFFSETOF__DomainLocalModule__m_pDataBlob + rdx], 1
-        jz      CallHelper
-
-        mov     rax, [rax + OFFSETOF__DomainLocalModule__m_pGCStatics]
-        ret
-
-    align 16
-    CallHelper:
-        ; Tail call Jit_GetSharedGCStaticBase_Helper
-        mov     rcx, rax
-        jmp     JIT_GetSharedGCStaticBase_Helper
-LEAF_END JIT_GetSharedGCStaticBase_InlineGetAppDomain, _TEXT
-
-LEAF_ENTRY JIT_GetSharedGCStaticBaseNoCtor_InlineGetAppDomain, _TEXT
-        ; Check if rcx (moduleDomainID) is not a moduleID
-        mov     rax, rcx
-        test    rax, 1
-        jz      HaveLocalModule
-
-        PATCHABLE_INLINE_GETAPPDOMAIN rax, JIT_GetSharedGCStaticBaseNoCtor__PatchTLSLabel
-
-        ; Get the LocalModule, rcx will always be odd, so: rcx * 4 - 4 <=> (rcx >> 1) * 8
-        mov     rax, [rax + OFFSETOF__AppDomain__m_sDomainLocalBlock + OFFSETOF__DomainLocalBlock__m_pModuleSlots]
-        mov     rax, [rax + rcx * 4 - 4]
-
-    HaveLocalModule:
-        mov     rax, [rax + OFFSETOF__DomainLocalModule__m_pGCStatics]
-        ret
-LEAF_END JIT_GetSharedGCStaticBaseNoCtor_InlineGetAppDomain, _TEXT
-
-        end
-
diff --git a/src/vm/amd64/JitHelpers_InlineGetThread.asm b/src/vm/amd64/JitHelpers_InlineGetThread.asm
index 40d63bf729..d9f58cc30f 100644
--- a/src/vm/amd64/JitHelpers_InlineGetThread.asm
+++ b/src/vm/amd64/JitHelpers_InlineGetThread.asm
@@ -2,11 +2,6 @@
 ; The .NET Foundation licenses this file to you under the MIT license.
 ; See the LICENSE file in the project root for more information.
 
-; ==++==
-;
-
-;
-; ==--==
 ; ***********************************************************************
 ; File: JitHelpers_InlineGetThread.asm, see history in jithelp.asm
 ;
@@ -21,15 +16,6 @@ include asmconstants.inc
 ; Min amount of stack space that a nested function should allocate.
 MIN_SIZE equ 28h
 
-; Macro to create a patchable inline GetAppdomain, if we decide to create patchable
-; high TLS inline versions then just change this macro to make sure to create enough
-; space in the asm to patch the high TLS getter instructions.
-PATCHABLE_INLINE_GETTHREAD macro Reg, PatchLabel
-PATCH_LABEL PatchLabel
-        mov     Reg, gs:[OFFSET__TEB__TlsSlots]
-        endm
-
-
 JIT_NEW                 equ     ?JIT_New@@YAPEAVObject@@PEAUCORINFO_CLASS_STRUCT_@@@Z
 Object__DEBUG_SetAppDomain equ ?DEBUG_SetAppDomain@Object@@QEAAXPEAVAppDomain@@@Z
 CopyValueClassUnchecked equ     ?CopyValueClassUnchecked@@YAXPEAX0PEAVMethodTable@@@Z
@@ -49,11 +35,6 @@ extern JIT_NewArr1:proc
 
 extern JIT_InternalThrow:proc
 
-ifdef _DEBUG
-extern DEBUG_TrialAllocSetAppDomain:proc
-extern DEBUG_TrialAllocSetAppDomain_NoScratchArea:proc
-endif
-
 ; IN: rcx: MethodTable*
 ; OUT: rax: new object
 LEAF_ENTRY JIT_TrialAllocSFastMP_InlineGetThread, _TEXT
@@ -61,7 +42,7 @@ LEAF_ENTRY JIT_TrialAllocSFastMP_InlineGetThread, _TEXT
 
         ; m_BaseSize is guaranteed to be a multiple of 8.
 
-        PATCHABLE_INLINE_GETTHREAD r11, JIT_TrialAllocSFastMP_InlineGetThread__PatchTLSOffset
+        INLINE_GETTHREAD r11
         mov     r10, [r11 + OFFSET__Thread__m_alloc_context__alloc_limit]
         mov     rax, [r11 + OFFSET__Thread__m_alloc_context__alloc_ptr]
 
@@ -73,10 +54,6 @@ LEAF_ENTRY JIT_TrialAllocSFastMP_InlineGetThread, _TEXT
         mov     [r11 + OFFSET__Thread__m_alloc_context__alloc_ptr], rdx
         mov     [rax], rcx
 
-ifdef _DEBUG
-        call    DEBUG_TrialAllocSetAppDomain_NoScratchArea
-endif ; _DEBUG
-
         ret
 
     AllocFailed:
@@ -95,7 +72,7 @@ NESTED_ENTRY JIT_BoxFastMP_InlineGetThread, _TEXT
 
         ; m_BaseSize is guaranteed to be a multiple of 8.
 
-        PATCHABLE_INLINE_GETTHREAD r11, JIT_BoxFastMPIGT__PatchTLSLabel
+        INLINE_GETTHREAD r11
         mov     r10, [r11 + OFFSET__Thread__m_alloc_context__alloc_limit]
         mov     rax, [r11 + OFFSET__Thread__m_alloc_context__alloc_ptr]
 
@@ -107,10 +84,6 @@ NESTED_ENTRY JIT_BoxFastMP_InlineGetThread, _TEXT
         mov     [r11 + OFFSET__Thread__m_alloc_context__alloc_ptr], r8
         mov     [rax], rcx
 
-ifdef _DEBUG
-        call    DEBUG_TrialAllocSetAppDomain_NoScratchArea
-endif ; _DEBUG
-
         ; Check whether the object contains pointers
         test    dword ptr [rcx + OFFSETOF__MethodTable__m_dwFlags], MethodTable__enum_flag_ContainsPointers
         jnz     ContainsPointers
@@ -169,7 +142,7 @@ LEAF_ENTRY AllocateStringFastMP_InlineGetThread, _TEXT
         lea     edx, [edx + ecx*2 + 7]
         and     edx, -8
 
-        PATCHABLE_INLINE_GETTHREAD r11, AllocateStringFastMP_InlineGetThread__PatchTLSOffset
+        INLINE_GETTHREAD r11
         mov     r10, [r11 + OFFSET__Thread__m_alloc_context__alloc_limit]
         mov     rax, [r11 + OFFSET__Thread__m_alloc_context__alloc_ptr]
 
@@ -183,10 +156,6 @@ LEAF_ENTRY AllocateStringFastMP_InlineGetThread, _TEXT
 
         mov     [rax + OFFSETOF__StringObject__m_StringLength], ecx
 
-ifdef _DEBUG
-        call    DEBUG_TrialAllocSetAppDomain_NoScratchArea
-endif ; _DEBUG
-
         ret
 
     OversizedString:
@@ -226,7 +195,7 @@ LEAF_ENTRY JIT_NewArr1VC_MP_InlineGetThread, _TEXT
         and     r8d, -8
 
 
-        PATCHABLE_INLINE_GETTHREAD r11, JIT_NewArr1VC_MP_InlineGetThread__PatchTLSOffset
+        INLINE_GETTHREAD r11
         mov     r10, [r11 + OFFSET__Thread__m_alloc_context__alloc_limit]
         mov     rax, [r11 + OFFSET__Thread__m_alloc_context__alloc_ptr]
 
@@ -241,10 +210,6 @@ LEAF_ENTRY JIT_NewArr1VC_MP_InlineGetThread, _TEXT
 
         mov     dword ptr [rax + OFFSETOF__ArrayBase__m_NumComponents], edx
 
-ifdef _DEBUG
-        call    DEBUG_TrialAllocSetAppDomain_NoScratchArea
-endif ; _DEBUG
-
         ret
 
     OversizedArray:
@@ -279,7 +244,7 @@ LEAF_ENTRY JIT_NewArr1OBJ_MP_InlineGetThread, _TEXT
         ; No need for rounding in this case - element size is 8, and m_BaseSize is guaranteed
         ; to be a multiple of 8.
 
-        PATCHABLE_INLINE_GETTHREAD r11, JIT_NewArr1OBJ_MP_InlineGetThread__PatchTLSOffset
+        INLINE_GETTHREAD r11
         mov     r10, [r11 + OFFSET__Thread__m_alloc_context__alloc_limit]
         mov     rax, [r11 + OFFSET__Thread__m_alloc_context__alloc_ptr]
 
@@ -293,10 +258,6 @@ LEAF_ENTRY JIT_NewArr1OBJ_MP_InlineGetThread, _TEXT
 
         mov     dword ptr [rax + OFFSETOF__ArrayBase__m_NumComponents], edx
 
-ifdef _DEBUG
-        call    DEBUG_TrialAllocSetAppDomain_NoScratchArea
-endif ; _DEBUG
-
         ret
 
     OversizedArray:
diff --git a/src/vm/amd64/JitHelpers_Slow.asm b/src/vm/amd64/JitHelpers_Slow.asm
index f86d429e33..0e26ae6dfd 100644
--- a/src/vm/amd64/JitHelpers_Slow.asm
+++ b/src/vm/amd64/JitHelpers_Slow.asm
@@ -42,7 +42,6 @@ EXTERN  g_GCShadowEnd:QWORD
 endif
 
 JIT_NEW                 equ     ?JIT_New@@YAPEAVObject@@PEAUCORINFO_CLASS_STRUCT_@@@Z
-Object__DEBUG_SetAppDomain equ ?DEBUG_SetAppDomain@Object@@QEAAXPEAVAppDomain@@@Z
 CopyValueClassUnchecked equ     ?CopyValueClassUnchecked@@YAXPEAX0PEAVMethodTable@@@Z
 JIT_Box                 equ     ?JIT_Box@@YAPEAVObject@@PEAUCORINFO_CLASS_STRUCT_@@PEAX@Z
 g_pStringClass          equ     ?g_pStringClass@@3PEAVMethodTable@@EA
@@ -162,290 +161,6 @@ endif
 LEAF_END_MARKED JIT_WriteBarrier_Debug, _TEXT
 endif
 
-NESTED_ENTRY JIT_TrialAllocSFastMP, _TEXT
-        alloc_stack      MIN_SIZE
-        END_PROLOGUE
-
-        CALL_GETTHREAD
-        mov     r11, rax
-
-        mov     r8d, [rcx + OFFSET__MethodTable__m_BaseSize]
-
-        ; m_BaseSize is guaranteed to be a multiple of 8.
-
-        mov     r10, [r11 + OFFSET__Thread__m_alloc_context__alloc_limit]
-        mov     rax, [r11 + OFFSET__Thread__m_alloc_context__alloc_ptr]
-
-        add     r8, rax
-
-        cmp     r8, r10
-        ja      AllocFailed
-
-        mov     [r11 + OFFSET__Thread__m_alloc_context__alloc_ptr], r8
-        mov     [rax], rcx
-
-ifdef _DEBUG
-        call    DEBUG_TrialAllocSetAppDomain
-endif ; _DEBUG
-
-        ; epilog
-        add     rsp, MIN_SIZE
-        ret
-
-    AllocFailed:
-        add     rsp, MIN_SIZE
-        jmp     JIT_NEW
-NESTED_END JIT_TrialAllocSFastMP, _TEXT
-
-
-; HCIMPL2(Object*, JIT_Box, CORINFO_CLASS_HANDLE type, void* unboxedData)
-NESTED_ENTRY JIT_BoxFastMP, _TEXT
-        alloc_stack MIN_SIZE
-        END_PROLOGUE
-
-        mov     rax, [rcx + OFFSETOF__MethodTable__m_pWriteableData]
-
-        ; Check whether the class has not been initialized
-        test    dword ptr [rax + OFFSETOF__MethodTableWriteableData__m_dwFlags], MethodTableWriteableData__enum_flag_Unrestored
-        jnz     ClassNotInited
-
-        CALL_GETTHREAD
-        mov     r11, rax
-
-        mov     r8d, [rcx + OFFSET__MethodTable__m_BaseSize]
-
-        ; m_BaseSize is guaranteed to be a multiple of 8.
-
-        mov     r10, [r11 + OFFSET__Thread__m_alloc_context__alloc_limit]
-        mov     rax, [r11 + OFFSET__Thread__m_alloc_context__alloc_ptr]
-
-        add     r8, rax
-
-        cmp     r8, r10
-        ja      AllocFailed
-
-        mov     [r11 + OFFSET__Thread__m_alloc_context__alloc_ptr], r8
-        mov     [rax], rcx
-
-ifdef _DEBUG
-        call    DEBUG_TrialAllocSetAppDomain
-endif ; _DEBUG
-
-        ; Check whether the object contains pointers
-        test    dword ptr [rcx + OFFSETOF__MethodTable__m_dwFlags], MethodTable__enum_flag_ContainsPointers
-        jnz     ContainsPointers
-
-        ; We have no pointers - emit a simple inline copy loop
-
-        mov     ecx, [rcx + OFFSET__MethodTable__m_BaseSize]
-        sub     ecx, 18h  ; sizeof(ObjHeader) + sizeof(Object) + last slot
-
-    CopyLoop:
-        mov     r8, [rdx+rcx]
-        mov     [rax+rcx+8], r8
-
-        sub     ecx, 8
-        jge     CopyLoop
-
-        add     rsp, MIN_SIZE
-        ret
-
-    ContainsPointers:
-        ; Do call to CopyValueClassUnchecked(object, data, pMT)
-
-        mov     [rsp+20h], rax
-
-        mov     r8, rcx
-        lea     rcx, [rax + 8]
-        call    CopyValueClassUnchecked
-
-        mov     rax, [rsp+20h]
-
-        add     rsp, MIN_SIZE
-        ret
-
-    ClassNotInited:
-    AllocFailed:
-        add     rsp, MIN_SIZE
-        jmp     JIT_Box
-NESTED_END JIT_BoxFastMP, _TEXT
-
-
-NESTED_ENTRY AllocateStringFastMP, _TEXT
-        alloc_stack MIN_SIZE
-        END_PROLOGUE
-
-        ; Instead of doing elaborate overflow checks, we just limit the number of elements
-        ; to (LARGE_OBJECT_SIZE - 256)/sizeof(WCHAR) or less.
-        ; This will avoid all overflow problems, as well as making sure
-        ; big string objects are correctly allocated in the big object heap.
-
-        cmp     ecx, (ASM_LARGE_OBJECT_SIZE - 256)/2
-        jae     OversizedString
-
-        CALL_GETTHREAD
-        mov     r11, rax
-
-        mov     rdx, [g_pStringClass]
-        mov     r8d, [rdx + OFFSET__MethodTable__m_BaseSize]
-
-        ; Calculate the final size to allocate.
-        ; We need to calculate baseSize + cnt*2, then round that up by adding 7 and anding ~7.
-
-        lea     r8d, [r8d + ecx*2 + 7]
-        and     r8d, -8
-
-        mov     r10, [r11 + OFFSET__Thread__m_alloc_context__alloc_limit]
-        mov     rax, [r11 + OFFSET__Thread__m_alloc_context__alloc_ptr]
-
-        add     r8, rax
-
-        cmp     r8, r10
-        ja      AllocFailed
-
-        mov     [r11 + OFFSET__Thread__m_alloc_context__alloc_ptr], r8
-        mov     [rax], rdx
-
-        mov     [rax + OFFSETOF__StringObject__m_StringLength], ecx
-
-ifdef _DEBUG
-        call    DEBUG_TrialAllocSetAppDomain
-endif ; _DEBUG
-
-        add     rsp, MIN_SIZE
-        ret
-
-    OversizedString:
-    AllocFailed:
-        add     rsp, MIN_SIZE
-        jmp     FramedAllocateString
-NESTED_END AllocateStringFastMP, _TEXT
-
-; HCIMPL2(Object*, JIT_NewArr1VC_MP, CORINFO_CLASS_HANDLE arrayMT, INT_PTR size)
-NESTED_ENTRY JIT_NewArr1VC_MP, _TEXT
-        alloc_stack MIN_SIZE
-        END_PROLOGUE
-
-        ; We were passed a (shared) method table in RCX, which contains the element type.
-
-        ; The element count is in RDX
-
-        ; NOTE: if this code is ported for CORINFO_HELP_NEWSFAST_ALIGN8, it will need
-        ; to emulate the double-specific behavior of JIT_TrialAlloc::GenAllocArray.
-
-        ; Do a conservative check here.  This is to avoid overflow while doing the calculations.  We don't
-        ; have to worry about "large" objects, since the allocation quantum is never big enough for
-        ; LARGE_OBJECT_SIZE.
-
-        ; For Value Classes, this needs to be 2^16 - slack (2^32 / max component size), 
-        ; The slack includes the size for the array header and round-up ; for alignment.  Use 256 for the
-        ; slack value out of laziness.
-
-        ; In both cases we do a final overflow check after adding to the alloc_ptr.
-
-        CALL_GETTHREAD
-        mov     r11, rax
-
-        cmp     rdx, (65535 - 256)
-        jae     OversizedArray
-
-        movzx   r8d, word ptr [rcx + OFFSETOF__MethodTable__m_dwFlags]  ; component size is low 16 bits
-        imul    r8d, edx  ; signed mul, but won't overflow due to length restriction above
-        add     r8d, dword ptr [rcx + OFFSET__MethodTable__m_BaseSize]
-
-        ; round the size to a multiple of 8
-
-        add     r8d, 7
-        and     r8d, -8
-
-        mov     r10, [r11 + OFFSET__Thread__m_alloc_context__alloc_limit]
-        mov     rax, [r11 + OFFSET__Thread__m_alloc_context__alloc_ptr]
-
-        add     r8, rax
-        jc      AllocFailed
-
-        cmp     r8, r10
-        ja      AllocFailed
-
-        mov     [r11 + OFFSET__Thread__m_alloc_context__alloc_ptr], r8
-        mov     [rax], rcx
-
-        mov     dword ptr [rax + OFFSETOF__ArrayBase__m_NumComponents], edx
-
-ifdef _DEBUG
-        call    DEBUG_TrialAllocSetAppDomain
-endif ; _DEBUG
-
-        add     rsp, MIN_SIZE
-        ret
-
-    OversizedArray:
-    AllocFailed:
-        add     rsp, MIN_SIZE
-        jmp     JIT_NewArr1
-NESTED_END JIT_NewArr1VC_MP, _TEXT
-
-
-; HCIMPL2(Object*, JIT_NewArr1OBJ_MP, CORINFO_CLASS_HANDLE arrayMT, INT_PTR size)
-NESTED_ENTRY JIT_NewArr1OBJ_MP, _TEXT
-        alloc_stack MIN_SIZE
-        END_PROLOGUE
-
-        ; We were passed a (shared) method table in RCX, which contains the element type.
-
-        ; The element count is in RDX
-
-        ; NOTE: if this code is ported for CORINFO_HELP_NEWSFAST_ALIGN8, it will need
-        ; to emulate the double-specific behavior of JIT_TrialAlloc::GenAllocArray.
-
-        ; Verifies that LARGE_OBJECT_SIZE fits in 32-bit.  This allows us to do array size
-        ; arithmetic using 32-bit registers.
-        .erre ASM_LARGE_OBJECT_SIZE lt 100000000h
-
-        cmp     rdx, (ASM_LARGE_OBJECT_SIZE - 256)/8
-        jae     OversizedArray
-
-        CALL_GETTHREAD
-        mov     r11, rax
-
-        ; In this case we know the element size is sizeof(void *), or 8 for x64
-        ; This helps us in two ways - we can shift instead of multiplying, and
-        ; there's no need to align the size either
-
-        mov     r8d, dword ptr [rcx + OFFSET__MethodTable__m_BaseSize]
-        lea     r8d, [r8d + edx * 8]
-
-        ; No need for rounding in this case - element size is 8, and m_BaseSize is guaranteed
-        ; to be a multiple of 8.
-
-        mov     r10, [r11 + OFFSET__Thread__m_alloc_context__alloc_limit]
-        mov     rax, [r11 + OFFSET__Thread__m_alloc_context__alloc_ptr]
-
-        add     r8, rax
-
-        cmp     r8, r10
-        ja      AllocFailed
-
-        mov     [r11 + OFFSET__Thread__m_alloc_context__alloc_ptr], r8
-        mov     [rax], rcx
-
-        mov     dword ptr [rax + OFFSETOF__ArrayBase__m_NumComponents], edx
-
-ifdef _DEBUG
-        call    DEBUG_TrialAllocSetAppDomain
-endif ; _DEBUG
-
-        add     rsp, MIN_SIZE
-        ret
-
-    OversizedArray:
-    AllocFailed:
-        add     rsp, MIN_SIZE
-        jmp     JIT_NewArr1
-NESTED_END JIT_NewArr1OBJ_MP, _TEXT
-
-
-
 
 extern g_global_alloc_lock:dword
 extern g_global_alloc_context:qword
@@ -471,10 +186,6 @@ LEAF_ENTRY JIT_TrialAllocSFastSP, _TEXT
         mov     [rax], rcx
         mov     [g_global_alloc_lock], -1
 
-ifdef _DEBUG
-        call    DEBUG_TrialAllocSetAppDomain_NoScratchArea
-endif ; _DEBUG
-
         ret
 
     AllocFailed:
@@ -511,10 +222,6 @@ NESTED_ENTRY JIT_BoxFastUP, _TEXT
         mov     [rax], rcx
         mov     [g_global_alloc_lock], -1
 
-ifdef _DEBUG
-        call    DEBUG_TrialAllocSetAppDomain_NoScratchArea
-endif ; _DEBUG
-
         ; Check whether the object contains pointers
         test    dword ptr [rcx + OFFSETOF__MethodTable__m_dwFlags], MethodTable__enum_flag_ContainsPointers
         jnz     ContainsPointers
@@ -594,10 +301,6 @@ LEAF_ENTRY AllocateStringFastUP, _TEXT
 
         mov     [rax + OFFSETOF__StringObject__m_StringLength], ecx
 
-ifdef _DEBUG
-        call    DEBUG_TrialAllocSetAppDomain_NoScratchArea
-endif ; _DEBUG
-
         ret
 
     AllocFailed:
@@ -655,10 +358,6 @@ LEAF_ENTRY JIT_NewArr1VC_UP, _TEXT
 
         mov     dword ptr [rax + OFFSETOF__ArrayBase__m_NumComponents], edx
 
-ifdef _DEBUG
-        call    DEBUG_TrialAllocSetAppDomain_NoScratchArea
-endif ; _DEBUG
-
         ret
 
     AllocFailed:
@@ -711,10 +410,6 @@ LEAF_ENTRY JIT_NewArr1OBJ_UP, _TEXT
 
         mov     dword ptr [rax + OFFSETOF__ArrayBase__m_NumComponents], edx
 
-ifdef _DEBUG
-        call    DEBUG_TrialAllocSetAppDomain_NoScratchArea
-endif ; _DEBUG
-
         ret
 
     AllocFailed:
@@ -725,181 +420,5 @@ endif ; _DEBUG
 LEAF_END JIT_NewArr1OBJ_UP, _TEXT
 
 
-NESTED_ENTRY JIT_GetSharedNonGCStaticBase_Slow, _TEXT
-        alloc_stack     MIN_SIZE
-        END_PROLOGUE
-
-        ; Check if rcx (moduleDomainID) is not a moduleID
-        test    rcx, 1
-        jz      HaveLocalModule
-
-        CALL_GETAPPDOMAIN
-
-        ; Get the LocalModule
-        mov     rax, [rax + OFFSETOF__AppDomain__m_sDomainLocalBlock + OFFSETOF__DomainLocalBlock__m_pModuleSlots]
-        ; rcx will always be odd, so: rcx * 4 - 4 <=> (rcx >> 1) * 8
-        mov     rcx, [rax + rcx * 4 - 4]
-
-    HaveLocalModule:
-        ; If class is not initialized, bail to C++ helper
-        test    [rcx + OFFSETOF__DomainLocalModule__m_pDataBlob + rdx], 1
-        jz      CallHelper
-
-        mov     rax, rcx
-        add     rsp, MIN_SIZE
-        ret
-
-    align 16
-    CallHelper:
-        ; Tail call Jit_GetSharedNonGCStaticBase_Helper
-        add     rsp, MIN_SIZE
-        jmp     JIT_GetSharedNonGCStaticBase_Helper
-NESTED_END JIT_GetSharedNonGCStaticBase_Slow, _TEXT
-
-NESTED_ENTRY JIT_GetSharedNonGCStaticBaseNoCtor_Slow, _TEXT
-        alloc_stack     MIN_SIZE
-        END_PROLOGUE
-
-        ; Check if rcx (moduleDomainID) is not a moduleID
-        test    rcx, 1
-        jz      HaveLocalModule
-
-        CALL_GETAPPDOMAIN
-
-        ; Get the LocalModule
-        mov     rax, [rax + OFFSETOF__AppDomain__m_sDomainLocalBlock + OFFSETOF__DomainLocalBlock__m_pModuleSlots]
-        ; rcx will always be odd, so: rcx * 4 - 4 <=> (rcx >> 1) * 8
-        mov     rax, [rax + rcx * 4 - 4]
-
-        add     rsp, MIN_SIZE
-        ret
-
-    align 16
-    HaveLocalModule:
-        mov     rax, rcx
-        add     rsp, MIN_SIZE
-        ret
-NESTED_END JIT_GetSharedNonGCStaticBaseNoCtor_Slow, _TEXT
-
-NESTED_ENTRY JIT_GetSharedGCStaticBase_Slow, _TEXT
-        alloc_stack     MIN_SIZE
-        END_PROLOGUE
-
-        ; Check if rcx (moduleDomainID) is not a moduleID
-        test    rcx, 1
-        jz      HaveLocalModule
-
-        CALL_GETAPPDOMAIN
-
-        ; Get the LocalModule
-        mov     rax, [rax + OFFSETOF__AppDomain__m_sDomainLocalBlock + OFFSETOF__DomainLocalBlock__m_pModuleSlots]
-        ; rcx will always be odd, so: rcx * 4 - 4 <=> (rcx >> 1) * 8
-        mov     rcx, [rax + rcx * 4 - 4]
-
-    HaveLocalModule:
-        ; If class is not initialized, bail to C++ helper
-        test    [rcx + OFFSETOF__DomainLocalModule__m_pDataBlob + rdx], 1
-        jz      CallHelper
-
-        mov     rax, [rcx + OFFSETOF__DomainLocalModule__m_pGCStatics]
-
-        add     rsp, MIN_SIZE
-        ret
-
-    align 16
-    CallHelper:
-        ; Tail call Jit_GetSharedGCStaticBase_Helper
-        add     rsp, MIN_SIZE
-        jmp     JIT_GetSharedGCStaticBase_Helper
-NESTED_END JIT_GetSharedGCStaticBase_Slow, _TEXT
-
-NESTED_ENTRY JIT_GetSharedGCStaticBaseNoCtor_Slow, _TEXT
-        alloc_stack     MIN_SIZE
-        END_PROLOGUE
-
-        ; Check if rcx (moduleDomainID) is not a moduleID
-        test    rcx, 1
-        jz      HaveLocalModule
-
-        CALL_GETAPPDOMAIN
-
-        ; Get the LocalModule
-        mov     rax, [rax + OFFSETOF__AppDomain__m_sDomainLocalBlock + OFFSETOF__DomainLocalBlock__m_pModuleSlots]
-        ; rcx will always be odd, so: rcx * 4 - 4 <=> (rcx >> 1) * 8
-        mov     rcx, [rax + rcx * 4 - 4]
-
-    HaveLocalModule:
-        mov     rax, [rcx + OFFSETOF__DomainLocalModule__m_pGCStatics]
-
-        add     rsp, MIN_SIZE
-        ret
-NESTED_END JIT_GetSharedGCStaticBaseNoCtor_Slow, _TEXT
-
-
-ifdef _DEBUG
-
-extern Object__DEBUG_SetAppDomain:proc
-
-;
-; IN: rax: new object needing the AppDomain ID set..
-; OUT: rax, returns original value at entry
-;
-; all integer register state is preserved
-;
-DEBUG_TrialAllocSetAppDomain_STACK_SIZE         equ MIN_SIZE + 10h
-NESTED_ENTRY DEBUG_TrialAllocSetAppDomain, _TEXT
-        push_vol_reg    rax
-        push_vol_reg    rcx
-        push_vol_reg    rdx
-        push_vol_reg    r8
-        push_vol_reg    r9
-        push_vol_reg    r10
-        push_vol_reg    r11
-        push_nonvol_reg rbx
-        alloc_stack     MIN_SIZE
-        END_PROLOGUE
-
-        mov             rbx, rax
-
-        ; get the app domain ptr
-        CALL_GETAPPDOMAIN
-
-        ; set the sync block app domain ID
-        mov             rcx, rbx
-        mov             rdx, rax
-        call            Object__DEBUG_SetAppDomain
-
-        ; epilog
-        add             rsp, MIN_SIZE
-        pop             rbx
-        pop             r11
-        pop             r10
-        pop             r9
-        pop             r8
-        pop             rdx
-        pop             rcx
-        pop             rax
-        ret
-NESTED_END DEBUG_TrialAllocSetAppDomain, _TEXT
-
-NESTED_ENTRY DEBUG_TrialAllocSetAppDomain_NoScratchArea, _TEXT
-
-        push_nonvol_reg rbp
-        set_frame       rbp, 0
-        END_PROLOGUE
-
-        sub             rsp, 20h
-        and             rsp, -16
-
-        call            DEBUG_TrialAllocSetAppDomain
-
-        lea             rsp, [rbp+0]
-        pop             rbp
-        ret
-NESTED_END DEBUG_TrialAllocSetAppDomain_NoScratchArea, _TEXT
-
-endif
-
-
         end
 
diff --git a/src/vm/amd64/RedirectedHandledJITCase.asm b/src/vm/amd64/RedirectedHandledJITCase.asm
index a6d3496357..5ab69475e2 100644
--- a/src/vm/amd64/RedirectedHandledJITCase.asm
+++ b/src/vm/amd64/RedirectedHandledJITCase.asm
@@ -195,9 +195,8 @@ NESTED_ENTRY RedirectForThrowControl2, _TEXT
         END_PROLOGUE
 
         ; Fetch rip from a CONTEXT, and store it as our return address.
-        CALL_GETTHREAD
+        INLINE_GETTHREAD rcx
 
-        mov             rcx, rax
         call            Thread__GetAbortContext
 
         mov             rax, [rax + OFFSETOF__CONTEXT__Rip]
diff --git a/src/vm/amd64/TlsGetters.asm b/src/vm/amd64/TlsGetters.asm
deleted file mode 100644
index 7b5a30844b..0000000000
--- a/src/vm/amd64/TlsGetters.asm
+++ /dev/null
@@ -1,120 +0,0 @@
-; Licensed to the .NET Foundation under one or more agreements.
-; The .NET Foundation licenses this file to you under the MIT license.
-; See the LICENSE file in the project root for more information.
-
-; ==++==
-;
-
-;
-; ==--==
-; ***********************************************************************
-; File: TlsGetters.asm, see history in jithelp.asm
-;
-; Notes: These TlsGetters (GetAppDomain(), GetThread()) are implemented
-;        in a generic fashion, but might be patched at runtime to contain
-;        a much faster implementation which goes straight to the TLS for
-;        the Thread* or AppDomain*. 
-;
-;        Note that the macro takes special care to not have these become 
-;        non-unwindable after the patching has overwritten the prologue of 
-;        the generic getter.
-; ***********************************************************************
-
-include AsmMacros.inc
-include asmconstants.inc
-
-; Min amount of stack space that a nested function should allocate.
-MIN_SIZE equ 28h
-
-
-; These generic TLS getters are used for GetThread() and GetAppDomain(), they do a little
-; extra work to ensure that certain registers are preserved, those include the following 
-; volatile registers
-; 
-;       rcx
-;       rdx
-;       r8
-;       r9
-;       r10
-;       r11
-;       
-; The return value is in rax as usual
-;
-; They DO NOT save scratch flowing point registers, if you need those you need to save them.
-
-ifdef ENABLE_GET_THREAD_GENERIC_FULL_CHECK
-GetThreadGenericFullCheck equ ?GetThreadGenericFullCheck@@YAPEAVThread@@XZ
-extern GetThreadGenericFullCheck:proc
-endif ; ENABLE_GET_THREAD_GENERIC_FULL_CHECK
-
-; Creates a generic TLS getter using the value from TLS slot gTLSIndex.  Set GenerateGetThread
-; when using this macro to generate GetThread, as that will cause special code to be generated which
-; enables additional debug-only checking, such as enforcement of EE_THREAD_NOT_REQUIRED contracts
-GenerateOptimizedTLSGetter macro name, GenerateGetThread
-
-extern g&name&TLSIndex:dword
-extern __imp_TlsGetValue:qword
-
-SIZEOF_PUSHED_ARGS equ 10h
-
-NESTED_ENTRY Get&name&Generic, _TEXT
-        push_vol_reg            r10
-        push_vol_reg            r11
-        alloc_stack             MIN_SIZE
-
-        ; save argument registers in shadow space
-        save_reg_postrsp        rcx, MIN_SIZE + 8h  + SIZEOF_PUSHED_ARGS
-        save_reg_postrsp        rdx, MIN_SIZE + 10h + SIZEOF_PUSHED_ARGS
-        save_reg_postrsp        r8,  MIN_SIZE + 18h + SIZEOF_PUSHED_ARGS
-        save_reg_postrsp        r9,  MIN_SIZE + 20h + SIZEOF_PUSHED_ARGS
-    END_PROLOGUE
-
-ifdef _DEBUG
-        cmp     dword ptr [g&name&TLSIndex], -1
-        jnz     @F
-        int     3
-@@:
-endif ; _DEBUG
-
-CALL_GET_THREAD_GENERIC_FULL_CHECK=0
-
-ifdef ENABLE_GET_THREAD_GENERIC_FULL_CHECK
-if GenerateGetThread
-
-; Generating the GetThread() tlsgetter, and GetThreadGenericFullCheck is
-; defined in C (in threads.cpp).  So we'll want to delegate directly to
-; GetThreadGenericFullCheck, which may choose to do additional checking, such
-; as enforcing EE_THREAD_NOT_REQUIRED contracts
-CALL_GET_THREAD_GENERIC_FULL_CHECK=1
-
-endif   ; GenerateGetThread
-endif   ; ENABLE_GET_THREAD_GENERIC_FULL_CHECK
-
-if CALL_GET_THREAD_GENERIC_FULL_CHECK
-        call    GetThreadGenericFullCheck
-else
-        ; Not generating the GetThread() tlsgetter (or there is no GetThreadGenericFullCheck
-        ; to call), so do nothing special--just look up the value stored at TLS slot gTLSIndex
-        mov     ecx, [g&name&TLSIndex]
-        call    [__imp_TlsGetValue]
-endif
-
-        ; restore arguments from shadow space
-        mov     rcx, [rsp + MIN_SIZE + 8h  + SIZEOF_PUSHED_ARGS]
-        mov     rdx, [rsp + MIN_SIZE + 10h + SIZEOF_PUSHED_ARGS]
-        mov     r8,  [rsp + MIN_SIZE + 18h + SIZEOF_PUSHED_ARGS]
-        mov     r9,  [rsp + MIN_SIZE + 20h + SIZEOF_PUSHED_ARGS]
-
-        ; epilog
-        add     rsp, MIN_SIZE
-        pop     r11
-        pop     r10
-        ret
-NESTED_END Get&name&Generic, _TEXT
-
-        endm    
-
-GenerateOptimizedTLSGetter Thread, 1
-GenerateOptimizedTLSGetter AppDomain, 0
-
-        end
diff --git a/src/vm/amd64/UMThunkStub.asm b/src/vm/amd64/UMThunkStub.asm
index dd6d4ebc3c..8b44e67a7e 100644
--- a/src/vm/amd64/UMThunkStub.asm
+++ b/src/vm/amd64/UMThunkStub.asm
@@ -166,14 +166,12 @@ UMThunkStubAMD64_FIXED_STACK_ALLOC_SIZE = UMThunkStubAMD64_STACK_FRAME_SIZE - (U
         ;
         ; Call GetThread()
         ;
-        CALL_GETTHREAD                          ; will not trash r10
-        test            rax, rax
+        INLINE_GETTHREAD r12                    ; will not trash r10
+        test            r12, r12
         jz              DoThreadSetup
 
 HaveThread:
 
-        mov             r12, rax                ; r12 <- Thread*
-
         ;FailFast if a native callable method invoked via ldftn and calli.
         cmp             dword ptr [r12 + OFFSETOF__Thread__m_fPreemptiveGCDisabled], 1
         jz              InvalidTransition
@@ -250,6 +248,8 @@ DoThreadSetup:
         movdqa          xmm1, xmmword ptr [rbp + UMThunkStubAMD64_XMM_SAVE_OFFSET + 10h]
         movdqa          xmm2, xmmword ptr [rbp + UMThunkStubAMD64_XMM_SAVE_OFFSET + 20h]
         movdqa          xmm3, xmmword ptr [rbp + UMThunkStubAMD64_XMM_SAVE_OFFSET + 30h]
+
+        mov             r12, rax
         
         jmp             HaveThread
         
diff --git a/src/vm/amd64/asmconstants.h b/src/vm/amd64/asmconstants.h
index 1fef80f66d..b6a3c712e7 100644
--- a/src/vm/amd64/asmconstants.h
+++ b/src/vm/amd64/asmconstants.h
@@ -535,22 +535,9 @@ ASMCONSTANTS_C_ASSERT(MethodDescClassification__mdcClassification == mdcClassifi
 ASMCONSTANTS_C_ASSERT(MethodDescClassification__mcInstantiated == mcInstantiated);
 
 #ifndef FEATURE_PAL
-
-#define OFFSET__TEB__TlsSlots 0x1480
-ASMCONSTANTS_C_ASSERT(OFFSET__TEB__TlsSlots == offsetof(TEB, TlsSlots));
-
-#define OFFSETOF__TEB__LastErrorValue 0x68
-ASMCONSTANTS_C_ASSERT(OFFSETOF__TEB__LastErrorValue == offsetof(TEB, LastErrorValue));
-
-#endif // !FEATURE_PAL
-
-#ifdef _DEBUG
-#define TLS_GETTER_MAX_SIZE_ASM 0x30
-#else
-#define TLS_GETTER_MAX_SIZE_ASM 0x18
+#define OFFSET__TEB__ThreadLocalStoragePointer 0x58
+ASMCONSTANTS_C_ASSERT(OFFSET__TEB__ThreadLocalStoragePointer == offsetof(TEB, ThreadLocalStoragePointer));
 #endif
-ASMCONSTANTS_C_ASSERT(TLS_GETTER_MAX_SIZE_ASM == TLS_GETTER_MAX_SIZE)
-
 
 // If you change these constants, you need to update code in
 // RedirectHandledJITCase.asm and ExcepAMD64.cpp.
diff --git a/src/vm/amd64/cgencpu.h b/src/vm/amd64/cgencpu.h
index 98e9770858..b83437039b 100644
--- a/src/vm/amd64/cgencpu.h
+++ b/src/vm/amd64/cgencpu.h
@@ -104,13 +104,6 @@ EXTERN_C void FastCallFinalizeWorker(Object *obj, PCODE funcPtr);
 #define X86RegFromAMD64Reg(extended_reg) \
             ((X86Reg)(((int)extended_reg) & X86_REGISTER_MASK))
 
-// Max size of optimized TLS helpers
-#ifdef _DEBUG
-// Debug build needs extra space for last error trashing
-#define TLS_GETTER_MAX_SIZE 0x30
-#else
-#define TLS_GETTER_MAX_SIZE 0x18
-#endif
 
 //=======================================================================
 // IMPORTANT: This value is used to figure out how much to allocate
@@ -538,26 +531,22 @@ inline BOOL ClrFlushInstructionCache(LPCVOID pCodeAddr, size_t sizeOfCode)
     return TRUE;
 }
 
-#ifndef FEATURE_IMPLICIT_TLS
 //
 // JIT HELPER ALIASING FOR PORTABILITY.
 //
 // Create alias for optimized implementations of helpers provided on this platform
 //
-#define JIT_GetSharedGCStaticBase           JIT_GetSharedGCStaticBase_InlineGetAppDomain
-#define JIT_GetSharedNonGCStaticBase        JIT_GetSharedNonGCStaticBase_InlineGetAppDomain
-#define JIT_GetSharedGCStaticBaseNoCtor     JIT_GetSharedGCStaticBaseNoCtor_InlineGetAppDomain
-#define JIT_GetSharedNonGCStaticBaseNoCtor  JIT_GetSharedNonGCStaticBaseNoCtor_InlineGetAppDomain
-#endif // FEATURE_IMPLICIT_TLS
+#define JIT_GetSharedGCStaticBase           JIT_GetSharedGCStaticBase_SingleAppDomain
+#define JIT_GetSharedNonGCStaticBase        JIT_GetSharedNonGCStaticBase_SingleAppDomain
+#define JIT_GetSharedGCStaticBaseNoCtor     JIT_GetSharedGCStaticBaseNoCtor_SingleAppDomain
+#define JIT_GetSharedNonGCStaticBaseNoCtor  JIT_GetSharedNonGCStaticBaseNoCtor_SingleAppDomain
 
 #ifndef FEATURE_PAL
-
 #define JIT_ChkCastClass            JIT_ChkCastClass
 #define JIT_ChkCastClassSpecial     JIT_ChkCastClassSpecial
 #define JIT_IsInstanceOfClass       JIT_IsInstanceOfClass
 #define JIT_ChkCastInterface        JIT_ChkCastInterface
 #define JIT_IsInstanceOfInterface   JIT_IsInstanceOfInterface
-
 #endif // FEATURE_PAL
 
 #define JIT_Stelem_Ref              JIT_Stelem_Ref
diff --git a/src/vm/appdomain.cpp b/src/vm/appdomain.cpp
index 7b0da7f5a2..edcc3df1d7 100644
--- a/src/vm/appdomain.cpp
+++ b/src/vm/appdomain.cpp
@@ -2406,10 +2406,6 @@ void SystemDomain::Init()
     _ASSERTE(curCtx->GetDomain() != NULL);
 #endif
 
-#ifdef _DEBUG
-    g_fVerifierOff = g_pConfig->IsVerifierOff();
-#endif
-
 #ifdef FEATURE_PREJIT
     if (CLRConfig::GetConfigValue(CLRConfig::EXTERNAL_ZapDisable) != 0)
         g_fAllowNativeImages = false;
diff --git a/src/vm/arm/asmconstants.h b/src/vm/arm/asmconstants.h
index 704fa28556..ab3d16b9a9 100644
--- a/src/vm/arm/asmconstants.h
+++ b/src/vm/arm/asmconstants.h
@@ -256,9 +256,6 @@ ASMCONSTANTS_C_ASSERT(VASigCookie__pNDirectILStub == offsetof(VASigCookie, pNDir
 #define CONTEXT_Pc 0x040
 ASMCONSTANTS_C_ASSERT(CONTEXT_Pc == offsetof(T_CONTEXT,Pc))
 
-#define TLS_GETTER_MAX_SIZE_ASM 0x10
-ASMCONSTANTS_C_ASSERT(TLS_GETTER_MAX_SIZE_ASM == TLS_GETTER_MAX_SIZE)
-
 #define CallDescrData__pSrc                0x00
 #define CallDescrData__numStackSlots       0x04
 #define CallDescrData__pArgumentRegisters  0x08
diff --git a/src/vm/arm/asmhelpers.asm b/src/vm/arm/asmhelpers.asm
index e5fd41a513..4d21d07462 100644
--- a/src/vm/arm/asmhelpers.asm
+++ b/src/vm/arm/asmhelpers.asm
@@ -1481,7 +1481,6 @@ stackProbe_loop
     NESTED_END
 
 
-#ifdef FEATURE_CORECLR
 ;
 ; JIT Static access helpers for single appdomain case
 ;
@@ -1543,7 +1542,6 @@ CallCppHelper3
     bx lr
     LEAF_END
 
-#endif
 
 ; ------------------------------------------------------------------
 ; __declspec(naked) void F_CALL_CONV JIT_Stelem_Ref(PtrArray* array, unsigned idx, Object* val)
diff --git a/src/vm/arm/cgencpu.h b/src/vm/arm/cgencpu.h
index 2a369d8f02..8da2b2b3cc 100644
--- a/src/vm/arm/cgencpu.h
+++ b/src/vm/arm/cgencpu.h
@@ -83,9 +83,6 @@ EXTERN_C void setFPReturn(int fpSize, INT64 retVal);
 #define CALLDESCR_ARGREGS                       1   // CallDescrWorker has ArgumentRegister parameter
 #define CALLDESCR_FPARGREGS                     1   // CallDescrWorker has FloatArgumentRegisters parameter
 
-// Max size of optimized TLS helpers
-#define TLS_GETTER_MAX_SIZE 0x10
-
 // Given a return address retrieved during stackwalk,
 // this is the offset by which it should be decremented to arrive at the callsite.
 #define STACKWALK_CONTROLPC_ADJUST_OFFSET 2
@@ -552,7 +549,7 @@ public:
         ThumbEmitJumpRegister(thumbRegLr);
     }
 
-    void ThumbEmitGetThread(TLSACCESSMODE mode, ThumbReg dest);
+    void ThumbEmitGetThread(ThumbReg dest);
 
     void ThumbEmitNop()
     {
@@ -1056,19 +1053,15 @@ inline BOOL ClrFlushInstructionCache(LPCVOID pCodeAddr, size_t sizeOfCode)
 #endif
 }
 
-#ifndef FEATURE_IMPLICIT_TLS
 //
 // JIT HELPER ALIASING FOR PORTABILITY.
 //
 // Create alias for optimized implementations of helpers provided on this platform
 //
-// optimized static helpers 
-#define JIT_GetSharedGCStaticBase           JIT_GetSharedGCStaticBase_InlineGetAppDomain
-#define JIT_GetSharedNonGCStaticBase        JIT_GetSharedNonGCStaticBase_InlineGetAppDomain
-#define JIT_GetSharedGCStaticBaseNoCtor     JIT_GetSharedGCStaticBaseNoCtor_InlineGetAppDomain
-#define JIT_GetSharedNonGCStaticBaseNoCtor  JIT_GetSharedNonGCStaticBaseNoCtor_InlineGetAppDomain
-
-#endif
+#define JIT_GetSharedGCStaticBase           JIT_GetSharedGCStaticBase_SingleAppDomain
+#define JIT_GetSharedNonGCStaticBase        JIT_GetSharedNonGCStaticBase_SingleAppDomain
+#define JIT_GetSharedGCStaticBaseNoCtor     JIT_GetSharedGCStaticBaseNoCtor_SingleAppDomain
+#define JIT_GetSharedNonGCStaticBaseNoCtor  JIT_GetSharedNonGCStaticBaseNoCtor_SingleAppDomain
 
 #ifndef FEATURE_PAL
 #define JIT_Stelem_Ref                      JIT_Stelem_Ref
diff --git a/src/vm/arm/patchedcode.S b/src/vm/arm/patchedcode.S
index 9335fe65fc..43adca25c8 100644
--- a/src/vm/arm/patchedcode.S
+++ b/src/vm/arm/patchedcode.S
@@ -20,22 +20,6 @@
     LEAF_END JIT_PatchedCodeStart, _TEXT
 
 // ------------------------------------------------------------------
-// Optimized TLS getters
-
-    LEAF_ENTRY GetTLSDummy, _TEXT
-        mov     r0, #0
-        bx      lr
-    LEAF_END GetTLSDummy, _TEXT
-
-        .align 4
-    LEAF_ENTRY ClrFlsGetBlock, _TEXT
-        // This will be overwritten at runtime with optimized ClrFlsGetBlock implementation
-        b       C_FUNC(GetTLSDummy)
-        // Just allocate space that will be filled in at runtime
-        .space (TLS_GETTER_MAX_SIZE_ASM - 2)
-    LEAF_END ClrFlsGetBlock, _TEXT
-
-// ------------------------------------------------------------------
 // GC write barrier support.
 //
 // GC Write barriers are defined in asmhelpers.asm. The following functions are used to define 
diff --git a/src/vm/arm/patchedcode.asm b/src/vm/arm/patchedcode.asm
index 9fdd60961d..c7e2322e38 100644
--- a/src/vm/arm/patchedcode.asm
+++ b/src/vm/arm/patchedcode.asm
@@ -13,52 +13,6 @@
 
 #include "asmmacros.h"
 
-    SETALIAS JIT_Box,?JIT_Box@@YAPAVObject@@PAUCORINFO_CLASS_STRUCT_@@PAX@Z
-    SETALIAS JIT_New, ?JIT_New@@YAPAVObject@@PAUCORINFO_CLASS_STRUCT_@@@Z
-    SETALIAS JIT_Box, ?JIT_Box@@YAPAVObject@@PAUCORINFO_CLASS_STRUCT_@@PAX@Z
-    SETALIAS FramedAllocateString, ?FramedAllocateString@@YAPAVStringObject@@K@Z
-    SETALIAS g_pStringClass, ?g_pStringClass@@3PAVMethodTable@@A
-    SETALIAS JIT_NewArr1, ?JIT_NewArr1@@YAPAVObject@@PAUCORINFO_CLASS_STRUCT_@@H@Z
-    SETALIAS CopyValueClassUnchecked, ?CopyValueClassUnchecked@@YAXPAX0PAVMethodTable@@@Z
-
-    IMPORT $JIT_New
-    IMPORT $JIT_Box
-    IMPORT $FramedAllocateString
-    IMPORT $g_pStringClass
-    IMPORT $JIT_NewArr1
-    IMPORT $CopyValueClassUnchecked
-    IMPORT SetAppDomainInObject
-
-
-    IMPORT JIT_GetSharedNonGCStaticBase_Helper
-    IMPORT JIT_GetSharedGCStaticBase_Helper
-
-
-    EXPORT JIT_TrialAllocSFastMP_InlineGetThread__PatchTLSOffset
-    EXPORT JIT_BoxFastMP_InlineGetThread__PatchTLSOffset
-    EXPORT AllocateStringFastMP_InlineGetThread__PatchTLSOffset
-    EXPORT JIT_NewArr1VC_MP_InlineGetThread__PatchTLSOffset
-    EXPORT JIT_NewArr1OBJ_MP_InlineGetThread__PatchTLSOffset
-
-    EXPORT JIT_GetSharedNonGCStaticBase__PatchTLSLabel
-    EXPORT JIT_GetSharedNonGCStaticBaseNoCtor__PatchTLSLabel
-    EXPORT JIT_GetSharedGCStaticBase__PatchTLSLabel
-    EXPORT JIT_GetSharedGCStaticBaseNoCtor__PatchTLSLabel
-
-    MACRO
-    PATCHABLE_INLINE_GETTHREAD $reg, $label
-$label
-    mrc p15, 0, $reg, c13, c0, 2
-    ldr $reg, [$reg, #0xe10]
-    MEND
-
-
-    MACRO
-    PATCHABLE_INLINE_GETAPPDOMAIN $reg, $label
-$label
-    mrc p15, 0, $reg, c13, c0, 2
-    ldr $reg, [$reg, #0xe10]
-    MEND
 
     TEXTAREA
 
@@ -70,38 +24,6 @@ $label
     LEAF_END
 
 ; ------------------------------------------------------------------
-; Optimized TLS getters
-
-        ALIGN 4
-    LEAF_ENTRY GetThread
-        ; This will be overwritten at runtime with optimized GetThread implementation
-        b       GetTLSDummy
-        ; Just allocate space that will be filled in at runtime
-        SPACE (TLS_GETTER_MAX_SIZE_ASM - 2)
-    LEAF_END
-
-        ALIGN 4
-    LEAF_ENTRY GetAppDomain
-        ; This will be overwritten at runtime with optimized GetThread implementation
-        b       GetTLSDummy
-        ; Just allocate space that will be filled in at runtime
-        SPACE (TLS_GETTER_MAX_SIZE_ASM - 2)
-    LEAF_END
-
-    LEAF_ENTRY GetTLSDummy
-        mov     r0, #0
-        bx      lr
-    LEAF_END
-
-        ALIGN 4
-    LEAF_ENTRY ClrFlsGetBlock
-        ; This will be overwritten at runtime with optimized ClrFlsGetBlock implementation
-        b       GetTLSDummy
-        ; Just allocate space that will be filled in at runtime
-        SPACE (TLS_GETTER_MAX_SIZE_ASM - 2)
-    LEAF_END
-
-; ------------------------------------------------------------------
 ; GC write barrier support.
 ;
 ; GC Write barriers are defined in asmhelpers.asm. The following functions are used to define 
@@ -134,443 +56,6 @@ $label
         bx lr
     LEAF_END
 
-; JIT Allocation helpers when TLS Index for Thread is low enough for fast helpers
-
-;---------------------------------------------------------------------------
-; IN: r0: MethodTable*
-;; OUT: r0: new object
-
-    LEAF_ENTRY JIT_TrialAllocSFastMP_InlineGetThread
-
-    ;get object size
-    ldr r1, [r0, #MethodTable__m_BaseSize]
-
-    ; m_BaseSize is guaranteed to be a multiple of 4.
-
-    ;getThread
-    PATCHABLE_INLINE_GETTHREAD r12, JIT_TrialAllocSFastMP_InlineGetThread__PatchTLSOffset
-
-    ;load current allocation pointers
-    ldr r2, [r12, #Thread__m_alloc_context__alloc_limit]
-    ldr r3, [r12, #Thread__m_alloc_context__alloc_ptr]
-
-    ;add object size to current pointer 
-    add r1, r3
-
-    ;if beyond the limit call c++ method
-    cmp r1, r2
-    bhi AllocFailed
-
-    ;r1 is the new alloc_ptr and r3 has object address
-    ;update the alloc_ptr in Thread
-    str r1, [r12, #Thread__m_alloc_context__alloc_ptr]
-
-    ;write methodTable in object
-    str r0, [r3]
-
-    ;return object in r0 
-    mov r0, r3
-
-#ifdef _DEBUG
-    ; Tail call to a helper that will set the current AppDomain index into the object header and then
-    ; return the object pointer back to our original caller.
-    b       SetAppDomainInObject
-#else
-    ;return
-    bx lr
-#endif
-
-AllocFailed
-    b       $JIT_New   
-    LEAF_END
-
-
-;---------------------------------------------------------------------------
-; HCIMPL2(Object*, JIT_Box, CORINFO_CLASS_HANDLE type, void* unboxedData)
-; IN: r0: MethodTable*
-; IN: r1: data pointer
-;; OUT: r0: new object
-
-    LEAF_ENTRY JIT_BoxFastMP_InlineGetThread
-
-    ldr r2, [r0, #MethodTable__m_pWriteableData]
-
-    ;Check whether the class has been initialized
-    ldr r2, [r2, #MethodTableWriteableData__m_dwFlags]
-    cmp r2, #MethodTableWriteableData__enum_flag_Unrestored
-    bne ClassNotInited
-
-    ; Check whether the object contains pointers
-    ldr r3, [r0, #MethodTable__m_dwFlags]
-    cmp r3, #MethodTable__enum_flag_ContainsPointers
-    bne ContainsPointers
-
-    ldr r2, [r0, #MethodTable__m_BaseSize]
-
-    ;m_BaseSize is guranteed to be a multiple of 4
-
-    ;GetThread
-    PATCHABLE_INLINE_GETTHREAD r12, JIT_BoxFastMP_InlineGetThread__PatchTLSOffset
-
-    ldr r3, [r12, #Thread__m_alloc_context__alloc_ptr]
-    add r3, r2
-
-    ldr r2, [r12, #Thread__m_alloc_context__alloc_limit]
-
-    cmp r3, r2
-    bhi AllocFailed2
-
-    ldr r2, [r12, #Thread__m_alloc_context__alloc_ptr]
-
-    ;advance alloc_ptr in Thread
-    str r3, [r12, #Thread__m_alloc_context__alloc_ptr]
-
-    ;write methodtable* in the object
-    str r0, [r2]
- 
-    ;copy the contents of value type in the object
-
-    ldr r3, [r0, #MethodTable__m_BaseSize]
-    sub r3, #0xc
-   
-    ;r3 = no of bytes to copy
-
-    ;move address of object to return register
-    mov r0, r2
-
-    ;advance r2 to skip methodtable location
-    add r2, #4
-
-CopyLoop
-    ldr r12, [r1, r3]
-    str r12, [r2, r3]
-    sub r3, #4
-    bne CopyLoop
-
-#ifdef _DEBUG
-    ; Tail call to a helper that will set the current AppDomain index into the object header and then
-    ; return the object pointer back to our original caller.
-    b       SetAppDomainInObject
-#else
-    ;return
-    bx lr
-#endif
-
-ContainsPointers
-ClassNotInited
-AllocFailed2
-    b       $JIT_Box       
-    LEAF_END
-
-
-;---------------------------------------------------------------------------
-; IN: r0: number of characters to allocate
-;; OUT: r0: address of newly allocated string
-
-    LEAF_ENTRY AllocateStringFastMP_InlineGetThread
-
-    ; Instead of doing elaborate overflow checks, we just limit the number of elements to
-    ; MAX_FAST_ALLOCATE_STRING_SIZE. This is picked (in asmconstants.h) to avoid any possibility of
-    ; overflow and to ensure we never try to allocate anything here that really should go on the large
-    ; object heap instead. Additionally the size has been selected so that it will encode into an
-    ; immediate in a single cmp instruction.
-
-    cmp     r0, #MAX_FAST_ALLOCATE_STRING_SIZE
-    bhs     OversizedString
-
-    ; Calculate total string size: Align(base size + (characters * 2), 4).
-    mov     r1, #(SIZEOF__BaseStringObject + 3) ; r1 == string base size + 3 for alignment round up
-    add     r1, r1, r0, lsl #1                  ; r1 += characters * 2
-    bic     r1, r1, #3                          ; r1 &= ~3; round size to multiple of 4
-
-    ;GetThread
-    PATCHABLE_INLINE_GETTHREAD r12, AllocateStringFastMP_InlineGetThread__PatchTLSOffset
-    ldr r2, [r12, #Thread__m_alloc_context__alloc_limit]
-    ldr r3, [r12, #Thread__m_alloc_context__alloc_ptr]
-
-    add r1, r3
-    cmp r1, r2
-    bhi AllocFailed3
-
-    ;can allocate
-
-    ;advance alloc_ptr
-    str r1, [r12, #Thread__m_alloc_context__alloc_ptr]
-
-    ; Write MethodTable pointer into new object.
-    ldr     r1, =$g_pStringClass
-    ldr     r1, [r1]
-    str     r1, [r3]
-
-    ; Write string length into new object.
-    str     r0, [r3, #StringObject__m_StringLength]
-
-    ;prepare to return new object address
-    mov     r0, r3
-
-#ifdef _DEBUG
-    ; Tail call to a helper that will set the current AppDomain index into the object header and then
-    ; return the object pointer back to our original caller.
-    b       SetAppDomainInObject
-#else
-    ;return
-    bx lr
-#endif
-
-
-OversizedString
-AllocFailed3
-    b       $FramedAllocateString
-
-    LEAF_END
-
-
-; HCIMPL2(Object*, JIT_NewArr1VC_MP_InlineGetThread, CORINFO_CLASS_HANDLE arrayMT, INT_PTR size)
-;---------------------------------------------------------------------------
-; IN: r0: a (shared) array method table, which contains the element type.
-; IN: r1: number of array elements 
-;; OUT: r0: address of newly allocated object
-
-    LEAF_ENTRY JIT_NewArr1VC_MP_InlineGetThread
-
-    ; Do a conservative check here for number of elements.  
-    ; This is to avoid overflow while doing the calculations.  We don't
-    ; have to worry about "large" objects, since the allocation quantum is never big enough for
-    ; LARGE_OBJECT_SIZE.
-
-    ; For Value Classes, this needs to be < (max_value_in_4byte - size_of_base_array)/(max_size_of_each_element)
-    ; This evaluates to (2^32-1 - 0xc)/2^16
-
-    ; Additionally the constant has been chosen such that it can be encoded in a
-    ; single Thumb2 CMP instruction.
-
-    cmp r1, #MAX_FAST_ALLOCATE_ARRAY_VC_SIZE
-    bhs OverSizedArray3
-
-    ;get element size - stored in low 16bits of m_dwFlags
-    ldrh r12, [r0, #MethodTable__m_dwFlags]
-
-    ; getting size of object to allocate
-
-    ; multiply number of elements with size of each element
-    mul r2, r12, r1
-
-    ; add the base array size and 3 to align total bytes at 4 byte boundary
-    add r2, r2, #SIZEOF__ArrayOfValueType + 3
-    bic r2, #3
-
-    ;GetThread
-    PATCHABLE_INLINE_GETTHREAD r12, JIT_NewArr1VC_MP_InlineGetThread__PatchTLSOffset
-    ldr r3, [r12, #Thread__m_alloc_context__alloc_ptr]
-
-    add r3, r2
-
-    ldr r2, [r12, #Thread__m_alloc_context__alloc_limit]
-
-    cmp r3, r2
-    bhi AllocFailed6
-
-    ; can allocate
-
-    ;r2 = address of new object
-    ldr r2, [r12, #Thread__m_alloc_context__alloc_ptr]
-
-    ;update pointer in allocation context
-    str r3, [r12, #Thread__m_alloc_context__alloc_ptr]
-
-    ;store number of elements
-    str r1, [r2, #ArrayBase__m_NumComponents]
-
-    ;store methodtable
-    str r0, [r2]
-
-    ;copy return value
-    mov r0, r2
-
-#ifdef _DEBUG
-    ; Tail call to a helper that will set the current AppDomain index into the object header and then
-    ; return the object pointer back to our original caller.
-    b       SetAppDomainInObject
-#else
-    ;return
-    bx lr
-#endif
-
-
-
-AllocFailed6
-OverSizedArray3
-    b $JIT_NewArr1
-
-    LEAF_END
-
-
-
-; HCIMPL2(Object*, JIT_NewArr1OBJ_MP_InlineGetThread, CORINFO_CLASS_HANDLE arrayMT, INT_PTR size)
-;---------------------------------------------------------------------------
-; IN: r0: a (shared) array method table, which contains the element type.
-; IN: r1: number of array elements 
-;; OUT: r0: address of newly allocated object
-
-    LEAF_ENTRY JIT_NewArr1OBJ_MP_InlineGetThread
-
-    cmp r1, #MAX_FAST_ALLOCATE_ARRAY_OBJECTREF_SIZE
-    bhs OverSizedArray
-
-    mov r2, #SIZEOF__ArrayOfObjectRef
-    add r2, r2, r1, lsl #2
- 
-    ;r2 will be a multiple of 4
-    
-
-    ;GetThread
-    PATCHABLE_INLINE_GETTHREAD r12, JIT_NewArr1OBJ_MP_InlineGetThread__PatchTLSOffset
-    ldr r3, [r12, #Thread__m_alloc_context__alloc_ptr]
-
-    add r3, r2
-
-    ldr r2, [r12, #Thread__m_alloc_context__alloc_limit]
-
-    cmp r3, r2
-    bhi AllocFailed4
-
-    ;can allocate
-
-    ;r2 = address of new object
-    ldr r2, [r12, #Thread__m_alloc_context__alloc_ptr]
-
-    ;update pointer in allocation context
-    str r3, [r12, #Thread__m_alloc_context__alloc_ptr]
-
-    ;store number of elements
-    str r1, [r2, #ArrayBase__m_NumComponents]
-
-    ;store methodtable
-    str r0, [r2]
-
-    ;copy return value
-    mov r0, r2
-
-#ifdef _DEBUG
-    ; Tail call to a helper that will set the current AppDomain index into the object header and then
-    ; return the object pointer back to our original caller.
-    b       SetAppDomainInObject
-#else
-    ;return
-    bx lr
-#endif
-    
-OverSizedArray
-AllocFailed4
-    b  $JIT_NewArr1
-    LEAF_END
-
-;
-; JIT Static access helpers when TLS Index for AppDomain is low enough for fast helpers
-;
-
-; ------------------------------------------------------------------
-; void* JIT_GetSharedNonGCStaticBase(SIZE_T moduleDomainID, DWORD dwClassDomainID)
-
-    LEAF_ENTRY JIT_GetSharedNonGCStaticBase_InlineGetAppDomain
-    ; Check if r0 (moduleDomainID) is not a moduleID
-    tst    r0, #1
-    beq      HaveLocalModule1
-
-    PATCHABLE_INLINE_GETAPPDOMAIN r2, JIT_GetSharedNonGCStaticBase__PatchTLSLabel
-
-    ; Get the LocalModule, r0 will always be odd, so: r0 * 2 - 2 <=> (r0 >> 1) * 4
-    ldr     r2, [r2 , #AppDomain__m_sDomainLocalBlock + DomainLocalBlock__m_pModuleSlots]
-    add     r2, r2, r0, LSL #1
-    ldr     r0, [r2, #-2]
-
-HaveLocalModule1
-    ; If class is not initialized, bail to C++ helper
-    add     r2, r0, #DomainLocalModule__m_pDataBlob
-    ldrb    r2, [r2, r1]
-    tst     r2, #1
-    beq      CallHelper1
-
-    bx      lr
-
-CallHelper1
-    ; Tail call JIT_GetSharedNonGCStaticBase_Helper
-    b     JIT_GetSharedNonGCStaticBase_Helper
-    LEAF_END
-
-
-; ------------------------------------------------------------------
-; void* JIT_GetSharedNonGCStaticBaseNoCtor(SIZE_T moduleDomainID, DWORD dwClassDomainID)
-
-    LEAF_ENTRY JIT_GetSharedNonGCStaticBaseNoCtor_InlineGetAppDomain
-    ; Check if r0 (moduleDomainID) is not a moduleID
-    tst    r0, #1
-    beq      HaveLocalModule2
-
-    PATCHABLE_INLINE_GETAPPDOMAIN r2, JIT_GetSharedNonGCStaticBaseNoCtor__PatchTLSLabel
-
-    ; Get the LocalModule, r0 will always be odd, so: r0 * 2 - 2 <=> (r0 >> 1) * 4
-    ldr     r2, [r2 , #AppDomain__m_sDomainLocalBlock + DomainLocalBlock__m_pModuleSlots]
-    add     r2, r2, r0, LSL #1
-    ldr     r0, [r2, #-2]
-
-
-HaveLocalModule2
-    bx lr
-    LEAF_END
-
-
-; ------------------------------------------------------------------
-; void* JIT_GetSharedGCStaticBase(SIZE_T moduleDomainID, DWORD dwClassDomainID)
-
-    LEAF_ENTRY JIT_GetSharedGCStaticBase_InlineGetAppDomain
-    ; Check if r0 (moduleDomainID) is not a moduleID
-    tst    r0, #1
-    beq      HaveLocalModule3
-
-    PATCHABLE_INLINE_GETAPPDOMAIN r2, JIT_GetSharedGCStaticBase__PatchTLSLabel
-
-    ; Get the LocalModule, r0 will always be odd, so: r0 * 2 - 2 <=> (r0 >> 1) * 4
-    ldr     r2, [r2 , #AppDomain__m_sDomainLocalBlock + DomainLocalBlock__m_pModuleSlots]
-    add     r2, r2, r0, LSL #1
-    ldr     r0, [r2, #-2]
-
-HaveLocalModule3
-    ; If class is not initialized, bail to C++ helper
-    add     r2, r0, #DomainLocalModule__m_pDataBlob
-    ldrb    r2, [r2, r1]
-    tst     r2, #1
-    beq      CallHelper3
-
-    ldr     r0, [r0, #DomainLocalModule__m_pGCStatics]
-    bx lr
-
-CallHelper3
-    ; Tail call Jit_GetSharedGCStaticBase_Helper
-    b     JIT_GetSharedGCStaticBase_Helper
-    LEAF_END
-
-
-; ------------------------------------------------------------------
-; void* JIT_GetSharedGCStaticBaseNoCtor(SIZE_T moduleDomainID, DWORD dwClassDomainID)
-
-    LEAF_ENTRY JIT_GetSharedGCStaticBaseNoCtor_InlineGetAppDomain
-    ; Check if r0 (moduleDomainID) is not a moduleID
-    tst    r0, #1
-    beq      HaveLocalModule4
-
-    PATCHABLE_INLINE_GETAPPDOMAIN r2, JIT_GetSharedGCStaticBaseNoCtor__PatchTLSLabel
-
-    ; Get the LocalModule, r0 will always be odd, so: r0 * 2 - 2 <=> (r0 >> 1) * 4
-    ldr     r2, [r2 , #AppDomain__m_sDomainLocalBlock + DomainLocalBlock__m_pModuleSlots]
-    add     r2, r2, r0, LSL #1
-    ldr     r0, [r2, #-2]
-
-HaveLocalModule4
-    ldr     r0, [r0, #DomainLocalModule__m_pGCStatics]
-    bx lr
-    LEAF_END
-
 ; ------------------------------------------------------------------
 ; End of the writeable code region
     LEAF_ENTRY JIT_PatchedCodeLast
diff --git a/src/vm/arm/stubs.cpp b/src/vm/arm/stubs.cpp
index 6d9b3d0dca..e0574762b1 100644
--- a/src/vm/arm/stubs.cpp
+++ b/src/vm/arm/stubs.cpp
@@ -1403,30 +1403,21 @@ Stub *GenerateInitPInvokeFrameHelper()
     ThumbReg regThread  = ThumbReg(5);
     ThumbReg regScratch = ThumbReg(6);
 
-#ifdef FEATURE_IMPLICIT_TLS
-    TLSACCESSMODE mode = TLSACCESS_GENERIC;
-#else
-    TLSACCESSMODE mode = GetTLSAccessMode(GetThreadTLSIndex());
-#endif
+#ifdef FEATURE_PAL
+    // Erect frame to perform call to GetThread
+    psl->ThumbEmitProlog(1, sizeof(ArgumentRegisters), FALSE); // Save r4 for aligned stack
 
+    // Save argument registers around the GetThread call. Don't bother with using ldm/stm since this inefficient path anyway.
+    for (int reg = 0; reg < 4; reg++)
+        psl->ThumbEmitStoreRegIndirect(ThumbReg(reg), thumbRegSp, offsetof(ArgumentRegisters, r[reg]));
+#endif
 
-    if (mode == TLSACCESS_GENERIC)
-    {
-        // Erect frame to perform call to GetThread
-        psl->ThumbEmitProlog(1, sizeof(ArgumentRegisters), FALSE); // Save r4 for aligned stack
-
-        // Save argument registers around the GetThread call. Don't bother with using ldm/stm since this inefficient path anyway.
-        for (int reg = 0; reg < 4; reg++)
-            psl->ThumbEmitStoreRegIndirect(ThumbReg(reg), thumbRegSp, offsetof(ArgumentRegisters, r[reg]));
-    }
-
-    psl->ThumbEmitGetThread(mode, regThread);
+    psl->ThumbEmitGetThread(regThread);
 
-    if (mode == TLSACCESS_GENERIC)
-    {
-        for (int reg = 0; reg < 4; reg++)
-            psl->ThumbEmitLoadRegIndirect(ThumbReg(reg), thumbRegSp, offsetof(ArgumentRegisters, r[reg]));
-    }
+#ifdef FEATURE_PAL
+    for (int reg = 0; reg < 4; reg++)
+        psl->ThumbEmitLoadRegIndirect(ThumbReg(reg), thumbRegSp, offsetof(ArgumentRegisters, r[reg]));
+#endif
 
     // mov [regFrame + FrameInfo.offsetOfGSCookie], GetProcessGSCookie()
     psl->ThumbEmitMovConstant(regScratch, GetProcessGSCookie());
@@ -1448,82 +1439,36 @@ Stub *GenerateInitPInvokeFrameHelper()
     psl->ThumbEmitMovConstant(regScratch, 0);
     psl->ThumbEmitStoreRegIndirect(regScratch, regFrame, FrameInfo.offsetOfReturnAddress - negSpace);
 
-    if (mode == TLSACCESS_GENERIC)
-    {
-        DWORD cbSavedRegs = sizeof(ArgumentRegisters) + 2 * 4; // r0-r3, r4, lr
-        psl->ThumbEmitAdd(regScratch, thumbRegSp, cbSavedRegs);
-        psl->ThumbEmitStoreRegIndirect(regScratch, regFrame, FrameInfo.offsetOfCallSiteSP - negSpace);
-    }
-    else
-    {
-        // str SP, [regFrame + FrameInfo.offsetOfCallSiteSP]
-        psl->ThumbEmitStoreRegIndirect(thumbRegSp, regFrame, FrameInfo.offsetOfCallSiteSP - negSpace);
-    }
+#ifdef FEATURE_PAL
+    DWORD cbSavedRegs = sizeof(ArgumentRegisters) + 2 * 4; // r0-r3, r4, lr
+    psl->ThumbEmitAdd(regScratch, thumbRegSp, cbSavedRegs);
+    psl->ThumbEmitStoreRegIndirect(regScratch, regFrame, FrameInfo.offsetOfCallSiteSP - negSpace);
+#else
+    // str SP, [regFrame + FrameInfo.offsetOfCallSiteSP]
+    psl->ThumbEmitStoreRegIndirect(thumbRegSp, regFrame, FrameInfo.offsetOfCallSiteSP - negSpace);
+#endif
 
     // mov [regThread + offsetof(Thread, m_pFrame)], regFrame
     psl->ThumbEmitStoreRegIndirect(regFrame, regThread, offsetof(Thread, m_pFrame));
 
     // leave current Thread in R4
 
-    if (mode == TLSACCESS_GENERIC)
-    {
-        psl->ThumbEmitEpilog();
-    }
-    else
-    {
-        // Return. The return address has been restored into LR at this point.
-        // bx lr
-        psl->ThumbEmitJumpRegister(thumbRegLr);
-    }
+#ifdef FEATURE_PAL
+    psl->ThumbEmitEpilog();
+#else
+    // Return. The return address has been restored into LR at this point.
+    // bx lr
+    psl->ThumbEmitJumpRegister(thumbRegLr);
+#endif
 
     // A single process-wide stub that will never unload
     RETURN psl->Link(SystemDomain::GetGlobalLoaderAllocator()->GetStubHeap());
 }
 
-void StubLinkerCPU::ThumbEmitGetThread(TLSACCESSMODE mode, ThumbReg dest)
+void StubLinkerCPU::ThumbEmitGetThread(ThumbReg dest)
 {
-#ifndef FEATURE_IMPLICIT_TLS
-    DWORD idxThread = GetThreadTLSIndex();
-
-    if (mode != TLSACCESS_GENERIC)
-    {
-        // mrc p15, 0, dest, c13, c0, 2
-        Emit16(0xee1d);
-        Emit16((WORD)(0x0f50 | (dest << 12)));
+#ifdef FEATURE_PAL
 
-        if (mode == TLSACCESS_WNT)
-        {
-            // ldr dest, [dest, #(WINNT_TLS_OFFSET + (idxThread * sizeof(void*)))]
-            ThumbEmitLoadRegIndirect(dest, dest, offsetof(TEB, TlsSlots) + (idxThread * sizeof(void*)));
-        }
-        else
-        {
-            _ASSERTE(mode == TLSACCESS_WNT_HIGH);
-
-            // ldr dest, [dest, #WINNT5_TLSEXPANSIONPTR_OFFSET]
-            ThumbEmitLoadRegIndirect(dest, dest, offsetof(TEB, TlsExpansionSlots));
-
-            // ldr dest, [dest + #(idxThread * 4)]
-            ThumbEmitLoadRegIndirect(dest, dest, (idxThread - TLS_MINIMUM_AVAILABLE) * sizeof(void*));
-        }
-    }
-    else
-    {
-        ThumbEmitMovConstant(ThumbReg(0), idxThread);
-
-#pragma push_macro("TlsGetValue")
-#undef TlsGetValue
-        ThumbEmitMovConstant(ThumbReg(1), (TADDR)TlsGetValue);
-#pragma pop_macro("TlsGetValue")
-
-        ThumbEmitCallRegister(ThumbReg(1));
-
-        if (dest != ThumbReg(0))
-        {
-            ThumbEmitMovRegReg(dest, ThumbReg(0));
-        }
-    }
-#else
     ThumbEmitMovConstant(ThumbReg(0), (TADDR)GetThread);
 
     ThumbEmitCallRegister(ThumbReg(0));
@@ -1532,7 +1477,20 @@ void StubLinkerCPU::ThumbEmitGetThread(TLSACCESSMODE mode, ThumbReg dest)
     {
         ThumbEmitMovRegReg(dest, ThumbReg(0));
     }
-#endif
+
+#else // FEATURE_PAL
+
+    // mrc p15, 0, dest, c13, c0, 2
+    Emit16(0xee1d);
+    Emit16((WORD)(0x0f50 | (dest << 12)));
+
+    ThumbEmitLoadRegIndirect(dest, dest, offsetof(TEB, ThreadLocalStoragePointer));
+
+    ThumbEmitLoadRegIndirect(dest, dest, sizeof(void *) * (g_TlsIndex & 0xFFFF));
+
+    ThumbEmitLoadRegIndirect(dest, dest, (g_TlsIndex & 0x7FFF0000) >> 16);
+
+#endif // FEATURE_PAL
 }
 #endif // CROSSGEN_COMPILE
 
@@ -2537,110 +2495,12 @@ void UMEntryThunkCode::Poison()
 
 #ifndef CROSSGEN_COMPILE
 
-
-EXTERN_C DWORD gThreadTLSIndex;
-EXTERN_C DWORD gAppDomainTLSIndex;
-
-
-EXTERN_C Object* JIT_TrialAllocSFastMP_InlineGetThread(CORINFO_CLASS_HANDLE typeHnd_);
-EXTERN_C Object* JIT_BoxFastMP_InlineGetThread (CORINFO_CLASS_HANDLE type, void* unboxedData);
-EXTERN_C Object* AllocateStringFastMP_InlineGetThread (CLR_I4 cch);
-EXTERN_C Object* JIT_NewArr1OBJ_MP_InlineGetThread (CORINFO_CLASS_HANDLE arrayTypeHnd_, INT_PTR size);
-EXTERN_C Object* JIT_NewArr1VC_MP_InlineGetThread (CORINFO_CLASS_HANDLE arrayTypeHnd_, INT_PTR size);
-
-EXTERN_C void JIT_TrialAllocSFastMP_InlineGetThread__PatchTLSOffset();
-EXTERN_C void JIT_BoxFastMP_InlineGetThread__PatchTLSOffset();
-EXTERN_C void AllocateStringFastMP_InlineGetThread__PatchTLSOffset();
-EXTERN_C void JIT_NewArr1VC_MP_InlineGetThread__PatchTLSOffset();
-EXTERN_C void JIT_NewArr1OBJ_MP_InlineGetThread__PatchTLSOffset();
-
 extern "C" void STDCALL JIT_PatchedCodeStart();
 extern "C" void STDCALL JIT_PatchedCodeLast();
 
-#ifndef FEATURE_IMPLICIT_TLS
-static const LPVOID InlineGetThreadLocations[] = {
-    (PVOID)JIT_TrialAllocSFastMP_InlineGetThread__PatchTLSOffset,
-    (PVOID)JIT_BoxFastMP_InlineGetThread__PatchTLSOffset,
-    (PVOID)AllocateStringFastMP_InlineGetThread__PatchTLSOffset,
-    (PVOID)JIT_NewArr1VC_MP_InlineGetThread__PatchTLSOffset,
-    (PVOID)JIT_NewArr1OBJ_MP_InlineGetThread__PatchTLSOffset,
-};
-#endif
-
-//EXTERN_C Object* JIT_TrialAllocSFastMP(CORINFO_CLASS_HANDLE typeHnd_);
-Object* JIT_TrialAllocSFastMP(CORINFO_CLASS_HANDLE typeHnd_);
-EXTERN_C Object* JIT_NewArr1OBJ_MP(CORINFO_CLASS_HANDLE arrayMT, INT_PTR size);
-EXTERN_C Object* AllocateStringFastMP(CLR_I4 cch);
-EXTERN_C Object* JIT_NewArr1VC_MP(CORINFO_CLASS_HANDLE arrayMT, INT_PTR size);
-EXTERN_C Object* JIT_BoxFastMP(CORINFO_CLASS_HANDLE type, void* unboxedData);
-
-
-EXTERN_C void JIT_GetSharedNonGCStaticBase__PatchTLSLabel();
-EXTERN_C void JIT_GetSharedNonGCStaticBaseNoCtor__PatchTLSLabel();
-EXTERN_C void JIT_GetSharedGCStaticBase__PatchTLSLabel();
-EXTERN_C void JIT_GetSharedGCStaticBaseNoCtor__PatchTLSLabel();
-
-EXTERN_C void JIT_GetSharedNonGCStaticBase_SingleAppDomain();
-EXTERN_C void JIT_GetSharedNonGCStaticBaseNoCtor_SingleAppDomain();
-EXTERN_C void JIT_GetSharedGCStaticBase_SingleAppDomain();
-EXTERN_C void JIT_GetSharedGCStaticBaseNoCtor_SingleAppDomain();
-
-
-static const LPVOID InlineGetAppDomainLocations[] = {
-    (PVOID)JIT_GetSharedNonGCStaticBase__PatchTLSLabel,
-    (PVOID)JIT_GetSharedNonGCStaticBaseNoCtor__PatchTLSLabel,
-    (PVOID)JIT_GetSharedGCStaticBase__PatchTLSLabel,
-    (PVOID)JIT_GetSharedGCStaticBaseNoCtor__PatchTLSLabel
-};
-
-#ifndef FEATURE_IMPLICIT_TLS
-void FixupInlineGetters(DWORD tlsSlot, const LPVOID * pLocations, int nLocations)
-{
-    STANDARD_VM_CONTRACT;
-
-    for (int i=0; i<nLocations; i++)
-    {
-        BYTE * pInlineGetter = (BYTE *)PCODEToPINSTR(GetEEFuncEntryPoint(pLocations[i]));
-
-        DWORD offset = (tlsSlot * sizeof(LPVOID) + offsetof(TEB, TlsSlots));
-
-        // ldr  r??, [r??, #offset]
-        _ASSERTE_ALL_BUILDS("clr/src/VM/arm/stubs.cpp",
-                            pInlineGetter[0] == 0x1d &&
-                            pInlineGetter[1] == 0xee &&
-                            pInlineGetter[2] == 0x50 &&
-                            pInlineGetter[5] == 0xf8 &&
-                            "Initialization failure while stomping instructions for the TLS slot offset: "
-                            "the instruction at the given offset did not match what we expect");
-
-        *((WORD*)(pInlineGetter + 6)) &= 0xf000;
-
-        _ASSERTE(offset <=4095);
-        *((WORD*)(pInlineGetter + 6)) |= (WORD)offset;
-    }
-}
-#endif
-
 void InitJITHelpers1()
 {
     STANDARD_VM_CONTRACT;
-    
-#ifndef FEATURE_IMPLICIT_TLS
-
-    if (gThreadTLSIndex < TLS_MINIMUM_AVAILABLE)
-    {
-        FixupInlineGetters(gThreadTLSIndex, InlineGetThreadLocations, COUNTOF(InlineGetThreadLocations));
-    }
-
-    if (gAppDomainTLSIndex < TLS_MINIMUM_AVAILABLE)
-    {
-        FixupInlineGetters(gAppDomainTLSIndex, InlineGetAppDomainLocations, COUNTOF(InlineGetAppDomainLocations));
-    }
-
-    if(gThreadTLSIndex < TLS_MINIMUM_AVAILABLE || gAppDomainTLSIndex < TLS_MINIMUM_AVAILABLE)
-    {
-        FlushInstructionCache(GetCurrentProcess(), JIT_PatchedCodeStart, (BYTE*)JIT_PatchedCodeLast - (BYTE*)JIT_PatchedCodeStart);
-    }
 
 #if CHECK_APP_DOMAIN_LEAKS
     if(g_pConfig->AppDomainLeaks())
@@ -2655,54 +2515,14 @@ void InitJITHelpers1()
 #endif // _DEBUG
         ))
     {
-
         _ASSERTE(GCHeapUtilities::UseThreadAllocationContexts());
-        // If the TLS for Thread is low enough use the super-fast helpers
-        if (gThreadTLSIndex < TLS_MINIMUM_AVAILABLE)
-        {
-            SetJitHelperFunction(CORINFO_HELP_NEWSFAST, JIT_TrialAllocSFastMP_InlineGetThread);
-            SetJitHelperFunction(CORINFO_HELP_BOX, JIT_BoxFastMP_InlineGetThread);
-            SetJitHelperFunction(CORINFO_HELP_NEWARR_1_VC, JIT_NewArr1VC_MP_InlineGetThread);
-            SetJitHelperFunction(CORINFO_HELP_NEWARR_1_OBJ, JIT_NewArr1OBJ_MP_InlineGetThread);
 
-            ECall::DynamicallyAssignFCallImpl(GetEEFuncEntryPoint(AllocateStringFastMP_InlineGetThread), ECall::FastAllocateString);
-        }
-        else
-        {
-/*
-            SetJitHelperFunction(CORINFO_HELP_NEWSFAST, JIT_TrialAllocSFastMP);
-            SetJitHelperFunction(CORINFO_HELP_BOX, JIT_BoxFastMP);
-            SetJitHelperFunction(CORINFO_HELP_NEWARR_1_VC, JIT_NewArr1VC_MP);
-            SetJitHelperFunction(CORINFO_HELP_NEWARR_1_OBJ, JIT_NewArr1OBJ_MP);
+        SetJitHelperFunction(CORINFO_HELP_NEWSFAST, JIT_NewS_MP_FastPortable);
+        SetJitHelperFunction(CORINFO_HELP_NEWARR_1_VC, JIT_NewArr1VC_MP_FastPortable);
+        SetJitHelperFunction(CORINFO_HELP_NEWARR_1_OBJ, JIT_NewArr1OBJ_MP_FastPortable);
 
-            ECall::DynamicallyAssignFCallImpl(GetEEFuncEntryPoint(AllocateStringFastMP), ECall::FastAllocateString);
-*/
-        }
+        ECall::DynamicallyAssignFCallImpl(GetEEFuncEntryPoint(AllocateString_MP_FastPortable), ECall::FastAllocateString);
     }
-
-
-    if(IsSingleAppDomain())
-    {
-        SetJitHelperFunction(CORINFO_HELP_GETSHARED_GCSTATIC_BASE,          JIT_GetSharedGCStaticBase_SingleAppDomain);
-        SetJitHelperFunction(CORINFO_HELP_GETSHARED_NONGCSTATIC_BASE,       JIT_GetSharedNonGCStaticBase_SingleAppDomain);
-        SetJitHelperFunction(CORINFO_HELP_GETSHARED_GCSTATIC_BASE_NOCTOR,   JIT_GetSharedGCStaticBaseNoCtor_SingleAppDomain);
-        SetJitHelperFunction(CORINFO_HELP_GETSHARED_NONGCSTATIC_BASE_NOCTOR,JIT_GetSharedNonGCStaticBaseNoCtor_SingleAppDomain);
-    }
-    else
-    if (gAppDomainTLSIndex >= TLS_MINIMUM_AVAILABLE)
-    {
-        SetJitHelperFunction(CORINFO_HELP_GETSHARED_GCSTATIC_BASE,          JIT_GetSharedGCStaticBase_Portable);
-        SetJitHelperFunction(CORINFO_HELP_GETSHARED_NONGCSTATIC_BASE,       JIT_GetSharedNonGCStaticBase_Portable);
-        SetJitHelperFunction(CORINFO_HELP_GETSHARED_GCSTATIC_BASE_NOCTOR,   JIT_GetSharedGCStaticBaseNoCtor_Portable);
-        SetJitHelperFunction(CORINFO_HELP_GETSHARED_NONGCSTATIC_BASE_NOCTOR,JIT_GetSharedNonGCStaticBaseNoCtor_Portable);
-    }
-#endif
-}
-
-extern "C" Object *SetAppDomainInObject(Object *pObject)
-{
-    pObject->SetAppDomain();
-    return pObject;
 }
 
 //              +64     stack-based arguments here
@@ -3041,19 +2861,13 @@ void StubLinkerCPU::EmitStubLinkFrame(TADDR pFrameVptr, int offsetOfFrame, int o
     //  str r6, [r4 + #offsetof(MulticastFrame, m_Next)]
     //  str r4, [r5 + #offsetof(Thread, m_pFrame)]
 
-#ifdef FEATURE_IMPLICIT_TLS
-    TLSACCESSMODE mode = TLSACCESS_GENERIC;
-#else
-    TLSACCESSMODE mode = GetTLSAccessMode(GetThreadTLSIndex());
+    ThumbEmitGetThread(ThumbReg(5));
+#ifdef FEATURE_PAL
+    // reload argument registers that could have been corrupted by the call
+    for (int reg = 0; reg < 4; reg++)
+        ThumbEmitLoadRegIndirect(ThumbReg(reg), ThumbReg(4), 
+            offsetOfTransitionBlock + TransitionBlock::GetOffsetOfArgumentRegisters() + offsetof(ArgumentRegisters, r[reg]));
 #endif
-    ThumbEmitGetThread(mode, ThumbReg(5));
-    if (mode == TLSACCESS_GENERIC)
-    {
-        // reload argument registers that could have been corrupted by the call
-        for (int reg = 0; reg < 4; reg++)
-            ThumbEmitLoadRegIndirect(ThumbReg(reg), ThumbReg(4), 
-                offsetOfTransitionBlock + TransitionBlock::GetOffsetOfArgumentRegisters() + offsetof(ArgumentRegisters, r[reg]));
-    }
 
     ThumbEmitLoadRegIndirect(ThumbReg(6), ThumbReg(5), Thread::GetOffsetOfCurrentFrame());
     ThumbEmitStoreRegIndirect(ThumbReg(6), ThumbReg(4), Frame::GetOffsetOfNextLink());
diff --git a/src/vm/arm64/cgencpu.h b/src/vm/arm64/cgencpu.h
index 8abe4de6ab..a168cdc162 100644
--- a/src/vm/arm64/cgencpu.h
+++ b/src/vm/arm64/cgencpu.h
@@ -84,7 +84,11 @@ typedef INT64 StackElemType;
 //
 // Create alias for optimized implementations of helpers provided on this platform
 //
-// optimized static helpers
+#define JIT_GetSharedGCStaticBase           JIT_GetSharedGCStaticBase_SingleAppDomain
+#define JIT_GetSharedNonGCStaticBase        JIT_GetSharedNonGCStaticBase_SingleAppDomain
+#define JIT_GetSharedGCStaticBaseNoCtor     JIT_GetSharedGCStaticBaseNoCtor_SingleAppDomain
+#define JIT_GetSharedNonGCStaticBaseNoCtor  JIT_GetSharedNonGCStaticBaseNoCtor_SingleAppDomain
+
 #define JIT_Stelem_Ref                      JIT_Stelem_Ref
 
 //**********************************************************************
@@ -435,10 +439,8 @@ public:
     void EmitUnboxMethodStub(MethodDesc* pRealMD);
     void EmitCallManagedMethod(MethodDesc *pMD, BOOL fTailCall);
     void EmitCallLabel(CodeLabel *target, BOOL fTailCall, BOOL fIndirect);
-    void EmitSecureDelegateInvoke(UINT_PTR hash);
-    static UINT_PTR HashMulticastInvoke(MetaSig* pSig);
+
     void EmitShuffleThunk(struct ShuffleEntry *pShuffleEntryArray);
-    void EmitGetThreadInlined(IntReg Xt);
 
 #ifdef _DEBUG
     void EmitNop() { Emit32(0xD503201F); }
diff --git a/src/vm/arm64/stubs.cpp b/src/vm/arm64/stubs.cpp
index c57fca94b1..3c56c382ea 100644
--- a/src/vm/arm64/stubs.cpp
+++ b/src/vm/arm64/stubs.cpp
@@ -16,10 +16,6 @@
 #include "jitinterface.h"
 #include "ecall.h"
 
-EXTERN_C void JIT_GetSharedNonGCStaticBase_SingleAppDomain();
-EXTERN_C void JIT_GetSharedNonGCStaticBaseNoCtor_SingleAppDomain();
-EXTERN_C void JIT_GetSharedGCStaticBase_SingleAppDomain();
-EXTERN_C void JIT_GetSharedGCStaticBaseNoCtor_SingleAppDomain();
 EXTERN_C void JIT_UpdateWriteBarrierState(bool skipEphemeralCheck);
 
 
@@ -1114,14 +1110,6 @@ void InitJITHelpers1()
     }
 #endif
 
-    if(IsSingleAppDomain())
-    {
-        SetJitHelperFunction(CORINFO_HELP_GETSHARED_GCSTATIC_BASE,          JIT_GetSharedGCStaticBase_SingleAppDomain);
-        SetJitHelperFunction(CORINFO_HELP_GETSHARED_NONGCSTATIC_BASE,       JIT_GetSharedNonGCStaticBase_SingleAppDomain);
-        SetJitHelperFunction(CORINFO_HELP_GETSHARED_GCSTATIC_BASE_NOCTOR,   JIT_GetSharedGCStaticBaseNoCtor_SingleAppDomain);
-        SetJitHelperFunction(CORINFO_HELP_GETSHARED_NONGCSTATIC_BASE_NOCTOR,JIT_GetSharedNonGCStaticBaseNoCtor_SingleAppDomain);
-    }
-
     JIT_UpdateWriteBarrierState(GCHeapUtilities::IsServerHeap());
 }
 #ifndef FEATURE_PAL // TODO-ARM64-WINDOWS #13592
@@ -1846,33 +1834,6 @@ void StubLinkerCPU::EmitCallManagedMethod(MethodDesc *pMD, BOOL fTailCall)
 
 #ifndef CROSSGEN_COMPILE
 
-EXTERN_C UINT32 _tls_index;
-void StubLinkerCPU::EmitGetThreadInlined(IntReg Xt)
-{
-#if defined(FEATURE_IMPLICIT_TLS) && !defined(FEATURE_PAL)
-    // Trashes x8.
-    IntReg X8 = IntReg(8);
-    _ASSERTE(Xt != X8);
-    
-    // Load the _tls_index
-    EmitLabelRef(NewExternalCodeLabel((LPVOID)&_tls_index), reinterpret_cast<LoadFromLabelInstructionFormat&>(gLoadFromLabelIF), X8);
-    
-    // Load Teb->ThreadLocalStoragePointer into x8
-    EmitLoadStoreRegImm(eLOAD, Xt, IntReg(18), offsetof(_TEB, ThreadLocalStoragePointer));
-
-    // index it with _tls_index, i.e Teb->ThreadLocalStoragePointer[_tls_index]. 
-    // This will give us the TLS section for the module on this thread's context
-    EmitLoadRegReg(Xt, Xt, X8, eLSL);
-
-    // read the Thread* from TLS section
-    EmitAddImm(Xt, Xt, OFFSETOF__TLS__tls_CurrentThread);
-    EmitLoadStoreRegImm(eLOAD, Xt, Xt, 0);
-#else
-    _ASSERTE(!"NYI:StubLinkerCPU::EmitGetThreadInlined");
-#endif
-
-}
-
 void StubLinkerCPU::EmitUnboxMethodStub(MethodDesc *pMD)
 {
     _ASSERTE(!pMD->RequiresInstMethodDescArg());
diff --git a/src/vm/ceeload.cpp b/src/vm/ceeload.cpp
index c64d9e9042..f5173196e3 100644
--- a/src/vm/ceeload.cpp
+++ b/src/vm/ceeload.cpp
@@ -10844,11 +10844,7 @@ void Module::LoadTokenTables()
         pEEInfo->emptyString = (CORINFO_Object **)StringObject::GetEmptyStringRefPtr();
     }
 
-#ifdef FEATURE_IMPLICIT_TLS
     pEEInfo->threadTlsIndex = TLS_OUT_OF_INDEXES;
-#else
-    pEEInfo->threadTlsIndex = GetThreadTLSIndex();
-#endif
     pEEInfo->rvaStaticTlsIndex = NULL;
 #endif // CROSSGEN_COMPILE
 }
diff --git a/src/vm/ceemain.cpp b/src/vm/ceemain.cpp
index 44f5d8eb1d..9dbe2b9dff 100644
--- a/src/vm/ceemain.cpp
+++ b/src/vm/ceemain.cpp
@@ -2714,17 +2714,7 @@ BOOL STDMETHODCALLTYPE EEDllMain( // TRUE on success, FALSE on error.
                                                                                          , TRUE
 #endif
                                                                                          );
-#ifdef FEATURE_IMPLICIT_TLS
                 Thread* thread = GetThread();
-#else
-                // Don't use GetThread because perhaps we didn't initialize yet, or we
-                // have already shutdown the EE.  Note that there is a race here.  We
-                // might ask for TLS from a slot we just released.  We are assuming that
-                // nobody re-allocates that same slot while we are doing this.  It just
-                // isn't worth locking for such an obscure case.
-                DWORD   tlsVal = GetThreadTLSIndex();
-                Thread  *thread = (tlsVal != (DWORD)-1)?(Thread *) UnsafeTlsGetValue(tlsVal):NULL;
-#endif
                 if (thread)
                 {
 #ifdef FEATURE_COMINTEROP
diff --git a/src/vm/ceemain.h b/src/vm/ceemain.h
index ccf763ac80..9a14af54ca 100644
--- a/src/vm/ceemain.h
+++ b/src/vm/ceemain.h
@@ -83,8 +83,6 @@ public:
     // Setup FLS simulation block, including ClrDebugState and StressLog.
     static void SetupTLSForThread(Thread *pThread);
 
-    static DWORD GetTlsIndex () {return TlsIndex;}
-
     static LPVOID* GetTlsData();
     static BOOL SetTlsData (void** ppTlsInfo);
 
@@ -92,12 +90,6 @@ public:
     // private implementation:
     //***************************************************************************
 private:
-
-    // The debugger needs access to the TlsIndex so that we can read it from OOP.
-    friend class EEDbgInterfaceImpl;
-
-    SVAL_DECL (DWORD, TlsIndex);
-
     static PTLS_CALLBACK_FUNCTION Callbacks[MAX_PREDEFINED_TLS_SLOT];
 
     //***************************************************************************
diff --git a/src/vm/corhost.cpp b/src/vm/corhost.cpp
index f6ca34267a..b1e5a8a0f3 100644
--- a/src/vm/corhost.cpp
+++ b/src/vm/corhost.cpp
@@ -51,7 +51,6 @@
 
 GVAL_IMPL_INIT(DWORD, g_fHostConfig, 0);
 
-#ifdef FEATURE_IMPLICIT_TLS
 #ifndef __llvm__
 EXTERN_C __declspec(thread) ThreadLocalInfo gCurrentThreadInfo;
 #else // !__llvm__
@@ -62,11 +61,6 @@ EXTERN_C UINT32 _tls_index;
 #else // FEATURE_PAL
 UINT32 _tls_index = 0;
 #endif // FEATURE_PAL
-SVAL_IMPL_INIT(DWORD, CExecutionEngine, TlsIndex, _tls_index);
-#else
-SVAL_IMPL_INIT(DWORD, CExecutionEngine, TlsIndex, TLS_OUT_OF_INDEXES);
-#endif
-
 
 #if defined(FEATURE_WINDOWSPHONE)
 SVAL_IMPL_INIT(ECustomDumpFlavor, CCLRErrorReportingManager, g_ECustomDumpFlavor, DUMP_FLAVOR_Default);
@@ -3168,7 +3162,6 @@ VOID WINAPI FlsCallback(
 #endif // HAS_FLS_SUPPORT
 
 
-#ifdef FEATURE_IMPLICIT_TLS
 void** CExecutionEngine::GetTlsData()
 {
     LIMITED_METHOD_CONTRACT;
@@ -3183,28 +3176,6 @@ BOOL CExecutionEngine::SetTlsData (void** ppTlsInfo)
     gCurrentThreadInfo.m_EETlsData = ppTlsInfo;
     return TRUE;
 }
-#else 
-void** CExecutionEngine::GetTlsData()
-{
-    LIMITED_METHOD_CONTRACT;
-
-    if (TlsIndex == TLS_OUT_OF_INDEXES)
-        return NULL;
-
-    void **ppTlsData = (void **)UnsafeTlsGetValue(TlsIndex);
-    return ppTlsData;
-}
-BOOL CExecutionEngine::SetTlsData (void** ppTlsInfo)
-{
-    LIMITED_METHOD_CONTRACT;
-
-    if (TlsIndex == TLS_OUT_OF_INDEXES)
-        return FALSE;
-
-    return UnsafeTlsSetValue(TlsIndex, ppTlsInfo);
-}
-
-#endif // FEATURE_IMPLICIT_TLS
 
 //---------------------------------------------------------------------------------------
 //
@@ -3295,25 +3266,6 @@ void **CExecutionEngine::CheckThreadState(DWORD slot, BOOL force)
     }
 #endif // HAS_FLS_SUPPORT
 
-#ifndef FEATURE_IMPLICIT_TLS
-    // Ensure we have a TLS Index
-    if (TlsIndex == TLS_OUT_OF_INDEXES)
-    {
-        DWORD tryTlsIndex = UnsafeTlsAlloc();
-        if (tryTlsIndex != TLS_OUT_OF_INDEXES)
-        {
-            if (FastInterlockCompareExchange((LONG*)&TlsIndex, tryTlsIndex, TLS_OUT_OF_INDEXES) != (LONG)TLS_OUT_OF_INDEXES)
-            {
-                UnsafeTlsFree(tryTlsIndex);
-            }
-        }
-        if (TlsIndex == TLS_OUT_OF_INDEXES)
-        {
-            COMPlusThrowOM();
-        }
-    }
-#endif // FEATURE_IMPLICIT_TLS
-
     void** pTlsData = CExecutionEngine::GetTlsData();
     BOOL fInTls = (pTlsData != NULL);
 
diff --git a/src/vm/crossgencompile.cpp b/src/vm/crossgencompile.cpp
index c4b9d3dfc3..411029becd 100644
--- a/src/vm/crossgencompile.cpp
+++ b/src/vm/crossgencompile.cpp
@@ -72,27 +72,6 @@ BOOL Debug_IsLockedViaThreadSuspension()
 }
 #endif // _DEBUG
 
-#if defined(FEATURE_MERGE_JIT_AND_ENGINE) && defined(FEATURE_IMPLICIT_TLS)
-void* theJitTls;
-
-extern "C"
-{
-
-void* GetJitTls()
-{
-    LIMITED_METHOD_CONTRACT
-
-    return theJitTls;
-}
-void SetJitTls(void* v)
-{
-    LIMITED_METHOD_CONTRACT
-    theJitTls = v;
-}
-
-}
-#endif
-
 //---------------------------------------------------------------------------------------
 //
 // All locks are nops because of there is always only one thread.
diff --git a/src/vm/eedbginterface.h b/src/vm/eedbginterface.h
index 8c8c44d3e2..241ef332e2 100644
--- a/src/vm/eedbginterface.h
+++ b/src/vm/eedbginterface.h
@@ -280,7 +280,6 @@ public:
    virtual void GetRuntimeOffsets(SIZE_T *pTLSIndex,
                                   SIZE_T *pTLSIsSpecialIndex,
                                   SIZE_T *pTLSCantStopIndex,
-                                  SIZE_T *pTLSIndexOfPredefs,
                                   SIZE_T *pEEThreadStateOffset,
                                   SIZE_T *pEEThreadStateNCOffset,
                                   SIZE_T *pEEThreadPGCDisabledOffset,
diff --git a/src/vm/eedbginterfaceimpl.cpp b/src/vm/eedbginterfaceimpl.cpp
index ede82c7780..e9f59d25d2 100644
--- a/src/vm/eedbginterfaceimpl.cpp
+++ b/src/vm/eedbginterfaceimpl.cpp
@@ -1387,14 +1387,11 @@ void EEDbgInterfaceImpl::DisableTraceCall(Thread *thread)
     thread->DecrementTraceCallCount();
 }
 
-#ifdef FEATURE_IMPLICIT_TLS
 EXTERN_C UINT32 _tls_index;
-#endif
 
 void EEDbgInterfaceImpl::GetRuntimeOffsets(SIZE_T *pTLSIndex,
                                            SIZE_T *pTLSIsSpecialIndex,
                                            SIZE_T *pTLSCantStopIndex,
-                                           SIZE_T* pTLSIndexOfPredefs,
                                            SIZE_T *pEEThreadStateOffset,
                                            SIZE_T *pEEThreadStateNCOffset,
                                            SIZE_T *pEEThreadPGCDisabledOffset,
@@ -1417,7 +1414,6 @@ void EEDbgInterfaceImpl::GetRuntimeOffsets(SIZE_T *pTLSIndex,
         PRECONDITION(CheckPointer(pTLSIndex));
         PRECONDITION(CheckPointer(pTLSIsSpecialIndex));
         PRECONDITION(CheckPointer(pEEThreadStateOffset));
-        PRECONDITION(CheckPointer(pTLSIndexOfPredefs));
         PRECONDITION(CheckPointer(pEEThreadStateNCOffset));
         PRECONDITION(CheckPointer(pEEThreadPGCDisabledOffset));
         PRECONDITION(CheckPointer(pEEThreadPGCDisabledValue));
@@ -1433,14 +1429,9 @@ void EEDbgInterfaceImpl::GetRuntimeOffsets(SIZE_T *pTLSIndex,
     }
     CONTRACTL_END;
     
-#ifdef FEATURE_IMPLICIT_TLS
-    *pTLSIndex = _tls_index;
-#else
-    *pTLSIndex = GetThreadTLSIndex();
-#endif
+    *pTLSIndex = g_TlsIndex;
     *pTLSIsSpecialIndex = TlsIdx_ThreadType;
     *pTLSCantStopIndex = TlsIdx_CantStopCount;
-    *pTLSIndexOfPredefs = CExecutionEngine::TlsIndex;
     *pEEThreadStateOffset = Thread::GetOffsetOfState();
     *pEEThreadStateNCOffset = Thread::GetOffsetOfStateNC();
     *pEEThreadPGCDisabledOffset = Thread::GetOffsetOfGCFlag();
diff --git a/src/vm/eedbginterfaceimpl.h b/src/vm/eedbginterfaceimpl.h
index 979c706fb2..7451246a21 100644
--- a/src/vm/eedbginterfaceimpl.h
+++ b/src/vm/eedbginterfaceimpl.h
@@ -272,7 +272,6 @@ public:
     void GetRuntimeOffsets(SIZE_T *pTLSIndex,
                            SIZE_T *pTLSIsSpecialIndex,
                            SIZE_T *pTLSCantStopIndex,
-                           SIZE_T *pTLSIndexOfPredefs,
                            SIZE_T *pEEThreadStateOffset,
                            SIZE_T *pEEThreadStateNCOffset,
                            SIZE_T *pEEThreadPGCDisabledOffset,
diff --git a/src/vm/i386/asmconstants.h b/src/vm/i386/asmconstants.h
index f7d5f709dc..aa420428f7 100644
--- a/src/vm/i386/asmconstants.h
+++ b/src/vm/i386/asmconstants.h
@@ -317,9 +317,6 @@ ASMCONSTANTS_C_ASSERT(ASM__VTABLE_SLOTS_PER_CHUNK == VTABLE_SLOTS_PER_CHUNK)
 #define ASM__VTABLE_SLOTS_PER_CHUNK_LOG2 3
 ASMCONSTANTS_C_ASSERT(ASM__VTABLE_SLOTS_PER_CHUNK_LOG2 == VTABLE_SLOTS_PER_CHUNK_LOG2)
 
-#define TLS_GETTER_MAX_SIZE_ASM DBG_FRE(0x20, 0x10)
-ASMCONSTANTS_C_ASSERT(TLS_GETTER_MAX_SIZE_ASM == TLS_GETTER_MAX_SIZE)
-
 #define JIT_TailCall_StackOffsetToFlags       0x08
 
 #define CallDescrData__pSrc                0x00
diff --git a/src/vm/i386/asmhelpers.asm b/src/vm/i386/asmhelpers.asm
index 9df33219ac..17d521fb92 100644
--- a/src/vm/i386/asmhelpers.asm
+++ b/src/vm/i386/asmhelpers.asm
@@ -47,23 +47,10 @@ endif ; FEATURE_COMINTEROP
 EXTERN __alloca_probe:PROC
 EXTERN _NDirectImportWorker@4:PROC
 EXTERN _UMThunkStubRareDisableWorker@8:PROC
-ifndef FEATURE_IMPLICIT_TLS
-ifdef ENABLE_GET_THREAD_GENERIC_FULL_CHECK
-; This is defined in C (threads.cpp) and enforces EE_THREAD_NOT_REQUIRED contracts
-GetThreadGenericFullCheck EQU ?GetThreadGenericFullCheck@@YGPAVThread@@XZ
-EXTERN  GetThreadGenericFullCheck:PROC
-endif ; ENABLE_GET_THREAD_GENERIC_FULL_CHECK
-
-EXTERN _gThreadTLSIndex:DWORD
-EXTERN _gAppDomainTLSIndex:DWORD
-endif ; FEATURE_IMPLICIT_TLS
 
 EXTERN _VarargPInvokeStubWorker@12:PROC
 EXTERN _GenericPInvokeCalliStubWorker@12:PROC
 
-EXTERN _GetThread@0:PROC
-EXTERN _GetAppDomain@0:PROC
-
 ifdef MDA_SUPPORTED
 EXTERN _PInvokeStackImbalanceWorker@8:PROC
 endif
@@ -716,56 +703,6 @@ doRet:
 FASTCALL_ENDFUNC HelperMethodFrameRestoreState
 
 
-ifndef FEATURE_IMPLICIT_TLS
-;---------------------------------------------------------------------------
-; Portable GetThread() function: used if no platform-specific optimizations apply.
-; This is in assembly code because we count on edx not getting trashed on calls
-; to this function.
-;---------------------------------------------------------------------------
-; Thread* __stdcall GetThreadGeneric(void);
-GetThreadGeneric PROC stdcall public USES ecx edx
-
-ifdef _DEBUG
-    cmp         dword ptr [_gThreadTLSIndex], -1
-    jnz         @F
-    int         3
-@@:
-endif
-ifdef ENABLE_GET_THREAD_GENERIC_FULL_CHECK
-    ; non-PAL, debug-only GetThreadGeneric should defer to GetThreadGenericFullCheck
-    ; to do extra contract enforcement.  (See GetThreadGenericFullCheck for details.)
-    ; This code is intentionally not added to asmhelper.s, as this enforcement is only
-    ; implemented for non-PAL builds.
-    call        GetThreadGenericFullCheck
-else
-    push        dword ptr [_gThreadTLSIndex]
-    call        dword ptr [__imp__TlsGetValue@4]
-endif
-    ret
-GetThreadGeneric ENDP
-
-;---------------------------------------------------------------------------
-; Portable GetAppdomain() function: used if no platform-specific optimizations apply.
-; This is in assembly code because we count on edx not getting trashed on calls
-; to this function.
-;---------------------------------------------------------------------------
-; Appdomain* __stdcall GetAppDomainGeneric(void);
-GetAppDomainGeneric PROC stdcall public USES ecx edx
-
-ifdef _DEBUG
-    cmp         dword ptr [_gAppDomainTLSIndex], -1
-    jnz         @F
-    int         3
-@@:
-endif
-
-    push        dword ptr [_gAppDomainTLSIndex]
-    call        dword ptr [__imp__TlsGetValue@4]
-    ret
-GetAppDomainGeneric ENDP
-endif
-
-
 ifdef FEATURE_HIJACK
 
 ; A JITted method's return address was hijacked to return to us here.  
diff --git a/src/vm/i386/cgencpu.h b/src/vm/i386/cgencpu.h
index 5360b3eb0e..ffdfb82b14 100644
--- a/src/vm/i386/cgencpu.h
+++ b/src/vm/i386/cgencpu.h
@@ -108,14 +108,6 @@ BOOL Runtime_Test_For_SSE2();
 #define ENREGISTERED_RETURNTYPE_INTEGER_MAXSIZE 4
 #define CALLDESCR_ARGREGS                       1   // CallDescrWorker has ArgumentRegister parameter
 
-// Max size of patched TLS helpers
-#ifdef _DEBUG
-// Debug build needs extra space for last error trashing
-#define TLS_GETTER_MAX_SIZE 0x20
-#else
-#define TLS_GETTER_MAX_SIZE 0x10
-#endif
-
 //=======================================================================
 // IMPORTANT: This value is used to figure out how much to allocate
 // for a fixed array of FieldMarshaler's. That means it must be at least
@@ -558,6 +550,12 @@ inline BOOL ClrFlushInstructionCache(LPCVOID pCodeAddr, size_t sizeOfCode)
     return TRUE;
 }
 
+//
+// JIT HELPER ALIASING FOR PORTABILITY.
+//
+// Create alias for optimized implementations of helpers provided on this platform
+//
+
 // optimized static helpers generated dynamically at runtime
 // #define JIT_GetSharedGCStaticBase
 // #define JIT_GetSharedNonGCStaticBase
@@ -573,4 +571,5 @@ inline BOOL ClrFlushInstructionCache(LPCVOID pCodeAddr, size_t sizeOfCode)
 #define JIT_NewCrossContext         JIT_NewCrossContext
 #define JIT_Stelem_Ref              JIT_Stelem_Ref
 #endif // FEATURE_PAL
+
 #endif // __cgenx86_h__
diff --git a/src/vm/i386/cgenx86.cpp b/src/vm/i386/cgenx86.cpp
index 3b523c9c28..c91a716137 100644
--- a/src/vm/i386/cgenx86.cpp
+++ b/src/vm/i386/cgenx86.cpp
@@ -1076,7 +1076,7 @@ Stub *GenerateInitPInvokeFrameHelper()
     unsigned negSpace = FrameInfo.offsetOfFrameVptr;
 
     // mov esi, GetThread()
-    psl->X86EmitCurrentThreadFetch(kESI, (1<<kEDI)|(1<<kEBX)|(1<<kECX)|(1<<kEDX));
+    psl->X86EmitCurrentThreadFetch(kESI);
 
     // mov [edi + FrameInfo.offsetOfGSCookie], GetProcessGSCookie()
     psl->X86EmitOffsetModRM(0xc7, (X86Reg)0x0, kEDI, FrameInfo.offsetOfGSCookie - negSpace);
diff --git a/src/vm/i386/jithelp.asm b/src/vm/i386/jithelp.asm
index 85e824040a..5d64b18c94 100644
--- a/src/vm/i386/jithelp.asm
+++ b/src/vm/i386/jithelp.asm
@@ -2,11 +2,6 @@
 ; The .NET Foundation licenses this file to you under the MIT license.
 ; See the LICENSE file in the project root for more information.
 
-; ==++==
-; 
-
-; 
-; ==--==
 ; ***********************************************************************
 ; File: JIThelp.asm
 ;
@@ -70,13 +65,11 @@ endif
 EXTERN _g_TailCallFrameVptr:DWORD
 EXTERN @JIT_FailFast@0:PROC
 EXTERN _s_gsCookie:DWORD
+EXTERN _GetThread@0:PROC
 EXTERN @JITutil_IsInstanceOfInterface@8:PROC
 EXTERN @JITutil_ChkCastInterface@8:PROC
 EXTERN @JITutil_IsInstanceOfAny@8:PROC
 EXTERN @JITutil_ChkCastAny@8:PROC
-ifdef FEATURE_IMPLICIT_TLS
-EXTERN _GetThread@0:PROC
-endif
 
 ifdef WRITE_BARRIER_CHECK 
 ; Those global variables are always defined, but should be 0 for Server GC
@@ -963,13 +956,15 @@ NewArgs         equ 20
 ; extra space is incremented as we push things on the stack along the way
 ExtraSpace      = 0
 
-        call    _GetThread@0; eax = Thread*
-        push    eax         ; Thread*
+        push    0           ; Thread*
 
         ; save ArgumentRegisters
         push    ecx
         push    edx
 
+        call    _GetThread@0; eax = Thread*
+        mov     [esp + 8], eax
+
 ExtraSpace      = 12    ; pThread, ecx, edx
 
 ifdef FEATURE_HIJACK
@@ -1247,44 +1242,8 @@ _JIT_PatchedCodeStart@0 proc public
 ret
 _JIT_PatchedCodeStart@0 endp
 
-;
-; Optimized TLS getters
-;
-
-            ALIGN 4
-            
-ifndef FEATURE_IMPLICIT_TLS
-_GetThread@0 proc public
-            ; This will be overwritten at runtime with optimized GetThread implementation
-            jmp short _GetTLSDummy@0
-            ; Just allocate space that will be filled in at runtime
-            db (TLS_GETTER_MAX_SIZE_ASM - 2) DUP (0CCh)
-_GetThread@0 endp
-
-            ALIGN 4
-
-_GetAppDomain@0 proc public
-            ; This will be overwritten at runtime with optimized GetAppDomain implementation
-            jmp short _GetTLSDummy@0
-            ; Just allocate space that will be filled in at runtime
-            db (TLS_GETTER_MAX_SIZE_ASM - 2) DUP (0CCh)
-_GetAppDomain@0 endp
-
-_GetTLSDummy@0 proc public
-            xor eax,eax
-            ret
-_GetTLSDummy@0 endp
-
             ALIGN 4
 
-_ClrFlsGetBlock@0 proc public
-            ; This will be overwritten at runtime with optimized ClrFlsGetBlock implementation
-            jmp short _GetTLSDummy@0
-            ; Just allocate space that will be filled in at runtime
-            db (TLS_GETTER_MAX_SIZE_ASM - 2) DUP (0CCh)
-_ClrFlsGetBlock@0 endp
-endif
-
 ;**********************************************************************
 ; Write barriers generated at runtime
 
diff --git a/src/vm/i386/jitinterfacex86.cpp b/src/vm/i386/jitinterfacex86.cpp
index 8463b084ac..1b470c9533 100644
--- a/src/vm/i386/jitinterfacex86.cpp
+++ b/src/vm/i386/jitinterfacex86.cpp
@@ -539,7 +539,7 @@ void JIT_TrialAlloc::EmitCore(CPUSTUBLINKER *psl, CodeLabel *noLock, CodeLabel *
                  && "EAX should contain size for allocation and it doesnt!!!");
 
         // Fetch current thread into EDX, preserving EAX and ECX
-        psl->X86EmitCurrentThreadFetch(kEDX, (1<<kEAX)|(1<<kECX));
+        psl->X86EmitCurrentThreadFetch(kEDX);
 
         // Try the allocation.
 
@@ -1253,7 +1253,7 @@ FastPrimitiveArrayAllocatorFuncPtr fastPrimitiveArrayAllocator = UnframedAllocat
 
 // "init" should be the address of a routine which takes an argument of
 // the module domain ID, the class domain ID, and returns the static base pointer
-void EmitFastGetSharedStaticBase(CPUSTUBLINKER *psl, CodeLabel *init, bool bCCtorCheck, bool bGCStatic, bool bSingleAppDomain)
+void EmitFastGetSharedStaticBase(CPUSTUBLINKER *psl, CodeLabel *init, bool bCCtorCheck, bool bGCStatic)
 {
     STANDARD_VM_CONTRACT;
 
@@ -1267,35 +1267,6 @@ void EmitFastGetSharedStaticBase(CPUSTUBLINKER *psl, CodeLabel *init, bool bCCto
     psl->Emit8(0x89);
     psl->Emit8(0xc8);
 
-    if(!bSingleAppDomain)
-    {
-        // Check tag
-        CodeLabel *cctorCheck = psl->NewCodeLabel();
-
-
-        // test eax, 1
-        psl->Emit8(0xa9);
-        psl->Emit32(1);
-
-        // jz cctorCheck
-        psl->X86EmitCondJump(cctorCheck, X86CondCode::kJZ);
-
-        // mov eax GetAppDomain()
-        psl->X86EmitCurrentAppDomainFetch(kEAX, (1<<kECX)|(1<<kEDX));
-
-        // mov eax [eax->m_sDomainLocalBlock.m_pModuleSlots]
-        psl->X86EmitIndexRegLoad(kEAX, kEAX, (__int32) AppDomain::GetOffsetOfModuleSlotsPointer());
-
-        // Note: weird address arithmetic effectively does:
-        // shift over 1 to remove tag bit (which is always 1), then multiply by 4.
-        // mov eax [eax + ecx*2 - 2]
-        psl->X86EmitOp(0x8b, kEAX, kEAX, -2, kECX, 2);
-
-        // cctorCheck:
-        psl->EmitLabel(cctorCheck);
-
-    }
-
     if (bCCtorCheck)
     {
         // test [eax + edx + offsetof(DomainLocalModule, m_pDataBlob], ClassInitFlags::INITIALIZED_FLAG       // Is class inited
@@ -1356,7 +1327,7 @@ void EmitFastGetSharedStaticBase(CPUSTUBLINKER *psl, CodeLabel *init, bool bCCto
 
 }
 
-void *GenFastGetSharedStaticBase(bool bCheckCCtor, bool bGCStatic, bool bSingleAppDomain)
+void *GenFastGetSharedStaticBase(bool bCheckCCtor, bool bGCStatic)
 {
     STANDARD_VM_CONTRACT;
 
@@ -1372,7 +1343,7 @@ void *GenFastGetSharedStaticBase(bool bCheckCCtor, bool bGCStatic, bool bSingleA
         init = sl.NewExternalCodeLabel((LPVOID)JIT_GetSharedNonGCStaticBase);
     }
 
-    EmitFastGetSharedStaticBase(&sl, init, bCheckCCtor, bGCStatic, bSingleAppDomain);
+    EmitFastGetSharedStaticBase(&sl, init, bCheckCCtor, bGCStatic);
 
     Stub *pStub = sl.Link(SystemDomain::GetGlobalLoaderAllocator()->GetExecutableHeap());
 
@@ -1521,16 +1492,14 @@ void InitJITHelpers1()
         //UnframedAllocateString;
     }
 
-    bool bSingleAppDomain = IsSingleAppDomain();
-
     // Replace static helpers with faster assembly versions
-    pMethodAddresses[6] = GenFastGetSharedStaticBase(true, true, bSingleAppDomain);
+    pMethodAddresses[6] = GenFastGetSharedStaticBase(true, true);
     SetJitHelperFunction(CORINFO_HELP_GETSHARED_GCSTATIC_BASE, pMethodAddresses[6]);
-    pMethodAddresses[7] = GenFastGetSharedStaticBase(true, false, bSingleAppDomain);
+    pMethodAddresses[7] = GenFastGetSharedStaticBase(true, false);
     SetJitHelperFunction(CORINFO_HELP_GETSHARED_NONGCSTATIC_BASE, pMethodAddresses[7]);
-    pMethodAddresses[8] = GenFastGetSharedStaticBase(false, true, bSingleAppDomain);
+    pMethodAddresses[8] = GenFastGetSharedStaticBase(false, true);
     SetJitHelperFunction(CORINFO_HELP_GETSHARED_GCSTATIC_BASE_NOCTOR, pMethodAddresses[8]);
-    pMethodAddresses[9] = GenFastGetSharedStaticBase(false, false, bSingleAppDomain);
+    pMethodAddresses[9] = GenFastGetSharedStaticBase(false, false);
     SetJitHelperFunction(CORINFO_HELP_GETSHARED_NONGCSTATIC_BASE_NOCTOR, pMethodAddresses[9]);
 
     ETW::MethodLog::StubsInitialized(pMethodAddresses, (PVOID *)pHelperNames, ETW_NUM_JIT_HELPERS);
diff --git a/src/vm/i386/stublinkerx86.cpp b/src/vm/i386/stublinkerx86.cpp
index b77609822b..3b376be26d 100644
--- a/src/vm/i386/stublinkerx86.cpp
+++ b/src/vm/i386/stublinkerx86.cpp
@@ -2320,7 +2320,7 @@ static const X86Reg c_argRegs[] = {
 
 #ifndef CROSSGEN_COMPILE
 
-#if defined(_DEBUG) && (defined(_TARGET_AMD64_) || defined(_TARGET_X86_)) && !defined(FEATURE_PAL)
+#if defined(_DEBUG) && !defined(FEATURE_PAL)
 void StubLinkerCPU::EmitJITHelperLoggingThunk(PCODE pJitHelper, LPVOID helperFuncCount)
 {
     STANDARD_VM_CONTRACT;
@@ -2358,181 +2358,36 @@ void StubLinkerCPU::EmitJITHelperLoggingThunk(PCODE pJitHelper, LPVOID helperFun
 #endif
     X86EmitTailcallWithSinglePop(NewExternalCodeLabel(pJitHelper), kECX);
 }
-#endif // _DEBUG && (_TARGET_AMD64_ || _TARGET_X86_) && !FEATURE_PAL
+#endif // _DEBUG && !FEATURE_PAL
 
-#ifndef FEATURE_IMPLICIT_TLS
-//---------------------------------------------------------------
-// Emit code to store the current Thread structure in dstreg
-// preservedRegSet is a set of registers to be preserved
-// TRASHES  EAX, EDX, ECX unless they are in preservedRegSet.
-// RESULTS  dstreg = current Thread
-//---------------------------------------------------------------
-VOID StubLinkerCPU::X86EmitTLSFetch(DWORD idx, X86Reg dstreg, unsigned preservedRegSet)
+#if !defined(FEATURE_STUBS_AS_IL)
+VOID StubLinkerCPU::X86EmitCurrentThreadFetch(X86Reg dstreg)
 {
     CONTRACTL
     {
         STANDARD_VM_CHECK;
 
-    // It doesn't make sense to have the destination register be preserved
-        PRECONDITION((preservedRegSet & (1<<dstreg)) == 0);
         AMD64_ONLY(PRECONDITION(dstreg < 8)); // code below doesn't support high registers
     }
     CONTRACTL_END;
 
-    TLSACCESSMODE mode = GetTLSAccessMode(idx);
-
-#ifdef _DEBUG
-    {
-        static BOOL f = TRUE;
-        f = !f;
-        if (f)
-        {
-           mode = TLSACCESS_GENERIC;
-        }
-    }
-#endif
-
-    switch (mode)
-    {
-        case TLSACCESS_WNT: 
-            {
-                unsigned __int32 tlsofs = offsetof(TEB, TlsSlots) + (idx * sizeof(void*));
-#ifdef _TARGET_AMD64_                
-                BYTE code[] = {0x65,0x48,0x8b,0x04,0x25};    // mov dstreg, qword ptr gs:[IMM32]
-                static const int regByteIndex = 3;
-#elif defined(_TARGET_X86_)
-                BYTE code[] = {0x64,0x8b,0x05};              // mov dstreg, dword ptr fs:[IMM32]
-                static const int regByteIndex = 2;
-#endif 
-                code[regByteIndex] |= (dstreg << 3);
-
-                EmitBytes(code, sizeof(code));
-                Emit32(tlsofs);
-            }
-            break;
-
-        case TLSACCESS_GENERIC:
-
-            X86EmitPushRegs(preservedRegSet & ((1<<kEAX)|(1<<kEDX)|(1<<kECX)));
-
-            X86EmitPushImm32(idx);
 #ifdef _TARGET_AMD64_
-            X86EmitPopReg (kECX);       // arg in reg
-#endif
-
-            // call TLSGetValue
-            X86EmitCall(NewExternalCodeLabel((LPVOID) TlsGetValue), sizeof(void*));
-
-            // mov dstreg, eax
-            X86EmitMovRegReg(dstreg, kEAX);
-
-            X86EmitPopRegs(preservedRegSet & ((1<<kEAX)|(1<<kEDX)|(1<<kECX)));
-
-            break;
-
-        default:
-            _ASSERTE(0);
-    }
-
-#ifdef _DEBUG
-    // Trash caller saved regs that we were not told to preserve, and that aren't the dstreg.
-    preservedRegSet |= 1<<dstreg;
-    if (!(preservedRegSet & (1<<kEAX)))
-        X86EmitDebugTrashReg(kEAX);
-    if (!(preservedRegSet & (1<<kEDX)))
-        X86EmitDebugTrashReg(kEDX);
-    if (!(preservedRegSet & (1<<kECX)))
-        X86EmitDebugTrashReg(kECX);
-#endif
-
-}
-#endif // FEATURE_IMPLICIT_TLS
-
-VOID StubLinkerCPU::X86EmitCurrentThreadFetch(X86Reg dstreg, unsigned preservedRegSet)
-{
-    CONTRACTL
-    {
-        STANDARD_VM_CHECK;
-
-    // It doesn't make sense to have the destination register be preserved
-        PRECONDITION((preservedRegSet & (1<<dstreg)) == 0);
-        AMD64_ONLY(PRECONDITION(dstreg < 8)); // code below doesn't support high registers
-    }
-    CONTRACTL_END;
-
-#ifdef FEATURE_IMPLICIT_TLS
-
-    X86EmitPushRegs(preservedRegSet & ((1<<kEAX)|(1<<kEDX)|(1<<kECX)));
-
-    //TODO: Inline the instruction instead of a call
-    // call GetThread
-    X86EmitCall(NewExternalCodeLabel((LPVOID) GetThread), sizeof(void*));
-
-    // mov dstreg, eax
-    X86EmitMovRegReg(dstreg, kEAX);
-
-    X86EmitPopRegs(preservedRegSet & ((1<<kEAX)|(1<<kEDX)|(1<<kECX)));
-
-#ifdef _DEBUG
-    // Trash caller saved regs that we were not told to preserve, and that aren't the dstreg.
-    preservedRegSet |= 1<<dstreg;
-    if (!(preservedRegSet & (1<<kEAX)))
-        X86EmitDebugTrashReg(kEAX);
-    if (!(preservedRegSet & (1<<kEDX)))
-        X86EmitDebugTrashReg(kEDX);
-    if (!(preservedRegSet & (1<<kECX)))
-        X86EmitDebugTrashReg(kECX);
-#endif // _DEBUG
-
-#else // FEATURE_IMPLICIT_TLS
-
-    X86EmitTLSFetch(GetThreadTLSIndex(), dstreg, preservedRegSet);
-    
-#endif // FEATURE_IMPLICIT_TLS
-
-}
-
-VOID StubLinkerCPU::X86EmitCurrentAppDomainFetch(X86Reg dstreg, unsigned preservedRegSet)
-{
-    CONTRACTL
-    {
-        STANDARD_VM_CHECK;
-
-    // It doesn't make sense to have the destination register be preserved
-        PRECONDITION((preservedRegSet & (1<<dstreg)) == 0);
-        AMD64_ONLY(PRECONDITION(dstreg < 8)); // code below doesn't support high registers
-    }
-    CONTRACTL_END;
-
-#ifdef FEATURE_IMPLICIT_TLS
-    X86EmitPushRegs(preservedRegSet & ((1<<kEAX)|(1<<kEDX)|(1<<kECX)));
-
-    //TODO: Inline the instruction instead of a call
-    // call GetThread
-    X86EmitCall(NewExternalCodeLabel((LPVOID) GetAppDomain), sizeof(void*));
-
-    // mov dstreg, eax
-    X86EmitMovRegReg(dstreg, kEAX);
-
-    X86EmitPopRegs(preservedRegSet & ((1<<kEAX)|(1<<kEDX)|(1<<kECX)));
-
-#ifdef _DEBUG
-    // Trash caller saved regs that we were not told to preserve, and that aren't the dstreg.
-    preservedRegSet |= 1<<dstreg;
-    if (!(preservedRegSet & (1<<kEAX)))
-        X86EmitDebugTrashReg(kEAX);
-    if (!(preservedRegSet & (1<<kEDX)))
-        X86EmitDebugTrashReg(kEDX);
-    if (!(preservedRegSet & (1<<kECX)))
-        X86EmitDebugTrashReg(kECX);
+    BYTE code[] = { 0x65,0x48,0x8b,0x04,0x25 };    // mov dstreg, qword ptr gs:[IMM32]
+    static const int regByteIndex = 3;
+#elif defined(_TARGET_X86_)
+    BYTE code[] = { 0x64,0x8b,0x05 };              // mov dstreg, dword ptr fs:[IMM32]
+    static const int regByteIndex = 2;
 #endif
+    code[regByteIndex] |= (dstreg << 3);
 
-#else // FEATURE_IMPLICIT_TLS
+    EmitBytes(code, sizeof(code));
+    Emit32(offsetof(TEB, ThreadLocalStoragePointer));
 
-    X86EmitTLSFetch(GetAppDomainTLSIndex(), dstreg, preservedRegSet);
+    X86EmitIndexRegLoad(dstreg, dstreg, sizeof(void *) * (g_TlsIndex & 0xFFFF));
 
-#endif // FEATURE_IMPLICIT_TLS
+    X86EmitIndexRegLoad(dstreg, dstreg, (g_TlsIndex & 0x7FFF0000) >> 16);
 }
+#endif // !FEATURE_STUBS_AS_IL
 
 #if defined(_TARGET_X86_)
 
@@ -2861,56 +2716,7 @@ VOID StubLinkerCPU::EmitSetup(CodeLabel *pForwardRef)
 {
     STANDARD_VM_CONTRACT;
 
-#ifdef FEATURE_IMPLICIT_TLS
-    DWORD idx = 0;
-    TLSACCESSMODE mode = TLSACCESS_GENERIC;
-#else
-    DWORD idx = GetThreadTLSIndex();
-    TLSACCESSMODE mode = GetTLSAccessMode(idx);
-#endif
-
-#ifdef _DEBUG
-    {
-        static BOOL f = TRUE;
-        f = !f;
-        if (f)
-        {
-           mode = TLSACCESS_GENERIC;
-        }
-    }
-#endif
-
-    switch (mode)
-    {
-        case TLSACCESS_WNT: 
-#ifndef FEATURE_PAL
-            {
-                unsigned __int32 tlsofs = offsetof(TEB, TlsSlots) + (idx * sizeof(void*));
-
-                static const BYTE code[] = {0x64,0x8b,0x1d};              // mov ebx, dword ptr fs:[IMM32]
-                EmitBytes(code, sizeof(code));
-                Emit32(tlsofs);
-            }
-#else  // !FEATURE_PAL
-            _ASSERTE("TLSACCESS_WNT mode is not supported");
-#endif // !FEATURE_PAL
-            break;
-
-        case TLSACCESS_GENERIC:
-#ifdef FEATURE_IMPLICIT_TLS
-            X86EmitCall(NewExternalCodeLabel((LPVOID) GetThread), sizeof(void*));
-#else
-            X86EmitPushImm32(idx);
-
-            // call TLSGetValue
-            X86EmitCall(NewExternalCodeLabel((LPVOID) TlsGetValue), sizeof(void*));
-#endif
-            // mov ebx,eax
-            Emit16(0xc389);
-            break;
-        default:
-            _ASSERTE(0);
-    }
+    X86EmitCurrentThreadFetch(kEBX);
 
     // cmp ebx, 0
     static const BYTE b[] = { 0x83, 0xFB, 0x0};
@@ -3150,8 +2956,7 @@ VOID StubLinkerCPU::EmitMethodStubProlog(TADDR pFrameVptr, int transitionBlockOf
 #endif // _TARGET_X86_
 
     // ebx <-- GetThread()
-    // Trashes X86TLSFetch_TRASHABLE_REGS
-    X86EmitCurrentThreadFetch(kEBX, 0);
+    X86EmitCurrentThreadFetch(kEBX);
 
 #if _DEBUG
 
diff --git a/src/vm/i386/stublinkerx86.h b/src/vm/i386/stublinkerx86.h
index 76d1f95845..d523bb0461 100644
--- a/src/vm/i386/stublinkerx86.h
+++ b/src/vm/i386/stublinkerx86.h
@@ -219,11 +219,7 @@ class StubLinkerCPU : public StubLinker
         VOID X86EmitLeaRIP(CodeLabel *target, X86Reg reg);
 #endif
 
-        static const unsigned X86TLSFetch_TRASHABLE_REGS = (1<<kEAX) | (1<<kEDX) | (1<<kECX);
-        VOID X86EmitTLSFetch(DWORD idx, X86Reg dstreg, unsigned preservedRegSet);
-
-        VOID X86EmitCurrentThreadFetch(X86Reg dstreg, unsigned preservedRegSet);
-        VOID X86EmitCurrentAppDomainFetch(X86Reg dstreg, unsigned preservedRegSet);
+        VOID X86EmitCurrentThreadFetch(X86Reg dstreg);
         
         VOID X86EmitIndexRegLoad(X86Reg dstreg, X86Reg srcreg, __int32 ofs = 0);
         VOID X86EmitIndexRegStore(X86Reg dstreg, __int32 ofs, X86Reg srcreg);
@@ -421,7 +417,7 @@ class StubLinkerCPU : public StubLinker
         VOID EmitDebugBreak();
 #endif // !FEATURE_STUBS_AS_IL
 
-#if defined(_DEBUG) && (defined(_TARGET_AMD64_) || defined(_TARGET_X86_)) && !defined(FEATURE_PAL)
+#if defined(_DEBUG) && !defined(FEATURE_PAL)
         //===========================================================================
         // Emits code to log JITHelper access
         void EmitJITHelperLoggingThunk(PCODE pJitHelper, LPVOID helperFuncCount);
diff --git a/src/vm/jitinterface.cpp b/src/vm/jitinterface.cpp
index d326a92ccb..f49ced56d0 100644
--- a/src/vm/jitinterface.cpp
+++ b/src/vm/jitinterface.cpp
@@ -9932,19 +9932,6 @@ DWORD CEEInfo::getThreadTLSIndex(void **ppIndirection)
     if (ppIndirection != NULL)
         *ppIndirection = NULL;
 
-    JIT_TO_EE_TRANSITION();
-
-#if !defined(CROSSGEN_COMPILE) && !defined(FEATURE_IMPLICIT_TLS)
-    result = GetThreadTLSIndex();
-
-    // The JIT can use the optimized TLS access only if the runtime is using it as well.
-    //  (This is necessaryto make managed code work well under appverifier.)
-    if (GetTLSAccessMode(result) == TLSACCESS_GENERIC)
-        result = (DWORD)-1;
-#endif
-
-    EE_TO_JIT_TRANSITION();
-
     return result;
 }
 
diff --git a/src/vm/jitinterfacegen.cpp b/src/vm/jitinterfacegen.cpp
index 38f1a7436a..b630e7f998 100644
--- a/src/vm/jitinterfacegen.cpp
+++ b/src/vm/jitinterfacegen.cpp
@@ -47,24 +47,10 @@ EXTERN_C Object* JIT_NewArr1OBJ_UP (CORINFO_CLASS_HANDLE arrayMT, INT_PTR size);
 EXTERN_C Object* JIT_NewArr1VC_MP (CORINFO_CLASS_HANDLE arrayMT, INT_PTR size);
 EXTERN_C Object* JIT_NewArr1VC_UP (CORINFO_CLASS_HANDLE arrayMT, INT_PTR size);
 
-extern "C" void* JIT_GetSharedNonGCStaticBase_Slow(SIZE_T moduleDomainID, DWORD dwModuleClassID);
-extern "C" void* JIT_GetSharedNonGCStaticBaseNoCtor_Slow(SIZE_T moduleDomainID, DWORD dwModuleClassID);
-extern "C" void* JIT_GetSharedGCStaticBase_Slow(SIZE_T moduleDomainID, DWORD dwModuleClassID);
-extern "C" void* JIT_GetSharedGCStaticBaseNoCtor_Slow(SIZE_T moduleDomainID, DWORD dwModuleClassID);
-
-extern "C" void* JIT_GetSharedNonGCStaticBase_SingleAppDomain(SIZE_T moduleDomainID, DWORD dwModuleClassID);
-extern "C" void* JIT_GetSharedNonGCStaticBaseNoCtor_SingleAppDomain(SIZE_T moduleDomainID, DWORD dwModuleClassID);
-extern "C" void* JIT_GetSharedGCStaticBase_SingleAppDomain(SIZE_T moduleDomainID, DWORD dwModuleClassID);
-extern "C" void* JIT_GetSharedGCStaticBaseNoCtor_SingleAppDomain(SIZE_T moduleDomainID, DWORD dwModuleClassID);
-
 #ifdef _TARGET_AMD64_
 extern WriteBarrierManager g_WriteBarrierManager;
 #endif // _TARGET_AMD64_
 
-#ifndef FEATURE_IMPLICIT_TLS
-EXTERN_C DWORD gThreadTLSIndex;
-EXTERN_C DWORD gAppDomainTLSIndex;
-#endif
 #endif // _WIN64
 
 /*********************************************************************/ 
@@ -73,99 +59,6 @@ EXTERN_C DWORD gAppDomainTLSIndex;
 /*********************************************************************/
 #ifndef _TARGET_X86_
 
-#if defined(_TARGET_AMD64_)
-
-void MakeIntoJumpStub(LPVOID pStubAddress, LPVOID pTarget)
-{
-    BYTE* pbStubAddress = (BYTE*)pStubAddress;
-    BYTE* pbTarget = (BYTE*)pTarget;
-
-    DWORD dwOldProtect;
-    if (!ClrVirtualProtect(pbStubAddress, 5, PAGE_EXECUTE_READWRITE, &dwOldProtect))
-    {
-        ThrowLastError();
-    }
-
-    DWORD diff = (DWORD)(pbTarget - (pbStubAddress + 5));
-
-    // Make sure that the offset fits in 32-bits
-    _ASSERTE( FitsInI4(pbTarget - (pbStubAddress + 5)) );
-        
-    // Write a jmp pcrel32 instruction
-    //
-    //      0xe9xxxxxxxx
-    pbStubAddress[0] = 0xE9;
-    *((DWORD*)&pbStubAddress[1]) = diff;
-
-    ClrVirtualProtect(pbStubAddress, 5, dwOldProtect, &dwOldProtect);
-}
-
-EXTERN_C void JIT_TrialAllocSFastMP_InlineGetThread__PatchTLSOffset();
-EXTERN_C void JIT_BoxFastMPIGT__PatchTLSLabel();
-EXTERN_C void AllocateStringFastMP_InlineGetThread__PatchTLSOffset();
-EXTERN_C void JIT_NewArr1VC_MP_InlineGetThread__PatchTLSOffset();
-EXTERN_C void JIT_NewArr1OBJ_MP_InlineGetThread__PatchTLSOffset();
-
-
-static const LPVOID InlineGetThreadLocations[] = {
-    (PVOID)JIT_TrialAllocSFastMP_InlineGetThread__PatchTLSOffset,
-    (PVOID)JIT_BoxFastMPIGT__PatchTLSLabel,
-    (PVOID)AllocateStringFastMP_InlineGetThread__PatchTLSOffset,
-    (PVOID)JIT_NewArr1VC_MP_InlineGetThread__PatchTLSOffset,
-    (PVOID)JIT_NewArr1OBJ_MP_InlineGetThread__PatchTLSOffset,
-};
-
-EXTERN_C void JIT_GetSharedNonGCStaticBase__PatchTLSLabel();
-EXTERN_C void JIT_GetSharedNonGCStaticBaseNoCtor__PatchTLSLabel();
-EXTERN_C void JIT_GetSharedGCStaticBase__PatchTLSLabel();
-EXTERN_C void JIT_GetSharedGCStaticBaseNoCtor__PatchTLSLabel();
-
-static const LPVOID InlineGetAppDomainLocations[] = {
-    (PVOID)JIT_GetSharedNonGCStaticBase__PatchTLSLabel,
-    (PVOID)JIT_GetSharedNonGCStaticBaseNoCtor__PatchTLSLabel,
-    (PVOID)JIT_GetSharedGCStaticBase__PatchTLSLabel,
-    (PVOID)JIT_GetSharedGCStaticBaseNoCtor__PatchTLSLabel
-};
-
-
-#endif // defined(_TARGET_AMD64_)
-
-#if defined(_WIN64) && !defined(FEATURE_IMPLICIT_TLS)
-void FixupInlineGetters(DWORD tlsSlot, const LPVOID * pLocations, int nLocations)
-{
-    BYTE* pInlineGetter;
-    DWORD dwOldProtect;
-    for (int i=0; i<nLocations; i++)
-    {
-        pInlineGetter = (BYTE*)GetEEFuncEntryPoint((BYTE*)pLocations[i]);
-
-        static const DWORD cbPatch = 9;
-        if (!ClrVirtualProtect(pInlineGetter, cbPatch, PAGE_EXECUTE_READWRITE, &dwOldProtect))
-        {
-            ThrowLastError();
-        }
-
-        DWORD offset = (tlsSlot * sizeof(LPVOID) + offsetof(TEB, TlsSlots));
-
-#if defined(_TARGET_AMD64_)
-        // mov  r??, gs:[TLS offset]
-        _ASSERTE_ALL_BUILDS("clr/src/VM/JITinterfaceGen.cpp",
-                            pInlineGetter[0] == 0x65 &&
-                            pInlineGetter[2] == 0x8B &&
-                            pInlineGetter[4] == 0x25 &&
-                            "Initialization failure while stomping instructions for the TLS slot offset: the instruction at the given offset did not match what we expect");
-
-        *((DWORD*)(pInlineGetter + 5)) = offset;
-#else // _TARGET_AMD64_
-        PORTABILITY_ASSERT("FixupInlineGetters");
-#endif //_TARGET_AMD64_
-
-        FlushInstructionCache(GetCurrentProcess(), pInlineGetter, cbPatch);
-        ClrVirtualProtect(pInlineGetter, cbPatch, dwOldProtect, &dwOldProtect);
-    }
-}
-#endif // defined(_WIN64) && !defined(FEATURE_IMPLICIT_TLS)
-
 void InitJITHelpers1()
 {
     STANDARD_VM_CONTRACT;
@@ -176,18 +69,6 @@ void InitJITHelpers1()
 
     g_WriteBarrierManager.Initialize();
 
-#ifndef FEATURE_IMPLICIT_TLS
-    if (gThreadTLSIndex < TLS_MINIMUM_AVAILABLE)
-    {
-        FixupInlineGetters(gThreadTLSIndex, InlineGetThreadLocations, COUNTOF(InlineGetThreadLocations));
-    }
-
-    if (gAppDomainTLSIndex < TLS_MINIMUM_AVAILABLE)
-    {
-        FixupInlineGetters(gAppDomainTLSIndex, InlineGetAppDomainLocations, COUNTOF(InlineGetAppDomainLocations));
-    }
-#endif // !FEATURE_IMPLICIT_TLS
-
     // Allocation helpers, faster but non-logging
     if (!((TrackAllocationsEnabled()) || 
         (LoggingOn(LF_GCALLOC, LL_INFO10))
@@ -196,43 +77,27 @@ void InitJITHelpers1()
 #endif // _DEBUG
         ))
     {
+#ifdef FEATURE_PAL
+        SetJitHelperFunction(CORINFO_HELP_NEWSFAST, JIT_NewS_MP_FastPortable);
+        SetJitHelperFunction(CORINFO_HELP_NEWSFAST_ALIGN8, JIT_NewS_MP_FastPortable);
+        SetJitHelperFunction(CORINFO_HELP_NEWARR_1_VC, JIT_NewArr1VC_MP_FastPortable);
+        SetJitHelperFunction(CORINFO_HELP_NEWARR_1_OBJ, JIT_NewArr1OBJ_MP_FastPortable);
+
+        ECall::DynamicallyAssignFCallImpl(GetEEFuncEntryPoint(AllocateString_MP_FastPortable), ECall::FastAllocateString);
+#else // FEATURE_PAL
         // if (multi-proc || server GC)
         if (GCHeapUtilities::UseThreadAllocationContexts())
         {
-#ifdef FEATURE_IMPLICIT_TLS
-            SetJitHelperFunction(CORINFO_HELP_NEWSFAST, JIT_NewS_MP_FastPortable);
-            SetJitHelperFunction(CORINFO_HELP_NEWSFAST_ALIGN8, JIT_NewS_MP_FastPortable);
-            SetJitHelperFunction(CORINFO_HELP_NEWARR_1_VC, JIT_NewArr1VC_MP_FastPortable);
-            SetJitHelperFunction(CORINFO_HELP_NEWARR_1_OBJ, JIT_NewArr1OBJ_MP_FastPortable);
-
-            ECall::DynamicallyAssignFCallImpl(GetEEFuncEntryPoint(AllocateString_MP_FastPortable), ECall::FastAllocateString);
-#else // !FEATURE_IMPLICIT_TLS
-            // If the TLS for Thread is low enough use the super-fast helpers
-            if (gThreadTLSIndex < TLS_MINIMUM_AVAILABLE)
-            {
-                SetJitHelperFunction(CORINFO_HELP_NEWSFAST, JIT_TrialAllocSFastMP_InlineGetThread);
-                SetJitHelperFunction(CORINFO_HELP_NEWSFAST_ALIGN8, JIT_TrialAllocSFastMP_InlineGetThread);
-                SetJitHelperFunction(CORINFO_HELP_BOX, JIT_BoxFastMP_InlineGetThread);
-                SetJitHelperFunction(CORINFO_HELP_NEWARR_1_VC, JIT_NewArr1VC_MP_InlineGetThread);
-                SetJitHelperFunction(CORINFO_HELP_NEWARR_1_OBJ, JIT_NewArr1OBJ_MP_InlineGetThread);
-
-                ECall::DynamicallyAssignFCallImpl(GetEEFuncEntryPoint(AllocateStringFastMP_InlineGetThread), ECall::FastAllocateString);
-            }
-            else
-            {
-                SetJitHelperFunction(CORINFO_HELP_NEWSFAST, JIT_TrialAllocSFastMP);
-                SetJitHelperFunction(CORINFO_HELP_NEWSFAST_ALIGN8, JIT_TrialAllocSFastMP);
-                SetJitHelperFunction(CORINFO_HELP_BOX, JIT_BoxFastMP);
-                SetJitHelperFunction(CORINFO_HELP_NEWARR_1_VC, JIT_NewArr1VC_MP);
-                SetJitHelperFunction(CORINFO_HELP_NEWARR_1_OBJ, JIT_NewArr1OBJ_MP);
+            SetJitHelperFunction(CORINFO_HELP_NEWSFAST, JIT_TrialAllocSFastMP_InlineGetThread);
+            SetJitHelperFunction(CORINFO_HELP_NEWSFAST_ALIGN8, JIT_TrialAllocSFastMP_InlineGetThread);
+            SetJitHelperFunction(CORINFO_HELP_BOX, JIT_BoxFastMP_InlineGetThread);
+            SetJitHelperFunction(CORINFO_HELP_NEWARR_1_VC, JIT_NewArr1VC_MP_InlineGetThread);
+            SetJitHelperFunction(CORINFO_HELP_NEWARR_1_OBJ, JIT_NewArr1OBJ_MP_InlineGetThread);
 
-                ECall::DynamicallyAssignFCallImpl(GetEEFuncEntryPoint(AllocateStringFastMP), ECall::FastAllocateString);
-            }
-#endif // FEATURE_IMPLICIT_TLS
+            ECall::DynamicallyAssignFCallImpl(GetEEFuncEntryPoint(AllocateStringFastMP_InlineGetThread), ECall::FastAllocateString);
         }
         else
         {
-#ifndef FEATURE_PAL
             // Replace the 1p slow allocation helpers with faster version
             //
             // When we're running Workstation GC on a single proc box we don't have 
@@ -244,27 +109,9 @@ void InitJITHelpers1()
             SetJitHelperFunction(CORINFO_HELP_NEWARR_1_OBJ, JIT_NewArr1OBJ_UP);
 
             ECall::DynamicallyAssignFCallImpl(GetEEFuncEntryPoint(AllocateStringFastUP), ECall::FastAllocateString);
-#endif // !FEATURE_PAL
         }
+#endif // FEATURE_PAL
     }
-
-    if(IsSingleAppDomain())
-    {
-        SetJitHelperFunction(CORINFO_HELP_GETSHARED_GCSTATIC_BASE,          JIT_GetSharedGCStaticBase_SingleAppDomain);
-        SetJitHelperFunction(CORINFO_HELP_GETSHARED_NONGCSTATIC_BASE,       JIT_GetSharedNonGCStaticBase_SingleAppDomain);
-        SetJitHelperFunction(CORINFO_HELP_GETSHARED_GCSTATIC_BASE_NOCTOR,   JIT_GetSharedGCStaticBaseNoCtor_SingleAppDomain);
-        SetJitHelperFunction(CORINFO_HELP_GETSHARED_NONGCSTATIC_BASE_NOCTOR,JIT_GetSharedNonGCStaticBaseNoCtor_SingleAppDomain);
-    }
-#ifndef FEATURE_IMPLICIT_TLS
-    else
-    if (gAppDomainTLSIndex >= TLS_MINIMUM_AVAILABLE)
-    {
-        SetJitHelperFunction(CORINFO_HELP_GETSHARED_GCSTATIC_BASE,          JIT_GetSharedGCStaticBase_Slow);
-        SetJitHelperFunction(CORINFO_HELP_GETSHARED_NONGCSTATIC_BASE,       JIT_GetSharedNonGCStaticBase_Slow);
-        SetJitHelperFunction(CORINFO_HELP_GETSHARED_GCSTATIC_BASE_NOCTOR,   JIT_GetSharedGCStaticBaseNoCtor_Slow);
-        SetJitHelperFunction(CORINFO_HELP_GETSHARED_NONGCSTATIC_BASE_NOCTOR,JIT_GetSharedNonGCStaticBaseNoCtor_Slow);
-    }
-#endif // !FEATURE_IMPLICIT_TLS
 #endif // _TARGET_AMD64_
 }
 
diff --git a/src/vm/threads.cpp b/src/vm/threads.cpp
index 56bdffb1f1..941d3645be 100644
--- a/src/vm/threads.cpp
+++ b/src/vm/threads.cpp
@@ -71,10 +71,6 @@ static CrstStatic s_initializeYieldProcessorNormalizedCrst;
 
 BOOL Thread::s_fCleanFinalizedThread = FALSE;
 
-#ifdef ENABLE_GET_THREAD_GENERIC_FULL_CHECK
-BOOL Thread::s_fEnforceEEThreadNotRequiredContracts = FALSE;
-#endif
-
 Volatile<LONG> Thread::s_threadPoolCompletionCountOverflow = 0;
 
 CrstStatic g_DeadlockAwareCrst;
@@ -290,9 +286,6 @@ bool Thread::DetectHandleILStubsForDebugger()
     return false;
 }
 
-
-#ifdef FEATURE_IMPLICIT_TLS
-
 extern "C" {
 #ifndef __llvm__
 __declspec(thread)
@@ -304,26 +297,15 @@ ThreadLocalInfo gCurrentThreadInfo =
                                                   NULL,    // m_pThread
                                                   NULL,    // m_pAppDomain
                                                   NULL,    // m_EETlsData
-#if defined(FEATURE_MERGE_JIT_AND_ENGINE)
-                                                  NULL,    // m_pCompiler
-#endif
                                               };
 } // extern "C"
 // index into TLS Array. Definition added by compiler
 EXTERN_C UINT32 _tls_index;
 
-#else // FEATURE_IMPLICIT_TLS
-extern "C" {
-GVAL_IMPL_INIT(DWORD, gThreadTLSIndex, TLS_OUT_OF_INDEXES);      // index ( (-1) == uninitialized )
-GVAL_IMPL_INIT(DWORD, gAppDomainTLSIndex, TLS_OUT_OF_INDEXES);   // index ( (-1) == uninitialized )
-}
-#endif // FEATURE_IMPLICIT_TLS
-
 #ifndef DACCESS_COMPILE
-#ifdef FEATURE_IMPLICIT_TLS
 BOOL SetThread(Thread* t)
 {
-	LIMITED_METHOD_CONTRACT
+    LIMITED_METHOD_CONTRACT
 
     gCurrentThreadInfo.m_pThread = t;
     return TRUE;
@@ -331,52 +313,12 @@ BOOL SetThread(Thread* t)
 
 BOOL SetAppDomain(AppDomain* ad)
 {
-	LIMITED_METHOD_CONTRACT
+    LIMITED_METHOD_CONTRACT
 
     gCurrentThreadInfo.m_pAppDomain = ad;
     return TRUE;
 }
 
-#if defined(FEATURE_MERGE_JIT_AND_ENGINE)
-extern "C"
-{
-
-void* GetJitTls()
-{
-    LIMITED_METHOD_CONTRACT
-
-    return gCurrentThreadInfo.m_pJitTls;
-}
-
-void SetJitTls(void* v)
-{
-    LIMITED_METHOD_CONTRACT
-    gCurrentThreadInfo.m_pJitTls = v;
-}
-
-}
-#endif // defined(FEATURE_MERGE_JIT_AND_ENGINE)
-
-#define ThreadInited()          (TRUE)
-
-#else // FEATURE_IMPLICIT_TLS
-BOOL SetThread(Thread* t)
-{
-    WRAPPER_NO_CONTRACT
-    return UnsafeTlsSetValue(GetThreadTLSIndex(), t);
-}
-
-BOOL SetAppDomain(AppDomain* ad)
-{
-    WRAPPER_NO_CONTRACT
-    return UnsafeTlsSetValue(GetAppDomainTLSIndex(), ad);
-}
-
-#define ThreadInited()          (gThreadTLSIndex != TLS_OUT_OF_INDEXES)
-
-#endif // FEATURE_IMPLICIT_TLS
-
-
 BOOL Thread::Alert ()
 {
     CONTRACTL {
@@ -681,7 +623,6 @@ Thread* SetupThread(BOOL fInternal)
     }
     CONTRACTL_END;
 
-    _ASSERTE(ThreadInited());
     Thread* pThread;
     if ((pThread = GetThread()) != NULL)
         return pThread;
@@ -782,18 +723,6 @@ Thread* SetupThread(BOOL fInternal)
         !pThread->PrepareApartmentAndContext())
         ThrowOutOfMemory();
 
-#ifndef FEATURE_IMPLICIT_TLS
-    // make sure we will not fail when we store in TLS in the future.
-    if (!UnsafeTlsSetValue(gThreadTLSIndex, NULL))
-    {
-        ThrowOutOfMemory();
-    }
-    if (!UnsafeTlsSetValue(GetAppDomainTLSIndex(), NULL))
-    {
-        ThrowOutOfMemory();
-    }
-#endif
-
     // reset any unstarted bits on the thread object
     FastInterlockAnd((ULONG *) &pThread->m_State, ~Thread::TS_Unstarted);
     FastInterlockOr((ULONG *) &pThread->m_State, Thread::TS_LegalToJoin);
@@ -915,7 +844,6 @@ Thread* SetupUnstartedThread(BOOL bRequiresTSL)
     }
     CONTRACTL_END;
 
-    _ASSERTE(ThreadInited());
     Thread* pThread = new Thread();
 
     FastInterlockOr((ULONG *) &pThread->m_State,
@@ -1112,42 +1040,11 @@ HRESULT Thread::DetachThread(BOOL fDLLThreadDetach)
     return S_OK;
 }
 
-#ifndef FEATURE_IMPLICIT_TLS
-//---------------------------------------------------------------------------
-// Returns the TLS index for the Thread. This is strictly for the use of
-// our ASM stub generators that generate inline code to access the Thread.
-// Normally, you should use GetThread().
-//---------------------------------------------------------------------------
-DWORD GetThreadTLSIndex()
-{
-    LIMITED_METHOD_CONTRACT;
-
-    return gThreadTLSIndex;
-}
-
-//---------------------------------------------------------------------------
-// Returns the TLS index for the AppDomain. This is strictly for the use of
-// our ASM stub generators that generate inline code to access the AppDomain.
-// Normally, you should use GetAppDomain().
-//---------------------------------------------------------------------------
-DWORD GetAppDomainTLSIndex()
-{
-    LIMITED_METHOD_CONTRACT;
-
-    return gAppDomainTLSIndex;
-}
-#endif
-
 DWORD GetRuntimeId()
 {
     LIMITED_METHOD_CONTRACT;
 
-#ifndef FEATURE_IMPLICIT_TLS
-    _ASSERTE(GetThreadTLSIndex() != TLS_OUT_OF_INDEXES);
-    return GetThreadTLSIndex() + 3;
-#else
     return _tls_index;
-#endif
 }
 
 //---------------------------------------------------------------------------
@@ -1191,165 +1088,6 @@ Thread* WINAPI CreateThreadBlockThrow()
 DWORD_PTR Thread::OBJREF_HASH = OBJREF_TABSIZE;
 #endif
 
-#ifndef FEATURE_IMPLICIT_TLS
-
-#ifdef ENABLE_GET_THREAD_GENERIC_FULL_CHECK
-
-// ----------------------------------------------------------------------------
-// GetThreadGenericFullCheck
-// 
-// Description:
-//     The non-PAL, x86 / x64 assembly versions of GetThreadGeneric call into this C
-//     function to optionally do some verification before returning the EE Thread object
-//     for the current thread. Currently the primary enforcement this function does is
-//     around the EE_THREAD_(NOT)_REQUIRED contracts. For a definition of these
-//     contracts, how they're used, and how temporary "safe" scopes may be created
-//     using BEGIN_GETTHREAD_ALLOWED / END_GETTHREAD_ALLOWED, see the comments at the top
-//     of contract.h.
-//     
-//     The EE_THREAD_(NOT)_REQUIRED contracts are enforced as follows:
-//         * code:EEContract::DoChecks enforces the following:
-//             * On entry to an EE_THREAD_REQUIRED function, GetThread() != NULL
-//             * An EE_THREAD_REQUIRED function may not be called from an
-//                 EE_THREAD_NOT_REQUIRED function, unless there is an intervening
-//                 BEGIN/END_GETTHREAD_ALLOWED scope
-//         * This function (GetThreadGenericFullCheck) enforces that an
-//             EE_THREAD_NOT_REQUIRED function may not call GetThread(), unless there is
-//             an intervening BEGIN/END_GETTHREAD_ALLOWED scope. While this enforcement
-//             is straightforward below, the tricky part is getting
-//             GetThreadGenericFullCheck() to actually be called when GetThread() is
-//             called, given the optimizations around GetThread():
-//             * code:InitThreadManager ensures that non-PAL, debug, x86/x64 builds that
-//                 run with COMPlus_EnforceEEThreadNotRequiredContracts set are forced to
-//                 use GetThreadGeneric instead of the dynamically generated optimized
-//                 TLS getter.
-//             * The non-PAL, debug, x86/x64 GetThreadGeneric() (implemented in the
-//                 processor-specific assembly files) knows to call
-//                 GetThreadGenericFullCheck() to do the enforcement.
-//    
-Thread * GetThreadGenericFullCheck()
-{
-    // Can not have a dynamic contract here.  Contract depends on GetThreadGeneric.
-    // Contract here causes stack overflow.
-    STATIC_CONTRACT_NOTHROW;
-    STATIC_CONTRACT_GC_NOTRIGGER;
-
-    if (!ThreadInited())
-    {
-        // #GTInfiniteRecursion
-        // 
-        // Normally, we'd want to assert here, but that could lead to infinite recursion.
-        // Bringing up the assert dialog requires a string lookup, which requires getting
-        // the Thread's UI culture ID, which, or course, requires getting the Thread. So
-        // we'll just break instead.
-        DebugBreak();
-    }
-
-    if (g_fEEStarted && 
-
-        // Using ShouldEnforceEEThreadNotRequiredContracts() instead
-        // of directly checking CLRConfig::GetConfigValue, as the latter contains a dynamic
-        // contract and therefore calls GetThread(), which would cause infinite recursion.
-        Thread::ShouldEnforceEEThreadNotRequiredContracts() &&
-
-        // The following verifies that it's safe to call GetClrDebugState() below without
-        // risk of its callees invoking extra error checking or fiber code that could
-        // recursively call GetThread() and overflow the stack
-        (CExecutionEngine::GetTlsData() != NULL))
-    {
-        // It's safe to peek into the debug state, so let's do so, to see if
-        // our caller is really allowed to be calling GetThread(). This enforces
-        // the EE_THREAD_NOT_REQUIRED contract.
-        ClrDebugState * pDbg = GetClrDebugState(FALSE);      // FALSE=don't allocate
-        if ((pDbg != NULL) && (!pDbg->IsGetThreadAllowed()))
-        {
-            // We need to bracket the ASSERTE with BEGIN/END_GETTHREAD_ALLOWED to avoid
-            // infinite recursion (see
-            // code:GetThreadGenericFullCheck#GTInfiniteRecursion). The ASSERTE here will
-            // cause us to reenter this function to get the thread (again). However,
-            // BEGIN/END_GETTHREAD_ALLOWED at least stops the recursion right then and
-            // there, as it prevents us from reentering this block yet again (since
-            // BEGIN/END_GETTHREAD_ALLOWED causes pDbg->IsGetThreadAllowed() to be TRUE).
-            // All such reentries to this function will quickly return the thread without
-            // executing the code below, so the original ASSERTE can proceed.
-            BEGIN_GETTHREAD_ALLOWED;
-            _ASSERTE(!"GetThread() called in a EE_THREAD_NOT_REQUIRED scope.  If the GetThread() call site has a clear code path for a return of NULL, then consider using GetThreadNULLOk() or BEGIN/END_GETTHREAD_ALLOWED");
-            END_GETTHREAD_ALLOWED;
-        }
-    }
-
-    Thread * pThread = (Thread *) UnsafeTlsGetValue(gThreadTLSIndex);
-
-    // set bogus last error to help find places that fail to save it across GetThread calls
-    ::SetLastError(LAST_ERROR_TRASH_VALUE);
-
-    return pThread;
-}
-
-#endif // ENABLE_GET_THREAD_GENERIC_FULL_CHECK
-
-#if defined(_TARGET_X86_) || defined(_TARGET_AMD64_)
-//
-// Some platforms have this implemented in assembly
-//
-EXTERN_C Thread* STDCALL GetThreadGeneric(VOID);
-EXTERN_C AppDomain* STDCALL GetAppDomainGeneric(VOID);
-#else
-Thread* STDCALL GetThreadGeneric()
-{
-    // Can not have contract here.  Contract depends on GetThreadGeneric.
-    // Contract here causes stack overflow.
-    //CONTRACTL {
-    //    NOTHROW;
-    //    GC_NOTRIGGER;
-    //}
-    //CONTRACTL_END;
-
-    // see code:GetThreadGenericFullCheck#GTInfiniteRecursion
-    _ASSERTE(ThreadInited());
-
-    Thread* pThread = (Thread*)UnsafeTlsGetValue(gThreadTLSIndex);
-
-    TRASH_LASTERROR;
-
-    return pThread;
-}
-
-AppDomain* STDCALL GetAppDomainGeneric()
-{
-    // No contract.  This function is called during ExitTask.
-    //CONTRACTL {
-    //    NOTHROW;
-    //    GC_NOTRIGGER;
-    //}
-    //CONTRACTL_END;
-
-    _ASSERTE(ThreadInited());
-
-    AppDomain* pAppDomain = (AppDomain*)UnsafeTlsGetValue(GetAppDomainTLSIndex());
-
-    TRASH_LASTERROR;
-
-    return pAppDomain;
-}
-#endif // defined(_TARGET_X86_) || defined(_TARGET_AMD64_)
-
-//
-// FLS getter to avoid unnecessary indirection via execution engine. It will be used if we get high TLS slot
-// from the OS where we cannot use the fast optimized assembly helpers. (It happens pretty often in hosted scenarios).
-//
-LPVOID* ClrFlsGetBlockDirect()
-{
-    LIMITED_METHOD_CONTRACT;
-
-    return (LPVOID*)UnsafeTlsGetValue(CExecutionEngine::GetTlsIndex());
-}
-
-extern "C" void * ClrFlsGetBlock();
-
-#endif // FEATURE_IMPLICIT_TLS
-
-
 extern "C" void STDCALL JIT_PatchedCodeStart();
 extern "C" void STDCALL JIT_PatchedCodeLast();
 
@@ -1387,98 +1125,24 @@ void InitThreadManager()
 
 #ifndef FEATURE_PAL
 
-#ifdef FEATURE_IMPLICIT_TLS
     _ASSERTE(GetThread() == NULL);
 
-    // Mscordbi calculates the address of currentThread pointer using OFFSETOF__TLS__tls_CurrentThread. Ensure that
-    // value is correct.
+    PTEB Teb = NtCurrentTeb();
+    BYTE** tlsArray = (BYTE**)Teb->ThreadLocalStoragePointer;
+    BYTE* tlsData = (BYTE*)tlsArray[_tls_index];
 
-    PTEB Teb;
-    BYTE* tlsData;
-    BYTE** tlsArray;
+    size_t offsetOfCurrentThreadInfo = (BYTE*)&gCurrentThreadInfo - tlsData;
 
-    Teb = NtCurrentTeb();
-    tlsArray = (BYTE**)Teb->ThreadLocalStoragePointer;
-    tlsData = (BYTE*)tlsArray[_tls_index];
+    _ASSERTE(offsetOfCurrentThreadInfo < 0x8000);
+    _ASSERTE(_tls_index < 0x10000);
 
-    Thread **ppThread = (Thread**) (tlsData + OFFSETOF__TLS__tls_CurrentThread);
-    _ASSERTE_ALL_BUILDS("clr/src/VM/Threads.cpp",
-                        (&(gCurrentThreadInfo.m_pThread) == ppThread) &&
-                        "Offset of m_pThread as specified by OFFSETOF__TLS__tls_CurrentThread is not correct. "
-                        "This can change due to addition/removal of declspec(Thread) thread local variables.");
+    // Save gCurrentThreadInfo location for debugger
+    g_TlsIndex = (DWORD)(_tls_index + (offsetOfCurrentThreadInfo << 16) + 0x80000000);
 
-   _ASSERTE_ALL_BUILDS("clr/src/VM/Threads.cpp",
-                       ((BYTE*)&(gCurrentThreadInfo.m_EETlsData) == tlsData + OFFSETOF__TLS__tls_EETlsData) &&
-                       "Offset of m_EETlsData as specified by OFFSETOF__TLS__tls_EETlsData is not correct. "
-                       "This can change due to addition/removal of declspec(Thread) thread local variables.");
-#else
-    _ASSERTE(gThreadTLSIndex == TLS_OUT_OF_INDEXES);
-#endif
     _ASSERTE(g_TrapReturningThreads == 0);
 #endif // !FEATURE_PAL
 
-    // Consult run-time switches that choose whether to use generic or optimized
-    // versions of GetThread and GetAppDomain
-
-    BOOL fUseGenericTlsGetters = FALSE;
-
-#ifdef ENABLE_GET_THREAD_GENERIC_FULL_CHECK
-    // Debug builds allow user to throw a switch to force use of the generic GetThread
-    // for the sole purpose of enforcing EE_THREAD_NOT_REQUIRED contracts
-    if (CLRConfig::GetConfigValue(CLRConfig::INTERNAL_EnforceEEThreadNotRequiredContracts) != 0)
-    {
-        // Set this static on Thread so this value can be safely read later on by
-        // code:GetThreadGenericFullCheck
-        Thread::s_fEnforceEEThreadNotRequiredContracts = TRUE;
-
-        fUseGenericTlsGetters = TRUE;
-    }
-#endif
-
-#ifndef FEATURE_IMPLICIT_TLS
-    // Now, we setup GetThread and GetAppDomain to point to their optimized or generic versions. Irrespective
-    // of the version they call into, we write opcode sequence into the dummy GetThread/GetAppDomain
-    // implementations (living in jithelp.s/.asm) via the MakeOptimizedTlsGetter calls below.
-    //
-    // For this to work, we must ensure that the dummy versions lie between the JIT_PatchedCodeStart
-    // and JIT_PatchedCodeLast code range (which lies in the .text section) so that when we change the protection
-    // above, we do so for GetThread and GetAppDomain as well.
-     
-    //---------------------------------------------------------------------------
-    // INITIALIZE GetThread
-    //---------------------------------------------------------------------------
-
-    // No backout necessary - part of the one time global initialization
-    gThreadTLSIndex = UnsafeTlsAlloc();
-    if (gThreadTLSIndex == TLS_OUT_OF_INDEXES)
-        COMPlusThrowWin32();
-
-    MakeOptimizedTlsGetter(gThreadTLSIndex, (PVOID)GetThread, TLS_GETTER_MAX_SIZE, (POPTIMIZEDTLSGETTER)GetThreadGeneric, fUseGenericTlsGetters);
-
-    //---------------------------------------------------------------------------
-    // INITIALIZE GetAppDomain
-    //---------------------------------------------------------------------------
-
-    // No backout necessary - part of the one time global initialization
-    gAppDomainTLSIndex = UnsafeTlsAlloc();
-    if (gAppDomainTLSIndex == TLS_OUT_OF_INDEXES)
-        COMPlusThrowWin32();
-
-    MakeOptimizedTlsGetter(gAppDomainTLSIndex, (PVOID)GetAppDomain, TLS_GETTER_MAX_SIZE, (POPTIMIZEDTLSGETTER)GetAppDomainGeneric, fUseGenericTlsGetters);
-
-    //---------------------------------------------------------------------------
-    // Switch general purpose TLS getter to more efficient one if possible
-    //---------------------------------------------------------------------------
-
-    // Make sure that the TLS index is allocated
-    CExecutionEngine::CheckThreadState(0, FALSE);
-
-    DWORD masterSlotIndex = CExecutionEngine::GetTlsIndex();
-    CLRFLSGETBLOCK pGetter = (CLRFLSGETBLOCK)MakeOptimizedTlsGetter(masterSlotIndex, (PVOID)ClrFlsGetBlock, TLS_GETTER_MAX_SIZE);
-    __ClrFlsGetBlock = pGetter ? pGetter : ClrFlsGetBlockDirect;
-#else
     __ClrFlsGetBlock = CExecutionEngine::GetTlsData;
-#endif // FEATURE_IMPLICIT_TLS
 
     IfFailThrow(Thread::CLRSetThreadStackGuarantee(Thread::STSGuarantee_Force));
 
diff --git a/src/vm/threads.h b/src/vm/threads.h
index 17cc1f305d..05e01b3004 100644
--- a/src/vm/threads.h
+++ b/src/vm/threads.h
@@ -586,8 +586,6 @@ enum ThreadpoolThreadType
 //
 // Public functions for ASM code generators
 //
-//      int GetThreadTLSIndex()         - returns TLS index used to point to Thread
-//      int GetAppDomainTLSIndex()      - returns TLS index used to point to AppDomain
 //      Thread* __stdcall CreateThreadBlockThrow() - creates new Thread on reverse p-invoke
 //
 // Public functions for one-time init/cleanup
@@ -629,14 +627,6 @@ Thread* SetupThreadNoThrow(HRESULT *phresult = NULL);
 Thread* SetupUnstartedThread(BOOL bRequiresTSL=TRUE);
 void    DestroyThread(Thread *th);
 
-
-//---------------------------------------------------------------------------
-//---------------------------------------------------------------------------
-#ifndef FEATURE_IMPLICIT_TLS
-DWORD GetThreadTLSIndex();
-DWORD GetAppDomainTLSIndex();
-#endif
-
 DWORD GetRuntimeId();
 
 EXTERN_C Thread* WINAPI CreateThreadBlockThrow();
@@ -3891,23 +3881,6 @@ private:
     ULONG  m_ulEnablePreemptiveGCCount;
 #endif  // _DEBUG
 
-#ifdef ENABLE_GET_THREAD_GENERIC_FULL_CHECK
-
-private:
-    // Set once on initialization, single-threaded, inside friend code:InitThreadManager,
-    // based on whether the user has set COMPlus_EnforceEEThreadNotRequiredContracts.
-    // This is then later accessed via public
-    // code:Thread::ShouldEnforceEEThreadNotRequiredContracts. See
-    // code:GetThreadGenericFullCheck for details.
-    static BOOL s_fEnforceEEThreadNotRequiredContracts;
-
-public:
-    static BOOL ShouldEnforceEEThreadNotRequiredContracts();
-
-#endif  // ENABLE_GET_THREAD_GENERIC_FULL_CHECK
-
-
-
 private:
     // For suspends:
     CLREvent        m_DebugSuspendEvent;
@@ -7386,7 +7359,6 @@ inline void SetTypeHandleOnThreadForAlloc(TypeHandle th)
 
 #endif // CROSSGEN_COMPILE
 
-#ifdef FEATURE_IMPLICIT_TLS
 class Compiler;
 // users of OFFSETOF__TLS__tls_CurrentThread macro expect the offset of these variables wrt to _tls_start to be stable. 
 // Defining each of the following thread local variable separately without the struct causes the offsets to change in 
@@ -7398,11 +7370,7 @@ struct ThreadLocalInfo
     Thread* m_pThread;
     AppDomain* m_pAppDomain;
     void** m_EETlsData; // ClrTlsInfo::data
-#ifdef FEATURE_MERGE_JIT_AND_ENGINE
-    void* m_pJitTls;
-#endif
 };
-#endif // FEATURE_IMPLICIT_TLS
 
 class ThreadStateHolder
 {
diff --git a/src/vm/threads.inl b/src/vm/threads.inl
index ee2aaacf94..f5a439c350 100644
--- a/src/vm/threads.inl
+++ b/src/vm/threads.inl
@@ -22,7 +22,6 @@
 #include "frames.h"
 
 #ifndef DACCESS_COMPILE
-#ifdef FEATURE_IMPLICIT_TLS
 
 #ifndef __llvm__
 EXTERN_C __declspec(thread) ThreadLocalInfo gCurrentThreadInfo;
@@ -40,18 +39,8 @@ EXTERN_C inline AppDomain* STDCALL GetAppDomain()
     return gCurrentThreadInfo.m_pAppDomain;
 }
 
-#endif // FEATURE_IMPLICIT_TLS
 #endif // !DACCESS_COMPILE
 
-#ifdef ENABLE_GET_THREAD_GENERIC_FULL_CHECK
-// See code:GetThreadGenericFullCheck
-inline /* static */ BOOL Thread::ShouldEnforceEEThreadNotRequiredContracts()
-{
-    LIMITED_METHOD_CONTRACT;
-    return s_fEnforceEEThreadNotRequiredContracts;
-}
-#endif // ENABLE_GET_THREAD_GENERIC_FULL_CHECK
-
 inline void Thread::IncLockCount()
 {
     LIMITED_METHOD_CONTRACT;
diff --git a/src/vm/vars.cpp b/src/vm/vars.cpp
index ff941d2101..464560c3a6 100644
--- a/src/vm/vars.cpp
+++ b/src/vm/vars.cpp
@@ -120,13 +120,10 @@ GPTR_IMPL_INIT(StressLog, g_pStressLog, &StressLog::theLog);
 GPTR_IMPL(RCWCleanupList,g_pRCWCleanupList);
 #endif // FEATURE_COMINTEROP
 
+GVAL_IMPL_INIT(DWORD, g_TlsIndex, TLS_OUT_OF_INDEXES);
 
 #ifndef DACCESS_COMPILE
 
-// <TODO> @TODO Remove eventually - </TODO> determines whether the verifier throws an exception when something fails
-bool                g_fVerifierOff;
-
-
 // <TODO> @TODO - PROMOTE. </TODO>
 OBJECTHANDLE         g_pPreallocatedOutOfMemoryException;
 OBJECTHANDLE         g_pPreallocatedStackOverflowException;
diff --git a/src/vm/vars.hpp b/src/vm/vars.hpp
index c9f4848692..a7b6fb6173 100644
--- a/src/vm/vars.hpp
+++ b/src/vm/vars.hpp
@@ -406,9 +406,7 @@ GPTR_DECL(MethodDesc,       g_pExecuteBackoutCodeHelperMethod);
 
 GPTR_DECL(MethodDesc,       g_pObjectFinalizerMD);
 
-//<TODO> @TODO Remove eventually - determines whether the verifier throws an exception when something fails</TODO>
-EXTERN bool                 g_fVerifierOff;
-
+GVAL_DECL(DWORD,            g_TlsIndex);
 
 // Global System Information
 extern SYSTEM_INFO g_SystemInfo;