[release/8.0] Make CoreCLR/NativeAOT assembly compile with .subsections_via_symbols on Apple platforms (#92544)

* Make CoreCLR/NativeAOT assembly compile with .subsections_via_symbols on Apple platforms

* Fix build with LSE_INSTRUCTIONS_ENABLED_BY_DEFAULT

---------

Co-authored-by: Filip Navara <navara@emclient.com>
Co-authored-by: Jan Kotas <jkotas@microsoft.com>

Author: 3 people

src/coreclr/nativeaot/Runtime/arm64/AllocFast.S

@@ -46,7 +46,7 @@ OFFSETOF__Thread__m_alloc_context__alloc_limit      = OFFSETOF__Thread__m_rgbAll
         add         x2, x2, x12
         ldr         x13, [x1, #OFFSETOF__Thread__m_alloc_context__alloc_limit]
         cmp         x2, x13
-        bhi         RhpNewFast_RarePath
+        bhi         LOCAL_LABEL(RhpNewFast_RarePath)
 
         // Update the alloc pointer to account for the allocation.
         str         x2, [x1, #OFFSETOF__Thread__m_alloc_context__alloc_ptr]
@@ -57,7 +57,7 @@ OFFSETOF__Thread__m_alloc_context__alloc_limit      = OFFSETOF__Thread__m_rgbAll
         mov         x0, x12
         ret
 
-RhpNewFast_RarePath:
+LOCAL_LABEL(RhpNewFast_RarePath):
         mov         x1, #0
         b           C_FUNC(RhpNewObject)
     LEAF_END RhpNewFast, _TEXT
@@ -88,12 +88,12 @@ RhpNewFast_RarePath:
         bl          C_FUNC(RhpGcAlloc)
 
         // Set the new objects MethodTable pointer on success.
-        cbz         x0, NewOutOfMemory
+        cbz         x0, LOCAL_LABEL(NewOutOfMemory)
 
         POP_COOP_PINVOKE_FRAME
         EPILOG_RETURN
 
-NewOutOfMemory:
+LOCAL_LABEL(NewOutOfMemory):
         // This is the OOM failure path. We are going to tail-call to a managed helper that will throw
         // an out of memory exception that the caller of this allocator understands.
 
@@ -113,7 +113,7 @@ NewOutOfMemory:
         movz        x2, MAX_STRING_LENGTH & 0xFFFF
         movk        x2, MAX_STRING_LENGTH >> 16, lsl 16
         cmp         x1, x2
-        bhi         StringSizeOverflow
+        bhi         LOCAL_LABEL(StringSizeOverflow)
 
         // Compute overall allocation size (align(base size + (element size * elements), 8)).
         mov         w2, #STRING_COMPONENT_SIZE
@@ -139,7 +139,7 @@ NewOutOfMemory:
         add         x2, x2, x12
         ldr         x12, [x3, #OFFSETOF__Thread__m_alloc_context__alloc_limit]
         cmp         x2, x12
-        bhi         C_FUNC(RhpNewArrayRare)
+        bhi         LOCAL_LABEL(RhNewString_Rare)
 
         // Reload new object address into r12.
         ldr         x12, [x3, #OFFSETOF__Thread__m_alloc_context__alloc_ptr]
@@ -156,14 +156,17 @@ NewOutOfMemory:
 
         ret
 
-StringSizeOverflow:
+LOCAL_LABEL(StringSizeOverflow):
         // We get here if the length of the final string object can not be represented as an unsigned
         // 32-bit value. We are going to tail-call to a managed helper that will throw
         // an OOM exception that the caller of this allocator understands.
 
         // x0 holds MethodTable pointer already
         mov         x1, #1                  // Indicate that we should throw OverflowException
         b           C_FUNC(RhExceptionHandling_FailedAllocation)
+
+LOCAL_LABEL(RhNewString_Rare):
+        b           C_FUNC(RhpNewArrayRare)
     LEAF_END    RhNewString, _Text
 
 // Allocate one dimensional, zero based array (SZARRAY).
@@ -177,7 +180,7 @@ StringSizeOverflow:
         // case (32 dimensional MdArray) is less than 0xffff, and thus the product fits in 64 bits.
         mov         x2, #0x7FFFFFFF
         cmp         x1, x2
-        bhi         ArraySizeOverflow
+        bhi         LOCAL_LABEL(ArraySizeOverflow)
 
         ldrh        w2, [x0, #OFFSETOF__MethodTable__m_usComponentSize]
         umull       x2, w1, w2
@@ -204,7 +207,7 @@ StringSizeOverflow:
         add         x2, x2, x12
         ldr         x12, [x3, #OFFSETOF__Thread__m_alloc_context__alloc_limit]
         cmp         x2, x12
-        bhi         C_FUNC(RhpNewArrayRare)
+        bhi         LOCAL_LABEL(RhpNewArray_Rare)
 
         // Reload new object address into x12.
         ldr         x12, [x3, #OFFSETOF__Thread__m_alloc_context__alloc_ptr]
@@ -221,14 +224,17 @@ StringSizeOverflow:
 
         ret
 
-ArraySizeOverflow:
+LOCAL_LABEL(ArraySizeOverflow):
         // We get here if the size of the final array object can not be represented as an unsigned
         // 32-bit value. We are going to tail-call to a managed helper that will throw
         // an overflow exception that the caller of this allocator understands.
 
         // x0 holds MethodTable pointer already
         mov         x1, #1                  // Indicate that we should throw OverflowException
         b           C_FUNC(RhExceptionHandling_FailedAllocation)
+
+LOCAL_LABEL(RhpNewArray_Rare):
+        b           C_FUNC(RhpNewArrayRare)
     LEAF_END    RhpNewArray, _TEXT
 
 // Allocate one dimensional, zero based array (SZARRAY) using the slow path that calls a runtime helper.
@@ -254,12 +260,12 @@ ArraySizeOverflow:
         bl          C_FUNC(RhpGcAlloc)
 
         // Set the new objects MethodTable pointer and length on success.
-        cbz         x0, ArrayOutOfMemory
+        cbz         x0, LOCAL_LABEL(ArrayOutOfMemory)
 
         POP_COOP_PINVOKE_FRAME
         EPILOG_RETURN
 
-ArrayOutOfMemory:
+LOCAL_LABEL(ArrayOutOfMemory):
         // This is the OOM failure path. We are going to tail-call to a managed helper that will throw
         // an out of memory exception that the caller of this allocator understands.
 

src/coreclr/nativeaot/Runtime/arm64/ExceptionHandling.S

@@ -275,7 +275,7 @@
         // where the tail-calling thread had saved LR, which may not match where we have saved LR.
 
         ldr         x1, [x2, #OFFSETOF__Thread__m_pvHijackedReturnAddress]
-        cbz         x1, NotHijacked
+        cbz         x1, LOCAL_LABEL(NotHijacked)
 
         ldr         x3, [x2, #OFFSETOF__Thread__m_ppvHijackedReturnAddressLocation]
 
@@ -286,13 +286,13 @@
 
         add         x12, sp, #(STACKSIZEOF_ExInfo + SIZEOF__PAL_LIMITED_CONTEXT)        // re-compute SP at callsite
         cmp         x3, x12             // if (m_ppvHijackedReturnAddressLocation < SP at callsite)
-        blo         TailCallWasHijacked
+        blo         LOCAL_LABEL(TailCallWasHijacked)
 
         // normal case where a valid return address location is hijacked
         str         x1, [x3]
-        b           ClearThreadState
+        b           LOCAL_LABEL(ClearThreadState)
 
-TailCallWasHijacked:
+LOCAL_LABEL(TailCallWasHijacked):
 
         // Abnormal case where the return address location is now invalid because we ended up here via a tail
         // call.  In this case, our hijacked return address should be the correct caller of this method.
@@ -302,13 +302,13 @@ TailCallWasHijacked:
         str         lr, [sp, #(rsp_offsetof_Context + OFFSETOF__PAL_LIMITED_CONTEXT__LR)]
         str         lr, [sp, #(rsp_offsetof_Context + OFFSETOF__PAL_LIMITED_CONTEXT__IP)]
 
-ClearThreadState:
+LOCAL_LABEL(ClearThreadState):
 
         // clear the Thread's hijack state
         str         xzr, [x2, #OFFSETOF__Thread__m_ppvHijackedReturnAddressLocation]
         str         xzr, [x2, #OFFSETOF__Thread__m_pvHijackedReturnAddress]
 
-NotHijacked:
+LOCAL_LABEL(NotHijacked):
 
         add         x1, sp, #rsp_offsetof_ExInfo                    // x1 <- ExInfo*
         str         xzr, [x1, #OFFSETOF__ExInfo__m_exception]       // pExInfo->m_exception = null
@@ -429,13 +429,13 @@ NotHijacked:
 
         add         x12, x5, #OFFSETOF__Thread__m_ThreadStateFlags
 
-ClearRetry_Catch:
+LOCAL_LABEL(ClearRetry_Catch):
         ldxr        w4, [x12]
         bic         w4, w4, #TSF_DoNotTriggerGc
         stxr        w6, w4, [x12]
-        cbz         w6, ClearSuccess_Catch
-        b           ClearRetry_Catch
-ClearSuccess_Catch:
+        cbz         w6, LOCAL_LABEL(ClearSuccess_Catch)
+        b           LOCAL_LABEL(ClearRetry_Catch)
+LOCAL_LABEL(ClearSuccess_Catch):
 
         //
         // set preserved regs to the values expected by the funclet
@@ -487,21 +487,21 @@ ClearSuccess_Catch:
         ldr         x3, [sp, #rsp_offset_x3]                    // x3 <- current ExInfo*
         ldr         x2, [x2, #OFFSETOF__REGDISPLAY__SP]         // x2 <- resume SP value
 
-PopExInfoLoop:
+LOCAL_LABEL(PopExInfoLoop):
         ldr         x3, [x3, #OFFSETOF__ExInfo__m_pPrevExInfo]  // x3 <- next ExInfo
-        cbz         x3, DonePopping                             // if (pExInfo == null) { we're done }
+        cbz         x3, LOCAL_LABEL(DonePopping)                // if (pExInfo == null) { we're done }
         cmp         x3, x2
-        blt         PopExInfoLoop                               // if (pExInfo < resume SP} { keep going }
+        blt         LOCAL_LABEL(PopExInfoLoop)                  // if (pExInfo < resume SP} { keep going }
 
-DonePopping:
+LOCAL_LABEL(DonePopping):
         str         x3, [x1, #OFFSETOF__Thread__m_pExInfoStackHead]     // store the new head on the Thread
 
         PREPARE_EXTERNAL_VAR_INDIRECT_W RhpTrapThreads, 3
 
-        tbz         x3, #TrapThreadsFlags_AbortInProgress_Bit, NoAbort
+        tbz         x3, #TrapThreadsFlags_AbortInProgress_Bit, LOCAL_LABEL(NoAbort)
 
         ldr         x3, [sp, #rsp_offset_is_not_handling_thread_abort]
-        cbnz        x3, NoAbort
+        cbnz        x3, LOCAL_LABEL(NoAbort)
 
         // It was the ThreadAbortException, so rethrow it
         // reset SP
@@ -510,7 +510,7 @@ DonePopping:
         mov         sp, x2
         b           C_FUNC(RhpThrowHwEx)
 
-NoAbort:
+LOCAL_LABEL(NoAbort):
         // reset SP and jump to continuation address
         mov         sp, x2
         br          x0
@@ -564,13 +564,13 @@ NoAbort:
 
         add         x12, x2, #OFFSETOF__Thread__m_ThreadStateFlags
 
-ClearRetry:
+LOCAL_LABEL(ClearRetry):
         ldxr        w4, [x12]
         bic         w4, w4, #TSF_DoNotTriggerGc
         stxr        w3, w4, [x12]
-        cbz         w3, ClearSuccess
-        b           ClearRetry
-ClearSuccess:
+        cbz         w3, LOCAL_LABEL(ClearSuccess)
+        b           LOCAL_LABEL(ClearRetry)
+LOCAL_LABEL(ClearSuccess):
 
         //
         // set preserved regs to the values expected by the funclet
@@ -602,13 +602,13 @@ ClearSuccess:
         ldr x2, [sp, rsp_FinallyFunclet_offset_thread]
 
         add         x12, x2, #OFFSETOF__Thread__m_ThreadStateFlags
-SetRetry:
+LOCAL_LABEL(SetRetry):
         ldxr        w1, [x12]
         orr         w1, w1, #TSF_DoNotTriggerGc
         stxr        w3, w1, [x12]
-        cbz         w3, SetSuccess
-        b           SetRetry
-SetSuccess:
+        cbz         w3, LOCAL_LABEL(SetSuccess)
+        b           LOCAL_LABEL(SetRetry)
+LOCAL_LABEL(SetSuccess):
 
         ldp         d8, d9,   [sp, #0x00]
         ldp         d10, d11, [sp, #0x10]
@@ -707,13 +707,13 @@ SetSuccess:
 
         add         x12, x5, #OFFSETOF__Thread__m_ThreadStateFlags
 
-ClearRetry_Propagate:
+LOCAL_LABEL(ClearRetry_Propagate):
         ldxr        w4, [x12]
         bic         w4, w4, #TSF_DoNotTriggerGc
         stxr        w6, w4, [x12]
-        cbz         w6, ClearSuccess_Propagate
-        b           ClearRetry_Propagate
-ClearSuccess_Propagate:
+        cbz         w6, LOCAL_LABEL(ClearSuccess_Propagate)
+        b           LOCAL_LABEL(ClearRetry_Propagate)
+LOCAL_LABEL(ClearSuccess_Propagate):
 
         //
         // set preserved regs to the values expected by the funclet
@@ -749,13 +749,13 @@ ClearSuccess_Propagate:
         ldr         x3, [sp, #rsp_offset_x3]                    // x3 <- current ExInfo*
         ldr         x2, [x2, #OFFSETOF__REGDISPLAY__SP]         // x2 <- resume SP value
 
-Propagate_PopExInfoLoop:
+LOCAL_LABEL(Propagate_PopExInfoLoop):
         ldr         x3, [x3, #OFFSETOF__ExInfo__m_pPrevExInfo]  // x3 <- next ExInfo
-        cbz         x3, Propagate_DonePopping                   // if (pExInfo == null) { we're done }
+        cbz         x3, LOCAL_LABEL(Propagate_DonePopping)      // if (pExInfo == null) { we're done }
         cmp         x3, x2
-        blt         Propagate_PopExInfoLoop                     // if (pExInfo < resume SP} { keep going }
+        blt         LOCAL_LABEL(Propagate_PopExInfoLoop)        // if (pExInfo < resume SP} { keep going }
 
-Propagate_DonePopping:
+LOCAL_LABEL(Propagate_DonePopping):
         str         x3, [x1, #OFFSETOF__Thread__m_pExInfoStackHead]     // store the new head on the Thread
 
         // restore preemptive mode

src/coreclr/nativeaot/Runtime/arm64/GcProbe.S

@@ -127,10 +127,10 @@ NESTED_ENTRY RhpGcProbeHijack, _TEXT, NoHandler
     FixupHijackedCallstack
 
     PREPARE_EXTERNAL_VAR_INDIRECT_W RhpTrapThreads, 3
-    tbnz        x3, #TrapThreadsFlags_TrapThreads_Bit, WaitForGC
+    tbnz        x3, #TrapThreadsFlags_TrapThreads_Bit, LOCAL_LABEL(WaitForGC)
     ret
 
-WaitForGC:
+LOCAL_LABEL(WaitForGC):
     orr         x12, x12, DEFAULT_FRAME_SAVE_FLAGS + PTFF_SAVE_X0 + PTFF_SAVE_X1
     b           C_FUNC(RhpWaitForGC)
 NESTED_END RhpGcProbeHijack
@@ -144,11 +144,11 @@ NESTED_ENTRY RhpWaitForGC, _TEXT, NoHandler
     bl          C_FUNC(RhpWaitForGC2)
 
     ldr         x2, [sp, #OFFSETOF__PInvokeTransitionFrame__m_Flags]
-    tbnz        x2, #PTFF_THREAD_ABORT_BIT, ThrowThreadAbort
+    tbnz        x2, #PTFF_THREAD_ABORT_BIT, LOCAL_LABEL(ThrowThreadAbort)
 
     POP_PROBE_FRAME
     EPILOG_RETURN
-ThrowThreadAbort:
+LOCAL_LABEL(ThrowThreadAbort):
     POP_PROBE_FRAME
     mov w0, #STATUS_REDHAWK_THREAD_ABORT
     mov x1, lr // return address as exception PC
@@ -159,8 +159,10 @@ NESTED_END RhpWaitForGC
 
 LEAF_ENTRY RhpGcPoll
     PREPARE_EXTERNAL_VAR_INDIRECT_W RhpTrapThreads, 0
-    cbnz    w0, C_FUNC(RhpGcPollRare) // TrapThreadsFlags_None = 0
+    cbnz    w0, LOCAL_LABEL(RhpGcPoll_Rare) // TrapThreadsFlags_None = 0
     ret
+LOCAL_LABEL(RhpGcPoll_Rare):
+    b C_FUNC(RhpGcPollRare)
 LEAF_END RhpGcPoll
 
 NESTED_ENTRY RhpGcPollRare, _TEXT, NoHandler