[ARM64] Add g_GCShadowEnd to JIT_WriteBarrier_Table (#91342)

* [ARM64] Add g_GCShadowEnd to JIT_WriteBarrier_Table

This change moves address of g_GCShadowEnd to JIT_WriteBarrier_Table like
others variables used in Write Barrier.

This fix simmilar to RISC-V one https://github.com/dotnet/runtime/pull/90036

* [ARM64] Move GCShadow vars to the end of the wbs block

* Update src/coreclr/vm/arm64/asmhelpers.asm

---------
Co-authored-by: NJan Kotas <jkotas@microsoft.com>

src/coreclr/vm/arm64/asmhelpers.S

@@ -208,7 +208,7 @@ LEAF_END ThePreStubPatch, _TEXT
 WRITE_BARRIER_ENTRY JIT_UpdateWriteBarrierState
     PROLOG_SAVE_REG_PAIR_INDEXED   fp, lr, -16
 
-    // x0-x7 will contain intended new state
+    // x0-x7, x10 will contain intended new state
     // x8 will preserve skipEphemeralCheck
     // x12 will be used for pointers
 
@@ -223,33 +223,36 @@ WRITE_BARRIER_ENTRY JIT_UpdateWriteBarrierState
     ldr  x1, [x12]
 #endif
 
-#ifdef WRITE_BARRIER_CHECK
-    PREPARE_EXTERNAL_VAR g_GCShadow, x12
-    ldr  x2, [x12]
-#endif
-
 #ifdef FEATURE_USE_SOFTWARE_WRITE_WATCH_FOR_GC_HEAP
     PREPARE_EXTERNAL_VAR g_sw_ww_table, x12
-    ldr  x3, [x12]
+    ldr  x2, [x12]
 #endif
 
     PREPARE_EXTERNAL_VAR g_ephemeral_low, x12
-    ldr  x4, [x12]
+    ldr  x3, [x12]
 
     PREPARE_EXTERNAL_VAR g_ephemeral_high, x12
-    ldr  x5, [x12]
+    ldr  x4, [x12]
 
     cbz  x8, LOCAL_LABEL(EphemeralCheckEnabled)
-    movz x4, #0
-    movn x5, #0
+    movz x3, #0
+    movn x4, #0
 LOCAL_LABEL(EphemeralCheckEnabled):
 
     PREPARE_EXTERNAL_VAR g_lowest_address, x12
-    ldr  x6, [x12]
+    ldr  x5, [x12]
 
     PREPARE_EXTERNAL_VAR g_highest_address, x12
+    ldr  x6, [x12]
+
+#ifdef WRITE_BARRIER_CHECK
+    PREPARE_EXTERNAL_VAR g_GCShadow, x12
     ldr  x7, [x12]
 
+    PREPARE_EXTERNAL_VAR g_GCShadowEnd, x12
+    ldr  x10, [x12]
+#endif
+
     // Update wbs state
     PREPARE_EXTERNAL_VAR JIT_WriteBarrier_Table_Loc, x12
     ldr  x12, [x12]
@@ -258,7 +261,11 @@ LOCAL_LABEL(EphemeralCheckEnabled):
     stp  x0, x1, [x12], 16
     stp  x2, x3, [x12], 16
     stp  x4, x5, [x12], 16
-    stp  x6, x7, [x12], 16
+    str  x6, [x12], 8
+#ifdef WRITE_BARRIER_CHECK
+    stp  x7, x10, [x12], 16
+#endif
+
 
     EPILOG_RESTORE_REG_PAIR_INDEXED fp, lr, 16
     EPILOG_RETURN
@@ -367,8 +374,7 @@ WRITE_BARRIER_ENTRY JIT_WriteBarrier
     add  x12, x13, x12
 
     // if (pShadow >= g_GCShadowEnd) goto end
-    PREPARE_EXTERNAL_VAR g_GCShadowEnd, x13
-    ldr  x13, [x13]
+    ldr  x13, LOCAL_LABEL(wbs_GCShadowEnd)
     cmp  x12, x13
     bhs  LOCAL_LABEL(ShadowUpdateEnd)
 
@@ -460,8 +466,6 @@ LOCAL_LABEL(wbs_card_table):
     .quad 0
 LOCAL_LABEL(wbs_card_bundle_table):
     .quad 0
-LOCAL_LABEL(wbs_GCShadow):
-    .quad 0
 LOCAL_LABEL(wbs_sw_ww_table):
     .quad 0
 LOCAL_LABEL(wbs_ephemeral_low):
@@ -472,6 +476,12 @@ LOCAL_LABEL(wbs_lowest_address):
     .quad 0
 LOCAL_LABEL(wbs_highest_address):
     .quad 0
+#ifdef WRITE_BARRIER_CHECK
+LOCAL_LABEL(wbs_GCShadow):
+    .quad 0
+LOCAL_LABEL(wbs_GCShadowEnd):
+    .quad 0
+#endif
 WRITE_BARRIER_END JIT_WriteBarrier_Table
 
 

src/coreclr/vm/arm64/asmhelpers.asm

@@ -278,7 +278,7 @@ ThePreStubPatchLabel
     WRITE_BARRIER_ENTRY JIT_UpdateWriteBarrierState
         PROLOG_SAVE_REG_PAIR   fp, lr, #-16!
 
-        ; x0-x7 will contain intended new state
+        ; x0-x7, x10 will contain intended new state
         ; x8 will preserve skipEphemeralCheck
         ; x12 will be used for pointers
 
@@ -293,33 +293,36 @@ ThePreStubPatchLabel
         ldr      x1, [x12, g_card_bundle_table]
 #endif
 
-#ifdef WRITE_BARRIER_CHECK
-        adrp     x12, $g_GCShadow
-        ldr      x2, [x12, $g_GCShadow]
-#endif
-
 #ifdef FEATURE_USE_SOFTWARE_WRITE_WATCH_FOR_GC_HEAP
         adrp     x12, g_sw_ww_table
-        ldr      x3, [x12, g_sw_ww_table]
+        ldr      x2, [x12, g_sw_ww_table]
 #endif
 
         adrp     x12, g_ephemeral_low
-        ldr      x4, [x12, g_ephemeral_low]
+        ldr      x3, [x12, g_ephemeral_low]
 
         adrp     x12, g_ephemeral_high
-        ldr      x5, [x12, g_ephemeral_high]
+        ldr      x4, [x12, g_ephemeral_high]
 
         ; Check skipEphemeralCheck
         cbz      x8, EphemeralCheckEnabled
-        movz     x4, #0
-        movn     x5, #0
+        movz     x3, #0
+        movn     x4, #0
 
 EphemeralCheckEnabled
         adrp     x12, g_lowest_address
-        ldr      x6, [x12, g_lowest_address]
+        ldr      x5, [x12, g_lowest_address]
 
         adrp     x12, g_highest_address
-        ldr      x7, [x12, g_highest_address]
+        ldr      x6, [x12, g_highest_address]
+
+#ifdef WRITE_BARRIER_CHECK
+        adrp     x12, $g_GCShadow
+        ldr      x7, [x12, $g_GCShadow]
+
+        adrp     x12, $g_GCShadowEnd
+        ldr      x10, [x12, $g_GCShadowEnd]
+#endif
 
         ; Update wbs state
         adrp     x12, JIT_WriteBarrier_Table_Loc
@@ -328,7 +331,10 @@ EphemeralCheckEnabled
         stp      x0, x1, [x12], 16
         stp      x2, x3, [x12], 16
         stp      x4, x5, [x12], 16
-        stp      x6, x7, [x12], 16
+        str      x6, [x12], 8
+#ifdef WRITE_BARRIER_CHECK
+        stp     x7, x10, [x12], 16
+#endif
 
         EPILOG_RESTORE_REG_PAIR fp, lr, #16!
         EPILOG_RETURN
@@ -343,8 +349,6 @@ wbs_card_table
         DCQ 0
 wbs_card_bundle_table
         DCQ 0
-wbs_GCShadow
-        DCQ 0
 wbs_sw_ww_table
         DCQ 0
 wbs_ephemeral_low
@@ -355,6 +359,12 @@ wbs_lowest_address
         DCQ 0
 wbs_highest_address
         DCQ 0
+#ifdef WRITE_BARRIER_CHECK
+wbs_GCShadow
+        DCQ 0
+wbs_GCShadowEnd
+        DCQ 0
+#endif
     WRITE_BARRIER_END JIT_WriteBarrier_Table
 
 ; void JIT_ByRefWriteBarrier
@@ -433,8 +443,7 @@ NotInHeap
         add      x12, x13, x12
 
         ; if (pShadow >= $g_GCShadowEnd) goto end
-        adrp     x13, $g_GCShadowEnd
-        ldr      x13, [x13, $g_GCShadowEnd]
+        ldr      x13, wbs_GCShadowEnd
         cmp      x12, x13
         bhs      ShadowUpdateEnd