Use 16 bytes to spill SIMD12 (#19237)

On 64-bit systems, SIMD12 locals are naturally rounded to 16 bytes, but the spill temp logic was not doing the same.

Fix #19197

2 files changed, 6 insertions, 4 deletions

diff --git a/src/jit/compiler.h b/src/jit/compiler.h
index e5bd4436a2..ee76c2810d 100644
--- a/src/jit/compiler.h
+++ b/src/jit/compiler.h
@@ -660,6 +660,7 @@ public:
         // For 32-bit architectures, we make local variable SIMD12 types 16 bytes instead of just 12. We can't do
         // this for arguments, which must be passed according the defined ABI. We don't want to do this for
         // dependently promoted struct fields, but we don't know that here. See lvaMapSimd12ToSimd16().
+        // (Note that for 64-bits, we are already rounding up to 16.)
         if ((lvType == TYP_SIMD12) && !lvIsParam)
         {
             assert(lvExactSize == 12);
diff --git a/src/jit/regset.cpp b/src/jit/regset.cpp
index 81887cc94e..8e41edfb0b 100644
--- a/src/jit/regset.cpp
+++ b/src/jit/regset.cpp
@@ -619,10 +619,11 @@ var_types RegSet::tmpNormalizeType(var_types type)
 {
     type = genActualType(type);
 
-#if defined(FEATURE_SIMD) && !defined(_TARGET_64BIT_)
-    // For SIMD on 32-bit platforms, we always spill SIMD12 to a 16-byte SIMD16 temp.
-    // This is because we don't have a single instruction to store 12 bytes. We also
-    // allocate non-argument locals as 16 bytes; see lvSize().
+#if defined(FEATURE_SIMD)
+    // We always spill SIMD12 to a 16-byte SIMD16 temp.
+    // This is because we don't have a single instruction to store 12 bytes, so we want
+    // to ensure that we always have the full 16 bytes for loading & storing the value.
+    // We also allocate non-argument locals as 16 bytes; see lvSize().
     if (type == TYP_SIMD12)
     {
         type = TYP_SIMD16;