[llvm-branch-commits] [llvm] [AArch64][SME] Handle zeroing ZA and ZT0 in functions with ZT0 state (PR #166361)

[Sander de Smalen via llvm-branch-commits]

================
@@ -356,20 +356,13 @@ define void @new_za_zt0_caller(ptr %callee) "aarch64_new_za" "aarch64_new_zt0" n
 
 ; Expect clear ZA on entry
 define void @new_za_shared_zt0_caller(ptr %callee) "aarch64_new_za" "aarch64_in_zt0" nounwind {
-; CHECK-LABEL: new_za_shared_zt0_caller:
-; CHECK:       // %bb.0:
-; CHECK-NEXT:    str x30, [sp, #-16]! // 8-byte Folded Spill
-; CHECK-NEXT:    zero {za}
-; CHECK-NEXT:    blr x0
-; CHECK-NEXT:    ldr x30, [sp], #16 // 8-byte Folded Reload
-; CHECK-NEXT:    ret
-;
-; CHECK-NEWLOWERING-LABEL: new_za_shared_zt0_caller:
-; CHECK-NEWLOWERING:       // %bb.0:
-; CHECK-NEWLOWERING-NEXT:    str x30, [sp, #-16]! // 8-byte Folded Spill
-; CHECK-NEWLOWERING-NEXT:    blr x0
----------------
sdesmalen-arm wrote:

Why wasn't ZA zeroed before?

https://github.com/llvm/llvm-project/pull/166361