Auto merge of #86416 - Amanieu:asm_clobber_only, r=nagisa

[rust.git] / compiler / rustc_codegen_llvm / src / asm.rs

diff --git a/compiler/rustc_codegen_llvm/src/asm.rs b/compiler/rustc_codegen_llvm/src/asm.rs
index ecf62ed213df82aec1daaf5b5b1cf958d0f9054c..7bd9397d64950d19eba45c753c9a9faa61d49294 100644 (file)
--- a/compiler/rustc_codegen_llvm/src/asm.rs
+++ b/compiler/rustc_codegen_llvm/src/asm.rs
@@ -128,6 +128,7 @@ fn codegen_inline_asm(
         let mut clobbers = vec![];
         let mut output_types = vec![];
         let mut op_idx = FxHashMap::default();
+        let mut clobbered_x87 = false;
         for (idx, op) in operands.iter().enumerate() {
             match *op {
                 InlineAsmOperandRef::Out { reg, late, place } => {
@@ -150,7 +151,27 @@ fn codegen_inline_asm(
                     let ty = if let Some(ref place) = place {
                         layout = Some(&place.layout);
                         llvm_fixup_output_type(self.cx, reg.reg_class(), &place.layout)
-                    } else if !is_target_supported(reg.reg_class()) {
+                    } else if matches!(
+                        reg.reg_class(),
+                        InlineAsmRegClass::X86(
+                            X86InlineAsmRegClass::mmx_reg | X86InlineAsmRegClass::x87_reg
+                        )
+                    ) {
+                        // Special handling for x87/mmx registers: we always
+                        // clobber the whole set if one register is marked as
+                        // clobbered. This is due to the way LLVM handles the
+                        // FP stack in inline assembly.
+                        if !clobbered_x87 {
+                            clobbered_x87 = true;
+                            clobbers.push("~{st}".to_string());
+                            for i in 1..=7 {
+                                clobbers.push(format!("~{{st({})}}", i));
+                            }
+                        }
+                        continue;
+                    } else if !is_target_supported(reg.reg_class())
+                        || reg.reg_class().is_clobber_only(asm_arch)
+                    {
                         // We turn discarded outputs into clobber constraints
                         // if the target feature needed by the register class is
                         // disabled. This is necessary otherwise LLVM will try
@@ -565,6 +586,9 @@ fn reg_to_llvm(reg: InlineAsmRegOrRegClass, layout: Option<&TyAndLayout<'tcx>>)
             InlineAsmRegClass::AArch64(AArch64InlineAsmRegClass::reg) => "r",
             InlineAsmRegClass::AArch64(AArch64InlineAsmRegClass::vreg) => "w",
             InlineAsmRegClass::AArch64(AArch64InlineAsmRegClass::vreg_low16) => "x",
+            InlineAsmRegClass::AArch64(AArch64InlineAsmRegClass::preg) => {
+                unreachable!("clobber-only")
+            }
             InlineAsmRegClass::Arm(ArmInlineAsmRegClass::reg) => "r",
             InlineAsmRegClass::Arm(ArmInlineAsmRegClass::reg_thumb) => "l",
             InlineAsmRegClass::Arm(ArmInlineAsmRegClass::sreg)
@@ -586,6 +610,9 @@ fn reg_to_llvm(reg: InlineAsmRegOrRegClass, layout: Option<&TyAndLayout<'tcx>>)
             InlineAsmRegClass::PowerPC(PowerPCInlineAsmRegClass::freg) => "f",
             InlineAsmRegClass::RiscV(RiscVInlineAsmRegClass::reg) => "r",
             InlineAsmRegClass::RiscV(RiscVInlineAsmRegClass::freg) => "f",
+            InlineAsmRegClass::RiscV(RiscVInlineAsmRegClass::vreg) => {
+                unreachable!("clobber-only")
+            }
             InlineAsmRegClass::X86(X86InlineAsmRegClass::reg) => "r",
             InlineAsmRegClass::X86(X86InlineAsmRegClass::reg_abcd) => "Q",
             InlineAsmRegClass::X86(X86InlineAsmRegClass::reg_byte) => "q",
@@ -593,6 +620,9 @@ fn reg_to_llvm(reg: InlineAsmRegOrRegClass, layout: Option<&TyAndLayout<'tcx>>)
             | InlineAsmRegClass::X86(X86InlineAsmRegClass::ymm_reg) => "x",
             InlineAsmRegClass::X86(X86InlineAsmRegClass::zmm_reg) => "v",
             InlineAsmRegClass::X86(X86InlineAsmRegClass::kreg) => "^Yk",
+            InlineAsmRegClass::X86(
+                X86InlineAsmRegClass::x87_reg | X86InlineAsmRegClass::mmx_reg,
+            ) => unreachable!("clobber-only"),
             InlineAsmRegClass::Wasm(WasmInlineAsmRegClass::local) => "r",
             InlineAsmRegClass::Bpf(BpfInlineAsmRegClass::reg) => "r",
             InlineAsmRegClass::Bpf(BpfInlineAsmRegClass::wreg) => "w",
@@ -617,6 +647,9 @@ fn modifier_to_llvm(
         | InlineAsmRegClass::AArch64(AArch64InlineAsmRegClass::vreg_low16) => {
             if modifier == Some('v') { None } else { modifier }
         }
+        InlineAsmRegClass::AArch64(AArch64InlineAsmRegClass::preg) => {
+            unreachable!("clobber-only")
+        }
         InlineAsmRegClass::Arm(ArmInlineAsmRegClass::reg)
         | InlineAsmRegClass::Arm(ArmInlineAsmRegClass::reg_thumb) => None,
         InlineAsmRegClass::Arm(ArmInlineAsmRegClass::sreg)
@@ -639,6 +672,9 @@ fn modifier_to_llvm(
         InlineAsmRegClass::PowerPC(_) => None,
         InlineAsmRegClass::RiscV(RiscVInlineAsmRegClass::reg)
         | InlineAsmRegClass::RiscV(RiscVInlineAsmRegClass::freg) => None,
+        InlineAsmRegClass::RiscV(RiscVInlineAsmRegClass::vreg) => {
+            unreachable!("clobber-only")
+        }
         InlineAsmRegClass::X86(X86InlineAsmRegClass::reg)
         | InlineAsmRegClass::X86(X86InlineAsmRegClass::reg_abcd) => match modifier {
             None if arch == InlineAsmArch::X86_64 => Some('q'),
@@ -663,6 +699,9 @@ fn modifier_to_llvm(
             _ => unreachable!(),
         },
         InlineAsmRegClass::X86(X86InlineAsmRegClass::kreg) => None,
+        InlineAsmRegClass::X86(X86InlineAsmRegClass::x87_reg | X86InlineAsmRegClass::mmx_reg) => {
+            unreachable!("clobber-only")
+        }
         InlineAsmRegClass::Wasm(WasmInlineAsmRegClass::local) => None,
         InlineAsmRegClass::Bpf(_) => None,
         InlineAsmRegClass::SpirV(SpirVInlineAsmRegClass::reg) => {
@@ -681,6 +720,9 @@ fn dummy_output_type(cx: &CodegenCx<'ll, 'tcx>, reg: InlineAsmRegClass) -> &'ll
         | InlineAsmRegClass::AArch64(AArch64InlineAsmRegClass::vreg_low16) => {
             cx.type_vector(cx.type_i64(), 2)
         }
+        InlineAsmRegClass::AArch64(AArch64InlineAsmRegClass::preg) => {
+            unreachable!("clobber-only")
+        }
         InlineAsmRegClass::Arm(ArmInlineAsmRegClass::reg)
         | InlineAsmRegClass::Arm(ArmInlineAsmRegClass::reg_thumb) => cx.type_i32(),
         InlineAsmRegClass::Arm(ArmInlineAsmRegClass::sreg)
@@ -704,6 +746,9 @@ fn dummy_output_type(cx: &CodegenCx<'ll, 'tcx>, reg: InlineAsmRegClass) -> &'ll
         InlineAsmRegClass::PowerPC(PowerPCInlineAsmRegClass::freg) => cx.type_f64(),
         InlineAsmRegClass::RiscV(RiscVInlineAsmRegClass::reg) => cx.type_i32(),
         InlineAsmRegClass::RiscV(RiscVInlineAsmRegClass::freg) => cx.type_f32(),
+        InlineAsmRegClass::RiscV(RiscVInlineAsmRegClass::vreg) => {
+            unreachable!("clobber-only")
+        }
         InlineAsmRegClass::X86(X86InlineAsmRegClass::reg)
         | InlineAsmRegClass::X86(X86InlineAsmRegClass::reg_abcd) => cx.type_i32(),
         InlineAsmRegClass::X86(X86InlineAsmRegClass::reg_byte) => cx.type_i8(),
@@ -711,6 +756,9 @@ fn dummy_output_type(cx: &CodegenCx<'ll, 'tcx>, reg: InlineAsmRegClass) -> &'ll
         | InlineAsmRegClass::X86(X86InlineAsmRegClass::ymm_reg)
         | InlineAsmRegClass::X86(X86InlineAsmRegClass::zmm_reg) => cx.type_f32(),
         InlineAsmRegClass::X86(X86InlineAsmRegClass::kreg) => cx.type_i16(),
+        InlineAsmRegClass::X86(X86InlineAsmRegClass::x87_reg | X86InlineAsmRegClass::mmx_reg) => {
+            unreachable!("clobber-only")
+        }
         InlineAsmRegClass::Wasm(WasmInlineAsmRegClass::local) => cx.type_i32(),
         InlineAsmRegClass::Bpf(BpfInlineAsmRegClass::reg) => cx.type_i64(),
         InlineAsmRegClass::Bpf(BpfInlineAsmRegClass::wreg) => cx.type_i32(),