Treat undef bytes as equal to any other byte

Author: oli-obk

compiler/rustc_codegen_gcc/src/common.rs

@@ -64,6 +64,11 @@ impl<'gcc, 'tcx> ConstCodegenMethods<'tcx> for CodegenCx<'gcc, 'tcx> {
         if type_is_pointer(typ) { self.context.new_null(typ) } else { self.const_int(typ, 0) }
     }
 
+    fn is_undef(&self, _val: RValue<'gcc>) -> bool {
+        // FIXME: actually check for undef
+        false
+    }
+
     fn const_undef(&self, typ: Type<'gcc>) -> RValue<'gcc> {
         let local = self.current_func.borrow().expect("func").new_local(None, typ, "undefined");
         if typ.is_struct().is_some() {

compiler/rustc_codegen_llvm/src/common.rs

@@ -126,6 +126,10 @@ impl<'ll, 'tcx> ConstCodegenMethods<'tcx> for CodegenCx<'ll, 'tcx> {
         unsafe { llvm::LLVMGetUndef(t) }
     }
 
+    fn is_undef(&self, v: &'ll Value) -> bool {
+        unsafe { llvm::LLVMIsUndef(v) == True }
+    }
+
     fn const_poison(&self, t: &'ll Type) -> &'ll Value {
         unsafe { llvm::LLVMGetPoison(t) }
     }

compiler/rustc_codegen_llvm/src/llvm/ffi.rs

@@ -917,6 +917,7 @@ unsafe extern "C" {
     pub fn LLVMMetadataTypeInContext(C: &Context) -> &Type;
 
     // Operations on all values
+    pub fn LLVMIsUndef(Val: &Value) -> Bool;
     pub fn LLVMTypeOf(Val: &Value) -> &Type;
     pub fn LLVMGetValueName2(Val: &Value, Length: *mut size_t) -> *const c_char;
     pub fn LLVMSetValueName2(Val: &Value, Name: *const c_char, NameLen: size_t);

compiler/rustc_codegen_ssa/src/mir/rvalue.rs

@@ -8,7 +8,7 @@ use rustc_middle::ty::{self, Instance, Ty, TyCtxt};
 use rustc_middle::{bug, mir, span_bug};
 use rustc_session::config::OptLevel;
 use rustc_span::{DUMMY_SP, Span};
-use tracing::{debug, instrument};
+use tracing::{debug, instrument, trace};
 
 use super::FunctionCx;
 use super::operand::{OperandRef, OperandValue};
@@ -93,6 +93,8 @@ impl<'a, 'tcx, Bx: BuilderMethods<'a, 'tcx>> FunctionCx<'a, 'tcx, Bx> {
                     return;
                 }
 
+                // If `v` is an integer constant whose value is just a single byte repeated N times,
+                // emit a `memset` filling the entire `dest` with that byte.
                 let try_init_all_same = |bx: &mut Bx, v| {
                     let start = dest.val.llval;
                     let size = bx.const_usize(dest.layout.size.bytes());
@@ -117,13 +119,33 @@ impl<'a, 'tcx, Bx: BuilderMethods<'a, 'tcx>> FunctionCx<'a, 'tcx, Bx> {
                     false
                 };
 
+                trace!(?cg_elem.val);
                 match cg_elem.val {
                     OperandValue::Immediate(v) => {
                         if try_init_all_same(bx, v) {
                             return;
                         }
                     }
-                    _ => (),
+                    OperandValue::Pair(a, b) => {
+                        let a_is_undef = bx.cx().is_undef(a);
+                        match (a_is_undef, bx.cx().is_undef(b)) {
+                            // Can happen for uninit unions
+                            (true, true) => {
+                                // FIXME: can we produce better output here?
+                            }
+                            (false, true) | (true, false) => {
+                                let val = if a_is_undef { b } else { a };
+                                if try_init_all_same(bx, val) {
+                                    return;
+                                }
+                            }
+                            (false, false) => {
+                                // FIXME: if both are the same value, use try_init_all_same
+                            }
+                        }
+                    }
+                    OperandValue::ZeroSized => unreachable!("checked above"),
+                    OperandValue::Ref(..) => {}
                 }
 
                 let count = self

compiler/rustc_codegen_ssa/src/traits/consts.rs

@@ -9,6 +9,7 @@ pub trait ConstCodegenMethods<'tcx>: BackendTypes {
     /// Generate an uninitialized value (matching uninitialized memory in MIR).
     /// Whether memory is initialized or not is tracked byte-for-byte.
     fn const_undef(&self, t: Self::Type) -> Self::Value;
+    fn is_undef(&self, v: Self::Value) -> bool;
     /// Generate a fake value. Poison always affects the entire value, even if just a single byte is
     /// poison. This can only be used in codepaths that are already UB, i.e., UB-free Rust code
     /// (including code that e.g. copies uninit memory with `MaybeUninit`) can never encounter a

tests/codegen/slice-init.rs

@@ -2,6 +2,8 @@
 
 #![crate_type = "lib"]
 
+use std::mem::MaybeUninit;
+
 // CHECK-LABEL: @zero_sized_elem
 #[no_mangle]
 pub fn zero_sized_elem() {
@@ -76,17 +78,64 @@ pub fn u16_init_one_bytes() -> [u16; N] {
     [const { u16::from_be_bytes([1, 1]) }; N]
 }
 
-// FIXME: undef bytes can just be initialized with the same value as the
-// defined bytes, if the defines bytes are all the same.
 // CHECK-LABEL: @option_none_init
 #[no_mangle]
 pub fn option_none_init() -> [Option<u8>; N] {
+    // CHECK-NOT: select
+    // CHECK-NOT: br
+    // CHECK-NOT: switch
+    // CHECK-NOT: icmp
+    // CHECK: call void @llvm.memset.p0
+    [None; N]
+}
+
+// If there is partial provenance or some bytes are initialized and some are not,
+// we can't really do better than initialize bytes or groups of bytes together.
+// CHECK-LABEL: @option_maybe_uninit_init
+#[no_mangle]
+pub fn option_maybe_uninit_init() -> [MaybeUninit<u16>; N] {
     // CHECK-NOT: select
     // CHECK: br label %repeat_loop_header{{.*}}
     // CHECK-NOT: switch
     // CHECK: icmp
     // CHECK-NOT: call void @llvm.memset.p0
-    [None; N]
+    [const {
+        let mut val: MaybeUninit<u16> = MaybeUninit::uninit();
+        let ptr = val.as_mut_ptr() as *mut u8;
+        unsafe {
+            ptr.write(0);
+        }
+        val
+    }; N]
+}
+
+#[repr(packed)]
+struct Packed {
+    start: u8,
+    ptr: &'static (),
+    rest: u16,
+    rest2: u8,
+}
+
+// If there is partial provenance or some bytes are initialized and some are not,
+// we can't really do better than initialize bytes or groups of bytes together.
+// CHECK-LABEL: @option_maybe_uninit_provenance
+#[no_mangle]
+pub fn option_maybe_uninit_provenance() -> [MaybeUninit<Packed>; N] {
+    // CHECK-NOT: select
+    // CHECK: br label %repeat_loop_header{{.*}}
+    // CHECK-NOT: switch
+    // CHECK: icmp
+    // CHECK-NOT: call void @llvm.memset.p0
+    [const {
+        let mut val: MaybeUninit<Packed> = MaybeUninit::uninit();
+        unsafe {
+            let ptr = &raw mut (*val.as_mut_ptr()).ptr;
+            static HAS_ADDR: () = ();
+            ptr.write_unaligned(&HAS_ADDR);
+        }
+        val
+    }; N]
 }
 
 // Use an opaque function to prevent rustc from removing useless drops.