Auto merge of #155506 - jhpratt:rollup-tMDKAQS, r=jhpratt

Rollup of 9 pull requests

Successful merges:

 - #155370 (Add regression test for dead code elimination with drop + panic)
 - #154823 (Replace the spdx-rs dependency with a minimal in-tree SPDX tag-value parser)
 - #154972 (Implement `core::arch::return_address` and tests)
 - #155294 (Add test for coalescing of diagnostic attribute duplicates)
 - #155352 (triagebot.toml: Sync `assign.owners` with `autolabel."T-compiler"`)
 - #155431 (Add temporary scope to assert_matches)
 - #152995 (ACP Implementation of PermissionsExt for Windows )
 - #153873 (deprecate `std::char` constants and functions)
 - #154865 (libtest: use binary search for --exact test filtering)

Author: bors

Cargo.lock

@@ -370,6 +370,8 @@ fn mk_tests_slice(cx: &TestCtxt<'_>, sp: Span) -> Box<ast::Expr> {
     let ecx = &cx.ext_cx;
 
     let mut tests = cx.test_cases.clone();
+    // Note that this sort is load-bearing: the libtest harness uses binary search to find tests by
+    // name.
     tests.sort_by(|a, b| a.name.as_str().cmp(b.name.as_str()));
 
     ecx.expr_array_ref(

compiler/rustc_builtin_macros/src/test_harness.rs

@@ -1529,6 +1529,12 @@ fn codegen_regular_intrinsic_call<'tcx>(
             fx.bcx.set_cold_block(fx.bcx.current_block().unwrap());
         }
 
+        sym::return_address => {
+            let val = fx.bcx.ins().get_return_address(fx.pointer_type);
+            let val = CValue::by_val(val, ret.layout());
+            ret.write_cvalue(fx, val);
+        }
+
         // Unimplemented intrinsics must have a fallback body. The fallback body is obtained
         // by converting the `InstanceKind::Intrinsic` to an `InstanceKind::Item`.
         _ => {

compiler/rustc_codegen_cranelift/src/intrinsics/mod.rs

@@ -788,6 +788,12 @@ impl<'ll, 'tcx> IntrinsicCallBuilderMethods<'tcx> for Builder<'_, 'll, 'tcx> {
                 }
             }
 
+            sym::return_address => {
+                let ty = self.type_ix(32);
+                let val = self.const_int(ty, 0);
+                self.call_intrinsic("llvm.returnaddress", &[], &[val])
+            }
+
             _ => {
                 debug!("unknown intrinsic '{}' -- falling back to default body", name);
                 // Call the fallback body instead of generating the intrinsic code

compiler/rustc_codegen_llvm/src/intrinsic.rs

@@ -120,7 +120,8 @@ impl<'a, 'tcx, Bx: BuilderMethods<'a, 'tcx>> FunctionCx<'a, 'tcx, Bx> {
                 | sym::contract_checks
                 | sym::atomic_fence
                 | sym::atomic_singlethreadfence
-                | sym::caller_location => {}
+                | sym::caller_location
+                | sym::return_address => {}
                 _ => {
                     span_bug!(
                         span,

compiler/rustc_codegen_ssa/src/mir/intrinsic.rs

@@ -179,6 +179,7 @@ fn intrinsic_operation_unsafety(tcx: TyCtxt<'_>, intrinsic_id: LocalDefId) -> hi
         | sym::ptr_guaranteed_cmp
         | sym::ptr_mask
         | sym::ptr_metadata
+        | sym::return_address
         | sym::rotate_left
         | sym::rotate_right
         | sym::round_ties_even_f16
@@ -801,6 +802,8 @@ pub(crate) fn check_intrinsic_type(
         | sym::atomic_xor => (2, 1, vec![Ty::new_mut_ptr(tcx, param(0)), param(1)], param(0)),
         sym::atomic_fence | sym::atomic_singlethreadfence => (0, 1, Vec::new(), tcx.types.unit),
 
+        sym::return_address => (0, 0, vec![], Ty::new_imm_ptr(tcx, tcx.types.unit)),
+
         other => {
             tcx.dcx().emit_err(UnrecognizedIntrinsicFunction { span, name: other });
             return;

compiler/rustc_hir_analysis/src/check/intrinsic.rs

@@ -1657,6 +1657,7 @@ symbols! {
         residual,
         result,
         result_ffi_guarantees,
+        return_address,
         return_position_impl_trait_in_trait,
         return_type_notation,
         riscv32,

compiler/rustc_span/src/symbol.rs

@@ -76,3 +76,29 @@ pub macro global_asm("assembly template", $(operands,)* $(options($(option),*))?
 pub fn breakpoint() {
     core::intrinsics::breakpoint();
 }
+
+/// The `core::arch::return_address!()` macro returns a pointer with an address that corresponds to the caller of the function that invoked the `return_address!()` macro.
+/// The pointer has no provenance, as if created by `core::ptr::without_provenance`. It cannot be used to read memory (other than ZSTs).
+///
+/// The value returned by the macro depends highly on the architecture and compiler (including any options set).
+/// In particular, it is allowed to be wrong (particularly if inlining is involved), or even contain a nonsense value.
+/// The result of this macro must not be relied upon for soundness or correctness, only for debugging purposes.
+///
+/// As a best effort, if a useful value cannot be determined (for example, due to limitations on the current codegen),
+/// this macro tries to return a null pointer instead of nonsense (this cannot be relied upon for correctness, however).
+///
+/// Formally, this function returns a pointer with a non-deterministic address and no provenance.
+///
+/// This is equivalent to the gcc `__builtin_return_address(0)` intrinsic (other forms of the intrinsic are not supported).
+/// Because the operation can be always performed by the compiler without crashing or causing undefined behaviour, invoking the macro is a safe operation.
+///
+/// ## Example
+/// ```
+/// #![feature(return_address)]
+///
+/// let addr = core::arch::return_address!();
+/// println!("Caller is {addr:p}");
+/// ```
+#[unstable(feature = "return_address", issue = "154966")]
+#[allow_internal_unstable(core_intrinsics)]
+pub macro return_address() {{ core::intrinsics::return_address() }}

library/core/src/arch.rs

@@ -93,31 +93,31 @@ const MAX_THREE_B: u32 = 0x10000;
 
 /// The highest valid code point a `char` can have, `'\u{10FFFF}'`. Use [`char::MAX`] instead.
 #[stable(feature = "rust1", since = "1.0.0")]
+#[deprecated(since = "TBD", note = "replaced by the `MAX` associated constant on `char`")]
 pub const MAX: char = char::MAX;
 
-/// The maximum number of bytes required to [encode](char::encode_utf8) a `char` to
-/// UTF-8 encoding.
-#[unstable(feature = "char_max_len", issue = "121714")]
-pub const MAX_LEN_UTF8: usize = char::MAX_LEN_UTF8;
-
-/// The maximum number of two-byte units required to [encode](char::encode_utf16) a `char`
-/// to UTF-16 encoding.
-#[unstable(feature = "char_max_len", issue = "121714")]
-pub const MAX_LEN_UTF16: usize = char::MAX_LEN_UTF16;
-
 /// `U+FFFD REPLACEMENT CHARACTER` (�) is used in Unicode to represent a
 /// decoding error. Use [`char::REPLACEMENT_CHARACTER`] instead.
 #[stable(feature = "decode_utf16", since = "1.9.0")]
+#[deprecated(
+    since = "TBD",
+    note = "replaced by the `REPLACEMENT_CHARACTER` associated constant on `char`"
+)]
 pub const REPLACEMENT_CHARACTER: char = char::REPLACEMENT_CHARACTER;
 
 /// The version of [Unicode](https://www.unicode.org/) that the Unicode parts of
 /// `char` and `str` methods are based on. Use [`char::UNICODE_VERSION`] instead.
 #[stable(feature = "unicode_version", since = "1.45.0")]
+#[deprecated(
+    since = "TBD",
+    note = "replaced by the `UNICODE_VERSION` associated constant on `char`"
+)]
 pub const UNICODE_VERSION: (u8, u8, u8) = char::UNICODE_VERSION;
 
 /// Creates an iterator over the UTF-16 encoded code points in `iter`, returning
 /// unpaired surrogates as `Err`s. Use [`char::decode_utf16`] instead.
 #[stable(feature = "decode_utf16", since = "1.9.0")]
+#[deprecated(since = "TBD", note = "replaced by the `decode_utf16` method on `char`")]
 #[inline]
 pub fn decode_utf16<I: IntoIterator<Item = u16>>(iter: I) -> DecodeUtf16<I::IntoIter> {
     self::decode::decode_utf16(iter)
@@ -126,6 +126,7 @@ pub fn decode_utf16<I: IntoIterator<Item = u16>>(iter: I) -> DecodeUtf16<I::Into
 /// Converts a `u32` to a `char`. Use [`char::from_u32`] instead.
 #[stable(feature = "rust1", since = "1.0.0")]
 #[rustc_const_stable(feature = "const_char_convert", since = "1.67.0")]
+#[deprecated(since = "TBD", note = "replaced by the `from_u32` method on `char`")]
 #[must_use]
 #[inline]
 pub const fn from_u32(i: u32) -> Option<char> {
@@ -136,6 +137,7 @@ pub const fn from_u32(i: u32) -> Option<char> {
 /// instead.
 #[stable(feature = "char_from_unchecked", since = "1.5.0")]
 #[rustc_const_stable(feature = "const_char_from_u32_unchecked", since = "1.81.0")]
+#[deprecated(since = "TBD", note = "replaced by the `from_u32_unchecked` method on `char`")]
 #[must_use]
 #[inline]
 pub const unsafe fn from_u32_unchecked(i: u32) -> char {
@@ -146,6 +148,7 @@ pub const unsafe fn from_u32_unchecked(i: u32) -> char {
 /// Converts a digit in the given radix to a `char`. Use [`char::from_digit`] instead.
 #[stable(feature = "rust1", since = "1.0.0")]
 #[rustc_const_stable(feature = "const_char_convert", since = "1.67.0")]
+#[deprecated(since = "TBD", note = "replaced by the `from_digit` method on `char`")]
 #[must_use]
 #[inline]
 pub const fn from_digit(num: u32, radix: u32) -> Option<char> {

library/core/src/char/mod.rs

@@ -3589,3 +3589,16 @@ pub const fn va_copy<'f>(src: &VaList<'f>) -> VaList<'f> {
 pub const unsafe fn va_end(ap: &mut VaList<'_>) {
     /* deliberately does nothing */
 }
+
+/// Returns the return address of the caller function (after inlining) in a best-effort manner or a null pointer if it is not supported on the current backend.
+/// Returning an accurate value is a quality-of-implementation concern, but no hard guarantees are
+/// made about the return value: formally, the intrinsic non-deterministically returns
+/// an arbitrary pointer without provenance.
+///
+/// Note that unlike most intrinsics, this is safe to call. This is because it only finds the return address of the immediate caller, which is guaranteed to be possible.
+/// Other forms of the corresponding gcc or llvm intrinsic (which can have wildly unpredictable results or even crash at runtime) are not exposed.
+#[rustc_intrinsic]
+#[rustc_nounwind]
+pub fn return_address() -> *const () {
+    core::ptr::null()
+}