Rollup merge of rust-lang#145670 - jdonszelmann:port-sanitize, r=lcnr

port `sanitize` attribute to the new parsing infrastructure

Author: samueltardieu

compiler/rustc_attr_parsing/src/attributes/codegen_attrs.rs

@@ -1,4 +1,4 @@
-use rustc_hir::attrs::{CoverageAttrKind, OptimizeAttr, UsedBy};
+use rustc_hir::attrs::{CoverageAttrKind, OptimizeAttr, SanitizerSet, UsedBy};
 use rustc_session::parse::feature_err;
 
 use super::prelude::*;
@@ -464,18 +464,119 @@ impl<S: Stage> CombineAttributeParser<S> for ForceTargetFeatureParser {
         was_forced: true,
     };
     const TEMPLATE: AttributeTemplate = template!(List: &["enable = \"feat1, feat2\""]);
+    const ALLOWED_TARGETS: AllowedTargets = AllowedTargets::AllowList(&[
+        Allow(Target::Fn),
+        Allow(Target::Method(MethodKind::Inherent)),
+        Allow(Target::Method(MethodKind::Trait { body: true })),
+        Allow(Target::Method(MethodKind::TraitImpl)),
+    ]);
 
     fn extend<'c>(
         cx: &'c mut AcceptContext<'_, '_, S>,
         args: &'c ArgParser<'_>,
     ) -> impl IntoIterator<Item = Self::Item> + 'c {
         parse_tf_attribute(cx, args)
     }
+}
 
-    const ALLOWED_TARGETS: AllowedTargets = AllowedTargets::AllowList(&[
-        Allow(Target::Fn),
-        Allow(Target::Method(MethodKind::Inherent)),
-        Allow(Target::Method(MethodKind::Trait { body: true })),
-        Allow(Target::Method(MethodKind::TraitImpl)),
+pub(crate) struct SanitizeParser;
+
+impl<S: Stage> SingleAttributeParser<S> for SanitizeParser {
+    const PATH: &[Symbol] = &[sym::sanitize];
+
+    // FIXME: still checked in check_attrs.rs
+    const ALLOWED_TARGETS: AllowedTargets = AllowedTargets::AllowList(ALL_TARGETS);
+
+    const TEMPLATE: AttributeTemplate = template!(List: &[
+        r#"address = "on|off""#,
+        r#"kernel_address = "on|off""#,
+        r#"cfi = "on|off""#,
+        r#"hwaddress = "on|off""#,
+        r#"kcfi = "on|off""#,
+        r#"memory = "on|off""#,
+        r#"memtag = "on|off""#,
+        r#"shadow_call_stack = "on|off""#,
+        r#"thread = "on|off""#
     ]);
+
+    const ATTRIBUTE_ORDER: AttributeOrder = AttributeOrder::KeepOutermost;
+    const ON_DUPLICATE: OnDuplicate<S> = OnDuplicate::Error;
+
+    fn convert(cx: &mut AcceptContext<'_, '_, S>, args: &ArgParser<'_>) -> Option<AttributeKind> {
+        let Some(list) = args.list() else {
+            cx.expected_list(cx.attr_span);
+            return None;
+        };
+
+        let mut on_set = SanitizerSet::empty();
+        let mut off_set = SanitizerSet::empty();
+
+        for item in list.mixed() {
+            let Some(item) = item.meta_item() else {
+                cx.expected_name_value(item.span(), None);
+                continue;
+            };
+
+            let path = item.path().word_sym();
+            let Some(value) = item.args().name_value() else {
+                cx.expected_name_value(item.span(), path);
+                continue;
+            };
+
+            let mut apply = |s: SanitizerSet| {
+                let is_on = match value.value_as_str() {
+                    Some(sym::on) => true,
+                    Some(sym::off) => false,
+                    Some(_) => {
+                        cx.expected_specific_argument_strings(
+                            value.value_span,
+                            &[sym::on, sym::off],
+                        );
+                        return;
+                    }
+                    None => {
+                        cx.expected_string_literal(value.value_span, Some(value.value_as_lit()));
+                        return;
+                    }
+                };
+
+                if is_on {
+                    on_set |= s;
+                } else {
+                    off_set |= s;
+                }
+            };
+
+            match path {
+                Some(sym::address) | Some(sym::kernel_address) => {
+                    apply(SanitizerSet::ADDRESS | SanitizerSet::KERNELADDRESS)
+                }
+                Some(sym::cfi) => apply(SanitizerSet::CFI),
+                Some(sym::kcfi) => apply(SanitizerSet::KCFI),
+                Some(sym::memory) => apply(SanitizerSet::MEMORY),
+                Some(sym::memtag) => apply(SanitizerSet::MEMTAG),
+                Some(sym::shadow_call_stack) => apply(SanitizerSet::SHADOWCALLSTACK),
+                Some(sym::thread) => apply(SanitizerSet::THREAD),
+                Some(sym::hwaddress) => apply(SanitizerSet::HWADDRESS),
+                _ => {
+                    cx.expected_specific_argument_strings(
+                        item.path().span(),
+                        &[
+                            sym::address,
+                            sym::cfi,
+                            sym::kcfi,
+                            sym::memory,
+                            sym::memtag,
+                            sym::shadow_call_stack,
+                            sym::thread,
+                            sym::hwaddress,
+                        ],
+                    );
+                    continue;
+                }
+            }
+        }
+
+        Some(AttributeKind::Sanitize { on_set, off_set, span: cx.attr_span })
+    }
 }

compiler/rustc_attr_parsing/src/context.rs

@@ -20,7 +20,8 @@ use crate::attributes::allow_unstable::{
 use crate::attributes::body::CoroutineParser;
 use crate::attributes::codegen_attrs::{
     ColdParser, CoverageParser, ExportNameParser, ForceTargetFeatureParser, NakedParser,
-    NoMangleParser, OptimizeParser, TargetFeatureParser, TrackCallerParser, UsedParser,
+    NoMangleParser, OptimizeParser, SanitizeParser, TargetFeatureParser, TrackCallerParser,
+    UsedParser,
 };
 use crate::attributes::confusables::ConfusablesParser;
 use crate::attributes::deprecation::DeprecationParser;
@@ -184,6 +185,7 @@ attribute_parsers!(
         Single<RustcLayoutScalarValidRangeEnd>,
         Single<RustcLayoutScalarValidRangeStart>,
         Single<RustcObjectLifetimeDefaultParser>,
+        Single<SanitizeParser>,
         Single<ShouldPanicParser>,
         Single<SkipDuringMethodDispatchParser>,
         Single<TransparencyParser>,

compiler/rustc_codegen_ssa/messages.ftl

@@ -171,9 +171,6 @@ codegen_ssa_invalid_monomorphization_unsupported_symbol = invalid monomorphizati
 
 codegen_ssa_invalid_monomorphization_unsupported_symbol_of_size = invalid monomorphization of `{$name}` intrinsic: unsupported {$symbol} from `{$in_ty}` with element `{$in_elem}` of size `{$size}` to `{$ret_ty}`
 
-codegen_ssa_invalid_sanitize = invalid argument for `sanitize`
-    .note = expected one of: `address`, `kernel_address`, `cfi`, `hwaddress`, `kcfi`, `memory`, `memtag`, `shadow_call_stack`, or `thread`
-
 codegen_ssa_invalid_windows_subsystem = invalid windows subsystem `{$subsystem}`, only `windows` and `console` are allowed
 
 codegen_ssa_ld64_unimplemented_modifier = `as-needed` modifier not implemented yet for ld64

compiler/rustc_codegen_ssa/src/codegen_attrs.rs

@@ -293,6 +293,9 @@ fn process_builtin_attrs(
                         codegen_fn_attrs.linkage = linkage;
                     }
                 }
+                AttributeKind::Sanitize { span, .. } => {
+                    interesting_spans.sanitize = Some(*span);
+                }
                 _ => {}
             }
         }
@@ -310,7 +313,6 @@ fn process_builtin_attrs(
                 codegen_fn_attrs.flags |= CodegenFnAttrFlags::ALLOCATOR_ZEROED
             }
             sym::thread_local => codegen_fn_attrs.flags |= CodegenFnAttrFlags::THREAD_LOCAL,
-            sym::sanitize => interesting_spans.sanitize = Some(attr.span()),
             sym::instruction_set => {
                 codegen_fn_attrs.instruction_set = parse_instruction_set_attr(tcx, attr)
             }
@@ -560,79 +562,9 @@ fn opt_trait_item(tcx: TyCtxt<'_>, def_id: DefId) -> Option<DefId> {
     }
 }
 
-/// For an attr that has the `sanitize` attribute, read the list of
-/// disabled sanitizers. `current_attr` holds the information about
-/// previously parsed attributes.
-fn parse_sanitize_attr(
-    tcx: TyCtxt<'_>,
-    attr: &Attribute,
-    current_attr: SanitizerSet,
-) -> SanitizerSet {
-    let mut result = current_attr;
-    if let Some(list) = attr.meta_item_list() {
-        for item in list.iter() {
-            let MetaItemInner::MetaItem(set) = item else {
-                tcx.dcx().emit_err(errors::InvalidSanitize { span: attr.span() });
-                break;
-            };
-            let segments = set.path.segments.iter().map(|x| x.ident.name).collect::<Vec<_>>();
-            match segments.as_slice() {
-                // Similar to clang, sanitize(address = ..) and
-                // sanitize(kernel_address = ..) control both ASan and KASan
-                // Source: https://reviews.llvm.org/D44981.
-                [sym::address] | [sym::kernel_address] if set.value_str() == Some(sym::off) => {
-                    result |= SanitizerSet::ADDRESS | SanitizerSet::KERNELADDRESS
-                }
-                [sym::address] | [sym::kernel_address] if set.value_str() == Some(sym::on) => {
-                    result &= !SanitizerSet::ADDRESS;
-                    result &= !SanitizerSet::KERNELADDRESS;
-                }
-                [sym::cfi] if set.value_str() == Some(sym::off) => result |= SanitizerSet::CFI,
-                [sym::cfi] if set.value_str() == Some(sym::on) => result &= !SanitizerSet::CFI,
-                [sym::kcfi] if set.value_str() == Some(sym::off) => result |= SanitizerSet::KCFI,
-                [sym::kcfi] if set.value_str() == Some(sym::on) => result &= !SanitizerSet::KCFI,
-                [sym::memory] if set.value_str() == Some(sym::off) => {
-                    result |= SanitizerSet::MEMORY
-                }
-                [sym::memory] if set.value_str() == Some(sym::on) => {
-                    result &= !SanitizerSet::MEMORY
-                }
-                [sym::memtag] if set.value_str() == Some(sym::off) => {
-                    result |= SanitizerSet::MEMTAG
-                }
-                [sym::memtag] if set.value_str() == Some(sym::on) => {
-                    result &= !SanitizerSet::MEMTAG
-                }
-                [sym::shadow_call_stack] if set.value_str() == Some(sym::off) => {
-                    result |= SanitizerSet::SHADOWCALLSTACK
-                }
-                [sym::shadow_call_stack] if set.value_str() == Some(sym::on) => {
-                    result &= !SanitizerSet::SHADOWCALLSTACK
-                }
-                [sym::thread] if set.value_str() == Some(sym::off) => {
-                    result |= SanitizerSet::THREAD
-                }
-                [sym::thread] if set.value_str() == Some(sym::on) => {
-                    result &= !SanitizerSet::THREAD
-                }
-                [sym::hwaddress] if set.value_str() == Some(sym::off) => {
-                    result |= SanitizerSet::HWADDRESS
-                }
-                [sym::hwaddress] if set.value_str() == Some(sym::on) => {
-                    result &= !SanitizerSet::HWADDRESS
-                }
-                _ => {
-                    tcx.dcx().emit_err(errors::InvalidSanitize { span: attr.span() });
-                }
-            }
-        }
-    }
-    result
-}
-
 fn disabled_sanitizers_for(tcx: TyCtxt<'_>, did: LocalDefId) -> SanitizerSet {
     // Backtrack to the crate root.
-    let disabled = match tcx.opt_local_parent(did) {
+    let mut disabled = match tcx.opt_local_parent(did) {
         // Check the parent (recursively).
         Some(parent) => tcx.disabled_sanitizers_for(parent),
         // We reached the crate root without seeing an attribute, so
@@ -641,8 +573,17 @@ fn disabled_sanitizers_for(tcx: TyCtxt<'_>, did: LocalDefId) -> SanitizerSet {
     };
 
     // Check for a sanitize annotation directly on this def.
-    if let Some(attr) = tcx.get_attr(did, sym::sanitize) {
-        return parse_sanitize_attr(tcx, attr, disabled);
+    if let Some((on_set, off_set)) = find_attr!(tcx.get_all_attrs(did), AttributeKind::Sanitize {on_set, off_set, ..} => (on_set, off_set))
+    {
+        // the on set is the set of sanitizers explicitly enabled.
+        // we mask those out since we want the set of disabled sanitizers here
+        disabled &= !*on_set;
+        // the off set is the set of sanitizers explicitly disabled.
+        // we or those in here.
+        disabled |= *off_set;
+        // the on set and off set are distjoint since there's a third option: unset.
+        // a node may not set the sanitizer setting in which case it inherits from parents.
+        // the code above in this function does this backtracking
     }
     disabled
 }

compiler/rustc_codegen_ssa/src/errors.rs

@@ -1120,14 +1120,6 @@ impl IntoDiagArg for ExpectedPointerMutability {
     }
 }
 
-#[derive(Diagnostic)]
-#[diag(codegen_ssa_invalid_sanitize)]
-#[note]
-pub(crate) struct InvalidSanitize {
-    #[primary_span]
-    pub span: Span,
-}
-
 #[derive(Diagnostic)]
 #[diag(codegen_ssa_target_feature_safe_trait)]
 pub(crate) struct TargetFeatureSafeTrait {

compiler/rustc_hir/src/attrs/data_structures.rs

@@ -10,6 +10,7 @@ use rustc_macros::{Decodable, Encodable, HashStable_Generic, PrintAttribute};
 use rustc_span::def_id::DefId;
 use rustc_span::hygiene::Transparency;
 use rustc_span::{Ident, Span, Symbol};
+pub use rustc_target::spec::SanitizerSet;
 use thin_vec::ThinVec;
 
 use crate::attrs::pretty_printing::PrintAttribute;
@@ -505,6 +506,12 @@ pub enum AttributeKind {
     /// Represents `#[rustc_object_lifetime_default]`.
     RustcObjectLifetimeDefault,
 
+    /// Represents `#[sanitize]`
+    ///
+    /// the on set and off set are distjoint since there's a third option: unset.
+    /// a node may not set the sanitizer setting in which case it inherits from parents.
+    Sanitize { on_set: SanitizerSet, off_set: SanitizerSet, span: Span },
+
     /// Represents `#[should_panic]`
     ShouldPanic { reason: Option<Symbol>, span: Span },
 

compiler/rustc_hir/src/attrs/encode_cross_crate.rs

@@ -73,6 +73,7 @@ impl AttributeKind {
             RustcLayoutScalarValidRangeEnd(..) => Yes,
             RustcLayoutScalarValidRangeStart(..) => Yes,
             RustcObjectLifetimeDefault => No,
+            Sanitize { .. } => No,
             ShouldPanic { .. } => No,
             SkipDuringMethodDispatch { .. } => No,
             SpecializationTrait(..) => No,

compiler/rustc_hir/src/attrs/pretty_printing.rs

@@ -6,6 +6,7 @@ use rustc_ast::{AttrStyle, IntTy, UintTy};
 use rustc_ast_pretty::pp::Printer;
 use rustc_span::hygiene::Transparency;
 use rustc_span::{ErrorGuaranteed, Ident, Span, Symbol};
+use rustc_target::spec::SanitizerSet;
 use thin_vec::ThinVec;
 
 /// This trait is used to print attributes in `rustc_hir_pretty`.
@@ -146,4 +147,14 @@ macro_rules! print_tup {
 print_tup!(A B C D E F G H);
 print_skip!(Span, (), ErrorGuaranteed);
 print_disp!(u16, bool, NonZero<u32>);
-print_debug!(Symbol, Ident, UintTy, IntTy, Align, AttrStyle, CommentKind, Transparency);
+print_debug!(
+    Symbol,
+    Ident,
+    UintTy,
+    IntTy,
+    Align,
+    AttrStyle,
+    CommentKind,
+    Transparency,
+    SanitizerSet,
+);