sanitizers: Implement support for the sanitize ignorelist

compiler/rustc_codegen_llvm/src/abi.rs

@@ -595,6 +595,7 @@ impl<'ll, 'tcx> FnAbiLlvmExt<'ll, 'tcx> for FnAbi<'tcx, Ty<'tcx>> {
                 llfn,
                 &cx.tcx.codegen_instance_attrs(instance.def),
                 Some(instance),
+                cx.sanitizer_ignorelist.as_ref(),
             );
         }
     }

compiler/rustc_codegen_llvm/src/allocator.rs

@@ -125,7 +125,7 @@ fn create_wrapper_function(
         ty,
     );
 
-    llfn_attrs_from_instance(cx, tcx, llfn, attrs, None);
+    llfn_attrs_from_instance(cx, tcx, llfn, attrs, None, None);
 
     let no_return = if no_return {
         // -> ! DIFlagNoReturn

compiler/rustc_codegen_llvm/src/attributes.rs

@@ -111,9 +111,9 @@ pub(crate) fn sanitize_attrs<'ll, 'tcx>(
     cx: &SimpleCx<'ll>,
     tcx: TyCtxt<'tcx>,
     sanitizer_fn_attr: SanitizerFnAttrs,
+    enabled: SanitizerSet,
 ) -> SmallVec<[&'ll Attribute; 4]> {
     let mut attrs = SmallVec::new();
-    let enabled = tcx.sess.sanitizers() - sanitizer_fn_attr.disabled;
     if enabled.contains(SanitizerSet::ADDRESS) || enabled.contains(SanitizerSet::KERNELADDRESS) {
         attrs.push(llvm::AttributeKind::SanitizeAddress.create_attr(cx.llcx));
     }
@@ -409,6 +409,7 @@ pub(crate) fn llfn_attrs_from_instance<'ll, 'tcx>(
     llfn: &'ll Value,
     codegen_fn_attrs: &CodegenFnAttrs,
     instance: Option<ty::Instance<'tcx>>,
+    sanitizer_ignorelist: Option<&crate::llvm::SanitizerIgnoreList>,
 ) {
     let sess = tcx.sess;
     let mut to_add = SmallVec::<[_; 16]>::new();
@@ -470,7 +471,67 @@ pub(crate) fn llfn_attrs_from_instance<'ll, 'tcx>(
         // not used.
     } else {
         // Do not set sanitizer attributes for naked functions.
-        to_add.extend(sanitize_attrs(cx, tcx, codegen_fn_attrs.sanitizers));
+        let mut enabled = tcx.sess.sanitizers() - codegen_fn_attrs.sanitizers.disabled;
+        if let Some(ignorelist) = sanitizer_ignorelist {
+            if let Some(instance) = instance {
+                let sym_name = tcx.symbol_name(instance).name;
+                let span = tcx.def_span(instance.def_id());
+                let source_map = tcx.sess.source_map();
+                let filename =
+                    source_map.span_to_filename(span).prefer_local_unconditionally().to_string();
+
+                let mainfile = tcx
+                    .sess
+                    .local_crate_source_file()
+                    .and_then(|path| path.local_path().map(|p| p.display().to_string()))
+                    .unwrap_or_default();
+
+                let demangled = rustc_middle::ty::print::with_no_trimmed_paths!(
+                    tcx.def_path_str(instance.def_id())
+                );
+                let is_ignored = |section: &std::ffi::CStr| -> bool {
+                    ignorelist.contains_prefix(section, c"fun", sym_name)
+                        || ignorelist.contains_prefix(section, c"fun", &demangled)
+                        || ignorelist.contains_prefix(section, c"src", &filename)
+                        || (!mainfile.is_empty()
+                            && ignorelist.contains_prefix(section, c"mainfile", &mainfile))
+                };
+
+                let ignore_address = is_ignored(c"address");
+                let ignore_kernel_address = ignore_address || is_ignored(c"kernel-address");
+                let ignore_hwaddress = is_ignored(c"hwaddress");
+                let ignore_kernel_hwaddress = ignore_hwaddress || is_ignored(c"kernel-hwaddress");
+
+                if enabled.contains(SanitizerSet::ADDRESS) && ignore_address {
+                    enabled.remove(SanitizerSet::ADDRESS);
+                }
+                if enabled.contains(SanitizerSet::KERNELADDRESS) && ignore_kernel_address {
+                    enabled.remove(SanitizerSet::KERNELADDRESS);
+                }
+                if enabled.contains(SanitizerSet::MEMORY) && is_ignored(c"memory") {
+                    enabled.remove(SanitizerSet::MEMORY);
+                }
+                if enabled.contains(SanitizerSet::THREAD) && is_ignored(c"thread") {
+                    enabled.remove(SanitizerSet::THREAD);
+                }
+                if enabled.contains(SanitizerSet::HWADDRESS) && ignore_hwaddress {
+                    enabled.remove(SanitizerSet::HWADDRESS);
+                }
+                if enabled.contains(SanitizerSet::KERNELHWADDRESS) && ignore_kernel_hwaddress {
+                    enabled.remove(SanitizerSet::KERNELHWADDRESS);
+                }
+                if enabled.contains(SanitizerSet::SAFESTACK) && is_ignored(c"safestack") {
+                    enabled.remove(SanitizerSet::SAFESTACK);
+                }
+                if is_ignored(c"cfi") {
+                    to_add.push(llvm::CreateAttrString(cx.llcx, "no-sanitize-cfi"));
+                }
+                if is_ignored(c"kcfi") {
+                    to_add.push(llvm::CreateAttrString(cx.llcx, "no-sanitize-kcfi"));
+                }
+            }
+        }
+        to_add.extend(sanitize_attrs(cx, tcx, codegen_fn_attrs.sanitizers, enabled));
 
         // For non-naked functions, set branch protection attributes on aarch64.
         if let Some(BranchProtection { bti, pac_ret, gcs }) =

compiler/rustc_codegen_llvm/src/base.rs

@@ -123,7 +123,12 @@ pub(crate) fn compile_codegen_unit(
             if let Some(entry) =
                 maybe_create_entry_wrapper::<Builder<'_, '_, '_>>(&cx, cx.codegen_unit)
             {
-                let attrs = attributes::sanitize_attrs(&cx, tcx, SanitizerFnAttrs::default());
+                let attrs = attributes::sanitize_attrs(
+                    &cx,
+                    tcx,
+                    SanitizerFnAttrs::default(),
+                    tcx.sess.sanitizers(),
+                );
                 attributes::apply_to_llfn(entry, llvm::AttributePlace::Function, &attrs);
             }
 

compiler/rustc_codegen_llvm/src/builder.rs

@@ -1832,7 +1832,7 @@ impl<'a, 'll, 'tcx> Builder<'a, 'll, 'tcx> {
 
         // Emit KCFI operand bundle
         let kcfi_bundle = self.kcfi_operand_bundle(fn_attrs, fn_abi, instance, llfn);
-        if let Some(kcfi_bundle) = kcfi_bundle.as_ref().map(|b| b.as_ref()) {
+        if let Some(kcfi_bundle) = kcfi_bundle.as_ref().map(|bundle| bundle.as_ref()) {
             bundles.push(kcfi_bundle);
         }
 
@@ -1875,6 +1875,9 @@ impl<'a, 'll, 'tcx> Builder<'a, 'll, 'tcx> {
             {
                 return;
             }
+            if crate::llvm::HasStringAttribute(self.llfn(), "no-sanitize-cfi") {
+                return;
+            }
 
             let mut options = cfi::TypeIdOptions::empty();
             if self.tcx.sess.is_sanitizer_cfi_generalize_pointers_enabled() {
@@ -1884,6 +1887,10 @@ impl<'a, 'll, 'tcx> Builder<'a, 'll, 'tcx> {
                 options.insert(cfi::TypeIdOptions::NORMALIZE_INTEGERS);
             }
 
+            if self.cx.is_sanitizer_type_ignored(c"cfi", fn_abi) {
+                return;
+            }
+
             let typeid = if let Some(instance) = instance {
                 cfi::typeid_for_instance(self.tcx, instance, options)
             } else {
@@ -1933,6 +1940,9 @@ impl<'a, 'll, 'tcx> Builder<'a, 'll, 'tcx> {
             {
                 return None;
             }
+            if crate::llvm::HasStringAttribute(self.llfn(), "no-sanitize-kcfi") {
+                return None;
+            }
 
             let mut options = kcfi::TypeIdOptions::empty();
             if self.tcx.sess.is_sanitizer_cfi_generalize_pointers_enabled() {
@@ -1942,6 +1952,10 @@ impl<'a, 'll, 'tcx> Builder<'a, 'll, 'tcx> {
                 options.insert(kcfi::TypeIdOptions::NORMALIZE_INTEGERS);
             }
 
+            if self.cx.is_sanitizer_type_ignored(c"kcfi", fn_abi) {
+                return None;
+            }
+
             let kcfi_typeid = if let Some(instance) = instance {
                 kcfi::typeid_for_instance(self.tcx, instance, options)
             } else {

compiler/rustc_codegen_llvm/src/common.rs

@@ -13,7 +13,7 @@ use rustc_hashes::Hash128;
 use rustc_hir::def_id::DefId;
 use rustc_middle::bug;
 use rustc_middle::mir::interpret::{GlobalAlloc, PointerArithmetic, Scalar};
-use rustc_middle::ty::TyCtxt;
+use rustc_middle::ty::{self, Ty, TyCtxt};
 use rustc_session::cstore::DllImport;
 use tracing::debug;
 
@@ -446,3 +446,17 @@ impl AsCCharPtr for [u8] {
         self.as_ptr().cast()
     }
 }
+
+pub(crate) fn type_name_for_ignore_list<'tcx>(
+    tcx: TyCtxt<'tcx>,
+    fn_abi: &rustc_target::callconv::FnAbi<'tcx, Ty<'tcx>>,
+) -> String {
+    let inputs: Vec<_> = fn_abi.args.iter().map(|arg| arg.layout.ty).collect();
+    let output = fn_abi.ret.layout.ty;
+    let mut fn_sig_kind = ty::FnSigKind::default();
+    fn_sig_kind = fn_sig_kind.set_safety(rustc_hir::Safety::Safe);
+    fn_sig_kind = fn_sig_kind.set_c_variadic(fn_abi.c_variadic);
+    let fn_sig = tcx.mk_fn_sig(inputs, output, fn_sig_kind);
+    let fn_ptr = Ty::new_fn_ptr(tcx, ty::Binder::dummy(fn_sig));
+    ty::print::with_no_trimmed_paths!(fn_ptr.to_string())
+}

compiler/rustc_codegen_llvm/src/consts.rs

@@ -509,6 +509,52 @@ impl<'ll> CodegenCx<'ll, '_> {
 
         base::set_variable_sanitizer_attrs(g, attrs);
 
+        if let Some(ignorelist) = &self.sanitizer_ignorelist {
+            let instance = ty::Instance::mono(self.tcx, def_id);
+            let sym_name = self.tcx.symbol_name(instance).name;
+            let span = self.tcx.def_span(def_id);
+            let source_map = self.tcx.sess.source_map();
+            let filename =
+                source_map.span_to_filename(span).prefer_local_unconditionally().to_string();
+            let ty_name = rustc_middle::ty::print::with_no_trimmed_paths!(
+                self.tcx.type_of(def_id).skip_binder().to_string()
+            );
+            let mainfile = self
+                .tcx
+                .sess
+                .local_crate_source_file()
+                .and_then(|path| path.local_path().map(|p| p.display().to_string()))
+                .unwrap_or_default();
+
+            let is_ignored = |section: &std::ffi::CStr| -> bool {
+                ignorelist.contains_prefix(section, c"global", sym_name)
+                    || ignorelist.contains_prefix(section, c"src", &filename)
+                    || (!mainfile.is_empty()
+                        && ignorelist.contains_prefix(section, c"mainfile", &mainfile))
+                    || ignorelist.contains_prefix(section, c"type", &ty_name)
+            };
+
+            let sanitizers = self.tcx.sess.sanitizers();
+            let ignore_address = is_ignored(c"address");
+            let ignore_kernel_address = ignore_address || is_ignored(c"kernel-address");
+            let ignore_hwaddress = is_ignored(c"hwaddress");
+            let ignore_kernel_hwaddress = ignore_hwaddress || is_ignored(c"kernel-hwaddress");
+
+            if (sanitizers.contains(rustc_target::spec::SanitizerSet::ADDRESS) && ignore_address)
+                || (sanitizers.contains(rustc_target::spec::SanitizerSet::KERNELADDRESS)
+                    && ignore_kernel_address)
+            {
+                unsafe { llvm::LLVMRustSetNoSanitizeAddress(g) };
+            }
+            if (sanitizers.contains(rustc_target::spec::SanitizerSet::HWADDRESS)
+                && ignore_hwaddress)
+                || (sanitizers.contains(rustc_target::spec::SanitizerSet::KERNELHWADDRESS)
+                    && ignore_kernel_hwaddress)
+            {
+                unsafe { llvm::LLVMRustSetNoSanitizeHWAddress(g) };
+            }
+        }
+
         if attrs.flags.contains(CodegenFnAttrFlags::USED_COMPILER) {
             // `USED` and `USED_LINKER` can't be used together.
             assert!(!attrs.flags.contains(CodegenFnAttrFlags::USED_LINKER));

compiler/rustc_codegen_llvm/src/context.rs

@@ -133,6 +133,7 @@ pub(crate) struct FullCx<'ll, 'tcx> {
     /// Extra per-CGU codegen state needed when coverage instrumentation is enabled.
     pub coverage_cx: Option<coverageinfo::CguCoverageContext<'ll, 'tcx>>,
     pub dbg_cx: Option<debuginfo::CodegenUnitDebugContext<'ll, 'tcx>>,
+    pub sanitizer_ignorelist: Option<crate::llvm::SanitizerIgnoreList>,
 
     eh_personality: Cell<Option<&'ll Value>>,
     pub rust_try_fn: Cell<Option<(&'ll Type, &'ll Value)>>,
@@ -650,6 +651,26 @@ impl<'ll, 'tcx> CodegenCx<'ll, 'tcx> {
             None
         };
 
+        // FIXME: This parses the ignorelist files for each CGU, which adds a performance overhead.
+        // Clang parses it once per frontend invocation. LLVM's `SpecialCaseList::inSection`
+        // mutates an internal `LazyInit` cache and is not thread-safe. We either need to wrap
+        // the queries in a lock or wait for LLVM to expose a thread-safe way to query it.
+        let sanitizer_ignorelist = if !tcx.sess.opts.unstable_opts.sanitizer_ignorelist.is_empty() {
+            for path in &tcx.sess.opts.unstable_opts.sanitizer_ignorelist {
+                let _ = tcx.sess.source_map().load_file(std::path::Path::new(path));
+            }
+            match crate::llvm::SanitizerIgnoreList::new(
+                &tcx.sess.opts.unstable_opts.sanitizer_ignorelist,
+            ) {
+                Ok(list) => Some(list),
+                Err(err) => {
+                    tcx.dcx().fatal(format!("failed to parse sanitizer ignorelist: {}", err));
+                }
+            }
+        } else {
+            None
+        };
+
         GenericCx(
             FullCx {
                 tcx,
@@ -669,6 +690,7 @@ impl<'ll, 'tcx> CodegenCx<'ll, 'tcx> {
                 scalar_lltypes: Default::default(),
                 coverage_cx,
                 dbg_cx,
+                sanitizer_ignorelist,
                 eh_personality: Cell::new(None),
                 rust_try_fn: Cell::new(None),
                 intrinsics: Default::default(),
@@ -766,6 +788,17 @@ impl<'ll, 'tcx> CodegenCx<'ll, 'tcx> {
             1 << 6,
         );
     }
+
+    pub(crate) fn is_sanitizer_type_ignored(
+        &self,
+        sanitizer: &std::ffi::CStr,
+        fn_abi: &rustc_target::callconv::FnAbi<'tcx, Ty<'tcx>>,
+    ) -> bool {
+        self.sanitizer_ignorelist.as_ref().is_some_and(|ignorelist| {
+            let type_name = crate::common::type_name_for_ignore_list(self.tcx, fn_abi);
+            ignorelist.contains_prefix(sanitizer, c"type", &type_name)
+        })
+    }
 }
 impl<'ll> SimpleCx<'ll> {
     pub(crate) fn get_type_of_global(&self, val: &'ll Value) -> &'ll Type {