coverage. Lowering MC/DC statements to llvm-ir

Author: zhuyunxing

compiler/rustc_codegen_llvm/src/builder.rs

@@ -17,7 +17,7 @@ use rustc_data_structures::small_c_str::SmallCStr;
 use rustc_hir::def_id::DefId;
 use rustc_middle::middle::codegen_fn_attrs::CodegenFnAttrs;
 use rustc_middle::ty::layout::{
-    FnAbiError, FnAbiOfHelpers, FnAbiRequest, LayoutError, LayoutOfHelpers, TyAndLayout,
+    FnAbiError, FnAbiOfHelpers, FnAbiRequest, HasTyCtxt, LayoutError, LayoutOfHelpers, TyAndLayout,
 };
 use rustc_middle::ty::{self, Instance, Ty, TyCtxt};
 use rustc_sanitizers::{cfi, kcfi};
@@ -1702,4 +1702,128 @@ impl<'a, 'll, 'tcx> Builder<'a, 'll, 'tcx> {
         };
         kcfi_bundle
     }
+
+    pub(crate) fn mcdc_parameters(
+        &mut self,
+        fn_name: &'ll Value,
+        hash: &'ll Value,
+        bitmap_bytes: &'ll Value,
+    ) -> &'ll Value {
+        debug!("mcdc_parameters() with args ({:?}, {:?}, {:?})", fn_name, hash, bitmap_bytes);
+
+        assert!(llvm_util::get_version() >= (18, 0, 0), "MCDC intrinsics require LLVM 18 or later");
+
+        let llfn = unsafe { llvm::LLVMRustGetInstrProfMCDCParametersIntrinsic(self.cx().llmod) };
+        let llty = self.cx.type_func(
+            &[self.cx.type_ptr(), self.cx.type_i64(), self.cx.type_i32()],
+            self.cx.type_void(),
+        );
+        let args = &[fn_name, hash, bitmap_bytes];
+        let args = self.check_call("call", llty, llfn, args);
+
+        unsafe {
+            let _ = llvm::LLVMRustBuildCall(
+                self.llbuilder,
+                llty,
+                llfn,
+                args.as_ptr() as *const &llvm::Value,
+                args.len() as c_uint,
+                [].as_ptr(),
+                0 as c_uint,
+            );
+            // Create condition bitmap named `mcdc.addr`.
+            let mut bx = Builder::with_cx(self.cx);
+            bx.position_at_start(llvm::LLVMGetFirstBasicBlock(self.llfn()));
+            let cond_bitmap = {
+                let alloca =
+                    llvm::LLVMBuildAlloca(bx.llbuilder, bx.cx.type_i32(), c"mcdc.addr".as_ptr());
+                llvm::LLVMSetAlignment(alloca, 4);
+                alloca
+            };
+            bx.store(self.const_i32(0), cond_bitmap, self.tcx().data_layout.i32_align.abi);
+            cond_bitmap
+        }
+    }
+
+    pub(crate) fn mcdc_tvbitmap_update(
+        &mut self,
+        fn_name: &'ll Value,
+        hash: &'ll Value,
+        bitmap_bytes: &'ll Value,
+        bitmap_index: &'ll Value,
+        mcdc_temp: &'ll Value,
+    ) {
+        debug!(
+            "mcdc_tvbitmap_update() with args ({:?}, {:?}, {:?}, {:?}, {:?})",
+            fn_name, hash, bitmap_bytes, bitmap_index, mcdc_temp
+        );
+        assert!(llvm_util::get_version() >= (18, 0, 0), "MCDC intrinsics require LLVM 18 or later");
+
+        let llfn =
+            unsafe { llvm::LLVMRustGetInstrProfMCDCTVBitmapUpdateIntrinsic(self.cx().llmod) };
+        let llty = self.cx.type_func(
+            &[
+                self.cx.type_ptr(),
+                self.cx.type_i64(),
+                self.cx.type_i32(),
+                self.cx.type_i32(),
+                self.cx.type_ptr(),
+            ],
+            self.cx.type_void(),
+        );
+        let args = &[fn_name, hash, bitmap_bytes, bitmap_index, mcdc_temp];
+        let args = self.check_call("call", llty, llfn, args);
+        unsafe {
+            let _ = llvm::LLVMRustBuildCall(
+                self.llbuilder,
+                llty,
+                llfn,
+                args.as_ptr() as *const &llvm::Value,
+                args.len() as c_uint,
+                [].as_ptr(),
+                0 as c_uint,
+            );
+        }
+        let i32_align = self.tcx().data_layout.i32_align.abi;
+        self.store(self.const_i32(0), mcdc_temp, i32_align);
+    }
+
+    pub(crate) fn mcdc_condbitmap_update(
+        &mut self,
+        fn_name: &'ll Value,
+        hash: &'ll Value,
+        cond_loc: &'ll Value,
+        mcdc_temp: &'ll Value,
+        bool_value: &'ll Value,
+    ) {
+        debug!(
+            "mcdc_condbitmap_update() with args ({:?}, {:?}, {:?}, {:?}, {:?})",
+            fn_name, hash, cond_loc, mcdc_temp, bool_value
+        );
+        assert!(llvm_util::get_version() >= (18, 0, 0), "MCDC intrinsics require LLVM 18 or later");
+        let llfn = unsafe { llvm::LLVMRustGetInstrProfMCDCCondBitmapIntrinsic(self.cx().llmod) };
+        let llty = self.cx.type_func(
+            &[
+                self.cx.type_ptr(),
+                self.cx.type_i64(),
+                self.cx.type_i32(),
+                self.cx.type_ptr(),
+                self.cx.type_i1(),
+            ],
+            self.cx.type_void(),
+        );
+        let args = &[fn_name, hash, cond_loc, mcdc_temp, bool_value];
+        self.check_call("call", llty, llfn, args);
+        unsafe {
+            let _ = llvm::LLVMRustBuildCall(
+                self.llbuilder,
+                llty,
+                llfn,
+                args.as_ptr() as *const &llvm::Value,
+                args.len() as c_uint,
+                [].as_ptr(),
+                0 as c_uint,
+            );
+        }
+    }
 }

compiler/rustc_codegen_llvm/src/coverageinfo/mod.rs

@@ -13,7 +13,7 @@ use rustc_codegen_ssa::traits::{
 use rustc_data_structures::fx::{FxHashMap, FxIndexMap};
 use rustc_llvm::RustString;
 use rustc_middle::bug;
-use rustc_middle::mir::coverage::CoverageKind;
+use rustc_middle::mir::coverage::{CoverageKind, FunctionCoverageInfo};
 use rustc_middle::ty::layout::HasTyCtxt;
 use rustc_middle::ty::Instance;
 use rustc_target::abi::Align;
@@ -30,13 +30,15 @@ pub struct CrateCoverageContext<'ll, 'tcx> {
     pub(crate) function_coverage_map:
         RefCell<FxIndexMap<Instance<'tcx>, FunctionCoverageCollector<'tcx>>>,
     pub(crate) pgo_func_name_var_map: RefCell<FxHashMap<Instance<'tcx>, &'ll llvm::Value>>,
+    pub(crate) mcdc_condition_bitmap_map: RefCell<FxHashMap<Instance<'tcx>, &'ll llvm::Value>>,
 }
 
 impl<'ll, 'tcx> CrateCoverageContext<'ll, 'tcx> {
     pub fn new() -> Self {
         Self {
             function_coverage_map: Default::default(),
             pgo_func_name_var_map: Default::default(),
+            mcdc_condition_bitmap_map: Default::default(),
         }
     }
 
@@ -45,6 +47,12 @@ impl<'ll, 'tcx> CrateCoverageContext<'ll, 'tcx> {
     ) -> FxIndexMap<Instance<'tcx>, FunctionCoverageCollector<'tcx>> {
         self.function_coverage_map.replace(FxIndexMap::default())
     }
+
+    /// LLVM use a temp value to record evaluated mcdc test vector of each decision, which is called condition bitmap.
+    /// This value is named `mcdc.addr` (same as clang) and is a 32-bit integer.
+    fn try_get_mcdc_condition_bitmap(&self, instance: &Instance<'tcx>) -> Option<&'ll llvm::Value> {
+        self.mcdc_condition_bitmap_map.borrow().get(instance).copied()
+    }
 }
 
 // These methods used to be part of trait `CoverageInfoMethods`, which no longer
@@ -90,6 +98,10 @@ impl<'tcx> CoverageInfoBuilderMethods<'tcx> for Builder<'_, '_, 'tcx> {
             return;
         };
 
+        if function_coverage_info.mcdc_bitmap_bytes > 0 {
+            ensure_mcdc_parameters(bx, instance, function_coverage_info);
+        }
+
         let Some(coverage_context) = bx.coverage_context() else { return };
         let mut coverage_map = coverage_context.function_coverage_map.borrow_mut();
         let func_coverage = coverage_map
@@ -131,10 +143,66 @@ impl<'tcx> CoverageInfoBuilderMethods<'tcx> for Builder<'_, '_, 'tcx> {
             CoverageKind::ExpressionUsed { id } => {
                 func_coverage.mark_expression_id_seen(id);
             }
+            CoverageKind::CondBitmapUpdate { id, value, .. } => {
+                drop(coverage_map);
+                assert_ne!(
+                    id.as_u32(),
+                    0,
+                    "ConditionId of evaluated conditions should never be zero"
+                );
+                let cond_bitmap = coverage_context
+                    .try_get_mcdc_condition_bitmap(&instance)
+                    .expect("mcdc cond bitmap should have been allocated for updating");
+                let cond_loc = bx.const_i32(id.as_u32() as i32 - 1);
+                let bool_value = bx.const_bool(value);
+                let fn_name = bx.get_pgo_func_name_var(instance);
+                let hash = bx.const_u64(function_coverage_info.function_source_hash);
+                bx.mcdc_condbitmap_update(fn_name, hash, cond_loc, cond_bitmap, bool_value);
+            }
+            CoverageKind::TestVectorBitmapUpdate { bitmap_idx } => {
+                drop(coverage_map);
+                let cond_bitmap = coverage_context
+                                    .try_get_mcdc_condition_bitmap(&instance)
+                                    .expect("mcdc cond bitmap should have been allocated for merging into the global bitmap");
+                let bitmap_bytes = bx.tcx().coverage_ids_info(instance.def).mcdc_bitmap_bytes;
+                assert!(bitmap_idx < bitmap_bytes, "bitmap index of the decision out of range");
+                assert!(
+                    bitmap_bytes <= function_coverage_info.mcdc_bitmap_bytes,
+                    "bitmap length disagreement: query says {bitmap_bytes} but function info only has {}",
+                    function_coverage_info.mcdc_bitmap_bytes
+                );
+
+                let fn_name = bx.get_pgo_func_name_var(instance);
+                let hash = bx.const_u64(function_coverage_info.function_source_hash);
+                let bitmap_bytes = bx.const_u32(bitmap_bytes);
+                let bitmap_index = bx.const_u32(bitmap_idx);
+                bx.mcdc_tvbitmap_update(fn_name, hash, bitmap_bytes, bitmap_index, cond_bitmap);
+            }
         }
     }
 }
 
+fn ensure_mcdc_parameters<'ll, 'tcx>(
+    bx: &mut Builder<'_, 'll, 'tcx>,
+    instance: Instance<'tcx>,
+    function_coverage_info: &FunctionCoverageInfo,
+) {
+    let Some(cx) = bx.coverage_context() else { return };
+    if cx.mcdc_condition_bitmap_map.borrow().contains_key(&instance) {
+        return;
+    }
+
+    let fn_name = bx.get_pgo_func_name_var(instance);
+    let hash = bx.const_u64(function_coverage_info.function_source_hash);
+    let bitmap_bytes = bx.const_u32(function_coverage_info.mcdc_bitmap_bytes);
+    let cond_bitmap = bx.mcdc_parameters(fn_name, hash, bitmap_bytes);
+    bx.coverage_context()
+        .expect("already checked above")
+        .mcdc_condition_bitmap_map
+        .borrow_mut()
+        .insert(instance, cond_bitmap);
+}
+
 /// Calls llvm::createPGOFuncNameVar() with the given function instance's
 /// mangled function name. The LLVM API returns an llvm::GlobalVariable
 /// containing the function name, with the specific variable name and linkage

compiler/rustc_codegen_llvm/src/llvm/ffi.rs

@@ -1631,6 +1631,10 @@ extern "C" {
 
     // Miscellaneous instructions
     pub fn LLVMRustGetInstrProfIncrementIntrinsic(M: &Module) -> &Value;
+    pub fn LLVMRustGetInstrProfMCDCParametersIntrinsic(M: &Module) -> &Value;
+    pub fn LLVMRustGetInstrProfMCDCTVBitmapUpdateIntrinsic(M: &Module) -> &Value;
+    pub fn LLVMRustGetInstrProfMCDCCondBitmapIntrinsic(M: &Module) -> &Value;
+
     pub fn LLVMRustBuildCall<'a>(
         B: &Builder<'a>,
         Ty: &'a Type,

compiler/rustc_llvm/llvm-wrapper/RustWrapper.cpp

@@ -1546,6 +1546,33 @@ extern "C" LLVMValueRef LLVMRustGetInstrProfIncrementIntrinsic(LLVMModuleRef M)
       unwrap(M), llvm::Intrinsic::instrprof_increment));
 }
 
+extern "C" LLVMValueRef LLVMRustGetInstrProfMCDCParametersIntrinsic(LLVMModuleRef M) {
+#if LLVM_VERSION_GE(18, 0)
+  return wrap(llvm::Intrinsic::getDeclaration(
+      unwrap(M), llvm::Intrinsic::instrprof_mcdc_parameters));
+#else
+  report_fatal_error("LLVM 18.0 is required for mcdc intrinsic functions");
+#endif
+}
+
+extern "C" LLVMValueRef LLVMRustGetInstrProfMCDCTVBitmapUpdateIntrinsic(LLVMModuleRef M) {
+#if LLVM_VERSION_GE(18, 0)
+  return wrap(llvm::Intrinsic::getDeclaration(
+      unwrap(M), llvm::Intrinsic::instrprof_mcdc_tvbitmap_update));
+#else
+  report_fatal_error("LLVM 18.0 is required for mcdc intrinsic functions");
+#endif
+}
+
+extern "C" LLVMValueRef LLVMRustGetInstrProfMCDCCondBitmapIntrinsic(LLVMModuleRef M) {
+#if LLVM_VERSION_GE(18, 0)
+  return wrap(llvm::Intrinsic::getDeclaration(
+      unwrap(M), llvm::Intrinsic::instrprof_mcdc_condbitmap_update));
+#else
+  report_fatal_error("LLVM 18.0 is required for mcdc intrinsic functions");
+#endif
+}
+
 extern "C" LLVMValueRef LLVMRustBuildMemCpy(LLVMBuilderRef B,
                                             LLVMValueRef Dst, unsigned DstAlign,
                                             LLVMValueRef Src, unsigned SrcAlign,

compiler/rustc_middle/src/mir/query.rs

@@ -361,4 +361,8 @@ pub struct CoverageIdsInfo {
     /// InstrumentCoverage MIR pass, if the highest-numbered counter increments
     /// were removed by MIR optimizations.
     pub max_counter_id: mir::coverage::CounterId,
+
+    /// Coverage codegen for mcdc needs to know the size of the global bitmap so that it can
+    /// set the `bytemap-bytes` argument of the `llvm.instrprof.mcdc.tvbitmap.update` intrinsic.
+    pub mcdc_bitmap_bytes: u32,
 }

compiler/rustc_mir_transform/src/coverage/query.rs

@@ -61,7 +61,17 @@ fn coverage_ids_info<'tcx>(
         .max()
         .unwrap_or(CounterId::ZERO);
 
-    CoverageIdsInfo { max_counter_id }
+    let mcdc_bitmap_bytes = mir_body
+        .coverage_branch_info
+        .as_deref()
+        .map(|info| {
+            info.mcdc_decision_spans
+                .iter()
+                .fold(0, |acc, decision| acc + (1_u32 << decision.conditions_num).div_ceil(8))
+        })
+        .unwrap_or_default();
+
+    CoverageIdsInfo { max_counter_id, mcdc_bitmap_bytes }
 }
 
 fn all_coverage_in_mir_body<'a, 'tcx>(