Populating LLVM stack maps into MSM/FSM.

Cargo.lock

@@ -1495,6 +1495,7 @@ dependencies = [
 name = "wasmer-llvm-backend"
 version = "0.5.6"
 dependencies = [
+ "byteorder 1.3.1 (registry+https://github.com/rust-lang/crates.io-index)",
  "capstone 0.5.0 (registry+https://github.com/rust-lang/crates.io-index)",
  "cc 1.0.37 (registry+https://github.com/rust-lang/crates.io-index)",
  "goblin 0.0.20 (registry+https://github.com/rust-lang/crates.io-index)",

lib/llvm-backend/src/backend.rs

@@ -244,7 +244,7 @@ impl LLVMBackend {
     pub fn new(
         module: Module,
         _intrinsics: Intrinsics,
-        _stackmaps: &StackmapRegistry,
+        stackmaps: &StackmapRegistry,
     ) -> (Self, LLVMCache) {
         Target::initialize_x86(&InitializationConfig {
             asm_parser: true,
@@ -303,8 +303,36 @@ impl LLVMBackend {
             )
         };
         if raw_stackmap.len() > 0 {
-            let map = stackmap::StackMap::parse(raw_stackmap);
+            let map = stackmap::StackMap::parse(raw_stackmap).unwrap();
             eprintln!("{:?}", map);
+
+            let (code_ptr, code_size) = unsafe {
+                (
+                    llvm_backend_get_code_ptr(module),
+                    llvm_backend_get_code_size(module),
+                )
+            };
+            let mut msm = ModuleStateMap {
+                local_functions: Default::default(),
+                total_size: code_size,
+            };
+            let mut map_record_idx: usize = 0;
+            for size_record in &map.stk_size_records {
+                for _ in 0..size_record.record_count {
+                    let map_record = &map.stk_map_records[map_record_idx];
+                    let map_entry = &stackmaps.entries[map_record_idx];
+                    assert_eq!(map_record.patchpoint_id, map_record_idx as u64);
+                    map_record_idx += 1;
+
+                    map_entry.populate_msm(
+                        code_ptr as usize,
+                        &map,
+                        size_record,
+                        map_record,
+                        &mut msm,
+                    );
+                }
+            }
         } else {
             eprintln!("WARNING: No stack map");
         }

lib/llvm-backend/src/code.rs

@@ -27,7 +27,7 @@ use wasmparser::{BinaryReaderError, MemoryImmediate, Operator, Type as WpType};
 use crate::backend::LLVMBackend;
 use crate::intrinsics::{CtxType, GlobalCache, Intrinsics, MemoryCache};
 use crate::read_info::{blocktype_to_type, type_to_type};
-use crate::stackmap::{StackmapEntry, StackmapEntryKind, StackmapRegistry};
+use crate::stackmap::{StackmapEntry, StackmapEntryKind, StackmapRegistry, ValueSemantic};
 use crate::state::{ControlFrame, IfElseState, State};
 use crate::trampolines::generate_trampolines;
 
@@ -310,6 +310,7 @@ fn emit_stack_map(
     kind: StackmapEntryKind,
     locals: &[PointerValue],
     state: &State,
+    opcode_offset: usize,
 ) {
     let stackmap_id = target.entries.len();
 
@@ -324,9 +325,13 @@ fn emit_stack_map(
     params.push(intrinsics.i32_ty.const_int(0, false).as_basic_value_enum());
 
     let locals: Vec<_> = locals.iter().map(|x| x.as_basic_value_enum()).collect();
+    let mut value_semantics: Vec<ValueSemantic> = vec![];
 
     params.extend_from_slice(&locals);
+    value_semantics.extend((0..locals.len()).map(ValueSemantic::WasmLocal));
+
     params.extend_from_slice(&state.stack);
+    value_semantics.extend((0..state.stack.len()).map(ValueSemantic::WasmStack));
 
     builder.build_call(intrinsics.experimental_stackmap, &params, &state.var_name());
 
@@ -335,6 +340,8 @@ fn emit_stack_map(
         local_function_id,
         local_count: locals.len(),
         stack_count: state.stack.len(),
+        opcode_offset,
+        value_semantics,
     });
 }
 
@@ -371,6 +378,7 @@ pub struct LLVMFunctionCodeGenerator {
     unreachable_depth: usize,
     stackmaps: Rc<RefCell<StackmapRegistry>>,
     index: usize,
+    opcode_offset: usize,
 }
 
 impl FunctionCodeGenerator<CodegenError> for LLVMFunctionCodeGenerator {
@@ -443,8 +451,13 @@ impl FunctionCodeGenerator<CodegenError> for LLVMFunctionCodeGenerator {
     }
 
     fn feed_event(&mut self, event: Event, module_info: &ModuleInfo) -> Result<(), CodegenError> {
+        let mut opcode_offset: Option<usize> = None;
         let op = match event {
-            Event::Wasm(x) => x,
+            Event::Wasm(x) => {
+                opcode_offset = Some(self.opcode_offset);
+                self.opcode_offset += 1;
+                x
+            }
             Event::Internal(_x) => {
                 return Ok(());
             }
@@ -530,7 +543,7 @@ impl FunctionCodeGenerator<CodegenError> for LLVMFunctionCodeGenerator {
 
                 builder.position_at_end(&loop_body);
 
-                {
+                if let Some(offset) = opcode_offset {
                     let mut stackmaps = self.stackmaps.borrow_mut();
                     emit_stack_map(
                         intrinsics,
@@ -540,6 +553,7 @@ impl FunctionCodeGenerator<CodegenError> for LLVMFunctionCodeGenerator {
                         StackmapEntryKind::Loop,
                         &self.locals,
                         state,
+                        offset,
                     )
                 }
 
@@ -804,7 +818,7 @@ impl FunctionCodeGenerator<CodegenError> for LLVMFunctionCodeGenerator {
                 // If llvm cannot prove that this is never touched,
                 // it will emit a `ud2` instruction on x86_64 arches.
 
-                {
+                if let Some(offset) = opcode_offset {
                     let mut stackmaps = self.stackmaps.borrow_mut();
                     emit_stack_map(
                         intrinsics,
@@ -814,6 +828,7 @@ impl FunctionCodeGenerator<CodegenError> for LLVMFunctionCodeGenerator {
                         StackmapEntryKind::Trappable,
                         &self.locals,
                         state,
+                        offset,
                     )
                 }
 
@@ -954,7 +969,7 @@ impl FunctionCodeGenerator<CodegenError> for LLVMFunctionCodeGenerator {
                 };
 
                 state.popn(func_sig.params().len())?;
-                {
+                if let Some(offset) = opcode_offset {
                     let mut stackmaps = self.stackmaps.borrow_mut();
                     emit_stack_map(
                         intrinsics,
@@ -964,6 +979,7 @@ impl FunctionCodeGenerator<CodegenError> for LLVMFunctionCodeGenerator {
                         StackmapEntryKind::Call,
                         &self.locals,
                         state,
+                        offset,
                     )
                 }
                 let call_site = builder.build_call(func_ptr, &params, &state.var_name());
@@ -1131,7 +1147,7 @@ impl FunctionCodeGenerator<CodegenError> for LLVMFunctionCodeGenerator {
                     "typed_func_ptr",
                 );
 
-                {
+                if let Some(offset) = opcode_offset {
                     let mut stackmaps = self.stackmaps.borrow_mut();
                     emit_stack_map(
                         intrinsics,
@@ -1141,6 +1157,7 @@ impl FunctionCodeGenerator<CodegenError> for LLVMFunctionCodeGenerator {
                         StackmapEntryKind::Call,
                         &self.locals,
                         state,
+                        offset,
                     )
                 }
                 let call_site = builder.build_call(typed_func_ptr, &args, "indirect_call");
@@ -2565,6 +2582,20 @@ impl ModuleCodeGenerator<LLVMFunctionCodeGenerator, LLVMBackend, CodegenError>
 
         let local_func_index = self.functions.len();
 
+        {
+            let mut stackmaps = self.stackmaps.borrow_mut();
+            emit_stack_map(
+                &intrinsics,
+                &builder,
+                local_func_index,
+                &mut *stackmaps,
+                StackmapEntryKind::FunctionHeader,
+                &locals,
+                &state,
+                ::std::usize::MAX,
+            );
+        }
+
         let code = LLVMFunctionCodeGenerator {
             state,
             context: Some(context),
@@ -2579,6 +2610,7 @@ impl ModuleCodeGenerator<LLVMFunctionCodeGenerator, LLVMBackend, CodegenError>
             unreachable_depth: 0,
             stackmaps: self.stackmaps.clone(),
             index: local_func_index,
+            opcode_offset: 0,
         };
         self.functions.push(code);
         Ok(self.functions.last_mut().unwrap())

lib/llvm-backend/src/stackmap.rs

@@ -1,7 +1,13 @@
 // https://llvm.org/docs/StackMaps.html#stackmap-section
 
 use byteorder::{LittleEndian, ReadBytesExt};
+use std::collections::HashMap;
 use std::io::{self, Cursor};
+use wasmer_runtime_core::state::{
+    x64::{new_machine_state, X64Register, GPR},
+    FunctionStateMap, MachineStateDiff, MachineValue, ModuleStateMap, OffsetInfo, RegisterIndex,
+    SuspendOffset, WasmAbstractValue,
+};
 
 #[derive(Default, Debug, Clone)]
 pub struct StackmapRegistry {
@@ -12,17 +18,231 @@ pub struct StackmapRegistry {
 pub struct StackmapEntry {
     pub kind: StackmapEntryKind,
     pub local_function_id: usize,
+    pub opcode_offset: usize,
+    pub value_semantics: Vec<ValueSemantic>,
     pub local_count: usize,
     pub stack_count: usize,
 }
 
+#[derive(Debug, Clone)]
+pub enum ValueSemantic {
+    WasmLocal(usize),
+    WasmStack(usize),
+}
+
 #[derive(Debug, Clone, Copy)]
 pub enum StackmapEntryKind {
+    FunctionHeader,
     Loop,
     Call,
     Trappable,
 }
 
+/*
+pub struct FunctionStateMap {
+    pub initial: MachineState,
+    pub local_function_id: usize,
+    pub locals: Vec<WasmAbstractValue>,
+    pub shadow_size: usize, // for single-pass backend, 32 bytes on x86-64
+    pub diffs: Vec<MachineStateDiff>,
+    pub wasm_function_header_target_offset: Option<SuspendOffset>,
+    pub wasm_offset_to_target_offset: BTreeMap<usize, SuspendOffset>,
+    pub loop_offsets: BTreeMap<usize, OffsetInfo>, /* suspend_offset -> info */
+    pub call_offsets: BTreeMap<usize, OffsetInfo>, /* suspend_offset -> info */
+    pub trappable_offsets: BTreeMap<usize, OffsetInfo>, /* suspend_offset -> info */
+}
+pub struct MachineStateDiff {
+    pub last: Option<usize>,
+    pub stack_push: Vec<MachineValue>,
+    pub stack_pop: usize,
+    pub reg_diff: Vec<(RegisterIndex, MachineValue)>,
+
+    pub wasm_stack_push: Vec<WasmAbstractValue>,
+    pub wasm_stack_pop: usize,
+    pub wasm_stack_private_depth: usize, // absolute value; not a diff.
+
+    pub wasm_inst_offset: usize, // absolute value; not a diff.
+}
+*/
+
+impl StackmapEntry {
+    pub fn populate_msm(
+        &self,
+        code_addr: usize,
+        llvm_map: &StackMap,
+        size_record: &StkSizeRecord,
+        map_record: &StkMapRecord,
+        msm: &mut ModuleStateMap,
+    ) {
+        #[derive(Copy, Clone, Debug)]
+        enum RuntimeOrConstant {
+            Runtime(MachineValue),
+            Constant(u64),
+        }
+
+        let fsm = msm
+            .local_functions
+            .entry(self.local_function_id)
+            .or_insert_with(|| {
+                FunctionStateMap::new(new_machine_state(), self.local_function_id, 0, vec![])
+            });
+
+        assert_eq!(self.value_semantics.len(), map_record.locations.len());
+        assert!(size_record.stack_size % 8 == 0);
+
+        let mut machine_stack_layout: Vec<MachineValue> =
+            vec![MachineValue::Undefined; (size_record.stack_size as usize) / 8];
+        let mut regs: Vec<(RegisterIndex, MachineValue)> = vec![];
+        let mut stack_constants: HashMap<usize, u64> = HashMap::new();
+
+        let mut wasm_locals: Vec<WasmAbstractValue> = vec![];
+        let mut wasm_stack: Vec<WasmAbstractValue> = vec![];
+
+        for (i, loc) in map_record.locations.iter().enumerate() {
+            let mv = match self.value_semantics[i] {
+                ValueSemantic::WasmLocal(x) => {
+                    if x != wasm_locals.len() {
+                        panic!("unordered local values");
+                    }
+                    wasm_locals.push(WasmAbstractValue::Runtime);
+                    MachineValue::WasmLocal(x)
+                }
+                ValueSemantic::WasmStack(x) => {
+                    if x != wasm_stack.len() {
+                        panic!("unordered stack values");
+                    }
+                    wasm_stack.push(WasmAbstractValue::Runtime);
+                    MachineValue::WasmStack(x)
+                }
+            };
+            match loc.ty {
+                LocationType::Register => {
+                    let index = X64Register::from_dwarf_regnum(loc.dwarf_regnum)
+                        .expect("invalid regnum")
+                        .to_index();
+                    regs.push((index, mv));
+                }
+                LocationType::Constant => {
+                    let v = loc.offset_or_small_constant as u32 as u64;
+                    match mv {
+                        MachineValue::WasmStack(x) => {
+                            stack_constants.insert(x, v);
+                            *wasm_stack.last_mut().unwrap() = WasmAbstractValue::Const(v);
+                        }
+                        _ => {} // TODO
+                    }
+                }
+                LocationType::ConstantIndex => {
+                    let v =
+                        llvm_map.constants[loc.offset_or_small_constant as usize].large_constant;
+                    match mv {
+                        MachineValue::WasmStack(x) => {
+                            stack_constants.insert(x, v);
+                            *wasm_stack.last_mut().unwrap() = WasmAbstractValue::Const(v);
+                        }
+                        _ => {} // TODO
+                    }
+                }
+                LocationType::Direct => match mv {
+                    MachineValue::WasmLocal(_) => {
+                        assert_eq!(loc.location_size, 8);
+                        assert!(loc.offset_or_small_constant < 0);
+                        assert!(
+                            X64Register::from_dwarf_regnum(loc.dwarf_regnum).unwrap()
+                                == X64Register::GPR(GPR::RBP)
+                        );
+                        let stack_offset = ((-loc.offset_or_small_constant) % 8) as usize;
+                        assert!(stack_offset > 0 && stack_offset <= machine_stack_layout.len());
+                        machine_stack_layout[stack_offset - 1] = mv;
+                    }
+                    _ => unreachable!(
+                        "Direct location type is not expected for values other than local"
+                    ),
+                },
+                LocationType::Indirect => {
+                    assert_eq!(loc.location_size, 8);
+                    assert!(loc.offset_or_small_constant < 0);
+                    assert!(
+                        X64Register::from_dwarf_regnum(loc.dwarf_regnum).unwrap()
+                            == X64Register::GPR(GPR::RBP)
+                    );
+                    let stack_offset = ((-loc.offset_or_small_constant) % 8) as usize;
+                    assert!(stack_offset > 0 && stack_offset <= machine_stack_layout.len());
+                    machine_stack_layout[stack_offset - 1] = mv;
+                }
+            }
+        }
+
+        assert_eq!(wasm_stack.len(), self.stack_count);
+        assert_eq!(wasm_locals.len(), self.local_count);
+
+        let diff = MachineStateDiff {
+            last: None,
+            stack_push: machine_stack_layout,
+            stack_pop: 0,
+            reg_diff: regs,
+            wasm_stack_push: wasm_stack,
+            wasm_stack_pop: 0,
+            wasm_stack_private_depth: 0,
+            wasm_inst_offset: self.opcode_offset,
+        };
+        let diff_id = fsm.diffs.len();
+        fsm.diffs.push(diff);
+
+        match self.kind {
+            StackmapEntryKind::FunctionHeader => {
+                fsm.locals = wasm_locals;
+            }
+            _ => {
+                assert_eq!(fsm.locals, wasm_locals);
+            }
+        }
+        let target_offset = (size_record.function_address as usize)
+            .checked_sub(code_addr)
+            .unwrap()
+            + map_record.instruction_offset as usize;
+
+        match self.kind {
+            StackmapEntryKind::Loop => {
+                fsm.wasm_offset_to_target_offset
+                    .insert(self.opcode_offset, SuspendOffset::Loop(target_offset));
+                fsm.loop_offsets.insert(
+                    target_offset,
+                    OffsetInfo {
+                        diff_id,
+                        activate_offset: target_offset,
+                    },
+                );
+            }
+            StackmapEntryKind::Call => {
+                fsm.wasm_offset_to_target_offset
+                    .insert(self.opcode_offset, SuspendOffset::Call(target_offset));
+                fsm.call_offsets.insert(
+                    target_offset,
+                    OffsetInfo {
+                        diff_id,
+                        activate_offset: target_offset,
+                    },
+                );
+            }
+            StackmapEntryKind::Trappable => {
+                fsm.wasm_offset_to_target_offset
+                    .insert(self.opcode_offset, SuspendOffset::Trappable(target_offset));
+                fsm.trappable_offsets.insert(
+                    target_offset,
+                    OffsetInfo {
+                        diff_id,
+                        activate_offset: target_offset,
+                    },
+                );
+            }
+            StackmapEntryKind::FunctionHeader => {
+                fsm.wasm_function_header_target_offset = Some(SuspendOffset::Loop(target_offset));
+            }
+        }
+    }
+}
+
 #[derive(Clone, Debug, Default)]
 pub struct StackMap {
     pub version: u8,

lib/runtime-core/src/state.rs

@@ -842,6 +842,7 @@ pub mod x64 {
         XMM7,
     }
 
+    #[derive(Copy, Clone, Debug, Eq, PartialEq)]
     pub enum X64Register {
         GPR(GPR),
         XMM(XMM),
@@ -854,5 +855,36 @@ pub mod x64 {
                 X64Register::XMM(x) => RegisterIndex(x as usize + 16),
             }
         }
+
+        pub fn from_dwarf_regnum(x: u16) -> Option<X64Register> {
+            Some(match x {
+                0 => X64Register::GPR(GPR::RAX),
+                1 => X64Register::GPR(GPR::RDX),
+                2 => X64Register::GPR(GPR::RCX),
+                3 => X64Register::GPR(GPR::RBX),
+                4 => X64Register::GPR(GPR::RSI),
+                5 => X64Register::GPR(GPR::RDI),
+                6 => X64Register::GPR(GPR::RBP),
+                7 => X64Register::GPR(GPR::RSP),
+                8 => X64Register::GPR(GPR::R8),
+                9 => X64Register::GPR(GPR::R9),
+                10 => X64Register::GPR(GPR::R10),
+                11 => X64Register::GPR(GPR::R11),
+                12 => X64Register::GPR(GPR::R12),
+                13 => X64Register::GPR(GPR::R13),
+                14 => X64Register::GPR(GPR::R14),
+                15 => X64Register::GPR(GPR::R15),
+
+                17 => X64Register::XMM(XMM::XMM0),
+                18 => X64Register::XMM(XMM::XMM1),
+                19 => X64Register::XMM(XMM::XMM2),
+                20 => X64Register::XMM(XMM::XMM3),
+                21 => X64Register::XMM(XMM::XMM4),
+                22 => X64Register::XMM(XMM::XMM5),
+                23 => X64Register::XMM(XMM::XMM6),
+                24 => X64Register::XMM(XMM::XMM7),
+                _ => return None,
+            })
+        }
     }
 }