wasmer_compiler_cranelift/translator/code_translator/

bounds_checks.rs

//! Implementation of Wasm to CLIF memory access translation.
//!
//! Given
//!
//! * a dynamic Wasm memory index operand,
//! * a static offset immediate, and
//! * a static access size,
//!
//! bounds check the memory access and translate it into a native memory access.
//!
//! !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
//! !!!                                                                      !!!
//! !!!    THIS CODE IS VERY SUBTLE, HAS MANY SPECIAL CASES, AND IS ALSO     !!!
//! !!!   ABSOLUTELY CRITICAL FOR MAINTAINING THE SAFETY OF THE WASM HEAP    !!!
//! !!!                             SANDBOX.                                 !!!
//! !!!                                                                      !!!
//! !!!    A good rule of thumb is to get two reviews on any substantive     !!!
//! !!!                         changes in here.                             !!!
//! !!!                                                                      !!!
//! !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

use super::Reachability;
use crate::{
    func_environ::FuncEnvironment,
    heap::{HeapData, HeapStyle},
    translator::materialize_global_value,
};
use Reachability::*;
use cranelift_codegen::{
    cursor::{Cursor, FuncCursor},
    ir::{self, InstBuilder, RelSourceLoc, condcodes::IntCC},
};
use cranelift_frontend::FunctionBuilder;
use wasmer_types::WasmResult;

/// Helper used to emit bounds checks (as necessary) and compute the native
/// address of a heap access.
///
/// Returns the `ir::Value` holding the native address of the heap access, or
/// `None` if the heap access will unconditionally trap.
pub fn bounds_check_and_compute_addr(
    builder: &mut FunctionBuilder,
    env: &mut FuncEnvironment<'_>,
    heap: &HeapData,
    // Dynamic operand indexing into the heap.
    index: ir::Value,
    // Static immediate added to the index.
    offset: u32,
    // Static size of the heap access.
    access_size: u8,
) -> WasmResult<Reachability<ir::Value>> {
    let index = cast_index_to_pointer_ty(
        index,
        heap.index_type,
        env.pointer_type(),
        &mut builder.cursor(),
    );
    let offset_and_size = offset_plus_size(offset, access_size);
    let spectre_mitigations_enabled = env.heap_access_spectre_mitigation();

    let host_page_size_log2 = env.target_config().page_size_align_log2;
    let can_use_virtual_memory = heap.page_size_log2 >= host_page_size_log2;

    let make_compare =
        |builder: &mut FunctionBuilder, compare_kind: IntCC, lhs: ir::Value, rhs: ir::Value| {
            builder.ins().icmp(compare_kind, lhs, rhs)
        };

    // We need to emit code that will trap (or compute an address that will trap
    // when accessed) if
    //
    //     index + offset + access_size > bound
    //
    // or if the `index + offset + access_size` addition overflows.
    //
    // Note that we ultimately want a 64-bit integer (we only target 64-bit
    // architectures at the moment) and that `offset` is a `u32` and
    // `access_size` is a `u8`. This means that we can add the latter together
    // as `u64`s without fear of overflow, and we only have to be concerned with
    // whether adding in `index` will overflow.
    //
    // Finally, the following right-hand sides of the matches do have a little
    // bit of duplicated code across them, but I think writing it this way is
    // worth it for readability and seeing very clearly each of our cases for
    // different bounds checks and optimizations of those bounds checks. It is
    // intentionally written in a straightforward case-matching style that will
    // hopefully make it easy to port to ISLE one day.
    Ok(match heap.style {
        // ====== Dynamic Memories ======
        //
        // 1. First special case for when `offset + access_size == 1`:
        //
        //            index + 1 > bound
        //        ==> index >= bound
        HeapStyle::Dynamic { .. } if offset_and_size == 1 => {
            let bound = get_dynamic_heap_bound(builder, env, heap);
            let oob = make_compare(builder, IntCC::UnsignedGreaterThanOrEqual, index, bound);
            Reachable(explicit_check_oob_condition_and_compute_addr(
                &mut builder.cursor(),
                heap,
                env.pointer_type(),
                index,
                offset,
                spectre_mitigations_enabled,
                oob,
            ))
        }

        // 2. Second special case for when we know that there are enough guard
        //    pages to cover the offset and access size.
        //
        //    The precise should-we-trap condition is
        //
        //        index + offset + access_size > bound
        //
        //    However, if we instead check only the partial condition
        //
        //        index > bound
        //
        //    then the most out of bounds that the access can be, while that
        //    partial check still succeeds, is `offset + access_size`.
        //
        //    However, when we have a guard region that is at least as large as
        //    `offset + access_size`, we can rely on the virtual memory
        //    subsystem handling these out-of-bounds errors at
        //    runtime. Therefore, the partial `index > bound` check is
        //    sufficient for this heap configuration.
        //
        //    Additionally, this has the advantage that a series of Wasm loads
        //    that use the same dynamic index operand but different static
        //    offset immediates -- which is a common code pattern when accessing
        //    multiple fields in the same struct that is in linear memory --
        //    will all emit the same `index > bound` check, which we can GVN.
        HeapStyle::Dynamic { .. }
            if can_use_virtual_memory && offset_and_size <= heap.offset_guard_size =>
        {
            let bound = get_dynamic_heap_bound(builder, env, heap);
            let oob = make_compare(builder, IntCC::UnsignedGreaterThan, index, bound);
            Reachable(explicit_check_oob_condition_and_compute_addr(
                &mut builder.cursor(),
                heap,
                env.pointer_type(),
                index,
                offset,
                spectre_mitigations_enabled,
                oob,
            ))
        }

        // 3. Third special case for when `offset + access_size <= min_size`.
        //
        //    We know that `bound >= min_size`, so we can do the following
        //    comparison, without fear of the right-hand side wrapping around:
        //
        //            index + offset + access_size > bound
        //        ==> index > bound - (offset + access_size)
        HeapStyle::Dynamic { .. } if offset_and_size <= heap.min_size => {
            let bound = get_dynamic_heap_bound(builder, env, heap);
            let adjustment = offset_and_size as i64;
            let adjustment_value = builder.ins().iconst(env.pointer_type(), adjustment);
            let adjusted_bound = builder.ins().isub(bound, adjustment_value);
            let oob = make_compare(builder, IntCC::UnsignedGreaterThan, index, adjusted_bound);
            Reachable(explicit_check_oob_condition_and_compute_addr(
                &mut builder.cursor(),
                heap,
                env.pointer_type(),
                index,
                offset,
                spectre_mitigations_enabled,
                oob,
            ))
        }

        // 4. General case for dynamic memories:
        //
        //        index + offset + access_size > bound
        //
        //    And we have to handle the overflow case in the left-hand side.
        HeapStyle::Dynamic { .. } => {
            let access_size_val = builder
                .ins()
                // Explicit cast from u64 to i64: we just want the raw
                // bits, and iconst takes an `Imm64`.
                .iconst(env.pointer_type(), offset_and_size as i64);
            let adjusted_index = builder.ins().uadd_overflow_trap(
                index,
                access_size_val,
                ir::TrapCode::HEAP_OUT_OF_BOUNDS,
            );
            let bound = get_dynamic_heap_bound(builder, env, heap);
            let oob = make_compare(builder, IntCC::UnsignedGreaterThan, adjusted_index, bound);
            Reachable(explicit_check_oob_condition_and_compute_addr(
                &mut builder.cursor(),
                heap,
                env.pointer_type(),
                index,
                offset,
                spectre_mitigations_enabled,
                oob,
            ))
        }

        // ====== Static Memories ======
        //
        // Static memories reserve the full wasm32 address space plus the offset
        // guard up front: omit explicit bounds checks and rely on virtual memory
        // protection to trap out-of-bounds accesses.
        HeapStyle::Static => {
            assert!(
                can_use_virtual_memory,
                "static memories require the ability to use virtual memory"
            );
            Reachable(compute_addr(
                &mut builder.cursor(),
                heap,
                env.pointer_type(),
                index,
                offset,
            ))
        }
    })
}

/// Get the bound of a dynamic heap as an `ir::Value`.
fn get_dynamic_heap_bound(
    builder: &mut FunctionBuilder,
    env: &mut FuncEnvironment<'_>,
    heap: &HeapData,
) -> ir::Value {
    match (heap.max_size, &heap.style) {
        // The heap has a constant size, no need to actually load the bound.
        (Some(max_size), HeapStyle::Dynamic { .. }) if heap.min_size == max_size => {
            builder.ins().iconst(env.pointer_type(), max_size as i64)
        }
        // Load the heap bound from its global variable.
        (_, HeapStyle::Dynamic { bound_gv }) => {
            materialize_global_value(&mut builder.cursor(), env.pointer_type(), *bound_gv)
        }
        (_, HeapStyle::Static) => unreachable!("not a dynamic heap"),
    }
}

fn cast_index_to_pointer_ty(
    index: ir::Value,
    index_ty: ir::Type,
    pointer_ty: ir::Type,
    pos: &mut FuncCursor,
) -> ir::Value {
    if index_ty == pointer_ty {
        return index;
    }
    // Note that using 64-bit heaps on a 32-bit host is not currently supported,
    // would require at least a bounds check here to ensure that the truncation
    // from 64-to-32 bits doesn't lose any upper bits. For now though we're
    // mostly interested in the 32-bit-heaps-on-64-bit-hosts cast.
    assert!(index_ty.bits() < pointer_ty.bits());

    // Convert `index` to `addr_ty`.
    let extended_index = pos.ins().uextend(pointer_ty, index);

    // Add debug value-label alias so that debuginfo can name the extended
    // value as the address
    let loc = pos.srcloc();
    let loc = RelSourceLoc::from_base_offset(pos.func.params.base_srcloc(), loc);
    pos.func
        .stencil
        .dfg
        .add_value_label_alias(extended_index, loc, index);

    extended_index
}

/// Emit explicit checks on the given out-of-bounds condition for the Wasm
/// address and return the native address.
///
/// This function deduplicates explicit bounds checks and Spectre mitigations
/// that inherently also implement bounds checking.
#[allow(clippy::too_many_arguments)]
fn explicit_check_oob_condition_and_compute_addr(
    pos: &mut FuncCursor,
    heap: &HeapData,
    addr_ty: ir::Type,
    index: ir::Value,
    offset: u32,
    // Whether Spectre mitigations are enabled for heap accesses.
    spectre_mitigations_enabled: bool,
    // The `i8` boolean value that is non-zero when the heap access is out of
    // bounds (and therefore we should trap) and is zero when the heap access is
    // in bounds (and therefore we can proceed).
    oob_condition: ir::Value,
) -> ir::Value {
    if !spectre_mitigations_enabled {
        pos.ins()
            .trapnz(oob_condition, ir::TrapCode::HEAP_OUT_OF_BOUNDS);
    }

    let mut addr = compute_addr(pos, heap, addr_ty, index, offset);

    if spectre_mitigations_enabled {
        let null = pos.ins().iconst(addr_ty, 0);
        addr = pos.ins().select_spectre_guard(oob_condition, null, addr);
    }

    addr
}

/// Emit code for the native address computation of a Wasm address,
/// without any bounds checks or overflow checks.
///
/// It is the caller's responsibility to ensure that any necessary bounds and
/// overflow checks are emitted, and that the resulting address is never used
/// unless they succeed.
fn compute_addr(
    pos: &mut FuncCursor,
    heap: &HeapData,
    addr_ty: ir::Type,
    index: ir::Value,
    offset: u32,
) -> ir::Value {
    debug_assert_eq!(pos.func.dfg.value_type(index), addr_ty);

    let heap_base = materialize_global_value(pos, addr_ty, heap.base);

    let base_and_index = pos.ins().iadd(heap_base, index);

    if offset == 0 {
        base_and_index
    } else {
        // NB: The addition of the offset immediate must happen *before* the
        // `select_spectre_guard`, if any. If it happens after, then we
        // potentially are letting speculative execution read the whole first
        // 4GiB of memory.
        let offset_val = pos.ins().iconst(addr_ty, i64::from(offset));

        pos.ins().iadd(base_and_index, offset_val)
    }
}

#[inline]
fn offset_plus_size(offset: u32, size: u8) -> u64 {
    // Cannot overflow because we are widening to `u64`.
    offset as u64 + size as u64
}