wasmer_vm/libcalls/eh/dwarf/

eh.rs

//! Parsing of GCC-style Language-Specific Data Area (LSDA)
//! For details se*const ():
//!  * <https://refspecs.linuxfoundation.org/LSB_3.0.0/LSB-PDA/LSB-PDA/ehframechpt.html>
//!  * <https://refspecs.linuxfoundation.org/LSB_5.0.0/LSB-Core-generic/LSB-Core-generic/dwarfext.html>
//!  * <https://itanium-cxx-abi.github.io/cxx-abi/exceptions.pdf>
//!  * <https://www.airs.com/blog/archives/460>
//!  * <https://www.airs.com/blog/archives/464>
//!
//! A reference implementation may be found in the GCC source tree
//! (`<root>/libgcc/unwind-c.c` as of this writing).

#![allow(non_upper_case_globals)]
#![allow(clippy::transmutes_expressible_as_ptr_casts)]
#![allow(clippy::comparison_chain)]
#![allow(unused)]

use core::{mem, ptr};

use gimli::DwEhPe;

use super::DwarfReader;

#[derive(Copy, Clone)]
pub struct EHContext<'a> {
    pub ip: *const u8,                             // Current instruction pointer
    pub func_start: *const u8,                     // Pointer to the current function
    pub get_text_start: &'a dyn Fn() -> *const u8, // Get pointer to the code section
    pub get_data_start: &'a dyn Fn() -> *const u8, // Get pointer to the data section
}

/// Landing pad.
type LPad = *const u8;

#[derive(Debug, Clone)]
pub enum EHAction {
    None,
    CatchAll { lpad: LPad },
    CatchSpecific { lpad: LPad, tags: Vec<u32> },
    CatchSpecificOrAll { lpad: LPad, tags: Vec<u32> },
    Terminate,
}

/// 32-bit ARM Darwin platforms uses SjLj exceptions.
///
/// The exception is watchOS armv7k (specifically that subarchitecture), which
/// instead uses DWARF Call Frame Information (CFI) unwinding.
///
/// <https://github.com/llvm/llvm-project/blob/llvmorg-18.1.4/clang/lib/Driver/ToolChains/Darwin.cpp#L3107-L3119>
pub const USING_SJLJ_EXCEPTIONS: bool = cfg!(all(
    target_vendor = "apple",
    not(target_os = "watchos"),
    target_arch = "arm"
));

/* change to true to enable logging from the personality function */
macro_rules! log {
    ($e: expr) => {
        if false {
            eprintln!($e)
        }

    };

    ($($e: expr),*) => {
        if false {
            eprintln!($($e),*)
        }

    };
}

pub unsafe fn find_eh_action(lsda: *const u8, context: &EHContext<'_>) -> Result<EHAction, ()> {
    if lsda.is_null() {
        log!("(pers) LSDA is null for IP {:?}", context.ip);
        return Ok(EHAction::None);
    }

    log!("(pers) Analysing LSDA at {lsda:?}");

    let func_start = context.func_start;
    let mut reader = DwarfReader::new(lsda);

    let lpad_start_encoding = unsafe { DwEhPe(reader.read::<u8>()) };
    log!("(pers) Read LP start encoding {lpad_start_encoding:?}");

    let lpad_base = unsafe {
        // base address for landing pad offsets
        if lpad_start_encoding != gimli::DW_EH_PE_omit {
            read_encoded_pointer(&mut reader, context, lpad_start_encoding)?
        } else {
            log!("(pers) (is omit)");
            func_start
        }
    };
    log!("(pers) read landingpad base: {lpad_base:?}");

    let types_table_encoding = unsafe { DwEhPe(reader.read::<u8>()) };
    log!("(pers) read ttype encoding: {types_table_encoding:?}");

    // If no value for types_table_encoding was given it means that there's no
    // types_table, therefore we can't possibly use this lpad.
    if types_table_encoding == gimli::DW_EH_PE_omit {
        log!("(pers) ttype is omit, returning None");
        return Ok(EHAction::None);
    }

    let types_table_base_offset = unsafe { reader.read_uleb128() };

    let types_table_base = unsafe {
        log!("(pers) read class_info offset {types_table_base_offset:?}");
        reader.ptr.wrapping_add(types_table_base_offset as _)
    };
    log!("(pers) read types_table_base sits at offset {types_table_base:?}");

    let call_site_table_encoding = unsafe { DwEhPe(reader.read::<u8>()) };
    log!("(pers) read call_site_table_encoding is {call_site_table_encoding:?}");

    let call_site_table_size = unsafe { reader.read_uleb128() };
    let action_table = unsafe {
        log!("(pers) read call_site has length {call_site_table_size:?}");
        reader.ptr.wrapping_add(call_site_table_size as usize)
    };

    log!("(pers) action table sits at offset {action_table:?}");
    let ip = context.ip;

    if !USING_SJLJ_EXCEPTIONS {
        // read the callsite table
        while reader.ptr < action_table {
            let call_site_record_reader = &mut reader;
            unsafe {
                // Offset of the call site relative to the previous call site, counted in number of 16-byte bundles
                let call_site_start =
                    read_encoded_offset(call_site_record_reader, call_site_table_encoding)?;
                let call_site_length =
                    read_encoded_offset(call_site_record_reader, call_site_table_encoding)?;
                // Offset of the landing pad, typically a byte offset relative to the LPStart address.
                let call_site_lpad =
                    read_encoded_offset(call_site_record_reader, call_site_table_encoding)?;
                // Offset of the first associated action record, relative to the start of the actions table.
                // This value is biased by 1 (1 indicates the start of the actions table), and 0 indicates that there are no actions.
                let call_site_action_entry = call_site_record_reader.read_uleb128();

                log!("(pers) read cs_start is {call_site_start:?}");
                log!("(pers) read cs_len is {call_site_length:?}");
                log!("(pers) read cs_lpad is {call_site_lpad:?}");
                log!("(pers) read cs_ae is {call_site_action_entry:?}");
                // Callsite table is sorted by cs_start, so if we've passed the ip, we
                // may stop searching.
                if ip < func_start.wrapping_add(call_site_start) {
                    break;
                }

                // Call site matches the current ip. It's a candidate.
                if ip < func_start.wrapping_add(call_site_start + call_site_length) {
                    log!(
                        "(pers) found a matching call site: {func_start:?} <= {ip:?} <= {:?}",
                        func_start.wrapping_add(call_site_start + call_site_length)
                    );
                    if call_site_lpad == 0 {
                        return Ok(EHAction::None);
                    } else {
                        let lpad = lpad_base.wrapping_add(call_site_lpad);
                        let mut catches = vec![];

                        log!("(pers) lpad sits at {lpad:?}");

                        if call_site_action_entry == 0 {
                            // We don't generate cleanup clauses, so this can't happen
                            return Ok(EHAction::Terminate);
                        }

                        log!("(pers) read cs_action_entry: {call_site_action_entry}");
                        log!("(pers) action_table: {action_table:?}");

                        // Convert 1-based byte offset into
                        let mut action_record: *const u8 =
                            action_table.wrapping_add((call_site_action_entry - 1) as usize);

                        log!("(pers) first action at: {action_record:?}");

                        loop {
                            // Read the action record.
                            let mut action_record_reader = DwarfReader::new(action_record);
                            // The two record kinds have the same format, with only small differences.
                            // They are distinguished by the "type_filter" field: Catch clauses have strictly positive switch values,
                            // and exception specifications have strictly negative switch values. Value 0 indicates a catch-all clause.
                            let type_filter = action_record_reader.read_sleb128();
                            log!(
                                "(pers) type_filter for action #{call_site_action_entry}: {type_filter:?}"
                            );

                            if type_filter > 0 {
                                // This is a catch clause so the type_filter is an index into the types table.
                                //
                                // Positive value, starting at 1.
                                // Index in the types table of the __typeinfo for the catch-clause type.
                                // 1 is the first word preceding TTBase, 2 is the second word, and so on.
                                // Used by the runtime to check if the thrown exception type matches the catch-clause type.
                                let types_table_index = type_filter;
                                if types_table_base.is_null() {
                                    panic!();
                                }

                                let tag_ptr = {
                                    let new_types_table_index =
                                        match DwEhPe(types_table_encoding.0 & 0x0f) {
                                            gimli::DW_EH_PE_absptr => {
                                                type_filter * (size_of::<*const u8>() as i64)
                                            }
                                            gimli::DW_EH_PE_sdata2 | gimli::DW_EH_PE_udata2 => {
                                                type_filter * 2
                                            }
                                            gimli::DW_EH_PE_sdata4 | gimli::DW_EH_PE_udata4 => {
                                                type_filter * 4
                                            }
                                            gimli::DW_EH_PE_sdata8 | gimli::DW_EH_PE_udata8 => {
                                                type_filter * 8
                                            }
                                            _ => panic!(),
                                        };

                                    log!(
                                        "(pers) new_types_table_index for action #{call_site_action_entry}: {new_types_table_index:?}"
                                    );

                                    let typeinfo = types_table_base
                                        .wrapping_sub(new_types_table_index as usize);
                                    log!("(pers) reading ttype info from {typeinfo:?}");
                                    read_encoded_pointer(
                                        // Basically just reader.read() a SLEB128.
                                        &mut DwarfReader::new(typeinfo),
                                        context,
                                        types_table_encoding,
                                    )
                                };
                                let tag_ptr = tag_ptr.unwrap();

                                if tag_ptr.is_null() {
                                    if catches.is_empty() {
                                        // No specifics so far, so we definitely have a catch-all we should use
                                        return Ok(EHAction::CatchAll { lpad });
                                    } else {
                                        // We do have catch clauses that *may* need to be used, so we must
                                        // defer to phase 2 anyway, but this catch-all will be used if
                                        // none of those clauses match, so we can return early.
                                        return Ok(EHAction::CatchSpecificOrAll {
                                            lpad,
                                            tags: catches,
                                        });
                                    }
                                }

                                let tag = std::mem::transmute::<*const u8, *const u32>(tag_ptr)
                                    .read_unaligned();
                                log!("(pers) read tag {tag:?}");

                                // Since we don't know what this tag corresponds to, we must defer
                                // the decision to the second phase.
                                catches.push(tag);
                            } else if type_filter == 0 {
                                // We don't create cleanup clauses, so this can't happen
                                return Ok(EHAction::Terminate);
                            }

                            let next_action_record = action_record_reader.clone().read_sleb128();
                            if next_action_record == 0 {
                                return Ok(if catches.is_empty() {
                                    EHAction::None
                                } else {
                                    EHAction::CatchSpecific {
                                        lpad,
                                        tags: catches,
                                    }
                                });
                            }

                            action_record = action_record_reader
                                .ptr
                                .wrapping_add(next_action_record as usize);
                        }
                    }
                }
            }
        }

        // Ip is not present in the table. This indicates a nounwind call.
        Ok(EHAction::Terminate)
    } else {
        todo!()
    }
}

#[inline]
fn round_up(unrounded: usize, align: usize) -> Result<usize, ()> {
    if align.is_power_of_two() {
        Ok(unrounded.next_multiple_of(align))
    } else {
        Err(())
    }
}

/// Reads an offset (`usize`) from `reader` whose encoding is described by `encoding`.
///
/// `encoding` must be a [DWARF Exception Header Encoding as described by the LSB spec][LSB-dwarf-ext].
/// In addition the upper ("application") part must be zero.
///
/// # Errors
/// Returns `Err` if `encoding`
/// * is not a valid DWARF Exception Header Encoding,
/// * is `DW_EH_PE_omit`, or
/// * has a non-zero application part.
///
/// [LSB-dwarf-ext]: https://refspecs.linuxfoundation.org/LSB_5.0.0/LSB-Core-generic/LSB-Core-generic/dwarfext.html
unsafe fn read_encoded_offset(reader: &mut DwarfReader, encoding: DwEhPe) -> Result<usize, ()> {
    if encoding == gimli::DW_EH_PE_omit || encoding.0 & 0xF0 != 0 {
        return Err(());
    }
    let result = unsafe {
        match DwEhPe(encoding.0 & 0x0F) {
            // despite the name, LLVM also uses absptr for offsets instead of pointers
            gimli::DW_EH_PE_absptr => reader.read::<usize>(),
            gimli::DW_EH_PE_uleb128 => reader.read_uleb128() as usize,
            gimli::DW_EH_PE_udata2 => reader.read::<u16>() as usize,
            gimli::DW_EH_PE_udata4 => reader.read::<u32>() as usize,
            gimli::DW_EH_PE_udata8 => reader.read::<u64>() as usize,
            gimli::DW_EH_PE_sleb128 => reader.read_sleb128() as usize,
            gimli::DW_EH_PE_sdata2 => reader.read::<i16>() as usize,
            gimli::DW_EH_PE_sdata4 => reader.read::<i32>() as usize,
            gimli::DW_EH_PE_sdata8 => reader.read::<i64>() as usize,
            _ => return Err(()),
        }
    };
    Ok(result)
}

/// Reads a pointer from `reader` whose encoding is described by `encoding`.
///
/// `encoding` must be a [DWARF Exception Header Encoding as described by the LSB spec][LSB-dwarf-ext].
///
/// # Errors
/// Returns `Err` if `encoding`
/// * is not a valid DWARF Exception Header Encoding,
/// * is `DW_EH_PE_omit`, or
/// * combines `DW_EH_PE_absptr` or `DW_EH_PE_aligned` application part with an integer encoding
///   (not `DW_EH_PE_absptr`) in the value format part.
///
/// [LSB-dwarf-ext]: https://refspecs.linuxfoundation.org/LSB_5.0.0/LSB-Core-generic/LSB-Core-generic/dwarfext.html
unsafe fn read_encoded_pointer(
    reader: &mut DwarfReader,
    context: &EHContext<'_>,
    encoding: DwEhPe,
) -> Result<*const u8, ()> {
    if encoding == gimli::DW_EH_PE_omit {
        return Err(());
    }

    log!("(pers) About to read encoded pointer at {:?}", reader.ptr);

    let base_ptr = match DwEhPe(encoding.0 & 0x70) {
        gimli::DW_EH_PE_absptr => {
            log!("(pers) encoding is: DW_EH_PE_absptr");
            core::ptr::null()
        }
        // relative to address of the encoded value, despite the name
        gimli::DW_EH_PE_pcrel => {
            log!("(pers) encoding is: DW_EH_PE_pcrel");
            reader.ptr
        }
        gimli::DW_EH_PE_funcrel => {
            log!("(pers) encoding is: DW_EH_PE_funcrel");
            if context.func_start.is_null() {
                return Err(());
            }
            context.func_start
        }
        gimli::DW_EH_PE_textrel => {
            log!("(pers) encoding is: DW_EH_PE_textrel");
            (*context.get_text_start)()
        }
        gimli::DW_EH_PE_datarel => {
            log!("(pers) encoding is: DW_EH_PE_datarel");

            (*context.get_data_start)()
        }
        // aligned means the value is aligned to the size of a pointer
        gimli::DW_EH_PE_aligned => {
            log!("(pers) encoding is: DW_EH_PE_aligned");
            reader.ptr = {
                let this = reader.ptr;
                let addr = round_up(
                    {
                        let this = reader.ptr;
                        unsafe { mem::transmute::<*const (), usize>(this.cast::<()>()) }
                    },
                    mem::size_of::<*const u8>(),
                )?;
                // In the mean-time, this operation is defined to be "as if" it was
                // a wrapping_offset, so we can emulate it as such. This should properly
                // restore pointer provenance even under today's compiler.
                let self_addr = unsafe { mem::transmute::<*const (), isize>(this.cast::<()>()) };
                let dest_addr = addr as isize;
                let offset = dest_addr.wrapping_sub(self_addr);

                // This is the canonical desugaring of this operation
                this.wrapping_byte_offset(offset)
            };
            core::ptr::null()
        }
        _ => return Err(()),
    };

    let mut ptr = if base_ptr.is_null() {
        // any value encoding other than absptr would be nonsensical here;
        // there would be no source of pointer provenance
        if DwEhPe(encoding.0 & 0x0f) != gimli::DW_EH_PE_absptr {
            return Err(());
        }
        unsafe { reader.read::<*const u8>() }
    } else {
        log!("(pers) since base_ptr is not null, we must an offset");
        let offset = unsafe { read_encoded_offset(reader, DwEhPe(encoding.0 & 0x0f))? };
        log!("(pers) read offset is {offset:x?}");
        // For relative encodings, a raw zero denotes a null pointer. Do not
        // apply the base or indirect dereference in that case.
        // Upstream implementation: https://github.com/llvm/llvm-project/blob/main/libcxxabi/src/cxa_personality.cpp#L341-L342.
        if offset == 0 {
            core::ptr::null()
        } else {
            base_ptr.wrapping_add(offset)
        }
    };

    log!("(pers) about to read from {ptr:?}");

    if !ptr.is_null() && encoding.0 & gimli::DW_EH_PE_indirect.0 != 0 {
        ptr = unsafe { ptr.cast::<*const u8>().read_unaligned() };
    }

    log!("(pers) returning ptr value {ptr:?}");

    Ok(ptr)
}