Apply relocations when parsing DWARF

6 files changed, 176 insertions, 212 deletions

diff --git a/Cargo.lock b/Cargo.lock
index c09b9122..6ed30022 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -727,6 +727,7 @@ dependencies = [
  "indexmap",
  "iset",
  "log",
+ "object 0.36.7",
  "regex",
 ]
 
@@ -747,6 +748,7 @@ dependencies = [
  "binaryninja",
  "binaryninjacore-sys",
  "gimli",
+ "object 0.36.7",
  "thiserror 2.0.12",
  "zstd",
 ]
@@ -1821,7 +1823,7 @@ dependencies = [
  "hashbrown 0.14.5",
  "indexmap",
  "memchr",
- "ruzstd",
+ "ruzstd 0.5.0",
 ]
 
 [[package]]
@@ -1830,7 +1832,9 @@ version = "0.36.7"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "62948e14d923ea95ea2c7c86c71013138b66525b86bdc08d2dcc262bdb497b87"
 dependencies = [
+ "flate2",
  "memchr",
+ "ruzstd 0.7.3",
 ]
 
 [[package]]
@@ -2421,6 +2425,15 @@ dependencies = [
 ]
 
 [[package]]
+name = "ruzstd"
+version = "0.7.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "fad02996bfc73da3e301efe90b1837be9ed8f4a462b6ed410aa35d00381de89f"
+dependencies = [
+ "twox-hash",
+]
+
+[[package]]
 name = "ryu"
 version = "1.0.20"
 source = "registry+https://github.com/rust-lang/crates.io-index"
diff --git a/plugins/dwarf/dwarf_import/Cargo.toml b/plugins/dwarf/dwarf_import/Cargo.toml
index 9527274a..ca06b11b 100644
--- a/plugins/dwarf/dwarf_import/Cargo.toml
+++ b/plugins/dwarf/dwarf_import/Cargo.toml
@@ -16,4 +16,5 @@ log = "0.4"
 iset = "0.2.2"
 cpp_demangle = "0.4.3"
 regex = "1"
-indexmap = "2.5.0"
\ No newline at end of file
+indexmap = "2.5.0"
+object = "0.36"
diff --git a/plugins/dwarf/dwarf_import/src/dwarfdebuginfo.rs b/plugins/dwarf/dwarf_import/src/dwarfdebuginfo.rs
index 0dc6b7fb..94f370a1 100644
--- a/plugins/dwarf/dwarf_import/src/dwarfdebuginfo.rs
+++ b/plugins/dwarf/dwarf_import/src/dwarfdebuginfo.rs
@@ -125,7 +125,7 @@ pub(crate) struct DebugInfoBuilderContext<R: ReaderType> {
 }
 
 impl<R: ReaderType> DebugInfoBuilderContext<R> {
-    pub(crate) fn new(view: &BinaryView, dwarf: &Dwarf<R>) -> Option<Self> {
+    pub(crate) fn new(default_address_size: usize, dwarf: &Dwarf<R>) -> Option<Self> {
         let mut units = vec![];
         let mut iter = dwarf.units();
         while let Ok(Some(header)) = iter.next() {
@@ -154,7 +154,7 @@ impl<R: ReaderType> DebugInfoBuilderContext<R> {
             units,
             sup_units,
             names: HashMap::new(),
-            default_address_size: view.address_size(),
+            default_address_size,
             total_die_count: 0,
             total_unit_size_bytes: 0,
         })
diff --git a/plugins/dwarf/dwarf_import/src/lib.rs b/plugins/dwarf/dwarf_import/src/lib.rs
index dcac4f0b..8ad5e442 100644
--- a/plugins/dwarf/dwarf_import/src/lib.rs
+++ b/plugins/dwarf/dwarf_import/src/lib.rs
@@ -32,9 +32,7 @@ use binaryninja::{
     settings::Settings,
     template_simplifier::simplify_str_to_str,
 };
-use dwarfreader::{
-    create_section_reader, get_endian, is_dwo_dwarf, is_non_dwo_dwarf, is_raw_dwo_dwarf,
-};
+use dwarfreader::{create_section_reader_object, get_endian, is_dwo_dwarf, is_non_dwo_dwarf};
 
 use functions::parse_lexical_block;
 use gimli::{
@@ -46,6 +44,7 @@ use binaryninja::logger::Logger;
 use helpers::{get_build_id, load_debug_info_for_build_id};
 use iset::IntervalMap;
 use log::{debug, error, warn};
+use object::{Object, ObjectSection};
 
 trait ReaderType: Reader<Offset = usize> {}
 impl<T: Reader<Offset = usize>> ReaderType for T {}
@@ -388,7 +387,7 @@ fn parse_unit<R: ReaderType>(
 }
 
 fn parse_unwind_section<R: Reader, U: UnwindSection<R>>(
-    view: &BinaryView,
+    file: &object::File,
     unwind_section: U,
 ) -> gimli::Result<iset::IntervalMap<u64, i64>>
 where
@@ -396,40 +395,22 @@ where
 {
     let mut bases = gimli::BaseAddresses::default();
 
-    // DWARF info is stored relative to the original image base (0 for relocatable images), normalize entries to the original image base
-    let section_adjustment = view.original_image_base().wrapping_sub(view.image_base());
-
-    if let Some(section) = view
-        .section_by_name(".eh_frame_hdr")
-        .or(view.section_by_name("__eh_frame_hdr"))
-    {
-        bases = bases.set_eh_frame_hdr(section.start().wrapping_add(section_adjustment));
+    if let Some(section) = file.section_by_name(".eh_frame_hdr") {
+        bases = bases.set_eh_frame_hdr(section.address());
     }
 
-    if let Some(section) = view
-        .section_by_name(".eh_frame")
-        .or(view.section_by_name("__eh_frame"))
-    {
-        bases = bases.set_eh_frame(section.start().wrapping_add(section_adjustment));
-    } else if let Some(section) = view
-        .section_by_name(".debug_frame")
-        .or(view.section_by_name("__debug_frame"))
-    {
-        bases = bases.set_eh_frame(section.start().wrapping_add(section_adjustment));
+    if let Some(section) = file.section_by_name(".eh_frame") {
+        bases = bases.set_eh_frame(section.address());
+    } else if let Some(section) = file.section_by_name(".debug_frame") {
+        bases = bases.set_eh_frame(section.address());
     }
 
-    if let Some(section) = view
-        .section_by_name(".text")
-        .or(view.section_by_name("__text"))
-    {
-        bases = bases.set_text(section.start().wrapping_add(section_adjustment));
+    if let Some(section) = file.section_by_name(".text") {
+        bases = bases.set_text(section.address());
     }
 
-    if let Some(section) = view
-        .section_by_name(".got")
-        .or(view.section_by_name("__got"))
-    {
-        bases = bases.set_got(section.start().wrapping_add(section_adjustment));
+    if let Some(section) = file.section_by_name(".got") {
+        bases = bases.set_got(section.address());
     }
 
     let mut cies = HashMap::new();
@@ -535,58 +516,49 @@ fn get_supplementary_build_id(bv: &BinaryView) -> Option<String> {
     }
 }
 
-fn parse_range_data_offsets(
-    bv: &BinaryView,
-    dwo_file: bool,
-) -> Option<Result<IntervalMap<u64, i64>, ()>> {
-    if bv.section_by_name(".eh_frame").is_some() || bv.section_by_name("__eh_frame").is_some() {
-        let eh_frame_endian = get_endian(bv);
-        let eh_frame_section_reader = |section_id: SectionId| -> _ {
-            create_section_reader(section_id, bv, eh_frame_endian, dwo_file)
-        };
-        let mut eh_frame = match gimli::EhFrame::load(eh_frame_section_reader) {
-            Ok(x) => x,
-            Err(e) => {
-                log::error!("Failed to load EH frame: {}", e);
-                return None;
-            }
-        };
-        if let Some(view_arch) = bv.default_arch() {
-            if view_arch.name().as_str() == "aarch64" {
-                eh_frame.set_vendor(gimli::Vendor::AArch64);
-            }
+fn parse_range_data_offsets(bv: &BinaryView) -> Result<IntervalMap<u64, i64>, String> {
+    let raw_view = bv.raw_view().unwrap();
+    let raw_view_data = raw_view.read_vec(0, raw_view.len() as usize);
+    let file =
+        object::File::parse(&*raw_view_data).map_err(|e| format!("Failed to parse bv: {}", e))?;
+    let dwo_file = file.section_by_name(".debug_info.dwo").is_some();
+    let endian = match file.endianness() {
+        object::Endianness::Little => gimli::RunTimeEndian::Little,
+        object::Endianness::Big => gimli::RunTimeEndian::Big,
+    };
+
+    let section_reader = |section_id: SectionId| -> _ {
+        create_section_reader_object(section_id, &file, endian, dwo_file)
+    };
+
+    if file.section_by_name(".eh_frame").is_some() {
+        let mut eh_frame = gimli::EhFrame::load(section_reader)
+            .map_err(|e| format!("Failed to load EH frame: {}", e))?;
+
+        if file.architecture() == object::Architecture::Aarch64 {
+            eh_frame.set_vendor(gimli::Vendor::AArch64);
         }
-        eh_frame.set_address_size(bv.address_size() as u8);
-        Some(
-            parse_unwind_section(bv, eh_frame)
-                .map_err(|e| error!("Error parsing .eh_frame: {}", e)),
-        )
-    } else if bv.section_by_name(".debug_frame").is_some()
-        || bv.section_by_name("__debug_frame").is_some()
-    {
-        let debug_frame_endian = get_endian(bv);
-        let debug_frame_section_reader = |section_id: SectionId| -> _ {
-            create_section_reader(section_id, bv, debug_frame_endian, dwo_file)
-        };
-        let mut debug_frame = match gimli::DebugFrame::load(debug_frame_section_reader) {
-            Ok(x) => x,
-            Err(e) => {
-                log::error!("Failed to load debug frame: {}", e);
-                return None;
-            }
-        };
-        if let Some(view_arch) = bv.default_arch() {
-            if view_arch.name().as_str() == "aarch64" {
-                debug_frame.set_vendor(gimli::Vendor::AArch64);
-            }
+
+        if let Some(address_size) = file.architecture().address_size() {
+            eh_frame.set_address_size(address_size.bytes());
         }
-        debug_frame.set_address_size(bv.address_size() as u8);
-        Some(
-            parse_unwind_section(bv, debug_frame)
-                .map_err(|e| error!("Error parsing .debug_frame: {}", e)),
-        )
+
+        parse_unwind_section(&file, eh_frame).map_err(|e| format!("Error parsing .eh_frame: {}", e))
+    } else if file.section_by_name(".debug_frame").is_some() {
+        let mut debug_frame = gimli::DebugFrame::load(section_reader)
+            .map_err(|e| format!("Failed to load debug frame: {}", e))?;
+
+        if file.architecture() == object::Architecture::Aarch64 {
+            debug_frame.set_vendor(gimli::Vendor::AArch64);
+        }
+
+        if let Some(address_size) = file.architecture().address_size() {
+            debug_frame.set_address_size(address_size.bytes());
+        }
+        parse_unwind_section(&file, debug_frame)
+            .map_err(|e| format!("Error parsing .debug_frame: {}", e))
     } else {
-        None
+        Ok(Default::default())
     }
 }
 
@@ -595,35 +567,43 @@ fn parse_dwarf(
     debug_bv: &BinaryView,
     supplementary_bv: Option<&BinaryView>,
     progress: Box<dyn Fn(usize, usize) -> Result<(), ()>>,
-) -> Result<DebugInfoBuilder, ()> {
+) -> Result<DebugInfoBuilder, String> {
     // TODO: warn if no supplementary file and .gnu_debugaltlink section present
 
     // Determine if this is a DWO
     // TODO : Make this more robust...some DWOs follow non-DWO conventions
 
     // Figure out if it's the given view or the raw view that has the dwarf info in it
-    let raw_view = &debug_bv.raw_view().ok_or(())?;
-    let view = if is_dwo_dwarf(debug_bv) || is_non_dwo_dwarf(debug_bv) {
-        debug_bv
+    let raw_view = &debug_bv
+        .raw_view()
+        .ok_or("Failed to get raw view for debug binary view".to_string())?;
+
+    let address_size = if is_dwo_dwarf(debug_bv) || is_non_dwo_dwarf(debug_bv) {
+        debug_bv.address_size()
     } else {
-        raw_view
+        raw_view.address_size()
     };
 
-    let dwo_file = is_dwo_dwarf(view) || is_raw_dwo_dwarf(view);
+    // Parse this early to reduce peak memory usage
+    let range_data_offsets = parse_range_data_offsets(bv).unwrap_or_default();
 
-    // gimli setup
-    let endian = get_endian(view);
-    let mut section_reader =
-        |section_id: SectionId| -> _ { create_section_reader(section_id, view, endian, dwo_file) };
+    // Read the raw view to an object::File so relocations get handled for us
+    let raw_view_data = raw_view.read_vec(0, raw_view.len() as usize);
+    let debug_file =
+        object::File::parse(&*raw_view_data).map_err(|e| format!("Failed to parse bv: {}", e))?;
+    let dwo_file = debug_file.section_by_name(".debug_info.dwo").is_some();
+    let endian = match debug_file.endianness() {
+        object::Endianness::Little => gimli::RunTimeEndian::Little,
+        object::Endianness::Big => gimli::RunTimeEndian::Big,
+    };
 
-    let mut dwarf = match Dwarf::load(&mut section_reader) {
-        Ok(x) => x,
-        Err(e) => {
-            error!("Failed to load DWARF info: {}", e);
-            return Err(());
-        }
+    let mut section_reader = |section_id: SectionId| -> _ {
+        create_section_reader_object(section_id, &debug_file, endian, dwo_file)
     };
 
+    let mut dwarf = Dwarf::load(&mut section_reader)
+        .map_err(|e| format!("Failed to load DWARF info: {}", e))?;
+
     if dwo_file {
         dwarf.file_type = DwarfFileType::Dwo;
     } else {
@@ -631,31 +611,22 @@ fn parse_dwarf(
     }
 
     if let Some(sup_bv) = supplementary_bv {
+        let sup_raw_view = sup_bv
+            .raw_view()
+            .ok_or_else(|| format!("Failed to get raw view for supplementary bv"))?;
+        let sup_view_data = sup_raw_view.read_vec(0, sup_raw_view.len() as usize);
+        let sup_file = object::File::parse(&*sup_view_data)
+            .map_err(|e| format!("Failed to parse supplementary bv: {}", e))?;
         let sup_endian = get_endian(sup_bv);
-        let sup_dwo_file = is_dwo_dwarf(sup_bv) || is_raw_dwo_dwarf(sup_bv);
+        let sup_dwo_file = sup_file.section_by_name(".debug_info.dwo").is_some();
         let sup_section_reader = |section_id: SectionId| -> _ {
-            create_section_reader(section_id, sup_bv, sup_endian, sup_dwo_file)
+            create_section_reader_object(section_id, &sup_file, sup_endian, sup_dwo_file)
         };
         if let Err(e) = dwarf.load_sup(sup_section_reader) {
             error!("Failed to load supplementary file: {}", e);
         }
     }
 
-    let range_data_offsets = match parse_range_data_offsets(bv, dwo_file) {
-        Some(x) => x?,
-        None => {
-            if let Some(raw_view) = bv.raw_view() {
-                if let Some(offsets) = parse_range_data_offsets(&raw_view, dwo_file) {
-                    offsets?
-                } else {
-                    Default::default()
-                }
-            } else {
-                Default::default()
-            }
-        }
-    };
-
     // Create debug info builder and recover name mapping first
     //  Since DWARF is stored as a tree with arbitrary implicit edges among leaves,
     //   it is not possible to correctly track namespaces while you're parsing "in order" without backtracking,
@@ -663,7 +634,8 @@ fn parse_dwarf(
     let mut debug_info_builder = DebugInfoBuilder::new();
     debug_info_builder.set_range_data_offsets(range_data_offsets);
 
-    if let Some(mut debug_info_builder_context) = DebugInfoBuilderContext::new(view, &dwarf) {
+    if let Some(mut debug_info_builder_context) = DebugInfoBuilderContext::new(address_size, &dwarf)
+    {
         calculate_total_unit_bytes(&dwarf, &mut debug_info_builder_context);
 
         let progress_weights = [0.5, 0.5];
@@ -770,7 +742,10 @@ impl CustomDebugInfoParser for DWARFParser {
                 builder.post_process(bv, debug_info).commit_info(debug_info);
                 true
             }
-            Err(_) => false,
+            Err(e) => {
+                log::error!("Failed to parse DWARF: {}", e);
+                false
+            }
         };
 
         if let (Some(ext), true) = (external_file, close_external) {
diff --git a/plugins/dwarf/shared/Cargo.toml b/plugins/dwarf/shared/Cargo.toml
index 83ecee93..19cb963e 100644
--- a/plugins/dwarf/shared/Cargo.toml
+++ b/plugins/dwarf/shared/Cargo.toml
@@ -11,3 +11,4 @@ binaryninjacore-sys.workspace = true
 gimli = "0.31"
 zstd = "0.13.2"
 thiserror = "2.0"
+object = "0.36"
diff --git a/plugins/dwarf/shared/src/lib.rs b/plugins/dwarf/shared/src/lib.rs
index 7aa3b486..634aa784 100644
--- a/plugins/dwarf/shared/src/lib.rs
+++ b/plugins/dwarf/shared/src/lib.rs
@@ -13,6 +13,7 @@
 // limitations under the License.
 
 use gimli::{EndianRcSlice, Endianity, RunTimeEndian, SectionId};
+use object::{Object, ObjectSection};
 
 use binaryninja::{
     binary_view::{BinaryView, BinaryViewBase, BinaryViewExt},
@@ -35,6 +36,9 @@ pub enum Error {
 
     #[error("{0}")]
     IoError(#[from] std::io::Error),
+
+    #[error("{0}")]
+    ObjectError(#[from] object::Error),
 }
 
 pub fn is_non_dwo_dwarf(view: &BinaryView) -> bool {
@@ -89,105 +93,75 @@ pub fn get_endian(view: &BinaryView) -> RunTimeEndian {
     }
 }
 
-pub fn create_section_reader<'a, Endian: 'a + Endianity>(
+pub fn create_section_reader<Endian: Endianity>(
     section_id: SectionId,
-    view: &'a BinaryView,
+    view: &BinaryView,
     endian: Endian,
-    dwo_file: bool,
-) -> Result<EndianRcSlice<Endian>, Error> {
-    let section_name = if dwo_file && section_id.dwo_name().is_some() {
-        section_id.dwo_name().unwrap()
-    } else {
-        section_id.name()
-    };
+    is_dwo: bool,
+) -> Result<RelocateOwned<gimli::EndianRcSlice<Endian>>, Error> {
+    let raw_view = view.raw_view().unwrap();
+    let view_data = raw_view.read_vec(0, raw_view.len() as usize);
+    let file = object::File::parse(&*view_data)?;
+    create_section_reader_object(section_id, &file, endian, is_dwo)
+}
 
-    if let Some(section) = view.section_by_name(section_name) {
-        // TODO : This is kinda broke....should add rust wrappers for some of this
-        if let Some(symbol) = view
-            .symbols()
-            .iter()
-            .find(|symbol| symbol.full_name().to_string_lossy() == "__elf_section_headers")
-        {
-            if let Some(data_var) = view
-                .data_variables()
-                .iter()
-                .find(|var| var.address == symbol.address())
-            {
-                // TODO : This should eventually be wrapped by some DataView sorta thingy thing, like how python does it
-                let data_type = &data_var.ty.contents;
-                let data = view.read_vec(data_var.address, data_type.width() as usize);
-                let element_type = data_type.element_type().unwrap().contents;
+#[derive(Debug, Clone)]
+pub struct RelocationMap(Rc<object::read::RelocationMap>);
 
-                if let Some(current_section_header) = data
-                    .chunks(element_type.width() as usize)
-                    .find(|section_header| {
-                        if view.address_size() == 4 {
-                            endian.read_u32(&section_header[16..20]) as u64 == section.start()
-                        } else {
-                            endian.read_u64(&section_header[24..32]) == section.start()
-                        }
-                    })
-                {
-                    let section_flags = if view.address_size() == 4 {
-                        endian.read_u32(&current_section_header[8..12]) as u64
-                    } else {
-                        endian.read_u64(&current_section_header[8..16])
-                    };
-                    // If the section has the compressed bit set
-                    if (section_flags & 2048) != 0 {
-                        // Get section, trim header, decompress, return
-                        let compressed_header_size = view.address_size() * 3;
+impl Default for RelocationMap {
+    fn default() -> Self {
+        Self(Rc::new(object::read::RelocationMap::default()))
+    }
+}
 
-                        let offset = section.start() + compressed_header_size as u64;
-                        let len = section.len() - compressed_header_size;
+impl RelocationMap {
+    fn add(&mut self, file: &object::File, section: &object::Section) -> Result<(), Error> {
+        let map =
+            Rc::get_mut(&mut self.0).expect("Failed to get mutable reference to RelocationMap");
+        for (offset, relocation) in section.relocations() {
+            map.add(file, offset, relocation)?
+        }
+        Ok(())
+    }
+}
 
-                        let ch_type_vec = view.read_vec(section.start(), 4);
-                        let ch_type = endian.read_u32(&ch_type_vec);
+impl gimli::read::Relocate for RelocationMap {
+    fn relocate_address(&self, offset: usize, value: u64) -> gimli::Result<u64> {
+        Ok(self.0.relocate(offset as u64, value))
+    }
 
-                        if let Ok(buffer) = view.read_buffer(offset, len) {
-                            match ch_type {
-                                1 => {
-                                    return Ok(EndianRcSlice::new(
-                                        buffer.zlib_decompress().get_data().into(),
-                                        endian,
-                                    ));
-                                }
-                                2 => {
-                                    return Ok(EndianRcSlice::new(
-                                        zstd::decode_all(buffer.get_data())?.as_slice().into(),
-                                        endian,
-                                    ));
-                                }
-                                x => {
-                                    return Err(Error::UnknownCompressionMethod(x));
-                                }
-                            }
-                        }
-                    }
-                }
-            }
-        }
-        let offset = section.start();
-        let len = section.len();
-        if len == 0 {
-            Ok(EndianRcSlice::new(Rc::from([]), endian))
-        } else {
-            Ok(EndianRcSlice::new(
-                Rc::from(view.read_vec(offset, len).as_slice()),
-                endian,
-            ))
-        }
+    fn relocate_offset(&self, offset: usize, value: usize) -> gimli::Result<usize> {
+        <usize as gimli::ReaderOffset>::from_u64(self.0.relocate(offset as u64, value as u64))
     }
-    // Truncate Mach-O section names to 16 bytes
-    else if let Some(section) = view.section_by_name(&format!(
-        "__{}",
-        &section_name[1..section_name.len().min(15)]
-    )) {
-        Ok(EndianRcSlice::new(
-            Rc::from(view.read_vec(section.start(), section.len()).as_slice()),
-            endian,
-        ))
+}
+
+type RelocateOwned<R> = gimli::RelocateReader<R, RelocationMap>;
+
+pub fn create_section_reader_object<Endian: gimli::Endianity>(
+    id: gimli::SectionId,
+    file: &object::File,
+    endian: Endian,
+    is_dwo: bool,
+) -> Result<RelocateOwned<gimli::EndianRcSlice<Endian>>, Error> {
+    let mut relocations = RelocationMap::default();
+    let name = if is_dwo {
+        id.dwo_name()
+    } else if file.format() == object::BinaryFormat::Xcoff {
+        id.xcoff_name()
     } else {
-        Ok(EndianRcSlice::new(Rc::from([]), endian))
-    }
+        Some(id.name())
+    };
+    let data = match name.and_then(|name| file.section_by_name(name)) {
+        Some(ref section) => {
+            // DWO sections never have relocations, so don't bother.
+            if !is_dwo {
+                relocations.add(file, section)?;
+            }
+            section.uncompressed_data()?.into_owned()
+        }
+        // Use a non-zero capacity so that `ReaderOffsetId`s are unique.
+        None => Vec::with_capacity(1),
+    };
+    let section = EndianRcSlice::new(Rc::from(data.into_boxed_slice()), endian);
+    Ok(gimli::RelocateReader::new(section, relocations))
 }