Merge pull request #529 from khuey/downstream

More split dwarf work

Author: philipc

examples/dwarfdump.rs

@@ -550,7 +550,10 @@ where
         reader: Default::default(),
     };
 
-    let dwarf = gimli::Dwarf::load(&mut load_section, |_| Ok(no_reader.clone())).unwrap();
+    let mut dwarf = gimli::Dwarf::load(&mut load_section, |_| Ok(no_reader.clone())).unwrap();
+    if flags.dwo {
+        dwarf.file_type = gimli::DwarfFileType::Dwo;
+    }
 
     let out = io::stdout();
     if flags.eh_frame {

src/common.rs

@@ -311,6 +311,7 @@ impl SectionId {
             SectionId::DebugLoc => ".debug_loc.dwo",
             SectionId::DebugLocLists => ".debug_loclists.dwo",
             SectionId::DebugMacro => ".debug_macro.dwo",
+            SectionId::DebugRngLists => ".debug_rnglists.dwo",
             SectionId::DebugStr => ".debug_str.dwo",
             SectionId::DebugStrOffsets => ".debug_str_offsets.dwo",
             _ => return None,
@@ -322,3 +323,20 @@ impl SectionId {
 /// split DWARF and linking a split compilation unit back together.
 #[derive(Debug, Clone, Copy, PartialEq, Eq, Hash)]
 pub struct DwoId(pub u64);
+
+/// The "type" of file with DWARF debugging information. This determines, among other things,
+/// which files DWARF sections should be loaded from.
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+pub enum DwarfFileType {
+    /// A normal executable or object file.
+    Main,
+    /// A .dwo split DWARF file.
+    Dwo,
+    // TODO: Supplementary files, .dwps?
+}
+
+impl Default for DwarfFileType {
+    fn default() -> Self {
+        DwarfFileType::Main
+    }
+}

src/read/dwarf.rs

@@ -3,8 +3,8 @@ use alloc::string::String;
 use crate::common::{
     DebugAddrBase, DebugAddrIndex, DebugInfoOffset, DebugLineStrOffset, DebugLocListsBase,
     DebugLocListsIndex, DebugRngListsBase, DebugRngListsIndex, DebugStrOffset, DebugStrOffsetsBase,
-    DebugStrOffsetsIndex, DebugTypesOffset, Encoding, LocationListsOffset, RangeListsOffset,
-    SectionId, UnitSectionOffset,
+    DebugStrOffsetsIndex, DebugTypesOffset, DwarfFileType, Encoding, LocationListsOffset,
+    RangeListsOffset, SectionId, UnitSectionOffset,
 };
 use crate::constants;
 use crate::read::{
@@ -50,6 +50,9 @@ pub struct Dwarf<R> {
 
     /// The range lists in the `.debug_ranges` and `.debug_rnglists` sections.
     pub ranges: RangeLists<R>,
+
+    /// The type of this file.
+    pub file_type: DwarfFileType,
 }
 
 impl<T> Dwarf<T> {
@@ -84,6 +87,7 @@ impl<T> Dwarf<T> {
             debug_types: Section::load(&mut section)?,
             locations: LocationLists::new(debug_loc, debug_loclists),
             ranges: RangeLists::new(debug_ranges, debug_rnglists),
+            file_type: DwarfFileType::Main,
         })
     }
 
@@ -129,6 +133,7 @@ impl<T> Dwarf<T> {
             debug_types: self.debug_types.borrow(&mut borrow),
             locations: self.locations.borrow(&mut borrow),
             ranges: self.ranges.borrow(&mut borrow),
+            file_type: self.file_type,
         }
     }
 }
@@ -376,13 +381,22 @@ impl<R: Reader> Dwarf<R> {
         unit: &Unit<R>,
         offset: LocationListsOffset<R::Offset>,
     ) -> Result<LocListIter<R>> {
-        self.locations.locations(
-            offset,
-            unit.encoding(),
-            unit.low_pc,
-            &self.debug_addr,
-            unit.addr_base,
-        )
+        match self.file_type {
+            DwarfFileType::Main => self.locations.locations(
+                offset,
+                unit.encoding(),
+                unit.low_pc,
+                &self.debug_addr,
+                unit.addr_base,
+            ),
+            DwarfFileType::Dwo => self.locations.locations_dwo(
+                offset,
+                unit.encoding(),
+                unit.low_pc,
+                &self.debug_addr,
+                unit.addr_base,
+            ),
+        }
     }
 
     /// Try to return an attribute value as a location list offset.
@@ -519,17 +533,26 @@ impl<R: Reader> Unit<R> {
     pub fn new(dwarf: &Dwarf<R>, header: UnitHeader<R>) -> Result<Self> {
         let abbreviations = header.abbreviations(&dwarf.debug_abbrev)?;
         let mut unit = Unit {
-            header,
             abbreviations,
             name: None,
             comp_dir: None,
             low_pc: 0,
-            // Defaults to 0 for GNU extensions.
-            str_offsets_base: DebugStrOffsetsBase(R::Offset::from_u8(0)),
+            str_offsets_base: DebugStrOffsetsBase::default_for_encoding_and_file(
+                header.encoding(),
+                dwarf.file_type,
+            ),
+            // NB: Because the .debug_addr section never lives in a .dwo, we can assume its base is always 0 or provided.
             addr_base: DebugAddrBase(R::Offset::from_u8(0)),
-            loclists_base: DebugLocListsBase(R::Offset::from_u8(0)),
-            rnglists_base: DebugRngListsBase(R::Offset::from_u8(0)),
+            loclists_base: DebugLocListsBase::default_for_encoding_and_file(
+                header.encoding(),
+                dwarf.file_type,
+            ),
+            rnglists_base: DebugRngListsBase::default_for_encoding_and_file(
+                header.encoding(),
+                dwarf.file_type,
+            ),
             line_program: None,
+            header,
         };
         let mut name = None;
         let mut comp_dir = None;
@@ -646,7 +669,9 @@ impl<R: Reader> Unit<R> {
     pub fn copy_relocated_attributes(&mut self, other: &Unit<R>) {
         self.low_pc = other.low_pc;
         self.addr_base = other.addr_base;
-        self.rnglists_base = other.rnglists_base;
+        if self.header.version() < 5 {
+            self.rnglists_base = other.rnglists_base;
+        }
     }
 }
 

src/read/lists.rs

@@ -0,0 +1,67 @@
+use crate::common::{Encoding, Format};
+use crate::read::{Error, Reader, Result};
+
+#[derive(Debug, Clone, Copy)]
+pub(crate) struct ListsHeader {
+    encoding: Encoding,
+    offset_entry_count: u32,
+}
+
+impl Default for ListsHeader {
+    fn default() -> Self {
+        ListsHeader {
+            encoding: Encoding {
+                format: Format::Dwarf32,
+                version: 5,
+                address_size: 0,
+            },
+            offset_entry_count: 0,
+        }
+    }
+}
+
+impl ListsHeader {
+    /// Return the serialized size of the table header.
+    #[allow(dead_code)]
+    #[inline]
+    fn size(self) -> u8 {
+        // initial_length + version + address_size + segment_selector_size + offset_entry_count
+        ListsHeader::size_for_encoding(self.encoding)
+    }
+
+    /// Return the serialized size of the table header.
+    #[inline]
+    pub(crate) fn size_for_encoding(encoding: Encoding) -> u8 {
+        // initial_length + version + address_size + segment_selector_size + offset_entry_count
+        encoding.format.initial_length_size() + 2 + 1 + 1 + 4
+    }
+}
+
+// TODO: add an iterator over headers in the appropriate sections section
+#[allow(dead_code)]
+fn parse_header<R: Reader>(input: &mut R) -> Result<ListsHeader> {
+    let (length, format) = input.read_initial_length()?;
+    input.truncate(length)?;
+
+    let version = input.read_u16()?;
+    if version != 5 {
+        return Err(Error::UnknownVersion(u64::from(version)));
+    }
+
+    let address_size = input.read_u8()?;
+    let segment_selector_size = input.read_u8()?;
+    if segment_selector_size != 0 {
+        return Err(Error::UnsupportedSegmentSize);
+    }
+    let offset_entry_count = input.read_u32()?;
+
+    let encoding = Encoding {
+        format,
+        version,
+        address_size,
+    };
+    Ok(ListsHeader {
+        encoding,
+        offset_entry_count,
+    })
+}