read: cache abbreviations at offset 0

This optimises for the case where there is a single set of
abbreviations that is used for all units.

src/read/abbrev.rs

@@ -1,16 +1,18 @@
 //! Functions for parsing DWARF debugging abbreviations.
 
 use alloc::collections::btree_map;
+use alloc::sync::Arc;
 use alloc::vec::Vec;
 use core::convert::TryFrom;
 use core::fmt::{self, Debug};
 use core::iter::FromIterator;
 use core::ops::Deref;
+use std::sync::Mutex;
 
 use crate::common::{DebugAbbrevOffset, Encoding, SectionId};
 use crate::constants;
 use crate::endianity::Endianity;
-use crate::read::{EndianSlice, Error, Reader, Result, Section, UnitHeader};
+use crate::read::{EndianSlice, Error, Reader, ReaderOffset, Result, Section, UnitHeader};
 
 /// The `DebugAbbrev` struct represents the abbreviations describing
 /// `DebuggingInformationEntry`s' attribute names and forms found in the
@@ -100,6 +102,44 @@ impl<R> From<R> for DebugAbbrev<R> {
     }
 }
 
+/// A cache of previously parsed `Abbreviations`.
+///
+/// Currently this only caches the abbreviations for offset 0,
+/// since this is a common case in which abbreviations are reused.
+/// This strategy may change in future if there is sufficient need.
+#[derive(Debug, Default)]
+pub struct AbbreviationsCache {
+    abbreviations: Mutex<Option<Arc<Abbreviations>>>,
+}
+
+impl AbbreviationsCache {
+    /// Create an empty abbreviations cache.
+    pub fn new() -> Self {
+        Self::default()
+    }
+
+    /// Parse a unit's abbreviations.
+    ///
+    /// This uses or updates the cache as required.
+    pub fn get<R: Reader>(
+        &self,
+        unit: &UnitHeader<R>,
+        debug_abbrev: &DebugAbbrev<R>,
+    ) -> Result<Arc<Abbreviations>> {
+        let offset = unit.debug_abbrev_offset();
+        if offset.0 != R::Offset::from_u8(0) {
+            return debug_abbrev.abbreviations(offset).map(Arc::new);
+        }
+        let mut cache = self.abbreviations.lock().unwrap();
+        if let Some(abbreviations) = &*cache {
+            return Ok(abbreviations.clone());
+        }
+        let abbreviations = debug_abbrev.abbreviations(offset).map(Arc::new)?;
+        *cache = Some(abbreviations.clone());
+        Ok(abbreviations)
+    }
+}
+
 /// A set of type abbreviations.
 ///
 /// Construct an `Abbreviations` instance with the

src/read/dwarf.rs

@@ -9,11 +9,11 @@ use crate::common::{
 };
 use crate::constants;
 use crate::read::{
-    Abbreviations, AttributeValue, DebugAbbrev, DebugAddr, DebugAranges, DebugCuIndex, DebugInfo,
-    DebugInfoUnitHeadersIter, DebugLine, DebugLineStr, DebugLoc, DebugLocLists, DebugRngLists,
-    DebugStr, DebugStrOffsets, DebugTuIndex, DebugTypes, DebugTypesUnitHeadersIter,
-    DebuggingInformationEntry, EntriesCursor, EntriesRaw, EntriesTree, Error,
-    IncompleteLineProgram, LocListIter, LocationLists, Range, RangeLists, RawLocListIter,
+    Abbreviations, AbbreviationsCache, AttributeValue, DebugAbbrev, DebugAddr, DebugAranges,
+    DebugCuIndex, DebugInfo, DebugInfoUnitHeadersIter, DebugLine, DebugLineStr, DebugLoc,
+    DebugLocLists, DebugRngLists, DebugStr, DebugStrOffsets, DebugTuIndex, DebugTypes,
+    DebugTypesUnitHeadersIter, DebuggingInformationEntry, EntriesCursor, EntriesRaw, EntriesTree,
+    Error, IncompleteLineProgram, LocListIter, LocationLists, Range, RangeLists, RawLocListIter,
     RawRngListIter, Reader, ReaderOffset, ReaderOffsetId, Result, RngListIter, Section, UnitHeader,
     UnitIndex, UnitIndexSectionIterator, UnitOffset, UnitType,
 };
@@ -59,6 +59,9 @@ pub struct Dwarf<R> {
 
     /// The DWARF sections for a supplementary object file.
     pub sup: Option<Arc<Dwarf<R>>>,
+
+    /// A cache of previously parsed abbreviations for units in this file.
+    pub abbreviations_cache: AbbreviationsCache,
 }
 
 impl<T> Dwarf<T> {
@@ -96,6 +99,7 @@ impl<T> Dwarf<T> {
             ranges: RangeLists::new(debug_ranges, debug_rnglists),
             file_type: DwarfFileType::Main,
             sup: None,
+            abbreviations_cache: AbbreviationsCache::new(),
         })
     }
 
@@ -157,6 +161,7 @@ impl<T> Dwarf<T> {
             ranges: self.ranges.borrow(&mut borrow),
             file_type: self.file_type,
             sup: self.sup().map(|sup| Arc::new(sup.borrow(borrow))),
+            abbreviations_cache: AbbreviationsCache::new(),
         }
     }
 
@@ -192,10 +197,9 @@ impl<R: Reader> Dwarf<R> {
     }
 
     /// Parse the abbreviations for a compilation unit.
-    // TODO: provide caching of abbreviations
     #[inline]
-    pub fn abbreviations(&self, unit: &UnitHeader<R>) -> Result<Abbreviations> {
-        unit.abbreviations(&self.debug_abbrev)
+    pub fn abbreviations(&self, unit: &UnitHeader<R>) -> Result<Arc<Abbreviations>> {
+        self.abbreviations_cache.get(unit, &self.debug_abbrev)
     }
 
     /// Return the string offset at the given index.
@@ -783,6 +787,7 @@ impl<R: Reader> DwarfPackage<R> {
             ranges: RangeLists::new(debug_ranges, debug_rnglists),
             file_type: DwarfFileType::Dwo,
             sup: None,
+            abbreviations_cache: AbbreviationsCache::new(),
         })
     }
 }
@@ -799,7 +804,7 @@ where
     pub header: UnitHeader<R, Offset>,
 
     /// The parsed abbreviations for the unit.
-    pub abbreviations: Abbreviations,
+    pub abbreviations: Arc<Abbreviations>,
 
     /// The `DW_AT_name` attribute of the unit.
     pub name: Option<R>,
@@ -833,7 +838,7 @@ impl<R: Reader> Unit<R> {
     /// Construct a new `Unit` from the given unit header.
     #[inline]
     pub fn new(dwarf: &Dwarf<R>, header: UnitHeader<R>) -> Result<Self> {
-        let abbreviations = header.abbreviations(&dwarf.debug_abbrev)?;
+        let abbreviations = dwarf.abbreviations(&header)?;
         let mut unit = Unit {
             abbreviations,
             name: None,

src/write/loc.rs

@@ -436,6 +436,7 @@ mod tests {
     };
     use crate::LittleEndian;
     use std::collections::HashMap;
+    use std::sync::Arc;
 
     #[test]
     fn test_loc_list() {
@@ -508,7 +509,7 @@ mod tests {
                             DebugInfoOffset(0).into(),
                             read::EndianSlice::default(),
                         ),
-                        abbreviations: read::Abbreviations::default(),
+                        abbreviations: Arc::new(read::Abbreviations::default()),
                         name: None,
                         comp_dir: None,
                         low_pc: 0,

src/write/op.rs

@@ -1071,6 +1071,7 @@ mod tests {
     };
     use crate::LittleEndian;
     use std::collections::HashMap;
+    use std::sync::Arc;
 
     #[test]
     fn test_operation() {
@@ -1578,7 +1579,7 @@ mod tests {
                             DebugInfoOffset(0).into(),
                             read::EndianSlice::new(&[], LittleEndian),
                         ),
-                        abbreviations: read::Abbreviations::default(),
+                        abbreviations: Arc::new(read::Abbreviations::default()),
                         name: None,
                         comp_dir: None,
                         low_pc: 0,

src/write/range.rs

@@ -315,6 +315,7 @@ mod tests {
     };
     use crate::LittleEndian;
     use std::collections::HashMap;
+    use std::sync::Arc;
 
     #[test]
     fn test_range() {
@@ -375,7 +376,7 @@ mod tests {
                             DebugInfoOffset(0).into(),
                             read::EndianSlice::default(),
                         ),
-                        abbreviations: read::Abbreviations::default(),
+                        abbreviations: Arc::new(read::Abbreviations::default()),
                         name: None,
                         comp_dir: None,
                         low_pc: 0,

src/write/unit.rs

@@ -1931,6 +1931,7 @@ mod tests {
     use crate::LittleEndian;
     use std::collections::HashMap;
     use std::mem;
+    use std::sync::Arc;
 
     #[test]
     #[allow(clippy::cyclomatic_complexity)]
@@ -2542,7 +2543,7 @@ mod tests {
 
                         let unit = read::Unit {
                             header: from_unit,
-                            abbreviations: read::Abbreviations::default(),
+                            abbreviations: Arc::new(read::Abbreviations::default()),
                             name: None,
                             comp_dir: None,
                             low_pc: 0,
@@ -3015,7 +3016,7 @@ mod tests {
 
                         let unit = read::Unit {
                             header: from_unit,
-                            abbreviations: read::Abbreviations::default(),
+                            abbreviations: Arc::new(read::Abbreviations::default()),
                             name: None,
                             comp_dir: None,
                             low_pc: 0,