indirect via object

src/common/loadable.rs

@@ -3,12 +3,10 @@ use std::iter::once;
 
 use crate::error::Error;
 use crate::{
-    common::{Relocatable, Section, SectionInfo, Symbol},
-    error::{LinkError, Trace, LE_GLOBAL_SYMBOL_DUPLICATE},
+    common::{Relocatable, Section, SectionInfo, SymbolIndex},
+    error::LinkError,
 };
 
-use super::Binding;
-
 pub enum SegmentType {
     Text,
     Data,
@@ -44,7 +42,7 @@ pub struct Loadable {
     rodata: SegmentSections,
     data: SegmentSections,
     bss: SegmentSections,
-    symbol_map: HashMap<String, Vec<Symbol>>,
+    symbol_map: HashMap<String, Vec<SymbolIndex>>,
 }
 
 impl Loadable {
@@ -59,15 +57,26 @@ impl Loadable {
         Ok(())
     }
 
-    pub fn process_symbol(&mut self, name: &str, symbol: Symbol) -> Result<(), Error> {
-        if symbol.binding == Binding::Local {
+    pub fn process_symbol<R>(&mut self, index: usize, object: &R) -> Result<(), Error>
+    where
+        R: Relocatable,
+    {
+        if !object.symbol_needs_resolving(index) {
             return Ok(()); // local symbols are not stored
         }
 
+        let name = object.symbol_name(index)?;
+        eprintln!("NAME: {}", name);
+
+        let si = SymbolIndex {
+            object_index: object.object_index(),
+            symbol_index: index,
+        };
+
         if let Some(existing) = self.symbol_map.get_mut(name) {
-            existing.push(symbol);
+            existing.push(si);
         } else {
-            self.symbol_map.insert(name.to_owned(), Vec::new());
+            self.symbol_map.insert(name.to_owned(), vec![si]);
         }
 
         Ok(())
@@ -151,8 +160,6 @@ impl Loadable {
     }
 
     pub fn start_offset(&self) -> Result<u64, Error> {
-        use crate::error::LinkError;
-
         self.start_offset.ok_or_else(|| {
             let link_error = LinkError {
                 message: "Program entrypoint not found".into(),

src/common/relocatable.rs

@@ -3,23 +3,27 @@ use std::{fmt::Display, path::PathBuf};
 
 use crate::error::Error;
 
-use super::{Section, Symbol};
+use super::Section;
 
 pub type SectionIterBox<'iter> = Box<dyn Iterator<Item = Result<Section, Error>> + 'iter>;
 
-pub type SymbolIterBox<'iter> = Box<dyn Iterator<Item = (&'iter str, Symbol)> + 'iter>;
-
 /// Contains all the needed getters to construct a final
 /// mushed and relocated executable from an object file
 pub trait Relocatable: Display + Sized {
     // + TryFrom<PathBuf, Error = Error> {
     fn new(origin: PathBuf, object_index: usize) -> Result<Self, Error>;
 
+    fn object_index(&self) -> usize;
+
     fn origin(&self) -> &Path; // not same as section's path since this one's supposed to be cannonical
 
     fn sections(&self) -> SectionIterBox;
 
-    fn symbols(&self) -> SymbolIterBox;
+    fn symbol_count(&self) -> usize;
+
+    fn symbol_name(&self, index: usize) -> Result<&str, Error>;
+
+    fn symbol_needs_resolving(&self, index: usize) -> bool;
 
     fn bytes(&self, offset: u64, size: u64) -> Result<&[u8], Error>;
 }

src/common/symbol.rs

@@ -1,19 +1,7 @@
-use elf_utilities::relocation;
+use std::fmt::Debug;
 
-use crate::error::Error;
-
-use super::Relocatable;
-
-#[derive(Clone, Copy, PartialEq, Eq)]
-pub enum Binding {
-    Local,
-    Global,
-    Weak,
-}
-
-pub struct Symbol {
+#[derive(Default, Debug)]
+pub struct SymbolIndex {
     pub object_index: usize,
-    pub binding: Binding,
-    pub address: u64,
-    pub size: u64,
+    pub symbol_index: usize,
 }

src/error.rs

@@ -11,6 +11,8 @@ pub enum Error {
     DataError(PathBuf), // generic "something wrong with data" error, not specific to parsing
     InvalidObjectType(u32),
     InvalidSectionName,
+    InvalidSymbolIndex,
+    InvalidSymbolName,
     InvalidSectionData,
     InvalidInput, // input is completely wrong (e.g. no magic bytes etc.)
     ParseError(Box<dyn std::error::Error>), // error on parsing input
@@ -41,6 +43,8 @@ impl Error {
             Error::DataError(_) => 800,
             Error::InvalidObjectType(_) => 200,
             Error::InvalidSectionName => 300,
+            Error::InvalidSymbolIndex => 325,
+            Error::InvalidSymbolName => 350,
             Error::InvalidSectionData => 400,
             Error::ParseError(_) => 500,
             Error::TryFromIntError => 600,
@@ -77,6 +81,8 @@ impl Display for Error {
             Error::DataError(origin) => write!(f, "Data error in {:?}", origin),
             Error::InvalidObjectType(ot) => write!(f, "Invalid object type: {}", ot),
             Error::InvalidSectionName => write!(f, "Invalid section name"),
+            Error::InvalidSymbolIndex => write!(f, "Invalid symbol index"),
+            Error::InvalidSymbolName => write!(f, "Invalid symbol name"),
             Error::InvalidSectionData => write!(f, "Invalid section data"),
             Error::ParseError(err) => write!(f, "Parse error: {}", err),
             Error::TryFromIntError => write!(f, "Integer conversion error"),

src/formats/elf/object.rs

@@ -4,12 +4,12 @@ use std::{
     path::{Path, PathBuf},
 };
 
-use crate::common::{Binding, Relocatable, Section, SectionInfo, Symbol, SymbolIterBox};
+use crate::common::{Relocatable, Section, SectionInfo};
 use crate::{common::SectionIterBox, error::Error};
 use elf_utilities::{
     header::Ehdr64,
-    section::{Contents64, Shdr64, Type},
-    symbol::{Bind, Symbol64},
+    section::{Shdr64, Type},
+    symbol::{Bind, Symbol64, Type as SymType},
 };
 use memmap::Mmap;
 
@@ -19,7 +19,7 @@ pub struct ElfObject {
     data: Mmap,
     ehdr: Ehdr64,
     shdrs: Vec<Shdr64>,
-    symbols: Vec<Symbol64>,
+    symbols: Vec<(usize, Symbol64)>,
 }
 
 impl ElfObject {
@@ -49,6 +49,16 @@ impl ElfObject {
     }
 
     fn resolve_name(&self, name_idx: usize, sh_index: usize) -> Result<&str, Error> {
+        if name_idx == 0 {
+            let error = if sh_index == self.ehdr.e_shstrndx.into() {
+                Error::InvalidSectionName
+            } else {
+                Error::InvalidSymbolName
+            };
+
+            return Err(error);
+        }
+
         let strtab = self.parse_strtab(sh_index)?;
         let strtab_offset: usize = strtab.sh_offset as usize;
         let start = strtab_offset + name_idx;
@@ -91,14 +101,21 @@ impl ElfObject {
                 let offset = shdr.sh_offset as usize;
                 let count = shdr.sh_size as usize / ent_size;
 
-                for s_i in 0..count {
+                for s_i in 1..count {
+                    // skip null-symbol
                     let start = offset + s_i * ent_size;
                     let s64 = Symbol64::deserialize(&self.data, start)
                         .map_err(|err| Error::ParseError(err))?;
-                    self.symbols.push(s64);
+
+                    match s64.get_type() {
+                        SymType::Object | SymType::Func | SymType::NoType => {
+                            self.symbols.push((shdr.sh_link as usize, s64))
                         }
+                        _ => {}
                     }
-            // self.symbols.push();
+                }
+            }
+
             self.shdrs.push(shdr);
         }
 
@@ -132,22 +149,6 @@ impl ElfObject {
             None
         }
     }
-
-    fn make_symbol(&self, s64: &Symbol64) -> Symbol {
-        let binding = match s64.get_bind() {
-            Bind::Global => Binding::Global,
-            Bind::Local => Binding::Local,
-            Bind::Weak => Binding::Weak,
-            _ => panic!("Unexpected binding type encountered on symbol"), // this is screened!
-        };
-
-        Symbol {
-            object_index: self.object_index,
-            binding,
-            address: s64.st_value,
-            size: s64.st_size,
-        }
-    }
 }
 
 impl Relocatable for ElfObject {
@@ -155,6 +156,10 @@ impl Relocatable for ElfObject {
         ElfObject::new(origin, object_index)
     }
 
+    fn object_index(&self) -> usize {
+        self.object_index
+    }
+
     fn origin(&self) -> &Path {
         &self.origin
     }
@@ -177,14 +182,23 @@ impl Relocatable for ElfObject {
         Ok(&self.data[o..o + s])
     }
 
-    fn symbols(&self) -> SymbolIterBox {
-        Box::new(self.symbols.iter().map(move |s| {
-            // TODO: change SymbolIterBox to allow errors!
-            let name = self
-                .resolve_name(s.st_name as usize, s.st_other as usize)
-                .expect("Symbol name resolution");
-            (name, self.make_symbol(s))
-        }))
+    fn symbol_count(&self) -> usize {
+        self.symbols.len()
+    }
+
+    fn symbol_name(&self, index: usize) -> Result<&str, Error> {
+        if let Some(symbol) = self.symbols.get(index) {
+            self.resolve_name(symbol.1.st_name as usize, symbol.0)
+        } else {
+            Err(Error::InvalidSymbolIndex)
+        }
+    }
+
+    fn symbol_needs_resolving(&self, index: usize) -> bool {
+        match self.symbols[index].1.get_bind() {
+            Bind::Global | Bind::Weak => true,
+            _ => false,
+        }
     }
 }
 

src/linker.rs

@@ -1,7 +1,7 @@
 use std::{fmt::Display, path::PathBuf};
 
 use crate::{
-    common::{Loadable, Output, Relocatable},
+    common::{Loadable, Output, Relocatable, SymbolIndex},
     error::Error,
 };
 
@@ -50,24 +50,15 @@ where
 
     fn process_symbols(&mut self) -> Result<(), Error> {
         for r in &self.relocatables {
-            for (name, symbol) in r.symbols() {
-                self.loadable.process_symbol(name, symbol)?;
+            let symbol_count = r.symbol_count();
+
+            for i in 0..symbol_count {
+                self.loadable.process_symbol(i, r)?;
             }
         }
 
         Ok(())
     }
-
-    // pub fn object_path(origin: &Path) -> Result<PathBuf, Error> {
-    //     let mut dest = std::fs::canonicalize(origin)?;
-    //     if !dest.pop() {
-    //         let err = std::io::Error::new(ErrorKind::Other, "Destination path invalid");
-    //         Err(Error::IOError(Box::new(err)))
-    //     } else {
-    //         dest.push("rld.out");
-    //         Ok(dest)
-    //     }
-    // }
 }
 
 impl<R, O> Display for Linker<R, O>