WIP

src/common/loadable.rs

@@ -1,7 +1,13 @@
+use std::collections::HashMap;
 use std::iter::once;
 
-use crate::common::{Relocatable, Section, SectionInfo, Symbol};
 use crate::error::Error;
+use crate::{
+    common::{Relocatable, Section, SectionInfo, Symbol},
+    error::{LinkError, Trace, LE_GLOBAL_SYMBOL_DUPLICATE},
+};
+
+use super::Binding;
 
 pub enum SegmentType {
     Text,
@@ -38,7 +44,7 @@ pub struct Loadable {
     rodata: SegmentSections,
     data: SegmentSections,
     bss: SegmentSections,
-    // symbol_map: HashMap<Rc<str>, Symbol>,
+    symbol_map: HashMap<String, Vec<Symbol>>,
 }
 
 impl Loadable {
@@ -53,36 +59,18 @@ impl Loadable {
         Ok(())
     }
 
-    pub fn process_symbol<R>(&mut self, _symbol: Symbol, _objects: &[R]) -> Result<(), Error>
+    pub fn process_symbol(&mut self, name: &str, symbol: Symbol) -> Result<(), Error> {
-    where
+        if symbol.binding == Binding::Local {
-        R: Relocatable,
+            return Ok(()); // local symbols are not stored
-    {
+        }
+
+        if let Some(existing) = self.symbol_map.get_mut(name) {
+            existing.push(symbol);
+        } else {
+            self.symbol_map.insert(name.to_owned(), Vec::new());
+        }
+
         Ok(())
-        // if symbol.binding == Binding::Local {
-        //     return Ok(()); // local symbols are not stored
-        // }
-
-        // let key = Rc::from(symbol.name.as_str());
-        // let object_index = symbol.object_index;
-
-        // if let Some(prev) = self.symbol_map.insert(key, symbol) {
-        //     let prev_trace = Trace {
-        //         origin: objects[prev.object_index].origin().into(),
-        //         ..Default::default()
-        //     };
-        //     let symbol_trace = Trace {
-        //         origin: objects[object_index].origin().into(),
-        //         ..Default::default()
-        //     };
-        //     let link_error = LinkError {
-        //         message: format!("Duplicate global symbol found: {}", prev.name),
-        //         code: LE_GLOBAL_SYMBOL_DUPLICATE,
-        //         traces: vec![prev_trace, symbol_trace],
-        //     };
-        //     Err(Error::LinkingError(link_error))
-        // } else {
-        //     Ok(())
-        // }
     }
 
     pub fn segment_sections(&self) -> impl Iterator<Item = &SegmentSections> {

src/common/relocatable.rs

@@ -7,7 +7,7 @@ use super::{Section, Symbol};
 
 pub type SectionIterBox<'iter> = Box<dyn Iterator<Item = Result<Section, Error>> + 'iter>;
 
-pub type SymbolIterBox<'iter> = Box<dyn Iterator<Item = Symbol> + '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

src/common/symbol.rs

@@ -1,3 +1,9 @@
+use elf_utilities::relocation;
+
+use crate::error::Error;
+
+use super::Relocatable;
+
 #[derive(Clone, Copy, PartialEq, Eq)]
 pub enum Binding {
     Local,
@@ -7,7 +13,6 @@ pub enum Binding {
 
 pub struct Symbol {
     pub object_index: usize,
-    pub name: String,
     pub binding: Binding,
     pub address: u64,
     pub size: u64,

src/error.rs

@@ -3,7 +3,7 @@ use std::{
     path::PathBuf,
 };
 
-// pub const LE_GLOBAL_SYMBOL_DUPLICATE: u32 = 10001;
+pub const LE_GLOBAL_SYMBOL_DUPLICATE: u32 = 10001;
 
 #[derive(Debug)]
 pub enum Error {

src/formats/elf/object.rs

@@ -8,7 +8,7 @@ use crate::common::{Binding, Relocatable, Section, SectionInfo, Symbol, SymbolIt
 use crate::{common::SectionIterBox, error::Error};
 use elf_utilities::{
     header::Ehdr64,
-    section::Shdr64,
+    section::{Contents64, Shdr64, Type},
     symbol::{Bind, Symbol64},
 };
 use memmap::Mmap;
@@ -18,7 +18,8 @@ pub struct ElfObject {
     origin: PathBuf,
     data: Mmap,
     ehdr: Ehdr64,
-    sh_name_offset: usize,
+    shdrs: Vec<Shdr64>,
+    symbols: Vec<Symbol64>,
 }
 
 impl ElfObject {
@@ -31,22 +32,26 @@ impl ElfObject {
         let file = File::open(str_origin)?;
         let data = unsafe { Mmap::map(&file)? };
         let ehdr = parse_elf_header(&data)?;
-        let shstrtab = parse_shstrtab(&data, &ehdr)?;
 
-        let result = ElfObject {
+        let mut result = ElfObject {
             object_index,
             origin,
             data,
             ehdr,
-            sh_name_offset: shstrtab.sh_offset as usize,
+            shdrs: Vec::new(),
+            symbols: Vec::new(),
         };
 
+        // TODO: make paralellization from outside possible?
+        result.populate()?;
+
         Ok(result)
     }
 
-    fn section_name(&self, shdr: &Shdr64) -> Result<&str, Error> {
+    fn resolve_name(&self, name_idx: usize, sh_index: usize) -> Result<&str, Error> {
-        let idx: usize = shdr.sh_name as usize;
+        let strtab = self.parse_strtab(sh_index)?;
-        let start = self.sh_name_offset + idx;
+        let strtab_offset: usize = strtab.sh_offset as usize;
+        let start = strtab_offset + name_idx;
 
         let mut i = start;
         while self.data[i] != 0 {
@@ -61,17 +66,57 @@ impl ElfObject {
         std::str::from_utf8(&self.data[start..i]).map_err(|err| Error::ParseError(err.into()))
     }
 
-    fn make_section(&self, offset: usize, sh: &Shdr64) -> Option<Result<Section, Error>> {
+    // TODO: consider caching these
-        let name = match self.section_name(sh) {
+    fn parse_strtab(&self, sh_index: usize) -> Result<Shdr64, Error> {
+        let sh_start: usize = self.ehdr.e_shoff as usize;
+        let sh_size: usize = self.ehdr.e_shentsize.into();
+        let offset: usize = sh_start + sh_size * sh_index;
+
+        bincode::deserialize(&self.data[offset..offset + sh_size])
+            .map_err(|err| Error::ParseError(err))
+    }
+
+    fn populate(&mut self) -> Result<(), Error> {
+        for i in 0..self.ehdr.e_shnum {
+            let i_usize: usize = i.into();
+            let sh_start = self.ehdr.e_shoff as usize;
+            let sh_size = usize::from(self.ehdr.e_shentsize);
+
+            let offset: usize = sh_start + i_usize * sh_size;
+            let shdr: Shdr64 = bincode::deserialize(&self.data[offset..offset + sh_size])
+                .map_err(|err| Error::ParseError(err))?;
+
+            if shdr.get_type() == Type::SymTab {
+                let ent_size = shdr.sh_entsize as usize;
+                let offset = shdr.sh_offset as usize;
+                let count = shdr.sh_size as usize / ent_size;
+
+                for s_i in 0..count {
+                    let start = offset + s_i * ent_size;
+                    let s64 = Symbol64::deserialize(&self.data, start)
+                        .map_err(|err| Error::ParseError(err))?;
+                    self.symbols.push(s64);
+                }
+            }
+            // self.symbols.push();
+            self.shdrs.push(shdr);
+        }
+
+        Ok(())
+    }
+
+    fn make_section(&self, shdr: &Shdr64) -> Option<Result<Section, Error>> {
+        let sh_index: usize = self.ehdr.e_shstrndx.into();
+        let name = match self.resolve_name(shdr.sh_name as usize, sh_index) {
             Ok(n) => n,
             Err(err) => return Some(Err(err)),
         };
 
         let mut si = SectionInfo {
             object_index: self.object_index,
-            file_size: sh.sh_size,
+            file_size: shdr.sh_size,
-            data_size: sh.sh_size,
+            data_size: shdr.sh_size,
-            offset: offset as u64,
+            offset: shdr.sh_offset,
         };
 
         if name.starts_with(".bss") {
@@ -88,7 +133,7 @@ impl ElfObject {
         }
     }
 
-    fn make_symbol(&self, s64: &Symbol64, strtab: &[u8]) -> Symbol {
+    fn make_symbol(&self, s64: &Symbol64) -> Symbol {
         let binding = match s64.get_bind() {
             Bind::Global => Binding::Global,
             Bind::Local => Binding::Local,
@@ -98,7 +143,6 @@ impl ElfObject {
 
         Symbol {
             object_index: self.object_index,
-            name: parse_strtab_name(&strtab, s64.st_name),
             binding,
             address: s64.st_value,
             size: s64.st_size,
@@ -116,20 +160,12 @@ impl Relocatable for ElfObject {
     }
 
     fn sections(&self) -> SectionIterBox {
-        let iter = (0..self.ehdr.e_shnum).into_iter().filter_map(move |i| {
+        let iter = self
-            let i_usize = usize::from(i);
+            .shdrs
-            let sh_start = self.ehdr.e_shoff as usize;
+            .iter()
-            let sh_size = usize::from(self.ehdr.e_shentsize);
+            .filter_map(move |shdr| match shdr.get_type() {
-
+                Type::ProgBits | Type::NoBits => self.make_section(shdr),
-            let offset: usize = sh_start + i_usize * sh_size;
+                _ => None,
-            let shr: Result<Shdr64, Error> =
-                bincode::deserialize(&self.data[offset..offset + sh_size])
-                    .map_err(|err| Error::ParseError(err));
-
-            match shr {
-                Err(err) => Some(Err(err)),
-                Ok(sh) => self.make_section(offset, &sh),
-            }
             });
 
         Box::new(iter)
@@ -142,36 +178,13 @@ impl Relocatable for ElfObject {
     }
 
     fn symbols(&self) -> SymbolIterBox {
-        // if let Some(strtab_section) = self
+        Box::new(self.symbols.iter().map(move |s| {
-        //     .elf
+            // TODO: change SymbolIterBox to allow errors!
-        //     .first_section_by(|s64| s64.header.get_type() == Type::StrTab && s64.name == ".strtab")
+            let name = self
-        // {
+                .resolve_name(s.st_name as usize, s.st_other as usize)
-        //     let strtab = match &strtab_section.contents {
+                .expect("Symbol name resolution");
-        //         Contents64::Raw(bytes) => bytes,
+            (name, self.make_symbol(s))
-        //         _ => panic!("Unexpected strtab content type"),
+        }))
-        //     };
-
-        //     let iter = self
-        //         .elf
-        //         .sections
-        //         .iter()
-        //         .filter_map(move |s| match &s.contents {
-        //             Contents64::Symbols(symbols) => {
-        //                 Some(symbols.iter().filter_map(move |sym| match sym.get_bind() {
-        //                     Bind::Global | Bind::Local | Bind::Weak => {
-        //                         Some(self.make_symbol(sym, strtab))
-        //                     }
-        //                     _ => None,
-        //                 }))
-        //             }
-        //             _ => None,
-        //         })
-        //         .flatten();
-
-        //     Box::new(iter)
-        // } else {
-        Box::new(std::iter::empty())
-        //}
     }
 }
 
@@ -221,12 +234,3 @@ fn parse_elf_header(data: &Mmap) -> Result<Ehdr64, Error> {
 
     Ok(ehdr)
 }
-
-fn parse_shstrtab(data: &Mmap, ehdr: &Ehdr64) -> Result<Shdr64, Error> {
-    let idx: usize = ehdr.e_shstrndx.into();
-    let sh_start: usize = ehdr.e_shoff as usize;
-    let sh_size: usize = ehdr.e_shentsize.into();
-    let offset: usize = sh_start + sh_size * idx;
-
-    bincode::deserialize(&data[offset..offset + sh_size]).map_err(|err| Error::ParseError(err))
-}

src/linker.rs

@@ -50,8 +50,8 @@ where
 
     fn process_symbols(&mut self) -> Result<(), Error> {
         for r in &self.relocatables {
-            for symbol in r.symbols() {
+            for (name, symbol) in r.symbols() {
-                self.loadable.process_symbol(symbol, &self.relocatables)?;
+                self.loadable.process_symbol(name, symbol)?;
             }
         }