add basics of symbol processing

src/common/loadable.rs

@@ -1,8 +1,12 @@
-use std::iter::once;
+use std::{collections::HashMap, iter::once, rc::Rc};
 
-use crate::common::SectionInfo;
-use crate::common::{Relocatable, Section};
 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,
@@ -40,6 +44,8 @@ pub struct Loadable {
     rodata: SegmentSections,
     data: SegmentSections,
     bss: SegmentSections,
+
+    symbol_map: HashMap<Rc<str>, Symbol>,
 }
 
 impl Loadable {
@@ -54,6 +60,38 @@ impl Loadable {
         Ok(())
     }
 
+    pub fn process_symbol<R>(&mut self, symbol: Symbol, objects: &[R]) -> Result<(), Error>
+    where
+        R: Relocatable,
+    {
+        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 const fn section_names() -> &'static [&'static str] {
         &[".text", ".rodata", ".data", ".bss"]
     }

src/common/symbol.rs

@@ -1,4 +1,4 @@
-#[derive(Clone, Copy)]
+#[derive(Clone, Copy, PartialEq, Eq)]
 pub enum Binding {
     Local,
     Global,
@@ -6,6 +6,7 @@ pub enum Binding {
 }
 
 pub struct Symbol {
+    pub object_index: usize,
     pub name: String,
     pub binding: Binding,
     pub address: u64,

src/error.rs

@@ -3,6 +3,8 @@ use std::{
     path::PathBuf,
 };
 
+pub const LE_GLOBAL_SYMBOL_DUPLICATE: u32 = 701;
+
 #[derive(Debug)]
 pub enum Error {
     IOError(Box<dyn std::error::Error>),
@@ -11,19 +13,24 @@ pub enum Error {
     InvalidSectionData,
     ParseError(Box<dyn std::error::Error>),
     TryFromIntError,
-    LinkingError(Trace),
+    LinkingError(LinkError),
 }
 
-// "backtrace" for error origin
 #[derive(Debug, Default)]
 pub struct Trace {
-    pub code: u32, // specific trace code, should be >= 700
-    pub message: &'static str,
     pub origin: PathBuf,
     pub offset: u64, // 0 indicates unknown/invalid
     pub source_info: Option<Box<dyn SourceInfo>>,
 }
 
+// "backtrace" for error origin
+#[derive(Debug, Default)]
+pub struct LinkError {
+    pub code: u32, // specific trace code, should be >= 700
+    pub message: String,
+    pub traces: Vec<Trace>,
+}
+
 impl Error {
     // used for programmatic input/output handling, IDE's etc.
     pub fn code(&self) -> u32 {
@@ -34,7 +41,7 @@ impl Error {
             Error::InvalidSectionData => 400,
             Error::ParseError(_) => 500,
             Error::TryFromIntError => 600,
-            Error::LinkingError(trace) => trace.code,
+            Error::LinkingError(le) => le.code,
         }
     }
 }
@@ -73,24 +80,26 @@ impl Display for Error {
     }
 }
 
-impl Display for Trace {
+impl Display for LinkError {
     fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result {
         let mut new_line = ""; // new line in case of previous info only
 
-        if let Some(origin) = self.origin.to_str() {
+        for trace in &self.traces {
+            if let Some(origin) = trace.origin.to_str() {
                 new_line = "\n";
                 write!(f, "\n\t{}", origin)?;
             }
 
-        if self.offset > 0 {
+            if trace.offset > 0 {
                 new_line = "\n";
-            write!(f, "\n@offset: {}", self.offset)?;
+                write!(f, "\n@offset: {}", trace.offset)?;
             }
 
-        if let Some(si) = &self.source_info {
+            if let Some(si) = &trace.source_info {
                 new_line = "\n";
                 write!(f, "\n{}", si)?;
             }
+        }
 
         write!(f, "{}{}", new_line, self.message)
     }

src/formats/elf/object.rs

@@ -39,6 +39,23 @@ impl ElfObject {
 
         Ok(result)
     }
+
+    fn make_symbol(&self, s64: &Symbol64, strtab: &[u8]) -> 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,
+            name: parse_strtab_name(&strtab, s64.st_name),
+            binding,
+            address: s64.st_value,
+            size: s64.st_size,
+        }
+    }
 }
 
 impl Relocatable for ElfObject {
@@ -124,7 +141,7 @@ impl Relocatable for ElfObject {
                     Contents64::Symbols(symbols) => {
                         Some(symbols.iter().filter_map(move |sym| match sym.get_bind() {
                             Bind::Global | Bind::Local | Bind::Weak => {
-                                Some(make_symbol(sym, strtab))
+                                Some(self.make_symbol(sym, strtab))
                             }
                             _ => None,
                         }))
@@ -158,23 +175,7 @@ impl Display for ElfObject {
     }
 }
 
-fn make_symbol(s64: &Symbol64, strtab: &Vec<u8>) -> Symbol {
+fn parse_strtab_name(strtab: &[u8], idx: u32) -> String {
-    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 {
-        name: parse_strtab_name(&strtab, s64.st_name),
-        binding,
-        address: s64.st_value,
-        size: s64.st_size,
-    }
-}
-
-fn parse_strtab_name(strtab: &Vec<u8>, idx: u32) -> String {
     let bytes: Vec<u8> = strtab
         .iter()
         .skip(idx as usize)

src/formats/elf/output.rs

@@ -216,15 +216,15 @@ fn make_strtab() -> (Vec<u8>, Vec<usize>) {
 }
 
 fn get_start_offset(loadable: &Loadable) -> Result<u64, Error> {
-    use crate::error::Trace;
+    use crate::error::LinkError;
 
     loadable.start_offset.ok_or_else(|| {
-        let trace = Trace {
+        let link_error = LinkError {
-            message: "Program entrypoint not found",
+            message: "Program entrypoint not found".into(),
             ..Default::default()
         };
 
-        Error::LinkingError(trace)
+        Error::LinkingError(link_error)
     })
 }
 

src/linker.rs

@@ -32,8 +32,8 @@ where
         let r = R::new(origin, DataIndex::for_object(self.relocatables.len()))?;
 
         // TODO: parallelize?
-        for s in r.sections() {
+        for section in r.sections() {
-            self.loadable.process_section(s?)?;
+            self.loadable.process_section(section?)?;
         }
 
         self.relocatables.push(r);
@@ -42,10 +42,22 @@ where
     }
 
     pub fn link(mut self) -> Result<PathBuf, Error> {
+        self.process_symbols()?;
+
         self.loadable.start_offset = Some(4096); // TODO: get from .start symbol location
         self.output.finalize(&self.relocatables, &self.loadable)
     }
 
+    fn process_symbols(&mut self) -> Result<(), Error> {
+        for r in &self.relocatables {
+            for symbol in r.symbols() {
+                self.loadable.process_symbol(symbol, &self.relocatables)?;
+            }
+        }
+
+        Ok(())
+    }
+
     // pub fn object_path(origin: &Path) -> Result<PathBuf, Error> {
     //     let mut dest = std::fs::canonicalize(origin)?;
     //     if !dest.pop() {