first draft implementation using xmas_elf

.gitignore

@@ -1,3 +1,5 @@
 /target
 
 /assets
+
+/docs

src/error.rs

@@ -1,3 +1,8 @@
+use std::{
+    fmt::{Debug, Display, Formatter},
+    path::PathBuf,
+};
+
 #[derive(Debug)]
 pub enum Error {
     IOError(std::io::Error),
@@ -6,6 +11,21 @@ pub enum Error {
     MissingSectionHeader(&'static str),
     MissingSectionData(&'static str),
     InvalidSectionData,
+    LinkingError(Trace),
+}
+
+// "backtrace" for error origin
+#[derive(Debug)]
+pub struct Trace {
+    origin: PathBuf,
+    offset: u64, // 0 indicates unknown/invalid
+    source_info: Option<Box<dyn SourceInfo>>,
+}
+
+pub trait SourceInfo: Debug + Display {
+    fn line(&self) -> u64;
+
+    // todo
 }
 
 impl From<std::io::Error> for Error {
@@ -13,3 +33,15 @@ impl From<std::io::Error> for Error {
         Self::IOError(err)
     }
 }
+
+impl Display for Error {
+    fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result {
+        write!(f, "TODO!")
+    }
+}
+
+impl Display for Trace {
+    fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result {
+        write!(f, "TODO!")
+    }
+}

src/formats/elf.rs

@@ -1,63 +1,84 @@
-use std::convert::TryFrom;
+use std::{
+    fmt::Display,
+    path::{Path, PathBuf},
+};
 
-use crate::error::Error;
-use crate::parts::{Lazy, Relocatable, Symbol, Text};
+use crate::linker::LinkState;
+use crate::parts::{Lazy, Relocatable, Storage, Symbol};
+use crate::{error::Error, parts::BSI};
 use xmas_elf::header::Type as ElfType;
-use xmas_elf::sections::*;
 use xmas_elf::ElfFile;
 
-impl<'a> Lazy<&'a str, ElfFile<'a>> for Symbol<'a> {
-    fn value(&self, src: &ElfFile<'a>) -> Result<&'a str, Error> {
-        src.get_string(self.index)
-            .map_err(|e| Error::StringError(e))
+mod section;
+mod section_iter;
+
+use section_iter::SectionIter;
+
+pub struct ElfObject<'data> {
+    origin: PathBuf,
+    elf: ElfFile<'data>,
 }
 
-    fn resolved(&self) -> bool {
-        self.str_ref.is_some()
+impl<'data> ElfObject<'data> {
+    pub fn new(storage: &'data mut Storage) -> Result<Self, Error> {
+        let origin = storage.origin()?;
+        let elf = ElfFile::new(storage.bytes()?).map_err(|_| Error::InvalidObjectType(0))?;
+
+        is_relocatable(&elf)?;
+        let result = ElfObject { origin, elf };
+
+        Ok(result)
     }
 }
 
-impl<'a> TryFrom<SectionData<'a>> for Text<'a> {
-    type Error = Error;
-
-    fn try_from(data: SectionData<'a>) -> Result<Self, Error> {
-        match data {
-            SectionData::Undefined(bytes) => Ok(Text { bytes }),
-            _ => Err(Error::InvalidSectionData), // TODO: more context needed
+impl Relocatable for ElfObject<'_> {
+    fn origin(&self) -> &Path {
+        &self.origin
     }
+
+    fn sections(&self) -> BSI<'_> {
+        Box::new(SectionIter {
+            elf: &self.elf,
+            iter: self.elf.section_iter(),
+        })
     }
 }
 
-impl<'a> Relocatable<'a> {
-    pub fn from_elf(origin: &'a str, elf: &'a ElfFile) -> Result<Self, Error> {
-        is_relocatable(elf)?;
-        let mut result = Relocatable::new(origin);
+impl Display for ElfObject<'_> {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        write!(
+            f,
+            "==={:?}===\n>Symbols:\n{}\n>Sections:\n",
+            self.origin().file_name().unwrap(),
+            "TODO"
+        )?;
 
-        for header in elf.section_iter() {
-            // .text sections > 0 only
-            if is_text_section(elf, &header)? {
-                let data = header
-                    .get_data(&elf)
-                    .map_err(|e| Error::MissingSectionData(e))?;
+        for section in self.sections() {
+            let u = section.unwrap();
+            write!(f, "{}", u)?;
+        }
 
-                result.texts.push(Text::try_from(data)?);
+        Ok(())
     }
 }
 
-        Ok(result)
+impl LinkState<'_> {
+    pub fn new() -> Self {
+        LinkState {
+            relocatables: Vec::new(), // it's a trait thing
+        }
     }
 }
 
-fn is_text_section(elf: &ElfFile, header: &SectionHeader) -> Result<bool, Error> {
-    let name = header.get_name(&elf).unwrap_or("null");
-    let h_type = header
-        .get_type()
-        .map_err(|e| Error::MissingSectionHeader(e))?;
+impl<'data> Lazy<&'data str, ElfFile<'data>> for Symbol<'data> {
+    fn value(&self, src: &ElfFile<'data>) -> Result<&'data str, Error> {
+        src.get_string(self.index)
+            .map_err(|e| Error::StringError(e))
+    }
 
-    Ok(h_type == ShType::ProgBits
-        && header.flags() == (SHF_EXECINSTR | SHF_ALLOC)
-        && name.starts_with(".text")
-        && header.size() > 0)
+    fn resolved(&self) -> bool {
+        self.str_ref.is_some()
+    }
 }
 
 fn is_relocatable(elf: &ElfFile) -> Result<(), Error> {

src/formats/elf/section.rs

@@ -0,0 +1,110 @@
+use std::iter::Cloned;
+
+use crate::{
+    error::Error,
+    parts::{Section, SectionInfo, SectionIterResult},
+};
+use xmas_elf::{
+    sections::{SectionData, SectionHeader, SectionIter as ElfIter, ShType},
+    ElfFile,
+};
+
+impl<'data> Section<'data> {
+    pub fn iterate(
+        elf: &'data ElfFile<'data>,
+        iter: &mut ElfIter<'data, 'data>,
+    ) -> SectionIterResult<'data> {
+        if let Some(h) = iter.next() {
+            let h_result = h.get_type().map_err(|e| Error::MissingSectionHeader(e));
+            return match h_result {
+                Ok(h_type) => match h_type {
+                    // text, [ro]data
+                    ShType::ProgBits => {
+                        let s_name = h.get_name(&elf).unwrap_or("null");
+                        match section_data_to_iterator(elf, h) {
+                            Ok(i) => Section::from_name(s_name, &h, Some(Box::new(i))),
+                            Err(err) => SectionIterResult::Err(err),
+                        }
+                    }
+                    // bss
+                    ShType::NoBits => {
+                        let si = SectionInfo {
+                            size: h.size(),
+                            data: None,
+                            offset: h.offset(),
+                        };
+                        let s = Section::Bss(si);
+                        SectionIterResult::Ok(s)
+                    }
+                    _ => SectionIterResult::Skip,
+                },
+                Err(err) => SectionIterResult::Err(err),
+            };
+        }
+
+        SectionIterResult::None
+    }
+
+    fn from_name(
+        name: &str,
+        header: &SectionHeader,
+        data: Option<Box<dyn Iterator<Item = u8> + 'data>>,
+    ) -> SectionIterResult<'data> {
+        if name.starts_with(".text") {
+            if header.size() == 0 {
+                return SectionIterResult::Skip; // empty texts
+            }
+            let iter = match data {
+                Some(i) => i,
+                None => return SectionIterResult::Err(Error::MissingSectionData(".text")),
+            };
+            let si = SectionInfo {
+                size: header.size(),
+                data: Some(iter),
+                offset: header.offset(),
+            };
+            return SectionIterResult::Ok(Section::Text(si));
+        } else if name.starts_with(".rodata") || name.starts_with(".data") {
+            if header.size() == 0 {
+                return SectionIterResult::Skip; // empty data
+            }
+            let iter = match data {
+                Some(i) => i,
+                None => return SectionIterResult::Err(Error::MissingSectionData(".data")),
+            };
+            let si = SectionInfo {
+                size: header.size(),
+                data: Some(iter),
+                offset: header.offset(),
+            };
+            return SectionIterResult::Ok(Section::Data(si));
+        } else if name.starts_with(".bss") {
+            let si = SectionInfo {
+                size: header.size(),
+                data: None,
+                offset: header.offset(),
+            };
+            return SectionIterResult::Ok(Section::Bss(si));
+        }
+
+        SectionIterResult::Skip
+    }
+}
+
+fn section_data_to_iterator<'file>(
+    elf: &'file ElfFile,
+    header: SectionHeader<'file>,
+) -> Result<Cloned<std::slice::Iter<'file, u8>>, Error> {
+    let data = header
+        .get_data(&elf)
+        .map_err(|e| Error::MissingSectionData(e))?;
+
+    match data {
+        SectionData::Empty => Ok(([]).iter().cloned()),
+        SectionData::Undefined(bytes) => {
+            let iter = bytes.iter().cloned();
+            Ok(iter)
+        }
+        _ => Err(Error::InvalidSectionData), // TODO: more context needed
+    }
+}

src/formats/elf/section_iter.rs

@@ -0,0 +1,24 @@
+use crate::{
+    error::Error,
+    parts::{Section, SectionIterResult},
+};
+use xmas_elf::sections::SectionIter as ElfIter;
+use xmas_elf::ElfFile;
+
+pub struct SectionIter<'data> {
+    pub elf: &'data ElfFile<'data>,
+    pub iter: ElfIter<'data, 'data>,
+}
+
+impl<'data> Iterator for SectionIter<'data> {
+    type Item = Result<Section<'data>, Error>;
+
+    fn next(&mut self) -> Option<Self::Item> {
+        match Section::iterate(&self.elf, &mut self.iter) {
+            SectionIterResult::Err(err) => Some(Err(err)),
+            SectionIterResult::None => None,
+            SectionIterResult::Skip => self.next(),
+            SectionIterResult::Ok(r) => Some(Ok(r)),
+        }
+    }
+}

src/linker.rs

@@ -1,27 +1,40 @@
 use std::fmt::Display;
 
-use crate::error::Error;
-use crate::parts::Relocatable;
+use crate::{error::Error, parts::Relocatable};
 
-#[derive(Default)]
-pub struct LinkState<'a> {
-    relocatables: Vec<Relocatable<'a>>,
+pub struct LinkState<'data> {
+    pub relocatables: Vec<Box<dyn Relocatable + 'data>>,
 }
 
-impl<'a> LinkState<'a> {
-    pub fn add_relocatable(&mut self, relocatable: Relocatable<'a>) -> Result<(), Error> {
+impl<'data> LinkState<'data> {
+    pub fn add_relocatable(
+        &mut self,
+        relocatable: Box<dyn Relocatable + 'data>,
+    ) -> Result<(), Error> {
         self.relocatables.push(relocatable);
 
         Ok(())
     }
+
+    pub fn total_size(&self) -> Result<u64, Error> {
+        let mut result = 0u64;
+
+        for o in self.relocatables.iter() {
+            for s in o.sections() {
+                result += s?.size()?;
+            }
+        }
+
+        Ok(result)
+    }
 }
 
 impl Display for LinkState<'_> {
     fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
         writeln!(f, "===Relocatables===")?;
 
-        for r in self.relocatables.iter() {
-            writeln!(f, "{}", r)?;
+        for r in &self.relocatables {
+            write!(f, "{}", r)?;
         }
 
         Ok(())

src/main.rs

@@ -1,13 +1,11 @@
-use std::fs;
-use xmas_elf::ElfFile;
-
 mod error;
 mod formats;
 mod linker;
 mod parts;
 
+use formats::ElfObject;
 use linker::LinkState;
-use parts::Relocatable;
+use parts::Storage;
 
 fn main() {
     let mut args = std::env::args().peekable();
@@ -15,25 +13,22 @@ fn main() {
     // can this miss?
     args.next().unwrap();
 
-    let mut contents: Vec<(String, Vec<u8>)> = Vec::new();
-    let mut parsed: Vec<(&str, ElfFile)> = Vec::new();
-    let mut linker = LinkState::default();
+    let mut contents: Vec<Storage> = Vec::new();
+    let mut linker = LinkState::new();
 
     while args.peek().is_some() {
         let fpath = args.next().expect("Unexpected peek-a-boo");
-        let raw = fs::read(&fpath).expect("Error reading file");
-        contents.push((fpath, raw));
+        let stored = Storage::new(fpath.into());
+        contents.push(stored);
     }
 
-    for data in contents.iter() {
-        let elf = ElfFile::new(&data.1).expect("Error parsing ELF file");
-        parsed.push((&data.0, elf));
+    for storage in contents.iter_mut() {
+        let relocatable = ElfObject::new(storage).unwrap();
+        linker.add_relocatable(Box::new(relocatable)).unwrap();
     }
 
-    for elf in parsed.iter() {
-        let relocatable = Relocatable::from_elf(elf.0, &elf.1).unwrap();
-        linker.add_relocatable(relocatable).unwrap();
-    }
+    let total_size = linker.total_size().expect("Size to be known");
+    println!("to allocate: {}", total_size);
 
-    println!("{}", linker);
+    println!("all: \n{}", linker);
 }

src/parts.rs

@@ -1,9 +1,11 @@
 mod lazy;
 mod relocatable;
+mod section;
+mod storage;
 mod symbol;
-mod segments;
 
 pub use lazy::*;
 pub use relocatable::*;
+pub use section::*;
+pub use storage::*;
 pub use symbol::*;
-pub use segments::*;

src/parts/lazy.rs

@@ -1,5 +1,4 @@
 use crate::error::Error;
-
 pub trait Lazy<VALUE, SOURCE> {
     /// Resolve VALUE, blocking load from the SOURCE
     fn value(&self, src: &SOURCE) -> Result<VALUE, Error>;

src/parts/relocatable.rs

@@ -1,33 +1,15 @@
-use crate::parts::{Symbol, Text, Data, Bss};
+use std::fmt::Display;
+use std::path::Path;
 
-/// Contains all the needed references to construct a final
-/// mushed and relocated executable from an object file
-pub struct Relocatable<'a> {
-    pub origin: &'a str,
-    pub symbols: Vec<Symbol<'a>>,
-    pub texts: Vec<Text<'a>>,
-    pub data: Vec<Data<'a>>,
-    pub bss: Vec<Bss<'a>>,
-}
+use crate::error::Error;
 
-impl<'a> Relocatable<'a> {
-    pub fn new(origin: &'a str) -> Self {
-        Self {
-            origin,
-            symbols: Vec::new(),
-            texts: Vec::new(),
-            data: Vec::new(),
-            bss: Vec::new(),
-        }
-    }
-}
+use super::Section;
 
-impl std::fmt::Display for Relocatable<'_> {
-    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
-        write!(
-            f,
-            "==={}===\nSymbols:\n{:?}\nTexts:\n??",
-            self.origin, self.symbols //, self.texts
-        )
-    }
+pub type BSI<'data> = Box<dyn Iterator<Item = Result<Section<'data>, Error>> + 'data>;
+/// Contains all the needed getters to construct a final
+/// mushed and relocated executable from an object file
+pub trait Relocatable: Display {
+    fn origin(&self) -> &Path; // not same as section's path since this one's supposed to be cannonical
+
+    fn sections(&self) -> BSI<'_>;
 }

src/parts/section.rs

@@ -0,0 +1,55 @@
+use crate::error::Error;
+use std::fmt::{Display, Formatter};
+
+pub type SectionDataIterator<'data> = Box<dyn Iterator<Item = u8> + 'data>;
+
+pub struct SectionInfo<'data> {
+    pub size: u64,
+    pub data: Option<SectionDataIterator<'data>>,
+    pub offset: u64,
+}
+
+pub enum Section<'data> {
+    Text(SectionInfo<'data>),
+    Data(SectionInfo<'data>),
+    Bss(SectionInfo<'data>),
+}
+
+impl Section<'_> {
+    pub fn size(&self) -> Result<u64, Error> {
+        match self {
+            Section::Text(s) => Ok(s.size),
+            Section::Data(s) => Ok(s.size),
+            Section::Bss(s) => Ok(s.size),
+        }
+    }
+}
+
+impl Display for SectionInfo<'_> {
+    fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result {
+        write!(
+            f,
+            "size: {}, data: {}, offset: {}",
+            self.size,
+            self.data.is_some(),
+            self.offset,
+        )
+    }
+}
+
+impl Display for Section<'_> {
+    fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result {
+        match self {
+            Section::Text(s) => writeln!(f, "TEXT {}", s),
+            Section::Data(s) => writeln!(f, "DATA {}", s),
+            Section::Bss(s) => writeln!(f, "BSS {}", s),
+        }
+    }
+}
+
+pub enum SectionIterResult<'data> {
+    Ok(Section<'data>),
+    Err(Error),
+    Skip,
+    None,
+}

src/parts/segments.rs

@@ -1,13 +0,0 @@
-
-
-pub struct Text<'a> {
-    pub bytes: &'a [u8],
-}
-
-pub struct Data<'a> {
-    pub bytes: &'a [u8],
-}
-
-pub struct Bss<'a> {
-    pub bytes: &'a [u8],
-}

src/parts/storage.rs

@@ -0,0 +1,33 @@
+use std::path::PathBuf;
+
+use crate::error::Error;
+
+pub struct Storage {
+    origin: PathBuf,
+    bytes: Vec<u8>, // for storing the total if needed
+}
+
+impl Storage {
+    pub fn new(origin: PathBuf) -> Self {
+        Self {
+            origin,
+            bytes: Vec::new(),
+        }
+    }
+
+    pub fn origin(&self) -> Result<PathBuf, Error> {
+        std::fs::canonicalize(&self.origin).map_err(|e| e.into())
+    }
+
+    pub fn iter(&mut self) -> Result<impl Iterator<Item = &u8>, Error> {
+        self.bytes = std::fs::read(&self.origin)?;
+
+        Ok(self.bytes.iter())
+    }
+
+    pub fn bytes(&mut self) -> Result<&[u8], Error> {
+        self.bytes = std::fs::read(&self.origin)?;
+
+        Ok(&self.bytes)
+    }
+}