Extra application function and plt data from platform

Cargo.lock

@@ -3323,6 +3323,7 @@ dependencies = [
  "clap 3.0.0-beta.1",
  "memmap2 0.3.1",
  "object 0.26.0",
+ "roc_collections",
 ]
 
 [[package]]

linker/Cargo.toml

@@ -18,6 +18,7 @@ test = false
 bench = false
 
 [dependencies]
+roc_collections = { path = "../compiler/collections" }
 bumpalo = { version = "3.6.1", features = ["collections"] }
 # TODO switch to clap 3.0.0 once it's out. Tried adding clap = "~3.0.0-beta.1" and cargo wouldn't accept it
 clap = { git = "https://github.com/rtfeldman/clap", branch = "master" }

linker/src/lib.rs

@@ -1,11 +1,17 @@
 use clap::{App, AppSettings, Arg, ArgMatches};
 use memmap2::Mmap;
-use object::Object;
+use object::{
+    Architecture, BinaryFormat, Object, ObjectSection, ObjectSymbol, Relocation, RelocationKind,
+    RelocationTarget, Symbol,
+};
+use roc_collections::all::MutMap;
 use std::fs;
 use std::io;
 
 pub const CMD_PREPROCESS: &str = "preprocess";
 pub const CMD_SURGERY: &str = "surgery";
+pub const FLAG_VERBOSE: &str = "verbose";
+
 pub const EXEC: &str = "EXEC";
 pub const SHARED_LIB: &str = "SHARED_LIB";
 
@@ -25,6 +31,13 @@ pub fn build_app<'a>() -> App<'a> {
                     Arg::with_name(SHARED_LIB)
                         .help("The dummy shared library representing the Roc application")
                         .required(true),
+                )
+                .arg(
+                    Arg::with_name(FLAG_VERBOSE)
+                        .long(FLAG_VERBOSE)
+                        .short('v')
+                        .help("enable verbose printing")
+                        .required(false),
                 ),
         )
         .subcommand(
@@ -33,22 +46,98 @@ pub fn build_app<'a>() -> App<'a> {
 }
 
 pub fn preprocess(matches: &ArgMatches) -> io::Result<i32> {
-    let _app_functions = application_functions(&matches.value_of(SHARED_LIB).unwrap())?;
+    let verbose = matches.is_present(FLAG_VERBOSE);
+
+    let app_functions = application_functions(&matches.value_of(SHARED_LIB).unwrap())?;
+    if verbose {
+        println!("Found app functions: {:?}", app_functions);
+    }
 
     let exec_file = fs::File::open(&matches.value_of(EXEC).unwrap())?;
     let exec_mmap = unsafe { Mmap::map(&exec_file)? };
-    let exec_obj = object::File::parse(&*exec_mmap).map_err(|err| {
+    let exec_obj = match object::File::parse(&*exec_mmap) {
-        io::Error::new(
+        Ok(obj) => obj,
-            io::ErrorKind::InvalidData,
+        Err(err) => {
-            format!("Failed to parse executable file: {}", err),
+            println!("Failed to parse executable file: {}", err);
-        )
+            return Ok(-1);
-    })?;
+        }
+    };
 
-    // TODO: Extract PLT related information for these functions.
+    // TODO: Deal with other file formats and architectures.
-    // The information need is really the address of each plt version of each application function.
+    let format = exec_obj.format();
-    // To find this, first get the dynmaic symbols for the app functions.
+    if format != BinaryFormat::Elf {
-    // Then reference them on the dynamic relocation table to figure out their plt function number.
+        println!("File Format, {:?}, not supported", format);
-    // Then with the plt base address and that function number(or scanning the code), it should be possible to find the address.
+        return Ok(-1);
+    }
+    let arch = exec_obj.architecture();
+    if arch != Architecture::X86_64 {
+        println!("Architecture, {:?}, not supported", arch);
+        return Ok(-1);
+    }
+
+    // Extract PLT related information for app functions.
+    let plt_address = match exec_obj.sections().find(|sec| sec.name() == Ok(".plt")) {
+        Some(section) => section.address(),
+        None => {
+            println!("Failed to find PLT section. Probably an malformed executable.");
+            return Ok(-1);
+        }
+    };
+    if verbose {
+        println!("PLT Address: {:x}", plt_address);
+    }
+
+    let plt_relocs: Vec<Relocation> = (match exec_obj.dynamic_relocations() {
+        Some(relocs) => relocs,
+        None => {
+            println!("Executable never calls any application functions.");
+            println!("No work to do. Probably an invalid input.");
+            return Ok(-1);
+        }
+    })
+    .map(|(_, reloc)| reloc)
+    .filter(|reloc| reloc.kind() == RelocationKind::Elf(7))
+    .collect();
+    if verbose {
+        println!();
+        println!("PLT relocations");
+        for reloc in plt_relocs.iter() {
+            println!("{:x?}", reloc);
+        }
+    }
+
+    let app_syms: Vec<Symbol> = exec_obj
+        .dynamic_symbols()
+        .filter(|sym| {
+            let name = sym.name();
+            name.is_ok() && app_functions.contains(&name.unwrap().to_string())
+        })
+        .collect();
+    if verbose {
+        println!();
+        println!("PLT Symbols for App Functions");
+        for symbol in app_syms.iter() {
+            println!("{}: {:x?}", symbol.index().0, symbol);
+        }
+    }
+
+    const PLT_ADDRESS_OFFSET: u64 = 0x10;
+
+    let mut app_func_addresses: MutMap<u64, &str> = MutMap::default();
+    for (i, reloc) in plt_relocs.into_iter().enumerate() {
+        for symbol in app_syms.iter() {
+            if reloc.target() == RelocationTarget::Symbol(symbol.index()) {
+                let func_address = (i as u64 + 1) * PLT_ADDRESS_OFFSET + plt_address;
+                app_func_addresses.insert(func_address, symbol.name().unwrap());
+                break;
+            }
+        }
+    }
+
+    if verbose {
+        println!();
+        println!("App Function Address Map: {:x?}", app_func_addresses);
+    }
 
     // TODO: For all text sections check for function calls to app functions.
     // This should just be disassembly and then scanning for jmp and call style ops that jump to the plt offsets we care about.
@@ -57,7 +146,7 @@ pub fn preprocess(matches: &ArgMatches) -> io::Result<i32> {
 
     // TODO: Store all this data in a nice format.
 
-    // TODO: Potentially create a version of the executable with certain dynamic and PLT information deleted.
+    // TODO: Potentially create a version of the executable with certain dynamic and PLT information deleted (changing offset may break stuff so be careful).
     // Remove shared library dependencies.
     // Delete extra plt entries, dynamic symbols, and dynamic relocations (might require updating other plt entries, may not worth it).
     // Add regular symbols pointing to 0 for the app functions (maybe not needed if it is just link metadata).