Merge pull request #138 from valida-xyz/morgan/issue/135

Loading ELF files

Cargo.toml

@@ -7,6 +7,7 @@ members = [
     "bus",
     "cpu",
     "derive",
+    "elf",
     "native_field",
     "machine",
     "memory",

basic/Cargo.toml

@@ -26,6 +26,7 @@ valida-assembler = { path = "../assembler" }
 valida-bus = { path = "../bus" }
 valida-cpu = { path = "../cpu" }
 valida-derive = { path = "../derive" }
+valida-elf = { path = "../elf" }
 valida-machine = { path = "../machine" }
 valida-memory = { path = "../memory" }
 valida-opcodes = { path = "../opcodes" }

basic/src/bin/valida.rs

@@ -1,4 +1,5 @@
 use clap::Parser;
+use std::fs;
 use std::fs::File;
 use std::io::{stdout, Write};
 
@@ -11,7 +12,9 @@ use valida_cpu::MachineWithCpuChip;
 use valida_machine::{Machine, MachineProof, ProgramROM, StdinAdviceProvider};
 use valida_memory::MachineWithMemoryChip;
 
+use valida_elf::{load_executable_file, Program};
 use valida_program::MachineWithProgramChip;
+use valida_static_data::MachineWithStaticDataChip;
 
 use p3_challenger::DuplexChallenger;
 use p3_dft::Radix2DitParallel;
@@ -283,16 +286,17 @@ fn main() {
     }
 
     let mut machine = BasicMachine::<BabyBear>::default();
-    let rom = match ProgramROM::from_file(&args.program) {
-        Ok(contents) => contents,
-        Err(e) => panic!("Failure to load file: {}. {}", &args.program, e),
-    };
-    machine.program_mut().set_program_rom(&rom);
+    let Program { code, data } = load_executable_file(
+        fs::read(&args.program)
+            .expect(format!("Failed to read executable file: {}", &args.program).as_str()),
+    );
+    machine.program_mut().set_program_rom(&code);
     machine.cpu_mut().fp = args.stack_height;
     machine.cpu_mut().save_register_state();
+    machine.static_data_mut().load(data);
 
     // Run the program
-    machine.run(&rom, &mut StdinAdviceProvider);
+    machine.run(&code, &mut StdinAdviceProvider);
 
     type Val = BabyBear;
     type Challenge = BinomialExtensionField<Val, 5>;

elf/Cargo.toml

@@ -0,0 +1,9 @@
+[package]
+name = "valida-elf"
+version = "0.1.0"
+edition = "2021"
+license = "MIT OR Apache-2.0"
+
+[dependencies]
+elf = "0.7.4"
+valida-machine = { path = "../machine" }

elf/src/lib.rs

@@ -0,0 +1,90 @@
+#![no_std]
+
+extern crate alloc;
+
+use alloc::collections::BTreeMap;
+use alloc::vec;
+use alloc::vec::Vec;
+use elf::abi;
+use elf::endian::AnyEndian;
+use elf::section::SectionHeader;
+use elf::ElfBytes;
+use valida_machine::{ProgramROM, Word};
+
+pub struct Program {
+    pub code: ProgramROM<i32>,
+    pub data: BTreeMap<u32, Word<u8>>,
+}
+
+pub fn load_executable_file(file: Vec<u8>) -> Program {
+    if file[0] == 0x7F && file[1] == 0x45 && file[2] == 0x4C && file[3] == 0x46 {
+        load_elf_object_file(file)
+    } else {
+        Program {
+            code: ProgramROM::from_machine_code(file.as_slice()),
+            data: BTreeMap::new(),
+        }
+    }
+}
+
+pub fn load_elf_object_file(file: Vec<u8>) -> Program {
+    let file = ElfBytes::<AnyEndian>::minimal_parse(file.as_slice()).unwrap();
+    let mut data_sections: Vec<(SectionHeader, &[u8])> = vec![];
+    let mut bss_sections: Vec<SectionHeader> = vec![];
+    let mut text_sections: Vec<(SectionHeader, &[u8])> = vec![];
+    for section_header in file.section_headers().unwrap().iter() {
+        let is_data: bool = section_header.sh_type == abi::SHT_PROGBITS
+            && section_header.sh_flags == (abi::SHF_ALLOC | abi::SHF_WRITE).into();
+        let is_bss: bool = section_header.sh_type == abi::SHT_NOBITS
+            && section_header.sh_flags == (abi::SHF_ALLOC | abi::SHF_WRITE).into();
+        let is_text: bool = section_header.sh_type == abi::SHT_PROGBITS
+            && section_header.sh_flags == (abi::SHF_ALLOC | abi::SHF_EXECINSTR).into();
+        let is_useful: bool = is_data || is_bss || is_text;
+        if is_useful {
+            if is_data || is_text {
+                let section_data = file.section_data(&section_header).unwrap();
+                match section_data {
+                    (section_data, None) => {
+                        if is_data {
+                            data_sections.push((section_header, section_data));
+                        } else if is_text {
+                            text_sections.push((section_header, section_data));
+                        }
+                    }
+                    _ => panic!("unsupported: compressed ELF section data"),
+                }
+            } else if is_bss {
+                bss_sections.push(section_header);
+            }
+        }
+    }
+    let code_size = text_sections
+        .iter()
+        .map(|(section_header, _)| section_header.sh_addr + section_header.sh_size)
+        .fold(0, |a, b| a.max(b));
+    let mut code: Vec<u8> = vec![0; code_size as usize];
+    for (section_header, section_data) in text_sections {
+        for i in 0..section_header.sh_size as usize {
+            code[i + section_header.sh_addr as usize] = section_data[i];
+        }
+    }
+    let mut data: BTreeMap<u32, Word<u8>> = BTreeMap::new();
+    for (section_header, section_data) in data_sections {
+        for i in 0..(section_header.sh_size / 4) as usize {
+            data.insert(
+                <u64 as TryInto<u32>>::try_into(section_header.sh_addr).unwrap()
+                    + <usize as TryInto<u32>>::try_into(i * 4).unwrap(),
+                Word([
+                    section_data[i * 4],
+                    section_data[i * 4 + 1],
+                    section_data[i * 4 + 2],
+                    section_data[i * 4 + 3],
+                ]),
+            );
+        }
+    }
+    Program {
+        code: ProgramROM::from_machine_code(code.as_slice()),
+        data: data,
+    }
+}

static_data/src/lib.rs

@@ -39,6 +39,10 @@ impl StaticDataChip {
         }
     }
 
+    pub fn load(&mut self, cells: BTreeMap<u32, Word<u8>>) {
+        self.cells = cells;
+    }
+
     pub fn write(&mut self, address: u32, value: Word<u8>) {
         self.cells.insert(address, value);
     }