Augment BPF binary dumps to resolve call instruction targets

sdk/cargo-build-bpf/Cargo.toml

@@ -12,6 +12,7 @@ publish = false
 [dependencies]
 bzip2 = "0.3.3"
 clap = "2.33.3"
+regex = "1.4.5"
 cargo_metadata = "0.12.0"
 solana-sdk = { path = "..", version = "=1.7.0" }
 solana-download-utils = { path = "../../download-utils", version = "=1.7.0" }

sdk/cargo-build-bpf/src/main.rs

@@ -3,16 +3,19 @@ use {
     clap::{
         crate_description, crate_name, crate_version, value_t, value_t_or_exit, values_t, App, Arg,
     },
+    regex::Regex,
     solana_download_utils::download_file,
     solana_sdk::signature::{write_keypair_file, Keypair},
     std::{
+        collections::HashMap,
         env,
         ffi::OsStr,
         fs::{self, File},
-        io::BufReader,
+        io::{prelude::*, BufReader, BufWriter},
         path::{Path, PathBuf},
         process::exit,
         process::{Command, Stdio},
+        str::FromStr,
     },
     tar::Archive,
 };
@@ -150,6 +153,103 @@ fn install_if_missing(
     Ok(())
 }
 
+// Process dump file attributing call instructions with callee function names
+fn postprocess_dump(program_dump: &Path) {
+    if !program_dump.exists() {
+        return;
+    }
+    let postprocessed_dump = program_dump.with_extension("postprocessed");
+    let head_re = Regex::new(r"(^[0-9a-f]{16}) (.+)").unwrap();
+    let insn_re = Regex::new(r"^ +([0-9]+)((\s[0-9a-f]{2})+)\s.+").unwrap();
+    let call_re = Regex::new(r"^ +([0-9]+)(\s[0-9a-f]{2})+\scall (-?)0x([0-9a-f]+)").unwrap();
+    let relo_re = Regex::new(r"^([0-9a-f]{16})  [0-9a-f]{16} R_BPF_64_32 +0{16} (.+)").unwrap();
+    let mut a2n: HashMap<i64, String> = HashMap::new();
+    let mut rel: HashMap<u64, String> = HashMap::new();
+    let mut name = String::from("");
+    let mut state = 0;
+    let file = match File::open(program_dump) {
+        Ok(x) => x,
+        _ => return,
+    };
+    for line_result in BufReader::new(file).lines() {
+        let line = line_result.unwrap();
+        let line = line.trim_end();
+        if line == "Disassembly of section .text" {
+            state = 1;
+        }
+        if state == 0 {
+            if relo_re.is_match(line) {
+                let captures = relo_re.captures(line).unwrap();
+                let address = u64::from_str_radix(&captures[1], 16).unwrap();
+                let symbol = captures[2].to_string();
+                rel.insert(address, symbol);
+            }
+        } else if state == 1 {
+            if head_re.is_match(line) {
+                state = 2;
+                let captures = head_re.captures(line).unwrap();
+                name = captures[2].to_string();
+            }
+        } else if state == 2 {
+            state = 1;
+            if insn_re.is_match(line) {
+                let captures = insn_re.captures(line).unwrap();
+                let address = i64::from_str(&captures[1]).unwrap();
+                a2n.insert(address, name.clone());
+            }
+        }
+    }
+    let file = match File::create(&postprocessed_dump) {
+        Ok(x) => x,
+        _ => return,
+    };
+    let mut out = BufWriter::new(file);
+    let file = match File::open(program_dump) {
+        Ok(x) => x,
+        _ => return,
+    };
+    let mut pc = 0u64;
+    let mut step = 0u64;
+    for line_result in BufReader::new(file).lines() {
+        let line = line_result.unwrap();
+        let line = line.trim_end();
+        if head_re.is_match(line) {
+            let captures = head_re.captures(line).unwrap();
+            pc = u64::from_str_radix(&captures[1], 16).unwrap();
+            writeln!(out, "{}", line).unwrap();
+            continue;
+        }
+        if insn_re.is_match(line) {
+            let captures = insn_re.captures(line).unwrap();
+            step = if captures[2].len() > 24 { 16 } else { 8 };
+        }
+        if call_re.is_match(line) {
+            if rel.contains_key(&pc) {
+                writeln!(out, "{} ; {}", line, rel[&pc]).unwrap();
+            } else {
+                let captures = call_re.captures(line).unwrap();
+                let pc = i64::from_str(&captures[1]).unwrap().checked_add(1).unwrap();
+                let offset = i64::from_str_radix(&captures[4], 16).unwrap();
+                let offset = if &captures[3] == "-" {
+                    offset.checked_neg().unwrap()
+                } else {
+                    offset
+                };
+                let address = pc.checked_add(offset).unwrap();
+                if a2n.contains_key(&address) {
+                    writeln!(out, "{} ; {}", line, a2n[&address]).unwrap();
+                } else {
+                    writeln!(out, "{}", line).unwrap();
+                }
+            }
+        } else {
+            writeln!(out, "{}", line).unwrap();
+        }
+        pc = pc.checked_add(step).unwrap();
+    }
+    fs::rename(postprocessed_dump, program_dump).unwrap();
+}
+
 // check whether custom BPF toolchain is linked, and link it if it is not.
 fn link_bpf_toolchain(config: &Config) {
     let toolchain_path = config
@@ -366,6 +466,7 @@ fn build_bpf_package(config: &Config, target_directory: &Path, package: &cargo_m
                 &config.bpf_sdk.join("scripts").join("dump.sh"),
                 &[&program_unstripped_so, &program_dump],
             );
+            postprocess_dump(&program_dump);
         }
 
         println!();