Add build script to extract a list of registered syscalls

The list is used in cargo-build-bpf to check generated .so modules for
undefined symbols that are not known run-time syscalls.

Cargo.lock

@@ -4314,6 +4314,7 @@ dependencies = [
  "num-traits",
  "rand 0.7.3",
  "rand_core 0.6.2",
+ "regex",
  "rustversion",
  "sha3",
  "solana-measure",

programs/bpf/Cargo.lock

@@ -2792,6 +2792,7 @@ dependencies = [
  "num-derive",
  "num-traits",
  "rand_core 0.6.2",
+ "regex",
  "sha3",
  "solana-measure",
  "solana-runtime",

programs/bpf_loader/Cargo.toml

@@ -9,6 +9,9 @@ homepage = "https://solana.com/"
 documentation = "https://docs.rs/solana-bpf-loader-program"
 edition = "2018"
 
+[build-dependencies]
+regex = "1.5.4"
+
 [dependencies]
 bincode = "1.3.1"
 byteorder = "1.3.4"

programs/bpf_loader/build.rs

@@ -0,0 +1,34 @@
+use regex::Regex;
+use std::{
+    fs::File,
+    io::{prelude::*, BufWriter, Read},
+    path::PathBuf,
+    process::exit,
+    str,
+};
+
+/**
+ * Extract a list of registered syscall names and save it in a file
+ * for distribution with the SDK.  This file is read by cargo-build-bpf
+ * to verify undefined symbols in a .so module that cargo-build-bpf has built.
+ */
+fn main() {
+    let path = PathBuf::from("src/syscalls.rs");
+    let mut file = match File::open(&path) {
+        Ok(x) => x,
+        _ => exit(1),
+    };
+    let mut text = vec![];
+    file.read_to_end(&mut text).unwrap();
+    let text = str::from_utf8(&text).unwrap();
+    let path = PathBuf::from("../../sdk/bpf/syscalls.txt");
+    let file = match File::create(&path) {
+        Ok(x) => x,
+        _ => exit(1),
+    };
+    let mut out = BufWriter::new(file);
+    let sysc_re = Regex::new(r#"register_syscall_by_name\([[:space:]]*b"([^"]+)","#).unwrap();
+    for caps in sysc_re.captures_iter(&text) {
+        writeln!(out, "{}", caps[1].to_string()).unwrap();
+    }
+}

sdk/bpf/.gitignore

@@ -7,3 +7,4 @@
 /dependencies/bin*
 /dependencies/.crates.toml
 /dependencies/.crates2.json
+/syscalls.txt

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

@@ -250,32 +250,17 @@ fn postprocess_dump(program_dump: &Path) {
 // Check whether the built .so file contains undefined symbols that are
 // not known to the runtime and warn about them if any.
 fn check_undefined_symbols(config: &Config, program: &Path) {
-    let syscalls: HashSet<String> = [
-        "abort",
-        "sol_panic_",
-        "sol_log_",
-        "sol_log_64_",
-        "sol_log_compute_units_",
-        "sol_log_pubkey",
-        "sol_create_program_address",
-        "sol_try_find_program_address",
-        "sol_sha256",
-        "sol_keccak256",
-        "sol_get_clock_sysvar",
-        "sol_get_epoch_schedule_sysvar",
-        "sol_get_fees_sysvar",
-        "sol_get_rent_sysvar",
-        "sol_memcpy_",
-        "sol_memmove_",
-        "sol_memcmp_",
-        "sol_memset_",
-        "sol_invoke_signed_c",
-        "sol_invoke_signed_rust",
-        "sol_alloc_free_",
-    ]
-    .iter()
-    .map(|&symbol| symbol.to_owned())
-    .collect();
+    let syscalls_txt = config.bpf_sdk.join("syscalls.txt");
+    let file = match File::open(syscalls_txt) {
+        Ok(x) => x,
+        _ => return,
+    };
+    let mut syscalls = HashSet::new();
+    for line_result in BufReader::new(file).lines() {
+        let line = line_result.unwrap();
+        let line = line.trim_end();
+        syscalls.insert(line.to_string());
+    }
     let entry =
         Regex::new(r"^ *[0-9]+: [0-9a-f]{16} +[0-9a-f]+ +NOTYPE +GLOBAL +DEFAULT +UND +(.+)")
             .unwrap();