Avoid shell out tar for genesis archive creation (solana-labs#3079)

Use the tar crate instead; doing so avoids spawning a subprocess and
also removes the variability of relying on whatever tar the caller has
in their path

ledger/Cargo.toml

@@ -14,6 +14,7 @@ assert_matches = { workspace = true }
 bincode = { workspace = true }
 bitflags = { workspace = true, features = ["serde"] }
 byteorder = { workspace = true }
+bzip2 = { workspace = true }
 chrono = { workspace = true, features = ["default", "serde"] }
 chrono-humanize = { workspace = true }
 crossbeam-channel = { workspace = true }
@@ -73,6 +74,7 @@ spl-token-2022 = { workspace = true, features = ["no-entrypoint"] }
 static_assertions = { workspace = true }
 strum = { workspace = true, features = ["derive"] }
 strum_macros = { workspace = true }
+tar = { workspace = true }
 tempfile = { workspace = true }
 thiserror = { workspace = true }
 tokio = { workspace = true, features = ["full"] }

ledger/src/blockstore.rs

@@ -71,7 +71,7 @@ use {
         },
         convert::TryInto,
         fmt::Write,
-        fs,
+        fs::{self, File},
         io::{Error as IoError, ErrorKind},
         ops::Bound,
         path::{Path, PathBuf},
@@ -81,6 +81,7 @@ use {
             Arc, Mutex, RwLock,
         },
     },
+    tar,
     tempfile::{Builder, TempDir},
     thiserror::Error,
     trees::{Tree, TreeWalk},
@@ -4835,32 +4836,12 @@ pub fn create_new_ledger(
     drop(blockstore);
 
     let archive_path = ledger_path.join(DEFAULT_GENESIS_ARCHIVE);
-    let args = vec![
-        "jcfhS",
-        archive_path.to_str().unwrap(),
-        "-C",
-        ledger_path.to_str().unwrap(),
-        DEFAULT_GENESIS_FILE,
-        blockstore_dir,
-    ];
-    let output = std::process::Command::new("tar")
-        .env("COPYFILE_DISABLE", "1")
-        .args(args)
-        .output()
-        .unwrap();
-    if !output.status.success() {
-        use std::str::from_utf8;
-        error!("tar stdout: {}", from_utf8(&output.stdout).unwrap_or("?"));
-        error!("tar stderr: {}", from_utf8(&output.stderr).unwrap_or("?"));
-
-        return Err(BlockstoreError::Io(IoError::new(
-            ErrorKind::Other,
-            format!(
-                "Error trying to generate snapshot archive: {}",
-                output.status
-            ),
-        )));
-    }
+    let archive_file = File::create(&archive_path)?;
+    let encoder = bzip2::write::BzEncoder::new(archive_file, bzip2::Compression::best());
+    let mut archive = tar::Builder::new(encoder);
+    archive.append_path_with_name(ledger_path.join(DEFAULT_GENESIS_FILE), DEFAULT_GENESIS_FILE)?;
+    archive.append_dir_all(blockstore_dir, ledger_path.join(blockstore_dir))?;
+    archive.into_inner()?;
 
     // ensure the genesis archive can be unpacked and it is under
     // max_genesis_archive_unpacked_size, immediately after creating it above.
@@ -5352,6 +5333,9 @@ pub mod tests {
         crossbeam_channel::unbounded,
         rand::{seq::SliceRandom, thread_rng},
         solana_account_decoder::parse_token::UiTokenAmount,
+        solana_accounts_db::hardened_unpack::{
+            open_genesis_config, MAX_GENESIS_ARCHIVE_UNPACKED_SIZE,
+        },
         solana_entry::entry::{next_entry, next_entry_mut},
         solana_runtime::bank::{Bank, RewardType},
         solana_sdk::{
@@ -5422,6 +5406,17 @@ pub mod tests {
         assert!(Path::new(ledger_path.path())
             .join(BLOCKSTORE_DIRECTORY_ROCKS_LEVEL)
             .exists());
+
+        assert_eq!(
+            genesis_config,
+            open_genesis_config(ledger_path.path(), MAX_GENESIS_ARCHIVE_UNPACKED_SIZE).unwrap()
+        );
+        // Remove DEFAULT_GENESIS_FILE to force extraction of DEFAULT_GENESIS_ARCHIVE
+        std::fs::remove_file(ledger_path.path().join(DEFAULT_GENESIS_FILE)).unwrap();
+        assert_eq!(
+            genesis_config,
+            open_genesis_config(ledger_path.path(), MAX_GENESIS_ARCHIVE_UNPACKED_SIZE).unwrap()
+        );
     }
 
     #[test]