From 03b596586c759b1e0d7afdf461b8938f5c33a59a Mon Sep 17 00:00:00 2001
From: HaoranYi <haoran.yi@gmail.com>
Date: Tue, 16 Apr 2024 16:04:52 -0500
Subject: [PATCH] Write default hash to account storage (#469)

* write default hash to apendvec

* fix get_pubkey_hash_for_slot

* fix tests

* fix rebase error

---------

Co-authored-by: HaoranYi <haoran.yi@solana.com>
---
 accounts-db/src/accounts_db.rs |  8 +++++---
 accounts-db/src/append_vec.rs  |  7 ++++---
 runtime/tests/accounts.rs      | 28 ++++++++++++++++++++++++----
 3 files changed, 33 insertions(+), 10 deletions(-)

diff --git a/accounts-db/src/accounts_db.rs b/accounts-db/src/accounts_db.rs
index b71603b4d8..b7a674746a 100644
--- a/accounts-db/src/accounts_db.rs
+++ b/accounts-db/src/accounts_db.rs
@@ -10215,14 +10215,16 @@ pub mod tests {
     );
 
     #[test]
-    #[should_panic(expected = "MismatchedAccountsHash")]
     fn test_accountsdb_scan_snapshot_stores_check_hash() {
         solana_logger::setup();
         let accounts_db = AccountsDb::new_single_for_tests();
         let (storages, _raw_expected) = sample_storages_and_accounts(&accounts_db);
         let max_slot = storages.iter().map(|storage| storage.slot()).max().unwrap();
 
-        let hash =
+        // bogus_hash is ignored during appendvec store. When we verify hashes
+        // later during scan, the test will recompute the "correct" hash and
+        // should not mismatch.
+        let bogus_hash =
             AccountHash(Hash::from_str("7JcmM6TFZMkcDkZe6RKVkGaWwN5dXciGC4fa3RxvqQc9").unwrap());
 
         // replace the sample storages, storing bogus hash values so that we trigger the hash mismatch
@@ -10242,7 +10244,7 @@ pub mod tests {
                     .collect::<Vec<_>>();
                 let slice = &accounts[..];
                 let account_data = (slot, slice);
-                let hashes = (0..account_data.len()).map(|_| &hash).collect();
+                let hashes = (0..account_data.len()).map(|_| &bogus_hash).collect();
                 let storable_accounts =
                     StorableAccountsWithHashes::new_with_hashes(&account_data, hashes);
                 copied_storage
diff --git a/accounts-db/src/append_vec.rs b/accounts-db/src/append_vec.rs
index c7feb0eaee..b963dee3df 100644
--- a/accounts-db/src/append_vec.rs
+++ b/accounts-db/src/append_vec.rs
@@ -20,6 +20,7 @@ use {
     solana_sdk::{
         account::{AccountSharedData, ReadableAccount, WritableAccount},
         clock::Slot,
+        hash::Hash,
         pubkey::Pubkey,
         stake_history::Epoch,
     },
@@ -752,8 +753,8 @@ impl AppendVec {
         skip: usize,
     ) -> Option<Vec<StoredAccountInfo>> {
         let _lock = self.append_lock.lock().unwrap();
+        let default_hash: Hash = Hash::default(); // [0_u8; 32];
         let mut offset = self.len();
-
         let len = accounts.accounts.len();
         // Here we have `len - skip` number of accounts.  The +1 extra capacity
         // is for storing the aligned offset of the last entry to that is used
@@ -765,7 +766,7 @@ impl AppendVec {
             if stop {
                 break;
             }
-            accounts.get(i, |account, hash| {
+            accounts.get(i, |account, _hash| {
                 let account_meta = AccountMeta {
                     lamports: account.lamports(),
                     owner: *account.owner(),
@@ -782,7 +783,7 @@ impl AppendVec {
                 let account_meta_ptr = &account_meta as *const AccountMeta;
                 let data_len = stored_meta.data_len as usize;
                 let data_ptr = account.data().as_ptr();
-                let hash_ptr = bytemuck::bytes_of(hash).as_ptr();
+                let hash_ptr = bytemuck::bytes_of(&default_hash).as_ptr();
                 let ptrs = [
                     (meta_ptr as *const u8, mem::size_of::<StoredMeta>()),
                     (account_meta_ptr as *const u8, mem::size_of::<AccountMeta>()),
diff --git a/runtime/tests/accounts.rs b/runtime/tests/accounts.rs
index 549839be5f..bcb6f7746a 100644
--- a/runtime/tests/accounts.rs
+++ b/runtime/tests/accounts.rs
@@ -4,11 +4,13 @@ use {
     rayon::prelude::*,
     solana_accounts_db::{
         accounts_db::{AccountsDb, LoadHint},
+        accounts_hash::AccountHash,
         ancestors::Ancestors,
     },
     solana_sdk::{
         account::{AccountSharedData, ReadableAccount, WritableAccount},
         clock::Slot,
+        hash::Hash,
         pubkey::Pubkey,
         sysvar::epoch_schedule::EpochSchedule,
     },
@@ -128,11 +130,29 @@ fn test_bad_bank_hash() {
         db.store_cached((some_slot, &account_refs[..]), None);
         for pass in 0..2 {
             for (key, account) in &account_refs {
-                assert_eq!(
-                    db.load_account_hash(&ancestors, key, Some(some_slot), LoadHint::Unspecified)
+                if pass == 1 {
+                    assert_eq!(
+                        db.load_account_hash(
+                            &ancestors,
+                            key,
+                            Some(some_slot),
+                            LoadHint::Unspecified
+                        )
                         .unwrap(),
-                    AccountsDb::hash_account(*account, key)
-                );
+                        AccountHash(Hash::default())
+                    );
+                } else {
+                    assert_eq!(
+                        db.load_account_hash(
+                            &ancestors,
+                            key,
+                            Some(some_slot),
+                            LoadHint::Unspecified
+                        )
+                        .unwrap(),
+                        AccountsDb::hash_account(*account, key)
+                    );
+                }
             }
             if pass == 0 {
                 // flush the write cache so we're reading from append vecs on the next iteration