From 287e26032e3f1bd3aa1dec13b8c410b5ca12c478 Mon Sep 17 00:00:00 2001
From: ananas-block <jorrit@lightprotocol.com>
Date: Sun, 15 Jun 2025 16:34:14 +0100
Subject: [PATCH] perf: optimize bloom filter hashing

---
 Cargo.lock                           | 18 +++++++++++-------
 program-libs/bloom-filter/Cargo.toml |  2 +-
 program-libs/bloom-filter/src/lib.rs | 23 ++++++++++++++---------
 program-libs/hasher/Cargo.toml       |  1 +
 program-libs/hasher/src/keccak.rs    | 26 +-------------------------
 5 files changed, 28 insertions(+), 42 deletions(-)

diff --git a/Cargo.lock b/Cargo.lock
index c88c5a087c..bfcd778e82 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -2030,12 +2030,6 @@ dependencies = [
  "wide",
 ]
 
-[[package]]
-name = "fastmurmur3"
-version = "0.2.0"
-source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "2d7e9bc68be4cdabbb8938140b01a8b5bc1191937f2c7e7ecc2fcebbe2d749df"
-
 [[package]]
 name = "fastrand"
 version = "2.3.0"
@@ -3270,11 +3264,11 @@ name = "light-bloom-filter"
 version = "0.2.0"
 dependencies = [
  "bitvec",
- "fastmurmur3",
  "light-hasher",
  "num-bigint 0.4.6",
  "pinocchio",
  "rand 0.8.5",
+ "solana-nostd-keccak",
  "solana-program-error",
  "thiserror 2.0.12",
 ]
@@ -3428,6 +3422,7 @@ dependencies = [
  "rand 0.8.5",
  "sha2 0.10.9",
  "sha3",
+ "solana-nostd-keccak",
  "solana-program-error",
  "solana-pubkey",
  "thiserror 2.0.12",
@@ -6893,6 +6888,15 @@ dependencies = [
  "solana-sdk-ids",
 ]
 
+[[package]]
+name = "solana-nostd-keccak"
+version = "0.1.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "d8ced70920435b1baa58f76e6f84bbc1110ddd1d6161ec76b6d731ae8431e9c4"
+dependencies = [
+ "sha3",
+]
+
 [[package]]
 name = "solana-offchain-message"
 version = "2.2.1"
diff --git a/program-libs/bloom-filter/Cargo.toml b/program-libs/bloom-filter/Cargo.toml
index 5ccf6a9214..e7314b666c 100644
--- a/program-libs/bloom-filter/Cargo.toml
+++ b/program-libs/bloom-filter/Cargo.toml
@@ -12,7 +12,7 @@ pinocchio = ["dep:pinocchio"]
 
 [dependencies]
 bitvec = "1.0.1"
-fastmurmur3 = "0.2.0"
+solana-nostd-keccak = "0.1.3"
 num-bigint = { workspace = true }
 solana-program-error = { workspace = true, optional = true }
 pinocchio = { workspace = true, optional = true }
diff --git a/program-libs/bloom-filter/src/lib.rs b/program-libs/bloom-filter/src/lib.rs
index f3bd8384d0..d5375bd82e 100644
--- a/program-libs/bloom-filter/src/lib.rs
+++ b/program-libs/bloom-filter/src/lib.rs
@@ -68,15 +68,20 @@ impl<'a> BloomFilter<'a> {
         })
     }
 
-    pub fn probe_index_fast_murmur(value_bytes: &[u8], iteration: usize, capacity: &u64) -> usize {
-        let iter_bytes = iteration.to_le_bytes();
-        let base_hash = fastmurmur3::hash(value_bytes);
-        let mut combined_bytes = [0u8; 24];
-        combined_bytes[..16].copy_from_slice(&base_hash.to_le_bytes());
-        combined_bytes[16..].copy_from_slice(&iter_bytes);
+    pub fn probe_index_keccak(value_bytes: &[u8; 32], iteration: usize, capacity: &u64) -> usize {
+        let iter_bytes: [u8; 8] = iteration.to_le_bytes();
+        let mut combined_bytes = [0u8; 40];
+        combined_bytes[..32].copy_from_slice(value_bytes);
+        combined_bytes[32..].copy_from_slice(&iter_bytes);
 
-        let combined_hash = fastmurmur3::hash(&combined_bytes);
-        (combined_hash % (*capacity as u128)) as usize
+        let hash = solana_nostd_keccak::hash(&combined_bytes);
+
+        let mut index = 0u64;
+        for chunk in hash.chunks(8) {
+            let value = u64::from_le_bytes(chunk.try_into().unwrap());
+            index = value.wrapping_add(index) % *capacity;
+        }
+        index as usize
     }
 
     pub fn insert(&mut self, value: &[u8; 32]) -> Result<(), BloomFilterError> {
@@ -98,7 +103,7 @@ impl<'a> BloomFilter<'a> {
 
         let bits = BitSlice::<u8, Msb0>::from_slice_mut(self.store);
         for i in 0..self.num_iters {
-            let probe_index = Self::probe_index_fast_murmur(value, i, &(self.capacity));
+            let probe_index = Self::probe_index_keccak(value, i, &(self.capacity));
             if bits[probe_index] {
                 continue;
             } else if insert {
diff --git a/program-libs/hasher/Cargo.toml b/program-libs/hasher/Cargo.toml
index 5f1c4cbb4b..4798f967bd 100644
--- a/program-libs/hasher/Cargo.toml
+++ b/program-libs/hasher/Cargo.toml
@@ -23,6 +23,7 @@ solana-program-error = { workspace = true, optional = true }
 solana-pubkey = { workspace = true, optional = true }
 pinocchio = { workspace = true, optional = true }
 borsh = { workspace = true }
+solana-nostd-keccak = "0.1.3"
 
 [target.'cfg(not(target_os = "solana"))'.dependencies]
 ark-bn254 = { workspace = true }
diff --git a/program-libs/hasher/src/keccak.rs b/program-libs/hasher/src/keccak.rs
index 1f16278bfc..81d81d810c 100644
--- a/program-libs/hasher/src/keccak.rs
+++ b/program-libs/hasher/src/keccak.rs
@@ -14,31 +14,7 @@ impl Hasher for Keccak {
     }
 
     fn hashv(vals: &[&[u8]]) -> Result<Hash, HasherError> {
-        #[cfg(not(target_os = "solana"))]
-        {
-            use sha3::{Digest, Keccak256};
-
-            let mut hasher = Keccak256::default();
-            for val in vals {
-                hasher.update(val);
-            }
-            Ok(hasher.finalize().into())
-        }
-        // Call via a system call to perform the calculation
-        #[cfg(target_os = "solana")]
-        {
-            use crate::HASH_BYTES;
-
-            let mut hash_result = [0; HASH_BYTES];
-            unsafe {
-                crate::syscalls::sol_keccak256(
-                    vals as *const _ as *const u8,
-                    vals.len() as u64,
-                    &mut hash_result as *mut _ as *mut u8,
-                );
-            }
-            Ok(hash_result)
-        }
+        Ok(solana_nostd_keccak::hashv(vals))
     }
 
     fn zero_bytes() -> ZeroBytes {