From 6859907df9f4d67fe86f870fdce5549d4f945589 Mon Sep 17 00:00:00 2001
From: Rob Walker <rob@solana.com>
Date: Sun, 14 Apr 2019 21:10:09 -0700
Subject: [PATCH] more rigorous erasure constants, comments (#3766)

* more rigorous erasure constants, comments

* new header size means new golden
---
 core/src/chacha.rs  |  2 +-
 core/src/erasure.rs | 15 ++++++++++-----
 core/src/packet.rs  |  6 +++---
 3 files changed, 14 insertions(+), 9 deletions(-)

diff --git a/core/src/chacha.rs b/core/src/chacha.rs
index 8ea5c8bf8d..09eb03d78c 100644
--- a/core/src/chacha.rs
+++ b/core/src/chacha.rs
@@ -164,7 +164,7 @@ mod tests {
         use bs58;
         //  golden needs to be updated if blob stuff changes....
         let golden = Hash::new(
-            &bs58::decode("P5ZZcpRopdqU4ZVZz1Kmck5zykiNSxc9T3iPZzF3rMc")
+            &bs58::decode("5NBn4cBZmNZRftkjxj3um8W1eyYPzn2RgUJSA3SVbHaJ")
                 .into_vec()
                 .unwrap(),
         );
diff --git a/core/src/erasure.rs b/core/src/erasure.rs
index e2150f946e..b7fdffc2b1 100644
--- a/core/src/erasure.rs
+++ b/core/src/erasure.rs
@@ -9,8 +9,6 @@ pub const NUM_DATA: usize = 16; // number of data blobs
 pub const NUM_CODING: usize = 4; // number of coding blobs, also the maximum number that can go missing
 pub const ERASURE_SET_SIZE: usize = NUM_DATA + NUM_CODING; // total number of blobs in an erasure set, includes data and coding blobs
 
-pub const JERASURE_ALIGN: usize = 4; // data size has to be a multiple of 4 bytes
-
 macro_rules! align {
     ($x:expr, $align:expr) => {
         $x + ($align - 1) & !($align - 1)
@@ -59,6 +57,7 @@ extern "C" {
 use std::sync::Once;
 static ERASURE_W_ONCE: Once = Once::new();
 
+// jerasure word size of 32
 fn w() -> i32 {
     let w = 32;
     unsafe {
@@ -70,6 +69,11 @@ fn w() -> i32 {
     w
 }
 
+// jerasure checks that arrays are a multiple of w()/8 in length
+fn wb() -> usize {
+    (w() / 8) as usize
+}
+
 fn get_matrix(m: i32, k: i32, w: i32) -> Vec<i32> {
     let mut matrix = vec![0; (m * k) as usize];
     for i in 0..m {
@@ -243,6 +247,7 @@ impl CodingGenerator {
     }
 
     // must be called with consecutive data blobs from previous invocation
+    // blobs from a new slot not start halfway through next_data
     pub fn next(&mut self, next_data: &[SharedBlob]) -> Vec<SharedBlob> {
         let mut next_coding =
             Vec::with_capacity((self.leftover.len() + next_data.len()) / NUM_DATA * NUM_CODING);
@@ -261,12 +266,12 @@ impl CodingGenerator {
             }
             self.leftover.clear();
 
-            // find max_data_size for the chunk
+            // find max_data_size for the chunk, round length up to a multiple of wb()
             let max_data_size = align!(
                 data_blobs
                     .iter()
                     .fold(0, |max, blob| cmp::max(blob.read().unwrap().meta.size, max)),
-                JERASURE_ALIGN
+                wb()
             );
 
             let data_locks: Vec<_> = data_blobs.iter().map(|b| b.read().unwrap()).collect();
@@ -514,7 +519,7 @@ pub mod test {
     }
 
     #[test]
-    fn test_generate_blocktree_with_coding() {
+    fn test_erasure_generate_blocktree_with_coding() {
         let cases = vec![
             (NUM_DATA, NUM_CODING, 7, 5),
             (NUM_DATA - 6, NUM_CODING - 1, 5, 7),
diff --git a/core/src/packet.rs b/core/src/packet.rs
index 2d7bcdc114..1bf6f810eb 100644
--- a/core/src/packet.rs
+++ b/core/src/packet.rs
@@ -25,7 +25,7 @@ pub type SharedBlobs = Vec<SharedBlob>;
 pub const NUM_PACKETS: usize = 1024 * 8;
 pub const BLOB_SIZE: usize = (64 * 1024 - 128); // wikipedia says there should be 20b for ipv4 headers
 pub const BLOB_DATA_SIZE: usize = BLOB_SIZE - (BLOB_HEADER_SIZE * 2);
-pub const BLOB_DATA_ALIGN: usize = 64;
+pub const BLOB_DATA_ALIGN: usize = 16; // safe for erasure input pointers, gf.c needs 16byte-aligned buffers
 pub const NUM_BLOBS: usize = (NUM_PACKETS * PACKET_DATA_SIZE) / BLOB_SIZE;
 
 #[derive(Clone, Default, Debug, PartialEq, Serialize, Deserialize)]
@@ -143,7 +143,7 @@ impl Packets {
     }
 }
 
-#[repr(align(64))] // 64 === BLOB_DATA_ALIGN
+#[repr(align(16))] // 16 === BLOB_DATA_ALIGN
 pub struct BlobData {
     pub data: [u8; BLOB_SIZE],
 }
@@ -372,7 +372,7 @@ macro_rules! align {
     };
 }
 
-pub const BLOB_HEADER_SIZE: usize = align!(SIZE_RANGE.end, BLOB_DATA_ALIGN);
+pub const BLOB_HEADER_SIZE: usize = align!(SIZE_RANGE.end, BLOB_DATA_ALIGN); // make sure data() is safe for erasure
 
 pub const BLOB_FLAG_IS_LAST_IN_SLOT: u32 = 0x2;