From ddc38452bea958b4ceeb00ff51550973fab35d8e Mon Sep 17 00:00:00 2001
From: Jef Roosens <roosensjef@gmail.com>
Date: Fri, 10 Nov 2023 11:34:55 +0100
Subject: [PATCH] feat(lsm): don't sync non-dirty entries

---
 lsm/src/_include/lsm/store_internal.h | 11 ++++-------
 lsm/src/store/lsm_store.c             |  4 ++++
 lsm/src/store/lsm_store_disk_read.c   |  6 +++++-
 lsm/src/store/lsm_store_entry.c       | 17 ++++++++++-------
 4 files changed, 23 insertions(+), 15 deletions(-)

diff --git a/lsm/src/_include/lsm/store_internal.h b/lsm/src/_include/lsm/store_internal.h
index cebb41b..ccafe99 100644
--- a/lsm/src/_include/lsm/store_internal.h
+++ b/lsm/src/_include/lsm/store_internal.h
@@ -59,8 +59,12 @@ void lsm_entry_wrapper_free(lsm_entry_wrapper *wrapper);
 struct lsm_entry_handle {
   lsm_entry_wrapper *wrapper;
   lsm_store *store;
+  // Either read or append, depending on how it was opened
   FILE *f;
+  // Current position in the file pointer
   uint64_t pos;
+  // Whether the entry's metadata has changed
+  bool dirty;
 };
 
 lsm_error lsm_entry_handle_init(lsm_entry_handle **out);
@@ -88,13 +92,6 @@ struct lsm_store {
  */
 lsm_error lsm_store_load_db(lsm_store *store);
 
-/**
- * Close & free the handle without updating the database
- *
- * @param handle handle to close
- */
-void lsm_entry_close_no_disk(lsm_entry_handle *handle);
-
 /**
  * Write a new insert to the database.
  *
diff --git a/lsm/src/store/lsm_store.c b/lsm/src/store/lsm_store.c
index 2da7c51..022e68b 100644
--- a/lsm/src/store/lsm_store.c
+++ b/lsm/src/store/lsm_store.c
@@ -150,6 +150,9 @@ lsm_error lsm_store_insert(lsm_entry_handle **out, lsm_store *store,
   handle->wrapper = wrapper;
   handle->store = store;
 
+  // Newly inserted entries are always dirty
+  handle->dirty = true;
+
   *out = handle;
 
   return lsm_error_ok;
@@ -189,6 +192,7 @@ lsm_error lsm_entry_data_append(lsm_entry_handle *handle, lsm_str *data) {
   }
 
   entry->data_len = new_len;
+  handle->dirty = true;
 
   return lsm_error_ok;
 }
diff --git a/lsm/src/store/lsm_store_disk_read.c b/lsm/src/store/lsm_store_disk_read.c
index 040708a..5c71dee 100644
--- a/lsm/src/store/lsm_store_disk_read.c
+++ b/lsm/src/store/lsm_store_disk_read.c
@@ -172,7 +172,11 @@ lsm_error lsm_store_load_db(lsm_store *store) {
       LSM_RES(lsm_entry_read_attrs(NULL, handle, store->db_file));
 
       handle->wrapper->entry->idx_file_offset = idx_file_offset;
-      lsm_entry_close_no_disk(handle);
+
+      // We explicitely set the dirty flag here to prevent writing to the datase
+      // when reading it in
+      handle->dirty = false;
+      lsm_entry_close(handle);
 
       store->db_file_size += db_dim[1];
     }
diff --git a/lsm/src/store/lsm_store_entry.c b/lsm/src/store/lsm_store_entry.c
index 45ead55..4623f36 100644
--- a/lsm/src/store/lsm_store_entry.c
+++ b/lsm/src/store/lsm_store_entry.c
@@ -47,19 +47,18 @@ lsm_error lsm_entry_handle_init(lsm_entry_handle **out) {
   return lsm_error_ok;
 }
 
-void lsm_entry_close_no_disk(lsm_entry_handle *handle) {
-  pthread_rwlock_unlock(&handle->wrapper->lock);
-  free(handle);
-}
-
 void lsm_entry_close(lsm_entry_handle *handle) {
   if (handle->f != NULL) {
     fclose(handle->f);
   }
 
   // TODO handle errors here
-  lsm_entry_disk_insert(handle);
-  lsm_entry_close_no_disk(handle);
+  if (handle->dirty) {
+    lsm_entry_disk_insert(handle);
+  }
+
+  pthread_rwlock_unlock(&handle->wrapper->lock);
+  free(handle);
 }
 
 bool lsm_entry_attr_present(lsm_entry_handle *handle, uint8_t type) {
@@ -158,6 +157,8 @@ lsm_error lsm_entry_attr_remove(lsm_str **out, lsm_entry_handle *handle,
   entry->attrs.count--;
   entry->attrs.bitmap[type / 64] &= ~(((uint64_t)1) << (type % 64));
 
+  handle->dirty = true;
+
   return lsm_error_ok;
 }
 
@@ -183,6 +184,8 @@ lsm_error lsm_entry_attr_insert(lsm_entry_handle *handle, uint8_t type,
   entry->attrs.count++;
   entry->attrs.bitmap[type / 64] |= ((uint64_t)1) << (type % 64);
 
+  handle->dirty = true;
+
   return lsm_error_ok;
 }