Rename big_write.block to location

blockstore_init.cpp

@@ -310,7 +310,7 @@ int blockstore_init_journal::handle_journal_part(void *buf, uint64_t len)
                 }, (dirty_entry){
                     .state = ST_D_META_SYNCED,
                     .flags = 0,
-                    .location = je->big_write.block,
+                    .location = je->big_write.location,
                     .offset = 0,
                     .size = bs->block_size,
                 });

blockstore_journal.h

@@ -49,7 +49,7 @@ struct __attribute__((__packed__)) journal_entry_big_write
     uint32_t crc32_prev;
     object_id oid;
     uint64_t version;
-    uint64_t block;
+    uint64_t location;
 };
 
 struct __attribute__((__packed__)) journal_entry_stable

blockstore_sync.cpp

@@ -79,7 +79,7 @@ int blockstore::continue_sync(blockstore_operation *op)
                 prefill_single_journal_entry(journal, JE_BIG_WRITE, sizeof(journal_entry_big_write));
             je->oid = it->oid;
             je->version = it->version;
-            je->block = dirty_db[*it].location;
+            je->location = dirty_db[*it].location;
             je->crc32 = je_crc32((journal_entry*)je);
             journal.crc32_last = je->crc32;
             it++;

blockstore_write.cpp

@@ -81,8 +81,8 @@ int blockstore::dequeue_write(blockstore_operation *op)
         if (op->version == 1 && op->len != block_size)
         {
             // zero fill newly allocated object
-            // FIXME: it's not so good because it turns new small writes into big writes
-            // but it's the first and the simplest implementation
+            // This thing turns new small writes into big writes
+            // So FIXME: consider writing an empty big_write as version 1 instead of zero-filling here
             if (op->offset > 0)
                 op->iov_zerofill[vcnt++] = (struct iovec){ zero_object, op->offset };
             op->iov_zerofill[vcnt++] = (struct iovec){ op->buf, op->len };