Do not pad the final block of a file with zeroes.

To accomplish this, a resize operation was added for reference-counted
strings.

diff --git a/bluesky/bluesky.h b/bluesky/bluesky.h
index 7408849..dbb69bb 100644 (file)
--- a/bluesky/bluesky.h
+++ b/bluesky/bluesky.h
@@ -33,6 +33,7 @@ BlueSkyRCStr *bluesky_string_new(gpointer data, gsize len);
 void bluesky_string_ref(BlueSkyRCStr *string);
 void bluesky_string_unref(BlueSkyRCStr *string);
 BlueSkyRCStr *bluesky_string_dup(BlueSkyRCStr *string);
+void bluesky_string_resize(BlueSkyRCStr *string, gsize len);
 
 /* Cryptographic operations. */
 #define CRYPTO_BLOCK_SIZE 16        /* 128-bit AES */

diff --git a/bluesky/file.c b/bluesky/file.c
index 607d611..75768d1 100644 (file)
--- a/bluesky/file.c
+++ b/bluesky/file.c
@@ -21,10 +21,17 @@ void bluesky_block_touch(BlueSkyInode *inode, uint64_t i)
     g_return_if_fail(i < inode->blocks->len);
     BlueSkyBlock *block = &g_array_index(inode->blocks, BlueSkyBlock, i);
 
+    gsize block_len;
+    if (i < inode->blocks->len - 1) {
+        block_len = BLUESKY_BLOCK_SIZE;
+    } else {
+        block_len = inode->size - i * BLUESKY_BLOCK_SIZE;
+    }
+
     switch (block->type) {
     case BLUESKY_BLOCK_ZERO:
-        block->data = bluesky_string_new(g_malloc0(BLUESKY_BLOCK_SIZE),
-                                         BLUESKY_BLOCK_SIZE);
+        g_print("Allocating zero block of size %zd\n", block_len);
+        block->data = bluesky_string_new(g_malloc0(block_len), block_len);
         break;
     case BLUESKY_BLOCK_REF:
         bluesky_block_fetch(inode->fs, block);
@@ -65,17 +72,21 @@ void bluesky_file_truncate(BlueSkyInode *inode, uint64_t size)
         g_array_set_size(inode->blocks, blocks);
     }
 
-    /* If the file size is being decreased, ensure that any trailing data in
-     * the last block is zeroed. */
-    if (size < inode->size) {
+    /* Ensure the last block of the file is properly sized.  If the block is
+     * extended, newly-added bytes must be zeroed. */
+    if (blocks > 0) {
         BlueSkyBlock *b = &g_array_index(inode->blocks, BlueSkyBlock,
                                          blocks - 1);
+
         if (b->type != BLUESKY_BLOCK_ZERO) {
             bluesky_block_touch(inode, blocks - 1);
-            int end_offset = size % BLUESKY_BLOCK_SIZE;
-            if (end_offset > 0) {
-                memset(&b->data->data[end_offset], 0,
-                       BLUESKY_BLOCK_SIZE - end_offset);
+            gsize old_size = b->data->len;
+            gsize new_size = size - (blocks - 1) * BLUESKY_BLOCK_SIZE;
+
+            bluesky_string_resize(b->data, new_size);
+
+            if (new_size > old_size) {
+                memset(&b->data->data[old_size], 0, new_size - old_size);
             }
         }
     }

diff --git a/bluesky/store.c b/bluesky/store.c
index 50b47b0..52cc6b3 100644 (file)
--- a/bluesky/store.c
+++ b/bluesky/store.c
@@ -64,59 +64,6 @@ void bluesky_store_put(BlueSkyStore *store,
     store->impl->put(store->handle, key, val);
 }
 
-/* Create and return a new reference-counted string.  The reference count is
- * initially one.  The newly-returned string takes ownership of the memory
- * pointed at by data, and will call g_free on it when the reference count
- * drops to zero. */
-BlueSkyRCStr *bluesky_string_new(gpointer data, gsize len)
-{
-    BlueSkyRCStr *string = g_new(BlueSkyRCStr, 1);
-    string->data = data;
-    string->len = len;
-    g_atomic_int_set(&string->refcount, 1);
-    return string;
-}
-
-void bluesky_string_ref(BlueSkyRCStr *string)
-{
-    if (string == NULL)
-        return;
-
-    g_atomic_int_inc(&string->refcount);
-}
-
-void bluesky_string_unref(BlueSkyRCStr *string)
-{
-    if (string == NULL)
-        return;
-
-    if (g_atomic_int_dec_and_test(&string->refcount)) {
-        g_free(string->data);
-        g_free(string);
-    }
-}
-
-/* Duplicate and return a new reference-counted string, containing a copy of
- * the original data, with a reference count of 1.  As an optimization, if the
- * passed-in string already has a reference count of 1, the original is
- * returned.   Can be used to make a mutable copy of a shared string.  For this
- * to truly be safe, it is probably needed that there be some type of lock
- * protecting access to the string. */
-BlueSkyRCStr *bluesky_string_dup(BlueSkyRCStr *string)
-{
-    if (string == NULL)
-        return NULL;
-
-    if (g_atomic_int_dec_and_test(&string->refcount)) {
-        /* There are no other shared copies, so return this one. */
-        g_atomic_int_inc(&string->refcount);
-        return string;
-    } else {
-        return bluesky_string_new(g_memdup(string->data, string->len),
-                                  string->len);
-    }
-}
-
 /* Simple in-memory data store for test purposes. */
 typedef struct {
     GMutex *lock;

diff --git a/bluesky/util.c b/bluesky/util.c
index 1f98990..1f21bbf 100644 (file)
--- a/bluesky/util.c
+++ b/bluesky/util.c
@@ -35,3 +35,71 @@ gchar *bluesky_lowercase(const gchar *s)
     /* TODO: Unicode handling; for now just do ASCII. */
     return g_ascii_strdown(s, -1);
 }
+
+/**** Reference-counted strings. ****/
+
+/* Create and return a new reference-counted string.  The reference count is
+ * initially one.  The newly-returned string takes ownership of the memory
+ * pointed at by data, and will call g_free on it when the reference count
+ * drops to zero. */
+BlueSkyRCStr *bluesky_string_new(gpointer data, gsize len)
+{
+    BlueSkyRCStr *string = g_new(BlueSkyRCStr, 1);
+    string->data = data;
+    string->len = len;
+    g_atomic_int_set(&string->refcount, 1);
+    return string;
+}
+
+void bluesky_string_ref(BlueSkyRCStr *string)
+{
+    if (string == NULL)
+        return;
+
+    g_atomic_int_inc(&string->refcount);
+}
+
+void bluesky_string_unref(BlueSkyRCStr *string)
+{
+    if (string == NULL)
+        return;
+
+    if (g_atomic_int_dec_and_test(&string->refcount)) {
+        g_free(string->data);
+        g_free(string);
+    }
+}
+
+/* Duplicate and return a new reference-counted string, containing a copy of
+ * the original data, with a reference count of 1.  As an optimization, if the
+ * passed-in string already has a reference count of 1, the original is
+ * returned.   Can be used to make a mutable copy of a shared string.  For this
+ * to truly be safe, it is probably needed that there be some type of lock
+ * protecting access to the string. */
+BlueSkyRCStr *bluesky_string_dup(BlueSkyRCStr *string)
+{
+    if (string == NULL)
+        return NULL;
+
+    if (g_atomic_int_dec_and_test(&string->refcount)) {
+        /* There are no other shared copies, so return this one. */
+        g_atomic_int_inc(&string->refcount);
+        return string;
+    } else {
+        return bluesky_string_new(g_memdup(string->data, string->len),
+                                  string->len);
+    }
+}
+
+/* Resize the data block used by a BlueSkyRCStr.  The data pointer might change
+ * after making this call, so it should not be cached across calls to this
+ * function.  To avoid confusing any other users, the caller probably ought to
+ * hold the only reference to the string (by calling bluesky_string_dup first
+ * if needed). */
+void bluesky_string_resize(BlueSkyRCStr *string, gsize len)
+{
+    if (string->len == len)
+        return;
+
+    string->data = g_realloc(string->data, len);
+}