Preparatory work before implementing proper cloud writing.

[bluesky.git] / bluesky / cloudlog.c

diff --git a/bluesky/cloudlog.c b/bluesky/cloudlog.c
index 0c5db98..e1fd44a 100644 (file)
--- a/bluesky/cloudlog.c
+++ b/bluesky/cloudlog.c
@@ -6,6 +6,7 @@
  * TODO: Licensing
  */
 
+#include <stdio.h>
 #include <stdint.h>
 #include <glib.h>
 #include <string.h>
@@ -27,6 +28,40 @@ BlueSkyCloudID bluesky_cloudlog_new_id()
     return id;
 }
 
+gchar *bluesky_cloudlog_id_to_string(BlueSkyCloudID id)
+{
+    char buf[sizeof(BlueSkyCloudID) * 2 + 1];
+    buf[0] = '\0';
+
+    for (int i = 0; i < sizeof(BlueSkyCloudID); i++) {
+        sprintf(&buf[2*i], "%02x", (uint8_t)(id.bytes[i]));
+    }
+
+    return g_strdup(buf);
+}
+
+BlueSkyCloudID bluesky_cloudlog_id_from_string(const gchar *idstr)
+{
+    BlueSkyCloudID id;
+    memset(&id, 0, sizeof(id));
+    for (int i = 0; i < 2*sizeof(BlueSkyCloudID); i++) {
+        char c = idstr[i];
+        if (c == '\0') {
+            g_warning("Short cloud id: %s\n", idstr);
+            break;
+        }
+        int val = 0;
+        if (c >= '0' && c <= '9')
+            val = c - '0';
+        else if (c >= 'a' && c <= 'f')
+            val = c - 'a' + 10;
+        else
+            g_warning("Bad character in cloud id: %s\n", idstr);
+        id.bytes[i / 2] += val << (i % 2 ? 0 : 4);
+    }
+    return id;
+}
+
 gboolean bluesky_cloudlog_equal(gconstpointer a, gconstpointer b)
 {
     BlueSkyCloudID *id1 = (BlueSkyCloudID *)a, *id2 = (BlueSkyCloudID *)b;
@@ -54,6 +89,7 @@ BlueSkyCloudLog *bluesky_cloudlog_new(BlueSkyFS *fs)
     log->fs = fs;
     log->type = LOGTYPE_UNKNOWN;
     log->id = bluesky_cloudlog_new_id();
+    log->pointers = g_array_new(FALSE, TRUE, sizeof(BlueSkyCloudID));
     g_atomic_int_set(&log->refcount, 1);
 
     return log;
@@ -75,7 +111,7 @@ void bluesky_cloudlog_unref(BlueSkyCloudLog *log)
 BlueSkyLogItem *bluesky_cloudlog_sync(BlueSkyCloudLog *log)
 {
     BlueSkyLogItem *log_item = bluesky_log_item_new();
-    log_item->key = g_strdup("cloudlog");
+    log_item->key = bluesky_cloudlog_id_to_string(log->id);
     log_item->data = log->data;
     bluesky_string_ref(log->data);
     bluesky_log_item_submit(log_item, log->fs->log);
@@ -90,3 +126,112 @@ void bluesky_cloudlog_insert(BlueSkyCloudLog *log)
     g_hash_table_insert(log->fs->locations, &log->id, log);
     g_mutex_unlock(log->fs->lock);
 }
+
+struct log_header {
+    char magic[4];
+    uint32_t size;
+    BlueSkyCloudID id;
+    uint32_t pointer_count;
+} __attribute__((packed));
+
+struct logref {
+    BlueSkyCloudID id;
+    BlueSkyCloudPointer location;
+} __attribute__((packed));
+
+struct log_footer {
+    char refmagic[4];
+    struct logref refs[0];
+};
+
+BlueSkyCloudPointer bluesky_cloudlog_serialize(BlueSkyCloudLog *log,
+                                               BlueSkyCloudLogState *state)
+{
+    if (log->location_flags & CLOUDLOG_CLOUD) {
+        return log->location;
+    }
+
+    g_print("Flushing object %s to cloud...\n",
+            bluesky_cloudlog_id_to_string(log->id));
+
+    for (int i = 0; i < log->pointers->len; i++) {
+        BlueSkyCloudID id = g_array_index(log->pointers, BlueSkyCloudID, i);
+        g_print("  ...checking reference %s...\n",
+                bluesky_cloudlog_id_to_string(id));
+        g_mutex_lock(log->fs->lock);
+        BlueSkyCloudLog *log2
+            = (BlueSkyCloudLog *)g_hash_table_lookup(log->fs->locations, &id);
+        // TODO: refcount
+        g_mutex_unlock(log->fs->lock);
+        g_assert(log2 != NULL);
+        bluesky_cloudlog_serialize(log2, state);
+    }
+
+    g_assert(log->data != NULL);
+
+    log->location = state->location;
+    log->location.offset = state->data->len;
+    log->location.size
+        = sizeof(struct log_header) + sizeof(BlueSkyCloudID) * 0
+           + log->data->len;
+
+    struct log_header header;
+    memcpy(header.magic, "AgI ", 4);
+    header.size = GUINT32_TO_LE(log->location.size);
+    header.id = log->id;
+    header.pointer_count = GUINT32_TO_LE(0);
+
+    g_string_append_len(state->data, (const char *)&header, sizeof(header));
+    g_string_append_len(state->data, log->data->data, log->data->len);
+
+    log->location_flags |= CLOUDLOG_CLOUD;
+
+    return log->location;
+}
+
+static void find_inodes(gpointer key, gpointer value, gpointer user_data)
+{
+    BlueSkyCloudLogState *state = (BlueSkyCloudLogState *)user_data;
+    BlueSkyCloudLog *item = (BlueSkyCloudLog *)value;
+
+    if (item->type != LOGTYPE_INODE)
+        return;
+
+    bluesky_cloudlog_ref(item);
+    state->inode_list = g_list_prepend(state->inode_list, item);
+}
+
+void bluesky_cloudlog_write_log(BlueSkyFS *fs)
+{
+    g_print("Starting cloudlog write...\n");
+
+    BlueSkyCloudLogState *state = fs->log_state;
+    state->data = g_string_new("");
+
+    g_mutex_lock(fs->lock);
+    g_hash_table_foreach(fs->locations, find_inodes, state);
+    g_mutex_unlock(fs->lock);
+
+    while (state->inode_list != NULL) {
+        BlueSkyCloudLog *log = (BlueSkyCloudLog *)state->inode_list->data;
+        bluesky_cloudlog_serialize(log, state);
+        state->inode_list = g_list_delete_link(state->inode_list,
+                                               state->inode_list);
+    }
+
+    g_print("Serialized %zd bytes of data\n", state->data->len);
+
+    BlueSkyStoreAsync *async = bluesky_store_async_new(fs->store);
+    async->op = STORE_OP_PUT;
+    async->key = g_strdup_printf("log-%08d-%08d",
+                                 state->location.directory,
+                                 state->location.sequence);
+    async->data = bluesky_string_new_from_gstring(state->data);
+    bluesky_store_async_submit(async);
+    bluesky_store_async_wait(async);
+    bluesky_store_async_unref(async);
+
+    state->data = NULL;
+    state->location.sequence++;
+    state->location.offset = 0;
+}