Patchwork [RFC,V4,02/30] qcow2: Add deduplication structures and fields.

login
register
mail settings
Submitter Benoît Canet
Date Jan. 2, 2013, 4:16 p.m.
Message ID <1357143393-29832-3-git-send-email-benoit@irqsave.net>
Download mbox | patch
Permalink /patch/209074/
State New
Headers show

Comments

Benoît Canet - Jan. 2, 2013, 4:16 p.m.
Signed-off-by: Benoit Canet <benoit@irqsave.net>
---
 block/qcow2.h |   57 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 56 insertions(+), 1 deletion(-)

Patch

diff --git a/block/qcow2.h b/block/qcow2.h
index 718b52b..637c86a 100644
--- a/block/qcow2.h
+++ b/block/qcow2.h
@@ -58,6 +58,50 @@ 
 
 #define DEFAULT_CLUSTER_SIZE 65536
 
+#define HASH_LENGTH 32
+
+typedef enum {
+    QCOW_HASH_SHA256 = 0,
+    QCOW_HASH_SHA3   = 1,
+    QCOW_HASH_SKEIN  = 2,
+} QCowHashAlgo;
+
+typedef struct {
+    uint8_t data[HASH_LENGTH]; /* 32 bytes hash of a given cluster */
+} QCowHash;
+
+/* Used to keep a single precomputed hash between the calls of the dedup
+ * function
+ */
+typedef struct {
+    QCowHash hash;
+    bool reuse;                  /* The hash is precomputed reuse it */
+} QcowPersistantHash;
+
+/* deduplication node */
+typedef struct {
+    QCowHash hash;
+    uint64_t physical_sect;       /* where the cluster is stored on disk */
+    uint64_t first_logical_sect;  /* logical sector of the first occurence of
+                                   * this cluster
+                                   */
+} QCowHashNode;
+
+/* Undedupable hashes that must be written later to disk */
+typedef struct QCowHashElement {
+    QCowHash hash;
+    QTAILQ_ENTRY(QCowHashElement) next;
+} QCowHashElement;
+
+typedef struct {
+    QcowPersistantHash phash;  /* contains a hash persisting between calls of
+                                * qcow2_dedup()
+                                */
+    QTAILQ_HEAD(, QCowHashElement) undedupables;
+    int nb_clusters_processed;
+    int nb_undedupable_sectors;
+} QCowDedupState;
+
 typedef struct QCowHeader {
     uint32_t magic;
     uint32_t version;
@@ -114,8 +158,10 @@  enum {
 enum {
     QCOW2_INCOMPAT_DIRTY_BITNR   = 0,
     QCOW2_INCOMPAT_DIRTY         = 1 << QCOW2_INCOMPAT_DIRTY_BITNR,
+    QCOW2_INCOMPAT_DEDUP_BITNR   = 1,
+    QCOW2_INCOMPAT_DEDUP         = 1 << QCOW2_INCOMPAT_DEDUP_BITNR,
 
-    QCOW2_INCOMPAT_MASK          = QCOW2_INCOMPAT_DIRTY,
+    QCOW2_INCOMPAT_MASK          = QCOW2_INCOMPAT_DIRTY | QCOW2_INCOMPAT_DEDUP,
 };
 
 /* Compatible feature bits */
@@ -138,6 +184,7 @@  typedef struct BDRVQcowState {
     int cluster_sectors;
     int l2_bits;
     int l2_size;
+    int hash_block_size;
     int l1_size;
     int l1_vm_state_index;
     int csize_shift;
@@ -148,6 +195,7 @@  typedef struct BDRVQcowState {
 
     Qcow2Cache* l2_table_cache;
     Qcow2Cache* refcount_block_cache;
+    Qcow2Cache *dedup_cluster_cache;
 
     uint8_t *cluster_cache;
     uint8_t *cluster_data;
@@ -160,6 +208,13 @@  typedef struct BDRVQcowState {
     int64_t free_cluster_index;
     int64_t free_byte_offset;
 
+    bool has_dedup;
+    QCowHashAlgo dedup_hash_algo;
+    uint64_t *dedup_table;
+    uint64_t dedup_table_offset;
+    int32_t dedup_table_size;
+    GTree *dedup_tree_by_hash;
+    GTree *dedup_tree_by_sect;
     CoMutex lock;
 
     uint32_t crypt_method; /* current crypt method, 0 if no key yet */