MDEV-31669: use XXH3 as a digest instead of md5

mysql-test/include/check_digest.inc

@@ -11,7 +11,7 @@ returns tinyint
 not deterministic
 begin
   declare digest_exists tinyint;
-  if length(digest) != 32 or conv(digest, 16, 10) = 0 then
+  if length(digest) != 32 or digest is null then
     return 0;
   end if;
   select exists (select d from test._digests where d = digest) into digest_exists;

scripts/sys_schema/procedures/ps_trace_statement_digest.sql

@@ -33,7 +33,7 @@ CREATE DEFINER='mariadb.sys'@'localhost' PROCEDURE ps_trace_statement_digest (
 
              When finding a statement of interest within the 
              performance_schema.events_statements_summary_by_digest table, feed
-             the DIGEST MD5 value in to this procedure, set how long to poll for, 
+             the DIGEST hash value in to this procedure, set how long to poll for, 
              and at what interval to poll, and it will generate a report of all 
              statistics tracked within Performance Schema for that digest for the
              interval.

sql/sql_digest.cc

@@ -19,7 +19,7 @@
 */
 
 #include "mariadb.h"
-#include "my_md5.h"
+#include "../mysys/xxhash.h"
 #include "unireg.h"
 
 #include "sql_string.h"
@@ -157,11 +157,11 @@ inline void store_token_identifier(sql_digest_storage* digest_storage,
   }
 }
 
-void compute_digest_md5(const sql_digest_storage *digest_storage, unsigned char *md5)
+void compute_digest_hash(const sql_digest_storage *digest_storage, unsigned char *hash)
 {
-  compute_md5_hash(md5,
-                   (const char *) digest_storage->m_token_array,
-                   digest_storage->m_byte_count);
+  XXH128_hash_t res = XXH3_128bits(digest_storage->m_token_array, 
+                                digest_storage->m_byte_count);
+  memcpy(hash, &res, sizeof(res));
 }
 
 /*

sql/sql_digest.h

@@ -18,10 +18,11 @@
 
 #include <string.h>
 class String;
-#include "my_md5.h"
 
 #define MAX_DIGEST_STORAGE_SIZE (1024*1024)
 
+#define DIGEST_HASH_SIZE 16
+
 /**
   Structure to store token count/array for a statement
   on which digest is to be calculated.
@@ -30,7 +31,7 @@ struct sql_digest_storage
 {
   bool m_full;
   uint m_byte_count;
-  unsigned char m_md5[MD5_HASH_SIZE];
+  unsigned char m_hash[DIGEST_HASH_SIZE];
   /** Character set number. */
   uint m_charset_number;
   /**
@@ -66,7 +67,7 @@ struct sql_digest_storage
     m_full= false;
     m_byte_count= 0;
     m_charset_number= 0;
-    memset(m_md5, 0, MD5_HASH_SIZE);
+    memset(m_hash, 0, DIGEST_HASH_SIZE);
   }
 
   inline bool is_empty()
@@ -90,7 +91,7 @@ struct sql_digest_storage
       m_byte_count= byte_count_copy;
       m_charset_number= from->m_charset_number;
       memcpy(m_token_array, from->m_token_array, m_byte_count);
-      memcpy(m_md5, from->m_md5, MD5_HASH_SIZE);
+      memcpy(m_hash, from->m_hash, DIGEST_HASH_SIZE);
     }
     else
     {
@@ -105,9 +106,9 @@ typedef struct sql_digest_storage sql_digest_storage;
 /**
   Compute a digest hash.
   @param digest_storage The digest
-  @param [out] md5 The computed digest hash. This parameter is a buffer of size @c MD5_HASH_SIZE.
+  @param [out] hash The computed digest hash. This parameter is a buffer of size @c DIGEST_HASH_SIZE.
 */
-void compute_digest_md5(const sql_digest_storage *digest_storage, unsigned char *md5);
+void compute_digest_hash(const sql_digest_storage *digest_storage, unsigned char *hash);
 
 /**
   Compute a digest text.

storage/perfschema/pfs_column_types.h

@@ -61,7 +61,7 @@
 #define COL_SOURCE_SIZE 64
 
 /** Size of the DIGEST columns. */
-#define COL_DIGEST_SIZE 64
+#define COL_DIGEST_SIZE 32
 
 /**
   Enum values for the TIMER_NAME columns.

storage/perfschema/pfs_digest.cc

@@ -224,9 +224,9 @@ find_or_create_digest(PFS_thread *thread,
   */
   PFS_digest_key hash_key;
   memset(& hash_key, 0, sizeof(hash_key));
-  /* Compute MD5 Hash of the tokens received. */
-  compute_digest_md5(digest_storage, hash_key.m_md5);
-  memcpy((void*)& digest_storage->m_md5, &hash_key.m_md5, MD5_HASH_SIZE);
+  /* Comchpute digest hash of the tokens received. */
+  compute_digest_hash(digest_storage, hash_key.m_hash);
+  memcpy((void*)& digest_storage->m_hash, &hash_key.m_hash, DIGEST_HASH_SIZE);
   /* Add the current schema to the key */
   hash_key.m_schema_name_length= schema_name_length;
   if (schema_name_length > 0)

storage/perfschema/pfs_digest.h

@@ -39,11 +39,11 @@ extern ulong digest_lost;
 struct PFS_thread;
 
 /**
-  Structure to store a MD5 hash value (digest) for a statement.
+  Structure to store a hash value (digest) for a statement.
 */
 struct PFS_digest_key
 {
-  unsigned char m_md5[MD5_HASH_SIZE];
+  unsigned char m_hash[DIGEST_HASH_SIZE];
   char m_schema_name[NAME_LEN];
   uint m_schema_name_length;
 };
@@ -54,7 +54,7 @@ struct PFS_ALIGNED PFS_statements_digest_stat
   /** Internal lock. */
   pfs_lock m_lock;
 
-  /** Digest Schema + MD5 Hash. */
+  /** Digest Schema + Digest Hash. */
   PFS_digest_key m_digest_key;
 
   /** Digest Storage. */

storage/perfschema/table_events_statements.cc

@@ -34,7 +34,6 @@
 #include "pfs_timer.h"
 #include "sp_head.h" /* TYPE_ENUM_FUNCTION, ... */
 #include "table_helper.h"
-#include "my_md5.h"
 #include "pfs_buffer_container.h"
 
 THR_LOCK table_events_statements_current::m_table_lock;
@@ -368,10 +367,10 @@ void table_events_statements_common::make_row_part_2(const sql_digest_storage *d
   if (safe_byte_count > 0 &&
       safe_byte_count <= pfs_max_digest_length)
   {
-    /* Generate the DIGEST string from the MD5 digest  */
-    MD5_HASH_TO_STRING(digest->m_md5,
+    /* Generate the DIGEST string from the digest */
+    DIGEST_HASH_TO_STRING(digest->m_hash,
                        m_row.m_digest.m_digest);
-    m_row.m_digest.m_digest_length= MD5_HASH_TO_STRING_LENGTH;
+    m_row.m_digest.m_digest_length= DIGEST_HASH_TO_STRING_LENGTH;
 
     /* Generate the DIGEST_TEXT string from the token array */
     compute_digest_text(digest, &m_row.m_digest.m_digest_text);

storage/perfschema/table_helper.cc

@@ -132,11 +132,11 @@ int PFS_digest_row::make_row(PFS_statements_digest_stat* pfs)
   if (safe_byte_count > 0)
   {
     /*
-      Calculate digest from MD5 HASH collected to be shown as
+      Calculate digest from HASH collected to be shown as
       DIGEST in this row.
     */
-    MD5_HASH_TO_STRING(pfs->m_digest_storage.m_md5, m_digest);
-    m_digest_length= MD5_HASH_TO_STRING_LENGTH;
+    DIGEST_HASH_TO_STRING(pfs->m_digest_storage.m_hash, m_digest);
+    m_digest_length= DIGEST_HASH_TO_STRING_LENGTH;
 
     /*
       Calculate digest_text information from the token array collected

storage/perfschema/table_helper.h

@@ -31,18 +31,19 @@
 #include "pfs_digest.h"
 
 /*
-  Write MD5 hash value in a string to be used
+  Write XXH3 hash value in a string to be used
   as DIGEST for the statement.
 */
-#define MD5_HASH_TO_STRING(_hash, _str)                    \
+#define DIGEST_HASH_TO_STRING(_hash, _str)                 \
   sprintf(_str, "%02x%02x%02x%02x%02x%02x%02x%02x"         \
                 "%02x%02x%02x%02x%02x%02x%02x%02x",        \
           _hash[0], _hash[1], _hash[2], _hash[3],          \
           _hash[4], _hash[5], _hash[6], _hash[7],          \
           _hash[8], _hash[9], _hash[10], _hash[11],        \
           _hash[12], _hash[13], _hash[14], _hash[15])
 
-#define MD5_HASH_TO_STRING_LENGTH 32
+/* XXH3_128bits = 16 bytes of binary = 128 printable characters */
+#define DIGEST_HASH_TO_STRING_LENGTH 32
 
 struct PFS_host;
 struct PFS_user;
@@ -140,7 +141,7 @@ struct PFS_digest_row
   /** Length in bytes of @c m_schema_name. */
   uint m_schema_name_length;
   /** Column DIGEST. */
-  char m_digest[COL_DIGEST_SIZE];
+    char m_digest[DIGEST_HASH_TO_STRING_LENGTH + 1];
   /** Length in bytes of @c m_digest. */
   uint m_digest_length;
   /** Column DIGEST_TEXT. */

storage/perfschema/unittest/pfs_server_stubs.cc

@@ -37,7 +37,7 @@ uint lower_case_table_names= 0;
 CHARSET_INFO *files_charset_info= NULL;
 CHARSET_INFO *system_charset_info= NULL;
 
-void compute_digest_md5(const sql_digest_storage *, unsigned char *)
+void compute_digest_hash(const sql_digest_storage *, unsigned char *)
 {
 }