Merge lp:~sergei.glushchenko/percona-xtrabackup/2.2-xb-bug1391216 into lp:percona-xtrabackup/2.2

[Sergei Glushchenko]

    Bug 1391216: Xtrabackup 2.2.5 fails with non-default innodb_log_block_size in 5.6.19

    Port fix for Percona Server bug 1114612 to XtraBackup.

    Original commit message:

    Fix bug 1114612 (Failing assertion: n % srv_log_block_size == 0 in
    file os0file.c line 4269).

    The assertion happens because InnoDB I/O and AIO routines assume that
    all I/O is done in log block size blocks. This works OK with the
    default log block size of 512, but is likely to assert with large
    innodb_log_block_size values.

    Fix by asserting that all I/O is done in the minimum allowed log block
    size blocks, that is, 512 bytes.

http://jenkins.percona.com/view/PXB%202.2/job/percona-xtrabackup-2.2-param/254/

Note: I could not reproduce the crash, the fix is what seems logical to do.

[Alexey Kopytov (akopytov)]

review: Approve

=== modified file 'storage/innobase/fil/fil0fil.cc'
--- storage/innobase/fil/fil0fil.cc	2014-09-15 08:52:36 +0000
+++ storage/innobase/fil/fil0fil.cc	2014-12-01 10:25:44 +0000
@@ -5651,8 +5651,8 @@
 
 	/* Do aio */
 
-	ut_a(byte_offset % OS_FILE_LOG_BLOCK_SIZE == 0);
-	ut_a((len % OS_FILE_LOG_BLOCK_SIZE) == 0);
+	ut_a(byte_offset % OS_MIN_LOG_BLOCK_SIZE == 0);
+	ut_a((len % OS_MIN_LOG_BLOCK_SIZE) == 0);
 
 #ifdef UNIV_HOTBACKUP
 	/* In mysqlbackup do normal i/o, not aio */

=== modified file 'storage/innobase/os/os0file.cc'
--- storage/innobase/os/os0file.cc	2014-10-16 17:32:22 +0000
+++ storage/innobase/os/os0file.cc	2014-12-01 10:25:44 +0000
@@ -4521,8 +4521,8 @@
 	ut_ad(file);
 	ut_ad(buf);
 	ut_ad(n > 0);
-	ut_ad(n % OS_FILE_LOG_BLOCK_SIZE == 0);
-	ut_ad(offset % OS_FILE_LOG_BLOCK_SIZE == 0);
+	ut_ad(n % OS_MIN_LOG_BLOCK_SIZE == 0);
+	ut_ad(offset % OS_MIN_LOG_BLOCK_SIZE == 0);
 	ut_ad(os_aio_validate_skip());
 #ifdef WIN_ASYNC_IO
 	ut_ad((n & 0xFFFFFFFFUL) == n);