From 0e0ae0bb06f75d8d745d35b2470e01671ddbe7a8 Mon Sep 17 00:00:00 2001
From: Sachin Setiya <sachinsetia1001@gmail.com>
Date: Tue, 14 Mar 2017 13:13:22 +0530
Subject: MW-28, codership/mysql-wsrep#28 Fix sync_thread_levels debug assert
Introduced a new wsrep_trx_print_locking() which may be called
under lock_sys->mutex if the trx has locks.
Signed-off-by: Sachin Setiya <sachinsetia1001@gmail.com>
---
storage/innobase/include/trx0trx.h | 17 ++++++
storage/innobase/lock/lock0lock.cc | 15 ++---
storage/innobase/trx/trx0trx.cc | 112 +++++++++++++++++++++++++++++++++++++
3 files changed, 134 insertions(+), 10 deletions(-)
(limited to 'storage/innobase')
diff --git a/storage/innobase/include/trx0trx.h b/storage/innobase/include/trx0trx.h
index 07123cd6d86..c7e0dc5a551 100644
--- a/storage/innobase/include/trx0trx.h
+++ b/storage/innobase/include/trx0trx.h
@@ -333,6 +333,23 @@ trx_print_latched(
or 0 to use the default max length */
MY_ATTRIBUTE((nonnull));
+#ifdef WITH_WSREP
+/**********************************************************************//**
+Prints info about a transaction.
+Transaction information may be retrieved without having trx_sys->mutex acquired
+so it may not be completely accurate. The caller must own lock_sys->mutex
+and the trx must have some locks to make sure that it does not escape
+without locking lock_sys->mutex. */
+UNIV_INTERN
+void
+wsrep_trx_print_locking(
+/*==============*/
+ FILE* f, /*!< in: output stream */
+ const trx_t* trx, /*!< in: transaction */
+ ulint max_query_len) /*!< in: max query length to print,
+ or 0 to use the default max length */
+ MY_ATTRIBUTE((nonnull));
+#endif /* WITH_WSREP */
/**********************************************************************//**
Prints info about a transaction.
Acquires and releases lock_sys->mutex and trx_sys->mutex. */
diff --git a/storage/innobase/lock/lock0lock.cc b/storage/innobase/lock/lock0lock.cc
index c44aa490a81..c1f5d71e1b6 100644
--- a/storage/innobase/lock/lock0lock.cc
+++ b/storage/innobase/lock/lock0lock.cc
@@ -1704,16 +1704,14 @@ wsrep_kill_victim(
is in the queue*/
} else if (lock->trx != trx) {
if (wsrep_log_conflicts) {
- mutex_enter(&trx_sys->mutex);
- if (bf_this) {
- fputs("\n*** Priority TRANSACTION:\n",
+ if (bf_this)
+ fputs("\n*** Priority TRANSACTION:\n",
stderr);
- } else {
+ else {
fputs("\n*** Victim TRANSACTION:\n",
stderr);
}
-
- trx_print_latched(stderr, trx, 3000);
+ wsrep_trx_print_locking(stderr, trx, 3000);
if (bf_other) {
fputs("\n*** Priority TRANSACTION:\n",
@@ -1722,10 +1720,7 @@ wsrep_kill_victim(
fputs("\n*** Victim TRANSACTION:\n",
stderr);
}
-
- trx_print_latched(stderr, lock->trx, 3000);
-
- mutex_exit(&trx_sys->mutex);
+ wsrep_trx_print_locking(stderr, lock->trx, 3000);
fputs("*** WAITING FOR THIS LOCK TO BE GRANTED:\n",
stderr);
diff --git a/storage/innobase/trx/trx0trx.cc b/storage/innobase/trx/trx0trx.cc
index de3afcdd8b0..3b39f685e6d 100644
--- a/storage/innobase/trx/trx0trx.cc
+++ b/storage/innobase/trx/trx0trx.cc
@@ -1876,6 +1876,118 @@ trx_print_latched(
mem_heap_get_size(trx->lock.lock_heap));
}
+#ifdef WITH_WSREP
+/**********************************************************************//**
+Prints info about a transaction.
+Transaction information may be retrieved without having trx_sys->mutex acquired
+so it may not be completely accurate. The caller must own lock_sys->mutex
+and the trx must have some locks to make sure that it does not escape
+without locking lock_sys->mutex. */
+UNIV_INTERN
+void
+wsrep_trx_print_locking(
+/*==========*/
+ FILE* f,
+ /*!< in: output stream */
+ const trx_t* trx,
+ /*!< in: transaction */
+ ulint max_query_len)
+ /*!< in: max query length to print,
+ or 0 to use the default max length */
+{
+ ibool newline;
+ const char* op_info;
+
+ ut_ad(lock_mutex_own());
+ ut_ad(trx->lock.trx_locks.count > 0);
+
+ fprintf(f, "TRANSACTION " TRX_ID_FMT, trx->id);
+
+ /* trx->state may change since trx_sys->mutex is not required */
+ switch (trx->state) {
+ case TRX_STATE_NOT_STARTED:
+ fputs(", not started", f);
+ goto state_ok;
+ case TRX_STATE_ACTIVE:
+ fprintf(f, ", ACTIVE %lu sec",
+ (ulong) difftime(time(NULL), trx->start_time));
+ goto state_ok;
+ case TRX_STATE_PREPARED:
+ fprintf(f, ", ACTIVE (PREPARED) %lu sec",
+ (ulong) difftime(time(NULL), trx->start_time));
+ goto state_ok;
+ case TRX_STATE_COMMITTED_IN_MEMORY:
+ fputs(", COMMITTED IN MEMORY", f);
+ goto state_ok;
+ }
+ fprintf(f, ", state %lu", (ulong) trx->state);
+ ut_ad(0);
+state_ok:
+
+ /* prevent a race condition */
+ op_info = trx->op_info;
+
+ if (*op_info) {
+ putc(' ', f);
+ fputs(op_info, f);
+ }
+
+ if (trx->is_recovered) {
+ fputs(" recovered trx", f);
+ }
+
+ if (trx->declared_to_be_inside_innodb) {
+ fprintf(f, ", thread declared inside InnoDB %lu",
+ (ulong) trx->n_tickets_to_enter_innodb);
+ }
+
+ putc('\n', f);
+
+ if (trx->n_mysql_tables_in_use > 0 || trx->mysql_n_tables_locked > 0) {
+ fprintf(f, "mysql tables in use %lu, locked %lu\n",
+ (ulong) trx->n_mysql_tables_in_use,
+ (ulong) trx->mysql_n_tables_locked);
+ }
+
+ newline = TRUE;
+
+ /* trx->lock.que_state of an ACTIVE transaction may change
+ while we are not holding trx->mutex. We perform a dirty read
+ for performance reasons. */
+
+ switch (trx->lock.que_state) {
+ case TRX_QUE_RUNNING:
+ newline = FALSE; break;
+ case TRX_QUE_LOCK_WAIT:
+ fputs("LOCK WAIT ", f); break;
+ case TRX_QUE_ROLLING_BACK:
+ fputs("ROLLING BACK ", f); break;
+ case TRX_QUE_COMMITTING:
+ fputs("COMMITTING ", f); break;
+ default:
+ fprintf(f, "que state %lu ", (ulong) trx->lock.que_state);
+ }
+
+ if (trx->has_search_latch) {
+ newline = TRUE;
+ fputs(", holds adaptive hash latch", f);
+ }
+
+ if (trx->undo_no != 0) {
+ newline = TRUE;
+ fprintf(f, ", undo log entries " TRX_ID_FMT, trx->undo_no);
+ }
+
+ if (newline) {
+ putc('\n', f);
+ }
+
+ if (trx->mysql_thd != NULL) {
+ innobase_mysql_print_thd(
+ f, trx->mysql_thd, static_cast<uint>(max_query_len));
+ }
+}
+#endif /* WITH_WSREP */
/**********************************************************************//**
Prints info about a transaction.
Acquires and releases lock_sys->mutex and trx_sys->mutex. */
--
cgit v1.2.1
From 53c6195eede2b9959331bf11fa7679a108896bde Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Jan=20Lindstr=C3=B6m?= <jan.lindstrom@mariadb.com>
Date: Mon, 20 Mar 2017 10:17:13 +0200
Subject: Fixed test failure on galere_wsrep_log_conflicts on XtraDB.
Problem was that trx_sys->mutex was acquired to print trx info
even when we already hold trx_sys->mutex. Fixed similarly as
in InnoDB, i.e. with wsrep_trx_print_locking() function that
does not acquire trx_sys->mutex.
---
storage/innobase/include/trx0trx.h | 1 +
storage/innobase/lock/lock0lock.cc | 10 ++++++----
storage/innobase/trx/trx0trx.cc | 1 +
3 files changed, 8 insertions(+), 4 deletions(-)
(limited to 'storage/innobase')
diff --git a/storage/innobase/include/trx0trx.h b/storage/innobase/include/trx0trx.h
index c7e0dc5a551..6db87efaf91 100644
--- a/storage/innobase/include/trx0trx.h
+++ b/storage/innobase/include/trx0trx.h
@@ -1,6 +1,7 @@
/*****************************************************************************
Copyright (c) 1996, 2016, Oracle and/or its affiliates. All Rights Reserved.
+Copyright (c) 2015, 2017, MariaDB Corporation.
This program is free software; you can redistribute it and/or modify it under
the terms of the GNU General Public License as published by the Free Software
diff --git a/storage/innobase/lock/lock0lock.cc b/storage/innobase/lock/lock0lock.cc
index c1f5d71e1b6..b927efaa286 100644
--- a/storage/innobase/lock/lock0lock.cc
+++ b/storage/innobase/lock/lock0lock.cc
@@ -1,7 +1,7 @@
/*****************************************************************************
Copyright (c) 1996, 2016, Oracle and/or its affiliates. All Rights Reserved.
-Copyright (c) 2014, 2016, MariaDB Corporation
+Copyright (c) 2014, 2017, MariaDB Corporation
This program is free software; you can redistribute it and/or modify it under
the terms of the GNU General Public License as published by the Free Software
@@ -1704,13 +1704,14 @@ wsrep_kill_victim(
is in the queue*/
} else if (lock->trx != trx) {
if (wsrep_log_conflicts) {
- if (bf_this)
- fputs("\n*** Priority TRANSACTION:\n",
+ if (bf_this) {
+ fputs("\n*** Priority TRANSACTION:\n",
stderr);
- else {
+ } else {
fputs("\n*** Victim TRANSACTION:\n",
stderr);
}
+
wsrep_trx_print_locking(stderr, trx, 3000);
if (bf_other) {
@@ -1720,6 +1721,7 @@ wsrep_kill_victim(
fputs("\n*** Victim TRANSACTION:\n",
stderr);
}
+
wsrep_trx_print_locking(stderr, lock->trx, 3000);
fputs("*** WAITING FOR THIS LOCK TO BE GRANTED:\n",
diff --git a/storage/innobase/trx/trx0trx.cc b/storage/innobase/trx/trx0trx.cc
index bfc1c546510..5fa176d99e1 100644
--- a/storage/innobase/trx/trx0trx.cc
+++ b/storage/innobase/trx/trx0trx.cc
@@ -1,6 +1,7 @@
/*****************************************************************************
Copyright (c) 1996, 2016, Oracle and/or its affiliates. All Rights Reserved.
+Copyright (c) 2015, 2017, MariaDB Corporation.
This program is free software; you can redistribute it and/or modify it under
the terms of the GNU General Public License as published by the Free Software
--
cgit v1.2.1
From 2cb94aa1b70fa01d9e5c1d8a7625415a0220a3d9 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Marko=20M=C3=A4kel=C3=A4?= <marko.makela@mariadb.com>
Date: Mon, 29 May 2017 13:07:23 +0300
Subject: MDEV-11626 innodb.innodb-change-buffer-recovery fails for xtradb
buf_page_get_gen(): Remove the error log messages about
page flushing and eviction when
innodb_change_buffering_debug=1 is in effect.
---
storage/innobase/buf/buf0buf.c | 6 ------
1 file changed, 6 deletions(-)
(limited to 'storage/innobase')
diff --git a/storage/innobase/buf/buf0buf.c b/storage/innobase/buf/buf0buf.c
index a5e1c045b7b..80a15ffe0d5 100644
--- a/storage/innobase/buf/buf0buf.c
+++ b/storage/innobase/buf/buf0buf.c
@@ -2560,14 +2560,8 @@ wait_until_unfixed:
}
}
buf_pool_mutex_exit(buf_pool);
- fprintf(stderr,
- "innodb_change_buffering_debug evict %u %u\n",
- (unsigned) space, (unsigned) offset);
return(NULL);
} else if (buf_flush_page_try(buf_pool, block)) {
- fprintf(stderr,
- "innodb_change_buffering_debug flush %u %u\n",
- (unsigned) space, (unsigned) offset);
guess = block;
goto loop;
}
--
cgit v1.2.1
From 1af8bf39ca2513bfdf43a55af0b6af10d32dcebb Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Jan=20Lindstr=C3=B6m?= <jan.lindstrom@mariadb.com>
Date: Tue, 30 May 2017 11:30:43 +0300
Subject: MDEV-12113: install_db shows corruption for rest encryption with
innodb_data_file_path=ibdata1:3M;
Problem was that FIL_PAGE_FLUSH_LSN_OR_KEY_VERSION field that for
encrypted pages even in system datafiles should contain key_version
except very first page (0:0) is after encryption overwritten with
flush lsn.
Ported WL#7990 Repurpose FIL_PAGE_FLUSH_LSN to 10.1
The field FIL_PAGE_FLUSH_LSN_OR_KEY_VERSION is consulted during
InnoDB startup.
At startup, InnoDB reads the FIL_PAGE_FLUSH_LSN_OR_KEY_VERSION
from the first page of each file in the InnoDB system tablespace.
If there are multiple files, the minimum and maximum LSN can differ.
These numbers are passed to InnoDB startup.
Having the number in other files than the first file of the InnoDB
system tablespace is not providing much additional value. It is
conflicting with other use of the field, such as on InnoDB R-tree
index pages and encryption key_version.
This worklog will stop writing FIL_PAGE_FLUSH_LSN_OR_KEY_VERSION to
other files than the first file of the InnoDB system tablespace
(page number 0:0) when system tablespace is encrypted. If tablespace
is not encrypted we continue writing FIL_PAGE_FLUSH_LSN_OR_KEY_VERSION
to all first pages of system tablespace to avoid unnecessary
warnings on downgrade.
open_or_create_data_files(): pass only one flushed_lsn parameter
xb_load_tablespaces(): pass only one flushed_lsn parameter.
buf_page_create(): Improve comment about where
FIL_PAGE_FIL_FLUSH_LSN_OR_KEY_VERSION is set.
fil_write_flushed_lsn(): A new function, merged from
fil_write_lsn_and_arch_no_to_file() and
fil_write_flushed_lsn_to_data_files().
Only write to the first page of the system tablespace (page 0:0)
if tablespace is encrypted, or write all first pages of system
tablespace and invoke fil_flush_file_spaces(FIL_TYPE_TABLESPACE)
afterwards.
fil_read_first_page(): read flush_lsn and crypt_data only from
first datafile.
fil_open_single_table_tablespace(): Remove output of LSN, because it
was only valid for the system tablespace and the undo tablespaces, not
user tablespaces.
fil_validate_single_table_tablespace(): Remove output of LSN.
checkpoint_now_set(): Use fil_write_flushed_lsn and output
a error if operation fails.
Remove lsn variable from fsp_open_info.
recv_recovery_from_checkpoint_start(): Remove unnecessary second
flush_lsn parameter.
log_empty_and_mark_files_at_shutdown(): Use fil_writte_flushed_lsn
and output error if it fails.
open_or_create_data_files(): Pass only one flushed_lsn variable.
---
storage/innobase/buf/buf0buf.cc | 10 +-
storage/innobase/fil/fil0fil.cc | 269 ++++++++++++++++------------------
storage/innobase/handler/ha_innodb.cc | 11 +-
storage/innobase/include/fil0fil.h | 63 ++++----
storage/innobase/include/log0recv.h | 36 +++--
storage/innobase/log/log0log.cc | 13 +-
storage/innobase/log/log0recv.cc | 37 +++--
storage/innobase/srv/srv0start.cc | 134 ++++++++---------
8 files changed, 279 insertions(+), 294 deletions(-)
(limited to 'storage/innobase')
diff --git a/storage/innobase/buf/buf0buf.cc b/storage/innobase/buf/buf0buf.cc
index 5cc09287350..2cb2673f7fb 100644
--- a/storage/innobase/buf/buf0buf.cc
+++ b/storage/innobase/buf/buf0buf.cc
@@ -4291,11 +4291,11 @@ buf_page_create(
memset(frame + FIL_PAGE_NEXT, 0xff, 4);
mach_write_to_2(frame + FIL_PAGE_TYPE, FIL_PAGE_TYPE_ALLOCATED);
- /* Reset to zero the file flush lsn field in the page; if the first
- page of an ibdata file is 'created' in this function into the buffer
- pool then we lose the original contents of the file flush lsn stamp.
- Then InnoDB could in a crash recovery print a big, false, corruption
- warning if the stamp contains an lsn bigger than the ib_logfile lsn. */
+ /* FIL_PAGE_FILE_FLUSH_LSN_OR_KEY_VERSION is only used on the
+ following pages:
+ (1) The first page of the InnoDB system tablespace (page 0:0)
+ (2) FIL_RTREE_SPLIT_SEQ_NUM on R-tree pages
+ (3) key_version on encrypted pages (not page 0:0) */
memset(frame + FIL_PAGE_FILE_FLUSH_LSN_OR_KEY_VERSION, 0, 8);
diff --git a/storage/innobase/fil/fil0fil.cc b/storage/innobase/fil/fil0fil.cc
index fa606e89828..9d4a464460f 100644
--- a/storage/innobase/fil/fil0fil.cc
+++ b/storage/innobase/fil/fil0fil.cc
@@ -2211,99 +2211,70 @@ fil_set_max_space_id_if_bigger(
mutex_exit(&fil_system->mutex);
}
-/****************************************************************//**
-Writes the flushed lsn and the latest archived log number to the page header
-of the first page of a data file of the system tablespace (space 0),
-which is uncompressed. */
-static MY_ATTRIBUTE((warn_unused_result))
+/** Write the flushed LSN to the page header of the first page in the
+system tablespace.
+@param[in] lsn flushed LSN
+@return DB_SUCCESS or error number */
dberr_t
-fil_write_lsn_and_arch_no_to_file(
-/*==============================*/
- ulint space, /*!< in: space to write to */
- ulint sum_of_sizes, /*!< in: combined size of previous files
- in space, in database pages */
- lsn_t lsn, /*!< in: lsn to write */
- ulint arch_log_no MY_ATTRIBUTE((unused)))
- /*!< in: archived log number to write */
+fil_write_flushed_lsn(
+ lsn_t lsn)
{
byte* buf1;
byte* buf;
dberr_t err;
- buf1 = static_cast<byte*>(mem_alloc(2 * UNIV_PAGE_SIZE));
+ buf1 = static_cast<byte*>(ut_malloc(2 * UNIV_PAGE_SIZE));
buf = static_cast<byte*>(ut_align(buf1, UNIV_PAGE_SIZE));
- err = fil_read(TRUE, space, 0, sum_of_sizes, 0,
- UNIV_PAGE_SIZE, buf, NULL, 0);
- if (err == DB_SUCCESS) {
- mach_write_to_8(buf + FIL_PAGE_FILE_FLUSH_LSN_OR_KEY_VERSION,
- lsn);
-
- err = fil_write(TRUE, space, 0, sum_of_sizes, 0,
- UNIV_PAGE_SIZE, buf, NULL, 0);
- }
-
- mem_free(buf1);
-
- return(err);
-}
+ /* Acquire system tablespace */
+ fil_space_t* space = fil_space_acquire(0);
-/****************************************************************//**
-Writes the flushed lsn and the latest archived log number to the page
-header of the first page of each data file in the system tablespace.
-@return DB_SUCCESS or error number */
-UNIV_INTERN
-dberr_t
-fil_write_flushed_lsn_to_data_files(
-/*================================*/
- lsn_t lsn, /*!< in: lsn to write */
- ulint arch_log_no) /*!< in: latest archived log file number */
-{
- fil_space_t* space;
- fil_node_t* node;
- dberr_t err;
+ /* If tablespace is not encrypted, stamp flush_lsn to
+ first page of all system tablespace datafiles to avoid
+ unnecessary error messages on possible downgrade. */
+ if (space->crypt_data->min_key_version == 0) {
+ fil_node_t* node;
+ ulint sum_of_sizes = 0;
- mutex_enter(&fil_system->mutex);
-
- for (space = UT_LIST_GET_FIRST(fil_system->space_list);
- space != NULL;
- space = UT_LIST_GET_NEXT(space_list, space)) {
-
- /* We only write the lsn to all existing data files which have
- been open during the lifetime of the mysqld process; they are
- represented by the space objects in the tablespace memory
- cache. Note that all data files in the system tablespace 0
- and the UNDO log tablespaces (if separate) are always open. */
-
- if (space->purpose == FIL_TABLESPACE
- && !fil_is_user_tablespace_id(space->id)) {
- ulint sum_of_sizes = 0;
-
- for (node = UT_LIST_GET_FIRST(space->chain);
- node != NULL;
- node = UT_LIST_GET_NEXT(chain, node)) {
-
- mutex_exit(&fil_system->mutex);
+ for (node = UT_LIST_GET_FIRST(space->chain);
+ node != NULL;
+ node = UT_LIST_GET_NEXT(chain, node)) {
- err = fil_write_lsn_and_arch_no_to_file(
- space->id, sum_of_sizes, lsn,
- arch_log_no);
+ err = fil_read(TRUE, 0, 0, sum_of_sizes, 0,
+ UNIV_PAGE_SIZE, buf, NULL, 0);
- if (err != DB_SUCCESS) {
+ if (err == DB_SUCCESS) {
+ mach_write_to_8(buf + FIL_PAGE_FILE_FLUSH_LSN_OR_KEY_VERSION,
+ lsn);
- return(err);
- }
-
- mutex_enter(&fil_system->mutex);
+ err = fil_write(TRUE, 0, 0, sum_of_sizes, 0,
+ UNIV_PAGE_SIZE, buf, NULL, 0);
sum_of_sizes += node->size;
}
}
+ } else {
+ /* When system tablespace is encrypted stamp flush_lsn to
+ only the first page of the first datafile (rest of pages
+ are encrypted). */
+ err = fil_read(TRUE, 0, 0, 0, 0,
+ UNIV_PAGE_SIZE, buf, NULL, 0);
+
+ if (err == DB_SUCCESS) {
+ mach_write_to_8(buf + FIL_PAGE_FILE_FLUSH_LSN_OR_KEY_VERSION,
+ lsn);
+
+ err = fil_write(TRUE, 0, 0, 0, 0,
+ UNIV_PAGE_SIZE, buf, NULL, 0);
+ }
}
- mutex_exit(&fil_system->mutex);
+ fil_flush_file_spaces(FIL_TABLESPACE);
+ fil_space_release(space);
- return(DB_SUCCESS);
+ ut_free(buf1);
+
+ return(err);
}
/** Check the consistency of the first data page of a tablespace
@@ -2356,36 +2327,37 @@ fil_check_first_page(const page_t* page, ulint space_id, ulint flags)
return("inconsistent data in space header");
}
-/*******************************************************************//**
-Reads the flushed lsn, arch no, space_id and tablespace flag fields from
-the first page of a data file at database startup.
+/** Reads the flushed lsn, arch no, space_id and tablespace flag fields from
+the first page of a first data file at database startup.
+@param[in] data_file open data file
+@param[in] one_read_only true if first datafile is already
+ read
+@param[out] flags FSP_SPACE_FLAGS
+@param[out] space_id tablepspace ID
+@param[out] min_arch_log_no min of archived log numbers in
+ data files
+@param[out] max_arch_log_no max of archived log numbers in
+ data files
+@param[out] flushed_lsn flushed lsn value
+@param[out] crypt_data encryption crypt data
@retval NULL on success, or if innodb_force_recovery is set
@return pointer to an error message string */
UNIV_INTERN
const char*
fil_read_first_page(
-/*================*/
- pfs_os_file_t data_file, /*!< in: open data file */
- ibool one_read_already, /*!< in: TRUE if min and max
- parameters below already
- contain sensible data */
- ulint* flags, /*!< out: FSP_SPACE_FLAGS */
- ulint* space_id, /*!< out: tablespace ID */
+ pfs_os_file_t data_file,
+ ibool one_read_already,
+ ulint* flags,
+ ulint* space_id,
#ifdef UNIV_LOG_ARCHIVE
- ulint* min_arch_log_no, /*!< out: min of archived
- log numbers in data files */
- ulint* max_arch_log_no, /*!< out: max of archived
- log numbers in data files */
+ ulint* min_arch_log_no,
+ ulint* max_arch_log_no,
#endif /* UNIV_LOG_ARCHIVE */
- lsn_t* min_flushed_lsn, /*!< out: min of flushed
- lsn values in data files */
- lsn_t* max_flushed_lsn, /*!< out: max of flushed
- lsn values in data files */
- fil_space_crypt_t** crypt_data) /*< out: crypt data */
+ lsn_t* flushed_lsn,
+ fil_space_crypt_t** crypt_data)
{
byte* buf;
byte* page;
- lsn_t flushed_lsn;
const char* check_msg = NULL;
fil_space_crypt_t* cdata;
@@ -2407,6 +2379,11 @@ fil_read_first_page(
*space_id = fsp_header_get_space_id(page);
*flags = fsp_header_get_flags(page);
+ if (flushed_lsn) {
+ *flushed_lsn = mach_read_from_8(page +
+ FIL_PAGE_FILE_FLUSH_LSN_OR_KEY_VERSION);
+ }
+
if (!fsp_flags_is_valid(*flags)) {
ulint cflags = fsp_flags_convert_from_101(*flags);
if (cflags == ULINT_UNDEFINED) {
@@ -2420,33 +2397,33 @@ fil_read_first_page(
}
check_msg = fil_check_first_page(page, *space_id, *flags);
- }
-
- flushed_lsn = mach_read_from_8(page +
- FIL_PAGE_FILE_FLUSH_LSN_OR_KEY_VERSION);
- ulint space = fsp_header_get_space_id(page);
- ulint offset = fsp_header_get_crypt_offset(
- fsp_flags_get_zip_size(*flags));
+ /* Possible encryption crypt data is also stored only to first page
+ of the first datafile. */
- cdata = fil_space_read_crypt_data(space, page, offset);
+ ulint offset = fsp_header_get_crypt_offset(
+ fsp_flags_get_zip_size(*flags));
- if (crypt_data) {
- *crypt_data = cdata;
- }
+ cdata = fil_space_read_crypt_data(*space_id, page, offset);
- /* If file space is encrypted we need to have at least some
- encryption service available where to get keys */
- if (cdata && cdata->should_encrypt()) {
+ if (crypt_data) {
+ *crypt_data = cdata;
+ }
- if (!encryption_key_id_exists(cdata->key_id)) {
- ib_logf(IB_LOG_LEVEL_ERROR,
- "Tablespace id %ld is encrypted but encryption service"
- " or used key_id %u is not available. Can't continue opening tablespace.",
- space, cdata->key_id);
+ /* If file space is encrypted we need to have at least some
+ encryption service available where to get keys */
+ if (cdata && cdata->should_encrypt()) {
- return ("table encrypted but encryption service not available.");
+ if (!encryption_key_id_exists(cdata->key_id)) {
+ ib_logf(IB_LOG_LEVEL_ERROR,
+ "Tablespace id " ULINTPF
+ " is encrypted but encryption service"
+ " or used key_id %u is not available. "
+ "Can't continue opening tablespace.",
+ *space_id, cdata->key_id);
+ return ("table encrypted but encryption service not available.");
+ }
}
}
@@ -2457,8 +2434,6 @@ fil_read_first_page(
}
if (!one_read_already) {
- *min_flushed_lsn = flushed_lsn;
- *max_flushed_lsn = flushed_lsn;
#ifdef UNIV_LOG_ARCHIVE
*min_arch_log_no = arch_log_no;
*max_arch_log_no = arch_log_no;
@@ -2466,16 +2441,11 @@ fil_read_first_page(
return(NULL);
}
- if (*min_flushed_lsn > flushed_lsn) {
- *min_flushed_lsn = flushed_lsn;
- }
- if (*max_flushed_lsn < flushed_lsn) {
- *max_flushed_lsn = flushed_lsn;
- }
#ifdef UNIV_LOG_ARCHIVE
if (*min_arch_log_no > arch_log_no) {
*min_arch_log_no = arch_log_no;
}
+
if (*max_arch_log_no < arch_log_no) {
*max_arch_log_no = arch_log_no;
}
@@ -4173,6 +4143,7 @@ fil_open_single_table_tablespace(
def.file = os_file_create_simple_no_error_handling(
innodb_file_data_key, def.filepath, OS_FILE_OPEN,
OS_FILE_READ_ONLY, &def.success, atomic_writes);
+
if (def.success) {
tablespaces_found++;
}
@@ -4187,11 +4158,11 @@ fil_open_single_table_tablespace(
/* Read the first page of the datadir tablespace, if found. */
if (def.success) {
def.check_msg = fil_read_first_page(
- def.file, FALSE, &def.flags, &def.id,
+ def.file, false, &def.flags, &def.id,
#ifdef UNIV_LOG_ARCHIVE
&space_arch_log_no, &space_arch_log_no,
#endif /* UNIV_LOG_ARCHIVE */
- &def.lsn, &def.lsn, &def.crypt_data);
+ NULL, &def.crypt_data);
if (table) {
table->crypt_data = def.crypt_data;
@@ -4200,6 +4171,7 @@ fil_open_single_table_tablespace(
def.valid = !def.check_msg && def.id == id
&& fsp_flags_match(flags, def.flags);
+
if (def.valid) {
valid_tablespaces_found++;
} else {
@@ -4213,11 +4185,11 @@ fil_open_single_table_tablespace(
/* Read the first page of the remote tablespace */
if (remote.success) {
remote.check_msg = fil_read_first_page(
- remote.file, FALSE, &remote.flags, &remote.id,
+ remote.file, false, &remote.flags, &remote.id,
#ifdef UNIV_LOG_ARCHIVE
&remote.arch_log_no, &remote.arch_log_no,
#endif /* UNIV_LOG_ARCHIVE */
- &remote.lsn, &remote.lsn, &remote.crypt_data);
+ NULL, &remote.crypt_data);
if (table) {
table->crypt_data = remote.crypt_data;
@@ -4227,6 +4199,7 @@ fil_open_single_table_tablespace(
/* Validate this single-table-tablespace with SYS_TABLES. */
remote.valid = !remote.check_msg && remote.id == id
&& fsp_flags_match(flags, remote.flags);
+
if (remote.valid) {
valid_tablespaces_found++;
} else {
@@ -4241,11 +4214,11 @@ fil_open_single_table_tablespace(
/* Read the first page of the datadir tablespace, if found. */
if (dict.success) {
dict.check_msg = fil_read_first_page(
- dict.file, FALSE, &dict.flags, &dict.id,
+ dict.file, false, &dict.flags, &dict.id,
#ifdef UNIV_LOG_ARCHIVE
&dict.arch_log_no, &dict.arch_log_no,
#endif /* UNIV_LOG_ARCHIVE */
- &dict.lsn, &dict.lsn, &dict.crypt_data);
+ NULL, &dict.crypt_data);
if (table) {
table->crypt_data = dict.crypt_data;
@@ -4289,26 +4262,32 @@ fil_open_single_table_tablespace(
ib_logf(IB_LOG_LEVEL_ERROR,
"A tablespace for %s has been found in "
"multiple places;", tablename);
+
if (def.success) {
ib_logf(IB_LOG_LEVEL_ERROR,
- "Default location; %s, LSN=" LSN_PF
- ", Space ID=%lu, Flags=%lu",
- def.filepath, def.lsn,
- (ulong) def.id, (ulong) def.flags);
+ "Default location; %s"
+ ", Space ID=" ULINTPF " , Flags=" ULINTPF " .",
+ def.filepath,
+ def.id,
+ def.flags);
}
+
if (remote.success) {
ib_logf(IB_LOG_LEVEL_ERROR,
- "Remote location; %s, LSN=" LSN_PF
- ", Space ID=%lu, Flags=%lu",
- remote.filepath, remote.lsn,
- (ulong) remote.id, (ulong) remote.flags);
+ "Remote location; %s"
+ ", Space ID=" ULINTPF " , Flags=" ULINTPF " .",
+ remote.filepath,
+ remote.id,
+ remote.flags);
}
+
if (dict.success) {
ib_logf(IB_LOG_LEVEL_ERROR,
- "Dictionary location; %s, LSN=" LSN_PF
- ", Space ID=%lu, Flags=%lu",
- dict.filepath, dict.lsn,
- (ulong) dict.id, (ulong) dict.flags);
+ "Dictionary location; %s"
+ ", Space ID=" ULINTPF " , Flags=" ULINTPF " .",
+ dict.filepath,
+ dict.id,
+ dict.flags);
}
/* Force-recovery will allow some tablespaces to be
@@ -4341,6 +4320,7 @@ fil_open_single_table_tablespace(
os_file_close(def.file);
tablespaces_found--;
}
+
if (dict.success && !dict.valid) {
dict.success = false;
os_file_close(dict.file);
@@ -4348,6 +4328,7 @@ fil_open_single_table_tablespace(
can be corrected below. */
tablespaces_found--;
}
+
if (remote.success && !remote.valid) {
remote.success = false;
os_file_close(remote.file);
@@ -4698,7 +4679,7 @@ check_first_page:
#ifdef UNIV_LOG_ARCHIVE
&fsp->arch_log_no, &fsp->arch_log_no,
#endif /* UNIV_LOG_ARCHIVE */
- &fsp->lsn, &fsp->lsn, &fsp->crypt_data)) {
+ NULL, &fsp->crypt_data)) {
ib_logf(IB_LOG_LEVEL_ERROR,
"%s in tablespace %s (table %s)",
check_msg, fsp->filepath, tablename);
@@ -4928,11 +4909,11 @@ will_not_choose:
if (def.success && remote.success) {
ib_logf(IB_LOG_LEVEL_ERROR,
"Tablespaces for %s have been found in two places;\n"
- "Location 1: SpaceID: %lu LSN: %lu File: %s\n"
- "Location 2: SpaceID: %lu LSN: %lu File: %s\n"
+ "Location 1: SpaceID: " ULINTPF " File: %s\n"
+ "Location 2: SpaceID: " ULINTPF " File: %s\n"
"You must delete one of them.",
- tablename, (ulong) def.id, (ulong) def.lsn,
- def.filepath, (ulong) remote.id, (ulong) remote.lsn,
+ tablename, def.id,
+ def.filepath, remote.id,
remote.filepath);
def.success = FALSE;
diff --git a/storage/innobase/handler/ha_innodb.cc b/storage/innobase/handler/ha_innodb.cc
index 012d81380f1..012539d1ace 100644
--- a/storage/innobase/handler/ha_innodb.cc
+++ b/storage/innobase/handler/ha_innodb.cc
@@ -18221,8 +18221,15 @@ checkpoint_now_set(
log_make_checkpoint_at(LSN_MAX, TRUE);
fil_flush_file_spaces(FIL_LOG);
}
- fil_write_flushed_lsn_to_data_files(log_sys->lsn, 0);
- fil_flush_file_spaces(FIL_TABLESPACE);
+
+ dberr_t err = fil_write_flushed_lsn(log_sys->lsn);
+
+ if (err != DB_SUCCESS) {
+ ib_logf(IB_LOG_LEVEL_ERROR,
+ "Failed to write flush lsn to the "
+ "system tablespace at checkpoint err=%s",
+ ut_strerr(err));
+ }
}
}
diff --git a/storage/innobase/include/fil0fil.h b/storage/innobase/include/fil0fil.h
index d546bab6862..ba440cb2a1c 100644
--- a/storage/innobase/include/fil0fil.h
+++ b/storage/innobase/include/fil0fil.h
@@ -215,7 +215,6 @@ struct fsp_open_info {
ibool valid; /*!< Is the tablespace valid? */
pfs_os_file_t file; /*!< File handle */
char* filepath; /*!< File path to open */
- lsn_t lsn; /*!< Flushed LSN from header page */
ulint id; /*!< Space ID */
ulint flags; /*!< Tablespace flags */
ulint encryption_error; /*!< if an encryption error occurs */
@@ -637,17 +636,17 @@ void
fil_set_max_space_id_if_bigger(
/*===========================*/
ulint max_id);/*!< in: maximum known id */
+
#ifndef UNIV_HOTBACKUP
-/****************************************************************//**
-Writes the flushed lsn and the latest archived log number to the page
-header of the first page of each data file in the system tablespace.
-@return DB_SUCCESS or error number */
-UNIV_INTERN
+
+/** Write the flushed LSN to the page header of the first page in the
+system tablespace.
+@param[in] lsn flushed LSN
+@return DB_SUCCESS or error number */
dberr_t
-fil_write_flushed_lsn_to_data_files(
-/*================================*/
- lsn_t lsn, /*!< in: lsn to write */
- ulint arch_log_no); /*!< in: latest archived log file number */
+fil_write_flushed_lsn(
+ lsn_t lsn)
+ MY_ATTRIBUTE((warn_unused_result));
/** Acquire a tablespace when it could be dropped concurrently.
Used by background threads that do not necessarily hold proper locks
@@ -793,35 +792,37 @@ private:
fil_space_t* m_space;
};
-/*******************************************************************//**
-Reads the flushed lsn, arch no, and tablespace flag fields from a data
-file at database startup.
+/** Reads the flushed lsn, arch no, space_id and tablespace flag fields from
+the first page of a first data file at database startup.
+@param[in] data_file open data file
+@param[in] one_read_only true if first datafile is already
+ read
+@param[out] flags FSP_SPACE_FLAGS
+@param[out] space_id tablepspace ID
+@param[out] min_arch_log_no min of archived log numbers in
+ data files
+@param[out] max_arch_log_no max of archived log numbers in
+ data files
+@param[out] flushed_lsn flushed lsn value
+@param[out] crypt_data encryption crypt data
@retval NULL on success, or if innodb_force_recovery is set
@return pointer to an error message string */
UNIV_INTERN
const char*
fil_read_first_page(
-/*================*/
- pfs_os_file_t data_file, /*!< in: open data file */
- ibool one_read_already, /*!< in: TRUE if min and max
- parameters below already
- contain sensible data */
- ulint* flags, /*!< out: FSP_SPACE_FLAGS */
- ulint* space_id, /*!< out: tablespace ID */
+ pfs_os_file_t data_file,
+ ibool one_read_already,
+ ulint* flags,
+ ulint* space_id,
#ifdef UNIV_LOG_ARCHIVE
- ulint* min_arch_log_no, /*!< out: min of archived
- log numbers in data files */
- ulint* max_arch_log_no, /*!< out: max of archived
- log numbers in data files */
+ ulint* min_arch_log_no,
+ ulint* max_arch_log_no,
#endif /* UNIV_LOG_ARCHIVE */
- lsn_t* min_flushed_lsn, /*!< out: min of flushed
- lsn values in data files */
- lsn_t* max_flushed_lsn, /*!< out: max of flushed
- lsn values in data files */
- fil_space_crypt_t** crypt_data) /*!< out: crypt data */
-
- __attribute__((warn_unused_result));
+ lsn_t* flushed_lsn,
+ fil_space_crypt_t** crypt_data)
+ MY_ATTRIBUTE((warn_unused_result));
#endif /* !UNIV_HOTBACKUP */
+
/*******************************************************************//**
Parses the body of a log record written about an .ibd file operation. That is,
the log record part after the standard (type, space id, page no) header of the
diff --git a/storage/innobase/include/log0recv.h b/storage/innobase/include/log0recv.h
index 9ca1c46d72b..dca99f2a7a5 100644
--- a/storage/innobase/include/log0recv.h
+++ b/storage/innobase/include/log0recv.h
@@ -123,26 +123,25 @@ a freshly read page)
*/
# define recv_recover_page(jri, block) recv_recover_page_func(block)
#endif /* !UNIV_HOTBACKUP */
-/********************************************************//**
-Recovers from a checkpoint. When this function returns, the database is able
+
+/** Recovers from a checkpoint. When this function returns, the database is able
to start processing of new user transactions, but the function
recv_recovery_from_checkpoint_finish should be called later to complete
the recovery and free the resources used in it.
+@param[in] type LOG_CHECKPOINT or LOG_ARCHIVE
+@param[in] limit_lsn recover up to this lsn if possible
+@param[in] flushed_lsn flushed lsn from first data file
@return error code or DB_SUCCESS */
UNIV_INTERN
dberr_t
recv_recovery_from_checkpoint_start_func(
-/*=====================================*/
#ifdef UNIV_LOG_ARCHIVE
- ulint type, /*!< in: LOG_CHECKPOINT or
- LOG_ARCHIVE */
- lsn_t limit_lsn, /*!< in: recover up to this lsn
- if possible */
+ ulint type,
+ lsn_t limit_lsn,
#endif /* UNIV_LOG_ARCHIVE */
- lsn_t min_flushed_lsn,/*!< in: min flushed lsn from
- data files */
- lsn_t max_flushed_lsn);/*!< in: max flushed lsn from
- data files */
+ lsn_t flushed_lsn)
+ MY_ATTRIBUTE((warn_unused_result));
+
#ifdef UNIV_LOG_ARCHIVE
/** Wrapper for recv_recovery_from_checkpoint_start_func().
Recovers from a checkpoint. When this function returns, the database is able
@@ -151,11 +150,10 @@ recv_recovery_from_checkpoint_finish should be called later to complete
the recovery and free the resources used in it.
@param type in: LOG_CHECKPOINT or LOG_ARCHIVE
@param lim in: recover up to this log sequence number if possible
-@param min in: minimum flushed log sequence number from data files
-@param max in: maximum flushed log sequence number from data files
+@param lsn in: flushed log sequence number from first data file
@return error code or DB_SUCCESS */
-# define recv_recovery_from_checkpoint_start(type,lim,min,max) \
- recv_recovery_from_checkpoint_start_func(type,lim,min,max)
+# define recv_recovery_from_checkpoint_start(type,lim,lsn) \
+ recv_recovery_from_checkpoint_start_func(type,lim,lsn)
#else /* UNIV_LOG_ARCHIVE */
/** Wrapper for recv_recovery_from_checkpoint_start_func().
Recovers from a checkpoint. When this function returns, the database is able
@@ -164,12 +162,12 @@ recv_recovery_from_checkpoint_finish should be called later to complete
the recovery and free the resources used in it.
@param type ignored: LOG_CHECKPOINT or LOG_ARCHIVE
@param lim ignored: recover up to this log sequence number if possible
-@param min in: minimum flushed log sequence number from data files
-@param max in: maximum flushed log sequence number from data files
+@param lsn in: flushed log sequence number from first data file
@return error code or DB_SUCCESS */
-# define recv_recovery_from_checkpoint_start(type,lim,min,max) \
- recv_recovery_from_checkpoint_start_func(min,max)
+# define recv_recovery_from_checkpoint_start(type,lim,lsn) \
+ recv_recovery_from_checkpoint_start_func(lsn)
#endif /* UNIV_LOG_ARCHIVE */
+
/********************************************************//**
Completes recovery from a checkpoint. */
UNIV_INTERN
diff --git a/storage/innobase/log/log0log.cc b/storage/innobase/log/log0log.cc
index 14d1f4544cd..0dc06fd4b1b 100644
--- a/storage/innobase/log/log0log.cc
+++ b/storage/innobase/log/log0log.cc
@@ -3246,7 +3246,9 @@ logs_empty_and_mark_files_at_shutdown(void)
/*=======================================*/
{
lsn_t lsn;
+#ifdef UNIV_LOG_ARCHIVE
ulint arch_log_no;
+#endif
ulint count = 0;
ulint pending_io;
ibool server_busy;
@@ -3454,9 +3456,9 @@ wait_suspend_loop:
goto loop;
}
+#ifdef UNIV_LOG_ARCHIVE
arch_log_no = 0;
-#ifdef UNIV_LOG_ARCHIVE
UT_LIST_GET_FIRST(log_sys->log_groups)->archived_file_no;
if (!UT_LIST_GET_FIRST(log_sys->log_groups)->archived_offset) {
@@ -3511,9 +3513,14 @@ wait_suspend_loop:
srv_shutdown_lsn = lsn;
if (!srv_read_only_mode) {
- fil_write_flushed_lsn_to_data_files(lsn, arch_log_no);
+ dberr_t err = fil_write_flushed_lsn(lsn);
- fil_flush_file_spaces(FIL_TABLESPACE);
+ if (err != DB_SUCCESS) {
+ ib_logf(IB_LOG_LEVEL_ERROR,
+ "Failed to write flush lsn to the "
+ "system tablespace at shutdown err=%s",
+ ut_strerr(err));
+ }
}
fil_close_all_files();
diff --git a/storage/innobase/log/log0recv.cc b/storage/innobase/log/log0recv.cc
index 2dc2d419e3f..bf00cd8e8d9 100644
--- a/storage/innobase/log/log0recv.cc
+++ b/storage/innobase/log/log0recv.cc
@@ -2929,22 +2929,22 @@ recv_init_crash_recovery(void)
}
}
-/********************************************************//**
-Recovers from a checkpoint. When this function returns, the database is able
+/** Recovers from a checkpoint. When this function returns, the database is able
to start processing of new user transactions, but the function
recv_recovery_from_checkpoint_finish should be called later to complete
the recovery and free the resources used in it.
+@param[in] type LOG_CHECKPOINT or LOG_ARCHIVE
+@param[in] limit_lsn recover up to this lsn if possible
+@param[in] flushed_lsn flushed lsn from first data file
@return error code or DB_SUCCESS */
UNIV_INTERN
dberr_t
recv_recovery_from_checkpoint_start_func(
-/*=====================================*/
#ifdef UNIV_LOG_ARCHIVE
- ulint type, /*!< in: LOG_CHECKPOINT or LOG_ARCHIVE */
- lsn_t limit_lsn, /*!< in: recover up to this lsn if possible */
+ ulint type,
+ lsn_t limit_lsn,
#endif /* UNIV_LOG_ARCHIVE */
- lsn_t min_flushed_lsn,/*!< in: min flushed lsn from data files */
- lsn_t max_flushed_lsn)/*!< in: max flushed lsn from data files */
+ lsn_t flushed_lsn)
{
log_group_t* group;
log_group_t* max_cp_group;
@@ -3165,6 +3165,7 @@ recv_recovery_from_checkpoint_start_func(
group = UT_LIST_GET_NEXT(log_groups, group);
}
+
/* Done with startup scan. Clear the flag. */
recv_log_scan_is_startup_type = FALSE;
@@ -3177,10 +3178,9 @@ recv_recovery_from_checkpoint_start_func(
there is something wrong we will print a message to the
user about recovery: */
- if (checkpoint_lsn != max_flushed_lsn
- || checkpoint_lsn != min_flushed_lsn) {
+ if (checkpoint_lsn != flushed_lsn) {
- if (checkpoint_lsn < max_flushed_lsn) {
+ if (checkpoint_lsn < flushed_lsn) {
ib_logf(IB_LOG_LEVEL_WARN,
"The log sequence number "
@@ -3191,24 +3191,21 @@ recv_recovery_from_checkpoint_start_func(
"ib_logfiles to start up the database. "
"Log sequence number in the "
"ib_logfiles is " LSN_PF ", log"
- "sequence numbers stamped "
- "to ibdata file headers are between "
- "" LSN_PF " and " LSN_PF ".",
+ "sequence number stamped "
+ "to ibdata file header is " LSN_PF ".",
checkpoint_lsn,
- min_flushed_lsn,
- max_flushed_lsn);
+ flushed_lsn);
}
if (!recv_needed_recovery) {
ib_logf(IB_LOG_LEVEL_INFO,
- "The log sequence numbers "
- LSN_PF " and " LSN_PF
- " in ibdata files do not match"
+ "The log sequence number "
+ LSN_PF
+ " in ibdata file do not match"
" the log sequence number "
LSN_PF
" in the ib_logfiles!",
- min_flushed_lsn,
- max_flushed_lsn,
+ flushed_lsn,
checkpoint_lsn);
if (!srv_read_only_mode) {
diff --git a/storage/innobase/srv/srv0start.cc b/storage/innobase/srv/srv0start.cc
index be65e4a6d17..f601ff17ebc 100644
--- a/storage/innobase/srv/srv0start.cc
+++ b/storage/innobase/srv/srv0start.cc
@@ -786,27 +786,26 @@ open_log_file(
return(DB_SUCCESS);
}
-/*********************************************************************//**
-Creates or opens database data files and closes them.
+/** Creates or opens database data files and closes them.
+@param[out] create_new_db true = create new database
+@param[out] min_arch_log_no min of archived log numbers in
+ data files
+@param[out] max_arch_log_no max of archived log numbers in
+ data files
+@param[out] flushed_lsn flushed lsn in fist datafile
+@param[out] sum_of_new_sizes sum of sizes of the new files
+ added
@return DB_SUCCESS or error code */
static MY_ATTRIBUTE((nonnull, warn_unused_result))
dberr_t
open_or_create_data_files(
-/*======================*/
- ibool* create_new_db, /*!< out: TRUE if new database should be
- created */
+ bool* create_new_db,
#ifdef UNIV_LOG_ARCHIVE
- ulint* min_arch_log_no,/*!< out: min of archived log
- numbers in data files */
- ulint* max_arch_log_no,/*!< out: max of archived log
- numbers in data files */
+ ulint* min_arch_log_no,
+ ulint* max_arch_log_no,
#endif /* UNIV_LOG_ARCHIVE */
- lsn_t* min_flushed_lsn,/*!< out: min of flushed lsn
- values in data files */
- lsn_t* max_flushed_lsn,/*!< out: max of flushed lsn
- values in data files */
- ulint* sum_of_new_sizes)/*!< out: sum of sizes of the
- new files added */
+ lsn_t* flushed_lsn,
+ ulint* sum_of_new_sizes)
{
ibool ret;
ulint i;
@@ -830,7 +829,7 @@ open_or_create_data_files(
*sum_of_new_sizes = 0;
- *create_new_db = FALSE;
+ *create_new_db = false;
srv_normalize_path_for_win(srv_data_home);
@@ -918,12 +917,13 @@ open_or_create_data_files(
}
const char* check_msg;
+
check_msg = fil_read_first_page(
files[i], FALSE, &flags, &space,
#ifdef UNIV_LOG_ARCHIVE
min_arch_log_no, max_arch_log_no,
#endif /* UNIV_LOG_ARCHIVE */
- min_flushed_lsn, max_flushed_lsn, NULL);
+ flushed_lsn, NULL);
/* If first page is valid, don't overwrite DB.
It prevents overwriting DB when mysql_install_db
@@ -954,6 +954,7 @@ open_or_create_data_files(
name);
return(DB_ERROR);
}
+
if (srv_data_file_is_raw_partition[i] == SRV_OLD_RAW) {
ut_a(!srv_read_only_mode);
files[i] = os_file_create(
@@ -973,7 +974,6 @@ open_or_create_data_files(
}
if (!ret) {
-
os_file_get_last_error(true);
ib_logf(IB_LOG_LEVEL_ERROR,
@@ -983,7 +983,6 @@ open_or_create_data_files(
}
if (srv_data_file_is_raw_partition[i] == SRV_OLD_RAW) {
-
goto skip_size_check;
}
@@ -1010,16 +1009,15 @@ size_check:
"auto-extending "
"data file %s is "
"of a different size "
- "%lu pages (rounded "
+ ULINTPF " pages (rounded "
"down to MB) than specified "
"in the .cnf file: "
- "initial %lu pages, "
- "max %lu (relevant if "
+ "initial " ULINTPF " pages, "
+ "max " ULINTPF " (relevant if "
"non-zero) pages!",
name,
- (ulong) rounded_size_pages,
- (ulong) srv_data_file_sizes[i],
- (ulong)
+ rounded_size_pages,
+ srv_data_file_sizes[i],
srv_last_file_size_max);
return(DB_ERROR);
@@ -1032,12 +1030,12 @@ size_check:
ib_logf(IB_LOG_LEVEL_ERROR,
"Data file %s is of a different "
- "size %lu pages (rounded down to MB) "
+ "size " ULINTPF " pages (rounded down to MB) "
"than specified in the .cnf file "
- "%lu pages!",
+ ULINTPF " pages!",
name,
- (ulong) rounded_size_pages,
- (ulong) srv_data_file_sizes[i]);
+ rounded_size_pages,
+ srv_data_file_sizes[i]);
return(DB_ERROR);
}
@@ -1057,7 +1055,7 @@ check_first_page:
#ifdef UNIV_LOG_ARCHIVE
min_arch_log_no, max_arch_log_no,
#endif /* UNIV_LOG_ARCHIVE */
- min_flushed_lsn, max_flushed_lsn, &crypt_data);
+ flushed_lsn, &crypt_data);
if (check_msg) {
@@ -1094,9 +1092,9 @@ check_first_page:
!= fsp_flags_get_page_size(flags)) {
ib_logf(IB_LOG_LEVEL_ERROR,
- "Data file \"%s\" uses page size %lu,"
+ "Data file \"%s\" uses page size " ULINTPF " ,"
"but the start-up parameter "
- "is --innodb-page-size=%lu",
+ "is --innodb-page-size=" ULINTPF " .",
name,
fsp_flags_get_page_size(flags),
UNIV_PAGE_SIZE);
@@ -1127,9 +1125,9 @@ check_first_page:
}
ib_logf(IB_LOG_LEVEL_INFO,
- "Setting file %s size to %lu MB",
+ "Setting file %s size to " ULINTPF " MB",
name,
- (ulong) (srv_data_file_sizes[i]
+ (srv_data_file_sizes[i]
>> (20 - UNIV_PAGE_SIZE_SHIFT)));
ret = os_file_set_size(
@@ -1592,9 +1590,8 @@ dberr_t
innobase_start_or_create_for_mysql(void)
/*====================================*/
{
- ibool create_new_db;
- lsn_t min_flushed_lsn;
- lsn_t max_flushed_lsn;
+ bool create_new_db;
+ lsn_t flushed_lsn;
#ifdef UNIV_LOG_ARCHIVE
ulint min_arch_log_no;
ulint max_arch_log_no;
@@ -2152,7 +2149,7 @@ innobase_start_or_create_for_mysql(void)
#ifdef UNIV_LOG_ARCHIVE
&min_arch_log_no, &max_arch_log_no,
#endif /* UNIV_LOG_ARCHIVE */
- &min_flushed_lsn, &max_flushed_lsn,
+ &flushed_lsn,
&sum_of_new_sizes);
if (err == DB_FAIL) {
@@ -2197,12 +2194,12 @@ innobase_start_or_create_for_mysql(void)
bool success = buf_flush_list(ULINT_MAX, LSN_MAX, NULL);
ut_a(success);
- min_flushed_lsn = max_flushed_lsn = log_get_lsn();
+ flushed_lsn = log_get_lsn();
buf_flush_wait_batch_end(NULL, BUF_FLUSH_LIST);
err = create_log_files(create_new_db, logfilename, dirnamelen,
- max_flushed_lsn, logfile0);
+ flushed_lsn, logfile0);
if (err != DB_SUCCESS) {
return(err);
@@ -2222,19 +2219,8 @@ innobase_start_or_create_for_mysql(void)
if (err == DB_NOT_FOUND) {
if (i == 0) {
- if (max_flushed_lsn
- != min_flushed_lsn) {
- ib_logf(IB_LOG_LEVEL_ERROR,
- "Cannot create"
- " log files because"
- " data files are"
- " corrupt or"
- " not in sync"
- " with each other");
- return(DB_ERROR);
- }
- if (max_flushed_lsn < (lsn_t) 1000) {
+ if (flushed_lsn < (lsn_t) 1000) {
ib_logf(IB_LOG_LEVEL_ERROR,
"Cannot create"
" log files because"
@@ -2249,14 +2235,14 @@ innobase_start_or_create_for_mysql(void)
err = create_log_files(
create_new_db, logfilename,
- dirnamelen, max_flushed_lsn,
+ dirnamelen, flushed_lsn,
logfile0);
if (err == DB_SUCCESS) {
err = create_log_files_rename(
logfilename,
dirnamelen,
- max_flushed_lsn,
+ flushed_lsn,
logfile0);
}
@@ -2266,8 +2252,7 @@ innobase_start_or_create_for_mysql(void)
/* Suppress the message about
crash recovery. */
- max_flushed_lsn = min_flushed_lsn
- = log_get_lsn();
+ flushed_lsn = log_get_lsn();
goto files_checked;
} else if (i < 2) {
/* must have at least 2 log files */
@@ -2422,17 +2407,19 @@ files_checked:
bool success = buf_flush_list(ULINT_MAX, LSN_MAX, NULL);
ut_a(success);
- min_flushed_lsn = max_flushed_lsn = log_get_lsn();
+ flushed_lsn = log_get_lsn();
buf_flush_wait_batch_end(NULL, BUF_FLUSH_LIST);
/* Stamp the LSN to the data files. */
- fil_write_flushed_lsn_to_data_files(max_flushed_lsn, 0);
+ err = fil_write_flushed_lsn(flushed_lsn);
- fil_flush_file_spaces(FIL_TABLESPACE);
+ if (err != DB_SUCCESS) {
+ return(err);
+ }
err = create_log_files_rename(logfilename, dirnamelen,
- max_flushed_lsn, logfile0);
+ flushed_lsn, logfile0);
if (err != DB_SUCCESS) {
return(err);
@@ -2504,7 +2491,7 @@ files_checked:
err = recv_recovery_from_checkpoint_start(
LOG_CHECKPOINT, LSN_MAX,
- min_flushed_lsn, max_flushed_lsn);
+ flushed_lsn);
if (err == DB_SUCCESS) {
/* Initialize the change buffer. */
@@ -2672,7 +2659,7 @@ files_checked:
DBUG_EXECUTE_IF("innodb_log_abort_1",
return(DB_ERROR););
- min_flushed_lsn = max_flushed_lsn = log_get_lsn();
+ flushed_lsn = log_get_lsn();
ib_logf(IB_LOG_LEVEL_WARN,
"Resizing redo log from %u*%u to %u*%u pages"
@@ -2681,7 +2668,7 @@ files_checked:
(unsigned) srv_log_file_size,
(unsigned) srv_n_log_files,
(unsigned) srv_log_file_size_requested,
- max_flushed_lsn);
+ flushed_lsn);
buf_flush_wait_batch_end(NULL, BUF_FLUSH_LIST);
@@ -2691,7 +2678,7 @@ files_checked:
we need to explicitly flush the log buffers. */
fil_flush(SRV_LOG_SPACE_FIRST_ID);
- ut_ad(max_flushed_lsn == log_get_lsn());
+ ut_ad(flushed_lsn == log_get_lsn());
/* Prohibit redo log writes from any other
threads until creating a log checkpoint at the
@@ -2703,8 +2690,7 @@ files_checked:
return(DB_ERROR););
/* Stamp the LSN to the data files. */
- fil_write_flushed_lsn_to_data_files(
- max_flushed_lsn, 0);
+ err = fil_write_flushed_lsn(flushed_lsn);
DBUG_EXECUTE_IF("innodb_log_abort_4", err = DB_ERROR;);
@@ -2712,8 +2698,6 @@ files_checked:
return(err);
}
- fil_flush_file_spaces(FIL_TABLESPACE);
-
/* Close and free the redo log files, so that
we can replace them. */
fil_close_log_files(true);
@@ -2730,13 +2714,23 @@ files_checked:
srv_log_file_size = srv_log_file_size_requested;
err = create_log_files(create_new_db, logfilename,
- dirnamelen, max_flushed_lsn,
+ dirnamelen, flushed_lsn,
logfile0);
+ if (err != DB_SUCCESS) {
+ return(err);
+ }
+
+ /* create_log_files() can increase system lsn that is
+ why FIL_PAGE_FILE_FLUSH_LSN have to be updated */
+ flushed_lsn = log_get_lsn();
+
+ err = fil_write_flushed_lsn(flushed_lsn);
+
if (err == DB_SUCCESS) {
err = create_log_files_rename(
logfilename, dirnamelen,
- max_flushed_lsn, logfile0);
+ flushed_lsn, logfile0);
}
if (err != DB_SUCCESS) {
--
cgit v1.2.1
From 6b6987154a23a8eba72fd58cbff915ae6d17189f Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Jan=20Lindstr=C3=B6m?= <jan.lindstrom@mariadb.com>
Date: Tue, 30 May 2017 12:02:42 +0300
Subject: MDEV-12114: install_db shows corruption for rest encryption and
innodb_checksum_algorithm=strict_none
Problem was that checksum check resulted false positives that page is
both not encrypted and encryted when checksum_algorithm was
strict_none.
Encrypton checksum will use only crc32 regardless of setting.
buf_zip_decompress: If compression fails report a error message
containing the space name if available (not available during import).
And note if space could be encrypted.
buf_page_get_gen: Do not assert if decompression fails,
instead unfix the page and return NULL to upper layer.
fil_crypt_calculate_checksum: Use only crc32 method.
fil_space_verify_crypt_checksum: Here we need to check
crc32, innodb and none method for old datafiles.
fil_space_release_for_io: Allow null space.
encryption.innodb-compressed-blob is now run with crc32 and none
combinations.
Note that with none and strict_none method there is not really
a way to detect page corruptions and page corruptions after
decrypting the page with incorrect key.
New test innodb-checksum-algorithm to test different checksum
algorithms with encrypted, row compressed and page compressed
tables.
---
storage/innobase/buf/buf0buf.cc | 101 +++++++++++++++++++++++++++++---------
storage/innobase/fil/fil0crypt.cc | 34 ++-----------
2 files changed, 82 insertions(+), 53 deletions(-)
(limited to 'storage/innobase')
diff --git a/storage/innobase/buf/buf0buf.cc b/storage/innobase/buf/buf0buf.cc
index 2cb2673f7fb..9b9d0b37f13 100644
--- a/storage/innobase/buf/buf0buf.cc
+++ b/storage/innobase/buf/buf0buf.cc
@@ -354,6 +354,15 @@ bool
buf_page_decrypt_after_read(buf_page_t* bpage, fil_space_t* space)
MY_ATTRIBUTE((nonnull));
+/********************************************************************//**
+Mark a table with the specified space pointed by bpage->space corrupted.
+Also remove the bpage from LRU list.
+@param[in,out] bpage Block */
+static
+void
+buf_mark_space_corrupt(
+ buf_page_t* bpage);
+
/* prototypes for new functions added to ha_innodb.cc */
trx_t* innobase_get_trx();
@@ -2526,17 +2535,26 @@ buf_zip_decompress(
{
const byte* frame = block->page.zip.data;
ulint size = page_zip_get_size(&block->page.zip);
+ /* Space is not found if this function is called during IMPORT */
+ fil_space_t* space = fil_space_acquire_for_io(block->page.space);
+ const unsigned key_version = mach_read_from_4(frame +
+ FIL_PAGE_FILE_FLUSH_LSN_OR_KEY_VERSION);
+ fil_space_crypt_t* crypt_data = space ? space->crypt_data : NULL;
+ const bool encrypted = crypt_data
+ && crypt_data->type != CRYPT_SCHEME_UNENCRYPTED
+ && (!crypt_data->is_default_encryption()
+ || srv_encrypt_tables);
ut_ad(buf_block_get_zip_size(block));
ut_a(buf_block_get_space(block) != 0);
if (UNIV_UNLIKELY(check && !page_zip_verify_checksum(frame, size))) {
- ut_print_timestamp(stderr);
- fprintf(stderr,
- " InnoDB: compressed page checksum mismatch"
- " (space %u page %u): stored: %lu, crc32: %lu "
- "innodb: %lu, none: %lu\n",
+ ib_logf(IB_LOG_LEVEL_ERROR,
+ "Compressed page checksum mismatch"
+ " for %s [%u:%u]: stored: " ULINTPF ", crc32: " ULINTPF
+ " innodb: " ULINTPF ", none: " ULINTPF ".",
+ space ? space->chain.start->name : "N/A",
block->page.space, block->page.offset,
mach_read_from_4(frame + FIL_PAGE_SPACE_OR_CHKSUM),
page_zip_calc_checksum(frame, size,
@@ -2545,22 +2563,28 @@ buf_zip_decompress(
SRV_CHECKSUM_ALGORITHM_INNODB),
page_zip_calc_checksum(frame, size,
SRV_CHECKSUM_ALGORITHM_NONE));
- return(FALSE);
+ goto err_exit;
}
switch (fil_page_get_type(frame)) {
- case FIL_PAGE_INDEX:
+ case FIL_PAGE_INDEX: {
+
if (page_zip_decompress(&block->page.zip,
block->frame, TRUE)) {
+ if (space) {
+ fil_space_release_for_io(space);
+ }
return(TRUE);
}
- fprintf(stderr,
- "InnoDB: unable to decompress space %u page %u\n",
+ ib_logf(IB_LOG_LEVEL_ERROR,
+ "Unable to decompress space %s [%u:%u]",
+ space ? space->chain.start->name : "N/A",
block->page.space,
block->page.offset);
- return(FALSE);
+ goto err_exit;
+ }
case FIL_PAGE_TYPE_ALLOCATED:
case FIL_PAGE_INODE:
case FIL_PAGE_IBUF_BITMAP:
@@ -2571,14 +2595,36 @@ buf_zip_decompress(
/* Copy to uncompressed storage. */
memcpy(block->frame, frame,
buf_block_get_zip_size(block));
+
+ if (space) {
+ fil_space_release_for_io(space);
+ }
+
return(TRUE);
}
- ut_print_timestamp(stderr);
- fprintf(stderr,
- " InnoDB: unknown compressed page"
- " type %lu\n",
- fil_page_get_type(frame));
+ ib_logf(IB_LOG_LEVEL_ERROR,
+ "Unknown compressed page in %s [%u:%u]"
+ " type %s [" ULINTPF "].",
+ space ? space->chain.start->name : "N/A",
+ block->page.space, block->page.offset,
+ fil_get_page_type_name(fil_page_get_type(frame)), fil_page_get_type(frame));
+
+err_exit:
+ if (encrypted) {
+ ib_logf(IB_LOG_LEVEL_INFO,
+ "Row compressed page could be encrypted with key_version %u.",
+ key_version);
+ block->page.encrypted = true;
+ dict_set_encrypted_by_space(block->page.space);
+ } else {
+ dict_set_corrupted_by_space(block->page.space);
+ }
+
+ if (space) {
+ fil_space_release_for_io(space);
+ }
+
return(FALSE);
}
@@ -3031,9 +3077,9 @@ loop:
}
ib_logf(IB_LOG_LEVEL_FATAL, "Unable"
- " to read tablespace %lu page no"
- " %lu into the buffer pool after"
- " %lu attempts"
+ " to read tablespace " ULINTPF " page no "
+ ULINTPF " into the buffer pool after "
+ ULINTPF " attempts."
" The most probable cause"
" of this error may be that the"
" table has been corrupted."
@@ -3232,12 +3278,21 @@ got_block:
/* Decompress the page while not holding
buf_pool->mutex or block->mutex. */
- /* Page checksum verification is already done when
- the page is read from disk. Hence page checksum
- verification is not necessary when decompressing the page. */
{
- bool success = buf_zip_decompress(block, FALSE);
- ut_a(success);
+ bool success = buf_zip_decompress(block, TRUE);
+
+ if (!success) {
+ buf_pool_mutex_enter(buf_pool);
+ buf_block_mutex_enter(fix_block);
+ buf_block_set_io_fix(fix_block, BUF_IO_NONE);
+ buf_block_mutex_exit(fix_block);
+
+ --buf_pool->n_pend_unzip;
+ buf_block_unfix(fix_block);
+ buf_pool_mutex_exit(buf_pool);
+ rw_lock_x_unlock(&fix_block->lock);
+ return NULL;
+ }
}
if (!recv_no_ibuf_operations) {
diff --git a/storage/innobase/fil/fil0crypt.cc b/storage/innobase/fil/fil0crypt.cc
index a6d85fb89bf..2131a936656 100644
--- a/storage/innobase/fil/fil0crypt.cc
+++ b/storage/innobase/fil/fil0crypt.cc
@@ -887,7 +887,7 @@ fil_space_decrypt(
Calculate post encryption checksum
@param[in] zip_size zip_size or 0
@param[in] dst_frame Block where checksum is calculated
-@return page checksum or BUF_NO_CHECKSUM_MAGIC
+@return page checksum
not needed. */
UNIV_INTERN
ulint
@@ -896,30 +896,13 @@ fil_crypt_calculate_checksum(
const byte* dst_frame)
{
ib_uint32_t checksum = 0;
- srv_checksum_algorithm_t algorithm =
- static_cast<srv_checksum_algorithm_t>(srv_checksum_algorithm);
+ /* For encrypted tables we use only crc32 and strict_crc32 */
if (zip_size == 0) {
- switch (algorithm) {
- case SRV_CHECKSUM_ALGORITHM_CRC32:
- case SRV_CHECKSUM_ALGORITHM_STRICT_CRC32:
- checksum = buf_calc_page_crc32(dst_frame);
- break;
- case SRV_CHECKSUM_ALGORITHM_INNODB:
- case SRV_CHECKSUM_ALGORITHM_STRICT_INNODB:
- checksum = (ib_uint32_t) buf_calc_page_new_checksum(
- dst_frame);
- break;
- case SRV_CHECKSUM_ALGORITHM_NONE:
- case SRV_CHECKSUM_ALGORITHM_STRICT_NONE:
- checksum = BUF_NO_CHECKSUM_MAGIC;
- break;
- /* no default so the compiler will emit a warning
- * if new enum is added and not handled here */
- }
+ checksum = buf_calc_page_crc32(dst_frame);
} else {
checksum = page_zip_calc_checksum(dst_frame, zip_size,
- algorithm);
+ SRV_CHECKSUM_ALGORITHM_CRC32);
}
return checksum;
@@ -953,14 +936,6 @@ fil_space_verify_crypt_checksum(
return(false);
}
- srv_checksum_algorithm_t algorithm =
- static_cast<srv_checksum_algorithm_t>(srv_checksum_algorithm);
-
- /* If no checksum is used, can't continue checking. */
- if (algorithm == SRV_CHECKSUM_ALGORITHM_NONE) {
- return(true);
- }
-
/* Read stored post encryption checksum. */
ib_uint32_t checksum = mach_read_from_4(
page + FIL_PAGE_FILE_FLUSH_LSN_OR_KEY_VERSION + 4);
@@ -1044,7 +1019,6 @@ fil_space_verify_crypt_checksum(
checksum2 = mach_read_from_4(
page + UNIV_PAGE_SIZE - FIL_PAGE_END_LSN_OLD_CHKSUM);
valid = (buf_page_is_checksum_valid_crc32(page,checksum1,checksum2)
- || buf_page_is_checksum_valid_none(page,checksum1,checksum2)
|| buf_page_is_checksum_valid_innodb(page,checksum1, checksum2));
}
--
cgit v1.2.1
From 112b21da37dad0fbb28bc65f9ab5a3ba6c0c2186 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Jan=20Lindstr=C3=B6m?= <jan.lindstrom@mariadb.com>
Date: Tue, 30 May 2017 11:55:11 +0300
Subject: MDEV-12600: crash during install_db with innodb_page_size=32K and
ibdata1=3M;
Problem was that all doublewrite buffer pages must fit to first
system datafile.
Ported commit 27a34df7882b1f8ed283f22bf83e8bfc523cbfde
Author: Shaohua Wang <shaohua.wang@oracle.com>
Date: Wed Aug 12 15:55:19 2015 +0800
BUG#21551464 - SEGFAULT WHILE INITIALIZING DATABASE WHEN
INNODB_DATA_FILE SIZE IS SMALL
To 10.1 (with extended error printout).
btr_create(): If ibuf header page allocation fails report error and
return FIL_NULL. Similarly if root page allocation fails return a error.
dict_build_table_def_step: If fsp_header_init fails return
error code.
fsp_header_init: returns true if header initialization succeeds
and false if not.
fseg_create_general: report error if segment or page allocation fails.
innobase_init: If first datafile is smaller than 3M and could not
contain all doublewrite buffer pages report error and fail to
initialize InnoDB plugin.
row_truncate_table_for_mysql: report error if fsp header init
fails.
srv_init_abort: New function to report database initialization errors.
srv_undo_tablespaces_init, innobase_start_or_create_for_mysql: If
database initialization fails report error and abort.
trx_rseg_create: If segment header creation fails return.
---
storage/innobase/btr/btr0btr.cc | 19 +++++++++++++
storage/innobase/dict/dict0crea.cc | 6 +++-
storage/innobase/fil/fil0fil.cc | 42 ++++++++++++++--------------
storage/innobase/fsp/fsp0fsp.cc | 37 ++++++++++++++++++-------
storage/innobase/handler/ha_innodb.cc | 14 ++++++++++
storage/innobase/include/fsp0fsp.h | 20 ++++++++------
storage/innobase/row/row0mysql.cc | 8 +++++-
storage/innobase/srv/srv0start.cc | 52 +++++++++++++++++++++++++++++++++--
storage/innobase/trx/trx0rseg.cc | 5 +++-
9 files changed, 159 insertions(+), 44 deletions(-)
(limited to 'storage/innobase')
diff --git a/storage/innobase/btr/btr0btr.cc b/storage/innobase/btr/btr0btr.cc
index cccae031b88..e200a2b9677 100644
--- a/storage/innobase/btr/btr0btr.cc
+++ b/storage/innobase/btr/btr0btr.cc
@@ -1701,6 +1701,12 @@ btr_create(
space, 0,
IBUF_HEADER + IBUF_TREE_SEG_HEADER, mtr);
+ if (ibuf_hdr_block == NULL) {
+ ib_logf(IB_LOG_LEVEL_ERROR,
+ "Allocation of the first ibuf header page failed.");
+ return (FIL_NULL);
+ }
+
buf_block_dbg_add_level(
ibuf_hdr_block, SYNC_IBUF_TREE_NODE_NEW);
@@ -1714,6 +1720,11 @@ btr_create(
+ IBUF_HEADER + IBUF_TREE_SEG_HEADER,
IBUF_TREE_ROOT_PAGE_NO,
FSP_UP, mtr);
+
+ if (!block) {
+ ib_logf(IB_LOG_LEVEL_ERROR,
+ "Allocation of the tree root page segment failed.");
+ }
ut_ad(buf_block_get_page_no(block) == IBUF_TREE_ROOT_PAGE_NO);
} else {
#ifdef UNIV_BLOB_DEBUG
@@ -1726,6 +1737,12 @@ btr_create(
#endif /* UNIV_BLOB_DEBUG */
block = fseg_create(space, 0,
PAGE_HEADER + PAGE_BTR_SEG_TOP, mtr);
+
+ if (!block) {
+ ib_logf(IB_LOG_LEVEL_ERROR,
+ "Allocation of the btree segment failed.");
+ }
+
}
if (block == NULL) {
@@ -1754,6 +1771,8 @@ btr_create(
segment before return. */
btr_free_root(space, zip_size, page_no, mtr);
+ ib_logf(IB_LOG_LEVEL_ERROR,
+ "Allocation of the non-ibuf tree segment for leaf pages failed.");
return(FIL_NULL);
}
diff --git a/storage/innobase/dict/dict0crea.cc b/storage/innobase/dict/dict0crea.cc
index f6cd294884b..1ec7123bbc2 100644
--- a/storage/innobase/dict/dict0crea.cc
+++ b/storage/innobase/dict/dict0crea.cc
@@ -321,9 +321,13 @@ dict_build_table_def_step(
mtr_start(&mtr);
- fsp_header_init(table->space, FIL_IBD_FILE_INITIAL_SIZE, &mtr);
+ bool res = fsp_header_init(table->space, FIL_IBD_FILE_INITIAL_SIZE, &mtr);
mtr_commit(&mtr);
+
+ if (!res) {
+ return (DB_ERROR);
+ }
} else {
/* Create in the system tablespace: disallow Barracuda
features by keeping only the first bit which says whether
diff --git a/storage/innobase/fil/fil0fil.cc b/storage/innobase/fil/fil0fil.cc
index 9d4a464460f..e19ef110b4f 100644
--- a/storage/innobase/fil/fil0fil.cc
+++ b/storage/innobase/fil/fil0fil.cc
@@ -5812,19 +5812,21 @@ fil_report_invalid_page_access(
ulint len, /*!< in: I/O length */
ulint type) /*!< in: I/O type */
{
- fprintf(stderr,
- "InnoDB: Error: trying to access page number %lu"
- " in space %lu,\n"
- "InnoDB: space name %s,\n"
- "InnoDB: which is outside the tablespace bounds.\n"
- "InnoDB: Byte offset %lu, len %lu, i/o type %lu.\n"
- "InnoDB: If you get this error at mysqld startup,"
- " please check that\n"
- "InnoDB: your my.cnf matches the ibdata files"
- " that you have in the\n"
- "InnoDB: MySQL server.\n",
- (ulong) block_offset, (ulong) space_id, space_name,
- (ulong) byte_offset, (ulong) len, (ulong) type);
+ ib_logf(IB_LOG_LEVEL_ERROR,
+ "Trying to access page number " ULINTPF
+ " in space " ULINTPF
+ " space name %s,"
+ " which is outside the tablespace bounds."
+ " Byte offset " ULINTPF ", len " ULINTPF " i/o type " ULINTPF ".",
+ block_offset, space_id, space_name,
+ byte_offset, len, type);
+
+ ib_logf(IB_LOG_LEVEL_FATAL,
+ "If you get this error at mysqld startup,"
+ " please check that"
+ " your my.cnf matches the ibdata files"
+ " that you have in the"
+ " MySQL server.");
}
/********************************************************************//**
@@ -6043,11 +6045,10 @@ fil_io(
mutex_exit(&fil_system->mutex);
return(DB_ERROR);
}
+
fil_report_invalid_page_access(
block_offset, space_id, space->name,
byte_offset, len, type);
-
- ut_error;
}
/* Open file if closed */
@@ -6059,10 +6060,11 @@ fil_io(
ib_logf(IB_LOG_LEVEL_ERROR,
"Trying to do i/o to a tablespace which "
"exists without .ibd data file. "
- "i/o type %lu, space id %lu, page no %lu, "
- "i/o length %lu bytes",
- (ulong) type, (ulong) space_id,
- (ulong) block_offset, (ulong) len);
+ "i/o type " ULINTPF ", space id "
+ ULINTPF ", page no " ULINTPF ", "
+ "i/o length " ULINTPF " bytes",
+ type, space_id,
+ block_offset, len);
return(DB_TABLESPACE_DELETED);
}
@@ -6082,8 +6084,6 @@ fil_io(
fil_report_invalid_page_access(
block_offset, space_id, space->name, byte_offset,
len, type);
-
- ut_error;
}
/* Now we have made the changes in the data structures of fil_system */
diff --git a/storage/innobase/fsp/fsp0fsp.cc b/storage/innobase/fsp/fsp0fsp.cc
index a40283412e2..98cd11f3369 100644
--- a/storage/innobase/fsp/fsp0fsp.cc
+++ b/storage/innobase/fsp/fsp0fsp.cc
@@ -670,16 +670,18 @@ fsp_header_init_fields(
}
#ifndef UNIV_HOTBACKUP
-/**********************************************************************//**
-Initializes the space header of a new created space and creates also the
-insert buffer tree root if space == 0. */
+/** Initializes the space header of a new created space and creates also the
+insert buffer tree root if space == 0.
+@param[in] space_id space id
+@param[in] size current size in blocks
+@param[in,out] mtr min-transaction
+@return true on success, otherwise false. */
UNIV_INTERN
-void
+bool
fsp_header_init(
-/*============*/
- ulint space_id, /*!< in: space id */
- ulint size, /*!< in: current size in blocks */
- mtr_t* mtr) /*!< in/out: mini-transaction */
+ ulint space_id,
+ ulint size,
+ mtr_t* mtr)
{
fsp_header_t* header;
buf_block_t* block;
@@ -722,11 +724,15 @@ fsp_header_init(
flst_init(header + FSP_SEG_INODES_FREE, mtr);
mlog_write_ull(header + FSP_SEG_ID, 1, mtr);
+
if (space_id == 0) {
fsp_fill_free_list(FALSE, space_id, header, mtr);
- btr_create(DICT_CLUSTERED | DICT_UNIVERSAL | DICT_IBUF,
+
+ if (btr_create(DICT_CLUSTERED | DICT_UNIVERSAL | DICT_IBUF,
0, 0, DICT_IBUF_ID_MIN + space_id,
- dict_ind_redundant, mtr);
+ dict_ind_redundant, mtr) == FIL_NULL) {
+ return (false);
+ }
} else {
fsp_fill_free_list(TRUE, space_id, header, mtr);
}
@@ -739,6 +745,8 @@ fsp_header_init(
}
fil_space_release(space);
+
+ return (true);
}
#endif /* !UNIV_HOTBACKUP */
@@ -2057,6 +2065,10 @@ fseg_create_general(
success = fsp_reserve_free_extents(&n_reserved, space, 2,
FSP_NORMAL, mtr);
if (!success) {
+ ib_logf(IB_LOG_LEVEL_ERROR,
+ "Reserving %d free extents failed"
+ " could reserve only " ULINTPF " extents.",
+ 2, n_reserved);
return(NULL);
}
}
@@ -2066,6 +2078,8 @@ fseg_create_general(
inode = fsp_alloc_seg_inode(space_header, mtr);
if (inode == NULL) {
+ ib_logf(IB_LOG_LEVEL_ERROR,
+ "Allocation of a new file segment inode page failed.");
goto funct_exit;
}
@@ -2095,6 +2109,9 @@ fseg_create_general(
inode, 0, FSP_UP, mtr, mtr);
if (block == NULL) {
+ ib_logf(IB_LOG_LEVEL_ERROR,
+ "Allocation of a free page from space " ULINTPF " failed.",
+ space);
fsp_free_seg_inode(space, zip_size, inode, mtr);
diff --git a/storage/innobase/handler/ha_innodb.cc b/storage/innobase/handler/ha_innodb.cc
index 012539d1ace..b68a96c8846 100644
--- a/storage/innobase/handler/ha_innodb.cc
+++ b/storage/innobase/handler/ha_innodb.cc
@@ -3353,6 +3353,7 @@ innobase_init(
char *default_path;
uint format_id;
ulong num_pll_degree;
+ ulint min_size = 0;
DBUG_ENTER("innobase_init");
handlerton *innobase_hton= (handlerton*) p;
@@ -3563,6 +3564,19 @@ mem_free_and_error:
goto error;
}
+ /* All doublewrite buffer pages must fit to first system
+ datafile and first datafile must be at least 3M. */
+ min_size = ut_max((3*1024*1024U), (192U*UNIV_PAGE_SIZE));
+
+ if ((srv_data_file_sizes[0]*1024*1024) < min_size) {
+ sql_print_error(
+ "InnoDB: first datafile is too small current=" ULINTPF
+ "M it should be at least " ULINTPF "M.",
+ srv_data_file_sizes[0],
+ min_size / (1024 * 1024));
+ goto mem_free_and_error;
+ }
+
/* -------------- All log files ---------------------------*/
/* The default dir for log files is the datadir of MySQL */
diff --git a/storage/innobase/include/fsp0fsp.h b/storage/innobase/include/fsp0fsp.h
index 2a162100dc1..905e98cc1e6 100644
--- a/storage/innobase/include/fsp0fsp.h
+++ b/storage/innobase/include/fsp0fsp.h
@@ -520,16 +520,20 @@ fsp_header_init_fields(
ulint space_id, /*!< in: space id */
ulint flags); /*!< in: tablespace flags (FSP_SPACE_FLAGS):
0, or table->flags if newer than COMPACT */
-/**********************************************************************//**
-Initializes the space header of a new created space and creates also the
-insert buffer tree root if space == 0. */
+/** Initializes the space header of a new created space and creates also the
+insert buffer tree root if space == 0.
+@param[in] space_id space id
+@param[in] size current size in blocks
+@param[in,out] mtr min-transaction
+@return true on success, otherwise false. */
UNIV_INTERN
-void
+bool
fsp_header_init(
-/*============*/
- ulint space, /*!< in: space id */
- ulint size, /*!< in: current size in blocks */
- mtr_t* mtr); /*!< in/out: mini-transaction */
+ ulint space_id,
+ ulint size,
+ mtr_t* mtr)
+ MY_ATTRIBUTE((warn_unused_result));
+
/**********************************************************************//**
Increases the space size field of a space. */
UNIV_INTERN
diff --git a/storage/innobase/row/row0mysql.cc b/storage/innobase/row/row0mysql.cc
index 82938995e93..6ca9443dc7d 100644
--- a/storage/innobase/row/row0mysql.cc
+++ b/storage/innobase/row/row0mysql.cc
@@ -3590,9 +3590,15 @@ row_truncate_table_for_mysql(
} while (index);
mtr_start_trx(&mtr, trx);
- fsp_header_init(space_id,
+ bool ret = fsp_header_init(space_id,
FIL_IBD_FILE_INITIAL_SIZE, &mtr);
mtr_commit(&mtr);
+
+ if (!ret) {
+ table->file_unreadable = true;
+ err = DB_ERROR;
+ goto funct_exit;
+ }
}
} else {
/* Lock all index trees for this table, as we will
diff --git a/storage/innobase/srv/srv0start.cc b/storage/innobase/srv/srv0start.cc
index f601ff17ebc..032b902c633 100644
--- a/storage/innobase/srv/srv0start.cc
+++ b/storage/innobase/srv/srv0start.cc
@@ -188,6 +188,39 @@ UNIV_INTERN mysql_pfs_key_t srv_master_thread_key;
UNIV_INTERN mysql_pfs_key_t srv_purge_thread_key;
#endif /* UNIV_PFS_THREAD */
+/** Innobase start-up aborted. Perform cleanup actions.
+@param[in] create_new_db TRUE if new db is being created
+@param[in] file File name
+@param[in] line Line number
+@param[in] err Reason for aborting InnoDB startup
+@return DB_SUCCESS or error code. */
+static
+dberr_t
+srv_init_abort(
+ bool create_new_db,
+ const char* file,
+ ulint line,
+ dberr_t err)
+{
+ if (create_new_db) {
+ ib_logf(IB_LOG_LEVEL_ERROR,
+ "Database creation was aborted"
+ " at %s [" ULINTPF "]"
+ " with error %s. You may need"
+ " to delete the ibdata1 file before trying to start"
+ " up again.",
+ file, line, ut_strerr(err));
+ } else {
+ ib_logf(IB_LOG_LEVEL_ERROR,
+ "Plugin initialization aborted"
+ " at %s [" ULINTPF "]"
+ " with error %s.",
+ file, line, ut_strerr(err));
+ }
+
+ return(err);
+}
+
/*********************************************************************//**
Convert a numeric string that optionally ends in G or M or K, to a number
containing megabytes.
@@ -1528,18 +1561,26 @@ srv_undo_tablespaces_init(
if (create_new_db) {
mtr_t mtr;
+ bool ret=true;
mtr_start(&mtr);
/* The undo log tablespace */
for (i = 0; i < n_undo_tablespaces; ++i) {
- fsp_header_init(
+ ret = fsp_header_init(
undo_tablespace_ids[i],
SRV_UNDO_TABLESPACE_SIZE_IN_PAGES, &mtr);
+ if (!ret) {
+ break;
+ }
}
mtr_commit(&mtr);
+
+ if (!ret) {
+ return (srv_init_abort(create_new_db, __FILE__, __LINE__, DB_ERROR));
+ }
}
return(DB_SUCCESS);
@@ -2378,10 +2419,14 @@ files_checked:
mtr_start(&mtr);
- fsp_header_init(0, sum_of_new_sizes, &mtr);
+ bool ret = fsp_header_init(0, sum_of_new_sizes, &mtr);
mtr_commit(&mtr);
+ if (!ret) {
+ return (srv_init_abort(create_new_db, __FILE__, __LINE__, DB_ERROR));
+ }
+
/* To maintain backward compatibility we create only
the first rollback segment before the double write buffer.
All the remaining rollback segments will be created later,
@@ -2809,6 +2854,9 @@ files_checked:
/* Can only happen if server is read only. */
ut_a(srv_read_only_mode);
srv_undo_logs = ULONG_UNDEFINED;
+ } else if (srv_available_undo_logs < srv_undo_logs) {
+ /* Should due to out of file space. */
+ return (srv_init_abort(create_new_db, __FILE__, __LINE__, DB_ERROR));
}
if (!srv_read_only_mode) {
diff --git a/storage/innobase/trx/trx0rseg.cc b/storage/innobase/trx/trx0rseg.cc
index 003d1036a8c..21dbab98e48 100644
--- a/storage/innobase/trx/trx0rseg.cc
+++ b/storage/innobase/trx/trx0rseg.cc
@@ -323,7 +323,10 @@ trx_rseg_create(
page_no = trx_rseg_header_create(
space, 0, ULINT_MAX, slot_no, &mtr);
- ut_a(page_no != FIL_NULL);
+ if (page_no == FIL_NULL) {
+ mtr_commit(&mtr);
+ return (rseg);
+ }
sys_header = trx_sysf_get(&mtr);
--
cgit v1.2.1
From 151daaf4805d4e8ed30d2d871ac99ff1c1873bdc Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Marko=20M=C3=A4kel=C3=A4?= <marko.makela@mariadb.com>
Date: Mon, 5 Jun 2017 15:16:15 +0300
Subject: MDEV-12994 innodb_fast_shutdown=0 skips change buffer merge; fast
shutdown does it
srv_master_thread(): Pass the correct parameter to srv_shutdown().
This bug was introduced in MDEV-12052, and it affects the MariaDB 10.1.24
release.
---
storage/innobase/srv/srv0srv.cc | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
(limited to 'storage/innobase')
diff --git a/storage/innobase/srv/srv0srv.cc b/storage/innobase/srv/srv0srv.cc
index 93ed302a236..bad1579070c 100644
--- a/storage/innobase/srv/srv0srv.cc
+++ b/storage/innobase/srv/srv0srv.cc
@@ -2403,7 +2403,7 @@ suspend_thread:
case SRV_SHUTDOWN_CLEANUP:
if (srv_shutdown_state == SRV_SHUTDOWN_CLEANUP
&& srv_fast_shutdown < 2) {
- srv_shutdown(srv_fast_shutdown == 1);
+ srv_shutdown(srv_fast_shutdown == 0);
}
srv_suspend_thread(slot);
my_thread_end();
--
cgit v1.2.1
From 68890fe7d48b6fee505f294400224fe01107950c Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Marko=20M=C3=A4kel=C3=A4?= <marko.makela@mariadb.com>
Date: Tue, 6 Jun 2017 15:24:43 +0300
Subject: Revert part of MDEV-12113
commit 1af8bf39ca2513bfdf43a55af0b6af10d32dcebb added unnecessary
calls to fil_write_flushed_lsn() during redo log resizing at
InnoDB server startup.
Because fil_write_flushed_lsn() is neither redo-logged nor doublewrite
buffered, the call is risky and should be avoided, because if the
server killed during the write call, the whole InnoDB instance can
become inaccessible (corrupted page 0 in the system tablespace).
In the best case, this call might prevent a diagnostic message from
being emitted to the error log on the next startup.
---
storage/innobase/srv/srv0start.cc | 13 ++-----------
1 file changed, 2 insertions(+), 11 deletions(-)
(limited to 'storage/innobase')
diff --git a/storage/innobase/srv/srv0start.cc b/storage/innobase/srv/srv0start.cc
index 032b902c633..4e59ddd1f87 100644
--- a/storage/innobase/srv/srv0start.cc
+++ b/storage/innobase/srv/srv0start.cc
@@ -2766,17 +2766,8 @@ files_checked:
return(err);
}
- /* create_log_files() can increase system lsn that is
- why FIL_PAGE_FILE_FLUSH_LSN have to be updated */
- flushed_lsn = log_get_lsn();
-
- err = fil_write_flushed_lsn(flushed_lsn);
-
- if (err == DB_SUCCESS) {
- err = create_log_files_rename(
- logfilename, dirnamelen,
- flushed_lsn, logfile0);
- }
+ err = create_log_files_rename(logfilename, dirnamelen,
+ log_get_lsn(), logfile0);
if (err != DB_SUCCESS) {
return(err);
--
cgit v1.2.1
From fbeb9489cd7d6ad859a49ae5ab8f876f3d988470 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Marko=20M=C3=A4kel=C3=A4?= <marko.makela@mariadb.com>
Date: Tue, 6 Jun 2017 14:59:42 +0300
Subject: Cleanup of MDEV-12600: crash during install_db with
innodb_page_size=32K and ibdata1=3M
The doublewrite buffer pages must fit in the first InnoDB system
tablespace data file. The checks that were added in the initial patch
(commit 112b21da37dad0fbb28bc65f9ab5a3ba6c0c2186)
were at too high level and did not cover all cases.
innodb.log_data_file_size: Test all innodb_page_size combinations.
fsp_header_init(): Never return an error. Move the change buffer creation
to the only caller that needs to do it.
btr_create(): Clean up the logic. Remove the error log messages.
buf_dblwr_create(): Try to return an error on non-fatal failure.
Check that the first data file is big enough for creating the
doublewrite buffers.
buf_dblwr_process(): Check if the doublewrite buffer is available.
Display the message only if it is available.
recv_recovery_from_checkpoint_start_func(): Remove a redundant message
about FIL_PAGE_FILE_FLUSH_LSN mismatch when crash recovery has already
been initiated.
fil_report_invalid_page_access(): Simplify the message.
fseg_create_general(): Do not emit messages to the error log.
innobase_init(): Revert the changes.
trx_rseg_create(): Refactor (no functional change).
---
storage/innobase/btr/btr0btr.cc | 54 ++++++++-----------------
storage/innobase/buf/buf0dblwr.cc | 74 ++++++++++++++++++++++++-----------
storage/innobase/dict/dict0crea.cc | 6 +--
storage/innobase/fil/fil0fil.cc | 20 +++++-----
storage/innobase/fsp/fsp0fsp.cc | 37 +++---------------
storage/innobase/handler/ha_innodb.cc | 14 -------
storage/innobase/include/buf0dblwr.h | 16 ++++----
storage/innobase/include/fsp0fsp.h | 14 ++-----
storage/innobase/include/trx0rseg.h | 11 +++---
storage/innobase/log/log0recv.cc | 23 -----------
storage/innobase/row/row0mysql.cc | 8 +---
storage/innobase/srv/srv0start.cc | 33 ++++++++--------
storage/innobase/trx/trx0rseg.cc | 34 +++++++---------
13 files changed, 134 insertions(+), 210 deletions(-)
(limited to 'storage/innobase')
diff --git a/storage/innobase/btr/btr0btr.cc b/storage/innobase/btr/btr0btr.cc
index e200a2b9677..b7afcf12b39 100644
--- a/storage/innobase/btr/btr0btr.cc
+++ b/storage/innobase/btr/btr0btr.cc
@@ -2,7 +2,7 @@
Copyright (c) 1994, 2016, Oracle and/or its affiliates. All Rights Reserved.
Copyright (c) 2012, Facebook Inc.
-Copyright (c) 2014, 2017, MariaDB Corporation
+Copyright (c) 2014, 2017, MariaDB Corporation.
This program is free software; you can redistribute it and/or modify it under
the terms of the GNU General Public License as published by the Free Software
@@ -1684,9 +1684,7 @@ btr_create(
dict_index_t* index, /*!< in: index */
mtr_t* mtr) /*!< in: mini-transaction handle */
{
- ulint page_no;
buf_block_t* block;
- buf_frame_t* frame;
page_t* page;
page_zip_des_t* page_zip;
@@ -1702,9 +1700,7 @@ btr_create(
IBUF_HEADER + IBUF_TREE_SEG_HEADER, mtr);
if (ibuf_hdr_block == NULL) {
- ib_logf(IB_LOG_LEVEL_ERROR,
- "Allocation of the first ibuf header page failed.");
- return (FIL_NULL);
+ return(FIL_NULL);
}
buf_block_dbg_add_level(
@@ -1721,11 +1717,16 @@ btr_create(
IBUF_TREE_ROOT_PAGE_NO,
FSP_UP, mtr);
- if (!block) {
- ib_logf(IB_LOG_LEVEL_ERROR,
- "Allocation of the tree root page segment failed.");
+ if (block == NULL) {
+ return(FIL_NULL);
}
+
ut_ad(buf_block_get_page_no(block) == IBUF_TREE_ROOT_PAGE_NO);
+
+ buf_block_dbg_add_level(block, SYNC_IBUF_TREE_NODE_NEW);
+
+ flst_init(block->frame + PAGE_HEADER + PAGE_BTR_IBUF_FREE_LIST,
+ mtr);
} else {
#ifdef UNIV_BLOB_DEBUG
if ((type & DICT_CLUSTERED) && !index->blobs) {
@@ -1738,41 +1739,18 @@ btr_create(
block = fseg_create(space, 0,
PAGE_HEADER + PAGE_BTR_SEG_TOP, mtr);
- if (!block) {
- ib_logf(IB_LOG_LEVEL_ERROR,
- "Allocation of the btree segment failed.");
+ if (block == NULL) {
+ return(FIL_NULL);
}
- }
-
- if (block == NULL) {
-
- return(FIL_NULL);
- }
-
- page_no = buf_block_get_page_no(block);
- frame = buf_block_get_frame(block);
-
- if (type & DICT_IBUF) {
- /* It is an insert buffer tree: initialize the free list */
- buf_block_dbg_add_level(block, SYNC_IBUF_TREE_NODE_NEW);
-
- ut_ad(page_no == IBUF_TREE_ROOT_PAGE_NO);
-
- flst_init(frame + PAGE_HEADER + PAGE_BTR_IBUF_FREE_LIST, mtr);
- } else {
- /* It is a non-ibuf tree: create a file segment for leaf
- pages */
buf_block_dbg_add_level(block, SYNC_TREE_NODE_NEW);
- if (!fseg_create(space, page_no,
+ if (!fseg_create(space, buf_block_get_page_no(block),
PAGE_HEADER + PAGE_BTR_SEG_LEAF, mtr)) {
/* Not enough space for new segment, free root
segment before return. */
- btr_free_root(space, zip_size, page_no, mtr);
-
- ib_logf(IB_LOG_LEVEL_ERROR,
- "Allocation of the non-ibuf tree segment for leaf pages failed.");
+ btr_free_root(space, zip_size,
+ buf_block_get_page_no(block), mtr);
return(FIL_NULL);
}
@@ -1816,7 +1794,7 @@ btr_create(
ut_ad(page_get_max_insert_size(page, 2) > 2 * BTR_PAGE_MAX_REC_SIZE);
- return(page_no);
+ return(buf_block_get_page_no(block));
}
/************************************************************//**
diff --git a/storage/innobase/buf/buf0dblwr.cc b/storage/innobase/buf/buf0dblwr.cc
index e2c7ae9bae1..8ff5428d3a5 100644
--- a/storage/innobase/buf/buf0dblwr.cc
+++ b/storage/innobase/buf/buf0dblwr.cc
@@ -175,13 +175,14 @@ buf_dblwr_init(
mem_zalloc(buf_size * sizeof(void*)));
}
-/****************************************************************//**
-Creates the doublewrite buffer to a new InnoDB installation. The header of the
-doublewrite buffer is placed on the trx system header page. */
+/** Create the doublewrite buffer if the doublewrite buffer header
+is not present in the TRX_SYS page.
+@return whether the operation succeeded
+@retval true if the doublewrite buffer exists or was created
+@retval false if the creation failed (too small first data file) */
UNIV_INTERN
-void
-buf_dblwr_create(void)
-/*==================*/
+bool
+buf_dblwr_create()
{
buf_block_t* block2;
buf_block_t* new_block;
@@ -194,8 +195,7 @@ buf_dblwr_create(void)
if (buf_dblwr) {
/* Already inited */
-
- return;
+ return(true);
}
start_again:
@@ -213,39 +213,59 @@ start_again:
mtr_commit(&mtr);
buf_dblwr_being_created = FALSE;
- return;
+ return(true);
}
- ib_logf(IB_LOG_LEVEL_INFO,
- "Doublewrite buffer not found: creating new");
-
if (buf_pool_get_curr_size()
< ((TRX_SYS_DOUBLEWRITE_BLOCKS * TRX_SYS_DOUBLEWRITE_BLOCK_SIZE
+ FSP_EXTENT_SIZE / 2 + 100)
* UNIV_PAGE_SIZE)) {
- ib_logf(IB_LOG_LEVEL_FATAL,
- "Cannot create doublewrite buffer: you must "
- "increase your buffer pool size. Cannot continue "
- "operation.");
+ ib_logf(IB_LOG_LEVEL_ERROR,
+ "Cannot create doublewrite buffer: "
+ "innodb_buffer_pool_size is too small.");
+ mtr_commit(&mtr);
+ return(false);
+ } else {
+ fil_space_t* space = fil_space_acquire(TRX_SYS_SPACE);
+ const bool fail = UT_LIST_GET_FIRST(space->chain)->size
+ < 3 * FSP_EXTENT_SIZE;
+ fil_space_release(space);
+
+ if (fail) {
+ goto too_small;
+ }
}
block2 = fseg_create(TRX_SYS_SPACE, TRX_SYS_PAGE_NO,
TRX_SYS_DOUBLEWRITE
+ TRX_SYS_DOUBLEWRITE_FSEG, &mtr);
+ if (block2 == NULL) {
+too_small:
+ ib_logf(IB_LOG_LEVEL_ERROR,
+ "Cannot create doublewrite buffer: "
+ "the first file in innodb_data_file_path"
+ " must be at least %luM.",
+ 3 * (FSP_EXTENT_SIZE * UNIV_PAGE_SIZE) >> 20);
+ mtr_commit(&mtr);
+ return(false);
+ }
+
+ ib_logf(IB_LOG_LEVEL_INFO,
+ "Doublewrite buffer not found: creating new");
+
+ /* FIXME: After this point, the doublewrite buffer creation
+ is not atomic. The doublewrite buffer should not exist in
+ the InnoDB system tablespace file in the first place.
+ It could be located in separate optional file(s) in a
+ user-specified location. */
+
/* fseg_create acquires a second latch on the page,
therefore we must declare it: */
buf_block_dbg_add_level(block2, SYNC_NO_ORDER_CHECK);
- if (block2 == NULL) {
- ib_logf(IB_LOG_LEVEL_FATAL,
- "Cannot create doublewrite buffer: you must "
- "increase your tablespace size. "
- "Cannot continue operation.");
- }
-
fseg_header = doublewrite + TRX_SYS_DOUBLEWRITE_FSEG;
prev_page_no = 0;
@@ -482,6 +502,14 @@ buf_dblwr_process()
byte* unaligned_read_buf;
recv_dblwr_t& recv_dblwr = recv_sys->dblwr;
+ if (!buf_dblwr) {
+ return;
+ }
+
+ ib_logf(IB_LOG_LEVEL_INFO,
+ "Restoring possible half-written data pages "
+ "from the doublewrite buffer...");
+
unaligned_read_buf = static_cast<byte*>(ut_malloc(2 * UNIV_PAGE_SIZE));
read_buf = static_cast<byte*>(
diff --git a/storage/innobase/dict/dict0crea.cc b/storage/innobase/dict/dict0crea.cc
index 1ec7123bbc2..f6cd294884b 100644
--- a/storage/innobase/dict/dict0crea.cc
+++ b/storage/innobase/dict/dict0crea.cc
@@ -321,13 +321,9 @@ dict_build_table_def_step(
mtr_start(&mtr);
- bool res = fsp_header_init(table->space, FIL_IBD_FILE_INITIAL_SIZE, &mtr);
+ fsp_header_init(table->space, FIL_IBD_FILE_INITIAL_SIZE, &mtr);
mtr_commit(&mtr);
-
- if (!res) {
- return (DB_ERROR);
- }
} else {
/* Create in the system tablespace: disallow Barracuda
features by keeping only the first bit which says whether
diff --git a/storage/innobase/fil/fil0fil.cc b/storage/innobase/fil/fil0fil.cc
index e19ef110b4f..f88bb2add59 100644
--- a/storage/innobase/fil/fil0fil.cc
+++ b/storage/innobase/fil/fil0fil.cc
@@ -1,7 +1,7 @@
/*****************************************************************************
Copyright (c) 1995, 2017, Oracle and/or its affiliates. All Rights Reserved.
-Copyright (c) 2014, 2017, MariaDB Corporation. All Rights Reserved.
+Copyright (c) 2014, 2017, MariaDB Corporation.
This program is free software; you can redistribute it and/or modify it under
the terms of the GNU General Public License as published by the Free Software
@@ -5812,21 +5812,19 @@ fil_report_invalid_page_access(
ulint len, /*!< in: I/O length */
ulint type) /*!< in: I/O type */
{
- ib_logf(IB_LOG_LEVEL_ERROR,
+ ib_logf(IB_LOG_LEVEL_FATAL,
"Trying to access page number " ULINTPF
" in space " ULINTPF
" space name %s,"
" which is outside the tablespace bounds."
- " Byte offset " ULINTPF ", len " ULINTPF " i/o type " ULINTPF ".",
+ " Byte offset " ULINTPF ", len " ULINTPF
+ " i/o type " ULINTPF ".%s",
block_offset, space_id, space_name,
- byte_offset, len, type);
-
- ib_logf(IB_LOG_LEVEL_FATAL,
- "If you get this error at mysqld startup,"
- " please check that"
- " your my.cnf matches the ibdata files"
- " that you have in the"
- " MySQL server.");
+ byte_offset, len, type,
+ space_id == 0 && !srv_was_started
+ ? "Please check that the configuration matches"
+ " the InnoDB system tablespace location (ibdata files)"
+ : "");
}
/********************************************************************//**
diff --git a/storage/innobase/fsp/fsp0fsp.cc b/storage/innobase/fsp/fsp0fsp.cc
index 98cd11f3369..4f05549bc1c 100644
--- a/storage/innobase/fsp/fsp0fsp.cc
+++ b/storage/innobase/fsp/fsp0fsp.cc
@@ -670,18 +670,13 @@ fsp_header_init_fields(
}
#ifndef UNIV_HOTBACKUP
-/** Initializes the space header of a new created space and creates also the
-insert buffer tree root if space == 0.
+/** Initialize a tablespace header.
@param[in] space_id space id
@param[in] size current size in blocks
-@param[in,out] mtr min-transaction
-@return true on success, otherwise false. */
+@param[in,out] mtr mini-transaction */
UNIV_INTERN
-bool
-fsp_header_init(
- ulint space_id,
- ulint size,
- mtr_t* mtr)
+void
+fsp_header_init(ulint space_id, ulint size, mtr_t* mtr)
{
fsp_header_t* header;
buf_block_t* block;
@@ -725,17 +720,7 @@ fsp_header_init(
mlog_write_ull(header + FSP_SEG_ID, 1, mtr);
- if (space_id == 0) {
- fsp_fill_free_list(FALSE, space_id, header, mtr);
-
- if (btr_create(DICT_CLUSTERED | DICT_UNIVERSAL | DICT_IBUF,
- 0, 0, DICT_IBUF_ID_MIN + space_id,
- dict_ind_redundant, mtr) == FIL_NULL) {
- return (false);
- }
- } else {
- fsp_fill_free_list(TRUE, space_id, header, mtr);
- }
+ fsp_fill_free_list(space_id != TRX_SYS_SPACE, space_id, header, mtr);
fil_space_t* space = fil_space_acquire(space_id);
ut_ad(space);
@@ -745,8 +730,6 @@ fsp_header_init(
}
fil_space_release(space);
-
- return (true);
}
#endif /* !UNIV_HOTBACKUP */
@@ -2065,10 +2048,6 @@ fseg_create_general(
success = fsp_reserve_free_extents(&n_reserved, space, 2,
FSP_NORMAL, mtr);
if (!success) {
- ib_logf(IB_LOG_LEVEL_ERROR,
- "Reserving %d free extents failed"
- " could reserve only " ULINTPF " extents.",
- 2, n_reserved);
return(NULL);
}
}
@@ -2078,9 +2057,6 @@ fseg_create_general(
inode = fsp_alloc_seg_inode(space_header, mtr);
if (inode == NULL) {
- ib_logf(IB_LOG_LEVEL_ERROR,
- "Allocation of a new file segment inode page failed.");
-
goto funct_exit;
}
@@ -2109,9 +2085,6 @@ fseg_create_general(
inode, 0, FSP_UP, mtr, mtr);
if (block == NULL) {
- ib_logf(IB_LOG_LEVEL_ERROR,
- "Allocation of a free page from space " ULINTPF " failed.",
- space);
fsp_free_seg_inode(space, zip_size, inode, mtr);
diff --git a/storage/innobase/handler/ha_innodb.cc b/storage/innobase/handler/ha_innodb.cc
index b68a96c8846..012539d1ace 100644
--- a/storage/innobase/handler/ha_innodb.cc
+++ b/storage/innobase/handler/ha_innodb.cc
@@ -3353,7 +3353,6 @@ innobase_init(
char *default_path;
uint format_id;
ulong num_pll_degree;
- ulint min_size = 0;
DBUG_ENTER("innobase_init");
handlerton *innobase_hton= (handlerton*) p;
@@ -3564,19 +3563,6 @@ mem_free_and_error:
goto error;
}
- /* All doublewrite buffer pages must fit to first system
- datafile and first datafile must be at least 3M. */
- min_size = ut_max((3*1024*1024U), (192U*UNIV_PAGE_SIZE));
-
- if ((srv_data_file_sizes[0]*1024*1024) < min_size) {
- sql_print_error(
- "InnoDB: first datafile is too small current=" ULINTPF
- "M it should be at least " ULINTPF "M.",
- srv_data_file_sizes[0],
- min_size / (1024 * 1024));
- goto mem_free_and_error;
- }
-
/* -------------- All log files ---------------------------*/
/* The default dir for log files is the datadir of MySQL */
diff --git a/storage/innobase/include/buf0dblwr.h b/storage/innobase/include/buf0dblwr.h
index 8e1b00db83c..7b7464761cc 100644
--- a/storage/innobase/include/buf0dblwr.h
+++ b/storage/innobase/include/buf0dblwr.h
@@ -1,7 +1,7 @@
/*****************************************************************************
Copyright (c) 1995, 2017, Oracle and/or its affiliates. All Rights Reserved.
-Copyright (c) 2017, MariaDB Corporation. All Rights Reserved.
+Copyright (c) 2017, MariaDB Corporation.
This program is free software; you can redistribute it and/or modify it under
the terms of the GNU General Public License as published by the Free Software
@@ -39,13 +39,15 @@ extern buf_dblwr_t* buf_dblwr;
/** Set to TRUE when the doublewrite buffer is being created */
extern ibool buf_dblwr_being_created;
-/****************************************************************//**
-Creates the doublewrite buffer to a new InnoDB installation. The header of the
-doublewrite buffer is placed on the trx system header page. */
+/** Create the doublewrite buffer if the doublewrite buffer header
+is not present in the TRX_SYS page.
+@return whether the operation succeeded
+@retval true if the doublewrite buffer exists or was created
+@retval false if the creation failed (too small first data file) */
UNIV_INTERN
-void
-buf_dblwr_create(void);
-/*==================*/
+bool
+buf_dblwr_create()
+ MY_ATTRIBUTE((warn_unused_result));
/****************************************************************//**
At a database startup initializes the doublewrite buffer memory structure if
diff --git a/storage/innobase/include/fsp0fsp.h b/storage/innobase/include/fsp0fsp.h
index 905e98cc1e6..d8d044ba2ec 100644
--- a/storage/innobase/include/fsp0fsp.h
+++ b/storage/innobase/include/fsp0fsp.h
@@ -520,19 +520,13 @@ fsp_header_init_fields(
ulint space_id, /*!< in: space id */
ulint flags); /*!< in: tablespace flags (FSP_SPACE_FLAGS):
0, or table->flags if newer than COMPACT */
-/** Initializes the space header of a new created space and creates also the
-insert buffer tree root if space == 0.
+/** Initialize a tablespace header.
@param[in] space_id space id
@param[in] size current size in blocks
-@param[in,out] mtr min-transaction
-@return true on success, otherwise false. */
+@param[in,out] mtr mini-transaction */
UNIV_INTERN
-bool
-fsp_header_init(
- ulint space_id,
- ulint size,
- mtr_t* mtr)
- MY_ATTRIBUTE((warn_unused_result));
+void
+fsp_header_init(ulint space_id, ulint size, mtr_t* mtr);
/**********************************************************************//**
Increases the space size field of a space. */
diff --git a/storage/innobase/include/trx0rseg.h b/storage/innobase/include/trx0rseg.h
index 185b05876b4..2fe5df14cee 100644
--- a/storage/innobase/include/trx0rseg.h
+++ b/storage/innobase/include/trx0rseg.h
@@ -1,6 +1,7 @@
/*****************************************************************************
Copyright (c) 1996, 2011, Oracle and/or its affiliates. All Rights Reserved.
+Copyright (c) 2017, MariaDB Corporation.
This program is free software; you can redistribute it and/or modify it under
the terms of the GNU General Public License as published by the Free Software
@@ -124,13 +125,13 @@ trx_rseg_mem_free(
/*==============*/
trx_rseg_t* rseg); /*!< in, own: instance to free */
-/*********************************************************************
-Creates a rollback segment. */
+/** Create a rollback segment.
+@param[in] space undo tablespace ID
+@return pointer to new rollback segment
+@retval NULL on failure */
UNIV_INTERN
trx_rseg_t*
-trx_rseg_create(
-/*============*/
- ulint space); /*!< in: id of UNDO tablespace */
+trx_rseg_create(ulint space);
/********************************************************************
Get the number of unique rollback tablespaces in use except space id 0.
diff --git a/storage/innobase/log/log0recv.cc b/storage/innobase/log/log0recv.cc
index bf00cd8e8d9..4e6e66e808e 100644
--- a/storage/innobase/log/log0recv.cc
+++ b/storage/innobase/log/log0recv.cc
@@ -2914,11 +2914,6 @@ recv_init_crash_recovery(void)
possible */
if (srv_force_recovery < SRV_FORCE_NO_LOG_REDO) {
-
- ib_logf(IB_LOG_LEVEL_INFO,
- "Restoring possible half-written data pages "
- "from the doublewrite buffer...");
-
buf_dblwr_process();
/* Spawn the background thread to flush dirty pages
@@ -3179,24 +3174,6 @@ recv_recovery_from_checkpoint_start_func(
user about recovery: */
if (checkpoint_lsn != flushed_lsn) {
-
- if (checkpoint_lsn < flushed_lsn) {
-
- ib_logf(IB_LOG_LEVEL_WARN,
- "The log sequence number "
- "in the ibdata files is higher "
- "than the log sequence number "
- "in the ib_logfiles! Are you sure "
- "you are using the right "
- "ib_logfiles to start up the database. "
- "Log sequence number in the "
- "ib_logfiles is " LSN_PF ", log"
- "sequence number stamped "
- "to ibdata file header is " LSN_PF ".",
- checkpoint_lsn,
- flushed_lsn);
- }
-
if (!recv_needed_recovery) {
ib_logf(IB_LOG_LEVEL_INFO,
"The log sequence number "
diff --git a/storage/innobase/row/row0mysql.cc b/storage/innobase/row/row0mysql.cc
index 6ca9443dc7d..82938995e93 100644
--- a/storage/innobase/row/row0mysql.cc
+++ b/storage/innobase/row/row0mysql.cc
@@ -3590,15 +3590,9 @@ row_truncate_table_for_mysql(
} while (index);
mtr_start_trx(&mtr, trx);
- bool ret = fsp_header_init(space_id,
+ fsp_header_init(space_id,
FIL_IBD_FILE_INITIAL_SIZE, &mtr);
mtr_commit(&mtr);
-
- if (!ret) {
- table->file_unreadable = true;
- err = DB_ERROR;
- goto funct_exit;
- }
}
} else {
/* Lock all index trees for this table, as we will
diff --git a/storage/innobase/srv/srv0start.cc b/storage/innobase/srv/srv0start.cc
index 4e59ddd1f87..c7026b4f04a 100644
--- a/storage/innobase/srv/srv0start.cc
+++ b/storage/innobase/srv/srv0start.cc
@@ -1561,26 +1561,18 @@ srv_undo_tablespaces_init(
if (create_new_db) {
mtr_t mtr;
- bool ret=true;
mtr_start(&mtr);
/* The undo log tablespace */
for (i = 0; i < n_undo_tablespaces; ++i) {
- ret = fsp_header_init(
+ fsp_header_init(
undo_tablespace_ids[i],
SRV_UNDO_TABLESPACE_SIZE_IN_PAGES, &mtr);
- if (!ret) {
- break;
- }
}
mtr_commit(&mtr);
-
- if (!ret) {
- return (srv_init_abort(create_new_db, __FILE__, __LINE__, DB_ERROR));
- }
}
return(DB_SUCCESS);
@@ -2419,14 +2411,24 @@ files_checked:
mtr_start(&mtr);
- bool ret = fsp_header_init(0, sum_of_new_sizes, &mtr);
+ fsp_header_init(0, sum_of_new_sizes, &mtr);
+ compile_time_assert(TRX_SYS_SPACE == 0);
+ compile_time_assert(IBUF_SPACE_ID == 0);
+
+ ulint ibuf_root = btr_create(
+ DICT_CLUSTERED | DICT_UNIVERSAL | DICT_IBUF,
+ 0, 0, DICT_IBUF_ID_MIN,
+ dict_ind_redundant, &mtr);
mtr_commit(&mtr);
- if (!ret) {
- return (srv_init_abort(create_new_db, __FILE__, __LINE__, DB_ERROR));
+ if (ibuf_root == FIL_NULL) {
+ return(srv_init_abort(true, __FILE__, __LINE__,
+ DB_ERROR));
}
+ ut_ad(ibuf_root == IBUF_TREE_ROOT_PAGE_NO);
+
/* To maintain backward compatibility we create only
the first rollback segment before the double write buffer.
All the remaining rollback segments will be created later,
@@ -2812,10 +2814,9 @@ files_checked:
/* fprintf(stderr, "Max allowed record size %lu\n",
page_get_free_space_of_empty() / 2); */
- if (buf_dblwr == NULL) {
- /* Create the doublewrite buffer to a new tablespace */
-
- buf_dblwr_create();
+ if (!buf_dblwr_create()) {
+ return(srv_init_abort(create_new_db, __FILE__, __LINE__,
+ DB_ERROR));
}
/* Here the double write buffer has already been created and so
diff --git a/storage/innobase/trx/trx0rseg.cc b/storage/innobase/trx/trx0rseg.cc
index 21dbab98e48..16fa334872b 100644
--- a/storage/innobase/trx/trx0rseg.cc
+++ b/storage/innobase/trx/trx0rseg.cc
@@ -1,6 +1,7 @@
/*****************************************************************************
Copyright (c) 1996, 2011, Oracle and/or its affiliates. All Rights Reserved.
+Copyright (c) 2017, MariaDB Corporation.
This program is free software; you can redistribute it and/or modify it under
the terms of the GNU General Public License as published by the Free Software
@@ -293,14 +294,13 @@ trx_rseg_create_instance(
}
}
-/*********************************************************************
-Creates a rollback segment.
-@return pointer to new rollback segment if create successful */
+/** Create a rollback segment.
+@param[in] space undo tablespace ID
+@return pointer to new rollback segment
+@retval NULL on failure */
UNIV_INTERN
trx_rseg_t*
-trx_rseg_create(
-/*============*/
- ulint space) /*!< in: id of UNDO tablespace */
+trx_rseg_create(ulint space)
{
mtr_t mtr;
ulint slot_no;
@@ -323,25 +323,21 @@ trx_rseg_create(
page_no = trx_rseg_header_create(
space, 0, ULINT_MAX, slot_no, &mtr);
- if (page_no == FIL_NULL) {
- mtr_commit(&mtr);
- return (rseg);
- }
-
- sys_header = trx_sysf_get(&mtr);
+ if (page_no != FIL_NULL) {
+ sys_header = trx_sysf_get(&mtr);
- id = trx_sysf_rseg_get_space(sys_header, slot_no, &mtr);
- ut_a(id == space);
+ id = trx_sysf_rseg_get_space(sys_header, slot_no, &mtr);
+ ut_a(id == space);
- zip_size = space ? fil_space_get_zip_size(space) : 0;
+ zip_size = space ? fil_space_get_zip_size(space) : 0;
- rseg = trx_rseg_mem_create(
- slot_no, space, zip_size, page_no,
- purge_sys->ib_bh, &mtr);
+ rseg = trx_rseg_mem_create(
+ slot_no, space, zip_size, page_no,
+ purge_sys->ib_bh, &mtr);
+ }
}
mtr_commit(&mtr);
-
return(rseg);
}
--
cgit v1.2.1