database/mariadb
1
0
Fork 0
mirror of https://github.com/MariaDB/server.git synced 2025-08-30 11:22:14 +03:00
Code Activity
Files
621db3d42ef19d8ff28da9968fcc47d952345a28
mariadb/storage/innodb_plugin/include/page0page.ic
Marko Mäkelä f77329ace9 Bug#13721257 RACE CONDITION IN UPDATES OR INSERTS OF WIDE RECORDS 
This bug was originally filed and fixed as Bug#12612184. The original
fix was buggy, and it was patched by Bug#12704861. Also that patch was
buggy (potentially breaking crash recovery), and both fixes were
reverted.

This fix was not ported to the built-in InnoDB of MySQL 5.1, because
the function signatures of many core functions are different from
InnoDB Plugin and later versions. The block allocation routines and
their callers would have to changed so that they handle block
descriptors instead of page frames.

When a record is updated so that its size grows, non-updated columns
can be selected for external (off-page) storage. The bug is that the
initially inserted updated record contains an all-zero BLOB pointer to
the field that was not updated. Only after the BLOB pages have been
allocated and written, the valid pointer can be written to the record.

Between the release of the page latch in mtr_commit(mtr) after
btr_cur_pessimistic_update() and the re-latching of the page in
btr_pcur_restore_position(), other threads can see the invalid BLOB
pointer consisting of 20 zero bytes. Moreover, if the system crashes
at this point, the situation could persist after crash recovery, and
the contents of the non-updated column would be permanently lost.

The problem is amplified by the ROW_FORMAT=DYNAMIC and
ROW_FORMAT=COMPRESSED that were introduced in
innodb_file_format=barracuda in InnoDB Plugin, but the bug does exist
in all InnoDB versions.

The fix is as follows. After a pessimistic B-tree operation that needs
to write out off-page columns, allocate the pages for these columns in
the mini-transaction that performed the B-tree operation (btr_mtr),
but write the pages in a separate mini-transaction (blob_mtr). Do
mtr_commit(blob_mtr) before mtr_commit(btr_mtr). A quirk: Do not reuse
pages that were previously freed in btr_mtr. Only write the off-page
columns to 'fresh' pages.

In this way, crash recovery will see redo log entries for blob_mtr
before any redo log entry for btr_mtr. It will apply the BLOB page
writes to pages that were marked free at that point. If crash recovery
fails to see all of the btr_mtr redo log, there will be some
unreachable BLOB data in free pages, but the B-tree will be in a
consistent state.

btr_page_alloc_low(): Renamed from btr_page_alloc(). Add the parameter
init_mtr. Return an allocated block, or NULL. If init_mtr!=mtr but
the page was already X-latched in mtr, do not initialize the page.

btr_page_alloc(): Wrapper for btr_page_alloc_for_ibuf() and
btr_page_alloc_low().

btr_page_free(): Add a debug assertion that the page was a B-tree page.

btr_lift_page_up(): Return the father block.

btr_compress(), btr_cur_compress_if_useful(): Add the parameter ibool
adjust, for adjusting the cursor position.

btr_cur_pessimistic_update(): Preserve the cursor position when
big_rec will be written and the new flag BTR_KEEP_POS_FLAG is defined.
Remove a duplicate rec_get_offsets() call. Keep the X-latch on
index->lock when big_rec is needed.

btr_store_big_rec_extern_fields(): Replace update_inplace with
an operation code, and local_mtr with btr_mtr. When not doing a
fresh insert and btr_mtr has freed pages, put aside any pages that
were previously X-latched in btr_mtr, and free the pages after
writing out all data. The data must be written to 'fresh' pages,
because btr_mtr will be committed and written to the redo log after
the BLOB writes have been written to the redo log.

btr_blob_op_is_update(): Check if an operation passed to
btr_store_big_rec_extern_fields() is an update or insert-by-update.

fseg_alloc_free_page_low(), fsp_alloc_free_page(),
fseg_alloc_free_extent(), fseg_alloc_free_page_general(): Add the
parameter init_mtr. Return an allocated block, or NULL. If
init_mtr!=mtr but the page was already X-latched in mtr, do not
initialize the page.

xdes_get_descriptor_with_space_hdr(): Assert that the file space
header is being X-latched.

fsp_alloc_from_free_frag(): Refactored from fsp_alloc_free_page().

fsp_page_create(): New function, for allocating, X-latching and
potentially initializing a page. If init_mtr!=mtr but the page was
already X-latched in mtr, do not initialize the page.

fsp_free_page(): Add ut_ad(0) to the error outcomes.

fsp_free_page(), fseg_free_page_low(): Increment mtr->n_freed_pages.

fsp_alloc_seg_inode_page(), fseg_create_general(): Assert that the
page was not previously X-latched in the mini-transaction. A file
segment or inode page should never be allocated in the middle of an
mini-transaction that frees pages, such as btr_cur_pessimistic_delete().

fseg_alloc_free_page_low(): If the hinted page was allocated, skip the
check if the tablespace should be extended. Return NULL instead of
FIL_NULL on failure. Remove the flag frag_page_allocated. Instead,
return directly, because the page would already have been initialized.

fseg_find_free_frag_page_slot() would return ULINT_UNDEFINED on error,
not FIL_NULL. Correct a bogus assertion.

fseg_alloc_free_page(): Redefine as a wrapper macro around
fseg_alloc_free_page_general().

buf_block_buf_fix_inc(): Move the definition from the buf0buf.ic to
buf0buf.h, so that it can be called from other modules.

mtr_t: Add n_freed_pages (number of pages that have been freed).

page_rec_get_nth_const(), page_rec_get_nth(): The inverse function of
page_rec_get_n_recs_before(), get the nth record of the record
list. This is faster than iterating the linked list. Refactored from
page_get_middle_rec().

trx_undo_rec_copy(): Add a debug assertion for the length.

trx_undo_add_page(): Return a block descriptor or NULL instead of a
page number or FIL_NULL.

trx_undo_report_row_operation(): Add debug assertions.

trx_sys_create_doublewrite_buf(): Assert that each page was not
previously X-latched.

page_cur_insert_rec_zip_reorg(): Make use of page_rec_get_nth().

row_ins_clust_index_entry_by_modify(): Pass BTR_KEEP_POS_FLAG, so that
the repositioning of the cursor can be avoided.

row_ins_index_entry_low(): Add DEBUG_SYNC points before and after
writing off-page columns. If inserting by updating a delete-marked
record, do not reposition the cursor or commit the mini-transaction
before writing the off-page columns.

row_build(): Tighten a debug assertion about null BLOB pointers.

row_upd_clust_rec(): Add DEBUG_SYNC points before and after writing
off-page columns. Do not reposition the cursor or commit the
mini-transaction before writing the off-page columns.

rb:939 approved by Jimmy Yang
2012-02-17 11:42:04 +02:00

1104 lines
30 KiB
Plaintext
Raw Blame History

/*****************************************************************************
Copyright (c) 1994, 2012, Oracle and/or its affiliates. All Rights Reserved.
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
Foundation; version 2 of the License.
This program is distributed in the hope that it will be useful, but WITHOUT
ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
You should have received a copy of the GNU General Public License along with
this program; if not, write to the Free Software Foundation, Inc.,
51 Franklin Street, Suite 500, Boston, MA 02110-1335 USA
*****************************************************************************/
/**************************************************//**
@file include/page0page.ic
Index page routines
Created 2/2/1994 Heikki Tuuri
*******************************************************/
#include "mach0data.h"
#ifdef UNIV_DEBUG
# include "log0recv.h"
#endif /* !UNIV_DEBUG */
#ifndef UNIV_HOTBACKUP
# include "rem0cmp.h"
#endif /* !UNIV_HOTBACKUP */
#include "mtr0log.h"
#include "page0zip.h"
#ifdef UNIV_MATERIALIZE
#undef UNIV_INLINE
#define UNIV_INLINE
#endif
/************************************************************//**
Gets the start of a page.
@return start of the page */
UNIV_INLINE
page_t*
page_align(
/*=======*/
const void* ptr) /*!< in: pointer to page frame */
{
return((page_t*) ut_align_down(ptr, UNIV_PAGE_SIZE));
}
/************************************************************//**
Gets the offset within a page.
@return offset from the start of the page */
UNIV_INLINE
ulint
page_offset(
/*========*/
const void* ptr) /*!< in: pointer to page frame */
{
return(ut_align_offset(ptr, UNIV_PAGE_SIZE));
}
/*************************************************************//**
Returns the max trx id field value. */
UNIV_INLINE
trx_id_t
page_get_max_trx_id(
/*================*/
const page_t* page) /*!< in: page */
{
ut_ad(page);
return(mach_read_from_8(page + PAGE_HEADER + PAGE_MAX_TRX_ID));
}
/*************************************************************//**
Sets the max trx id field value if trx_id is bigger than the previous
value. */
UNIV_INLINE
void
page_update_max_trx_id(
/*===================*/
buf_block_t* block, /*!< in/out: page */
page_zip_des_t* page_zip,/*!< in/out: compressed page whose
uncompressed part will be updated, or NULL */
trx_id_t trx_id, /*!< in: transaction id */
mtr_t* mtr) /*!< in/out: mini-transaction */
{
ut_ad(block);
ut_ad(mtr_memo_contains(mtr, block, MTR_MEMO_PAGE_X_FIX));
/* During crash recovery, this function may be called on
something else than a leaf page of a secondary index or the
insert buffer index tree (dict_index_is_sec_or_ibuf() returns
TRUE for the dummy indexes constructed during redo log
application). In that case, PAGE_MAX_TRX_ID is unused,
and trx_id is usually zero. */
ut_ad(!ut_dulint_is_zero(trx_id) || recv_recovery_is_on());
ut_ad(page_is_leaf(buf_block_get_frame(block)));
if (ut_dulint_cmp(page_get_max_trx_id(buf_block_get_frame(block)),
trx_id) < 0) {
page_set_max_trx_id(block, page_zip, trx_id, mtr);
}
}
/*************************************************************//**
Reads the given header field. */
UNIV_INLINE
ulint
page_header_get_field(
/*==================*/
const page_t* page, /*!< in: page */
ulint field) /*!< in: PAGE_LEVEL, ... */
{
ut_ad(page);
ut_ad(field <= PAGE_INDEX_ID);
return(mach_read_from_2(page + PAGE_HEADER + field));
}
/*************************************************************//**
Sets the given header field. */
UNIV_INLINE
void
page_header_set_field(
/*==================*/
page_t* page, /*!< in/out: page */
page_zip_des_t* page_zip,/*!< in/out: compressed page whose
uncompressed part will be updated, or NULL */
ulint field, /*!< in: PAGE_N_DIR_SLOTS, ... */
ulint val) /*!< in: value */
{
ut_ad(page);
ut_ad(field <= PAGE_N_RECS);
ut_ad(field == PAGE_N_HEAP || val < UNIV_PAGE_SIZE);
ut_ad(field != PAGE_N_HEAP || (val & 0x7fff) < UNIV_PAGE_SIZE);
mach_write_to_2(page + PAGE_HEADER + field, val);
if (UNIV_LIKELY_NULL(page_zip)) {
page_zip_write_header(page_zip,
page + PAGE_HEADER + field, 2, NULL);
}
}
/*************************************************************//**
Returns the offset stored in the given header field.
@return offset from the start of the page, or 0 */
UNIV_INLINE
ulint
page_header_get_offs(
/*=================*/
const page_t* page, /*!< in: page */
ulint field) /*!< in: PAGE_FREE, ... */
{
ulint offs;
ut_ad(page);
ut_ad((field == PAGE_FREE)
|| (field == PAGE_LAST_INSERT)
|| (field == PAGE_HEAP_TOP));
offs = page_header_get_field(page, field);
ut_ad((field != PAGE_HEAP_TOP) || offs);
return(offs);
}
/*************************************************************//**
Sets the pointer stored in the given header field. */
UNIV_INLINE
void
page_header_set_ptr(
/*================*/
page_t* page, /*!< in: page */
page_zip_des_t* page_zip,/*!< in/out: compressed page whose
uncompressed part will be updated, or NULL */
ulint field, /*!< in: PAGE_FREE, ... */
const byte* ptr) /*!< in: pointer or NULL*/
{
ulint offs;
ut_ad(page);
ut_ad((field == PAGE_FREE)
|| (field == PAGE_LAST_INSERT)
|| (field == PAGE_HEAP_TOP));
if (ptr == NULL) {
offs = 0;
} else {
offs = ptr - page;
}
ut_ad((field != PAGE_HEAP_TOP) || offs);
page_header_set_field(page, page_zip, field, offs);
}
#ifndef UNIV_HOTBACKUP
/*************************************************************//**
Resets the last insert info field in the page header. Writes to mlog
about this operation. */
UNIV_INLINE
void
page_header_reset_last_insert(
/*==========================*/
page_t* page, /*!< in/out: page */
page_zip_des_t* page_zip,/*!< in/out: compressed page whose
uncompressed part will be updated, or NULL */
mtr_t* mtr) /*!< in: mtr */
{
ut_ad(page && mtr);
if (UNIV_LIKELY_NULL(page_zip)) {
mach_write_to_2(page + (PAGE_HEADER + PAGE_LAST_INSERT), 0);
page_zip_write_header(page_zip,
page + (PAGE_HEADER + PAGE_LAST_INSERT),
2, mtr);
} else {
mlog_write_ulint(page + (PAGE_HEADER + PAGE_LAST_INSERT), 0,
MLOG_2BYTES, mtr);
}
}
#endif /* !UNIV_HOTBACKUP */
/************************************************************//**
Determine whether the page is in new-style compact format.
@return nonzero if the page is in compact format, zero if it is in
old-style format */
UNIV_INLINE
ulint
page_is_comp(
/*=========*/
const page_t* page) /*!< in: index page */
{
return(UNIV_EXPECT(page_header_get_field(page, PAGE_N_HEAP) & 0x8000,
0x8000));
}
/************************************************************//**
TRUE if the record is on a page in compact format.
@return nonzero if in compact format */
UNIV_INLINE
ulint
page_rec_is_comp(
/*=============*/
const rec_t* rec) /*!< in: record */
{
return(page_is_comp(page_align(rec)));
}
/***************************************************************//**
Returns the heap number of a record.
@return heap number */
UNIV_INLINE
ulint
page_rec_get_heap_no(
/*=================*/
const rec_t* rec) /*!< in: the physical record */
{
if (page_rec_is_comp(rec)) {
return(rec_get_heap_no_new(rec));
} else {
return(rec_get_heap_no_old(rec));
}
}
/************************************************************//**
Determine whether the page is a B-tree leaf.
@return TRUE if the page is a B-tree leaf */
UNIV_INLINE
ibool
page_is_leaf(
/*=========*/
const page_t* page) /*!< in: page */
{
return(!*(const uint16*) (page + (PAGE_HEADER + PAGE_LEVEL)));
}
/************************************************************//**
Gets the offset of the first record on the page.
@return offset of the first record in record list, relative from page */
UNIV_INLINE
ulint
page_get_infimum_offset(
/*====================*/
const page_t* page) /*!< in: page which must have record(s) */
{
ut_ad(page);
ut_ad(!page_offset(page));
if (page_is_comp(page)) {
return(PAGE_NEW_INFIMUM);
} else {
return(PAGE_OLD_INFIMUM);
}
}
/************************************************************//**
Gets the offset of the last record on the page.
@return offset of the last record in record list, relative from page */
UNIV_INLINE
ulint
page_get_supremum_offset(
/*=====================*/
const page_t* page) /*!< in: page which must have record(s) */
{
ut_ad(page);
ut_ad(!page_offset(page));
if (page_is_comp(page)) {
return(PAGE_NEW_SUPREMUM);
} else {
return(PAGE_OLD_SUPREMUM);
}
}
/************************************************************//**
TRUE if the record is a user record on the page.
@return TRUE if a user record */
UNIV_INLINE
ibool
page_rec_is_user_rec_low(
/*=====================*/
ulint offset) /*!< in: record offset on page */
{
ut_ad(offset >= PAGE_NEW_INFIMUM);
#if PAGE_OLD_INFIMUM < PAGE_NEW_INFIMUM
# error "PAGE_OLD_INFIMUM < PAGE_NEW_INFIMUM"
#endif
#if PAGE_OLD_SUPREMUM < PAGE_NEW_SUPREMUM
# error "PAGE_OLD_SUPREMUM < PAGE_NEW_SUPREMUM"
#endif
#if PAGE_NEW_INFIMUM > PAGE_OLD_SUPREMUM
# error "PAGE_NEW_INFIMUM > PAGE_OLD_SUPREMUM"
#endif
#if PAGE_OLD_INFIMUM > PAGE_NEW_SUPREMUM
# error "PAGE_OLD_INFIMUM > PAGE_NEW_SUPREMUM"
#endif
#if PAGE_NEW_SUPREMUM > PAGE_OLD_SUPREMUM_END
# error "PAGE_NEW_SUPREMUM > PAGE_OLD_SUPREMUM_END"
#endif
#if PAGE_OLD_SUPREMUM > PAGE_NEW_SUPREMUM_END
# error "PAGE_OLD_SUPREMUM > PAGE_NEW_SUPREMUM_END"
#endif
ut_ad(offset <= UNIV_PAGE_SIZE - PAGE_EMPTY_DIR_START);
return(UNIV_LIKELY(offset != PAGE_NEW_SUPREMUM)
&& UNIV_LIKELY(offset != PAGE_NEW_INFIMUM)
&& UNIV_LIKELY(offset != PAGE_OLD_INFIMUM)
&& UNIV_LIKELY(offset != PAGE_OLD_SUPREMUM));
}
/************************************************************//**
TRUE if the record is the supremum record on a page.
@return TRUE if the supremum record */
UNIV_INLINE
ibool
page_rec_is_supremum_low(
/*=====================*/
ulint offset) /*!< in: record offset on page */
{
ut_ad(offset >= PAGE_NEW_INFIMUM);
ut_ad(offset <= UNIV_PAGE_SIZE - PAGE_EMPTY_DIR_START);
return(UNIV_UNLIKELY(offset == PAGE_NEW_SUPREMUM)
|| UNIV_UNLIKELY(offset == PAGE_OLD_SUPREMUM));
}
/************************************************************//**
TRUE if the record is the infimum record on a page.
@return TRUE if the infimum record */
UNIV_INLINE
ibool
page_rec_is_infimum_low(
/*====================*/
ulint offset) /*!< in: record offset on page */
{
ut_ad(offset >= PAGE_NEW_INFIMUM);
ut_ad(offset <= UNIV_PAGE_SIZE - PAGE_EMPTY_DIR_START);
return(UNIV_UNLIKELY(offset == PAGE_NEW_INFIMUM)
|| UNIV_UNLIKELY(offset == PAGE_OLD_INFIMUM));
}
/************************************************************//**
TRUE if the record is a user record on the page.
@return TRUE if a user record */
UNIV_INLINE
ibool
page_rec_is_user_rec(
/*=================*/
const rec_t* rec) /*!< in: record */
{
return(page_rec_is_user_rec_low(page_offset(rec)));
}
/************************************************************//**
TRUE if the record is the supremum record on a page.
@return TRUE if the supremum record */
UNIV_INLINE
ibool
page_rec_is_supremum(
/*=================*/
const rec_t* rec) /*!< in: record */
{
return(page_rec_is_supremum_low(page_offset(rec)));
}
/************************************************************//**
TRUE if the record is the infimum record on a page.
@return TRUE if the infimum record */
UNIV_INLINE
ibool
page_rec_is_infimum(
/*================*/
const rec_t* rec) /*!< in: record */
{
return(page_rec_is_infimum_low(page_offset(rec)));
}
/************************************************************//**
Returns the nth record of the record list.
This is the inverse function of page_rec_get_n_recs_before().
@return nth record */
UNIV_INLINE
rec_t*
page_rec_get_nth(
/*=============*/
page_t* page, /*!< in: page */
ulint nth) /*!< in: nth record */
{
return((rec_t*) page_rec_get_nth_const(page, nth));
}
#ifndef UNIV_HOTBACKUP
/************************************************************//**
Returns the middle record of the records on the page. If there is an
even number of records in the list, returns the first record of the
upper half-list.
@return middle record */
UNIV_INLINE
rec_t*
page_get_middle_rec(
/*================*/
page_t* page) /*!< in: page */
{
ulint middle = (page_get_n_recs(page) + PAGE_HEAP_NO_USER_LOW) / 2;
return(page_rec_get_nth(page, middle));
}
/*************************************************************//**
Compares a data tuple to a physical record. Differs from the function
cmp_dtuple_rec_with_match in the way that the record must reside on an
index page, and also page infimum and supremum records can be given in
the parameter rec. These are considered as the negative infinity and
the positive infinity in the alphabetical order.
@return 1, 0, -1, if dtuple is greater, equal, less than rec,
respectively, when only the common first fields are compared */
UNIV_INLINE
int
page_cmp_dtuple_rec_with_match(
/*===========================*/
const dtuple_t* dtuple, /*!< in: data tuple */
const rec_t* rec, /*!< in: physical record on a page; may also
be page infimum or supremum, in which case
matched-parameter values below are not
affected */
const ulint* offsets,/*!< in: array returned by rec_get_offsets() */
ulint* matched_fields, /*!< in/out: number of already completely
matched fields; when function returns
contains the value for current comparison */
ulint* matched_bytes) /*!< in/out: number of already matched
bytes within the first field not completely
matched; when function returns contains the
value for current comparison */
{
ulint rec_offset;
ut_ad(dtuple_check_typed(dtuple));
ut_ad(rec_offs_validate(rec, NULL, offsets));
ut_ad(!rec_offs_comp(offsets) == !page_rec_is_comp(rec));
rec_offset = page_offset(rec);
if (UNIV_UNLIKELY(rec_offset == PAGE_NEW_INFIMUM)
|| UNIV_UNLIKELY(rec_offset == PAGE_OLD_INFIMUM)) {
return(1);
}
if (UNIV_UNLIKELY(rec_offset == PAGE_NEW_SUPREMUM)
|| UNIV_UNLIKELY(rec_offset == PAGE_OLD_SUPREMUM)) {
return(-1);
}
return(cmp_dtuple_rec_with_match(dtuple, rec, offsets,
matched_fields,
matched_bytes));
}
#endif /* !UNIV_HOTBACKUP */
/*************************************************************//**
Gets the page number.
@return page number */
UNIV_INLINE
ulint
page_get_page_no(
/*=============*/
const page_t* page) /*!< in: page */
{
ut_ad(page == page_align((page_t*) page));
return(mach_read_from_4(page + FIL_PAGE_OFFSET));
}
/*************************************************************//**
Gets the tablespace identifier.
@return space id */
UNIV_INLINE
ulint
page_get_space_id(
/*==============*/
const page_t* page) /*!< in: page */
{
ut_ad(page == page_align((page_t*) page));
return(mach_read_from_4(page + FIL_PAGE_ARCH_LOG_NO_OR_SPACE_ID));
}
/*************************************************************//**
Gets the number of user records on page (infimum and supremum records
are not user records).
@return number of user records */
UNIV_INLINE
ulint
page_get_n_recs(
/*============*/
const page_t* page) /*!< in: index page */
{
return(page_header_get_field(page, PAGE_N_RECS));
}
/*************************************************************//**
Gets the number of dir slots in directory.
@return number of slots */
UNIV_INLINE
ulint
page_dir_get_n_slots(
/*=================*/
const page_t* page) /*!< in: index page */
{
return(page_header_get_field(page, PAGE_N_DIR_SLOTS));
}
/*************************************************************//**
Sets the number of dir slots in directory. */
UNIV_INLINE
void
page_dir_set_n_slots(
/*=================*/
page_t* page, /*!< in/out: page */
page_zip_des_t* page_zip,/*!< in/out: compressed page whose
uncompressed part will be updated, or NULL */
ulint n_slots)/*!< in: number of slots */
{
page_header_set_field(page, page_zip, PAGE_N_DIR_SLOTS, n_slots);
}
/*************************************************************//**
Gets the number of records in the heap.
@return number of user records */
UNIV_INLINE
ulint
page_dir_get_n_heap(
/*================*/
const page_t* page) /*!< in: index page */
{
return(page_header_get_field(page, PAGE_N_HEAP) & 0x7fff);
}
/*************************************************************//**
Sets the number of records in the heap. */
UNIV_INLINE
void
page_dir_set_n_heap(
/*================*/
page_t* page, /*!< in/out: index page */
page_zip_des_t* page_zip,/*!< in/out: compressed page whose
uncompressed part will be updated, or NULL.
Note that the size of the dense page directory
in the compressed page trailer is
n_heap * PAGE_ZIP_DIR_SLOT_SIZE. */
ulint n_heap) /*!< in: number of records */
{
ut_ad(n_heap < 0x8000);
ut_ad(!page_zip || n_heap
== (page_header_get_field(page, PAGE_N_HEAP) & 0x7fff) + 1);
page_header_set_field(page, page_zip, PAGE_N_HEAP, n_heap
| (0x8000
& page_header_get_field(page, PAGE_N_HEAP)));
}
#ifdef UNIV_DEBUG
/*************************************************************//**
Gets pointer to nth directory slot.
@return pointer to dir slot */
UNIV_INLINE
page_dir_slot_t*
page_dir_get_nth_slot(
/*==================*/
const page_t* page, /*!< in: index page */
ulint n) /*!< in: position */
{
ut_ad(page_dir_get_n_slots(page) > n);
return((page_dir_slot_t*)
page + UNIV_PAGE_SIZE - PAGE_DIR
- (n + 1) * PAGE_DIR_SLOT_SIZE);
}
#endif /* UNIV_DEBUG */
/**************************************************************//**
Used to check the consistency of a record on a page.
@return TRUE if succeed */
UNIV_INLINE
ibool
page_rec_check(
/*===========*/
const rec_t* rec) /*!< in: record */
{
const page_t* page = page_align(rec);
ut_a(rec);
ut_a(page_offset(rec) <= page_header_get_field(page, PAGE_HEAP_TOP));
ut_a(page_offset(rec) >= PAGE_DATA);
return(TRUE);
}
/***************************************************************//**
Gets the record pointed to by a directory slot.
@return pointer to record */
UNIV_INLINE
const rec_t*
page_dir_slot_get_rec(
/*==================*/
const page_dir_slot_t* slot) /*!< in: directory slot */
{
return(page_align(slot) + mach_read_from_2(slot));
}
/***************************************************************//**
This is used to set the record offset in a directory slot. */
UNIV_INLINE
void
page_dir_slot_set_rec(
/*==================*/
page_dir_slot_t* slot, /*!< in: directory slot */
rec_t* rec) /*!< in: record on the page */
{
ut_ad(page_rec_check(rec));
mach_write_to_2(slot, page_offset(rec));
}
/***************************************************************//**
Gets the number of records owned by a directory slot.
@return number of records */
UNIV_INLINE
ulint
page_dir_slot_get_n_owned(
/*======================*/
const page_dir_slot_t* slot) /*!< in: page directory slot */
{
const rec_t* rec = page_dir_slot_get_rec(slot);
if (page_rec_is_comp(slot)) {
return(rec_get_n_owned_new(rec));
} else {
return(rec_get_n_owned_old(rec));
}
}
/***************************************************************//**
This is used to set the owned records field of a directory slot. */
UNIV_INLINE
void
page_dir_slot_set_n_owned(
/*======================*/
page_dir_slot_t*slot, /*!< in/out: directory slot */
page_zip_des_t* page_zip,/*!< in/out: compressed page, or NULL */
ulint n) /*!< in: number of records owned by the slot */
{
rec_t* rec = (rec_t*) page_dir_slot_get_rec(slot);
if (page_rec_is_comp(slot)) {
rec_set_n_owned_new(rec, page_zip, n);
} else {
ut_ad(!page_zip);
rec_set_n_owned_old(rec, n);
}
}
/************************************************************//**
Calculates the space reserved for directory slots of a given number of
records. The exact value is a fraction number n * PAGE_DIR_SLOT_SIZE /
PAGE_DIR_SLOT_MIN_N_OWNED, and it is rounded upwards to an integer. */
UNIV_INLINE
ulint
page_dir_calc_reserved_space(
/*=========================*/
ulint n_recs) /*!< in: number of records */
{
return((PAGE_DIR_SLOT_SIZE * n_recs + PAGE_DIR_SLOT_MIN_N_OWNED - 1)
/ PAGE_DIR_SLOT_MIN_N_OWNED);
}
/************************************************************//**
Gets the pointer to the next record on the page.
@return pointer to next record */
UNIV_INLINE
const rec_t*
page_rec_get_next_low(
/*==================*/
const rec_t* rec, /*!< in: pointer to record */
ulint comp) /*!< in: nonzero=compact page layout */
{
ulint offs;
const page_t* page;
ut_ad(page_rec_check(rec));
page = page_align(rec);
offs = rec_get_next_offs(rec, comp);
if (UNIV_UNLIKELY(offs >= UNIV_PAGE_SIZE)) {
fprintf(stderr,
"InnoDB: Next record offset is nonsensical %lu"
" in record at offset %lu\n"
"InnoDB: rec address %p, space id %lu, page %lu\n",
(ulong)offs, (ulong) page_offset(rec),
(void*) rec,
(ulong) page_get_space_id(page),
(ulong) page_get_page_no(page));
buf_page_print(page, 0);
ut_error;
}
if (UNIV_UNLIKELY(offs == 0)) {
return(NULL);
}
return(page + offs);
}
/************************************************************//**
Gets the pointer to the next record on the page.
@return pointer to next record */
UNIV_INLINE
rec_t*
page_rec_get_next(
/*==============*/
rec_t* rec) /*!< in: pointer to record */
{
return((rec_t*) page_rec_get_next_low(rec, page_rec_is_comp(rec)));
}
/************************************************************//**
Gets the pointer to the next record on the page.
@return pointer to next record */
UNIV_INLINE
const rec_t*
page_rec_get_next_const(
/*====================*/
const rec_t* rec) /*!< in: pointer to record */
{
return(page_rec_get_next_low(rec, page_rec_is_comp(rec)));
}
/************************************************************//**
Sets the pointer to the next record on the page. */
UNIV_INLINE
void
page_rec_set_next(
/*==============*/
rec_t* rec, /*!< in: pointer to record,
must not be page supremum */
rec_t* next) /*!< in: pointer to next record,
must not be page infimum */
{
ulint offs;
ut_ad(page_rec_check(rec));
ut_ad(!page_rec_is_supremum(rec));
ut_ad(rec != next);
ut_ad(!next || !page_rec_is_infimum(next));
ut_ad(!next || page_align(rec) == page_align(next));
if (UNIV_LIKELY(next != NULL)) {
offs = page_offset(next);
} else {
offs = 0;
}
if (page_rec_is_comp(rec)) {
rec_set_next_offs_new(rec, offs);
} else {
rec_set_next_offs_old(rec, offs);
}
}
/************************************************************//**
Gets the pointer to the previous record.
@return pointer to previous record */
UNIV_INLINE
const rec_t*
page_rec_get_prev_const(
/*====================*/
const rec_t* rec) /*!< in: pointer to record, must not be page
infimum */
{
const page_dir_slot_t* slot;
ulint slot_no;
const rec_t* rec2;
const rec_t* prev_rec = NULL;
const page_t* page;
ut_ad(page_rec_check(rec));
page = page_align(rec);
ut_ad(!page_rec_is_infimum(rec));
slot_no = page_dir_find_owner_slot(rec);
ut_a(slot_no != 0);
slot = page_dir_get_nth_slot(page, slot_no - 1);
rec2 = page_dir_slot_get_rec(slot);
if (page_is_comp(page)) {
while (rec != rec2) {
prev_rec = rec2;
rec2 = page_rec_get_next_low(rec2, TRUE);
}
} else {
while (rec != rec2) {
prev_rec = rec2;
rec2 = page_rec_get_next_low(rec2, FALSE);
}
}
ut_a(prev_rec);
return(prev_rec);
}
/************************************************************//**
Gets the pointer to the previous record.
@return pointer to previous record */
UNIV_INLINE
rec_t*
page_rec_get_prev(
/*==============*/
rec_t* rec) /*!< in: pointer to record, must not be page
infimum */
{
return((rec_t*) page_rec_get_prev_const(rec));
}
/***************************************************************//**
Looks for the record which owns the given record.
@return the owner record */
UNIV_INLINE
rec_t*
page_rec_find_owner_rec(
/*====================*/
rec_t* rec) /*!< in: the physical record */
{
ut_ad(page_rec_check(rec));
if (page_rec_is_comp(rec)) {
while (rec_get_n_owned_new(rec) == 0) {
rec = page_rec_get_next(rec);
}
} else {
while (rec_get_n_owned_old(rec) == 0) {
rec = page_rec_get_next(rec);
}
}
return(rec);
}
/**********************************************************//**
Returns the base extra size of a physical record. This is the
size of the fixed header, independent of the record size.
@return REC_N_NEW_EXTRA_BYTES or REC_N_OLD_EXTRA_BYTES */
UNIV_INLINE
ulint
page_rec_get_base_extra_size(
/*=========================*/
const rec_t* rec) /*!< in: physical record */
{
#if REC_N_NEW_EXTRA_BYTES + 1 != REC_N_OLD_EXTRA_BYTES
# error "REC_N_NEW_EXTRA_BYTES + 1 != REC_N_OLD_EXTRA_BYTES"
#endif
return(REC_N_NEW_EXTRA_BYTES + (ulint) !page_rec_is_comp(rec));
}
/************************************************************//**
Returns the sum of the sizes of the records in the record list, excluding
the infimum and supremum records.
@return data in bytes */
UNIV_INLINE
ulint
page_get_data_size(
/*===============*/
const page_t* page) /*!< in: index page */
{
ulint ret;
ret = (ulint)(page_header_get_field(page, PAGE_HEAP_TOP)
- (page_is_comp(page)
? PAGE_NEW_SUPREMUM_END
: PAGE_OLD_SUPREMUM_END)
- page_header_get_field(page, PAGE_GARBAGE));
ut_ad(ret < UNIV_PAGE_SIZE);
return(ret);
}
/************************************************************//**
Allocates a block of memory from the free list of an index page. */
UNIV_INLINE
void
page_mem_alloc_free(
/*================*/
page_t* page, /*!< in/out: index page */
page_zip_des_t* page_zip,/*!< in/out: compressed page with enough
space available for inserting the record,
or NULL */
rec_t* next_rec,/*!< in: pointer to the new head of the
free record list */
ulint need) /*!< in: number of bytes allocated */
{
ulint garbage;
#ifdef UNIV_DEBUG
const rec_t* old_rec = page_header_get_ptr(page, PAGE_FREE);
ulint next_offs;
ut_ad(old_rec);
next_offs = rec_get_next_offs(old_rec, page_is_comp(page));
ut_ad(next_rec == (next_offs ? page + next_offs : NULL));
#endif
page_header_set_ptr(page, page_zip, PAGE_FREE, next_rec);
garbage = page_header_get_field(page, PAGE_GARBAGE);
ut_ad(garbage >= need);
page_header_set_field(page, page_zip, PAGE_GARBAGE, garbage - need);
}
/*************************************************************//**
Calculates free space if a page is emptied.
@return free space */
UNIV_INLINE
ulint
page_get_free_space_of_empty(
/*=========================*/
ulint comp) /*!< in: nonzero=compact page layout */
{
if (UNIV_LIKELY(comp)) {
return((ulint)(UNIV_PAGE_SIZE
- PAGE_NEW_SUPREMUM_END
- PAGE_DIR
- 2 * PAGE_DIR_SLOT_SIZE));
}
return((ulint)(UNIV_PAGE_SIZE
- PAGE_OLD_SUPREMUM_END
- PAGE_DIR
- 2 * PAGE_DIR_SLOT_SIZE));
}
/************************************************************//**
Each user record on a page, and also the deleted user records in the heap
takes its size plus the fraction of the dir cell size /
PAGE_DIR_SLOT_MIN_N_OWNED bytes for it. If the sum of these exceeds the
value of page_get_free_space_of_empty, the insert is impossible, otherwise
it is allowed. This function returns the maximum combined size of records
which can be inserted on top of the record heap.
@return maximum combined size for inserted records */
UNIV_INLINE
ulint
page_get_max_insert_size(
/*=====================*/
const page_t* page, /*!< in: index page */
ulint n_recs) /*!< in: number of records */
{
ulint occupied;
ulint free_space;
if (page_is_comp(page)) {
occupied = page_header_get_field(page, PAGE_HEAP_TOP)
- PAGE_NEW_SUPREMUM_END
+ page_dir_calc_reserved_space(
n_recs + page_dir_get_n_heap(page) - 2);
free_space = page_get_free_space_of_empty(TRUE);
} else {
occupied = page_header_get_field(page, PAGE_HEAP_TOP)
- PAGE_OLD_SUPREMUM_END
+ page_dir_calc_reserved_space(
n_recs + page_dir_get_n_heap(page) - 2);
free_space = page_get_free_space_of_empty(FALSE);
}
/* Above the 'n_recs +' part reserves directory space for the new
inserted records; the '- 2' excludes page infimum and supremum
records */
if (occupied > free_space) {
return(0);
}
return(free_space - occupied);
}
/************************************************************//**
Returns the maximum combined size of records which can be inserted on top
of the record heap if a page is first reorganized.
@return maximum combined size for inserted records */
UNIV_INLINE
ulint
page_get_max_insert_size_after_reorganize(
/*======================================*/
const page_t* page, /*!< in: index page */
ulint n_recs) /*!< in: number of records */
{
ulint occupied;
ulint free_space;
occupied = page_get_data_size(page)
+ page_dir_calc_reserved_space(n_recs + page_get_n_recs(page));
free_space = page_get_free_space_of_empty(page_is_comp(page));
if (occupied > free_space) {
return(0);
}
return(free_space - occupied);
}
/************************************************************//**
Puts a record to free list. */
UNIV_INLINE
void
page_mem_free(
/*==========*/
page_t* page, /*!< in/out: index page */
page_zip_des_t* page_zip,/*!< in/out: compressed page, or NULL */
rec_t* rec, /*!< in: pointer to the (origin of) record */
dict_index_t* index, /*!< in: index of rec */
const ulint* offsets)/*!< in: array returned by rec_get_offsets() */
{
rec_t* free;
ulint garbage;
ut_ad(rec_offs_validate(rec, index, offsets));
free = page_header_get_ptr(page, PAGE_FREE);
page_rec_set_next(rec, free);
page_header_set_ptr(page, page_zip, PAGE_FREE, rec);
garbage = page_header_get_field(page, PAGE_GARBAGE);
page_header_set_field(page, page_zip, PAGE_GARBAGE,
garbage + rec_offs_size(offsets));
if (UNIV_LIKELY_NULL(page_zip)) {
page_zip_dir_delete(page_zip, rec, index, offsets, free);
} else {
page_header_set_field(page, page_zip, PAGE_N_RECS,
page_get_n_recs(page) - 1);
}
}
#ifdef UNIV_MATERIALIZE
#undef UNIV_INLINE
#define UNIV_INLINE UNIV_INLINE_ORIGINAL
#endif
View Git Blame Copy Permalink