mariadb/storage/xtradb/row/row0sel.cc

/*****************************************************************************

Copyright (c) 1997, 2013, Oracle and/or its affiliates. All Rights Reserved.
Copyright (c) 2008, Google Inc.

Portions of this file contain modifications contributed and copyrighted by
Google, Inc. Those modifications are gratefully acknowledged and are described
briefly in the InnoDB documentation. The contributions by Google are
incorporated with their permission, and subject to the conditions contained in
the file COPYING.Google.

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 row/row0sel.cc
Select

Created 12/19/1997 Heikki Tuuri
*******************************************************/

#include "row0sel.h"

#ifdef UNIV_NONINL
#include "row0sel.ic"
#endif

#include "dict0dict.h"
#include "dict0boot.h"
#include "trx0undo.h"
#include "trx0trx.h"
#include "btr0btr.h"
#include "btr0cur.h"
#include "btr0sea.h"
#include "mach0data.h"
#include "que0que.h"
#include "row0upd.h"
#include "row0row.h"
#include "row0vers.h"
#include "rem0cmp.h"
#include "lock0lock.h"
#include "eval0eval.h"
#include "pars0sym.h"
#include "pars0pars.h"
#include "row0mysql.h"
#include "read0read.h"
#include "buf0lru.h"
#include "srv0srv.h"
#include "ha_prototypes.h"
#include "srv0start.h"
#include "m_string.h" /* for my_sys.h */
#include "my_sys.h" /* DEBUG_SYNC_C */

#include "my_compare.h" /* enum icp_result */

/* Maximum number of rows to prefetch; MySQL interface has another parameter */
#define SEL_MAX_N_PREFETCH	16

/* Number of rows fetched, after which to start prefetching; MySQL interface
has another parameter */
#define SEL_PREFETCH_LIMIT	1

/* When a select has accessed about this many pages, it returns control back
to que_run_threads: this is to allow canceling runaway queries */

#define SEL_COST_LIMIT	100

/* Flags for search shortcut */
#define SEL_FOUND	0
#define	SEL_EXHAUSTED	1
#define SEL_RETRY	2

/********************************************************************//**
Returns TRUE if the user-defined column in a secondary index record
is alphabetically the same as the corresponding BLOB column in the clustered
index record.
NOTE: the comparison is NOT done as a binary comparison, but character
fields are compared with collation!
@return	TRUE if the columns are equal */
static
ibool
row_sel_sec_rec_is_for_blob(
/*========================*/
	ulint		mtype,		/*!< in: main type */
	ulint		prtype,		/*!< in: precise type */
	ulint		mbminmaxlen,	/*!< in: minimum and maximum length of
					a multi-byte character */
	const byte*	clust_field,	/*!< in: the locally stored part of
					the clustered index column, including
					the BLOB pointer; the clustered
					index record must be covered by
					a lock or a page latch to protect it
					against deletion (rollback or purge) */
	ulint		clust_len,	/*!< in: length of clust_field */
	const byte*	sec_field,	/*!< in: column in secondary index */
	ulint		sec_len,	/*!< in: length of sec_field */
	ulint		prefix_len,	/*!< in: index column prefix length
					in bytes */
	dict_table_t*	table)		/*!< in: table */
{
	ulint	len;
	byte	buf[REC_VERSION_56_MAX_INDEX_COL_LEN];
	ulint	zip_size = dict_tf_get_zip_size(table->flags);

	/* This function should never be invoked on an Antelope format
	table, because they should always contain enough prefix in the
	clustered index record. */
	ut_ad(dict_table_get_format(table) >= UNIV_FORMAT_B);
	ut_a(clust_len >= BTR_EXTERN_FIELD_REF_SIZE);
	ut_ad(prefix_len >= sec_len);
	ut_ad(prefix_len > 0);
	ut_a(prefix_len <= sizeof buf);

	if (UNIV_UNLIKELY
	    (!memcmp(clust_field + clust_len - BTR_EXTERN_FIELD_REF_SIZE,
		     field_ref_zero, BTR_EXTERN_FIELD_REF_SIZE))) {
		/* The externally stored field was not written yet.
		This record should only be seen by
		recv_recovery_rollback_active() or any
		TRX_ISO_READ_UNCOMMITTED transactions. */
		return(FALSE);
	}

	len = btr_copy_externally_stored_field_prefix(buf, prefix_len,
						      zip_size,
						      clust_field, clust_len);

	if (UNIV_UNLIKELY(len == 0)) {
		/* The BLOB was being deleted as the server crashed.
		There should not be any secondary index records
		referring to this clustered index record, because
		btr_free_externally_stored_field() is called after all
		secondary index entries of the row have been purged. */
		return(FALSE);
	}

	len = dtype_get_at_most_n_mbchars(prtype, mbminmaxlen,
					  prefix_len, len, (const char*) buf);

	return(!cmp_data_data(mtype, prtype, buf, len, sec_field, sec_len));
}

/********************************************************************//**
Returns TRUE if the user-defined column values in a secondary index record
are alphabetically the same as the corresponding columns in the clustered
index record.
NOTE: the comparison is NOT done as a binary comparison, but character
fields are compared with collation!
@return TRUE if the secondary record is equal to the corresponding
fields in the clustered record, when compared with collation;
FALSE if not equal or if the clustered record has been marked for deletion */
static
ibool
row_sel_sec_rec_is_for_clust_rec(
/*=============================*/
	const rec_t*	sec_rec,	/*!< in: secondary index record */
	dict_index_t*	sec_index,	/*!< in: secondary index */
	const rec_t*	clust_rec,	/*!< in: clustered index record;
					must be protected by a lock or
					a page latch against deletion
					in rollback or purge */
	dict_index_t*	clust_index)	/*!< in: clustered index */
{
	const byte*	sec_field;
	ulint		sec_len;
	const byte*	clust_field;
	ulint		n;
	ulint		i;
	mem_heap_t*	heap		= NULL;
	ulint		clust_offsets_[REC_OFFS_NORMAL_SIZE];
	ulint		sec_offsets_[REC_OFFS_SMALL_SIZE];
	ulint*		clust_offs	= clust_offsets_;
	ulint*		sec_offs	= sec_offsets_;
	ibool		is_equal	= TRUE;

	rec_offs_init(clust_offsets_);
	rec_offs_init(sec_offsets_);

	if (rec_get_deleted_flag(clust_rec,
				 dict_table_is_comp(clust_index->table))) {

		/* The clustered index record is delete-marked;
		it is not visible in the read view.  Besides,
		if there are any externally stored columns,
		some of them may have already been purged. */
		return(FALSE);
	}

	clust_offs = rec_get_offsets(clust_rec, clust_index, clust_offs,
				     ULINT_UNDEFINED, &heap);
	sec_offs = rec_get_offsets(sec_rec, sec_index, sec_offs,
				   ULINT_UNDEFINED, &heap);

	n = dict_index_get_n_ordering_defined_by_user(sec_index);

	for (i = 0; i < n; i++) {
		const dict_field_t*	ifield;
		const dict_col_t*	col;
		ulint			clust_pos;
		ulint			clust_len;
		ulint			len;

		ifield = dict_index_get_nth_field(sec_index, i);
		col = dict_field_get_col(ifield);
		clust_pos = dict_col_get_clust_pos(col, clust_index);

		clust_field = rec_get_nth_field(
			clust_rec, clust_offs, clust_pos, &clust_len);
		sec_field = rec_get_nth_field(sec_rec, sec_offs, i, &sec_len);

		len = clust_len;

		if (ifield->prefix_len > 0 && len != UNIV_SQL_NULL
		    && sec_len != UNIV_SQL_NULL) {

			if (rec_offs_nth_extern(clust_offs, clust_pos)) {
				len -= BTR_EXTERN_FIELD_REF_SIZE;
			}

			len = dtype_get_at_most_n_mbchars(
				col->prtype, col->mbminmaxlen,
				ifield->prefix_len, len, (char*) clust_field);

			if (rec_offs_nth_extern(clust_offs, clust_pos)
			    && len < sec_len) {
				if (!row_sel_sec_rec_is_for_blob(
					    col->mtype, col->prtype,
					    col->mbminmaxlen,
					    clust_field, clust_len,
					    sec_field, sec_len,
					    ifield->prefix_len,
					    clust_index->table)) {
					goto inequal;
				}

				continue;
			}
		}

		if (0 != cmp_data_data(col->mtype, col->prtype,
				       clust_field, len,
				       sec_field, sec_len)) {
inequal:
			is_equal = FALSE;
			goto func_exit;
		}
	}

func_exit:
	if (UNIV_LIKELY_NULL(heap)) {
		mem_heap_free(heap);
	}
	return(is_equal);
}

/*********************************************************************//**
Creates a select node struct.
@return	own: select node struct */
UNIV_INTERN
sel_node_t*
sel_node_create(
/*============*/
	mem_heap_t*	heap)	/*!< in: memory heap where created */
{
	sel_node_t*	node;

	node = static_cast<sel_node_t*>(
		mem_heap_alloc(heap, sizeof(sel_node_t)));

	node->common.type = QUE_NODE_SELECT;
	node->state = SEL_NODE_OPEN;

	node->plans = NULL;

	return(node);
}

/*********************************************************************//**
Frees the memory private to a select node when a query graph is freed,
does not free the heap where the node was originally created. */
UNIV_INTERN
void
sel_node_free_private(
/*==================*/
	sel_node_t*	node)	/*!< in: select node struct */
{
	ulint	i;
	plan_t*	plan;

	if (node->plans != NULL) {
		for (i = 0; i < node->n_tables; i++) {
			plan = sel_node_get_nth_plan(node, i);

			btr_pcur_close(&(plan->pcur));
			btr_pcur_close(&(plan->clust_pcur));

			if (plan->old_vers_heap) {
				mem_heap_free(plan->old_vers_heap);
			}
		}
	}
}

/*********************************************************************//**
Evaluates the values in a select list. If there are aggregate functions,
their argument value is added to the aggregate total. */
UNIV_INLINE
void
sel_eval_select_list(
/*=================*/
	sel_node_t*	node)	/*!< in: select node */
{
	que_node_t*	exp;

	exp = node->select_list;

	while (exp) {
		eval_exp(exp);

		exp = que_node_get_next(exp);
	}
}

/*********************************************************************//**
Assigns the values in the select list to the possible into-variables in
SELECT ... INTO ... */
UNIV_INLINE
void
sel_assign_into_var_values(
/*=======================*/
	sym_node_t*	var,	/*!< in: first variable in a list of
				variables */
	sel_node_t*	node)	/*!< in: select node */
{
	que_node_t*	exp;

	if (var == NULL) {

		return;
	}

	for (exp = node->select_list;
	     var != 0;
	     var = static_cast<sym_node_t*>(que_node_get_next(var))) {

		ut_ad(exp);

		eval_node_copy_val(var->alias, exp);

		exp = que_node_get_next(exp);
	}
}

/*********************************************************************//**
Resets the aggregate value totals in the select list of an aggregate type
query. */
UNIV_INLINE
void
sel_reset_aggregate_vals(
/*=====================*/
	sel_node_t*	node)	/*!< in: select node */
{
	func_node_t*	func_node;

	ut_ad(node->is_aggregate);

	for (func_node = static_cast<func_node_t*>(node->select_list);
	     func_node != 0;
	     func_node = static_cast<func_node_t*>(
		     	que_node_get_next(func_node))) {

		eval_node_set_int_val(func_node, 0);
	}

	node->aggregate_already_fetched = FALSE;
}

/*********************************************************************//**
Copies the input variable values when an explicit cursor is opened. */
UNIV_INLINE
void
row_sel_copy_input_variable_vals(
/*=============================*/
	sel_node_t*	node)	/*!< in: select node */
{
	sym_node_t*	var;

	var = UT_LIST_GET_FIRST(node->copy_variables);

	while (var) {
		eval_node_copy_val(var, var->alias);

		var->indirection = NULL;

		var = UT_LIST_GET_NEXT(col_var_list, var);
	}
}

/*********************************************************************//**
Fetches the column values from a record. */
static
void
row_sel_fetch_columns(
/*==================*/
	dict_index_t*	index,	/*!< in: record index */
	const rec_t*	rec,	/*!< in: record in a clustered or non-clustered
				index; must be protected by a page latch */
	const ulint*	offsets,/*!< in: rec_get_offsets(rec, index) */
	sym_node_t*	column)	/*!< in: first column in a column list, or
				NULL */
{
	dfield_t*	val;
	ulint		index_type;
	ulint		field_no;
	const byte*	data;
	ulint		len;

	ut_ad(rec_offs_validate(rec, index, offsets));

	if (dict_index_is_clust(index)) {
		index_type = SYM_CLUST_FIELD_NO;
	} else {
		index_type = SYM_SEC_FIELD_NO;
	}

	while (column) {
		mem_heap_t*	heap = NULL;
		ibool		needs_copy;

		field_no = column->field_nos[index_type];

		if (field_no != ULINT_UNDEFINED) {

			if (UNIV_UNLIKELY(rec_offs_nth_extern(offsets,
							      field_no))) {

				/* Copy an externally stored field to the
				temporary heap, if possible. */

				heap = mem_heap_create(1);

				data = btr_rec_copy_externally_stored_field(
					rec, offsets,
					dict_table_zip_size(index->table),
					field_no, &len, heap);

				/* data == NULL means that the
				externally stored field was not
				written yet. This record
				should only be seen by
				recv_recovery_rollback_active() or any
				TRX_ISO_READ_UNCOMMITTED
				transactions. The InnoDB SQL parser
				(the sole caller of this function)
				does not implement READ UNCOMMITTED,
				and it is not involved during rollback. */
				ut_a(data);
				ut_a(len != UNIV_SQL_NULL);

				needs_copy = TRUE;
			} else {
				data = rec_get_nth_field(rec, offsets,
							 field_no, &len);

				needs_copy = column->copy_val;
			}

			if (needs_copy) {
				eval_node_copy_and_alloc_val(column, data,
							     len);
			} else {
				val = que_node_get_val(column);
				dfield_set_data(val, data, len);
			}

			if (UNIV_LIKELY_NULL(heap)) {
				mem_heap_free(heap);
			}
		}

		column = UT_LIST_GET_NEXT(col_var_list, column);
	}
}

/*********************************************************************//**
Allocates a prefetch buffer for a column when prefetch is first time done. */
static
void
sel_col_prefetch_buf_alloc(
/*=======================*/
	sym_node_t*	column)	/*!< in: symbol table node for a column */
{
	sel_buf_t*	sel_buf;
	ulint		i;

	ut_ad(que_node_get_type(column) == QUE_NODE_SYMBOL);

	column->prefetch_buf = static_cast<sel_buf_t*>(
		mem_alloc(SEL_MAX_N_PREFETCH * sizeof(sel_buf_t)));

	for (i = 0; i < SEL_MAX_N_PREFETCH; i++) {
		sel_buf = column->prefetch_buf + i;

		sel_buf->data = NULL;
		sel_buf->len = 0;
		sel_buf->val_buf_size = 0;
	}
}

/*********************************************************************//**
Frees a prefetch buffer for a column, including the dynamically allocated
memory for data stored there. */
UNIV_INTERN
void
sel_col_prefetch_buf_free(
/*======================*/
	sel_buf_t*	prefetch_buf)	/*!< in, own: prefetch buffer */
{
	sel_buf_t*	sel_buf;
	ulint		i;

	for (i = 0; i < SEL_MAX_N_PREFETCH; i++) {
		sel_buf = prefetch_buf + i;

		if (sel_buf->val_buf_size > 0) {

			mem_free(sel_buf->data);
		}
	}

	mem_free(prefetch_buf);
}

/*********************************************************************//**
Pops the column values for a prefetched, cached row from the column prefetch
buffers and places them to the val fields in the column nodes. */
static
void
sel_dequeue_prefetched_row(
/*=======================*/
	plan_t*	plan)	/*!< in: plan node for a table */
{
	sym_node_t*	column;
	sel_buf_t*	sel_buf;
	dfield_t*	val;
	byte*		data;
	ulint		len;
	ulint		val_buf_size;

	ut_ad(plan->n_rows_prefetched > 0);

	column = UT_LIST_GET_FIRST(plan->columns);

	while (column) {
		val = que_node_get_val(column);

		if (!column->copy_val) {
			/* We did not really push any value for the
			column */

			ut_ad(!column->prefetch_buf);
			ut_ad(que_node_get_val_buf_size(column) == 0);
			ut_d(dfield_set_null(val));

			goto next_col;
		}

		ut_ad(column->prefetch_buf);
		ut_ad(!dfield_is_ext(val));

		sel_buf = column->prefetch_buf + plan->first_prefetched;

		data = sel_buf->data;
		len = sel_buf->len;
		val_buf_size = sel_buf->val_buf_size;

		/* We must keep track of the allocated memory for
		column values to be able to free it later: therefore
		we swap the values for sel_buf and val */

		sel_buf->data = static_cast<byte*>(dfield_get_data(val));
		sel_buf->len = dfield_get_len(val);
		sel_buf->val_buf_size = que_node_get_val_buf_size(column);

		dfield_set_data(val, data, len);
		que_node_set_val_buf_size(column, val_buf_size);
next_col:
		column = UT_LIST_GET_NEXT(col_var_list, column);
	}

	plan->n_rows_prefetched--;

	plan->first_prefetched++;
}

/*********************************************************************//**
Pushes the column values for a prefetched, cached row to the column prefetch
buffers from the val fields in the column nodes. */
UNIV_INLINE
void
sel_enqueue_prefetched_row(
/*=======================*/
	plan_t*	plan)	/*!< in: plan node for a table */
{
	sym_node_t*	column;
	sel_buf_t*	sel_buf;
	dfield_t*	val;
	byte*		data;
	ulint		len;
	ulint		pos;
	ulint		val_buf_size;

	if (plan->n_rows_prefetched == 0) {
		pos = 0;
		plan->first_prefetched = 0;
	} else {
		pos = plan->n_rows_prefetched;

		/* We have the convention that pushing new rows starts only
		after the prefetch stack has been emptied: */

		ut_ad(plan->first_prefetched == 0);
	}

	plan->n_rows_prefetched++;

	ut_ad(pos < SEL_MAX_N_PREFETCH);

	for (column = UT_LIST_GET_FIRST(plan->columns);
	     column != 0;
	     column = UT_LIST_GET_NEXT(col_var_list, column)) {

		if (!column->copy_val) {
			/* There is no sense to push pointers to database
			page fields when we do not keep latch on the page! */
			continue;
		}

		if (!column->prefetch_buf) {
			/* Allocate a new prefetch buffer */

			sel_col_prefetch_buf_alloc(column);
		}

		sel_buf = column->prefetch_buf + pos;

		val = que_node_get_val(column);

		data = static_cast<byte*>(dfield_get_data(val));
		len = dfield_get_len(val);
		val_buf_size = que_node_get_val_buf_size(column);

		/* We must keep track of the allocated memory for
		column values to be able to free it later: therefore
		we swap the values for sel_buf and val */

		dfield_set_data(val, sel_buf->data, sel_buf->len);
		que_node_set_val_buf_size(column, sel_buf->val_buf_size);

		sel_buf->data = data;
		sel_buf->len = len;
		sel_buf->val_buf_size = val_buf_size;
	}
}

/*********************************************************************//**
Builds a previous version of a clustered index record for a consistent read
@return	DB_SUCCESS or error code */
static __attribute__((nonnull, warn_unused_result))
dberr_t
row_sel_build_prev_vers(
/*====================*/
	read_view_t*	read_view,	/*!< in: read view */
	dict_index_t*	index,		/*!< in: plan node for table */
	rec_t*		rec,		/*!< in: record in a clustered index */
	ulint**		offsets,	/*!< in/out: offsets returned by
					rec_get_offsets(rec, plan->index) */
	mem_heap_t**	offset_heap,	/*!< in/out: memory heap from which
					the offsets are allocated */
	mem_heap_t**    old_vers_heap,  /*!< out: old version heap to use */
	rec_t**		old_vers,	/*!< out: old version, or NULL if the
					record does not exist in the view:
					i.e., it was freshly inserted
					afterwards */
	mtr_t*		mtr)		/*!< in: mtr */
{
	dberr_t	err;

	if (*old_vers_heap) {
		mem_heap_empty(*old_vers_heap);
	} else {
		*old_vers_heap = mem_heap_create(512);
	}

	err = row_vers_build_for_consistent_read(
		rec, mtr, index, offsets, read_view, offset_heap,
		*old_vers_heap, old_vers);
	return(err);
}

/*********************************************************************//**
Builds the last committed version of a clustered index record for a
semi-consistent read. */
static __attribute__((nonnull))
void
row_sel_build_committed_vers_for_mysql(
/*===================================*/
	dict_index_t*	clust_index,	/*!< in: clustered index */
	row_prebuilt_t*	prebuilt,	/*!< in: prebuilt struct */
	const rec_t*	rec,		/*!< in: record in a clustered index */
	ulint**		offsets,	/*!< in/out: offsets returned by
					rec_get_offsets(rec, clust_index) */
	mem_heap_t**	offset_heap,	/*!< in/out: memory heap from which
					the offsets are allocated */
	const rec_t**	old_vers,	/*!< out: old version, or NULL if the
					record does not exist in the view:
					i.e., it was freshly inserted
					afterwards */
	mtr_t*		mtr)		/*!< in: mtr */
{
	if (prebuilt->old_vers_heap) {
		mem_heap_empty(prebuilt->old_vers_heap);
	} else {
		prebuilt->old_vers_heap = mem_heap_create(
			rec_offs_size(*offsets));
	}

	row_vers_build_for_semi_consistent_read(
		rec, mtr, clust_index, offsets, offset_heap,
		prebuilt->old_vers_heap, old_vers);
}

/*********************************************************************//**
Tests the conditions which determine when the index segment we are searching
through has been exhausted.
@return	TRUE if row passed the tests */
UNIV_INLINE
ibool
row_sel_test_end_conds(
/*===================*/
	plan_t*	plan)	/*!< in: plan for the table; the column values must
			already have been retrieved and the right sides of
			comparisons evaluated */
{
	func_node_t*	cond;

	/* All conditions in end_conds are comparisons of a column to an
	expression */

	for (cond = UT_LIST_GET_FIRST(plan->end_conds);
	     cond != 0;
	     cond = UT_LIST_GET_NEXT(cond_list, cond)) {

		/* Evaluate the left side of the comparison, i.e., get the
		column value if there is an indirection */

		eval_sym(static_cast<sym_node_t*>(cond->args));

		/* Do the comparison */

		if (!eval_cmp(cond)) {

			return(FALSE);
		}
	}

	return(TRUE);
}

/*********************************************************************//**
Tests the other conditions.
@return	TRUE if row passed the tests */
UNIV_INLINE
ibool
row_sel_test_other_conds(
/*=====================*/
	plan_t*	plan)	/*!< in: plan for the table; the column values must
			already have been retrieved */
{
	func_node_t*	cond;

	cond = UT_LIST_GET_FIRST(plan->other_conds);

	while (cond) {
		eval_exp(cond);

		if (!eval_node_get_ibool_val(cond)) {

			return(FALSE);
		}

		cond = UT_LIST_GET_NEXT(cond_list, cond);
	}

	return(TRUE);
}

/*********************************************************************//**
Retrieves the clustered index record corresponding to a record in a
non-clustered index. Does the necessary locking.
@return	DB_SUCCESS or error code */
static __attribute__((nonnull, warn_unused_result))
dberr_t
row_sel_get_clust_rec(
/*==================*/
	sel_node_t*	node,	/*!< in: select_node */
	plan_t*		plan,	/*!< in: plan node for table */
	rec_t*		rec,	/*!< in: record in a non-clustered index */
	que_thr_t*	thr,	/*!< in: query thread */
	rec_t**		out_rec,/*!< out: clustered record or an old version of
				it, NULL if the old version did not exist
				in the read view, i.e., it was a fresh
				inserted version */
	mtr_t*		mtr)	/*!< in: mtr used to get access to the
				non-clustered record; the same mtr is used to
				access the clustered index */
{
	dict_index_t*	index;
	rec_t*		clust_rec;
	rec_t*		old_vers;
	dberr_t		err;
	mem_heap_t*	heap		= NULL;
	ulint		offsets_[REC_OFFS_NORMAL_SIZE];
	ulint*		offsets		= offsets_;
	rec_offs_init(offsets_);

	*out_rec = NULL;

	offsets = rec_get_offsets(rec,
				  btr_pcur_get_btr_cur(&plan->pcur)->index,
				  offsets, ULINT_UNDEFINED, &heap);

	row_build_row_ref_fast(plan->clust_ref, plan->clust_map, rec, offsets);

	index = dict_table_get_first_index(plan->table);

	btr_pcur_open_with_no_init(index, plan->clust_ref, PAGE_CUR_LE,
				   BTR_SEARCH_LEAF, &plan->clust_pcur,
				   0, mtr);

	clust_rec = btr_pcur_get_rec(&(plan->clust_pcur));

	/* Note: only if the search ends up on a non-infimum record is the
	low_match value the real match to the search tuple */

	if (!page_rec_is_user_rec(clust_rec)
	    || btr_pcur_get_low_match(&(plan->clust_pcur))
	    < dict_index_get_n_unique(index)) {

		ut_a(rec_get_deleted_flag(rec,
					  dict_table_is_comp(plan->table)));
		ut_a(node->read_view);

		/* In a rare case it is possible that no clust rec is found
		for a delete-marked secondary index record: if in row0umod.cc
		in row_undo_mod_remove_clust_low() we have already removed
		the clust rec, while purge is still cleaning and removing
		secondary index records associated with earlier versions of
		the clustered index record. In that case we know that the
		clustered index record did not exist in the read view of
		trx. */

		goto func_exit;
	}

	offsets = rec_get_offsets(clust_rec, index, offsets,
				  ULINT_UNDEFINED, &heap);

	if (!node->read_view) {
		/* Try to place a lock on the index record */
		ulint	lock_type;
		trx_t*	trx;

		trx = thr_get_trx(thr);

		/* If innodb_locks_unsafe_for_binlog option is used
		or this session is using READ COMMITTED or lower isolation level
		we lock only the record, i.e., next-key locking is
		not used. */
		if (srv_locks_unsafe_for_binlog
		    || trx->isolation_level <= TRX_ISO_READ_COMMITTED) {
			lock_type = LOCK_REC_NOT_GAP;
		} else {
			lock_type = LOCK_ORDINARY;
		}

		err = lock_clust_rec_read_check_and_lock(
			0, btr_pcur_get_block(&plan->clust_pcur),
			clust_rec, index, offsets,
			static_cast<enum lock_mode>(node->row_lock_mode),
			lock_type,
			thr);

		switch (err) {
		case DB_SUCCESS:
		case DB_SUCCESS_LOCKED_REC:
			/* Declare the variable uninitialized in Valgrind.
			It should be set to DB_SUCCESS at func_exit. */
			UNIV_MEM_INVALID(&err, sizeof err);
			break;
		default:
			goto err_exit;
		}
	} else {
		/* This is a non-locking consistent read: if necessary, fetch
		a previous version of the record */

		old_vers = NULL;

		if (!lock_clust_rec_cons_read_sees(clust_rec, index, offsets,
						   node->read_view)) {

			err = row_sel_build_prev_vers(
				node->read_view, index, clust_rec,
				&offsets, &heap, &plan->old_vers_heap,
				&old_vers, mtr);

			if (err != DB_SUCCESS) {

				goto err_exit;
			}

			clust_rec = old_vers;

			if (clust_rec == NULL) {
				goto func_exit;
			}
		}

		/* If we had to go to an earlier version of row or the
		secondary index record is delete marked, then it may be that
		the secondary index record corresponding to clust_rec
		(or old_vers) is not rec; in that case we must ignore
		such row because in our snapshot rec would not have existed.
		Remember that from rec we cannot see directly which transaction
		id corresponds to it: we have to go to the clustered index
		record. A query where we want to fetch all rows where
		the secondary index value is in some interval would return
		a wrong result if we would not drop rows which we come to
		visit through secondary index records that would not really
		exist in our snapshot. */

		if ((old_vers
		     || rec_get_deleted_flag(rec, dict_table_is_comp(
						     plan->table)))
		    && !row_sel_sec_rec_is_for_clust_rec(rec, plan->index,
							 clust_rec, index)) {
			goto func_exit;
		}
	}

	/* Fetch the columns needed in test conditions.  The clustered
	index record is protected by a page latch that was acquired
	when plan->clust_pcur was positioned.  The latch will not be
	released until mtr_commit(mtr). */

	ut_ad(!rec_get_deleted_flag(clust_rec, rec_offs_comp(offsets)));
	row_sel_fetch_columns(index, clust_rec, offsets,
			      UT_LIST_GET_FIRST(plan->columns));
	*out_rec = clust_rec;
func_exit:
	err = DB_SUCCESS;
err_exit:
	if (UNIV_LIKELY_NULL(heap)) {
		mem_heap_free(heap);
	}
	return(err);
}

/*********************************************************************//**
Sets a lock on a record.
@return	DB_SUCCESS, DB_SUCCESS_LOCKED_REC, or error code */
UNIV_INLINE
dberr_t
sel_set_rec_lock(
/*=============*/
	const buf_block_t*	block,	/*!< in: buffer block of rec */
	const rec_t*		rec,	/*!< in: record */
	dict_index_t*		index,	/*!< in: index */
	const ulint*		offsets,/*!< in: rec_get_offsets(rec, index) */
	ulint			mode,	/*!< in: lock mode */
	ulint			type,	/*!< in: LOCK_ORDINARY, LOCK_GAP, or
					LOC_REC_NOT_GAP */
	que_thr_t*		thr)	/*!< in: query thread */
{
	trx_t*		trx;
	dberr_t		err;

	trx = thr_get_trx(thr);

	if (UT_LIST_GET_LEN(trx->lock.trx_locks) > 10000) {
		if (buf_LRU_buf_pool_running_out()) {

			return(DB_LOCK_TABLE_FULL);
		}
	}

	if (dict_index_is_clust(index)) {
		err = lock_clust_rec_read_check_and_lock(
			0, block, rec, index, offsets,
			static_cast<enum lock_mode>(mode), type, thr);
	} else {
		err = lock_sec_rec_read_check_and_lock(
			0, block, rec, index, offsets,
			static_cast<enum lock_mode>(mode), type, thr);
	}

	return(err);
}

/*********************************************************************//**
Opens a pcur to a table index. */
static
void
row_sel_open_pcur(
/*==============*/
	plan_t*		plan,		/*!< in: table plan */
	ibool		search_latch_locked,
					/*!< in: TRUE if the thread currently
					has the search latch locked in
					s-mode */
	mtr_t*		mtr)		/*!< in: mtr */
{
	dict_index_t*	index;
	func_node_t*	cond;
	que_node_t*	exp;
	ulint		n_fields;
	ulint		has_search_latch = 0;	/* RW_S_LATCH or 0 */
	ulint		i;

	if (search_latch_locked) {
		has_search_latch = RW_S_LATCH;
	}

	index = plan->index;

	/* Calculate the value of the search tuple: the exact match columns
	get their expressions evaluated when we evaluate the right sides of
	end_conds */

	cond = UT_LIST_GET_FIRST(plan->end_conds);

	while (cond) {
		eval_exp(que_node_get_next(cond->args));

		cond = UT_LIST_GET_NEXT(cond_list, cond);
	}

	if (plan->tuple) {
		n_fields = dtuple_get_n_fields(plan->tuple);

		if (plan->n_exact_match < n_fields) {
			/* There is a non-exact match field which must be
			evaluated separately */

			eval_exp(plan->tuple_exps[n_fields - 1]);
		}

		for (i = 0; i < n_fields; i++) {
			exp = plan->tuple_exps[i];

			dfield_copy_data(dtuple_get_nth_field(plan->tuple, i),
					 que_node_get_val(exp));
		}

		/* Open pcur to the index */

		btr_pcur_open_with_no_init(index, plan->tuple, plan->mode,
					   BTR_SEARCH_LEAF, &plan->pcur,
					   has_search_latch, mtr);
	} else {
		/* Open the cursor to the start or the end of the index
		(FALSE: no init) */

		btr_pcur_open_at_index_side(plan->asc, index, BTR_SEARCH_LEAF,
					    &(plan->pcur), false, 0, mtr);
	}

	ut_ad(plan->n_rows_prefetched == 0);
	ut_ad(plan->n_rows_fetched == 0);
	ut_ad(plan->cursor_at_end == FALSE);

	plan->pcur_is_open = TRUE;
}

/*********************************************************************//**
Restores a stored pcur position to a table index.
@return TRUE if the cursor should be moved to the next record after we
return from this function (moved to the previous, in the case of a
descending cursor) without processing again the current cursor
record */
static
ibool
row_sel_restore_pcur_pos(
/*=====================*/
	plan_t*		plan,	/*!< in: table plan */
	mtr_t*		mtr)	/*!< in: mtr */
{
	ibool	equal_position;
	ulint	relative_position;

	ut_ad(!plan->cursor_at_end);

	relative_position = btr_pcur_get_rel_pos(&(plan->pcur));

	equal_position = btr_pcur_restore_position(BTR_SEARCH_LEAF,
						   &(plan->pcur), mtr);

	/* If the cursor is traveling upwards, and relative_position is

	(1) BTR_PCUR_BEFORE: this is not allowed, as we did not have a lock
	yet on the successor of the page infimum;
	(2) BTR_PCUR_AFTER: btr_pcur_restore_position placed the cursor on the
	first record GREATER than the predecessor of a page supremum; we have
	not yet processed the cursor record: no need to move the cursor to the
	next record;
	(3) BTR_PCUR_ON: btr_pcur_restore_position placed the cursor on the
	last record LESS or EQUAL to the old stored user record; (a) if
	equal_position is FALSE, this means that the cursor is now on a record
	less than the old user record, and we must move to the next record;
	(b) if equal_position is TRUE, then if
	plan->stored_cursor_rec_processed is TRUE, we must move to the next
	record, else there is no need to move the cursor. */

	if (plan->asc) {
		if (relative_position == BTR_PCUR_ON) {

			if (equal_position) {

				return(plan->stored_cursor_rec_processed);
			}

			return(TRUE);
		}

		ut_ad(relative_position == BTR_PCUR_AFTER
		      || relative_position == BTR_PCUR_AFTER_LAST_IN_TREE);

		return(FALSE);
	}

	/* If the cursor is traveling downwards, and relative_position is

	(1) BTR_PCUR_BEFORE: btr_pcur_restore_position placed the cursor on
	the last record LESS than the successor of a page infimum; we have not
	processed the cursor record: no need to move the cursor;
	(2) BTR_PCUR_AFTER: btr_pcur_restore_position placed the cursor on the
	first record GREATER than the predecessor of a page supremum; we have
	processed the cursor record: we should move the cursor to the previous
	record;
	(3) BTR_PCUR_ON: btr_pcur_restore_position placed the cursor on the
	last record LESS or EQUAL to the old stored user record; (a) if
	equal_position is FALSE, this means that the cursor is now on a record
	less than the old user record, and we need not move to the previous
	record; (b) if equal_position is TRUE, then if
	plan->stored_cursor_rec_processed is TRUE, we must move to the previous
	record, else there is no need to move the cursor. */

	if (relative_position == BTR_PCUR_BEFORE
	    || relative_position == BTR_PCUR_BEFORE_FIRST_IN_TREE) {

		return(FALSE);
	}

	if (relative_position == BTR_PCUR_ON) {

		if (equal_position) {

			return(plan->stored_cursor_rec_processed);
		}

		return(FALSE);
	}

	ut_ad(relative_position == BTR_PCUR_AFTER
	      || relative_position == BTR_PCUR_AFTER_LAST_IN_TREE);

	return(TRUE);
}

/*********************************************************************//**
Resets a plan cursor to a closed state. */
UNIV_INLINE
void
plan_reset_cursor(
/*==============*/
	plan_t*	plan)	/*!< in: plan */
{
	plan->pcur_is_open = FALSE;
	plan->cursor_at_end = FALSE;
	plan->n_rows_fetched = 0;
	plan->n_rows_prefetched = 0;
}

/*********************************************************************//**
Tries to do a shortcut to fetch a clustered index record with a unique key,
using the hash index if possible (not always).
@return	SEL_FOUND, SEL_EXHAUSTED, SEL_RETRY */
static
ulint
row_sel_try_search_shortcut(
/*========================*/
	sel_node_t*	node,	/*!< in: select node for a consistent read */
	plan_t*		plan,	/*!< in: plan for a unique search in clustered
				index */
	ibool		search_latch_locked,
				/*!< in: whether the search holds
				btr_search_latch */
	mtr_t*		mtr)	/*!< in: mtr */
{
	dict_index_t*	index;
	rec_t*		rec;
	mem_heap_t*	heap		= NULL;
	ulint		offsets_[REC_OFFS_NORMAL_SIZE];
	ulint*		offsets		= offsets_;
	ulint		ret;
	rec_offs_init(offsets_);

	index = plan->index;

	ut_ad(node->read_view);
	ut_ad(plan->unique_search);
	ut_ad(!plan->must_get_clust);
#ifdef UNIV_SYNC_DEBUG
	if (search_latch_locked) {
		ut_ad(rw_lock_own(btr_search_get_latch(index),
				  RW_LOCK_SHARED));
	}
#endif /* UNIV_SYNC_DEBUG */

	row_sel_open_pcur(plan, search_latch_locked, mtr);

	rec = btr_pcur_get_rec(&(plan->pcur));

	if (!page_rec_is_user_rec(rec)) {

		return(SEL_RETRY);
	}

	ut_ad(plan->mode == PAGE_CUR_GE);

	/* As the cursor is now placed on a user record after a search with
	the mode PAGE_CUR_GE, the up_match field in the cursor tells how many
	fields in the user record matched to the search tuple */

	if (btr_pcur_get_up_match(&(plan->pcur)) < plan->n_exact_match) {

		return(SEL_EXHAUSTED);
	}

	/* This is a non-locking consistent read: if necessary, fetch
	a previous version of the record */

	offsets = rec_get_offsets(rec, index, offsets, ULINT_UNDEFINED, &heap);

	if (dict_index_is_clust(index)) {
		if (!lock_clust_rec_cons_read_sees(rec, index, offsets,
						   node->read_view)) {
			ret = SEL_RETRY;
			goto func_exit;
		}
	} else if (!lock_sec_rec_cons_read_sees(rec, node->read_view)) {

		ret = SEL_RETRY;
		goto func_exit;
	}

	/* Test the deleted flag. */

	if (rec_get_deleted_flag(rec, dict_table_is_comp(plan->table))) {

		ret = SEL_EXHAUSTED;
		goto func_exit;
	}

	/* Fetch the columns needed in test conditions.  The index
	record is protected by a page latch that was acquired when
	plan->pcur was positioned.  The latch will not be released
	until mtr_commit(mtr). */

	row_sel_fetch_columns(index, rec, offsets,
			      UT_LIST_GET_FIRST(plan->columns));

	/* Test the rest of search conditions */

	if (!row_sel_test_other_conds(plan)) {

		ret = SEL_EXHAUSTED;
		goto func_exit;
	}

	ut_ad(plan->pcur.latch_mode == BTR_SEARCH_LEAF);

	plan->n_rows_fetched++;
	ret = SEL_FOUND;
func_exit:
	if (UNIV_LIKELY_NULL(heap)) {
		mem_heap_free(heap);
	}
	return(ret);
}

/*********************************************************************//**
Performs a select step.
@return	DB_SUCCESS or error code */
static __attribute__((nonnull, warn_unused_result))
dberr_t
row_sel(
/*====*/
	sel_node_t*	node,	/*!< in: select node */
	que_thr_t*	thr)	/*!< in: query thread */
{
	dict_index_t*	index;
	plan_t*		plan;
	mtr_t		mtr;
	ibool		moved;
	rec_t*		rec;
	rec_t*		old_vers;
	rec_t*		clust_rec;
	ibool		search_latch_locked;
	ibool		consistent_read;

	/* The following flag becomes TRUE when we are doing a
	consistent read from a non-clustered index and we must look
	at the clustered index to find out the previous delete mark
	state of the non-clustered record: */

	ibool		cons_read_requires_clust_rec	= FALSE;
	ulint		cost_counter			= 0;
	ibool		cursor_just_opened;
	ibool		must_go_to_next;
	ibool		mtr_has_extra_clust_latch	= FALSE;
	/* TRUE if the search was made using
	a non-clustered index, and we had to
	access the clustered record: now &mtr
	contains a clustered index latch, and
	&mtr must be committed before we move
	to the next non-clustered record */
	ulint		found_flag;
	dberr_t		err;
	mem_heap_t*	heap				= NULL;
	ulint		offsets_[REC_OFFS_NORMAL_SIZE];
	ulint*		offsets				= offsets_;
	rec_offs_init(offsets_);

	ut_ad(thr->run_node == node);

	search_latch_locked = FALSE;

	if (node->read_view) {
		/* In consistent reads, we try to do with the hash index and
		not to use the buffer page get. This is to reduce memory bus
		load resulting from semaphore operations. The search latch
		will be s-locked when we access an index with a unique search
		condition, but not locked when we access an index with a
		less selective search condition. */

		consistent_read = TRUE;
	} else {
		consistent_read = FALSE;
	}

table_loop:
	/* TABLE LOOP
	----------
	This is the outer major loop in calculating a join. We come here when
	node->fetch_table changes, and after adding a row to aggregate totals
	and, of course, when this function is called. */

	ut_ad(mtr_has_extra_clust_latch == FALSE);

	plan = sel_node_get_nth_plan(node, node->fetch_table);
	index = plan->index;

	if (plan->n_rows_prefetched > 0) {
		sel_dequeue_prefetched_row(plan);

		goto next_table_no_mtr;
	}

	if (plan->cursor_at_end) {
		/* The cursor has already reached the result set end: no more
		rows to process for this table cursor, as also the prefetch
		stack was empty */

		ut_ad(plan->pcur_is_open);

		goto table_exhausted_no_mtr;
	}

	/* Open a cursor to index, or restore an open cursor position */

	mtr_start(&mtr);

	if (consistent_read && plan->unique_search && !plan->pcur_is_open
	    && !plan->must_get_clust
	    && !plan->table->big_rows) {
		if (!search_latch_locked) {
			rw_lock_s_lock(btr_search_get_latch(index));

			search_latch_locked = TRUE;
		} else if (rw_lock_get_writer(btr_search_get_latch(index))
			   == RW_LOCK_WAIT_EX) {

			/* There is an x-latch request waiting: release the
			s-latch for a moment; as an s-latch here is often
			kept for some 10 searches before being released,
			a waiting x-latch request would block other threads
			from acquiring an s-latch for a long time, lowering
			performance significantly in multiprocessors. */

			rw_lock_s_unlock(btr_search_get_latch(index));
			rw_lock_s_lock(btr_search_get_latch(index));
		}

		found_flag = row_sel_try_search_shortcut(node, plan,
							 search_latch_locked,
							 &mtr);

		if (found_flag == SEL_FOUND) {

			goto next_table;

		} else if (found_flag == SEL_EXHAUSTED) {

			goto table_exhausted;
		}

		ut_ad(found_flag == SEL_RETRY);

		plan_reset_cursor(plan);

		mtr_commit(&mtr);
		mtr_start(&mtr);
	}

	if (search_latch_locked) {
		rw_lock_s_unlock(btr_search_get_latch(index));

		search_latch_locked = FALSE;
	}

	if (!plan->pcur_is_open) {
		/* Evaluate the expressions to build the search tuple and
		open the cursor */

		row_sel_open_pcur(plan, search_latch_locked, &mtr);

		cursor_just_opened = TRUE;

		/* A new search was made: increment the cost counter */
		cost_counter++;
	} else {
		/* Restore pcur position to the index */

		must_go_to_next = row_sel_restore_pcur_pos(plan, &mtr);

		cursor_just_opened = FALSE;

		if (must_go_to_next) {
			/* We have already processed the cursor record: move
			to the next */

			goto next_rec;
		}
	}

rec_loop:
	/* RECORD LOOP
	-----------
	In this loop we use pcur and try to fetch a qualifying row, and
	also fill the prefetch buffer for this table if n_rows_fetched has
	exceeded a threshold. While we are inside this loop, the following
	holds:
	(1) &mtr is started,
	(2) pcur is positioned and open.

	NOTE that if cursor_just_opened is TRUE here, it means that we came
	to this point right after row_sel_open_pcur. */

	ut_ad(mtr_has_extra_clust_latch == FALSE);

	rec = btr_pcur_get_rec(&(plan->pcur));

	/* PHASE 1: Set a lock if specified */

	if (!node->asc && cursor_just_opened
	    && !page_rec_is_supremum(rec)) {

		/* When we open a cursor for a descending search, we must set
		a next-key lock on the successor record: otherwise it would
		be possible to insert new records next to the cursor position,
		and it might be that these new records should appear in the
		search result set, resulting in the phantom problem. */

		if (!consistent_read) {
			rec_t*	next_rec = page_rec_get_next(rec);
			ulint	lock_type;
			trx_t*	trx;

			trx = thr_get_trx(thr);

			offsets = rec_get_offsets(next_rec, index, offsets,
						  ULINT_UNDEFINED, &heap);

			/* If innodb_locks_unsafe_for_binlog option is used
			or this session is using READ COMMITTED or lower isolation
			level, we lock only the record, i.e., next-key
			locking is not used. */
			if (srv_locks_unsafe_for_binlog
			    || trx->isolation_level
			    <= TRX_ISO_READ_COMMITTED) {

				if (page_rec_is_supremum(next_rec)) {

					goto skip_lock;
				}

				lock_type = LOCK_REC_NOT_GAP;
			} else {
				lock_type = LOCK_ORDINARY;
			}

			err = sel_set_rec_lock(btr_pcur_get_block(&plan->pcur),
					       next_rec, index, offsets,
					       node->row_lock_mode,
					       lock_type, thr);

			switch (err) {
			case DB_SUCCESS_LOCKED_REC:
				err = DB_SUCCESS;
			case DB_SUCCESS:
				break;
			default:
				/* Note that in this case we will store in pcur
				the PREDECESSOR of the record we are waiting
				the lock for */
				goto lock_wait_or_error;
			}
		}
	}

skip_lock:
	if (page_rec_is_infimum(rec)) {

		/* The infimum record on a page cannot be in the result set,
		and neither can a record lock be placed on it: we skip such
		a record. We also increment the cost counter as we may have
		processed yet another page of index. */

		cost_counter++;

		goto next_rec;
	}

	if (!consistent_read) {
		/* Try to place a lock on the index record */
		ulint	lock_type;
		trx_t*	trx;

		offsets = rec_get_offsets(rec, index, offsets,
					  ULINT_UNDEFINED, &heap);

		trx = thr_get_trx(thr);

		/* If innodb_locks_unsafe_for_binlog option is used
		or this session is using READ COMMITTED or lower isolation level,
		we lock only the record, i.e., next-key locking is
		not used. */
		if (srv_locks_unsafe_for_binlog
		    || trx->isolation_level <= TRX_ISO_READ_COMMITTED) {

			if (page_rec_is_supremum(rec)) {

				goto next_rec;
			}

			lock_type = LOCK_REC_NOT_GAP;
		} else {
			lock_type = LOCK_ORDINARY;
		}

		err = sel_set_rec_lock(btr_pcur_get_block(&plan->pcur),
				       rec, index, offsets,
				       node->row_lock_mode, lock_type, thr);

		switch (err) {
		case DB_SUCCESS_LOCKED_REC:
			err = DB_SUCCESS;
		case DB_SUCCESS:
			break;
		default:
			goto lock_wait_or_error;
		}
	}

	if (page_rec_is_supremum(rec)) {

		/* A page supremum record cannot be in the result set: skip
		it now when we have placed a possible lock on it */

		goto next_rec;
	}

	ut_ad(page_rec_is_user_rec(rec));

	if (cost_counter > SEL_COST_LIMIT) {

		/* Now that we have placed the necessary locks, we can stop
		for a while and store the cursor position; NOTE that if we
		would store the cursor position BEFORE placing a record lock,
		it might happen that the cursor would jump over some records
		that another transaction could meanwhile insert adjacent to
		the cursor: this would result in the phantom problem. */

		goto stop_for_a_while;
	}

	/* PHASE 2: Check a mixed index mix id if needed */

	if (plan->unique_search && cursor_just_opened) {

		ut_ad(plan->mode == PAGE_CUR_GE);

		/* As the cursor is now placed on a user record after a search
		with the mode PAGE_CUR_GE, the up_match field in the cursor
		tells how many fields in the user record matched to the search
		tuple */

		if (btr_pcur_get_up_match(&(plan->pcur))
		    < plan->n_exact_match) {
			goto table_exhausted;
		}

		/* Ok, no need to test end_conds or mix id */

	}

	/* We are ready to look at a possible new index entry in the result
	set: the cursor is now placed on a user record */

	/* PHASE 3: Get previous version in a consistent read */

	cons_read_requires_clust_rec = FALSE;
	offsets = rec_get_offsets(rec, index, offsets, ULINT_UNDEFINED, &heap);

	if (consistent_read) {
		/* This is a non-locking consistent read: if necessary, fetch
		a previous version of the record */

		if (dict_index_is_clust(index)) {

			if (!lock_clust_rec_cons_read_sees(rec, index, offsets,
							   node->read_view)) {

				err = row_sel_build_prev_vers(
					node->read_view, index, rec,
					&offsets, &heap, &plan->old_vers_heap,
					&old_vers, &mtr);

				if (err != DB_SUCCESS) {

					goto lock_wait_or_error;
				}

				if (old_vers == NULL) {
					/* The record does not exist
					in our read view. Skip it, but
					first attempt to determine
					whether the index segment we
					are searching through has been
					exhausted. */

					offsets = rec_get_offsets(
						rec, index, offsets,
						ULINT_UNDEFINED, &heap);

					/* Fetch the columns needed in
					test conditions. The clustered
					index record is protected by a
					page latch that was acquired
					by row_sel_open_pcur() or
					row_sel_restore_pcur_pos().
					The latch will not be released
					until mtr_commit(mtr). */

					row_sel_fetch_columns(
						index, rec, offsets,
						UT_LIST_GET_FIRST(
							plan->columns));

					if (!row_sel_test_end_conds(plan)) {

						goto table_exhausted;
					}

					goto next_rec;
				}

				rec = old_vers;
			}
		} else if (!lock_sec_rec_cons_read_sees(rec,
							node->read_view)) {
			cons_read_requires_clust_rec = TRUE;
		}
	}

	/* PHASE 4: Test search end conditions and deleted flag */

	/* Fetch the columns needed in test conditions.  The record is
	protected by a page latch that was acquired by
	row_sel_open_pcur() or row_sel_restore_pcur_pos().  The latch
	will not be released until mtr_commit(mtr). */

	row_sel_fetch_columns(index, rec, offsets,
			      UT_LIST_GET_FIRST(plan->columns));

	/* Test the selection end conditions: these can only contain columns
	which already are found in the index, even though the index might be
	non-clustered */

	if (plan->unique_search && cursor_just_opened) {

		/* No test necessary: the test was already made above */

	} else if (!row_sel_test_end_conds(plan)) {

		goto table_exhausted;
	}

	if (rec_get_deleted_flag(rec, dict_table_is_comp(plan->table))
	    && !cons_read_requires_clust_rec) {

		/* The record is delete marked: we can skip it if this is
		not a consistent read which might see an earlier version
		of a non-clustered index record */

		if (plan->unique_search) {

			goto table_exhausted;
		}

		goto next_rec;
	}

	/* PHASE 5: Get the clustered index record, if needed and if we did
	not do the search using the clustered index */

	if (plan->must_get_clust || cons_read_requires_clust_rec) {

		/* It was a non-clustered index and we must fetch also the
		clustered index record */

		err = row_sel_get_clust_rec(node, plan, rec, thr, &clust_rec,
					    &mtr);
		mtr_has_extra_clust_latch = TRUE;

		if (err != DB_SUCCESS) {

			goto lock_wait_or_error;
		}

		/* Retrieving the clustered record required a search:
		increment the cost counter */

		cost_counter++;

		if (clust_rec == NULL) {
			/* The record did not exist in the read view */
			ut_ad(consistent_read);

			goto next_rec;
		}

		if (rec_get_deleted_flag(clust_rec,
					 dict_table_is_comp(plan->table))) {

			/* The record is delete marked: we can skip it */

			goto next_rec;
		}

		if (node->can_get_updated) {

			btr_pcur_store_position(&(plan->clust_pcur), &mtr);
		}
	}

	/* PHASE 6: Test the rest of search conditions */

	if (!row_sel_test_other_conds(plan)) {

		if (plan->unique_search) {

			goto table_exhausted;
		}

		goto next_rec;
	}

	/* PHASE 7: We found a new qualifying row for the current table; push
	the row if prefetch is on, or move to the next table in the join */

	plan->n_rows_fetched++;

	ut_ad(plan->pcur.latch_mode == BTR_SEARCH_LEAF);

	if ((plan->n_rows_fetched <= SEL_PREFETCH_LIMIT)
	    || plan->unique_search || plan->no_prefetch
	    || plan->table->big_rows) {

		/* No prefetch in operation: go to the next table */

		goto next_table;
	}

	sel_enqueue_prefetched_row(plan);

	if (plan->n_rows_prefetched == SEL_MAX_N_PREFETCH) {

		/* The prefetch buffer is now full */

		sel_dequeue_prefetched_row(plan);

		goto next_table;
	}

next_rec:
	ut_ad(!search_latch_locked);

	if (mtr_has_extra_clust_latch) {

		/* We must commit &mtr if we are moving to the next
		non-clustered index record, because we could break the
		latching order if we would access a different clustered
		index page right away without releasing the previous. */

		goto commit_mtr_for_a_while;
	}

	if (node->asc) {
		moved = btr_pcur_move_to_next(&(plan->pcur), &mtr);
	} else {
		moved = btr_pcur_move_to_prev(&(plan->pcur), &mtr);
	}

	if (!moved) {

		goto table_exhausted;
	}

	cursor_just_opened = FALSE;

	/* END OF RECORD LOOP
	------------------ */
	goto rec_loop;

next_table:
	/* We found a record which satisfies the conditions: we can move to
	the next table or return a row in the result set */

	ut_ad(btr_pcur_is_on_user_rec(&plan->pcur));

	if (plan->unique_search && !node->can_get_updated) {

		plan->cursor_at_end = TRUE;
	} else {
		ut_ad(!search_latch_locked);

		plan->stored_cursor_rec_processed = TRUE;

		btr_pcur_store_position(&(plan->pcur), &mtr);
	}

	mtr_commit(&mtr);

	mtr_has_extra_clust_latch = FALSE;

next_table_no_mtr:
	/* If we use 'goto' to this label, it means that the row was popped
	from the prefetched rows stack, and &mtr is already committed */

	if (node->fetch_table + 1 == node->n_tables) {

		sel_eval_select_list(node);

		if (node->is_aggregate) {

			goto table_loop;
		}

		sel_assign_into_var_values(node->into_list, node);

		thr->run_node = que_node_get_parent(node);

		err = DB_SUCCESS;
		goto func_exit;
	}

	node->fetch_table++;

	/* When we move to the next table, we first reset the plan cursor:
	we do not care about resetting it when we backtrack from a table */

	plan_reset_cursor(sel_node_get_nth_plan(node, node->fetch_table));

	goto table_loop;

table_exhausted:
	/* The table cursor pcur reached the result set end: backtrack to the
	previous table in the join if we do not have cached prefetched rows */

	plan->cursor_at_end = TRUE;

	mtr_commit(&mtr);

	mtr_has_extra_clust_latch = FALSE;

	if (plan->n_rows_prefetched > 0) {
		/* The table became exhausted during a prefetch */

		sel_dequeue_prefetched_row(plan);

		goto next_table_no_mtr;
	}

table_exhausted_no_mtr:
	if (node->fetch_table == 0) {
		err = DB_SUCCESS;

		if (node->is_aggregate && !node->aggregate_already_fetched) {

			node->aggregate_already_fetched = TRUE;

			sel_assign_into_var_values(node->into_list, node);

			thr->run_node = que_node_get_parent(node);
		} else {
			node->state = SEL_NODE_NO_MORE_ROWS;

			thr->run_node = que_node_get_parent(node);
		}

		goto func_exit;
	}

	node->fetch_table--;

	goto table_loop;

stop_for_a_while:
	/* Return control for a while to que_run_threads, so that runaway
	queries can be canceled. NOTE that when we come here, we must, in a
	locking read, have placed the necessary (possibly waiting request)
	record lock on the cursor record or its successor: when we reposition
	the cursor, this record lock guarantees that nobody can meanwhile have
	inserted new records which should have appeared in the result set,
	which would result in the phantom problem. */

	ut_ad(!search_latch_locked);

	plan->stored_cursor_rec_processed = FALSE;
	btr_pcur_store_position(&(plan->pcur), &mtr);

	mtr_commit(&mtr);

#ifdef UNIV_SYNC_DEBUG
	ut_ad(sync_thread_levels_empty_except_dict());
#endif /* UNIV_SYNC_DEBUG */
	err = DB_SUCCESS;
	goto func_exit;

commit_mtr_for_a_while:
	/* Stores the cursor position and commits &mtr; this is used if
	&mtr may contain latches which would break the latching order if
	&mtr would not be committed and the latches released. */

	plan->stored_cursor_rec_processed = TRUE;

	ut_ad(!search_latch_locked);
	btr_pcur_store_position(&(plan->pcur), &mtr);

	mtr_commit(&mtr);

	mtr_has_extra_clust_latch = FALSE;

#ifdef UNIV_SYNC_DEBUG
	ut_ad(sync_thread_levels_empty_except_dict());
#endif /* UNIV_SYNC_DEBUG */

	goto table_loop;

lock_wait_or_error:
	/* See the note at stop_for_a_while: the same holds for this case */

	ut_ad(!btr_pcur_is_before_first_on_page(&plan->pcur) || !node->asc);
	ut_ad(!search_latch_locked);

	plan->stored_cursor_rec_processed = FALSE;
	btr_pcur_store_position(&(plan->pcur), &mtr);

	mtr_commit(&mtr);

#ifdef UNIV_SYNC_DEBUG
	ut_ad(sync_thread_levels_empty_except_dict());
#endif /* UNIV_SYNC_DEBUG */

func_exit:
	if (search_latch_locked) {
		rw_lock_s_unlock(btr_search_get_latch(index));
	}
	if (UNIV_LIKELY_NULL(heap)) {
		mem_heap_free(heap);
	}
	return(err);
}

/**********************************************************************//**
Performs a select step. This is a high-level function used in SQL execution
graphs.
@return	query thread to run next or NULL */
UNIV_INTERN
que_thr_t*
row_sel_step(
/*=========*/
	que_thr_t*	thr)	/*!< in: query thread */
{
	sel_node_t*	node;

	ut_ad(thr);

	node = static_cast<sel_node_t*>(thr->run_node);

	ut_ad(que_node_get_type(node) == QUE_NODE_SELECT);

	/* If this is a new time this node is executed (or when execution
	resumes after wait for a table intention lock), set intention locks
	on the tables, or assign a read view */

	if (node->into_list && (thr->prev_node == que_node_get_parent(node))) {

		node->state = SEL_NODE_OPEN;
	}

	if (node->state == SEL_NODE_OPEN) {

		/* It may be that the current session has not yet started
		its transaction, or it has been committed: */

		trx_start_if_not_started_xa(thr_get_trx(thr));

		plan_reset_cursor(sel_node_get_nth_plan(node, 0));

		if (node->consistent_read) {
			/* Assign a read view for the query */
			node->read_view = trx_assign_read_view(
				thr_get_trx(thr));
		} else {
			sym_node_t*	table_node;
			enum lock_mode	i_lock_mode;

			if (node->set_x_locks) {
				i_lock_mode = LOCK_IX;
			} else {
				i_lock_mode = LOCK_IS;
			}

			for (table_node = node->table_list;
			     table_node != 0;
			     table_node = static_cast<sym_node_t*>(
					que_node_get_next(table_node))) {

				dberr_t	err = lock_table(
					0, table_node->table, i_lock_mode,
					thr);

				if (err != DB_SUCCESS) {
					trx_t*	trx;

					trx = thr_get_trx(thr);
					trx->error_state = err;

					return(NULL);
				}
			}
		}

		/* If this is an explicit cursor, copy stored procedure
		variable values, so that the values cannot change between
		fetches (currently, we copy them also for non-explicit
		cursors) */

		if (node->explicit_cursor
		    && UT_LIST_GET_FIRST(node->copy_variables)) {

			row_sel_copy_input_variable_vals(node);
		}

		node->state = SEL_NODE_FETCH;
		node->fetch_table = 0;

		if (node->is_aggregate) {
			/* Reset the aggregate total values */
			sel_reset_aggregate_vals(node);
		}
	}

	dberr_t	err = row_sel(node, thr);

	/* NOTE! if queries are parallelized, the following assignment may
	have problems; the assignment should be made only if thr is the
	only top-level thr in the graph: */

	thr->graph->last_sel_node = node;

	if (err != DB_SUCCESS) {
		thr_get_trx(thr)->error_state = err;

		return(NULL);
	}

	return(thr);
}

/**********************************************************************//**
Performs a fetch for a cursor.
@return	query thread to run next or NULL */
UNIV_INTERN
que_thr_t*
fetch_step(
/*=======*/
	que_thr_t*	thr)	/*!< in: query thread */
{
	sel_node_t*	sel_node;
	fetch_node_t*	node;

	ut_ad(thr);

	node = static_cast<fetch_node_t*>(thr->run_node);
	sel_node = node->cursor_def;

	ut_ad(que_node_get_type(node) == QUE_NODE_FETCH);

	if (thr->prev_node != que_node_get_parent(node)) {

		if (sel_node->state != SEL_NODE_NO_MORE_ROWS) {

			if (node->into_list) {
				sel_assign_into_var_values(node->into_list,
							   sel_node);
			} else {
				ibool ret = (*node->func->func)(
					sel_node, node->func->arg);

				if (!ret) {
					sel_node->state
						 = SEL_NODE_NO_MORE_ROWS;
				}
			}
		}

		thr->run_node = que_node_get_parent(node);

		return(thr);
	}

	/* Make the fetch node the parent of the cursor definition for
	the time of the fetch, so that execution knows to return to this
	fetch node after a row has been selected or we know that there is
	no row left */

	sel_node->common.parent = node;

	if (sel_node->state == SEL_NODE_CLOSED) {
		fprintf(stderr,
			"InnoDB: Error: fetch called on a closed cursor\n");

		thr_get_trx(thr)->error_state = DB_ERROR;

		return(NULL);
	}

	thr->run_node = sel_node;

	return(thr);
}

/****************************************************************//**
Sample callback function for fetch that prints each row.
@return	always returns non-NULL */
UNIV_INTERN
void*
row_fetch_print(
/*============*/
	void*	row,		/*!< in:  sel_node_t* */
	void*	user_arg)	/*!< in:  not used */
{
	que_node_t*	exp;
	ulint		i = 0;
	sel_node_t*	node = static_cast<sel_node_t*>(row);

	UT_NOT_USED(user_arg);

	fprintf(stderr, "row_fetch_print: row %p\n", row);

	for (exp = node->select_list;
	     exp != 0;
	     exp = que_node_get_next(exp), i++) {

		dfield_t*	dfield = que_node_get_val(exp);
		const dtype_t*	type = dfield_get_type(dfield);

		fprintf(stderr, " column %lu:\n", (ulong) i);

		dtype_print(type);
		putc('\n', stderr);

		if (dfield_get_len(dfield) != UNIV_SQL_NULL) {
			ut_print_buf(stderr, dfield_get_data(dfield),
				     dfield_get_len(dfield));
			putc('\n', stderr);
		} else {
			fputs(" <NULL>;\n", stderr);
		}
	}

	return((void*)42);
}

/***********************************************************//**
Prints a row in a select result.
@return	query thread to run next or NULL */
UNIV_INTERN
que_thr_t*
row_printf_step(
/*============*/
	que_thr_t*	thr)	/*!< in: query thread */
{
	row_printf_node_t*	node;
	sel_node_t*		sel_node;
	que_node_t*		arg;

	ut_ad(thr);

	node = static_cast<row_printf_node_t*>(thr->run_node);

	sel_node = node->sel_node;

	ut_ad(que_node_get_type(node) == QUE_NODE_ROW_PRINTF);

	if (thr->prev_node == que_node_get_parent(node)) {

		/* Reset the cursor */
		sel_node->state = SEL_NODE_OPEN;

		/* Fetch next row to print */

		thr->run_node = sel_node;

		return(thr);
	}

	if (sel_node->state != SEL_NODE_FETCH) {

		ut_ad(sel_node->state == SEL_NODE_NO_MORE_ROWS);

		/* No more rows to print */

		thr->run_node = que_node_get_parent(node);

		return(thr);
	}

	arg = sel_node->select_list;

	while (arg) {
		dfield_print_also_hex(que_node_get_val(arg));

		fputs(" ::: ", stderr);

		arg = que_node_get_next(arg);
	}

	putc('\n', stderr);

	/* Fetch next row to print */

	thr->run_node = sel_node;

	return(thr);
}

/****************************************************************//**
Converts a key value stored in MySQL format to an Innobase dtuple. The last
field of the key value may be just a prefix of a fixed length field: hence
the parameter key_len. But currently we do not allow search keys where the
last field is only a prefix of the full key field len and print a warning if
such appears. A counterpart of this function is
ha_innobase::store_key_val_for_row() in ha_innodb.cc. */
UNIV_INTERN
void
row_sel_convert_mysql_key_to_innobase(
/*==================================*/
	dtuple_t*	tuple,		/*!< in/out: tuple where to build;
					NOTE: we assume that the type info
					in the tuple is already according
					to index! */
	byte*		buf,		/*!< in: buffer to use in field
					conversions; NOTE that dtuple->data
					may end up pointing inside buf so
					do not discard that buffer while
					the tuple is being used. See
					row_mysql_store_col_in_innobase_format()
					in the case of DATA_INT */
	ulint		buf_len,	/*!< in: buffer length */
	dict_index_t*	index,		/*!< in: index of the key value */
	const byte*	key_ptr,	/*!< in: MySQL key value */
	ulint		key_len,	/*!< in: MySQL key value length */
	trx_t*		trx)		/*!< in: transaction */
{
	byte*		original_buf	= buf;
	const byte*	original_key_ptr = key_ptr;
	dict_field_t*	field;
	dfield_t*	dfield;
	ulint		data_offset;
	ulint		data_len;
	ulint		data_field_len;
	ibool		is_null;
	const byte*	key_end;
	ulint		n_fields = 0;

	/* For documentation of the key value storage format in MySQL, see
	ha_innobase::store_key_val_for_row() in ha_innodb.cc. */

	key_end = key_ptr + key_len;

	/* Permit us to access any field in the tuple (ULINT_MAX): */

	dtuple_set_n_fields(tuple, ULINT_MAX);

	dfield = dtuple_get_nth_field(tuple, 0);
	field = dict_index_get_nth_field(index, 0);

	if (UNIV_UNLIKELY(dfield_get_type(dfield)->mtype == DATA_SYS)) {
		/* A special case: we are looking for a position in the
		generated clustered index which InnoDB automatically added
		to a table with no primary key: the first and the only
		ordering column is ROW_ID which InnoDB stored to the key_ptr
		buffer. */

		ut_a(key_len == DATA_ROW_ID_LEN);

		dfield_set_data(dfield, key_ptr, DATA_ROW_ID_LEN);

		dtuple_set_n_fields(tuple, 1);

		return;
	}

	while (key_ptr < key_end) {

		ulint	type = dfield_get_type(dfield)->mtype;
		ut_a(field->col->mtype == type);

		data_offset = 0;
		is_null = FALSE;

		if (!(dfield_get_type(dfield)->prtype & DATA_NOT_NULL)) {
			/* The first byte in the field tells if this is
			an SQL NULL value */

			data_offset = 1;

			if (*key_ptr != 0) {
				dfield_set_null(dfield);

				is_null = TRUE;
			}
		}

		/* Calculate data length and data field total length */

		if (type == DATA_BLOB) {
			/* The key field is a column prefix of a BLOB or
			TEXT */

			ut_a(field->prefix_len > 0);

			/* MySQL stores the actual data length to the first 2
			bytes after the optional SQL NULL marker byte. The
			storage format is little-endian, that is, the most
			significant byte at a higher address. In UTF-8, MySQL
			seems to reserve field->prefix_len bytes for
			storing this field in the key value buffer, even
			though the actual value only takes data_len bytes
			from the start. */

			data_len = key_ptr[data_offset]
				+ 256 * key_ptr[data_offset + 1];
			data_field_len = data_offset + 2 + field->prefix_len;

			data_offset += 2;

			/* Now that we know the length, we store the column
			value like it would be a fixed char field */

		} else if (field->prefix_len > 0) {
			/* Looks like MySQL pads unused end bytes in the
			prefix with space. Therefore, also in UTF-8, it is ok
			to compare with a prefix containing full prefix_len
			bytes, and no need to take at most prefix_len / 3
			UTF-8 characters from the start.
			If the prefix is used as the upper end of a LIKE
			'abc%' query, then MySQL pads the end with chars
			0xff. TODO: in that case does it any harm to compare
			with the full prefix_len bytes. How do characters
			0xff in UTF-8 behave? */

			data_len = field->prefix_len;
			data_field_len = data_offset + data_len;
		} else {
			data_len = dfield_get_type(dfield)->len;
			data_field_len = data_offset + data_len;
		}

		if (UNIV_UNLIKELY
		    (dtype_get_mysql_type(dfield_get_type(dfield))
		     == DATA_MYSQL_TRUE_VARCHAR)
		    && UNIV_LIKELY(type != DATA_INT)) {
			/* In a MySQL key value format, a true VARCHAR is
			always preceded by 2 bytes of a length field.
			dfield_get_type(dfield)->len returns the maximum
			'payload' len in bytes. That does not include the
			2 bytes that tell the actual data length.

			We added the check != DATA_INT to make sure we do
			not treat MySQL ENUM or SET as a true VARCHAR! */

			data_len += 2;
			data_field_len += 2;
		}

		/* Storing may use at most data_len bytes of buf */

		if (UNIV_LIKELY(!is_null)) {
			ut_a(buf + data_len <= original_buf + buf_len);
			row_mysql_store_col_in_innobase_format(
				dfield, buf,
				FALSE, /* MySQL key value format col */
				key_ptr + data_offset, data_len,
				dict_table_is_comp(index->table));
			buf += data_len;
		}

		key_ptr += data_field_len;

		if (UNIV_UNLIKELY(key_ptr > key_end)) {
			/* The last field in key was not a complete key field
			but a prefix of it.

			Print a warning about this! HA_READ_PREFIX_LAST does
			not currently work in InnoDB with partial-field key
			value prefixes. Since MySQL currently uses a padding
			trick to calculate LIKE 'abc%' type queries there
			should never be partial-field prefixes in searches. */

			ut_print_timestamp(stderr);

			fputs("  InnoDB: Warning: using a partial-field"
			      " key prefix in search.\n"
			      "InnoDB: ", stderr);
			dict_index_name_print(stderr, trx, index);
			fprintf(stderr, ". Last data field length %lu bytes,\n"
				"InnoDB: key ptr now exceeds"
				" key end by %lu bytes.\n"
				"InnoDB: Key value in the MySQL format:\n",
				(ulong) data_field_len,
				(ulong) (key_ptr - key_end));
			fflush(stderr);
			ut_print_buf(stderr, original_key_ptr, key_len);
			putc('\n', stderr);

			if (!is_null) {
				ulint	len = dfield_get_len(dfield);
				dfield_set_len(dfield, len
					       - (ulint) (key_ptr - key_end));
			}
                        ut_ad(0);
		}

		n_fields++;
		field++;
		dfield++;
	}

	DBUG_EXECUTE_IF("innodb_srch_key_buffer_full",
		ut_a(buf == (original_buf + buf_len)););

	ut_a(buf <= original_buf + buf_len);

	/* We set the length of tuple to n_fields: we assume that the memory
	area allocated for it is big enough (usually bigger than n_fields). */

	dtuple_set_n_fields(tuple, n_fields);
}

/**************************************************************//**
Stores the row id to the prebuilt struct. */
static
void
row_sel_store_row_id_to_prebuilt(
/*=============================*/
	row_prebuilt_t*		prebuilt,	/*!< in/out: prebuilt */
	const rec_t*		index_rec,	/*!< in: record */
	const dict_index_t*	index,		/*!< in: index of the record */
	const ulint*		offsets)	/*!< in: rec_get_offsets
						(index_rec, index) */
{
	const byte*	data;
	ulint		len;

	ut_ad(rec_offs_validate(index_rec, index, offsets));

	data = rec_get_nth_field(
		index_rec, offsets,
		dict_index_get_sys_col_pos(index, DATA_ROW_ID), &len);

	if (UNIV_UNLIKELY(len != DATA_ROW_ID_LEN)) {
		fprintf(stderr,
			"InnoDB: Error: Row id field is"
			" wrong length %lu in ", (ulong) len);
		dict_index_name_print(stderr, prebuilt->trx, index);
		fprintf(stderr, "\n"
			"InnoDB: Field number %lu, record:\n",
			(ulong) dict_index_get_sys_col_pos(index,
							   DATA_ROW_ID));
		rec_print_new(stderr, index_rec, offsets);
		putc('\n', stderr);
		ut_error;
	}

	ut_memcpy(prebuilt->row_id, data, len);
}

#ifdef UNIV_DEBUG
/** Convert a non-SQL-NULL field from Innobase format to MySQL format. */
# define row_sel_field_store_in_mysql_format(dest,templ,idx,field,src,len) \
	row_sel_field_store_in_mysql_format_func(dest,templ,idx,field,src,len)
#else /* UNIV_DEBUG */
/** Convert a non-SQL-NULL field from Innobase format to MySQL format. */
# define row_sel_field_store_in_mysql_format(dest,templ,idx,field,src,len) \
	row_sel_field_store_in_mysql_format_func(dest,templ,src,len)
#endif /* UNIV_DEBUG */

/**************************************************************//**
Stores a non-SQL-NULL field in the MySQL format. The counterpart of this
function is row_mysql_store_col_in_innobase_format() in row0mysql.cc. */
static __attribute__((nonnull))
void
row_sel_field_store_in_mysql_format_func(
/*=====================================*/
	byte*		dest,	/*!< in/out: buffer where to store; NOTE
				that BLOBs are not in themselves
				stored here: the caller must allocate
				and copy the BLOB into buffer before,
				and pass the pointer to the BLOB in
				'data' */
	const mysql_row_templ_t* templ,
				/*!< in: MySQL column template.
				Its following fields are referenced:
				type, is_unsigned, mysql_col_len,
				mbminlen, mbmaxlen */
#ifdef UNIV_DEBUG
	const dict_index_t* index,
				/*!< in: InnoDB index */
	ulint		field_no,
				/*!< in: templ->rec_field_no or
				templ->clust_rec_field_no or
				templ->icp_rec_field_no */
#endif /* UNIV_DEBUG */
	const byte*	data,	/*!< in: data to store */
	ulint		len)	/*!< in: length of the data */
{
	byte*			ptr;
#ifdef UNIV_DEBUG
	const dict_field_t*	field
		= dict_index_get_nth_field(index, field_no);
#endif /* UNIV_DEBUG */

	ut_ad(len != UNIV_SQL_NULL);
	UNIV_MEM_ASSERT_RW(data, len);
	UNIV_MEM_ASSERT_W(dest, templ->mysql_col_len);
	UNIV_MEM_INVALID(dest, templ->mysql_col_len);

	switch (templ->type) {
		const byte*	field_end;
		byte*		pad;
	case DATA_INT:
		/* Convert integer data from Innobase to a little-endian
		format, sign bit restored to normal */

		ptr = dest + len;

		for (;;) {
			ptr--;
			*ptr = *data;
			if (ptr == dest) {
				break;
			}
			data++;
		}

		if (!templ->is_unsigned) {
			dest[len - 1] = (byte) (dest[len - 1] ^ 128);
		}

		ut_ad(templ->mysql_col_len == len);
		break;

	case DATA_VARCHAR:
	case DATA_VARMYSQL:
	case DATA_BINARY:
		field_end = dest + templ->mysql_col_len;

		if (templ->mysql_type == DATA_MYSQL_TRUE_VARCHAR) {
			/* This is a >= 5.0.3 type true VARCHAR. Store the
			length of the data to the first byte or the first
			two bytes of dest. */

			dest = row_mysql_store_true_var_len(
				dest, len, templ->mysql_length_bytes);
			/* Copy the actual data. Leave the rest of the
			buffer uninitialized. */
			memcpy(dest, data, len);
			break;
		}

		/* Copy the actual data */
		ut_memcpy(dest, data, len);

		/* Pad with trailing spaces. */

		pad = dest + len;

		ut_ad(templ->mbminlen <= templ->mbmaxlen);

		/* We treat some Unicode charset strings specially. */
		switch (templ->mbminlen) {
		case 4:
			/* InnoDB should never have stripped partial
			UTF-32 characters. */
			ut_a(!(len & 3));
			break;
		case 2:
			/* A space char is two bytes,
			0x0020 in UCS2 and UTF-16 */

			if (UNIV_UNLIKELY(len & 1)) {
				/* A 0x20 has been stripped from the column.
				Pad it back. */

				if (pad < field_end) {
					*pad++ = 0x20;
				}
			}
		}

		row_mysql_pad_col(templ->mbminlen, pad, field_end - pad);
		break;

	case DATA_BLOB:
		/* Store a pointer to the BLOB buffer to dest: the BLOB was
		already copied to the buffer in row_sel_store_mysql_rec */

		row_mysql_store_blob_ref(dest, templ->mysql_col_len, data,
					 len);
		break;

	case DATA_MYSQL:
		memcpy(dest, data, len);

		ut_ad(templ->mysql_col_len >= len);
		ut_ad(templ->mbmaxlen >= templ->mbminlen);

		/* If field_no equals to templ->icp_rec_field_no,
		we are examining a row pointed by "icp_rec_field_no".
		There is possibility that icp_rec_field_no refers to
		a field in a secondary index while templ->rec_field_no
		points to field in a primary index. The length
		should still be equal, unless the field pointed
		by icp_rec_field_no has a prefix */
		ut_ad(templ->mbmaxlen > templ->mbminlen
		      || templ->mysql_col_len == len
		      || (field_no == templ->icp_rec_field_no
			  && field->prefix_len > 0));

		/* The following assertion would fail for old tables
		containing UTF-8 ENUM columns due to Bug #9526. */
		ut_ad(!templ->mbmaxlen
		      || !(templ->mysql_col_len % templ->mbmaxlen));
		ut_ad(len * templ->mbmaxlen >= templ->mysql_col_len
		      || (field_no == templ->icp_rec_field_no
			  && field->prefix_len > 0));
		ut_ad(!(field->prefix_len % templ->mbmaxlen));

		if (templ->mbminlen == 1 && templ->mbmaxlen != 1) {
			/* Pad with spaces. This undoes the stripping
			done in row0mysql.cc, function
			row_mysql_store_col_in_innobase_format(). */

			memset(dest + len, 0x20, templ->mysql_col_len - len);
		}
		break;

	default:
#ifdef UNIV_DEBUG
	case DATA_SYS_CHILD:
	case DATA_SYS:
		/* These column types should never be shipped to MySQL. */
		ut_ad(0);

	case DATA_CHAR:
	case DATA_FIXBINARY:
	case DATA_FLOAT:
	case DATA_DOUBLE:
	case DATA_DECIMAL:
		/* Above are the valid column types for MySQL data. */
#endif /* UNIV_DEBUG */
		ut_ad(field->prefix_len
		      ? field->prefix_len == len
		      : templ->mysql_col_len == len);
		memcpy(dest, data, len);
	}
}

#ifdef UNIV_DEBUG
/** Convert a field from Innobase format to MySQL format. */
# define row_sel_store_mysql_field(m,p,r,i,o,f,t) \
	row_sel_store_mysql_field_func(m,p,r,i,o,f,t)
#else /* UNIV_DEBUG */
/** Convert a field from Innobase format to MySQL format. */
# define row_sel_store_mysql_field(m,p,r,i,o,f,t) \
	row_sel_store_mysql_field_func(m,p,r,o,f,t)
#endif /* UNIV_DEBUG */
/**************************************************************//**
Convert a field in the Innobase format to a field in the MySQL format. */
static __attribute__((warn_unused_result))
ibool
row_sel_store_mysql_field_func(
/*===========================*/
	byte*			mysql_rec,	/*!< out: record in the
						MySQL format */
	row_prebuilt_t*		prebuilt,	/*!< in/out: prebuilt struct */
	const rec_t*		rec,		/*!< in: InnoDB record;
						must be protected by
						a page latch */
#ifdef UNIV_DEBUG
	const dict_index_t*	index,		/*!< in: index of rec */
#endif
	const ulint*		offsets,	/*!< in: array returned by
						rec_get_offsets() */
	ulint			field_no,	/*!< in: templ->rec_field_no or
						templ->clust_rec_field_no or
						templ->icp_rec_field_no */
	const mysql_row_templ_t*templ)		/*!< in: row template */
{
	const byte*	data;
	ulint		len;

	ut_ad(prebuilt->default_rec);
	ut_ad(templ);
	ut_ad(templ >= prebuilt->mysql_template);
	ut_ad(templ < &prebuilt->mysql_template[prebuilt->n_template]);
	ut_ad(field_no == templ->clust_rec_field_no
	      || field_no == templ->rec_field_no
	      || field_no == templ->icp_rec_field_no);
	ut_ad(rec_offs_validate(rec, index, offsets));

	if (UNIV_UNLIKELY(rec_offs_nth_extern(offsets, field_no))) {

		mem_heap_t*	heap;
		/* Copy an externally stored field to a temporary heap */

		ut_a(!prebuilt->trx->has_search_latch);
#ifdef UNIV_SYNC_DEBUG
		ut_ad(!btr_search_own_any());
#endif
		ut_ad(field_no == templ->clust_rec_field_no);

		if (UNIV_UNLIKELY(templ->type == DATA_BLOB)) {
			if (prebuilt->blob_heap == NULL) {
				prebuilt->blob_heap = mem_heap_create(
					UNIV_PAGE_SIZE);
			}

			heap = prebuilt->blob_heap;
		} else {
			heap = mem_heap_create(UNIV_PAGE_SIZE);
		}

		/* NOTE: if we are retrieving a big BLOB, we may
		already run out of memory in the next call, which
		causes an assert */

		data = btr_rec_copy_externally_stored_field(
			rec, offsets,
			dict_table_zip_size(prebuilt->table),
			field_no, &len, heap);

		if (UNIV_UNLIKELY(!data)) {
			/* The externally stored field was not written
			yet. This record should only be seen by
			recv_recovery_rollback_active() or any
			TRX_ISO_READ_UNCOMMITTED transactions. */

			if (heap != prebuilt->blob_heap) {
				mem_heap_free(heap);
			}

			ut_a(prebuilt->trx->isolation_level
			     == TRX_ISO_READ_UNCOMMITTED);
			return(FALSE);
		}

		ut_a(len != UNIV_SQL_NULL);

		row_sel_field_store_in_mysql_format(
			mysql_rec + templ->mysql_col_offset,
			templ, index, field_no, data, len);

		if (heap != prebuilt->blob_heap) {
			mem_heap_free(heap);
		}
	} else {
		/* Field is stored in the row. */

		data = rec_get_nth_field(rec, offsets, field_no, &len);

		if (len == UNIV_SQL_NULL) {
			/* MySQL assumes that the field for an SQL
			NULL value is set to the default value. */
			ut_ad(templ->mysql_null_bit_mask);

			UNIV_MEM_ASSERT_RW(prebuilt->default_rec
					   + templ->mysql_col_offset,
					   templ->mysql_col_len);
			mysql_rec[templ->mysql_null_byte_offset]
				|= (byte) templ->mysql_null_bit_mask;
			memcpy(mysql_rec + templ->mysql_col_offset,
			       (const byte*) prebuilt->default_rec
			       + templ->mysql_col_offset,
			       templ->mysql_col_len);
			return(TRUE);
		}

		if (UNIV_UNLIKELY(templ->type == DATA_BLOB)) {

			/* It is a BLOB field locally stored in the
			InnoDB record: we MUST copy its contents to
			prebuilt->blob_heap here because
			row_sel_field_store_in_mysql_format() stores a
			pointer to the data, and the data passed to us
			will be invalid as soon as the
			mini-transaction is committed and the page
			latch on the clustered index page is
			released. */

			if (prebuilt->blob_heap == NULL) {
				prebuilt->blob_heap = mem_heap_create(
					UNIV_PAGE_SIZE);
			}

			data = static_cast<byte*>(
				mem_heap_dup(prebuilt->blob_heap, data, len));
		}

		row_sel_field_store_in_mysql_format(
			mysql_rec + templ->mysql_col_offset,
			templ, index, field_no, data, len);
	}

	ut_ad(len != UNIV_SQL_NULL);

	if (templ->mysql_null_bit_mask) {
		/* It is a nullable column with a non-NULL
		value */
		mysql_rec[templ->mysql_null_byte_offset]
			&= ~(byte) templ->mysql_null_bit_mask;
	}

	return(TRUE);
}

/**************************************************************//**
Convert a row in the Innobase format to a row in the MySQL format.
Note that the template in prebuilt may advise us to copy only a few
columns to mysql_rec, other columns are left blank. All columns may not
be needed in the query.
@return TRUE on success, FALSE if not all columns could be retrieved */
static __attribute__((warn_unused_result))
ibool
row_sel_store_mysql_rec(
/*====================*/
	byte*		mysql_rec,	/*!< out: row in the MySQL format */
	row_prebuilt_t*	prebuilt,	/*!< in: prebuilt struct */
	const rec_t*	rec,		/*!< in: Innobase record in the index
					which was described in prebuilt's
					template, or in the clustered index;
					must be protected by a page latch */
	ibool		rec_clust,	/*!< in: TRUE if rec is in the
					clustered index instead of
					prebuilt->index */
	const dict_index_t* index,	/*!< in: index of rec */
	const ulint*	offsets)	/*!< in: array returned by
 					rec_get_offsets(rec) */
{
	ulint	i;

	ut_ad(rec_clust || index == prebuilt->index);
	ut_ad(!rec_clust || dict_index_is_clust(index));

	if (UNIV_LIKELY_NULL(prebuilt->blob_heap)) {
		mem_heap_free(prebuilt->blob_heap);
		prebuilt->blob_heap = NULL;
	}

	for (i = 0; i < prebuilt->n_template; i++) {
		const mysql_row_templ_t*templ = &prebuilt->mysql_template[i];
		const ulint		field_no
			= rec_clust
			? templ->clust_rec_field_no
			: templ->rec_field_no;
		/* We should never deliver column prefixes to MySQL,
		except for evaluating innobase_index_cond(). */
		/* ...actually, we do want to do this in order to
		support the prefix query optimization.

		ut_ad(dict_index_get_nth_field(index, field_no)->prefix_len
		      == 0);

		...so we disable this assert. */

		if (!row_sel_store_mysql_field(mysql_rec, prebuilt,
					       rec, index, offsets,
					       field_no, templ)) {
			return(FALSE);
		}
	}

	/* FIXME: We only need to read the doc_id if an FTS indexed
	column is being updated.
	NOTE, the record must be cluster index record. Secondary index
	might not have the Doc ID */
	if (dict_table_has_fts_index(prebuilt->table)
	    && dict_index_is_clust(index)) {

		prebuilt->fts_doc_id = fts_get_doc_id_from_rec(
			prebuilt->table, rec, NULL);
	}

	return(TRUE);
}

/*********************************************************************//**
Builds a previous version of a clustered index record for a consistent read
@return	DB_SUCCESS or error code */
static __attribute__((nonnull, warn_unused_result))
dberr_t
row_sel_build_prev_vers_for_mysql(
/*==============================*/
	read_view_t*	read_view,	/*!< in: read view */
	dict_index_t*	clust_index,	/*!< in: clustered index */
	row_prebuilt_t*	prebuilt,	/*!< in: prebuilt struct */
	const rec_t*	rec,		/*!< in: record in a clustered index */
	ulint**		offsets,	/*!< in/out: offsets returned by
					rec_get_offsets(rec, clust_index) */
	mem_heap_t**	offset_heap,	/*!< in/out: memory heap from which
					the offsets are allocated */
	rec_t**		old_vers,	/*!< out: old version, or NULL if the
					record does not exist in the view:
					i.e., it was freshly inserted
					afterwards */
	mtr_t*		mtr)		/*!< in: mtr */
{
	dberr_t	err;

	if (prebuilt->old_vers_heap) {
		mem_heap_empty(prebuilt->old_vers_heap);
	} else {
		prebuilt->old_vers_heap = mem_heap_create(200);
	}

	err = row_vers_build_for_consistent_read(
		rec, mtr, clust_index, offsets, read_view, offset_heap,
		prebuilt->old_vers_heap, old_vers);
	return(err);
}

/*********************************************************************//**
Retrieves the clustered index record corresponding to a record in a
non-clustered index. Does the necessary locking. Used in the MySQL
interface.
@return	DB_SUCCESS, DB_SUCCESS_LOCKED_REC, or error code */
static __attribute__((nonnull, warn_unused_result))
dberr_t
row_sel_get_clust_rec_for_mysql(
/*============================*/
	row_prebuilt_t*	prebuilt,/*!< in: prebuilt struct in the handle */
	dict_index_t*	sec_index,/*!< in: secondary index where rec resides */
	const rec_t*	rec,	/*!< in: record in a non-clustered index; if
				this is a locking read, then rec is not
				allowed to be delete-marked, and that would
				not make sense either */
	que_thr_t*	thr,	/*!< in: query thread */
	const rec_t**	out_rec,/*!< out: clustered record or an old version of
				it, NULL if the old version did not exist
				in the read view, i.e., it was a fresh
				inserted version */
	ulint**		offsets,/*!< in: offsets returned by
				rec_get_offsets(rec, sec_index);
				out: offsets returned by
				rec_get_offsets(out_rec, clust_index) */
	mem_heap_t**	offset_heap,/*!< in/out: memory heap from which
				the offsets are allocated */
	mtr_t*		mtr)	/*!< in: mtr used to get access to the
				non-clustered record; the same mtr is used to
				access the clustered index */
{
	dict_index_t*	clust_index;
	const rec_t*	clust_rec;
	rec_t*		old_vers;
	dberr_t		err;
	trx_t*		trx;

	srv_stats.n_sec_rec_cluster_reads.inc();

	*out_rec = NULL;
	trx = thr_get_trx(thr);

	row_build_row_ref_in_tuple(prebuilt->clust_ref, rec,
				   sec_index, *offsets, trx);

	clust_index = dict_table_get_first_index(sec_index->table);

	btr_pcur_open_with_no_init(clust_index, prebuilt->clust_ref,
				   PAGE_CUR_LE, BTR_SEARCH_LEAF,
				   &prebuilt->clust_pcur, 0, mtr);

	clust_rec = btr_pcur_get_rec(&prebuilt->clust_pcur);

	prebuilt->clust_pcur.trx_if_known = trx;

	/* Note: only if the search ends up on a non-infimum record is the
	low_match value the real match to the search tuple */

	if (!page_rec_is_user_rec(clust_rec)
	    || btr_pcur_get_low_match(&prebuilt->clust_pcur)
	    < dict_index_get_n_unique(clust_index)) {

		/* In a rare case it is possible that no clust rec is found
		for a delete-marked secondary index record: if in row0umod.cc
		in row_undo_mod_remove_clust_low() we have already removed
		the clust rec, while purge is still cleaning and removing
		secondary index records associated with earlier versions of
		the clustered index record. In that case we know that the
		clustered index record did not exist in the read view of
		trx. */

		if (!rec_get_deleted_flag(rec,
					  dict_table_is_comp(sec_index->table))
		    || prebuilt->select_lock_type != LOCK_NONE) {
			ut_print_timestamp(stderr);
			fputs("  InnoDB: error clustered record"
			      " for sec rec not found\n"
			      "InnoDB: ", stderr);
			dict_index_name_print(stderr, trx, sec_index);
			fputs("\n"
			      "InnoDB: sec index record ", stderr);
			rec_print(stderr, rec, sec_index);
			fputs("\n"
			      "InnoDB: clust index record ", stderr);
			rec_print(stderr, clust_rec, clust_index);
			putc('\n', stderr);
			trx_print(stderr, trx, 600);
			fputs("\n"
			      "InnoDB: Submit a detailed bug report"
			      " to http://bugs.mysql.com\n", stderr);
			ut_ad(0);
		}

		clust_rec = NULL;

		err = DB_SUCCESS;
		goto func_exit;
	}

	*offsets = rec_get_offsets(clust_rec, clust_index, *offsets,
				   ULINT_UNDEFINED, offset_heap);

	if (prebuilt->select_lock_type != LOCK_NONE) {
		/* Try to place a lock on the index record; we are searching
		the clust rec with a unique condition, hence
		we set a LOCK_REC_NOT_GAP type lock */

		err = lock_clust_rec_read_check_and_lock(
			0, btr_pcur_get_block(&prebuilt->clust_pcur),
			clust_rec, clust_index, *offsets,
			static_cast<enum lock_mode>(prebuilt->select_lock_type),
			LOCK_REC_NOT_GAP,
			thr);

		switch (err) {
		case DB_SUCCESS:
		case DB_SUCCESS_LOCKED_REC:
			break;
		default:
			goto err_exit;
		}
	} else {
		/* This is a non-locking consistent read: if necessary, fetch
		a previous version of the record */

		old_vers = NULL;

		/* If the isolation level allows reading of uncommitted data,
		then we never look for an earlier version */

		if (trx->isolation_level > TRX_ISO_READ_UNCOMMITTED
		    && !lock_clust_rec_cons_read_sees(
			    clust_rec, clust_index, *offsets,
			    trx->read_view)) {

			/* The following call returns 'offsets' associated with
			'old_vers' */
			err = row_sel_build_prev_vers_for_mysql(
				trx->read_view, clust_index, prebuilt,
				clust_rec, offsets, offset_heap, &old_vers,
				mtr);

			if (err != DB_SUCCESS || old_vers == NULL) {

				goto err_exit;
			}

			clust_rec = old_vers;
		}

		/* If we had to go to an earlier version of row or the
		secondary index record is delete marked, then it may be that
		the secondary index record corresponding to clust_rec
		(or old_vers) is not rec; in that case we must ignore
		such row because in our snapshot rec would not have existed.
		Remember that from rec we cannot see directly which transaction
		id corresponds to it: we have to go to the clustered index
		record. A query where we want to fetch all rows where
		the secondary index value is in some interval would return
		a wrong result if we would not drop rows which we come to
		visit through secondary index records that would not really
		exist in our snapshot. */

		if (clust_rec
		    && (old_vers
			|| trx->isolation_level <= TRX_ISO_READ_UNCOMMITTED
			|| rec_get_deleted_flag(rec, dict_table_is_comp(
							sec_index->table)))
		    && !row_sel_sec_rec_is_for_clust_rec(
			    rec, sec_index, clust_rec, clust_index)) {
			clust_rec = NULL;
#ifdef UNIV_SEARCH_DEBUG
		} else {
			ut_a(clust_rec == NULL
			     || row_sel_sec_rec_is_for_clust_rec(
				     rec, sec_index, clust_rec, clust_index));
#endif
		}

		err = DB_SUCCESS;
	}

func_exit:
	*out_rec = clust_rec;

	/* Store the current position if select_lock_type is not
	LOCK_NONE or if we are scanning using InnoDB APIs */
	if (prebuilt->select_lock_type != LOCK_NONE
	    || prebuilt->innodb_api) {
		/* We may use the cursor in update or in unlock_row():
		store its position */

		btr_pcur_store_position(&prebuilt->clust_pcur, mtr);
	}

err_exit:
	return(err);
}

/********************************************************************//**
Restores cursor position after it has been stored. We have to take into
account that the record cursor was positioned on may have been deleted.
Then we may have to move the cursor one step up or down.
@return TRUE if we may need to process the record the cursor is now
positioned on (i.e. we should not go to the next record yet) */
static
ibool
sel_restore_position_for_mysql(
/*===========================*/
	ibool*		same_user_rec,	/*!< out: TRUE if we were able to restore
					the cursor on a user record with the
					same ordering prefix in in the
					B-tree index */
	ulint		latch_mode,	/*!< in: latch mode wished in
					restoration */
	btr_pcur_t*	pcur,		/*!< in: cursor whose position
					has been stored */
	ibool		moves_up,	/*!< in: TRUE if the cursor moves up
					in the index */
	mtr_t*		mtr)		/*!< in: mtr; CAUTION: may commit
					mtr temporarily! */
{
	ibool		success;

	success = btr_pcur_restore_position(latch_mode, pcur, mtr);

	*same_user_rec = success;

	ut_ad(!success || pcur->rel_pos == BTR_PCUR_ON);
#ifdef UNIV_DEBUG
	if (pcur->pos_state == BTR_PCUR_IS_POSITIONED_OPTIMISTIC) {
		ut_ad(pcur->rel_pos == BTR_PCUR_BEFORE
		      || pcur->rel_pos == BTR_PCUR_AFTER);
	} else {
		ut_ad(pcur->pos_state == BTR_PCUR_IS_POSITIONED);
		ut_ad((pcur->rel_pos == BTR_PCUR_ON)
		      == btr_pcur_is_on_user_rec(pcur));
	}
#endif

	/* The position may need be adjusted for rel_pos and moves_up. */

	switch (pcur->rel_pos) {
	case BTR_PCUR_ON:
		if (!success && moves_up) {
next:
			btr_pcur_move_to_next(pcur, mtr);
			return(TRUE);
		}
		return(!success);
	case BTR_PCUR_AFTER_LAST_IN_TREE:
	case BTR_PCUR_BEFORE_FIRST_IN_TREE:
		return(TRUE);
	case BTR_PCUR_AFTER:
		/* positioned to record after pcur->old_rec. */
		pcur->pos_state = BTR_PCUR_IS_POSITIONED;
prev:
		if (btr_pcur_is_on_user_rec(pcur) && !moves_up) {
			btr_pcur_move_to_prev(pcur, mtr);
		}
		return(TRUE);
	case BTR_PCUR_BEFORE:
		/* For non optimistic restoration:
		The position is now set to the record before pcur->old_rec.

		For optimistic restoration:
		The position also needs to take the previous search_mode into
		consideration. */

		switch (pcur->pos_state) {
		case BTR_PCUR_IS_POSITIONED_OPTIMISTIC:
			pcur->pos_state = BTR_PCUR_IS_POSITIONED;
			if (pcur->search_mode == PAGE_CUR_GE) {
				/* Positioned during Greater or Equal search
				with BTR_PCUR_BEFORE. Optimistic restore to
				the same record. If scanning for lower then
				we must move to previous record.
				This can happen with:
				HANDLER READ idx a = (const);
				HANDLER READ idx PREV; */
				goto prev;
			}
			return(TRUE);
		case BTR_PCUR_IS_POSITIONED:
			if (moves_up && btr_pcur_is_on_user_rec(pcur)) {
				goto next;
			}
			return(TRUE);
		case BTR_PCUR_WAS_POSITIONED:
		case BTR_PCUR_NOT_POSITIONED:
			break;
		}
	}
	ut_ad(0);
	return(TRUE);
}

/********************************************************************//**
Copies a cached field for MySQL from the fetch cache. */
static
void
row_sel_copy_cached_field_for_mysql(
/*================================*/
	byte*			buf,	/*!< in/out: row buffer */
	const byte*		cache,	/*!< in: cached row */
	const mysql_row_templ_t*templ)	/*!< in: column template */
{
	ulint	len;

	buf += templ->mysql_col_offset;
	cache += templ->mysql_col_offset;

	UNIV_MEM_ASSERT_W(buf, templ->mysql_col_len);

	if (templ->mysql_type == DATA_MYSQL_TRUE_VARCHAR
	    && templ->type != DATA_INT) {
		/* Check for != DATA_INT to make sure we do
		not treat MySQL ENUM or SET as a true VARCHAR!
		Find the actual length of the true VARCHAR field. */
		row_mysql_read_true_varchar(
			&len, cache, templ->mysql_length_bytes);
		len += templ->mysql_length_bytes;
		UNIV_MEM_INVALID(buf, templ->mysql_col_len);
	} else {
		len = templ->mysql_col_len;
	}

	ut_memcpy(buf, cache, len);
}

/********************************************************************//**
Pops a cached row for MySQL from the fetch cache. */
UNIV_INLINE
void
row_sel_dequeue_cached_row_for_mysql(
/*=================================*/
	byte*		buf,		/*!< in/out: buffer where to copy the
					row */
	row_prebuilt_t*	prebuilt)	/*!< in: prebuilt struct */
{
	ulint			i;
	const mysql_row_templ_t*templ;
	const byte*		cached_rec;
	ut_ad(prebuilt->n_fetch_cached > 0);
	ut_ad(prebuilt->mysql_prefix_len <= prebuilt->mysql_row_len);

	UNIV_MEM_ASSERT_W(buf, prebuilt->mysql_row_len);

	cached_rec = prebuilt->fetch_cache[prebuilt->fetch_cache_first];

	if (UNIV_UNLIKELY(prebuilt->keep_other_fields_on_keyread)) {
		/* Copy cache record field by field, don't touch fields that
		are not covered by current key */

		for (i = 0; i < prebuilt->n_template; i++) {
			templ = prebuilt->mysql_template + i;
			row_sel_copy_cached_field_for_mysql(
				buf, cached_rec, templ);
			/* Copy NULL bit of the current field from cached_rec
			to buf */
			if (templ->mysql_null_bit_mask) {
				buf[templ->mysql_null_byte_offset]
					^= (buf[templ->mysql_null_byte_offset]
					    ^ cached_rec[templ->mysql_null_byte_offset])
					& (byte) templ->mysql_null_bit_mask;
			}
		}
	} else if (prebuilt->mysql_prefix_len > 63) {
		/* The record is long. Copy it field by field, in case
		there are some long VARCHAR column of which only a
		small length is being used. */
		UNIV_MEM_INVALID(buf, prebuilt->mysql_prefix_len);

		/* First copy the NULL bits. */
		ut_memcpy(buf, cached_rec, prebuilt->null_bitmap_len);
		/* Then copy the requested fields. */

		for (i = 0; i < prebuilt->n_template; i++) {
			row_sel_copy_cached_field_for_mysql(
				buf, cached_rec, prebuilt->mysql_template + i);
		}
	} else {
		ut_memcpy(buf, cached_rec, prebuilt->mysql_prefix_len);
	}

	prebuilt->n_fetch_cached--;
	prebuilt->fetch_cache_first++;

	if (prebuilt->n_fetch_cached == 0) {
		prebuilt->fetch_cache_first = 0;
	}
}

/********************************************************************//**
Initialise the prefetch cache. */
UNIV_INLINE
void
row_sel_prefetch_cache_init(
/*========================*/
	row_prebuilt_t*	prebuilt)	/*!< in/out: prebuilt struct */
{
	ulint	i;
	ulint	sz;
	byte*	ptr;

	/* Reserve space for the magic number. */
	sz = UT_ARR_SIZE(prebuilt->fetch_cache) * (prebuilt->mysql_row_len + 8);
	ptr = static_cast<byte*>(mem_alloc(sz));

	for (i = 0; i < UT_ARR_SIZE(prebuilt->fetch_cache); i++) {

		/* A user has reported memory corruption in these
		buffers in Linux. Put magic numbers there to help
		to track a possible bug. */

		mach_write_to_4(ptr, ROW_PREBUILT_FETCH_MAGIC_N);
		ptr += 4;

		prebuilt->fetch_cache[i] = ptr;
		ptr += prebuilt->mysql_row_len;

		mach_write_to_4(ptr, ROW_PREBUILT_FETCH_MAGIC_N);
		ptr += 4;
	}
}

/********************************************************************//**
Get the last fetch cache buffer from the queue.
@return pointer to buffer. */
UNIV_INLINE
byte*
row_sel_fetch_last_buf(
/*===================*/
	row_prebuilt_t*	prebuilt)	/*!< in/out: prebuilt struct */
{
	ut_ad(!prebuilt->templ_contains_blob);
	ut_ad(prebuilt->n_fetch_cached < MYSQL_FETCH_CACHE_SIZE);

	if (prebuilt->fetch_cache[0] == NULL) {
		/* Allocate memory for the fetch cache */
		ut_ad(prebuilt->n_fetch_cached == 0);

		row_sel_prefetch_cache_init(prebuilt);
	}

	ut_ad(prebuilt->fetch_cache_first == 0);
	UNIV_MEM_INVALID(prebuilt->fetch_cache[prebuilt->n_fetch_cached],
			 prebuilt->mysql_row_len);

	return(prebuilt->fetch_cache[prebuilt->n_fetch_cached]);
}

/********************************************************************//**
Pushes a row for MySQL to the fetch cache. */
UNIV_INLINE
void
row_sel_enqueue_cache_row_for_mysql(
/*================================*/
	byte*		mysql_rec,	/*!< in/out: MySQL record */
	row_prebuilt_t*	prebuilt)	/*!< in/out: prebuilt struct */
{
	/* For non ICP code path the row should already exist in the
	next fetch cache slot. */

	if (prebuilt->idx_cond != NULL) {
		byte*	dest = row_sel_fetch_last_buf(prebuilt);

		ut_memcpy(dest, mysql_rec, prebuilt->mysql_row_len);
	}

	++prebuilt->n_fetch_cached;
}

/*********************************************************************//**
Tries to do a shortcut to fetch a clustered index record with a unique key,
using the hash index if possible (not always). We assume that the search
mode is PAGE_CUR_GE, it is a consistent read, there is a read view in trx,
btr search latch has been locked in S-mode if AHI is enabled.
@return	SEL_FOUND, SEL_EXHAUSTED, SEL_RETRY */
static
ulint
row_sel_try_search_shortcut_for_mysql(
/*==================================*/
	const rec_t**	out_rec,/*!< out: record if found */
	row_prebuilt_t*	prebuilt,/*!< in: prebuilt struct */
	ulint**		offsets,/*!< in/out: for rec_get_offsets(*out_rec) */
	mem_heap_t**	heap,	/*!< in/out: heap for rec_get_offsets() */
	mtr_t*		mtr)	/*!< in: started mtr */
{
	dict_index_t*	index		= prebuilt->index;
	const dtuple_t*	search_tuple	= prebuilt->search_tuple;
	btr_pcur_t*	pcur		= &prebuilt->pcur;
	trx_t*		trx		= prebuilt->trx;
	const rec_t*	rec;

	ut_ad(dict_index_is_clust(index));
	ut_ad(!prebuilt->templ_contains_blob);

#ifndef UNIV_SEARCH_DEBUG
	ut_ad(trx->has_search_latch);

	btr_pcur_open_with_no_init(index, search_tuple, PAGE_CUR_GE,
				   BTR_SEARCH_LEAF, pcur,
				   RW_S_LATCH,
				   mtr);
#else /* UNIV_SEARCH_DEBUG */
	btr_pcur_open_with_no_init(index, search_tuple, PAGE_CUR_GE,
				   BTR_SEARCH_LEAF, pcur,
				   0,
				   mtr);
#endif /* UNIV_SEARCH_DEBUG */
	rec = btr_pcur_get_rec(pcur);

	if (!page_rec_is_user_rec(rec)) {

		return(SEL_RETRY);
	}

	/* As the cursor is now placed on a user record after a search with
	the mode PAGE_CUR_GE, the up_match field in the cursor tells how many
	fields in the user record matched to the search tuple */

	if (btr_pcur_get_up_match(pcur) < dtuple_get_n_fields(search_tuple)) {

		return(SEL_EXHAUSTED);
	}

	/* This is a non-locking consistent read: if necessary, fetch
	a previous version of the record */

	*offsets = rec_get_offsets(rec, index, *offsets,
				   ULINT_UNDEFINED, heap);

	if (!lock_clust_rec_cons_read_sees(rec, index,
					   *offsets, trx->read_view)) {

		return(SEL_RETRY);
	}

	if (rec_get_deleted_flag(rec, dict_table_is_comp(index->table))) {

		return(SEL_EXHAUSTED);
	}

	*out_rec = rec;

	return(SEL_FOUND);
}

/*********************************************************************//**
Check a pushed-down index condition.
@return ICP_NO_MATCH, ICP_MATCH, or ICP_OUT_OF_RANGE */
static
enum icp_result
row_search_idx_cond_check(
/*======================*/
	byte*			mysql_rec,	/*!< out: record
						in MySQL format (invalid unless
						prebuilt->idx_cond!=NULL and
						we return ICP_MATCH) */
	row_prebuilt_t*		prebuilt,	/*!< in/out: prebuilt struct
						for the table handle */
	const rec_t*		rec,		/*!< in: InnoDB record */
	const ulint*		offsets)	/*!< in: rec_get_offsets() */
{
	enum icp_result result;
	ulint		i;

	ut_ad(rec_offs_validate(rec, prebuilt->index, offsets));

	if (!prebuilt->idx_cond) {
		return(ICP_MATCH);
	}

	MONITOR_INC(MONITOR_ICP_ATTEMPTS);

	/* Convert to MySQL format those fields that are needed for
	evaluating the index condition. */

	if (UNIV_LIKELY_NULL(prebuilt->blob_heap)) {
		mem_heap_empty(prebuilt->blob_heap);
	}

	for (i = 0; i < prebuilt->idx_cond_n_cols; i++) {
		const mysql_row_templ_t*templ = &prebuilt->mysql_template[i];

		if (!row_sel_store_mysql_field(mysql_rec, prebuilt,
					       rec, prebuilt->index, offsets,
					       templ->icp_rec_field_no,
					       templ)) {
			return(ICP_NO_MATCH);
		}
	}

	/* We assume that the index conditions on
	case-insensitive columns are case-insensitive. The
	case of such columns may be wrong in a secondary
	index, if the case of the column has been updated in
	the past, or a record has been deleted and a record
	inserted in a different case. */
	result = innobase_index_cond(prebuilt->idx_cond);
	switch (result) {
	case ICP_MATCH:
		/* Convert the remaining fields to MySQL format.
		If this is a secondary index record, we must defer
		this until we have fetched the clustered index record. */
		if (!prebuilt->need_to_access_clustered
		    || dict_index_is_clust(prebuilt->index)) {
			if (!row_sel_store_mysql_rec(
				    mysql_rec, prebuilt, rec, FALSE,
				    prebuilt->index, offsets)) {
				ut_ad(dict_index_is_clust(prebuilt->index));
				return(ICP_NO_MATCH);
			}
		}
		MONITOR_INC(MONITOR_ICP_MATCH);
		return(result);
	case ICP_NO_MATCH:
		MONITOR_INC(MONITOR_ICP_NO_MATCH);
		return(result);
	case ICP_OUT_OF_RANGE:
		MONITOR_INC(MONITOR_ICP_OUT_OF_RANGE);
		return(result);
        case ICP_ERROR:
        case ICP_ABORTED_BY_USER:
		return(result);
	}

	ut_error;
	return(result);
}

/********************************************************************//**
Searches for rows in the database. This is used in the interface to
MySQL. This function opens a cursor, and also implements fetch next
and fetch prev. NOTE that if we do a search with a full key value
from a unique index (ROW_SEL_EXACT), then we will not store the cursor
position and fetch next or fetch prev must not be tried to the cursor!
@return DB_SUCCESS, DB_RECORD_NOT_FOUND, DB_END_OF_INDEX, DB_DEADLOCK,
DB_LOCK_TABLE_FULL, DB_CORRUPTION, DB_SEARCH_ABORTED_BY_USER or
DB_TOO_BIG_RECORD */
UNIV_INTERN
dberr_t
row_search_for_mysql(
/*=================*/
	byte*		buf,		/*!< in/out: buffer for the fetched
					row in the MySQL format */
	ulint		mode,		/*!< in: search mode PAGE_CUR_L, ... */
	row_prebuilt_t*	prebuilt,	/*!< in: prebuilt struct for the
					table handle; this contains the info
					of search_tuple, index; if search
					tuple contains 0 fields then we
					position the cursor at the start or
					the end of the index, depending on
					'mode' */
	ulint		match_mode,	/*!< in: 0 or ROW_SEL_EXACT or
					ROW_SEL_EXACT_PREFIX */
	ulint		direction)	/*!< in: 0 or ROW_SEL_NEXT or
					ROW_SEL_PREV; NOTE: if this is != 0,
					then prebuilt must have a pcur
					with stored position! In opening of a
					cursor 'direction' should be 0. */
{
	dict_index_t*	index		= prebuilt->index;
	ibool		comp		= dict_table_is_comp(index->table);
	const dtuple_t*	search_tuple	= prebuilt->search_tuple;
	btr_pcur_t*	pcur		= &prebuilt->pcur;
	trx_t*		trx		= prebuilt->trx;
	dict_index_t*	clust_index;
	que_thr_t*	thr;
	const rec_t*	rec;
	const rec_t*	result_rec = NULL;
	const rec_t*	clust_rec;
	dberr_t		err				= DB_SUCCESS;
	ibool		unique_search			= FALSE;
	ibool		mtr_has_extra_clust_latch	= FALSE;
	ibool		moves_up			= FALSE;
	ibool		set_also_gap_locks		= TRUE;
	/* if the query is a plain locking SELECT, and the isolation level
	is <= TRX_ISO_READ_COMMITTED, then this is set to FALSE */
	ibool		did_semi_consistent_read	= FALSE;
	/* if the returned record was locked and we did a semi-consistent
	read (fetch the newest committed version), then this is set to
	TRUE */
#ifdef UNIV_SEARCH_DEBUG
	ulint		cnt				= 0;
#endif /* UNIV_SEARCH_DEBUG */
	ulint		next_offs;
	ibool		same_user_rec;
	mtr_t		mtr;
	mem_heap_t*	heap				= NULL;
	ulint		offsets_[REC_OFFS_NORMAL_SIZE];
	ulint*		offsets				= offsets_;
	ibool		table_lock_waited		= FALSE;
	byte*		next_buf			= 0;
	ibool		use_clustered_index		= FALSE;

	rec_offs_init(offsets_);

	ut_ad(index && pcur && search_tuple);

	/* We don't support FTS queries from the HANDLER interfaces, because
	we implemented FTS as reversed inverted index with auxiliary tables.
	So anything related to traditional index query would not apply to
	it. */
	if (index->type & DICT_FTS) {
		return(DB_END_OF_INDEX);
	}

	ut_ad(!trx->has_search_latch);
#ifdef UNIV_SYNC_DEBUG
	ut_ad(!btr_search_own_any());
	ut_ad(!sync_thread_levels_nonempty_trx(trx->has_search_latch));
#endif /* UNIV_SYNC_DEBUG */

	if (dict_table_is_discarded(prebuilt->table)) {

		return(DB_TABLESPACE_DELETED);

	} else if (prebuilt->table->ibd_file_missing) {

		return(DB_TABLESPACE_NOT_FOUND);

	} else if (!prebuilt->index_usable) {

		return(DB_MISSING_HISTORY);

	} else if (dict_index_is_corrupted(index)) {

		return(DB_CORRUPTION);

	} else if (prebuilt->magic_n != ROW_PREBUILT_ALLOCATED) {
		fprintf(stderr,
			"InnoDB: Error: trying to free a corrupt\n"
			"InnoDB: table handle. Magic n %lu, table name ",
			(ulong) prebuilt->magic_n);
		ut_print_name(stderr, trx, TRUE, prebuilt->table->name);
		putc('\n', stderr);

		mem_analyze_corruption(prebuilt);

		ut_error;
	}

	/* init null bytes with default values as they might be
	left uninitialized in some cases and these uninited bytes
	might be copied into mysql record buffer that leads to
	valgrind warnings */
	memcpy(buf, prebuilt->default_rec, prebuilt->null_bitmap_len);

#if 0
	/* August 19, 2005 by Heikki: temporarily disable this error
	print until the cursor lock count is done correctly.
	See bugs #12263 and #12456!*/

	if (trx->n_mysql_tables_in_use == 0
	    && UNIV_UNLIKELY(prebuilt->select_lock_type == LOCK_NONE)) {
		/* Note that if MySQL uses an InnoDB temp table that it
		created inside LOCK TABLES, then n_mysql_tables_in_use can
		be zero; in that case select_lock_type is set to LOCK_X in
		::start_stmt. */

		fputs("InnoDB: Error: MySQL is trying to perform a SELECT\n"
		      "InnoDB: but it has not locked"
		      " any tables in ::external_lock()!\n",
		      stderr);
		trx_print(stderr, trx, 600);
		fputc('\n', stderr);
	}
#endif

#if 0
	fprintf(stderr, "Match mode %lu\n search tuple ",
		(ulong) match_mode);
	dtuple_print(search_tuple);
	fprintf(stderr, "N tables locked %lu\n",
		(ulong) trx->mysql_n_tables_locked);
#endif
	/* Reset the new record lock info if srv_locks_unsafe_for_binlog
	is set or session is using a READ COMMITED isolation level. Then
	we are able to remove the record locks set here on an individual
	row. */
	prebuilt->new_rec_locks = 0;

	/*-------------------------------------------------------------*/
	/* PHASE 1: Try to pop the row from the prefetch cache */

	if (UNIV_UNLIKELY(direction == 0)) {
		trx->op_info = "starting index read";

		prebuilt->n_rows_fetched = 0;
		prebuilt->n_fetch_cached = 0;
		prebuilt->fetch_cache_first = 0;

		if (prebuilt->sel_graph == NULL) {
			/* Build a dummy select query graph */
			row_prebuild_sel_graph(prebuilt);
		}
	} else {
		trx->op_info = "fetching rows";

		if (prebuilt->n_rows_fetched == 0) {
			prebuilt->fetch_direction = direction;
		}

		if (UNIV_UNLIKELY(direction != prebuilt->fetch_direction)) {
			if (UNIV_UNLIKELY(prebuilt->n_fetch_cached > 0)) {
				ut_error;
				/* TODO: scrollable cursor: restore cursor to
				the place of the latest returned row,
				or better: prevent caching for a scroll
				cursor! */
			}

			prebuilt->n_rows_fetched = 0;
			prebuilt->n_fetch_cached = 0;
			prebuilt->fetch_cache_first = 0;

		} else if (UNIV_LIKELY(prebuilt->n_fetch_cached > 0)) {
			row_sel_dequeue_cached_row_for_mysql(buf, prebuilt);

			prebuilt->n_rows_fetched++;

			err = DB_SUCCESS;
			goto func_exit;
		}

		if (prebuilt->fetch_cache_first > 0
		    && prebuilt->fetch_cache_first < MYSQL_FETCH_CACHE_SIZE) {

			/* The previous returned row was popped from the fetch
			cache, but the cache was not full at the time of the
			popping: no more rows can exist in the result set */

			err = DB_RECORD_NOT_FOUND;
			goto func_exit;
		}

		prebuilt->n_rows_fetched++;

		if (prebuilt->n_rows_fetched > 1000000000) {
			/* Prevent wrap-over */
			prebuilt->n_rows_fetched = 500000000;
		}

		mode = pcur->search_mode;
	}

	/* In a search where at most one record in the index may match, we
	can use a LOCK_REC_NOT_GAP type record lock when locking a
	non-delete-marked matching record.

	Note that in a unique secondary index there may be different
	delete-marked versions of a record where only the primary key
	values differ: thus in a secondary index we must use next-key
	locks when locking delete-marked records. */

	if (match_mode == ROW_SEL_EXACT
	    && dict_index_is_unique(index)
	    && dtuple_get_n_fields(search_tuple)
	    == dict_index_get_n_unique(index)
	    && (dict_index_is_clust(index)
		|| !dtuple_contains_null(search_tuple))) {

		/* Note above that a UNIQUE secondary index can contain many
		rows with the same key value if one of the columns is the SQL
		null. A clustered index under MySQL can never contain null
		columns because we demand that all the columns in primary key
		are non-null. */

		unique_search = TRUE;

		/* Even if the condition is unique, MySQL seems to try to
		retrieve also a second row if a primary key contains more than
		1 column. Return immediately if this is not a HANDLER
		command. */

		if (UNIV_UNLIKELY(direction != 0
				  && !prebuilt->used_in_HANDLER)) {

			err = DB_RECORD_NOT_FOUND;
			goto func_exit;
		}
	}

	mtr_start(&mtr);

	/*-------------------------------------------------------------*/
	/* PHASE 2: Try fast adaptive hash index search if possible */

	/* Next test if this is the special case where we can use the fast
	adaptive hash index to try the search. Since we must release the
	search system latch when we retrieve an externally stored field, we
	cannot use the adaptive hash index in a search in the case the row
	may be long and there may be externally stored fields */

	if (UNIV_UNLIKELY(direction == 0)
	    && unique_search
	    && dict_index_is_clust(index)
	    && !prebuilt->templ_contains_blob
	    && !prebuilt->used_in_HANDLER
	    && (prebuilt->mysql_row_len < UNIV_PAGE_SIZE / 8)
	    && !prebuilt->innodb_api) {

		mode = PAGE_CUR_GE;

		if (trx->mysql_n_tables_locked == 0
		    && prebuilt->select_lock_type == LOCK_NONE
		    && trx->isolation_level > TRX_ISO_READ_UNCOMMITTED
		    && trx->read_view) {

			/* This is a SELECT query done as a consistent read,
			and the read view has already been allocated:
			let us try a search shortcut through the hash
			index.
			NOTE that we must also test that
			mysql_n_tables_locked == 0, because this might
			also be INSERT INTO ... SELECT ... or
			CREATE TABLE ... SELECT ... . Our algorithm is
			NOT prepared to inserts interleaved with the SELECT,
			and if we try that, we can deadlock on the adaptive
			hash index semaphore! */

#ifndef UNIV_SEARCH_DEBUG
			ut_ad(!trx->has_search_latch);
			rw_lock_s_lock(btr_search_get_latch(index));
			trx->has_search_latch = TRUE;
#endif
			switch (row_sel_try_search_shortcut_for_mysql(
					&rec, prebuilt, &offsets, &heap,
					&mtr)) {
			case SEL_FOUND:
#ifdef UNIV_SEARCH_DEBUG
				ut_a(0 == cmp_dtuple_rec(search_tuple,
							 rec, offsets));
#endif
				/* At this point, rec is protected by
				a page latch that was acquired by
				row_sel_try_search_shortcut_for_mysql().
				The latch will not be released until
				mtr_commit(&mtr). */
				ut_ad(!rec_get_deleted_flag(rec, comp));

				if (prebuilt->idx_cond) {
					switch (row_search_idx_cond_check(
							buf, prebuilt,
							rec, offsets)) {
					case ICP_NO_MATCH:
					case ICP_OUT_OF_RANGE:
                                        case ICP_ABORTED_BY_USER:
                                        case ICP_ERROR:
						goto shortcut_mismatch;
					case ICP_MATCH:
						goto shortcut_match;
					}
				}

				if (!row_sel_store_mysql_rec(
					    buf, prebuilt,
					    rec, FALSE, index, offsets)) {
					/* Only fresh inserts may contain
					incomplete externally stored
					columns. Pretend that such
					records do not exist. Such
					records may only be accessed
					at the READ UNCOMMITTED
					isolation level or when
					rolling back a recovered
					transaction. Rollback happens
					at a lower level, not here. */

					/* Proceed as in case SEL_RETRY. */
					break;
				}

			shortcut_match:
				mtr_commit(&mtr);

				/* ut_print_name(stderr, index->name);
				fputs(" shortcut\n", stderr); */

				err = DB_SUCCESS;
				goto release_search_latch;

			case SEL_EXHAUSTED:
			shortcut_mismatch:
				mtr_commit(&mtr);

				/* ut_print_name(stderr, index->name);
				fputs(" record not found 2\n", stderr); */

				err = DB_RECORD_NOT_FOUND;
release_search_latch:
				rw_lock_s_unlock(
					btr_search_get_latch(index));
				trx->has_search_latch = FALSE;

				/* NOTE that we do NOT store the cursor
				position */
				goto func_exit;

			case SEL_RETRY:
				break;

			default:
				ut_ad(0);
			}

			mtr_commit(&mtr);
			mtr_start(&mtr);

			rw_lock_s_unlock(btr_search_get_latch(index));
			trx->has_search_latch = FALSE;
		}
	}

	/*-------------------------------------------------------------*/
	/* PHASE 3: Open or restore index cursor position */

	ut_ad(!trx->has_search_latch);
#ifdef UNIV_SYNC_DEBUG
	ut_ad(!btr_search_own_any());
#endif

	/* The state of a running trx can only be changed by the
	thread that is currently serving the transaction. Because we
	are that thread, we can read trx->state without holding any
	mutex. */
	ut_ad(prebuilt->sql_stat_start || trx->state == TRX_STATE_ACTIVE);

	ut_ad(trx->state == TRX_STATE_NOT_STARTED
	      || trx->state == TRX_STATE_ACTIVE);

	ut_ad(prebuilt->sql_stat_start
	      || prebuilt->select_lock_type != LOCK_NONE
	      || trx->read_view);

	trx_start_if_not_started(trx);

	if (trx->isolation_level <= TRX_ISO_READ_COMMITTED
	    && prebuilt->select_lock_type != LOCK_NONE
	    && trx->mysql_thd != NULL
	    && thd_is_select(trx->mysql_thd)) {
		/* It is a plain locking SELECT and the isolation
		level is low: do not lock gaps */

		set_also_gap_locks = FALSE;
	}

	/* Note that if the search mode was GE or G, then the cursor
	naturally moves upward (in fetch next) in alphabetical order,
	otherwise downward */

	if (UNIV_UNLIKELY(direction == 0)) {
		if (mode == PAGE_CUR_GE || mode == PAGE_CUR_G) {
			moves_up = TRUE;
		}
	} else if (direction == ROW_SEL_NEXT) {
		moves_up = TRUE;
	}

	thr = que_fork_get_first_thr(prebuilt->sel_graph);

	que_thr_move_to_run_state_for_mysql(thr, trx);

	clust_index = dict_table_get_first_index(index->table);

	/* Do some start-of-statement preparations */

	if (!prebuilt->sql_stat_start) {
		/* No need to set an intention lock or assign a read view */

		if (UNIV_UNLIKELY
		    (trx->read_view == NULL
		     && prebuilt->select_lock_type == LOCK_NONE)) {

			fputs("InnoDB: Error: MySQL is trying to"
			      " perform a consistent read\n"
			      "InnoDB: but the read view is not assigned!\n",
			      stderr);
			trx_print(stderr, trx, 600);
			fputc('\n', stderr);
			ut_error;
		}
	} else if (prebuilt->select_lock_type == LOCK_NONE) {
		/* This is a consistent read */
		/* Assign a read view for the query */

		trx_assign_read_view(trx);
		prebuilt->sql_stat_start = FALSE;
	} else {
wait_table_again:
		err = lock_table(0, index->table,
				 prebuilt->select_lock_type == LOCK_S
				 ? LOCK_IS : LOCK_IX, thr);

		if (err != DB_SUCCESS) {

			table_lock_waited = TRUE;
			goto lock_table_wait;
		}
		prebuilt->sql_stat_start = FALSE;
	}

	/* Open or restore index cursor position */

	if (UNIV_LIKELY(direction != 0)) {
		ibool	need_to_process = sel_restore_position_for_mysql(
			&same_user_rec, BTR_SEARCH_LEAF,
			pcur, moves_up, &mtr);

		if (UNIV_UNLIKELY(need_to_process)) {
			if (UNIV_UNLIKELY(prebuilt->row_read_type
					  == ROW_READ_DID_SEMI_CONSISTENT)) {
				/* We did a semi-consistent read,
				but the record was removed in
				the meantime. */
				prebuilt->row_read_type
					= ROW_READ_TRY_SEMI_CONSISTENT;
			}
		} else if (UNIV_LIKELY(prebuilt->row_read_type
				       != ROW_READ_DID_SEMI_CONSISTENT)) {

			/* The cursor was positioned on the record
			that we returned previously.  If we need
			to repeat a semi-consistent read as a
			pessimistic locking read, the record
			cannot be skipped. */

			goto next_rec;
		}

	} else if (dtuple_get_n_fields(search_tuple) > 0) {

		btr_pcur_open_with_no_init(index, search_tuple, mode,
					   BTR_SEARCH_LEAF,
					   pcur, 0, &mtr);

		pcur->trx_if_known = trx;

		rec = btr_pcur_get_rec(pcur);

		if (!moves_up
		    && !page_rec_is_supremum(rec)
		    && set_also_gap_locks
		    && !(srv_locks_unsafe_for_binlog
			 || trx->isolation_level <= TRX_ISO_READ_COMMITTED)
		    && prebuilt->select_lock_type != LOCK_NONE) {

			/* Try to place a gap lock on the next index record
			to prevent phantoms in ORDER BY ... DESC queries */
			const rec_t*	next_rec = page_rec_get_next_const(rec);

			offsets = rec_get_offsets(next_rec, index, offsets,
						  ULINT_UNDEFINED, &heap);
			err = sel_set_rec_lock(btr_pcur_get_block(pcur),
					       next_rec, index, offsets,
					       prebuilt->select_lock_type,
					       LOCK_GAP, thr);

			switch (err) {
			case DB_SUCCESS_LOCKED_REC:
				err = DB_SUCCESS;
			case DB_SUCCESS:
				break;
			default:
				goto lock_wait_or_error;
			}
		}
	} else if (mode == PAGE_CUR_G || mode == PAGE_CUR_L) {
		btr_pcur_open_at_index_side(
			mode == PAGE_CUR_G, index, BTR_SEARCH_LEAF,
			pcur, false, 0, &mtr);
	}

rec_loop:
	DEBUG_SYNC_C("row_search_rec_loop");
	if (trx_is_interrupted(trx)) {
		btr_pcur_store_position(pcur, &mtr);
		err = DB_INTERRUPTED;
		goto normal_return;
	}

	/*-------------------------------------------------------------*/
	/* PHASE 4: Look for matching records in a loop */

	rec = btr_pcur_get_rec(pcur);

	SRV_CORRUPT_TABLE_CHECK(rec,
	{
		err = DB_CORRUPTION;
		goto lock_wait_or_error;
	});

	ut_ad(!!page_rec_is_comp(rec) == comp);
#ifdef UNIV_SEARCH_DEBUG
	/*
	fputs("Using ", stderr);
	dict_index_name_print(stderr, trx, index);
	fprintf(stderr, " cnt %lu ; Page no %lu\n", cnt,
	page_get_page_no(page_align(rec)));
	rec_print(stderr, rec, index);
	printf("delete-mark: %lu\n",
	       rec_get_deleted_flag(rec, page_rec_is_comp(rec)));
	*/
#endif /* UNIV_SEARCH_DEBUG */

	if (page_rec_is_infimum(rec)) {

		/* The infimum record on a page cannot be in the result set,
		and neither can a record lock be placed on it: we skip such
		a record. */

		goto next_rec;
	}

	if (page_rec_is_supremum(rec)) {

		if (set_also_gap_locks
		    && !(srv_locks_unsafe_for_binlog
			 || trx->isolation_level <= TRX_ISO_READ_COMMITTED)
		    && prebuilt->select_lock_type != LOCK_NONE) {

			/* Try to place a lock on the index record */

			/* If innodb_locks_unsafe_for_binlog option is used
			or this session is using a READ COMMITTED or lower isolation
			level we do not lock gaps. Supremum record is really
			a gap and therefore we do not set locks there. */

			offsets = rec_get_offsets(rec, index, offsets,
						  ULINT_UNDEFINED, &heap);
			err = sel_set_rec_lock(btr_pcur_get_block(pcur),
					       rec, index, offsets,
					       prebuilt->select_lock_type,
					       LOCK_ORDINARY, thr);

			switch (err) {
			case DB_SUCCESS_LOCKED_REC:
				err = DB_SUCCESS;
			case DB_SUCCESS:
				break;
			default:
				goto lock_wait_or_error;
			}
		}
		/* A page supremum record cannot be in the result set: skip
		it now that we have placed a possible lock on it */

		goto next_rec;
	}

	/*-------------------------------------------------------------*/
	/* Do sanity checks in case our cursor has bumped into page
	corruption */

	if (comp) {
		next_offs = rec_get_next_offs(rec, TRUE);
		if (UNIV_UNLIKELY(next_offs < PAGE_NEW_SUPREMUM)) {

			goto wrong_offs;
		}
	} else {
		next_offs = rec_get_next_offs(rec, FALSE);
		if (UNIV_UNLIKELY(next_offs < PAGE_OLD_SUPREMUM)) {

			goto wrong_offs;
		}
	}

	if (UNIV_UNLIKELY(next_offs >= UNIV_PAGE_SIZE - PAGE_DIR)) {

wrong_offs:
		if (srv_pass_corrupt_table && index->table->space != 0 &&
		    index->table->space < SRV_LOG_SPACE_FIRST_ID) {
			index->table->is_corrupt = TRUE;
			fil_space_set_corrupt(index->table->space);
		}

		if ((srv_force_recovery == 0 || moves_up == FALSE)
		    && srv_pass_corrupt_table <= 1) {
			ut_print_timestamp(stderr);
			buf_page_print(page_align(rec), 0,
				       BUF_PAGE_PRINT_NO_CRASH);
			fprintf(stderr,
				"\nInnoDB: rec address %p,"
				" buf block fix count %lu\n",
				(void*) rec, (ulong)
				btr_cur_get_block(btr_pcur_get_btr_cur(pcur))
				->page.buf_fix_count);
			fprintf(stderr,
				"InnoDB: Index corruption: rec offs %lu"
				" next offs %lu, page no %lu,\n"
				"InnoDB: ",
				(ulong) page_offset(rec),
				(ulong) next_offs,
				(ulong) page_get_page_no(page_align(rec)));
			dict_index_name_print(stderr, trx, index);
			fputs(". Run CHECK TABLE. You may need to\n"
			      "InnoDB: restore from a backup, or"
			      " dump + drop + reimport the table.\n",
			      stderr);
			ut_ad(0);
			err = DB_CORRUPTION;

			goto lock_wait_or_error;
		} else {
			/* The user may be dumping a corrupt table. Jump
			over the corruption to recover as much as possible. */

			fprintf(stderr,
				"InnoDB: Index corruption: rec offs %lu"
				" next offs %lu, page no %lu,\n"
				"InnoDB: ",
				(ulong) page_offset(rec),
				(ulong) next_offs,
				(ulong) page_get_page_no(page_align(rec)));
			dict_index_name_print(stderr, trx, index);
			fputs(". We try to skip the rest of the page.\n",
			      stderr);

			btr_pcur_move_to_last_on_page(pcur, &mtr);

			goto next_rec;
		}
	}
	/*-------------------------------------------------------------*/

	/* Calculate the 'offsets' associated with 'rec' */

	ut_ad(fil_page_get_type(btr_pcur_get_page(pcur)) == FIL_PAGE_INDEX);
	ut_ad(btr_page_get_index_id(btr_pcur_get_page(pcur)) == index->id);

	offsets = rec_get_offsets(rec, index, offsets, ULINT_UNDEFINED, &heap);

	if (UNIV_UNLIKELY(srv_force_recovery > 0
			  || (index->table->is_corrupt &&
			      srv_pass_corrupt_table == 2))) {
		if (!rec_validate(rec, offsets)
		    || !btr_index_rec_validate(rec, index, FALSE)) {
			fprintf(stderr,
				"InnoDB: Index corruption: rec offs %lu"
				" next offs %lu, page no %lu,\n"
				"InnoDB: ",
				(ulong) page_offset(rec),
				(ulong) next_offs,
				(ulong) page_get_page_no(page_align(rec)));
			dict_index_name_print(stderr, trx, index);
			fputs(". We try to skip the record.\n",
			      stderr);

			goto next_rec;
		}
	}

	/* Note that we cannot trust the up_match value in the cursor at this
	place because we can arrive here after moving the cursor! Thus
	we have to recompare rec and search_tuple to determine if they
	match enough. */

	if (match_mode == ROW_SEL_EXACT) {
		/* Test if the index record matches completely to search_tuple
		in prebuilt: if not, then we return with DB_RECORD_NOT_FOUND */

		/* fputs("Comparing rec and search tuple\n", stderr); */

		if (0 != cmp_dtuple_rec(search_tuple, rec, offsets)) {

			if (set_also_gap_locks
			    && !(srv_locks_unsafe_for_binlog
				 || trx->isolation_level
				 <= TRX_ISO_READ_COMMITTED)
			    && prebuilt->select_lock_type != LOCK_NONE) {

				/* Try to place a gap lock on the index
				record only if innodb_locks_unsafe_for_binlog
				option is not set or this session is not
				using a READ COMMITTED or lower isolation level. */

				err = sel_set_rec_lock(
					btr_pcur_get_block(pcur),
					rec, index, offsets,
					prebuilt->select_lock_type, LOCK_GAP,
					thr);

				switch (err) {
				case DB_SUCCESS_LOCKED_REC:
				case DB_SUCCESS:
					break;
				default:
					goto lock_wait_or_error;
				}
			}

			btr_pcur_store_position(pcur, &mtr);

			/* The found record was not a match, but may be used
			as NEXT record (index_next). Set the relative position
			to BTR_PCUR_BEFORE, to reflect that the position of
			the persistent cursor is before the found/stored row
			(pcur->old_rec). */
			ut_ad(pcur->rel_pos == BTR_PCUR_ON);
			pcur->rel_pos = BTR_PCUR_BEFORE;

			err = DB_RECORD_NOT_FOUND;
#if 0
			ut_print_name(stderr, trx, FALSE, index->name);
			fputs(" record not found 3\n", stderr);
#endif

			goto normal_return;
		}

	} else if (match_mode == ROW_SEL_EXACT_PREFIX) {

		if (!cmp_dtuple_is_prefix_of_rec(search_tuple, rec, offsets)) {

			if (set_also_gap_locks
			    && !(srv_locks_unsafe_for_binlog
				 || trx->isolation_level
				 <= TRX_ISO_READ_COMMITTED)
			    && prebuilt->select_lock_type != LOCK_NONE) {

				/* Try to place a gap lock on the index
				record only if innodb_locks_unsafe_for_binlog
				option is not set or this session is not
				using a READ COMMITTED or lower isolation level. */

				err = sel_set_rec_lock(
					btr_pcur_get_block(pcur),
					rec, index, offsets,
					prebuilt->select_lock_type, LOCK_GAP,
					thr);

				switch (err) {
				case DB_SUCCESS_LOCKED_REC:
				case DB_SUCCESS:
					break;
				default:
					goto lock_wait_or_error;
				}
			}

			btr_pcur_store_position(pcur, &mtr);

			/* The found record was not a match, but may be used
			as NEXT record (index_next). Set the relative position
			to BTR_PCUR_BEFORE, to reflect that the position of
			the persistent cursor is before the found/stored row
			(pcur->old_rec). */
			ut_ad(pcur->rel_pos == BTR_PCUR_ON);
			pcur->rel_pos = BTR_PCUR_BEFORE;

			err = DB_RECORD_NOT_FOUND;
#if 0
			ut_print_name(stderr, trx, FALSE, index->name);
			fputs(" record not found 4\n", stderr);
#endif

			goto normal_return;
		}
	}

	/* We are ready to look at a possible new index entry in the result
	set: the cursor is now placed on a user record */

	if (prebuilt->select_lock_type != LOCK_NONE) {
		/* Try to place a lock on the index record; note that delete
		marked records are a special case in a unique search. If there
		is a non-delete marked record, then it is enough to lock its
		existence with LOCK_REC_NOT_GAP. */

		/* If innodb_locks_unsafe_for_binlog option is used
		or this session is using a READ COMMITED isolation
		level we lock only the record, i.e., next-key locking is
		not used. */

		ulint	lock_type;

		if (!set_also_gap_locks
		    || srv_locks_unsafe_for_binlog
		    || trx->isolation_level <= TRX_ISO_READ_COMMITTED
		    || (unique_search && !rec_get_deleted_flag(rec, comp))) {

			goto no_gap_lock;
		} else {
			lock_type = LOCK_ORDINARY;
		}

		/* If we are doing a 'greater or equal than a primary key
		value' search from a clustered index, and we find a record
		that has that exact primary key value, then there is no need
		to lock the gap before the record, because no insert in the
		gap can be in our search range. That is, no phantom row can
		appear that way.

		An example: if col1 is the primary key, the search is WHERE
		col1 >= 100, and we find a record where col1 = 100, then no
		need to lock the gap before that record. */

		if (index == clust_index
		    && mode == PAGE_CUR_GE
		    && direction == 0
		    && dtuple_get_n_fields_cmp(search_tuple)
		    == dict_index_get_n_unique(index)
		    && 0 == cmp_dtuple_rec(search_tuple, rec, offsets)) {
no_gap_lock:
			lock_type = LOCK_REC_NOT_GAP;
		}

		err = sel_set_rec_lock(btr_pcur_get_block(pcur),
				       rec, index, offsets,
				       prebuilt->select_lock_type,
				       lock_type, thr);

		switch (err) {
			const rec_t*	old_vers;
		case DB_SUCCESS_LOCKED_REC:
			if (srv_locks_unsafe_for_binlog
			    || trx->isolation_level
			    <= TRX_ISO_READ_COMMITTED) {
				/* Note that a record of
				prebuilt->index was locked. */
				prebuilt->new_rec_locks = 1;
			}
			err = DB_SUCCESS;
		case DB_SUCCESS:
			break;
		case DB_LOCK_WAIT:
			/* Never unlock rows that were part of a conflict. */
			prebuilt->new_rec_locks = 0;

			if (UNIV_LIKELY(prebuilt->row_read_type
					!= ROW_READ_TRY_SEMI_CONSISTENT)
			    || unique_search
			    || index != clust_index) {

				goto lock_wait_or_error;
			}

			/* The following call returns 'offsets'
			associated with 'old_vers' */
			row_sel_build_committed_vers_for_mysql(
				clust_index, prebuilt, rec,
				&offsets, &heap, &old_vers, &mtr);

			/* Check whether it was a deadlock or not, if not
			a deadlock and the transaction had to wait then
			release the lock it is waiting on. */

			err = lock_trx_handle_wait(trx);

			switch (err) {
			case DB_SUCCESS:
				/* The lock was granted while we were
				searching for the last committed version.
				Do a normal locking read. */

				offsets = rec_get_offsets(
					rec, index, offsets, ULINT_UNDEFINED,
					&heap);
				goto locks_ok;
			case DB_DEADLOCK:
				goto lock_wait_or_error;
			case DB_LOCK_WAIT:
				err = DB_SUCCESS;
				break;
			default:
				ut_error;
			}

			if (old_vers == NULL) {
				/* The row was not yet committed */

				goto next_rec;
			}

			did_semi_consistent_read = TRUE;
			rec = old_vers;
			break;
		default:

			goto lock_wait_or_error;
		}
	} else {
		/* This is a non-locking consistent read: if necessary, fetch
		a previous version of the record */

		if (trx->isolation_level == TRX_ISO_READ_UNCOMMITTED) {

			/* Do nothing: we let a non-locking SELECT read the
			latest version of the record */

		} else if (index == clust_index) {

			/* Fetch a previous version of the row if the current
			one is not visible in the snapshot; if we have a very
			high force recovery level set, we try to avoid crashes
			by skipping this lookup */

			if (UNIV_LIKELY(srv_force_recovery < 5)
			    && !lock_clust_rec_cons_read_sees(
				    rec, index, offsets, trx->read_view)) {

				rec_t*	old_vers;
				/* The following call returns 'offsets'
				associated with 'old_vers' */
				err = row_sel_build_prev_vers_for_mysql(
					trx->read_view, clust_index,
					prebuilt, rec, &offsets, &heap,
					&old_vers, &mtr);

				if (err != DB_SUCCESS) {

					goto lock_wait_or_error;
				}

				if (old_vers == NULL) {
					/* The row did not exist yet in
					the read view */

					goto next_rec;
				}

				rec = old_vers;
			}
		} else {
			/* We are looking into a non-clustered index,
			and to get the right version of the record we
			have to look also into the clustered index: this
			is necessary, because we can only get the undo
			information via the clustered index record. */

			ut_ad(!dict_index_is_clust(index));

			if (!lock_sec_rec_cons_read_sees(
				    rec, trx->read_view)) {
				/* We should look at the clustered index.
				However, as this is a non-locking read,
				we can skip the clustered index lookup if
				the condition does not match the secondary
				index entry. */
				switch (row_search_idx_cond_check(
						buf, prebuilt, rec, offsets)) {
				case ICP_NO_MATCH:
					goto next_rec;
				case ICP_OUT_OF_RANGE:
					err = DB_RECORD_NOT_FOUND;
					goto idx_cond_failed;
                                case ICP_ABORTED_BY_USER:
				        err = DB_SEARCH_ABORTED_BY_USER;
					goto idx_cond_failed;
                                case ICP_ERROR:
				        err = DB_ERROR;
					goto idx_cond_failed;
				case ICP_MATCH:
					goto requires_clust_rec;
				}

				ut_error;
			}
		}
	}

locks_ok:
	/* NOTE that at this point rec can be an old version of a clustered
	index record built for a consistent read. We cannot assume after this
	point that rec is on a buffer pool page. Functions like
	page_rec_is_comp() cannot be used! */

	if (rec_get_deleted_flag(rec, comp)) {

		/* The record is delete-marked: we can skip it */

		if ((srv_locks_unsafe_for_binlog
		     || trx->isolation_level <= TRX_ISO_READ_COMMITTED)
		    && prebuilt->select_lock_type != LOCK_NONE
		    && !did_semi_consistent_read) {

			/* No need to keep a lock on a delete-marked record
			if we do not want to use next-key locking. */

			row_unlock_for_mysql(prebuilt, TRUE);
		}

		/* This is an optimization to skip setting the next key lock
		on the record that follows this delete-marked record. This
		optimization works because of the unique search criteria
		which precludes the presence of a range lock between this
		delete marked record and the record following it.

		For now this is applicable only to clustered indexes while
		doing a unique search except for HANDLER queries because
		HANDLER allows NEXT and PREV even in unique search on
		clustered index. There is scope for further optimization
		applicable to unique secondary indexes. Current behaviour is
		to widen the scope of a lock on an already delete marked record
		if the same record is deleted twice by the same transaction */
		if (index == clust_index && unique_search
		    && !prebuilt->used_in_HANDLER) {

			err = DB_RECORD_NOT_FOUND;

			goto normal_return;
		}

		goto next_rec;
	}

	/* Check if the record matches the index condition. */
	switch (row_search_idx_cond_check(buf, prebuilt, rec, offsets)) {
	case ICP_NO_MATCH:
		if (did_semi_consistent_read) {
			row_unlock_for_mysql(prebuilt, TRUE);
		}
		goto next_rec;
        case ICP_ABORTED_BY_USER:
	        err = DB_SEARCH_ABORTED_BY_USER;
		goto idx_cond_failed;
        case ICP_ERROR:
	        err = DB_ERROR;
		goto idx_cond_failed;
	case ICP_OUT_OF_RANGE:
		err = DB_RECORD_NOT_FOUND;
		goto idx_cond_failed;
	case ICP_MATCH:
		break;
	}

	/* Get the clustered index record if needed, if we did not do the
	search using the clustered index... */

	use_clustered_index =
		(index != clust_index && prebuilt->need_to_access_clustered);

	if (use_clustered_index && srv_prefix_index_cluster_optimization
	    && prebuilt->n_template <= index->n_fields) {
		/* ...but, perhaps avoid the clustered index lookup if
		all of the following are true:
		1) all columns are in the secondary index
		2) all values for columns that are prefix-only
		   indexes are shorter than the prefix size
		This optimization can avoid many IOs for certain schemas.
		*/
		ibool row_contains_all_values = TRUE;
		int i;
		for (i = 0; i < prebuilt->n_template; i++) {
			/* Condition (1) from above: is the field in the
			index (prefix or not)? */
			mysql_row_templ_t* templ =
				prebuilt->mysql_template + i;
			ulint secondary_index_field_no =
				templ->rec_prefix_field_no;
			if (secondary_index_field_no == ULINT_UNDEFINED) {
				row_contains_all_values = FALSE;
				break;
			}
			/* Condition (2) from above: if this is a
			prefix, is this row's value size shorter
			than the prefix? */
			if (templ->rec_field_is_prefix) {
				ulint record_size = rec_offs_nth_size(
					offsets,
					secondary_index_field_no);
				const dict_field_t *field =
					dict_index_get_nth_field(
						index,
						secondary_index_field_no);
				ut_a(field->prefix_len > 0);
				if (record_size >= field->prefix_len) {
					row_contains_all_values = FALSE;
					break;
				}
			}
		}
		/* If (1) and (2) were true for all columns above, use
		rec_prefix_field_no instead of rec_field_no, and skip
		the clustered lookup below. */
		if (row_contains_all_values) {
			for (i = 0; i < prebuilt->n_template; i++) {
				mysql_row_templ_t* templ =
					prebuilt->mysql_template + i;
				templ->rec_field_no =
					templ->rec_prefix_field_no;
				ut_a(templ->rec_field_no != ULINT_UNDEFINED);
			}
			use_clustered_index = FALSE;
			srv_stats.n_sec_rec_cluster_reads_avoided.inc();
		}
	}

	if (use_clustered_index) {

requires_clust_rec:
		ut_ad(index != clust_index);
		/* We use a 'goto' to the preceding label if a consistent
		read of a secondary index record requires us to look up old
		versions of the associated clustered index record. */

		ut_ad(rec_offs_validate(rec, index, offsets));

		/* It was a non-clustered index and we must fetch also the
		clustered index record */

		mtr_has_extra_clust_latch = TRUE;

		/* The following call returns 'offsets' associated with
		'clust_rec'. Note that 'clust_rec' can be an old version
		built for a consistent read. */

		err = row_sel_get_clust_rec_for_mysql(prebuilt, index, rec,
						      thr, &clust_rec,
						      &offsets, &heap, &mtr);
		switch (err) {
		case DB_SUCCESS:
			if (clust_rec == NULL) {
				/* The record did not exist in the read view */
				ut_ad(prebuilt->select_lock_type == LOCK_NONE);

				goto next_rec;
			}
			break;
		case DB_SUCCESS_LOCKED_REC:
			ut_a(clust_rec != NULL);
			if (srv_locks_unsafe_for_binlog
			     || trx->isolation_level
			    <= TRX_ISO_READ_COMMITTED) {
				/* Note that the clustered index record
				was locked. */
				prebuilt->new_rec_locks = 2;
			}
			err = DB_SUCCESS;
			break;
		default:
			goto lock_wait_or_error;
		}

		if (rec_get_deleted_flag(clust_rec, comp)) {

			/* The record is delete marked: we can skip it */

			if ((srv_locks_unsafe_for_binlog
			     || trx->isolation_level <= TRX_ISO_READ_COMMITTED)
			    && prebuilt->select_lock_type != LOCK_NONE) {

				/* No need to keep a lock on a delete-marked
				record if we do not want to use next-key
				locking. */

				row_unlock_for_mysql(prebuilt, TRUE);
			}

			goto next_rec;
		}

		result_rec = clust_rec;
		ut_ad(rec_offs_validate(result_rec, clust_index, offsets));

		if (prebuilt->idx_cond) {
			/* Convert the record to MySQL format. We were
			unable to do this in row_search_idx_cond_check(),
			because the condition is on the secondary index
			and the requested column is in the clustered index.
			We convert all fields, including those that
			may have been used in ICP, because the
			secondary index may contain a column prefix
			rather than the full column. Also, as noted
			in Bug #56680, the column in the secondary
			index may be in the wrong case, and the
			authoritative case is in result_rec, the
			appropriate version of the clustered index record. */
			if (!row_sel_store_mysql_rec(
				    buf, prebuilt, result_rec,
				    TRUE, clust_index, offsets)) {
				goto next_rec;
			}
		}
	} else {
		result_rec = rec;
	}

	/* We found a qualifying record 'result_rec'. At this point,
	'offsets' are associated with 'result_rec'. */

	ut_ad(rec_offs_validate(result_rec,
				result_rec != rec ? clust_index : index,
				offsets));
	ut_ad(!rec_get_deleted_flag(result_rec, comp));

	/* At this point, the clustered index record is protected
	by a page latch that was acquired when pcur was positioned.
	The latch will not be released until mtr_commit(&mtr). */

	if ((match_mode == ROW_SEL_EXACT
	     || prebuilt->n_rows_fetched >= MYSQL_FETCH_CACHE_THRESHOLD)
	    && prebuilt->select_lock_type == LOCK_NONE
	    && !prebuilt->templ_contains_blob
	    && !prebuilt->clust_index_was_generated
	    && !prebuilt->used_in_HANDLER
	    && !prebuilt->innodb_api
	    && prebuilt->template_type
	    != ROW_MYSQL_DUMMY_TEMPLATE
	    && !prebuilt->in_fts_query) {

		/* Inside an update, for example, we do not cache rows,
		since we may use the cursor position to do the actual
		update, that is why we require ...lock_type == LOCK_NONE.
		Since we keep space in prebuilt only for the BLOBs of
		a single row, we cannot cache rows in the case there
		are BLOBs in the fields to be fetched. In HANDLER we do
		not cache rows because there the cursor is a scrollable
		cursor. */

		ut_a(prebuilt->n_fetch_cached < MYSQL_FETCH_CACHE_SIZE);

		/* We only convert from InnoDB row format to MySQL row
		format when ICP is disabled. */

		if (!prebuilt->idx_cond) {

			/* We use next_buf to track the allocation of buffers
			where we store and enqueue the buffers for our
			pre-fetch optimisation.

			If next_buf == 0 then we store the converted record
			directly into the MySQL record buffer (buf). If it is
			!= 0 then we allocate a pre-fetch buffer and store the
			converted record there.

			If the conversion fails and the MySQL record buffer
			was not written to then we reset next_buf so that
			we can re-use the MySQL record buffer in the next
			iteration. */

			next_buf = next_buf
				 ? row_sel_fetch_last_buf(prebuilt) : buf;

			if (!row_sel_store_mysql_rec(
				next_buf, prebuilt, result_rec,
				result_rec != rec,
				result_rec != rec ? clust_index : index,
				offsets)) {

				if (next_buf == buf) {
					ut_a(prebuilt->n_fetch_cached == 0);
					next_buf = 0;
				}

				/* Only fresh inserts may contain incomplete
				externally stored columns. Pretend that such
				records do not exist. Such records may only be
				accessed at the READ UNCOMMITTED isolation
				level or when rolling back a recovered
				transaction. Rollback happens at a lower
				level, not here. */
				goto next_rec;
			}

			if (next_buf != buf) {
				row_sel_enqueue_cache_row_for_mysql(
					next_buf, prebuilt);
			}
		} else {
			row_sel_enqueue_cache_row_for_mysql(buf, prebuilt);
		}

		if (prebuilt->n_fetch_cached < MYSQL_FETCH_CACHE_SIZE) {
			goto next_rec;
		}

	} else {
		if (UNIV_UNLIKELY
		    (prebuilt->template_type == ROW_MYSQL_DUMMY_TEMPLATE)) {
			/* CHECK TABLE: fetch the row */

			if (result_rec != rec
			    && !prebuilt->need_to_access_clustered) {
				/* We used 'offsets' for the clust
				rec, recalculate them for 'rec' */
				offsets = rec_get_offsets(rec, index, offsets,
							  ULINT_UNDEFINED,
							  &heap);
				result_rec = rec;
			}

			memcpy(buf + 4, result_rec
			       - rec_offs_extra_size(offsets),
			       rec_offs_size(offsets));
			mach_write_to_4(buf,
					rec_offs_extra_size(offsets) + 4);
		} else if (!prebuilt->idx_cond && !prebuilt->innodb_api) {
			/* The record was not yet converted to MySQL format. */
			if (!row_sel_store_mysql_rec(
				    buf, prebuilt, result_rec,
				    result_rec != rec,
				    result_rec != rec ? clust_index : index,
				    offsets)) {
				/* Only fresh inserts may contain
				incomplete externally stored
				columns. Pretend that such records do
				not exist. Such records may only be
				accessed at the READ UNCOMMITTED
				isolation level or when rolling back a
				recovered transaction. Rollback
				happens at a lower level, not here. */
				goto next_rec;
			}
		}

		if (prebuilt->clust_index_was_generated) {
			row_sel_store_row_id_to_prebuilt(
				prebuilt, result_rec,
				result_rec == rec ? index : clust_index,
				offsets);
		}
	}

	/* From this point on, 'offsets' are invalid. */

	/* We have an optimization to save CPU time: if this is a consistent
	read on a unique condition on the clustered index, then we do not
	store the pcur position, because any fetch next or prev will anyway
	return 'end of file'. Exceptions are locking reads and the MySQL
	HANDLER command where the user can move the cursor with PREV or NEXT
	even after a unique search. */

	err = DB_SUCCESS;

idx_cond_failed:
	if (!unique_search
	    || !dict_index_is_clust(index)
	    || direction != 0
	    || prebuilt->select_lock_type != LOCK_NONE
	    || prebuilt->used_in_HANDLER
	    || prebuilt->innodb_api) {

		/* Inside an update always store the cursor position */

		btr_pcur_store_position(pcur, &mtr);

		if (prebuilt->innodb_api) {
			prebuilt->innodb_api_rec = result_rec;
		}
	}

	goto normal_return;

next_rec:
	/* Reset the old and new "did semi-consistent read" flags. */
	if (UNIV_UNLIKELY(prebuilt->row_read_type
			  == ROW_READ_DID_SEMI_CONSISTENT)) {
		prebuilt->row_read_type = ROW_READ_TRY_SEMI_CONSISTENT;
	}
	did_semi_consistent_read = FALSE;
	prebuilt->new_rec_locks = 0;

	/*-------------------------------------------------------------*/
	/* PHASE 5: Move the cursor to the next index record */

	/* NOTE: For moves_up==FALSE, the mini-transaction will be
	committed and restarted every time when switching b-tree
	pages. For moves_up==TRUE in index condition pushdown, we can
	scan an entire secondary index tree within a single
	mini-transaction. As long as the prebuilt->idx_cond does not
	match, we do not need to consult the clustered index or
	return records to MySQL, and thus we can avoid repositioning
	the cursor. What prevents us from buffer-fixing all leaf pages
	within the mini-transaction is the btr_leaf_page_release()
	call in btr_pcur_move_to_next_page(). Only the leaf page where
	the cursor is positioned will remain buffer-fixed. */

	if (UNIV_UNLIKELY(mtr_has_extra_clust_latch)) {
		/* We must commit mtr if we are moving to the next
		non-clustered index record, because we could break the
		latching order if we would access a different clustered
		index page right away without releasing the previous. */

		btr_pcur_store_position(pcur, &mtr);

		mtr_commit(&mtr);
		mtr_has_extra_clust_latch = FALSE;

		mtr_start(&mtr);
		if (sel_restore_position_for_mysql(&same_user_rec,
						   BTR_SEARCH_LEAF,
						   pcur, moves_up, &mtr)) {
#ifdef UNIV_SEARCH_DEBUG
			cnt++;
#endif /* UNIV_SEARCH_DEBUG */

			goto rec_loop;
		}
	}

	if (moves_up) {
		if (UNIV_UNLIKELY(!btr_pcur_move_to_next(pcur, &mtr))) {
not_moved:
			btr_pcur_store_position(pcur, &mtr);

			if (match_mode != 0) {
				err = DB_RECORD_NOT_FOUND;
			} else {
				err = DB_END_OF_INDEX;
			}

			goto normal_return;
		}
	} else {
		if (UNIV_UNLIKELY(!btr_pcur_move_to_prev(pcur, &mtr))) {
			goto not_moved;
		}
	}

#ifdef UNIV_SEARCH_DEBUG
	cnt++;
#endif /* UNIV_SEARCH_DEBUG */

	goto rec_loop;

lock_wait_or_error:
	/* Reset the old and new "did semi-consistent read" flags. */
	if (UNIV_UNLIKELY(prebuilt->row_read_type
			  == ROW_READ_DID_SEMI_CONSISTENT)) {
		prebuilt->row_read_type = ROW_READ_TRY_SEMI_CONSISTENT;
	}
	did_semi_consistent_read = FALSE;

	/*-------------------------------------------------------------*/

	btr_pcur_store_position(pcur, &mtr);

lock_table_wait:
	mtr_commit(&mtr);
	mtr_has_extra_clust_latch = FALSE;

	trx->error_state = err;

	/* The following is a patch for MySQL */

	que_thr_stop_for_mysql(thr);

	thr->lock_state = QUE_THR_LOCK_ROW;

	if (row_mysql_handle_errors(&err, trx, thr, NULL)) {
		/* It was a lock wait, and it ended */

		thr->lock_state = QUE_THR_LOCK_NOLOCK;
		mtr_start(&mtr);

		/* Table lock waited, go try to obtain table lock
		again */
		if (table_lock_waited) {
			table_lock_waited = FALSE;

			goto wait_table_again;
		}

		sel_restore_position_for_mysql(&same_user_rec,
					       BTR_SEARCH_LEAF, pcur,
					       moves_up, &mtr);

		if ((srv_locks_unsafe_for_binlog
		     || trx->isolation_level <= TRX_ISO_READ_COMMITTED)
		    && !same_user_rec) {

			/* Since we were not able to restore the cursor
			on the same user record, we cannot use
			row_unlock_for_mysql() to unlock any records, and
			we must thus reset the new rec lock info. Since
			in lock0lock.cc we have blocked the inheriting of gap
			X-locks, we actually do not have any new record locks
			set in this case.

			Note that if we were able to restore on the 'same'
			user record, it is still possible that we were actually
			waiting on a delete-marked record, and meanwhile
			it was removed by purge and inserted again by some
			other user. But that is no problem, because in
			rec_loop we will again try to set a lock, and
			new_rec_lock_info in trx will be right at the end. */

			prebuilt->new_rec_locks = 0;
		}

		mode = pcur->search_mode;

		goto rec_loop;
	}

	thr->lock_state = QUE_THR_LOCK_NOLOCK;

#ifdef UNIV_SEARCH_DEBUG
	/*	fputs("Using ", stderr);
	dict_index_name_print(stderr, index);
	fprintf(stderr, " cnt %lu ret value %lu err\n", cnt, err); */
#endif /* UNIV_SEARCH_DEBUG */
	goto func_exit;

normal_return:
	/*-------------------------------------------------------------*/
	que_thr_stop_for_mysql_no_error(thr, trx);

	mtr_commit(&mtr);

	if (prebuilt->idx_cond != 0) {

		/* When ICP is active we don't write to the MySQL buffer
		directly, only to buffers that are enqueued in the pre-fetch
		queue. We need to dequeue the first buffer and copy the contents
		to the record buffer that was passed in by MySQL. */

		if (prebuilt->n_fetch_cached > 0) {
			row_sel_dequeue_cached_row_for_mysql(buf, prebuilt);
			err = DB_SUCCESS;
		}

	} else if (next_buf != 0) {

		/* We may or may not have enqueued some buffers to the
		pre-fetch queue, but we definitely wrote to the record
		buffer passed to use by MySQL. */

		DEBUG_SYNC_C("row_search_cached_row");
		err = DB_SUCCESS;
	}

#ifdef UNIV_SEARCH_DEBUG
	/*	fputs("Using ", stderr);
	dict_index_name_print(stderr, index);
	fprintf(stderr, " cnt %lu ret value %lu err\n", cnt, err); */
#endif /* UNIV_SEARCH_DEBUG */

func_exit:
	trx->op_info = "";
	if (UNIV_LIKELY_NULL(heap)) {
		mem_heap_free(heap);
	}

	/* Set or reset the "did semi-consistent read" flag on return.
	The flag did_semi_consistent_read is set if and only if
	the record being returned was fetched with a semi-consistent read. */
	ut_ad(prebuilt->row_read_type != ROW_READ_WITH_LOCKS
	      || !did_semi_consistent_read);

	if (UNIV_UNLIKELY(prebuilt->row_read_type != ROW_READ_WITH_LOCKS)) {
		if (UNIV_UNLIKELY(did_semi_consistent_read)) {
			prebuilt->row_read_type = ROW_READ_DID_SEMI_CONSISTENT;
		} else {
			prebuilt->row_read_type = ROW_READ_TRY_SEMI_CONSISTENT;
		}
	}

	ut_ad(!trx->has_search_latch);
#ifdef UNIV_SYNC_DEBUG
	ut_ad(!btr_search_own_any());
	ut_ad(!sync_thread_levels_nonempty_trx(trx->has_search_latch));
#endif /* UNIV_SYNC_DEBUG */

	DEBUG_SYNC_C("innodb_row_search_for_mysql_exit");

	return(err);
}

/*******************************************************************//**
Checks if MySQL at the moment is allowed for this table to retrieve a
consistent read result, or store it to the query cache.
@return	TRUE if storing or retrieving from the query cache is permitted */
UNIV_INTERN
ibool
row_search_check_if_query_cache_permitted(
/*======================================*/
	trx_t*		trx,		/*!< in: transaction object */
	const char*	norm_name)	/*!< in: concatenation of database name,
					'/' char, table name */
{
	dict_table_t*	table;
	ibool		ret	= FALSE;

	/* Disable query cache altogether for all tables if recovered XA
	transactions in prepared state exist. This is because we do not
	restore the table locks for those transactions and we may wrongly
	set ret=TRUE above if "lock_table_get_n_locks(table) == 0". See
	"Bug#14658648 XA ROLLBACK (DISTRIBUTED DATABASE) NOT WORKING WITH
	QUERY CACHE ENABLED".
	Read trx_sys->n_prepared_recovered_trx without mutex protection,
	not possible to end up with a torn read since n_prepared_recovered_trx
	is word size. */
	if (trx_sys->n_prepared_recovered_trx > 0) {

		return(FALSE);
	}

	table = dict_table_open_on_name(norm_name, FALSE, FALSE,
					DICT_ERR_IGNORE_NONE);

	if (table == NULL) {

		return(FALSE);
	}

	/* Start the transaction if it is not started yet */

	trx_start_if_not_started(trx);

	/* If there are locks on the table or some trx has invalidated the
	cache up to our trx id, then ret = FALSE.
	We do not check what type locks there are on the table, though only
	IX type locks actually would require ret = FALSE. */

	if (lock_table_get_n_locks(table) == 0
	    && trx->id >= table->query_cache_inv_trx_id) {

		ret = TRUE;

		/* If the isolation level is high, assign a read view for the
		transaction if it does not yet have one */

		if (trx->isolation_level >= TRX_ISO_REPEATABLE_READ
		    && !trx->read_view) {

			trx->read_view =
				read_view_open_now(trx->id,
						   trx->prebuilt_view);
			trx->global_read_view = trx->read_view;
		}
	}

	dict_table_close(table, FALSE, FALSE);

	return(ret);
}

/*******************************************************************//**
Read the AUTOINC column from the current row. If the value is less than
0 and the type is not unsigned then we reset the value to 0.
@return	value read from the column */
static
ib_uint64_t
row_search_autoinc_read_column(
/*===========================*/
	dict_index_t*	index,		/*!< in: index to read from */
	const rec_t*	rec,		/*!< in: current rec */
	ulint		col_no,		/*!< in: column number */
	ulint		mtype,		/*!< in: column main type */
	ibool		unsigned_type)	/*!< in: signed or unsigned flag */
{
	ulint		len;
	const byte*	data;
	ib_uint64_t	value;
	mem_heap_t*	heap = NULL;
	ulint		offsets_[REC_OFFS_NORMAL_SIZE];
	ulint*		offsets	= offsets_;

	rec_offs_init(offsets_);

	offsets = rec_get_offsets(rec, index, offsets, col_no + 1, &heap);

	if (rec_offs_nth_sql_null(offsets, col_no)) {
		/* There is no non-NULL value in the auto-increment column. */
		value = 0;
		goto func_exit;
	}

	data = rec_get_nth_field(rec, offsets, col_no, &len);

	switch (mtype) {
	case DATA_INT:
		ut_a(len <= sizeof value);
		value = mach_read_int_type(data, len, unsigned_type);
		break;

	case DATA_FLOAT:
		ut_a(len == sizeof(float));
		value = (ib_uint64_t) mach_float_read(data);
		break;

	case DATA_DOUBLE:
		ut_a(len == sizeof(double));
		value = (ib_uint64_t) mach_double_read(data);
		break;

	default:
		ut_error;
	}

	if (!unsigned_type && (ib_int64_t) value < 0) {
		value = 0;
	}

func_exit:
	if (UNIV_LIKELY_NULL(heap)) {
		mem_heap_free(heap);
	}

	return(value);
}

/** Get the maximum and non-delete-marked record in an index.
@param[in]	index	index tree
@param[in,out]	mtr	mini-transaction (may be committed and restarted)
@return maximum record, page s-latched in mtr
@retval NULL if there are no records, or if all of them are delete-marked */
static
const rec_t*
row_search_get_max_rec(
	dict_index_t*	index,
	mtr_t*		mtr)
{
	btr_pcur_t	pcur;
	const rec_t*	rec;
	/* Open at the high/right end (false), and init cursor */
	btr_pcur_open_at_index_side(
		false, index, BTR_SEARCH_LEAF, &pcur, true, 0, mtr);

	do {
		const page_t*	page;

		page = btr_pcur_get_page(&pcur);
		rec = page_find_rec_max_not_deleted(page);

		if (page_rec_is_user_rec(rec)) {
			break;
		} else {
			rec = NULL;
		}
		btr_pcur_move_before_first_on_page(&pcur);
	} while (btr_pcur_move_to_prev(&pcur, mtr));

	btr_pcur_close(&pcur);

	return(rec);
}

/*******************************************************************//**
Read the max AUTOINC value from an index.
@return DB_SUCCESS if all OK else error code, DB_RECORD_NOT_FOUND if
column name can't be found in index */
UNIV_INTERN
dberr_t
row_search_max_autoinc(
/*===================*/
	dict_index_t*	index,		/*!< in: index to search */
	const char*	col_name,	/*!< in: name of autoinc column */
	ib_uint64_t*	value)		/*!< out: AUTOINC value read */
{
	dict_field_t*	dfield = dict_index_get_nth_field(index, 0);
	dberr_t		error = DB_SUCCESS;
	*value = 0;

	if (strcmp(col_name, dfield->name) != 0) {
		error = DB_RECORD_NOT_FOUND;
	} else {
		mtr_t		mtr;
		const rec_t*	rec;

		mtr_start(&mtr);

		rec = row_search_get_max_rec(index, &mtr);

		if (rec != NULL) {
			ibool unsigned_type = (
				dfield->col->prtype & DATA_UNSIGNED);

			*value = row_search_autoinc_read_column(
				index, rec, 0,
				dfield->col->mtype, unsigned_type);
		}

		mtr_commit(&mtr);
	}

	return(error);
}