mariadb/storage/myisam/mi_range.c
Monty f36ca142f7 Added page_range to records_in_range() to improve range statistics
Prototype change:
-  virtual ha_rows records_in_range(uint inx, key_range *min_key,
-                                   key_range *max_key)
+  virtual ha_rows records_in_range(uint inx, const key_range *min_key,
+                                   const key_range *max_key,
+                                   page_range *res)

The handler can ignore the page_range parameter. In the case the handler
updates the parameter, the optimizer can deduce the following:
- If previous range's last key is on the same block as next range's first
  key
- If the current key range is in one block
- We can also assume that the first and last block read are cached!
  This can be used for a better calculation of IO seeks when we
  estimate the cost of a range index scan.

The parameter is fully implemented for MyISAM, Aria and InnoDB.
A separate patch will update handler::multi_range_read_info_const() to
take the benefits of this change and also remove the double
records_in_range() calls that are not anymore needed.
2020-03-27 03:54:45 +02:00

327 lines
11 KiB
C

/*
Copyright (c) 2000, 2011, Oracle and/or its affiliates
Copyright (c) 2010, 2020, MariaDB Corporation.
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software 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 St, Fifth Floor, Boston, MA 02110-1335 USA */
/*
Gives a approximated number of how many records there is between two keys.
Used when optimizing querries.
*/
#include "myisamdef.h"
#include "rt_index.h"
static double _mi_record_pos(MI_INFO *, const uchar *, key_part_map,
enum ha_rkey_function, ulonglong *);
static double _mi_search_pos(MI_INFO *,MI_KEYDEF *,uchar *, uint,uint,
my_off_t,my_bool, ulonglong *);
static uint _mi_keynr(MI_INFO *info,MI_KEYDEF *,uchar *, uchar *,uint *);
/*
Estimate how many records there is in a given range
SYNOPSIS
mi_records_in_range()
info MyISAM handler
inx Index to use
min_key Min key. Is = 0 if no min range
max_key Max key. Is = 0 if no max range
NOTES
We should ONLY return 0 if there is no rows in range
RETURN
HA_POS_ERROR error (or we can't estimate number of rows)
number Estimated number of rows
*/
ha_rows mi_records_in_range(MI_INFO *info, int inx,
const key_range *min_key, const key_range *max_key,
page_range *pages)
{
ha_rows res;
double start_pos,end_pos,diff;
DBUG_ENTER("mi_records_in_range");
if ((inx = _mi_check_index(info,inx)) < 0)
DBUG_RETURN(HA_POS_ERROR);
if (fast_mi_readinfo(info))
DBUG_RETURN(HA_POS_ERROR);
info->update&= (HA_STATE_CHANGED+HA_STATE_ROW_CHANGED);
if (info->s->concurrent_insert)
mysql_rwlock_rdlock(&info->s->key_root_lock[inx]);
switch(info->s->keyinfo[inx].key_alg){
#ifdef HAVE_RTREE_KEYS
case HA_KEY_ALG_RTREE:
{
uchar * key_buff;
uint start_key_len;
/*
The problem is that the optimizer doesn't support
RTree keys properly at the moment.
Hope this will be fixed some day.
But now NULL in the min_key means that we
didn't make the task for the RTree key
and expect BTree functionality from it.
As it's not able to handle such request
we return the error.
*/
if (!min_key)
{
res= HA_POS_ERROR;
break;
}
key_buff= info->lastkey+info->s->base.max_key_length;
start_key_len= _mi_pack_key(info,inx, key_buff,
(uchar*) min_key->key, min_key->keypart_map,
(HA_KEYSEG**) 0);
res= rtree_estimate(info, inx, key_buff, start_key_len,
myisam_read_vec[min_key->flag]);
res= res ? res : 1; /* Don't return 0 */
break;
}
#endif
case HA_KEY_ALG_BTREE:
default:
start_pos= (min_key ?_mi_record_pos(info, min_key->key,
min_key->keypart_map, min_key->flag,
&pages->first_page)
: (double) 0);
end_pos= (max_key ? _mi_record_pos(info, max_key->key,
max_key->keypart_map, max_key->flag,
&pages->last_page)
: (double) info->state->records);
res= (end_pos < start_pos ? (ha_rows) 0 :
(end_pos == start_pos ? (ha_rows) 1 : (ha_rows) (end_pos-start_pos)));
if (start_pos == (double) HA_POS_ERROR || end_pos == (double) HA_POS_ERROR)
res=HA_POS_ERROR;
else
{
diff= end_pos - start_pos;
if (diff >= 0)
{
if (!(res= (ha_rows) (diff + 0.5)))
res= 1;
}
else
res= 0;
}
}
if (info->s->concurrent_insert)
mysql_rwlock_unlock(&info->s->key_root_lock[inx]);
fast_mi_writeinfo(info);
DBUG_PRINT("info",("records: %ld",(ulong) (res)));
DBUG_RETURN(res);
}
/*
To find an approximate relative position of a key tuple among all index
key tuples would not be hard if we considered B-trees where all key
tuples were contained only in leaf nodes. If we consider a B-tree where
key tuples are stored also in non-leaf nodes we have to convert such
tree into the tree of the first type. The transformation procedure is
simple: the key tuple k goes alter the last key tuple in the most right
sub-tree pointer to which is coupled with k. As a result of this
transformation each leaf node except the most right one in the tree will
contain one extra key tuple following those originally belonging to
the leaf.
*/
/* Find relative position (in records) for key in index-tree */
static double _mi_record_pos(MI_INFO *info, const uchar *key,
key_part_map keypart_map,
enum ha_rkey_function search_flag,
ulonglong *final_page)
{
uint inx=(uint) info->lastinx, nextflag, key_len;
MI_KEYDEF *keyinfo=info->s->keyinfo+inx;
uchar *key_buff;
double pos;
DBUG_ENTER("_mi_record_pos");
DBUG_PRINT("enter",("search_flag: %d",search_flag));
DBUG_ASSERT(keypart_map);
key_buff=info->lastkey+info->s->base.max_key_length;
key_len=_mi_pack_key(info,inx,key_buff,(uchar*) key, keypart_map,
(HA_KEYSEG**) 0);
DBUG_EXECUTE("key",_mi_print_key(DBUG_FILE,keyinfo->seg,
(uchar*) key_buff,key_len););
nextflag=myisam_read_vec[search_flag];
if (!(nextflag & (SEARCH_FIND | SEARCH_NO_FIND | SEARCH_LAST)))
key_len=USE_WHOLE_KEY;
/*
my_handler.c:ha_compare_text() has a flag 'skip_end_space'.
This is set in my_handler.c:ha_key_cmp() in dependence on the
compare flags 'nextflag' and the column type.
TEXT columns are of type HA_KEYTYPE_VARTEXT. In this case the
condition is skip_end_space= ((nextflag & (SEARCH_FIND |
SEARCH_UPDATE)) == SEARCH_FIND).
SEARCH_FIND is used for an exact key search. The combination
SEARCH_FIND | SEARCH_UPDATE is used in write/update/delete
operations with a comment like "Not real duplicates", whatever this
means. From the condition above we can see that 'skip_end_space' is
always false for these operations. The result is that trailing space
counts in key comparison and hence, empty strings ('', string length
zero, but not NULL) compare less that strings starting with control
characters and these in turn compare less than strings starting with
blanks.
When estimating the number of records in a key range, we request an
exact search for the minimum key. This translates into a plain
SEARCH_FIND flag. Using this alone would lead to a 'skip_end_space'
compare. Empty strings would be expected above control characters.
Their keys would not be found because they are located below control
characters.
This is the reason that we add the SEARCH_UPDATE flag here. It makes
the key estimation compare in the same way like key write operations
do. Only so we will find the keys where they have been inserted.
Adding the flag unconditionally does not hurt as it is used in the
above mentioned condition only. So it can safely be used together
with other flags.
*/
pos=_mi_search_pos(info,keyinfo,key_buff,key_len,
nextflag | SEARCH_SAVE_BUFF | SEARCH_UPDATE,
info->s->state.key_root[inx], TRUE,
final_page);
if (pos >= 0.0)
{
DBUG_PRINT("exit",("pos: %g",(pos*info->state->records)));
DBUG_RETURN(pos*info->state->records);
}
DBUG_RETURN((double) (HA_POS_ERROR));
}
/* This is a modified version of _mi_search */
/* Returns offset for key in indextable (decimal 0.0 <= x <= 1.0) */
static double _mi_search_pos(register MI_INFO *info,
register MI_KEYDEF *keyinfo,
uchar *key, uint key_len, uint nextflag,
register my_off_t pos, my_bool last_in_level,
ulonglong *final_page)
{
int flag;
uint nod_flag,keynr,UNINIT_VAR(max_keynr);
my_bool after_key;
uchar *keypos,*buff;
double offset;
DBUG_ENTER("_mi_search_pos");
if (pos == HA_OFFSET_ERROR)
DBUG_RETURN(0.5);
if (!(buff=_mi_fetch_keypage(info,keyinfo,pos,DFLT_INIT_HITS,info->buff,1)))
goto err;
*final_page= pos;
flag=(*keyinfo->bin_search)(info,keyinfo,buff,key,key_len,nextflag,
&keypos,info->lastkey, &after_key);
nod_flag=mi_test_if_nod(buff);
keynr=_mi_keynr(info,keyinfo,buff,keypos,&max_keynr);
if (flag)
{
if (flag == MI_FOUND_WRONG_KEY)
DBUG_RETURN(-1); /* error */
/*
Didn't found match. keypos points at next (bigger) key
Try to find a smaller, better matching key.
Matches keynr + [0-1]
*/
if (flag > 0 && ! nod_flag)
offset= 1.0;
else if ((offset=_mi_search_pos(info,keyinfo,key,key_len,nextflag,
_mi_kpos(nod_flag,keypos),
last_in_level && after_key,
final_page)) < 0)
DBUG_RETURN(offset);
}
else
{
/*
Found match. Keypos points at the start of the found key
Matches keynr+1
*/
offset=1.0; /* Matches keynr+1 */
if ((nextflag & SEARCH_FIND) && nod_flag &&
((keyinfo->flag & (HA_NOSAME | HA_NULL_PART)) != HA_NOSAME ||
key_len != USE_WHOLE_KEY))
{
/*
There may be identical keys in the tree. Try to match on of those.
Matches keynr + [0-1]
*/
if ((offset=_mi_search_pos(info,keyinfo,key,key_len,SEARCH_FIND,
_mi_kpos(nod_flag,keypos),
last_in_level && after_key,
final_page)) < 0)
DBUG_RETURN(offset); /* Read error */
}
}
DBUG_PRINT("info",("keynr: %d offset: %g max_keynr: %d nod: %d flag: %d",
keynr,offset,max_keynr,nod_flag,flag));
DBUG_RETURN((keynr+offset-MY_TEST(!nod_flag))/
(max_keynr+MY_TEST(nod_flag || !last_in_level)));
err:
DBUG_PRINT("exit",("Error: %d",my_errno));
DBUG_RETURN (-1.0);
}
/* Get keynummer of current key and max number of keys in nod */
static uint _mi_keynr(MI_INFO *info, register MI_KEYDEF *keyinfo, uchar *page,
uchar *keypos, uint *ret_max_key)
{
uint nod_flag,keynr,max_key;
uchar t_buff[HA_MAX_KEY_BUFF],*end;
end= page+mi_getint(page);
nod_flag=mi_test_if_nod(page);
page+=2+nod_flag;
if (!(keyinfo->flag & (HA_VAR_LENGTH_KEY | HA_BINARY_PACK_KEY)))
{
*ret_max_key= (uint) (end-page)/(keyinfo->keylength+nod_flag);
return (uint) (keypos-page)/(keyinfo->keylength+nod_flag);
}
max_key=keynr=0;
t_buff[0]=0; /* Safety */
while (page < end)
{
if (!(*keyinfo->get_key)(keyinfo,nod_flag,&page,t_buff))
return 0; /* Error */
max_key++;
if (page == keypos)
keynr=max_key;
}
*ret_max_key=max_key;
return(keynr);
}