diff options
| author | Sergey Petrunya <psergey@askmonty.org> | 2012-01-25 18:33:57 +0400 |
|---|---|---|
| committer | Sergey Petrunya <psergey@askmonty.org> | 2012-01-25 18:33:57 +0400 |
| commit | 73cc529b51dd4262e81df86491e3f2803946339c (patch) | |
| tree | 2706fa7435c85b4c80e9161d75876204441a32e3 /sql/opt_subselect.cc | |
| parent | e10816118a35cdb67cd9e1e7b671c946741d3f7c (diff) | |
| download | mariadb-git-73cc529b51dd4262e81df86491e3f2803946339c.tar.gz | |
BUG#920255: Wrong result (extra rows) with loosescan and IN subquery
The problem was that LooseScan execution code assumed that tab->key holds
the index used for looseScan. This is only true when range or full index
scan are used. In case of ref access, the index is in tab->ref.key (and
tab->index==0 which explains how LooseScan passed tests with ref access: they
used one index)
Fixed by setting/using loosescan_key, which always the correct index#.
Diffstat (limited to 'sql/opt_subselect.cc')
| -rw-r--r-- | sql/opt_subselect.cc | 1 |
1 files changed, 1 insertions, 0 deletions
diff --git a/sql/opt_subselect.cc b/sql/opt_subselect.cc index 8e043b17bcf..63a90891720 100644 --- a/sql/opt_subselect.cc +++ b/sql/opt_subselect.cc @@ -4107,6 +4107,7 @@ int setup_semijoin_dups_elimination(JOIN *join, ulonglong options, for (uint kp=0; kp < pos->loosescan_picker.loosescan_parts; kp++) keylen += tab->table->key_info[keyno].key_part[kp].store_length; + tab->loosescan_key= keyno; tab->loosescan_key_len= keylen; if (pos->n_sj_tables > 1) tab[pos->n_sj_tables - 1].do_firstmatch= tab; |