From: Roy Lyseng
Date: March 2 2011 4:16pm
Subject: Re: bzr commit into mysql-trunk branch (roy.lyseng:3343) Bug#11766739
List-Archive: http://lists.mysql.com/commits/132291
Message-Id: <4D6E6D54.9000300@oracle.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 8bit

Hi Øystein,

thank you for reviewing.

On 22.02.11 08.53, Øystein Grøvlen wrote:
> Hi Roy,
>
> Thanks for fixing this. The patch is approved. Some minor suggestions in-line.
>
>
> On 02/14/2011 04:01 PM, Roy Lyseng wrote:
>  > #At file:///home/rl136806/mysql/repo/mysql-work5/ based on
> revid:roy.lyseng@stripped
>  >
>  > 3343 Roy Lyseng 2011-02-14
>  > Bug#11766739: Crash in tmp_table_param::init() with semijoin=on
>  >
>  > Used to be bug#59919.
>  >
>  > A crash occurs when a certain subquery is transformed with a semi join
>  > and a semi join materialization operation is attempted.
>  > The subquery contains a nested join structure (an outer join where
>  > the inner part is itself an inner join).
>  > Function setup_sj_materialization() tries to lookup the embedding
>  > semi join nest for an inner table of the subquery. However, it uses
>  > the pointer tab->table->pos_in_table_list->embedding, pointing to
>  > an outer join nest, instead of the correct pointer tab->emb_sj_nest.
>  > When dereferencing this pointer, a crash occurs.
>  >
>  > mysql-test/include/subquery_sj.inc
>  > Added test case for bug#11766739
>  >
>  > mysql-test/r/subquery_sj_all.result
>  > mysql-test/r/subquery_sj_all_jcl6.result
>  > mysql-test/r/subquery_sj_all_jcl7.result
>  > mysql-test/r/subquery_sj_dupsweed.result
>  > mysql-test/r/subquery_sj_dupsweed_jcl6.result
>  > mysql-test/r/subquery_sj_dupsweed_jcl7.result
>  > mysql-test/r/subquery_sj_firstmatch.result
>  > mysql-test/r/subquery_sj_firstmatch_jcl6.result
>  > mysql-test/r/subquery_sj_firstmatch_jcl7.result
>  > mysql-test/r/subquery_sj_loosescan.result
>  > mysql-test/r/subquery_sj_loosescan_jcl6.result
>  > mysql-test/r/subquery_sj_loosescan_jcl7.result
>  > mysql-test/r/subquery_sj_mat.result
>  > mysql-test/r/subquery_sj_mat_jcl6.result
>  > mysql-test/r/subquery_sj_mat_jcl7.result
>  > mysql-test/r/subquery_sj_mat_nosj.result
>  > mysql-test/r/subquery_sj_none.result
>  > mysql-test/r/subquery_sj_none_jcl6.result
>  > mysql-test/r/subquery_sj_none_jcl7.result
>  > Added test results for bug#11766739
>  >
>  > sql/sql_select.cc
>  > In setup_sj_materialization(), replaced use of pointer
>  > tab->table->pos_in_table_list->embedding with
>  > tab->emb_sj_nest.
>  >
>  > modified:
>  > mysql-test/include/subquery_sj.inc
>  > mysql-test/r/subquery_sj_all.result
>  > mysql-test/r/subquery_sj_all_jcl6.result
>  > mysql-test/r/subquery_sj_all_jcl7.result
>  > mysql-test/r/subquery_sj_dupsweed.result
>  > mysql-test/r/subquery_sj_dupsweed_jcl6.result
>  > mysql-test/r/subquery_sj_dupsweed_jcl7.result
>  > mysql-test/r/subquery_sj_firstmatch.result
>  > mysql-test/r/subquery_sj_firstmatch_jcl6.result
>  > mysql-test/r/subquery_sj_firstmatch_jcl7.result
>  > mysql-test/r/subquery_sj_loosescan.result
>  > mysql-test/r/subquery_sj_loosescan_jcl6.result
>  > mysql-test/r/subquery_sj_loosescan_jcl7.result
>  > mysql-test/r/subquery_sj_mat.result
>  > mysql-test/r/subquery_sj_mat_jcl6.result
>  > mysql-test/r/subquery_sj_mat_jcl7.result
>  > mysql-test/r/subquery_sj_mat_nosj.result
>  > mysql-test/r/subquery_sj_none.result
>  > mysql-test/r/subquery_sj_none_jcl6.result
>  > mysql-test/r/subquery_sj_none_jcl7.result
>  > sql/sql_select.cc
>  > === modified file 'mysql-test/include/subquery_sj.inc'
>  > --- a/mysql-test/include/subquery_sj.inc 2011-01-27 11:38:22 +0000
>  > +++ b/mysql-test/include/subquery_sj.inc 2011-02-14 15:01:33 +0000
>  > @@ -3504,3 +3504,25 @@ deallocate prepare s;
>  > DROP TABLE ot1, ot2, ot3, it1;
>  >
>  > --echo # End of the test for bug#57623.
>  > +
>  > +--echo #
>  > +--echo # Bug#11766739: Crash in tmp_table_param::init() with semijoin=on
>  > +--echo #
>  > +
>  > +CREATE TABLE a (f1 INTEGER NOT NULL) ENGINE=MyISAM;
>  > +CREATE TABLE b (f1 INTEGER NOT NULL, f2 INTEGER) ENGINE=MyISAM;
>
> Suggestions, which you may feel free to ignore:
> - Use t1, t2 for table names.

Done.

> - "NOT NULL" is not significant and can be dropped to get a more
> minimal test case.

Done.

> - Since MyISAM is default in this test, it should not be necessary
> to specify engine. (I acknowledge that this error does not occur
> with Innodb since constant optimizations will not be done.)

I prefer to be explicit on this one.
>
>  > +
>  > +INSERT INTO a VALUES (1);
>  > +INSERT INTO b VALUES (1,1), (2,1);
>  > +
>  > +let $query=
>  > +SELECT * FROM b
>  > +WHERE f2 IN (SELECT a.f1
>  > + FROM a LEFT OUTER JOIN (b AS b1 JOIN b AS b2 ON 1=1) ON 1=1);
>
> How aboue "ON TRUE" instead of "ON 1=1"?

Done.
>
>  > +
>  > +eval EXPLAIN $query;
>  > +eval $query;
>  > +
>  > +DROP TABLE a, b;
>  > +
>  > +--echo # End of the test for bug#11766739.
>  >
>
> ...
>
> --
> Øystein
>