MySQL Bugs: #35273: Corrupted table if using GIS index

Bug #35273	Corrupted table if using GIS index
Submitted:	13 Mar 2008 16:02	Modified:	17 Mar 2008 22:11
Reporter:	Guilhem Bichot	Email Updates:
Status:	Closed	Impact on me:	None
Category:	MySQL Server: Maria storage engine	Severity:	S3 (Non-critical)
Version:	5.1-maria	OS:	Any
Assigned to:	Guilhem Bichot	CPU Architecture:	Any

Description:
Use this (taken from gis-rtree.test):

-- source include/have_geometry.inc

set storage_engine=maria;

#
# test of rtree (using with spatial data)
#
--disable_warnings
DROP TABLE IF EXISTS t1, t2;
--enable_warnings

CREATE TABLE t1 (
  fid INT NOT NULL AUTO_INCREMENT PRIMARY KEY, 
  g GEOMETRY NOT NULL,
  SPATIAL KEY(g)
) Transactional=0;

SHOW CREATE TABLE t1;

let $1=150;
let $2=150;
while ($1)
{
  eval INSERT INTO t1 (g) VALUES (GeomFromText('LineString($1 $1, $2 $2)'));
  dec $1;
  inc $2;
}

It will say in the error log that t1.MAI is corrupted

How to repeat:
see description

Suggested fix:
at the end of one INSERT, the table is unlocked, which causes a flush of all index pages (expected, table is not transactional) and this flush fails because it finds one page is pinned. This page was pinned by a ma_write_keypage(). Either it should not have been pinned, or it should have been unpinned before the flush (I assume the latter is the good solution).

When this bug is fixed, please run the attached maria-gis-rtree.test to see if there are next bugs.

test file for GIS Maria bugs

Attachment: maria-g (application/octet-stream, text), 0 bytes.

test file for GIS Maria bugs

Attachment: maria-gis-rtree.test (application/octet-stream, text), 41.47 KiB.

A patch for this bug has been committed. After review, it may
be pushed to the relevant source trees for release in the next
version. You can access the patch from:

  http://lists.mysql.com/commits/44157

ChangeSet@1.2617, 2008-03-17 22:31:50+01:00, guilhem@gbichot4.local +3 -0
  Fix for BUG#35273 "Corrupted table if using GIS index":
  when modifying R-tree indices we forgot to unpin pages (causing
  flush at-end-of-statement to fail - because page is pinned - and thus
  mark table corrupted) and also to unlock key_del (causing assertion
  failure info->used_key_del==0 at maria_close()).