[postgis-devel] [PostGIS] #1697: Crash in GiST index with empty MultiPolygons

[PostGIS]

#1697: Crash in GiST index with empty MultiPolygons
---------------------------+------------------------------------------------
 Reporter:  realityexists  |       Owner:  pramsey      
     Type:  defect         |      Status:  new          
 Priority:  critical       |   Milestone:  PostGIS 2.0.0
Component:  postgis        |     Version:  trunk        
 Keywords:                 |  
---------------------------+------------------------------------------------

Comment(by pramsey):

 So, several hours of debugging later, no real change. The problem happens
 in picksplit, it happens right at the start, as PostgreSQL hands in the
 user code a bogus box_current. The box contains a crazy size header which
 leads to a failed palloc in the gidx_copy function, and finally the failed
 memcpy that the stacktrace reports. None of this happens if you use
 geometry instead of geography but still use an nd index. Same code, same
 input objects, very different results.

 I thought perhaps the gist compress function (which all the objects
 traverse as they are added to the node) might do something different
 between geometry and geography, but it doesn't. In both cases, compress
 finds no calculateable bounding box (everything is empty) so it returns
 back the input.

-- 
Ticket URL: <http://trac.osgeo.org/postgis/ticket/1697#comment:7>
PostGIS <http://trac.osgeo.org/postgis/>
The PostGIS Trac is used for bug, enhancement & task tracking, a user and developer wiki, and a view into the subversion code repository of PostGIS project.