[postgis-users] Problem with st_endpoint in postgis 1.3.1 for windows

[Mark Cave-Ayland]

On Tue, 2007-08-21 at 23:41 +0200, Francois Eric wrote:
> Hello,
> I have a problem when I run st_endpoint function : the query kill the
> process.
> The same query with st_startpoint works fine. I'm also able to run
> st_endpoint with postgis 1.2.1 with no problem.
> My conf 
> Windows XP
> "PostgreSQL 8.2.4 on i686-pc-mingw32, compiled by GCC gcc.exe (GCC)
> 3.4.2 (mingw-special)"
> "POSTGIS="1.3.1" GEOS="3.0.0rc4-CAPI-1.3.3" PROJ="Rel. 4.5.0, 22 Oct
> 2006" USE_STATS"
> The log
> 2007-08-21 23:11:04 STATEMENT: select y(st _endpoint(the_geom)) from
> streets limit 10
> 2007-08-21 23:11:16 LOG: server process (PID 4988) exited with exit
> code -1073741819
> 2007-08-21 23:11:16 LOG: terminating any other active server processes
> 2007-08-21 23:11:16 WARNING: terminating connection because of crash
> of another server process
> 2007-08-21 23:11:16 DETAIL: The postmaster has commanded this server
> process to roll back the current transaction and exit, because another
> server process exited abnormally and possibly corrupted shared memory.
> 2007-08-21 23:11:16 HINT: In a moment you should be able to reconnect
> to the database and repeat your command.
> 2007-08-21 23:11:16 WARNING: terminating connection because of crash
> of another server process
> 2007-08-21 23:11:16 DETAIL: The postmaster has commanded this server
> process to roll back the current transaction and exit, because another
> server process exited abnormally and possibly corrupted shared memory.
> 2007-08-21 23:11:16 HINT: In a moment you should be able to reconnect
> to the database and repeat your command.
> 2007-08-21 23:11:16 LOG: all server processes terminated;
> reinitializing
> 2007-08-21 23:11:16 LOG: database system was interrupted at 2007-08-21
> 23:09:07 Paris, Madrid
> 2007-08-21 23:11:16 LOG: checkpoint record is at 5/D3ECC1D0
> 2007-08-21 23:11:16 LOG: redo record is at 5/D3ECC1D0; undo record is
> at 0/0; shutdown FALSE
> 2007-08-21 23:11:16 LOG: next transaction ID: 0/1226346; next OID:
> 101454
> 2007-08-21 23:11:16 LOG: next MultiXactId: 1; next MultiXactOffset: 0
> 2007-08-21 23:11:16 LOG: database system was not properly shut down;
> automatic recovery in progress
> 2007-08-21 23:11:16 LOG: redo starts at 5/D3ECC220
> 2007-08-21 23:11:16 LOG: record with zero length at 5/D42FAF58
> 2007-08-21 23:11:16 LOG: redo done at 5/D42FAF28
> 2007-08-21 23:11:18 LOG: database system is ready
> Thanks for your help

Yuck :(  There was a double-free condition in LWGEOM_endpoint_linestring
which was causing the backend to crash just before the function exit.
I've verified the bug, and applied a fix to the latest SVN - thanks for
the bug report!


Kind regards,

Mark.
________________________________________________________________________
-- 
ILande - Open Source Consultancy
http://www.ilande.co.uk