query log corrupted-looking entries

after an attempt at stress-testing my app i started seeing some
corrupted-looking entries in the postgresql query log. for example:

...
2006-10-10 21:33:24 PDT [31175]: [509-1] LOG: statem2006-10-10 21:33:24
PDT [31166]: [4206-1] LOG: duration: 0.197 ms
ent: EXECUTE <unnamed> [PREPARE: SELECT P.IS_DISPLAY, P.IS_ACTIVE,
P.CITY_POSTAL_CODE_ID,P.property_name,P.address_1,P.address_2,
...
2006-10-10 21:35:11 PDT [31163]: [20258-1] LOG: statement2006-10-10
21:35:11 PDT [31179]: [218-1] LOG: duration: 57.264 ms
2006-10-10 21:35:11 PDT [31182]: [1446-1] LOG: statement: <BIND>
...

this wories me in general, but in particular messes up pgfouine attempts
at log analysis.

any thoughts on where this might be coming from, how to deal with it,
etc.?

the relevant config settings are (show all | grep ' log_' output):

log_connections | on
log_destination | stderr
log_directory | pg_log
log_disconnections | on
log_duration | on
log_error_verbosity | default
log_executor_stats | off
log_filename | postgresql-%a-%H.log
log_hostname | off
log_line_prefix | %t [%p]: [%l-1]
log_min_duration_statement | -1
log_min_error_statement | panic
log_min_messages | notice
log_parser_stats | off
log_planner_stats | off
log_rotation_age | 60
log_rotation_size | 0
log_statement | all
log_statement_stats | off
log_truncate_on_rotation | on