Here is a trivial fix of locking issue in pgstattuple(). It was locking buffers around PageGetHeapFreeSpace() in the heap scan loop, but not in the scanning of the tail. I think we need locks in the tail, too.
Regards, --- ITAGAKI Takahiro NTT Open Source Software Center
pgstattuple_lock.patch
Description: Binary data
---------------------------(end of broadcast)--------------------------- TIP 5: don't forget to increase your free space map settings
