Johann Spies <jspies@sun.ac.za> writes:
> Breakpoint 1, 0x0827eef6 in errfinish ()
> (gdb) bt
> #0 0x0827eef6 in errfinish ()
> #1 0x0828039c in elog_finish ()
> #2 0x082993b3 in MemoryContextAlloc ()
> #3 0x0828d437 in pg_mb2wchar ()
> #4 0x0811c288 in CreateCopyDestReceiver ()
> #5 0x0811ea43 in DoCopy ()
> #6 0x081f9d00 in ProcessUtility ()
> #7 0x081f6ed9 in PostgresMain ()
> #8 0x081f7cac in FreeQueryDesc ()
> #9 0x081f84d9 in PortalRun ()
> #10 0x081f3b51 in pg_parse_query ()
> #11 0x081f5296 in PostgresMain ()
> #12 0x081c8a1f in ClosePostmasterPorts ()
> #13 0x081c9924 in PostmasterMain ()
> #14 0x081823b5 in main ()
Hmmm ... this is interesting, but unfortunately backtraces from
non-debug builds are often not very trustworthy, and that seems to have
happened here. CreateCopyDestReceiver doesn't call pg_mb2wchar, and
pg_mb2wchar doesn't call MemoryContextAlloc, so we're being lied to
right at the spot where we might've hoped to learn something.
Could you please rebuild the executables with --enable-debug and try
again?
regards, tom lane