Corrupt records in WAL archive

2014-08-26 06:16:27 PDT : LOG: invalid contrecord length 2 in log file
545, segment 146, offset 0
2014-08-26 06:16:32 PDT : LOG: restored log file
"000000010000022100000091" from archive
2014-08-26 06:16:32 PDT : LOG: restored log file
"000000010000022100000092" from archive
2014-08-26 06:16:32 PDT : LOG: invalid contrecord length 2 in log file
545, segment 146, offset 0
2014-08-26 06:16:37 PDT : LOG: restored log file
"000000010000022100000091" from archive
2014-08-26 06:16:37 PDT : LOG: restored log file
"000000010000022100000092" from archive
2014-08-26 06:16:37 PDT : LOG: invalid contrecord length 2 in log file
545, segment 146, offset 0
2014-08-26 06:16:42 PDT : LOG: restored log file
"000000010000022100000091" from archive
2014-08-26 06:16:42 PDT : LOG: restored log file
"000000010000022100000092" from archive

What's this about? I found several other complaints about this, but no
answers other than, "do another basebackup, or maybe your whole database is
corrupt."

We're trying to set up a reliable warm standby. Each time this happens
we're vulnerable until the next basebackup is complete -- the better part
of a day. It's not reassuring.

This is 9.2.4 running on Linux. We're going to update to the latest 9.2
before we start the next basebackup. It would be very reassuring to know
this is fixed.

Thanks,
Craig