slow log reporting invalid Rows_read

[aeva black]

On some occasions, the value for Rows_read in the slow query log is 18446744073709551615, even for simple queries. I am seeing this in 5.1.55-percona-server, built from source on gentoo. I have not been able to determine what is common denominator between the many queries which show this value -- some are single row PK lookup, others are range scan, JOIN, subquery, etc. So there is wide variety of types of queries. But always, it is the same number (which looks like BIGINT -1 wrapping error).

Here is sample bit of slow log. Contact me directly to get more details.

> # Thread_id: 338264 Schema:xxxx Last_errno: 0 Killed: 0
> # Query_time: 0.000161 Lock_time: 0.000038 Rows_sent: 0 Rows_examined: 1 Rows_affected: 0 Rows_read: 18446744073709551615
> # Bytes_sent: 465 Tmp_tables: 0 Tmp_disk_tables: 0 Tmp_table_sizes: 0