pgsql: Fix a bug with SSI and prepared transactions:

Fix a bug with SSI and prepared transactions:

If there's a dangerous structure T0 ---> T1 ---> T2, and T2 commits first,
we need to abort something. If T2 commits before both conflicts appear,
then it should be caught by OnConflict_CheckForSerializationFailure. If
both conflicts appear before T2 commits, it should be caught by
PreCommit_CheckForSerializationFailure. But that is actually run when
T2 *prepares*. Fix that in OnConflict_CheckForSerializationFailure, by
treating a prepared T2 as if it committed already.

This is mostly a problem for prepared transactions, which are in prepared
state for some time, but also for regular transactions because they also go
through the prepared state in the SSI code for a short moment when they're
committed.

Kevin Grittner and Dan Ports

Branch
------
REL9_1_STABLE

Details
-------
http://git.postgresql.org/pg/commitdiff/046d52f7d319419d338fa605f9d6e00b21c5c5ff

Modified Files
--------------
src/backend/storage/lmgr/predicate.c           |   38 ++++++++++++++++++++++--
src/test/regress/expected/prepared_xacts.out   |    6 ++--
src/test/regress/expected/prepared_xacts_1.out |    2 +-
src/test/regress/sql/prepared_xacts.sql        |    2 +-
4 files changed, 40 insertions(+), 8 deletions(-)