In CLUSTER, VACUUM FULL and ALTER TABLE SET TABLESPACE
I erroneously set checksum before log_newpage, which
sets the LSN and invalidates the checksum. So set
checksum immediately *after* log_newpage.

Bug report Fujii Masao, Fix and patch by Jeff Davis

contrib/pg_trgm's make_trigrams() was coded to ignore multibyte character
boundaries and just make trigrams from bytes if USE_WIDE_UPPER_LOWER wasn't
defined.  This is a bit odd, since there's no obvious reason why trigram
compaction rules should depend on the presence of towlower() and friends.
What's more, there was an Assert() that would fail if that code path was
fed any multibyte characters.

We need to do something about this since the pending regex-indexing patch
has an assumption that you get just one "trgm" from any three characters.
The best solution seems to be to remove the USE_WIDE_UPPER_LOWER
dependency, which shouldn't really have been there in the first place.
The second loop in make_trigrams() is now just a fast path and not a
potentially incompatible algorithm.

If there is anybody still using Postgres on machines without wcstombs() or
towlower(), and they have non-ASCII data indexed by pg_trgm, they'll need
to REINDEX those indexes after pg_upgrade to 9.3, else searches may fail
incorrectly. It seems likely that there are no such installations, though.

In passing, rename cnt_trigram to compact_trigram, which seems to better
describe its functionality, and improve make_trigrams' test for whether it
has to use the slow path or not (per a suggestion from Alexander Korotkov).

Per report from Jaime Casanova.  Very curious that no one else has seen
this failure ... but the code is clearly wrong as-is.

KaiGai Kohei, with comment and doc wordsmithing by me

There's been some confusion expressed about this point, so clarify.
Extended version of a patch by David Wheeler.

Counting newlines shows that quite a few recent patches have neglected
to update the output-lines count given to PageOutput().  Fortunately
it's not terribly critical that this be exact, since we long since
exceeded the height of most people's terminal windows.  Still, maybe
we ought to think of a way to not have to maintain this manually anymore.

This allows convenient re-execution of commands.

Will Leinweber, reviewed by Peter Eisentraut, Daniel Farina, and Tom Lane

Bug report by David Wheeler, diagnosis assistance from Tom Lane.

'strdup' the PSQLRC environment variable value before calling a routine
that might free() it.

Backpatch to 9.2, where the bug first appeared.

Throw an error instead.

Backpatch to all supported branches.

Dickson S. Guedes

The old formula didn't take into account that each WAL sender process needs
a spinlock. We had also already exceeded the fixed number of spinlocks
reserved for misc purposes (10). Bump that to 30.

Backpatch to 9.0, where WAL senders were introduced. If I counted correctly,
9.0 had exactly 10 predefined spinlocks, and 9.1 exceeded that, but bump the
limit in 9.0 too because 10 is uncomfortably close to the edge.

The point of turning off track_activities is to avoid this reporting
overhead, but a thinko in commit 4f42b546
caused pgstat_report_activity() to perform half of its updates anyway.
Fix that, and also make sure that we clear all the now-disabled fields
when transitioning to the non-reporting state.

Laurenz Albe

Notice and complain about PQcancel() failures.  Also, don't dump core if
an error PGresult doesn't contain severity and message subfields, as it
might not if it was generated by libpq itself.  (We have a longstanding
TODO item to improve that, but in the meantime isolationtester had better
cope.)

I tripped across the latter item while investigating a trouble report on
buildfarm member spoonbill.  As for the former, there's no evidence that
PQcancel failure is actually involved in spoonbill's problem, but it still
seems like a bad idea to ignore an error return code.

Security: CVE-2013-1899, CVE-2013-1901

An oversight in commit e710b65c allowed
database names beginning with "-" to be treated as though they were secure
command-line switches; and this switch processing occurs before client
authentication, so that even an unprivileged remote attacker could exploit
the bug, needing only connectivity to the postmaster's port.  Assorted
exploits for this are possible, some requiring a valid database login,
some not.  The worst known problem is that the "-r" switch can be invoked
to redirect the process's stderr output, so that subsequent error messages
will be appended to any file the server can write.  This can for example be
used to corrupt the server's configuration files, so that it will fail when
next restarted.  Complete destruction of database tables is also possible.

Fix by keeping the database name extracted from a startup packet fully
separate from command-line switches, as had already been done with the
user name field.

The Postgres project thanks Mitsumasa Kondo for discovering this bug,
Kyotaro Horiguchi for drafting the fix, and Noah Misch for recognizing
the full extent of the danger.

Security: CVE-2013-1899

The pg_start_backup() and pg_stop_backup() functions checked the privileges
of the initially-authenticated user rather than the current user, which is
wrong.  For example, a user-defined index function could successfully call
these functions when executed by ANALYZE within autovacuum.  This could
allow an attacker with valid but low-privilege database access to interfere
with creation of routine backups.  Reported and fixed by Noah Misch.

Security: CVE-2013-1901

This reverts commit 3780fc67.

HP-UX didn't like it.  There would probably be a way to fix that, but
since the net effect of all of this is zero because ecpg ends up using
libpq anyway, it's not worth bothering further.

In commit 0f61d4dd, I added code to copy up
column width estimates for each column of a subquery.  That code supposed
that the subquery couldn't have any output columns that didn't correspond
to known columns of the current query level --- which is true when a query
is parsed from scratch, but the assumption fails when planning a view that
depends on another view that's been redefined (adding output columns) since
the upper view was made.  This results in an assertion failure or even a
crash, as per bug #8025 from lindebg.  Remove the Assert and instead skip
the column if its resno is out of the expected range.

This will hopefully be easier to use than pg_config for users who are
already used to the pkg-config interface.  It also works better for
multi-arch installations.

reviewed by Tom Lane

It doesn't actually use libpq.  But we need to keep libpq in the
CPPFLAGS for building, because compatlib uses ecpglib.h which uses
libpq-fe.h, but we don't need to refer to libpq for linking.

reviewed by Tom Lane

Now that pg_dump no longer dumps invalid indexes, per commit
683abc73, have pg_upgrade also skip
them.  Previously pg_upgrade threw an error if invalid indexes existed.

Backpatch to 9.2, 9.1, and 9.0 (where pg_upgrade was added to git)

The modern incarnation of md.c is by no means specific to magnetic disk
technology, but every so often we hear from someone who's misled by the
label.  Try to clarify that it will work for anything that supports
standard filesystem operations.  Per suggestion from Andrew Dunstan.

Windows sometimes gets upset if we rename a large directory and then try
to use the old name quickly, as seen in occasional buildfarm failures.
So we avoid that by building the old version in the intended
destination in the first place instead of renaming it, similar to the
change made for the same reason in commit b7f8465c.

In some parallel make situations, the install-headers target could be
called before the installation directories are created by installdirs,
causing the installation to fail.  Fix that by making install-headers
depend on installdirs.

The JSON parser is converted into a recursive descent parser, and
exposed for use by other modules such as extensions. The API provides
hooks for all the significant parser event such as the beginning and end
of objects and arrays, and providing functions to handle these hooks
allows for fairly simple construction of a wide variety of JSON
processing functions. A set of new basic processing functions and
operators is also added, which use this API, including operations to
extract array elements, object fields, get the length of arrays and the
set of keys of a field, deconstruct an object into a set of key/value
pairs, and create records from JSON objects and arrays of objects.

Catalog version bumped.

Andrew Dunstan, with some documentation assistance from Merlin Moncure.

I changed this in commit fd15dba5, but
missed the fact that the SGML documentation of the function specified
exactly what it did.  Well, one of the two places where it's specified
documented that --- probably I looked at the other place and thought
nothing needed to be done.  Sync the two places where encode() and
decode() are described.

9.2 uses a kluge representation of "indislive"; we have to account for
that when examining pg_index.  Simplest solution is to check indisready
for 9.0 and 9.1 as well; that's harmless though unnecessary, so it's
not worth making a version distinction for.

Fixes oversight in commit 683abc73,
as noted by Andres Freund.

Covers commits through today.  Not back-patching into back branches
yet, since this is just for people to review in advance.

Fixes by me, per griping by Thom Brown.

The main change here is to call security_compute_create_name_raw()
rather than security_compute_create_raw().  This ups the minimum
requirement for libselinux from 2.0.99 to 2.1.10, but it looks
like most distributions will have picked that up before 9.3 is out.

KaiGai Kohei

DST law changes in Chile, Haiti, Morocco, Paraguay, some Russian areas.
Historical corrections for numerous places.

On older-model gcc, the original coding of UTILITY_BEGIN_QUERY() can
draw this error because of multiple assignments to _needCleanup.
Rather than mark that variable volatile, we can suppress the warning
by arranging to have just one unconditional assignment before PG_TRY.

This event takes place just before ddl_command_end, and is fired if and
only if at least one object has been dropped by the command.  (For
instance, DROP TABLE IF EXISTS of a table that does not in fact exist
will not lead to such a trigger firing).  Commands that drop multiple
objects (such as DROP SCHEMA or DROP OWNED BY) will cause a single event
to fire.  Some firings might be surprising, such as
ALTER TABLE DROP COLUMN.

The trigger is fired after the drop has taken place, because that has
been deemed the safest design, to avoid exposing possibly-inconsistent
internal state (system catalogs as well as current transaction) to the
user function code.  This means that careful tracking of object
identification is required during the object removal phase.

Like other currently existing events, there is support for tag
filtering.

To support the new event, add a new pg_event_trigger_dropped_objects()
set-returning function, which returns a set of rows comprising the
objects affected by the command.  This is to be used within the user
function code, and is mostly modelled after the recently introduced
pg_identify_object() function.

Catalog version bumped due to the new function.

Dimitri Fontaine and Álvaro Herrera
Review by Robert Haas, Tom Lane