WTIH -> WITH

Pointed out by Dmitriy Olshevskiy

Backpatch to 9.5 where create_policy.sgml was added.

Per buildfarm member crake.

Modify pg_dump to restore postgres/template1 databases to non-default
tablespaces by switching out of the database to be moved, then switching
back.

Also, to fix potentially cases where the old/new tablespaces might not
match, fix pg_upgrade to process new/old tablespaces separately in all
cases.

Report by Marti Raudsepp

Patch by Marti Raudsepp, me

Backpatch through 9.0

The previous way didn't work for vpath builds, and make distprep was
busted too.

Reported off-list by Andres Freund.

I think this particular branch is actually dead, but the analysis to
prove that is not trivial, so instead take the weasel way.

Reported by Jinyu Zhang
Backpatch to 9.5, where BRIN was introduced.

Commit 45ba424f improved foreign key lookups during bulk updates
when the FK value does not change.  When restoring a schema dump
from a database with many (say 100,000) foreign keys, this cache
would grow very big and every ALTER TABLE command was causing an
InvalidateConstraintCacheCallBack(), which uses a sequential hash
table scan.  This could cause a severe performance regression in
restoring a schema dump (including during pg_upgrade).

The patch uses a heuristic method of detecting when the hash table
should be destroyed and recreated.
InvalidateConstraintCacheCallBack() adds the current size of the
hash table to a counter.  When that sum reaches 1,000,000, the hash
table is flushed.  This fixes the regression without noticeable
harm to the bulk update use case.

Jan Wieck
Backpatch to 9.3 where the performance regression was introduced.

Naming the individual lwlocks seems like something that may be useful
for other types of debugging, monitoring, or instrumentation output,
but this commit just implements it for the specific case of
trace_lwlocks.

Patch by me, reviewed by Amit Kapila and Kyotaro Horiguchi

Back-patch to 9.3 where PageHeaderData layout was changed.

Michael Paquier

The "typo" alleged in commit 1e460d4b was actually a comment that was
correct when written, but I missed updating it in commit b5282aa8.
Use a slightly less specific (and hopefully more future-proof) description
of what is collected.  Back-patch to 9.2 where that commit appeared, and
revert the comment to its then-entirely-correct state before that.

We're adding OIDs, not TIDs, to invalItems.

Pointed out by Etsuro Fujita.

Back-patch to all supported branches.

The list-wrangling here was done wrong, allowing the same state to get
put into the list twice.  The following loop then would clone it twice.
The second clone would wind up with no inarcs, so that there was no
observable misbehavior AFAICT, but a useless state in the finished NFA
isn't an especially good thing.

Declaration of varibale inside ÓÝ×Õ

Mark index tuple as dead if it's pointed by kill_prior_tuple during
ordinary (search) scan and remove it during insert process if there is no
enough space for new tuple to insert. This improves select performance
because index will not return tuple marked as dead and improves insert
performance because it reduces number of page split.

Anastasia Lubennikova <a.lubennikova@postgrespro.ru> with
 minor editorialization by me

This commit makes postmaster forcibly remove the files signaling
a standby promotion request. Otherwise, the existence of those files
can trigger a promotion too early, whether a user wants that or not.

This removal of files is usually unnecessary because they can exist
only during a few moments during a standby promotion. However
there is a race condition: if pg_ctl promote is executed and creates
the files during a promotion, the files can stay around even after
the server is brought up to new master. Then, if new standby starts
by using the backup taken from that master, the files can exist
at the server startup and should be removed in order to avoid
an unexpected promotion.

Back-patch to 9.1 where promote signal file was introduced.

Problem reported by Feike Steenbergen.
Original patch by Michael Paquier, modified by me.

Discussion: 20150528100705.4686.91426@wrigleys.postgresql.org

Even views considered "simple" enough to be automatically updatable may
have mulitple relations involved (eg: in a where clause).  We need to
make sure and lock those relations when rewriting the query.

Back-patch to 9.3 where updatable views were added.

Pointed out by Andres, patch thanks to Dean Rasheed.

Author: Pavel Stehule
Reviewed-By: Andres Freund
Discussion: 5594FE7A.5050205@iki.fi

This was forgotten in 8a3631f8 (commit that originally added the parameter)
and 0ca9907c (commit that added the documentation later that year).

Back-patch to all supported versions.

Thomas Munro, reviewed by Michael Paquier, modified by me.

Per discussion, nowadays it is possible to have tablespaces that have
wildly different I/O characteristics from others.  Setting different
effective_io_concurrency parameters for those has been measured to
improve performance.

Author: Julien Rouhaud
Reviewed by: Andres Freund

This new function provides information about SSL extensions present in
the X509 certificate used for the current connection.

Extension version updated to version 1.1.

Author: Дмитрий Воронин (Dmitry Voronin)
Reviewed by: Michael Paquier, Heikki Linnakangas, Álvaro Herrera

If a transaction never reaches the standby, later tests find unexpected
cluster state.  A "tail-copy: query result matches" test failure has
been the usual symptom.  Among the buildfarm members having run this
test suite, most have exhibited that symptom at least once.  Back-patch
to 9.5, where pg_rewind was introduced.

Michael Paquier, reported by Christoph Berg.

Commit f828654e introduced the 'n' option, but it invoked
gettimeofday() independently of the 'm' option. If both options were
in use (or multiple 'n' options), or if 'n' was in use along with
csvlog, then the reported times could be different for the same log
message.

To fix, initialize a global variable with gettimeofday() once per log
message, and use that for both formats.

Don't bother coordinating the time for the 't' option, which has much
lower resolution.

Per complaint by Alvaro Herrera.

We were missing a few return checks on OpenSSL calls.  Should be pretty
harmless, since we haven't seen any user reports about problems, and
this is not a high-traffic module anyway; still, a bug is a bug, so
backpatch this all the way back to 9.0.

Author: Michael Paquier, while reviewing another sslinfo patch

Prints time as Unix epoch with milliseconds.

Tomas Vondra, reviewed by Fabien Coelho.

Cleanup process could be called by ordinary insert/update and could take a lot
of time. Add vacuum_delay_point() to make this process interruptable. Under
vacuum this call will also throttle a vacuum process to decrease system load,
called from insert/update it will not throttle, and that reduces a latency.

Backpatch for all supported branches.

Jeff Janes <jeff.janes@gmail.com>

Add pages deleted from GIN's pending list during cleanup to free space map
immediately. Clean up process could be initiated by ordinary insert but adding
page to FSM might occur only at vacuum. On some workload like never-vacuumed
insert-only tables it could cause a huge bloat.

Jeff Janes <jeff.janes@gmail.com>

Recent commit 0426f349 changed handling of error context reports
in such a way to have a minor effect on the sepgsql regression
output. Adapt the expected output file to suit. Since that commit
was HEAD only, so is this one.

Previous limit was 16 characters, due to lack of support for multiple passes
of encryption.

Marko Tiikkaja

Since 6fcd8851 it is possible to immediately reserve WAL when creating a
slot via pg_create_physical_replication_slot(). Extend the replication
protocol to allow that as well.

Although, in contrast to the SQL interface, it is possible to update the
reserved location via the replication interface, it is still useful
being able to reserve upon creation there. Otherwise the logic in
ReplicationSlotReserveWal() has to be repeated in slot employing
clients.

Author: Michael Paquier
Discussion: CAB7nPqT0Wc1W5mdYGeJ_wbutbwNN+3qgrFR64avXaQCiJMGaYA@mail.gmail.com

RESERV. RESERV is meant for tokens like "now" and having them in that
category throws errors like these when used as an input date:

stark=# SELECT 'doy'::timestamptz;
ERROR:  unexpected dtype 33 while parsing timestamptz "doy"
LINE 1: SELECT 'doy'::timestamptz;
               ^
stark=# SELECT 'dow'::timestamptz;
ERROR:  unexpected dtype 32 while parsing timestamptz "dow"
LINE 1: SELECT 'dow'::timestamptz;
               ^

Found by LLVM's Libfuzzer

This has been broken since 9.3 (commit 82b1b213 to be exact),
which suggests that nobody is any longer using a Windows build system that
doesn't provide a symlink emulation.  Still, it's wrong on its own terms,
so repair.

YUriy Zhuravlev

Remove the code in plpgsql that suppressed the innermost line of CONTEXT
for messages emitted by RAISE commands.  That was never more than a quick
backwards-compatibility hack, and it's pretty silly in cases where the
RAISE is nested in several levels of function.  What's more, it violated
our design theory that verbosity of error reports should be controlled
on the client side not the server side.

To alleviate the resulting noise increase, introduce a feature in libpq
and psql whereby the CONTEXT field of messages can be suppressed, either
always or only for non-error messages.  Printing CONTEXT for errors only
is now their default behavior.

The actual code changes here are pretty small, but the effects on the
regression test outputs are widespread.  I had to edit some of the
alternative expected outputs by hand; hopefully the buildfarm will soon
find anything I fat-fingered.

In passing, fix up (again) the output line counts in psql's various
help displays.  Add some commentary about how to verify them.

Pavel Stehule, reviewed by Petr Jelínek, Jeevan Chalke, and others

Oskari Saarenmaa. Backpatch to stable branches where applicable.

If the number of heap blocks is not multiples of pages per range, the
summarizing produces wrong summary information for the last brin index
tuple while vacuuming.

Problem reported by Tatsuo Ishii and fixed by Amit Langote.

Discussion at "[HACKERS] BRIN INDEX value (message id :20150903.174935.1946402199422994347.t-ishii@sraoss.co.jp)
Backpatched to 9.5 in which brin index was added.

Formerly, we treated only portals created in the current subtransaction as
having failed during subtransaction abort.  However, if the error occurred
while running a portal created in an outer subtransaction (ie, a cursor
declared before the last savepoint), that has to be considered broken too.

To allow reliable detection of which ones those are, add a bookkeeping
field to struct Portal that tracks the innermost subtransaction in which
each portal has actually been executed.  (Without this, we'd end up
failing portals containing functions that had called the subtransaction,
thereby breaking plpgsql exception blocks completely.)

In addition, when we fail an outer-subtransaction Portal, transfer its
resources into the subtransaction's resource owner, so that they're
released early in cleanup of the subxact.  This fixes a problem reported by
Jim Nasby in which a function executed in an outer-subtransaction cursor
could cause an Assert failure or crash by referencing a relation created
within the inner subtransaction.

The proximate cause of the Assert failure is that AtEOSubXact_RelationCache
assumed it could blow away a relcache entry without first checking that the
entry had zero refcount.  That was a bad idea on its own terms, so add such
a check there, and to the similar coding in AtEOXact_RelationCache.  This
provides an independent safety measure in case there are still ways to
provoke the situation despite the Portal-level changes.

This has been broken since subtransactions were invented, so back-patch
to all supported branches.

Tom Lane and Michael Paquier