This was broken in 5a593153.

The original code was overly optimistic about the cost of scanning a
BRIN index, leading to BRIN indexes being selected when they'd be a
worse choice than some other index.  This complete rewrite should be
more accurate.

Author: David Rowley, based on an earlier patch by Emre Hasegeli
Reviewed-by: Emre Hasegeli
Discussion: https://postgr.es/m/CAKJS1f9n-Wapop5Xz1dtGdpdqmzeGqQK4sV2MK-zZugfC14Xtw@mail.gmail.com

This displays the number of workers launched, thus the test is
dependant on configuration to some degree.  We'll see whether that
turns out ot be a problem.

Author: Rafia Sabih
Discussion: https://postgr.es/m/20170331185540.zmsue4ndvqtnayqw@alap3.anarazel.de

Author: Dilip Kumar
Discussion: https://postgr.es/m/20170331184603.qcp7t4md5bzxbx32@alap3.anarazel.de

When sending a tuple attribute, the previous coding erroneously sent the
length byte before encoding conversion, which would lead to protocol
failures on the receiving side if the length did not match the following
string.

To fix that, use pq_sendcountedtext() for sending tuple attributes,
which takes care of all of that internally.  To match the API of
pq_sendcountedtext(), send even text values without a trailing zero byte
and have the receiving end put it in place instead.  This matches how
the standard FE/BE protocol behaves.
Reported-by: Kyotaro HORIGUCHI <horiguchi.kyotaro@lab.ntt.co.jp>

Also add opr_sanity check that all preloaded immutable functions are
parallel safe.  (Per discussion, this does not necessarily have to be
true for all possible such functions, but deviations would be unlikely
enough that maintaining such a test is reasonable.)
Reported-by: David Rowley <david.rowley@2ndquadrant.com>
Reviewed-by: Robert Haas <robertmhaas@gmail.com>
Reviewed-by: Tom Lane <tgl@sss.pgh.pa.us>

In struct _archiveHandle, some of the fields have the same name as a
typedef.  This is kind of confusing, so rename the types so they have
names distinct from the struct fields.  In C++, the previous coding
changes the meaning of the typedef in the scope of the struct, causing
warnings and possibly other problems.
Reviewed-by: Andres Freund <andres@anarazel.de>

Remove unnecessary casts, safely schema-qualify the ones that remain,
lose an unnecessary level of sub-SELECT, reformat for tidiness.

Not currently a problem, but could be with stricter bool behavior under
stdbool or C++.
Reviewed-by: Andres Freund <andres@anarazel.de>

Using AclResult as a bool or vice versa works by accident, but it's
unusual and possibly confusing style, so write it out more explicitly.

Clean up some code comments in new extended statistics code, from
7b504eb2.

It was not used for what the comment claimed, at all. It was actually used
as the 'base' argument to strtol(), when reading the iteration count. We
don't need a constant for base-10, so remove it.

Avoid corner cases during 2PC with 6bad580d

This is the SQL standard-conforming variant of PostgreSQL's serial
columns.  It fixes a few usability issues that serial columns have:

- CREATE TABLE / LIKE copies default but refers to same sequence
- cannot add/drop serialness with ALTER TABLE
- dropping default does not drop sequence
- need to grant separate privileges to sequence
- other slight weirdnesses because serial is some kind of special macro
Reviewed-by: Vitaly Burovoy <vitaly.burovoy@gmail.com>

For normal commits and aborts we already reset PgXact->xmin,
so we can simply avoid running SnapshotResetXmin() twice.

During performance tests by Alexander Korotkov, diagnosis
by Andres Freund showed PgXact array as a bottleneck. After
manual analysis by me of the code paths that touch those
memory locations, I was able to identify extraneous code
in the main transaction commit path.

Avoiding touching highly contented shmem improves concurrent
performance slightly on all workloads, confirmed by tests
run by Ashutosh Sharma and Alexander Korotkov.

Simon Riggs

Discussion: CANP8+jJdXE9b+b9F8CQT-LuxxO0PBCB-SZFfMVAdp+akqo4zfg@mail.gmail.com

HandleFunctionRequest() is no longer responsible for reading the protocol
message from the client, since commit 2b3a8b20. Fix the outdated
comments.

HandleFunctionRequest() now always returns 0, because the code that used
to return EOF was moved in 2b3a8b20. Therefore, the caller no longer
needs to check the return value.

Reported by Andres Freund. Backpatch to all supported versions, even though
this doesn't have any user-visible effect, to make backporting future
patches in this area easier.

Discussion: https://www.postgresql.org/message-id/20170405010525.rt5azbya5fkbhvrx@alap3.anarazel.de

When using integer timestamps, the interval-comparison functions tried
to compute the overall magnitude of an interval as an int64 number of
microseconds.  As reported by Frazer McLean, this overflows for intervals
exceeding about 296000 years, which is bad since we nominally allow
intervals many times larger than that.  That results in wrong comparison
results, and possibly in corrupted btree indexes for columns containing
such large interval values.

To fix, compute the magnitude as int128 instead.  Although some compilers
have native support for int128 calculations, many don't, so create our
own support functions that can do 128-bit addition and multiplication
if the compiler support isn't there.  These support functions are designed
with an eye to allowing the int128 code paths in numeric.c to be rewritten
for use on all platforms, although this patch doesn't do that, or even
provide all the int128 primitives that will be needed for it.

Back-patch as far as 9.4.  Earlier releases did not guard against overflow
of interval values at all (commit 146604ec fixed that), so it seems not
very exciting to worry about overly-large intervals for them.

Before 9.6, we did not assume that unreferenced "static inline" functions
would not draw compiler warnings, so omit functions not directly referenced
by timestamp.c, the only present consumer of int128.h.  (We could have
omitted these functions in HEAD too, but since they were written and
debugged on the way to the present patch, and they look likely to be needed
by numeric.c, let's keep them in HEAD.)  I did not bother to try to prevent
such warnings in a --disable-integer-datetimes build, though.

Before 9.5, configure will never define HAVE_INT128, so the part of
int128.h that exploits a native int128 implementation is dead code in the
9.4 branch.  I didn't bother to remove it, thinking that keeping the file
looking similar in different branches is more useful.

In HEAD only, add a simple test harness for int128.h in src/tools/.

In back branches, this does not change the float-timestamps code path.
That's not subject to the same kind of overflow risk, since it computes
the interval magnitude as float8.  (No doubt, when this code was originally
written, overflow was disregarded for exactly that reason.)  There is a
precision hazard instead :-(, but we'll avert our eyes from that question,
since no complaints have been reported and that code's deprecated anyway.

Kyotaro Horiguchi and Tom Lane

Discussion: https://postgr.es/m/1490104629.422698.918452336.26FA96B7@webmail.messagingengine.com

In line with other lock reductions related to planning.

Simon Riggs

Follow on patch in the multi-variate statistics patch series.

CREATE STATISTICS s1 WITH (dependencies) ON (a, b) FROM t;
ANALYZE;
will collect dependency stats on (a, b) and then use the measured
dependency in subsequent query planning.

Commit 7b504eb2 added
CREATE STATISTICS with n-distinct coefficients. These are now
specified using the mutually exclusive option WITH (ndistinct).

Author: Tomas Vondra, David Rowley
Reviewed-by: Kyotaro HORIGUCHI, Álvaro Herrera, Dean Rasheed, Robert Haas
and many other comments and contributions
Discussion: https://postgr.es/m/56f40b20-c464-fad2-ff39-06b668fac47c@2ndquadrant.com

Other part of Beena Emerson's patch to allow testing

Make every page in a hash index which isn't all-zeroes have a valid
special space, so that tools like pageinspect don't error out.

Also, make pageinspect cope with all-zeroes pages, because
_hash_alloc_buckets can leave behind large numbers of those until
they're consumed by splits.

Ashutosh Sharma and Robert Haas, reviewed by Amit Kapila.
Original trouble report from Jeff Janes.

Discussion: http://postgr.es/m/CAMkU=1y6NjKmqbJ8wLMhr=F74WzcMALYWcVFhEpm7i=mV=XsOg@mail.gmail.com

All error messages use the American English spelling of recognize,
apply to the single one not doing so to be consistent.

Author: Daniel Gustafsson <daniel@yesql.se>

They were not rendered with DSSSL, but now they show up by default using
XSLT.  Just remove them, since they are not useful.

The size of the data is not the same thing as the size of the size of
the data.

Reported off-list by Tushar Ahuja.  Fix by Ashutosh Sharma, reviewed
by Amit Kapila.

Discussion: http://postgr.es/m/CAE9k0PnmPDXfvf8HDObme7q_Ewc4E26ukHXUBPySoOs0ObqqaQ@mail.gmail.com

This improves code coverage and lays a foundation for testing
similar issues in a distributed environment.

Author: Thomas Munro <thomas.munro@enterprisedb.com>
Reviewed-by: Michael Paquier <michael.paquier@gmail.com>

suggested by Tom Lane

Author: Daniel Gustafsson <daniel@yesql.se>

Author: Tomas Vondra <tomas.vondra@2ndquadrant.com>
Reviewed-by: Ashutosh Sharma <ashu.coek88@gmail.com>

Commit 59702716 added transition table support to PL/pgsql so that
SQL queries in trigger functions could access those transient
tables.  In order to provide the same level of support for PL/perl,
PL/python and PL/tcl, refactor the relevant code into a new
function SPI_register_trigger_data.  Call the new function in the
trigger handler of all four PLs, and document it as a public SPI
function so that authors of out-of-tree PLs can do the same.

Also get rid of a second QueryEnvironment object that was
maintained by PL/pgsql.  That was previously used to deal with
cursors, but the same approach wasn't appropriate for PLs that are
less tangled up with core code.  Instead, have SPI_cursor_open
install the connection's current QueryEnvironment, as already
happens for SPI_execute_plan.

While in the docs, remove the note that transition tables were only
supported in C and PL/pgSQL triggers, and correct some ommissions.

Thomas Munro with some work by Kevin Grittner (mostly docs)

Previously they were defined using multiples of XLogSegSize.
Remove GUC_UNIT_XSEGS. Introduce GUC_UNIT_MB

Extracted from patch series on XLogSegSize infrastructure.

Beena Emerson

This was missed in a924c327 ff.

During allocation VALGRIND_MAKE_MEM_DEFINED was called with a pointer
as size. That kind of works, but makes valgrind exceedingly slow for
workloads involving the slab allocator.

Secondly there was an access to memory marked as unreachable within
SlabCheck(). Fix that too.

Author: Tomas Vondra
Discussion: https://postgr.es/m/a6543b6d-6015-99b1-63ef-3ed55a76a730@2ndquadrant.com

Per gripe from Sven R. Kunze

2PC state info held in shmem at PREPARE, then cleaned at COMMIT PREPARED/ABORT PREPARED,
avoiding writing/fsyncing any state information to disk in the normal path, greatly enhancing replay speed.
Prepared transactions that live past one checkpoint redo horizon will be written to disk as now.
Similar conceptually to 978b2f65 and building upon
the infrastructure created by that commit.

Authors, in equal measure: Stas Kelvich, Nikhil Sontakke and Michael Paquier
Discussion: https://postgr.es/m/CAMGcDxf8Bn9ZPBBJZba9wiyQq-Qk5uqq=VjoMnRnW5s+fKST3w@mail.gmail.com

When changing the type of a sequence, adjust the min/max values of the
sequence if it looks like the previous values were the default values.
Previously, it would leave the old values in place, requiring manual
adjustments even in the usual/default cases.
Reviewed-by: Michael Paquier <michael.paquier@gmail.com>
Reviewed-by: Vitaly Burovoy <vitaly.burovoy@gmail.com>