In commit 71450d7f, we added code to inform
suitably-intelligent compilers that ereport() doesn't return if the elevel
is ERROR or higher.  This patch extends that to elog(), and also fixes a
double-evaluation hazard that the previous commit created in ereport(),
as well as reducing the emitted code size.

The elog() improvement requires the compiler to support __VA_ARGS__, which
should be available in just about anything nowadays since it's required by
C99.  But our minimum language baseline is still C89, so add a configure
test for that.

The previous commit assumed that ereport's elevel could be evaluated twice,
which isn't terribly safe --- there are already counterexamples in xlog.c.
On compilers that have __builtin_constant_p, we can use that to protect the
second test, since there's no possible optimization gain if the compiler
doesn't know the value of elevel.  Otherwise, use a local variable inside
the macros to prevent double evaluation.  The local-variable solution is
inferior because (a) it leads to useless code being emitted when elevel
isn't constant, and (b) it increases the optimization level needed for the
compiler to recognize that subsequent code is unreachable.  But it seems
better than not teaching non-gcc compilers about unreachability at all.

Lastly, if the compiler has __builtin_unreachable(), we can use that
instead of abort(), resulting in a noticeable code savings since no
function call is actually emitted.  However, it seems wise to do this only
in non-assert builds.  In an assert build, continue to use abort(), so that
the behavior will be predictable and debuggable if the "impossible"
happens.

These changes involve making the ereport and elog macros emit do-while
statement blocks not just expressions, which forces small changes in
a few call sites.

Andres Freund, Tom Lane, Heikki Linnakangas

I notice that plperl's makefile adds the -I for $perl_archlibexp/CORE
at the end of CPPFLAGS not the beginning.  It seems somewhat unlikely
that the include search order has anything to do with why buildfarm
member okapi is failing, but I'm about out of other ideas.

It appears that perl_embed_ldflags should already mention all the libraries
that are required by libperl.so itself.  So let's try the test link with
just those and not the other LIBS we've found up to now.  This should
more nearly reproduce what will happen when plperl is linked, and perhaps
will fix buildfarm member okapi's problem.

Although most platforms seem to package Perl in such a way that these files
are present even in basic Perl installations, Debian does not.  Hence, make
an effort to fail during configure rather than build if --with-perl was
given and these files are lacking.  Per gripe from Josh Berkus.

Fully update git head, and update back branches in ./COPYRIGHT and
legal.sgml files.

Some versions of libedit expose bogus definitions of setproctitle(),
optreset, and perhaps other symbols that we don't want configure to pick up
on.  There was a previous report of similar problems with strlcpy(), which
we addressed in commit 59cf88da, but the
problem has evidently grown in scope since then.  In hopes of not having to
deal with it again in future, rearrange configure's tests for supplied
functions so that we ignore libedit/libreadline except when probing
specifically for functions we expect them to provide.

Per report from Christoph Berg, though this is slightly more aggressive
than his proposed patch.

It creates too many warnings with GCC 4.3 and 4.4.

Get rid of the fundamentally indefensible assumption that "long long int"
exists and is exactly 64 bits wide on every platform Postgres runs on.
Instead let the configure script select the type to use for "pg_int64".

This is a bit of a pain in the rear since we do not want to pollute client
namespace with all the random symbols that pg_config.h defines; instead
we have to create a separate generated header file, "pg_config_ext.h".
But now that the infrastructure is there, we might have the ability to
add some other stuff that's long been wanting in this area.

Currently, the macros only work with fairly recent gcc versions, but there
is room to expand them to other compilers that have comparable features.

Heavily revised and autoconfiscated version of a patch by Andres Freund.

We previously supposed that any given platform would supply both or neither
of these functions, so that one configure test would be sufficient.  It now
appears that at least on AIX this is not the case ... which is likely an
AIX bug, but nonetheless we need to cope with it.  So use separate tests.
Per bug #6758; thanks to Andrew Hastie for doing the followup testing
needed to confirm what was happening.

Backpatch to 9.1, where we began using these functions.

port.  The last use was QNX, per Peter Eisentraut.

We had put a test for libxml2's xmlStructuredErrorContext variable in
configure, but of course that doesn't work on Windows builds.  The next
best alternative seems to be to test the LIBXML_VERSION symbol provided
by xmlversion.h.

Per report from Talha Bin Rizwan, though this fixes it in a different way
than his proposed patch.

Historically we have not worried about fsync'ing anything during initdb
(in fact, initdb intentionally passes -F to each backend launch to prevent
it from fsync'ing).  But with filesystems getting more aggressive about
caching data, that's not such a good plan anymore.  Make initdb do a pass
over the finished data directory tree to fsync everything.  For testing
purposes, the -N/--nosync flag can be used to restore the old behavior.

Also, testing shows that on Linux, sync_file_range() is much faster than
posix_fadvise() for hinting to the kernel that an fsync is coming,
apparently because the latter blocks on a rather small request queue while
the former doesn't.  So use this function if available in initdb, and also
in the backend's pg_flush_data() (where it currently will affect only the
speed of CREATE DATABASE's cloning step).

We will later make pg_regress invoke initdb with the --nosync flag
to avoid slowing down cases such as "make check" in contrib.  But
let's not do so until we've shaken out any portability issues in this
patch.

Jeff Davis, reviewed by Andres Freund

All Unix-oid platforms that we currently support should have waitpid(),
since it's in V2 of the Single Unix Spec.  Our git history shows that
the wait3 code was added to support NextStep, which we officially dropped
support for as of 9.2.  So get rid of the configure test, and simplify the
macro spaghetti in reaper().  Per suggestion from Fujii Masao.

The $(or) make function was introduced in GNU make 3.81, so the
previous coding didn't work in 3.80.  Write it differently, and
improve the variable naming to make more sense in the new coding.

configure handles INSTALL as a substitution variable specially, and
apparently it gets confused when it's set to empty.  Use INSTALL_
instead as a workaround to avoid the issue.

In a3176dac we switched to using
install-sh unconditionally, because the configure check
AC_PROG_INSTALL would pick up any random program named install, which
has caused failure reports
(http://archives.postgresql.org/pgsql-hackers/2001-03/msg00312.php).
Now the configure check is much improved and should avoid false
positives.  It has also been shown that using a system install program
can significantly reduce "make install" times, so it's worth trying.

Let the hacking begin ...

ENABLE_DTRACE            unused as of a7b7b07a
HAVE_ERR_SET_MARK        unused as of 4ed4b6c5
HAVE_FCVT                unused as of 4553e1d8
HAVE_STRUCT_SOCKADDR_UN  unused as of b4cea00a
HAVE_SYSCONF             unused as of f83356c7
TM_IN_SYS_TIME           never used, obsolescent per Autoconf documentation

These should have been removed when the BeOS port was removed in
44f90212.

These were apparently never used.  The AC_SUBST was probably just
added in a copy-and-paste manner.  (The shell variables continue to be
used inside configure.  The change is just that we don't need them
outside of configure.)

These should not be needed anymore, at least after the recent port
removals.  So let's see whether we can do without them.

Postgres 9.2, and perhaps no existing users either.

Remove the following ports:

- dgux
- nextstep
- sunos4
- svr4
- ultrix4
- univel

These are obsolete and not worth rescuing.  In most cases, there is
circumstantial evidence that they wouldn't work anymore anyway.

In the Fedora variant of MinGW, the openssl libraries have their normal
names, not libeay32 and libssleay32.  Adjust configure probes to allow
that, per bug #6486.

Tomasz Ostrowski

The immediate impetus for this is that Noah Misch's patch to elide
unnecessary table and index rebuilds when changing typmod for temporal
types uses it; and this is extracted from that patch, with some
further commentary by me.  But it seems logically separate from the
remainder of the patch, so I'm committing it separately; this is not
the first time someone has wanted fls() in the backend and probably
won't be the last.

If we end up using this in more performance-critical spots it may be
worthwhile to add some architecture-specific optimizations to our
src/port version of fls() - e.g. any x86 platform can implement this
using the assembly instruction BSRL.  But performance won't matter
a bit for assessing typmod changes, so I'm not worried about that
right now.

Historically we've used the SWPB instruction for TAS() on ARM, but this
is deprecated and not available on ARMv6 and later.  Instead, make use
of a GCC builtin if available.  We'll still fall back to SWPB if not,
so as not to break existing ports using older GCC versions.

Eventually we might want to try using __sync_lock_test_and_set() on some
other architectures too, but for now that seems to present only risk and
not reward.

Back-patch to all supported versions, since people might want to use any
of them on more recent ARM chips.

Martin Pitt

The hint bit makes for a small but measurable performance improvement
in access to contended spinlocks.

On the other hand, some PPC chips give an illegal-instruction failure.
There doesn't seem to be a completely bulletproof way to tell whether the
hint bit will cause an illegal-instruction failure other than by trying
it; but most if not all 64-bit PPC machines should accept it, so follow
the Linux kernel's lead and assume it's okay to use it in 64-bit builds.
Of course we must also check whether the assembler accepts the command,
since even with a recent CPU the toolchain could be old.

Patch by Manabu Ori, significantly modified by me.

All supported platforms support the C89 standard function atexit()
(SunOS 4 probably being the last one not to), and supporting both
makes the code clumsy.

Suggested solution from Tom Lane. Problem discovered, probably not
for the first time, while testing the mingw-w64 32 bit compiler.

Backpatched to all live branches.

Original patch by Lars Kanis, reviewed by Nishiyama Tomoaki and tweaked some by me.

This compiler, or at least the latest version of it, is currently broken, and
only passes the regression tests if built with -O0.

--- we were not using the scandir pattern filtering anyway.  This also
removes the scandir requirement in configure.

Add __attribute__ decorations for printf format checking to the places that
were missing them.  Fix the resulting warnings.  Add
-Wmissing-format-attribute to the standard set of warnings for GCC, so these
don't happen again.

The warning fixes here are relatively harmless.  The one serious problem
discovered by this was already committed earlier in
cf15fb5c.

KaiGai Kohei

on Windows. ecpglib doesn't link with libpgport, but picks and compiles
the .c files it needs individually. To cope with that, move the setlocale()
wrapper from chklocale.c to a separate setlocale.c file, and include that
in ecpglib.

Without this, it's not very easy to tell which compiler version a buildfarm
animal is actually using at the moment.