Tom Lane [Thu, 30 Jan 2014 23:10:01 +0000 (18:10 -0500)]
Fix potential coredump on bad locale value in pg_upgrade.
Thinko in error report (and a typo in the message text, too). We're
failing anyway, but it would be good to print something useful first.
Noted while reviewing a patch to make pg_upgrade's locale code laxer.
Tom Lane [Thu, 30 Jan 2014 19:51:16 +0000 (14:51 -0500)]
Fix bogus handling of "postponed" lateral quals.
When pulling a "postponed" qual from a LATERAL subquery up into the quals
of an outer join, we must make sure that the postponed qual is included
in those seen by make_outerjoininfo(). Otherwise we might compute a
too-small min_lefthand or min_righthand for the outer join, leading to
"JOIN qualification cannot refer to other relations" failures from
distribute_qual_to_rels. Subtler errors in the created plan seem possible,
too, if the extra qual would only affect join ordering constraints.
Per bug #9041 from David Leverton. Back-patch to 9.3.
Bruce Momjian [Thu, 30 Jan 2014 14:41:43 +0000 (09:41 -0500)]
Add checks for interval overflow/underflow
New checks include input, month/day/time internal adjustments, addition,
subtraction, multiplication, and negation. Also adjust docs to
correctly specify interval size in bytes.
Tom Lane [Thu, 30 Jan 2014 01:03:57 +0000 (20:03 -0500)]
Fix unsafe references to errno within error messaging logic.
Various places were supposing that errno could be expected to hold still
within an ereport() nest or similar contexts. This isn't true necessarily,
though in some cases it accidentally failed to fail depending on how the
compiler chanced to order the subexpressions. This class of thinko
explains recent reports of odd failures on clang-built versions, typically
missing or inappropriate HINT fields in messages.
Problem identified by Christian Kruse, who also submitted the patch this
commit is based on. (I fixed a few issues in his patch and found a couple
of additional places with the same disease.)
Back-patch as appropriate to all supported branches.
Robert Haas [Wed, 29 Jan 2014 21:04:19 +0000 (16:04 -0500)]
Include planning time in EXPLAIN ANALYZE output.
This doesn't work for prepared queries, but it's not too easy to get
the information in that case and there's some debate as to exactly
what the right thing to measure is, so just do this for now.
When skipping over some items in a posting tree, re-find the new location
by descending the tree from root, rather than walking the right links.
This can save a lot of I/O.
Heavily modified from Alexander Korotkov's fast scan patch.
Allow skipping some items in a multi-key GIN search.
In a multi-key search, ie. something like "col @> 'foo' AND col @> 'bar'",
as soon as we find the next item that matches the first criteria, we don't
need to check the second criteria for TIDs smaller the first match. That
saves a lot of effort, especially if one of the terms is rare, while the
second occurs very frequently.
Based on ideas from Alexander Korotkov's fast scan patch.
This patch adds an option, huge_tlb_pages, which allows requesting the
shared memory segment to be allocated using huge pages, by using the
MAP_HUGETLB flag in mmap(). This can improve performance.
The default is 'try', which means that we will attempt using huge pages,
and fall back to non-huge pages if it doesn't work. Currently, only Linux
has MAP_HUGETLB. On other platforms, the default 'try' behaves the same as
'off'.
In the passing, don't try to round the mmap() size to a multiple of
pagesize. mmap() doesn't require that, and there's no particular reason for
PostgreSQL to do that either. When using MAP_HUGETLB, however, round the
request size up to nearest 2MB boundary. This is to work around a bug in
some Linux kernel versions, but also to avoid wasting memory, because the
kernel will round the size up anyway.
Many people were involved in writing this patch, including Christian Kruse,
Richard Poole, Abhijit Menon-Sen, reviewed by Peter Geoghegan, Andres Freund
and me.
Andrew Dunstan [Tue, 28 Jan 2014 22:48:21 +0000 (17:48 -0500)]
New json functions.
json_build_array() and json_build_object allow for the construction of
arbitrarily complex json trees. json_object() turns a one or two
dimensional array, or two separate arrays, into a json_object of
name/value pairs, similarly to the hstore() function.
json_object_agg() aggregates its two arguments into a single json object
as name value pairs.
Stephen Frost [Tue, 28 Jan 2014 13:40:41 +0000 (08:40 -0500)]
Revert dup2() checking in syslogger.c
Per the expanded comment-
As we're just trying to reset these to go to DEVNULL, there's not
much point in checking for failure from the close/dup2 calls here,
if they fail then presumably the file descriptors are closed and
any writes will go into the bitbucket anyway.
Tom Lane [Tue, 28 Jan 2014 02:04:09 +0000 (21:04 -0500)]
Log a detail message for auth failures due to missing or expired password.
It's worth distinguishing these cases from run-of-the-mill wrong-password
problems, since users have been known to waste lots of time pursuing the
wrong theory about what's failing. Now, our longstanding policy about how
to report authentication failures is that we don't really want to tell the
*client* such things, since that might be giving information to a bad guy.
But there's nothing wrong with reporting the details to the postmaster log,
and indeed the comments in this area of the code contemplate that
interesting details should be so reported. We just weren't handling these
particular interesting cases usefully.
To fix, add infrastructure allowing subroutines of ClientAuthentication()
to return a string to be added to the errdetail_log field of the main
authentication-failed error report. We might later want to use this to
report other subcases of authentication failure the same way, but for the
moment I just dealt with password cases.
Per discussion of a patch from Josh Drake, though this is not what
he proposed.
Tom Lane [Mon, 27 Jan 2014 20:37:54 +0000 (15:37 -0500)]
Keep pg_stat_statements' query texts in a file, not in shared memory.
This change allows us to eliminate the previous limit on stored query
length, and it makes the shared-memory hash table very much smaller,
allowing more statements to be tracked. (The default value of
pg_stat_statements.max is therefore increased from 1000 to 5000.)
In typical scenarios, the hash table can be large enough to hold all the
statements commonly issued by an application, so that there is little
"churn" in the set of tracked statements, and thus little need to do I/O
to the file.
To further reduce the need for I/O to the query-texts file, add a way
to retrieve all the columns of the pg_stat_statements view except for
the query text column. This is probably not of much interest for human
use but it could be exploited by programs, which will prefer using the
queryid anyway.
Ordinarily, we'd need to bump the extension version number for the latter
change. But since we already advanced pg_stat_statements' version number
from 1.1 to 1.2 in the 9.4 development cycle, it seems all right to just
redefine what 1.2 means.
Robert Haas [Mon, 27 Jan 2014 16:07:44 +0000 (11:07 -0500)]
Relax the requirement that all lwlocks be stored in a single array.
This makes it possible to store lwlocks as part of some other data
structure in the main shared memory segment, or in a dynamic shared
memory segment. There is still a main LWLock array and this patch does
not move anything out of it, but it provides necessary infrastructure
for doing that in the future.
This change is likely to increase the size of LWLockPadded on some
platforms, especially 32-bit platforms where it was previously only
16 bytes.
Patch by me. Review by Andres Freund and KaiGai Kohei.
Tom Lane [Mon, 27 Jan 2014 05:05:49 +0000 (00:05 -0500)]
Code review for auto-tuned effective_cache_size.
Fix integer overflow issue noted by Magnus Hagander, as well as a bunch
of other infelicities in commit ee1e5662d8d8330726eaef7d3110cb7add24d058
and its unreasonably large number of followups.
Stephen Frost [Sun, 26 Jan 2014 22:58:48 +0000 (17:58 -0500)]
Fix minor leak in pg_dump
Move allocation to after we check the remote server version, to avoid
a possible, very minor, memory leak. This makes us more consistent
throughout as most places in pg_dump are done in the same way (due, in
part, to previous fixes like this).
Stephen Frost [Sun, 26 Jan 2014 21:26:18 +0000 (16:26 -0500)]
Check dup2() results in syslogger
Consistently check the dup2() call results throughout syslogger.c.
It's pretty unlikely that they'll error out, but if they do,
ereport(FATAL) instead of blissfully continuing on.
Stephen Frost [Sun, 26 Jan 2014 14:40:34 +0000 (09:40 -0500)]
Use E, not e, for escaping in example docs
From the Department of Nitpicking, be consistent with other escaping
and use 'E' instead of 'e' to escape the string in the example docs
for GET DISAGNOSTICS stack = PG_CONTEXT.
Bruce Momjian [Sat, 25 Jan 2014 19:32:48 +0000 (14:32 -0500)]
docs: mention CREATE TABLE LIKE linkage using INCLUDING DEFAULTS
Mention that CREATE TABLE LIKE INCLUDING DEFAULTS creates a link between
the original and new tables if a default function modifies the database,
like nextval().
This allows ending recovery as a consistent state has been reached. Without
this, there was no easy way to e.g restore an online backup, without
replaying any extra WAL after the backup ended.
Noah Misch [Sat, 25 Jan 2014 00:29:06 +0000 (19:29 -0500)]
libpq: Support TLS versions beyond TLSv1.
Per report from Jeffrey Walton, libpq has been accepting only TLSv1
exactly. Along the lines of the backend code, libpq will now support
new versions as OpenSSL adds them.
Stephen Frost [Fri, 24 Jan 2014 20:10:08 +0000 (15:10 -0500)]
Avoid minor leak in parallel pg_dump
During parallel pg_dump, a worker process closing the connection caused
a minor memory leak (particularly minor as we are likely about to exit
anyway). Instead, free the memory in this case prior to returning NULL
to indicate connection closed.
The maxoff field is not used in the new, compressed page format. Let's
reset it when converting an old-format page to the new format. The code
won't care either way, but this makes it possible to use the field for
something else in the future.
In GIN recompression code, use mmemove rather than memcpy, for vacuum.
When vacuuming a data leaf page, any compressed posting lists that are not
modified, are copied back to the buffer from a later location in the same
buffer rather than from a palloc'd copy. IOW, they are just moved
downwards in the same buffer. Because the source and destination addresses
can overlap, we must use memmove rather than memcpy.
Stephen Frost [Fri, 24 Jan 2014 04:52:40 +0000 (23:52 -0500)]
ALTER TABLESPACE ... MOVE ... OWNED BY
Add the ability to specify the objects to move by who those objects are
owned by (as relowner) and change ALL to mean ALL objects. This
makes the command always operate against a well-defined set of objects
and not have the objects-to-be-moved based on the role of the user
running the command.
Tom Lane [Thu, 23 Jan 2014 22:18:23 +0000 (17:18 -0500)]
Allow use of "z" flag in our printf calls, and use it where appropriate.
Since C99, it's been standard for printf and friends to accept a "z" size
modifier, meaning "whatever size size_t has". Up to now we've generally
dealt with printing size_t values by explicitly casting them to unsigned
long and using the "l" modifier; but this is really the wrong thing on
platforms where pointers are wider than longs (such as Win64). So let's
start using "z" instead. To ensure we can do that on all platforms, teach
src/port/snprintf.c to understand "z", and add a configure test to force
use of that implementation when the platform's version doesn't handle "z".
Having done that, modify a bunch of places that were using the
unsigned-long hack to use "z" instead. This patch doesn't pretend to have
gotten everyplace that could benefit, but it catches many of them. I made
an effort in particular to ensure that all uses of the same error message
text were updated together, so as not to increase the number of
translatable strings.
It's possible that this change will result in format-string warnings from
pre-C99 compilers. We might have to reconsider if there are any popular
compilers that will warn about this; but let's start by seeing what the
buildfarm thinks.
Andres Freund, with a little additional work by me
Fix alignment of GIN in-line posting lists stored in entry tuples.
The Sparc machines in the buildfarm are crashing because of misaligned
access to posting lists stored in entry tuples.
I accidentally removed a critical SHORTALIGN() from ginFormTuple, as part
of the packed posting lists patch. Perhaps I thought it was unnecessary,
because the index_form_tuple() call above the SHORTALIGN already aligned
the size, missing the fact that the null-category byte makes it misaligned
again (I think the SHORTALIGN is indeed unnecessary if there's no null-
category byte, but let's just play it safe...)
Alvaro Herrera [Thu, 23 Jan 2014 17:40:29 +0000 (14:40 -0300)]
Make DROP IF EXISTS more consistently not fail
Some cases were still reporting errors and aborting, instead of a NOTICE
that the object was being skipped. This makes it more difficult to
cleanly handle pg_dump --clean, so change that to instead skip missing
objects properly.
Per bug #7873 reported by Dave Rolsky; apparently this affects a large
number of users.
Authors: Pavel Stehule and Dean Rasheed. Some tweaks by Álvaro Herrera
Fujii Masao [Thu, 23 Jan 2014 17:32:39 +0000 (02:32 +0900)]
Add libpq function PQhostaddr().
There was a bug in the psql's meta command \conninfo. When the
IP address was specified in the hostaddr and psql used it to create
a connection (i.e., psql -d "hostaddr=xxx"), \conninfo could not
display that address. This is because \conninfo got the connection
information only from PQhost() which could not return hostaddr.
This patch adds PQhostaddr(), and changes \conninfo so that it
can display not only the host name that PQhost() returns but also
the IP address which PQhostaddr() returns.
The bug has existed since 9.1 where \conninfo was introduced.
But it's too late to add new libpq function into the released versions,
so no backpatch.
Fujii Masao [Thu, 23 Jan 2014 13:48:12 +0000 (22:48 +0900)]
Fix bugs in PQhost().
In the platform that doesn't support Unix-domain socket, when
neither host nor hostaddr are specified, the default host
'localhost' is used to connect to the server and PQhost() must
return that, but it didn't. This patch fixes PQhost() so that
it returns the default host in that case.
Also this patch fixes PQhost() so that it doesn't return
Unix-domain socket directory path in the platform that doesn't
support Unix-domain socket.
Compress GIN posting lists, for smaller index size.
GIN posting lists are now encoded using varbyte-encoding, which allows them
to fit in much smaller space than the straight ItemPointer array format used
before. The new encoding is used for both the lists stored in-line in entry
tree items, and in posting tree leaf pages.
To maintain backwards-compatibility and keep pg_upgrade working, the code
can still read old-style pages and tuples. Posting tree leaf pages in the
new format are flagged with GIN_COMPRESSED flag, to distinguish old and new
format pages. Likewise, entry tree tuples in the new format have a
GIN_ITUP_COMPRESSED flag set in a bit that was previously unused.
This patch bumps GIN_CURRENT_VERSION from 1 to 2. New indexes created with
version 9.4 will therefore have version number 2 in the metapage, while old
pg_upgraded indexes will have version 1. The code treats them the same, but
it might be come handy in the future, if we want to drop support for the
uncompressed format.
Alexander Korotkov and me. Reviewed by Tomas Vondra and Amit Langote.
Stephen Frost [Wed, 22 Jan 2014 03:49:22 +0000 (22:49 -0500)]
Allow type_func_name_keywords in even more places
A while back, 2c92edad48796119c83d7dbe6c33425d1924626d allowed
type_func_name_keywords to be used in more places, including role
identifiers. Unfortunately, that commit missed out on cases where
name_list was used for lists-of-roles, eg: for DROP ROLE. This
resulted in the unfortunate situation that you could CREATE a role
with a type_func_name_keywords-allowed identifier, but not DROP it
(directly- ALTER could be used to rename it to something which
could be DROP'd).
This extends allowing type_func_name_keywords to places where role
lists can be used.
Tom Lane [Tue, 21 Jan 2014 21:34:28 +0000 (16:34 -0500)]
Tweak parse location assignment for CURRENT_DATE and related constructs.
All these constructs generate parse trees consisting of a Const and
a run-time type coercion (perhaps a FuncExpr or a CoerceViaIO). Modify
the raw parse output so that we end up with the original token's location
attached to the type coercion node while the Const has location -1;
before, it was the other way around. This makes no difference in terms
of what exprLocation() will say about the parse tree as a whole, so it
should not have any user-visible impact. The point of changing it is that
we do not want contrib/pg_stat_statements to treat these constructs as
replaceable constants. It will do the right thing if the Const has
location -1 rather than a valid location.
This is a pretty ugly hack, but then this code is ugly already; we should
someday replace this translation with special-purpose parse node(s) that
would allow ruleutils.c to reconstruct the original query text.
Robert Haas [Tue, 21 Jan 2014 17:38:53 +0000 (12:38 -0500)]
Add a cardinality function for arrays.
Unlike our other array functions, this considers the total number of
elements across all dimensions, and returns 0 rather than NULL when the
array has no elements. But it seems that both of those behaviors are
almost universally disliked, so hopefully that's OK.
Marko Tiikkaja, reviewed by Dean Rasheed and Pavel Stehule
Robert Haas [Tue, 21 Jan 2014 16:42:37 +0000 (11:42 -0500)]
Fix inadvertent semantics change in last patch to plug memory leaks.
Commit a5bca4ef034f71175d46462963af2329d22068c2 accidentally changed
the semantics when the "skipping missing configuration file" is
emitted, because it forced OK to true instead of leaving the value
untouched.
Tom Lane [Mon, 20 Jan 2014 18:45:51 +0000 (13:45 -0500)]
Fix to_timestamp/to_date's handling of consecutive spaces in format string.
When there are consecutive spaces (or other non-format-code characters) in
the format, we should advance over exactly that many characters of input.
The previous coding mistakenly did a "skip whitespace" action between such
characters, possibly allowing more input to be skipped than the user
intended. We only need to skip whitespace just before an actual field.
This is really a bug fix, but given the minimal number of field complaints
and the risk of breaking applications coded to expect the old behavior,
let's not back-patch it.
Simon Riggs [Mon, 20 Jan 2014 17:22:38 +0000 (17:22 +0000)]
Speed up COPY into tables with DEFAULT nextval()
Previously the presence of a nextval() prevented the
use of batch-mode COPY. This patch introduces a
special case just for nextval() functions. In future
we will introduce a general case solution for
labelling volatile functions as safe for use.
Magnus Hagander [Sun, 19 Jan 2014 16:07:15 +0000 (17:07 +0100)]
Rename msvc build option krb5 to gss
In the MSVC build system we've never separated krb5 from gss,
and always built them both. Since the removal of native krb5
support, this parameter only controls GSSAPI, so rename it
accordingly.
Magnus Hagander [Wed, 15 Jan 2014 16:24:01 +0000 (17:24 +0100)]
Remove support for native krb5 authentication
krb5 has been deprecated since 8.3, and the recommended way to do
Kerberos authentication is using the GSSAPI authentication method
(which is still fully supported).
libpq retains the ability to identify krb5 authentication, but only
gives an error message about it being unsupported. Since all authentication
is initiated from the backend, there is no need to keep it at all
in the backend.
Stephen Frost [Sun, 19 Jan 2014 01:59:31 +0000 (20:59 -0500)]
Add CREATE TABLESPACE ... WITH ... Options
Tablespaces have a few options which can be set on them to give PG hints
as to how the tablespace behaves (perhaps it's faster for sequential
scans, or better able to handle random access, etc). These options were
only available through the ALTER TABLESPACE command.
This adds the ability to set these options at CREATE TABLESPACE time,
removing the need to do both a CREATE TABLESPACE and ALTER TABLESPACE to
get the correct options set on the tablespace.
Tom Lane [Sun, 19 Jan 2014 00:24:20 +0000 (19:24 -0500)]
Fix VACUUM's reporting of dead-tuple counts to the stats collector.
Historically, VACUUM has just reported its new_rel_tuples estimate
(the same thing it puts into pg_class.reltuples) to the stats collector.
That number counts both live and dead-but-not-yet-reclaimable tuples.
This behavior may once have been right, but modern versions of the
pgstats code track live and dead tuple counts separately, so putting
the total into n_live_tuples and zero into n_dead_tuples is surely
pretty bogus. Fix it to report live and dead tuple counts separately.
This doesn't really do much for situations where updating transactions
commit concurrently with a VACUUM scan (possibly causing double-counting or
omission of the tuples they add or delete); but it's clearly an improvement
over what we were doing before.
Stephen Frost [Sat, 18 Jan 2014 23:56:40 +0000 (18:56 -0500)]
Add ALTER TABLESPACE ... MOVE command
This adds a 'MOVE' sub-command to ALTER TABLESPACE which allows moving sets of
objects from one tablespace to another. This can be extremely handy and avoids
a lot of error-prone scripting. ALTER TABLESPACE ... MOVE will only move
objects the user owns, will notify the user if no objects were found, and can
be used to move ALL objects or specific types of objects (TABLES, INDEXES, or
MATERIALIZED VIEWS).
Stephen Frost [Sat, 18 Jan 2014 23:41:52 +0000 (18:41 -0500)]
Allow SET TABLESPACE to database default
We've always allowed CREATE TABLE to create tables in the database's default
tablespace without checking for CREATE permissions on that tablespace.
Unfortunately, the original implementation of ALTER TABLE ... SET TABLESPACE
didn't pick up on that exception.
This changes ALTER TABLE ... SET TABLESPACE to allow the database's default
tablespace without checking for CREATE rights on that tablespace, just as
CREATE TABLE works today. Users could always do this through a series of
commands (CREATE TABLE ... AS SELECT * FROM ...; DROP TABLE ...; etc), so
let's fix the oversight in SET TABLESPACE's original implementation.
Tom Lane [Sat, 18 Jan 2014 21:04:11 +0000 (16:04 -0500)]
Make various variables const (read-only).
These changes should generally improve correctness/maintainability.
A nice side benefit is that several kilobytes move from initialized
data to text segment, allowing them to be shared across processes and
probably reducing copy-on-write overhead while forking a new backend.
Unfortunately this doesn't seem to help libpq in the same way (at least
not when it's compiled with -fpic on x86_64), but we can hope the linker
at least collects all nominally-const data together even if it's not
actually part of the text segment.
Also, make pg_encname_tbl[] static in encnames.c, since there seems
no very good reason for any other code to use it; per a suggestion
from Wim Lewis, who independently submitted a patch that was mostly
a subset of this one.
Oskari Saarenmaa, with some editorialization by me
Tom Lane [Fri, 17 Jan 2014 21:52:06 +0000 (16:52 -0500)]
Add gen_random_uuid() to contrib/pgcrypto.
This function provides a way of generating version 4 (pseudorandom) UUIDs
based on pgcrypto's PRNG. The main reason for doing this is that the
OSSP UUID library depended on by contrib/uuid-ossp is becoming more and
more of a porting headache, so we need an alternative for people who can't
install that. A nice side benefit though is that this implementation is
noticeably faster than uuid-ossp's uuid_generate_v4() function.
Magnus Hagander [Fri, 17 Jan 2014 11:37:57 +0000 (12:37 +0100)]
Define WIN32 when _WIN32 is set
_WIN32 is set by the compiler, whereas our code uses WIN32 that is
normally set through our build system. To make it possible to build
extensions out of tree we cannot rely on that, so set the WIN32
symbol explicitly whenever the compiler has set _WIN32.
Not setting this symbol causes double inclusion of pg_config_os.h,
and possibly other errors as well.
If --progress=2148 or higher was given, the calculation of the next time
to report overflowed, and pgbench would print a progress report very
frequently.
Fix Hot Standby feedback sending when streaming busily.
Commit 6f60fdd7015b032bf49273c99f80913d57eac284 accidentally removed a
call to XLogWalRcvSendHSFeedback() after flushing received WAL to disk.
The consequence is that when walsender is busy streaming WAL, it doesn't
send HS feedback messages. One is sent if nothing is received from the
master for 100ms, but if there's a steady stream of WAL, it never happens.