Seems a bit silly that many (in fact all, as of today) uses of
StaticAssertExpr would need to cast it to void to avoid warnings from
pickier compilers. Let's just do the cast right in the macro, instead.
In passing, change StaticAssertExpr to StaticAssertStmt in one
place where that seems more apropos.
Discussion: https://postgr.es/m/16161.1518715186@sss.pgh.pa.us
All of these are false positives, but in each case a fair amount of
analysis is needed to see that, and it's not too surprising that not all
compilers are smart enough. (In particular, in the logtape.c case, a
compiler lacking the knowledge provided by the Assert would almost surely
complain, so that this warning will be seen in any non-assert build.)
Some of these are of long standing while others are pretty recent,
but it only seems worth fixing them in HEAD.
Jaime Casanova, tweaked a bit by me
Discussion: https://postgr.es/m/CAJGNTeMcYAMJdPAom52dppLMtF-UnEZi0dooj==75OEv1EoBZA@mail.gmail.com
Formerly, DTYPE_REC was used only for variables declared as "record";
variables of named composite types used DTYPE_ROW, which is faster for
some purposes but much less flexible. In particular, the ROW code paths
are entirely incapable of dealing with DDL-caused changes to the number
or data types of the columns of a row variable, once a particular plpgsql
function has been parsed for the first time in a session. And, since the
stored representation of a ROW isn't a tuple, there wasn't any easy way
to deal with variables of domain-over-composite types, since the domain
constraint checking code would expect the value to be checked to be a
tuple. A lesser, but still real, annoyance is that ROW format cannot
represent a true NULL composite value, only a row of per-field NULL
values, which is not exactly the same thing.
Hence, switch to using DTYPE_REC for all composite-typed variables,
whether "record", named composite type, or domain over named composite
type. DTYPE_ROW remains but is used only for its native purpose, to
represent a fixed-at-compile-time list of variables, for instance the
targets of an INTO clause.
To accomplish this without taking significant performance losses, introduce
infrastructure that allows storing composite-type variables as "expanded
objects", similar to the "expanded array" infrastructure introduced in
commit 1dc5ebc90. A composite variable's value is thereby kept (most of
the time) in the form of separate Datums, so that field accesses and
updates are not much more expensive than they were in the ROW format.
This holds the line, more or less, on performance of variables of named
composite types in field-access-intensive microbenchmarks, and makes
variables declared "record" perform much better than before in similar
tests. In addition, the logic involved with enforcing composite-domain
constraints against updates of individual fields is in the expanded
record infrastructure not plpgsql proper, so that it might be reusable
for other purposes.
In further support of this, introduce a typcache feature for assigning a
unique-within-process identifier to each distinct tuple descriptor of
interest; in particular, DDL alterations on composite types result in a new
identifier for that type. This allows very cheap detection of the need to
refresh tupdesc-dependent data. This improves on the "tupDescSeqNo" idea
I had in commit 687f096ea: that assigned identifying sequence numbers to
successive versions of individual composite types, but the numbers were not
unique across different types, nor was there support for assigning numbers
to registered record types.
In passing, allow plpgsql functions to accept as well as return type
"record". There was no good reason for the old restriction, and it
was out of step with most of the other PLs.
Tom Lane, reviewed by Pavel Stehule
Discussion: https://postgr.es/m/8962.1514399547@sss.pgh.pa.us
The modern way is to use a missing_ok argument instead of two separate
almost-identical routines, so do that.
Author: Michaël Paquier
Reviewed-by: Álvaro Herrera
Discussion: https://postgr.es/m/20180201063212.GE6398@paquier.xyz
The previous code failed to realize that this setting effectively
disables parallelism, and would crash if it decided to attempt
parallelism anyway. Instead, treat it as a disabling condition.
Kyotaro Horiguchi, who also reported the issue. Reviewed by Michael
Paquier and Peter Geoghegan.
Discussion: http://postgr.es/m/20180209.170635.256350357.horiguchi.kyotaro@lab.ntt.co.jp
Prematurely freeing the EState used to evaluate CALL arguments led, in some
cases, to passing dangling pointers to the procedure. This was masked in
trivial cases because the argument pointers would point to Const nodes in
the original expression tree, and in some other cases because the result
value would end up in the standalone ExprContext rather than in memory
belonging to the EState --- but that wasn't exactly high quality
programming either, because the standalone ExprContext was never
explicitly freed, breaking assorted API contracts.
In addition, using a separate EState for each argument was just silly.
So let's use just one EState, and one ExprContext, and make the latter
belong to the former rather than be standalone, and clean up the EState
(and hence the ExprContext) post-call.
While at it, improve the function's commentary a bit.
Discussion: https://postgr.es/m/29173.1518282748@sss.pgh.pa.us
CALL statements cannot support sub-SELECTs in the arguments of the called
procedure, since they just use ExecEvalExpr to evaluate such arguments.
Teach transformSubLink() to reject the case, as it already does for other
contexts in which subqueries are not supported.
In passing, s/EXPR_KIND_CALL/EXPR_KIND_CALL_ARGUMENT/ to make that enum
symbol line up more closely with the phrasing of the error messages it is
associated with. And fix someone's weak grasp of English grammar in the
preceding EXPR_KIND_PARTITION_EXPRESSION addition. Also update an
incorrect comment in resolve_unique_index_expr (possibly it was correct
when written, but nowadays transformExpr definitely does reject SRFs here).
Per report from Pavel Stehule --- but this resolves only one of the bugs
he mentions.
Discussion: https://postgr.es/m/CAFj8pRDxOwPPzpA8i+AQeDQFj7bhVw-dR2==rfWZ3zMGkm568Q@mail.gmail.com
Doing so causes EXPLAIN ANALYZE to show trigger statistics multiple
times. Commit 2f17844104 seems to
be to blame for this.
Amit Langote, revieed by Amit Khandekar, Etsuro Fujita, and me.
When the previously-chosen plan was non-partial, all pa_finished
flags for partial plans are now set, and pa_next_plan has not yet
been set to INVALID_SUBPLAN_INDEX, the previous code could go into
an infinite loop.
Report by Rajkumar Raghuwanshi. Patch by Amit Khandekar and me.
Review by Kyotaro Horiguchi.
Discussion: http://postgr.es/m/CAJ3gD9cf43z78qY=U=H0HvOEN341qfRO-vLpnKPSviHeWgJQ5w@mail.gmail.com
This patch adds the ability to use "RANGE offset PRECEDING/FOLLOWING"
frame boundaries in window functions. We'd punted on that back in the
original patch to add window functions, because it was not clear how to
do it in a reasonably data-type-extensible fashion. That problem is
resolved here by adding the ability for btree operator classes to provide
an "in_range" support function that defines how to add or subtract the
RANGE offset value. Factoring it this way also allows the operator class
to avoid overflow problems near the ends of the datatype's range, if it
wishes to expend effort on that. (In the committed patch, the integer
opclasses handle that issue, but it did not seem worth the trouble to
avoid overflow failures for datetime types.)
The patch includes in_range support for the integer_ops opfamily
(int2/int4/int8) as well as the standard datetime types. Support for
other numeric types has been requested, but that seems like suitable
material for a follow-on patch.
In addition, the patch adds GROUPS mode which counts the offset in
ORDER-BY peer groups rather than rows, and it adds the frame_exclusion
options specified by SQL:2011. As far as I can see, we are now fully
up to spec on window framing options.
Existing behaviors remain unchanged, except that I changed the errcode
for a couple of existing error reports to meet the SQL spec's expectation
that negative "offset" values should be reported as SQLSTATE 22013.
Internally and in relevant parts of the documentation, we now consistently
use the terminology "offset PRECEDING/FOLLOWING" rather than "value
PRECEDING/FOLLOWING", since the term "value" is confusingly vague.
Oliver Ford, reviewed and whacked around some by me
Discussion: https://postgr.es/m/CAGMVOdu9sivPAxbNN0X+q19Sfv9edEPv=HibOJhB14TJv_RCQg@mail.gmail.com
LogicalTapeFreeze() may write out its first block when it is dirty but
not full, and then immediately read the first block back in from its
BufFile as a BLCKSZ-width block. This can only occur in rare cases
where very few tuples were written out, which is currently only
possible with parallel external tuplesorts. To avoid valgrind
complaints, tell it to treat the tail of logtape.c's buffer as
defined.
Commit 9da0cc3528 exposed this problem
but did not create it. LogicalTapeFreeze() has always tended to write
out some amount of garbage bytes, but previously never wrote less than
one block of data in total, so the problem was masked.
Per buildfarm members lousyjack and skink.
Peter Geoghegan, based on a suggestion from Tom Lane and me. Some
comment revisions by me.
Up to now, useful info for writing a new btree opclass has been buried
in the backend's nbtree/README file. Let's move it into the SGML docs,
in preparation for extending it with info about "in_range" functions
in the upcoming window RANGE patch.
To do this, I chose to create a new chapter for btree indexes in Part VII
(Internals), parallel to the chapters that exist for the newer index AMs.
This is a pretty short chapter as-is. At some point somebody might care
to flesh it out with more detail about btree internals, but that is
beyond the scope of my ambition for today.
Discussion: https://postgr.es/m/23141.1517874668@sss.pgh.pa.us
The previous code assumed that we'd always succeed in creating
child-joins for a joinrel for which partition-wise join was considered,
but that's not guaranteed, at least in the case where dummy rels
are involved.
Ashutosh Bapat, with some wordsmithing by me.
Discussion: http://postgr.es/m/CAFjFpRf8=uyMYYfeTBjWDMs1tR5t--FgOe2vKZPULxxdYQ4RNw@mail.gmail.com
Failure to advance the list pointer while reading partition expressions
from a list results in invoking an input function with inappropriate data,
possibly leading to crashes or, with carefully crafted input, disclosure
of arbitrary backend memory.
Bug discovered independently by Álvaro Herrera and David Rowley.
This patch is by Álvaro but owes something to David's proposed fix.
Back-patch to v10 where the issue was introduced.
Security: CVE-2018-1052
We don't need to set up the shared space for hash join instrumentation data
if instrumentation hasn't been requested. Let's follow the example of the
similar Sort node code and save a few cycles by skipping that when we can.
This reverts commit d59ff4ab3 and instead allows us to use the safer choice
of passing noError = false to shm_toc_lookup in ExecHashInitializeWorker,
since if we reach that call there should be a TOC entry to be found.
Thomas Munro
Discussion: https://postgr.es/m/E1ehkoZ-0005uW-43%40gemulon.postgresql.org
One or another author of commit 5bcf389ec seems to have thought that
computing an offset from a NULL pointer would yield another NULL pointer.
There may possibly be architectures where that works, but common machines
don't work like that. Per a quick code review of places calling
shm_toc_lookup and not using noError = false.
Commit 445dbd82a basically missed the point of commit d46633506,
which was that we shouldn't allow shm_toc_lookup() failure to lead
to a core dump or assertion crash, because the odds of such a
failure should never be considered negligible. It's correct that
we can't expect the PARALLEL_KEY_ERROR_QUEUE TOC entry to be there
if we have no workers. But if we have no workers, we're not going
to do anything in this function with the lookup result anyway,
so let's just skip it. That lets the code use the easy-to-prove-safe
noError=false case, rather than anything requiring effort to review.
Back-patch to v10, like the previous commit.
Discussion: https://postgr.es/m/3647.1517601675@sss.pgh.pa.us
Investigation of 2d2d06b7e2 revealed that
identity values were not applied in some further cases, including
logical replication subscribers, VALUES RTEs, and ALTER TABLE ... ADD
COLUMN. To fix all that, apply the identity column expression in
build_column_default() instead of repeating the same logic at each call
site.
For ALTER TABLE ... ADD COLUMN ... IDENTITY, the previous coding
completely ignored that existing rows for the new column should have
values filled in from the identity sequence. The coding using
build_column_default() fails for this because the sequence ownership
isn't registered until after ALTER TABLE, and we can't do it before
because we don't have the column in the catalog yet. So we specially
remember in ColumnDef the sequence name that we decided on and build a
custom NextValueExpr using that.
Reviewed-by: Michael Paquier <michael.paquier@gmail.com>
To make this work, tuplesort.c and logtape.c must also support
parallelism, so this patch adds that infrastructure and then applies
it to the particular case of parallel btree index builds. Testing
to date shows that this can often be 2-3x faster than a serial
index build.
The model for deciding how many workers to use is fairly primitive
at present, but it's better than not having the feature. We can
refine it as we get more experience.
Peter Geoghegan with some help from Rushabh Lathia. While Heikki
Linnakangas is not an author of this patch, he wrote other patches
without which this feature would not have been possible, and
therefore the release notes should possibly credit him as an author
of this feature. Reviewed by Claudio Freire, Heikki Linnakangas,
Thomas Munro, Tels, Amit Kapila, me.
Discussion: http://postgr.es/m/CAM3SWZQKM=Pzc=CAHzRixKjp2eO5Q0Jg1SoFQqeXFQ647JiwqQ@mail.gmail.com
Discussion: http://postgr.es/m/CAH2-Wz=AxWqDoVvGU7dq856S4r6sJAj6DBn7VMtigkB33N5eyg@mail.gmail.com
Remove partition_bound_cmp() and partition_bound_bsearch(), whose
void * argument could be, depending on the situation, of any of
three different types: PartitionBoundSpec *, PartitionRangeBound *,
Datum *.
Instead, introduce separate bound-searching functions for each
situation: partition_list_bsearch, partition_range_bsearch,
partition_range_datum_bsearch, and partition_hash_bsearch. This
requires duplicating the code for binary search, but it makes the
code much more type safe, involves fewer branches at runtime, and
at least in my opinion, is much easier to understand.
Along the way, add an option to partition_range_datum_bsearch
allowing the number of keys to be specified, so that we can search
for partitions based on a prefix of the full list of partition
keys. This is important for pending work to improve partition
pruning.
Amit Langote, per a suggestion from me.
Discussion: http://postgr.es/m/CA+TgmoaVLDLc8=YESRwD32gPhodU_ELmXyKs77gveiYp+JE4vQ@mail.gmail.com
Once this function has been called, we know that all workers have
started and attached to their error queues -- so if any of them
subsequently exit uncleanly, we'll be sure to throw an ERROR promptly.
Otherwise, users of the ParallelContext machinery must be careful not
to wait forever for a worker that has failed to start. Parallel query
manages to work without needing this for reasons explained in new
comments added by this patch, but it's a useful primitive for other
parallel operations, such as the pending patch to make creating a
btree index run in parallel.
Amit Kapila, revised by me. Additional review by Peter Geoghegan.
Discussion: http://postgr.es/m/CAA4eK1+e2MzyouF5bg=OtyhDSX+=Ao=3htN=T-r_6s3gCtKFiw@mail.gmail.com
The old code generated always generated a constraint of the form
col = ANY(ARRAY[val1, val2, ...]), but that's invalid when col is an
array type. Instead, generate col = val when there's only one value,
col = val1 OR col = val2 OR ... when there are multiple values and
col is of array type, and the old form when there are multiple values
and col is not of an array type.
As a side benefit, this makes constraint exclusion able to prune
a list partition declared to accept a single Boolean value, which
didn't work before.
Amit Langote, reviewed by Etsuro Fujita
Discussion: http://postgr.es/m/97267195-e235-89d1-a41a-c110198dfce9@lab.ntt.co.jp
pg_hba_file_rules erroneously reported this as scram-sha256. Fix that.
To avoid future errors and confusion, also adjust documentation links
and internal symbols to have a separator between "sha" and "256".
Reported-by: Christophe Courtois <christophe.courtois@dalibo.com>
Author: Michael Paquier <michael.paquier@gmail.com>
It's a common task to evaluate a qual and reset the corresponding
expression context. Currently that requires storing the result of the
qual eval, resetting the context, and then reacting on the result. As
that's awkward several places only reset the context next time through
a node. That's not great, so introduce a helper that evaluates and
resets.
It's a bit ugly that it currently uses MemoryContextReset() instead of
ResetExprContext(), but that seems easier than reordering all of
executor.h.
Author: Andres Freund
Discussion: https://postgr.es/m/20180109222544.f7loxrunqh3xjl5f@alap3.anarazel.de
There's never any value in giving a fully specified cache key to
SearchCatCacheList: you might as well call SearchCatCache instead,
since there could be only one match. So the maximum useful number of
key arguments is one less than the supported number of key columns.
We might as well remove the useless extra argument and save some few
bytes per call site, as well as a cycle or so per call.
I believe the reason it was coded like this is that originally, callers
had to write out all the dummy arguments in each call, and so it seemed
less confusing if SearchCatCache and SearchCatCacheList took the same
number of key arguments. But since commit e26c539e9, callers only write
their live arguments explicitly, making that a non-factor; and there's
surely been enough time for third-party modules to adapt to that coding
style. So this is only an ABI break not an API break for callers.
Per discussion with Oliver Ford, this might also make it less confusing
how to use SearchCatCacheList correctly.
Discussion: https://postgr.es/m/27788.1517069693@sss.pgh.pa.us
ExecPushExprSlots didn't initialize ExprEvalStep's resvalue/resnull
steps as it didn't use them. That caused wrong valgrind warnings for
an upcoming patch, so zero-intialize.
Also zero-initialize all scratch ExprEvalStep's allocated on the
stack, to avoid issues with similar future omissions of non-critial
data.
The changes in b81b5a96f4 did not fully
address the issue, because the bit-mixing of the IV into the final
hash-key didn't prevent clustering in the input-data survive in the
output data.
This didn't cause a lot of problems because of the additional growth
conditions added d4c62a6b62. But as we
want to rein those in due to explosive growth in some edges, this
needs to be fixed.
Author: Andres Freund
Discussion: https://postgr.es/m/20171127185700.1470.20362@wrigleys.postgresql.org
Backpatch: 10, where simplehash was introduced
create_plan_recurse lacked any stack depth check. This is not per
our normal coding rules, but I'd supposed it was safe because earlier
planner processing is more complex and presumably should eat more
stack. But bug #15033 from Andrew Grossman shows this isn't true,
at least not for queries having the form of a many-thousand-way
INTERSECT stack.
Further testing showed that recurse_set_operations is also capable
of being crashed in this way, since it likewise will recurse to the
bottom of a parsetree before calling any support functions that
might themselves contain any stack checks. However, its stack
consumption is only perhaps a third of create_plan_recurse's.
It's possible that this particular problem with create_plan_recurse can
only manifest in 9.6 and later, since before that we didn't build a Path
tree for set operations. But having seen this example, I now have no
faith in the proposition that create_plan_recurse doesn't need a stack
check, so back-patch to all supported branches.
Discussion: https://postgr.es/m/20180127050845.28812.58244@wrigleys.postgresql.org
Commit 09529a70b changed nodeIndexscan.c and nodeIndexonlyscan.c to
postpone initialization of the indexscan proper until the first tuple
fetch. It overlooked the question of mark/restore behavior, which means
that if some caller attempts to mark the scan before the first tuple fetch,
you get a null pointer dereference.
The only existing user of mark/restore is nodeMergejoin.c, which (somewhat
accidentally) will never attempt to set a mark before the first inner tuple
unless the inner child node is a Material node. Hence the case can't arise
normally, so it seems sufficient to document the assumption at both ends.
However, during an EvalPlanQual recheck, ExecScanFetch doesn't call
IndexNext but just returns the jammed-in test tuple. Therefore, if we're
doing a recheck in a plan tree with a mergejoin with inner indexscan,
it's possible to reach ExecIndexMarkPos with iss_ScanDesc still null,
as reported by Guo Xiang Tan in bug #15032.
Really, when there's a test tuple supplied during an EPQ recheck, touching
the index at all is the wrong thing: rather, the behavior of mark/restore
ought to amount to saving and restoring the es_epqScanDone flag. We can
avoid finding a place to actually save the flag, for the moment, because
given the assumption that no caller will set a mark before fetching a
tuple, es_epqScanDone must always be set by the time we try to mark.
So the actual behavior change required is just to not reach the index
access if a test tuple is supplied.
The set of plan node types that need to consider this issue are those
that support EPQ test tuples (i.e., call ExecScan()) and also support
mark/restore; which is to say, IndexScan, IndexOnlyScan, and perhaps
CustomScan. It's tempting to try to fix the problem in one place by
teaching ExecMarkPos() itself about EPQ; but ExecMarkPos supports some
plan types that aren't Scans, and also it seems risky to make assumptions
about what a CustomScan wants to do here. Also, the most likely future
change here is to decide that we do need to support marks placed before
the first tuple, which would require additional work in IndexScan and
IndexOnlyScan in any case. Hence, fix the EPQ issue in nodeIndexscan.c
and nodeIndexonlyscan.c, accepting the small amount of code duplicated
thereby, and leave it to CustomScan providers to fix this bug if they
have it.
Back-patch to v10 where commit 09529a70b came in. In earlier branches,
the index_markpos() call is a waste of cycles when EPQ is active, but
no more than that, so it doesn't seem appropriate to back-patch further.
Discussion: https://postgr.es/m/20180126074932.3098.97815@wrigleys.postgresql.org
We have a lot of code in which option names, which from the user's
viewpoint are logically keywords, are passed through the grammar as plain
identifiers, and then matched to string literals during command execution.
This approach avoids making words into lexer keywords unnecessarily. Some
places matched these strings using plain strcmp, some using pg_strcasecmp.
But the latter should be unnecessary since identifiers would have been
downcased on their way through the parser. Aside from any efficiency
concerns (probably not a big factor), the lack of consistency in this area
creates a hazard of subtle bugs due to different places coming to different
conclusions about whether two option names are the same or different.
Hence, standardize on using strcmp() to match any option names that are
expected to have been fed through the parser.
This does create a user-visible behavioral change, which is that while
formerly all of these would work:
alter table foo set (fillfactor = 50);
alter table foo set (FillFactor = 50);
alter table foo set ("fillfactor" = 50);
alter table foo set ("FillFactor" = 50);
now the last case will fail because that double-quoted identifier is
different from the others. However, none of our documentation says that
you can use a quoted identifier in such contexts at all, and we should
discourage doing so since it would break if we ever decide to parse such
constructs as true lexer keywords rather than poor man's substitutes.
So this shouldn't create a significant compatibility issue for users.
Daniel Gustafsson, reviewed by Michael Paquier, small changes by me
Discussion: https://postgr.es/m/29405B24-564E-476B-98C0-677A29805B84@yesql.se
This is preparatory refactoring to prepare the way for partition-wise
aggregate, which will reuse the new subroutines for child grouping
rels. It also does not seem like a bad idea on general principle,
as the function was getting pretty long.
Jeevan Chalke. The larger patch series of which this patch is a part
was reviewed and tested by Antonin Houska, Rajkumar Raghuwanshi,
Ashutosh Bapat, David Rowley, Dilip Kumar, Konstantin Knizhnik,
Pascal Legrand, and me. Some cosmetic changes by me.
Discussion: http://postgr.es/m/CAM2+6=V64_xhstVHie0Rz=KPEQnLJMZt_e314P0jaT_oJ9MR8A@mail.gmail.com
This clause was superseded by SQL-standard syntax back in 7.3.
We've kept it around for backwards-compatibility purposes ever since;
but 15 years seems like long enough for that, especially seeing that
there are undocumented weirdnesses in how it interacts with the
SQL-standard syntax for specifying the same options.
Michael Paquier, per an observation by Daniel Gustafsson;
some small cosmetic adjustments to nearby code by me.
Discussion: https://postgr.es/m/20180115022748.GB1724@paquier.xyz
The existing "connection authorized" server log messages used OpenSSL
API calls directly, even though similar abstracted API calls exist.
Change to use the latter instead.
Change the function prototype for the functions that return the TLS
version and the cipher to return const char * directly instead of
copying into a buffer. That makes them slightly easier to use.
Add bits= to the message. psql shows that, so we might as well show the
same information on the client and server.
Reviewed-by: Daniel Gustafsson <daniel@yesql.se>
Reviewed-by: Michael Paquier <michael.paquier@gmail.com>
get_relation_info() was too optimistic about opening indexes in
partitioned tables, which would raise errors when any queries were
planned on such tables. Fix by ignoring any indexes of the partitioned
kind.
CLUSTER (and ALTER TABLE CLUSTER ON) had a similar problem. Fix by
disallowing these commands in partitioned tables.
Fallout from 8b08f7d482.
These were introduced in 4cbb646334, but
after further analysis and testing, they should not be necessary and
probably weren't the part of that commit that fixed anything.
Reviewed-by: Michael Paquier <michael.paquier@gmail.com>
Avoid compiler warnings on MSVC (which doesn't want to see both
__forceinline and inline) and ancient GCC (which doesn't have
__attribute__((always_inline))).
Don't force inline-ing when building at -O0, as the programmer is probably
hoping for exact source-to-object-line correspondence in that case.
(For the moment this only works for GCC; maybe we can extend it later.)
Make pg_attribute_always_inline be syntactically a drop-in replacement
for inline, rather than an additional wart.
And improve the comments.
Thomas Munro and Michail Nikolaev, small tweaks by me
Discussion: https://postgr.es/m/32278.1514863068@sss.pgh.pa.us
Discussion: https://postgr.es/m/CANtu0oiYp74brgntKOxgg1FK5+t8uQ05guSiFU6FYz_5KUhr6Q@mail.gmail.com
If we're inside a lateral subquery, there may be no unparameterized paths
for a particular child relation of an appendrel, in which case we *must*
be able to create similarly-parameterized paths for each other child
relation, else the planner will fail with "could not devise a query plan
for the given query". This means that there are situations where we'd
better be able to reparameterize at least one path for each child.
This calls into question the assumption in reparameterize_path() that
it can just punt if it feels like it. However, the only case that is
known broken right now is where the child is itself an appendrel so that
all its paths are AppendPaths. (I think possibly I disregarded that in
the original coding on the theory that nested appendrels would get folded
together --- but that only happens *after* reparameterize_path(), so it's
not excused from handling a child AppendPath.) Given that this code's been
like this since 9.3 when LATERAL was introduced, it seems likely we'd have
heard of other cases by now if there were a larger problem.
Per report from Elvis Pranskevichus. Back-patch to 9.3.
Discussion: https://postgr.es/m/5981018.zdth1YWmNy@hammer.magicstack.net
Since 9.6, heavyweight locking is not an abstract and unhandled
concern of the parallel machinery, but rather something to which
we have a specific approach.
Commit 28724fd90d fixed things so that
if a background worker fails to start due to fork() failure or because
it is terminated before startup succeeds, BGWH_STOPPED will be
reported. However, that only helps if the code that uses the
background worker machinery notices the change in status, and the code
in parallel.c did not.
To fix that, do two things. First, make sure that when a worker
exits, it triggers the leader to read from error queues. That way, if
a worker which has attached to an error queue exits uncleanly, the
leader is sure to throw some error, either the contents of the
ErrorResponse sent by the worker, or "lost connection to parallel
worker" if it exited without sending one. To cover the case where
the worker never starts up in the first place or exits before
attaching to the error queue, the ParallelContext now keeps track
of which workers have sent at least one message via the error
queue. A worker which sends no messages by the time the parallel
operation finishes will be checked to see whether it exited before
attaching to the error queue; if so, a new error message, "parallel
worker failed to initialize", will be reported. If not, we'll
continue to wait until it either starts up and exits cleanly, starts
up and exits uncleanly, or fails to start, and then take the
appropriate action.
Patch by me, reviewed by Amit Kapila.
Discussion: http://postgr.es/m/CA+TgmoYnBgXgdTu6wk5YPdWhmgabYc9nY_pFLq=tB=FSLYkD8Q@mail.gmail.com
Some things in be-secure-openssl.c and fe-secure-openssl.c were not
actually specific to OpenSSL but could also be used by other
implementations. In order to avoid copy-and-pasting, move some of that
code to common files.
Move the documentation of the SSL API calls are supposed to do into the
headers files, instead of keeping them in the files for the OpenSSL
implementation. That way, they don't have to be duplicated or be
inconsistent when other implementations are added.
Split the "Authentication and Security" section into two separate
sections "Authentication" and "SSL". The latter part has gotten much
longer over time, and doesn't primarily have to do with authentication.
Also, the row_security parameter was inconsistently categorized, so
clean that up while we're here.
In each of the supplied procedural languages (PL/pgSQL, PL/Perl,
PL/Python, PL/Tcl), add language-specific commit and rollback
functions/commands to control transactions in procedures in that
language. Add similar underlying functions to SPI. Some additional
cleanup so that transaction commit or abort doesn't blow away data
structures still used by the procedure call. Add execution context
tracking to CALL and DO statements so that transaction control commands
can only be issued in top-level procedure and block calls, not function
calls or other procedure or block calls.
- SPI
Add a new function SPI_connect_ext() that is like SPI_connect() but
allows passing option flags. The only option flag right now is
SPI_OPT_NONATOMIC. A nonatomic SPI connection can execute transaction
control commands, otherwise it's not allowed. This is meant to be
passed down from CALL and DO statements which themselves know in which
context they are called. A nonatomic SPI connection uses different
memory management. A normal SPI connection allocates its memory in
TopTransactionContext. For nonatomic connections we use PortalContext
instead. As the comment in SPI_connect_ext() (previously SPI_connect())
indicates, one could potentially use PortalContext in all cases, but it
seems safest to leave the existing uses alone, because this stuff is
complicated enough already.
SPI also gets new functions SPI_start_transaction(), SPI_commit(), and
SPI_rollback(), which can be used by PLs to implement their transaction
control logic.
- portalmem.c
Some adjustments were made in the code that cleans up portals at
transaction abort. The portal code could already handle a command
*committing* a transaction and continuing (e.g., VACUUM), but it was not
quite prepared for a command *aborting* a transaction and continuing.
In AtAbort_Portals(), remove the code that marks an active portal as
failed. As the comment there already predicted, this doesn't work if
the running command wants to keep running after transaction abort. And
it's actually not necessary, because pquery.c is careful to run all
portal code in a PG_TRY block and explicitly runs MarkPortalFailed() if
there is an exception. So the code in AtAbort_Portals() is never used
anyway.
In AtAbort_Portals() and AtCleanup_Portals(), we need to be careful not
to clean up active portals too much. This mirrors similar code in
PreCommit_Portals().
- PL/Perl
Gets new functions spi_commit() and spi_rollback()
- PL/pgSQL
Gets new commands COMMIT and ROLLBACK.
Update the PL/SQL porting example in the documentation to reflect that
transactions are now possible in procedures.
- PL/Python
Gets new functions plpy.commit and plpy.rollback.
- PL/Tcl
Gets new commands commit and rollback.
Reviewed-by: Andrew Dunstan <andrew.dunstan@2ndquadrant.com>
Add support for huge pages (called large pages on Windows) to the
Windows build.
This (probably) breaks compatibility with Windows versions prior to
Windows 2003 or Windows Vista.
Authors: Takayuki Tsunakawa and Thomas Munro
Reviewed by: Magnus Hagander, Amit Kapila
Apparently, Peter's compiler has faith that the switch test values here
could never not be valid values of their enums. Mine does not, and
I tend to agree with it.
When an UPDATE causes a row to no longer match the partition
constraint, try to move it to a different partition where it does
match the partition constraint. In essence, the UPDATE is split into
a DELETE from the old partition and an INSERT into the new one. This
can lead to surprising behavior in concurrency scenarios because
EvalPlanQual rechecks won't work as they normally did; the known
problems are documented. (There is a pending patch to improve the
situation further, but it needs more review.)
Amit Khandekar, reviewed and tested by Amit Langote, David Rowley,
Rajkumar Raghuwanshi, Dilip Kumar, Amul Sul, Thomas Munro, Álvaro
Herrera, Amit Kapila, and me. A few final revisions by me.
Discussion: http://postgr.es/m/CAJ3gD9do9o2ccQ7j7+tSgiE1REY65XRiMb=yJO3u3QhyP8EEPQ@mail.gmail.com
When an index column is an expression, it makes no sense to compare its
attribute numbers.
This seems to account for remaining buildfarm fallout from 8b08f7d482.
At least, it solves the issue in my local 32bit VM -- let's see what the
rest thinks.
AclObjectKind was basically just another enumeration for object types,
and we already have a preferred one for that. It's only used in
aclcheck_error. By using ObjectType instead, we can also give some more
precise error messages, for example "index" instead of "relation".
Reviewed-by: Michael Paquier <michael.paquier@gmail.com>
There used to be a lot of different *Type and *Kind symbol groups to
address objects within different commands, most of which have been
replaced by ObjectType, starting with
b256f24264. But this conversion was never
done for the ACL commands until now.
This change ends up being just a plain replacement of the types and
symbols, without any code restructuring needed, except deleting some now
redundant code.
Reviewed-by: Michael Paquier <michael.paquier@gmail.com>
Reviewed-by: Stephen Frost <sfrost@snowman.net>
When CREATE INDEX is run on a partitioned table, create catalog entries
for an index on the partitioned table (which is just a placeholder since
the table proper has no data of its own), and recurse to create actual
indexes on the existing partitions; create them in future partitions
also.
As a convenience gadget, if the new index definition matches some
existing index in partitions, these are picked up and used instead of
creating new ones. Whichever way these indexes come about, they become
attached to the index on the parent table and are dropped alongside it,
and cannot be dropped on isolation unless they are detached first.
To support pg_dump'ing these indexes, add commands
CREATE INDEX ON ONLY <table>
(which creates the index on the parent partitioned table, without
recursing) and
ALTER INDEX ATTACH PARTITION
(which is used after the indexes have been created individually on each
partition, to attach them to the parent index). These reconstruct prior
database state exactly.
Reviewed-by: (in alphabetical order) Peter Eisentraut, Robert Haas, Amit
Langote, Jesper Pedersen, Simon Riggs, David Rowley
Discussion: https://postgr.es/m/20171113170646.gzweigyrgg6pwsg4@alvherre.pgsql
For no apparent reason, this function was using a 16bit-wide inhseqno
value, rather than the correct 32 bit width which is what is stored in
the pg_inherits catalog. This becomes evident if you try to create a
table with more than 65535 parents, because this error appears:
ERROR: duplicate key value violates unique constraint «pg_inherits_relid_seqno_index»
DETAIL: Key (inhrelid, inhseqno)=(329371, 0) already exists.
Needless to say, having so many parents is an uncommon situations, which
explains why this error has never been reported despite being having
been introduced with the Postgres95 1.01 sources in commit d31084e9d111:
https://git.postgresql.org/gitweb/?p=postgresql.git;a=blob;f=src/backend/commands/creatinh.c;hb=d31084e9d111#l349
Backpatch all the way back.
David Rowley noticed this while reviewing a patch of mine.
Discussion: https://postgr.es/m/CAKJS1f8Dn7swSEhOWwzZzssW7747YB=2Hi+T7uGud40dur69-g@mail.gmail.com
node->partitioned_rels is only set in UPDATE/DELETE cases, but
ExecInitModifyTable only uses its "rel" variable in INSERT cases,
so the extra logic to find the root rel is just a waste of complexity
and cycles.
Etsuro Fujita, reviewed by Amit Langote
Discussion: https://postgr.es/m/93cf9816-2f7d-0f67-8ed2-4a4e497a6ab8@lab.ntt.co.jp
Ability to advance both physical and logical replication slots using a
new user function pg_replication_slot_advance().
For logical advance that means records are consumed as fast as possible
and changes are not given to output plugin for sending. Makes 2nd phase
(after we reached SNAPBUILD_FULL_SNAPSHOT) of replication slot creation
faster, especially when there are big transactions as the reorder buffer
does not have to deal with data changes and does not have to spill to
disk.
Author: Petr Jelinek
Reviewed-by: Simon Riggs
The creates a single function JsonEncodeDateTime which will format these
data types in an efficient and consistent manner. This will be all the
more important when we come to jsonpath so we don't have to implement yet
more code doing the same thing in two more places.
This also extends the code to handle time and timetz types which were
not previously handled specially. This requires exposing the time2tm and
timetz2tm functions.
Patch from Nikita Glukhov
If a query against an inheritance tree runs concurrently with an ALTER
TABLE that's disinheriting one of the tree members, it's possible to get
a "could not find inherited attribute" error because after obtaining lock
on the removed member, make_inh_translation_list sees that its columns
have attinhcount=0 and decides they aren't the columns it's looking for.
An ideal fix, perhaps, would avoid including such a just-removed member
table in the query at all; but there seems no way to accomplish that
without adding expensive catalog rechecks or creating a likelihood of
deadlocks. Instead, let's just drop the check on attinhcount. In this
way, a query that's included a just-disinherited child will still
succeed, which is not a completely unreasonable behavior.
This problem has existed for a long time, so back-patch to all supported
branches. Also add an isolation test verifying related behaviors.
Patch by me; the new isolation test is based on Kyotaro Horiguchi's work.
Discussion: https://postgr.es/m/20170626.174612.23936762.horiguchi.kyotaro@lab.ntt.co.jp
If we flatten a subquery whose target list contains constants or
expressions, when those output columns are used in GROUPING SET columns,
the planner was capable of doing the wrong thing by merging a pulled-up
expression into the surrounding expression during const-simplification.
Then the late processing that attempts to match subexpressions to grouping
sets would fail to match those subexpressions to grouping sets, with the
effect that they'd not go to null when expected.
To fix, wrap such subquery outputs in PlaceHolderVars, ensuring that
they preserve their separate identity throughout the planner's expression
processing. This is a bit of a band-aid, because the wrapper defeats
const-simplification even in places where it would be safe to allow.
But a nicer fix would likely be too invasive to back-patch, and the
consequences of the missed optimizations probably aren't large in most
cases.
Back-patch to 9.5 where grouping sets were introduced.
Heikki Linnakangas, with small mods and better test cases by me;
additional review by Andrew Gierth
Discussion: https://postgr.es/m/7dbdcf5c-b5a6-ef89-4958-da212fe10176@iki.fi
Add the ability to label a column's default value in the catalog header,
and implement this for pg_attribute. A new function in Catalog.pm is
used to fill in a tuple with defaults. The build process will complain
loudly if a catalog entry is incomplete,
Commit 8137f2c323 labeled variable length columns for the C preprocessor.
Expose that label to genbki.pl so we can exclude those columns from schema
macros in a general fashion. Also, format schema macro entries according
to their types.
This means slightly less code maintenance, but more importantly it's a
proving ground for mechanisms intended to be used in later commits.
While at it, I (Álvaro) couldn't resist making some changes in
genbki.pl: rename some functions to actually indicate their purpose
instead of actively misleading onlookers; and don't iterate on the whole
of pg_type to find the entry for each catalog row, using a hash instead
of an array.
Author: John Naylor, some changes by Álvaro Herrera
Discussion: https://postgr.es/m/CAJVSVGVJHwD8sfDfZW9TbCHWKf=C1YDRM-rF=2JenRU_y+VcFg@mail.gmail.com
This should have been done in commit 18ce3a4ab, which added that parameter
to ExplainOneQuery, but it was overlooked. This makes it impossible for
a user of the hook to pass the queryEnv down to ExplainOnePlan.
It's too late to change this API in v10, I suppose, but fortunately
passing NULL to ExplainOnePlan will work in nearly all interesting
cases in v10. That might not be true forever, so we'd better fix it.
Tatsuro Yamada, reviewed by Thomas Munro
Discussion: https://postgr.es/m/890e8dd9-c1c7-a422-6892-874f5eaee048@lab.ntt.co.jp
It seems incorrect to assume that the list of CkptSortItems can never
contain duplicate page numbers: concurrent activity could result in some
page getting dropped from a low-numbered buffer and later loaded into a
high-numbered buffer while BufferSync is scanning the buffer pool.
If that happened, the comparator would give self-inconsistent results,
potentially confusing qsort(). Saving one comparison step is not worth
possibly getting the sort wrong.
So far as I can tell, nothing would actually go wrong given our current
implementation of qsort(). It might get a bit slower than expected
if there were a large number of duplicates of one value, but that's
surely a probability-epsilon case. Still, the comment is wrong,
and if we ever switched to another sort implementation it might be
less forgiving.
In passing, avoid casting away const-ness of the argument pointers;
I've not seen any compiler complaints from that, but it seems likely
that some compilers would not like it.
Back-patch to 9.6 where this code came in, just in case I've underestimated
the possible consequences.
Discussion: https://postgr.es/m/18437.1515607610@sss.pgh.pa.us
PL/pgSQL "pins" internally generated (unnamed) portals so that user code
cannot close them by guessing their names. This logic is also useful in
other languages and really for any code. So move that logic into SPI.
An unnamed portal obtained through SPI_cursor_open() and related
functions is now automatically pinned, and SPI_cursor_close()
automatically unpins a portal that is pinned.
In the core distribution, this affects PL/Perl and PL/Python, preventing
users from manually closing cursors created by spi_query and
plpy.cursor, respectively. (PL/Tcl does not currently offer any cursor
functionality.)
Reviewed-by: Andrew Dunstan <andrew.dunstan@2ndquadrant.com>
The previous code gave the same error message for attempting to drop
pinned and active portals, but those are separate states, so give
separate error messages.
Previously aggregate transition and combination functions were invoked
by special case code in nodeAgg.c, evaluating input and filters
separately using the expression evaluation machinery. That turns out
to not be great for performance for several reasons:
- repeated expression evaluations have some cost
- the transition functions invocations are poorly predicted, as
commonly there are multiple aggregates in a query, resulting in the
same call-stack invoking different functions.
- filter and input computation had to be done separately
- the special case code made it hard to implement JITing of the whole
transition function invocation
Address this by building one large expression that computes input,
evaluates filters, and invokes transition functions.
This leads to moderate speedups in queries bottlenecked by aggregate
computations, and enables large speedups for similar cases once JITing
is done.
There's potential for further improvement:
- It'd be nice if we could simplify the somewhat expensive
aggstate->all_pergroups lookups.
- right now there's still an advance_transition_function invocation in
nodeAgg.c, leading to some code duplication.
Author: Andres Freund
Discussion: https://postgr.es/m/20170901064131.tazjxwus3k2w3ybh@alap3.anarazel.de
As src/backend/access/transam/README says, PageGetLSN may only be called
by processes holding either exclusive lock on buffer, or a shared lock
on buffer plus buffer header lock. Therefore any place that only holds
a shared buffer lock must use BufferGetLSNAtomic instead of PageGetLSN,
which internally obtains buffer header lock prior to reading the LSN.
A few callsites failed to comply with this rule. This was detected by
running all tests under a new (not committed) assertion that verifies
PageGetLSN locking contract. All but one of the callsites that failed
the assertion are fixed by this patch. Remaining callsites were
inspected manually and determined not to need any change.
The exception (unfixed callsite) is in TestForOldSnapshot, which only
has a Page argument, making it impossible to access the corresponding
Buffer from it. Fixing that seems a much larger patch that will have to
be done separately; and that's just as well, since it was only
introduced in 9.6 and other bugs are much older.
Some of these bugs are ancient; backpatch all the way back to 9.3.
Authors: Jacob Champion, Asim Praveen, Ashwin Agrawal
Reviewed-by: Michaël Paquier
Discussion: https://postgr.es/m/CABAq_6GXgQDVu3u12mK9O5Xt5abBZWQ0V40LZCE+oUf95XyNFg@mail.gmail.com
These are compatible with Oracle and required for the datetime template
language for jsonpath in an upcoming patch.
Nikita Glukhov and Andrew Dunstan, reviewed by Pavel Stehule.
After having gotten rid of PortalGetHeapMemory(), there seems little
reason to keep one Portal access macro around that offers no actual
abstraction and isn't consistently used anyway.
Reviewed-by: Andrew Dunstan <andrew.dunstan@2ndquadrant.com>
Reviewed-by: Alvaro Herrera <alvherre@alvh.no-ip.org>
Rename PortalMemory to TopPortalContext, to avoid confusion with
PortalContext and align naming with similar top-level memory contexts.
Rename PortalData's "heap" field to portalContext. The "heap" naming
seems quite antiquated and confusing. Also get rid of the
PortalGetHeapMemory() macro and access the field directly, which we do
for other portal fields, so this abstraction doesn't buy anything.
Reviewed-by: Andrew Dunstan <andrew.dunstan@2ndquadrant.com>
Reviewed-by: Alvaro Herrera <alvherre@alvh.no-ip.org>
The initial implementation of list_qsort(), from commit ab7271677,
re-used the ListCells of the input list while not touching the List
header. This meant that anybody who still had a pointer to the
original header would now be in possession of a corrupted list,
a problem that seems sure to bite us eventually.
One possible solution is to re-use the original List header as well,
giving the function the semantics of update-in-place. However, that
doesn't seem like a very good idea either given the way that the
function is used in the planner: create_path functions aren't normally
supposed to modify their input lists. It doesn't look like there would
be a problem today, but it's not hard to foresee a time when modifying
a list of Paths in-place could have side-effects on some other append
path.
On the whole, and in view of the likelihood that this function might
be used in other contexts in the future, it seems best to get rid of
the micro-optimization of re-using the input list cells. Just build
a new list.
Discussion: https://postgr.es/m/16912.1515449066@sss.pgh.pa.us
Commit ab7271677 introduced code that attempts to order the child
scans of a Parallel Append node in a way that will minimize execution
time, based on total cost and startup cost. However, it failed to
think hard about what to do when estimated costs are exactly equal;
a case that's particularly likely to occur when comparing on startup
cost. In such a case the ordering of the child paths would be left
to the whims of qsort, an algorithm that isn't even stable.
We can improve matters by applying the rule used elsewhere in the
planner: if total costs are equal, sort on startup cost, and
vice versa. When both cost estimates are exactly equal, rather
than letting qsort do something unpredictable, sort based on the
child paths' relids, which should typically result in sorting in
inheritance order. (The latter provision requires inventing a
qsort-style comparator for bitmapsets, but maybe we'll have use
for that for other reasons in future.)
This results in a few plan changes in the select_parallel test,
but those all look more reasonable than before, when the actual
underlying cost numbers are taken into account.
Discussion: https://postgr.es/m/4944.1515446989@sss.pgh.pa.us
The general assumption for postmaster child processes is that they
should just exit(1), reasonably promptly, if the postmaster disappears.
condition_variable.c neglected this consideration and could be left
waiting forever, if the counterpart process it is waiting for has
done the right thing and exited.
We had some discussion of adjusting the WaitEventSet API to make it
harder to make this type of mistake in future; but for the moment,
and for v10, let's make this narrow fix.
Discussion: https://postgr.es/m/20412.1515456143@sss.pgh.pa.us
replorigin_drop() misunderstood the API for condition variables: it
had ConditionVariablePrepareToSleep and ConditionVariableCancelSleep
inside its test-and-sleep loop, rather than outside the loop as
intended. The net effect is a narrow race-condition window wherein,
if the process using a replication slot releases it immediately after
replorigin_drop() releases the ReplicationOriginLock, replorigin_drop()
would get into the condition variable's wait list too late and then
wait indefinitely for a signal that won't come.
Because there's a different CV for each replication slot, we can't
just move the ConditionVariablePrepareToSleep call to above the
test-and-sleep loop. What we can do, in the wake of commit 13db3b936,
is drop the ConditionVariablePrepareToSleep call entirely. This fix
depends on that commit because (at least in principle) the slot matching
the target replication origin might move around, so that once in a blue
moon successive loop iterations might involve different CVs. We can now
cope with such a scenario, at the cost of an extra trip through the
retry loop.
(There are ways we could fix this bug without depending on that commit,
but they're all a lot more complicated than this way.)
While at it, upgrade the rather skimpy comments in this function.
Back-patch to v10 where this code came in.
Discussion: https://postgr.es/m/19947.1515455433@sss.pgh.pa.us
The original coding here insisted that callers manually cancel any prepared
sleep for one condition variable before starting a sleep on another one.
While that's not a huge burden today, it seems like a gotcha that will bite
us in future if the use of condition variables increases; anything we can
do to make the use of this API simpler and more robust is attractive.
Hence, allow these functions to automatically switch their attention to
a different CV when required. This is safe for the same reason it was OK
for commit aced5a92b to let a broadcast operation cancel any prepared CV
sleep: whenever we return to the other test-and-sleep loop, we will
automatically re-prepare that CV, paying at most an extra test of that
loop's exit condition.
Back-patch to v10 where condition variables were introduced. Ordinarily
we would probably not back-patch a change like this, but since it does not
invalidate any coding pattern that was legal before, it seems safe enough.
Furthermore, there's an open bug in replorigin_drop() for which the
simplest fix requires this. Even if we chose to fix that in some more
complicated way, the hazard would remain that we might back-patch some
other bug fix that requires this behavior.
Patch by me, reviewed by Thomas Munro.
Discussion: https://postgr.es/m/2437.1515368316@sss.pgh.pa.us
There are plans to extend the syntax for ANALYZE, so we need to break
the link between VacuumStmt and AnalyzeStmt. But apart from that, the
syntax above is undocumented and, if discovered by users, might give
the impression that the VERBOSE option for VACUUM differs from the
verbose option from ANALYZE, which it does not.
Nathan Bossart, reviewed by Michael Paquier and Masahiko Sawada
Discussion: http://postgr.es/m/D3FC73E2-9B1A-4DB4-8180-55F57D116B4E@amazon.com
Previously, although the initial state of a proclist_node is expected
to be next == prev == 0, proclist_delete_offset would reset nodes to
next == prev == INVALID_PGPROCNO when removing them from a list.
This is the same state that a node in a singleton list has, so that
it's impossible to distinguish not-in-a-list from in-a-list. Change
proclist_delete_offset to reset removed nodes to next == prev == 0,
making it possible to distinguish those cases, and then add Asserts
to the list add and delete functions that the supplied node isn't
or is in a list at entry. Also tighten assertions about the node
being in the particular list (not some other one) where it is possible
to check that in O(1) time.
In ConditionVariablePrepareToSleep, since we don't expect the process's
cvWaitLink to already be in a list, remove the more-or-less-useless
proclist_contains check; we'd rather have proclist_push_tail's new
assertion fire if that happens.
Improve various comments related to proclists, too.
Patch by me, reviewed by Thomas Munro. This isn't back-patchable, since
there could theoretically be inlined copies of proclist_delete_offset in
third-party modules. But it's only improving debuggability anyway.
Discussion: https://postgr.es/m/CAEepm=0NWKehYw7NDoUSf8juuKOPRnCyY3vuaSvhrEWsOTAa3w@mail.gmail.com
In commit 561885db0, as part of normalizing RemovePgTempFiles's error
handling, I removed its behavior of silently ignoring ENOENT failures
during directory opens. Thomas Munro points out that this is a bad idea at
the top level, because we don't create pgsql_tmp directories until needed.
Thus this coding could produce LOG messages in perfectly normal situations,
which isn't what I intended. Restore the suppression of ENOENT logging,
but only at top level --- it would still be unexpected for a nested temp
directory to disappear between seeing it in the parent directory and
opening it.
Discussion: https://postgr.es/m/CAEepm=2y06SehAkTnd5sU_eVqdv5P-=Srt1y5vYNQk6yVDVaPw@mail.gmail.com
25fff40798 introduced
default monitoring roles. Apply these corrections:
* Allow access to pg_stat_get_wal_senders()
by role pg_read_all_stats
* Correct comment in pg_stat_get_wal_receiver()
to show it is no longer superuser-only.
Author: Feike Steenbergen
Reviewed-by: Michael Paquier
Apply to HEAD, then later backpatch to 10
In the wake of commit aced5a92b, the semantics of these results are
a bit squishy: we can tell whether we signaled some other process(es),
but we do not know which ones were real waiters versus mere sentinels
for ConditionVariableBroadcast operations. It does not help much that
ConditionVariableBroadcast will attempt to pass on the signal to the
next real waiter, because (a) there might not be one, and (b) that will
only happen awhile later, anyway. So these results could overstate how
much effect the calls really had.
However, no existing caller of either function pays any attention to its
result value, so it seems reasonable to just define that as a required
property of a correct algorithm. To encourage correctness and save some
tiny number of cycles, change both functions to return void.
Patch by me, per an observation by Thomas Munro. No back-patch, since
if any third parties happen to be using these functions, they might not
appreciate an API break in a minor release.
Discussion: https://postgr.es/m/CAEepm=0NWKehYw7NDoUSf8juuKOPRnCyY3vuaSvhrEWsOTAa3w@mail.gmail.com
In the admittedly-very-unlikely case that AddWaitEventToSet fails,
ConditionVariablePrepareToSleep would error out after already having
set cv_sleep_target, which is probably bad, and after having already
set cv_wait_event_set, which is very bad. Transaction abort might or
might not clean up cv_sleep_target properly; but there is nothing
that would be aware that the WaitEventSet wasn't fully constructed,
so that all future condition variable sleeps would be broken.
We can easily guard against these hazards with slight restructuring.
Back-patch to v10 where condition_variable.c was introduced.
Discussion: https://postgr.es/m/CAEepm=0NWKehYw7NDoUSf8juuKOPRnCyY3vuaSvhrEWsOTAa3w@mail.gmail.com
The original implementation of ConditionVariableBroadcast was, per its
self-description, "the dumbest way possible". Thomas Munro found out
it was a bit too dumb. An awakened process may immediately re-queue
itself, if the specific condition it's waiting for is not yet satisfied.
If this happens before ConditionVariableBroadcast is able to see the wait
queue as empty, then ConditionVariableBroadcast will re-awaken the same
process, repeating the cycle. Given unlucky timing this back-and-forth
can repeat indefinitely; loops lasting thousands of seconds have been
seen in testing.
To fix, add our own process to the end of the wait queue to serve as a
sentinel, and exit the broadcast loop once our process is not there
anymore. There are various special considerations described in the
comments, the principal disadvantage being that wakers can no longer
be sure whether they awakened a real waiter or just a sentinel. But in
practice nobody pays attention to the result of ConditionVariableSignal
or ConditionVariableBroadcast anyway, so that problem seems hypothetical.
Back-patch to v10 where condition_variable.c was introduced.
Tom Lane and Thomas Munro
Discussion: https://postgr.es/m/CAEepm=0NWKehYw7NDoUSf8juuKOPRnCyY3vuaSvhrEWsOTAa3w@mail.gmail.com
At present, we always raise an ERROR if the partition constraint
is violated, but a pending patch for UPDATE tuple routing will
consider instead moving the tuple to the correct partition.
Refactor to make that simpler.
Amit Khandekar, reviewed by Amit Langote, David Rowley, and me.
Discussion: http://postgr.es/m/CAJ3gD9cue54GbEzfV-61nyGpijvjZgCcghvLsB0_nL8Nm8HzCA@mail.gmail.com
Logical decoding's reorderbuffer.c may spill transaction files to disk
when transactions are large. These are supposed to be removed when they
become "too old" by xid; but file removal requires the boundary LSNs of
the transaction to be known. The final_lsn is only set when we see the
commit or abort record for the transaction, but nothing sets the value
for transactions that crash, so the removal code misbehaves -- in
assertion-enabled builds, it crashes by a failed assertion.
To fix, modify the final_lsn of transactions that don't have a value
set, to the LSN of the very latest change in the transaction. This
causes the spilled files to be removed appropriately.
Author: Atsushi Torikoshi
Reviewed-by: Kyotaro HORIGUCHI, Craig Ringer, Masahiko Sawada
Discussion: https://postgr.es/m/54e4e488-186b-a056-6628-50628e4e4ebc@lab.ntt.co.jp
It seems we can't easily work around the lack of
X509_get_signature_nid(), so revert the previous attempts and just
disable the tls-server-end-point feature if we don't have it.
Generalize is_partition_attr to has_partition_attrs and make it
accessible from outside tablecmds.c. Change map_partition_varattnos
to clarify that it can be used for mapping between any two relations
in a partitioning hierarchy, not just parent -> child.
Amit Khandekar, reviewed by Amit Langote, David Rowley, and me.
Some comment changes by me.
Discussion: http://postgr.es/m/CAJ3gD9fWfxgKC+PfJZF3hkgAcNOy-LpfPxVYitDEXKHjeieWQQ@mail.gmail.com
Instead of having ExecSetupPartitionTupleRouting return multiple out
parameters, have it return a pointer to a structure containing all of
those different things. Also, provide and use a cleanup function,
ExecCleanupTupleRouting, instead of cleaning up all of the resources
allocated by ExecSetupPartitionTupleRouting individually.
Amit Khandekar, reviewed by Amit Langote, David Rowley, and me
Discussion: http://postgr.es/m/CAJ3gD9fWfxgKC+PfJZF3hkgAcNOy-LpfPxVYitDEXKHjeieWQQ@mail.gmail.com
This adds a second standard channel binding type for SCRAM. It is
mainly intended for third-party clients that cannot implement
tls-unique, for example JDBC.
Author: Michael Paquier <michael.paquier@gmail.com>
As things stand now, channel binding data is fetched from OpenSSL and
saved into the SCRAM exchange context for any SSL connection attempted
for a SCRAM authentication, resulting in data fetched but not used if no
channel binding is used or if a different channel binding type is used
than what the data is here for.
Refactor the code in such a way that binding data is fetched from the
SSL stack only when a specific channel binding is used for both the
frontend and the backend. In order to achieve that, save the libpq
connection context directly in the SCRAM exchange state, and add a
dependency to SSL in the low-level SCRAM routines.
This makes the interface in charge of initializing the SCRAM context
cleaner as all its data comes from either PGconn* (for frontend) or
Port* (for the backend).
Author: Michael Paquier <michael.paquier@gmail.com>
Some versions of Windows don't define LDAPS_PORT.
Also, Windows' ldap_sslinit() is documented to use LDAPS even if you
said secure=0 when the port number happens to be 636 or 3269. Let's
avoid using the port number to imply that you want LDAPS, so that
connection strings have the same meaning on Windows and Unix.
Author: Thomas Munro
Discussion: https://postgr.es/m/CAEepm%3D23B7GV4AUz3MYH1TKpTv030VHxD2Sn%2BLYWDv8d-qWxww%40mail.gmail.com
- Remove unnecessary #include mistakenly added in execnodes.h.
- Fix mistake in comment in choose_next_subplan_for_leader.
- Adjust row estimates in cost_append for a possibly-different
parallel divisor.
- Clamp row estimates in cost_append after operations that may
not produce integers.
Amit Kapila, with cosmetic adjustments by me.
Discussion: http://postgr.es/m/CAA4eK1+qcbeai3coPpRW=GFCzFeLUsuY4T-AKHqMjxpEGZBPQg@mail.gmail.com
TupleDescCopy needs to have the same effects as CreateTupleDescCopy in
that, since it doesn't copy constraints, it should clear the per-attribute
fields associated with them. Oversight in commit cc5f81366.
Since TupleDescCopy has already established the presumption that it
can just flat-copy the entire attribute array in one go, propagate
that approach into CreateTupleDescCopy and CreateTupleDescCopyConstr.
(I'm suspicious that this would lead to valgrind complaints if we
had any trailing padding in the struct, but we do not, and anyway
fixing that seems like a job for a separate commit.)
Add some better comments.
Thomas Munro, reviewed by Vik Fearing, some additional hacking by me
Discussion: https://postgr.es/m/CAEepm=0NvOGZ8B6GbQyQe2C_c2m3LKJ9w=8OMBaYRLgZ_Gw6Nw@mail.gmail.com
XactLockTableWait assumed that its xid argument has already added itself
to the lock table. That assumption led to another assumption that if
locking the xid has succeeded but the xid is reported as still in
progress, then the input xid must have been a subtransaction.
These assumptions hold true for the original uses of this code in
locking related to on-disk tuples, but they break down in logical
replication slot snapshot building -- in particular, when a standby
snapshot logged contains an xid that's already in ProcArray but not yet
in the lock table. This leads to assertion failures that can be
reproduced all the way back to 9.4, when logical decoding was
introduced.
To fix, change SubTransGetParent to SubTransGetTopmostTransaction which
has a slightly different API: it returns the argument Xid if there is no
parent, and it goes all the way to the top instead of moving up the
levels one by one. Also, to avoid busy-waiting, add a 1ms sleep to give
the other process time to register itself in the lock table.
For consistency, change ConditionalXactLockTableWait the same way.
Author: Petr Jelínek
Discussion: https://postgr.es/m/1B3E32D8-FCF4-40B4-AEF9-5C0E3AC57969@postgrespro.ru
Reported-by: Konstantin Knizhnik
Diagnosed-by: Stas Kelvich, Petr Jelínek
Reviewed-by: Andres Freund, Robert Haas
Correct ExecParallelHashTuplePrealloc's estimate of whether the
space_allowed limit is exceeded. Be more consistent about tuples that
are exactly HASH_CHUNK_THRESHOLD in size (they're "small", not "large").
Neither of these things explain the current buildfarm unhappiness, but
they're still bugs.
Thomas Munro, per gripe by me
Discussion: https://postgr.es/m/CAEepm=34PDuR69kfYVhmZPgMdy8pSA-MYbpesEN1SR+2oj3Y+w@mail.gmail.com
Add some infrastructure (mostly macros) to make it easier to write
typical cases for constant-expression simplification. Add simplification
processing for ArrayRef, RowExpr, and ScalarArrayOpExpr node types,
which formerly went unsimplified even if all their inputs were constants.
Also teach it to simplify FieldSelect from a composite constant.
Make use of the new infrastructure to reduce the amount of code needed
for the existing ArrayExpr and ArrayCoerceExpr cases.
One existing test case changes output as a result of the fact that
RowExpr can now be folded to a constant. All the new code is exercised
by existing test cases according to gcov, so I feel no need to add
additional tests.
Tom Lane, reviewed by Dmitry Dolgov
Discussion: https://postgr.es/m/3be3b82c-e29c-b674-2163-bf47d98817b1@iki.fi
While ldaptls=1 provides an RFC 4513 conforming way to do LDAP
authentication with TLS encryption, there was an earlier de facto
standard way to do LDAP over SSL called LDAPS. Even though it's not
enshrined in a standard, it's still widely used and sometimes required
by organizations' network policies. There seems to be no reason not to
support it when available in the client library. Therefore, add support
when using OpenLDAP 2.4+ or Windows. It can be configured with
ldapscheme=ldaps or ldapurl=ldaps://...
Add tests for both ways of requesting LDAPS and a test for the
pre-existing ldaptls=1. Modify the 001_auth.pl test for "diagnostic
messages", which was previously relying on the server rejecting
ldaptls=1.
Author: Thomas Munro
Reviewed-By: Peter Eisentraut
Discussion: https://postgr.es/m/CAEepm=1s+pA-LZUjQ-9GQz0Z4rX_eK=DFXAF1nBQ+ROPimuOYQ@mail.gmail.com
Previously aggregate transition values for hash and other forms of
aggregation (i.e. sort and no group by) were represented
differently. Hash based aggregation used a grouping set indexed array
pointing to an array of transition values, whereas other forms of
aggregation used one flattened array with the index being computed out
of grouping set and transition offsets.
That made upcoming changes hard, so represent both as grouping set
indexed array of per-group data.
As a nice side-effect this also makes aggregation slightly faster,
because computing offsets with `transno + (setno * numTrans)` turns
out not to be that cheap (too big for x86 lea for example).
Author: Andres Freund
Discussion: https://postgr.es/m/20171128003121.nmxbm2ounxzb6n2t@alap3.anarazel.de
The previous coding relied (without any documentation) on the data[]
member of HashMemoryChunkData being at a MAXALIGN'ed offset. If it
was not, the tuples would not be maxaligned either, leading to failures
on alignment-picky machines. While there seems to be no live bug on any
platform we support, this is clearly pretty fragile: any addition to or
rearrangement of the fields in HashMemoryChunkData could break it.
Let's remove the hazard by getting rid of the data[] member and instead
using pointer arithmetic with an explicitly maxalign'ed offset.
Discussion: https://postgr.es/m/14483.1514938129@sss.pgh.pa.us
Multiple sessions doing CREATE INDEX CONCURRENTLY simultaneously are
supposed to be able to work in parallel, as evidenced by fixes in commit
c3d09b3bd2 specifically to support this case. In reality, one of the
sessions would be aborted by a misterious "deadlock detected" error.
Jeff Janes diagnosed that this is because of leftover snapshots used for
system catalog scans -- this was broken by 8aa3e47510 keeping track of
(registering) the catalog snapshot. To fix the deadlocks, it's enough
to de-register that snapshot prior to waiting.
Backpatch to 9.4, which introduced MVCC catalog scans.
Include an isolationtester spec that 8 out of 10 times reproduces the
deadlock with the unpatched code for me (Álvaro).
Author: Jeff Janes
Diagnosed-by: Jeff Janes
Reported-by: Jeremy Finzel
Discussion: https://postgr.es/m/CAMa1XUhHjCv8Qkx0WOr1Mpm_R4qxN26EibwCrj0Oor2YBUFUTg%40mail.gmail.com
The original idea was that we could use an isNull-style bool array
directly as a GinNullCategory array. However, the existing code already
acknowledges that that doesn't really work, because of the possibility
that bool as currently defined can have arbitrary bit patterns for true
values. So it has to loop through the nullFlags array to set each bool
value to an acceptable value. But if we are looping through the whole
array anyway, we might as well build a proper GinNullCategory array
instead and abandon the type casting. That makes the code much safer in
case bool is ever changed to something else.
Reviewed-by: Michael Paquier <michael.paquier@gmail.com>
In a race case, EXPLAIN ANALYZE could fail to display correct nbatch
and size information. Refactor so that participants report only on
batches they worked on rather than trying to report on all of them,
and teach explain.c to consider the HashInstrumentation object from
all participants instead of picking the first one it can find. This
should fix an occasional build farm failure in the "join" regression
test.
Author: Thomas Munro
Reviewed-By: Andres Freund
Discussion: https://postgr.es/m/30219.1514428346%40sss.pgh.pa.us
This reduces code duplication a bit, but the primary benefit that it
makes JITing expression evaluation easier. When doing so we can't, as
previously done in the interpreted case, really change opcode without
recompiling. Nor dow we just carry around unnecessary branches to
avoid re-checking over and over.
As a minor side-effect this makes ExecEvalStepOp() O(log(N)) rather
than O(N).
Author: Andres Freund
Discussion: https://postgr.es/m/20170901064131.tazjxwus3k2w3ybh@alap3.anarazel.de
This is useful because it gets rid of the sole direct user of
ExecAssignResultType(). A future commit will likely make use of that
and combine creating the targetlist with the initialization of the
result slot. But it seems like good code hygiene anyway.
Author: Andres Freund
Discussion: https://postgr.es/m/20170901064131.tazjxwus3k2w3ybh@alap3.anarazel.de
When walsenders were included in pg_stat_activity, only the ones
actually streaming WAL were listed as active when they were active. In
particular, the connections sending base backups were listed as being
idle. Which means that a regular pg_basebackup would show up with one
active and one idle connection, when both were active.
This patch updates to set all walsenders to active when they are
(including those doing very fast things like IDENTIFY_SYSTEM), and then
back to idle. Details about exactly what they are doing is available in
pg_stat_replication.
Patch by me, review by Michael Paquier and David Steele.
A momentary window exists when synchronous_standby_names
changes that allows commands issued after the change to
continue to act as async until the change becomes visible.
Remove the race by using more appropriate test in syncrep.c
Author: Asim Rama Praveen and Ashwin Agrawal
Reported-by: Xin Zhang, Ashwin Agrawal, and Asim Rama Praveen
Reviewed-by: Michael Paquier, Masahiko Sawada
When a backend runs out of inner tuples to hash, it should detach from
grow_batch_barrier only after it has flushed all batches to disk and
merged counters, not before. Otherwise a concurrent backend in
ExecParallelHashIncreaseNumBatches() could stop waiting for this
backend and try to read tuples before they have been written. This
commit reorders those operations and should fix the assertion failures
seen occasionally on the build farm since commit
1804284042.
Author: Thomas Munro
Discussion: https://postgr.es/m/E1eRwXy-0004IK-TO%40gemulon.postgresql.org
Hash index depends on estimation of numbers of tuples and pages of relations,
incorrect value could be a reason of significantly growing of index. Vacuum
full recreates heap and reindex all indexes before renewal stats. The patch
fixes that, so indexes will see correct values.
Backpatch to v10 only because earlier versions haven't usable hash index and
growing of hash index is a single user-visible symptom.
Author: Amit Kapila
Reviewed-by: Ashutosh Sharma, me
Discussion: https://www.postgresql.org/message-id/flat/20171115232922.5tomkxnw3iq6jsg7@inml.weebeastie.net
Polygon opclass uses compress method feature of SP-GiST added earlier. For now
it's a single operator class which uses this feature. SP-GiST actually indexes
a bounding boxes of input polygons, so part of supported operations are lossy.
Opclass uses most methods of corresponding opclass over boxes of SP-GiST and
treats bounding boxes as point in 4D-space.
Bump catalog version.
Authors: Nikita Glukhov, Alexander Korotkov with minor editorization by me
Reviewed-By: all authors + Darafei Praliaskouski
Discussion: https://www.postgresql.org/message-id/flat/54907069.1030506@sigaev.ru
Since 9.4, we've allowed the syntax "select union select" and variants
of that. However, the planner wasn't expecting a no-column set operation
and ended up treating the set operation as if it were UNION ALL.
Turns out it's trivial to fix in v10 and later; we just need to be careful
about not generating a Sort node with no sort keys. However, since a weird
corner case like this is never going to be exercised by developers, we'd
better have thorough regression tests if we want to consider it supported.
Per report from Victor Yegorov.
Discussion: https://postgr.es/m/CAGnEbojGJrRSOgJwNGM7JSJZpVAf8xXcVPbVrGdhbVEHZ-BUMw@mail.gmail.com
currentCommandIdUsed is only used to skip redundant increments of the
command counter, and CommandCounterIncrement() is categorically denied
under parallelism anyway. Therefore, it's OK for
GetCurrentCommandId() to mark the counter value used, as long as it
happens in the leader, not a worker.
Prior to commit e9baa5e9fa, the slightly
incorrect check didn't matter, but now it does. A test case added by
commit 1804284042 uncovered the problem
by accident; it caused failures with force_parallel_mode=on/regress.
Report and review by Andres Freund. Patch by me.
Discussion: http://postgr.es/m/20171221143106.5lhtygohvmazli3x@alap3.anarazel.de
This patch does three interrelated things:
* Create a new expression execution step type EEOP_PARAM_CALLBACK
and add the infrastructure needed for add-on modules to generate that.
As discussed, the best control mechanism for that seems to be to add
another hook function to ParamListInfo, which will be called by
ExecInitExpr if it's supplied and a PARAM_EXTERN Param is found.
For stand-alone expressions, we add a new entry point to allow the
ParamListInfo to be specified directly, since it can't be retrieved
from the parent plan node's EState.
* Redesign the API for the ParamListInfo paramFetch hook so that the
ParamExternData array can be entirely virtual. This also lets us get rid
of ParamListInfo.paramMask, instead leaving it to the paramFetch hook to
decide which param IDs should be accessible or not. plpgsql_param_fetch
was already doing the identical masking check, so having callers do it too
seemed redundant. While I was at it, I added a "speculative" flag to
paramFetch that the planner can specify as TRUE to avoid unwanted failures.
This solves an ancient problem for plpgsql that it couldn't provide values
of non-DTYPE_VAR variables to the planner for fear of triggering premature
"record not assigned yet" or "field not found" errors during planning.
* Rework plpgsql to get rid of the need for "unshared" parameter lists,
by dint of turning the single ParamListInfo per estate into a nearly
read-only data structure that doesn't instantiate any per-variable data.
Instead, the paramFetch hook controls access to per-variable data and can
make the right decisions on the fly, replacing the cases that we used to
need multiple ParamListInfos for. This might perhaps have been a
performance loss on its own, but by using a paramCompile hook we can
bypass plpgsql_param_fetch entirely during normal query execution.
(It's now only called when, eg, we copy the ParamListInfo into a cursor
portal. copyParamList() or SerializeParamList() effectively instantiate
the virtual parameter array as a simple physical array without a
paramFetch hook, which is what we want in those cases.) This allows
reverting most of commit 6c82d8d1f, though I kept the cosmetic
code-consolidation aspects of that (eg the assign_simple_var function).
Performance testing shows this to be at worst a break-even change,
and it can provide wins ranging up to 20% in test cases involving
accesses to fields of "record" variables. The fact that values of
such variables can now be exposed to the planner might produce wins
in some situations, too, but I've not pursued that angle.
In passing, remove the "parent" pointer from the arguments to
ExecInitExprRec and related functions, instead storing that pointer in a
transient field in ExprState. The ParamListInfo pointer for a stand-alone
expression is handled the same way; we'd otherwise have had to add
yet another recursively-passed-down argument in expression compilation.
Discussion: https://postgr.es/m/32589.1513706441@sss.pgh.pa.us
That function currently exists to avoid leaking memory in
CacheMemoryContext in case of trouble while the partition key is being
built, but there's a better way: allocate everything in a memcxt that
goes away if the current (sub)transaction fails, and once the partition
key is built and no further errors can occur, make the memcxt permanent
by making it a child of CacheMemoryContext.
Reviewed-by: Tom Lane
Discussion: https://postgr.es/m/20171027172730.eh2domlkpn4ja62m@alvherre.pgsql
Various Perl scripts we use to generate files were in the habit of
printing things like "generated by $0" into their output files.
That looks like a fine idea at first glance, but it results in
non-reproducible output, because in VPATH builds $0 won't be just
the name of the script file, but a full path for it. We'd prefer
that you get identical results whether using VPATH or not, so this
is a bad thing.
Some of these places also printed their input file name(s), causing
an additional hazard of the same type.
Hence, establish a policy that thou shalt not print $0, nor input file
pathnames, into output files (they're still allowed in error messages,
though). Instead just write the script name verbatim. While we are at
it, we can make these annotations more useful by giving the script's
full relative path name within the PG source tree, eg instead of
Gen_fmgrtab.pl let's print src/backend/utils/Gen_fmgrtab.pl.
Not all of the changes made here actually affect any files shipped
in finished tarballs today, but it seems best to apply the policy
everyplace so that nobody copies unsafe code into places where it
could matter.
Christoph Berg and Tom Lane
Discussion: https://postgr.es/m/20171215102223.GB31812@msg.df7cb.de
Generally, error recovery paths that need to do things like
LWLockReleaseAll and pgstat_report_wait_end also need to call
ConditionVariableCancelSleep, but AbortSubTransaction was missed.
Since subtransaction abort might destroy up the DSM segment that
contains the ConditionVariable stored in cv_sleep_target, this
can result in a crash for anything using condition variables.
Reported and diagnosed by Andres Freund.
Discussion: http://postgr.es/m/20171221110048.rxk6464azzl5t2fi@alap3.anarazel.de
Introduce parallel-aware hash joins that appear in EXPLAIN plans as Parallel
Hash Join with Parallel Hash. While hash joins could already appear in
parallel queries, they were previously always parallel-oblivious and had a
partial subplan only on the outer side, meaning that the work of the inner
subplan was duplicated in every worker.
After this commit, the planner will consider using a partial subplan on the
inner side too, using the Parallel Hash node to divide the work over the
available CPU cores and combine its results in shared memory. If the join
needs to be split into multiple batches in order to respect work_mem, then
workers process different batches as much as possible and then work together
on the remaining batches.
The advantages of a parallel-aware hash join over a parallel-oblivious hash
join used in a parallel query are that it:
* avoids wasting memory on duplicated hash tables
* avoids wasting disk space on duplicated batch files
* divides the work of building the hash table over the CPUs
One disadvantage is that there is some communication between the participating
CPUs which might outweigh the benefits of parallelism in the case of small
hash tables. This is avoided by the planner's existing reluctance to supply
partial plans for small scans, but it may be necessary to estimate
synchronization costs in future if that situation changes. Another is that
outer batch 0 must be written to disk if multiple batches are required.
A potential future advantage of parallel-aware hash joins is that right and
full outer joins could be supported, since there is a single set of matched
bits for each hashtable, but that is not yet implemented.
A new GUC enable_parallel_hash is defined to control the feature, defaulting
to on.
Author: Thomas Munro
Reviewed-By: Andres Freund, Robert Haas
Tested-By: Rafia Sabih, Prabhat Sahu
Discussion:
https://postgr.es/m/CAEepm=2W=cOkiZxcg6qiFQP-dHUe09aqTrEMM7yJDrHMhDv_RA@mail.gmail.comhttps://postgr.es/m/CAEepm=37HKyJ4U6XOLi=JgfSHM3o6B-GaeO-6hkOmneTDkH+Uw@mail.gmail.com
Tom Lane
Peter Eisentraut
Alvaro Herrera
Robert Haas
Andres Freund
Bruce Momjian
Magnus Hagander
Simon Riggs
Andrew Dunstan
Teodor Sigaev