Previously, pull_varnos() took the relids of a PlaceHolderVar as being
equal to the relids in its contents, but that fails to account for the
possibility that we have to postpone evaluation of the PHV due to outer
joins. This could result in a malformed plan. The known cases end up
triggering the "failed to assign all NestLoopParams to plan nodes"
sanity check in createplan.c, but other symptoms may be possible.
The right value to use is the join level we actually intend to evaluate
the PHV at. We can get that from the ph_eval_at field of the associated
PlaceHolderInfo. However, there are some places that call pull_varnos()
before the PlaceHolderInfos have been created; in that case, fall back
to the conservative assumption that the PHV will be evaluated at its
syntactic level. (In principle this might result in missing some legal
optimization, but I'm not aware of any cases where it's an issue in
practice.) Things are also a bit ticklish for calls occurring during
deconstruct_jointree(), but AFAICS the ph_eval_at fields should have
reached their final values by the time we need them.
The main problem in making this work is that pull_varnos() has no
way to get at the PlaceHolderInfos. We can fix that easily, if a
bit tediously, in HEAD by passing it the planner "root" pointer.
In the back branches that'd cause an unacceptable API/ABI break for
extensions, so leave the existing entry points alone and add new ones
with the additional parameter. (If an old entry point is called and
encounters a PHV, it'll fall back to using the syntactic level,
again possibly missing some valid optimization.)
Back-patch to v12. The computation is surely also wrong before that,
but it appears that we cannot reach a bad plan thanks to join order
restrictions imposed on the subquery that the PlaceHolderVar came from.
The error only became reachable when commit 4be058fe9 allowed trivial
subqueries to be collapsed out completely, eliminating their join order
restrictions.
Per report from Stephan Springl.
Discussion: https://postgr.es/m/171041.1610849523@sss.pgh.pa.us
Every core SLRU wraps around. With the exception of pg_notify, the wrap
point can fall in the middle of a page. Account for this in the
PagePrecedes callback specification and in SimpleLruTruncate()'s use of
said callback. Update each callback implementation to fit the new
specification. This changes SerialPagePrecedesLogically() from the
style of asyncQueuePagePrecedes() to the style of CLOGPagePrecedes().
(Whereas pg_clog and pg_serial share a key space, pg_serial is nothing
like pg_notify.) The bug fixed here has the same symptoms and user
followup steps as 592a589a04. Back-patch
to 9.5 (all supported versions).
Reviewed by Andrey Borodin and (in earlier versions) by Tom Lane.
Discussion: https://postgr.es/m/20190202083822.GC32531@gust.leadboat.com
When a tablespace is used in a partitioned relation (per commits
ca4103025d in pg12 for tables and 33e6c34c32 in pg11 for indexes),
it is possible to drop the tablespace, potentially causing various
problems. One such was reported in bug #16577, where a rewriting ALTER
TABLE causes a server crash.
Protect against this by using pg_shdepend to keep track of tablespaces
when used for relations that don't keep physical files; we now abort a
tablespace if we see that the tablespace is referenced from any
partitioned relations.
Backpatch this to 11, where this problem has been latent all along. We
don't try to create pg_shdepend entries for existing partitioned
indexes/tables, but any ones that are modified going forward will be
protected.
Note slight behavior change: when trying to drop a tablespace that
contains both regular tables as well as partitioned ones, you'd
previously get ERRCODE_OBJECT_NOT_IN_PREREQUISITE_STATE and now you'll
get ERRCODE_DEPENDENT_OBJECTS_STILL_EXIST. Arguably, the latter is more
correct.
It is possible to add protecting pg_shdepend entries for existing
tables/indexes, by doing
ALTER TABLE ONLY some_partitioned_table SET TABLESPACE pg_default;
ALTER TABLE ONLY some_partitioned_table SET TABLESPACE original_tablespace;
for each partitioned table/index that is not in the database default
tablespace. Because these partitioned objects do not have storage, no
file needs to be actually moved, so it shouldn't take more time than
what's required to acquire locks.
This query can be used to search for such relations:
SELECT ... FROM pg_class WHERE relkind IN ('p', 'I') AND reltablespace <> 0
Reported-by: Alexander Lakhin <exclusion@gmail.com>
Discussion: https://postgr.es/m/16577-881633a9f9894fd5@postgresql.org
Author: Álvaro Herrera <alvherre@alvh.no-ip.org>
Reviewed-by: Michael Paquier <michael@paquier.xyz>
The deadlocks that the recovery conflict on lock is involved in can
happen between hot-standby backends and the startup process.
If a backend takes an access exclusive lock on the table and which
finally triggers the deadlock, that deadlock can be detected
as expected. On the other hand, previously, if the startup process
took an access exclusive lock and which finally triggered the deadlock,
that deadlock could not be detected and could remain even after
deadlock_timeout passed. This is a bug.
The cause of this bug was that the code for handling the recovery
conflict on lock didn't take care of deadlock case at all. It assumed
that deadlocks involving the startup process and backends were able
to be detected by the deadlock detector invoked within backends.
But this assumption was incorrect. The startup process also should
have invoked the deadlock detector if necessary.
To fix this bug, this commit makes the startup process invoke
the deadlock detector if deadlock_timeout is reached while handling
the recovery conflict on lock. Specifically, in that case, the startup
process requests all the backends holding the conflicting locks to
check themselves for deadlocks.
Back-patch to v9.6. v9.5 has also this bug, but per discussion we decided
not to back-patch the fix to v9.5. Because v9.5 doesn't have some
infrastructure codes (e.g., 37c54863cf) that this bug fix patch depends on.
We can apply those codes for the back-patch, but since the next minor
version release is the final one for v9.5, it's risky to do that. If we
unexpectedly introduce new bug to v9.5 by the back-patch, there is no
chance to fix that. We determined that the back-patch to v9.5 would give
more risk than gain.
Author: Fujii Masao
Reviewed-by: Bertrand Drouvot, Masahiko Sawada, Kyotaro Horiguchi
Discussion: https://postgr.es/m/4041d6b6-cf24-a120-36fa-1294220f8243@oss.nttdata.com
Unrecoverable errors detected by GSSAPI encryption can't just be
reported with elog(ERROR) or elog(FATAL), because attempting to
send the error report to the client is likely to lead to infinite
recursion or loss of protocol sync. Instead make this code do what
the SSL encryption code has long done, which is to just report any
such failure to the server log (with elevel COMMERROR), then pretend
we've lost the connection by returning errno = ECONNRESET.
Along the way, fix confusion about whether message translation is done
by pg_GSS_error() or its callers (the latter should do it), and make
the backend version of that function work more like the frontend
version.
Avoid allocating the port->gss struct until it's needed; we surely
don't need to allocate it in the postmaster.
Improve logging of "connection authorized" messages with GSS enabled.
(As part of this, I back-patched the code changes from dc11f31a1.)
Make BackendStatusShmemSize() account for the GSS-related space that
will be allocated by CreateSharedBackendStatus(). This omission
could possibly cause out-of-shared-memory problems with very high
max_connections settings.
Remove arbitrary, pointless restriction that only GSS authentication
can be used on a GSS-encrypted connection.
Improve documentation; notably, document the fact that libpq now
prefers GSS encryption over SSL encryption if both are possible.
Per report from Mikael Gustavsson. Back-patch to v12 where
this code was introduced.
Discussion: https://postgr.es/m/e5b0b6ed05764324a2f3fe7acfc766d5@smhi.se
If a database shutdown (smart or fast) is commanded between the time
some process decides to request a new background worker and the time
that the postmaster can launch that worker, then nothing happens
because the postmaster won't launch any bgworkers once it's exited
PM_RUN state. This is fine ... unless the requesting process is
waiting for that worker to finish (or even for it to start); in that
case the requestor is stuck, and only manual intervention will get us
to the point of being able to shut down.
To fix, cancel pending requests for workers when the postmaster sends
shutdown (SIGTERM) signals, and similarly cancel any new requests that
arrive after that point. (We can optimize things slightly by only
doing the cancellation for workers that have waiters.) To fit within
the existing bgworker APIs, the "cancel" is made to look like the
worker was started and immediately stopped, causing deregistration of
the bgworker entry. Waiting processes would have to deal with
premature worker exit anyway, so this should introduce no bugs that
weren't there before. We do have a side effect that registration
records for restartable bgworkers might disappear when theoretically
they should have remained in place; but since we're shutting down,
that shouldn't matter.
Back-patch to v10. There might be value in putting this into 9.6
as well, but the management of bgworkers is a bit different there
(notably see 8ff518699) and I'm not convinced it's worth the effort
to validate the patch for that branch.
Discussion: https://postgr.es/m/661570.1608673226@sss.pgh.pa.us
As it stood, expandTableLikeClause() re-did the same relation_openrv
call that transformTableLikeClause() had done. However there are
scenarios where this would not find the same table as expected.
We hold lock on the LIKE source table, so it can't be renamed or
dropped, but another table could appear before it in the search path.
This explains the odd behavior reported in bug #16758 when cloning a
table as a temp table of the same name. This case worked as expected
before commit 502898192 introduced the need to open the source table
twice, so we should fix it.
To make really sure we get the same table, let's re-open it by OID not
name. That requires adding an OID field to struct TableLikeClause,
which is a little nervous-making from an ABI standpoint, but as long
as it's at the end I don't think there's any serious risk.
Per bug #16758 from Marc Boeren. Like the previous patch,
back-patch to all supported branches.
Discussion: https://postgr.es/m/16758-840e84a6cfab276d@postgresql.org
Previously this code assumed that all IndexScan nodes supported
mark/restore, which is not true since it depends on optional index AM
support functions. This could lead to errors about missing support
functions in rare edge cases of mergejoins with no sort keys, where an
unordered non-btree index scan was placed on the inner path without a
protecting Materialize node. (Normally, the fact that merge join
requires ordered input would avoid this error.)
Backpatch all the way since this bug is ancient.
Per report from Eugen Konkov on irc.
Discussion: https://postgr.es/m/87o8jn50be.fsf@news-spur.riddles.org.uk
Introduce TimestampDifferenceMilliseconds() to simplify callers
that would rather have the difference in milliseconds, instead of
the select()-oriented seconds-and-microseconds format. This gets
rid of at least one integer division per call, and it eliminates
some apparently-easy-to-mess-up arithmetic.
Two of these call sites were in fact wrong:
* pg_prewarm's autoprewarm_main() forgot to multiply the seconds
by 1000, thus ending up with a delay 1000X shorter than intended.
That doesn't quite make it a busy-wait, but close.
* postgres_fdw's pgfdw_get_cleanup_result() thought it needed to compute
microseconds not milliseconds, thus ending up with a delay 1000X longer
than intended. Somebody along the way had noticed this problem but
misdiagnosed the cause, and imposed an ad-hoc 60-second limit rather
than fixing the units. This was relatively harmless in context, because
we don't care that much about exactly how long this delay is; still,
it's wrong.
There are a few more callers of TimestampDifference() that don't
have a direct need for seconds-and-microseconds, but can't use
TimestampDifferenceMilliseconds() either because they do need
microsecond precision or because they might possibly deal with
intervals long enough to overflow 32-bit milliseconds. It might be
worth inventing another API to improve that, but that seems outside
the scope of this patch; so those callers are untouched here.
Given the fact that we are fixing some bugs, and the likelihood
that future patches might want to back-patch code that uses this
new API, back-patch to all supported branches.
Alexey Kondratov and Tom Lane
Discussion: https://postgr.es/m/3b1c053a21c07c1ed5e00be3b2b855ef@postgrespro.ru
This back-patches commit 20d3fe900 into the v12 and v13 branches.
At the time I thought that commit was not fixing any observable
bug, but Bertrand Drouvot showed otherwise: adding a dropped column
to the previously-considered scenario crashes v12 and v13, unless the
dropped column happens to be an integer. That is, of course, because
the tupdesc we derive from the plan output tlist fails to describe
the dropped column accurately, so that we'll do the wrong thing with
a tuple in which that column isn't NULL.
There is no bug in pre-v12 branches because they already did use
the table's real tuple descriptor for any trigger-returned tuple.
It seems that this set of bugs can be blamed on the changes that
removed es_trig_tuple_slot, though I've not attempted to pin that
down precisely.
Although there's no code change needed in HEAD, update the test case
to include a dropped column there too.
Discussion: https://postgr.es/m/db5d97c8-f48a-51e2-7b08-b73d5434d425@amazon.com
Discussion: https://postgr.es/m/16644-5da7ef98a7ac4545@postgresql.org
fill_hba_line() thought it could get away with passing sizeof(struct
sockaddr_storage) rather than the actual addrlen previously returned
by getaddrinfo(). While that appears to work on many platforms,
it does not work on FreeBSD 11: you get back a failure, which leads
to the view showing NULL for the address and netmask columns in all
rows. The POSIX spec for getnameinfo() is pretty clearly on
FreeBSD's side here: you should pass the actual address length.
So it seems plausible that there are other platforms where this
coding also fails, and we just hadn't noticed.
Also, IMO the fact that getnameinfo() failure leads to a NULL output
is pretty bogus in itself. Our pg_getnameinfo_all() wrapper is
careful to emit "???" on failure, and we should use that in such
cases. NULL should only be emitted in rows that don't have IP
addresses.
Per bug #16695 from Peter Vandivier. Back-patch to v10 where this
code was added.
Discussion: https://postgr.es/m/16695-a665558e2f630be7@postgresql.org
This is useful for checks of relation pages without having to load the
pages into the shared buffers, and two cases can make use of that: page
verification in base backups and the online, lock-safe, flavor.
Compatibility is kept with past versions using a routine that calls the
new extended routine with the set of options compatible with the
original version. Contrary to d401c576, a macro cannot be used as there
may be external code relying on the presence of the original routine.
This is applied down to 11, where this will be used by a follow-up
commit addressing a set of issues with page verification in base
backups.
Extracted from a larger patch by the same author.
Author: Anastasia Lubennikova
Reviewed-by: Michael Paquier, Julien Rouhaud
Discussion: https://postgr.es/m/608f3476-0598-2514-2c03-e05c7d2b0cbd@postgrespro.ru
Backpatch-through: 11
Statistics associated to an index got lost after running REINDEX
CONCURRENTLY, while the non-concurrent case preserves these correctly.
The concurrent and non-concurrent operations need to be consistent for
the end-user, and missing statistics would force to wait for a new
analyze to happen, which could take some time depending on the activity
of the existing autovacuum workers. This issue is fixed by copying any
existing entries in pg_statistic associated to the old index to the new
one. Note that this copy is already done with the data of the index in
the stats collector.
Reported-by: Fabrízio de Royes Mello
Author: Michael Paquier, Fabrízio de Royes Mello
Reviewed-by: Justin Pryzby
Discussion: https://postgr.es/m/CAFcNs+qpFPmiHd1oTXvcPdvAHicJDA9qBUSujgAhUMJyUMb+SA@mail.gmail.com
Backpatch-through: 12
It's unsafe to do this at parse time because addition of generated
columns to a table would not invalidate stored rules containing
UPDATEs on the table ... but there might now be dependent generated
columns that were not there when the rule was made. This also fixes
an oversight that rewriteTargetView failed to update extraUpdatedCols
when transforming an UPDATE on an updatable view. (Since the new
calculation is downstream of that, rewriteTargetView doesn't actually
need to do anything; but before, there was a demonstrable bug there.)
In v13 and HEAD, this leads to easily-visible bugs because (since
commit c6679e4fc) we won't recalculate generated columns that aren't
listed in extraUpdatedCols. In v12 this bitmap is mostly just used
for trigger-firing decisions, so you'd only notice a problem if a
trigger cared whether a generated column had been updated.
I'd complained about this back in May, but then forgot about it
until bug #16671 from Michael Paul Killian revived the issue.
Back-patch to v12 where this field was introduced. If existing
stored rules contain any extraUpdatedCols values, they'll be
ignored because the rewriter will overwrite them, so the bug will
be fixed even for existing rules. (But note that if someone were
to update to 13.1 or 12.5, store some rules with UPDATEs on tables
having generated columns, and then downgrade to a prior minor version,
they might observe issues similar to what this patch fixes. That
seems unlikely enough to not be worth going to a lot of effort to fix.)
Discussion: https://postgr.es/m/10206.1588964727@sss.pgh.pa.us
Discussion: https://postgr.es/m/16671-2fa55851859fb166@postgresql.org
Unfortunately in LLVM 3.9 LLVMGetAttributeCountAtIndex(func, index)
crashes when called with an index that has 0 attributes. Since there's
no way to work around this in the C API, add a small C++ wrapper doing
so.
The only reason this didn't fail before 72559438f9 is that there
always are function attributes...
Author: Andres Freund <andres@anarazel.de>
Discussion: https://postgr.es/m/20201016001254.w2nfj7gd74jmb5in@alap3.anarazel.de
Backpatch: 11-, like 72559438f9
POSIX sigaction(2) can be told to block a set of signals while a
signal handler executes. Make use of that instead of manually
blocking and unblocking signals in the postmaster's signal handlers.
This should save a few cycles, but more importantly it prevents
recursive invocation of signal handlers when many signals arrive in
close succession. (Assuming that the platform's signal infrastructure
is designed to avoid consuming stack space in that case, but this is
demonstrably true at least on Linux.) The existing code has been seen
to recurse to the point of stack overflow, either in the postmaster
or in a forked-off child.
Back-patch of commit 9abb2bfc0. At the time, we'd only seen excess
postmaster stack consumption in the buildfarm; but we now have a
user report of it, and that commit has aged enough to have a fair
amount of confidence that it doesn't break anything.
This still doesn't change anything about the way that it works on
Windows. Perhaps someone else would like to fix that?
Per bug #16673 from David Geier. Back-patch to 9.6. Although
the problem exists in principle before that, we've only seen it
actually materialize in connection with heavy use of parallel
workers, so it doesn't seem necessary to do anything in 9.5;
and the relevant code is different there, too.
Discussion: https://postgr.es/m/16673-d278c604f8e34ec0@postgresql.org
Discussion: https://postgr.es/m/14878.1570820201@sss.pgh.pa.us
The code recorded cache invalidation events by zeroing the "localreloid"
field of affected cache entries. However, it's possible for an inval
event to occur even while we have the entry open and locked. So an
ill-timed inval could result in "cache lookup failed for relation 0"
errors, if the worker's code tried to use the cleared field. We can
fix that by creating a separate bool field to record whether the entry
needs to be revalidated. (In the back branches, cram the bool into
what had been padding space, to avoid an ABI break in the somewhat
unlikely event that any extension is looking at this struct.)
Also, rearrange the logic in logicalrep_rel_open so that it
does the right thing in cases where table_open would fail.
We should retry the lookup by name in that case, but we didn't.
The real-world impact of this is probably small. In the first place,
the error conditions are very low probability, and in the second place,
the worker would just exit and get restarted. We only noticed because
in a CLOBBER_CACHE_ALWAYS build, the failure can occur repeatedly,
preventing the worker from making progress. Nonetheless, it's clearly
a bug, and it impedes a useful type of testing; so back-patch to v10
where this code was introduced.
Discussion: https://postgr.es/m/1032727.1600096803@sss.pgh.pa.us
If a CREATE TABLE command uses both LIKE and traditional inheritance,
Vars in CHECK constraints and expression indexes that are absorbed
from a LIKE parent table tended to get mis-numbered, resulting in
wrong answers and/or bizarre error messages (though probably not any
actual crashes, thanks to validation occurring in the executor).
In v12 and up, the same could happen to Vars in GENERATED expressions,
even in cases with no LIKE clause but multiple traditional-inheritance
parents.
The cause of the problem for LIKE is that parse_utilcmd.c supposed
it could renumber such Vars correctly during transformCreateStmt(),
which it cannot since we have not yet accounted for columns added via
inheritance. Fix that by postponing processing of LIKE INCLUDING
CONSTRAINTS, DEFAULTS, GENERATED, INDEXES till after we've performed
DefineRelation().
The error with GENERATED and multiple inheritance is a simple oversight
in MergeAttributes(); it knows it has to renumber Vars in inherited
CHECK constraints, but forgot to apply the same processing to inherited
GENERATED expressions (a/k/a defaults).
Per bug #16272 from Tom Gottfried. The non-GENERATED variants of the
issue are ancient, presumably dating right back to the addition of
CREATE TABLE LIKE; hence back-patch to all supported branches.
Discussion: https://postgr.es/m/16272-6e32da020e9a9381@postgresql.org
Commit a477bfc1d fixed eval_const_expressions() to ensure that it
didn't generate unnecessary RelabelType nodes, but I failed to notice
that some other places in the planner had the same issue. Really
noplace in the planner should be using plain makeRelabelType(), for
fear of generating expressions that should be equal() to semantically
equivalent trees, but aren't.
An example is that because canonicalize_ec_expression() failed
to be careful about this, we could end up with an equivalence class
containing both a plain Const, and a Const-with-RelabelType
representing exactly the same value. So far as I can tell this led to
no visible misbehavior, but we did waste a bunch of cycles generating
and evaluating "Const = Const-with-RelabelType" to prove such entries
are redundant.
Hence, move the support function added by a477bfc1d to where it can
be more generally useful, and use it in the places where planner code
previously used makeRelabelType.
Back-patch to v12, like the previous patch. While I have no concrete
evidence of any real misbehavior here, it's certainly possible that
I overlooked a case where equivalent expressions that aren't equal()
could cause a user-visible problem. In any case carrying extra
RelabelType nodes through planning to execution isn't very desirable.
Discussion: https://postgr.es/m/1311836.1597781384@sss.pgh.pa.us
Several PGXN modules reference LockTagType values; renumbering would
force a recompile of those modules. Oversight in back-patch of today's
commit 566372b3d6. Back-patch to released
branches, v12 through 9.5.
Reported by Tom Lane.
Discussion: https://postgr.es/m/921383.1597523945@sss.pgh.pa.us
The SimpleLruTruncate() header comment states the new coding rule. To
achieve this, add locktype "frozenid" and two LWLocks. This closes a
rare opportunity for data loss, which manifested as "apparent
wraparound" or "could not access status of transaction" errors. Data
loss is more likely in pg_multixact, due to released branches' thin
margin between multiStopLimit and multiWrapLimit. If a user's physical
replication primary logged ": apparent wraparound" messages, the user
should rebuild standbys of that primary regardless of symptoms. At less
risk is a cluster having emitted "not accepting commands" errors or
"must be vacuumed" warnings at some point. One can test a cluster for
this data loss by running VACUUM FREEZE in every database. Back-patch
to 9.5 (all supported versions).
Discussion: https://postgr.es/m/20190218073103.GA1434723@rfd.leadboat.com
nodeSubplan.c expects that the testexpr for a hashable ANY SubPlan
has the form of one or more OpExprs whose LHS is an expression of the
outer query's, while the RHS is an expression over Params representing
output columns of the subquery. However, the planner only went as far
as verifying that the clauses were all binary OpExprs. This works
99.99% of the time, because the clauses have the right shape when
emitted by the parser --- but it's possible for function inlining to
break that, as reported by PegoraroF10. To fix, teach the planner
to check that the LHS and RHS contain the right things, or more
accurately don't contain the wrong things. Given that this has been
broken for years without anyone noticing, it seems sufficient to just
give up hashing when it happens, rather than go to the trouble of
commuting the clauses back again (which wouldn't necessarily work
anyway).
While poking at that, I also noticed that nodeSubplan.c had a baked-in
assumption that the number of hash clauses is identical to the number
of subquery output columns. Again, that's fine as far as parser output
goes, but it's not hard to break it via function inlining. There seems
little reason for that assumption though --- AFAICS, the only thing
it's buying us is not having to store the number of hash clauses
explicitly. Adding code to the planner to reject such cases would take
more code than getting nodeSubplan.c to cope, so I fixed it that way.
This has been broken for as long as we've had hashable SubPlans,
so back-patch to all supported branches.
Discussion: https://postgr.es/m/1549209182255-0.post@n3.nabble.com
Up to now, upon receipt of a SIGTERM ("smart shutdown" command), the
postmaster has immediately killed all "optional" background processes,
and subsequently refused to launch new ones while it's waiting for
foreground client processes to exit. No doubt this seemed like an OK
policy at some point; but it's a pretty bad one now, because it makes
for a seriously degraded environment for the remaining clients:
* Parallel queries are killed, and new ones fail to launch. (And our
parallel-query infrastructure utterly fails to deal with the case
in a reasonable way --- it just hangs waiting for workers that are
not going to arrive. There is more work needed in that area IMO.)
* Autovacuum ceases to function. We can tolerate that for awhile,
but if bulk-update queries continue to run in the surviving client
sessions, there's eventually going to be a mess. In the worst case
the system could reach a forced shutdown to prevent XID wraparound.
* The bgwriter and walwriter are also stopped immediately, likely
resulting in performance degradation.
Hence, let's rearrange things so that the only immediate change in
behavior is refusing to let in new normal connections. Once the last
normal connection is gone, shut everything down as though we'd received
a "fast" shutdown. To implement this, remove the PM_WAIT_BACKUP and
PM_WAIT_READONLY states, instead staying in PM_RUN or PM_HOT_STANDBY
while normal connections remain. A subsidiary state variable tracks
whether or not we're letting in new connections in those states.
This also allows having just one copy of the logic for killing child
processes in smart and fast shutdown modes. I moved that logic into
PostmasterStateMachine() by inventing a new state PM_STOP_BACKENDS.
Back-patch to 9.6 where parallel query was added. In principle
this'd be a good idea in 9.5 as well, but the risk/reward ratio
is not as good there, since lack of autovacuum is not a problem
during typical uses of smart shutdown.
Per report from Bharath Rupireddy.
Patch by me, reviewed by Thomas Munro
Discussion: https://postgr.es/m/CALj2ACXAZ5vKxT9P7P89D87i3MDO9bfS+_bjMHgnWJs8uwUOOw@mail.gmail.com
Any libpq client can use the header. Clients include backend components
postgres_fdw, dblink, and logical replication apply worker. Back-patch
to v10, because another fix needs this. In released branches, just copy
the header and keep the original.
The previous coding imagined that it could call before_shmem_exit()
when a non-exclusive backup began and then remove the previously-added
handler by calling cancel_before_shmem_exit() when that backup
ended. However, this only works provided that nothing else in the
system has registered a before_shmem_exit() hook in the interim,
because cancel_before_shmem_exit() is documented to remove a callback
only if it is the latest callback registered. It also only works
if nothing can ERROR out between the time that sessionBackupState
is reset and the time that cancel_before_shmem_exit(), which doesn't
seem to be strictly true.
To fix, leave the handler installed for the lifetime of the session,
arrange to install it just once, and teach it to quietly do nothing if
there isn't a non-exclusive backup in process.
This was originally committed to master as
303640199d, but I did not back-patch
at the time because the consequences were minor. However, now
there's been a second report of this causing trouble with a slightly
different test case than the one I reported originally, so now
I'm back-patching as far as v11 where JIT was introduced.
Patch by me, reviewed by Kyotaro Horiguchi, Michael Paquier (who
preferred a different approach, but got outvoted), Fujii Masao,
and Tom Lane, and with comments by various others. New problem
report from Bharath Rupireddy.
Discussion: http://postgr.es/m/CA+TgmobMjnyBfNhGTKQEDbqXYE3_rXWpc4CM63fhyerNCes3mA@mail.gmail.com
Discussion: http://postgr.es/m/CALj2ACWk7j4F2v2fxxYfrroOF=AdFNPr1WsV+AGtHAFQOqm_pw@mail.gmail.com
This coding technique is unsafe, since we'd be accessing off the end
of the tuple if the field is null. SIGSEGV is pretty improbable, but
perhaps not impossible. Also, returning garbage for the LSN doesn't
seem like a great idea, even if callers aren't looking at it today.
Also update docs to point out explicitly that
pg_subscription.subslotname and pg_subscription_rel.srsublsn
can be null.
Perhaps we should mark these two fields BKI_FORCE_NULL, so that
they'd be correctly labeled in databases that are initdb'd in the
future. But we can't force that for existing databases, and on
balance it's not too clear that having a mix of different catalog
contents in the field would be wise.
Apply to v10 (where this code came in) through v12. Already
fixed in v13 and HEAD.
Discussion: https://postgr.es/m/732838.1595278439@sss.pgh.pa.us
Incorrect function names were referenced. As this fixes some portions
of tableam.h, that is mentioned in the docs as something to look at when
implementing a table AM, backpatch down to 12 where this has been
introduced.
Author: Hironobu Suzuki
Discussion: https://postgr.es/m/8fe6d672-28dd-3f1d-7aed-ac2f6d599d3f@interdb.jp
Backpatch-through: 12
Previously we used pg_atomic_write_64_impl inside
pg_atomic_init_u64. That works correctly, but on platforms without
64bit single copy atomicity it could trigger spurious valgrind errors
about uninitialized memory, because we use compare_and_swap for atomic
writes on such platforms.
I previously suppressed one instance of this problem (6c878edc1d),
but as Tom reports that wasn't enough. As the atomic variable cannot
yet be concurrently accessible during initialization, it seems better
to have pg_atomic_init_64_impl set the value directly.
Change pg_atomic_init_u32_impl for symmetry.
Reported-By: Tom Lane
Author: Andres Freund
Discussion: https://postgr.es/m/1714601.1591503815@sss.pgh.pa.us
Backpatch: 9.5-
Commit 5e0928005 changed the planner so that, instead of blindly using
DEFAULT_COLLATION_OID when invoking operators for selectivity estimation,
it would use the collation of the column whose statistics we're
considering. This was recognized as still being not quite the right
thing, but it seemed like a good incremental improvement. However,
shortly thereafter we introduced nondeterministic collations, and that
creates cases where operators can fail if they're passed the wrong
collation. We don't want planning to fail in cases where the query itself
would work, so this means that we *must* use the query's collation when
invoking operators for estimation purposes.
The only real problem this creates is in ineq_histogram_selectivity, where
the binary search might produce a garbage answer if we perform comparisons
using a different collation than the column's histogram is ordered with.
However, when the query's collation is significantly different from the
column's default collation, the estimate we previously generated would be
pretty irrelevant anyway; so it's not clear that this will result in
noticeably worse estimates in practice. (A follow-on patch will improve
this situation in HEAD, but it seems too invasive for back-patch.)
The patch requires changing the signatures of mcv_selectivity and allied
functions, which are exported and very possibly are used by extensions.
In HEAD, I just did that, but an API/ABI break of this sort isn't
acceptable in stable branches. Therefore, in v12 the patch introduces
"mcv_selectivity_ext" and so on, with signatures matching HEAD, and makes
the old functions into wrappers that assume DEFAULT_COLLATION_OID should
be used. That does not match the prior behavior, but it should avoid risk
of failure in most cases. (In practice, I think most extension datatypes
aren't collation-aware, so the change probably doesn't matter to them.)
Per report from James Lucas. Back-patch to v12 where the problem was
introduced.
Discussion: https://postgr.es/m/CAAFmbbOvfi=wMM=3qRsPunBSLb8BFREno2oOzSBS=mzfLPKABw@mail.gmail.com
It's intentional that we don't allow values greater than 24 hours,
while we do allow "24:00:00" as well as "23:59:60" as inputs.
However, the range check was miscoded in such a way that it would
accept "23:59:60.nnn" with a nonzero fraction. For time or timetz,
the stored result would then be greater than "24:00:00" which would
fail dump/reload, not to mention possibly confusing other operations.
Fix by explicitly calculating the result and making sure it does not
exceed 24 hours. (This calculation is redundant with what will happen
later in tm2time or tm2timetz. Maybe someday somebody will find that
annoying enough to justify refactoring to avoid the duplication; but
that seems too invasive for a back-patched bug fix, and the cost is
probably unmeasurable anyway.)
Note that this change also rejects such input as the time portion
of a timestamp(tz) value.
Back-patch to v10. The bug is far older, but to change this pre-v10
we'd need to ensure that the logic behaves sanely with float timestamps,
which is possibly nontrivial due to roundoff considerations.
Doesn't really seem worth troubling with.
Per report from Christoph Berg.
Discussion: https://postgr.es/m/20200520125807.GB296739@msg.df7cb.de
Fix some more violations of the "only straight-line code inside a
spinlock" rule. These are hazardous not only because they risk
holding the lock for an excessively long time, but because it's
possible for palloc to throw elog(ERROR), leaving a stuck spinlock
behind.
copy_replication_slot() had two separate places that did pallocs
while holding a spinlock. We can make the code simpler and safer
by copying the whole ReplicationSlot struct into a local variable
while holding the spinlock, and then referencing that copy.
(While that's arguably more cycles than we really need to spend
holding the lock, the struct isn't all that big, and this way seems
far more maintainable than copying fields piecemeal. Anyway this
is surely much cheaper than a palloc.) That bug goes back to v12.
InvalidateObsoleteReplicationSlots() not only did a palloc while
holding a spinlock, but for extra sloppiness then leaked the memory
--- probably for the lifetime of the checkpointer process, though
I didn't try to verify that. Fortunately that silliness is new
in HEAD.
pg_get_replication_slots() had a cosmetic violation of the rule,
in that it only assumed it's safe to call namecpy() while holding
a spinlock. Still, that's a hazard waiting to bite somebody, and
there were some other cosmetic coding-rule violations in the same
function, so clean it up. I back-patched this as far as v10; the
code exists before that but it looks different, and this didn't
seem important enough to adapt the patch further back.
Discussion: https://postgr.es/m/20200602.161518.1399689010416646074.horikyota.ntt@gmail.com
We were acquiring object locks then deleting objects one by one, instead
of acquiring all object locks first, ignoring those that did not exist,
and then deleting all objects together. The latter is the correct
protocol to use, and what this commits changes to code to do. Failing
to follow that leads to "cache lookup failed for relation XYZ" error
reports when DROP OWNED runs concurrently with other DDL -- for example,
a session termination that removes some temp tables.
Author: Álvaro Herrera
Reported-by: Mithun Chicklore Yogendra (Mithun CY)
Reviewed-by: Ahsan Hadi, Tom Lane
Discussion: https://postgr.es/m/CADq3xVZTbzK4ZLKq+dn_vB4QafXXbmMgDP3trY-GuLnib2Ai1w@mail.gmail.com
Writing a trailing semicolon in a macro is almost never the right thing,
because you almost always want to write a semicolon after each macro
call instead. (Even if there was some reason to prefer not to, pgindent
would probably make a hash of code formatted that way; so within PG the
rule should basically be "don't do it".) Thus, if we have a semi inside
the macro, the compiler sees "something;;". Much of the time the extra
empty statement is harmless, but it could lead to mysterious syntax
errors at call sites. In perhaps an overabundance of neatnik-ism, let's
run around and get rid of the excess semicolons whereever possible.
The only thing worse than a mysterious syntax error is a mysterious
syntax error that only happens in the back branches; therefore,
backpatch these changes where relevant, which is most of them because
most of these mistakes are old. (The lack of reported problems shows
that this is largely a hypothetical issue, but still, it could bite
us in some future patch.)
John Naylor and Tom Lane
Discussion: https://postgr.es/m/CACPNZCs0qWTqJ2QUSGJ07B7uvAvzMb-KbG2q+oo+J3tsWN5cqw@mail.gmail.com
The logic for determining how many nbtree pages in an index are deleted
pages sometimes undercounted pages. Pages that were deleted by the
current VACUUM operation (as opposed to some previous VACUUM operation
whose deleted pages have yet to be reused) were sometimes overlooked.
The final count is exposed to users through VACUUM VERBOSE's "%u index
pages have been deleted" output.
btvacuumpage() avoided double-counting when _bt_pagedel() deleted more
than one page by assuming that only one page was deleted, and that the
additional deleted pages would get picked up during a future call to
btvacuumpage() by the same VACUUM operation. _bt_pagedel() can
legitimately delete pages that the btvacuumscan() scan will not visit
again, though, so that assumption was slightly faulty.
Fix the accounting by teaching _bt_pagedel() about its caller's
requirements. It now only reports on pages that it knows btvacuumscan()
won't visit again (including the current btvacuumpage() page), so
everything works out in the end.
This bug has been around forever. Only backpatch to v11, though, to
keep _bt_pagedel() is sync on the branches that have today's bugfix
commit b0229f26da. Note that this commit changes the signature of
_bt_pagedel(), just like commit b0229f26da.
Author: Peter Geoghegan
Reviewed-By: Masahiko Sawada
Discussion: https://postgr.es/m/CAH2-WzkrXBcMQWAYUJMFTTvzx_r4q=pYSjDe07JnUXhe+OZnJA@mail.gmail.com
Backpatch: 11-
Commit 857f9c36cd (which taught nbtree VACUUM to skip a scan of the
index from btcleanup in situations where it doesn't seem worth it) made
VACUUM maintain the oldest btpo.xact among all deleted pages for the
index as a whole. It failed to handle all the details surrounding pages
that are deleted by the current VACUUM operation correctly (though pages
deleted by some previous VACUUM operation were processed correctly).
The most immediate problem was that the special area of the page was
examined without a buffer pin at one point. More fundamentally, the
handling failed to account for the full range of _bt_pagedel()
behaviors. For example, _bt_pagedel() sometimes deletes internal pages
in passing, as part of deleting an entire subtree with btvacuumpage()
caller's page as the leaf level page. The original leaf page passed to
_bt_pagedel() might not be the page that it deletes first in cases where
deletion can take place.
It's unclear how disruptive this bug may have been, or what symptoms
users might want to look out for. The issue was spotted during
unrelated code review.
To fix, push down the logic for maintaining the oldest btpo.xact to
_bt_pagedel(). btvacuumpage() is now responsible for pages that were
fully deleted by a previous VACUUM operation, while _bt_pagedel() is now
responsible for pages that were deleted by the current VACUUM operation
(this includes half-dead pages from a previous interrupted VACUUM
operation that become fully deleted in _bt_pagedel()). Note that
_bt_pagedel() should never encounter an existing deleted page.
This commit theoretically breaks the ABI of a stable release by changing
the signature of _bt_pagedel(). However, if any third party extension
is actually affected by this, then it must already be completely broken
(since there are numerous assumptions made in _bt_pagedel() that cannot
be met outside of VACUUM). It seems highly unlikely that such an
extension actually exists, in any case.
Author: Peter Geoghegan
Reviewed-By: Masahiko Sawada
Discussion: https://postgr.es/m/CAH2-WzkrXBcMQWAYUJMFTTvzx_r4q=pYSjDe07JnUXhe+OZnJA@mail.gmail.com
Backpatch: 11-, where the "skip full scan" feature was introduced.
78ea8b5 has fixed an issue related to the recycling of WAL segments on
standbys depending on archive_mode. However, it has introduced a
regression with the handling of WAL segments ready to be archived during
crash recovery, causing those files to be recycled without getting
archived.
This commit fixes the regression by tracking in shared memory if a live
cluster is either in crash recovery or archive recovery as the handling
of WAL segments ready to be archived is different in both cases (those
WAL segments should not be removed during crash recovery), and by using
this new shared memory state to decide if a segment can be recycled or
not. Previously, it was not possible to know if a cluster was in crash
recovery or archive recovery as the shared state was able to track only
if recovery was happening or not, leading to the problem.
A set of TAP tests is added to close the gap here, making sure that WAL
segments ready to be archived are correctly handled when a cluster is in
archive or crash recovery with archive_mode set to "on" or "always", for
both standby and primary.
Reported-by: Benoît Lobréau
Author: Jehan-Guillaume de Rorthais
Reviewed-by: Kyotaro Horiguchi, Fujii Masao, Michael Paquier
Discussion: https://postgr.es/m/20200331172229.40ee00dc@firost
Backpatch-through: 9.5
index.c supposed that it could just use a PG_TRY block to clean up the
state associated with an active REINDEX operation. However, that code
doesn't run if we do a FATAL exit --- for example, due to a SIGTERM
shutdown signal --- while the REINDEX is happening. And that state does
get consulted during catalog accesses, which makes it problematic if we
do any catalog accesses during shutdown --- for example, to clean up any
temp tables created in the session.
If this combination of circumstances occurred, we could find ourselves
trying to access already-freed memory. In debug builds that'd fairly
reliably cause an assertion failure. In production we might often
get away with it, but with some bad luck it could cause a core dump.
Another possible bad outcome is an erroneous conclusion that an
index-to-be-accessed is being reindexed; but it looks like that would
be unlikely to have any consequences worse than failing to drop temp
tables right away. (They'd still get dropped by the next session that
uses that temp schema.)
To fix, get rid of the use of PG_TRY here, and instead hook into
the transaction abort mechanisms to clean up reindex state.
Per bug #16378 from Alexander Lakhin. This has been wrong for a
very long time, so back-patch to all supported branches.
Discussion: https://postgr.es/m/16378-7a70ca41b3ec2009@postgresql.org
Working on commit 1c455078b led me to check through FunctionCallInvoke
call sites to see if every one was being honest about (a) making sure
that fcinfo.isnull is initially false, and (b) checking its state after
the call. Sure enough, I found some violations.
The main one is that finalize_partialaggregate re-used serialfn_fcinfo
without resetting isnull, even though it clearly intends to cater for
serialfns that return NULL. There would only be an issue with a
non-strict serialfn, since it's unlikely that a serialfn would return
NULL for non-null input. We have no non-strict serialfns in core, and
there may be none in the wild either, which would account for the lack
of complaints. Still, it's clearly wrong, so back-patch that fix to
9.6 where finalize_partialaggregate was introduced.
Also, arrayfuncs.c and rowtypes.c contained various callers that were
not bothering to check for result nulls. While what's being called is
a comparison or hash function that probably *shouldn't* return null,
that's a lousy excuse for not having any check at all. There are
existing places that just Assert(!fcinfo->isnull) in comparable
situations, so I added that to the places that were calling btree
comparison or hash support functions. In the places calling
boolean-returning equality functions, it's quite cheap to have them
treat isnull as FALSE, so make those places do that. Also remove some
"locfcinfo->isnull = false" assignments that are unnecessary given the
assumption that no previous call returned null. These changes seem like
mostly neatnik-ism or debugging support, so I didn't back-patch.
We have repeatedly seen the buildfarm reach the Assert(false) in
SyncRepGetSyncStandbysPriority. This apparently is due to failing to
consider the possibility that the sync_standby_priority values in
shared memory might be inconsistent; but they will be whenever only
some of the walsenders have updated their values after a change in
the synchronous_standby_names setting. That function is vastly too
complex for what it does, anyway, so rewriting it seems better than
trying to apply a band-aid fix.
Furthermore, the API of SyncRepGetSyncStandbys is broken by design:
it returns a list of WalSnd array indexes, but there is nothing
guaranteeing that the contents of the WalSnd array remain stable.
Thus, if some walsender exits and then a new walsender process
takes over that WalSnd array slot, a caller might make use of
WAL position data that it should not, potentially leading to
incorrect decisions about whether to release transactions that
are waiting for synchronous commit.
To fix, replace SyncRepGetSyncStandbys with a new function
SyncRepGetCandidateStandbys that copies all the required data
from shared memory while holding the relevant mutexes. If the
associated walsender process then exits, this data is still safe to
make release decisions with, since we know that that much WAL *was*
sent to a valid standby server. This incidentally means that we no
longer need to treat sync_standby_priority as protected by the
SyncRepLock rather than the per-walsender mutex.
SyncRepGetSyncStandbys is no longer used by the core code, so remove
it entirely in HEAD. However, it seems possible that external code is
relying on that function, so do not remove it from the back branches.
Instead, just remove the known-incorrect Assert. When the bug occurs,
the function will return a too-short list, which callers should treat
as meaning there are not enough sync standbys, which seems like a
reasonably safe fallback until the inconsistent state is resolved.
Moreover it's bug-compatible with what has been happening in non-assert
builds. We cannot do anything about the walsender-replacement race
condition without an API/ABI break.
The bogus assertion exists back to 9.6, but 9.6 is sufficiently
different from the later branches that the patch doesn't apply at all.
I chose to just remove the bogus assertion in 9.6, feeling that the
probability of a bad outcome from the walsender-replacement race
condition is too low to justify rewriting the whole patch for 9.6.
Discussion: https://postgr.es/m/21519.1585272409@sss.pgh.pa.us
A table rewritten by ALTER TABLE would lose tracking of an index usable
for CLUSTER. This setting is tracked by pg_index.indisclustered and is
controlled by ALTER TABLE, so some extra work was needed to restore it
properly. Note that ALTER TABLE only marks the index that can be used
for clustering, and does not do the actual operation.
Author: Amit Langote, Justin Pryzby
Reviewed-by: Ibrar Ahmed, Michael Paquier
Discussion: https://postgr.es/m/20200202161718.GI13621@telsasoft.com
Backpatch-through: 9.5
Now that we require C99, we can depend on __VA_ARGS__ to work, and
revising ereport() to use it has several significant benefits:
* The extra parentheses around the auxiliary function calls are now
optional. Aside from being a bit less ugly, this removes a common
gotcha for new contributors, because in some cases the compiler errors
you got from forgetting them were unintelligible.
* The auxiliary function calls are now evaluated as a comma expression
list rather than as extra arguments to errfinish(). This means that
compilers can be expected to warn about no-op expressions in the list,
allowing detection of several other common mistakes such as forgetting
to add errmsg(...) when converting an elog() call to ereport().
* Unlike the situation with extra function arguments, comma expressions
are guaranteed to be evaluated left-to-right, so this removes platform
dependency in the order of the auxiliary function calls. While that
dependency hasn't caused us big problems in the past, this change does
allow dropping some rather shaky assumptions around errcontext() domain
handling.
There's no intention to make wholesale changes of existing ereport
calls, but as proof-of-concept this patch removes the extra parens
from a couple of calls in postgres.c.
While new code can be written either way, code intended to be
back-patched will need to use extra parens for awhile yet. It seems
worth back-patching this change into v12, so as to reduce the window
where we have to be careful about that by one year. Hence, this patch
is careful to preserve ABI compatibility; a followup HEAD-only patch
will make some additional simplifications.
Andres Freund and Tom Lane
Discussion: https://postgr.es/m/CA+fd4k6N8EjNvZpM8nme+y+05mz-SM8Z_BgkixzkA34R+ej0Kw@mail.gmail.com
