Sloppy loop coding in set_status_by_pages() resulted in fetching one array
element more than it should from the subxids[] array. The odds of this
resulting in SIGSEGV are pretty small, but we've certainly seen that happen
with similar mistakes elsewhere. While at it, we can get rid of an extra
TransactionIdToPage() calculation per loop.
Per report from David Binderman. Back-patch to all supported branches,
since this code is quite old.
Discussion: https://postgr.es/m/HE1PR0802MB2331CBA919CBFFF0C465EB429C710@HE1PR0802MB2331.eurprd08.prod.outlook.com
When some tuple versions in an update chain are frozen due to them being
older than freeze_min_age, the xmax/xmin trail can become broken. This
breaks HOT (and probably other things). A subsequent VACUUM can break
things in more serious ways, such as leaving orphan heap-only tuples
whose root HOT redirect items were removed. This can be seen because
index creation (or REINDEX) complain like
ERROR: XX000: failed to find parent tuple for heap-only tuple at (0,7) in table "t"
Because of relfrozenxid contraints, we cannot avoid the freezing of the
early tuples, so we must cope with the results: whenever we see an Xmin
of FrozenTransactionId, consider it a match for whatever the previous
Xmax value was.
This problem seems to have appeared in 9.3 with multixact changes,
though strictly speaking it seems unrelated.
Since 9.4 we have commit 37484ad2a "Change the way we mark tuples as
frozen", so the fix is simple: just compare the raw Xmin (still stored
in the tuple header, since freezing merely set an infomask bit) to the
Xmax. But in 9.3 we rewrite the Xmin value to FrozenTransactionId, so
the original value is lost and we have nothing to compare the Xmax with.
To cope with that case we need to compare the Xmin with FrozenXid,
assume it's a match, and hope for the best. Sadly, since you can
pg_upgrade a 9.3 instance containing half-frozen pages to newer
releases, we need to keep the old check in newer versions too, which
seems a bit brittle; I hope we can somehow get rid of that.
I didn't optimize the new function for performance. The new coding is
probably a bit slower than before, since there is a function call rather
than a straight comparison, but I'd rather have it work correctly than
be fast but wrong.
This is a followup after 20b6552242 fixed a few related problems.
Apparently, in 9.6 and up there are more ways to get into trouble, but
in 9.3 - 9.5 I cannot reproduce a problem anymore with this patch, so
there must be a separate bug.
Reported-by: Peter Geoghegan
Diagnosed-by: Peter Geoghegan, Michael Paquier, Daniel Wood,
Yi Wen Wong, Álvaro
Discussion: https://postgr.es/m/CAH2-Wznm4rCrhFAiwKPWTpEw2bXDtgROZK7jWWGucXeH3D1fmA@mail.gmail.com
1. Since commit b1a9bad9e7 we had pstrdup() inside a
spinlock-protected critical section; reported by Andreas Seltenreich.
Turn those into strlcpy() to stack-allocated variables instead.
Backpatch to 9.6.
2. Since commit 9ed551e0a4 we had a pfree() uselessly inside a
spinlock-protected critical section. Tom Lane noticed in code review.
Move down. Backpatch to 9.6.
3. Since commit 64233902d2 we had GetCurrentTimestamp() (a kernel
call) inside a spinlock-protected critical section. Tom Lane noticed in
code review. Move it up. Backpatch to 9.2.
4. Since commit 1bb2558046 we did elog(PANIC) while holding spinlock.
Tom Lane noticed in code review. Release spinlock before dying.
Backpatch to 9.2.
Discussion: https://postgr.es/m/87h8vhtgj2.fsf@ansel.ydns.eu
Vacuum calls page-level HOT prune to remove dead HOT tuples before doing
liveness checks (HeapTupleSatisfiesVacuum) on the remaining tuples. But
concurrent transaction commit/abort may turn DEAD some of the HOT tuples
that survived the prune, before HeapTupleSatisfiesVacuum tests them.
This happens to activate the code that decides to freeze the tuple ...
which resuscitates it, duplicating data.
(This is especially bad if there's any unique constraints, because those
are now internally violated due to the duplicate entries, though you
won't know until you try to REINDEX or dump/restore the table.)
One possible fix would be to simply skip doing anything to the tuple,
and hope that the next HOT prune would remove it. But there is a
problem: if the tuple is older than freeze horizon, this would leave an
unfrozen XID behind, and if no HOT prune happens to clean it up before
the containing pg_clog segment is truncated away, it'd later cause an
error when the XID is looked up.
Fix the problem by having the tuple freezing routines cope with the
situation: don't freeze the tuple (and keep it dead). In the cases that
the XID is older than the freeze age, set the HEAP_XMAX_COMMITTED flag
so that there is no need to look up the XID in pg_clog later on.
An isolation test is included, authored by Michael Paquier, loosely
based on Daniel Wood's original reproducer. It only tests one
particular scenario, though, not all the possible ways for this problem
to surface; it be good to have a more reliable way to test this more
fully, but it'd require more work.
In message https://postgr.es/m/20170911140103.5akxptyrwgpc25bw@alvherre.pgsql
I outlined another test case (more closely matching Dan Wood's) that
exposed a few more ways for the problem to occur.
Backpatch all the way back to 9.3, where this problem was introduced by
multixact juggling. In branches 9.3 and 9.4, this includes a backpatch
of commit e5ff9fefcd50 (of 9.5 era), since the original is not
correctable without matching the coding pattern in 9.5 up.
Reported-by: Daniel Wood
Diagnosed-by: Daniel Wood
Reviewed-by: Yi Wen Wong, Michaël Paquier
Discussion: https://postgr.es/m/E5711E62-8FDF-4DCA-A888-C200BF6B5742@amazon.com
float8_numeric() and float4_numeric() failed to consider the possibility
that the input is an IEEE infinity. The results depended on the
platform-specific behavior of sprintf(): on most platforms you'd get
something like
ERROR: invalid input syntax for type numeric: "inf"
but at least on Windows it's possible for the conversion to succeed and
deliver a finite value (typically 1), due to a nonstandard output format
from sprintf and lack of syntax error checking in these functions.
Since our numeric type lacks the concept of infinity, a suitable conversion
is impossible; the best thing to do is throw an explicit error before
letting sprintf do its thing.
While at it, let's use snprintf not sprintf. Overrunning the buffer
should be impossible if sprintf does what it's supposed to, but this
is cheap insurance against a stack smash if it doesn't.
Problem reported by Taiki Kondo. Patch by me based on fix suggestion
from KaiGai Kohei. Back-patch to all supported branches.
Discussion: https://postgr.es/m/12A9442FBAE80D4E8953883E0B84E088C8C7A2@BPXM01GP.gisp.nec.co.jp
In two cases, we set a different umask for some piece of code and
restore it afterwards. But if the contained code errors out, the umask
is not restored. So add TRY/CATCH blocks to fix that.
Previously, the code didn't think about this case and would just try to
analyze such a column twice. That would fail at the point of inserting
the second version of the pg_statistic row, with obscure error messsages
like "duplicate key value violates unique constraint" or "tuple already
updated by self", depending on context and PG version. We could allow
the case by ignoring duplicate column specifications, but it seems better
to reject it explicitly.
The bogus error messages seem like arguably a bug, so back-patch to
all supported versions.
Nathan Bossart, per a report from Michael Paquier, and whacked
around a bit by me.
Discussion: https://postgr.es/m/E061A8E3-5E3D-494D-94F0-E8A9B312BBFC@amazon.com
AfterTriggerEndQuery correctly notes that the query_stack could get
repalloc'd during a trigger firing, but it nonetheless passes the address
of a query_stack entry to afterTriggerInvokeEvents, so that if such a
repalloc occurs, afterTriggerInvokeEvents is already working with an
obsolete dangling pointer while it scans the rest of the events. Oops.
The only code at risk is its "delete_ok" cleanup code, so we can
prevent unsafe behavior by passing delete_ok = false instead of true.
However, that could have a significant performance penalty, because the
point of passing delete_ok = true is to not have to re-scan possibly
a large number of dead trigger events on the next time through the loop.
There's more than one way to skin that cat, though. What we can do is
delete all the "chunks" in the event list except the last one, since
we know all events in them must be dead. Deleting the chunks is work
we'd have had to do later in AfterTriggerEndQuery anyway, and it ends
up saving rescanning of just about the same events we'd have gotten
rid of with delete_ok = true.
In v10 and HEAD, we also have to be careful to mop up any per-table
after_trig_events pointers that would become dangling. This is slightly
annoying, but I don't think that normal use-cases will traverse this code
path often enough for it to be a performance problem.
It's pretty hard to hit this in practice because of the unlikelihood
of the query_stack getting resized at just the wrong time. Nonetheless,
it's definitely a live bug of ancient standing, so back-patch to all
supported branches.
Discussion: https://postgr.es/m/2891.1505419542@sss.pgh.pa.us
This appears to have been an omission in the original commit
0d692a0dc9. All related information_schema views already include
foreign tables.
Reported-by: Nicolas Thauvin <nicolas.thauvin@dalibo.com>
Stress testing by Andreas Seltenreich disclosed longstanding problems that
occur if a FATAL exit (e.g. due to receipt of SIGTERM) occurs while we are
trying to execute a ROLLBACK of an already-failed transaction. In such a
case, xact.c is in TBLOCK_ABORT state, so that AbortOutOfAnyTransaction
would skip AbortTransaction and go straight to CleanupTransaction. This
led to an assert failure in an assert-enabled build (due to the ROLLBACK's
portal still having a cleanup hook) or without assertions, to a FATAL exit
complaining about "cannot drop active portal". The latter's not
disastrous, perhaps, but it's messy enough to want to improve it.
We don't really want to run all of AbortTransaction in this code path.
The minimum required to clean up the open portal safely is to do
AtAbort_Memory and AtAbort_Portals. It seems like a good idea to
do AtAbort_Memory unconditionally, to be entirely sure that we are
starting with a safe CurrentMemoryContext. That means that if the
main loop in AbortOutOfAnyTransaction does nothing, we need an extra
step at the bottom to restore CurrentMemoryContext = TopMemoryContext,
which I chose to do by invoking AtCleanup_Memory. This'll result in
calling AtCleanup_Memory twice in many of the paths through this function,
but that seems harmless and reasonably inexpensive.
The original motivation for the assertion in AtCleanup_Portals was that
we wanted to be sure that any user-defined code executed as a consequence
of the cleanup hook runs during AbortTransaction not CleanupTransaction.
That still seems like a valid concern, and now that we've seen one case
of the assertion firing --- which means that exactly that would have
happened in a production build --- let's replace the Assert with a runtime
check. If we see the cleanup hook still set, we'll emit a WARNING and
just drop the hook unexecuted.
This has been like this a long time, so back-patch to all supported
branches.
Discussion: https://postgr.es/m/877ey7bmun.fsf@ansel.ydns.eu
The sole useful effect of this function, to check that no catcache
entries have positive refcounts at transaction end, has really been
obsolete since we introduced ResourceOwners in PG 8.1. We reduced the
checks to assertions years ago, so that the function was a complete
no-op in production builds. There have been previous discussions about
removing it entirely, but consensus up to now was that it had some small
value as a cross-check for bugs in the ResourceOwner logic.
However, it now emerges that it's possible to trigger these assertions
if you hit an assert-enabled backend with SIGTERM during a call to
SearchCatCacheList, because that function temporarily increases the
refcounts of entries it's intending to add to a catcache list construct.
In a normal ERROR scenario, the extra refcounts are cleaned up by
SearchCatCacheList's PG_CATCH block; but in a FATAL exit we do a
transaction abort and exit without ever executing PG_CATCH handlers.
There's a case to be made that this is a generic hazard and we should
consider restructuring elog(FATAL) handling so that pending PG_CATCH
handlers do get run. That's pretty scary though: it could easily create
more problems than it solves. Preliminary stress testing by Andreas
Seltenreich suggests that there are not many live problems of this ilk,
so we rejected that idea.
There are more-localized ways to fix the problem; the most principled
one would be to use PG_ENSURE_ERROR_CLEANUP instead of plain PG_TRY.
But adding cycles to SearchCatCacheList isn't very appealing. We could
also weaken the assertions in AtEOXact_CatCache in some more or less
ad-hoc way, but that just makes its raison d'etre even less compelling.
In the end, the most reasonable solution seems to be to just remove
AtEOXact_CatCache altogether, on the grounds that it's not worth trying
to fix it. It hasn't found any bugs for us in many years.
Per report from Jeevan Chalke. Back-patch to all supported branches.
Discussion: https://postgr.es/m/CAM2+6=VEE30YtRQCZX7_sCFsEpoUkFBV1gZazL70fqLn8rcvBA@mail.gmail.com
find_composite_type_dependencies correctly found columns that are of
the specified type, and columns that are of arrays of that type, but
not columns that are domains or ranges over the given type, its array
type, etc. The most general way to handle this seems to be to assume
that any type that is directly dependent on the specified type can be
treated as a container type, and processed recursively (allowing us
to handle nested cases such as ranges over domains over arrays ...).
Since a type's array type already has such a dependency, we can drop
the existing special case for the array type.
The very similar logic in get_rels_with_domain was likewise a few
bricks shy of a load, as it supposed that a directly dependent type
could *only* be a sub-domain. This is already wrong for ranges over
domains, and it'll someday be wrong for arrays over domains.
Add test cases illustrating the problems, and back-patch to all
supported branches.
Discussion: https://postgr.es/m/15268.1502309024@sss.pgh.pa.us
Commit 3eefc51053 claimed to make
pg_user_mappings enforce the qualifications user_mapping_options had
been enforcing, but its removal of a longstanding restriction left them
distinct when the current user is the subject of a mapping yet has no
server privileges. user_mapping_options emits no rows for such a
mapping, but pg_user_mappings includes full umoptions. Change
pg_user_mappings to show null for umoptions. Back-patch to 9.2, like
the above commit.
Reviewed by Tom Lane. Reported by Jeff Janes.
Security: CVE-2017-7547
Some authentication methods allowed it, others did not. In the client-side,
libpq does not even try to authenticate with an empty password, which makes
using empty passwords hazardous: an administrator might think that an
account with an empty password cannot be used to log in, because psql
doesn't allow it, and not realize that a different client would in fact
allow it. To clear that confusion and to be be consistent, disallow empty
passwords in all authentication methods.
All the authentication methods that used plaintext authentication over the
wire, except for BSD authentication, already checked that the password
received from the user was not empty. To avoid forgetting it in the future
again, move the check to the recv_password_packet function. That only
forbids using an empty password with plaintext authentication, however.
MD5 and SCRAM need a different fix:
* In stable branches, check that the MD5 hash stored for the user does not
not correspond to an empty string. This adds some overhead to MD5
authentication, because the server needs to compute an extra MD5 hash, but
it is not noticeable in practice.
* In HEAD, modify CREATE and ALTER ROLE to clear the password if an empty
string, or a password hash that corresponds to an empty string, is
specified. The user-visible behavior is the same as in the stable branches,
the user cannot log in, but it seems better to stop the empty password from
entering the system in the first place. Secondly, it is fairly expensive to
check that a SCRAM hash doesn't correspond to an empty string, because
computing a SCRAM hash is much more expensive than an MD5 hash by design,
so better avoid doing that on every authentication.
We could clear the password on CREATE/ALTER ROLE also in stable branches,
but we would still need to check at authentication time, because even if we
prevent empty passwords from being stored in pg_authid, there might be
existing ones there already.
Reported by Jeroen van der Ham, Ben de Graaff and Jelte Fennema.
Security: CVE-2017-7546
We don't actually support session tickets, since we do not create an SSL
session identifier. But it seems that OpenSSL will issue a session ticket
on-demand anyway, which will then fail when used. This results in
reconnection failures when using ticket-aware client-side SSL libraries
(such as the Npgsql .NET driver), as reported by Shay Rojansky.
To fix, just tell OpenSSL not to issue tickets. At some point in the
far future, we might consider enabling tickets instead. But the security
implications of that aren't entirely clear; and besides it would have
little benefit except for very short-lived database connections, which is
Something We're Bad At anyhow. It would take a lot of other work to get
to a point where that would really be an exciting thing to do.
While at it, also tell OpenSSL not to use a session cache. This doesn't
really do anything, since a backend would never populate the cache anyway,
but it might gain some micro-efficiencies and/or reduce security
exposures.
Patch by me, per discussion with Heikki Linnakangas and Shay Rojansky.
Back-patch to all supported versions.
Discussion: https://postgr.es/m/CADT4RqBU8N-csyZuzaook-c795dt22Zcwg1aHWB6tfVdAkodZA@mail.gmail.com
If several sessions are concurrently locking a tuple update chain with
nonconflicting lock modes using an old snapshot, and they all succeed,
it may happen that some of them fail because of restarting the loop (due
to a concurrent Xmax change) and getting an error in the subsequent pass
while trying to obtain a tuple lock that they already have in some tuple
version.
This can only happen with very high concurrency (where a row is being
both updated and FK-checked by multiple transactions concurrently), but
it's been observed in the field and can have unpleasant consequences
such as an FK check failing to see a tuple that definitely exists:
ERROR: insert or update on table "child_table" violates foreign key constraint "fk_constraint_name"
DETAIL: Key (keyid)=(123456) is not present in table "parent_table".
(where the key is observably present in the table).
Discussion: https://postgr.es/m/20170714210011.r25mrff4nxjhmf3g@alvherre.pgsql
Trading a little too heavily on letting the code path be the same whether
we were creating shared data structures or only attaching to them,
InitPredicateLocks() inserted the "scratch" PredicateLockTargetHash entry
unconditionally. This is just wrong if we're in a postmaster child,
which would only reach this code in EXEC_BACKEND builds. Most of the
time, the hash_search(HASH_ENTER) call would simply report that the
entry already existed, causing no visible effect since the code did not
bother to check for that possibility. However, if this happened while
some other backend had transiently removed the "scratch" entry, then
that other backend's eventual RestoreScratchTarget would suffer an
assert failure; this appears to be the explanation for a recent failure
on buildfarm member culicidae. In non-assert builds, there would be
no visible consequences there either. But nonetheless this is a pretty
bad bug for EXEC_BACKEND builds, for two reasons:
1. Each new backend would perform the hash_search(HASH_ENTER) call
without holding any lock that would prevent concurrent access to the
PredicateLockTargetHash hash table. This creates a low but certainly
nonzero risk of corruption of that hash table.
2. In the event that the race condition occurred, by reinserting the
scratch entry too soon, we were defeating the entire purpose of the
scratch entry, namely to guarantee that transaction commit could move
hash table entries around with no risk of out-of-memory failure.
The odds of an actual OOM failure are quite low, but not zero, and if
it did happen it would again result in corruption of the hash table.
The user-visible symptoms of such corruption are a little hard to predict,
but would presumably amount to misbehavior of SERIALIZABLE transactions
that'd require a crash or postmaster restart to fix.
To fix, just skip the hash insertion if IsUnderPostmaster. I also
inserted a bunch of assertions that the expected things happen
depending on whether IsUnderPostmaster is true. That might be overkill,
since most comparable code in other functions isn't quite that paranoid,
but once burnt twice shy.
In passing, also move a couple of lines to places where they seemed
to make more sense.
Diagnosis of problem by Thomas Munro, patch by me. Back-patch to
all supported branches.
Discussion: https://postgr.es/m/10593.1500670709@sss.pgh.pa.us
Various cases involving renaming of view columns are handled by having
make_viewdef pass down the view's current relation tupledesc to
get_query_def, which then takes care to use the column names from the
tupledesc for the output column names of the SELECT. For some reason
though, we'd missed teaching make_ruledef to do similarly when it is
printing an ON SELECT rule, even though this is exactly the same case.
The results from pg_get_ruledef would then be different and arguably wrong.
In particular, this breaks pre-v10 versions of pg_dump, which in some
situations would define views by means of emitting a CREATE RULE ... ON
SELECT command. Third-party tools might not be happy either.
In passing, clean up some crufty code in make_viewdef; we'd apparently
modernized the equivalent code in make_ruledef somewhere along the way,
and missed this copy.
Per report from Gilles Darold. Back-patch to all supported versions.
Discussion: https://postgr.es/m/ec05659a-40ff-4510-fc45-ca9d965d0838@dalibo.com
Normally, a JoinExpr would have empty "quals" only if it came from CROSS
JOIN syntax. However, it's possible to get to this state by specifying
NATURAL JOIN between two tables with no common column names, and there
might be other ways too. The code previously printed no ON clause if
"quals" was empty; that's right for CROSS JOIN but syntactically invalid
if it's some type of outer join. Fix by printing ON TRUE in that case.
This got broken by commit 2ffa740be, which stopped using NATURAL JOIN
syntax in ruleutils output due to its brittleness in the face of
column renamings. Back-patch to 9.3 where that commit appeared.
Per report from Tushar Ahuja.
Discussion: https://postgr.es/m/98b283cd-6dda-5d3f-f8ac-87db8c76a3da@enterprisedb.com
The grammar will only accept something syntactically similar to a function
call in a function-in-FROM expression. However, there are various ways
to input something that ruleutils.c won't deparse that way, potentially
leading to a view or rule that fails dump/reload. Fix by inserting a
dummy CAST around anything that isn't going to deparse as a function
(which is one of the ways to get something like that in there in the
first place).
In HEAD, also make use of the infrastructure added by this to avoid
emitting unnecessary parentheses in CREATE INDEX deparsing. I did
not change that in back branches, thinking that people might find it
to be unexpected/unnecessary behavioral change.
In HEAD, also fix incorrect logic for when to add extra parens to
partition key expressions. Somebody apparently thought they could
get away with simpler logic than pg_get_indexdef_worker has, but
they were wrong --- a counterexample is PARTITION BY LIST ((a[1])).
Ignoring the prettyprint flag for partition expressions isn't exactly
a nice solution anyway.
This has been broken all along, so back-patch to all supported branches.
Discussion: https://postgr.es/m/10477.1499970459@sss.pgh.pa.us
The race condition goes like this:
1. GetNewTransactionId advances nextXid e.g. from 100 to 101
2. GetOldestActiveTransactionId reads the new nextXid, 101
3. GetOldestActiveTransactionId loops through the proc array. There are no
active XIDs there, so it returns 101 as the oldest active XID.
4. GetNewTransactionid stores XID 100 to MyPgXact->xid
So, GetOldestActiveTransactionId returned XID 101, even though 100 only
just started and is surely still running.
This would be hard to hit in practice, and even harder to spot any ill
effect if it happens. GetOldestActiveTransactionId is only used when
creating a checkpoint in a master server, and the race condition can only
happen on an online checkpoint, as there are no backends running during a
shutdown checkpoint. The oldestActiveXid value of an online checkpoint is
only used when starting up a hot standby server, to determine the starting
point where pg_subtrans is initialized from. For the race condition to
happen, there must be no other XIDs in the proc array that would hold back
the oldest-active XID value, which means that the missed XID must be a top
transaction's XID. However, pg_subtrans is not used for top XIDs, so I
believe an off-by-one error is in fact inconsequential. Nevertheless, let's
fix it, as it's clearly wrong and the fix is simple.
This has been wrong ever since hot standby was introduced, so backport to
all supported versions.
Discussion: https://www.postgresql.org/message-id/e7258662-82b6-7a45-56d4-99b337a32bf7@iki.fi
Further investigation shows that ruleutils isn't quite up to speed either
for cases where we have a domain-over-array: it needs to be prepared to
look past a CoerceToDomain at the top level of field and element
assignments, else it decompiles them incorrectly. Potentially this would
result in failure to dump/reload a rule, if it looked like the one in the
new test case. (I also added a test for EXPLAIN; that output isn't broken,
but clearly we need more test coverage here.)
Like commit b1cb32fb6, this bug is reachable in cases we already support,
so back-patch all the way.
compute_tsvector_stats() detoasted and kept in memory every tsvector value
in the sample, but that can be a lot of memory. The original bug report
described a case using over 10 gigabytes, with statistics target of 10000
(the maximum).
To fix, allocate a separate copy of just the lexemes that we keep around,
and free the detoasted tsvector values as we go. This adds some palloc/pfree
overhead, when you have a lot of distinct lexemes in the sample, but it's
better than running out of memory.
Fixes bug #14654 reported by James C. Reviewed by Tom Lane. Backport to
all supported versions.
Discussion: https://www.postgresql.org/message-id/20170514200602.1451.46797@wrigleys.postgresql.org
We allow INSERT and UPDATE commands to assign to the same column more than
once, as long as the assignments are to subfields or elements rather than
the whole column. However, this failed when the target column was a domain
over array rather than plain array. Fix by teaching process_matched_tle()
to look through CoerceToDomain nodes, and add relevant test cases.
Also add a group of test cases exercising domains over array of composite.
It's doubtless accidental that CREATE DOMAIN allows this case while not
allowing straight domain over composite; but it does, so we'd better make
sure we don't break it. (I could not find any documentation mentioning
either side of that, so no doc changes.)
It's been like this for a long time, so back-patch to all supported
branches.
Discussion: https://postgr.es/m/4206.1499798337@sss.pgh.pa.us
We've heard occasional reports of backend launch failing because
pgwin32_ReserveSharedMemoryRegion() fails, indicating that something
has already used that address space in the child process. It's not
very clear what, given that we disable ASLR in Windows builds, but
suspicion falls on antivirus products. It'd be better if we didn't
have to disable ASLR, anyway. So let's try to ameliorate the problem
by retrying the process launch after such a failure, up to 100 times.
Patch by me, based on previous work by Amit Kapila and others.
This is a longstanding issue, so back-patch to all supported branches.
Discussion: https://postgr.es/m/CAA4eK1+R6hSx6t_yvwtx+NRzneVp+MRqXAdGJZChcau8Uij-8g@mail.gmail.com
Buildfarm evidence shows that TCP_KEEPALIVE_THRESHOLD doesn't exist
after all on Solaris < 11. This means we need to take positive action to
prevent the TCP_KEEPALIVE code path from being taken on that platform.
I've chosen to limit it with "&& defined(__darwin__)", since it's unclear
that anyone else would follow Apple's precedent of spelling the symbol
that way.
Also, follow a suggestion from Michael Paquier of eliminating code
duplication by defining a couple of intermediate symbols for the
socket option.
In passing, make some effort to reduce the number of translatable messages
by replacing "setsockopt(foo) failed" with "setsockopt(%s) failed", etc,
throughout the affected files. And update relevant documentation so
that it doesn't claim to provide an exhaustive list of the possible
socket option names.
Like the previous commit (f0256c774), back-patch to all supported branches.
Discussion: https://postgr.es/m/20170627163757.25161.528@wrigleys.postgresql.org
Turns out that the socket option for this is named TCP_KEEPALIVE_THRESHOLD,
at least according to the tcp(7P) man page for Solaris 11. (But since that
text refers to "SunOS", it's likely pretty ancient.) It appears that the
symbol TCP_KEEPALIVE does get defined on that platform, but it doesn't
seem to represent a valid protocol-level socket option. This leads to
bleats in the postmaster log, and no tcp_keepalives_idle functionality.
Per bug #14720 from Andrey Lizenko, as well as an earlier report from
Dhiraj Chawla that nobody had followed up on. The issue's been there
since we added the TCP_KEEPALIVE code path in commit 5acd417c8, so
back-patch to all supported branches.
Discussion: https://postgr.es/m/20170627163757.25161.528@wrigleys.postgresql.org
check_agg_arguments_walker threw an error upon seeing a SRF or window
function, but that is too aggressive: if the function is within a
sub-select then it's perfectly fine. I broke the SRF case in commit
0436f6bde by copying the logic for window functions ... but that was
broken too, and had been since commit eaccfded9.
Repair both cases in HEAD, and the window function case back to 9.3.
9.2 gets this right.
When a walreceiver dies, the startup process will notice that and send
a PMSIGNAL_START_WALRECEIVER signal to the postmaster, asking for a new
walreceiver to be launched. There's a race condition, which at least
in HEAD is very easy to hit, whereby the postmaster might see that
signal before it processes the SIGCHLD from the walreceiver process.
In that situation, sigusr1_handler() just dropped the start request
on the floor, reasoning that it must be redundant. Eventually, after
10 seconds (WALRCV_STARTUP_TIMEOUT), the startup process would make a
fresh request --- but that's a long time if the connection could have
been re-established almost immediately.
Fix it by setting a state flag inside the postmaster that we won't
clear until we do launch a walreceiver. In cases where that results
in an extra walreceiver launch, it's up to the walreceiver to realize
it's unwanted and go away --- but we have, and need, that logic anyway
for the opposite race case.
I came across this through investigating unexpected delays in the
src/test/recovery TAP tests: it manifests there in test cases where
a master server is stopped and restarted while leaving streaming
slaves active.
This logic has been broken all along, so back-patch to all supported
branches.
Discussion: https://postgr.es/m/21344.1498494720@sss.pgh.pa.us
The stats collector disregards inquiry messages that bear a cutoff_time
before when it last wrote the relevant stats file. That's fine, but at
startup when it reads the "permanent" stats files, it absorbed their
timestamps as if they were the times at which the corresponding temporary
stats files had been written. In reality, of course, there's no data
out there at all. This led to disregarding inquiry messages soon after
startup if the postmaster had been shut down and restarted within less
than PGSTAT_STAT_INTERVAL; which is a pretty common scenario, both for
testing and in the field. Requesting backends would hang for 10 seconds
and then report failure to read statistics, unless they got bailed out
by some other backend coming along and making a newer request within
that interval.
I came across this through investigating unexpected delays in the
src/test/recovery TAP tests: it manifests there because the autovacuum
launcher hangs for 10 seconds when it can't get statistics at startup,
thus preventing a second shutdown from occurring promptly. We might
want to do some things in the autovac code to make it less prone to
getting stuck that way, but this change is a good bug fix regardless.
In passing, also fix pgstat_read_statsfiles() to ensure that it
re-zeroes its global stats variables if they are corrupted by a
short read from the stats file. (Other reads in that function
go into temp variables, so that the issue doesn't arise.)
This has been broken since we created the separation between permanent
and temporary stats files in 8.4, so back-patch to all supported branches.
Discussion: https://postgr.es/m/16860.1498442626@sss.pgh.pa.us
When promoting a standby just after a XLOG_SWITCH record was replayed,
and next segment(s) are already are locally available (via walsender,
restore_command + trigger/recovery target), that segment could
accidentally be recycled onto the past of the new timeline. Later
checkpointer would create a .ready file for it, assuming there was an
error during creation, and it would get archived. That causes trouble
if another standby is later brought up from a basebackup from before
the timeline creation, because it would try to read the
segment, because XLogFileReadAnyTLI just tries all possible timelines,
which doesn't have valid contents. Thus replay would fail.
The problem, if already occurred, can be fixed by removing the segment
and/or having restore_command filter it out.
The reason for the creation of such "phantom" segments was, that after
an XLOG_SWITCH record the EndOfLog variable points to the beginning of
the next segment, and RemoveXlogFile() used XLByteToPrevSeg().
Normally RemoveXlogFile() doing so is harmless, because the last
segment will still exist preventing InstallXLogFileSegment() from
causing harm, but just after promotion there's no previous segment on
the new timeline.
Fix that by using XLByteToSeg() instead of XLByteToPrevSeg().
Author: Andres Freund
Reported-By: Greg Burek
Discussion: https://postgr.es/m/20170619073026.zcwpe6mydsaz5ygd@alap3.anarazel.de
Backpatch: 9.2-, bug older than all supported versions
When a new base type is created using the old-style procedure of first
creating the input/output functions with "opaque" in place of the base
type, the "opaque" argument/return type is changed to the final base type,
on CREATE TYPE. However, we did not create a pg_depend record when doing
that, so the functions were left not depending on the type.
Fixes bug #14706, reported by Karen Huddleston.
Discussion: https://www.postgresql.org/message-id/20170614232259.1424.82774@wrigleys.postgresql.org
We had three occurrences of essentially the same coding pattern
wherein we tried to retrieve a query result from a libpq connection
without blocking. In the case where PQconsumeInput failed (typically
indicating a lost connection), all three loops simply gave up and
returned, forgetting to clear any previously-collected PGresult
object. Since those are malloc'd not palloc'd, the oversight results
in a process-lifespan memory leak.
One instance, in libpqwalreceiver, is of little significance because
the walreceiver process would just quit anyway if its connection fails.
But we might as well fix it.
The other two instances, in postgres_fdw, are somewhat more worrisome
because at least in principle the scenario could be repeated, allowing
the amount of memory leaked to build up to something worth worrying
about. Moreover, in these cases the loops contain CHECK_FOR_INTERRUPTS
calls, as well as other calls that could potentially elog(ERROR),
providing another way to exit without having cleared the PGresult.
Here we need to add PG_TRY logic similar to what exists in quite a
few other places in postgres_fdw.
Coverity noted the libpqwalreceiver bug; I found the other two cases
by checking all calls of PQconsumeInput.
Back-patch to all supported versions as appropriate (9.2 lacks
postgres_fdw, so this is really quite unexciting for that branch).
Discussion: https://postgr.es/m/22620.1497486981@sss.pgh.pa.us
Because walsender and normal backends share the same main loop it's
problematic to have two different flag variables, set in signal
handlers, indicating a pending configuration reload. Only certain
walsender commands reach code paths checking for the
variable (START_[LOGICAL_]REPLICATION, CREATE_REPLICATION_SLOT
... LOGICAL, notably not base backups).
This is a bug present since the introduction of walsender, but has
gotten worse in releases since then which allow walsender to do more.
A later patch, not slated for v10, will similarly unify SIGHUP
handling in other types of processes as well.
Author: Petr Jelinek, Andres Freund
Reviewed-By: Michael Paquier
Discussion: https://postgr.es/m/20170423235941.qosiuoyqprq4nu7v@alap3.anarazel.de
Backpatch: 9.2-, bug is present since 9.0
The NumericOnly grammar production accepted ICONST, + ICONST, - ICONST,
FCONST, and - FCONST, but for some reason not + FCONST. This led to
strange inconsistencies like
regression=# set random_page_cost = +4;
SET
regression=# set random_page_cost = 4000000000;
SET
regression=# set random_page_cost = +4000000000;
ERROR: syntax error at or near "4000000000"
(because 4000000000 is too large to be an ICONST). While there's
no actual functional reason to need to write a "+", if we allow
it for integers it seems like we should allow it for numerics too.
It's been like that forever, so back-patch to all supported branches.
Discussion: https://postgr.es/m/30908.1496006184@sss.pgh.pa.us
Commit 9aa3c782c added code to allow CREATE TABLE/CREATE TYPE to not fail
when the desired type name conflicts with an autogenerated array type, by
dint of renaming the array type out of the way. But I (tgl) overlooked
that the same case arises in ALTER TABLE/TYPE RENAME. Fix that too.
Back-patch to all supported branches.
Report and patch by Vik Fearing, modified a bit by me
Discussion: https://postgr.es/m/0f4ade49-4f0b-a9a3-c120-7589f01d1eb8@2ndquadrant.com
This patch replaces isspace() calls with scanner_isspace() in functions
that are likely to be presented with non-ASCII input. isspace() has
the small advantage that it will correctly recognize no-break space
in single-byte encodings (such as LATIN1); but it cannot work successfully
for any multibyte character, and depending on platform it might return
false positive results for some fragments of multibyte characters. That's
disastrous for functions that are trying to discard whitespace between
valid strings, as noted in bug #14662 from Justin Muise. Even treating
no-break space as whitespace is pretty questionable for the usages touched
here, because the core scanner would think it is an identifier character.
Affected functions are parse_ident(), parseNameAndArgTypes (underlying
regprocedurein() and siblings), SplitIdentifierString (used for parsing
GUCs and options that are qualified names or lists of names), and
SplitDirectoriesString (used for parsing GUCs that are lists of
directories).
All the functions adjusted here are parsing SQL identifiers and similar
constructs, so it's reasonable to insist that their definition of
whitespace match the core scanner. So we can hope that this won't cause
many backwards-compatibility problems. I've left alone isspace() calls
in places that aren't really expecting any non-ASCII input characters,
such as float8in().
Back-patch to all supported branches.
Discussion: https://postgr.es/m/10129.1495302480@sss.pgh.pa.us
The cash_div_intX functions applied rint() to the result of the division.
That's not merely useless (because the result is already an integer) but
it causes precision loss for values larger than 2^52 or so, because of
the forced conversion to float8.
On the other hand, the cash_mul_fltX functions neglected to apply rint() to
their multiplication results, thus possibly causing off-by-one outputs.
Per C standard, arithmetic between any integral value and a float value is
performed in float format. Thus, cash_mul_flt4 and cash_div_flt4 produced
answers good to only about six digits, even when the float value is exact.
We can improve matters noticeably by widening the float inputs to double.
(It's tempting to consider using "long double" arithmetic if available,
but that's probably too much of a stretch for a back-patched fix.)
Also, document that cash_div_intX operators truncate rather than round.
Per bug #14663 from Richard Pistole. Back-patch to all supported branches.
Discussion: https://postgr.es/m/22403.1495223615@sss.pgh.pa.us
