a new macro, DatumGetInetPP(), that does not. This brings these macros
in line with other DatumGet*P() macros.
Backpatch to 8.3, where 1-byte header varlenas were introduced.
This assumption can be wrong when the toaster is passed a raw on-disk
tuple, because the tuple might pre-date an ALTER TABLE ADD COLUMN operation
that added columns without rewriting the table. In such a case the tuple's
natts value is smaller than what we expect from the tuple descriptor, and
so its t_hoff value could be smaller too. In fact, the tuple might not
have a null bitmap at all, and yet our current opinion of it is that it
contains some trailing nulls.
In such a situation, toast_insert_or_update did the wrong thing, because
to save a few lines of code it would use the old t_hoff value as the offset
where heap_fill_tuple should start filling data. This did not leave enough
room for the new nulls bitmap, with the result that the first few bytes of
data could be overwritten with null flag bits, as in a recent report from
Hubert Depesz Lubaczewski.
The particular case reported requires ALTER TABLE ADD COLUMN followed by
CREATE TABLE AS SELECT * FROM ... or INSERT ... SELECT * FROM ..., and
further requires that there be some out-of-line toasted fields in one of
the tuples to be copied; else we'll not reach the troublesome code.
The problem can only manifest in this form in 8.4 and later, because
before commit a77eaa6a95009a3441e0d475d1980259d45da072, CREATE TABLE AS or
INSERT/SELECT wouldn't result in raw disk tuples getting passed directly
to heap_insert --- there would always have been at least a junkfilter in
between, and that would reconstitute the tuple header with an up-to-date
t_natts and hence t_hoff. But I'm backpatching the tuptoaster change all
the way anyway, because I'm not convinced there are no older code paths
that present a similar risk.
Both dict_int and dict_xsyn were blithely assuming that whatever memory
palloc gives back will be pre-zeroed. This would typically work for
just about long enough to run their regression tests, and no longer :-(.
The pre-9.0 code in dict_xsyn was even lamer than that, as it would
happily give back a pointer to the result of palloc(0), encouraging
its caller to access off the end of memory. Again, this would just
barely fail to fail as long as memory contained nothing but zeroes.
Per a report from Rodrigo Hjort that code based on these examples
didn't work reliably.
inline_set_returning_function failed to distinguish functions returning
generic RECORD (which require a column list in the RTE, as well as run-time
type checking) from those with multiple OUT parameters (which do not).
This prevented inlining from happening. Per complaint from Jay Levitt.
Back-patch to 8.4 where this capability was introduced.
This reverts commit d23165bbc306005f687756a20d9af807d665a1f2.
As pointed out by Naoya Anzai, we need to do more work to make that
idea handle end-of-index cases, and it is looking like too much risk
for a back-patch. So bug #6278 is only going to be fixed in HEAD.
If a tuple in a syscache contains an out-of-line toasted field, and we
try to fetch that field shortly after some other transaction has committed
an update or deletion of the tuple, there is a race condition: vacuum
could come along and remove the toast tuples before we can fetch them.
This leads to transient failures like "missing chunk number 0 for toast
value NNNNN in pg_toast_2619", as seen in recent reports from Andrew
Hammond and Tim Uckun.
The design idea of syscache is that access to stale syscache entries
should be prevented by relation-level locks, but that fails for at least
two cases where toasted fields are possible: ANALYZE updates pg_statistic
rows without locking out sessions that might want to plan queries on the
same table, and CREATE OR REPLACE FUNCTION updates pg_proc rows without
any meaningful lock at all.
The least risky fix seems to be an idea that Heikki suggested when we
were dealing with a related problem back in August: forcibly detoast any
out-of-line fields before putting a tuple into syscache in the first place.
This avoids the problem because at the time we fetch the parent tuple from
the catalog, we should be holding an MVCC snapshot that will prevent
removal of the toast tuples, even if the parent tuple is outdated
immediately after we fetch it. (Note: I'm not convinced that this
statement holds true at every instant where we could be fetching a syscache
entry at all, but it does appear to hold true at the times where we could
fetch an entry that could have a toasted field. We will need to be a bit
wary of adding toast tables to low-level catalogs that don't have them
already.) An additional benefit is that subsequent uses of the syscache
entry should be faster, since they won't have to detoast the field.
Back-patch to all supported versions. The problem is significantly harder
to reproduce in pre-9.0 releases, because of their willingness to flush
every entry in a syscache whenever the underlying catalog is vacuumed
(cf CatalogCacheFlushRelation); but there is still a window for trouble.
The existing scan-direction-sensitive tests were overly complex, and
failed to stop the scan in cases where it's perfectly legitimate to do so.
Per bug #6278 from Maksym Boguk.
Back-patch to 8.3, which is as far back as the patch applies easily.
Doesn't seem worth sweating over a relatively minor performance issue in
8.2 at this late date. (But note that this was a performance regression
from 8.1 and before, so 8.2 is being left as an outlier.)
cash_out failed to handle multiple-byte thousands separators, as per bug
#6277 from Alexander Law. In addition, cash_in didn't handle that either,
nor could it handle multiple-byte positive_sign. Both routines failed to
support multiple-byte mon_decimal_point, which I did not think was worth
changing, but at least now they check for the possibility and fall back to
using '.' rather than emitting invalid output. Also, make cash_in handle
trailing negative signs, which formerly it would reject. Since cash_out
generates trailing negative signs whenever the locale tells it to, this
last omission represents a fail-to-reload-dumped-data bug. IMO that
justifies patching this all the way back.
When a foreign-key constraint references another column of the same table,
row updates will queue both the PK's ON UPDATE action and the FK's CHECK
action in the same event. The ON UPDATE action must execute first, else
the CHECK will check a non-final state of the row and possibly throw an
inappropriate error, as seen in bug #6268 from Roman Lytovchenko.
Now, the firing order of multiple triggers for the same event is determined
by the sort order of their pg_trigger.tgnames, and the auto-generated names
we use for FK triggers are "RI_ConstraintTrigger_NNNN" where NNNN is the
trigger OID. So most of the time the firing order is the same as creation
order, and so rearranging the creation order fixes it.
This patch will fail to fix the problem if the OID counter wraps around or
adds a decimal digit (eg, from 99999 to 100000) while we are creating the
triggers for an FK constraint. Given the small odds of that, and the low
usage of self-referential FKs, we'll live with that solution in the back
branches. A better fix is to change the auto-generated names for FK
triggers, but it seems unwise to do that in stable branches because there
may be client code that depends on the naming convention. We'll fix it
that way in HEAD in a separate patch.
Back-patch to all supported branches, since this bug has existed for a long
time.
The heuristic for when to dump a cast failed for a cast between table
rowtypes, as reported by Frédéric Rejol. Fix it by setting
the "dump" flag for such a type the same way as the flag is set for the
underlying table or base type. This won't result in the auto-generated
type appearing in the output, since setting its objType to DO_DUMMY_TYPE
unconditionally suppresses that. But it will result in dumpCast doing what
was intended.
Back-patch to 8.3. The 8.2 code is rather different in this area, and it
doesn't seem worth any risk to fix a corner case that nobody has stumbled
on before.
This view was being insufficiently careful about matching the FK constraint
to the depended-on primary or unique key constraint. That could result in
failure to show an FK constraint at all, or showing it multiple times, or
claiming that it depended on a different constraint than the one it really
does. Fix by joining via pg_depend to ensure that we find only the correct
dependency.
Back-patch, but don't bump catversion because we can't force initdb in back
branches. The next minor-version release notes should explain that if you
need to fix this in an existing installation, you can drop the
information_schema schema then re-create it by sourcing
$SHAREDIR/information_schema.sql in each database (as a superuser of
course).
The documentation neglected to explain its behavior in a script file
(it only ends execution of the script, not psql as a whole), and failed
to mention the long form \quit either.
transform_null_equals is only supposed to affect "foo = NULL" expressions
given directly by the user, not the internal "foo = NULL" expression
generated from CASE-WHEN.
This fixes bug #6242, reported by Sergey. Backpatch to all supported
branches.
A similar problem for pgstattuple() was fixed in April of 2010 by commit
33065ef8bc52253ae855bc959576e52d8a28ba06, but pgstatindex() seems to have
been overlooked.
Back-patch all the way, as with that commit, though not to 7.4 through
8.1, since those are now EOL.
This oversight meant that on Windows, the pg_settings view would not
display source file or line number information for values coming from
postgresql.conf, unless the backend had received a SIGHUP since starting.
In passing, also make the error detection in read_nondefault_variables a
tad more thorough, and fix it to not lose precision on float GUCs (these
changes are already in HEAD as of my previous commit).
Thus, an object referenced in a default expression could be dropped while
the function remained present. This was unaccountably missed in the
original patch to add default parameters for functions. Reported by
Pavel Stehule.
This mode still exists for backwards compatibility, making
sslmode=require the same as sslmode=verify-ca when the file is present,
but not causing an error when it isn't.
Per bug 6189, reported by Srinivas Aji
We were mapping "Central America Standard Time" to "CST6CDT", which seems
entirely wrong, because according to the Olson timezone database noplace
in Central America observes daylight savings time on any regular basis ---
and certainly not according to the USA DST rules that are implied by
"CST6CDT". (Mexico is an exception, but they can be disregarded since
they have a separate timezone name in Windows.) So, map this zone name to
plain "CST6", which will provide a fixed UTC offset.
As written, this patch will also result in mapping "Central America
Daylight Time" to CST6. I considered hacking things so that would still
map to CST6CDT, but it seems it would confuse win32tzlist.pl to put those
two names in separate entries. Since there's little evidence that any
such zone name is used in the wild, much less that CST6CDT would be a good
match for it, I'm not too worried about what we do with it.
Per complaint from Pratik Chirania.
This oversight led to a massive memory leak --- upwards of 10KB per tuple
--- during creation-time verification of an exclusion constraint based on a
GIST index. In most other scenarios it'd just be a leak of 10KB that would
be recovered at end of query, so not too significant; though perhaps the
leak would be noticeable in a situation where a GIST index was being used
in a nestloop inner indexscan. In any case, it's a real leak of long
standing, so patch all supported branches. Per report from Harald Fuchs.
Trailing-zero stripping applied by the FM specifier could strip zeroes
to the left of the decimal point, for a format with no digit positions
after the decimal point (such as "FM999.").
Reported and diagnosed by Marti Raudsepp, though I didn't use his patch.
The code in shift_jis_20042euc_jis_2004() would fetch two bytes even when
only one remained in the string. Since conversion functions aren't
supposed to assume null-terminated input, this poses a small risk of
fetching past the end of memory and incurring SIGSEGV. No such crash has
been identified in the field, but we've certainly seen the equivalent
happen in other code paths, so patch this one all the way back.
Report and patch by Noah Misch.
Since the last couple of columns of pg_type are often NULL,
sizeof(FormData_pg_type) can be an overestimate of the actual size of the
tuple data part. Therefore memcpy'ing that much out of the catalog cache,
as analyze.c was doing, poses a small risk of copying past the end of
memory and incurring SIGSEGV. No such crash has been identified in the
field, but we've certainly seen the equivalent happen in other code paths,
so patch this one all the way back.
Per valgrind testing by Noah Misch, though this is not his proposed patch.
I chose to use SearchSysCacheCopy1 rather than inventing special-purpose
infrastructure for copying only the minimal part of a pg_type tuple.
This example wasn't updated when we changed the behavior of bpcharlen()
in 8.0, nor when we changed the number of parameters taken by the bpchar()
cast function in 7.3. Per report from lsliang.
">" should be ">>". This typo results in failure to use all of the bits
of the provided seed.
This might rise to the level of a security bug if we were relying on
srand48 for any security-critical purposes, but we are not --- in fact,
it's not used at all unless the platform lacks srandom(), which is
improbable. Even on such a platform the exposure seems minimal.
Reported privately by Andres Freund.
ifdef block. It has nothing to do with whether the replacement snprintf
function is used. It caused no live bug, because the replacement snprintf
function is always used on Win32, but it was nevertheless misplaced.
Examination of examples provided by Mark Kirkwood and others has convinced
me that actually commit 7f3eba30c9d622d1981b1368f2d79ba0999cdff2 was quite
a few bricks shy of a load. The useful part of that patch was clamping
ndistinct for the inner side of a semi or anti join, and the reason why
that's needed is that it's the only way that restriction clauses
eliminating rows from the inner relation can affect the estimated size of
the join result. I had not clearly understood why the clamping was
appropriate, and so mis-extrapolated to conclude that we should clamp
ndistinct for the outer side too, as well as for both sides of regular
joins. These latter actions were all wrong, and are reverted with this
patch. In addition, the clamping logic is now made to affect the behavior
of both paths in eqjoinsel_semi, with or without MCV lists to compare.
When we have MCVs, we suppose that the most common values are the ones
that are most likely to survive the decimation resulting from a lower
restriction clause, so we think of the clamping as eliminating non-MCV
values, or potentially even the least-common MCVs for the inner relation.
Back-patch to 8.4, same as previous fixes in this area.
This patch fixes an oversight in my commit
7f3eba30c9d622d1981b1368f2d79ba0999cdff2 of 2008-10-23. That patch
accounted for baserel restriction clauses that reduced the number of rows
coming out of a table (and hence the number of possibly-distinct values of
a join variable), but not for join restriction clauses that might have been
applied at a lower level of join. To account for the latter, look up the
sizes of the min_lefthand and min_righthand inputs of the current join,
and clamp with those in the same way as for the base relations.
Noted while investigating a complaint from Ben Chobot, although this in
itself doesn't seem to explain his report.
Back-patch to 8.4; previous versions used different estimation methods
for which this heuristic isn't relevant.
It is possible for VACUUM to scan no pages at all, if the visibility map
shows that all pages are all-visible. In this situation VACUUM has no new
information to report about the relation's tuple density, so it wasn't
changing pg_class.reltuples ... but it updated pg_class.relpages anyway.
That's wrong in general, since there is no evidence to justify changing the
density ratio reltuples/relpages, but it's particularly bad if the previous
state was relpages=reltuples=0, which means "unknown tuple density".
We just replaced "unknown" with "zero". ANALYZE would eventually recover
from this, but it could take a lot of repetitions of ANALYZE to do so if
the relation size is much larger than the maximum number of pages ANALYZE
will scan, because of the moving-average behavior introduced by commit
b4b6923e03f4d29636a94f6f4cc2f5cf6298b8c8.
The only known situation where we could have relpages=reltuples=0 and yet
the visibility map asserts everything's visible is immediately following
a pg_upgrade. It might be advisable for pg_upgrade to try to preserve the
relpages/reltuples statistics; but in any case this code is wrong on its
own terms, so fix it. Per report from Sergey Koposov.
Back-patch to 8.4, where the visibility map was introduced, same as the
previous change.
On closer inspection, whining in restore_toc_entries_parallel is really
much too late for any user-facing error case. The right place to do it
is at the start of RestoreArchive(), before we've done anything interesting
(suh as trying to DROP all the targets ...)
Back-patch to 8.4, where parallel restore was introduced.
If we are unable to do a parallel restore because the input file is stdin
or is otherwise unseekable, we should complain and fail immediately, not
after having done some of the restore. Complaining once per thread isn't
so cool either, and the messages should be worded to make it clear this is
an unsupported case not some weird race-condition bug. Per complaint from
Lonni Friedman.
Back-patch to 8.4, where parallel restore was introduced.
These days, such a response is far more likely to signify a server-side
problem, such as fork failure. Reporting "server does not support SSL"
(in sslmode=require) could be quite misleading. But the results could
be even worse in sslmode=prefer: if the problem was transient and the
next connection attempt succeeds, we'll have silently fallen back to
protocol version 2.0, possibly disabling features the user needs.
Hence, it seems best to just eliminate the assumption that backing off
to non-SSL/2.0 protocol is the way to recover from an "E" response, and
instead treat the server error the same as we would in non-SSL cases.
I tested this change against a pre-7.0 server, and found that there
was a second logic bug in the "prefer" path: the test to decide whether
to make a fallback connection attempt assumed that we must have opened
conn->ssl, which in fact does not happen given an "E" response. After
fixing that, the code does indeed connect successfully to pre-7.0,
as long as you didn't set sslmode=require. (If you did, you get
"Unsupported frontend protocol", which isn't completely off base
given the server certainly doesn't support SSL.)
Since there seems no reason to believe that pre-7.0 servers exist anymore
in the wild, back-patch to all supported branches.
There are assorted situations wherein PQconnectPoll() will abandon a
connection attempt and try again with different parameters (eg, SSL versus
not SSL). However, the code forgot to discard any pending data in libpq's
I/O buffers when doing this. In at least one case (server returns E
message during SSL negotiation), there is unread input data which bollixes
the next connection attempt. I have not checked to see whether this is
possible in the other cases where we close the socket and retry, but it
seems like a matter of good defensive programming to add explicit
buffer-flushing code to all of them.
This is one of several issues exposed by Daniel Farina's report of
misbehavior after a server-side fork failure.
This has been wrong since forever, so back-patch to all supported branches.
tsvector_concat() allocated its result workspace using the "conservative"
estimate of the sum of the two input tsvectors' sizes. Unfortunately that
wasn't so conservative as all that, because it supposed that the number of
pad bytes required could not grow. Which it can, as per test case from
Jesper Krogh, if there's a mix of lexemes with positions and lexemes
without them in the input data. The fix is to assume that we might add
a not-previously-present pad byte for each and every lexeme in the two
inputs; which really is conservative, but it doesn't seem worthwhile to
try to be more precise.
This is an aboriginal bug in tsvector_concat, so back-patch to all
versions containing it.
For an empty index, the pgstatindex() function would compute 0.0/0.0 for
its avg_leaf_density and leaf_fragmentation outputs. On machines that
follow the IEEE float arithmetic standard with any care, that results in
a NaN. However, per report from Rushabh Lathia, Microsoft couldn't
manage to get this right, so you'd get a bizarre error on Windows.
Fix by forcing the results to be NaN explicitly, rather than relying on
the division operator to give that or the snprintf function to print it
correctly. I have some doubts that this is really the most useful
definition, but it seems better to remain backward-compatible with
those platforms for which the behavior wasn't completely broken.
Back-patch to 8.2, since the code is like that in all current releases.
