Given these routines are heap specific, and that there will be more
generic visibility support in via table AM, it makes sense to move the
prototypes to heapam.h (routines like HeapTupleSatisfiesVacuum will
not be exposed in a generic fashion, because they are too storage
specific).
Similarly, the code in tqual.c is specific to heap, so moving it into
access/heap/ makes sense.
Author: Andres Freund
Discussion: https://postgr.es/m/20180703070645.wchpu5muyto5n647@alap3.anarazel.de
Most of these had been obsoleted by 568d4138c / the SnapshotNow
removal.
This is is preparation for moving most of tqual.[ch] into either
snapmgr.h or heapam.h, which in turn is in preparation for pluggable
table AMs.
Author: Andres Freund
Discussion: https://postgr.es/m/20180703070645.wchpu5muyto5n647@alap3.anarazel.de
This removes a portion of infrastructure introduced by fe0a0b5 to allow
compilation of Postgres in environments where no strong random source is
available, meaning that there is no linking to OpenSSL and no
/dev/urandom (Windows having its own CryptoAPI). No systems shipped
this century lack /dev/urandom, and the buildfarm is actually not
testing this switch at all, so just remove it. This simplifies
particularly some backend code which included a fallback implementation
using shared memory, and removes a set of alternate regression output
files from pgcrypto.
Author: Michael Paquier
Reviewed-by: Tom Lane
Discussion: https://postgr.es/m/20181230063219.GG608@paquier.xyz
Move the responsibility for checking for and reporting a failure from
the only current BufFileSize() caller, logtape.c, to BufFileSize()
itself. Code within buffile.c is generally responsible for interfacing
with fd.c to report irrecoverable failures. This seems like a
convention that's worth sticking to.
Reorganizing things this way makes it easy to make the error message
raised in the event of BufFileSize() failure descriptive of the
underlying problem. We're now clear on the distinction between
temporary file name and BufFile name, and can show errno, confident that
its value actually relates to the error being reported. In passing, an
existing, similar buffile.c ereport() + errcode_for_file_access() site
is changed to follow the same conventions.
The API of the function BufFileSize() is changed by this commit, despite
already being in a stable release (Postgres 11). This seems acceptable,
since the BufFileSize() ABI was changed by commit aa55183042, which
hasn't made it into a point release yet. Besides, it's difficult to
imagine a third party BufFileSize() caller not just raising an error
anyway, since BufFile state should be considered corrupt when
BufFileSize() fails.
Per complaint from Tom Lane.
Discussion: https://postgr.es/m/26974.1540826748@sss.pgh.pa.us
Backpatch: 11-, where shared BufFiles were introduced.
If you extend a relation, it should count as a block written, not
read (we write a zero-filled block). If you ask for a zero-filled
buffer, it shouldn't be counted as read or written.
Later we might consider counting zero-filled buffers with a separate
counter, if they become more common due to future work.
Author: Thomas Munro
Reviewed-by: Haribabu Kommi, Kyotaro Horiguchi, David Rowley
Discussion: https://postgr.es/m/CAEepm%3D3JytB3KPpvSwXzkY%2Bdwc5zC8P8Lk7Nedkoci81_0E9rA%40mail.gmail.com
Users of the WaitEventSet and WaitLatch() APIs can now choose between
asking for WL_POSTMASTER_DEATH and then handling it explicitly, or asking
for WL_EXIT_ON_PM_DEATH to trigger immediate exit on postmaster death.
This reduces code duplication, since almost all callers want the latter.
Repair all code that was previously ignoring postmaster death completely,
or requesting the event but ignoring it, or requesting the event but then
doing an unconditional PostmasterIsAlive() call every time through its
event loop (which is an expensive syscall on platforms for which we don't
have USE_POSTMASTER_DEATH_SIGNAL support).
Assert that callers of WaitLatchXXX() under the postmaster remember to
ask for either WL_POSTMASTER_DEATH or WL_EXIT_ON_PM_DEATH, to prevent
future bugs.
The only process that doesn't handle postmaster death is syslogger. It
waits until all backends holding the write end of the syslog pipe
(including the postmaster) have closed it by exiting, to be sure to
capture any parting messages. By using the WaitEventSet API directly
it avoids the new assertion, and as a by-product it may be slightly
more efficient on platforms that have epoll().
Author: Thomas Munro
Reviewed-by: Kyotaro Horiguchi, Heikki Linnakangas, Tom Lane
Discussion: https://postgr.es/m/CAEepm%3D1TCviRykkUb69ppWLr_V697rzd1j3eZsRMmbXvETfqbQ%40mail.gmail.com,
https://postgr.es/m/CAEepm=2LqHzizbe7muD7-2yHUbTOoF7Q+qkSD5Q41kuhttRTwA@mail.gmail.com
Previously tables declared WITH OIDS, including a significant fraction
of the catalog tables, stored the oid column not as a normal column,
but as part of the tuple header.
This special column was not shown by default, which was somewhat odd,
as it's often (consider e.g. pg_class.oid) one of the more important
parts of a row. Neither pg_dump nor COPY included the contents of the
oid column by default.
The fact that the oid column was not an ordinary column necessitated a
significant amount of special case code to support oid columns. That
already was painful for the existing, but upcoming work aiming to make
table storage pluggable, would have required expanding and duplicating
that "specialness" significantly.
WITH OIDS has been deprecated since 2005 (commit ff02d0a05280e0).
Remove it.
Removing includes:
- CREATE TABLE and ALTER TABLE syntax for declaring the table to be
WITH OIDS has been removed (WITH (oids[ = true]) will error out)
- pg_dump does not support dumping tables declared WITH OIDS and will
issue a warning when dumping one (and ignore the oid column).
- restoring an pg_dump archive with pg_restore will warn when
restoring a table with oid contents (and ignore the oid column)
- COPY will refuse to load binary dump that includes oids.
- pg_upgrade will error out when encountering tables declared WITH
OIDS, they have to be altered to remove the oid column first.
- Functionality to access the oid of the last inserted row (like
plpgsql's RESULT_OID, spi's SPI_lastoid, ...) has been removed.
The syntax for declaring a table WITHOUT OIDS (or WITH (oids = false)
for CREATE TABLE) is still supported. While that requires a bit of
support code, it seems unnecessary to break applications / dumps that
do not use oids, and are explicit about not using them.
The biggest user of WITH OID columns was postgres' catalog. This
commit changes all 'magic' oid columns to be columns that are normally
declared and stored. To reduce unnecessary query breakage all the
newly added columns are still named 'oid', even if a table's column
naming scheme would indicate 'reloid' or such. This obviously
requires adapting a lot code, mostly replacing oid access via
HeapTupleGetOid() with access to the underlying Form_pg_*->oid column.
The bootstrap process now assigns oids for all oid columns in
genbki.pl that do not have an explicit value (starting at the largest
oid previously used), only oids assigned later by oids will be above
FirstBootstrapObjectId. As the oid column now is a normal column the
special bootstrap syntax for oids has been removed.
Oids are not automatically assigned during insertion anymore, all
backend code explicitly assigns oids with GetNewOidWithIndex(). For
the rare case that insertions into the catalog via SQL are called for
the new pg_nextoid() function can be used (which only works on catalog
tables).
The fact that oid columns on system tables are now normal columns
means that they will be included in the set of columns expanded
by * (i.e. SELECT * FROM pg_class will now include the table's oid,
previously it did not). It'd not technically be hard to hide oid
column by default, but that'd mean confusing behavior would either
have to be carried forward forever, or it'd cause breakage down the
line.
While it's not unlikely that further adjustments are needed, the
scope/invasiveness of the patch makes it worthwhile to get merge this
now. It's painful to maintain externally, too complicated to commit
after the code code freeze, and a dependency of a number of other
patches.
Catversion bump, for obvious reasons.
Author: Andres Freund, with contributions by John Naylor
Discussion: https://postgr.es/m/20180930034810.ywp2c7awz7opzcfr@alap3.anarazel.de
Previously, any program launched by COPY TO/FROM PROGRAM inherited the
server's setting of SIGPIPE handling, i.e. SIG_IGN. Hence, if we were
doing COPY FROM PROGRAM and closed the pipe early, the child process
would see EPIPE on its output file and typically would treat that as
a fatal error, in turn causing the COPY to report error. Similarly,
one could get a failure report from a query that didn't read all of
the output from a contrib/file_fdw foreign table that uses file_fdw's
PROGRAM option.
To fix, ensure that child programs inherit SIG_DFL not SIG_IGN processing
of SIGPIPE. This seems like an all-around better situation since if
the called program wants some non-default treatment of SIGPIPE, it would
expect to have to set that up for itself. Then in COPY, if it's COPY
FROM PROGRAM and we stop reading short of detecting EOF, treat a SIGPIPE
exit from the called program as a non-error condition. This still allows
us to report an error for any case where the called program gets SIGPIPE
on some other file descriptor.
As coded, we won't report a SIGPIPE if we stop reading as a result of
seeing an in-band EOF marker (e.g. COPY BINARY EOF marker). It's
somewhat debatable whether we should complain if the called program
continues to transmit data after an EOF marker. However, it seems like
we should avoid throwing error in any questionable cases, especially in a
back-patched fix, and anyway it would take additional code to make such
an error get reported consistently.
Back-patch to v10. We could go further back, since COPY FROM PROGRAM
has been around awhile, but AFAICS the only way to reach this situation
using core or contrib is via file_fdw, which has only supported PROGRAM
sources since v10. The COPY statement per se has no feature whereby
it'd stop reading without having hit EOF or an error already. Therefore,
I don't see any upside to back-patching further that'd outweigh the
risk of complaints about behavioral change.
Per bug #15449 from Eric Cyr.
Patch by me, review by Etsuro Fujita and Kyotaro Horiguchi
Discussion: https://postgr.es/m/15449-1cf737dd5929450e@postgresql.org
On some operating systems, it doesn't make sense to retry fsync(),
because dirty data cached by the kernel may have been dropped on
write-back failure. In that case the only remaining copy of the
data is in the WAL. A subsequent fsync() could appear to succeed,
but not have flushed the data. That means that a future checkpoint
could apparently complete successfully but have lost data.
Therefore, violently prevent any future checkpoint attempts by
panicking on the first fsync() failure. Note that we already
did the same for WAL data; this change extends that behavior to
non-temporary data files.
Provide a GUC data_sync_retry to control this new behavior, for
users of operating systems that don't eject dirty data, and possibly
forensic/testing uses. If it is set to on and the write-back error
was transient, a later checkpoint might genuinely succeed (on a
system that does not throw away buffers on failure); if the error is
permanent, later checkpoints will continue to fail. The GUC defaults
to off, meaning that we panic.
Back-patch to all supported releases.
There is still a narrow window for error-loss on some operating
systems: if the file is closed and later reopened and a write-back
error occurs in the intervening time, but the inode has the bad
luck to be evicted due to memory pressure before we reopen, we could
miss the error. A later patch will address that with a scheme
for keeping files with dirty data open at all times, but we judge
that to be too complicated to back-patch.
Author: Craig Ringer, with some adjustments by Thomas Munro
Reported-by: Craig Ringer
Reviewed-by: Robert Haas, Thomas Munro, Andres Freund
Discussion: https://postgr.es/m/20180427222842.in2e4mibx45zdth5%40alap3.anarazel.de
If fsync() fails, md.c must keep the request in its bitmap, so that
future attempts will try again.
Back-patch to all supported releases.
Author: Thomas Munro
Reviewed-by: Amit Kapila
Reported-by: Andrew Gierth
Discussion: https://postgr.es/m/87y3i1ia4w.fsf%40news-spur.riddles.org.uk
BufFileSize() can't use off_t, because it's only 32 bits wide on
some systems. BufFile objects can have many 1GB segments so the
total size can exceed 2^31. The only known client of the function
is parallel CREATE INDEX, which was reported to fail when building
large indexes on Windows.
Though this is technically an ABI break on platforms with a 32 bit
off_t and we might normally avoid back-patching it, the function is
brand new and thus unlikely to have been discovered by extension
authors yet, and it's fairly thoroughly broken on those platforms
anyway, so just fix it.
Defect in 9da0cc35. Bug #15460. Back-patch to 11, where this
function landed.
Author: Thomas Munro
Reported-by: Paul van der Linden, Pavel Oskin
Reviewed-by: Peter Geoghegan
Discussion: https://postgr.es/m/15460-b6db80de822fa0ad%40postgresql.org
Discussion: https://postgr.es/m/CAHDGBJP_GsESbTt4P3FZA8kMUKuYxjg57XHF7NRBoKnR%3DCAR-g%40mail.gmail.com
group clearing mechanism.
Commits 0e141c0fbb and baaf272ac9 introduced initialization of atomic
variables in InitProcess which means that it's not safe to look at those
for backends that aren't currently in use. Fix that by initializing them
during postmaster startup.
Reported-by: Andres Freund
Author: Amit Kapila
Backpatch-through: 9.6
Discussion: https://postgr.es/m/20181027104138.qmbbelopvy7cw2qv@alap3.anarazel.de
The use of volatiles in procarray.c largely originated from the time
when postgres did not have reliable compiler and memory
barriers. That's not the case anymore, so we can do better.
Several of the functions in procarray.c can be bottlenecks, and
removal of volatile yields mildly better code.
The new state, with explicit memory barriers, is also more
correct. The previous use of volatile did not actually deliver
sufficient guarantees on weakly ordered machines, in particular the
logic in GetNewTransactionId() does not look safe. It seems unlikely
to be a problem in practice, but worth fixing.
Thomas and I independently wrote a patch for this.
Reported-By: Andres Freund and Thomas Munro
Author: Andres Freund, with cherrypicked changes from a patch by Thomas Munro
Discussion:
https://postgr.es/m/20181005172955.wyjb4fzcdzqtaxjq@alap3.anarazel.dehttps://postgr.es/m/CAEepm=1nff0x=7i3YQO16jLA2qw-F9O39YmUew4oq-xcBQBs0g@mail.gmail.com
Previously the code checked PROC_IN_LOGICAL_DECODING and
PROC_IN_VACUUM separately. As the relevant variable is marked as
volatile, the compiler cannot combine the two tests. As
GetSnapshotData() is pretty hot in a number of workloads, it's
worthwhile to fix that.
It'd also be a good idea to get rid of the volatiles altogether. But
for one that's a larger patch, and for another, the code after this
change still seems at least as easy to read as before.
Author: Andres Freund
Discussion: https://postgr.es/m/20181005172955.wyjb4fzcdzqtaxjq@alap3.anarazel.de
This allows the compiler / linker to mark affected pages as read-only.
There's other cases, but they're a bit more invasive, and should go
through some review. These are easy.
They were found with
objdump -j .data -t src/backend/postgres|awk '{print $4, $5, $6}'|sort -r|less
Discussion: https://postgr.es/m/20181015200754.7y7zfuzsoux2c4ya@alap3.anarazel.de
On a primary, sets of XLOG_RUNNING_XACTS records are generated on a
periodic basis to allow recovery to build the initial state of
transactions for a hot standby. The set of transaction IDs is created
by scanning all the entries in ProcArray. However it happens that its
logic never counted on the fact that two-phase transactions finishing to
prepare can put ProcArray in a state where there are two entries with
the same transaction ID, one for the initial transaction which gets
cleared when prepare finishes, and a second, dummy, entry to track that
the transaction is still running after prepare finishes. This way
ensures a continuous presence of the transaction so as callers of for
example TransactionIdIsInProgress() are always able to see it as alive.
So, if a XLOG_RUNNING_XACTS takes a standby snapshot while a two-phase
transaction finishes to prepare, the record can finish with duplicated
XIDs, which is a state expected by design. If this record gets applied
on a standby to initial its recovery state, then it would simply fail,
so the odds of facing this failure are very low in practice. It would
be tempting to change the generation of XLOG_RUNNING_XACTS so as
duplicates are removed on the source, but this requires to hold on
ProcArrayLock for longer and this would impact all workloads,
particularly those using heavily two-phase transactions.
XLOG_RUNNING_XACTS is also actually used only to initialize the standby
state at recovery, so instead the solution is taken to discard
duplicates when applying the initial snapshot.
Diagnosed-by: Konstantin Knizhnik
Author: Michael Paquier
Discussion: https://postgr.es/m/0c96b653-4696-d4b4-6b5d-78143175d113@postgrespro.ru
Backpatch-through: 9.3
This moves the system administration functions for signalling backends
from backend/utils/adt/misc.c into a separate file dedicated to backend
signalling. No new functionality is introduced in this commit.
Author: Daniel Gustafsson
Reviewed-by: Michael Paquier, Álvaro Herrera
Discussion: https://postgr.es/m/C2C7C3EC-CC5F-44B6-9C78-637C88BD7D14@yesql.se
Opening a relation with no lock at all is unsafe; there's no guarantee
that we'll see a consistent state of the relevant catalog entries.
While use of MVCC scans to read the catalogs partially addresses that
complaint, it's still possible to switch to a new catalog snapshot
partway through loading the relcache entry. Moreover, whether or not
you trust the reasoning behind sometimes using less than
AccessExclusiveLock for ALTER TABLE, that reasoning is certainly not
valid if concurrent users of the table don't hold a lock corresponding
to the operation they want to perform.
Hence, add some assertion-build-only checks that require any caller
of relation_open(x, NoLock) to hold at least AccessShareLock. This
isn't a full solution, since we can't verify that the lock level is
semantically appropriate for the action --- but it's definitely of
some use, because it's already caught two bugs.
We can also assert that callers of addRangeTableEntryForRelation()
hold at least the lock level specified for the new RTE.
Amit Langote and Tom Lane
Discussion: https://postgr.es/m/16565.1538327894@sss.pgh.pa.us
Commit 37c54863c removed the code in StandbyAcquireAccessExclusiveLock
that checked the return value of LockAcquireExtended. That created a
bug, because it's still passing reportMemoryError = false to
LockAcquireExtended, meaning that LOCKACQUIRE_NOT_AVAIL will be returned
if we're out of shared memory for the lock table.
In such a situation, the startup process would believe it had acquired an
exclusive lock even though it hadn't, with potentially dire consequences.
To fix, just drop the use of reportMemoryError = false, which allows us
to simplify the call into a plain LockAcquire(). It's unclear that the
locktable-full situation arises often enough that it's worth having a
better recovery method than crash-and-restart. (I strongly suspect that
the only reason the code path existed at all was that it was relatively
simple to do in the pre-37c54863c implementation. But now it's not.)
LockAcquireExtended's reportMemoryError parameter is now dead code and
could be removed. I refrained from doing so, however, because there
was some interest in resurrecting the behavior if we do get reports of
locktable-full failures in the field. Also, it seems unwise to remove
the parameter concurrently with shipping commit f868a8143, which added a
parameter; if there are any third-party callers of LockAcquireExtended,
we want them to get a wrong-number-of-parameters compile error rather
than a possibly-silent misinterpretation of its last parameter.
Back-patch to 9.6 where the bug was introduced.
Discussion: https://postgr.es/m/6202.1536359835@sss.pgh.pa.us
Chris Travers reported that the startup process can repeatedly try to
cancel a backend that is in a posix_fallocate()/EINTR loop and cause it
to loop forever. Teach the retry loop to give up if an interrupt is
pending. Don't actually check for interrupts in that loop though,
because a non-local exit would skip some clean-up code in the caller.
Back-patch to 9.4 where DSM was added (and posix_fallocate() was later
back-patched).
Author: Chris Travers
Reviewed-by: Ildar Musin, Murat Kabilov, Oleksii Kliukin
Tested-by: Oleksii Kliukin
Discussion: https://postgr.es/m/CAN-RpxB-oeZve_J3SM_6%3DHXPmvEG%3DHX%2B9V9pi8g2YR7YW0rBBg%40mail.gmail.com
LockRelationOid and sibling routines supposed that, if our session already
holds the lock they were asked to acquire, they could skip calling
AcceptInvalidationMessages on the grounds that we must have already read
any remote sinval messages issued against the relation being locked.
This is normally true, but there's a critical special case where it's not:
processing inside AcceptInvalidationMessages might attempt to access system
relations, resulting in a recursive call to acquire a relation lock.
Hence, if the outer call had acquired that same system catalog lock, we'd
fall through, despite the possibility that there's an as-yet-unread sinval
message for that system catalog. This could, for example, result in
failure to access a system catalog or index that had just been processed
by VACUUM FULL. This is the explanation for buildfarm failures we've been
seeing intermittently for the past three months. The bug is far older
than that, but commits a54e1f158 et al added a new recursion case within
AcceptInvalidationMessages that is apparently easier to hit than any
previous case.
To fix this, we must not skip calling AcceptInvalidationMessages until
we have *finished* a call to it since acquiring a relation lock, not
merely acquired the lock. (There's already adequate logic inside
AcceptInvalidationMessages to deal with being called recursively.)
Fortunately, we can implement that at trivial cost, by adding a flag
to LOCALLOCK hashtable entries that tracks whether we know we have
completed such a call.
There is an API hazard added by this patch for external callers of
LockAcquire: if anything is testing for LOCKACQUIRE_ALREADY_HELD,
it might be fooled by the new return code LOCKACQUIRE_ALREADY_CLEAR
into thinking the lock wasn't already held. This should be a fail-soft
condition, though, unless something very bizarre is being done in
response to the test.
Also, I added an additional output argument to LockAcquireExtended,
assuming that that probably isn't called by any outside code given
the very limited usefulness of its additional functionality.
Back-patch to all supported branches.
Discussion: https://postgr.es/m/12259.1532117714@sss.pgh.pa.us
There's a project policy against using plain "char buf[BLCKSZ]" local
or static variables as page buffers; preferred style is to palloc or
malloc each buffer to ensure it is MAXALIGN'd. However, that policy's
been ignored in an increasing number of places. We've apparently got
away with it so far, probably because (a) relatively few people use
platforms on which misalignment causes core dumps and/or (b) the
variables chance to be sufficiently aligned anyway. But this is not
something to rely on. Moreover, even if we don't get a core dump,
we might be paying a lot of cycles for misaligned accesses.
To fix, invent new union types PGAlignedBlock and PGAlignedXLogBlock
that the compiler must allocate with sufficient alignment, and use
those in place of plain char arrays.
I used these types even for variables where there's no risk of a
misaligned access, since ensuring proper alignment should make
kernel data transfers faster. I also changed some places where
we had been palloc'ing short-lived buffers, for coding style
uniformity and to save palloc/pfree overhead.
Since this seems to be a live portability hazard (despite the lack
of field reports), back-patch to all supported versions.
Patch by me; thanks to Michael Paquier for review.
Discussion: https://postgr.es/m/1535618100.1286.3.camel@credativ.de
Commit dafa084, added in 10, made the removal of temporary orphaned
tables more aggressive. This commit makes an extra step into the
aggressiveness by adding a flag in each backend's MyProc which tracks
down any temporary namespace currently in use. The flag is set when the
namespace gets created and can be reset if the temporary namespace has
been created in a transaction or sub-transaction which is aborted. The
flag value assignment is assumed to be atomic, so this can be done in a
lock-less fashion like other flags already present in PGPROC like
databaseId or backendId, still the fact that the temporary namespace and
table created are still locked until the transaction creating those
commits acts as a barrier for other backends.
This new flag gets used by autovacuum to discard more aggressively
orphaned tables by additionally checking for the database a backend is
connected to as well as its temporary namespace in-use, removing
orphaned temporary relations even if a backend reuses the same slot as
one which created temporary relations in a past session.
The base idea of this patch comes from Robert Haas, has been written in
its first version by Tsunakawa Takayuki, then heavily reviewed by me.
Author: Tsunakawa Takayuki
Reviewed-by: Michael Paquier, Kyotaro Horiguchi, Andres Freund
Discussion: https://postgr.es/m/0A3221C70F24FB45833433255569204D1F8A4DC6@G01JPEXMBYT05
Backpatch: 11-, as PGPROC gains a new flag and we don't want silent ABI
breakages on already released versions.
When reading an existing FSM or VM page that was found to be corrupt by the
buffer manager, the code applied PageInit() to reinitialize the page, but
did so without any locking. There is thus a hazard that two backends might
concurrently do PageInit, which in itself would still be OK, but the slower
one might then zero over subsequent data changes applied by the faster one.
Even that is unlikely to be fatal; but it's not desirable, so add locking
to prevent it.
This does not add any locking overhead in the normal code path where the
page is OK. It's not immediately obvious that that's safe, but I believe
it is, for reasons explained in the added comments.
Problem noted by R P Asim. It's been like this for a long time, so
back-patch to all supported branches.
Discussion: https://postgr.es/m/CANXE4Te4G0TGq6cr0-TvwP0H4BNiK_-hB5gHe8mF+nz0mcYfMQ@mail.gmail.com
In ad9a274778, shmem_exit_inprogress was
introduced. But we need to reset it after shmem_exit(), because unlike
the similar proc_exit(), shmem_exit() can also be called for cleanup
when the process will not exit.
Reported-by: Andrew Gierth <andrew@tao11.riddles.org.uk>
Header comment of shm_mq.c was mistakenly specifying path to shm_mq.h.
It was introduced in ec9037df. So, theoretically it could be
backpatched to 9.4, but it doesn't seem to worth it.
PostgreSQL nowadays offers some kind of dynamic shared memory feature on
all supported platforms. Having the choice of "none" prevents us from
relying on DSM in core features. So this patch removes the choice of
"none".
Author: Kyotaro Horiguchi <horiguchi.kyotaro@lab.ntt.co.jp>
When multiple relations are deleted at the same transaction,
the files of those relations are deleted by one call to smgrdounlinkall(),
which leads to scan whole shared_buffers only one time. OTOH,
previously, during recovery, smgrdounlink() (not smgrdounlinkall()) was
called for each file to delete, which led to scan shared_buffers
multiple times. Obviously this could cause to increase the WAL replay
time very much especially when shared_buffers was huge.
To alleviate this situation, this commit changes the recovery so that
it also calls smgrdounlinkall() only one time to delete multiple
relation files.
This is just fix for oversight of commit 279628a0a7, not new feature.
So, per discussion on pgsql-hackers, we concluded to backpatch this
to all supported versions.
Author: Fujii Masao
Reviewed-by: Michael Paquier, Andres Freund, Thomas Munro, Kyotaro Horiguchi, Takayuki Tsunakawa
Discussion: https://postgr.es/m/CAHGQGwHVQkdfDqtvGVkty+19cQakAydXn1etGND3X0PHbZ3+6w@mail.gmail.com
Issues relate only to subtransactions that hold AccessExclusiveLocks
when replayed on standby.
Prior to PG10, aborting subtransactions that held an
AccessExclusiveLock failed to release the lock until top level commit or
abort. 49bff5300d fixed that.
However, 49bff5300d also introduced a similar bug where subtransaction
commit would fail to release an AccessExclusiveLock, leaving the lock to
be removed sometimes early and sometimes late. This commit fixes
that bug also. Backpatch to PG10 needed.
Tested by observation. Note need for multi-node isolationtester to improve
test coverage for this and other HS cases.
Reported-by: Simon Riggs
Author: Simon Riggs
Also this commit unifies some duplicated code in makeBufFile() and
BufFileOpenShared() into new function makeBufFileCommon().
Author: Antonin Houska
Reviewed-By: Thomas Munro, Tatsuo Ishii
Discussion: https://postgr.es/m/16139.1529049566%40localhost
GetRunningTransactionData() suggested that subxids were not worth
optimizing away if overflowed, yet they have already been removed
for that case.
Changes to LogAccessExclusiveLock() API forgot to remove the
prior comment when it was copied to LockAcquire().
32ac7a118f tried to fix a Hot Standby issue
reported by Greg Stark, but in doing so caused
a different bug to appear, noted by Andres Freund.
Revoke the core changes from 32ac7a118f,
leaving in its place a minor change in code
ordering and comments to explain for the future.
GetRunningTransactionData() should ignore VACUUM procs because in some
cases they are assigned xids. This could lead to holding back xmin via
the route of passing the xid to standby and then having that hold back
xmin on master via feedback.
Backpatch to 9.1 needed, but will only do so on supported versions.
Backpatch once proven on the buildfarm.
Reported-by: Greg Stark
Author: Simon Riggs
Reviewed-by: Amit Kapila
Discussion: https://postgr.es/m/CANP8+jJBYt=4PpTfiPb0UrH1_iPhzsxKH5Op_Wec634F0ohnAw@mail.gmail.com
Commit c37b3d08c dropped its added GetDataDirectoryCreatePerm call into
the wrong place in pg_resetwal.c, namely after the chdir to DataDir.
That broke invocations using a relative path, as reported by Tushar Ahuja.
We could have left it where it was and changed the argument to be ".",
but that'd result in a rather confusing error message in event of a
failure, so re-ordering seems like a better solution.
Similarly reorder operations in pg_rewind.c. The issue there is that
it doesn't seem like a good idea to do any actual operations before the
not-root check (on Unix) or the restricted token acquisition (on Windows).
I don't know that this is an actual bug, but I'm definitely not convinced
that it isn't, either.
Assorted other code review for c37b3d08c and da9b580d8: fix some
misspelled or otherwise badly worded comments, put the #include for
<sys/stat.h> where it actually belongs, etc.
Discussion: https://postgr.es/m/aeb9c3a7-3c3f-a57f-1a18-c8d4fcdc2a1f@enterprisedb.com
The principle behind the locking was not very well thought-out, and not
documented. Add a section in the README to explain how it's supposed to
work, and change the code so that it actually works that way.
This fixes two bugs:
1. If fast update was turned on concurrently, subsequent inserts to the
pending list would not conflict with predicate locks that were acquired
earlier, on entry pages. The included 'predicate-gin-fastupdate' test
demonstrates that. To fix, make all scans acquire a predicate lock on
the metapage. That lock represents a scan of the pending list, whether
or not there is a pending list at the moment. Forget about the
optimization to skip locking/checking for locks, when fastupdate=off.
2. If a scan finds no match, it still needs to lock the entry page. The
point of predicate locks is to lock the gabs between values, whether
or not there is a match. The included 'predicate-gin-nomatch' test
tests that case.
In addition to those two bug fixes, this removes some unnecessary locking,
following the principle laid out in the README. Because all items in
a posting tree have the same key value, a lock on the posting tree root is
enough to cover all the items. (With a very large posting tree, it would
possibly be better to lock the posting tree leaf pages instead, so that a
"skip scan" with a query like "A & B", you could avoid unnecessary conflict
if a new tuple is inserted with A but !B. But let's keep this simple.)
Also, some spelling fixes.
Author: Heikki Linnakangas with some editorization by me
Review: Andrey Borodin, Alexander Korotkov
Discussion: https://www.postgresql.org/message-id/0b3ad2c2-2692-62a9-3a04-5724f2af9114@iki.fi
Failure to use DatumGetFoo/FooGetDatum macros correctly, or at all,
causes some warnings about sign conversion. This is just cosmetic
at the moment but in principle it's a type violation, so clean up
the instances I could find.
autoprewarm.c and sharedfileset.c contained code that unportably
assumed that pid_t is the same size as int. We've variously dealt
with this by casting pid_t to int or to unsigned long for printing
purposes; I went with the latter.
Fix uninitialized-variable warning in RestoreGUCState. This is
a live bug in some sense, but of no great significance given that
nobody is very likely to care what "line number" is associated with
a GUC that hasn't got a source file recorded.
There were three related issues:
* BufFileAppend() incorrectly reset the seek position on the 'source' file.
As a result, if you had called BufFileRead() on the file before calling
BufFileAppend(), it got confused, and subsequent calls would read/write
at wrong position.
* BufFileSize() did not work with files opened with BufFileOpenShared().
* FileGetSize() only worked on temporary files.
To fix, change the way BufFileSize() works so that it works on shared
files. Remove FileGetSize() altogether, as it's no longer needed. Remove
buffilesize from TapeShare struct, as the leader process can simply call
BufFileSize() to get the tape's size, there's no need to pass it through
shared memory anymore.
Discussion: https://www.postgresql.org/message-id/CAH2-WznEDYe_NZXxmnOfsoV54oFkTdMy7YLE2NPBLuttO96vTQ@mail.gmail.com
This change makes this module act more like most of our other low-level
resource management modules. It's a caller error if something is not
explicitly closed by the end of a successful transaction, so issue
a WARNING about it. This would not actually have caught the file leak
bug fixed in commit 231bcd080, because that was in a transaction-abort
path; but it still seems like a good, and pretty cheap, cross-check.
Discussion: https://postgr.es/m/152056616579.4966.583293218357089052@wrigleys.postgresql.org
One improbable error-exit path in this function used close() where
it should have used CloseTransientFile(). This is unlikely to be
hit in the field, and I think the consequences wouldn't be awful
(just an elog(LOG) bleat later). But a bug is a bug, so back-patch
to 9.4 where this code came in.
Pan Bian
Discussion: https://postgr.es/m/152056616579.4966.583293218357089052@wrigleys.postgresql.org
In commit 9c0a0de4c, I'd failed to notice that catalog/catalog.h
should also be considered a frontend-unsafe header, because it includes
(and needs) the full form of pg_class.h, not to mention relcache.h.
However, various frontend code was depending on it to get
TABLESPACE_VERSION_DIRECTORY, so refactoring of some sort is called for.
The cleanest answer seems to be to move TABLESPACE_VERSION_DIRECTORY,
as well as the OIDCHARS symbol, to common/relpath.h. Do that, and mop up
inclusions as necessary. (I found that quite a few current users of
catalog/catalog.h don't seem to need it at all anymore, apparently as a
result of the refactorings that created common/relpath.[hc]. And
initdb.c needed it only as a route to pg_class_d.h.)
Discussion: https://postgr.es/m/6629.1523294509@sss.pgh.pa.us
This reverts the backend sides of commit 1fde38beaa.
I have, at least for now, left the pg_verify_checksums tool in place, as
this tool can be very valuable without the rest of the patch as well,
and since it's a read-only tool that only runs when the cluster is down
it should be a lot safer.
Consolidate directory and file create permissions for tools which work
with the PG data directory by adding a new module (common/file_perm.c)
that contains variables (pg_file_create_mode, pg_dir_create_mode) and
constants to initialize them (0600 for files and 0700 for directories).
Convert mkdir() calls in the backend to MakePGDirectory() if the
original call used default permissions (always the case for regular PG
directories).
Add tests to make sure permissions in PGDATA are set correctly by the
tools which modify the PG data directory.
Authors: David Steele <david@pgmasters.net>,
Adam Brightwell <adam.brightwell@crunchydata.com>
Reviewed-By: Michael Paquier, with discussion amongst many others.
Discussion: https://postgr.es/m/ad346fe6-b23e-59f1-ecb7-0e08390ad629%40pgmasters.net
Hash index searches acquire predicate locks on the primary
page of a bucket. It acquires a lock on both the old and new buckets
for scans that happen concurrently with page splits. During a bucket
split, a predicate lock is copied from the primary page of an old
bucket to the primary page of a new bucket.
Author: Shubham Barai, Amit Kapila
Reviewed by: Amit Kapila, Alexander Korotkov, Thomas Munro
Discussion: https://www.postgresql.org/message-id/flat/CALxAEPvNsM2GTiXdRgaaZ1Pjd1bs+sxfFsf7Ytr+iq+5JJoYXA@mail.gmail.com
This makes it possible to turn checksums on in a live cluster, without
the previous need for dump/reload or logical replication (and to turn it
off).
Enabling checkusm starts a background process in the form of a
launcher/worker combination that goes through the entire database and
recalculates checksums on each and every page. Only when all pages have
been checksummed are they fully enabled in the cluster. Any failure of
the process will revert to checksums off and the process has to be
started.
This adds a new WAL record that indicates the state of checksums, so
the process works across replicated clusters.
Authors: Magnus Hagander and Daniel Gustafsson
Review: Tomas Vondra, Michael Banck, Heikki Linnakangas, Andrey Borodin
When SSI was developed, slru.c was limited to segment files with names in
the range 0000-FFFF. This didn't allow enough space for predicate.c to
store every possible XID when spilling old transactions to disk, so it
would wrap around sooner and print warnings. Since commits 638cf09e and
73c986ad increased the number of segment files slru.c could manage, that
behavior is unnecessary. Therefore remove that code.
Also remove the macro OldSerXidSegment, which has been unused since
4cd3fb6e.
Thomas Munro, reviewed by Anastasia Lubennikova
Discussion: https://postgr.es/m/CAEepm=3XfsTSxgEbEOmxu0QDiXy0o18NUg2nC89JZcCGE+XFPA@mail.gmail.com
Predicate locks are used on per page basis only if fastupdate = off, in
opposite case predicate lock on pending list will effectively lock whole index,
to reduce locking overhead, just lock a relation. Entry and posting trees are
essentially B-tree, so locks are acquired on leaf pages only.
Author: Shubham Barai with some editorization by me and Dmitry Ivanov
Review by: Alexander Korotkov, Dmitry Ivanov, Fedor Sigaev
Discussion: https://www.postgresql.org/message-id/flat/CALxAEPt5sWW+EwTaKUGFL5_XFcZ0MuGBcyJ70oqbWqr42YKR8Q@mail.gmail.com
Just noticed that these were quite redundant, since we're holding the
page address in a local variable anyway, and we have pin on the buffer
throughout.
Also improve a comment.
FreeSpaceMapVacuumRange has the same effect, is more efficient if many
pages are involved, and makes fewer assumptions about how it's used.
Notably, Claudio Freire pointed out that UpdateFreeSpaceMap could fail
if the specified freespace value isn't the maximum possible. This isn't
a problem for the single existing user, but the function represents an
attractive nuisance IMO, because it's named as though it were a
general-purpose update function and its limitations are undocumented.
In any case we don't need multiple ways to get the same result.
In passing, do some code review and cleanup in RelationAddExtraBlocks.
In particular, I see no excuse for it to omit the PageIsNew safety check
that's done in the mainline extension path in RelationGetBufferForTuple.
Discussion: https://postgr.es/m/CAGTBQpYR0uJCNTt3M5GOzBRHo+-GccNO1nCaQ8yEJmZKSW5q1A@mail.gmail.com
VACUUM updates leaf-level FSM entries immediately after cleaning the
corresponding heap blocks. fsmpage.c updates the intra-page search trees
on the leaf-level FSM pages when this happens, but it does not touch the
upper-level FSM pages, so that the released space might not actually be
findable by searchers. Previously, updating the upper-level pages happened
only at the conclusion of the VACUUM run, in a single FreeSpaceMapVacuum()
call. This is bad because the VACUUM might get canceled before ever
reaching that point, so that from the point of view of searchers no space
has been freed at all, leading to table bloat.
We can improve matters by updating the upper pages immediately after each
cycle of index-cleaning and heap-cleaning, processing just the FSM pages
corresponding to the range of heap blocks we have now fully cleaned.
This adds a small amount of extra work, since the FSM pages leading down
to each range boundary will be touched twice, but it's pretty negligible
compared to everything else going on in a large VACUUM.
If there are no indexes, VACUUM doesn't work in cycles but just cleans
each heap page on first visit. In that case we just arbitrarily update
upper FSM pages after each 8GB of heap. That maintains the goal of not
letting all this work slide until the very end, and it doesn't seem worth
expending extra complexity on a case that so seldom occurs in practice.
In either case, the FSM is fully up to date before any attempt is made
to truncate the relation, so that the most likely scenario for VACUUM
cancellation no longer results in out-of-date upper FSM pages. When
we do successfully truncate, adjusting the FSM to reflect that is now
fully handled within FreeSpaceMapTruncateRel.
Claudio Freire, reviewed by Masahiko Sawada and Jing Wang, some additional
tweaks by me
Discussion: https://postgr.es/m/CAGTBQpYR0uJCNTt3M5GOzBRHo+-GccNO1nCaQ8yEJmZKSW5q1A@mail.gmail.com
Add page-level predicate locking, due to gist's code organization, patch seems
close to trivial: add check before page changing, add predicate lock before page
scanning. Although choosing right place to check is not simple: it should not
be called during index build, it should support insertion of new downlink and so
on.
Author: Shubham Barai with editorization by me and Alexander Korotkov
Reviewed by: Alexander Korotkov, Andrey Borodin, me
Discussion: https://www.postgresql.org/message-id/flat/CALxAEPtdcANpw5ePU3LvnTP8HCENFw6wygupQAyNBgD-sG3h0g@mail.gmail.com
For years, our makefiles have correctly observed that "there is no correct
way to write a rule that generates two files". However, what we did is to
provide empty rules that "generate" the secondary output files from the
primary one, and that's not right either. Depending on the details of
the creating process, the primary file might end up timestamped later than
one or more secondary files, causing subsequent make runs to consider the
secondary file(s) out of date. That's harmless in a plain build, since
make will just re-execute the empty rule and nothing happens. But it's
fatal in a VPATH build, since make will expect the secondary file to be
rebuilt in the build directory. This would manifest as "file not found"
failures during VPATH builds from tarballs, if we were ever unlucky enough
to ship a tarball with apparently out-of-date secondary files. (It's not
clear whether that has ever actually happened, but it definitely could.)
To ensure that secondary output files have timestamps >= their primary's,
change our makefile convention to be that we provide a "touch $@" action
not an empty rule. Also, make sure that this rule actually gets invoked
during a distprep run, else the hazard remains.
It's been like this a long time, so back-patch to all supported branches.
In HEAD, I skipped the changes in src/backend/catalog/Makefile, because
those rules are due to get replaced soon in the bootstrap data format
patch, and there seems no need to create a merge issue for that patch.
If for some reason we fail to land that patch in v11, we'll need to
back-fill the changes in that one makefile from v10.
Discussion: https://postgr.es/m/18556.1521668179@sss.pgh.pa.us
Commit 34db06ef9a adopted a lock-free
design for shm_mq.c, but it introduced a race condition that could
lose messages. When shm_mq_receive_bytes() detects that the other end
has detached, it must make sure that it has seen the final version of
mq_bytes_written, or it might miss a message sent before detaching.
Thomas Munro
Discussion: https://postgr.es/m/CAEepm%3D2myZ4qxpt1a%3DC%2BwEv3o188K13K3UvD-44FK0SdAzHy%2Bw%40mail.gmail.com
Instead of marking data from the ringer buffer consumed and setting the
sender's latch for every message, do it only when the amount of data we
can consume is at least 1/4 of the size of the ring buffer, or when no
data remains in the ring buffer. This is dramatically faster in my
testing; apparently, the savings from sending signals less frequently
outweighs the benefit of letting the sender know about available buffer
space sooner.
Patch by me, reviewed by Andres Freund and tested by Rafia Sabih.
Discussion: http://postgr.es/m/CA+TgmoYK7RFj6r7KLEfSGtYZCi3zqTRhAz8mcsDbUAjEmLOZ3Q@mail.gmail.com
Previously, mq_bytes_read and mq_bytes_written were protected by the
spinlock, but that turns out to cause pretty serious spinlock
contention on queries which send many tuples through a Gather or
Gather Merge node. This patches changes things so that we instead
read and write those values using 8-byte atomics. Since mq_bytes_read
can only be changed by the receiver and mq_bytes_written can only be
changed by the sender, the only purpose of the spinlock is to prevent
reads and writes of these values from being torn on platforms where
8-byte memory access is not atomic, making the conversion fairly
straightforward.
Testing shows that this produces some slowdown if we're using emulated
64-bit atomics, but since they should be available on any platform
where performance is a primary concern, that seems OK. It's faster,
sometimes a lot faster, on platforms where such atomics are available.
Patch by me, reviewed by Andres Freund, who also suggested the
design. Also tested by Rafia Sabih.
Discussion: http://postgr.es/m/CA+TgmoYuK0XXxmUNTFT9TSNiBtWnRwasBcHHRCOK9iYmDLQVPg@mail.gmail.com
elog(FATAL) would end up calling PortalCleanup(), which would call
executor shutdown code, which could fail and crash, especially under
parallel query. This was introduced by
8561e4840c, which did not want to mark an
active portal as failed by a normal transaction abort anymore. But we
do need to do that for an elog(FATAL) exit. Introduce a variable
shmem_exit_inprogress similar to the existing proc_exit_inprogress, so
we can tell whether we are in the FATAL exit scenario.
Reported-by: Andres Freund <andres@anarazel.de>
Seems a bit silly that many (in fact all, as of today) uses of
StaticAssertExpr would need to cast it to void to avoid warnings from
pickier compilers. Let's just do the cast right in the macro, instead.
In passing, change StaticAssertExpr to StaticAssertStmt in one
place where that seems more apropos.
Discussion: https://postgr.es/m/16161.1518715186@sss.pgh.pa.us
To make this work, tuplesort.c and logtape.c must also support
parallelism, so this patch adds that infrastructure and then applies
it to the particular case of parallel btree index builds. Testing
to date shows that this can often be 2-3x faster than a serial
index build.
The model for deciding how many workers to use is fairly primitive
at present, but it's better than not having the feature. We can
refine it as we get more experience.
Peter Geoghegan with some help from Rushabh Lathia. While Heikki
Linnakangas is not an author of this patch, he wrote other patches
without which this feature would not have been possible, and
therefore the release notes should possibly credit him as an author
of this feature. Reviewed by Claudio Freire, Heikki Linnakangas,
Thomas Munro, Tels, Amit Kapila, me.
Discussion: http://postgr.es/m/CAM3SWZQKM=Pzc=CAHzRixKjp2eO5Q0Jg1SoFQqeXFQ647JiwqQ@mail.gmail.com
Discussion: http://postgr.es/m/CAH2-Wz=AxWqDoVvGU7dq856S4r6sJAj6DBn7VMtigkB33N5eyg@mail.gmail.com
It seems incorrect to assume that the list of CkptSortItems can never
contain duplicate page numbers: concurrent activity could result in some
page getting dropped from a low-numbered buffer and later loaded into a
high-numbered buffer while BufferSync is scanning the buffer pool.
If that happened, the comparator would give self-inconsistent results,
potentially confusing qsort(). Saving one comparison step is not worth
possibly getting the sort wrong.
So far as I can tell, nothing would actually go wrong given our current
implementation of qsort(). It might get a bit slower than expected
if there were a large number of duplicates of one value, but that's
surely a probability-epsilon case. Still, the comment is wrong,
and if we ever switched to another sort implementation it might be
less forgiving.
In passing, avoid casting away const-ness of the argument pointers;
I've not seen any compiler complaints from that, but it seems likely
that some compilers would not like it.
Back-patch to 9.6 where this code came in, just in case I've underestimated
the possible consequences.
Discussion: https://postgr.es/m/18437.1515607610@sss.pgh.pa.us
The general assumption for postmaster child processes is that they
should just exit(1), reasonably promptly, if the postmaster disappears.
condition_variable.c neglected this consideration and could be left
waiting forever, if the counterpart process it is waiting for has
done the right thing and exited.
We had some discussion of adjusting the WaitEventSet API to make it
harder to make this type of mistake in future; but for the moment,
and for v10, let's make this narrow fix.
Discussion: https://postgr.es/m/20412.1515456143@sss.pgh.pa.us
The original coding here insisted that callers manually cancel any prepared
sleep for one condition variable before starting a sleep on another one.
While that's not a huge burden today, it seems like a gotcha that will bite
us in future if the use of condition variables increases; anything we can
do to make the use of this API simpler and more robust is attractive.
Hence, allow these functions to automatically switch their attention to
a different CV when required. This is safe for the same reason it was OK
for commit aced5a92b to let a broadcast operation cancel any prepared CV
sleep: whenever we return to the other test-and-sleep loop, we will
automatically re-prepare that CV, paying at most an extra test of that
loop's exit condition.
Back-patch to v10 where condition variables were introduced. Ordinarily
we would probably not back-patch a change like this, but since it does not
invalidate any coding pattern that was legal before, it seems safe enough.
Furthermore, there's an open bug in replorigin_drop() for which the
simplest fix requires this. Even if we chose to fix that in some more
complicated way, the hazard would remain that we might back-patch some
other bug fix that requires this behavior.
Patch by me, reviewed by Thomas Munro.
Discussion: https://postgr.es/m/2437.1515368316@sss.pgh.pa.us
Previously, although the initial state of a proclist_node is expected
to be next == prev == 0, proclist_delete_offset would reset nodes to
next == prev == INVALID_PGPROCNO when removing them from a list.
This is the same state that a node in a singleton list has, so that
it's impossible to distinguish not-in-a-list from in-a-list. Change
proclist_delete_offset to reset removed nodes to next == prev == 0,
making it possible to distinguish those cases, and then add Asserts
to the list add and delete functions that the supplied node isn't
or is in a list at entry. Also tighten assertions about the node
being in the particular list (not some other one) where it is possible
to check that in O(1) time.
In ConditionVariablePrepareToSleep, since we don't expect the process's
cvWaitLink to already be in a list, remove the more-or-less-useless
proclist_contains check; we'd rather have proclist_push_tail's new
assertion fire if that happens.
Improve various comments related to proclists, too.
Patch by me, reviewed by Thomas Munro. This isn't back-patchable, since
there could theoretically be inlined copies of proclist_delete_offset in
third-party modules. But it's only improving debuggability anyway.
Discussion: https://postgr.es/m/CAEepm=0NWKehYw7NDoUSf8juuKOPRnCyY3vuaSvhrEWsOTAa3w@mail.gmail.com
In commit 561885db0, as part of normalizing RemovePgTempFiles's error
handling, I removed its behavior of silently ignoring ENOENT failures
during directory opens. Thomas Munro points out that this is a bad idea at
the top level, because we don't create pgsql_tmp directories until needed.
Thus this coding could produce LOG messages in perfectly normal situations,
which isn't what I intended. Restore the suppression of ENOENT logging,
but only at top level --- it would still be unexpected for a nested temp
directory to disappear between seeing it in the parent directory and
opening it.
Discussion: https://postgr.es/m/CAEepm=2y06SehAkTnd5sU_eVqdv5P-=Srt1y5vYNQk6yVDVaPw@mail.gmail.com
In the wake of commit aced5a92b, the semantics of these results are
a bit squishy: we can tell whether we signaled some other process(es),
but we do not know which ones were real waiters versus mere sentinels
for ConditionVariableBroadcast operations. It does not help much that
ConditionVariableBroadcast will attempt to pass on the signal to the
next real waiter, because (a) there might not be one, and (b) that will
only happen awhile later, anyway. So these results could overstate how
much effect the calls really had.
However, no existing caller of either function pays any attention to its
result value, so it seems reasonable to just define that as a required
property of a correct algorithm. To encourage correctness and save some
tiny number of cycles, change both functions to return void.
Patch by me, per an observation by Thomas Munro. No back-patch, since
if any third parties happen to be using these functions, they might not
appreciate an API break in a minor release.
Discussion: https://postgr.es/m/CAEepm=0NWKehYw7NDoUSf8juuKOPRnCyY3vuaSvhrEWsOTAa3w@mail.gmail.com
In the admittedly-very-unlikely case that AddWaitEventToSet fails,
ConditionVariablePrepareToSleep would error out after already having
set cv_sleep_target, which is probably bad, and after having already
set cv_wait_event_set, which is very bad. Transaction abort might or
might not clean up cv_sleep_target properly; but there is nothing
that would be aware that the WaitEventSet wasn't fully constructed,
so that all future condition variable sleeps would be broken.
We can easily guard against these hazards with slight restructuring.
Back-patch to v10 where condition_variable.c was introduced.
Discussion: https://postgr.es/m/CAEepm=0NWKehYw7NDoUSf8juuKOPRnCyY3vuaSvhrEWsOTAa3w@mail.gmail.com
The original implementation of ConditionVariableBroadcast was, per its
self-description, "the dumbest way possible". Thomas Munro found out
it was a bit too dumb. An awakened process may immediately re-queue
itself, if the specific condition it's waiting for is not yet satisfied.
If this happens before ConditionVariableBroadcast is able to see the wait
queue as empty, then ConditionVariableBroadcast will re-awaken the same
process, repeating the cycle. Given unlucky timing this back-and-forth
can repeat indefinitely; loops lasting thousands of seconds have been
seen in testing.
To fix, add our own process to the end of the wait queue to serve as a
sentinel, and exit the broadcast loop once our process is not there
anymore. There are various special considerations described in the
comments, the principal disadvantage being that wakers can no longer
be sure whether they awakened a real waiter or just a sentinel. But in
practice nobody pays attention to the result of ConditionVariableSignal
or ConditionVariableBroadcast anyway, so that problem seems hypothetical.
Back-patch to v10 where condition_variable.c was introduced.
Tom Lane and Thomas Munro
Discussion: https://postgr.es/m/CAEepm=0NWKehYw7NDoUSf8juuKOPRnCyY3vuaSvhrEWsOTAa3w@mail.gmail.com
XactLockTableWait assumed that its xid argument has already added itself
to the lock table. That assumption led to another assumption that if
locking the xid has succeeded but the xid is reported as still in
progress, then the input xid must have been a subtransaction.
These assumptions hold true for the original uses of this code in
locking related to on-disk tuples, but they break down in logical
replication slot snapshot building -- in particular, when a standby
snapshot logged contains an xid that's already in ProcArray but not yet
in the lock table. This leads to assertion failures that can be
reproduced all the way back to 9.4, when logical decoding was
introduced.
To fix, change SubTransGetParent to SubTransGetTopmostTransaction which
has a slightly different API: it returns the argument Xid if there is no
parent, and it goes all the way to the top instead of moving up the
levels one by one. Also, to avoid busy-waiting, add a 1ms sleep to give
the other process time to register itself in the lock table.
For consistency, change ConditionalXactLockTableWait the same way.
Author: Petr Jelínek
Discussion: https://postgr.es/m/1B3E32D8-FCF4-40B4-AEF9-5C0E3AC57969@postgrespro.ru
Reported-by: Konstantin Knizhnik
Diagnosed-by: Stas Kelvich, Petr Jelínek
Reviewed-by: Andres Freund, Robert Haas
SharedTuplestore allows multiple participants to write into it and
then read the tuples back from it in parallel. Each reader receives
partial results.
For now it always uses disk files, but other buffering policies and
other kinds of scans (ie each reader receives complete results) may be
useful in future.
The upcoming parallel hash join feature will use this facility.
Author: Thomas Munro
Reviewed-By: Peter Geoghegan, Andres Freund, Robert Haas
Discussion: https://postgr.es/m/CAEepm=2W=cOkiZxcg6qiFQP-dHUe09aqTrEMM7yJDrHMhDv_RA@mail.gmail.com
Crash restarts currently don't clean up temporary files, as a debugging aid.
If a left-over file happens to have the same name as a segment file we're
trying to create, we'll just truncate and reuse it, but there is a problem:
BufFileOpenShared() determines how many segment files exist by trying to open
.0, .1, .2, ... until it finds no more files. It might be confused by a junk
file that has the next segment number. To defend against that, make sure we
always create a gap after the end file by unlinking the following name if it
exists. Also make it an error to try to open a BufFile that doesn't exist
(has no segment 0), so as not to encourage the development of client code
that depends on an interface that we can't reliably provide.
Author: Thomas Munro
Reviewed-By: Andres Freund
Discussion: https://postgr.es/m/CAEepm%3D2jhCbC_GFQJaaDhWxLB4EXtT3vVd5czuRNaqF5CWSTog%40mail.gmail.com
When we create an Append node, we can spread out the workers over the
subplans instead of piling on to each subplan one at a time, which
should typically be a bit more efficient, both because the startup
cost of any plan executed entirely by one worker is paid only once and
also because of reduced contention. We can also construct Append
plans using a mix of partial and non-partial subplans, which may allow
for parallelism in places that otherwise couldn't support it.
Unfortunately, this patch doesn't handle the important case of
parallelizing UNION ALL by running each branch in a separate worker;
the executor infrastructure is added here, but more planner work is
needed.
Amit Khandekar, Robert Haas, Amul Sul, reviewed and tested by
Ashutosh Bapat, Amit Langote, Rafia Sabih, Amit Kapila, and
Rajkumar Raghuwanshi.
Discussion: http://postgr.es/m/CAJ3gD9dy0K_E8r727heqXoBmWZ83HwLFwdcaSSmBQ1+S+vRuUQ@mail.gmail.com
Previously, this code just reported such problems at LOG level and kept
going. The problem with this approach is that transient failures (e.g.,
ENFILE) could prevent us from resetting unlogged relations to empty,
yet allow recovery to appear to complete successfully. That seems like
a data corruption hazard large enough to treat such problems as reasons
to fail startup.
For the same reason, treat unlink failures for unlogged files as hard
errors not just LOG messages. It's a little odd that we did it like that
when file-level errors in other steps (copy_file, fsync_fname) are ERRORs.
The sole case that I left alone is that ENOENT failure on a tablespace
(not database) directory is not an error, though it will now be logged
rather than just silently ignored. This is to cover the scenario where
a previous DROP TABLESPACE removed the tablespace directory but failed
before removing the pg_tblspc symlink. I'm not sure that that's very
likely in practice, but that seems like the only real excuse for the
old behavior here, so let's allow for it. (As coded, this will also
allow ENOENT on $PGDATA/base/. But since we'll fail soon enough if
that's gone, I don't think we need to complicate this code by
distinguishing that from a true tablespace case.)
Discussion: https://postgr.es/m/21040.1512418508@sss.pgh.pa.us
Modify this function and its subsidiaries so that syscall failures are
reported via ereport(LOG), rather than silently ignored as before.
We don't want to throw a hard ERROR, as that would prevent database
startup, and getting rid of leftover temporary files is not important
enough for that. On the other hand, not reporting trouble at all
seems like an odd choice not in line with current project norms,
especially since any failure here is quite unexpected.
On the same reasoning, adjust these functions' AllocateDir/ReadDir calls
so that failure to scan a directory results in LOG not ERROR. I also
removed the previous practice of silently ignoring ENOENT failures during
directory opens --- there are some corner cases where that could happen
given a previous database crash, but that seems like a bad excuse for
ignoring a condition that isn't expected in most cases. A LOG message
during postmaster start seems OK in such situations, and better than
no output at all.
In passing, make RemovePgTempRelationFiles' test for "is the file name
all digits" look more like the way it's done elsewhere.
Discussion: https://postgr.es/m/19907.1512402254@sss.pgh.pa.us
This patch fixes a couple of low-probability bugs that could lead to
reporting an irrelevant errno value (and hence possibly a wrong SQLSTATE)
concerning directory-open or file-open failures. It also fixes places
where we took shortcuts in reporting such errors, either by using elog
instead of ereport or by using ereport but forgetting to specify an
errcode. And it eliminates a lot of just plain redundant error-handling
code.
In service of all this, export fd.c's formerly-static function
ReadDirExtended, so that external callers can make use of the coding
pattern
dir = AllocateDir(path);
while ((de = ReadDirExtended(dir, path, LOG)) != NULL)
if they'd like to treat directory-open failures as mere LOG conditions
rather than errors. Also fix FreeDir to be a no-op if we reach it
with dir == NULL, as such a coding pattern would cause.
Then, remove code at many call sites that was throwing an error or log
message for AllocateDir failure, as ReadDir or ReadDirExtended can handle
that job just fine. Aside from being a net code savings, this gets rid of
a lot of not-quite-up-to-snuff reports, as mentioned above. (In some
places these changes result in replacing a custom error message such as
"could not open tablespace directory" with more generic wording "could not
open directory", but it was agreed that the custom wording buys little as
long as we report the directory name.) In some other call sites where we
can't just remove code, change the error reports to be fully
project-style-compliant.
Also reorder code in restoreTwoPhaseData that was acquiring a lock
between AllocateDir and ReadDir; in the unlikely but surely not
impossible case that LWLockAcquire changes errno, AllocateDir failures
would be misreported. There is no great value in opening the directory
before acquiring TwoPhaseStateLock, so just do it in the other order.
Also fix CheckXLogRemoved to guarantee that it preserves errno,
as quite a number of call sites are implicitly assuming. (Again,
it's unlikely but I think not impossible that errno could change
during a SpinLockAcquire. If so, this function was broken for its
own purposes as well as breaking callers.)
And change a few places that were using not-per-project-style messages,
such as "could not read directory" when "could not open directory" is
more correct.
Back-patch the exporting of ReadDirExtended, in case we have occasion
to back-patch some fix that makes use of it; it's not needed right now
but surely making it global is pretty harmless. Also back-patch the
restoreTwoPhaseData and CheckXLogRemoved fixes. The rest of this is
essentially cosmetic and need not get back-patched.
Michael Paquier, with a bit of additional work by me
Discussion: https://postgr.es/m/CAB7nPqRpOCxjiirHmebEFhXVTK7V5Jvw4bz82p7Oimtsm3TyZA@mail.gmail.com
Provide support for dynamic or static parties of processes to wait for
all processes to reach point in the code before continuing.
This is similar to the mechanism of the same name in POSIX threads and
MPI, though has explicit phasing and dynamic party support like the
Java core library's Phaser.
This will be used by an upcoming patch adding support for parallel
hash joins.
Author: Thomas Munro
Reviewed-By: Andres Freund
Discussion: https://postgr.es/m/CAEepm=2_y7oi01OjA_wLvYcWMc9_d=LaoxrY3eiROCZkB_qakA@mail.gmail.com
Commit 11e264517 removed BufFile's isTemp flag, thereby eliminating
the possibility of resurrecting BufFileCreate(). But it left that
function in place, as well as a bunch of comments describing how things
worked for the non-temp-file case. At best, that's now a source of
confusion. So remove the long-since-commented-out function and change
relevant comments.
I (tgl) wanted to rename BufFileCreateTemp() to BufFileCreate(), but
that seems not to be the consensus position, so leave it as-is.
In passing, fix commit f0828b2fc's failure to update BufFileSeek's
comment to match the change of its argument type from long to off_t.
(I think that might actually have been intentional at the time, but
now that 64-bit off_t is nearly universal, it looks anachronistic.)
Thomas Munro and Tom Lane
Discussion: https://postgr.es/m/E1eFVyl-0008J1-RO@gemulon.postgresql.org
Up to now, ACL checks for large objects happened at the level of
the SQL-callable functions, which led to CVE-2017-7548 because of a
missing check. Push them down to be enforced in inv_api.c as much
as possible, in hopes of preventing future bugs. This does have the
effect of moving read and write permission errors to happen at lo_open
time not loread or lowrite time, but that seems acceptable.
Michael Paquier and Tom Lane
Discussion: https://postgr.es/m/CAB7nPqRHmNOYbETnc_2EjsuzSM00Z+BWKv9sy6tnvSd5gWT_JA@mail.gmail.com
The point of having separate ResourceOwnerEnlargeFoo and
ResourceOwnerRememberFoo functions is so that resource allocation
can happen in between. Doing it in some other order is just wrong.
OpenTemporaryFile() did open(), enlarge, remember, which would leak the
open file if the enlarge step ran out of memory. Because fd.c has its own
layer of resource-remembering, the consequences look like they'd be limited
to an intratransaction FD leak, but it's still not good.
IncrBufferRefCount() did enlarge, remember, incr-refcount, which would blow
up if the incr-refcount step ever failed. It was safe enough when written,
but since the introduction of PrivateRefCountHash, I think the assumption
that no error could happen there is pretty shaky.
The odds of real problems from either bug are probably small, but still,
back-patch to supported branches.
Thomas Munro and Tom Lane, per a comment from Andres Freund
The lower case spellings are C and C++ standard and are used in most
parts of the PostgreSQL sources. The upper case spellings are only used
in some files/modules. So standardize on the standard spellings.
The APIs for ICU, Perl, and Windows define their own TRUE and FALSE, so
those are left as is when using those APIs.
In code comments, we use the lower-case spelling for the C concepts and
keep the upper-case spelling for the SQL concepts.
Reviewed-by: Michael Paquier <michael.paquier@gmail.com>
There doesn't seem to be any good reason to do the filling of the
itemidbase[] array separately from the first traversal of the pointers.
It's certainly not a win if there are any line pointers with storage,
and even if there aren't, this change doesn't insert code into the part
of the first loop that will be traversed in that case. So let's just
merge the two loops.
Yura Sokolov, reviewed by Claudio Freire
Discussion: https://postgr.es/m/e49befcc6f1d7099834c6fdf5c675a60@postgrespro.ru
resowner/README contained advice to use a PG_TRY block to restore the
old CurrentResourceOwner value anywhere that that variable is transiently
changed. That advice was only inconsistently followed, however, and
on reflection it seems like unnecessary overhead. We don't bother
with such a convention for transient CurrentMemoryContext changes,
on the grounds that any (sub)transaction abort will start out by
resetting CurrentMemoryContext to what it wants. But the same is
true of CurrentResourceOwner, so there seems no need to treat it
differently.
Hence, remove PG_TRY blocks that exist only to restore CurrentResourceOwner
before re-throwing the error. There are a couple of places that restore
it along with some other actions, and I left those alone; the restore is
probably unnecessary but no noticeable gain will result from removing it.
Discussion: https://postgr.es/m/5236.1507583529@sss.pgh.pa.us
copy_file() reads and writes data 64KB at a time (with default BLCKSZ),
and historically has issued a pg_flush_data request after each write.
This turns out to interact really badly with macOS's new APFS file
system: a large file copy takes over 100X longer than it ought to on
APFS, as reported by Brent Dearth. While that's arguably a macOS bug,
it's not clear whether Apple will do anything about it in the near
future, and in any case experimentation suggests that issuing flushes
a bit less often can be helpful on other platforms too.
Hence, rearrange the logic in copy_file() so that flush requests are
issued once per N writes rather than every time through the loop.
I set the FLUSH_DISTANCE to 32MB on macOS (any less than that still
results in a noticeable speed degradation on APFS), but 1MB elsewhere.
In limited testing on Linux and FreeBSD, this seems slightly faster
than the previous code, and certainly no worse. It helps noticeably
on macOS even with the older HFS filesystem.
A simpler change would have been to just increase the size of the
copy buffer without changing the loop logic, but that seems likely
to trash the processor cache without really helping much.
Back-patch to 9.6 where we introduced msync() as an implementation
option for pg_flush_data(). The problem seems specific to APFS's
mmap/msync support, so I don't think we need to go further back.
Discussion: https://postgr.es/m/CADkxhTNv-j2jw2g8H57deMeAbfRgYBoLmVuXkC=YCFBXRuCOww@mail.gmail.com
posix_fallocate() is not quite a drop-in replacement for fallocate(),
because it is defined to return the error code as its function result,
not in "errno". I (tgl) missed this because RHEL6's version seems
to set errno as well. That is not the case on more modern Linuxen,
though, as per buildfarm results.
Aside from fixing the return-convention confusion, remove the test
for ENOSYS; we expect that glibc will mask that for posix_fallocate,
though it does not for fallocate. Keep the test for EINTR, because
POSIX specifies that as a possible result, and buildfarm results
suggest that it can happen in practice.
Back-patch to 9.4, like the previous commit.
Thomas Munro
Discussion: https://postgr.es/m/1002664500.12301802.1471008223422.JavaMail.yahoo@mail.yahoo.com
On Linux, shared memory segments created with shm_open() are backed by
swap files created in tmpfs. If the swap file needs to be extended,
but there's no tmpfs space left, you get a very unfriendly SIGBUS trap.
To avoid this, force allocation of the full request size when we create
the segment. This adds a few cycles, but none that we wouldn't expend
later anyway, assuming the request isn't hugely bigger than the actual
need.
Make this code #ifdef __linux__, because (a) there's not currently a
reason to think the same problem exists on other platforms, and (b)
applying posix_fallocate() to an FD created by shm_open() isn't very
portable anyway.
Back-patch to 9.4 where the DSM code came in.
Thomas Munro, per a bug report from Amul Sul
Discussion: https://postgr.es/m/1002664500.12301802.1471008223422.JavaMail.yahoo@mail.yahoo.com
The file handling functions from fd.c were called with a diverse mix of
notations for the file permissions when they were opening new files.
Almost all files created by the server should have the same permissions
set. So change the API so that e.g. OpenTransientFile() automatically
uses the standard permissions set, and OpenTransientFilePerm() is a new
function that takes an explicit permissions set for the few cases where
it is needed. This also saves an unnecessary argument for call sites
that are just opening an existing file.
While we're reviewing these APIs, get rid of the FileName typedef and
use the standard const char * for the file name and mode_t for the file
mode. This makes these functions match other file handling functions
and removes an unnecessary layer of mysteriousness. We can also get rid
of a few casts that way.
Author: David Steele <david@pgmasters.net>
Tuples can have type RECORDOID and a typmod number that identifies a blessed
TupleDesc in a backend-private cache. To support the sharing of such tuples
through shared memory and temporary files, provide a typmod registry in
shared memory.
To achieve that, introduce per-session DSM segments, created on demand when a
backend first runs a parallel query. The per-session DSM segment has a
table-of-contents just like the per-query DSM segment, and initially the
contents are a shared record typmod registry and a DSA area to provide the
space it needs to grow.
State relating to the current session is accessed via a Session object
reached through global variable CurrentSession that may require significant
redesign further down the road as we figure out what else needs to be shared
or remodelled.
Author: Thomas Munro
Reviewed-By: Andres Freund
Discussion: https://postgr.es/m/CAEepm=0ZtQ-SpsgCyzzYpsXS6e=kZWqk3g5Ygn3MDV7A8dabUA@mail.gmail.com
It is equivalent in ANSI C to write (*funcptr) () and funcptr(). These
two styles have been applied inconsistently. After discussion, we'll
use the more verbose style for plain function pointer variables, to make
it clear that it's a variable, and the shorter style when the function
pointer is in a struct (s.func() or s->func()), because then it's clear
that it's not a plain function name, and otherwise the excessive
punctuation makes some of those invocations hard to read.
Discussion: https://www.postgresql.org/message-id/f52c16db-14ed-757d-4b48-7ef360b1631d@2ndquadrant.com
Commit 0e141c0fbb introduced a mechanism
to reduce contention on ProcArrayLock by having a single process clear
XIDs in the procArray on behalf of multiple processes, reducing the
need to hand the lock around. A previous attempt to introduce a similar
mechanism for CLogControlLock in ccce90b398
crashed and burned, but the design problem which resulted in those
failures is believed to have been corrected in this version.
Amit Kapila, with some cosmetic changes by me. See the previous commit
message for additional credits.
Discussion: http://postgr.es/m/CAA4eK1KudxzgWhuywY_X=yeSAhJMT4DwCjroV5Ay60xaeB2Eew@mail.gmail.com
The logic around shm_mq_detach was a few bricks shy of a load, because
(contrary to the comments for shm_mq_attach) all it did was update the
shared shm_mq state. That left us leaking a bit of process-local
memory, but much worse, the on_dsm_detach callback for shm_mq_detach
was still armed. That means that whenever we ultimately detach from
the DSM segment, we'd run shm_mq_detach again for already-detached,
possibly long-dead queues. This accidentally fails to fail today,
because we only ever re-use a shm_mq's memory for another shm_mq, and
multiple detach attempts on the last such shm_mq are fairly harmless.
But it's gonna bite us someday, so let's clean it up.
To do that, change shm_mq_detach's API so it takes a shm_mq_handle
not the underlying shm_mq. This makes the callers simpler in most
cases anyway. Also fix a few places in parallel.c that were just
pfree'ing the handle structs rather than doing proper cleanup.
Back-patch to v10 because of the risk that the revenant shm_mq_detach
callbacks would cause a live bug sometime. Since this is an API
change, it's too late to do it in 9.6. (We could make a variant
patch that preserves API, but I'm not excited enough to do that.)
Discussion: https://postgr.es/m/8670.1504192177@sss.pgh.pa.us
Periodically while the server is running, and at shutdown, write out a
list of blocks in shared buffers. When the server reaches consistency
-- unfortunatey, we can't do it before that point without breaking
things -- reload those blocks into any still-unused shared buffers.
Mithun Cy and Robert Haas, reviewed and tested by Beena Emerson,
Amit Kapila, Jim Nasby, and Rafia Sabih.
Discussion: http://postgr.es/m/CAD__OugubOs1Vy7kgF6xTjmEqTR4CrGAv8w+ZbaY_+MZeitukw@mail.gmail.com
Previously, if you passed a non-aligned size to shm_toc_create(), the
memory returned by shm_toc_allocate() would be similarly non-aligned.
This was exposed by commit 3cda10f41b, which allocated structs containing
a pg_atomic_uint64 field with shm_toc_allocate(). On systems with
MAXIMUM_ALIGNOF = 4, such structs still need to be 8-bytes aligned, but
the memory returned by shm_toc_allocate() was only 4-bytes aligned.
It's quite bogus that we abuse BUFFERALIGN to align the structs for
pg_atomic_uint64. It doesn't really have anything to do with buffers. But
that's a separate issue.
This ought to fix the buildfarm failures on 32-bit x86 systems.
Discussion: https://www.postgresql.org/message-id/7e0a73a5-0df9-1859-b8ae-9acf122dc38d@iki.fi
Although not confirmed and probably rare, if the newly allocated memory
is not already zero, this could possibly have caused some problems.
Also reorder the initializations slightly so they match the order of the
struct definition.
Author: Wong, Yi Wen <yiwong@amazon.com>
Reviewed-by: Masahiko Sawada <sawada.mshk@gmail.com>
The API for WaitLatch and friends followed the Unix convention in which
waiting for a socket connection to complete is identical to waiting for
the socket to accept a write. While Windows provides a select(2)
emulation that agrees with that, the native WaitForMultipleObjects API
treats them as quite different --- and for some bizarre reason, it will
report a not-yet-connected socket as write-ready. libpq itself has so
far escaped dealing with this because it waits with select(), but in
libpqwalreceiver.c we want to wait using WaitLatchOrSocket. The semantics
mismatch resulted in replication connection failures on Windows, but only
for remote connections (apparently, localhost connections complete
immediately, or at least too fast for anyone to have noticed the problem
in single-machine testing).
To fix, introduce an additional WL_SOCKET_CONNECTED wait flag for
WaitLatchOrSocket, which is identical to WL_SOCKET_WRITEABLE on
non-Windows, but results in waiting for FD_CONNECT events on Windows.
Ideally, we would also distinguish the two conditions in the API for
PQconnectPoll(), but changing that API at this point seems infeasible.
Instead, cheat by checking for PQstatus() == CONNECTION_STARTED to
determine that we're still waiting for the connection to complete.
(This is a cheat mainly because CONNECTION_STARTED is documented as an
internal state rather than something callers should rely on. Perhaps
we ought to change the documentation ... but this patch doesn't.)
Per reports from Jobin Augustine and Igor Neyman. Back-patch to v10
where commit 1e8a85009 exposed this longstanding shortcoming.
Andres Freund, minor fix and some code review/beautification by me
Discussion: https://postgr.es/m/CAHBggj8g2T+ZDcACZ2FmzX9CTxkWjKBsHd6NkYB4i9Ojf6K1Fw@mail.gmail.com
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
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
Don't move parenthesized lines to the left, even if that means they
flow past the right margin.
By default, BSD indent lines up statement continuation lines that are
within parentheses so that they start just to the right of the preceding
left parenthesis. However, traditionally, if that resulted in the
continuation line extending to the right of the desired right margin,
then indent would push it left just far enough to not overrun the margin,
if it could do so without making the continuation line start to the left of
the current statement indent. That makes for a weird mix of indentations
unless one has been completely rigid about never violating the 80-column
limit.
This behavior has been pretty universally panned by Postgres developers.
Hence, disable it with indent's new -lpl switch, so that parenthesized
lines are always lined up with the preceding left paren.
This patch is much less interesting than the first round of indent
changes, but also bulkier, so I thought it best to separate the effects.
Discussion: https://postgr.es/m/E1dAmxK-0006EE-1r@gemulon.postgresql.org
Discussion: https://postgr.es/m/30527.1495162840@sss.pgh.pa.us
Change pg_bsd_indent to follow upstream rules for placement of comments
to the right of code, and remove pgindent hack that caused comments
following #endif to not obey the general rule.
Commit e3860ffa4d wasn't actually using
the published version of pg_bsd_indent, but a hacked-up version that
tried to minimize the amount of movement of comments to the right of
code. The situation of interest is where such a comment has to be
moved to the right of its default placement at column 33 because there's
code there. BSD indent has always moved right in units of tab stops
in such cases --- but in the previous incarnation, indent was working
in 8-space tab stops, while now it knows we use 4-space tabs. So the
net result is that in about half the cases, such comments are placed
one tab stop left of before. This is better all around: it leaves
more room on the line for comment text, and it means that in such
cases the comment uniformly starts at the next 4-space tab stop after
the code, rather than sometimes one and sometimes two tabs after.
Also, ensure that comments following #endif are indented the same
as comments following other preprocessor commands such as #else.
That inconsistency turns out to have been self-inflicted damage
from a poorly-thought-through post-indent "fixup" in pgindent.
This patch is much less interesting than the first round of indent
changes, but also bulkier, so I thought it best to separate the effects.
Discussion: https://postgr.es/m/E1dAmxK-0006EE-1r@gemulon.postgresql.org
Discussion: https://postgr.es/m/30527.1495162840@sss.pgh.pa.us
The new indent version includes numerous fixes thanks to Piotr Stefaniak.
The main changes visible in this commit are:
* Nicer formatting of function-pointer declarations.
* No longer unexpectedly removes spaces in expressions using casts,
sizeof, or offsetof.
* No longer wants to add a space in "struct structname *varname", as
well as some similar cases for const- or volatile-qualified pointers.
* Declarations using PG_USED_FOR_ASSERTS_ONLY are formatted more nicely.
* Fixes bug where comments following declarations were sometimes placed
with no space separating them from the code.
* Fixes some odd decisions for comments following case labels.
* Fixes some cases where comments following code were indented to less
than the expected column 33.
On the less good side, it now tends to put more whitespace around typedef
names that are not listed in typedefs.list. This might encourage us to
put more effort into typedef name collection; it's not really a bug in
indent itself.
There are more changes coming after this round, having to do with comment
indentation and alignment of lines appearing within parentheses. I wanted
to limit the size of the diffs to something that could be reviewed without
one's eyes completely glazing over, so it seemed better to split up the
changes as much as practical.
Discussion: https://postgr.es/m/E1dAmxK-0006EE-1r@gemulon.postgresql.org
Discussion: https://postgr.es/m/30527.1495162840@sss.pgh.pa.us
Previously we required every exported transaction to have an xid
assigned. That was used to check that the exporting transaction is
still running, which in turn is needed to guarantee that that
necessary rows haven't been removed in between exporting and importing
the snapshot.
The exported xid caused unnecessary problems with logical decoding,
because slot creation has to wait for all concurrent xid to finish,
which in turn serializes concurrent slot creation. It also
prohibited snapshots to be exported on hot-standby replicas.
Instead export the virtual transactionid, which avoids the unnecessary
serialization and the inability to export snapshots on standbys. This
changes the file name of the exported snapshot, but since we never
documented what that one means, that seems ok.
Author: Petr Jelinek, slightly editorialized by me
Reviewed-By: Andres Freund
Discussion: https://postgr.es/m/f598b4b8-8cd7-0d54-0939-adda763d8c34@2ndquadrant.com
The larger part of this patch replaces usages of MyProc->procLatch
with MyLatch. The latter works even early during backend startup,
where MyProc->procLatch doesn't yet. While the affected code
shouldn't run in cases where it's not initialized, it might get copied
into places where it might. Using MyLatch is simpler and a bit faster
to boot, so there's little point to stick with the previous coding.
While doing so I noticed some weaknesses around newly introduced uses
of latches that could lead to missed events, and an omitted
CHECK_FOR_INTERRUPTS() call in worker_spi.
As all the actual bugs are in v10 code, there doesn't seem to be
sufficient reason to backpatch this.
Author: Andres Freund
Discussion:
https://postgr.es/m/20170606195321.sjmenrfgl2nu6j63@alap3.anarazel.dehttps://postgr.es/m/20170606210405.sim3yl6vpudhmufo@alap3.anarazel.de
Backpatch: -
When the checkpointer writes the shutdown checkpoint, it checks
afterwards whether any WAL has been written since it started and
throws a PANIC if so. At that point, only walsenders are still
active, so one might think this could not happen, but walsenders can
also generate WAL, for instance in BASE_BACKUP and logical decoding
related commands (e.g. via hint bits). So they can trigger this panic
if such a command is run while the shutdown checkpoint is being
written.
To fix this, divide the walsender shutdown into two phases. First,
checkpointer, itself triggered by postmaster, sends a
PROCSIG_WALSND_INIT_STOPPING signal to all walsenders. If the backend
is idle or runs an SQL query this causes the backend to shutdown, if
logical replication is in progress all existing WAL records are
processed followed by a shutdown. Otherwise this causes the walsender
to switch to the "stopping" state. In this state, the walsender will
reject any further replication commands. The checkpointer begins the
shutdown checkpoint once all walsenders are confirmed as
stopping. When the shutdown checkpoint finishes, the postmaster sends
us SIGUSR2. This instructs walsender to send any outstanding WAL,
including the shutdown checkpoint record, wait for it to be replicated
to the standby, and then exit.
Author: Andres Freund, based on an earlier patch by Michael Paquier
Reported-By: Fujii Masao, Andres Freund
Reviewed-By: Michael Paquier
Discussion: https://postgr.es/m/20170602002912.tqlwn4gymzlxpvs2@alap3.anarazel.de
Backpatch: 9.4, where logical decoding was introduced
Otherwise code that uses this will abort with an assertion failure,
because postmaster_alive_fds are not initialized.
Reported-by: tushar <tushar.ahuja@enterprisedb.com>
Declare the toc_nentry field as uint32 not Size. Since shm_toc_lookup()
reads the field without any lock, it has to be atomically readable, and
we do not assume that for fields wider than 32 bits. Performance would
be impossibly bad for entry counts approaching 2^32 anyway, so there is
no need to try to preserve maximum width here.
This is probably an academic issue, because even if reading int64 isn't
atomic, the high order half would never change in practice. Still, it's
a coding rule violation, so let's fix it.
Adjust some other not-terribly-well-chosen data types too, and copy-edit
some comments. Make shm_toc_attach's Asserts consistent with
shm_toc_create's.
None of this looks to be a live bug, so no need for back-patch.
Discussion: https://postgr.es/m/16984.1496679541@sss.pgh.pa.us
Given the possibility of race conditions and so on, it seems entirely
unsafe to just assume that shm_toc_lookup() always finds the key it's
looking for --- but that was exactly what all but one call site were
doing. To fix, add a "bool noError" argument, similarly to what we
have in many other functions, and throw an error on an unexpected
lookup failure. Remove now-redundant Asserts that a rather random
subset of call sites had.
I doubt this will throw any light on buildfarm member lorikeet's
recent failures, because if an unnoticed lookup failure were involved,
you'd kind of expect a null-pointer-dereference crash rather than the
observed symptom. But you never know ... and this is better coding
practice even if it never catches anything.
Discussion: https://postgr.es/m/9697.1496675981@sss.pgh.pa.us
Commit fa31b6f4e supposed that we didn't have to worry about that
anymore, but it seems that RHEL5 is like that, and that's still
a supported platform. Put back the prior coding under an #ifdef,
adding an explicit fcntl() to retain the desired CLOEXEC property.
Discussion: https://postgr.es/m/12307.1493325329@sss.pgh.pa.us
The logical decoding machinery already preserved all the required
catalog tuples, which is sufficient in the course of normal logical
decoding, but did not guarantee that non-catalog tuples were preserved
during computation of the initial snapshot when creating a slot over
the replication protocol.
This could cause a corrupted initial snapshot being exported. The
time window for issues is usually not terribly large, but on a busy
server it's perfectly possible to it hit it. Ongoing decoding is not
affected by this bug.
To avoid increased overhead for the SQL API, only retain additional
tuples when a logical slot is being created over the replication
protocol. To do so this commit changes the signature of
CreateInitDecodingContext(), but it seems unlikely that it's being
used in an extension, so that's probably ok.
In a drive-by fix, fix handling of
ReplicationSlotsComputeRequiredXmin's already_locked argument, which
should only apply to ProcArrayLock, not ReplicationSlotControlLock.
Reported-By: Erik Rijkers
Analyzed-By: Petr Jelinek
Author: Petr Jelinek, heavily editorialized by Andres Freund
Reviewed-By: Andres Freund
Discussion: https://postgr.es/m/9a897b86-46e1-9915-ee4c-da02e4ff6a95@2ndquadrant.com
Backport: 9.4, where logical decoding was introduced.
Although the postmaster doesn't currently create a self-pipe or any
latches, there's discussion of it doing so in future. It's also
conceivable that a shared_preload_libraries extension would try to
create such a thing in the postmaster process today. In that case
the self-pipe FDs would be inherited by forked child processes.
latch.c was entirely unprepared for such a case and could suffer an
assertion failure, or worse try to use the inherited pipe if somebody
called WaitLatch without having called InitializeLatchSupport in that
process. Make it keep track of whether InitializeLatchSupport has been
called in the *current* process, and do the right thing if state has
been inherited from a parent.
Apply FD_CLOEXEC to file descriptors created in latch.c (the self-pipe,
as well as epoll event sets). This ensures that child processes spawned
in backends, the archiver, etc cannot accidentally or intentionally mess
with these FDs. It also ensures that we end up with the right state
for the self-pipe in EXEC_BACKEND processes, which otherwise wouldn't
know to close the postmaster's self-pipe FDs.
Back-patch to 9.6, mainly to keep latch.c looking similar in all branches
it exists in.
Discussion: https://postgr.es/m/8322.1493240739@sss.pgh.pa.us
poll.h is mandated by Single Unix Spec v2, the usual baseline for
postgres on unix. None of the unixoid buildfarms animals has
sys/poll.h but not poll.h. Therefore there's not much point to test
for sys/poll.h's existence and include it optionally.
Author: Andres Freund, per suggestion from Tom Lane
Discussion: https://postgr.es/m/20505.1492723662@sss.pgh.pa.us
poll(2) is required by Single Unix Spec v2, the usual baseline for
postgres (leaving windows aside). There's not been any buildfarm
animals without poll(2) for a long while, leaving the select(2)
implementation to be largely untested.
On windows, including mingw, poll() is not available, but we have a
special case implementation for windows anyway.
Author: Andres Freund
Discussion: https://postgr.es/m/20170420003611.7r2sdvehesdyiz2i@alap3.anarazel.de
The POSIX standard does not say that the success return value for
fcntl(F_SETFD) and fcntl(F_SETFL) is zero; it says only that it's not -1.
We had several calls that were making the stronger assumption. Adjust
them to test specifically for -1 for strict spec compliance.
The standard further leaves open the possibility that the O_NONBLOCK
flag bit is not the only active one in F_SETFL's argument. Formally,
therefore, one ought to get the current flags with F_GETFL and store
them back with only the O_NONBLOCK bit changed when trying to change
the nonblock state. In port/noblock.c, we were doing the full pushup
in pg_set_block but not in pg_set_noblock, which is just weird. Make
both of them do it properly, since they have little business making
any assumptions about the socket they're handed. The other places
where we're issuing F_SETFL are working with FDs we just got from
pipe(2), so it's reasonable to assume the FDs' properties are all
default, so I didn't bother adding F_GETFL steps there.
Also, while pg_set_block deserves some points for trying to do things
right, somebody had decided that it'd be even better to cast fcntl's
third argument to "long". Which is completely loony, because POSIX
clearly says the third argument for an F_SETFL call is "int".
Given the lack of field complaints, these missteps apparently are not
of significance on any common platforms. But they're still wrong,
so back-patch to all supported branches.
Discussion: https://postgr.es/m/30882.1492800880@sss.pgh.pa.us
Commit 4deb41381 modified isolationtester's query to see whether a
session is blocked to also check for waits occurring in GetSafeSnapshot.
However, it did that in a way that enormously increased the query's
runtime under CLOBBER_CACHE_ALWAYS, causing the buildfarm members
that use that to run about four times slower than before, and in some
cases fail entirely. To fix, push the entire logic into a dedicated
backend function. This should actually reduce the CLOBBER_CACHE_ALWAYS
runtime from what it was previously, though I've not checked that.
In passing, expose a SQL function to check for safe-snapshot blockage,
comparable to pg_blocking_pids. This is more or less free given the
infrastructure built to solve the other problem, so we might as well.
Thomas Munro
Discussion: https://postgr.es/m/20170407165749.pstcakbc637opkax@alap3.anarazel.de
Defaults match the fixed behavior of prior releases, but now DBAs
have better options to tune serializable workloads.
It might be nice to be able to set this per relation, but that part
will need to wait for another release.
Author: Dagfinn Ilmari Mannsåker
dsm_create and dsm_attach assumed that a current resource owner was
always in place. Exploration with the API show that this is
inconvenient: sometimes one must create a dummy resowner, create/attach
the DSM, only to pin the mapping later, which is wasteful. Change
create/attach so that if there is no current resowner, the dsm is
effectively pinned right from the start.
Discussion: https://postgr.es/m/20170324232710.32acsfsvjqfgc6ud@alvherre.pgsql
Reviewed by Thomas Munro.
There are no functional changes here; this simply encapsulates knowledge
of the ItemPointerData struct so that a future patch can change things
without more breakage.
All direct users of ip_blkid and ip_posid are changed to use existing
macros ItemPointerGetBlockNumber and ItemPointerGetOffsetNumber
respectively. For callers where that's inappropriate (because they
Assert that the itempointer is is valid-looking), add
ItemPointerGetBlockNumberNoCheck and ItemPointerGetOffsetNumberNoCheck,
which lack the assertion but are otherwise identical.
Author: Pavan Deolasee
Discussion: https://postgr.es/m/CABOikdNnFon4cJiL=h1mZH3bgUeU+sWHuU4Yr8AB=j3A2p1GiA@mail.gmail.com
If file was created/deleted just before powerloss it's possible that
file system will miss that. To prevent it, call fsync() where creating/
unlinkg file is critical.
Author: Michael Paquier
Reviewed-by: Ashutosh Bapat, Takayuki Tsunakawa, me
Fix all perlcritic warnings of severity level 5, except in
src/backend/utils/Gen_dummy_probes.pl, which is automatically generated.
Reviewed-by: Dagfinn Ilmari Mannsåker <ilmari@ilmari.org>
Reviewed-by: Daniel Gustafsson <daniel@yesql.se>
Previously, auxiliary processes and background workers not connected
to a database (such as the logical replication launcher) weren't
shown. Include them, so that we can see the associated wait state
information. Add a new column to identify the processes type, so that
people can filter them out easily using SQL if they wish.
Before this patch was written, there was discussion about whether we
should expose this information in a separate view, so as to avoid
contaminating pg_stat_activity with things people might not want to
see. But putting everything in pg_stat_activity was a more popular
choice, so that's what the patch does.
Kuntal Ghosh, reviewed by Amit Langote and Michael Paquier. Some
revisions and bug fixes by me.
Discussion: http://postgr.es/m/CA+TgmoYES5nhkEGw9nZXU8_FhA8XEm8NTm3-SO+3ML1B81Hkww@mail.gmail.com
If the upstream walsender is using a physical replication slot, store the
catalog_xmin in the slot's catalog_xmin field. If the upstream doesn't use a
slot and has only a PGPROC entry behaviour doesn't change, as we store the
combined xmin and catalog_xmin in the PGPROC entry.
Author: Craig Ringer
This provides infrastructure for looking up arbitrary, user-supplied
XIDs without a risk of scary-looking failures from within the clog
module. Normally, the oldest XID that can be safely looked up in CLOG
is the same as the oldest XID that can reused without causing
wraparound, and the latter is already tracked. However, while
truncation is in progress, the values are different, so we must
keep track of them separately.
Craig Ringer, reviewed by Simon Riggs and by me.
Discussion: http://postgr.es/m/CAMsr+YHQiWNEi0daCTboS40T+V5s_+dst3PYv_8v2wNVH+Xx4g@mail.gmail.com
Previously AELs were registered against the top-level xid, which could
cause locks to be held much longer than necessary in some cases during
Hot Standby replay. We now record locks directly against their appropriate
xids. Requires few code changes because original code allowed for this
situation but didn’t fully implement it.
Discussion: CAKJS1f9vJ841HY=wonnLVbfkTWGYWdPN72VMxnArcGCjF3SywA@mail.gmail.com
Author: Simon Riggs and David Rowley
A hot standby replica keeps a list of Access Exclusive locks for a top
level transaction. These locks are released when the top level transaction
ends. Searching of this list is O(N^2), and each transaction had to pay the
price of searching this list for locks, even if it didn't take any AE
locks itself.
This patch optimizes this case by having the master server track which
transactions took AE locks, and passes that along to the standby server in
the commit/abort record. This allows the standby to only try to release
locks for transactions which actually took any, avoiding the majority of
the performance issue.
Refactor MyXactAccessedTempRel into MyXactFlags to allow minimal additional
cruft with this.
Analysis and initial patch by David Rowley
Author: David Rowley and Simon Riggs
this makes buffer access strategy have no effect.
Change was a part of commit 48354581a4 during 9.6
release cycle, so backpath to 9.6
Reported-by: Jim Nasby
Author: Alexander Korotkov
Reviewed-by: Jim Nasby, Andres Freund
https://commitfest.postgresql.org/13/1029/
Previous commits, notably 53be0b1add and
6f3bd98ebf, made it possible to see from
pg_stat_activity when a backend was stuck waiting for another backend,
but it's also fairly common for a backend to be stuck waiting for an
I/O. Add wait events for those operations, too.
Rushabh Lathia, with further hacking by me. Reviewed and tested by
Michael Paquier, Amit Kapila, Rajkumar Raghuwanshi, and Rahila Syed.
Discussion: http://postgr.es/m/CAGPqQf0LsYHXREPAZqYGVkDqHSyjf=KsD=k0GTVPAuzyThh-VQ@mail.gmail.com
Windows apparently will not detect socket write-ready events unless a
preceding send attempt returned WSAEWOULDBLOCK. In many usage patterns
that's satisfied by the caller of WaitEvenSetWait(), but not always.
Apply the same solution that we already had in pgwin32_select(), namely to
perform a dummy WSASend() call with len=0. This will return WSAEWOULDBLOCK
if there's no buffer space (even though it could legitimately do nothing
and report success, which makes me a bit nervous about this solution;
but since it's been working fine in libpq, let's roll with it).
In passing, improve the comments about this in pgwin32_select(), and remove
duplicated code there.
Back-patch to 9.6 where WaitEventSetWait() was introduced. We might need
to back-patch something similar into predecessor code. But given the lack
of complaints so far, it's not clear that the case ever gets exercised
in the back branches, so I'm not going to expend effort on it right now.
This should resolve recurring failures on buildfarm member bowerbird,
which has been failing since 1e8a85009 went in.
Diagnosis and patch by Petr Jelinek, cosmetic adjustments by me.
Discussion: https://postgr.es/m/5b6a6d6d-fb45-0afb-2e95-5600063c3dbd@2ndquadrant.com
This could result in corruption of the init fork of an unlogged index
if the ambuildempty routine for that index used shared buffers to
create the init fork, which was true for brin, gin, gist, and hash
indexes.
Patch by me, based on an earlier patch by Michael Paquier, who also
reviewed this one. This also incorporates an idea from Artur
Zakirov.
Discussion: http://postgr.es/m/CACYUyc8yccE4xfxhqxfh_Mh38j7dRFuxfaK1p6dSNAEUakxUyQ@mail.gmail.com
This reverts commit ccce90b398. This
optimization is unsafe, at least, of rollbacks and rollbacks to
savepoints, but I'm concerned there may be other problematic cases as
well. Therefore, I've decided to revert this pending further
investigation.
Commit 0e141c0fbb introduced a mechanism
to reduce contention on ProcArrayLock by having a single process clear
XIDs in the procArray on behalf of multiple processes, reducing the
need to hand the lock around. Use a similar mechanism to reduce
contention on CLogControlLock. Testing shows that this very
significantly reduces the amount of time waiting for CLogControlLock
on high-concurrency pgbench tests run on a large multi-socket
machines; whether that translates into a TPS improvement depends on
how much of that contention is simply shifted to some other lock,
particularly WALWriteLock.
Amit Kapila, with some cosmetic changes by me. Extensively reviewed,
tested, and benchmarked over a period of about 15 months by Simon
Riggs, Robert Haas, Andres Freund, Jesper Pedersen, and especially by
Tomas Vondra and Dilip Kumar.
Discussion: http://postgr.es/m/CAA4eK1L_snxM_JcrzEstNq9P66++F4kKFce=1r5+D1vzPofdtg@mail.gmail.com
Discussion: http://postgr.es/m/CAA4eK1LyR2A+m=RBSZ6rcPEwJ=rVi1ADPSndXHZdjn56yqO6Vg@mail.gmail.com
Discussion: http://postgr.es/m/91d57161-d3ea-0cc2-6066-80713e4f90d7@2ndquadrant.com
Any logical rep workers must have their subscription entries in
pg_subscription. To ensure this, we need to prevent the launcher
from starting new worker corresponding to the subscription that
DROP SUBSCRIPTION command is removing. To implement this,
previously LogicalRepLauncherLock was introduced and held until
the end of transaction running DROP SUBSCRIPTION. But using
LWLock for that purpose was not valid.
Instead, this commit changes DROP SUBSCRIPTION so that it takes
AccessExclusiveLock on pg_subscription, in order to ensure that
the launcher cannot see any subscriptions being removed. Also this
commit gets rid of LogicalRepLauncherLock.
Patch by me, reviewed by Petr Jelinek
Discussion: https://www.postgresql.org/message-id/CAHGQGwHPi8ky-yANFfe0sgmhKtsYcQLTnKx07bW9S7-Rn1746w@mail.gmail.com
When a shared iterator is used, each call to tbm_shared_iterate()
returns a result that has not yet been returned to any process
attached to the shared iterator. In other words, each cooperating
processes gets a disjoint subset of the full result set, but all
results are returned exactly once.
This is infrastructure for parallel bitmap heap scan.
Dilip Kumar. The larger patch set of which this is a part has been
reviewed and tested by (at least) Andres Freund, Amit Khandekar,
Tushar Ahuja, Rafia Sabih, Haribabu Kommi, and Thomas Munro.
Discussion: http://postgr.es/m/CAFiTN-uc4=0WxRGfCzs-xfkMYcSEWUC-Fon6thkJGjkh9i=13A@mail.gmail.com
In preparation for adding write-ahead logging to hash indexes,
refactor _hash_freeovflpage and _hash_squeezebucket so that all
related page modifications happen in a single section of code. The
previous coding assumed that it would be fine to move tuples one at a
time, and also that the various operations involved in freeing an
overflow page didn't necessarily all need to be done together, all
of which is true if you don't care about write-ahead logging.
Amit Kapila, with slight changes by me.
c.h #includes a number of core libc header files, such as <stdio.h>.
There's no point in re-including these after having read postgres.h,
postgres_fe.h, or c.h; so remove code that did so.
While at it, also fix some places that were ignoring our standard pattern
of "include postgres[_fe].h, then system header files, then other Postgres
header files". While there's not any great magic in doing it that way
rather than system headers last, it's silly to have just a few files
deviating from the general pattern. (But I didn't attempt to enforce this
globally, only in files I was touching anyway.)
I'd be the first to say that this is mostly compulsive neatnik-ism,
but over time it might save enough compile cycles to be useful.
Several places in fd.c had badly-thought-through handling of error returns
from lseek() and close(). The fact that those would seldom fail on valid
FDs is probably the reason we've not noticed this up to now; but if they
did fail, we'd get quite confused.
LruDelete and LruInsert actually just Assert'd that lseek never fails,
which is pretty awful on its face.
In LruDelete, we indeed can't throw an error, because that's likely to get
called during error abort and so throwing an error would probably just lead
to an infinite loop. But by the same token, throwing an error from the
close() right after that was ill-advised, not to mention that it would've
left the LRU state corrupted since we'd already unlinked the VFD from the
list. I also noticed that really, most of the time, we should know the
current seek position and it shouldn't be necessary to do an lseek here at
all. As patched, if we don't have a seek position and an lseek attempt
doesn't give us one, we'll close the file but then subsequent re-open
attempts will fail (except in the somewhat-unlikely case that a
FileSeek(SEEK_SET) call comes between and allows us to re-establish a known
target seek position). This isn't great but it won't result in any state
corruption.
Meanwhile, having an Assert instead of an honest test in LruInsert is
really dangerous: if that lseek failed, a subsequent read or write would
read or write from the start of the file, not where the caller expected,
leading to data corruption.
In both LruDelete and FileClose, if close() fails, just LOG that and mark
the VFD closed anyway. Possibly leaking an FD is preferable to getting
into an infinite loop or corrupting the VFD list. Besides, as far as I can
tell from the POSIX spec, it's unspecified whether or not the file has been
closed, so treating it as still open could be the wrong thing anyhow.
I also fixed a number of other places that were being sloppy about
behaving correctly when the seekPos is unknown.
Also, I changed FileSeek to return -1 with EINVAL for the cases where it
detects a bad offset, rather than throwing a hard elog(ERROR). It seemed
pretty inconsistent that some bad-offset cases would get a failure return
while others got elog(ERROR). It was missing an offset validity check for
the SEEK_CUR case on a closed file, too.
Back-patch to all supported branches, since all this code is fundamentally
identical in all of them.
Discussion: https://postgr.es/m/2982.1487617365@sss.pgh.pa.us
Doing so doesn't seem to be within the purpose of the per user
connection limits, and has particularly unfortunate effects in
conjunction with parallel queries.
Backpatch to 9.6 where parallel queries were introduced.
David Rowley, reviewed by Robert Haas and Albe Laurenz.
Add CatalogTupleInsertWithInfo and CatalogTupleUpdateWithInfo to let
callers use the CatalogTupleXXX abstraction layer even in cases where
we want to share the results of CatalogOpenIndexes across multiple
inserts/updates for efficiency. This finishes the job begun in commit
2f5c9d9c9, by allowing some remaining simple_heap_insert/update
calls to be replaced. The abstraction layer is now complete enough
that we don't have to export CatalogIndexInsert at all anymore.
Also, this fixes several places in which 2f5c9d9c9 introduced performance
regressions by using retail CatalogTupleInsert or CatalogTupleUpdate even
though the previous coding had been able to amortize CatalogOpenIndexes
work across multiple tuples.
A possible future improvement is to arrange for the indexing.c functions
to cache the CatalogIndexState somewhere, maybe in the relcache, in which
case we could get rid of CatalogTupleInsertWithInfo and
CatalogTupleUpdateWithInfo again. But that's a task for another day.
Discussion: https://postgr.es/m/27502.1485981379@sss.pgh.pa.us
This extends the work done in commit 2f5c9d9c9 to provide a more nearly
complete abstraction layer hiding the details of index updating for catalog
changes. That commit only invented abstractions for catalog inserts and
updates, leaving nearby code for catalog deletes still calling the
heap-level routines directly. That seems rather ugly from here, and it
does little to help if we ever want to shift to a storage system in which
indexing work is needed at delete time.
Hence, create a wrapper function CatalogTupleDelete(), and replace calls
of simple_heap_delete() on catalog tuples with it. There are now very
few direct calls of [simple_]heap_delete remaining in the tree.
Discussion: https://postgr.es/m/462.1485902736@sss.pgh.pa.us
Split the existing CatalogUpdateIndexes into two different routines,
CatalogTupleInsert and CatalogTupleUpdate, which do both the heap
insert/update plus the index update. This removes over 300 lines of
boilerplate code all over src/backend/catalog/ and src/backend/commands.
The resulting code is much more pleasing to the eye.
Also, by encapsulating what happens in detail during an UPDATE, this
facilitates the upcoming WARM patch, which is going to add a few more
lines to the update case making the boilerplate even more boring.
The original CatalogUpdateIndexes is removed; there was only one use
left, and since it's just three lines, we can as well expand it in place
there. We could keep it, but WARM is going to break all the UPDATE
out-of-core callsites anyway, so there seems to be no benefit in doing
so.
Author: Pavan Deolasee
Discussion: https://www.postgr.es/m/CABOikdOcFYSZ4vA2gYfs=M2cdXzXX4qGHeEiW3fu9PCfkHLa2A@mail.gmail.com
Gen_fmgrtab.pl creates a new file fmgrprotos.h, which contains
prototypes for all functions registered in pg_proc.h. This avoids
having to manually maintain these prototypes across a random variety of
header files. It also automatically enforces a correct function
signature, and since there are warnings about missing prototypes, it
will detect functions that are defined but not registered in
pg_proc.h (or otherwise used).
Reviewed-by: Pavel Stehule <pavel.stehule@gmail.com>
Commit 4aec49899e reorganized the order
of operations here so that we no longer increment the number of "extra
waits" before locking the semaphore, but it did not change the
starting value of extraWaits from 0 to -1 to compensate. In the worst
case, this could leak a semaphore count, but that seems to be unlikely
in practice.
Discussion: http://postgr.es/m/CAA4eK1JyVqXiMba+-a589Rk0pyHsyKkGxeumVKjU6Y74hdrVLQ@mail.gmail.com
Amit Kapila, per an off-list report by Dilip Kumar. Reviewed by me.
With the old code, a backend that read pg_stat_activity without ever
having executed a parallel query might see a backend in the midst of
executing one waiting on a DSA LWLock, resulting in a crash. The
solution is for backends to register the tranche at startup time, not
the first time a parallel query is executed.
Report by Andreas Seltenreich. Patch by me, reviewed by Thomas Munro.
Some background activity (like checkpoints, archive timeout, standby
snapshots) is not supposed to happen on an idle system. Unfortunately
so far it was not easy to determine when a system is idle, which
defeated some of the attempts to avoid redundant activity on an idle
system.
To make that easier, allow to make individual WAL insertions as not
being "important". By checking whether any important activity happened
since the last time an activity was performed, it now is easy to check
whether some action needs to be repeated.
Use the new facility for checkpoints, archive timeout and standby
snapshots.
The lack of a facility causes some issues in older releases, but in my
opinion the consequences (superflous checkpoints / archived segments)
aren't grave enough to warrant backpatching.
Author: Michael Paquier, editorialized by Andres Freund
Reviewed-By: Andres Freund, David Steele, Amit Kapila, Kyotaro HORIGUCHI
Bug: #13685
Discussion:
https://www.postgresql.org/message-id/20151016203031.3019.72930@wrigleys.postgresql.orghttps://www.postgresql.org/message-id/CAB7nPqQcPqxEM3S735Bd2RzApNqSNJVietAC=6kfkYv_45dKwA@mail.gmail.com
Backpatch: -
If we do not reset the FD_READ event, WaitForMultipleObjects won't
return it again again unless we've meanwhile read from the socket,
which is generally true but not guaranteed. WaitEventSetWaitBlock
itself may fail to return the event to the caller if the latch is
also set, and even if we changed that, the caller isn't obliged to
handle all returned events at once. On non-Windows systems, the
socket-read event is purely level-triggered, so this issue does
not exist. To fix, make Windows reset the event when needed.
This bug was introduced by 98a64d0bd7,
and causes hangs when trying to use the pldebugger extension.
Patch by Amit Kapial. Reported and tested by Ashutosh Sharma, who
also provided some analysis. Further analysis by Michael Paquier.
array_base and array_stride were added so that we could identify the
offset of an LWLock within a tranche, but this facility is only very
marginally used apart from the main tranche. So, give every lock in
the main tranche its own tranche ID and get rid of array_base,
array_stride, and all that's attached. For debugging facilities
(Trace_lwlocks and LWLOCK_STATS) print the pointer address of the
LWLock using %p instead of the offset. This is arguably more useful,
and certainly a lot cheaper. Drop the offset-within-tranche from
the information reported to dtrace and from one can't-happen message
inside lwlock.c.
The main user-visible impact of this change is that pg_stat_activity
will now report all waits for LWLocks as "LWLock" rather than
reporting some as "LWLockTranche" and others as "LWLockNamed".
The main motivation for this change is that the need to specify an
array_base and an array_stride is awkward for parallel query. There
is only a very limited supply of tranche IDs so we can't just keep
allocating new ones, and if we try to use the same tranche IDs every
time then we run into trouble when multiple parallel contexts are
use simultaneously. So if we didn't get rid of this mechanism we'd
have to make it even more complicated. By simplifying it in this
way, we instead reduce the size of the generated code for lwlock.c
by about 5%.
Discussion: http://postgr.es/m/CA+TgmoYsFn6NUW1x0AZtupJGUAs1UDY4dJtCN47_Q6D0sP80PA@mail.gmail.com
Previously, the "sem" field of PGPROC varied in size depending on which
kernel semaphore API we were using. That was okay as long as there was
only one likely choice per platform, but in the wake of commit ecb0d20a9,
that assumption seems rather shaky. It doesn't seem out of the question
anymore that an extension compiled against one API choice might be loaded
into a postmaster built with another choice. Moreover, this prevents any
possibility of selecting the semaphore API at postmaster startup, which
might be something we want to do in future.
Hence, change PGPROC.sem to be PGSemaphore (i.e. a pointer) for all Unix
semaphore APIs, and turn the pointed-to data into an opaque struct whose
contents are only known within the responsible modules.
For the SysV and unnamed-POSIX APIs, the pointed-to data has to be
allocated elsewhere in shared memory, which takes a little bit of
rejiggering of the InitShmemAllocation code sequence. (I invented a
ShmemAllocUnlocked() function to make that a little cleaner than it used
to be. That function is not meant for any uses other than the ones it
has now, but it beats having InitShmemAllocation() know explicitly about
allocation of space for semaphores and spinlocks.) This change means an
extra indirection to access the semaphore data, but since we only touch
that when blocking or awakening a process, there shouldn't be any
meaningful performance penalty. Moreover, at least for the unnamed-POSIX
case on Linux, the sem_t type is quite a bit wider than a pointer, so this
reduces sizeof(PGPROC) which seems like a good thing.
For the named-POSIX API, there's effectively no change: the PGPROC.sem
field was and still is a pointer to something returned by sem_open() in
the postmaster's memory space. Document and check the pre-existing
limitation that this case can't work in EXEC_BACKEND mode.
It did not seem worth unifying the Windows semaphore ABI with the Unix
cases, since there's no likelihood of needing ABI compatibility much less
runtime switching across those cases. However, we can simplify the Windows
code a bit if we define PGSemaphore as being directly a HANDLE, rather than
pointer to HANDLE, so let's do that while we're here. (This also ends up
being no change in what's physically stored in PGPROC.sem. We're just
moving the HANDLE fetch from callees to callers.)
It would take a bunch of additional code shuffling to get to the point of
actually choosing a semaphore API at postmaster start, but the effects
of that would now be localized in the port/XXX_sema.c files, so it seems
like fit material for a separate patch. The need for it is unproven as
yet, anyhow, whereas the ABI risk to extensions seems real enough.
Discussion: https://postgr.es/m/4029.1481413370@sss.pgh.pa.us
This allows creating temporary replication slots that are removed
automatically at the end of the session or on error.
From: Petr Jelinek <petr.jelinek@2ndquadrant.com>
Rearrange a bit of code to ensure that 'mode' in LWLockRelease is
obviously always set, which seems a bit cleaner and avoids a compiler
warning (thanks to Robert for the suggestion!).
In GetCachedPlan(), initialize 'plan' to silence a compiler warning, but
also add an Assert() to make sure we don't ever actually fall through
with 'plan' still being set to NULL, since we are about to dereference
it.
Neither of these appear to be live bugs but at least gcc
5.4.0-6ubuntu1~16.04.4 doesn't quite have the smarts to realize that.
Discussion: https://www.postgresql.org/message-id/20161129152102.GR13284%40tamriel.snowman.net
This adds a new routine, pg_strong_random() for generating random bytes,
for use in both frontend and backend. At the moment, it's only used in
the backend, but the upcoming SCRAM authentication patches need strong
random numbers in libpq as well.
pg_strong_random() is based on, and replaces, the existing implementation
in pgcrypto. It can acquire strong random numbers from a number of sources,
depending on what's available:
- OpenSSL RAND_bytes(), if built with OpenSSL
- On Windows, the native cryptographic functions are used
- /dev/urandom
Unlike the current pgcrypto function, the source is chosen by configure.
That makes it easier to test different implementations, and ensures that
we don't accidentally fall back to a less secure implementation, if the
primary source fails. All of those methods are quite reliable, it would be
pretty surprising for them to fail, so we'd rather find out by failing
hard.
If no strong random source is available, we fall back to using erand48(),
seeded from current timestamp, like PostmasterRandom() was. That isn't
cryptographically secure, but allows us to still work on platforms that
don't have any of the above stronger sources. Because it's not very secure,
the built-in implementation is only used if explicitly requested with
--disable-strong-random.
This replaces the more complicated Fortuna algorithm we used to have in
pgcrypto, which is unfortunate, but all modern platforms have /dev/urandom,
so it doesn't seem worth the maintenance effort to keep that. pgcrypto
functions that require strong random numbers will be disabled with
--disable-strong-random.
Original patch by Magnus Hagander, tons of further work by Michael Paquier
and me.
Discussion: https://www.postgresql.org/message-id/CAB7nPqRy3krN8quR9XujMVVHYtXJ0_60nqgVc6oUk8ygyVkZsA@mail.gmail.com
Discussion: https://www.postgresql.org/message-id/CAB7nPqRWkNYRRPJA7-cF+LfroYV10pvjdz6GNvxk-Eee9FypKA@mail.gmail.com
A new thing also called a "barrier" is proposed, but whether we decide
to take that patch or not, this file seems to have outlived its
usefulness.
Thomas Munro
Condition variables provide a flexible way to sleep until a
cooperating process causes an arbitrary condition to become true. In
simple cases, this can be accomplished with a WaitLatch/ResetLatch
loop; the cooperating process can call SetLatch after performing work
that might cause the condition to be satisfied, and the waiting
process can recheck the condition each time. However, if the process
performing the work doesn't have an easy way to identify which
processes might be waiting, this doesn't work, because it can't
identify which latches to set. Condition variables solve that problem
by internally maintaining a list of waiters; a process that may have
caused some waiter's condition to be satisfied must "signal" or
"broadcast" on the condition variable.
Robert Haas and Thomas Munro
Previously, the handle for the control segment could not be zero, but
some other DSM segment could potentially have a handle value of zero.
However, that means that if someone wanted to store a dsm_handle that
might or might not be valid, they would need a separate boolean to
keep track of whether the associated value is legal. That's annoying,
so change things so that no DSM segment can ever have a handle of 0 -
or as we call it here, DSM_HANDLE_INVALID.
Thomas Munro. This was submitted as part of a much larger patch to
add an malloc-like allocator for dynamic shared memory, but this part
seems like a good idea independently of the rest of the patch.
LockBufferForCleanup() acquires a cleanup lock unconditionally, and
ConditionalLockBufferForCleanup() acquires a cleanup lock if it is
possible to do so without waiting; this patch adds a new API,
IsBufferCleanupOK(), which tests whether an exclusive lock already
held happens to be a cleanup lock. This is possible because a cleanup
lock simply means an exclusive lock plus the assurance any other pins
on the buffer are newer than our own pin. Therefore, just as the
existing functions decide that the exclusive lock that they've just
taken is a cleanup lock if they observe the pin count to be 1, this
new function allows us to observe that the pin count is 1 on a buffer
we've already locked.
This is useful in situations where a backend definitely wishes to
modify the buffer and also wishes to perform cleanup operations if
possible. The patch to eliminate heavyweight locking by hash indexes
uses this, and it may have other applications as well.
Amit Kapila, per a suggestion from me. Some comment adjustments by me
as well.
"xlog" is not a particularly clear abbreviation for "write-ahead log",
and it sometimes confuses users into believe that the contents of the
"pg_xlog" directory are not critical data, leading to unpleasant
consequences. So, rename the directory to "pg_wal".
This patch modifies pg_upgrade and pg_basebackup to understand both
the old and new directory layouts; the former is necessary given the
purpose of the tool, while the latter merely avoids an unnecessary
backward-compatibility break.
We may wish to consider renaming other programs, switches, and
functions which still use the old "xlog" naming to also refer to
"wal". However, that's still under discussion, so let's do just this
much for now.
Discussion: CAB7nPqTeC-8+zux8_-4ZD46V7YPwooeFxgndfsq5Rg8ibLVm1A@mail.gmail.com
Michael Paquier
When a relation is truncated, it is important that the FSM is truncated as
well. Otherwise, after recovery, the FSM can return a page that has been
truncated away, leading to errors like:
ERROR: could not read block 28991 in file "base/16390/572026": read only 0
of 8192 bytes
We were using MarkBufferDirtyHint() to dirty the buffer holding the last
remaining page of the FSM, but during recovery, that might in fact not
dirty the page, and the FSM update might be lost.
To fix, use the stronger MarkBufferDirty() function. MarkBufferDirty()
requires us to do WAL-logging ourselves, to protect from a torn page, if
checksumming is enabled.
Also fix an oversight in visibilitymap_truncate: it also needs to WAL-log
when checksumming is enabled.
Analysis by Pavan Deolasee.
Discussion: <CABOikdNr5vKucqyZH9s1Mh0XebLs_jRhKv6eJfNnD2wxTn=_9A@mail.gmail.com>
I somehow had assumed that in the spinlock (in turn possibly using
semaphores) based fallback atomics implementation 32 bit writes could be
done without a lock. As far as the write goes that's correct, since
postgres supports only platforms with single-copy atomicity for aligned
32bit writes. But writing without holding the spinlock breaks
read-modify-write operations like pg_atomic_compare_exchange_u32(),
since they'll potentially "miss" a concurrent write, which can't happen
in actual hardware implementations.
In 9.6+ when using the fallback atomics implementation this could lead
to buffer header locks not being properly marked as released, and
potentially some related state corruption. I don't see a related danger
in 9.5 (earliest release with the API), because pg_atomic_write_u32()
wasn't used in a concurrent manner there.
The state variable of local buffers, before this change, were
manipulated using pg_atomic_write_u32(), to avoid unnecessary
synchronization overhead. As that'd not be the case anymore, introduce
and use pg_atomic_unlocked_write_u32(), which does not correctly
interact with RMW operations.
This bug only caused issues when postgres is compiled on platforms
without atomics support (i.e. no common new platform), or when compiled
with --disable-atomics, which explains why this wasn't noticed in
testing.
Reported-By: Tom Lane
Discussion: <14947.1475690465@sss.pgh.pa.us>
Backpatch: 9.5-, where the atomic operations API was introduced.
Windows apparently has a constant named WAIT_TIMEOUT, and some of these
other names are pretty generic, too. Insert "PG_" at the front of each
name in order to disambiguate.
Michael Paquier
WaitLatch, WaitLatchOrSocket, and WaitEventSetWait now taken an
additional wait_event_info parameter; legal values are defined in
pgstat.h. This makes it possible to uniquely identify every point in
the core code where we are waiting for a latch; extensions can pass
WAIT_EXTENSION.
Because latches were the major wait primitive not previously covered
by this patch, it is now possible to see information in
pg_stat_activity on a large number of important wait events not
previously addressed, such as ClientRead, ClientWrite, and SyncRep.
Unfortunately, many of the wait events added by this patch will fail
to appear in pg_stat_activity because they're only used in background
processes which don't currently appear in pg_stat_activity. We should
fix this either by creating a separate view for such information, or
else by deciding to include them in pg_stat_activity after all.
Michael Paquier and Robert Haas, reviewed by Alexander Korotkov and
Thomas Munro.
We must test GetLastError() even when CreateFileMapping() returns a
non-null handle. If that value were left over from some previous system
call, we might be fooled into thinking the segment already existed.
Experimentation on Windows 7 suggests that CreateFileMapping() clears
the error code on success, but it is not documented to do so, so let's
not rely on that happening in all Windows releases.
Amit Kapila
Discussion: <20811.1474390987@sss.pgh.pa.us>
Commits 470d886c3 et al intended to fix the problem that the postmaster
selected the same "random" DSM control segment ID on every start. But
using PostmasterRandom() for that destroys the intended property that the
delay between random_start_time and random_stop_time will be unpredictable.
(Said delay is probably already more predictable than we could wish, but
that doesn't mean that reducing it by a couple orders of magnitude is OK.)
Revert the previous patch and add a comment warning against misuse of
PostmasterRandom. Fix the original problem by calling srandom() early in
PostmasterMain, using a low-security seed that will later be overwritten
by PostmasterRandom.
Discussion: <20789.1474390434@sss.pgh.pa.us>
Otherwise, attempts to run multiple postmasters running on the same
machine may fail, because Windows sometimes returns ERROR_ACCESS_DENIED
rather than ERROR_ALREADY_EXISTS when there is an existing segment.
Hitting this bug is much more likely because of another defect not
fixed by this patch, namely that dsm_postmaster_startup() uses
random() which returns the same value every time. But that's not
a reason not to fix this.
Kyotaro Horiguchi and Amit Kapila, reviewed by Michael Paquier
Discussion: <CAA4eK1JyNdMeF-dgrpHozDecpDfsRZUtpCi+1AbtuEkfG3YooQ@mail.gmail.com>
The full generality of deleting an arbitrary number of tuples is no longer
needed, so let's save some code and cycles by replacing the original coding
with an implementation based on PageIndexTupleDelete.
We can always get back the old code from git if we need it again for new
callers (though I don't care for its willingness to mess with line pointers
it wasn't told to mess with).
Discussion: <552.1473445163@sss.pgh.pa.us>
Nowadays this is just a backwards-compatibility wrapper around
PageAddItemExtended, so let's avoid the extra level of function call.
In addition, because pretty much all callers are passing constants
for the two bool arguments, compilers will be able to constant-fold
the conversion to a flags bitmask.
Discussion: <552.1473445163@sss.pgh.pa.us>
PageIndexTupleOverwrite performs approximately the same function as
PageIndexTupleDelete (or PageIndexDeleteNoCompact) followed by PageAddItem
targeting the same item pointer offset. But in the case where the new
tuple is the same size as the old, it avoids shuffling other data around on
the page, because the new tuple is placed where the old one was rather than
being appended to the end of the page. This has been shown to provide a
substantial speedup for some GiST use-cases.
Also, this change allows some API simplifications: we can get rid of
the rather klugy and error-prone PAI_ALLOW_FAR_OFFSET flag for
PageAddItemExtended, since that was used only to cover a corner case
for BRIN that's better expressed by using PageIndexTupleOverwrite.
Note that this patch causes a rather subtle WAL incompatibility: the
physical page content change represented by certain WAL records is now
different than it was before, because while the tuples have the same
itempointer line numbers, the tuples themselves are in different places.
I have not bumped the WAL version number because I think it doesn't matter
unless you are trying to do bitwise comparisons of original and replayed
pages, and in any case we're early in a devel cycle and there will probably
be more WAL changes before v10 gets out the door.
There is probably room to make use of PageIndexTupleOverwrite in SP-GiST
and GIN too, but that is left for a future patch.
Andrey Borodin, reviewed by Anastasia Lubennikova, whacked around a bit
by me
Discussion: <CAJEAwVGQjGGOj6mMSgMwGvtFd5Kwe6VFAxY=uEPZWMDjzbn4VQ@mail.gmail.com>
PageAddItem stores the item length as-is. It MAXALIGN's the amount of
space actually allocated for each tuple, but not the stored length.
PageRepairFragmentation, PageIndexMultiDelete, and PageIndexDeleteNoCompact
are all on board with this and MAXALIGN item lengths after fetching them.
But PageIndexTupleDelete expects the stored length to be a MAXALIGN
multiple already. This accidentally works for existing index AMs because
they all maxalign their tuple sizes internally; but we don't do that for
heap tuples, and it shouldn't be a requirement for index tuples either.
So, sync PageIndexTupleDelete with the rest of bufpage.c by having it
maxalign the item size after fetching.
Also add a check that pd_special is maxaligned, to ensure that the test
"(offset + size) > phdr->pd_special" is still doing the right thing.
(If offset and pd_special are aligned, it doesn't matter whether size is.)
Again, this is in sync with the rest of the routines here, except for
PageAddItem which doesn't test because it doesn't actually do anything
for which pd_special alignment matters.
This shouldn't have any immediate functional impact; it just adds the
flexibility to use PageIndexTupleDelete on index tuples with non-aligned
lengths.
Discussion: <3814.1473366762@sss.pgh.pa.us>
So far md.c used a linked list of segments. That proved to be a problem
when processing large relations, because every smgr.c/md.c level access
to a page incurred walking through a linked list of all preceding
segments. Thus making accessing pages O(#segments).
Replace the linked list of segments hanging off SMgrRelationData with an
array of opened segments. That allows O(1) access to individual
segments, if they've previously been opened.
Discussion: <20140331101001.GE13135@alap3.anarazel.de>
Reviewed-By: Peter Geoghegan, Tom Lane (in an older version)
That's primarily useful for testing very large relations, using sparse
files.
Discussion: <20140331101001.GE13135@alap3.anarazel.de>
Reviewed-By: Peter Geoghegan
mdtruncate() forgot to FileClose() a segment's mdfd_vfd, when deleting
it. That lead to a fd.c handle to a truncated file being kept open until
backend exit.
The issue appears to have been introduced way back in 1a5c450f30,
before that the handle was closed inside FileUnlink().
The impact of this bug is limited - only VACUUM and ON COMMIT TRUNCATE
for temporary tables, truncate files in place (i.e. TRUNCATE itself is
not affected), and the relation has to be bigger than 1GB. The
consequences of a leaked fd.c handle aren't severe either.
Discussion: <20160908220748.oqh37ukwqqncbl3n@alap3.anarazel.de>
Backpatch: all supported releases
A majority of callers seem to have believed that this was the API spec
already, because they omitted any check for a NULL result, and hence
would crash on an out-of-shared-memory failure. The original proposal
was to just add such error checks everywhere, but that does nothing to
prevent similar omissions in future. Instead, let's make ShmemAlloc()
throw the error (so we can remove the caller-side checks that do exist),
and introduce a new function ShmemAllocNoError() that has the previous
behavior of returning NULL, for the small number of callers that need
that and are prepared to do the right thing. This also lets us remove
the rather wishy-washy behavior of printing a WARNING for out-of-shmem,
which never made much sense: either the caller has a strategy for
dealing with that, or it doesn't. It's not ShmemAlloc's business to
decide whether a warning is appropriate.
The v10 release notes will need to call this out as a significant
source-code change. It's likely that it will be a bug fix for
extension callers too, but if not, they'll need to change to using
ShmemAllocNoError().
This is nominally a bug fix, but the odds that it's fixing any live
bug are actually rather small, because in general the requests
being made by the unchecked callers were already accounted for in
determining the overall shmem size, so really they ought not fail.
Between that and the possible impact on extensions, no back-patch.
Discussion: <24843.1472563085@sss.pgh.pa.us>
I found that half a dozen (nearly 5%) of our AllocSetContextCreate calls
had typos in the context-sizing parameters. While none of these led to
especially significant problems, they did create minor inefficiencies,
and it's now clear that expecting people to copy-and-paste those calls
accurately is not a great idea. Let's reduce the risk of future errors
by introducing single macros that encapsulate the common use-cases.
Three such macros are enough to cover all but two special-purpose contexts;
those two calls can be left as-is, I think.
While this patch doesn't in itself improve matters for third-party
extensions, it doesn't break anything for them either, and they can
gradually adopt the simplified notation over time.
In passing, change TopMemoryContext to use the default allocation
parameters. Formerly it could only be extended 8K at a time. That was
probably reasonable when this code was written; but nowadays we create
many more contexts than we did then, so that it's not unusual to have a
couple hundred K in TopMemoryContext, even without considering various
dubious code that sticks other things there. There seems no good reason
not to let it use growing blocks like most other contexts.
Back-patch to 9.6, mostly because that's still close enough to HEAD that
it's easy to do so, and keeping the branches in sync can be expected to
avoid some future back-patching pain. The bugs fixed by these changes
don't seem to be significant enough to justify fixing them further back.
Discussion: <21072.1472321324@sss.pgh.pa.us>
If you have previously pinned a segment and decide that you don't
actually want to keep it around until shutdown, this new API lets you
remove the pin. This is pretty trivial except on Windows, where it
requires closing the duplicate handle that was used to implement the
pin.
Thomas Munro and Amit Kapila, reviewed by Amit Kapila and by me.
Prior to commit 7882c3b0b9, it was
possible to use LWLocks within DSM segments, but that commit broke
this use case by switching from a doubly linked list to a circular
linked list. Switch back, using a new bit of general infrastructure
for maintaining lists of PGPROCs.
Thomas Munro, reviewed by me.
This coding pattern creates a race condition, because if an interesting
interrupt happens after we've checked InterruptPending but before we reset
our latch, the latch-setting done by the signal handler would get lost,
and then we might block at WaitLatch in the next iteration without ever
noticing the interrupt condition. You can put the CHECK_FOR_INTERRUPTS
before WaitLatch or after ResetLatch, but not between them.
Aside from fixing the bugs, add some explanatory comments to latch.h
to perhaps forestall the next person from making the same mistake.
In HEAD, also replace gather_readnext's direct call of
HandleParallelMessages with CHECK_FOR_INTERRUPTS. It does not seem clean
or useful for this one caller to bypass ProcessInterrupts and go straight
to HandleParallelMessages; not least because that fails to consider the
InterruptPending flag, resulting in useless work both here
(if InterruptPending isn't set) and in the next CHECK_FOR_INTERRUPTS call
(if it is).
This thinko seems to have been introduced in the initial coding of
storage/ipc/shm_mq.c (commit ec9037df2), and then blindly copied into all
the subsequent parallel-query support logic. Back-patch relevant hunks
to 9.4 to extirpate the error everywhere.
Discussion: <1661.1469996911@sss.pgh.pa.us>
Examination of the results from anole and gharial suggests that we're
only managing to track the size of one of the two stacks of IA64 machines.
Some googling gave the answer: on HPUX11, the register stack is reported
as a page type I don't see in pstat.h on my HPUX10 box. Let's try
testing for that.
After a look at preliminary results from commit 88cf37d2a8,
I realized it'd be a good idea to spew out the maximum depth measurement
seen by check_stack_depth. So add some quick-n-dirty code to do that.
Like the previous commit, this will be reverted once we've gathered
a set of buildfarm runs with it.
This patch is meant to gather information from the buildfarm members, and
will be reverted in a day or so. The idea is to try to find out the
high-water stack consumption while running the regression tests,
particularly on IA64 which is suspected to use much more stack than other
architectures. On machines with pmap, we can use that; but the IA64 farm
members are running HPUX, so also include some bespoke code for HPUX.
(I've tested the latter on HPUX 10/HPPA; not entirely sure it will work
on HPUX 11/IA64, but we'll soon find out.)
Discussion: <CAM-w4HMwwcwaVvYcAH0_FGtG5GeXdYVRfvG81pXnSJWHnCfosQ@mail.gmail.com>
Since indexes are created without valid LSNs, an index created
while a snapshot older than old_snapshot_threshold existed could
cause queries to return incorrect results when those old snapshots
were used, if any relevant rows had been subject to early pruning
before the index was built. Prevent usage of a newly created index
until all such snapshots are released, for relations where this can
happen.
Questions about the interaction of "snapshot too old" with index
creation were initially raised by Andres Freund.
Reviewed by Robert Haas.
Transmit the leader's temp-namespace state to workers. This is important
because without it, the workers do not really have the same search path
as the leader. For example, there is no good reason (and no extant code
either) to prevent a worker from executing a temp function that the
leader created previously; but as things stood it would fail to find the
temp function, and then either fail or execute the wrong function entirely.
We still prohibit a worker from creating a temp namespace on its own.
In effect, a worker can only see the session's temp namespace if the leader
had created it before starting the worker, which seems like the right
semantics.
Also, transmit the leader's BackendId to workers, and arrange for workers
to use that when determining the physical file path of a temp relation
belonging to their session. While the original intent was to prevent such
accesses entirely, there were a number of holes in that, notably in places
like dbsize.c which assume they can safely access temp rels of other
sessions anyway. We might as well get this right, as a small down payment
on someday allowing workers to access the leader's temp tables. (With
this change, directly using "MyBackendId" as a relation or buffer backend
ID is deprecated; you should use BackendIdForTempRelations() instead.
I left a couple of such uses alone though, as they're not going to be
reachable in parallel workers until we do something about localbuf.c.)
Move the thou-shalt-not-access-thy-leader's-temp-tables prohibition down
into localbuf.c, which is where it actually matters, instead of having it
in relation_open(). This amounts to recognizing that access to temp
tables' catalog entries is perfectly safe in a worker, it's only the data
in local buffers that is problematic.
Having done all that, we can get rid of the test in has_parallel_hazard()
that says that use of a temp table's rowtype is unsafe in parallel workers.
That test was unduly expensive, and if we really did need such a
prohibition, that was not even close to being a bulletproof guard for it.
(For example, any user-defined function executed in a parallel worker
might have attempted such access.)
Prior to this patch, it was occasionally possible, after shm_mq_sendv
had previously returned SHM_MQ_DETACHED, for a later shm_mq_sendv
operation to fail an assertion instead of just again returning
SHM_MQ_ATTACHED. From the shm_mq code's point of view, it was
expecting to be called again with the same arguments, since the
previous operation had only partially completed. However, a caller
who isn't using non-blocking mode won't be prepared to repeat the call
with the same arguments, and this code shouldn't expect that they
will. Repair in such a way that we'll be OK whether the next call
uses the same arguments or not.
Found by Andreas Seltenreich. Analysis and sketch of fix by Amit
Kapila. Patch by me, reviewed by Amit Kapila.
Mostly these are just comments but there are a few in documentation
and a handful in code and tests. Hopefully this doesn't cause too much
unnecessary pain for backpatching. I relented from some of the most
common like "thru" for that reason. The rest don't seem numerous
enough to cause problems.
Thanks to Kevin Lyda's tool https://pypi.python.org/pypi/misspellings
Use MAXALIGN size/alignment to guarantee that later uses of memory are
aligned correctly. E.g. epoll_event might need 8 byte alignment on some
platforms, but earlier allocations like WaitEventSet and WaitEvent might
not sized to guarantee that when purely using sizeof().
Found by myself while testing on an Sun Ultra 5 (Sparc IIi) with some
editorializing by Andres Freund.
In passing fix a couple typos in the area
BRIN was relying on the ability to remove a tuple from an index page,
then putting another tuple in the same line pointer. But PageAddItem
refuses to add a tuple beyond the first free item past the last used
item, and in particular, it rejects an attempt to add an item to an
empty page anywhere other than the first line pointer. PageAddItem
issues a WARNING and indicates to the caller that it failed, which in
turn causes the BRIN calling code to issue a PANIC, so the whole
sequence looks like this:
WARNING: specified item offset is too large
PANIC: failed to add BRIN tuple
To fix, create a new function PageAddItemExtended which is like
PageAddItem except that the two boolean arguments become a flags bitmap;
the "overwrite" and "is_heap" boolean flags in PageAddItem become
PAI_OVERWITE and PAI_IS_HEAP flags in the new function, and a new flag
PAI_ALLOW_FAR_OFFSET enables the behavior required by BRIN.
PageAddItem() retains its original signature, for compatibility with
third-party modules (other callers in core code are not modified,
either).
Also, in the belt-and-suspenders spirit, I added a new sanity check in
brinGetTupleForHeapBlock to raise an error if an TID found in the revmap
is not marked as live by the page header. This causes it to react with
"ERROR: corrupted BRIN index" to the bug at hand, rather than a hard
crash.
Backpatch to 9.5.
Bug reported by Andreas Seltenreich as detected by his handy sqlsmith
fuzzer.
Discussion: https://www.postgresql.org/message-id/87mvni77jh.fsf@elite.ansel.ydns.eu
Commit 1aba62ec moved the range check of that option form guc.c into
bufmgr.c, but introduced a bug by changing a >= 0.0 to > 0.0, which made
the value 0 no longer accepted. Put it back.
Reported by Jeff Janes, diagnosed by Tom Lane
Unfortunately the segment size checks from 72a98a6395 had the negative
side-effect of breaking a corner case in mdsync(): When processing a
fsync request for a truncated away segment mdsync() could fail with
"could not fsync file" (if previous segment < RELSEG_SIZE) because
_mdfd_getseg() now wouldn't return the relevant segment anymore.
The cleanest fix seems to be to allow the caller of _mdfd_getseg() to
specify whether checks for RELSEG_SIZE are performed. To allow doing so,
change the ExtensionBehavior enum into a bitmask. Besides allowing for
the addition of EXTENSION_DONT_CHECK_SIZE, this makes for a nicer
implementation of EXTENSION_REALLY_RETURN_NULL.
Besides mdsync() the only callsite that should change behaviour due to
this is mdprefetch() which now doesn't create segments anymore, even in
recovery. Given the uses of mdprefetch() that seems better.
Reported-By: Thom Brown
Discussion: CAA-aLv72QazLvPdKZYpVn4a_Eh+i4_cxuB03k+iCuZM_xjc+6Q@mail.gmail.com
Before this commit _mdfd_getseg(), in contrast to mdnblocks(), did not
verify whether all segments leading up to the to-be-opened one, were
RELSEG_SIZE sized. That is e.g. not the case after truncating a
relation, because later segments just get truncated to zero length, not
removed.
Once a "non-existent" segment has been opened in a session, mdnblocks()
will return wrong results, causing errors like "could not read block %u
in file" when accessing blocks. Closing the session, or the later
arrival of relevant invalidation messages, would "fix" the problem.
That, so far, was mostly harmless, because most segment accesses are
only done after an mdnblocks() call. But since 428b1d6b29 we try to
open segments that might have been deleted, to trigger kernel writeback
from a backend's queue of recent writes.
To fix check segment sizes in _mdfd_getseg() when opening previously
unopened segments. In practice this shouldn't imply a lot of additional
lseek() calls, because mdnblocks() will most of the time already have
opened all relevant segments.
This commit also fixes a second problem, namely that _mdfd_getseg(
EXTENSION_RETURN_NULL) extends files during recovery, which is not
desirable for the mdwriteback() case. Add EXTENSION_REALLY_RETURN_NULL,
which does not behave that way, and use it.
Reported-By: Thom Brown
Author: Andres Freund, Abhijit Menon-Sen
Reviewd-By: Robert Haas, Fabien Coehlo
Discussion: CAA-aLv6Dp_ZsV-44QA-2zgkqWKQq=GedBX2dRSrWpxqovXK=Pg@mail.gmail.com
Fixes: 428b1d6b29
So far, when a transaction with pending invalidations, but without an
assigned xid, committed, we simply ignored those invalidation
messages. That's problematic, because those are actually sent for a
reason.
Known symptoms of this include that existing sessions on a hot-standby
replica sometimes fail to notice new concurrently built indexes and
visibility map updates.
The solution is to WAL log such invalidations in transactions without an
xid. We considered to alternatively force-assign an xid, but that'd be
problematic for vacuum, which might be run in systems with few xids.
Important: This adds a new WAL record, but as the patch has to be
back-patched, we can't bump the WAL page magic. This means that standbys
have to be updated before primaries; otherwise
"PANIC: standby_redo: unknown op code 32" errors can be encountered.
XXX:
Reported-By: Васильев Дмитрий, Masahiko Sawada
Discussion:
CAB-SwXY6oH=9twBkXJtgR4UC1NqT-vpYAtxCseME62ADwyK5OA@mail.gmail.comCAD21AoDpZ6Xjg=gFrGPnSn4oTRRcwK1EBrWCq9OqOHuAcMMC=w@mail.gmail.com
That commit increased all shared memory allocations to the next higher
multiple of PG_CACHE_LINE_SIZE, but it didn't ensure that allocation
started on a cache line boundary. It also failed to remove a couple
other pieces of now-useless code.
BUFFERALIGN() is perhaps obsolete at this point, and likely should be
removed at some point, too, but that seems like it can be left to a
future cleanup.
Mistakes all pointed out by Andres Freund. The patch is mine, with
a few extra assertions which I adopted from his version of this fix.
Even with old_snapshot_threshold = -1 (which disables the "snapshot
too old" feature), performance regressions were seen at moderate to
high concurrency. For example, a one-socket, four-core system
running 200 connections at saturation could see up to a 2.3%
regression, with larger regressions possible on NUMA machines.
By inlining the early (smaller, faster) tests in the
TestForOldSnapshot() function, the i7 case dropped to a 0.2%
regression, which could easily just be noise, and is clearly an
improvement. Further testing will show whether more is needed.
The reverted changes were intended to force a choice of whether any
newly-added BufferGetPage() calls needed to be accompanied by a
test of the snapshot age, to support the "snapshot too old"
feature. Such an accompanying test is needed in about 7% of the
cases, where the page is being used as part of a scan rather than
positioning for other purposes (such as DML or vacuuming). The
additional effort required for back-patching, and the doubt whether
the intended benefit would really be there, have indicated it is
best just to rely on developers to do the right thing based on
comments and existing usage, as we do with many other conventions.
This change should have little or no effect on generated executable
code.
Motivated by the back-patching pain of Tom Lane and Robert Haas
Coverity complained that oldPartitionLock was possibly dereferenced after
having been set to NULL. That actually can't happen, because we'd only use
it if (oldFlags & BM_TAG_VALID) is true. But nonetheless Coverity is
justified in complaining, because at line 1275 we actually overwrite
oldFlags, and then still expect its BM_TAG_VALID bit to be a safe guide to
whether to release the oldPartitionLock. Thus, the code would be incorrect
if someone else had changed the buffer's BM_TAG_VALID flag meanwhile.
That should not happen, since we hold pin on the buffer throughout this
sequence, but it's starting to look like a rather shaky chain of logic.
And there's no need for such assumptions, because we can simply replace
the (oldFlags & BM_TAG_VALID) tests with (oldPartitionLock != NULL),
which has identical results and makes it plain to all comers that we don't
dereference a null pointer. A small side benefit is that the range of
liveness of oldFlags is greatly reduced, possibly allowing the compiler
to save a register.
This is just cleanup, not an actual bug fix, so there seems no need
for a back-patch.
We've had repeated troubles over the years with failures to initialize
spinlocks correctly; see 6b93fcd14 for a recent example. Most of the time,
on most platforms, such oversights can escape notice because all-zeroes is
the expected initial content of an slock_t variable. The only platform
we have where the initialized state of an slock_t isn't zeroes is HPPA,
and that's practically gone in the wild. To make it easier to catch such
errors without needing one of those, adjust the --disable-spinlocks code
so that zero is not a valid value for an slock_t for it.
In passing, remove a bunch of unnecessary #include's from spin.c;
commit daa7527afc removed all the intermodule coupling that
made them necessary.
pg_xlogdump includes bufmgr.h. With a compiler that emits code for
static inline functions even when they're unreferenced, that leads
to unresolved external references in the new static-inline version
of BufferGetPage(). So hide it with #ifndef FRONTEND, as we've done
for similar issues elsewhere. Per buildfarm member pademelon.
My previous attempt at doing so, in 80abbeba23, was not sufficient. While that
fixed the problem for bufmgr.c and lwlock.c , s_lock.c still has non-constant
expressions in the struct initializer, because the file/line/function
information comes from the caller of s_lock().
Give up on using a macro, and use a static inline instead.
Discussion: 4369.1460435533@sss.pgh.pa.us
The current definition of init_spin_delay (introduced recently in
48354581a) wasn't C89 compliant. It's not legal to refer to refer to
non-constant expressions, and the ptr argument was one. This, as
reported by Tom, lead to a failure on buildfarm animal pademelon.
The pointer, especially on system systems with ASLR, isn't super helpful
anyway, though. So instead of making init_spin_delay into an inline
function, make s_lock_stuck() report the function name in addition to
file:line and change init_spin_delay() accordingly. While not a direct
replacement, the function name is likely more useful anyway (line
numbers are often hard to interpret in third party reports).
This also fixes what file/line number is reported for waits via
s_lock().
As PG_FUNCNAME_MACRO is now used outside of elog.h, move it to c.h.
Reported-By: Tom Lane
Discussion: 4369.1460435533@sss.pgh.pa.us
The recent patch to make Pin/UnpinBuffer lockfree in the hot
path (48354581a), accidentally used pg_atomic_fetch_or_u32() in
MarkLocalBufferDirty(). Other code operating on local buffers was
careful to only use pg_atomic_read/write_u32 which just read/write from
memory; to avoid unnecessary overhead.
On its own that'd just make MarkLocalBufferDirty() slightly less
efficient, but in addition InitLocalBuffers() doesn't call
pg_atomic_init_u32() - thus the spinlock fallback for the atomic
operations isn't initialized. That in turn caused, as reported by Tom,
buildfarm animal gaur to fail. As those errors are actually useful
against this type of error, continue to omit - intentionally this time -
initialization of the atomic variable.
In addition, add an explicit note about only using pg_atomic_read/write
on local buffers's state to BufferDesc's description.
Reported-By: Tom Lane
Discussion: 1881.1460431476@sss.pgh.pa.us
It's silly to define these counts as narrower than they might someday
need to be. Also, I believe that the BLCKSZ * nflush calculation in
mdwriteback was capable of overflowing an int.
Commit 428b1d6b29 introduced the use of
msync() for flushing dirty data from the kernel's file buffers. Several
portability issues were overlooked, though:
* Not all implementations of mmap() think that nbytes == 0 means "map
the whole file". To fix, use lseek() to find out the true length.
Fix callers of pg_flush_data to be aware that nbytes == 0 may result
in trashing the file's seek position.
* Not all implementations of mmap() will accept partial-page mmap
requests. To fix, round down the length request to whatever sysconf()
says the page size is. (I think this is OK from a portability standpoint,
because sysconf() is required by SUS v2, and we aren't trying to compile
this part on Windows anyway. Buildfarm should let us know if not.)
* On 32-bit machines, the file size might exceed the available free
address space, or even exceed what will fit in size_t. Check for
the latter explicitly to avoid passing a false request size to mmap().
If mmap fails, silently fall through to the next implementation method,
rather than bleating to the postmaster log and giving up.
* mmap'ing directories fails on some platforms, and even if it works,
msync'ing the directory is quite unlikely to help, as for that matter are
the other flush implementations. In pre_sync_fname(), just skip flush
attempts on directories.
In passing, copy-edit the comments a bit.
Stas Kelvich and myself
On a big NUMA machine with 1000 connections in saturation load
there was a performance regression due to spinlock contention, for
acquiring values which were never used. Just fill with dummy
values if we're not going to use them.
This patch has not been benchmarked yet on a big NUMA machine, but
it seems like a good idea on general principle, and it seemed to
prevent an apparent 2.2% regression on a single-socket i7 box
running 200 connections at saturation load.
I was initially concerned that the some of the hundreds of
references to BufferGetPage() where the literal
BGP_NO_SNAPSHOT_TEST were passed might not optimize as well as a
macro, leading to some hard-to-find performance regressions in
corner cases. Inspection of disassembled code has shown identical
code at all inspected locations, and the size difference doesn't
amount to even one byte per such call. So make it readable.
Per gripes from Álvaro Herrera and Tom Lane
Previously we used a spinlock, in adition to the atomically manipulated
->state field, to protect the wait queue. But it's pretty simple to
instead perform the locking using a flag in state.
Due to 6150a1b0 BufferDescs, on platforms (like PPC) with > 1 byte
spinlocks, increased their size above 64byte. As 64 bytes are the size
we pad allocated BufferDescs to, this can increase false sharing;
causing performance problems in turn. Together with the previous commit
this reduces the size to <= 64 bytes on all common platforms.
Author: Andres Freund
Discussion: CAA4eK1+ZeB8PMwwktf+3bRS0Pt4Ux6Rs6Aom0uip8c6shJWmyg@mail.gmail.com20160327121858.zrmrjegmji2ymnvr@alap3.anarazel.de
Pinning/Unpinning a buffer is a very frequent operation; especially in
read-mostly cache resident workloads. Benchmarking shows that in various
scenarios the spinlock protecting a buffer header's state becomes a
significant bottleneck. The problem can be reproduced with pgbench -S on
larger machines, but can be considerably worse for queries which touch
the same buffers over and over at a high frequency (e.g. nested loops
over a small inner table).
To allow atomic operations to be used, cram BufferDesc's flags,
usage_count, buf_hdr_lock, refcount into a single 32bit atomic variable;
that allows to manipulate them together using 32bit compare-and-swap
operations. This requires reducing MAX_BACKENDS to 2^18-1 (which could
be lifted by using a 64bit field, but it's not a realistic configuration
atm).
As not all operations can easily implemented in a lockfree manner,
implement the previous buf_hdr_lock via a flag bit in the atomic
variable. That way we can continue to lock the header in places where
it's needed, but can get away without acquiring it in the more frequent
hot-paths. There's some additional operations which can be done without
the lock, but aren't in this patch; but the most important places are
covered.
As bufmgr.c now essentially re-implements spinlocks, abstract the delay
logic from s_lock.c into something more generic. It now has already two
users, and more are coming up; there's a follupw patch for lwlock.c at
least.
This patch is based on a proof-of-concept written by me, which Alexander
Korotkov made into a fully working patch; the committed version is again
revised by me. Benchmarking and testing has, amongst others, been
provided by Dilip Kumar, Alexander Korotkov, Robert Haas.
On a large x86 system improvements for readonly pgbench, with a high
client count, of a factor of 8 have been observed.
Author: Alexander Korotkov and Andres Freund
Discussion: 2400449.GjM57CE0Yg@dinodell
This feature is controlled by a new old_snapshot_threshold GUC. A
value of -1 disables the feature, and that is the default. The
value of 0 is just intended for testing. Above that it is the
number of minutes a snapshot can reach before pruning and vacuum
are allowed to remove dead tuples which the snapshot would
otherwise protect. The xmin associated with a transaction ID does
still protect dead tuples. A connection which is using an "old"
snapshot does not get an error unless it accesses a page modified
recently enough that it might not be able to produce accurate
results.
This is similar to the Oracle feature, and we use the same SQLSTATE
and error message for compatibility.
This patch is a no-op patch which is intended to reduce the chances
of failures of omission once the functional part of the "snapshot
too old" patch goes in. It adds parameters for snapshot, relation,
and an enum to specify whether the snapshot age check needs to be
done for the page at this point. This initial patch passes NULL
for the first two new parameters and BGP_NO_SNAPSHOT_TEST for the
third. The follow-on patch will change the places where the test
needs to be made.
Contention on the relation extension lock can become quite fierce when
multiple processes are inserting data into the same relation at the same
time at a high rate. Experimentation shows the extending the relation
multiple blocks at a time improves scalability.
Dilip Kumar, reviewed by Petr Jelinek, Amit Kapila, and me.
Experimentation shows this only costs about 6kB, which seems well
worth it given the major performance effects that can be caused
by insufficient alignment, especially on larger systems.
Discussion: 14166.1458924422@sss.pgh.pa.us
If archive_timeout > 0 we should avoid logging XLOG_RUNNING_XACTS if idle.
Bug 13685 reported by Laurence Rowe, investigated in detail by Michael Paquier,
though this is not his proposed fix.
20151016203031.3019.72930@wrigleys.postgresql.org
Simple non-invasive patch to allow later backpatch to 9.4 and 9.5
Commit ac1d794 ("Make idle backends exit if the postmaster dies.")
introduced a regression on, at least, large linux systems. Constantly
adding the same postmaster_alive_fds to the OSs internal datastructures
for implementing poll/select can cause significant contention; leading
to a performance regression of nearly 3x in one example.
This can be avoided by using e.g. linux' epoll, which avoids having to
add/remove file descriptors to the wait datastructures at a high rate.
Unfortunately the current latch interface makes it hard to allocate any
persistent per-backend resources.
Replace, with a backward compatibility layer, WaitLatchOrSocket with a
new WaitEventSet API. Users can allocate such a Set across multiple
calls, and add more than one file-descriptor to wait on. The latter has
been added because there's upcoming postgres features where that will be
helpful.
In addition to the previously existing poll(2), select(2),
WaitForMultipleObjects() implementations also provide an epoll_wait(2)
based implementation to address the aforementioned performance
problem. Epoll is only available on linux, but that is the most likely
OS for machines large enough (four sockets) to reproduce the problem.
To actually address the aforementioned regression, create and use a
long-lived WaitEventSet for FE/BE communication. There are additional
places that would benefit from a long-lived set, but that's a task for
another day.
Thanks to Amit Kapila, who helped make the windows code I blindly wrote
actually work.
Reported-By: Dmitry Vasilyev Discussion:
CAB-SwXZh44_2ybvS5Z67p_CDz=XFn4hNAD=CnMEF+QqkXwFrGg@mail.gmail.com20160114143931.GG10941@awork2.anarazel.de
Previously latches for windows and unix had been implemented in
different files. A later patch introduce an expanded wait
infrastructure, keeping the implementation separate would introduce too
much duplication.
This basically just moves the functions, without too much change. The
reason to keep this separate is that it allows blame to continue working
a little less badly; and to make review a tiny bit easier.
Discussion: 20160114143931.GG10941@awork2.anarazel.de
Buildfarm members gaur and pademelon are old enough not to know about
MAP_FAILED; which is used in 428b1d6. Include portability/mem.h to fix;
as already done in a bunch of other places.
The old code is wrong, because it returns a pointer to an automatic
variable. And it's also more clever than we really need to be
considering that the case it's worrying about should never happen.
Up to now checkpoints were written in the order they're in the
BufferDescriptors. That's nearly random in a lot of cases, which
performs badly on rotating media, but even on SSDs it causes slowdowns.
To avoid that, sort checkpoints before writing them out. We currently
sort by tablespace, relfilenode, fork and block number.
One of the major reasons that previously wasn't done, was fear of
imbalance between tablespaces. To address that balance writes between
tablespaces.
The other prime concern was that the relatively large allocation to sort
the buffers in might fail, preventing checkpoints from happening. Thus
pre-allocate the required memory in shared memory, at server startup.
This particularly makes it more efficient to have checkpoint flushing
enabled, because that'll often result in a lot of writes that can be
coalesced into one flush.
Discussion: alpine.DEB.2.10.1506011320000.28433@sto
Author: Fabien Coelho and Andres Freund
Currently writes to the main data files of postgres all go through the
OS page cache. This means that some operating systems can end up
collecting a large number of dirty buffers in their respective page
caches. When these dirty buffers are flushed to storage rapidly, be it
because of fsync(), timeouts, or dirty ratios, latency for other reads
and writes can increase massively. This is the primary reason for
regular massive stalls observed in real world scenarios and artificial
benchmarks; on rotating disks stalls on the order of hundreds of seconds
have been observed.
On linux it is possible to control this by reducing the global dirty
limits significantly, reducing the above problem. But global
configuration is rather problematic because it'll affect other
applications; also PostgreSQL itself doesn't always generally want this
behavior, e.g. for temporary files it's undesirable.
Several operating systems allow some control over the kernel page
cache. Linux has sync_file_range(2), several posix systems have msync(2)
and posix_fadvise(2). sync_file_range(2) is preferable because it
requires no special setup, whereas msync() requires the to-be-flushed
range to be mmap'ed. For the purpose of flushing dirty data
posix_fadvise(2) is the worst alternative, as flushing dirty data is
just a side-effect of POSIX_FADV_DONTNEED, which also removes the pages
from the page cache. Thus the feature is enabled by default only on
linux, but can be enabled on all systems that have any of the above
APIs.
While desirable and likely possible this patch does not contain an
implementation for windows.
With the infrastructure added, writes made via checkpointer, bgwriter
and normal user backends can be flushed after a configurable number of
writes. Each of these sources of writes controlled by a separate GUC,
checkpointer_flush_after, bgwriter_flush_after and backend_flush_after
respectively; they're separate because the number of flushes that are
good are separate, and because the performance considerations of
controlled flushing for each of these are different.
A later patch will add checkpoint sorting - after that flushes from the
ckeckpoint will almost always be desirable. Bgwriter flushes are most of
the time going to be random, which are slow on lots of storage hardware.
Flushing in backends works well if the storage and bgwriter can keep up,
but if not it can have negative consequences. This patch is likely to
have negative performance consequences without checkpoint sorting, but
unfortunately so has sorting without flush control.
Discussion: alpine.DEB.2.10.1506011320000.28433@sto
Author: Fabien Coelho and Andres Freund
Earlier version committed in 9.0 caused spurious waits in some cases.
New infrastructure for lock waits in 9.3 used to correct and improve this.
Jeff Janes based upon a proposal by Simon Riggs, who also reviewed
Additional review comments from Amit Kapila
When a process is waiting for a heavyweight lock, we will now indicate
the type of heavyweight lock for which it is waiting. Also, you can
now see when a process is waiting for a lightweight lock - in which
case we will indicate the individual lock name or the tranche, as
appropriate - or for a buffer pin.
Amit Kapila, Ildus Kurbangaliev, reviewed by me. Lots of helpful
discussion and suggestions by many others, including Alexander
Korotkov, Vladimir Borodin, and many others.
Renaming a file using rename(2) is not guaranteed to be durable in face
of crashes; especially on filesystems like xfs and ext4 when mounted
with data=writeback. To be certain that a rename() atomically replaces
the previous file contents in the face of crashes and different
filesystems, one has to fsync the old filename, rename the file, fsync
the new filename, fsync the containing directory. This sequence is not
generally adhered to currently; which exposes us to data loss risks. To
avoid having to repeat this arduous sequence, introduce
durable_rename(), which wraps all that.
Also add durable_link_or_rename(). Several places use link() (with a
fallback to rename()) to rename a file, trying to avoid replacing the
target file out of paranoia. Some of those rename sequences need to be
durable as well. There seems little reason extend several copies of the
same logic, so centralize the link() callers.
This commit does not yet make use of the new functions; they're used in
a followup commit.
Author: Michael Paquier, Andres Freund
Discussion: 56583BDD.9060302@2ndquadrant.com
Backpatch: All supported branches
For example, if you want to perform an ioctl() on a file descriptor
opened through the fd.c routines, there's no way to do that without
being able to get at the underlying fd.
KaiGai Kohei
The new bit indicates whether every tuple on the page is already frozen.
It is cleared only when the all-visible bit is cleared, and it can be
set only when we vacuum a page and find that every tuple on that page is
both visible to every transaction and in no need of any future
vacuuming.
A future commit will use this new bit to optimize away full-table scans
that would otherwise be triggered by XID wraparound considerations. A
page which is merely all-visible must still be scanned in that case, but
a page which is all-frozen need not be. This commit does not attempt
that optimization, although that optimization is the goal here. It
seems better to get the basic infrastructure in place first.
Per discussion, it's very desirable for pg_upgrade to automatically
migrate existing VM forks from the old format to the new format. That,
too, will be handled in a follow-on patch.
Masahiko Sawada, reviewed by Kyotaro Horiguchi, Fujii Masao, Amit
Kapila, Simon Riggs, Andres Freund, and others, and substantially
revised by me.
This patch introduces "pg_blocking_pids(int) returns int[]", which returns
the PIDs of any sessions that are blocking the session with the given PID.
Historically people have obtained such information using a self-join on
the pg_locks view, but it's unreasonably tedious to do it that way with any
modicum of correctness, and the addition of parallel queries has pretty
much broken that approach altogether. (Given some more columns in the view
than there are today, you could imagine handling parallel-query cases with
a 4-way join; but ugh.)
The new function has the following behaviors that are painful or impossible
to get right via pg_locks:
1. Correctly understands which lock modes block which other ones.
2. In soft-block situations (two processes both waiting for conflicting lock
modes), only the one that's in front in the wait queue is reported to
block the other.
3. In parallel-query cases, reports all sessions blocking any member of
the given PID's lock group, and reports a session by naming its leader
process's PID, which will be the pg_backend_pid() value visible to
clients.
The motivation for doing this right now is mostly to fix the isolation
tests. Commit 38f8bdcac4 lobotomized
isolationtester's is-it-waiting query by removing its ability to recognize
nonconflicting lock modes, as a crude workaround for the inability to
handle soft-block situations properly. But even without the lock mode
tests, the old query was excessively slow, particularly in
CLOBBER_CACHE_ALWAYS builds; some of our buildfarm animals fail the new
deadlock-hard test because the deadlock timeout elapses before they can
probe the waiting status of all eight sessions. Replacing the pg_locks
self-join with use of pg_blocking_pids() is not only much more correct, but
a lot faster: I measure it at about 9X faster in a typical dev build with
Asserts, and 3X faster in CLOBBER_CACHE_ALWAYS builds. That should provide
enough headroom for the slower CLOBBER_CACHE_ALWAYS animals to pass the
test, without having to lengthen deadlock_timeout yet more and thus slow
down the test for everyone else.
We don't really need this field, because it's either zero or redundant with
PGPROC.pid. The use of zero to mark "not a group leader" is not necessary
since we can just as well test whether lockGroupLeader is NULL. This does
not save very much, either as to code or data, but the simplification seems
worthwhile anyway.
In 4b4b680c3 I accidentally used sizeof(PrivateRefCountArray) instead of
sizeof(PrivateRefCountEntry) when creating the refcount overflow
hashtable. As the former is bigger than the latter, this luckily only
resulted in a slightly increased memory usage when many buffers are
pinned in a backend.
Reported-By: Takashi Horikawa
Discussion: 73FA3881462C614096F815F75628AFCD035A48C3@BPXM01GP.gisp.nec.co.jp
Backpatch: 9.5, where thew new ref count infrastructure was introduced
Reflow text in lock manager README so that it fits within 80 columns.
Correct some mistakes. Expand the README to explain not only why group
locking exists but also the data structures that support it. Improve
comments related to group locking several files. Change the name of a
macro argument for improved clarity.
Most of these problems were reported by Tom Lane, but I found a few
of them myself.
Robert Haas and Tom Lane
The API spec for this function was changed completely (and for the better)
by commit 3cba8999b3, but it didn't bother
with anything as mundane as updating the comments.
This finishes the work - spread across many commits over the last
several months - of putting each type of lock other than the named
individual locks into a separate tranche.
Amit Kapila
Commit 0e141c0fbb introduced a new
facility to reduce ProcArrayLock contention by clearing several XIDs
from the ProcArray under a single lock acquisition. The names
initially chosen were deemed not to be very good choices, so commit
4aec49899e renamed them. But now it
seems like we still didn't get it right. A pending patch wants to
add similar infrastructure for batching CLOG updates, so the names
need to be clear enough to allow a new set of structure members with
a related purpose.
Amit Kapila
As of commit c1772ad922, there's no
longer any way of requesting additional LWLocks in the main tranche,
so we don't need NumLWLocks() or LWLockAssign() any more. Also,
some of the allocation counters that we had previously aren't needed
any more either.
Amit Kapila
Further investigation says that there may be some slow operations after
we've finished ShutdownXLOG(), so add some more log messages to try to
isolate that. This is all temporary code too.
Early returns from the buildfarm show that there's a bit of a gap in the
logging I added in 3971f64843: the portion of CreateCheckPoint()
after CheckPointGuts() can take a fair amount of time. Add a few more
log messages in that section of code. This too shall be reverted later.
This is a quick hack, due to be reverted when its purpose has been served,
to try to gather information about why some of the buildfarm critters
regularly fail with "postmaster does not shut down" complaints. Maybe they
are just really overloaded, but maybe something else is going on. Hence,
instrument pg_ctl to print the current time when it starts waiting for
postmaster shutdown and when it gives up, and add a lot of logging of the
current time in the server's checkpoint and shutdown code paths.
No attempt has been made to make this pretty. I'm not even totally sure
if it will build on Windows, but we'll soon find out.
For locking purposes, we now regard heavyweight locks as mutually
non-conflicting between cooperating parallel processes. There are some
possible pitfalls to this approach that are not to be taken lightly,
but it works OK for now and can be changed later if we find a better
approach. Without this, it's very easy for parallel queries to
silently self-deadlock if the user backend holds strong relation locks.
Robert Haas, with help from Amit Kapila. Thanks to Noah Misch and
Andres Freund for extensive discussion of possible issues with this
approach.
The previous RequestAddinLWLocks() method had several disadvantages.
First, the locks would be in the main tranche; we've recently decided
that it's useful for LWLocks used for separate purposes to have
separate tranche IDs. Second, there wasn't any correlation between
what code called RequestAddinLWLocks() and what code called
LWLockAssign(); when multiple modules are in use, it could become
quite difficult to troubleshoot problems where LWLockAssign() ran out
of locks. To fix, create a concept of named LWLock tranches which
can be used either by extension or by core code.
Amit Kapila and Robert Haas
This is following in a long train of similar changes and for the same
reasons - see b319356f0e and
fe702a7b3f inter alia.
Author: Amit Kapila
Reviewed-by: Alexander Korotkov, Robert Haas
Previously, each PGPROC's backendLock was part of the main tranche,
and the PGPROC just contained a pointer. Now, the actual LWLock is
part of the PGPROC.
As with previous, similar patches, this makes it significantly easier
to identify these lwlocks in LWLOCK_STATS or Trace_lwlocks output
and improves modularity.
Author: Ildus Kurbangaliev
Reviewed-by: Amit Kapila, Robert Haas
It's entirely surprising that mdnblocks() has the side effect of
creating new files on disk, so let's make it not do that. One
consequence of the old behavior is that, if running on a damaged
cluster that is missing a file, mdnblocks() can recreate the file
and allow a subsequent _mdfd_getseg() for a higher segment to succeed.
This happens because, while mdnblocks() stops when it finds a segment
that is shorter than 1GB, _mdfd_getseg() has no such check, and thus
the empty file created by mdnblocks() can allow it to continue its
traversal and find higher-numbered segments which remain.
It might be a good idea for _mdfd_getseg() to actually verify that
each segment it finds is exactly 1GB before proceeding to the next
one, but that would involve some additional system calls, so for
now I'm just doing this much.
Patch by me, per off-list analysis by Kevin Grittner and Rahila Syed.
Review by Andres Freund.
Move the content lock directly into the BufferDesc, so that locking and
pinning a buffer touches only one cache line rather than two. Adjust
the definition of BufferDesc slightly so that this doesn't make the
BufferDesc any larger than one cache line (at least on platforms where
a spinlock is only 1 or 2 bytes).
We can't fit the I/O locks into the BufferDesc and stay within one
cache line, so move those to a completely separate tranche. This
leaves a relatively limited number of LWLocks in the main tranche, so
increase the padding of those remaining locks to a full cache line,
rather than allowing adjacent locks to share a cache line, hopefully
reducing false sharing.
Performance testing shows that these changes make little difference
on laptop-class machines, but help significantly on larger servers,
especially those with more than 2 sockets.
Andres Freund, originally based on an earlier patch by Simon Riggs.
Review and cosmetic adjustments (including heavy rewriting of the
comments) by me.
It's a bit cumbersome to use LWLockNewTrancheId(), because the returned
value needs to be shared between backends so that each backend can call
LWLockRegisterTranche() with the correct ID. So, for built-in tranches,
use a hard-coded value instead.
This is motivated by an upcoming patch adding further built-in tranches.
Andres Freund and Robert Haas
e3f4cfc7 introduced a LWLockHeldByMe() call, without the corresponding
Assert() surrounding it.
Spotted by Coverity.
Backpatch: 9.1+, like the previous commit
At the end of crash recovery, unlogged relations are reset to the empty
state, using their init fork as the template. The init fork is copied to
the main fork without going through shared buffers. Unfortunately WAL
replay so far has not necessarily flushed writes from shared buffers to
disk at that point. In normal crash recovery, and before the
introduction of 'fast promotions' in fd4ced523 / 9.3, the
END_OF_RECOVERY checkpoint flushes the buffers out in time. But with
fast promotions that's not the case anymore.
To fix, force WAL writes targeting the init fork to be flushed
immediately (using the new FlushOneBuffer() function). In 9.5+ that
flush can centrally be triggered from the code dealing with restoring
full page writes (XLogReadBufferForRedoExtended), in earlier releases
that responsibility is in the hands of XLOG_HEAP_NEWPAGE's replay
function.
Backpatch to 9.1, even if this currently is only known to trigger in
9.3+. Flushing earlier is more robust, and it is advantageous to keep
the branches similar.
Typical symptoms of this bug are errors like
'ERROR: index "..." contains unexpected zero page at block 0'
shortly after promoting a node.
Reported-By: Thom Brown
Author: Andres Freund and Michael Paquier
Discussion: 20150326175024.GJ451@alap3.anarazel.de
Backpatch: 9.1-
Prior to commit 0709b7ee72, access to
variables within a spinlock-protected critical section had to be done
through a volatile pointer, but that should no longer be necessary.
Review by Andres Freund
This makes it significantly easier to identify these lwlocks in
LWLOCK_STATS or Trace_lwlocks output. It's also arguably better
from a modularity standpoint, since lwlock.c no longer needs to
know anything about the LWLock needs of the higher-level SLRU
facility.
Ildus Kurbangaliev, reviewd by Álvaro Herrera and by me.
Commit a1480ec1d3 purported to fix the
problems with commit b2ccb5f4e6, but it
didn't completely fix them. The problem is that the checks were
performed in the wrong order, leading to a race condition. If the
sender attached, sent a message, and detached after the receiver
called shm_mq_get_sender and before the receiver called
shm_mq_counterparty_gone, we'd incorrectly return SHM_MQ_DETACHED
before all messages were read. Repair by reversing the order of
operations, and add a long comment explaining why this new logic is
(hopefully) correct.
On insert the CheckForSerializableConflictIn() test was performed
before the page(s) which were going to be modified had been locked
(with an exclusive buffer content lock). If another process
acquired a relation SIReadLock on the heap and scanned to a page on
which an insert was going to occur before the page was so locked,
a rw-conflict would be missed, which could allow a serialization
anomaly to be missed. The window between the check and the page
lock was small, so the bug was generally not noticed unless there
was high concurrency with multiple processes inserting into the
same table.
This was reported by Peter Bailis as bug #11732, by Sean Chittenden
as bug #13667, and by others.
The race condition was eliminated in heap_insert() by moving the
check down below the acquisition of the buffer lock, which had been
the very next statement. Because of the loop locking and unlocking
multiple buffers in heap_multi_insert() a check was added after all
inserts were completed. The check before the start of the inserts
was left because it might avoid a large amount of work to detect a
serialization anomaly before performing the all of the inserts and
the related WAL logging.
While investigating this bug, other SSI bugs which were even harder
to hit in practice were noticed and fixed, an unnecessary check
(covered by another check, so redundant) was removed from
heap_update(), and comments were improved.
Back-patch to all supported branches.
Kevin Grittner and Thomas Munro
If the counterparty writes some data into the queue and then detaches,
it's wrong to return SHM_MQ_DETACHED right away. If we do that, we
fail to read whatever was written.
The shm_mq mechanism was intended to optionally notice when the process
on the other end of the queue fails to attach to the queue. It does
this by allowing the user to pass a BackgroundWorkerHandle; if the
background worker in question is launched and dies without attaching
to the queue, then we know it never will. This logic works OK in
blocking mode, but when called with nowait = true we fail to notice
that this has happened due to an asymmetry in the logic. Repair.
Reported off-list by Rushabh Lathia. Patch by me.
Prior to commit 0709b7ee72, access to
variables within a spinlock-protected critical section had to be done
through a volatile pointer, but that should no longer be necessary.
Michael Paquier
Prior to commit 0709b7ee72, access to
variables within a spinlock-protected critical section had to be done
through a volatile pointer, but that should no longer be necessary.
Thomas Munro
This flag has proven to be a recipe for bugs, and it doesn't seem like
it can really buy anything in terms of performance. So let's just
*always* set the process latch when we receive SIGUSR1 instead of
trying to do it only when needed.
Per my recent proposal on pgsql-hackers.
The fact that multixact truncations are not WAL logged has caused a fair
share of problems. Amongst others it requires to do computations during
recovery while the database is not in a consistent state, delaying
truncations till checkpoints, and handling members being truncated, but
offset not.
We tried to put bandaids on lots of these issues over the last years,
but it seems time to change course. Thus this patch introduces WAL
logging for multixact truncations.
This allows:
1) to perform the truncation directly during VACUUM, instead of delaying it
to the checkpoint.
2) to avoid looking at the offsets SLRU for truncation during recovery,
we can just use the master's values.
3) simplify a fair amount of logic to keep in memory limits straight,
this has gotten much easier
During the course of fixing this a bunch of additional bugs had to be
fixed:
1) Data was not purged from memory the member's SLRU before deleting
segments. This happened to be hard or impossible to hit due to the
interlock between checkpoints and truncation.
2) find_multixact_start() relied on SimpleLruDoesPhysicalPageExist - but
that doesn't work for offsets that haven't yet been flushed to
disk. Add code to flush the SLRUs to fix. Not pretty, but it feels
slightly safer to only make decisions based on actual on-disk state.
3) find_multixact_start() could be called concurrently with a truncation
and thus fail. Via SetOffsetVacuumLimit() that could lead to a round
of emergency vacuuming. The problem remains in
pg_get_multixact_members(), but that's quite harmless.
For now this is going to only get applied to 9.5+, leaving the issues in
the older branches in place. It is quite possible that we need to
backpatch at a later point though.
For the case this gets backpatched we need to handle that an updated
standby may be replaying WAL from a not-yet upgraded primary. We have to
recognize that situation and use "old style" truncation (i.e. looking at
the SLRUs) during WAL replay. In contrast to before, this now happens in
the startup process, when replaying a checkpoint record, instead of the
checkpointer. Doing truncation in the restartpoint is incorrect, they
can happen much later than the original checkpoint, thereby leading to
wraparound. To avoid "multixact_redo: unknown op code 48" errors
standbys would have to be upgraded before primaries.
A later patch will bump the WAL page magic, and remove the legacy
truncation codepaths. Legacy truncation support is just included to make
a possible future backpatch easier.
Discussion: 20150621192409.GA4797@alap3.anarazel.de
Reviewed-By: Robert Haas, Alvaro Herrera, Thomas Munro
Backpatch: 9.5 for now
RemoveLocalLock() must consider the possibility that LockAcquireExtended()
failed to palloc the initial space for a locallock's lockOwners array.
I had evidently meant to cope with this hazard when the code was originally
written (commit 1785acebf2), but missed that
the pfree needed to be protected with an if-test. Just to make sure things
are left in a clean state, reset numLockOwners as well.
Per low-memory testing by Andreas Seltenreich. Back-patch to all supported
branches.
The shm_mq mechanism was built to send error (and notice) messages and
tuples between backends. However, shm_mq itself only deals in raw
bytes. Since commit 2bd9e412f9, we have
had infrastructure for one message to redirect protocol messages to a
queue and for another backend to parse them and do useful things with
them. This commit introduces a somewhat analogous facility for tuples
by adding a new type of DestReceiver, DestTupleQueue, which writes
each tuple generated by a query into a shm_mq, and a new
TupleQueueFunnel facility which reads raw tuples out of the queue and
reconstructs the HeapTuple format expected by the executor.
The TupleQueueFunnel abstraction supports reading from multiple tuple
streams at the same time, but only in round-robin fashion. Someone
could imaginably want other policies, but this should be good enough
to meet our short-term needs related to parallel query, and we can
always extend it later.
This also makes one minor addition to the shm_mq API that didn'
seem worth breaking out as a separate patch.
Extracted from Amit Kapila's parallel sequential scan patch. This
code was originally written by me, and then it was revised by Amit,
and then it was revised some more by me.
Naming the individual lwlocks seems like something that may be useful
for other types of debugging, monitoring, or instrumentation output,
but this commit just implements it for the specific case of
trace_lwlocks.
Patch by me, reviewed by Amit Kapila and Kyotaro Horiguchi
Per discussion, nowadays it is possible to have tablespaces that have
wildly different I/O characteristics from others. Setting different
effective_io_concurrency parameters for those has been measured to
improve performance.
Author: Julien Rouhaud
Reviewed by: Andres Freund
Post-commit review by Andres Freund discovered a couple of concurrency
bugs in the original patch: specifically, if the leader cleared a
follower's XID before it reached PGSemaphoreLock, the semaphore would be
left in the wrong state; and if another process did PGSemaphoreUnlock
for some unrelated reason, we might resume execution before the fact
that our XID was cleared was globally visible.
Also, improve the wording of some comments, rename nextClearXidElem
to firstClearXidElem in PROC_HDR for clarity, and drop some volatile
qualifiers that aren't necessary.
Amit Kapila, reviewed and slightly revised by me.
This fixes a bunch of somewhat pedantic warnings with new
compilers. Since by far the majority of other functions definitions use
the (void) style it just seems to be consistent to do so as well in the
remaining few places.
In 4b4b680c I passed a buffer index number (starting from 0) instead of
a proper Buffer id (which start from 1 for shared buffers) in two
places.
This wasn't noticed so far as one of those locations isn't compiled at
all (PrintPinnedBufs) and the other one (InvalidBuffer) requires a
unlikely, but possible, set of circumstances to trigger a symptom.
To reduce the likelihood of such incidents a bit also convert existing
open coded mappings from buffer descriptors to buffer ids with
BufferDescriptorGetBuffer().
Author: Qingqing Zhou
Reported-By: Qingqing Zhou
Discussion: CAJjS0u2ai9ooUisKtkV8cuVUtEkMTsbK8c7juNAjv8K11zeCQg@mail.gmail.com
Backpatch: 9.5 where the private ref count infrastructure was introduced
If some, but not all, of the length word has already been read, and the
next attempt to read sees exactly the number of bytes needed to complete
the length word, or fewer, then we'll incorrectly read less than all of
the available data.
Antonin Houska
When a write transaction commits, it must clear its XID advertised via
the ProcArray, which requires that we hold ProcArrayLock in exclusive
mode in order to prevent concurrent processes running GetSnapshotData
from seeing inconsistent results. When many processes try to commit
at once, ProcArrayLock must change hands repeatedly, with each
concurrent process trying to commit waking up to acquire the lock in
turn. To make things more efficient, when more than one backend is
trying to commit a write transaction at the same time, have just one
of them acquire ProcArrayLock in exclusive mode and clear the XIDs of
all processes in the group. Benchmarking reveals that this is much
more efficient at very high client counts.
Amit Kapila, heavily revised by me, with some review also from Pavan
Deolasee.
LWLockAttemptLock pointlessly read the lock's state in every loop
iteration, even though pg_atomic_compare_exchange_u32() returns the old
value. Instead do that only once before the loop iteration.
Additionally there's no need to have the expected_state variable,
old_state mostly had the same value anyway.
Noticed-By: Heikki Linnakangas
Backpatch: 9.5, no reason to let the branches diverge at this point
The lwlock scalability work introduced two race conditions into the
lwlock variable support provided for xlog.c. First, and harmlessly on
most platforms, it set/read the variable without the spinlock in some
places. Secondly, due to the removal of the spinlock, it was possible
that a backend missed changes to the variable's state if it changed in
the wrong moment because checking the lock's state, the variable's state
and the queuing are not protected by a single spinlock acquisition
anymore.
To fix first move resetting the variable's from LWLockAcquireWithVar to
WALInsertLockRelease, via a new function LWLockReleaseClearVar. That
prevents issues around waiting for a variable's value to change when a
new locker has acquired the lock, but not yet set the value. Secondly
re-check that the variable hasn't changed after enqueing, that prevents
the issue that the lock has been released and already re-acquired by the
time the woken up backend checks for the lock's state.
Reported-By: Jeff Janes
Analyzed-By: Heikki Linnakangas
Reviewed-By: Heikki Linnakangas
Discussion: 5592DB35.2060401@iki.fi
Backpatch: 9.5, where the lwlock scalability went in
Don't print a WARNING if we get ESRCH from a kill() that's attempting
to cancel an autovacuum worker. It's possible (and has been seen in the
buildfarm) that the worker is already gone by the time we are able to
execute the kill, in which case the failure is harmless. About the only
plausible reason for reporting such cases would be to help debug corrupted
lock table contents, but this is hardly likely to be the most important
symptom if that happens. Moreover issuing a WARNING might scare users
more than is warranted.
Also, since sending a signal to an autovacuum worker is now entirely a
routine thing, and the worker will log the query cancel on its end anyway,
reduce the message saying we're doing that from LOG to DEBUG1 level.
Very minor cosmetic cleanup as well.
Since the main practical reason for doing this is to avoid unnecessary
buildfarm failures, back-patch to all active branches.
This code was originally written as part of parallel query effort, but
it seems to have independent value, because if we make one decision
about where to get a PGPROC when we allocate and then put it back on a
different list at backend-exit time, bad things happen. This isn't
just a theoretical risk; we fixed an actual problem of this type in
commit e280c630a8.
1. Replay of the WAL record for setting a bit in the visibility map
contained an assertion that a full-page image of that record type can only
occur with checksums enabled. But it can also happen with wal_log_hints, so
remove the assertion. Unlike checksums, wal_log_hints can be changed on the
fly, so it would be complicated to figure out if it was enabled at the time
that the WAL record was generated.
2. wal_log_hints has the same effect on the locking needed to read the LSN
of a page as data checksums. BufferGetLSNAtomic() didn't get the memo.
Backpatch to 9.4, where wal_log_hints was added.
The argument that this is a sufficiently-expected case to be silently
ignored seems pretty thin. Andres had brought it up back when we were
still considering that most fsync failures should be hard errors, and it
probably would be legit not to fail hard for ETXTBSY --- but the same is
true for EROFS and other cases, which is why we gave up on hard failures.
ETXTBSY is surely not a normal case, so logging the failure seems fine
from here.
Commit 2ce439f337 introduced a rather serious
regression, namely that if its scan of the data directory came across any
un-fsync-able files, it would fail and thereby prevent database startup.
Worse yet, symlinks to such files also caused the problem, which meant that
crash restart was guaranteed to fail on certain common installations such
as older Debian.
After discussion, we agreed that (1) failure to start is worse than any
consequence of not fsync'ing is likely to be, therefore treat all errors
in this code as nonfatal; (2) we should not chase symlinks other than
those that are expected to exist, namely pg_xlog/ and tablespace links
under pg_tblspc/. The latter restriction avoids possibly fsync'ing a
much larger part of the filesystem than intended, if the user has left
random symlinks hanging about in the data directory.
This commit takes care of that and also does some code beautification,
mainly moving the relevant code into fd.c, which seems a much better place
for it than xlog.c, and making sure that the conditional compilation for
the pre_sync_fname pass has something to do with whether pg_flush_data
works.
I also relocated the call site in xlog.c down a few lines; it seems a
bit silly to be doing this before ValidateXLOGDirectoryStructure().
The similar logic in initdb.c ought to be made to match this, but that
change is noncritical and will be dealt with separately.
Back-patch to all active branches, like the prior commit.
Abhijit Menon-Sen and Tom Lane
Ensure that we null-terminate the result string (one place in pg_rewind).
Be paranoid about out-of-range results from readlink() (should not happen,
but there is no good reason for some call sites to be careful about it and
others not). Consistently use the whole buffer, not sometimes one byte
less. Ensure we emit an appropriate errcode() in all cases. Spell the
error messages the same way.
The only serious bug here is the missing null-termination in pg_rewind,
which is new code, so no need for a back-patch.
Abhijit Menon-Sen and Tom Lane
We want this struct to be exactly a series of 3 int16 words, no more
and no less. Historically, at least, some ARM compilers preferred to
pad it to 8 bytes unless coerced. Our old way of doing that was just
to use __attribute__((packed)), but as pointed out by Piotr Stefaniak,
that does too much: it also licenses the compiler to give the struct
only byte-alignment. We don't want that because it adds access overhead,
possibly quite significant overhead. According to the GCC manual, what
we want requires also specifying __attribute__((align(2))). It's not
entirely clear if all the relevant compilers accept this pragma as well,
but we can hope the buildfarm will tell us if not. We can also add a
static assertion that should fire if the compiler padded the struct.
Since the combination of these pragmas should define exactly what we
want on any compiler that accepts them, let's try using them wherever
we think they exist, not only for __arm__. (This is likely to expose
that the conditional definitions in c.h are inadequate, but finding
that out would be a good thing.)
The immediate motivation for this is that the current definition of
ExecRowMark allows its curCtid field to be misaligned. It is not clear
whether there are any other uses of ItemPointerData with a similar hazard.
We could change the definition of ExecRowMark if this doesn't work, but
it would be far better to have a future-proof fix.
Piotr Stefaniak, some further hacking by me
Use "a" and "an" correctly, mostly in comments. Two error messages were
also fixed (they were just elogs, so no translation work required). Two
function comments in pg_proc.h were also fixed. Etsuro Fujita reported one
of these, but I found a lot more with grep.
Also fix a few other typos spotted while grepping for the a/an typos.
For example, "consists out of ..." -> "consists of ...". Plus a "though"/
"through" mixup reported by Euler Taveira.
Many of these typos were in old code, which would be nice to backpatch to
make future backpatching easier. But much of the code was new, and I didn't
feel like crafting separate patches for each branch. So no backpatching.
The point of the assertion is to ensure that the arrays allocated in stack
are large enough, but the check was one item short.
This won't matter in practice because MaxIndexTuplesPerPage is an
overestimate, so you can't have that many items on a page in reality.
But let's be tidy.
Spotted by Anastasia Lubennikova. Backpatch to all supported versions, like
the patch that added the assertion.
The newly added ON CONFLICT clause allows to specify an alternative to
raising a unique or exclusion constraint violation error when inserting.
ON CONFLICT refers to constraints that can either be specified using a
inference clause (by specifying the columns of a unique constraint) or
by naming a unique or exclusion constraint. DO NOTHING avoids the
constraint violation, without touching the pre-existing row. DO UPDATE
SET ... [WHERE ...] updates the pre-existing tuple, and has access to
both the tuple proposed for insertion and the existing tuple; the
optional WHERE clause can be used to prevent an update from being
executed. The UPDATE SET and WHERE clauses have access to the tuple
proposed for insertion using the "magic" EXCLUDED alias, and to the
pre-existing tuple using the table name or its alias.
This feature is often referred to as upsert.
This is implemented using a new infrastructure called "speculative
insertion". It is an optimistic variant of regular insertion that first
does a pre-check for existing tuples and then attempts an insert. If a
violating tuple was inserted concurrently, the speculatively inserted
tuple is deleted and a new attempt is made. If the pre-check finds a
matching tuple the alternative DO NOTHING or DO UPDATE action is taken.
If the insertion succeeds without detecting a conflict, the tuple is
deemed inserted.
To handle the possible ambiguity between the excluded alias and a table
named excluded, and for convenience with long relation names, INSERT
INTO now can alias its target table.
Bumps catversion as stored rules change.
Author: Peter Geoghegan, with significant contributions from Heikki
Linnakangas and Andres Freund. Testing infrastructure by Jeff Janes.
Reviewed-By: Heikki Linnakangas, Andres Freund, Robert Haas, Simon Riggs,
Dean Rasheed, Stephen Frost and many others.
pg_win32_is_junction() was a typo for pgwin32_is_junction(). open()
was used not only in a two-argument form, which breaks on Windows,
but also where BasicOpenFile() should have been used.
Per reports from Andrew Dunstan and David Rowley.
Otherwise, if there's another crash, some writes from after the first
crash might make it to disk while writes from before the crash fail
to make it to disk. This could lead to data corruption.
Back-patch to all supported versions.
Abhijit Menon-Sen, reviewed by Andres Freund and slightly revised
by me.
This does four basic things. First, it provides convenience routines
to coordinate the startup and shutdown of parallel workers. Second,
it synchronizes various pieces of state (e.g. GUCs, combo CID
mappings, transaction snapshot) from the parallel group leader to the
worker processes. Third, it prohibits various operations that would
result in unsafe changes to that state while parallelism is active.
Finally, it propagates events that would result in an ErrorResponse,
NoticeResponse, or NotifyResponse message being sent to the client
from the parallel workers back to the master, from which they can then
be sent on to the client.
Robert Haas, Amit Kapila, Noah Misch, Rushabh Lathia, Jeevan Chalke.
Suggestions and review from Andres Freund, Heikki Linnakangas, Noah
Misch, Simon Riggs, Euler Taveira, and Jim Nasby.
When implementing a replication solution ontop of logical decoding, two
related problems exist:
* How to safely keep track of replication progress
* How to change replication behavior, based on the origin of a row;
e.g. to avoid loops in bi-directional replication setups
The solution to these problems, as implemented here, consist out of
three parts:
1) 'replication origins', which identify nodes in a replication setup.
2) 'replication progress tracking', which remembers, for each
replication origin, how far replay has progressed in a efficient and
crash safe manner.
3) The ability to filter out changes performed on the behest of a
replication origin during logical decoding; this allows complex
replication topologies. E.g. by filtering all replayed changes out.
Most of this could also be implemented in "userspace", e.g. by inserting
additional rows contain origin information, but that ends up being much
less efficient and more complicated. We don't want to require various
replication solutions to reimplement logic for this independently. The
infrastructure is intended to be generic enough to be reusable.
This infrastructure also replaces the 'nodeid' infrastructure of commit
timestamps. It is intended to provide all the former capabilities,
except that there's only 2^16 different origins; but now they integrate
with logical decoding. Additionally more functionality is accessible via
SQL. Since the commit timestamp infrastructure has also been introduced
in 9.5 (commit 73c986add) changing the API is not a problem.
For now the number of origins for which the replication progress can be
tracked simultaneously is determined by the max_replication_slots
GUC. That GUC is not a perfect match to configure this, but there
doesn't seem to be sufficient reason to introduce a separate new one.
Bumps both catversion and wal page magic.
Author: Andres Freund, with contributions from Petr Jelinek and Craig Ringer
Reviewed-By: Heikki Linnakangas, Petr Jelinek, Robert Haas, Steve Singer
Discussion: 20150216002155.GI15326@awork2.anarazel.de,
20140923182422.GA15776@alap3.anarazel.de,
20131114172632.GE7522@alap2.anarazel.de
BackendIdGetTransactionIds() neglected the possibility that the PROC
pointer in a ProcState array entry is null. In current usage, this could
only crash if the other backend had exited since pgstat_read_current_status
saw it as active, which is a pretty narrow window. But it's reachable in
the field, per bug #12918 from Vladimir Borodin.
Back-patch to 9.4 where the faulty code was introduced.
Even though the main benefit of the Lehman and Yao algorithm for
btrees is that no locks need be held between page reads in an
index search, we were holding a buffer pin on each leaf page after
it was read until we were ready to read the next one. The reason
was so that we could treat this as a weak lock to create an
"interlock" with vacuum's deletion of heap line pointers, even
though our README file pointed out that this was not necessary for
a scan using an MVCC snapshot.
The main goal of this patch is to reduce the blocking of vacuum
processes by in-progress btree index scans (including a cursor
which is idle), but the code rearrangement also allows for one
less buffer content lock to be taken when a forward scan steps from
one page to the next, which results in a small but consistent
performance improvement in many workloads.
This patch leaves behavior unchanged for some cases, which can be
addressed separately so that each case can be evaluated on its own
merits. These unchanged cases are when a scan uses a non-MVCC
snapshot, an index-only scan, and a scan of a btree index for which
modifications are not WAL-logged. If later patches allow all of
these cases to drop the buffer pin after reading a leaf page, then
the btree vacuum process can be simplified; it will no longer need
the "super-exclusive" lock to delete tuples from a page.
Reviewed by Heikki Linnakangas and Kyotaro Horiguchi
dsm_control->nitems never decreases, so this is testing whether the
server has *ever* run out of DSM segments, not whether it is
*currently* out of DSM segments.
Reported off-list by Amit Kapila.
Right now, there's only one flag, DSM_CREATE_NULL_IF_MAXSEGMENTS,
which suppresses the error that would normally be thrown when the
maximum number of segments already exists, instead returning NULL.
It might be useful to add more flags in the future, such as one to
ignore allocation errors, but I haven't done that here.
When LockBufferForCleanup() has to wait for getting a cleanup lock on a
buffer it does so by setting a flag in the buffer header and then wait
for other backends to signal it using ProcWaitForSignal().
Unfortunately LockBufferForCleanup() missed that ProcWaitForSignal() can
return for other reasons than the signal it is hoping for. If such a
spurious signal arrives the wait flags on the buffer header will still
be set. That then triggers "ERROR: multiple backends attempting to wait
for pincount 1".
The fix is simple, unset the flag if still set when retrying. That
implies an additional spinlock acquisition/release, but that's unlikely
to matter given the cost of waiting for a cleanup lock. Alternatively
it'd have been possible to move responsibility for maintaining the
relevant flag to the waiter all together, but that might have had
negative consequences due to possible floods of signals. Besides being
more invasive.
This looks to be a very longstanding bug. The relevant code in
LockBufferForCleanup() hasn't changed materially since its introduction
and ProcWaitForSignal() was documented to return for unrelated reasons
since 8.2. The master only patch series removing ImmediateInterruptOK
made it much easier to hit though, as ProcSendSignal/ProcWaitForSignal
now uses a latch shared with other tasks.
Per discussion with Kevin Grittner, Tom Lane and me.
Backpatch to all supported branches.
Discussion: 11553.1423805224@sss.pgh.pa.us
The remaining caller (lwlocks) doesn't need that facility, and we plan
to remove ImmedidateInterruptOK entirely. That means that interrupts
can't be serviced race-free and portably anyway, so there's little
reason for keeping the feature.
Reviewed-By: Heikki Linnakangas
Deadlock checking was performed inside signal handlers up to
now. While it's a remarkable feat to have made this work reliably,
it's quite complex to understand why that is the case. Partially it
worked due to the assumption that semaphores are signal safe - which
is not actually documented to be the case for sysv semaphores.
The reason we had to rely on performing this work inside signal
handlers is that semaphores aren't guaranteed to be interruptable by
signals on all platforms. But now that latches provide a somewhat
similar API, which actually has the guarantee of being interruptible,
we can avoid doing so.
Signalling between ProcSleep, ProcWakeup, ProcWaitForSignal and
ProcSendSignal is now done using latches. This increases the
likelihood of spurious wakeups. As spurious wakeup already were
possible and aren't likely to be frequent enough to be an actual
problem, this seems acceptable.
This change would allow for further simplification of the deadlock
checking, now that it doesn't have to run in a signal handler. But
even if I were motivated to do so right now, it would still be better
to do that separately. Such a cleanup shouldn't have to be reviewed a
the same time as the more fundamental changes in this commit.
There is one possible usability regression due to this commit. Namely
it is more likely than before that log_lock_waits messages are output
more than once.
Reviewed-By: Heikki Linnakangas
Up to now large swathes of backend code ran inside signal handlers
while reading commands from the client, to allow for speedy reaction to
asynchronous events. Most prominently shared invalidation and NOTIFY
handling. That means that complex code like the starting/stopping of
transactions is run in signal handlers... The required code was
fragile and verbose, and is likely to contain bugs.
That approach also severely limited what could be done while
communicating with the client. As the read might be from within
openssl it wasn't safely possible to trigger an error, e.g. to cancel
a backend in idle-in-transaction state. We did that in some cases,
namely fatal errors, nonetheless.
Now that FE/BE communication in the backend employs non-blocking
sockets and latches to block, we can quite simply interrupt reads from
signal handlers by setting the latch. That allows us to signal an
interrupted read, which is supposed to be retried after returning from
within the ssl library.
As signal handlers now only need to set the latch to guarantee timely
interrupt processing, remove a fair amount of complicated & fragile
code from async.c and sinval.c.
We could now actually start to process some kinds of interrupts, like
sinval ones, more often that before, but that seems better done
separately.
This work will hopefully allow to handle cases like being blocked by
sending data, interrupting idle transactions and similar to be
implemented without too much effort. In addition to allowing getting
rid of ImmediateInterruptOK, that is.
Author: Andres Freund
Reviewed-By: Heikki Linnakangas
The logic to compact away removed tuples from page was duplicated with
small differences in PageRepairFragmentation, PageIndexMultiDelete, and
PageIndexDeleteNoCompact. Put it into a common function.
Reviewed by Peter Geoghegan.
If any error occurred while we were in the middle of reading a protocol
message from the client, we could lose sync, and incorrectly try to
interpret a part of another message as a new protocol message. That will
usually lead to an "invalid frontend message" error that terminates the
connection. However, this is a security issue because an attacker might
be able to deliberately cause an error, inject a Query message in what's
supposed to be just user data, and have the server execute it.
We were quite careful to not have CHECK_FOR_INTERRUPTS() calls or other
operations that could ereport(ERROR) in the middle of processing a message,
but a query cancel interrupt or statement timeout could nevertheless cause
it to happen. Also, the V2 fastpath and COPY handling were not so careful.
It's very difficult to recover in the V2 COPY protocol, so we will just
terminate the connection on error. In practice, that's what happened
previously anyway, as we lost protocol sync.
To fix, add a new variable in pqcomm.c, PqCommReadingMsg, that is set
whenever we're in the middle of reading a message. When it's set, we cannot
safely ERROR out and continue running, because we might've read only part
of a message. PqCommReadingMsg acts somewhat similarly to critical sections
in that if an error occurs while it's set, the error handler will force the
connection to be terminated, as if the error was FATAL. It's not
implemented by promoting ERROR to FATAL in elog.c, like ERROR is promoted
to PANIC in critical sections, because we want to be able to use
PG_TRY/CATCH to recover and regain protocol sync. pq_getmessage() takes
advantage of that to prevent an OOM error from terminating the connection.
To prevent unnecessary connection terminations, add a holdoff mechanism
similar to HOLD/RESUME_INTERRUPTS() that can be used hold off query cancel
interrupts, but still allow die interrupts. The rules on which interrupts
are processed when are now a bit more complicated, so refactor
ProcessInterrupts() and the calls to it in signal handlers so that the
signal handlers always call it if ImmediateInterruptOK is set, and
ProcessInterrupts() can decide to not do anything if the other conditions
are not met.
Reported by Emil Lenngren. Patch reviewed by Noah Misch and Andres Freund.
Backpatch to all supported versions.
Security: CVE-2015-0244
GetLockConflicts() has for a long time not properly terminated the
returned array. During normal processing the returned array is zero
initialized which, while not pretty, is sufficient to be recognized as
a invalid virtual transaction id. But the HotStandby case is more than
aesthetically broken: The allocated (and reused) array is neither
zeroed upon allocation, nor reinitialized, nor terminated.
Not having a terminating element means that the end of the array will
not be recognized and that recovery conflict handling will thus read
ahead into adjacent memory. Only terminating when hitting memory
content that looks like a invalid virtual transaction id. Luckily
this seems so far not have caused significant problems, besides making
recovery conflict more expensive.
Discussion: 20150127142713.GD29457@awork2.anarazel.de
Backpatch into all supported branches.
Benchmarks has shown that aligning the buffer descriptor array to
cache lines is important for scalability; especially on bigger,
multi-socket, machines.
Currently the array sometimes already happens to be aligned by
happenstance, depending how large previous shared memory allocations
were. That can lead to wildly varying performance results after minor
configuration changes.
In addition to aligning the start of descriptor array, also force the
size of individual descriptors to be of a common cache line size (64
bytes). That happens to already be the case on 64bit platforms, but
this way we can change the struct BufferDesc more easily.
As the alignment primarily matters in highly concurrent workloads
which probably all are 64bit these days, and the space wastage of
element alignment would be a bit more noticeable on 32bit systems, we
don't force the stride to be cacheline sized on 32bit platforms for
now. If somebody does actual performance testing, we can reevaluate
that decision by changing the definition of BUFFERDESC_PADDED_SIZE.
Discussion: 20140202151319.GD32123@awork2.anarazel.de
Per discussion with Bruce Momjan, Tom Lane, Robert Haas, and Peter
Geoghegan.
As noted by Tom Lane the improvements in 4b4b680c3d had the problem
that in some situations we searched, entered and modified entries in
the private refcount hash while holding a spinlock. I had tried to
keep the logic entirely local to PinBuffer_Locked(), but that's not
really possible given it's called with a spinlock held...
Besides being disadvantageous from a performance point of view, this
also has problems with error handling safety. If we failed inserting
an entry into the hashtable due to an out of memory error, we'd error
out with a held spinlock. Not good.
Change the way private refcounts are manipulated: Before a buffer can
be tracked an entry has to be reserved using
ReservePrivateRefCountEntry(); then, if a entry is not found using
GetPrivateRefCountEntry(), it can be entered with
NewPrivateRefCountEntry().
Also take advantage of the fact that PinBuffer_Locked() currently is
never called for buffers that already have been pinned by the current
backend and don't search the private refcount entries for preexisting
local pins. That results in a small, but measurable, performance
improvement.
Additionally make ReleaseBuffer() always call UnpinBuffer() for shared
buffers. That avoids duplicating work in an eventual UnpinBuffer()
call that already has been done in ReleaseBuffer() and also saves some
code.
Per discussion with Tom Lane.
Discussion: 15028.1418772313@sss.pgh.pa.us
To do so, move InitializeLatchSupport() into the new common process
initialization functions, and add a new global variable MyLatch.
MyLatch is usable as soon InitPostmasterChild() has been called
(i.e. very early during startup). Initially it points to a process
local latch that exists in all processes. InitProcess/InitAuxiliaryProcess
then replaces that local latch with PGPROC->procLatch. During shutdown
the reverse happens.
This is primarily advantageous for two reasons: For one it simplifies
dealing with the shared process latch, especially in signal handlers,
because instead of having to check for MyProc, MyLatch can be used
unconditionally. For another, a later patch that makes FEs/BE
communication use latches, now can rely on the existence of a latch,
even before having gone through InitProcess.
Discussion: 20140927191243.GD5423@alap3.anarazel.de
Since their introduction latches have required barriers in SetLatch
and ResetLatch - but when they were introduced there wasn't any
barrier abstraction. Instead latches were documented to rely on the
callsites to provide barrier semantics.
Now that the barrier support looks halfway complete, add the necessary
barriers to both latch implementations.
Also remove a now superflous lock acquisition from syncrep.c and a
superflous (and insufficient) barrier from freelist.c. There might be
other cases that can now be simplified, but those are the only ones
I've seen on a quick scan.
We might want to backpatch this at some later point, but right now the
barrier infrastructure in the backbranches isn't totally on par with
master.
Discussion: 20150112154026.GB2092@awork2.anarazel.de
Back in ed0b409, PGPROC was split and moved to static variables in
procarray.c, with procs in ProcArrayStruct replaced by an array of
integers representing process numbers (pgprocnos), with -1 indicating a
dead process which has yet to be removed. Access to procArray is
generally done under ProcArrayLock and therefore most code does not have
to concern itself with -1 entries.
However, MinimumActiveBackends intentionally does not take
ProcArrayLock, which means it has to be extra careful when accessing
procArray. Prior to ed0b409, this was handled by checking for a NULL
in the pointer array, but that check was no longer valid after the
split. Coverity pointed out that the check could never happen and so
it was removed in 5592eba. That didn't make anything worse, but it
didn't fix the issue either.
The correct fix is to check for pgprocno == -1 and skip over that entry
if it is encountered.
Back-patch to 9.2, since there can be attempts to access the arrays
prior to their start otherwise. Note that the changes prior to 9.4 will
look a bit different due to the change in 5592eba.
Note that MinimumActiveBackends only returns a bool for heuristic
purposes and any pre-array accesses are strictly read-only and so there
is no security implication and the lack of fields complaints indicates
it's very unlikely to run into issues due to this.
Pointed out by Noah.
StrategyGetBuffer() has proven to be a bottleneck in a number of
buffer acquisition heavy workloads. To some degree this has already
been alleviated by 5d7962c6, but it still can be quite a heavy
bottleneck. The problem is that in unfortunate usage patterns a
single StrategyGetBuffer() call will have to look at a large number of
buffers - in turn making it likely that the process will be put to
sleep while still holding the spinlock.
Replace most of the usage of the buffer_strategy_lock spinlock for the
clock sweep by a atomic nextVictimBuffer variable. That variable,
modulo NBuffers, is the current hand of the clock sweep. The buffer
clock-sweep then only needs to acquire the spinlock after a
wraparound. And even then only in the process that did the wrapping
around. That alleviates nearly all the contention on the relevant
spinlock, although significant contention on the cacheline can still
exist.
Reviewed-By: Robert Haas and Amit Kapila
Discussion: 20141010160020.GG6670@alap3.anarazel.de,
20141027133218.GA2639@awork2.anarazel.de
The old LWLock implementation had the problem that concurrent lock
acquisitions required exclusively acquiring a spinlock. Often that
could lead to acquirers waiting behind the spinlock, even if the
actual LWLock was free.
The new implementation doesn't acquire the spinlock when acquiring the
lock itself. Instead the new atomic operations are used to atomically
manipulate the state. Only the waitqueue, used solely in the slow
path, is still protected by the spinlock. Check lwlock.c's header for
an explanation about the used algorithm.
For some common workloads on larger machines this can yield
significant performance improvements. Particularly in read mostly
workloads.
Reviewed-By: Amit Kapila and Robert Haas
Author: Andres Freund
Discussion: 20130926225545.GB26663@awork2.anarazel.de
Besides being shorter and much easier to read it changes the logic in
LWLockRelease() to release all shared lockers when waking up any. This
can yield some significant performance improvements - and the fairness
isn't really much worse than before, as we always allowed new shared
lockers to jump the queue.
In LWLockRelease() (and in 9.4+ LWLockUpdateVar()) we release enqueued
waiters using PGSemaphoreUnlock(). As there are other sources of such
unlocks backends only wake up if MyProc->lwWaiting is set to false;
which is only done in the aforementioned functions.
Before this commit there were dangers because the store to lwWaitLink
could become visible before the store to lwWaitLink. This could both
happen due to compiler reordering (on most compilers) and on some
platforms due to the CPU reordering stores.
The possible consequence of this is that a backend stops waiting
before lwWaitLink is set to NULL. If that backend then tries to
acquire another lock and has to wait there the list could become
corrupted once the lwWaitLink store is finally performed.
Add a write memory barrier to prevent that issue.
Unfortunately the barrier support has been only added in 9.2. Given
that the issue has not knowingly been observed in praxis it seems
sufficient to prohibit compiler reordering using volatile for 9.0 and
9.1. Actual problems due to compiler reordering are more likely
anyway.
Discussion: 20140210134625.GA15246@awork2.anarazel.de
Previously, if you wanted anything besides C-string hash keys, you had to
specify a custom hashing function to hash_create(). Nearly all such
callers were specifying tag_hash or oid_hash; which is tedious, and rather
error-prone, since a caller could easily miss the opportunity to optimize
by using hash_uint32 when appropriate. Replace this with a design whereby
callers using simple binary-data keys just specify HASH_BLOBS and don't
need to mess with specific support functions. hash_create() itself will
take care of optimizing when the key size is four bytes.
This nets out saving a few hundred bytes of code space, and offers
a measurable performance improvement in tidbitmap.c (which was not
exploiting the opportunity to use hash_uint32 for its 4-byte keys).
There might be some wins elsewhere too, I didn't analyze closely.
In future we could look into offering a similar optimized hashing function
for 8-byte keys. Under this design that could be done in a centralized
and machine-independent fashion, whereas getting it right for keys of
platform-dependent sizes would've been notationally painful before.
For the moment, the old way still works fine, so as not to break source
code compatibility for loadable modules. Eventually we might want to
remove tag_hash and friends from the exported API altogether, since there's
no real need for them to be explicitly referenced from outside dynahash.c.
Teodor Sigaev and Tom Lane
Transactions can now set their commit timestamp directly as they commit,
or an external transaction commit timestamp can be fed from an outside
system using the new function TransactionTreeSetCommitTsData(). This
data is crash-safe, and truncated at Xid freeze point, same as pg_clog.
This module is disabled by default because it causes a performance hit,
but can be enabled in postgresql.conf requiring only a server restart.
A new test in src/test/modules is included.
Catalog version bumped due to the new subdirectory within PGDATA and a
couple of new SQL functions.
Authors: Álvaro Herrera and Petr Jelínek
Reviewed to varying degrees by Michael Paquier, Andres Freund, Robert
Haas, Amit Kapila, Fujii Masao, Jaime Casanova, Simon Riggs, Steven
Singer, Peter Eisentraut
Each WAL record now carries information about the modified relation and
block(s) in a standardized format. That makes it easier to write tools that
need that information, like pg_rewind, prefetching the blocks to speed up
recovery, etc.
There's a whole new API for building WAL records, replacing the XLogRecData
chains used previously. The new API consists of XLogRegister* functions,
which are called for each buffer and chunk of data that is added to the
record. The new API also gives more control over when a full-page image is
written, by passing flags to the XLogRegisterBuffer function.
This also simplifies the XLogReadBufferForRedo() calls. The function can dig
the relation and block number from the WAL record, so they no longer need to
be passed as arguments.
For the convenience of redo routines, XLogReader now disects each WAL record
after reading it, copying the main data part and the per-block data into
MAXALIGNed buffers. The data chunks are not aligned within the WAL record,
but the redo routines can assume that the pointers returned by XLogRecGet*
functions are. Redo routines are now passed the XLogReaderState, which
contains the record in the already-disected format, instead of the plain
XLogRecord.
The new record format also makes the fixed size XLogRecord header smaller,
by removing the xl_len field. The length of the "main data" portion is now
stored at the end of the WAL record, and there's a separate header after
XLogRecord for it. The alignment padding at the end of XLogRecord is also
removed. This compansates for the fact that the new format would otherwise
be more bulky than the old format.
Reviewed by Andres Freund, Amit Kapila, Michael Paquier, Alvaro Herrera,
Fujii Masao.
Unlogged relations are only reset when performing a unclean
restart. That means they have to be synced to disk during clean
shutdowns. During normal processing that's achieved by registering a
buffer's file to be fsynced at the next checkpoint when flushed. But
ResetUnloggedRelations() doesn't go through the buffer manager, so
nothing will force reset relations to disk before the next shutdown
checkpoint.
So just make ResetUnloggedRelations() fsync the newly created main
forks to disk.
Discussion: 20140912112246.GA4984@alap3.anarazel.de
Backpatch to 9.1 where unlogged tables were introduced.
Abhijit Menon-Sen and Andres Freund
There was a window in RestoreBackupBlock where a page would be zeroed out,
but not yet locked. If a backend pinned and locked the page in that window,
it saw the zeroed page instead of the old page or new page contents, which
could lead to missing rows in a result set, or errors.
To fix, replace RBM_ZERO with RBM_ZERO_AND_LOCK, which atomically pins,
zeroes, and locks the page, if it's not in the buffer cache already.
In stable branches, the old RBM_ZERO constant is renamed to RBM_DO_NOT_USE,
to avoid breaking any 3rd party extensions that might use RBM_ZERO. More
importantly, this avoids renumbering the other enum values, which would
cause even bigger confusion in extensions that use ReadBufferExtended, but
haven't been recompiled.
Backpatch to all supported versions; this has been racy since hot standby
was introduced.
BRIN is a new index access method intended to accelerate scans of very
large tables, without the maintenance overhead of btrees or other
traditional indexes. They work by maintaining "summary" data about
block ranges. Bitmap index scans work by reading each summary tuple and
comparing them with the query quals; all pages in the range are returned
in a lossy TID bitmap if the quals are consistent with the values in the
summary tuple, otherwise not. Normal index scans are not supported
because these indexes do not store TIDs.
As new tuples are added into the index, the summary information is
updated (if the block range in which the tuple is added is already
summarized) or not; in the latter case, a subsequent pass of VACUUM or
the brin_summarize_new_values() function will create the summary
information.
For data types with natural 1-D sort orders, the summary info consists
of the maximum and the minimum values of each indexed column within each
page range. This type of operator class we call "Minmax", and we
supply a bunch of them for most data types with B-tree opclasses.
Since the BRIN code is generalized, other approaches are possible for
things such as arrays, geometric types, ranges, etc; even for things
such as enum types we could do something different than minmax with
better results. In this commit I only include minmax.
Catalog version bumped due to new builtin catalog entries.
There's more that could be done here, but this is a good step forwards.
Loosely based on ideas from Simon Riggs; code mostly by Álvaro Herrera,
with contribution by Heikki Linnakangas.
Patch reviewed by: Amit Kapila, Heikki Linnakangas, Robert Haas.
Testing help from Jeff Janes, Erik Rijkers, Emanuel Calvo.
PS:
The research leading to these results has received funding from the
European Union's Seventh Framework Programme (FP7/2007-2013) under
grant agreement n° 318633.
xlog.c is huge, this makes it a little bit smaller, which is nice. Functions
related to putting together the WAL record are in xloginsert.c, and the
lower level stuff for managing WAL buffers and such are in xlog.c.
Also move the definition of XLogRecord to a separate header file. This
causes churn in the #includes of all the files that write WAL records, and
redo routines, but it avoids pulling in xlog.h into most places.
Reviewed by Michael Paquier, Alvaro Herrera, Andres Freund and Amit Kapila.
This reassociates a dynamic shared memory handle previous passed to
dsm_pin_mapping with the current resource owner, so that it will be
cleaned up at the end of the current query.
Patch by me. Review of the function name by Andres Freund, Amit
Kapila, Jim Nasby, Petr Jelinek, and Álvaro Herrera.
Nobody seemed concerned about this naming when it originally went in,
but there's a pending patch that implements the opposite of
dsm_keep_mapping, and the term "unkeep" was judged unpalatable.
"unpin" has existing precedent in the PostgreSQL code base, and the
English language, so use this terminology instead.
Per discussion, back-patch to 9.4.
CREATE DATABASE and ALTER DATABASE .. SET TABLESPACE copy the source
database directory on the filesystem level. To ensure the on disk
state is consistent they block out users of the affected database and
force a checkpoint to flush out all data to disk. Unfortunately, up to
now, that checkpoint didn't flush out dirty buffers from unlogged
relations.
That bug means there could be leftover dirty buffers in either the
template database, or the database in its old location. Leading to
problems when accessing relations in an inconsistent state; and to
possible problems during shutdown in the SET TABLESPACE case because
buffers belonging files that don't exist anymore are flushed.
This was reported in bug #10675 by Maxim Boguk.
Fix by Pavan Deolasee, modified somewhat by me. Reviewed by MauMau and
Fujii Masao.
Backpatch to 9.1 where unlogged tables were introduced.
LWLockRelease should release all backends waiting with LWLockWaitForVar,
even when another backend has already been woken up to acquire the lock,
i.e. when releaseOK is false. LWLockWaitForVar can return as soon as the
protected value changes, even if the other backend will acquire the lock.
Fix that by resetting releaseOK to true in LWLockWaitForVar, whenever
adding itself to the wait queue.
This should fix the bug reported by MauMau, where the system occasionally
hangs when there is a lot of concurrent WAL activity and a checkpoint.
Backpatch to 9.4, where this code was added.
shm_mq_sendv sends a message to the queue assembled from multiple
locations. This is expected to be used by forthcoming patches to
allow frontend/backend protocol messages to be sent via shm_mq, but
might be useful for other purposes as well.
shm_mq_set_handle associates a BackgroundWorkerHandle with an
already-existing shm_mq_handle. This solves a timing problem when
creating a shm_mq to communicate with a newly-launched background
worker: if you attach to the queue first, and the background worker
fails to start, you might block forever trying to do I/O on the queue;
but if you start the background worker first, but then die before
attaching to the queue, the background worrker might block forever
trying to do I/O on the queue. This lets you attach before starting
the worker (so that the worker is protected) and then associate the
BackgroundWorkerHandle later (so that you are also protected).
Patch by me, reviewed by Stephen Frost.
Testing by Amit Kapila, Andres Freund, and myself, with and without
other patches that also aim to improve scalability, seems to indicate
that this change is a significant win over the current value and over
smaller values such as 64. It's not clear how high we can push this
value before it starts to have negative side-effects elsewhere, but
going this far looks OK.
Several upcoming performance/scalability improvements require atomic
operations. This new API avoids the need to splatter compiler and
architecture dependent code over all the locations employing atomic
ops.
For several of the potential usages it'd be problematic to maintain
both, a atomics using implementation and one using spinlocks or
similar. In all likelihood one of the implementations would not get
tested regularly under concurrency. To avoid that scenario the new API
provides a automatic fallback of atomic operations to spinlocks. All
properties of atomic operations are maintained. This fallback -
obviously - isn't as fast as just using atomic ops, but it's not bad
either. For one of the future users the atomics ontop spinlocks
implementation was actually slightly faster than the old purely
spinlock using implementation. That's important because it reduces the
fear of regressing older platforms when improving the scalability for
new ones.
The API, loosely modeled after the C11 atomics support, currently
provides 'atomic flags' and 32 bit unsigned integers. If the platform
efficiently supports atomic 64 bit unsigned integers those are also
provided.
To implement atomics support for a platform/architecture/compiler for
a type of atomics 32bit compare and exchange needs to be
implemented. If available and more efficient native support for flags,
32 bit atomic addition, and corresponding 64 bit operations may also
be provided. Additional useful atomic operations are implemented
generically ontop of these.
The implementation for various versions of gcc, msvc and sun studio have
been tested. Additional existing stub implementations for
* Intel icc
* HUPX acc
* IBM xlc
are included but have never been tested. These will likely require
fixes based on buildfarm and user feedback.
As atomic operations also require barriers for some operations the
existing barrier support has been moved into the atomics code.
Author: Andres Freund with contributions from Oskari Saarenmaa
Reviewed-By: Amit Kapila, Robert Haas, Heikki Linnakangas and Álvaro Herrera
Discussion: CA+TgmoYBW+ux5-8Ja=Mcyuy8=VXAnVRHp3Kess6Pn3DMXAPAEA@mail.gmail.com,
20131015123303.GH5300@awork2.anarazel.de,
20131028205522.GI20248@awork2.anarazel.de
Previously, we used an lwlock that was held from the time we began
seeking a candidate buffer until the time when we found and pinned
one, which is disastrous for concurrency. Instead, use a spinlock
which is held just long enough to pop the freelist or advance the
clock sweep hand, and then released. If we need to advance the clock
sweep further, we reacquire the spinlock once per buffer.
This represents a significant increase in atomic operations around
buffer eviction, but it still wins on many workloads. On others, it
may result in no gain, or even cause a regression, unless the number
of buffer mapping locks is also increased. However, that seems like
material for a separate commit. We may also need to consider other
methods of mitigating contention on this spinlock, such as splitting
it into multiple locks or jumping the clock sweep hand more than one
buffer at a time, but those, too, seem like separate improvements.
Patch by me, inspired by a much larger patch from Amit Kapila.
Reviewed by Andres Freund.
Now that spinlocks (hopefully!) act as compiler barriers, as of commit
0709b7ee72, this should be safe. This
serves as a demonstration of the new coding style, and may be optimized
better on some machines as well.
Previously, they functioned as barriers against CPU reordering but not
compiler reordering, an odd API that required extensive use of volatile
everywhere that spinlocks are used. That's error-prone and has negative
implications for performance, so change it.
In theory, this makes it safe to remove many of the uses of volatile
that we currently have in our code base, but we may find that there are
some bugs in this effort when we do. In the long run, though, this
should make for much more maintainable code.
Patch by me. Review by Andres Freund.
68a2e52bba has introduced LWLockAcquireCommon() containing the
previous contents of LWLockAcquire() plus added functionality. The
latter then calls it, just like LWLockAcquireWithVar(). Because the
majority of callers don't need the added functionality, declare the
common code as inline. The compiler then can optimize away the unused
code. Doing so is also useful when looking at profiles, to
differentiate the users.
Backpatch to 9.4, the first branch to contain LWLockAcquireCommon().
Since the dawn of time (aka Postgres95) multiple pins of the same
buffer by one backend have been optimized not to modify the shared
refcount more than once. This optimization has always used a NBuffer
sized array in each backend keeping track of a backend's pins.
That array (PrivateRefCount) was one of the biggest per-backend memory
allocations, depending on the shared_buffers setting. Besides the
waste of memory it also has proven to be a performance bottleneck when
assertions are enabled as we make sure that there's no remaining pins
left at the end of transactions. Also, on servers with lots of memory
and a correspondingly high shared_buffers setting the amount of random
memory accesses can also lead to poor cpu cache efficiency.
Because of these reasons a backend's buffers pins are now kept track
of in a small statically sized array that overflows into a hash table
when necessary. Benchmarks have shown neutral to positive performance
results with considerably lower memory usage.
Patch by me, review by Robert Haas.
Discussion: 20140321182231.GA17111@alap3.anarazel.de
shm_mq_send_bytes didn't invariably initialize *bytes_written before
returning, which would cause shm_mq_send to read from uninitialized
memory and add the value it found there to mqh->mqh_partial_bytes.
This could cause the next attempt to send a message via the queue to
fail an assertion (if the queue was detached) or copy data from a
garbage pointer value into the queue (if non-blocking mode was in use).
get_raw_page tried to validate the supplied block number against
RelationGetNumberOfBlocks(), which of course is only right when
accessing the main fork. In most cases, the main fork is longer
than the others, so that the check was too weak (allowing a
lower-level error to be reported, but no real harm to be done).
However, very small tables could have an FSM larger than their heap,
in which case the mistake prevented access to some FSM pages.
Per report from Torsten Foertsch.
In passing, make the bad-block-number error into an ereport not elog
(since it's certainly not an internal error); and fix sloppily
maintained comment for RelationGetNumberOfBlocksInFork.
This has been wrong since we invented relation forks, so back-patch
to all supported branches.
The assertion failed if WAL_DEBUG or LWLOCK_STATS was enabled; fix that by
using separate memory contexts for the allocations made within those code
blocks.
This patch introduces a mechanism for marking any memory context as allowed
in a critical section. Previously ErrorContext was exempt as a special case.
Instead of a blanket exception of the checkpointer process, only exempt the
memory context used for the pending ops hash table.
The existance of the assert_enabled variable (backing the
debug_assertions GUC) reduced the amount of knowledge some static code
checkers (like coverity and various compilers) could infer from the
existance of the assertion. That could have been solved by optionally
removing the assertion_enabled variable from the Assert() et al macros
at compile time when some special macro is defined, but the resulting
complication doesn't seem to be worth the gain from having
debug_assertions. Recompiling is fast enough.
The debug_assertions GUC is still available, but readonly, as it's
useful when diagnosing problems. The commandline/client startup option
-A, which previously also allowed to enable/disable assertions, has
been removed as it doesn't serve a purpose anymore.
While at it, reduce code duplication in bufmgr.c and localbuf.c
assertions checking for spurious buffer pins. That code had to be
reindented anyway to cope with the assert_enabled removal.
It's critical that the backend's idea of LOBLKSIZE match the way data has
actually been divided up in pg_largeobject. While we don't provide any
direct way to adjust that value, doing so is a one-line source code change
and various people have expressed interest recently in changing it. So,
just as with TOAST_MAX_CHUNK_SIZE, it seems prudent to record the value in
pg_control and cross-check that the backend's compiled-in setting matches
the on-disk data.
Also tweak the code in inv_api.c so that fetches from pg_largeobject
explicitly verify that the length of the data field is not more than
LOBLKSIZE. Formerly we just had Asserts() for that, which is no protection
at all in production builds. In some of the call sites an overlength data
value would translate directly to a security-relevant stack clobber, so it
seems worth one extra runtime comparison to be sure.
In the back branches, we can't change the contents of pg_control; but we
can still make the extra checks in inv_api.c, which will offer some amount
of protection against running with the wrong value of LOBLKSIZE.
The previous coding would potentially cause attaching to segment A to
fail if segment B was at the same time in the process of going away.
Andres Freund, with a comment tweak by me
Commit fad153ec45 modified sinval.c to reduce
the number of calls into sinvaladt.c (which require taking a shared lock)
by keeping a local buffer of collected-but-not-yet-processed messages.
However, if processing of the last message in a batch resulted in a
recursive call to ReceiveSharedInvalidMessages, we could overwrite that
message with a new one while the outer invalidation function was still
working on it. This would be likely to lead to invalidation of the wrong
cache entry, allowing subsequent processing to use stale cache data.
The fix is just to make a local copy of each message while we're processing
it.
Spotted by Andres Freund. Back-patch to 8.4 where the bug was introduced.
This was intended to work always, but the previous code only allowed
it if at least one message was successfully read by the receiver
before the sender detached the queue.
Report by Petr Jelinek. Patch by me.
Commit a730183926 created rather a mess by
putting dependencies on backend-only include files into include/common.
We really shouldn't do that. To clean it up:
* Move TABLESPACE_VERSION_DIRECTORY back to its longtime home in
catalog/catalog.h. We won't consider this symbol part of the FE/BE API.
* Push enum ForkNumber from relfilenode.h into relpath.h. We'll consider
relpath.h as the source of truth for fork numbers, since relpath.c was
already partially serving that function, and anyway relfilenode.h was
kind of a random place for that enum.
* So, relfilenode.h now includes relpath.h rather than vice-versa. This
direction of dependency is fine. (That allows most, but not quite all,
of the existing explicit #includes of relpath.h to go away again.)
* Push forkname_to_number from catalog.c to relpath.c, just to centralize
fork number stuff a bit better.
* Push GetDatabasePath from catalog.c to relpath.c; it was rather odd
that the previous commit didn't keep this together with relpath().
* To avoid needing relfilenode.h in common/, redefine the underlying
function (now called GetRelationPath) as taking separate OID arguments,
and make the APIs using RelFileNode or RelFileNodeBackend into macro
wrappers. (The macros have a potential multiple-eval risk, but none of
the existing call sites have an issue with that; one of them had such a
risk already anyway.)
* Fix failure to follow the directions when "init" fork type was added;
specifically, the errhint in forkname_to_number wasn't updated, and neither
was the SGML documentation for pg_relation_size().
* Fix tablespace-path-too-long check in CreateTableSpace() to account for
fork-name component of maximum-length pathnames. This requires putting
FORKNAMECHARS into a header file, but it was rather useless (and
actually unreferenced) where it was.
The last couple of items are potentially back-patchable bug fixes,
if anyone is sufficiently excited about them; but personally I'm not.
Per a gripe from Christoph Berg about how include/common wasn't
self-contained.
Original coding failed to enlarge the array as required if
the requested tranche_id was equal to LWLockTranchesAllocated.
In passing, fix poor style of not casting the result of (re)palloc.
Apparently, Windows can sometimes return an error code even when the
operation actually worked just fine. Rearrange the order of checks
according to what appear to be the best practices in this area.
Amit Kapila
Instead of storing the ID of the dynamic shared memory control
segment in a file within the data directory, store it in the main
control segment. This avoids a number of nasty corner cases,
most seriously that doing an online backup and then using it on
the same machine (e.g. to fire up a standby) would result in the
standby clobbering all of the master's dynamic shared memory
segments.
Per complaints from Heikki Linnakangas, Fujii Masao, and Tom
Lane.
The one existing assertion of this type has tripped a few times in the
buildfarm lately, but it's not clear whether the problem is really
originating there or whether it's leftovers from a trip through one
of the other two paths that lack a matching assertion. So add one.
Since the same bug(s) most likely exist(s) in the back-branches also,
back-patch to 9.2, where the fast-path lock mechanism was added.
Otherwise, the compiler might decide to move modifications to data
within this structure outside the enclosing SpinLockAcquire /
SpinLockRelease pair, leading to shared memory corruption.
This may or may not explain a recent lmgr-related buildfarm failure
on prairiedog, but it needs to be fixed either way.
Previously, such buffers weren't counted, with the possible result that
EXPLAIN (BUFFERS) and pg_stat_statements would understate the true
number of blocks dirtied by an SQL statement.
Back-patch to 9.2, where this counter was introduced.
Amit Kapila
Also fix the name of the dtrace probe for LWLockAcquireOrWait(). The
function was renamed from LWLockWaitUntilFree to LWLockAqcuireOrWait, but
the dtrace probe was neglected.
Pointed out by Andres Freund and the buildfarm.
Clear errno before calling readdir() and handle old MinGW errno bug
while adding full test coverage for readdir/closedir failures.
Backpatch through 8.4.
The special feature the XLogInsert slots had over regular LWLocks is the
insertingAt value that was updated atomically with releasing backends
waiting on it. Add new functions to the LWLock API to do that, and replace
the slots with LWLocks. This reduces the amount of duplicated code.
(There's still some duplication, but at least it's all in lwlock.c now.)
Reviewed by Andres Freund.
With this in place, a session blocking behind another one because of
tuple locks will get a context line mentioning the relation name, tuple
TID, and operation being done on tuple. For example:
LOG: process 11367 still waiting for ShareLock on transaction 717 after 1000.108 ms
DETAIL: Process holding the lock: 11366. Wait queue: 11367.
CONTEXT: while updating tuple (0,2) in relation "foo"
STATEMENT: UPDATE foo SET value = 3;
Most usefully, the new line is displayed by log entries due to
log_lock_waits, although of course it will be printed by any other log
message as well.
Author: Christian Kruse, some tweaks by Álvaro Herrera
Reviewed-by: Amit Kapila, Andres Freund, Tom Lane, Robert Haas
Revise the original decision to expose a uint64-based interface and
use Size everywhere possible. Avoid assuming that MAXIMUM_ALIGNOF is
8, or making any assumption about the relationship between that value
and sizeof(Size). If MAXIMUM_ALIGNOF is bigger, we'll now insert
padding after the length word; if it's smaller, we are now prepared
to read and write the length word in chunks.
Per discussion with Tom Lane.
The new function dsm_detach_all() can be used either by postmaster
children that don't wish to take any risk of accidentally corrupting
shared memory; or by forked children of regular backends with
the same need. This patch also updates the postmaster children that
already do PGSharedMemoryDetach() to do dsm_detach_all() as well.
Per discussion with Tom Lane.
If a postmaster child invokes fork() and then calls on_exit_reset, that
should be sufficient to let it exit() without breaking anything, but
dynamic shared memory broke that by not updating on_exit_reset() to
discard callbacks registered with dynamic shared memory segments.
Per investigation of a complaint from Tom Lane.
A fake relcache entry can "own" a SmgrRelation object, like a regular
relcache entry. But when it was free'd, the owner field in SmgrRelation
was not cleared, so it was left pointing to free'd memory.
Amazingly this apparently hasn't caused crashes in practice, or we would've
heard about it earlier. Andres found this with Valgrind.
Report and fix by Andres Freund, with minor modifications by me. Backpatch
to all supported versions.
This feature, building on previous commits, allows the write-ahead log
stream to be decoded into a series of logical changes; that is,
inserts, updates, and deletes and the transactions which contain them.
It is capable of handling decoding even across changes to the schema
of the effected tables. The output format is controlled by a
so-called "output plugin"; an example is included. To make use of
this in a real replication system, the output plugin will need to be
modified to produce output in the format appropriate to that system,
and to perform filtering.
Currently, information can be extracted from the logical decoding
system only via SQL; future commits will add the ability to stream
changes via walsender.
Andres Freund, with review and other contributions from many other
people, including Álvaro Herrera, Abhijit Menon-Sen, Peter Gheogegan,
Kevin Grittner, Robert Haas, Heikki Linnakangas, Fujii Masao, Abhijit
Menon-Sen, Michael Paquier, Simon Riggs, Craig Ringer, and Steve
Singer.
Additional non-security issues/improvements spotted by Coverity.
In backend/libpq, no sense trying to protect against port->hba being
NULL after we've already dereferenced it in the switch() statement.
Prevent against possible overflow due to 32bit arithmitic in
basebackup throttling (not yet released, so no security concern).
Remove nonsensical check of array pointer against NULL in procarray.c,
looks to be a holdover from 9.1 and earlier when there were pointers
being used but now it's just an array.
Remove pointer check-against-NULL in tsearch/spell.c as we had already
dereferenced it above (in the strcmp()).
Remove dead code from adt/orderedsetaggs.c, isnull is checked
immediately after each tuplesort_getdatum() call and if true we return,
so no point checking it again down at the bottom.
Remove recently added minor error-condition memory leak in pg_regress.
A number of issues were identified by the Coverity scanner and are
addressed in this patch. None of these appear to be security issues
and many are mostly cosmetic changes.
Short comments for each of the changes follows.
Correct the semi-colon placement in be-secure.c regarding SSL retries.
Remove a useless comparison-to-NULL in proc.c (value is dereferenced
prior to this check and therefore can't be NULL).
Add checking of chmod() return values to initdb.
Fix a couple minor memory leaks in initdb.
Fix memory leak in pg_ctl- involves free'ing the config file contents.
Use an int to capture fgetc() return instead of an enum in pg_dump.
Fix minor memory leaks in pg_dump.
(note minor change to convertOperatorReference()'s API)
Check fclose()/remove() return codes in psql.
Check fstat(), find_my_exec() return codes in psql.
Various ECPG memory leak fixes.
Check find_my_exec() return in ECPG.
Explicitly ignore pqFlush return in libpq error-path.
Change PQfnumber() to avoid doing an strdup() when no changes required.
Remove a few useless check-against-NULL's (value deref'd beforehand).
Check rmtree(), malloc() results in pg_regress.
Also check get_alternative_expectfile() return in pg_regress.
Replication slots are a crash-safe data structure which can be created
on either a master or a standby to prevent premature removal of
write-ahead log segments needed by a standby, as well as (with
hot_standby_feedback=on) pruning of tuples whose removal would cause
replication conflicts. Slots have some advantages over existing
techniques, as explained in the documentation.
In a few places, we refer to the type of replication slots introduced
by this patch as "physical" slots, because forthcoming patches for
logical decoding will also have slots, but with somewhat different
properties.
Andres Freund and Robert Haas
Evidence from buildfarm member crake suggests that the new test_shm_mq
module is routinely crashing the server due to the arrival of a SIGUSR1
after the shared memory segment has been unmapped. Although processes
using the new dynamic background worker facilities are more likely to
receive a SIGUSR1 around this time, the problem is also possible on older
branches, so I'm back-patching the parts of this change that apply to
older branches as far as they apply.
It's already generally the case that code checks whether these pointers
are NULL before deferencing them, so the important thing is mostly to
make sure that they do get set to NULL before they become invalid. But
in master, there's one case in procsignal_sigusr1_handler that lacks a
NULL guard, so add that.
Patch by me; review by Tom Lane.
This makes it possible to store lwlocks as part of some other data
structure in the main shared memory segment, or in a dynamic shared
memory segment. There is still a main LWLock array and this patch does
not move anything out of it, but it provides necessary infrastructure
for doing that in the future.
This change is likely to increase the size of LWLockPadded on some
platforms, especially 32-bit platforms where it was previously only
16 bytes.
Patch by me. Review by Andres Freund and KaiGai Kohei.
Since C99, it's been standard for printf and friends to accept a "z" size
modifier, meaning "whatever size size_t has". Up to now we've generally
dealt with printing size_t values by explicitly casting them to unsigned
long and using the "l" modifier; but this is really the wrong thing on
platforms where pointers are wider than longs (such as Win64). So let's
start using "z" instead. To ensure we can do that on all platforms, teach
src/port/snprintf.c to understand "z", and add a configure test to force
use of that implementation when the platform's version doesn't handle "z".
Having done that, modify a bunch of places that were using the
unsigned-long hack to use "z" instead. This patch doesn't pretend to have
gotten everyplace that could benefit, but it catches many of them. I made
an effort in particular to ensure that all uses of the same error message
text were updated together, so as not to increase the number of
translatable strings.
It's possible that this change will result in format-string warnings from
pre-C99 compilers. We might have to reconsider if there are any popular
compilers that will warn about this; but let's start by seeing what the
buildfarm thinks.
Andres Freund, with a little additional work by me
Previously, we did this just once per checkpoint, but that could make
Hot Standby take a long time to initialize. To avoid busying an
otherwise-idle system, we don't do this if no WAL has been written
since we did it last.
Andres Freund
In ordinary operation, VACUUM must be careful to take a cleanup lock on
each leaf page of a btree index; this ensures that no indexscans could
still be "in flight" to heap tuples due to be deleted. (Because of
possible index-tuple motion due to concurrent page splits, it's not enough
to lock only the pages we're deleting index tuples from.) In Hot Standby,
the WAL replay process must likewise lock every leaf page. There were
several bugs in the code for that:
* The replay scan might come across unused, all-zero pages in the index.
While btree_xlog_vacuum itself did the right thing (ie, nothing) with
such pages, xlogutils.c supposed that such pages must be corrupt and
would throw an error. This accounts for various reports of replication
failures with "PANIC: WAL contains references to invalid pages". To
fix, add a ReadBufferMode value that instructs XLogReadBufferExtended
not to complain when we're doing this.
* btree_xlog_vacuum performed the extra locking if standbyState ==
STANDBY_SNAPSHOT_READY, but that's not the correct test: we won't open up
for hot standby queries until the database has reached consistency, and
we don't want to do the extra locking till then either, for fear of reading
corrupted pages (which bufmgr.c would complain about). Fix by exporting a
new function from xlog.c that will report whether we're actually in hot
standby replay mode.
* To ensure full coverage of the index in the replay scan, btvacuumscan
would emit a dummy WAL record for the last page of the index, if no
vacuuming work had been done on that page. However, if the last page
of the index is all-zero, that would result in corruption of said page,
since the functions called on it weren't prepared to handle that case.
There's no need to lock any such pages, so change the logic to target
the last normal leaf page instead.
The first two of these bugs were diagnosed by Andres Freund, the other one
by me. Fixes based on ideas from Heikki Linnakangas and myself.
This has been wrong since Hot Standby was introduced, so back-patch to 9.0.
This code provides infrastructure for user backends to communicate
relatively easily with background workers. The message queue is
structured as a ring buffer and allows messages of arbitary length
to be sent and received.
Patch by me. Review by KaiGai Kohei and Andres Freund.
This interface is intended to make it simple to divide a dynamic shared
memory segment into different regions with distinct purposes. It
therefore serves much the same purpose that ShmemIndex accomplishes for
the main shared memory segment, but it is intended to be more
lightweight.
Patch by me. Review by Andres Freund.
Minor improvement to commit daa7527afc:
s_lock.h no longer has any need to mention PGSemaphoreData, so we can
rip out the #include that supplies that. In a non-HAVE_SPINLOCKS
build, this doesn't really buy much since we still need the #include
in spin.h --- but everywhere else, this reduces #include footprint by
some trifle, and helps keep the different locking facilities separate.
Instead of allocating a semaphore from the operating system for every
spinlock, allocate a fixed number of semaphores (by default, 1024)
from the operating system and multiplex all the spinlocks that get
created onto them. This could self-deadlock if a process attempted
to acquire more than one spinlock at a time, but since processes
aren't supposed to execute anything other than short stretches of
straight-line code while holding a spinlock, that shouldn't happen.
One motivation for this change is that, with the introduction of
dynamic shared memory, it may be desirable to create spinlocks that
last for less than the lifetime of the server. Without this change,
attempting to use such facilities under --disable-spinlocks would
quickly exhaust any supply of available semaphores. Quite apart
from that, it's desirable to contain the quantity of semaphores
needed to run the server simply on convenience grounds, since using
too many may make it harder to get PostgreSQL running on a new
platform, which is mostly the point of --disable-spinlocks in the
first place.
Patch by me; review by Tom Lane.
Instead of changing the tuple xmin to FrozenTransactionId, the combination
of HEAP_XMIN_COMMITTED and HEAP_XMIN_INVALID, which were previously never
set together, is now defined as HEAP_XMIN_FROZEN. A variety of previous
proposals to freeze tuples opportunistically before vacuum_freeze_min_age
is reached have foundered on the objection that replacing xmin by
FrozenTransactionId might hinder debugging efforts when things in this
area go awry; this patch is intended to solve that problem by keeping
the XID around (but largely ignoring the value to which it is set).
Third-party code that checks for HEAP_XMIN_INVALID on tuples where
HEAP_XMIN_COMMITTED might be set will be broken by this change. To fix,
use the new accessor macros in htup_details.h rather than consulting the
bits directly. HeapTupleHeaderGetXmin has been modified to return
FrozenTransactionId when the infomask bits indicate that the tuple is
frozen; use HeapTupleHeaderGetRawXmin when you already know that the
tuple isn't marked commited or frozen, or want the raw value anyway.
We currently do this in routines that display the xmin for user consumption,
in tqual.c where it's known to be safe and important for the avoidance of
extra cycles, and in the function-caching code for various procedural
languages, which shouldn't invalidate the cache just because the tuple
gets frozen.
Robert Haas and Andres Freund
Just as backends must clean up their shared memory state (releasing
lwlocks, buffer pins, etc.) before exiting, they must also perform
any similar cleanups related to dynamic shared memory segments they
have mapped before unmapping those segments. So add a mechanism to
ensure that.
Existing on_shmem_exit hooks include both "user level" cleanup such
as transaction abort and removal of leftover temporary relations and
also "low level" cleanup that forcibly released leftover shared
memory resources. On-detach callbacks should run after the first
group but before the second group, so create a new before_shmem_exit
function for registering the early callbacks and keep on_shmem_exit
for the regular callbacks. (An earlier draft of this patch added an
additional argument to on_shmem_exit, but that had a much larger
footprint and probably a substantially higher risk of breaking third
party code for no real gain.)
Patch by me, reviewed by KaiGai Kohei and Andres Freund.
This prevents a possible longjmp out of the signal handler if a timeout
or SIGINT occurs while something within the handler has transiently set
ImmediateInterruptOK. For safety we must hold off the timeout or cancel
error until we're back in mainline, or at least till we reach the end of
the signal handler when ImmediateInterruptOK was true at entry. This
syncs these functions with the logic now present in handle_sig_alarm.
AFAICT there is no live bug here in 9.0 and up, because I don't think we
currently can wait for any heavyweight lock inside these functions, and
there is no other code (except read-from-client) that will turn on
ImmediateInterruptOK. However, that was not true pre-9.0: in older
branches ProcessIncomingNotify might block trying to lock pg_listener, and
then a SIGINT could lead to undesirable control flow. It might be all
right anyway given the relatively narrow code ranges in which NOTIFY
interrupts are enabled, but for safety's sake I'm back-patching this.
WAL records of hint bit updates is useful to tools that want to examine
which pages have been modified. In particular, this is required to make
the pg_rewind tool safe (without checksums).
This can also be used to test how much extra WAL-logging would occur if
you enabled checksums, without actually enabling them (which you can't
currently do without re-initdb'ing).
Sawada Masahiko, docs by Samrat Revagade. Reviewed by Dilip Kumar, with
further changes by me.
The various places that transferred fast-path locks to the main lock table
neglected to release the PGPROC's backendLock if SetupLockInTable failed
due to being out of shared memory. In most cases this is no big deal since
ensuing error cleanup would release all held LWLocks anyway. But there are
some hot-standby functions that don't consider failure of
FastPathTransferRelationLocks to be a hard error, and in those cases this
oversight could lead to system lockup. For consistency, make all of these
places look the same as FastPathTransferRelationLocks.
Noted while looking for the cause of Dan Wood's bugs --- this wasn't it,
but it's a bug anyway.
Prevent handle_sig_alarm from losing control partway through due to a query
cancel (either an asynchronous SIGINT, or a cancel triggered by one of the
timeout handler functions). That would at least result in failure to
schedule any required future interrupt, and might result in actual
corruption of timeout.c's data structures, if the interrupt happened while
we were updating those.
We could still lose control if an asynchronous SIGINT arrives just as the
function is entered. This wouldn't break any data structures, but it would
have the same effect as if the SIGALRM interrupt had been silently lost:
we'd not fire any currently-due handlers, nor schedule any new interrupt.
To forestall that scenario, forcibly reschedule any pending timer interrupt
during AbortTransaction and AbortSubTransaction. We can avoid any extra
kernel call in most cases by not doing that until we've allowed
LockErrorCleanup to kill the DEADLOCK_TIMEOUT and LOCK_TIMEOUT events.
Another hazard is that some platforms (at least Linux and *BSD) block a
signal before calling its handler and then unblock it on return. When we
longjmp out of the handler, the unblock doesn't happen, and the signal is
left blocked indefinitely. Again, we can fix that by forcibly unblocking
signals during AbortTransaction and AbortSubTransaction.
These latter two problems do not manifest when the longjmp reaches
postgres.c, because the error recovery code there kills all pending timeout
events anyway, and it uses sigsetjmp(..., 1) so that the appropriate signal
mask is restored. So errors thrown outside any transaction should be OK
already, and cleaning up in AbortTransaction and AbortSubTransaction should
be enough to fix these issues. (We're assuming that any code that catches
a query cancel error and doesn't re-throw it will do at least a
subtransaction abort to clean up; but that was pretty much required already
by other subsystems.)
Lastly, ProcSleep should not clear the LOCK_TIMEOUT indicator flag when
disabling that event: if a lock timeout interrupt happened after the lock
was granted, the ensuing query cancel is still going to happen at the next
CHECK_FOR_INTERRUPTS, and we want to report it as a lock timeout not a user
cancel.
Per reports from Dan Wood.
Back-patch to 9.3 where the new timeout handling infrastructure was
introduced. We may at some point decide to back-patch the signal
unblocking changes further, but I'll desist from that until we hear
actual field complaints about it.
We have for a long time checked the head pointer of each of the backend's
proclock lists and skipped acquiring the corresponding locktable partition
lock if the head pointer was NULL. This was safe enough in the days when
proclock lists were changed only by the owning backend, but it is pretty
questionable now that the fast-path patch added cases where backends add
entries to other backends' proclock lists. However, we don't really wish
to revert to locking each partition lock every time, because in simple
transactions that would add a lot of useless lock/unlock cycles on
already-heavily-contended LWLocks. Fortunately, the only way that another
backend could be modifying our proclock list at this point would be if it
was promoting a formerly fast-path lock of ours; and any such lock must be
one that we'd decided not to delete in the previous loop over the locallock
table. So it's okay if we miss seeing it in this loop; we'd just decide
not to delete it again. However, once we've detected a non-empty list,
we'd better re-fetch the list head pointer after acquiring the partition
lock. This guards against possibly fetching a corrupt-but-non-null pointer
if pointer fetch/store isn't atomic. It's not clear if any practical
architectures are like that, but we've never assumed that before and don't
wish to start here. In any case, the situation certainly deserves a code
comment.
While at it, refactor the partition traversal loop to use a for() construct
instead of a while() loop with goto's.
Back-patch, just in case the risk is real and not hypothetical.
When acquiring a lock in fast-path mode, we must reset the locallock
object's lock and proclock fields to NULL. They are not necessarily that
way to start with, because the locallock could be left over from a failed
lock acquisition attempt earlier in the transaction. Failure to do this
led to all sorts of interesting misbehaviors when LockRelease tried to
clean up no-longer-related lock and proclock objects in shared memory.
Per report from Dan Wood.
In passing, modify LockRelease to elog not just Assert if it doesn't find
lock and proclock objects for a formerly fast-path lock, matching the code
in FastPathGetRelationLockEntry and LockRefindAndRelease. This isn't a
bug but it will help in diagnosing any future bugs in this area.
Also, modify FastPathTransferRelationLocks and FastPathGetRelationLockEntry
to break out of their loops over the fastpath array once they've found the
sole matching entry. This was inconsistently done in some search loops
and not others.
Improve assorted related comments, too.
Back-patch to 9.2 where the fast-path mechanism was introduced.
Correct an obsolete statement that no backend touches another backend's
PROCLOCK lists. This was probably wrong even when written (the deadlock
checker looks at everybody's lists), and it's certainly quite wrong now
that fast-path locking can require creation of lock and proclock objects
on behalf of another backend. Also improve some statements in the hot
standby explanation, and do one or two other trivial bits of wordsmithing/
reformatting.
These bugs can cause data loss on standbys started with hot_standby=on at
the moment they start to accept read only queries, by marking committed
transactions as uncommited. The likelihood of such corruptions is small
unless the primary has a high transaction rate.
5a031a5556 fixed bugs in HS's startup logic
by maintaining less state until at least STANDBY_SNAPSHOT_PENDING state
was reached, missing the fact that both clog and subtrans are written to
before that. This only failed to fail in common cases because the usage
of ExtendCLOG in procarray.c was superflous since clog extensions are
actually WAL logged.
f44eedc3f0f347a856eea8590730769125964597/I then tried to fix the missing
extensions of pg_subtrans due to the former commit's changes - which are
not WAL logged - by performing the extensions when switching to a state
> STANDBY_INITIALIZED and not performing xid assignments before that -
again missing the fact that ExtendCLOG is unneccessary - but screwed up
twice: Once because latestObservedXid wasn't updated anymore in that
state due to the earlier commit and once by having an off-by-one error in
the loop performing extensions. This means that whenever a
CLOG_XACTS_PER_PAGE (32768 with default settings) boundary was crossed
between the start of the checkpoint recovery started from and the first
xl_running_xact record old transactions commit bits in pg_clog could be
overwritten if they started and committed in that window.
Fix this mess by not performing ExtendCLOG() in HS at all anymore since
it's unneeded and evidently dangerous and by performing subtrans
extensions even before reaching STANDBY_SNAPSHOT_PENDING.
Analysis and patch by Andres Freund. Reported by Christophe Pettus.
Backpatch down to 9.0, like the previous commit that caused this.
Callers expect that they only have to set the right resource owner when
creating a BufFile, not during subsequent operations on it. While we could
insist this be fixed at the caller level, it seems more sensible for the
BufFile to take care of it. Without this, some temp files belonging to
a BufFile can go away too soon, eg at the end of a subtransaction,
leading to errors or crashes.
Reported and fixed by Andres Freund. Back-patch to all active branches.
Apparently, shifts greater than or equal to the width of the type
are undefined, and can surprisingly produce a non-zero value.
Amit Kapila, with a comment by me.
To wit,
bgworker.c: In function `RegisterDynamicBackgroundWorker':
bgworker.c:761: warning: `generation' might be used uninitialized in this function
dsm_impl.c: In function `dsm_impl_op':
dsm_impl.c:197: warning: control reaches end of non-void function
Neither of these represent actual bugs, but we may as well tweak the code
so that more compilers can tell that. This won't change the generated code
on compilers that do recognize that the cases are unreachable.
All of these platforms are very much obsolete.
As far as I can determine, the last version of SINIX, later renamed
Reliant, occurred some time between 2002 and 2005.
The last release of SunOS that would run on a sun3 was released in
November of 1991; the last release of OpenBSD which supported that
platform was in 2001. The highest clock speed of any processor in
the family was 25MHz.
The NS32K (national semiconductor 320xx) architecture was retired
in 1990.
Support can be re-added if a maintainer emerges for any of these
platforms, but it seems unlikely.
Reviewed by Andres Freund.
1. In heap_hot_search_buffer(), the PredicateLockTuple() call is passed
wrong offset number. heapTuple->t_self is set to the tid of the first
tuple in the chain that's visited, not the one actually being read.
2. CheckForSerializableConflictIn() uses the tuple's t_ctid field
instead of t_self to check for exiting predicate locks on the tuple. If
the tuple was updated, but the updater rolled back, t_ctid points to the
aborted dead tuple.
Reported by Hannu Krosing. Backpatch to 9.1.
If a tuple was frozen while its predicate locks mattered,
read-write dependencies could be missed, resulting in failure to
detect conflicts which could lead to anomalies in committed
serializable transactions.
This field was added to the tag when we still thought that it was
necessary to carry locks forward to a new version of an updated
row. That was later proven to be unnecessary, which allowed
simplification of the code, but elimination of xmin from the tag
was missed at the time.
Per report and analysis by Heikki Linnakangas.
Backpatch to 9.1.
lo_open registers the currently active snapshot, and checks if the
large object exists after that. Normally, snapshots registered by lo_open
are unregistered at end of transaction when the lo descriptor is closed, but
if we error out before the lo descriptor is added to the list of open
descriptors, it is leaked. Fix by moving the snapshot registration to after
checking if the large object exists.
Reported by Pavel Stehule. Backpatch to 8.4. The snapshot registration
system was introduced in 8.4, so prior versions are not affected (and not
supported, anyway).
Using the infrastructure provided by this patch, it's possible either
to wait for the startup of a dynamically-registered background worker,
or to poll the status of such a worker without waiting. In either
case, the current PID of the worker process can also be obtained.
As usual, worker_spi is updated to demonstrate the new functionality.
Patch by me. Review by Andres Freund.
Previously, these functions took a HeapTupleHeader, but upcoming
patches for logical replication will introduce new a new snapshot
type under which the tuple's TID will be used to lookup (CMIN, CMAX)
for visibility determination purposes. This makes that information
available. Code churn is minimal since HeapTupleSatisfiesVisibility
took the HeapTuple anyway, and deferenced it before calling the
satisfies function.
Independently of logical replication, this allows t_tableOid and
t_self to be cross-checked via assertions in tqual.c. This seems
like a useful way to make sure that all callers are setting these
values properly, which has been previously put forward as
desirable.
Andres Freund, reviewed by Álvaro Herrera
There is a new API, RegisterDynamicBackgroundWorker, which allows
an ordinary user backend to register a new background writer during
normal running. This means that it's no longer necessary for all
background workers to be registered during processing of
shared_preload_libraries, although the option of registering workers
at that time remains available.
When a background worker exits and will not be restarted, the
slot previously used by that background worker is automatically
released and becomes available for reuse. Slots used by background
workers that are configured for automatic restart can't (yet) be
released without shutting down the system.
This commit adds a new source file, bgworker.c, and moves some
of the existing control logic for background workers there.
Previously, there was little enough logic that it made sense to
keep everything in postmaster.c, but not any more.
This commit also makes the worker_spi contrib module into an
extension and adds a new function, worker_spi_launch, which can
be used to demonstrate the new facility.
This patch replaces WALInsertLock with a number of WAL insertion slots,
allowing multiple backends to insert WAL records to the WAL buffers
concurrently. This is particularly useful for parallel loading large amounts
of data on a system with many CPUs.
This has one user-visible change: switching to a new WAL segment with
pg_switch_xlog() now fills the remaining unused portion of the segment with
zeros. This potentially adds some overhead, but it has been a very common
practice by DBA's to clear the "tail" of the segment with an external
pg_clearxlogtail utility anyway, to make the WAL files compress better.
With this patch, it's no longer necessary to do that.
This patch adds a new GUC, xloginsert_slots, to tune the number of WAL
insertion slots. Performance testing suggests that the default, 8, works
pretty well for all kinds of worklods, but I left the GUC in place to allow
others with different hardware to test that easily. We might want to remove
that before release.
Reviewed by Andres Freund.
In 9.3, there's no particular limit on the number of bgworkers;
instead, we just count up the number that are actually registered,
and use that to set MaxBackends. However, that approach causes
problems for Hot Standby, which needs both MaxBackends and the
size of the lock table to be the same on the standby as on the
master, yet it may not be desirable to run the same bgworkers in
both places. 9.3 handles that by failing to notice the problem,
which will probably work fine in nearly all cases anyway, but is
not theoretically sound.
A further problem with simply counting the number of registered
workers is that new workers can't be registered without a
postmaster restart. This is inconvenient for administrators,
since bouncing the postmaster causes an interruption of service.
Moreover, there are a number of applications for background
processes where, by necessity, the background process must be
started on the fly (e.g. parallel query). While this patch
doesn't actually make it possible to register new background
workers after startup time, it's a necessary prerequisite.
Patch by me. Review by Michael Paquier.
SnapshotNow scans have the undesirable property that, in the face of
concurrent updates, the scan can fail to see either the old or the new
versions of the row. In many cases, we work around this by requiring
DDL operations to hold AccessExclusiveLock on the object being
modified; in some cases, the existing locking is inadequate and random
failures occur as a result. This commit doesn't change anything
related to locking, but will hopefully pave the way to allowing lock
strength reductions in the future.
The major issue has held us back from making this change in the past
is that taking an MVCC snapshot is significantly more expensive than
using a static special snapshot such as SnapshotNow. However, testing
of various worst-case scenarios reveals that this problem is not
severe except under fairly extreme workloads. To mitigate those
problems, we avoid retaking the MVCC snapshot for each new scan;
instead, we take a new snapshot only when invalidation messages have
been processed. The catcache machinery already requires that
invalidation messages be sent before releasing the related heavyweight
lock; else other backends might rely on locally-cached data rather
than scanning the catalog at all. Thus, making snapshot reuse
dependent on the same guarantees shouldn't break anything that wasn't
already subtly broken.
Patch by me. Review by Michael Paquier and Andres Freund.
On many platforms the OS will round the sleep time to millisecond
resolution, but there is no reason for us to pre-emptively round the
argument to pg_usleep.
When the delay was measured in milliseconds and started from 1 ms, it
sometimes took many attempts until the logic that increases the delay by
multiplying with a random value between 1 and 2 actually managed to bump it
from 1 ms to 2 ms. That lead to a sequence of 1 ms waits until the delay
started to increase. This wasn't really a problem but it looked odd if you
observed the waits. There is no measurable difference in performance, but
it's more readable this way.
Jeff Janes
Valgrind "client requests" in aset.c and mcxt.c teach Valgrind and its
Memcheck tool about the PostgreSQL allocator. This makes Valgrind
roughly as sensitive to memory errors involving palloc chunks as it is
to memory errors involving malloc chunks. Further client requests in
PageAddItem() and printtup() verify that all bits being added to a
buffer page or furnished to an output function are predictably-defined.
Those tests catch failures of C-language functions to fully initialize
the bits of a Datum, which in turn stymie optimizations that rely on
_equalConst(). Define the USE_VALGRIND symbol in pg_config_manual.h to
enable these additions. An included "suppression file" silences nominal
errors we don't plan to fix.
Reviewed in earlier versions by Peter Geoghegan and Korry Douglas.
In some cases with higher numbers of subtransactions
it was possible for us to incorrectly initialize
subtrans leading to complaints of missing pages.
Bug report by Sergey Konoplev
Analysis and fix by Andres Freund
MarkBufferDirtyHint() writes WAL, and should know if it's got a
standard buffer or not. Currently, the only callers where buffer_std
is false are related to the FSM.
In passing, rename XLOG_HINT to XLOG_FPI, which is more descriptive.
Back-patch to 9.3.
pg_filedump and other external utility programs are likely to want to be
able to check Postgres page checksums. To avoid messy duplication of code,
move the checksumming functionality into an exported header file, much as
we did awhile back for the CRC code.
In passing, get rid of an unportable assumption that a static char[] array
will be word-aligned, and do some other minor code beautification.
Since commit f21bb9cfb5, this function
ignores the caller-provided length and loops until it finds a
terminator, which GetVirtualXIDsDelayingChkpt() never adds. Restore the
previous loop control logic. In passing, revert the addition of an
unused variable by the same commit, presumably a debugging relic.
AllocateFile(), AllocateDir(), and some sister routines share a small array
for remembering requests, so that the files can be closed on transaction
failure. Previously that array had a fixed size, MAX_ALLOCATED_DESCS (32).
While historically that had seemed sufficient, Steve Toutant pointed out
that this meant you couldn't scan more than 32 file_fdw foreign tables in
one query, because file_fdw depends on the COPY code which uses
AllocateFile(). There are probably other cases, or will be in the future,
where this nonconfigurable limit impedes users.
We can't completely remove any such limit, at least not without a lot of
work, since each such request requires a kernel file descriptor and most
platforms limit the number we can have. (In principle we could
"virtualize" these descriptors, as fd.c already does for the main VFD pool,
but not without an additional layer of overhead and a lot of notational
impact on the calling code.) But we can at least let the array size be
configurable. Hence, change the code to allow up to max_safe_fds/2
allocated file requests. On modern platforms this should allow several
hundred concurrent file_fdw scans, or more if one increases the value of
max_files_per_process. To go much further than that, we'd need to do some
more work on the data structure, since the current code for closing
requests has potentially O(N^2) runtime; but it should still be all right
for request counts in this range.
Back-patch to 9.1 where contrib/file_fdw was introduced.
The array allocated by GetRunningTransactionLocks() needs to be pfree'd
when we're done with it. Otherwise we leak some memory during each
checkpoint, if wal_level = hot_standby. This manifests as memory bloat
in the checkpointer process, or in bgwriter in versions before we made
the checkpointer separate.
Reported and fixed by Naoya Anzai. Back-patch to 9.0 where the issue
was introduced.
In passing, improve comments for GetRunningTransactionLocks(), and add
an Assert that we didn't overrun the palloc'd array.
PathNameOpenFile failed to ensure that the correct value of errno was
returned to its caller after a failure (because it incorrectly supposed
that free() can never change errno). In some cases this would result
in a user-visible failure because an expected ENOENT errno was replaced
with something else. Bogus EINVAL failures have been observed on OS X,
for example.
There were also a couple of places that could mangle an important value
of errno if FDDEBUG was defined. While the usefulness of that debug
support is highly debatable, we might as well make it safe to use,
so add errno save/restore logic to the DO_DB macro.
Per bug #8167 from Nelson Minar, diagnosed by RhodiumToad.
Back-patch to all supported branches.
Isolate checksum calculation to its own module, so that bufpage
knows little if anything about the details of the calculation.
This implementation is a modified FNV-1a hash checksum, details
of which are given in the new checksum.c header comments.
Basic implementation only, so we fix the output value.
Later related commits will add version numbers to pg_control,
compiler optimization flags and memory barriers.
Ants Aasma, reviewed by Jeff Davis and Simon Riggs
We copy the buffer before inserting an XLOG_HINT to avoid WAL CRC errors
caused by concurrent hint writes to buffer while share locked. To make this work
we refactor RestoreBackupBlock() to allow an XLOG_HINT to avoid the normal
path for backup blocks, which assumes the underlying buffer is exclusive locked.
Resulting code completely changes layout of XLOG_HINT WAL records, but
this isn't even beta code, so this is a low impact change.
In passing, avoid taking WALInsertLock for full page writes on checksummed
hints, remove related cruft from XLogInsert() and improve xlog_desc record for
XLOG_HINT.
Andres Freund
Bug report by Fujii Masao, testing by Jeff Janes and Jaime Casanova,
review by Jeff Davis and Simon Riggs. Applied with changes from review
and some comment editing.
The old formula didn't take into account that each WAL sender process needs
a spinlock. We had also already exceeded the fixed number of spinlocks
reserved for misc purposes (10). Bump that to 30.
Backpatch to 9.0, where WAL senders were introduced. If I counted correctly,
9.0 had exactly 10 predefined spinlocks, and 9.1 exceeded that, but bump the
limit in 9.0 too because 10 is uncomfortably close to the edge.
The modern incarnation of md.c is by no means specific to magnetic disk
technology, but every so often we hear from someone who's misled by the
label. Try to clarify that it will work for anything that supports
standard filesystem operations. Per suggestion from Andrew Dunstan.
Checksums are set immediately prior to flush out of shared buffers
and checked when pages are read in again. Hint bit setting will
require full page write when block is dirtied, which causes various
infrastructure changes. Extensive comments, docs and README.
WARNING message thrown if checksum fails on non-all zeroes page;
ERROR thrown but can be disabled with ignore_checksum_failure = on.
Feature enabled by an initdb option, since transition from option off
to option on is long and complex and has not yet been implemented.
Default is not to use checksums.
Checksum used is WAL CRC-32 truncated to 16-bits.
Simon Riggs, Jeff Davis, Greg Smith
Wide input and assistance from many community members. Thank you.
This GUC allows limiting the time spent waiting to acquire any one
heavyweight lock.
In support of this, improve the recently-added timeout infrastructure
to permit efficiently enabling or disabling multiple timeouts at once.
That reduces the performance hit from turning on lock_timeout, though
it's still not zero.
Zoltán Böszörményi, reviewed by Tom Lane,
Stephen Frost, and Hari Babu
Formerly we just Assert'ed that each refcount was zero, which was quick
and easy but failed to provide a good overview of what was wrong.
Change the code so that we'll call PrintBufferLeakWarning() for each
buffer with a nonzero refcount, and then Assert at the end of the loop.
This costs nothing in runtime and might ease diagnosis of some bugs.
Greg Smith, reviewed by Satoshi Nagayasu, further tweaked by me
A materialized view has a rule just like a view and a heap and
other physical properties like a table. The rule is only used to
populate the table, references in queries refer to the
materialized data.
This is a minimal implementation, but should still be useful in
many cases. Currently data is only populated "on demand" by the
CREATE MATERIALIZED VIEW and REFRESH MATERIALIZED VIEW statements.
It is expected that future releases will add incremental updates
with various timings, and that a more refined concept of defining
what is "fresh" data will be developed. At some point it may even
be possible to have queries use a materialized in place of
references to underlying tables, but that requires the other
above-mentioned features to be working first.
Much of the documentation work by Robert Haas.
Review by Noah Misch, Thom Brown, Robert Haas, Marko Tiikkaja
Security review by KaiGai Kohei, with a decision on how best to
implement sepgsql still pending.
This includes backend "COPY TO/FROM PROGRAM '...'" syntax, and corresponding
psql \copy syntax. Like with reading/writing files, the backend version is
superuser-only, and in the psql version, the program is run in the client.
In the passing, the psql \copy STDIN/STDOUT syntax is subtly changed: if you
the stdin/stdout is quoted, it's now interpreted as a filename. For example,
"\copy foo from 'stdin'" now reads from a file called 'stdin', not from
standard input. Before this, there was no way to specify a filename called
stdin, stdout, pstdin or pstdout.
This creates a new function in pgport, wait_result_to_str(), which can
be used to convert the exit status of a process, as returned by wait(3),
to a human-readable string.
Etsuro Fujita, reviewed by Amit Kapila.
This enables non-backend code, such as pg_xlogdump, to use it easily.
The previous location, in src/backend/catalog/catalog.c, made that
essentially impossible because that file depends on many backend-only
facilities; so this needs to live separately.
libpgcommon is a new static library to allow sharing code among the
various frontend programs and backend; this lets us eliminate duplicate
implementations of common routines. We avoid libpgport, because that's
intended as a place for porting issues; per discussion, it seems better
to keep them separate.
The first use case, and the only implemented by this patch, is pg_malloc
and friends, which many frontend programs were already using.
At the same time, we can use this to provide palloc emulation functions
for the frontend; this way, some palloc-using files in the backend can
also be used by the frontend cleanly. To do this, we change palloc() in
the backend to be a function instead of a macro on top of
MemoryContextAlloc(). This was previously believed to cause loss of
performance, but this implementation has been tweaked by Tom and Andres
so that on modern compilers it provides a slight improvement over the
previous one.
This lets us clean up some places that were already with
localized hacks.
Most of the pg_malloc/palloc changes in this patch were authored by
Andres Freund. Zoltán Böszörményi also independently provided a form of
that. libpgcommon infrastructure was authored by Álvaro.
The reason this wasn't supported before was that GiST indexes need an
increasing sequence to detect concurrent page-splits. In a regular WAL-
logged GiST index, the LSN of the page-split record is used for that
purpose, and in a temporary index, we can get away with a backend-local
counter. Neither of those methods works for an unlogged relation.
To provide such an increasing sequence of numbers, create a "fake LSN"
counter that is saved and restored across shutdowns. On recovery, unlogged
relations are blown away, so the counter doesn't need to survive that
either.
Jeevan Chalke, based on discussions with Robert Haas, Tom Lane and me.
This patch introduces two additional lock modes for tuples: "SELECT FOR
KEY SHARE" and "SELECT FOR NO KEY UPDATE". These don't block each
other, in contrast with already existing "SELECT FOR SHARE" and "SELECT
FOR UPDATE". UPDATE commands that do not modify the values stored in
the columns that are part of the key of the tuple now grab a SELECT FOR
NO KEY UPDATE lock on the tuple, allowing them to proceed concurrently
with tuple locks of the FOR KEY SHARE variety.
Foreign key triggers now use FOR KEY SHARE instead of FOR SHARE; this
means the concurrency improvement applies to them, which is the whole
point of this patch.
The added tuple lock semantics require some rejiggering of the multixact
module, so that the locking level that each transaction is holding can
be stored alongside its Xid. Also, multixacts now need to persist
across server restarts and crashes, because they can now represent not
only tuple locks, but also tuple updates. This means we need more
careful tracking of lifetime of pg_multixact SLRU files; since they now
persist longer, we require more infrastructure to figure out when they
can be removed. pg_upgrade also needs to be careful to copy
pg_multixact files over from the old server to the new, or at least part
of multixact.c state, depending on the versions of the old and new
servers.
Tuple time qualification rules (HeapTupleSatisfies routines) need to be
careful not to consider tuples with the "is multi" infomask bit set as
being only locked; they might need to look up MultiXact values (i.e.
possibly do pg_multixact I/O) to find out the Xid that updated a tuple,
whereas they previously were assured to only use information readily
available from the tuple header. This is considered acceptable, because
the extra I/O would involve cases that would previously cause some
commands to block waiting for concurrent transactions to finish.
Another important change is the fact that locking tuples that have
previously been updated causes the future versions to be marked as
locked, too; this is essential for correctness of foreign key checks.
This causes additional WAL-logging, also (there was previously a single
WAL record for a locked tuple; now there are as many as updated copies
of the tuple there exist.)
With all this in place, contention related to tuples being checked by
foreign key rules should be much reduced.
As a bonus, the old behavior that a subtransaction grabbing a stronger
tuple lock than the parent (sub)transaction held on a given tuple and
later aborting caused the weaker lock to be lost, has been fixed.
Many new spec files were added for isolation tester framework, to ensure
overall behavior is sane. There's probably room for several more tests.
There were several reviewers of this patch; in particular, Noah Misch
and Andres Freund spent considerable time in it. Original idea for the
patch came from Simon Riggs, after a problem report by Joel Jacobson.
Most code is from me, with contributions from Marti Raudsepp, Alexander
Shulgin, Noah Misch and Andres Freund.
This patch was discussed in several pgsql-hackers threads; the most
important start at the following message-ids:
AANLkTimo9XVcEzfiBR-ut3KVNDkjm2Vxh+t8kAmWjPuv@mail.gmail.com1290721684-sup-3951@alvh.no-ip.org1294953201-sup-2099@alvh.no-ip.org1320343602-sup-2290@alvh.no-ip.org1339690386-sup-8927@alvh.no-ip.org4FE5FF020200002500048A3D@gw.wicourts.gov4FEAB90A0200002500048B7D@gw.wicourts.gov
This got broken in the original fast-path locking patch, because
I failed to account for the fact that Hot Standby startup process
might take a strong relation lock on a relation in a database to
which it is not bound, and confused MyDatabaseId with the database
ID of the relation being locked.
Report and diagnosis by Andres Freund. Final form of patch by me.
When relations are dropped, at end of transaction we need to remove the
files and clean the buffer pool of buffers containing pages of those
relations. Previously we would scan the buffer pool once per relation
to clean up buffers. When there are many relations to drop, the
repeated scans make this process slow; so we now instead pass a list of
relations to drop and scan the pool once, checking each buffer against
the passed list. When the number of relations is larger than a
threshold (which as of this patch is being set to 20 relations) we sort
the array before starting, and bsearch the array; when it's smaller, we
simply scan the array linearly each time, because that's faster. The
exact optimal threshold value depends on many factors, but the
difference is not likely to be significant enough to justify making it
user-settable.
This has been measured to be a significant win (a 15x win when dropping
100,000 relations; an extreme case, but reportedly a real one).
Author: Tomas Vondra, some tweaks by me
Reviewed by: Robert Haas, Shigeru Hanada, Andres Freund, Álvaro Herrera
The code in PostPrepare_Locks supposed that it could reassign locks to
the prepared transaction's dummy PGPROC by deleting the PROCLOCK table
entries and immediately creating new ones. This was safe when that code
was written, but since we invented partitioning of the shared lock table,
it's not safe --- another process could steal away the PROCLOCK entry in
the short interval when it's on the freelist. Then, if we were otherwise
out of shared memory, PostPrepare_Locks would have to PANIC, since it's
too late to back out of the PREPARE at that point.
Fix by inventing a dynahash.c function to atomically update a hashtable
entry's key. (This might possibly have other uses in future.)
This is an ancient bug that in principle we ought to back-patch, but the
odds of someone hitting it in the field seem really tiny, because (a) the
risk window is small, and (b) nobody runs servers with maxed-out lock
tables for long, because they'll be getting non-PANIC out-of-memory errors
anyway. So fixing it in HEAD seems sufficient, at least until the new
code has gotten some testing.
If VirtualXactLock() has to wait for a transaction that holds its VXID lock
as a fast-path lock, it must first convert the fast-path lock to a regular
lock. It failed to take the required "partition" lock on the main
shared-memory lock table while doing so. This is the direct cause of the
assert failure in GetLockStatusData() recently observed in the buildfarm,
but more worryingly it could result in arbitrary corruption of the shared
lock table if some other process were concurrently engaged in modifying the
same partition of the lock table. Fortunately, VirtualXactLock() is only
used by CREATE INDEX CONCURRENTLY and DROP INDEX CONCURRENTLY, so the
opportunities for failure are fewer than they might have been.
In passing, improve some comments and be a bit more consistent about
order of operations.
In situations where there are over 8MB of empty pages at the end of
a table, the truncation work for trailing empty pages takes longer
than deadlock_timeout, and there is frequent access to the table by
processes other than autovacuum, there was a problem with the
autovacuum worker process being canceled by the deadlock checking
code. The truncation work done by autovacuum up that point was
lost, and the attempt tried again by a later autovacuum worker. The
attempts could continue indefinitely without making progress,
consuming resources and blocking other processes for up to
deadlock_timeout each time.
This patch has the autovacuum worker checking whether it is
blocking any other thread at 20ms intervals. If such a condition
develops, the autovacuum worker will persist the work it has done
so far, release its lock on the table, and sleep in 50ms intervals
for up to 5 seconds, hoping to be able to re-acquire the lock and
try again. If it is unable to get the lock in that time, it moves
on and a worker will try to continue later from the point this one
left off.
While this patch doesn't change the rules about when and what to
truncate, it does cause the truncation to occur sooner, with less
blocking, and with the consumption of fewer resources when there is
contention for the table's lock.
The only user-visible change other than improved performance is
that the table size during truncation may change incrementally
instead of just once.
This problem exists in all supported versions but is infrequently
reported, although some reports of performance problems when
autovacuum runs might be caused by this. Initial commit is just the
master branch, but this should probably be backpatched once the
build farm and general developer usage confirm that there are no
surprising effects.
Jan Wieck
Background workers are postmaster subprocesses that run arbitrary
user-specified code. They can request shared memory access as well as
backend database connections; or they can just use plain libpq frontend
database connections.
Modules listed in shared_preload_libraries can register background
workers in their _PG_init() function; this is early enough that it's not
necessary to provide an extra GUC option, because the necessary extra
resources can be allocated early on. Modules can install more than one
bgworker, if necessary.
Care is taken that these extra processes do not interfere with other
postmaster tasks: only one such process is started on each ServerLoop
iteration. This means a large number of them could be waiting to be
started up and postmaster is still able to quickly service external
connection requests. Also, shutdown sequence should not be impacted by
a worker process that's reasonably well behaved (i.e. promptly responds
to termination signals.)
The current implementation lets worker processes specify their start
time, i.e. at what point in the server startup process they are to be
started: right after postmaster start (in which case they mustn't ask
for shared memory access), when consistent state has been reached
(useful during recovery in a HOT standby server), or when recovery has
terminated (i.e. when normal backends are allowed).
In case of a bgworker crash, actions to take depend on registration
data: if shared memory was requested, then all other connections are
taken down (as well as other bgworkers), just like it were a regular
backend crashing. The bgworker itself is restarted, too, within a
configurable timeframe (which can be configured to be never).
More features to add to this framework can be imagined without much
effort, and have been discussed, but this seems good enough as a useful
unit already.
An elementary sample module is supplied.
Author: Álvaro Herrera
This patch is loosely based on prior patches submitted by KaiGai Kohei,
and unsubmitted code by Simon Riggs.
Reviewed by: KaiGai Kohei, Markus Wanner, Andres Freund,
Heikki Linnakangas, Simon Riggs, Amit Kapila
Rename PGXACT->inCommit flag into delayChkpt flag,
and generalise comments to allow use in other situations,
such as the forthcoming potential use in checksum patch.
Replace wait loop to look for VXIDs with delayChkpt set.
No user visible changes, not behaviour changes at present.
Simon Riggs, reviewed and rebased by Jeff Davis
This reverts commit c11130690d in favor of
actually fixing the problem: namely, that we should never have been
modifying the checkpoint record's nextXid at this point to begin with.
The nextXid should match the state as of the checkpoint's logical WAL
position (ie the redo point), not the state as of its physical position.
It's especially bogus to advance it in some wal_levels and not others.
In any case there is no need for the checkpoint record to carry the
same nextXid shown in the XLOG_RUNNING_XACTS record just emitted by
LogStandbySnapshot, as any replay operation will already have adopted
that value as current.
This fixes bug #7710 from Tarvi Pillessaar, and probably also explains bug
#6291 from Daniel Farina, in that if a checkpoint were in progress at the
instant of XID wraparound, the epoch bump would be lost as reported.
(And, of course, these days there's at least a 50-50 chance of a checkpoint
being in progress at any given instant.)
Diagnosed by me and independently by Andres Freund. Back-patch to all
branches supporting hot standby.
Previously we stored all xids mixed together.
Now we store top-level xids first, followed
by all subxids. Also skip logging any subxids
if the snapshot is suboverflowed, since there
are potentially large numbers of them and they
are not useful in that case anyway. Has value
in the envisaged design for decoding of WAL.
No planned effect on Hot Standby.
Andres Freund, reviewed by me
Files opened with BasicOpenFile or PathNameOpenFile are not automatically
cleaned up on error. That puts unnecessary burden on callers that only want
to keep the file open for a short time. There is AllocateFile, but that
returns a buffered FILE * stream, which in many cases is not the nicest API
to work with. So add function called OpenTransientFile, which returns a
unbuffered fd that's cleaned up like the FILE* returned by AllocateFile().
This plugs a few rare fd leaks in error cases:
1. copy_file() - fixed by by using OpenTransientFile instead of BasicOpenFile
2. XLogFileInit() - fixed by adding close() calls to the error cases. Can't
use OpenTransientFile here because the fd is supposed to persist over
transaction boundaries.
3. lo_import/lo_export - fixed by using OpenTransientFile instead of
PathNameOpenFile.
In addition to plugging those leaks, this replaces many BasicOpenFile() calls
with OpenTransientFile() that were not leaking, because the code meticulously
closed the file on error. That wasn't strictly necessary, but IMHO it's good
for robustness.
The same leaks exist in older versions, but given the rarity of the issues,
I'm not backpatching this. Not yet, anyway - it might be good to backpatch
later, after this mechanism has had some more testing in master branch.
errcontext() is typically used in an error context callback function, not
within an ereport() invocation like e.g errmsg and errdetail are. That means
that the message domain that the TEXTDOMAIN magic in ereport() determines
is not the right one for the errcontext() calls. The message domain needs to
be determined by the C file containing the errcontext() call, not the file
containing the ereport() call.
Fix by turning errcontext() into a macro that passes the TEXTDOMAIN to use
for the errcontext message. "errcontext" was used in a few places as a
variable or struct field name, I had to rename those out of the way, now
that errcontext is a macro.
We've had this problem all along, but this isn't doesn't seem worth
backporting. It's a fairly minor issue, and turning errcontext from a
function to a macro requires at least a recompile of any external code that
calls errcontext().
If an SMgrRelation is not "owned" by a relcache entry, don't allow it to
live past transaction end. This design allows the same SMgrRelation to be
used for blind writes of multiple blocks during a transaction, but ensures
that we don't hold onto such an SMgrRelation indefinitely. Because an
SMgrRelation typically corresponds to open file descriptors at the fd.c
level, leaving it open when there's no corresponding relcache entry can
mean that we prevent the kernel from reclaiming deleted disk space.
(While CacheInvalidateSmgr messages usually fix that, there are cases
where they're not issued, such as DROP DATABASE. We might want to add
some more sinval messaging for that, but I'd be inclined to keep this
type of logic anyway, since allowing VFDs to accumulate indefinitely
for blind-written relations doesn't seem like a good idea.)
This code replaces a previous attempt towards the same goal that proved
to be unreliable. Back-patch to 9.1 where the previous patch was added.
This reverts commit fba105b109.
That approach had problems with the smgr-level state not tracking what
we really want to happen, and with the VFD-level state not tracking the
smgr-level state very well either. In consequence, it was still possible
to hold kernel file descriptors open for long-gone tables (as in recent
report from Tore Halset), and yet there were also cases of FDs being closed
undesirably soon. A replacement implementation will follow.
In the previous coding, new backend processes would attempt to create their
self-pipe during the OwnLatch call in InitProcess. However, pipe creation
could fail if the kernel is short of resources; and the system does not
recover gracefully from a FATAL error right there, since we have armed the
dead-man switch for this process and not yet set up the on_shmem_exit
callback that would disarm it. The postmaster then forces an unnecessary
database-wide crash and restart, as reported by Sean Chittenden.
There are various ways we could rearrange the code to fix this, but the
simplest and sanest seems to be to split out creation of the self-pipe into
a new function InitializeLatchSupport, which must be called from a place
where failure is allowed. For most processes that gets called in
InitProcess or InitAuxiliaryProcess, but processes that don't call either
but still use latches need their own calls.
Back-patch to 9.1, which has only a part of the latch logic that 9.2 and
HEAD have, but nonetheless includes this bug.
Do read/write permissions checks at most once per large object descriptor,
not once per lo_read or lo_write call as before. The repeated tests were
quite useless in the read case since the snapshot-based tests were
guaranteed to produce the same answer every time. In the write case,
the extra tests could in principle detect revocation of write privileges
after a series of writes has started --- but there's a race condition there
anyway, since we'd check privileges before performing and certainly before
committing the write. So there's no real advantage to checking every
single time, and we might as well redefine it as "only check the first
time".
On the same reasoning, remove the LargeObjectExists checks in inv_write
and inv_truncate. We already checked existence when the descriptor was
opened, and checking again doesn't provide any real increment of safety
that would justify the cost.
Fix broken-on-bigendian-machines byte-swapping functions, add missed update
of alternate regression expected file, improve error reporting, remove some
unnecessary code, sync testlo64.c with current testlo.c (it seems to have
been cloned from a very old copy of that), assorted cosmetic improvements.
4TB large objects (standard 8KB BLCKSZ case). For this purpose new
libpq API lo_lseek64, lo_tell64 and lo_truncate64 are added. Also
corresponding new backend functions lo_lseek64, lo_tell64 and
lo_truncate64 are added. inv_api.c is changed to handle 64-bit
offsets.
Patch contributed by Nozomi Anzai (backend side) and Yugo Nagata
(frontend side, docs, regression tests and example program). Reviewed
by Kohei Kaigai. Committed by Tatsuo Ishii with minor editings.
Recovery code documents clearly that a shutdown checkpoint is executed at
end of recovery - a shutdown checkpoint WAL record is written but the buffer
manager had been altered to treat end of recovery as a normal checkpoint.
This bug exacerbates the bufmgr relpersistence bug.
Bug spotted by Andres Freund, patch by me.
This reduces unnecessary exposure of other headers through htup.h, which
is very widely included by many files.
I have chosen to move the function prototypes to the new file as well,
because that means htup.h no longer needs to include tupdesc.h. In
itself this doesn't have much effect in indirect inclusion of tupdesc.h
throughout the tree, because it's also required by execnodes.h; but it's
something to explore in the future, and it seemed best to do the htup.h
change now while I'm busy with it.
The GUC check hooks for transaction_read_only and transaction_isolation
tried to check RecoveryInProgress(), so as to disallow setting read/write
mode or serializable isolation level (respectively) in hot standby
sessions. However, GUC check hooks can be called in many situations where
we're not connected to shared memory at all, resulting in a crash in
RecoveryInProgress(). Among other cases, this results in EXEC_BACKEND
builds crashing during child process start if default_transaction_isolation
is serializable, as reported by Heikki Linnakangas. Protect those calls
by silently allowing any setting when not inside a transaction; which is
okay anyway since these GUCs are always reset at start of transaction.
Also, add a check to GetSerializableTransactionSnapshot() to complain
if we are in hot standby. We need that check despite the one in
check_XactIsoLevel() because default_transaction_isolation could be
serializable. We don't want to complain any sooner than this in such
cases, since that would prevent running transactions at all in such a
state; but a transaction can be run, if SET TRANSACTION ISOLATION is done
before setting a snapshot. Per report some months ago from Robert Haas.
Back-patch to 9.1, since these problems were introduced by the SSI patch.
Kevin Grittner and Tom Lane, with ideas from Heikki Linnakangas
Move discussion of why our algorithm for taking snapshots in recovery
to a more appropriate location in the function, and delete incorrect
mention of taking a lock.
In the original coding of the autovacuum cancel feature, commit
acac68b2bc, an autovacuum process was
considered a target for cancellation if it was found to hard-block any
process examined in the deadlock search. This patch tightens the test so
that the autovacuum must directly hard-block the current process. This
should make the behavior more predictable in general, and in particular
it ensures that an autovacuum will not be canceled with less than
deadlock_timeout grace period. In the old coding, it was possible for an
autovacuum to be canceled almost instantly, given unfortunate timing of two
or more other processes' lock attempts.
This also justifies the logging methodology in the recent commit
d7318d43d891bd63e82dcfc27948113ed7b1db80; without this restriction, that
patch isn't providing enough information to see the connection of the
canceling process to the autovacuum. Like that one, patch all the way
back.
The old message was at DEBUG2, so typically it didn't show up in the
log at all. As a result, in most cases where autovacuum was canceled,
the only information that was logged was the table being vacuumed,
with no indication as to what problem caused the cancel. Crank up
the level to LOG and add some more details to assist with debugging.
Back-patch all the way, per discussion on pgsql-hackers.
We should avoid calling sync_file_range or posix_fadvise in this case,
since (a) we don't really care if the data gets synced, and might as
well save the kernel calls; (b) at least on Linux we know that the
kernel might block us until it's scheduled the write.
Also, avoid making a useless second traversal of the directory tree
if we're not actually going to call fsync(2) after all.
Instead of having one hash table entry per relation/fork/segment, just have
one per relation, and use bitmapsets to represent which specific segments
need to be fsync'd. This eliminates the need to scan the whole hash table
to implement FORGET_RELATION_FSYNC, which fixes the O(N^2) behavior
recently demonstrated by Jeff Janes for cases involving lots of TRUNCATE or
DROP TABLE operations during a single checkpoint cycle. Per an idea from
Robert Haas.
(FORGET_DATABASE_FSYNC still sucks, but since dropping a database is a
pretty expensive operation anyway, we'll live with that.)
In passing, improve the delayed-unlink code: remove the pass over the list
in mdpreckpt, since it wasn't doing anything for us except supporting a
useless Assert in mdpostckpt, and fix mdpostckpt so that it will absorb
fsync requests every so often when clearing a large backlog of deletion
requests.
We were sending one per fork, but a little bit of refactoring allows us
to send just one request with forknum == InvalidForkNumber. This not only
reduces pressure on the shared-memory request queue, but saves repeated
traversals of the checkpointer's hash table.
mdinit() was misusing IsBootstrapProcessingMode() to decide whether to
create an fsync pending-operations table in the current process. This led
to creating a table not only in the startup and checkpointer processes as
intended, but also in the bgwriter process, not to mention other auxiliary
processes such as walwriter and walreceiver. Creation of the table in the
bgwriter is fatal, because it absorbs fsync requests that should have gone
to the checkpointer; instead they just sit in bgwriter local memory and are
never acted on. So writes performed by the bgwriter were not being fsync'd
which could result in data loss after an OS crash. I think there is no
live bug with respect to walwriter and walreceiver because those never
perform any writes of shared buffers; but the potential is there for
future breakage in those processes too.
To fix, make AuxiliaryProcessMain() export the current process's
AuxProcType as a global variable, and then make mdinit() test directly for
the types of aux process that should have a pendingOpsTable. Having done
that, we might as well also get rid of the random bool flags such as
am_walreceiver that some of the aux processes had grown. (Note that we
could not have fixed the bug by examining those variables in mdinit(),
because it's called from BaseInit() which is run by AuxiliaryProcessMain()
before entering any of the process-type-specific code.)
Back-patch to 9.2, where the problem was introduced by the split-up of
bgwriter and checkpointer processes. The bogus pendingOpsTable exists
in walwriter and walreceiver processes in earlier branches, but absent
any evidence that it causes actual problems there, I'll leave the older
branches alone.
In all branches back to 8.3, this patch fixes a questionable assumption in
CompactCheckpointerRequestQueue/CompactBgwriterRequestQueue that there are
no uninitialized pad bytes in the request queue structs. This would only
cause trouble if (a) there were such pad bytes, which could happen in 8.4
and up if the compiler makes enum ForkNumber narrower than 32 bits, but
otherwise would require not-currently-planned changes in the widths of
other typedefs; and (b) the kernel has not uniformly initialized the
contents of shared memory to zeroes. Still, it seems a tad risky, and we
can easily remove any risk by pre-zeroing the request array for ourselves.
In addition to that, we need to establish a coding rule that struct
RelFileNode can't contain any padding bytes, since such structs are copied
into the request array verbatim. (There are other places that are assuming
this anyway, it turns out.)
In 9.1 and up, the risk was a bit larger because we were also effectively
assuming that struct RelFileNodeBackend contained no pad bytes, and with
fields of different types in there, that would be much easier to break.
However, there is no good reason to ever transmit fsync or delete requests
for temp files to the bgwriter/checkpointer, so we can revert the request
structs to plain RelFileNode, getting rid of the padding risk and saving
some marginal number of bytes and cycles in fsync queue manipulation while
we are at it. The savings might be more than marginal during deletion of
a temp relation, because the old code transmitted an entirely useless but
nonetheless expensive-to-process ForgetRelationFsync request to the
background process, and also had the background process perform the file
deletion even though that can safely be done immediately.
In addition, make some cleanup of nearby comments and small improvements to
the code in CompactCheckpointerRequestQueue/CompactBgwriterRequestQueue.
Management of timeouts was getting a little cumbersome; what we
originally had was more than enough back when we were only concerned
about deadlocks and query cancel; however, when we added timeouts for
standby processes, the code got considerably messier. Since there are
plans to add more complex timeouts, this seems a good time to introduce
a central timeout handling module.
External modules register their timeout handlers during process
initialization, and later enable and disable them as they see fit using
a simple API; timeout.c is in charge of keeping track of which timeouts
are in effect at any time, installing a common SIGALRM signal handler,
and calling setitimer() as appropriate to ensure timely firing of
external handlers.
timeout.c additionally supports pluggable modules to add their own
timeouts, though this capability isn't exercised anywhere yet.
Additionally, as of this commit, walsender processes are aware of
timeouts; we had a preexisting bug there that made those ignore SIGALRM,
thus being subject to unhandled deadlocks, particularly during the
authentication phase. This has already been fixed in back branches in
commit 0bf8eb2a, which see for more details.
Main author: Zoltán Böszörményi
Some review and cleanup by Álvaro Herrera
Extensive reworking by Tom Lane
The Solaris Studio compiler warns about these instances, unlike more
mainstream compilers such as gcc. But manual inspection showed that
the code is clearly not reachable, and we hope no worthy compiler will
complain about removing this code.
Historically we have not worried about fsync'ing anything during initdb
(in fact, initdb intentionally passes -F to each backend launch to prevent
it from fsync'ing). But with filesystems getting more aggressive about
caching data, that's not such a good plan anymore. Make initdb do a pass
over the finished data directory tree to fsync everything. For testing
purposes, the -N/--nosync flag can be used to restore the old behavior.
Also, testing shows that on Linux, sync_file_range() is much faster than
posix_fadvise() for hinting to the kernel that an fsync is coming,
apparently because the latter blocks on a rather small request queue while
the former doesn't. So use this function if available in initdb, and also
in the backend's pg_flush_data() (where it currently will affect only the
speed of CREATE DATABASE's cloning step).
We will later make pg_regress invoke initdb with the --nosync flag
to avoid slowing down cases such as "make check" in contrib. But
let's not do so until we've shaken out any portability issues in this
patch.
Jeff Davis, reviewed by Andres Freund
This simplifies code that needs to do arithmetic on XLogRecPtrs.
To avoid changing on-disk format of data pages, the LSN on data pages is
still stored in the old format. That should keep pg_upgrade happy. However,
we have XLogRecPtrs embedded in the control file, and in the structs that
are sent over the replication protocol, so this changes breaks compatibility
of pg_basebackup and server. I didn't do anything about this in this patch,
per discussion on -hackers, the right thing to do would to be to change the
replication protocol to be architecture-independent, so that you could use
a newer version of pg_receivexlog, for example, against an older server
version.
This speeds up reassigning locks to the parent owner, when the transaction
holds a lot of locks, but only a few of them belong to the current resource
owner. This is particularly helps pg_dump when dumping a large number of
objects.
The cache can hold up to 15 locks in each resource owner. After that, the
cache is marked as overflowed, and we fall back to the old method of
scanning the whole local lock table. The tradeoff here is that the cache has
to be scanned whenever a lock is released, so if the cache is too large,
lock release becomes more expensive. 15 seems enough to cover pg_dump, and
doesn't have much impact on lock release.
Jeff Janes, reviewed by Amit Kapila and Heikki Linnakangas.
When HS startup is deferred because of overflowed subtransactions, ensure
that we re-initialize KnownAssignedXids for when both existing and incoming
snapshots have non-zero qualifying xids.
Fixes bug #6661 reported by Valentine Gogichashvili.
Analysis and fix by Andres Freund
This provides a speedup of about 4X when NBuffers is large enough.
There is also a useful reduction in sinval traffic, since we
only do CacheInvalidateSmgr() once not once per fork.
Simon Riggs, reviewed and somewhat revised by Tom Lane
DropRelFileNodeBuffers, DropDatabaseBuffers, FlushRelationBuffers, and
FlushDatabaseBuffers have to scan the whole shared_buffers pool because
we have no index structure that would find the target buffers any more
efficiently than that. This gets expensive with large NBuffers. We can
shave some cycles from these loops by prechecking to see if the current
buffer is interesting before we acquire the buffer header lock.
Ordinarily such a test would be unsafe, but in these cases it should be
safe because we are already assuming that the caller holds a lock that
prevents any new target pages from being loaded into the buffer pool
concurrently. Therefore, no buffer tag should be changing to a value of
interest, only away from a value of interest. So a false negative match
is impossible, while a false positive is safe because we'll recheck after
acquiring the buffer lock. Initial testing says that this speeds these
loops by a factor of 2X to 3X on common Intel hardware.
Patch for DropRelFileNodeBuffers by Jeff Janes (based on an idea of
Heikki's); extended to the remaining sequential scans by Tom Lane
First, the previous code failed to account for the fact that, during Hot
Standby operation, the startup process takes AccessExclusiveLocks on
relations without setting MyDatabaseId. This resulted in fast path
strong lock counts failing to be incremented with the startup process
took locks, which in turn allowed conflicting lock requests to succeed
when they should not have. Report by Erik Rijkers, diagnosis by Heikki
Linnakangas.
Second, LockReleaseAll() failed to honor the allLocks and lockmethodid
restrictions with respect to fast-path locks. It's not clear to me
whether this produces any user-visible breakage at the moment, but it's
certainly wrong. Rearrange order of operations in LockReleaseAll to fix.
Noted by Tom Lane.
In commit d526575f89, we changed things so
that buffer usage counts are incremented when the buffer is pinned, rather
than when it is unpinned, but the README file didn't get the memo.
Report by Amit Kapila.
When the "hot" members of PGPROC were split off to separate PGXACT structs,
many PGPROC fields referred to in comments were moved to PGXACT, but the
comments were neglected in the commit. Mostly this is just a search/replace
of PGPROC with PGXACT, but the way the dummy PGPROC entries are created for
prepared transactions changed more, making some of the comments totally
bogus.
Noah Misch
Commit 6d90eaaa89 added a hibernation mode
to the bgwriter to reduce the server's idle-power consumption. However,
its interaction with the detailed behavior of BgBufferSync's feedback
control loop wasn't very well thought out. That control loop depends
primarily on the rate of buffer allocation, not the rate of buffer
dirtying, so the hibernation mode has to be designed to operate only when
no new buffer allocations are happening. Also, the check for whether the
system is effectively idle was not quite right and would fail to detect
a constant low level of activity, thus allowing the bgwriter to go into
hibernation mode in a way that would let the cycle time vary quite a bit,
possibly further confusing the feedback loop. To fix, move the wakeup
support from MarkBufferDirty and SetBufferCommitInfoNeedsSave into
StrategyGetBuffer, and prevent the bgwriter from entering hibernation mode
unless no buffer allocations have happened recently.
In addition, fix the delaying logic to remove the problem of possibly not
responding to signals promptly, which was basically caused by trying to use
the process latch's is_set flag for multiple purposes. I can't prove it
but I'm suspicious that that hack was responsible for the intermittent
"postmaster does not shut down" failures we've been seeing in the buildfarm
lately. In any case it did nothing to improve the readability or
robustness of the code.
In passing, express the hibernation sleep time as a multiplier on
BgWriterDelay, not a constant. I'm not sure whether there's any value in
exposing the longer sleep time as an independently configurable setting,
but we can at least make it act like this for little extra code.
This patch modifies the walwriter process so that, when it has not found
anything useful to do for many consecutive wakeup cycles, it extends its
sleep time to reduce the server's idle power consumption. It reverts to
normal as soon as it's done any successful flushes. It's still true that
during any async commit, backends check for completed, unflushed pages of
WAL and signal the walwriter if there are any; so that in practice the
walwriter can get awakened and returned to normal operation sooner than the
sleep time might suggest.
Also, improve the checkpointer so that it uses a latch and a computed delay
time to not wake up at all except when it has something to do, replacing a
previous hardcoded 0.5 sec wakeup cycle. This also is primarily useful for
reducing the server's power consumption when idle.
In passing, get rid of the dedicated latch for signaling the walwriter in
favor of using its procLatch, since that comports better with possible
generic signal handlers using that latch. Also, fix a pre-existing bug
with failure to save/restore errno in walwriter's signal handlers.
Peter Geoghegan, somewhat simplified by Tom
Commit 62c7bd31c8 had assorted problems, most
visibly that it broke PREPARE TRANSACTION in the presence of session-level
advisory locks (which should be ignored by PREPARE), as per a recent
complaint from Stephen Rees. More abstractly, the patch made the
LockMethodData.transactional flag not merely useless but outright
dangerous, because in point of fact that flag no longer tells you anything
at all about whether a lock is held transactionally. This fix therefore
removes that flag altogether. We now rely entirely on the convention
already in use in lock.c that transactional lock holds must be owned by
some ResourceOwner, while session holds are never so owned. Setting the
locallock struct's owner link to NULL thus denotes a session hold, and
there is no redundant marker for that.
PREPARE TRANSACTION now works again when there are session-level advisory
locks, and it is also able to transfer transactional advisory locks to the
prepared transaction, but for implementation reasons it throws an error if
we hold both types of lock on a single lockable object. Perhaps it will be
worth improving that someday.
Assorted other minor cleanup and documentation editing, as well.
Back-patch to 9.1, except that in the 9.1 branch I did not remove the
LockMethodData.transactional flag for fear of causing an ABI break for
any external code that might be examining those structs.
The previous code could cause a backend crash after BEGIN; SAVEPOINT a;
LOCK TABLE foo (interrupted by ^C or statement timeout); ROLLBACK TO
SAVEPOINT a; LOCK TABLE foo, and might have leaked strong-lock counts
in other situations.
Report by Zoltán Böszörményi; patch review by Jeff Davis.
Ants Aasma's original patch to add timing information for buffer I/O
requests exposed this data at the relation level, which was judged too
costly. I've here exposed it at the database level instead.
Postmaster sets max_safe_fds by testing how many open file descriptors it
can open, and that is normally inherited by all child processes at fork().
Not so on EXEC_BACKEND, ie. Windows, however. Because of that, we
effectively ignored max_files_per_process on Windows, and always assumed
a conservative default of 32 simultaneous open files. That could have an
impact on performance, if you need to access a lot of different files
in a query. After this patch, the value is passed to child processes by
save/restore_backend_variables() among many other global variables.
It has been like this forever, but given the lack of complaints about it,
I'm not backpatching this.
Currently, the only way to see the numbers this gathers is via
EXPLAIN (ANALYZE, BUFFERS), but the plan is to add visibility through
the stats collector and pg_stat_statements in subsequent patches.
Ants Aasma, reviewed by Greg Smith, with some further changes by me.
For those variables only used when asserts are enabled, use a new
macro PG_USED_FOR_ASSERTS_ONLY, which expands to
__attribute__((unused)) when asserts are not enabled.
This allows loadable modules to get control at drop time, perhaps for the
purpose of performing additional security checks or to log the event.
The initial purpose of this code is to support sepgsql, but other
applications should be possible as well.
KaiGai Kohei, reviewed by me.
A prepared transaction can get new conflicts in and out after preparing, so
we cannot rely on the in- and out-flags stored in the statefile at prepare-
time. As a quick fix, make the conservative assumption that after a restart,
all prepared transactions are considered to have both in- and out-conflicts.
That can lead to unnecessary rollbacks after a crash, but that shouldn't be
a big problem in practice; you don't want prepared transactions to hang
around for a long time anyway.
Dan Ports
If LWLockWaitUntilFree was called before the first LWLockAcquire call, you
would either crash because of access to uninitialized array or account the
acquisition incorrectly. LWLockConditionalAcquire doesn't have this problem
because it doesn't update the lwlock stats.
In practice, this never happens because there is no codepath where you would
call LWLockWaitUntilfree before LWLockAcquire after a new process is
launched. But that's just accidental, there's no guarantee that that's
always going to be true in the future.
Spotted by Jeff Janes.
Originally, most of this code assumed that no Postgres backends could be
running concurrently with it, and so no locking could be needed. That
assumption fails in Hot Standby. While it's still true that Hot Standby
backends should never change values like nextXid, they can examine them,
and consistency is important in some cases such as when computing a
snapshot. Therefore, prudence requires that WAL replay code obtain the
relevant locks when modifying such variables, even though it can examine
them without taking a lock. We were following that coding rule in some
places but not all. This commit applies the coding rule uniformly to all
updates of ShmemVariableCache and MultiXactState fields; a search of the
replay routines did not find any other cases that seemed to be at risk.
In addition, this commit fixes a longstanding thinko in replay of NEXTOID
and checkpoint records: we tried to advance nextOid only if it was behind
the value in the WAL record, but the comparison would draw the wrong
conclusion if OID wraparound had occurred since the previous value.
Better to just unconditionally assign the new value, since OID assignment
shouldn't be happening during replay anyway.
The additional locking seems to be more in the nature of future-proofing
than fixing any live bug, so I am not going to back-patch it. The NEXTOID
fix will be back-patched separately.
All other WAL redo routines either call RestoreBkpBlocks() or Assert that
they haven't been passed any backup blocks. Make this one do likewise.
Also, fix incorrect routine name in its failure message.
If there was a wait-until-free process in the head of the wait queue,
followed by an exclusive locker, the exclusive locker was not be woken up
as it should.
When a backend needs to flush the WAL, and someone else is already flushing
the WAL, wait until it releases the WALInsertLock and check if we still need
to do the flush or if the other backend already did the work for us, before
acquiring WALInsertLock. This helps group commit, because when the WAL flush
finishes, all the backends that were waiting for it can be woken up in one
go, and the can all concurrently observe that they're done, rather than
waking them up one by one in a cascading fashion.
This is based on a new LWLock function, LWLockWaitUntilFree(), which has
peculiar semantics. If the lock is immediately free, it grabs the lock and
returns true. If it's not free, it waits until it is released, but then
returns false without grabbing the lock. This is used in XLogFlush(), so
that when the lock is acquired, the backend flushes the WAL, but if it's
not, the backend first checks the current flush location before retrying.
Original patch and benchmarking by Peter Geoghegan and Simon Riggs, although
this patch as committed ended up being very different from that.
