Commit Graph

29 Commits

Author SHA1 Message Date
Andres Freund
48354581a4 Allow Pin/UnpinBuffer to operate in a lockfree manner.
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
2016-04-10 20:12:32 -07:00
Bruce Momjian
807b9e0dff pgindent run for 9.5 2015-05-23 21:35:49 -04:00
Heikki Linnakangas
4fc72cc7bb Collection of typo fixes.
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.
2015-05-20 16:56:22 +03:00
Andres Freund
ed127002d8 Align buffer descriptors to cache line boundaries.
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.
2015-01-29 22:48:45 +01:00
Andres Freund
4b4b680c3d Make backend local tracking of buffer pins memory efficient.
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
2014-08-30 14:03:21 +02:00
Andres Freund
ebf20f65d9 Fix newly introduced misspelling of existence in pg_buffercache.
Peter Geoghegan
2014-08-22 17:49:47 +02:00
Andres Freund
f57791985a Add pinning_backends column to the pg_buffercache extension.
The new column shows how many backends have a buffer pinned. That can
be useful during development or to diagnose production issues
e.g. caused by vacuum waiting for cleanup locks.

To handle upgrades transparently - the extension might be used in
views - deal with callers expecting the old number of columns.

Reviewed by Fujii Masao and Rajeev rastogi.
2014-08-22 00:28:37 +02:00
Bruce Momjian
6a605cd6bd Adjust blank lines around PG_MODULE_MAGIC defines, for consistency
Report by Robert Haas
2014-07-10 14:02:08 -04:00
Peter Eisentraut
e7128e8dbb Create function prototype as part of PG_FUNCTION_INFO_V1 macro
Because of gcc -Wmissing-prototypes, all functions in dynamically
loadable modules must have a separate prototype declaration.  This is
meant to detect global functions that are not declared in header files,
but in cases where the function is called via dfmgr, this is redundant.
Besides filling up space with boilerplate, this is a frequent source of
compiler warnings in extension modules.

We can fix that by creating the function prototype as part of the
PG_FUNCTION_INFO_V1 macro, which such modules have to use anyway.  That
makes the code of modules cleaner, because there is one less place where
the entry points have to be listed, and creates an additional check that
functions have the right prototype.

Remove now redundant prototypes from contrib and other modules.
2014-04-18 00:03:19 -04:00
Robert Haas
ea9df812d8 Relax the requirement that all lwlocks be stored in a single array.
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.
2014-01-27 11:07:44 -05:00
Alvaro Herrera
c219d9b0a5 Split tuple struct defs from htup.h to htup_details.h
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.
2012-08-30 16:52:35 -04:00
Bruce Momjian
6416a82a62 Remove unnecessary #include references, per pgrminclude script. 2011-09-01 10:04:27 -04:00
Magnus Hagander
9f2e211386 Remove cvs keywords from all files. 2010-09-20 22:08:53 +02:00
Bruce Momjian
d747140279 8.4 pgindent run, with new combined Linux/FreeBSD/MinGW typedef list
provided by Andrew.
2009-06-11 14:49:15 +00:00
Heikki Linnakangas
ef1c807c25 pg_buffercache needs to be taught about relation forks, as Greg Stark
pointed out.
2008-08-14 12:56:41 +00:00
Bruce Momjian
fdf5a5efb7 pgindent run for 8.3. 2007-11-15 21:14:46 +00:00
Tom Lane
2a275e6d3c Fix pg_buffercache to release buffer partition locks in reverse order,
and add a note about why.  This is not tremendously important right now,
probably, but it will get more urgent if NUM_BUFFER_PARTITIONS is increased
as much as proposed.
2007-07-16 21:20:36 +00:00
Bruce Momjian
b0194ab110 Add the usage count statistics to the information available in
contrib/pgbuffercache.

Greg Smith
2007-04-07 16:09:14 +00:00
Tom Lane
bd95c74985 Code cleanup for pg_buffercache, from Mark Kirkwood. 2006-10-22 17:49:21 +00:00
Tom Lane
def651f48f Clean up local redeclarations of variables with DLLIMPORT, per report
from Magnus that MSVC complains about this.
2006-10-19 18:32:48 +00:00
Bruce Momjian
f99a569a2e pgindent run for 8.2. 2006-10-04 00:30:14 +00:00
Tom Lane
10b9ca3d05 Split the buffer mapping table into multiple separately lockable
partitions, as per discussion.  Passes functionality checks, but
I don't have any performance data yet.
2006-07-23 03:07:58 +00:00
Tom Lane
a0ffab351e Magic blocks don't do us any good unless we use 'em ... so install one
in every shared library.
2006-05-30 22:12:16 +00:00
Bruce Momjian
1dc3498251 Standard pgindent run for 8.1. 2005-10-15 02:49:52 +00:00
Tom Lane
07eeb9d109 Do all accesses to shared buffer headers through volatile-qualified
pointers, to ensure that compilers won't rearrange accesses to occur
while we're not holding the buffer header spinlock.  It's probably
not necessary to mark volatile in every single place in bufmgr.c,
but better safe than sorry.  Per trouble report from Kevin Grittner.
2005-10-12 16:45:14 +00:00
Tom Lane
12a323b7a8 Change relblocknumber field of pg_buffercache view from numeric to int8
for efficiency's sake.  Mark Kirkwood.
2005-05-31 00:07:47 +00:00
Neil Conway
f5dbbdcf00 Remove pg_buffercache_pages.h -- patch from Mark Kirkwood. 2005-04-01 08:41:19 +00:00
Neil Conway
e1b398c6f5 Fix build error in pg_buffercache on Win32 (native and cygwin). From
Mark Kirkwood.
2005-03-31 00:45:26 +00:00
Neil Conway
de004e44e2 Add pg_buffercache contrib module, from Mark Kirkwood. 2005-03-12 15:36:24 +00:00