database/postgres
1
0
Fork 0
mirror of https://github.com/postgres/postgres.git synced 2025-07-17 06:41:09 +03:00
Code Activity

Replace the BufMgrLock with separate locks on the lookup hashtable and

Browse Source 
the freelist, plus per-buffer spinlocks that protect access to individual
shared buffer headers.  This requires abandoning a global freelist (since
the freelist is a global contention point), which shoots down ARC and 2Q
as well as plain LRU management.  Adopt a clock sweep algorithm instead.
Preliminary results show substantial improvement in multi-backend situations.
This commit is contained in:
Tom Lane
2005-03-04 20:21:07 +00:00
parent 5592a6cf46
commit 5d5087363d
18 changed files with 1410 additions and 1932 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
src/backend/catalog/index.c
View File

@ -8,7 +8,7 @@
*
*
* IDENTIFICATION
* $PostgreSQL: pgsql/src/backend/catalog/index.c,v 1.244 2005/01/10 20:02:19 tgl Exp $
* $PostgreSQL: pgsql/src/backend/catalog/index.c,v 1.245 2005/03/04 20:21:05 tgl Exp $
*
*
* INTERFACE ROUTINES
@ -1060,7 +1060,6 @@ setRelhasindex(Oid relid, bool hasindex, bool isprimary, Oid reltoastidxid)
/* Send out shared cache inval if necessary */
if (!IsBootstrapProcessingMode())
CacheInvalidateHeapTuple(pg_class, tuple);
BufferSync(-1, -1);
}
else if (dirty)
{

6
src/backend/commands/dbcommands.c
View File

@ -15,7 +15,7 @@
*
*
* IDENTIFICATION
* $PostgreSQL: pgsql/src/backend/commands/dbcommands.c,v 1.151 2005/02/26 18:43:33 tgl Exp $
* $PostgreSQL: pgsql/src/backend/commands/dbcommands.c,v 1.152 2005/03/04 20:21:05 tgl Exp $
*
*-------------------------------------------------------------------------
*/
@ -339,7 +339,7 @@ createdb(const CreatedbStmt *stmt)
* up-to-date for the copy. (We really only need to flush buffers for
* the source database, but bufmgr.c provides no API for that.)
*/
BufferSync(-1, -1);
BufferSync();
/*
* Close virtual file descriptors so the kernel has more available for
@ -1201,7 +1201,7 @@ dbase_redo(XLogRecPtr lsn, XLogRecord *record)
* up-to-date for the copy. (We really only need to flush buffers for
* the source database, but bufmgr.c provides no API for that.)
*/
BufferSync(-1, -1);
BufferSync();
#ifndef WIN32

3
src/backend/commands/vacuum.c
View File

@ -13,7 +13,7 @@
*
*
* IDENTIFICATION
* $PostgreSQL: pgsql/src/backend/commands/vacuum.c,v 1.302 2005/02/26 18:43:33 tgl Exp $
* $PostgreSQL: pgsql/src/backend/commands/vacuum.c,v 1.303 2005/03/04 20:21:06 tgl Exp $
*
*-------------------------------------------------------------------------
*/
@ -36,7 +36,6 @@
#include "commands/vacuum.h"
#include "executor/executor.h"
#include "miscadmin.h"
#include "storage/buf_internals.h"
#include "storage/freespace.h"
#include "storage/sinval.h"
#include "storage/smgr.h"

21
src/backend/postmaster/bgwriter.c
View File

@ -37,7 +37,7 @@
*
*
* IDENTIFICATION
* $PostgreSQL: pgsql/src/backend/postmaster/bgwriter.c,v 1.14 2005/02/19 23:16:15 tgl Exp $
* $PostgreSQL: pgsql/src/backend/postmaster/bgwriter.c,v 1.15 2005/03/04 20:21:06 tgl Exp $
*
*-------------------------------------------------------------------------
*/
@ -116,9 +116,6 @@ static BgWriterShmemStruct *BgWriterShmem;
* GUC parameters
*/
int BgWriterDelay = 200;
int BgWriterPercent = 1;
int BgWriterMaxPages = 100;
int CheckPointTimeout = 300;
int CheckPointWarning = 30;
@ -274,7 +271,6 @@ BackgroundWriterMain(void)
bool force_checkpoint = false;
time_t now;
int elapsed_secs;
int n;
long udelay;
/*
@ -365,16 +361,13 @@ BackgroundWriterMain(void)
* checkpoints happen at a predictable spacing.
*/
last_checkpoint_time = now;
/* Nap for configured time before rechecking */
n = 1;
}
else
n = BufferSync(BgWriterPercent, BgWriterMaxPages);
BgBufferSync();
/*
* Nap for the configured time or sleep for 10 seconds if there
* was nothing to do at all.
* Nap for the configured time, or sleep for 10 seconds if there
* is no bgwriter activity configured.
*
* On some platforms, signals won't interrupt the sleep. To ensure
* we respond reasonably promptly when someone signals us, break
@ -383,7 +376,11 @@ BackgroundWriterMain(void)
*
* We absorb pending requests after each short sleep.
*/
udelay = ((n > 0) ? BgWriterDelay : 10000) * 1000L;
if ((bgwriter_all_percent > 0.0 && bgwriter_all_maxpages > 0) ||
(bgwriter_lru_percent > 0.0 && bgwriter_lru_maxpages > 0))
udelay = BgWriterDelay * 1000L;
else
udelay = 10000000L;
while (udelay > 1000000L)
{
if (got_SIGHUP || checkpoint_requested || shutdown_requested)

251
src/backend/storage/buffer/README
View File

@ -1,12 +1,12 @@
$PostgreSQL: pgsql/src/backend/storage/buffer/README,v 1.7 2004/04/19 23:27:17 tgl Exp $
$PostgreSQL: pgsql/src/backend/storage/buffer/README,v 1.8 2005/03/04 20:21:06 tgl Exp $
Notes about shared buffer access rules
--------------------------------------
There are two separate access control mechanisms for shared disk buffers:
reference counts (a/k/a pin counts) and buffer locks. (Actually, there's
a third level of access control: one must hold the appropriate kind of
lock on a relation before one can legally access any page belonging to
reference counts (a/k/a pin counts) and buffer content locks. (Actually,
there's a third level of access control: one must hold the appropriate kind
of lock on a relation before one can legally access any page belonging to
the relation. Relation-level locks are not discussed here.)
Pins: one must "hold a pin on" a buffer (increment its reference count)
@ -26,7 +26,7 @@ handled by waiting to obtain the relation-level lock, which is why you'd
better hold one first.) Pins may not be held across transaction
boundaries, however.
Buffer locks: there are two kinds of buffer locks, shared and exclusive,
Buffer content locks: there are two kinds of buffer lock, shared and exclusive,
which act just as you'd expect: multiple backends can hold shared locks on
the same buffer, but an exclusive lock prevents anyone else from holding
either shared or exclusive lock. (These can alternatively be called READ
@ -38,12 +38,12 @@ the same buffer. One must pin a buffer before trying to lock it.
Buffer access rules:
1. To scan a page for tuples, one must hold a pin and either shared or
exclusive lock. To examine the commit status (XIDs and status bits) of
a tuple in a shared buffer, one must likewise hold a pin and either shared
exclusive content lock. To examine the commit status (XIDs and status bits)
of a tuple in a shared buffer, one must likewise hold a pin and either shared
or exclusive lock.
2. Once one has determined that a tuple is interesting (visible to the
current transaction) one may drop the buffer lock, yet continue to access
current transaction) one may drop the content lock, yet continue to access
the tuple's data for as long as one holds the buffer pin. This is what is
typically done by heap scans, since the tuple returned by heap_fetch
contains a pointer to tuple data in the shared buffer. Therefore the
@ -52,9 +52,9 @@ change, but that is assumed not to matter after the initial determination
of visibility is made.
3. To add a tuple or change the xmin/xmax fields of an existing tuple,
one must hold a pin and an exclusive lock on the containing buffer.
one must hold a pin and an exclusive content lock on the containing buffer.
This ensures that no one else might see a partially-updated state of the
tuple.
tuple while they are doing visibility checks.
4. It is considered OK to update tuple commit status bits (ie, OR the
values HEAP_XMIN_COMMITTED, HEAP_XMIN_INVALID, HEAP_XMAX_COMMITTED, or
@ -76,7 +76,7 @@ no other backend can be holding a reference to an existing tuple that it
might expect to examine again. Note that another backend might pin the
buffer (increment the refcount) while one is performing the cleanup, but
it won't be able to actually examine the page until it acquires shared
or exclusive lock.
or exclusive content lock.
VACUUM FULL ignores rule #5, because it instead acquires exclusive lock at
@ -97,149 +97,142 @@ for VACUUM's use, since we don't allow multiple VACUUMs concurrently on a
single relation anyway.
Buffer replacement strategy interface
-------------------------------------
Buffer manager's internal locking
---------------------------------
The file freelist.c contains the buffer cache replacement strategy.
The interface to the strategy is:
Before PostgreSQL 8.1, all operations of the shared buffer manager itself
were protected by a single system-wide lock, the BufMgrLock, which
unsurprisingly proved to be a source of contention. The new locking scheme
avoids grabbing system-wide exclusive locks in common code paths. It works
like this:
BufferDesc *StrategyBufferLookup(BufferTag *tagPtr, bool recheck,
int *cdb_found_index)
* There is a system-wide LWLock, the BufMappingLock, that notionally
protects the mapping from buffer tags (page identifiers) to buffers.
(Physically, it can be thought of as protecting the hash table maintained
by buf_table.c.) To look up whether a buffer exists for a tag, it is
sufficient to obtain share lock on the BufMappingLock. Note that one
must pin the found buffer, if any, before releasing the BufMappingLock.
To alter the page assignment of any buffer, one must hold exclusive lock
on the BufMappingLock. This lock must be held across adjusting the buffer's
header fields and changing the buf_table hash table. The only common
operation that needs exclusive lock is reading in a page that was not
in shared buffers already, which will require at least a kernel call
and usually a wait for I/O, so it will be slow anyway.
This is always the first call made by the buffer manager to check if a disk
page is in memory. If so, the function returns the buffer descriptor and no
further action is required. If the page is not in memory,
StrategyBufferLookup() returns NULL.
* A separate system-wide LWLock, the BufFreelistLock, provides mutual
exclusion for operations that access the buffer free list or select
buffers for replacement. This is always taken in exclusive mode since
there are no read-only operations on those data structures. The buffer
management policy is designed so that BufFreelistLock need not be taken
except in paths that will require I/O, and thus will be slow anyway.
(Details appear below.) It is never necessary to hold the BufMappingLock
and the BufFreelistLock at the same time.
The flag recheck tells the strategy that this is a second lookup after
flushing a dirty block. If the buffer manager has to evict another buffer,
it will release the bufmgr lock while doing the write IO. During this time,
another backend could possibly fault in the same page this backend is after,
so we have to check again after the IO is done if the page is in memory now.
* Each buffer header contains a spinlock that must be taken when examining
or changing fields of that buffer header. This allows operations such as
ReleaseBuffer to make local state changes without taking any system-wide
lock. We use a spinlock, not an LWLock, since there are no cases where
the lock needs to be held for more than a few instructions.
*cdb_found_index is set to the index of the found CDB, or -1 if none.
This is not intended to be used by the caller, except to pass to
StrategyReplaceBuffer().
Note that a buffer header's spinlock does not control access to the data
held within the buffer. Each buffer header also contains an LWLock, the
"buffer content lock", that *does* represent the right to access the data
in the buffer. It is used per the rules above.
BufferDesc *StrategyGetBuffer(int *cdb_replace_index)
The buffer manager calls this function to get an unpinned cache buffer whose
content can be evicted. The returned buffer might be empty, clean or dirty.
The returned buffer is only a candidate for replacement. It is possible that
while the buffer is being written, another backend finds and modifies it, so
that it is dirty again. The buffer manager will then have to call
StrategyGetBuffer() again to ask for another candidate.
*cdb_replace_index is set to the index of the candidate CDB, or -1 if none
(meaning we are using a previously free buffer). This is not intended to be
used by the caller, except to pass to StrategyReplaceBuffer().
void StrategyReplaceBuffer(BufferDesc *buf, BufferTag *newTag,
int cdb_found_index, int cdb_replace_index)
Called by the buffer manager at the time it is about to change the association
of a buffer with a disk page.
Before this call, StrategyBufferLookup() still has to find the buffer under
its old tag, even if it was returned by StrategyGetBuffer() as a candidate
for replacement.
After this call, this buffer must be returned for a lookup of the new page
identified by *newTag.
cdb_found_index and cdb_replace_index must be the auxiliary values
returned by previous calls to StrategyBufferLookup and StrategyGetBuffer.
void StrategyInvalidateBuffer(BufferDesc *buf)
Called by the buffer manager to inform the strategy that the content of this
buffer is being thrown away. This happens for example in the case of dropping
a relation. The buffer must be clean and unpinned on call.
If the buffer was associated with a disk page, StrategyBufferLookup()
must not return it for this page after the call.
void StrategyHintVacuum(bool vacuum_active)
Because VACUUM reads all relations of the entire database through the buffer
manager, it can greatly disturb the buffer replacement strategy. This function
is used by VACUUM to inform the strategy that subsequent buffer lookups are
(or are not) caused by VACUUM scanning relations.
There is yet another set of per-buffer LWLocks, the io_in_progress locks,
that are used to wait for I/O on a buffer to complete. The process doing
a read or write takes exclusive lock for the duration, and processes that
need to wait for completion try to take shared locks (which they release
immediately upon obtaining). XXX on systems where an LWLock represents
nontrivial resources, it's fairly annoying to need so many locks. Possibly
we could use per-backend LWLocks instead (a buffer header would then contain
a field to show which backend is doing its I/O).
Buffer replacement strategy
---------------------------
The buffer replacement strategy actually used in freelist.c is a version of
the Adaptive Replacement Cache (ARC) specially tailored for PostgreSQL.
There is a "free list" of buffers that are prime candidates for replacement.
In particular, buffers that are completely free (contain no valid page) are
always in this list. We may also throw buffers into this list if we
consider their pages unlikely to be needed soon. The list is singly-linked
using fields in the buffer headers; we maintain head and tail pointers in
global variables. (Note: although the list links are in the buffer headers,
they are considered to be protected by the BufFreelistLock, not the
buffer-header spinlocks.) To choose a victim buffer to recycle when there
are no free buffers available, we use a simple clock-sweep algorithm, which
avoids the need to take system-wide locks during common operations. It
works like this:
The algorithm works as follows:
Each buffer header contains a usage counter, which is incremented (up to a
small limit value) whenever the buffer is unpinned. (This requires only the
buffer header spinlock, which would have to be taken anyway to decrement the
buffer reference count, so it's nearly free.)
C is the size of the cache in number of pages (a/k/a shared_buffers or
NBuffers). ARC uses 2*C Cache Directory Blocks (CDB). A cache directory block
is always associated with one unique file page. It may point to one shared
buffer, or may indicate that the file page is not in a buffer but has been
accessed recently.
The "clock hand" is a buffer index, NextVictimBuffer, that moves circularly
through all the available buffers. NextVictimBuffer is protected by the
BufFreelistLock.
All CDB entries are managed in 4 LRU lists named T1, T2, B1 and B2. The T1 and
T2 lists are the "real" cache entries, linking a file page to a memory buffer
where the page is currently cached. Consequently T1len+T2len <= C. B1 and B2
are ghost cache directories that extend T1 and T2 so that the strategy
remembers pages longer. The strategy tries to keep B1len+T1len and B2len+T2len
both at C. T1len and T2len vary over the runtime depending on the lookup
pattern and its resulting cache hits. The desired size of T1len is called
T1target.
The algorithm for a process that needs to obtain a victim buffer is:
Assuming we have a full cache, one of 5 cases happens on a lookup:
1. Obtain BufFreelistLock.
MISS On a cache miss, depending on T1target and the actual T1len
the LRU buffer of either T1 or T2 is evicted. Its CDB is removed
from the T list and added as MRU of the corresponding B list.
The now free buffer is replaced with the requested page
and added as MRU of T1.
2. If buffer free list is nonempty, remove its head buffer. If the buffer
is pinned or has a nonzero usage count, it cannot be used; ignore it and
return to the start of step 2. Otherwise, pin the buffer, release
BufFreelistLock, and return the buffer.
T1 hit The T1 CDB is moved to the MRU position of the T2 list.
3. Otherwise, select the buffer pointed to by NextVictimBuffer, and
circularly advance NextVictimBuffer for next time.
T2 hit The T2 CDB is moved to the MRU position of the T2 list.
4. If the selected buffer is pinned or has a nonzero usage count, it cannot
be used. Decrement its usage count (if nonzero) and return to step 3 to
examine the next buffer.
B1 hit This means that a buffer that was evicted from the T1
list is now requested again, indicating that T1target is
too small (otherwise it would still be in T1 and thus in
memory). The strategy raises T1target, evicts a buffer
depending on T1target and T1len and places the CDB at
MRU of T2.
5. Pin the selected buffer, release BufFreelistLock, and return the buffer.
B2 hit This means the opposite of B1, the T2 list is probably too
small. So the strategy lowers T1target, evicts a buffer
and places the CDB at MRU of T2.
(Note that if the selected buffer is dirty, we will have to write it out
before we can recycle it; if someone else pins the buffer meanwhile we will
have to give up and try another buffer. This however is not a concern
of the basic select-a-victim-buffer algorithm.)
Thus, every page that is found on lookup in any of the four lists
ends up as the MRU of the T2 list. The T2 list therefore is the
"frequency" cache, holding frequently requested pages.
A special provision is that while running VACUUM, a backend does not
increment the usage count on buffers it accesses. In fact, if ReleaseBuffer
sees that it is dropping the pin count to zero and the usage count is zero,
then it appends the buffer to the tail of the free list. (This implies that
VACUUM, but only VACUUM, must take the BufFreelistLock during ReleaseBuffer;
this shouldn't create much of a contention problem.) This provision
encourages VACUUM to work in a relatively small number of buffers rather
than blowing out the entire buffer cache. It is reasonable since a page
that has been touched only by VACUUM is unlikely to be needed again soon.
Every page that is seen for the first time ends up as the MRU of the T1
list. The T1 list is the "recency" cache, holding recent newcomers.
The tailoring done for PostgreSQL has to do with the way the query executor
works. A typical UPDATE or DELETE first scans the relation, searching for the
tuples and then calls heap_update() or heap_delete(). This causes at least 2
lookups for the block in the same statement. In the case of multiple matches
in one block even more often. As a result, every block touched in an UPDATE or
DELETE would directly jump into the T2 cache, which is wrong. To prevent this
the strategy remembers which transaction added a buffer to the T1 list and
will not promote it from there into the T2 cache during the same transaction.
Another specialty is the change of the strategy during VACUUM. Lookups during
VACUUM do not represent application needs, and do not suggest that the page
will be hit again soon, so it would be wrong to change the cache balance
T1target due to that or to cause massive cache evictions. Therefore, a page
read in to satisfy vacuum is placed at the LRU position of the T1 list, for
immediate reuse. Also, if we happen to get a hit on a CDB entry during
VACUUM, we do not promote the page above its current position in the list.
Since VACUUM usually requests many pages very fast, the effect of this is that
it will get back the very buffers it filled and possibly modified on the next
call and will therefore do its work in a few shared memory buffers, while
being able to use whatever it finds in the cache already. This also implies
that most of the write traffic caused by a VACUUM will be done by the VACUUM
itself and not pushed off onto other processes.
Background writer's processing
------------------------------
The background writer is designed to write out pages that are likely to be
recycled soon, thereby offloading the writing work from active backends.
To do this, it scans forward circularly from the current position of
NextVictimBuffer (which it does not change!), looking for buffers that are
dirty and not pinned nor marked with a positive usage count. It pins,
writes, and releases any such buffer.
If we can assume that reading NextVictimBuffer is an atomic action, then
the writer doesn't even need to take the BufFreelistLock in order to look
for buffers to write; it needs only to spinlock each buffer header for long
enough to check the dirtybit. Even without that assumption, the writer
only needs to take the lock long enough to read the variable value, not
while scanning the buffers. (This is a very substantial improvement in
the contention cost of the writer compared to PG 8.0.)
During a checkpoint, the writer's strategy must be to write every dirty
buffer (pinned or not!). We may as well make it start this scan from
NextVictimBuffer, however, so that the first-to-be-written pages are the
ones that backends might otherwise have to write for themselves soon.

76
src/backend/storage/buffer/buf_init.c
View File

@ -8,7 +8,7 @@
*
*
* IDENTIFICATION
* $PostgreSQL: pgsql/src/backend/storage/buffer/buf_init.c,v 1.71 2005/02/03 23:29:11 tgl Exp $
* $PostgreSQL: pgsql/src/backend/storage/buffer/buf_init.c,v 1.72 2005/03/04 20:21:06 tgl Exp $
*
*-------------------------------------------------------------------------
*/
@ -22,6 +22,8 @@ BufferDesc *BufferDescriptors;
Block *BufferBlockPointers;
int32 *PrivateRefCount;
static char *BufferBlocks;
/* statistics counters */
long int ReadBufferCount;
long int ReadLocalBufferCount;
@ -50,16 +52,11 @@ long int LocalBufferFlushCount;
*
* Synchronization/Locking:
*
* BufMgrLock lock -- must be acquired before manipulating the
* buffer search datastructures (lookup/freelist, as well as the
* flag bits of any buffer). Must be released
* before exit and before doing any IO.
*
* IO_IN_PROGRESS -- this is a flag in the buffer descriptor.
* It must be set when an IO is initiated and cleared at
* the end of the IO. It is there to make sure that one
* process doesn't start to use a buffer while another is
* faulting it in. see IOWait/IOSignal.
* faulting it in. see WaitIO and related routines.
*
* refcount -- Counts the number of processes holding pins on a buffer.
* A buffer is pinned during IO and immediately after a BufferAlloc().
@ -85,10 +82,8 @@ long int LocalBufferFlushCount;
void
InitBufferPool(void)
{
char *BufferBlocks;
bool foundBufs,
foundDescs;
int i;
BufferDescriptors = (BufferDesc *)
ShmemInitStruct("Buffer Descriptors",
@ -102,52 +97,42 @@ InitBufferPool(void)
{
/* both should be present or neither */
Assert(foundDescs && foundBufs);
/* note: this path is only taken in EXEC_BACKEND case */
}
else
{
BufferDesc *buf;
char *block;
/*
* It's probably not really necessary to grab the lock --- if
* there's anyone else attached to the shmem at this point, we've
* got problems.
*/
LWLockAcquire(BufMgrLock, LW_EXCLUSIVE);
int i;
buf = BufferDescriptors;
block = BufferBlocks;
/*
* Initialize all the buffer headers.
*/
for (i = 0; i < NBuffers; block += BLCKSZ, buf++, i++)
for (i = 0; i < NBuffers; buf++, i++)
{
Assert(ShmemIsValid((unsigned long) block));
/*
* The bufNext fields link together all totally-unused buffers.
* Subsequent management of this list is done by
* StrategyGetBuffer().
*/
buf->bufNext = i + 1;
CLEAR_BUFFERTAG(buf->tag);
buf->flags = 0;
buf->usage_count = 0;
buf->refcount = 0;
buf->wait_backend_id = 0;
SpinLockInit(&buf->buf_hdr_lock);
buf->buf_id = i;
buf->data = MAKE_OFFSET(block);
buf->flags = 0;
buf->refcount = 0;
/*
* Initially link all the buffers together as unused.
* Subsequent management of this list is done by freelist.c.
*/
buf->freeNext = i + 1;
buf->io_in_progress_lock = LWLockAssign();
buf->cntx_lock = LWLockAssign();
buf->cntxDirty = false;
buf->wait_backend_id = 0;
buf->content_lock = LWLockAssign();
}
/* Correct last entry of linked list */
BufferDescriptors[NBuffers - 1].bufNext = -1;
LWLockRelease(BufMgrLock);
BufferDescriptors[NBuffers - 1].freeNext = FREENEXT_END_OF_LIST;
}
/* Init other shared buffer-management stuff */
@ -162,12 +147,13 @@ InitBufferPool(void)
* buffer pool.
*
* NB: this is called before InitProcess(), so we do not have a PGPROC and
* cannot do LWLockAcquire; hence we can't actually access the bufmgr's
* cannot do LWLockAcquire; hence we can't actually access stuff in
* shared memory yet. We are only initializing local data here.
*/
void
InitBufferPoolAccess(void)
{
char *block;
int i;
/*
@ -179,12 +165,18 @@ InitBufferPoolAccess(void)
sizeof(*PrivateRefCount));
/*
* Convert shmem offsets into addresses as seen by this process. This
* is just to speed up the BufferGetBlock() macro. It is OK to do this
* without any lock since the data pointers never change.
* Construct addresses for the individual buffer data blocks. We do
* this just to speed up the BufferGetBlock() macro. (Since the
* addresses should be the same in every backend, we could inherit
* this data from the postmaster --- but in the EXEC_BACKEND case
* that doesn't work.)
*/
block = BufferBlocks;
for (i = 0; i < NBuffers; i++)
BufferBlockPointers[i] = (Block) MAKE_PTR(BufferDescriptors[i].data);
{
BufferBlockPointers[i] = (Block) block;
block += BLCKSZ;
}
}
/*

38
src/backend/storage/buffer/buf_table.c
View File

@ -3,12 +3,9 @@
* buf_table.c
* routines for mapping BufferTags to buffer indexes.
*
* NOTE: this module is called only by freelist.c, and the "buffer IDs"
* it deals with are whatever freelist.c needs them to be; they may not be
* directly equivalent to Buffer numbers.
*
* Note: all routines in this file assume that the BufMgrLock is held
* by the caller, so no synchronization is needed.
* Note: the routines in this file do no locking of their own. The caller
* must hold a suitable lock on the BufMappingLock, as specified in the
* comments.
*
*
* Portions Copyright (c) 1996-2005, PostgreSQL Global Development Group
@ -16,7 +13,7 @@
*
*
* IDENTIFICATION
* $PostgreSQL: pgsql/src/backend/storage/buffer/buf_table.c,v 1.39 2005/02/03 23:29:11 tgl Exp $
* $PostgreSQL: pgsql/src/backend/storage/buffer/buf_table.c,v 1.40 2005/03/04 20:21:06 tgl Exp $
*
*-------------------------------------------------------------------------
*/
@ -74,17 +71,17 @@ InitBufTable(int size)
/*
* BufTableLookup
* Lookup the given BufferTag; return buffer ID, or -1 if not found
*
* Caller must hold at least share lock on BufMappingLock
*/
int
BufTableLookup(BufferTag *tagPtr)
{
BufferLookupEnt *result;
if (tagPtr->blockNum == P_NEW)
return -1;
result = (BufferLookupEnt *)
hash_search(SharedBufHash, (void *) tagPtr, HASH_FIND, NULL);
if (!result)
return -1;
@ -93,14 +90,23 @@ BufTableLookup(BufferTag *tagPtr)
/*
* BufTableInsert
* Insert a hashtable entry for given tag and buffer ID
* Insert a hashtable entry for given tag and buffer ID,
* unless an entry already exists for that tag
*
* Returns -1 on successful insertion. If a conflicting entry exists
* already, returns the buffer ID in that entry.
*
* Caller must hold write lock on BufMappingLock
*/
void
int
BufTableInsert(BufferTag *tagPtr, int buf_id)
{
BufferLookupEnt *result;
bool found;
Assert(buf_id >= 0); /* -1 is reserved for not-in-table */
Assert(tagPtr->blockNum != P_NEW); /* invalid tag */
result = (BufferLookupEnt *)
hash_search(SharedBufHash, (void *) tagPtr, HASH_ENTER, &found);
@ -109,15 +115,19 @@ BufTableInsert(BufferTag *tagPtr, int buf_id)
(errcode(ERRCODE_OUT_OF_MEMORY),
errmsg("out of shared memory")));
if (found) /* found something already in the table? */
elog(ERROR, "shared buffer hash table corrupted");
if (found) /* found something already in the table */
return result->id;
result->id = buf_id;
return -1;
}
/*
* BufTableDelete
* Delete the hashtable entry for given tag (which must exist)
*
* Caller must hold write lock on BufMappingLock
*/
void
BufTableDelete(BufferTag *tagPtr)

1615
src/backend/storage/buffer/bufmgr.c
View File

File diff suppressed because it is too large Load Diff

939
src/backend/storage/buffer/freelist.c
View File

File diff suppressed because it is too large Load Diff

76
src/backend/storage/buffer/localbuf.c
View File

@ -9,7 +9,7 @@
*
*
* IDENTIFICATION
* $PostgreSQL: pgsql/src/backend/storage/buffer/localbuf.c,v 1.62 2005/01/10 20:02:21 tgl Exp $
* $PostgreSQL: pgsql/src/backend/storage/buffer/localbuf.c,v 1.63 2005/03/04 20:21:06 tgl Exp $
*
*-------------------------------------------------------------------------
*/
@ -24,6 +24,10 @@
/*#define LBDEBUG*/
/* Note: this macro only works on local buffers, not shared ones! */
#define LocalBufHdrGetBlock(bufHdr) \
LocalBufferBlockPointers[-((bufHdr)->buf_id + 2)]
/* should be a GUC parameter some day */
int NLocBuffer = 64;
@ -39,7 +43,7 @@ static int nextFreeLocalBuf = 0;
* allocate a local buffer. We do round robin allocation for now.
*
* API is similar to bufmgr.c's BufferAlloc, except that we do not need
* to have the BufMgrLock since this is all local. Also, IO_IN_PROGRESS
* to do any locking since this is all local. Also, IO_IN_PROGRESS
* does not get set.
*/
BufferDesc *
@ -47,11 +51,12 @@ LocalBufferAlloc(Relation reln, BlockNumber blockNum, bool *foundPtr)
{
BufferTag newTag; /* identity of requested block */
int i;
int trycounter;
BufferDesc *bufHdr;
INIT_BUFFERTAG(newTag, reln, blockNum);
/* a low tech search for now -- not optimized for scans */
/* a low tech search for now -- should use a hashtable */
for (i = 0; i < NLocBuffer; i++)
{
bufHdr = &LocalBufferDescriptors[i];
@ -81,32 +86,44 @@ LocalBufferAlloc(Relation reln, BlockNumber blockNum, bool *foundPtr)
RelationGetRelid(reln), blockNum, -nextFreeLocalBuf - 1);
#endif
/* need to get a new buffer (round robin for now) */
bufHdr = NULL;
for (i = 0; i < NLocBuffer; i++)
/*
* Need to get a new buffer. We use a clock sweep algorithm
* (essentially the same as what freelist.c does now...)
*/
trycounter = NLocBuffer;
for (;;)
{
int b = (nextFreeLocalBuf + i) % NLocBuffer;
int b = nextFreeLocalBuf;
if (LocalRefCount[b] == 0)
if (++nextFreeLocalBuf >= NLocBuffer)
nextFreeLocalBuf = 0;
bufHdr = &LocalBufferDescriptors[b];
if (LocalRefCount[b] == 0 && bufHdr->usage_count == 0)
{
bufHdr = &LocalBufferDescriptors[b];
LocalRefCount[b]++;
ResourceOwnerRememberBuffer(CurrentResourceOwner,
BufferDescriptorGetBuffer(bufHdr));
nextFreeLocalBuf = (b + 1) % NLocBuffer;
BufferDescriptorGetBuffer(bufHdr));
break;
}
if (bufHdr->usage_count > 0)
{
bufHdr->usage_count--;
trycounter = NLocBuffer;
}
else if (--trycounter == 0)
ereport(ERROR,
(errcode(ERRCODE_INSUFFICIENT_RESOURCES),
errmsg("no empty local buffer available")));
}
if (bufHdr == NULL)
ereport(ERROR,
(errcode(ERRCODE_INSUFFICIENT_RESOURCES),
errmsg("no empty local buffer available")));
/*
* this buffer is not referenced but it might still be dirty. if
* that's the case, write it out before reusing it!
*/
if (bufHdr->flags & BM_DIRTY || bufHdr->cntxDirty)
if (bufHdr->flags & BM_DIRTY)
{
SMgrRelation oreln;
@ -116,7 +133,7 @@ LocalBufferAlloc(Relation reln, BlockNumber blockNum, bool *foundPtr)
/* And write... */
smgrwrite(oreln,
bufHdr->tag.blockNum,
(char *) MAKE_PTR(bufHdr->data),
(char *) LocalBufHdrGetBlock(bufHdr),
true);
LocalBufferFlushCount++;
@ -129,7 +146,7 @@ LocalBufferAlloc(Relation reln, BlockNumber blockNum, bool *foundPtr)
* use, so it's okay to do it (and possibly error out) before marking
* the buffer as not dirty.
*/
if (bufHdr->data == (SHMEM_OFFSET) 0)
if (LocalBufHdrGetBlock(bufHdr) == NULL)
{
char *data = (char *) malloc(BLCKSZ);
@ -138,17 +155,10 @@ LocalBufferAlloc(Relation reln, BlockNumber blockNum, bool *foundPtr)
(errcode(ERRCODE_OUT_OF_MEMORY),
errmsg("out of memory")));
/*
* This is a bit of a hack: bufHdr->data needs to be a shmem
* offset for consistency with the shared-buffer case, so make it
* one even though it's not really a valid shmem offset.
*/
bufHdr->data = MAKE_OFFSET(data);
/*
* Set pointer for use by BufferGetBlock() macro.
*/
LocalBufferBlockPointers[-(bufHdr->buf_id + 2)] = (Block) data;
LocalBufHdrGetBlock(bufHdr) = (Block) data;
}
/*
@ -156,7 +166,8 @@ LocalBufferAlloc(Relation reln, BlockNumber blockNum, bool *foundPtr)
*/
bufHdr->tag = newTag;
bufHdr->flags &= ~(BM_VALID | BM_DIRTY | BM_JUST_DIRTIED | BM_IO_ERROR);
bufHdr->cntxDirty = false;
bufHdr->flags |= BM_TAG_VALID;
bufHdr->usage_count = 0;
*foundPtr = FALSE;
return bufHdr;
@ -170,6 +181,7 @@ void
WriteLocalBuffer(Buffer buffer, bool release)
{
int bufid;
BufferDesc *bufHdr;
Assert(BufferIsLocal(buffer));
@ -178,12 +190,18 @@ WriteLocalBuffer(Buffer buffer, bool release)
#endif
bufid = -(buffer + 1);
LocalBufferDescriptors[bufid].flags |= BM_DIRTY;
Assert(LocalRefCount[bufid] > 0);
bufHdr = &LocalBufferDescriptors[bufid];
bufHdr->flags |= BM_DIRTY;
if (release)
{
Assert(LocalRefCount[bufid] > 0);
LocalRefCount[bufid]--;
if (LocalRefCount[bufid] == 0 &&
bufHdr->usage_count < BM_MAX_USAGE_COUNT)
bufHdr->usage_count++;
ResourceOwnerForgetBuffer(CurrentResourceOwner, buffer);
}
}

46
src/backend/utils/misc/guc.c
View File

@ -10,7 +10,7 @@
* Written by Peter Eisentraut <peter_e@gmx.net>.
*
* IDENTIFICATION
* $PostgreSQL: pgsql/src/backend/utils/misc/guc.c,v 1.253 2005/03/01 20:23:34 tgl Exp $
* $PostgreSQL: pgsql/src/backend/utils/misc/guc.c,v 1.254 2005/03/04 20:21:06 tgl Exp $
*
*--------------------------------------------------------------------
*/
@ -77,7 +77,6 @@ extern bool Log_disconnections;
extern DLLIMPORT bool check_function_bodies;
extern int CommitDelay;
extern int CommitSiblings;
extern int DebugSharedBuffers;
extern char *default_tablespace;
static const char *assign_log_destination(const char *value,
@ -1230,15 +1229,6 @@ static struct config_int ConfigureNamesInt[] =
-1, -1, INT_MAX / 1000, NULL, NULL
},
{
{"debug_shared_buffers", PGC_POSTMASTER, STATS_MONITORING,
gettext_noop("Interval to report shared buffer status in seconds"),
NULL
},
&DebugSharedBuffers,
0, 0, 600, NULL, NULL
},
{
{"bgwriter_delay", PGC_SIGHUP, RESOURCES,
gettext_noop("Background writer sleep time between rounds in milliseconds"),
@ -1249,21 +1239,21 @@ static struct config_int ConfigureNamesInt[] =
},
{
{"bgwriter_percent", PGC_SIGHUP, RESOURCES,
gettext_noop("Background writer percentage of dirty buffers to flush per round"),
{"bgwriter_lru_maxpages", PGC_SIGHUP, RESOURCES,
gettext_noop("Background writer maximum number of all pages to flush per round"),
NULL
},
&BgWriterPercent,
1, 0, 100, NULL, NULL
&bgwriter_lru_maxpages,
5, 0, 1000, NULL, NULL
},
{
{"bgwriter_maxpages", PGC_SIGHUP, RESOURCES,
gettext_noop("Background writer maximum number of pages to flush per round"),
{"bgwriter_all_maxpages", PGC_SIGHUP, RESOURCES,
gettext_noop("Background writer maximum number of LRU pages to flush per round"),
NULL
},
&BgWriterMaxPages,
100, 0, 1000, NULL, NULL
&bgwriter_all_maxpages,
5, 0, 1000, NULL, NULL
},
{
@ -1394,6 +1384,24 @@ static struct config_real ConfigureNamesReal[] =
MAX_GEQO_SELECTION_BIAS, NULL, NULL
},
{
{"bgwriter_lru_percent", PGC_SIGHUP, RESOURCES,
gettext_noop("Background writer percentage of LRU buffers to flush per round"),
NULL
},
&bgwriter_lru_percent,
1.0, 0.0, 100.0, NULL, NULL
},
{
{"bgwriter_all_percent", PGC_SIGHUP, RESOURCES,
gettext_noop("Background writer percentage of all buffers to flush per round"),
NULL
},
&bgwriter_all_percent,
0.333, 0.0, 100.0, NULL, NULL
},
{
{"seed", PGC_USERSET, UNGROUPED,
gettext_noop("Sets the seed for random-number generation."),

6
src/backend/utils/misc/postgresql.conf.sample
View File

@ -99,8 +99,10 @@
# - Background writer -
#bgwriter_delay = 200 # 10-10000 milliseconds between rounds
#bgwriter_percent = 1 # 0-100% of dirty buffers in each round
#bgwriter_maxpages = 100 # 0-1000 buffers max per round
#bgwriter_lru_percent = 1.0 # 0-100% of LRU buffers scanned in each round
#bgwriter_lru_maxpages = 5 # 0-1000 buffers max written per round
#bgwriter_all_percent = 0.333 # 0-100% of all buffers scanned in each round
#bgwriter_all_maxpages = 5 # 0-1000 buffers max written per round
#---------------------------------------------------------------------------

9
src/backend/utils/resowner/resowner.c
View File

@ -14,7 +14,7 @@
*
*
* IDENTIFICATION
* $PostgreSQL: pgsql/src/backend/utils/resowner/resowner.c,v 1.9 2004/12/31 22:02:50 pgsql Exp $
* $PostgreSQL: pgsql/src/backend/utils/resowner/resowner.c,v 1.10 2005/03/04 20:21:06 tgl Exp $
*
*-------------------------------------------------------------------------
*/
@ -200,12 +200,7 @@ ResourceOwnerReleaseInternal(ResourceOwner owner,
* that would indicate failure to clean up the executor correctly ---
* so issue warnings. In the abort case, just clean up quietly.
*
* XXX this is fairly inefficient due to multiple BufMgrLock
* grabs if there are lots of buffers to be released, but we
* don't expect many (indeed none in the success case) so it's
* probably not worth optimizing.
*
* We are however careful to release back-to-front, so as to
* We are careful to do the releasing back-to-front, so as to
* avoid O(N^2) behavior in ResourceOwnerForgetBuffer().
*/
while (owner->nbuffers > 0)