database/postgres
1
0
Fork 0
mirror of https://github.com/postgres/postgres.git synced 2025-09-02 04:21:28 +03:00
Code Activity

Fix up rickety handling of relation-truncation interlocks.

Browse Source 
Move rd_targblock, rd_fsm_nblocks, and rd_vm_nblocks from relcache to the smgr
relation entries, so that they will get reset to InvalidBlockNumber whenever
an smgr-level flush happens.  Because we now send smgr invalidation messages
immediately (not at end of transaction) when a relation truncation occurs,
this ensures that other backends will reset their values before they next
access the relation.  We no longer need the unreliable assumption that a
VACUUM that's doing a truncation will hold its AccessExclusive lock until
commit --- in fact, we can intentionally release that lock as soon as we've
completed the truncation.  This patch therefore reverts (most of) Alvaro's
patch of 2009-11-10, as well as my marginal hacking on it yesterday.  We can
also get rid of assorted no-longer-needed relcache flushes, which are far more
expensive than an smgr flush because they kill a lot more state.

In passing this patch fixes smgr_redo's failure to perform visibility-map
truncation, and cleans up some rather dubious assumptions in freespace.c and
visibilitymap.c about when rd_fsm_nblocks and rd_vm_nblocks can be out of
date.
This commit is contained in:
Tom Lane
2010-02-09 21:43:30 +00:00
parent 79647eed86
commit cbe9d6beb4
15 changed files with 210 additions and 219 deletions
Download Patch File Download Diff File Expand all files Collapse all files

45
src/backend/storage/smgr/smgr.c
View File

@@ -11,7 +11,7 @@
*
*
* IDENTIFICATION
* $PostgreSQL: pgsql/src/backend/storage/smgr/smgr.c,v 1.119 2010/02/03 01:14:17 tgl Exp $
* $PostgreSQL: pgsql/src/backend/storage/smgr/smgr.c,v 1.120 2010/02/09 21:43:30 tgl Exp $
*
*-------------------------------------------------------------------------
*/
@@ -161,6 +161,9 @@ smgropen(RelFileNode rnode)
/* hash_search already filled in the lookup key */
reln->smgr_owner = NULL;
reln->smgr_targblock = InvalidBlockNumber;
reln->smgr_fsm_nblocks = InvalidBlockNumber;
reln->smgr_vm_nblocks = InvalidBlockNumber;
reln->smgr_which = 0; /* we only have md.c at present */
/* mark it not open */
@@ -351,6 +354,16 @@ smgr_internal_unlink(RelFileNode rnode, ForkNumber forknum,
* anyway).
*/
/*
* Send a shared-inval message to force other backends to close any
* dangling smgr references they may have for this rel. We should do
* this before starting the actual unlinking, in case we fail partway
* through that step. Note that the sinval message will eventually come
* back to this backend, too, and thereby provide a backstop that we
* closed our own smgr rel.
*/
CacheInvalidateSmgr(rnode);
/*
* Delete the physical file(s).
*
@@ -359,14 +372,6 @@ smgr_internal_unlink(RelFileNode rnode, ForkNumber forknum,
* xact.
*/
(*(smgrsw[which].smgr_unlink)) (rnode, forknum, isRedo);
/*
* Lastly, send a shared-inval message to force other backends to close
* any dangling smgr references they may have for this rel. We do this
* last because the sinval will eventually come back to this backend, too,
* and thereby provide a backstop that we closed our own smgr rel.
*/
CacheInvalidateSmgr(rnode);
}
/*
@@ -459,21 +464,23 @@ smgrtruncate(SMgrRelation reln, ForkNumber forknum, BlockNumber nblocks,
*/
DropRelFileNodeBuffers(reln->smgr_rnode, forknum, isTemp, nblocks);
/*
* Send a shared-inval message to force other backends to close any smgr
* references they may have for this rel. This is useful because they
* might have open file pointers to segments that got removed, and/or
* smgr_targblock variables pointing past the new rel end. (The inval
* message will come back to our backend, too, causing a
* probably-unnecessary local smgr flush. But we don't expect that this
* is a performance-critical path.) As in the unlink code, we want to
* be sure the message is sent before we start changing things on-disk.
*/
CacheInvalidateSmgr(reln->smgr_rnode);
/*
* Do the truncation.
*/
(*(smgrsw[reln->smgr_which].smgr_truncate)) (reln, forknum, nblocks,
isTemp);
/*
* Send a shared-inval message to force other backends to close any smgr
* references they may have for this rel. This is useful because they
* might have open file pointers to segments that got removed. (The inval
* message will come back to our backend, too, causing a
* probably-unnecessary smgr flush. But we don't expect that this is
* a performance-critical path.)
*/
CacheInvalidateSmgr(reln->smgr_rnode);
}
/*