summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorHeikki Linnakangas2009-01-20 18:59:37 +0000
committerHeikki Linnakangas2009-01-20 18:59:37 +0000
commit5403a7851823fd2686a3f29d12208414eb6dd1b0 (patch)
tree4592784c29dc8dd0d331eae766608665475253c0
parent290fd9389d980ec1d70732f12257a60c23701d5a (diff)
Add a new option to RestoreBkpBlocks() to indicate if a cleanup lock should
be used instead of the normal exclusive lock, and make WAL redo functions responsible for calling RestoreBkpBlocks(). They know better what kind of a lock they need. At the moment, this just moves things around with no functional change, but makes the hot standby patch that's under review cleaner.
-rw-r--r--src/backend/access/gin/ginxlog.c2
-rw-r--r--src/backend/access/gist/gistxlog.c3
-rw-r--r--src/backend/access/heap/heapam.c5
-rw-r--r--src/backend/access/nbtree/nbtxlog.c2
-rw-r--r--src/backend/access/transam/clog.c3
-rw-r--r--src/backend/access/transam/multixact.c3
-rw-r--r--src/backend/access/transam/xact.c3
-rw-r--r--src/backend/access/transam/xlog.c21
-rw-r--r--src/backend/access/transam/xlogutils.c43
-rw-r--r--src/backend/catalog/storage.c3
-rw-r--r--src/backend/commands/dbcommands.c3
-rw-r--r--src/backend/commands/sequence.c3
-rw-r--r--src/backend/commands/tablespace.c3
-rw-r--r--src/backend/storage/freespace/freespace.c2
-rw-r--r--src/include/access/xlog.h2
15 files changed, 80 insertions, 21 deletions
diff --git a/src/backend/access/gin/ginxlog.c b/src/backend/access/gin/ginxlog.c
index d2cff392e5..8382576d3c 100644
--- a/src/backend/access/gin/ginxlog.c
+++ b/src/backend/access/gin/ginxlog.c
@@ -438,6 +438,8 @@ gin_redo(XLogRecPtr lsn, XLogRecord *record)
{
uint8 info = record->xl_info & ~XLR_INFO_MASK;
+ RestoreBkpBlocks(lsn, record, false);
+
topCtx = MemoryContextSwitchTo(opCtx);
switch (info)
{
diff --git a/src/backend/access/gist/gistxlog.c b/src/backend/access/gist/gistxlog.c
index 116449e3c8..4a20d905d4 100644
--- a/src/backend/access/gist/gistxlog.c
+++ b/src/backend/access/gist/gistxlog.c
@@ -394,9 +394,10 @@ void
gist_redo(XLogRecPtr lsn, XLogRecord *record)
{
uint8 info = record->xl_info & ~XLR_INFO_MASK;
-
MemoryContext oldCxt;
+ RestoreBkpBlocks(lsn, record, false);
+
oldCxt = MemoryContextSwitchTo(opCtx);
switch (info)
{
diff --git a/src/backend/access/heap/heapam.c b/src/backend/access/heap/heapam.c
index 4af17c8079..52115cf64e 100644
--- a/src/backend/access/heap/heapam.c
+++ b/src/backend/access/heap/heapam.c
@@ -4777,6 +4777,8 @@ heap_redo(XLogRecPtr lsn, XLogRecord *record)
{
uint8 info = record->xl_info & ~XLR_INFO_MASK;
+ RestoreBkpBlocks(lsn, record, false);
+
switch (info & XLOG_HEAP_OPMASK)
{
case XLOG_HEAP_INSERT:
@@ -4816,12 +4818,15 @@ heap2_redo(XLogRecPtr lsn, XLogRecord *record)
switch (info & XLOG_HEAP_OPMASK)
{
case XLOG_HEAP2_FREEZE:
+ RestoreBkpBlocks(lsn, record, false);
heap_xlog_freeze(lsn, record);
break;
case XLOG_HEAP2_CLEAN:
+ RestoreBkpBlocks(lsn, record, true);
heap_xlog_clean(lsn, record, false);
break;
case XLOG_HEAP2_CLEAN_MOVE:
+ RestoreBkpBlocks(lsn, record, true);
heap_xlog_clean(lsn, record, true);
break;
default:
diff --git a/src/backend/access/nbtree/nbtxlog.c b/src/backend/access/nbtree/nbtxlog.c
index a874e491f5..517c4b90ce 100644
--- a/src/backend/access/nbtree/nbtxlog.c
+++ b/src/backend/access/nbtree/nbtxlog.c
@@ -714,6 +714,8 @@ btree_redo(XLogRecPtr lsn, XLogRecord *record)
{
uint8 info = record->xl_info & ~XLR_INFO_MASK;
+ RestoreBkpBlocks(lsn, record, false);
+
switch (info)
{
case XLOG_BTREE_INSERT_LEAF:
diff --git a/src/backend/access/transam/clog.c b/src/backend/access/transam/clog.c
index c80867c472..528a219db4 100644
--- a/src/backend/access/transam/clog.c
+++ b/src/backend/access/transam/clog.c
@@ -684,6 +684,9 @@ clog_redo(XLogRecPtr lsn, XLogRecord *record)
{
uint8 info = record->xl_info & ~XLR_INFO_MASK;
+ /* Backup blocks are not used in clog records */
+ Assert(!(record->xl_info & XLR_BKP_BLOCK_MASK));
+
if (info == CLOG_ZEROPAGE)
{
int pageno;
diff --git a/src/backend/access/transam/multixact.c b/src/backend/access/transam/multixact.c
index 78d47154c9..7314341101 100644
--- a/src/backend/access/transam/multixact.c
+++ b/src/backend/access/transam/multixact.c
@@ -1870,6 +1870,9 @@ multixact_redo(XLogRecPtr lsn, XLogRecord *record)
{
uint8 info = record->xl_info & ~XLR_INFO_MASK;
+ /* Backup blocks are not used in multixact records */
+ Assert(!(record->xl_info & XLR_BKP_BLOCK_MASK));
+
if (info == XLOG_MULTIXACT_ZERO_OFF_PAGE)
{
int pageno;
diff --git a/src/backend/access/transam/xact.c b/src/backend/access/transam/xact.c
index 23f6ea0a8b..c94e2a2251 100644
--- a/src/backend/access/transam/xact.c
+++ b/src/backend/access/transam/xact.c
@@ -4308,6 +4308,9 @@ xact_redo(XLogRecPtr lsn, XLogRecord *record)
{
uint8 info = record->xl_info & ~XLR_INFO_MASK;
+ /* Backup blocks are not used in xact records */
+ Assert(!(record->xl_info & XLR_BKP_BLOCK_MASK));
+
if (info == XLOG_XACT_COMMIT)
{
xl_xact_commit *xlrec = (xl_xact_commit *) XLogRecGetData(record);
diff --git a/src/backend/access/transam/xlog.c b/src/backend/access/transam/xlog.c
index 7debb34072..bd6035d4a6 100644
--- a/src/backend/access/transam/xlog.c
+++ b/src/backend/access/transam/xlog.c
@@ -2922,9 +2922,15 @@ CleanupBackupHistory(void)
* page might not be. This will force us to replay all subsequent
* modifications of the page that appear in XLOG, rather than possibly
* ignoring them as already applied, but that's not a huge drawback.
+ *
+ * If 'cleanup' is true, a cleanup lock is used when restoring blocks.
+ * Otherwise, a normal exclusive lock is used. At the moment, that's just
+ * pro forma, because there can't be any regular backends in the system
+ * during recovery. The 'cleanup' argument applies to all backup blocks
+ * in the WAL record, that suffices for now.
*/
-static void
-RestoreBkpBlocks(XLogRecord *record, XLogRecPtr lsn)
+void
+RestoreBkpBlocks(XLogRecPtr lsn, XLogRecord *record, bool cleanup)
{
Buffer buffer;
Page page;
@@ -2944,6 +2950,11 @@ RestoreBkpBlocks(XLogRecord *record, XLogRecPtr lsn)
buffer = XLogReadBufferExtended(bkpb.node, bkpb.fork, bkpb.block,
RBM_ZERO);
Assert(BufferIsValid(buffer));
+ if (cleanup)
+ LockBufferForCleanup(buffer);
+ else
+ LockBuffer(buffer, BUFFER_LOCK_EXCLUSIVE);
+
page = (Page) BufferGetPage(buffer);
if (bkpb.hole_length == 0)
@@ -5199,9 +5210,6 @@ StartupXLOG(void)
TransactionIdAdvance(ShmemVariableCache->nextXid);
}
- if (record->xl_info & XLR_BKP_BLOCK_MASK)
- RestoreBkpBlocks(record, EndRecPtr);
-
RmgrTable[record->xl_rmid].rm_redo(EndRecPtr, record);
/* Pop the error context stack */
@@ -6233,6 +6241,9 @@ xlog_redo(XLogRecPtr lsn, XLogRecord *record)
{
uint8 info = record->xl_info & ~XLR_INFO_MASK;
+ /* Backup blocks are not used in xlog records */
+ Assert(!(record->xl_info & XLR_BKP_BLOCK_MASK));
+
if (info == XLOG_NEXTOID)
{
Oid nextOid;
diff --git a/src/backend/access/transam/xlogutils.c b/src/backend/access/transam/xlogutils.c
index ef5fe4cef7..4a86271019 100644
--- a/src/backend/access/transam/xlogutils.c
+++ b/src/backend/access/transam/xlogutils.c
@@ -217,8 +217,19 @@ XLogCheckInvalidPages(void)
/*
* XLogReadBuffer
- * A shorthand of XLogReadBufferExtended(), for reading from the main
- * fork.
+ * Read a page during XLOG replay.
+ *
+ * This is a shorthand of XLogReadBufferExtended() followed by
+ * LockBuffer(buffer, BUFFER_LOCK_EXCLUSIVE), for reading from the main
+ * fork.
+ *
+ * (Getting the lock is not really necessary, since we expect that this is
+ * only used during single-process XLOG replay, but some subroutines such
+ * as MarkBufferDirty will complain if we don't. And hopefully we'll get
+ * hot standby support in the future, where there will be backends running
+ * read-only queries during XLOG replay.)
+ *
+ * The returned buffer is exclusively-locked.
*
* For historical reasons, instead of a ReadBufferMode argument, this only
* supports RBM_ZERO (init == true) and RBM_NORMAL (init == false) modes.
@@ -226,22 +237,21 @@ XLogCheckInvalidPages(void)
Buffer
XLogReadBuffer(RelFileNode rnode, BlockNumber blkno, bool init)
{
- return XLogReadBufferExtended(rnode, MAIN_FORKNUM, blkno,
- init ? RBM_ZERO : RBM_NORMAL);
+ Buffer buf;
+ buf = XLogReadBufferExtended(rnode, MAIN_FORKNUM, blkno,
+ init ? RBM_ZERO : RBM_NORMAL);
+ if (BufferIsValid(buf))
+ LockBuffer(buf, BUFFER_LOCK_EXCLUSIVE);
+
+ return buf;
}
/*
* XLogReadBufferExtended
* Read a page during XLOG replay
*
- * This is functionally comparable to ReadBuffer followed by
- * LockBuffer(buffer, BUFFER_LOCK_EXCLUSIVE): you get back a pinned
- * and locked buffer. (Getting the lock is not really necessary, since we
- * expect that this is only used during single-process XLOG replay, but
- * some subroutines such as MarkBufferDirty will complain if we don't.)
- *
- * There's some differences in the behavior wrt. the "mode" argument,
- * compared to ReadBufferExtended:
+ * This is functionally comparable to ReadBufferExtended. There's some
+ * differences in the behavior wrt. the "mode" argument:
*
* In RBM_NORMAL mode, if the page doesn't exist, or contains all-zeroes, we
* return InvalidBuffer. In this case the caller should silently skip the
@@ -306,16 +316,19 @@ XLogReadBufferExtended(RelFileNode rnode, ForkNumber forknum,
Assert(BufferGetBlockNumber(buffer) == blkno);
}
- LockBuffer(buffer, BUFFER_LOCK_EXCLUSIVE);
-
if (mode == RBM_NORMAL)
{
/* check that page has been initialized */
Page page = (Page) BufferGetPage(buffer);
+ /*
+ * We assume that PageIsNew is safe without a lock. During recovery,
+ * there should be no other backends that could modify the buffer at
+ * the same time.
+ */
if (PageIsNew(page))
{
- UnlockReleaseBuffer(buffer);
+ ReleaseBuffer(buffer);
log_invalid_page(rnode, forknum, blkno, true);
return InvalidBuffer;
}
diff --git a/src/backend/catalog/storage.c b/src/backend/catalog/storage.c
index e9008e3f2c..309fa469ad 100644
--- a/src/backend/catalog/storage.c
+++ b/src/backend/catalog/storage.c
@@ -401,6 +401,9 @@ smgr_redo(XLogRecPtr lsn, XLogRecord *record)
{
uint8 info = record->xl_info & ~XLR_INFO_MASK;
+ /* Backup blocks are not used in smgr records */
+ Assert(!(record->xl_info & XLR_BKP_BLOCK_MASK));
+
if (info == XLOG_SMGR_CREATE)
{
xl_smgr_create *xlrec = (xl_smgr_create *) XLogRecGetData(record);
diff --git a/src/backend/commands/dbcommands.c b/src/backend/commands/dbcommands.c
index 50cb6780d5..7e065762a8 100644
--- a/src/backend/commands/dbcommands.c
+++ b/src/backend/commands/dbcommands.c
@@ -1941,6 +1941,9 @@ dbase_redo(XLogRecPtr lsn, XLogRecord *record)
{
uint8 info = record->xl_info & ~XLR_INFO_MASK;
+ /* Backup blocks are not used in dbase records */
+ Assert(!(record->xl_info & XLR_BKP_BLOCK_MASK));
+
if (info == XLOG_DBASE_CREATE)
{
xl_dbase_create_rec *xlrec = (xl_dbase_create_rec *) XLogRecGetData(record);
diff --git a/src/backend/commands/sequence.c b/src/backend/commands/sequence.c
index c80f2a669a..46d7683377 100644
--- a/src/backend/commands/sequence.c
+++ b/src/backend/commands/sequence.c
@@ -1339,6 +1339,9 @@ seq_redo(XLogRecPtr lsn, XLogRecord *record)
xl_seq_rec *xlrec = (xl_seq_rec *) XLogRecGetData(record);
sequence_magic *sm;
+ /* Backup blocks are not used in seq records */
+ Assert(!(record->xl_info & XLR_BKP_BLOCK_MASK));
+
if (info != XLOG_SEQ_LOG)
elog(PANIC, "seq_redo: unknown op code %u", info);
diff --git a/src/backend/commands/tablespace.c b/src/backend/commands/tablespace.c
index 7fa4bfba62..ee0c6428ef 100644
--- a/src/backend/commands/tablespace.c
+++ b/src/backend/commands/tablespace.c
@@ -1276,6 +1276,9 @@ tblspc_redo(XLogRecPtr lsn, XLogRecord *record)
{
uint8 info = record->xl_info & ~XLR_INFO_MASK;
+ /* Backup blocks are not used in tblspc records */
+ Assert(!(record->xl_info & XLR_BKP_BLOCK_MASK));
+
if (info == XLOG_TBLSPC_CREATE)
{
xl_tblspc_create_rec *xlrec = (xl_tblspc_create_rec *) XLogRecGetData(record);
diff --git a/src/backend/storage/freespace/freespace.c b/src/backend/storage/freespace/freespace.c
index 8ef11be652..7fcd51e638 100644
--- a/src/backend/storage/freespace/freespace.c
+++ b/src/backend/storage/freespace/freespace.c
@@ -212,6 +212,8 @@ XLogRecordPageWithFreeSpace(RelFileNode rnode, BlockNumber heapBlk,
/* If the page doesn't exist already, extend */
buf = XLogReadBufferExtended(rnode, FSM_FORKNUM, blkno, RBM_ZERO_ON_ERROR);
+ LockBuffer(buf, BUFFER_LOCK_EXCLUSIVE);
+
page = BufferGetPage(buf);
if (PageIsNew(page))
PageInit(page, BLCKSZ, 0);
diff --git a/src/include/access/xlog.h b/src/include/access/xlog.h
index 5b7374ee66..6913f7c800 100644
--- a/src/include/access/xlog.h
+++ b/src/include/access/xlog.h
@@ -194,6 +194,8 @@ extern bool XLogNeedsFlush(XLogRecPtr RecPtr);
extern void XLogSetAsyncCommitLSN(XLogRecPtr record);
+extern void RestoreBkpBlocks(XLogRecPtr lsn, XLogRecord *record, bool cleanup);
+
extern void xlog_redo(XLogRecPtr lsn, XLogRecord *record);
extern void xlog_desc(StringInfo buf, uint8 xl_info, char *rec);