Skip to content

P
Postgres FD Implementation
Commit cf23b75b authored by Teodor Sigaev's avatar Teodor Sigaev
Browse files
Options

Fix using too many LWLocks bug, reported by Craig Ringer 

<craig@postnewspapers.com.au>.
It was my mistake, I missed limitation of number of held locks, now GIN doesn't
use continiuous locks, but still hold buffers pinned to prevent interference
with vacuum's deletion algorithm.

Backpatch is needed.
parent fc507eb0
Show whitespace changes
Inline Side-by-side
Showing with 165 additions and 211 deletions
src/backend/access/gin/ginget.c
View file @ cf23b75b
......@@ -8,7 +8,7 @@
* Portions Copyright (c) 1994, Regents of the University of California
*
* IDENTIFICATION
* $PostgreSQL: pgsql/src/backend/access/gin/ginget.c,v 1.13 2008/04/14 17:05:33 tgl Exp $
* $PostgreSQL: pgsql/src/backend/access/gin/ginget.c,v 1.14 2008/04/22 17:52:43 teodor Exp $
*-------------------------------------------------------------------------
*/
......@@ -29,58 +29,16 @@ findItemInPage(Page page, ItemPointer item, OffsetNumber *off)
/* page was deleted by concurrent vacuum */
return false;
if (*off > maxoff || *off == InvalidOffsetNumber)
res = -1;
else
res = compareItemPointers(item, (ItemPointer) GinDataPageGetItem(page, *off));
if (res == 0)
{
/* page isn't changed */
return true;
}
else if (res > 0)
{
/*
* some items was added before our position, look further to find it
* or first greater
*/
(*off)++;
for (; *off <= maxoff; (*off)++)
{
res = compareItemPointers(item, (ItemPointer) GinDataPageGetItem(page, *off));
if (res == 0)
return true;
if (res < 0)
{
(*off)--;
return true;
}
}
}
else
{
/*
* some items was deleted before our position, look from begining to
* find it or first greater
* scan page to find equal or first greater value
*/
for (*off = FirstOffsetNumber; *off <= maxoff; (*off)++)
{
res = compareItemPointers(item, (ItemPointer) GinDataPageGetItem(page, *off));
if (res == 0)
if (res <= 0)
return true;
if (res < 0)
{
(*off)--;
return true;
}
}
}
return false;
......@@ -91,24 +49,23 @@ findItemInPage(Page page, ItemPointer item, OffsetNumber *off)
* Stop* functions unlock buffer (but don't release!)
*/
static void
startScanEntry(Relation index, GinState *ginstate, GinScanEntry entry, bool firstCall)
startScanEntry(Relation index, GinState *ginstate, GinScanEntry entry)
{
GinBtreeData btreeEntry;
GinBtreeStack *stackEntry;
Page page;
bool needUnlock = TRUE;
if (entry->master != NULL)
{
entry->isFinished = entry->master->isFinished;
return;
}
if (firstCall)
{
/*
* at first call we should find entry, and begin scan of posting tree
* we should find entry, and begin scan of posting tree
* or just store posting list in memory
*/
GinBtreeData btreeEntry;
GinBtreeStack *stackEntry;
Page page;
bool needUnlock = TRUE;
prepareEntryScan(&btreeEntry, index, entry->entry, ginstate);
btreeEntry.searchMode = TRUE;
......@@ -138,11 +95,25 @@ startScanEntry(Relation index, GinState *ginstate, GinScanEntry entry, bool firs
gdi = prepareScanPostingTree(index, rootPostingTree, TRUE);
entry->buffer = scanBeginPostingTree(gdi);
/*
* We keep buffer pinned because we need to prevent deletition
* page during scan. See GIN's vacuum implementation. RefCount
* is increased to keep buffer pinned after freeGinBtreeStack() call.
*/
IncrBufferRefCount(entry->buffer);
page = BufferGetPage(entry->buffer);
entry->predictNumberResult = gdi->stack->predictNumber * GinPageGetOpaque(page)->maxoff;
/*
* Keep page content in memory to prevent durable page locking
*/
entry->list = (ItemPointerData *) palloc( BLCKSZ );
entry->nlist = GinPageGetOpaque(page)->maxoff;
memcpy( entry->list, GinDataPageGetItem(page, FirstOffsetNumber),
GinPageGetOpaque(page)->maxoff * sizeof(ItemPointerData) );
LockBuffer(entry->buffer, GIN_UNLOCK);
freeGinBtreeStack(gdi->stack);
pfree(gdi);
entry->isFinished = FALSE;
......@@ -159,51 +130,6 @@ startScanEntry(Relation index, GinState *ginstate, GinScanEntry entry, bool firs
if (needUnlock)
LockBuffer(stackEntry->buffer, GIN_UNLOCK);
freeGinBtreeStack(stackEntry);
}
else if (entry->buffer != InvalidBuffer)
{
/* we should find place where we was stopped */
BlockNumber blkno;
Page page;
LockBuffer(entry->buffer, GIN_SHARE);
if (!ItemPointerIsValid(&entry->curItem))
/* start position */
return;
Assert(entry->offset != InvalidOffsetNumber);
page = BufferGetPage(entry->buffer);
/* try to find curItem in current buffer */
if (findItemInPage(page, &entry->curItem, &entry->offset))
return;
/* walk to right */
while ((blkno = GinPageGetOpaque(page)->rightlink) != InvalidBlockNumber)
{
LockBuffer(entry->buffer, GIN_UNLOCK);
entry->buffer = ReleaseAndReadBuffer(entry->buffer, index, blkno);
LockBuffer(entry->buffer, GIN_SHARE);
page = BufferGetPage(entry->buffer);
entry->offset = InvalidOffsetNumber;
if (findItemInPage(page, &entry->curItem, &entry->offset))
return;
}
/*
* curItem and any greated items was deleted by concurrent vacuum, so
* we finished scan with currrent entry
*/
}
}
static void
stopScanEntry(GinScanEntry entry)
{
if (entry->buffer != InvalidBuffer)
LockBuffer(entry->buffer, GIN_UNLOCK);
}
static void
......@@ -211,11 +137,12 @@ startScanKey(Relation index, GinState *ginstate, GinScanKey key)
{
uint32 i;
if (!key->firstCall)
return;
for (i = 0; i < key->nentries; i++)
startScanEntry(index, ginstate, key->scanEntry + i, key->firstCall);
startScanEntry(index, ginstate, key->scanEntry + i);
if (key->firstCall)
{
memset(key->entryRes, TRUE, sizeof(bool) * key->nentries);
key->isFinished = FALSE;
key->firstCall = FALSE;
......@@ -240,16 +167,6 @@ startScanKey(Relation index, GinState *ginstate, GinScanKey key)
key->scanEntry[i].reduceResult = TRUE;
}
}
}
}
static void
stopScanKey(GinScanKey key)
{
uint32 i;
for (i = 0; i < key->nentries; i++)
stopScanEntry(key->scanEntry + i);
}
static void
......@@ -262,44 +179,82 @@ startScan(IndexScanDesc scan)
startScanKey(scan->indexRelation, &so->ginstate, so->keys + i);
}
static void
stopScan(IndexScanDesc scan)
{
uint32 i;
GinScanOpaque so = (GinScanOpaque) scan->opaque;
for (i = 0; i < so->nkeys; i++)
stopScanKey(so->keys + i);
}
/*
* Gets next ItemPointer from PostingTree. Note, that we copy
* page into GinScanEntry->list array and unlock page, but keep it pinned
* to prevent interference with vacuum
*/
static void
entryGetNextItem(Relation index, GinScanEntry entry)
{
Page page = BufferGetPage(entry->buffer);
Page page;
BlockNumber blkno;
for(;;)
{
entry->offset++;
if (entry->offset <= GinPageGetOpaque(page)->maxoff && GinPageGetOpaque(page)->maxoff >= FirstOffsetNumber)
if (entry->offset <= entry->nlist)
{
entry->curItem = *(ItemPointerData *) GinDataPageGetItem(page, entry->offset);
entry->curItem = entry->list[entry->offset - 1];
return;
}
else
LockBuffer(entry->buffer, GIN_SHARE);
page = BufferGetPage(entry->buffer);
for(;;)
{
BlockNumber blkno = GinPageGetOpaque(page)->rightlink;
/*
* It's needed to go by right link. During that we should refind
* first ItemPointer greater that stored
*/
blkno = GinPageGetOpaque(page)->rightlink;
LockBuffer(entry->buffer, GIN_UNLOCK);
if (blkno == InvalidBlockNumber)
{
ReleaseBuffer(entry->buffer);
ItemPointerSet(&entry->curItem, InvalidBlockNumber, InvalidOffsetNumber);
entry->buffer = InvalidBuffer;
entry->isFinished = TRUE;
return;
}
else
{
entry->buffer = ReleaseAndReadBuffer(entry->buffer, index, blkno);
LockBuffer(entry->buffer, GIN_SHARE);
page = BufferGetPage(entry->buffer);
entry->offset = InvalidOffsetNumber;
entryGetNextItem(index, entry);
if (!ItemPointerIsValid(&entry->curItem) || findItemInPage(page, &entry->curItem, &entry->offset))
{
/*
* Found position equal to or greater than stored
*/
entry->nlist = GinPageGetOpaque(page)->maxoff;
memcpy( entry->list, GinDataPageGetItem(page, FirstOffsetNumber),
GinPageGetOpaque(page)->maxoff * sizeof(ItemPointerData) );
LockBuffer(entry->buffer, GIN_UNLOCK);
if ( !ItemPointerIsValid(&entry->curItem) ||
compareItemPointers( &entry->curItem, entry->list + entry->offset - 1 ) == 0 )
{
/*
* First pages are deleted or empty, or we found exact position,
* so break inner loop and continue outer one.
*/
break;
}
/*
* Find greater than entry->curItem position, store it.
*/
entry->curItem = entry->list[entry->offset - 1];
return;
}
}
}
}
......@@ -319,7 +274,7 @@ entryGetItem(Relation index, GinScanEntry entry)
entry->isFinished = entry->master->isFinished;
entry->curItem = entry->master->curItem;
}
else if (entry->list)
else if (!BufferIsValid(entry->buffer))
{
entry->offset++;
if (entry->offset <= entry->nlist)
......@@ -527,8 +482,6 @@ gingetbitmap(PG_FUNCTION_ARGS)
ntids++;
}
stopScan(scan);
PG_RETURN_INT64(ntids);
}
......@@ -550,7 +503,6 @@ gingettuple(PG_FUNCTION_ARGS)
startScan(scan);
res = scanGetItem(scan, &scan->xs_ctup.t_self, &scan->xs_recheck);
stopScan(scan);
PG_RETURN_BOOL(res);
}
src/include/access/gin.h
View file @ cf23b75b
......@@ -4,7 +4,7 @@
*
* Copyright (c) 2006-2008, PostgreSQL Global Development Group
*
* $PostgreSQL: pgsql/src/include/access/gin.h,v 1.17 2008/04/10 22:25:25 tgl Exp $
* $PostgreSQL: pgsql/src/include/access/gin.h,v 1.18 2008/04/22 17:52:43 teodor Exp $
*--------------------------------------------------------------------------
*/
......@@ -356,14 +356,16 @@ typedef struct GinScanEntryData
/* entry, got from extractQueryFn */
Datum entry;
/* current ItemPointer to heap, its offset in buffer and buffer */
ItemPointerData curItem;
OffsetNumber offset;
/* Current page in posting tree */
Buffer buffer;
/* in case of Posing list */
/* current ItemPointer to heap */
ItemPointerData curItem;
/* used for Posting list and one page in Posting tree */
ItemPointerData *list;
uint32 nlist;
OffsetNumber offset;
bool isFinished;
bool reduceResult;
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment