When GIN vacuum deletes a posting tree page, it assumes that no concurrent
searchers can access it, thanks to ginStepRight() locking two pages at once.
However, since 9.4 searches can skip parts of posting trees descending from the
root. That leads to the risk that page is deleted and reclaimed before
concurrent search can access it.
This commit prevents the risk of above by waiting for every transaction, which
might wait to reference this page, to finish. Due to binary compatibility
we can't change GinPageOpaqueData to store corresponding transaction id.
Instead we reuse page header pd_prune_xid field, which is unused in index pages.
Discussion: https://postgr.es/m/
31a702a.14dd.
166c1366ac1.Coremail.chjischj%40163.com
Author: Andrey Borodin, Alexander Korotkov
Reviewed-by: Alexander Korotkov
Backpatch-through: 9.4
The downlink is more tricky. A search descending the tree must release the
lock on the parent page before locking the child, or it could deadlock with
a concurrent split of the child page; a page split locks the parent, while
-already holding a lock on the child page. However, posting trees are only
-fully searched from left to right, starting from the leftmost leaf. (The
-tree-structure is only needed by insertions, to quickly find the correct
-insert location). So as long as we don't delete the leftmost page on each
-level, a search can never follow a downlink to page that's about to be
-deleted.
+already holding a lock on the child page. So, deleted page cannot be reclaimed
+immediately. Instead, we have to wait for every transaction, which might wait
+to reference this page, to finish. Corresponding processes must observe that
+the page is marked deleted and recover accordingly.
The previous paragraph's reasoning only applies to searches, and only to
posting trees. To protect from inserters following a downlink to a deleted
*/
if (ConditionalLockBuffer(buffer))
{
- Page page = BufferGetPage(buffer);
-
- if (PageIsNew(page))
- return buffer; /* OK to use, if never initialized */
-
- if (GinPageIsDeleted(page))
+ if (GinPageIsRecyclable(BufferGetPage(buffer)))
return buffer; /* OK to use */
LockBuffer(buffer, GIN_UNLOCK);
page = BufferGetPage(dBuffer);
rightlink = GinPageGetOpaque(page)->rightlink;
+ /* For deleted page remember last xid which could knew its address */
+ GinPageSetDeleteXid(page, ReadNewTransactionId());
+
/*
* Any insert which would have gone on the leaf block will now go to its
* right sibling.
data.parentOffset = myoff;
data.rightLink = GinPageGetOpaque(page)->rightlink;
+ data.deleteXid = GinPageGetDeleteXid(page);
XLogRegisterData((char *) &data, sizeof(ginxlogDeletePage));
LockBuffer(buffer, GIN_SHARE);
page = (Page) BufferGetPage(buffer);
- if (PageIsNew(page) || GinPageIsDeleted(page))
+ if (GinPageIsRecyclable(page))
{
Assert(blkno != GIN_ROOT_BLKNO);
RecordFreeIndexPage(index, blkno);
page = BufferGetPage(dbuffer);
Assert(GinPageIsData(page));
GinPageGetOpaque(page)->flags = GIN_DELETED;
+ GinPageSetDeleteXid(page, data->deleteXid);
PageSetLSN(page, lsn);
MarkBufferDirty(dbuffer);
}
#ifndef GINBLOCK_H
#define GINBLOCK_H
+#include "access/transam.h"
#include "storage/block.h"
#include "storage/itemptr.h"
#include "storage/off.h"
#define GinPageRightMost(page) ( GinPageGetOpaque(page)->rightlink == InvalidBlockNumber)
+/*
+ * We should reclaim deleted page only once every transaction started before
+ * its deletion is over.
+ */
+#define GinPageGetDeleteXid(page) ( ((PageHeader) (page))->pd_prune_xid )
+#define GinPageSetDeleteXid(page, xid) ( ((PageHeader) (page))->pd_prune_xid = xid)
+#define GinPageIsRecyclable(page) ( PageIsNew(page) || (GinPageIsDeleted(page) \
+ && TransactionIdPrecedes(GinPageGetDeleteXid(page), RecentGlobalXmin)))
+
/*
* We use our own ItemPointerGet(BlockNumber|OffsetNumber)
* to avoid Asserts, since sometimes the ip_posid isn't "valid"
{
OffsetNumber parentOffset;
BlockNumber rightLink;
+ TransactionId deleteXid; /* last Xid which could see this page in scan */
} ginxlogDeletePage;
#define XLOG_GIN_UPDATE_META_PAGE 0x60