From d6c9e05cb7db64239887fac65b243229594f331d Mon Sep 17 00:00:00 2001 From: Tom Lane Date: Sat, 13 Aug 2016 22:24:48 -0400 Subject: [PATCH] Fix assorted bugs in contrib/bloom. In blinsert(), cope with the possibility that a page we pull from the notFullPage list is marked BLOOM_DELETED. This could happen if VACUUM recently marked it deleted but hasn't (yet) updated the metapage. We can re-use such a page safely, but we *must* reinitialize it so that it's no longer marked deleted. Fix blvacuum() so that it updates the notFullPage list even if it's going to update it to empty. The previous "optimization" of skipping the update seems pretty dubious, since it means that the next blinsert() will uselessly visit whatever pages we left in the list. Uniformly treat PageIsNew pages the same as deleted pages. This should allow proper recovery if a crash occurs just after relation extension. Properly use vacuum_delay_point, not assorted ad-hoc CHECK_FOR_INTERRUPTS calls, in the blvacuum() main loop. Fix broken tuple-counting logic: blvacuum.c counted the number of live index tuples over again in each scan, leading to VACUUM VERBOSE reporting some multiple of the actual number of surviving index tuples after any vacuum that removed any tuples (since they'd be counted in blvacuum, maybe more than once, and then again in blvacuumcleanup, without ever zeroing the counter). It's sufficient to count them in blvacuumcleanup. stats->estimated_count is a boolean, not a counter, and we don't want to set it true, so don't add tuple counts to it. Add a couple of Asserts that we don't overrun available space on a bloom page. I don't think there's any bug there today, but the way the FreeBlockNumberArray size calculation is set up is scarily fragile, and BloomPageGetFreeSpace isn't much better. The Asserts should help catch any future mistakes. Per investigation of a report from Jeff Janes. I think the first item above may explain his report; the other changes were things I noticed while casting about for an explanation. Report: --- contrib/bloom/blinsert.c | 11 ++++++ contrib/bloom/blscan.c | 2 +- contrib/bloom/blutils.c | 13 +++++-- contrib/bloom/blvacuum.c | 75 ++++++++++++++++++++-------------------- 4 files changed, 60 insertions(+), 41 deletions(-) diff --git a/contrib/bloom/blinsert.c b/contrib/bloom/blinsert.c index 3953af996b..78eec5c67e 100644 --- a/contrib/bloom/blinsert.c +++ b/contrib/bloom/blinsert.c @@ -237,6 +237,13 @@ blinsert(Relation index, Datum *values, bool *isnull, state = GenericXLogStart(index); page = GenericXLogRegisterBuffer(state, buffer, 0); + /* + * We might have found a page that was recently deleted by VACUUM. If + * so, we can reuse it, but we must reinitialize it. + */ + if (PageIsNew(page) || BloomPageIsDeleted(page)) + BloomInitPage(page, 0); + if (BloomPageAddItem(&blstate, page, itup)) { /* Success! Apply the change, clean up, and exit */ @@ -295,6 +302,10 @@ blinsert(Relation index, Datum *values, bool *isnull, LockBuffer(buffer, BUFFER_LOCK_EXCLUSIVE); page = GenericXLogRegisterBuffer(state, buffer, 0); + /* Basically same logic as above */ + if (PageIsNew(page) || BloomPageIsDeleted(page)) + BloomInitPage(page, 0); + if (BloomPageAddItem(&blstate, page, itup)) { /* Success! Apply the changes, clean up, and exit */ diff --git a/contrib/bloom/blscan.c b/contrib/bloom/blscan.c index 0c954dc8d5..316a906734 100644 --- a/contrib/bloom/blscan.c +++ b/contrib/bloom/blscan.c @@ -135,7 +135,7 @@ blgetbitmap(IndexScanDesc scan, TIDBitmap *tbm) page = BufferGetPage(buffer); TestForOldSnapshot(scan->xs_snapshot, scan->indexRelation, page); - if (!BloomPageIsDeleted(page)) + if (!PageIsNew(page) && !BloomPageIsDeleted(page)) { OffsetNumber offset, maxOffset = BloomPageGetMaxOffset(page); diff --git a/contrib/bloom/blutils.c b/contrib/bloom/blutils.c index b5007ed6bb..debf4f46eb 100644 --- a/contrib/bloom/blutils.c +++ b/contrib/bloom/blutils.c @@ -299,7 +299,7 @@ BloomFormTuple(BloomState *state, ItemPointer iptr, Datum *values, bool *isnull) /* * Add new bloom tuple to the page. Returns true if new tuple was successfully - * added to the page. Returns false if it doesn't fit the page. + * added to the page. Returns false if it doesn't fit on the page. */ bool BloomPageAddItem(BloomState *state, Page page, BloomTuple *tuple) @@ -308,7 +308,10 @@ BloomPageAddItem(BloomState *state, Page page, BloomTuple *tuple) BloomPageOpaque opaque; Pointer ptr; - /* Does new tuple fit the page */ + /* We shouldn't be pointed to an invalid page */ + Assert(!PageIsNew(page) && !BloomPageIsDeleted(page)); + + /* Does new tuple fit on the page? */ if (BloomPageGetFreeSpace(state, page) < state->sizeOfBloomTuple) return false; @@ -322,6 +325,9 @@ BloomPageAddItem(BloomState *state, Page page, BloomTuple *tuple) ptr = (Pointer) BloomPageGetTuple(state, page, opaque->maxoff + 1); ((PageHeader) page)->pd_lower = ptr - page; + /* Assert we didn't overrun available space */ + Assert(((PageHeader) page)->pd_lower <= ((PageHeader) page)->pd_upper); + return true; } @@ -424,6 +430,9 @@ BloomFillMetapage(Relation index, Page metaPage) metadata->magickNumber = BLOOM_MAGICK_NUMBER; metadata->opts = *opts; ((PageHeader) metaPage)->pd_lower += sizeof(BloomMetaPageData); + + /* If this fails, probably FreeBlockNumberArray size calc is wrong: */ + Assert(((PageHeader) metaPage)->pd_lower <= ((PageHeader) metaPage)->pd_upper); } /* diff --git a/contrib/bloom/blvacuum.c b/contrib/bloom/blvacuum.c index 7c355463a2..482242f1c2 100644 --- a/contrib/bloom/blvacuum.c +++ b/contrib/bloom/blvacuum.c @@ -13,6 +13,7 @@ #include "postgres.h" #include "access/genam.h" +#include "bloom.h" #include "catalog/storage.h" #include "commands/vacuum.h" #include "miscadmin.h" @@ -21,7 +22,6 @@ #include "storage/indexfsm.h" #include "storage/lmgr.h" -#include "bloom.h" /* * Bulk deletion of all index entries pointing to a set of heap tuples. @@ -42,6 +42,7 @@ blbulkdelete(IndexVacuumInfo *info, IndexBulkDeleteResult *stats, BloomState state; Buffer buffer; Page page; + BloomMetaPageData *metaData; GenericXLogState *gxlogState; if (stats == NULL) @@ -60,6 +61,8 @@ blbulkdelete(IndexVacuumInfo *info, IndexBulkDeleteResult *stats, *itupPtr, *itupEnd; + vacuum_delay_point(); + buffer = ReadBufferExtended(index, MAIN_FORKNUM, blkno, RBM_NORMAL, info->strategy); @@ -67,15 +70,18 @@ blbulkdelete(IndexVacuumInfo *info, IndexBulkDeleteResult *stats, gxlogState = GenericXLogStart(index); page = GenericXLogRegisterBuffer(gxlogState, buffer, 0); - if (BloomPageIsDeleted(page)) + /* Ignore empty/deleted pages until blvacuumcleanup() */ + if (PageIsNew(page) || BloomPageIsDeleted(page)) { UnlockReleaseBuffer(buffer); GenericXLogAbort(gxlogState); - CHECK_FOR_INTERRUPTS(); continue; } - /* Iterate over the tuples */ + /* + * Iterate over the tuples. itup points to current tuple being + * scanned, itupPtr points to where to save next non-deleted tuple. + */ itup = itupPtr = BloomPageGetTuple(&state, page, FirstOffsetNumber); itupEnd = BloomPageGetTuple(&state, page, OffsetNumberNext(BloomPageGetMaxOffset(page))); @@ -84,36 +90,32 @@ blbulkdelete(IndexVacuumInfo *info, IndexBulkDeleteResult *stats, /* Do we have to delete this tuple? */ if (callback(&itup->heapPtr, callback_state)) { - stats->tuples_removed += 1; + /* Yes; adjust count of tuples that will be left on page */ BloomPageGetOpaque(page)->maxoff--; + stats->tuples_removed += 1; } else { + /* No; copy it to itupPtr++, but skip copy if not needed */ if (itupPtr != itup) - { - /* - * If we already delete something before, we have to move - * this tuple backward. - */ memmove((Pointer) itupPtr, (Pointer) itup, state.sizeOfBloomTuple); - } - stats->num_index_tuples++; itupPtr = BloomPageGetNextTuple(&state, itupPtr); } itup = BloomPageGetNextTuple(&state, itup); } + /* Assert that we counted correctly */ Assert(itupPtr == BloomPageGetTuple(&state, page, OffsetNumberNext(BloomPageGetMaxOffset(page)))); /* - * Add page to notFullPage list if we will not mark page as deleted - * and there is a free space on it + * Add page to new notFullPage list if we will not mark page as + * deleted and there is free space on it */ if (BloomPageGetMaxOffset(page) != 0 && - BloomPageGetFreeSpace(&state, page) > state.sizeOfBloomTuple && + BloomPageGetFreeSpace(&state, page) >= state.sizeOfBloomTuple && countPage < BloomMetaBlockN) notFullPage[countPage++] = blkno; @@ -134,27 +136,26 @@ blbulkdelete(IndexVacuumInfo *info, IndexBulkDeleteResult *stats, GenericXLogAbort(gxlogState); } UnlockReleaseBuffer(buffer); - CHECK_FOR_INTERRUPTS(); } - if (countPage > 0) - { - BloomMetaPageData *metaData; - - buffer = ReadBuffer(index, BLOOM_METAPAGE_BLKNO); - LockBuffer(buffer, BUFFER_LOCK_EXCLUSIVE); + /* + * Update the metapage's notFullPage list with whatever we found. Our + * info could already be out of date at this point, but blinsert() will + * cope if so. + */ + buffer = ReadBuffer(index, BLOOM_METAPAGE_BLKNO); + LockBuffer(buffer, BUFFER_LOCK_EXCLUSIVE); - gxlogState = GenericXLogStart(index); - page = GenericXLogRegisterBuffer(gxlogState, buffer, 0); + gxlogState = GenericXLogStart(index); + page = GenericXLogRegisterBuffer(gxlogState, buffer, 0); - metaData = BloomPageGetMeta(page); - memcpy(metaData->notFullPage, notFullPage, sizeof(BlockNumber) * countPage); - metaData->nStart = 0; - metaData->nEnd = countPage; + metaData = BloomPageGetMeta(page); + memcpy(metaData->notFullPage, notFullPage, sizeof(BlockNumber) * countPage); + metaData->nStart = 0; + metaData->nEnd = countPage; - GenericXLogFinish(gxlogState); - UnlockReleaseBuffer(buffer); - } + GenericXLogFinish(gxlogState); + UnlockReleaseBuffer(buffer); return stats; } @@ -170,7 +171,6 @@ blvacuumcleanup(IndexVacuumInfo *info, IndexBulkDeleteResult *stats) Relation index = info->index; BlockNumber npages, blkno; - BlockNumber totFreePages; if (info->analyze_only) return stats; @@ -183,7 +183,9 @@ blvacuumcleanup(IndexVacuumInfo *info, IndexBulkDeleteResult *stats) * statistics. */ npages = RelationGetNumberOfBlocks(index); - totFreePages = 0; + stats->num_pages = npages; + stats->pages_free = 0; + stats->num_index_tuples = 0; for (blkno = BLOOM_HEAD_BLKNO; blkno < npages; blkno++) { Buffer buffer; @@ -196,23 +198,20 @@ blvacuumcleanup(IndexVacuumInfo *info, IndexBulkDeleteResult *stats) LockBuffer(buffer, BUFFER_LOCK_SHARE); page = (Page) BufferGetPage(buffer); - if (BloomPageIsDeleted(page)) + if (PageIsNew(page) || BloomPageIsDeleted(page)) { RecordFreeIndexPage(index, blkno); - totFreePages++; + stats->pages_free++; } else { stats->num_index_tuples += BloomPageGetMaxOffset(page); - stats->estimated_count += BloomPageGetMaxOffset(page); } UnlockReleaseBuffer(buffer); } IndexFreeSpaceMapVacuum(info->index); - stats->pages_free = totFreePages; - stats->num_pages = RelationGetNumberOfBlocks(index); return stats; } -- 2.40.0