static void
pushStackIfSplited(Page page, GistBDItem *stack)
{
GISTPageOpaque opaque = GistPageGetOpaque(page);
if (stack->blkno != GIST_ROOT_BLKNO && !XLogRecPtrIsInvalid(stack->parentlsn) &&
(GistFollowRight(page) || stack->parentlsn < opaque->nsn) &&
opaque->rightlink != InvalidBlockNumber /* sanity check */ )
{
/* split page detected, install right link to the stack */
GistBDItem *ptr = (GistBDItem *) palloc(sizeof(GistBDItem));
ptr->blkno = opaque->rightlink;
ptr->parentlsn = stack->parentlsn;
ptr->next = stack->next;
stack->next = ptr;
}
}
/*
* Scan all items on the GiST index page identified by *pageItem, and insert
* them into the queue (or directly to output areas)
*
* scan: index scan we are executing
* pageItem: search queue item identifying an index page to scan
* myDistances: distances array associated with pageItem, or NULL at the root
* tbm: if not NULL, gistgetbitmap's output bitmap
* ntids: if not NULL, gistgetbitmap's output tuple counter
*
* If tbm/ntids aren't NULL, we are doing an amgetbitmap scan, and heap
* tuples should be reported directly into the bitmap. If they are NULL,
* we're doing a plain or ordered indexscan. For a plain indexscan, heap
* tuple TIDs are returned into so->pageData[]. For an ordered indexscan,
* heap tuple TIDs are pushed into individual search queue items.
*
* If we detect that the index page has split since we saw its downlink
* in the parent, we push its new right sibling onto the queue so the
* sibling will be processed next.
*/
static void
gistScanPage(IndexScanDesc scan, GISTSearchItem *pageItem, double *myDistances,
TIDBitmap *tbm, int64 *ntids)
{
GISTScanOpaque so = (GISTScanOpaque) scan->opaque;
Buffer buffer;
Page page;
GISTPageOpaque opaque;
OffsetNumber maxoff;
OffsetNumber i;
GISTSearchTreeItem *tmpItem = so->tmpTreeItem;
bool isNew;
MemoryContext oldcxt;
Assert(!GISTSearchItemIsHeap(*pageItem));
buffer = ReadBuffer(scan->indexRelation, pageItem->blkno);
LockBuffer(buffer, GIST_SHARE);
gistcheckpage(scan->indexRelation, buffer);
page = BufferGetPage(buffer);
opaque = GistPageGetOpaque(page);
/*
* Check if we need to follow the rightlink. We need to follow it if the
* page was concurrently split since we visited the parent (in which case
* parentlsn < nsn), or if the system crashed after a page split but
* before the downlink was inserted into the parent.
*/
if (!XLogRecPtrIsInvalid(pageItem->data.parentlsn) &&
(GistFollowRight(page) ||
pageItem->data.parentlsn < GistPageGetNSN(page)) &&
opaque->rightlink != InvalidBlockNumber /* sanity check */ )
{
/* There was a page split, follow right link to add pages */
GISTSearchItem *item;
/* This can't happen when starting at the root */
Assert(myDistances != NULL);
oldcxt = MemoryContextSwitchTo(so->queueCxt);
/* Create new GISTSearchItem for the right sibling index page */
item = palloc(sizeof(GISTSearchItem));
item->next = NULL;
item->blkno = opaque->rightlink;
item->data.parentlsn = pageItem->data.parentlsn;
/* Insert it into the queue using same distances as for this page */
tmpItem->head = item;
tmpItem->lastHeap = NULL;
memcpy(tmpItem->distances, myDistances,
sizeof(double) * scan->numberOfOrderBys);
(void) rb_insert(so->queue, (RBNode *) tmpItem, &isNew);
MemoryContextSwitchTo(oldcxt);
}
so->nPageData = so->curPageData = 0;
/*
* check all tuples on page
*/
maxoff = PageGetMaxOffsetNumber(page);
for (i = FirstOffsetNumber; i <= maxoff; i = OffsetNumberNext(i))
{
IndexTuple it = (IndexTuple) PageGetItem(page, PageGetItemId(page, i));
bool match;
bool recheck;
/*
* Must call gistindex_keytest in tempCxt, and clean up any leftover
* junk afterward.
*/
oldcxt = MemoryContextSwitchTo(so->giststate->tempCxt);
match = gistindex_keytest(scan, it, page, i, &recheck);
MemoryContextSwitchTo(oldcxt);
MemoryContextReset(so->giststate->tempCxt);
/* Ignore tuple if it doesn't match */
if (!match)
continue;
if (tbm && GistPageIsLeaf(page))
{
/*
* getbitmap scan, so just push heap tuple TIDs into the bitmap
* without worrying about ordering
*/
tbm_add_tuples(tbm, &it->t_tid, 1, recheck);
(*ntids)++;
}
else if (scan->numberOfOrderBys == 0 && GistPageIsLeaf(page))
{
/*
* Non-ordered scan, so report heap tuples in so->pageData[]
*/
so->pageData[so->nPageData].heapPtr = it->t_tid;
so->pageData[so->nPageData].recheck = recheck;
so->nPageData++;
}
else
{
/*
* Must push item into search queue. We get here for any lower
* index page, and also for heap tuples if doing an ordered
* search.
*/
GISTSearchItem *item;
oldcxt = MemoryContextSwitchTo(so->queueCxt);
/* Create new GISTSearchItem for this item */
item = palloc(sizeof(GISTSearchItem));
item->next = NULL;
if (GistPageIsLeaf(page))
{
/* Creating heap-tuple GISTSearchItem */
item->blkno = InvalidBlockNumber;
item->data.heap.heapPtr = it->t_tid;
item->data.heap.recheck = recheck;
}
else
{
/* Creating index-page GISTSearchItem */
item->blkno = ItemPointerGetBlockNumber(&it->t_tid);
/*
* LSN of current page is lsn of parent page for child. We
* only have a shared lock, so we need to get the LSN
* atomically.
*/
item->data.parentlsn = BufferGetLSNAtomic(buffer);
}
/* Insert it into the queue using new distance data */
tmpItem->head = item;
tmpItem->lastHeap = GISTSearchItemIsHeap(*item) ? item : NULL;
memcpy(tmpItem->distances, so->distances,
sizeof(double) * scan->numberOfOrderBys);
(void) rb_insert(so->queue, (RBNode *) tmpItem, &isNew);
MemoryContextSwitchTo(oldcxt);
}
}
UnlockReleaseBuffer(buffer);
}
/*
* Workhouse routine for doing insertion into a GiST index. Note that
* this routine assumes it is invoked in a short-lived memory context,
* so it does not bother releasing palloc'd allocations.
*/
void
gistdoinsert(Relation r, IndexTuple itup, Size freespace, GISTSTATE *giststate)
{
ItemId iid;
IndexTuple idxtuple;
GISTInsertStack firststack;
GISTInsertStack *stack;
GISTInsertState state;
bool xlocked = false;
memset(&state, 0, sizeof(GISTInsertState));
state.freespace = freespace;
state.r = r;
/* Start from the root */
firststack.blkno = GIST_ROOT_BLKNO;
firststack.lsn = 0;
firststack.parent = NULL;
firststack.downlinkoffnum = InvalidOffsetNumber;
state.stack = stack = &firststack;
/*
* Walk down along the path of smallest penalty, updating the parent
* pointers with the key we're inserting as we go. If we crash in the
* middle, the tree is consistent, although the possible parent updates
* were a waste.
*/
for (;;)
{
if (XLogRecPtrIsInvalid(stack->lsn))
stack->buffer = ReadBuffer(state.r, stack->blkno);
/*
* Be optimistic and grab shared lock first. Swap it for an exclusive
* lock later if we need to update the page.
*/
if (!xlocked)
{
LockBuffer(stack->buffer, GIST_SHARE);
gistcheckpage(state.r, stack->buffer);
}
stack->page = (Page) BufferGetPage(stack->buffer);
stack->lsn = PageGetLSN(stack->page);
Assert(!RelationNeedsWAL(state.r) || !XLogRecPtrIsInvalid(stack->lsn));
/*
* If this page was split but the downlink was never inserted to the
* parent because the inserting backend crashed before doing that, fix
* that now.
*/
if (GistFollowRight(stack->page))
{
if (!xlocked)
{
LockBuffer(stack->buffer, GIST_UNLOCK);
LockBuffer(stack->buffer, GIST_EXCLUSIVE);
xlocked = true;
/* someone might've completed the split when we unlocked */
if (!GistFollowRight(stack->page))
continue;
}
gistfixsplit(&state, giststate);
UnlockReleaseBuffer(stack->buffer);
xlocked = false;
state.stack = stack = stack->parent;
continue;
}
if (stack->blkno != GIST_ROOT_BLKNO &&
stack->parent->lsn < GistPageGetNSN(stack->page))
{
/*
* Concurrent split detected. There's no guarantee that the
* downlink for this page is consistent with the tuple we're
* inserting anymore, so go back to parent and rechoose the best
* child.
*/
UnlockReleaseBuffer(stack->buffer);
xlocked = false;
state.stack = stack = stack->parent;
continue;
}
if (!GistPageIsLeaf(stack->page))
{
/*
* This is an internal page so continue to walk down the tree.
* Find the child node that has the minimum insertion penalty.
*/
BlockNumber childblkno;
IndexTuple newtup;
GISTInsertStack *item;
OffsetNumber downlinkoffnum;
downlinkoffnum = gistchoose(state.r, stack->page, itup, giststate);
iid = PageGetItemId(stack->page, downlinkoffnum);
idxtuple = (IndexTuple) PageGetItem(stack->page, iid);
childblkno = ItemPointerGetBlockNumber(&(idxtuple->t_tid));
/*
* Check that it's not a leftover invalid tuple from pre-9.1
*/
if (GistTupleIsInvalid(idxtuple))
ereport(ERROR,
(errmsg("index \"%s\" contains an inner tuple marked as invalid",
RelationGetRelationName(r)),
errdetail("This is caused by an incomplete page split at crash recovery before upgrading to PostgreSQL 9.1."),
errhint("Please REINDEX it.")));
/*
* Check that the key representing the target child node is
* consistent with the key we're inserting. Update it if it's not.
*/
newtup = gistgetadjusted(state.r, idxtuple, itup, giststate);
if (newtup)
{
/*
* Swap shared lock for an exclusive one. Beware, the page may
* change while we unlock/lock the page...
*/
if (!xlocked)
{
LockBuffer(stack->buffer, GIST_UNLOCK);
LockBuffer(stack->buffer, GIST_EXCLUSIVE);
xlocked = true;
stack->page = (Page) BufferGetPage(stack->buffer);
if (PageGetLSN(stack->page) != stack->lsn)
{
/* the page was changed while we unlocked it, retry */
continue;
}
}
/*
* Update the tuple.
*
* We still hold the lock after gistinserttuple(), but it
* might have to split the page to make the updated tuple fit.
* In that case the updated tuple might migrate to the other
* half of the split, so we have to go back to the parent and
* descend back to the half that's a better fit for the new
* tuple.
*/
if (gistinserttuple(&state, stack, giststate, newtup,
downlinkoffnum))
{
/*
* If this was a root split, the root page continues to be
* the parent and the updated tuple went to one of the
* child pages, so we just need to retry from the root
* page.
*/
if (stack->blkno != GIST_ROOT_BLKNO)
{
UnlockReleaseBuffer(stack->buffer);
xlocked = false;
state.stack = stack = stack->parent;
}
continue;
}
}
LockBuffer(stack->buffer, GIST_UNLOCK);
xlocked = false;
/* descend to the chosen child */
item = (GISTInsertStack *) palloc0(sizeof(GISTInsertStack));
item->blkno = childblkno;
item->parent = stack;
item->downlinkoffnum = downlinkoffnum;
state.stack = stack = item;
}
else
{
/*
* Leaf page. Insert the new key. We've already updated all the
* parents on the way down, but we might have to split the page if
* it doesn't fit. gistinserthere() will take care of that.
*/
/*
* Swap shared lock for an exclusive one. Be careful, the page may
* change while we unlock/lock the page...
*/
if (!xlocked)
{
LockBuffer(stack->buffer, GIST_UNLOCK);
LockBuffer(stack->buffer, GIST_EXCLUSIVE);
xlocked = true;
stack->page = (Page) BufferGetPage(stack->buffer);
stack->lsn = PageGetLSN(stack->page);
if (stack->blkno == GIST_ROOT_BLKNO)
{
/*
* the only page that can become inner instead of leaf is
* the root page, so for root we should recheck it
*/
if (!GistPageIsLeaf(stack->page))
{
/*
* very rare situation: during unlock/lock index with
* number of pages = 1 was increased
*/
LockBuffer(stack->buffer, GIST_UNLOCK);
xlocked = false;
continue;
}
/*
* we don't need to check root split, because checking
* leaf/inner is enough to recognize split for root
*/
}
else if (GistFollowRight(stack->page) ||
stack->parent->lsn < GistPageGetNSN(stack->page))
{
/*
* The page was split while we momentarily unlocked the
* page. Go back to parent.
*/
UnlockReleaseBuffer(stack->buffer);
xlocked = false;
state.stack = stack = stack->parent;
continue;
}
}
/* now state.stack->(page, buffer and blkno) points to leaf page */
gistinserttuple(&state, stack, giststate, itup,
InvalidOffsetNumber);
LockBuffer(stack->buffer, GIST_UNLOCK);
/* Release any pins we might still hold before exiting */
for (; stack; stack = stack->parent)
ReleaseBuffer(stack->buffer);
break;
}
}
}
/*
* Actually do a base backup for the specified tablespaces.
*
* This is split out mainly to avoid complaints about "variable might be
* clobbered by longjmp" from stupider versions of gcc.
*/
static void
perform_base_backup(basebackup_options *opt, DIR *tblspcdir)
{
XLogRecPtr startptr;
XLogRecPtr endptr;
char *labelfile;
startptr = do_pg_start_backup(opt->label, opt->fastcheckpoint, &labelfile);
Assert(!XLogRecPtrIsInvalid(startptr));
elogif(!debug_basebackup, LOG,
"basebackup perform -- "
"Basebackup start xlog location = %X/%X",
startptr.xlogid, startptr.xrecoff);
/*
* Set xlogCleanUpTo so that checkpoint process knows
* which old xlog files should not be cleaned
*/
WalSndSetXLogCleanUpTo(startptr);
SIMPLE_FAULT_INJECTOR(BaseBackupPostCreateCheckpoint);
SendXlogRecPtrResult(startptr);
PG_ENSURE_ERROR_CLEANUP(base_backup_cleanup, (Datum) 0);
{
List *filespaces = NIL;
ListCell *lc;
/* Collect information about all filespaces, including pg_system */
filespaces = get_filespaces_to_send(opt);
/* Send filespace header */
SendBackupHeader(filespaces);
/* Send off our filespaces one by one */
foreach(lc, filespaces)
{
filespaceinfo *fi = (filespaceinfo *) lfirst(lc);
StringInfoData buf;
/* Send CopyOutResponse message */
pq_beginmessage(&buf, 'H');
pq_sendbyte(&buf, 0); /* overall format */
pq_sendint(&buf, 0, 2); /* natts */
pq_endmessage(&buf);
/* In the main tar, include the backup_label first. */
if (fi->primary_path == NULL)
sendFileWithContent(BACKUP_LABEL_FILE, labelfile);
sendDir(fi->primary_path == NULL ? "." : fi->primary_path,
fi->primary_path == NULL ? 1 : strlen(fi->primary_path),
opt->exclude, false);
/* In the main tar, include pg_control last. */
if (fi->primary_path == NULL)
{
struct stat statbuf;
if (lstat(XLOG_CONTROL_FILE, &statbuf) != 0)
{
ereport(ERROR,
(errcode_for_file_access(),
errmsg("could not stat control file \"%s\": %m",
XLOG_CONTROL_FILE)));
}
sendFile(XLOG_CONTROL_FILE, XLOG_CONTROL_FILE, &statbuf);
elogif(debug_basebackup, LOG,
"basebackup perform -- Sent file %s." , XLOG_CONTROL_FILE);
}
/*
* If we're including WAL, and this is the main data directory we
* don't terminate the tar stream here. Instead, we will append
* the xlog files below and terminate it then. This is safe since
* the main data directory is always sent *last*.
*/
if (opt->includewal && fi->xlogdir)
{
Assert(lnext(lc) == NULL);
}
else
pq_putemptymessage('c'); /* CopyDone */
}
}
/*
* Backup data file in the from_root directory to the to_root directory with
* same relative path.
* If lsn is not NULL, pages only which are modified after the lsn will be
* copied.
*/
bool
backup_data_file(const char *from_root, const char *to_root,
pgFile *file, const XLogRecPtr *lsn)
{
char to_path[MAXPGPATH];
FILE *in;
FILE *out;
BackupPageHeader header;
DataPage page; /* used as read buffer */
BlockNumber blknum = 0;
size_t read_len = 0;
pg_crc32 crc;
off_t offset;
INIT_CRC32C(crc);
/* reset size summary */
file->read_size = 0;
file->write_size = 0;
/* open backup mode file for read */
in = fopen(file->path, "r");
if (in == NULL)
{
FIN_CRC32C(crc);
file->crc = crc;
/* maybe vanished, it's not error */
if (errno == ENOENT)
return false;
elog(ERROR, "cannot open backup mode file \"%s\": %s",
file->path, strerror(errno));
}
/* open backup file for write */
if (check)
snprintf(to_path, lengthof(to_path), "%s/tmp", backup_path);
else
join_path_components(to_path, to_root, file->path + strlen(from_root) + 1);
out = fopen(to_path, "w");
if (out == NULL)
{
int errno_tmp = errno;
fclose(in);
elog(ERROR, "cannot open backup file \"%s\": %s",
to_path, strerror(errno_tmp));
}
/* confirm server version */
check_server_version();
/*
* Read each page and write the page excluding hole. If it has been
* determined that the page can be copied safely, but no page map
* has been built, it means that we are in presence of a relation
* file that needs to be completely scanned. If a page map is present
* only scan the blocks needed. In each case, pages are copied without
* their hole to ensure some basic level of compression.
*/
if (file->pagemap.bitmapsize == 0)
{
for (blknum = 0;
(read_len = fread(&page, 1, sizeof(page), in)) == sizeof(page);
++blknum)
{
XLogRecPtr page_lsn;
int upper_offset;
int upper_length;
header.block = blknum;
/*
* If an invalid data page was found, fallback to simple copy to ensure
* all pages in the file don't have BackupPageHeader.
*/
if (!parse_page(&page, &page_lsn,
&header.hole_offset, &header.hole_length))
{
elog(LOG, "%s fall back to simple copy", file->path);
fclose(in);
fclose(out);
file->is_datafile = false;
return copy_file(from_root, to_root, file);
}
file->read_size += read_len;
/* if the page has not been modified since last backup, skip it */
if (lsn && !XLogRecPtrIsInvalid(page_lsn) && page_lsn < *lsn)
continue;
upper_offset = header.hole_offset + header.hole_length;
upper_length = BLCKSZ - upper_offset;
/* write data page excluding hole */
if (fwrite(&header, 1, sizeof(header), out) != sizeof(header) ||
fwrite(page.data, 1, header.hole_offset, out) != header.hole_offset ||
fwrite(page.data + upper_offset, 1, upper_length, out) != upper_length)
{
int errno_tmp = errno;
/* oops */
fclose(in);
fclose(out);
elog(ERROR, "cannot write at block %u of \"%s\": %s",
blknum, to_path, strerror(errno_tmp));
}
/* update CRC */
COMP_CRC32C(crc, &header, sizeof(header));
COMP_CRC32C(crc, page.data, header.hole_offset);
COMP_CRC32C(crc, page.data + upper_offset, upper_length);
file->write_size += sizeof(header) + read_len - header.hole_length;
}
}
else
{
datapagemap_iterator_t *iter;
iter = datapagemap_iterate(&file->pagemap);
while (datapagemap_next(iter, &blknum))
{
XLogRecPtr page_lsn;
int upper_offset;
int upper_length;
int ret;
offset = blknum * BLCKSZ;
if (offset > 0)
{
ret = fseek(in, offset, SEEK_SET);
if (ret != 0)
elog(ERROR,
"Can't seek in file offset: %llu ret:%i\n",
(long long unsigned int) offset, ret);
}
read_len = fread(&page, 1, sizeof(page), in);
header.block = blknum;
/*
* If an invalid data page was found, fallback to simple copy to ensure
* all pages in the file don't have BackupPageHeader.
*/
if (!parse_page(&page, &page_lsn,
&header.hole_offset, &header.hole_length))
{
elog(LOG, "%s fall back to simple copy", file->path);
fclose(in);
fclose(out);
file->is_datafile = false;
return copy_file(from_root, to_root, file);
}
file->read_size += read_len;
/* if the page has not been modified since last backup, skip it */
if (lsn && !XLogRecPtrIsInvalid(page_lsn) && page_lsn < *lsn)
continue;
upper_offset = header.hole_offset + header.hole_length;
upper_length = BLCKSZ - upper_offset;
/* write data page excluding hole */
if (fwrite(&header, 1, sizeof(header), out) != sizeof(header) ||
fwrite(page.data, 1, header.hole_offset, out) != header.hole_offset ||
fwrite(page.data + upper_offset, 1, upper_length, out) != upper_length)
{
int errno_tmp = errno;
/* oops */
fclose(in);
fclose(out);
elog(ERROR, "cannot write at block %u of \"%s\": %s",
blknum, to_path, strerror(errno_tmp));
}
/* update CRC */
COMP_CRC32C(crc, &header, sizeof(header));
COMP_CRC32C(crc, page.data, header.hole_offset);
COMP_CRC32C(crc, page.data + upper_offset, upper_length);
file->write_size += sizeof(header) + read_len - header.hole_length;
}
pg_free(iter);
}
/*
* update file permission
* FIXME: Should set permission on open?
*/
if (!check && chmod(to_path, FILE_PERMISSION) == -1)
{
int errno_tmp = errno;
fclose(in);
fclose(out);
elog(ERROR, "cannot change mode of \"%s\": %s", file->path,
strerror(errno_tmp));
}
fclose(in);
fclose(out);
/* finish CRC calculation and store into pgFile */
FIN_CRC32C(crc);
file->crc = crc;
/* Treat empty file as not-datafile */
if (file->read_size == 0)
file->is_datafile = false;
/* We do not backup if all pages skipped. */
if (file->write_size == 0 && file->read_size > 0)
{
if (remove(to_path) == -1)
elog(ERROR, "cannot remove file \"%s\": %s", to_path,
strerror(errno));
return false;
}
/* remove $BACKUP_PATH/tmp created during check */
if (check)
remove(to_path);
return true;
}
/*
* Physical write of a page from a buffer slot
*
* On failure, we cannot just ereport(ERROR) since caller has put state in
* shared memory that must be undone. So, we return FALSE and save enough
* info in static variables to let SlruReportIOError make the report.
*
* For now, assume it's not worth keeping a file pointer open across
* independent read/write operations. We do batch operations during
* SimpleLruFlush, though.
*
* fdata is NULL for a standalone write, pointer to open-file info during
* SimpleLruFlush.
*/
static bool
SlruPhysicalWritePage(SlruCtl ctl, int pageno, int slotno, SlruFlush fdata)
{
SlruShared shared = ctl->shared;
int segno = pageno / SLRU_PAGES_PER_SEGMENT;
int rpageno = pageno % SLRU_PAGES_PER_SEGMENT;
int offset = rpageno * BLCKSZ;
char path[MAXPGPATH];
int fd = -1;
struct timeval tv;
/*
* Honor the write-WAL-before-data rule, if appropriate, so that we do not
* write out data before associated WAL records. This is the same action
* performed during FlushBuffer() in the main buffer manager.
*/
if (shared->group_lsn != NULL)
{
/*
* We must determine the largest async-commit LSN for the page. This
* is a bit tedious, but since this entire function is a slow path
* anyway, it seems better to do this here than to maintain a per-page
* LSN variable (which'd need an extra comparison in the
* transaction-commit path).
*/
XLogRecPtr max_lsn;
int lsnindex,
lsnoff;
lsnindex = slotno * shared->lsn_groups_per_page;
max_lsn = shared->group_lsn[lsnindex++];
for (lsnoff = 1; lsnoff < shared->lsn_groups_per_page; lsnoff++)
{
XLogRecPtr this_lsn = shared->group_lsn[lsnindex++];
if (XLByteLT(max_lsn, this_lsn))
max_lsn = this_lsn;
}
if (!XLogRecPtrIsInvalid(max_lsn))
{
/*
* As noted above, elog(ERROR) is not acceptable here, so if
* XLogFlush were to fail, we must PANIC. This isn't much of a
* restriction because XLogFlush is just about all critical
* section anyway, but let's make sure.
*/
START_CRIT_SECTION();
XLogFlush(max_lsn);
END_CRIT_SECTION();
}
}
/*
* During a Flush, we may already have the desired file open.
*/
if (fdata)
{
int i;
for (i = 0; i < fdata->num_files; i++)
{
if (fdata->segno[i] == segno)
{
fd = fdata->fd[i];
break;
}
}
}
if (fd < 0)
{
/*
* If the file doesn't already exist, we should create it. It is
* possible for this to need to happen when writing a page that's not
* first in its segment; we assume the OS can cope with that. (Note:
* it might seem that it'd be okay to create files only when
* SimpleLruZeroPage is called for the first page of a segment.
* However, if after a crash and restart the REDO logic elects to
* replay the log from a checkpoint before the latest one, then it's
* possible that we will get commands to set transaction status of
* transactions that have already been truncated from the commit log.
* Easiest way to deal with that is to accept references to
* nonexistent files here and in SlruPhysicalReadPage.)
*
* Note: it is possible for more than one backend to be executing this
* code simultaneously for different pages of the same file. Hence,
* don't use O_EXCL or O_TRUNC or anything like that.
*/
SlruFileName(ctl, path, segno);
fd = BasicOpenFile(path, O_RDWR | O_CREAT | PG_BINARY,
S_IRUSR | S_IWUSR);
if (fd < 0)
{
slru_errcause = SLRU_OPEN_FAILED;
slru_errno = errno;
return false;
}
if (fdata)
{
if (fdata->num_files < MAX_FLUSH_BUFFERS)
{
fdata->fd[fdata->num_files] = fd;
fdata->segno[fdata->num_files] = segno;
fdata->num_files++;
}
else
{
/*
* In the unlikely event that we exceed MAX_FLUSH_BUFFERS,
* fall back to treating it as a standalone write.
*/
fdata = NULL;
}
}
}
if (lseek(fd, (off_t) offset, SEEK_SET) < 0)
{
slru_errcause = SLRU_SEEK_FAILED;
slru_errno = errno;
if (!fdata)
close(fd);
return false;
}
errno = 0;
if (write(fd, shared->page_buffer[slotno], BLCKSZ) != BLCKSZ)
{
/* if write didn't set errno, assume problem is no disk space */
if (errno == 0)
errno = ENOSPC;
slru_errcause = SLRU_WRITE_FAILED;
slru_errno = errno;
if (!fdata)
close(fd);
return false;
}
#ifdef XP_TRACE_LRU_WRITE
gettimeofday(&tv, NULL);
ereport(TRACE_LEVEL,
(errmsg("%ld.%ld:\tWRITE:\tSlruPhysicalWritePage:\tfile:%s",
tv.tv_sec, tv.tv_usec, path)));
#endif
/*
* If not part of Flush, need to fsync now. We assume this happens
* infrequently enough that it's not a performance issue.
*/
if (!fdata)
{
if (ctl->do_fsync && pg_fsync(fd))
{
slru_errcause = SLRU_FSYNC_FAILED;
slru_errno = errno;
close(fd);
return false;
}
if (close(fd))
{
slru_errcause = SLRU_CLOSE_FAILED;
slru_errno = errno;
return false;
}
}
return true;
}
/*
* visibilitymap_set - set a bit on a previously pinned page
*
* recptr is the LSN of the XLOG record we're replaying, if we're in recovery,
* or InvalidXLogRecPtr in normal running. The page LSN is advanced to the
* one provided; in normal running, we generate a new___ XLOG record and set the
* page LSN to that value. cutoff_xid is the largest xmin on the page being
* marked all-visible; it is needed for Hot Standby, and can be
* InvalidTransactionId if the page contains no tuples.
*
* Caller is expected to set the heap page's PD_ALL_VISIBLE bit before calling
* this function. Except in recovery, caller should also pass the heap
* buffer. When checksums are enabled and we're not in recovery, we must add
* the heap buffer to the WAL chain to protect it from being torn.
*
* You must pass a buffer containing the correct map page to this function.
* Call visibilitymap_pin first to pin the right one. This function doesn't do
* any I/O.
*/
void
visibilitymap_set(Relation rel, BlockNumber heapBlk, Buffer heapBuf,
XLogRecPtr recptr, Buffer vmBuf, TransactionId cutoff_xid)
{
BlockNumber mapBlock = HEAPBLK_TO_MAPBLOCK(heapBlk);
uint32 mapByte = HEAPBLK_TO_MAPBYTE(heapBlk);
uint8 mapBit = HEAPBLK_TO_MAPBIT(heapBlk);
Page page;
char *map;
#ifdef TRACE_VISIBILITYMAP
elog(DEBUG1, "vm_set %s %d", RelationGetRelationName(rel), heapBlk);
#endif
Assert(InRecovery || XLogRecPtrIsInvalid(recptr));
Assert(InRecovery || BufferIsValid(heapBuf));
/* Check that we have the right heap page pinned, if present */
if (BufferIsValid(heapBuf) && BufferGetBlockNumber(heapBuf) != heapBlk)
elog(ERROR, "wrong heap buffer passed to visibilitymap_set");
/* Check that we have the right VM page pinned */
if (!BufferIsValid(vmBuf) || BufferGetBlockNumber(vmBuf) != mapBlock)
elog(ERROR, "wrong VM buffer passed to visibilitymap_set");
page = BufferGetPage(vmBuf);
map = PageGetContents(page);
LockBuffer(vmBuf, BUFFER_LOCK_EXCLUSIVE);
if (!(map[mapByte] & (1 << mapBit)))
{
START_CRIT_SECTION();
map[mapByte] |= (1 << mapBit);
MarkBufferDirty(vmBuf);
if (RelationNeedsWAL(rel))
{
if (XLogRecPtrIsInvalid(recptr))
{
Assert(!InRecovery);
recptr = log_heap_visible(rel->rd_node, heapBuf, vmBuf,
cutoff_xid);
/*
* If data checksums are enabled (or wal_log_hints=on), we
* need to protect the heap page from being torn.
*/
if (XLogHintBitIsNeeded())
{
Page heapPage = BufferGetPage(heapBuf);
/* caller is expected to set PD_ALL_VISIBLE first */
Assert(PageIsAllVisible(heapPage));
PageSetLSN(heapPage, recptr);
}
}
PageSetLSN(page, recptr);
}
END_CRIT_SECTION();
}
LockBuffer(vmBuf, BUFFER_LOCK_UNLOCK);
}
/*
* gistkillitems() -- set LP_DEAD state for items an indexscan caller has
* told us were killed.
*
* We re-read page here, so it's important to check page LSN. If the page
* has been modified since the last read (as determined by LSN), we cannot
* flag any entries because it is possible that the old entry was vacuumed
* away and the TID was re-used by a completely different heap tuple.
*/
static void
gistkillitems(IndexScanDesc scan)
{
GISTScanOpaque so = (GISTScanOpaque) scan->opaque;
Buffer buffer;
Page page;
OffsetNumber offnum;
ItemId iid;
int i;
bool killedsomething = false;
Assert(so->curBlkno != InvalidBlockNumber);
Assert(!XLogRecPtrIsInvalid(so->curPageLSN));
Assert(so->killedItems != NULL);
buffer = ReadBuffer(scan->indexRelation, so->curBlkno);
if (!BufferIsValid(buffer))
return;
LockBuffer(buffer, GIST_SHARE);
gistcheckpage(scan->indexRelation, buffer);
page = BufferGetPage(buffer);
/*
* If page LSN differs it means that the page was modified since the last
* read. killedItems could be not valid so LP_DEAD hints applying is not
* safe.
*/
if (BufferGetLSNAtomic(buffer) != so->curPageLSN)
{
UnlockReleaseBuffer(buffer);
so->numKilled = 0; /* reset counter */
return;
}
Assert(GistPageIsLeaf(page));
/*
* Mark all killedItems as dead. We need no additional recheck, because,
* if page was modified, pageLSN must have changed.
*/
for (i = 0; i < so->numKilled; i++)
{
offnum = so->killedItems[i];
iid = PageGetItemId(page, offnum);
ItemIdMarkDead(iid);
killedsomething = true;
}
if (killedsomething)
{
GistMarkPageHasGarbage(page);
MarkBufferDirtyHint(buffer, true);
}
UnlockReleaseBuffer(buffer);
/*
* Always reset the scan state, so we don't look for same items on other
* pages.
*/
so->numKilled = 0;
}
/*
* Scan all items on the GiST index page identified by *pageItem, and insert
* them into the queue (or directly to output areas)
*
* scan: index scan we are executing
* pageItem: search queue item identifying an index page to scan
* myDistances: distances array associated with pageItem, or NULL at the root
* tbm: if not NULL, gistgetbitmap's output bitmap
* ntids: if not NULL, gistgetbitmap's output tuple counter
*
* If tbm/ntids aren't NULL, we are doing an amgetbitmap scan, and heap
* tuples should be reported directly into the bitmap. If they are NULL,
* we're doing a plain or ordered indexscan. For a plain indexscan, heap
* tuple TIDs are returned into so->pageData[]. For an ordered indexscan,
* heap tuple TIDs are pushed into individual search queue items. In an
* index-only scan, reconstructed index tuples are returned along with the
* TIDs.
*
* If we detect that the index page has split since we saw its downlink
* in the parent, we push its new right sibling onto the queue so the
* sibling will be processed next.
*/
static void
gistScanPage(IndexScanDesc scan, GISTSearchItem *pageItem, double *myDistances,
TIDBitmap *tbm, int64 *ntids)
{
GISTScanOpaque so = (GISTScanOpaque) scan->opaque;
GISTSTATE *giststate = so->giststate;
Relation r = scan->indexRelation;
Buffer buffer;
Page page;
GISTPageOpaque opaque;
OffsetNumber maxoff;
OffsetNumber i;
MemoryContext oldcxt;
Assert(!GISTSearchItemIsHeap(*pageItem));
buffer = ReadBuffer(scan->indexRelation, pageItem->blkno);
LockBuffer(buffer, GIST_SHARE);
PredicateLockPage(r, BufferGetBlockNumber(buffer), scan->xs_snapshot);
gistcheckpage(scan->indexRelation, buffer);
page = BufferGetPage(buffer);
TestForOldSnapshot(scan->xs_snapshot, r, page);
opaque = GistPageGetOpaque(page);
/*
* Check if we need to follow the rightlink. We need to follow it if the
* page was concurrently split since we visited the parent (in which case
* parentlsn < nsn), or if the system crashed after a page split but
* before the downlink was inserted into the parent.
*/
if (!XLogRecPtrIsInvalid(pageItem->data.parentlsn) &&
(GistFollowRight(page) ||
pageItem->data.parentlsn < GistPageGetNSN(page)) &&
opaque->rightlink != InvalidBlockNumber /* sanity check */ )
{
/* There was a page split, follow right link to add pages */
GISTSearchItem *item;
/* This can't happen when starting at the root */
Assert(myDistances != NULL);
oldcxt = MemoryContextSwitchTo(so->queueCxt);
/* Create new GISTSearchItem for the right sibling index page */
item = palloc(SizeOfGISTSearchItem(scan->numberOfOrderBys));
item->blkno = opaque->rightlink;
item->data.parentlsn = pageItem->data.parentlsn;
/* Insert it into the queue using same distances as for this page */
memcpy(item->distances, myDistances,
sizeof(double) * scan->numberOfOrderBys);
pairingheap_add(so->queue, &item->phNode);
MemoryContextSwitchTo(oldcxt);
}
so->nPageData = so->curPageData = 0;
scan->xs_hitup = NULL; /* might point into pageDataCxt */
if (so->pageDataCxt)
MemoryContextReset(so->pageDataCxt);
/*
* We save the LSN of the page as we read it, so that we know whether it
* safe to apply LP_DEAD hints to the page later. This allows us to drop
* the pin for MVCC scans, which allows vacuum to avoid blocking.
*/
so->curPageLSN = BufferGetLSNAtomic(buffer);
/*
* check all tuples on page
*/
maxoff = PageGetMaxOffsetNumber(page);
for (i = FirstOffsetNumber; i <= maxoff; i = OffsetNumberNext(i))
{
ItemId iid = PageGetItemId(page, i);
IndexTuple it;
bool match;
bool recheck;
bool recheck_distances;
/*
* If the scan specifies not to return killed tuples, then we treat a
* killed tuple as not passing the qual.
*/
if (scan->ignore_killed_tuples && ItemIdIsDead(iid))
continue;
it = (IndexTuple) PageGetItem(page, iid);
/*
* Must call gistindex_keytest in tempCxt, and clean up any leftover
* junk afterward.
*/
oldcxt = MemoryContextSwitchTo(so->giststate->tempCxt);
match = gistindex_keytest(scan, it, page, i,
&recheck, &recheck_distances);
MemoryContextSwitchTo(oldcxt);
MemoryContextReset(so->giststate->tempCxt);
/* Ignore tuple if it doesn't match */
if (!match)
continue;
if (tbm && GistPageIsLeaf(page))
{
/*
* getbitmap scan, so just push heap tuple TIDs into the bitmap
* without worrying about ordering
*/
tbm_add_tuples(tbm, &it->t_tid, 1, recheck);
(*ntids)++;
}
else if (scan->numberOfOrderBys == 0 && GistPageIsLeaf(page))
{
/*
* Non-ordered scan, so report tuples in so->pageData[]
*/
so->pageData[so->nPageData].heapPtr = it->t_tid;
so->pageData[so->nPageData].recheck = recheck;
so->pageData[so->nPageData].offnum = i;
/*
* In an index-only scan, also fetch the data from the tuple. The
* reconstructed tuples are stored in pageDataCxt.
*/
if (scan->xs_want_itup)
{
oldcxt = MemoryContextSwitchTo(so->pageDataCxt);
so->pageData[so->nPageData].recontup =
gistFetchTuple(giststate, r, it);
MemoryContextSwitchTo(oldcxt);
}
so->nPageData++;
}
else
{
/*
* Must push item into search queue. We get here for any lower
* index page, and also for heap tuples if doing an ordered
* search.
*/
GISTSearchItem *item;
oldcxt = MemoryContextSwitchTo(so->queueCxt);
/* Create new GISTSearchItem for this item */
item = palloc(SizeOfGISTSearchItem(scan->numberOfOrderBys));
if (GistPageIsLeaf(page))
{
/* Creating heap-tuple GISTSearchItem */
item->blkno = InvalidBlockNumber;
item->data.heap.heapPtr = it->t_tid;
item->data.heap.recheck = recheck;
item->data.heap.recheckDistances = recheck_distances;
/*
* In an index-only scan, also fetch the data from the tuple.
*/
if (scan->xs_want_itup)
item->data.heap.recontup = gistFetchTuple(giststate, r, it);
}
else
{
/* Creating index-page GISTSearchItem */
item->blkno = ItemPointerGetBlockNumber(&it->t_tid);
/*
* LSN of current page is lsn of parent page for child. We
* only have a shared lock, so we need to get the LSN
* atomically.
*/
item->data.parentlsn = BufferGetLSNAtomic(buffer);
}
/* Insert it into the queue using new distance data */
memcpy(item->distances, so->distances,
sizeof(double) * scan->numberOfOrderBys);
pairingheap_add(so->queue, &item->phNode);
MemoryContextSwitchTo(oldcxt);
}
}
UnlockReleaseBuffer(buffer);
}
static void
gistfindleaf(GISTInsertState *state, GISTSTATE *giststate)
{
ItemId iid;
IndexTuple idxtuple;
GISTPageOpaque opaque;
MIRROREDLOCK_BUFMGR_MUST_ALREADY_BE_HELD;
/*
* walk down, We don't lock page for a long time, but so we should be
* ready to recheck path in a bad case... We remember, that page->lsn
* should never be invalid.
*/
for (;;)
{
if (XLogRecPtrIsInvalid(state->stack->lsn))
state->stack->buffer = ReadBuffer(state->r, state->stack->blkno);
LockBuffer(state->stack->buffer, GIST_SHARE);
gistcheckpage(state->r, state->stack->buffer);
state->stack->page = (Page) BufferGetPage(state->stack->buffer);
opaque = GistPageGetOpaque(state->stack->page);
state->stack->lsn = PageGetLSN(state->stack->page);
Assert(state->r->rd_istemp || !XLogRecPtrIsInvalid(state->stack->lsn));
if (state->stack->blkno != GIST_ROOT_BLKNO &&
XLByteLT(state->stack->parent->lsn, opaque->nsn))
{
/*
* caused split non-root page is detected, go up to parent to
* choose best child
*/
UnlockReleaseBuffer(state->stack->buffer);
state->stack = state->stack->parent;
continue;
}
if (!GistPageIsLeaf(state->stack->page))
{
/*
* This is an internal page, so continue to walk down the tree. We
* find the child node that has the minimum insertion penalty and
* recursively invoke ourselves to modify that node. Once the
* recursive call returns, we may need to adjust the parent node
* for two reasons: the child node split, or the key in this node
* needs to be adjusted for the newly inserted key below us.
*/
GISTInsertStack *item = (GISTInsertStack *) palloc0(sizeof(GISTInsertStack));
state->stack->childoffnum = gistchoose(state->r, state->stack->page, state->itup[0], giststate);
iid = PageGetItemId(state->stack->page, state->stack->childoffnum);
idxtuple = (IndexTuple) PageGetItem(state->stack->page, iid);
item->blkno = ItemPointerGetBlockNumber(&(idxtuple->t_tid));
LockBuffer(state->stack->buffer, GIST_UNLOCK);
item->parent = state->stack;
item->child = NULL;
if (state->stack)
state->stack->child = item;
state->stack = item;
}
else
{
/* be carefull, during unlock/lock page may be changed... */
LockBuffer(state->stack->buffer, GIST_UNLOCK);
LockBuffer(state->stack->buffer, GIST_EXCLUSIVE);
state->stack->page = (Page) BufferGetPage(state->stack->buffer);
opaque = GistPageGetOpaque(state->stack->page);
if (state->stack->blkno == GIST_ROOT_BLKNO)
{
/*
* the only page can become inner instead of leaf is a root
* page, so for root we should recheck it
*/
if (!GistPageIsLeaf(state->stack->page))
{
/*
* very rarely situation: during unlock/lock index with
* number of pages = 1 was increased
*/
LockBuffer(state->stack->buffer, GIST_UNLOCK);
continue;
}
/*
* we don't need to check root split, because checking
* leaf/inner is enough to recognize split for root
*/
}
else if (XLByteLT(state->stack->parent->lsn, opaque->nsn))
{
/*
* detecting split during unlock/lock, so we should find
* better child on parent
*/
/* forget buffer */
UnlockReleaseBuffer(state->stack->buffer);
state->stack = state->stack->parent;
continue;
}
state->stack->lsn = PageGetLSN(state->stack->page);
/* ok we found a leaf page and it X-locked */
break;
}
}
/* now state->stack->(page, buffer and blkno) points to leaf page */
}
/*
* Fetch a tuples that matchs the search key; this can be invoked
* either to fetch the first such tuple or subsequent matching
* tuples. Returns true iff a matching tuple was found.
*/
static int
gistnext(IndexScanDesc scan, ScanDirection dir, ItemPointer tids,
int maxtids, bool ignore_killed_tuples)
{
MIRROREDLOCK_BUFMGR_DECLARE;
Page p;
OffsetNumber n;
GISTScanOpaque so;
GISTSearchStack *stk;
IndexTuple it;
GISTPageOpaque opaque;
int ntids = 0;
so = (GISTScanOpaque) scan->opaque;
// -------- MirroredLock ----------
MIRROREDLOCK_BUFMGR_LOCK;
if ( so->qual_ok == false )
return 0;
if (ItemPointerIsValid(&so->curpos) == false)
{
/* Being asked to fetch the first entry, so start at the root */
Assert(so->curbuf == InvalidBuffer);
Assert(so->stack == NULL);
so->curbuf = ReadBuffer(scan->indexRelation, GIST_ROOT_BLKNO);
stk = so->stack = (GISTSearchStack *) palloc0(sizeof(GISTSearchStack));
stk->next = NULL;
stk->block = GIST_ROOT_BLKNO;
pgstat_count_index_scan(scan->indexRelation);
}
else if (so->curbuf == InvalidBuffer)
{
MIRROREDLOCK_BUFMGR_UNLOCK;
// -------- MirroredLock ----------
return 0;
}
/*
* check stored pointers from last visit
*/
if ( so->nPageData > 0 )
{
while( ntids < maxtids && so->curPageData < so->nPageData )
{
tids[ ntids ] = scan->xs_ctup.t_self = so->pageData[ so->curPageData ].heapPtr;
ItemPointerSet(&(so->curpos),
BufferGetBlockNumber(so->curbuf),
so->pageData[ so->curPageData ].pageOffset);
so->curPageData ++;
ntids++;
}
if ( ntids == maxtids )
{
MIRROREDLOCK_BUFMGR_UNLOCK;
// -------- MirroredLock ----------
return ntids;
}
/*
* Go to the next page
*/
stk = so->stack->next;
pfree(so->stack);
so->stack = stk;
/* If we're out of stack entries, we're done */
if (so->stack == NULL)
{
ReleaseBuffer(so->curbuf);
so->curbuf = InvalidBuffer;
MIRROREDLOCK_BUFMGR_UNLOCK;
// -------- MirroredLock ----------
return ntids;
}
so->curbuf = ReleaseAndReadBuffer(so->curbuf,
scan->indexRelation,
stk->block);
}
for (;;)
{
/* First of all, we need lock buffer */
Assert(so->curbuf != InvalidBuffer);
LockBuffer(so->curbuf, GIST_SHARE);
gistcheckpage(scan->indexRelation, so->curbuf);
p = BufferGetPage(so->curbuf);
opaque = GistPageGetOpaque(p);
/* remember lsn to identify page changed for tuple's killing */
so->stack->lsn = PageGetLSN(p);
/* check page split, occured from last visit or visit to parent */
if (!XLogRecPtrIsInvalid(so->stack->parentlsn) &&
XLByteLT(so->stack->parentlsn, opaque->nsn) &&
opaque->rightlink != InvalidBlockNumber /* sanity check */ &&
(so->stack->next == NULL || so->stack->next->block != opaque->rightlink) /* check if already
added */ )
{
/* detect page split, follow right link to add pages */
stk = (GISTSearchStack *) palloc(sizeof(GISTSearchStack));
stk->next = so->stack->next;
stk->block = opaque->rightlink;
stk->parentlsn = so->stack->parentlsn;
memset(&(stk->lsn), 0, sizeof(GistNSN));
so->stack->next = stk;
}
/* if page is empty, then just skip it */
if (PageIsEmpty(p))
{
LockBuffer(so->curbuf, GIST_UNLOCK);
stk = so->stack->next;
pfree(so->stack);
so->stack = stk;
if (so->stack == NULL)
{
ReleaseBuffer(so->curbuf);
so->curbuf = InvalidBuffer;
MIRROREDLOCK_BUFMGR_UNLOCK;
// -------- MirroredLock ----------
return ntids;
}
so->curbuf = ReleaseAndReadBuffer(so->curbuf, scan->indexRelation,
stk->block);
continue;
}
if (ScanDirectionIsBackward(dir))
n = PageGetMaxOffsetNumber(p);
else
n = FirstOffsetNumber;
/* wonderful, we can look at page */
so->nPageData = so->curPageData = 0;
for (;;)
{
n = gistfindnext(scan, n, dir);
if (!OffsetNumberIsValid(n))
{
while( ntids < maxtids && so->curPageData < so->nPageData )
{
tids[ ntids ] = scan->xs_ctup.t_self =
so->pageData[ so->curPageData ].heapPtr;
ItemPointerSet(&(so->curpos),
BufferGetBlockNumber(so->curbuf),
so->pageData[ so->curPageData ].pageOffset);
so->curPageData ++;
ntids++;
}
if ( ntids == maxtids )
{
LockBuffer(so->curbuf, GIST_UNLOCK);
MIRROREDLOCK_BUFMGR_UNLOCK;
// -------- MirroredLock ----------
return ntids;
}
/*
* We ran out of matching index entries on the current page,
* so pop the top stack entry and use it to continue the
* search.
*/
LockBuffer(so->curbuf, GIST_UNLOCK);
stk = so->stack->next;
pfree(so->stack);
so->stack = stk;
/* If we're out of stack entries, we're done */
if (so->stack == NULL)
{
ReleaseBuffer(so->curbuf);
so->curbuf = InvalidBuffer;
MIRROREDLOCK_BUFMGR_UNLOCK;
// -------- MirroredLock ----------
return ntids;
}
so->curbuf = ReleaseAndReadBuffer(so->curbuf,
scan->indexRelation,
stk->block);
/* XXX go up */
break;
}
if (GistPageIsLeaf(p))
{
/*
* We've found a matching index entry in a leaf page, so
* return success. Note that we keep "curbuf" pinned so that
* we can efficiently resume the index scan later.
*/
if (!(ignore_killed_tuples && ItemIdIsDead(PageGetItemId(p, n))))
{
it = (IndexTuple) PageGetItem(p, PageGetItemId(p, n));
so->pageData[ so->nPageData ].heapPtr = it->t_tid;
so->pageData[ so->nPageData ].pageOffset = n;
so->nPageData ++;
}
}
else
{
/*
* We've found an entry in an internal node whose key is
* consistent with the search key, so push it to stack
*/
stk = (GISTSearchStack *) palloc(sizeof(GISTSearchStack));
it = (IndexTuple) PageGetItem(p, PageGetItemId(p, n));
stk->block = ItemPointerGetBlockNumber(&(it->t_tid));
memset(&(stk->lsn), 0, sizeof(GistNSN));
stk->parentlsn = so->stack->lsn;
stk->next = so->stack->next;
so->stack->next = stk;
}
if (ScanDirectionIsBackward(dir))
n = OffsetNumberPrev(n);
else
n = OffsetNumberNext(n);
}
}
MIRROREDLOCK_BUFMGR_UNLOCK;
// -------- MirroredLock ----------
return ntids;
}
int
do_delete(pgBackupRange *range)
{
int i;
int ret;
parray *backup_list;
bool do_delete = false;
XLogRecPtr oldest_lsn = InvalidXLogRecPtr;
TimeLineID oldest_tli;
/* DATE are always required */
if (!pgBackupRangeIsValid(range))
elog(ERROR, "required delete range option not specified: delete DATE");
/* Lock backup catalog */
ret = catalog_lock();
if (ret == -1)
elog(ERROR, "can't lock backup catalog.");
else if (ret == 1)
elog(ERROR,
"another pg_arman is running, stop delete.");
/* Get complete list of backups */
backup_list = catalog_get_backup_list(NULL);
if (!backup_list)
elog(ERROR, "No backup list found, can't process any more.");
/* Find backups to be deleted */
for (i = 0; i < parray_num(backup_list); i++)
{
pgBackup *backup = (pgBackup *) parray_get(backup_list, i);
/* delete backup and update status to DELETED */
if (do_delete)
{
/* check for interrupt */
if (interrupted)
elog(ERROR, "interrupted during delete backup");
pgBackupDeleteFiles(backup);
continue;
}
/* Found the latest full backup */
if (backup->backup_mode >= BACKUP_MODE_FULL &&
backup->status == BACKUP_STATUS_OK &&
backup->start_time <= range->begin)
{
do_delete = true;
oldest_lsn = backup->start_lsn;
oldest_tli = backup->tli;
}
}
/* release catalog lock */
catalog_unlock();
/* cleanup */
parray_walk(backup_list, pgBackupFree);
parray_free(backup_list);
/*
* Delete in archive WAL segments that are not needed anymore. The oldest
* segment to be kept is the first segment that the oldest full backup
* found around needs to keep.
*/
if (!XLogRecPtrIsInvalid(oldest_lsn))
{
XLogSegNo targetSegNo;
char oldestSegmentNeeded[MAXFNAMELEN];
DIR *arcdir;
struct dirent *arcde;
char wal_file[MAXPGPATH];
int rc;
XLByteToSeg(oldest_lsn, targetSegNo);
XLogFileName(oldestSegmentNeeded, oldest_tli, targetSegNo);
elog(LOG, "Removing segments older than %s", oldestSegmentNeeded);
/*
* Now is time to do the actual work and to remove all the segments
* not needed anymore.
*/
if ((arcdir = opendir(arclog_path)) != NULL)
{
while (errno = 0, (arcde = readdir(arcdir)) != NULL)
{
/*
* We ignore the timeline part of the XLOG segment identifiers in
* deciding whether a segment is still needed. This ensures that
* we won't prematurely remove a segment from a parent timeline.
* We could probably be a little more proactive about removing
* segments of non-parent timelines, but that would be a whole lot
* more complicated.
*
* We use the alphanumeric sorting property of the filenames to
* decide which ones are earlier than the exclusiveCleanupFileName
* file. Note that this means files are not removed in the order
* they were originally written, in case this worries you.
*/
if ((IsXLogFileName(arcde->d_name) ||
IsPartialXLogFileName(arcde->d_name)) &&
strcmp(arcde->d_name + 8, oldestSegmentNeeded + 8) < 0)
{
/*
* Use the original file name again now, including any
* extension that might have been chopped off before testing
* the sequence.
*/
snprintf(wal_file, MAXPGPATH, "%s/%s",
arclog_path, arcde->d_name);
rc = unlink(wal_file);
if (rc != 0)
{
elog(WARNING, "could not remove file \"%s\": %s",
wal_file, strerror(errno));
break;
}
elog(LOG, "removed WAL segment \"%s\"", wal_file);
}
}
if (errno)
elog(WARNING, "could not read archive location \"%s\": %s",
arclog_path, strerror(errno));
if (closedir(arcdir))
elog(WARNING, "could not close archive location \"%s\": %s",
arclog_path, strerror(errno));
}
else
elog(WARNING, "could not open archive location \"%s\": %s",
arclog_path, strerror(errno));
}
return 0;
}
/*
* Attempt to retrieve the specified file from off-line archival storage.
* If successful, fill "path" with its complete path (note that this will be
* a temp file name that doesn't follow the normal naming convention), and
* return TRUE.
*
* If not successful, fill "path" with the name of the normal on-line file
* (which may or may not actually exist, but we'll try to use it), and return
* FALSE.
*
* For fixed-size files, the caller may pass the expected size as an
* additional crosscheck on successful recovery. If the file size is not
* known, set expectedSize = 0.
*/
bool
RestoreArchivedFile(char *path, const char *xlogfname,
const char *recovername, off_t expectedSize)
{
char xlogpath[MAXPGPATH];
char xlogRestoreCmd[MAXPGPATH];
char lastRestartPointFname[MAXPGPATH];
char *dp;
char *endp;
const char *sp;
int rc;
bool signaled;
struct stat stat_buf;
XLogSegNo restartSegNo;
XLogRecPtr restartRedoPtr;
TimeLineID restartTli;
/* In standby mode, restore_command might not be supplied */
if (recoveryRestoreCommand == NULL)
goto not_available;
/*
* When doing archive recovery, we always prefer an archived log file even
* if a file of the same name exists in XLOGDIR. The reason is that the
* file in XLOGDIR could be an old, un-filled or partly-filled version
* that was copied and restored as part of backing up $PGDATA.
*
* We could try to optimize this slightly by checking the local copy
* lastchange timestamp against the archived copy, but we have no API to
* do this, nor can we guarantee that the lastchange timestamp was
* preserved correctly when we copied to archive. Our aim is robustness,
* so we elect not to do this.
*
* If we cannot obtain the log file from the archive, however, we will try
* to use the XLOGDIR file if it exists. This is so that we can make use
* of log segments that weren't yet transferred to the archive.
*
* Notice that we don't actually overwrite any files when we copy back
* from archive because the restore_command may inadvertently
* restore inappropriate xlogs, or they may be corrupt, so we may wish to
* fallback to the segments remaining in current XLOGDIR later. The
* copy-from-archive filename is always the same, ensuring that we don't
* run out of disk space on long recoveries.
*/
snprintf(xlogpath, MAXPGPATH, XLOGDIR "/%s", recovername);
/*
* Make sure there is no existing file named recovername.
*/
if (stat(xlogpath, &stat_buf) != 0)
{
if (errno != ENOENT)
ereport(FATAL,
(errcode_for_file_access(),
errmsg("could not stat file \"%s\": %m",
xlogpath)));
}
else
{
if (unlink(xlogpath) != 0)
ereport(FATAL,
(errcode_for_file_access(),
errmsg("could not remove file \"%s\": %m",
xlogpath)));
}
/*
* Calculate the archive file cutoff point for use during log shipping
* replication. All files earlier than this point can be deleted from the
* archive, though there is no requirement to do so.
*
* We initialise this with the filename of an InvalidXLogRecPtr, which
* will prevent the deletion of any WAL files from the archive because of
* the alphabetic sorting property of WAL filenames.
*
* Once we have successfully located the redo pointer of the checkpoint
* from which we start recovery we never request a file prior to the redo
* pointer of the last restartpoint. When redo begins we know that we have
* successfully located it, so there is no need for additional status
* flags to signify the point when we can begin deleting WAL files from
* the archive.
*/
GetOldestRestartPoint(&restartRedoPtr, &restartTli);
if (!XLogRecPtrIsInvalid(restartRedoPtr))
{
XLByteToSeg(restartRedoPtr, restartSegNo);
XLogFileName(lastRestartPointFname, restartTli, restartSegNo);
/* we shouldn't need anything earlier than last restart point */
Assert(strcmp(lastRestartPointFname, xlogfname) <= 0);
}
else
XLogFileName(lastRestartPointFname, 0, 0L);
/*
* construct the command to be executed
*/
dp = xlogRestoreCmd;
endp = xlogRestoreCmd + MAXPGPATH - 1;
*endp = '\0';
for (sp = recoveryRestoreCommand; *sp; sp++)
{
if (*sp == '%')
{
switch (sp[1])
{
case 'p':
/* %p: relative path of target file */
sp++;
StrNCpy(dp, xlogpath, endp - dp);
make_native_path(dp);
dp += strlen(dp);
break;
case 'f':
/* %f: filename of desired file */
sp++;
StrNCpy(dp, xlogfname, endp - dp);
dp += strlen(dp);
break;
case 'r':
/* %r: filename of last restartpoint */
sp++;
StrNCpy(dp, lastRestartPointFname, endp - dp);
dp += strlen(dp);
break;
case '%':
/* convert %% to a single % */
sp++;
if (dp < endp)
*dp++ = *sp;
break;
default:
/* otherwise treat the % as not special */
if (dp < endp)
*dp++ = *sp;
break;
}
}
else
{
if (dp < endp)
*dp++ = *sp;
}
}
*dp = '\0';
ereport(DEBUG3,
(errmsg_internal("executing restore command \"%s\"",
xlogRestoreCmd)));
/*
* Check signals before restore command and reset afterwards.
*/
PreRestoreCommand();
/*
* Copy xlog from archival storage to XLOGDIR
*/
rc = system(xlogRestoreCmd);
PostRestoreCommand();
if (rc == 0)
{
/*
* command apparently succeeded, but let's make sure the file is
* really there now and has the correct size.
*/
if (stat(xlogpath, &stat_buf) == 0)
{
if (expectedSize > 0 && stat_buf.st_size != expectedSize)
{
int elevel;
/*
* If we find a partial file in standby mode, we assume it's
* because it's just being copied to the archive, and keep
* trying.
*
* Otherwise treat a wrong-sized file as FATAL to ensure the
* DBA would notice it, but is that too strong? We could try
* to plow ahead with a local copy of the file ... but the
* problem is that there probably isn't one, and we'd
* incorrectly conclude we've reached the end of WAL and we're
* done recovering ...
*/
if (StandbyMode && stat_buf.st_size < expectedSize)
elevel = DEBUG1;
else
elevel = FATAL;
ereport(elevel,
(errmsg("archive file \"%s\" has wrong size: %lu instead of %lu",
xlogfname,
(unsigned long) stat_buf.st_size,
(unsigned long) expectedSize)));
return false;
}
else
{
ereport(LOG,
(errmsg("restored log file \"%s\" from archive",
xlogfname)));
strcpy(path, xlogpath);
return true;
}
}
else
{
/* stat failed */
if (errno != ENOENT)
ereport(FATAL,
(errcode_for_file_access(),
errmsg("could not stat file \"%s\": %m",
xlogpath)));
}
}
/*
* Remember, we rollforward UNTIL the restore fails so failure here is
* just part of the process... that makes it difficult to determine
* whether the restore failed because there isn't an archive to restore,
* or because the administrator has specified the restore program
* incorrectly. We have to assume the former.
*
* However, if the failure was due to any sort of signal, it's best to
* punt and abort recovery. (If we "return false" here, upper levels will
* assume that recovery is complete and start up the database!) It's
* essential to abort on child SIGINT and SIGQUIT, because per spec
* system() ignores SIGINT and SIGQUIT while waiting; if we see one of
* those it's a good bet we should have gotten it too.
*
* On SIGTERM, assume we have received a fast shutdown request, and exit
* cleanly. It's pure chance whether we receive the SIGTERM first, or the
* child process. If we receive it first, the signal handler will call
* proc_exit, otherwise we do it here. If we or the child process received
* SIGTERM for any other reason than a fast shutdown request, postmaster
* will perform an immediate shutdown when it sees us exiting
* unexpectedly.
*
* Per the Single Unix Spec, shells report exit status > 128 when a called
* command died on a signal. Also, 126 and 127 are used to report
* problems such as an unfindable command; treat those as fatal errors
* too.
*/
if (WIFSIGNALED(rc) && WTERMSIG(rc) == SIGTERM)
proc_exit(1);
signaled = WIFSIGNALED(rc) || WEXITSTATUS(rc) > 125;
ereport(signaled ? FATAL : DEBUG2,
(errmsg("could not restore file \"%s\" from archive: return code %d",
xlogfname, rc)));
not_available:
/*
* if an archived file is not available, there might still be a version of
* this file in XLOGDIR, so return that as the filename to open.
*
* In many recovery scenarios we expect this to fail also, but if so that
* just means we've reached the end of WAL.
*/
snprintf(path, MAXPGPATH, XLOGDIR "/%s", xlogfname);
return false;
}
/*
* Main entry point for bgwriter process
*
* This is invoked from AuxiliaryProcessMain, which has already created the
* basic execution environment, but not enabled signals yet.
*/
void
BackgroundWriterMain(void)
{
sigjmp_buf local_sigjmp_buf;
MemoryContext bgwriter_context;
bool prev_hibernate;
WritebackContext wb_context;
/*
* Properly accept or ignore signals the postmaster might send us.
*
* bgwriter doesn't participate in ProcSignal signalling, but a SIGUSR1
* handler is still needed for latch wakeups.
*/
pqsignal(SIGHUP, BgSigHupHandler); /* set flag to read config file */
pqsignal(SIGINT, SIG_IGN);
pqsignal(SIGTERM, ReqShutdownHandler); /* shutdown */
pqsignal(SIGQUIT, bg_quickdie); /* hard crash time */
pqsignal(SIGALRM, SIG_IGN);
pqsignal(SIGPIPE, SIG_IGN);
pqsignal(SIGUSR1, bgwriter_sigusr1_handler);
pqsignal(SIGUSR2, SIG_IGN);
/*
* Reset some signals that are accepted by postmaster but not here
*/
pqsignal(SIGCHLD, SIG_DFL);
pqsignal(SIGTTIN, SIG_DFL);
pqsignal(SIGTTOU, SIG_DFL);
pqsignal(SIGCONT, SIG_DFL);
pqsignal(SIGWINCH, SIG_DFL);
/* We allow SIGQUIT (quickdie) at all times */
sigdelset(&BlockSig, SIGQUIT);
/*
* Create a resource owner to keep track of our resources (currently only
* buffer pins).
*/
CurrentResourceOwner = ResourceOwnerCreate(NULL, "Background Writer");
/*
* We just started, assume there has been either a shutdown or
* end-of-recovery snapshot.
*/
last_snapshot_ts = GetCurrentTimestamp();
/*
* Create a memory context that we will do all our work in. We do this so
* that we can reset the context during error recovery and thereby avoid
* possible memory leaks. Formerly this code just ran in
* TopMemoryContext, but resetting that would be a really bad idea.
*/
bgwriter_context = AllocSetContextCreate(TopMemoryContext,
"Background Writer",
ALLOCSET_DEFAULT_MINSIZE,
ALLOCSET_DEFAULT_INITSIZE,
ALLOCSET_DEFAULT_MAXSIZE);
MemoryContextSwitchTo(bgwriter_context);
WritebackContextInit(&wb_context, &bgwriter_flush_after);
/*
* If an exception is encountered, processing resumes here.
*
* See notes in postgres.c about the design of this coding.
*/
if (sigsetjmp(local_sigjmp_buf, 1) != 0)
{
/* Since not using PG_TRY, must reset error stack by hand */
error_context_stack = NULL;
/* Prevent interrupts while cleaning up */
HOLD_INTERRUPTS();
/* Report the error to the server log */
EmitErrorReport();
/*
* These operations are really just a minimal subset of
* AbortTransaction(). We don't have very many resources to worry
* about in bgwriter, but we do have LWLocks, buffers, and temp files.
*/
LWLockReleaseAll();
AbortBufferIO();
UnlockBuffers();
/* buffer pins are released here: */
ResourceOwnerRelease(CurrentResourceOwner,
RESOURCE_RELEASE_BEFORE_LOCKS,
false, true);
/* we needn't bother with the other ResourceOwnerRelease phases */
AtEOXact_Buffers(false);
AtEOXact_SMgr();
AtEOXact_Files();
AtEOXact_HashTables(false);
/*
* Now return to normal top-level context and clear ErrorContext for
* next time.
*/
MemoryContextSwitchTo(bgwriter_context);
FlushErrorState();
/* Flush any leaked data in the top-level context */
MemoryContextResetAndDeleteChildren(bgwriter_context);
/* re-initilialize to avoid repeated errors causing problems */
WritebackContextInit(&wb_context, &bgwriter_flush_after);
/* Now we can allow interrupts again */
RESUME_INTERRUPTS();
/*
* Sleep at least 1 second after any error. A write error is likely
* to be repeated, and we don't want to be filling the error logs as
* fast as we can.
*/
pg_usleep(1000000L);
/*
* Close all open files after any error. This is helpful on Windows,
* where holding deleted files open causes various strange errors.
* It's not clear we need it elsewhere, but shouldn't hurt.
*/
smgrcloseall();
/* Report wait end here, when there is no further possibility of wait */
pgstat_report_wait_end();
}
/* We can now handle ereport(ERROR) */
PG_exception_stack = &local_sigjmp_buf;
/*
* Unblock signals (they were blocked when the postmaster forked us)
*/
PG_SETMASK(&UnBlockSig);
/*
* Reset hibernation state after any error.
*/
prev_hibernate = false;
/*
* Loop forever
*/
for (;;)
{
bool can_hibernate;
int rc;
/* Clear any already-pending wakeups */
ResetLatch(MyLatch);
if (got_SIGHUP)
{
got_SIGHUP = false;
ProcessConfigFile(PGC_SIGHUP);
}
if (shutdown_requested)
{
/*
* From here on, elog(ERROR) should end with exit(1), not send
* control back to the sigsetjmp block above
*/
ExitOnAnyError = true;
/* Normal exit from the bgwriter is here */
proc_exit(0); /* done */
}
/*
* Do one cycle of dirty-buffer writing.
*/
can_hibernate = BgBufferSync(&wb_context);
/*
* Send off activity statistics to the stats collector
*/
pgstat_send_bgwriter();
if (FirstCallSinceLastCheckpoint())
{
/*
* After any checkpoint, close all smgr files. This is so we
* won't hang onto smgr references to deleted files indefinitely.
*/
smgrcloseall();
}
/*
* Log a new xl_running_xacts every now and then so replication can
* get into a consistent state faster (think of suboverflowed
* snapshots) and clean up resources (locks, KnownXids*) more
* frequently. The costs of this are relatively low, so doing it 4
* times (LOG_SNAPSHOT_INTERVAL_MS) a minute seems fine.
*
* We assume the interval for writing xl_running_xacts is
* significantly bigger than BgWriterDelay, so we don't complicate the
* overall timeout handling but just assume we're going to get called
* often enough even if hibernation mode is active. It's not that
* important that log_snap_interval_ms is met strictly. To make sure
* we're not waking the disk up unnecessarily on an idle system we
* check whether there has been any WAL inserted since the last time
* we've logged a running xacts.
*
* We do this logging in the bgwriter as its the only process that is
* run regularly and returns to its mainloop all the time. E.g.
* Checkpointer, when active, is barely ever in its mainloop and thus
* makes it hard to log regularly.
*/
if (XLogStandbyInfoActive() && !RecoveryInProgress())
{
TimestampTz timeout = 0;
TimestampTz now = GetCurrentTimestamp();
timeout = TimestampTzPlusMilliseconds(last_snapshot_ts,
LOG_SNAPSHOT_INTERVAL_MS);
/*
* only log if enough time has passed and some xlog record has
* been inserted.
*/
if (now >= timeout &&
last_snapshot_lsn != GetXLogInsertRecPtr())
{
XLogRecPtr log_standby_lsn = LogStandbySnapshot();
last_snapshot_ts = now;
if (!XLogRecPtrIsInvalid(log_standby_lsn))
last_snapshot_lsn = log_standby_lsn;
}
}
/*
* Sleep until we are signaled or BgWriterDelay has elapsed.
*
* Note: the feedback control loop in BgBufferSync() expects that we
* will call it every BgWriterDelay msec. While it's not critical for
* correctness that that be exact, the feedback loop might misbehave
* if we stray too far from that. Hence, avoid loading this process
* down with latch events that are likely to happen frequently during
* normal operation.
*/
rc = WaitLatch(MyLatch,
WL_LATCH_SET | WL_TIMEOUT | WL_POSTMASTER_DEATH,
BgWriterDelay /* ms */ );
/*
* If no latch event and BgBufferSync says nothing's happening, extend
* the sleep in "hibernation" mode, where we sleep for much longer
* than bgwriter_delay says. Fewer wakeups save electricity. When a
* backend starts using buffers again, it will wake us up by setting
* our latch. Because the extra sleep will persist only as long as no
* buffer allocations happen, this should not distort the behavior of
* BgBufferSync's control loop too badly; essentially, it will think
* that the system-wide idle interval didn't exist.
*
* There is a race condition here, in that a backend might allocate a
* buffer between the time BgBufferSync saw the alloc count as zero
* and the time we call StrategyNotifyBgWriter. While it's not
* critical that we not hibernate anyway, we try to reduce the odds of
* that by only hibernating when BgBufferSync says nothing's happening
* for two consecutive cycles. Also, we mitigate any possible
* consequences of a missed wakeup by not hibernating forever.
*/
if (rc == WL_TIMEOUT && can_hibernate && prev_hibernate)
{
/* Ask for notification at next buffer allocation */
StrategyNotifyBgWriter(MyProc->pgprocno);
/* Sleep ... */
rc = WaitLatch(MyLatch,
WL_LATCH_SET | WL_TIMEOUT | WL_POSTMASTER_DEATH,
BgWriterDelay * HIBERNATE_FACTOR);
/* Reset the notification request in case we timed out */
StrategyNotifyBgWriter(-1);
}
/*
* Emergency bailout if postmaster has died. This is to avoid the
* necessity for manual cleanup of all postmaster children.
*/
if (rc & WL_POSTMASTER_DEATH)
exit(1);
prev_hibernate = can_hibernate;
}
}
