Changeset: 6e85b1747188 for MonetDB
URL: http://dev.monetdb.org/hg/MonetDB?cmd=changeset;node=6e85b1747188
Modified Files:
clients/mapiclient/tomograph.c
gdk/gdk_atoms.c
gdk/gdk_bbp.c
gdk/gdk_heap.c
monetdb5/mal/mal_factory.c
monetdb5/mal/mal_instruction.h
monetdb5/mal/mal_profiler.c
monetdb5/mal/mal_readline.c
monetdb5/modules/mal/tablet.c
monetdb5/optimizer/opt_qep.c
Branch: default
Log Message:
Merge with Feb2013 branch.
diffs (truncated from 382 to 300 lines):
diff --git a/clients/mapiclient/tomograph.c b/clients/mapiclient/tomograph.c
--- a/clients/mapiclient/tomograph.c
+++ b/clients/mapiclient/tomograph.c
@@ -2157,7 +2157,7 @@ doProfile(void *d)
for (i = 0; i < MAXTHREADS; i++)
threads[i] = topbox++;
- /* sent single query */
+ /* send single query */
if (sqlstatement) {
doRequest(dbhsql, sqlstatement);
}
diff --git a/gdk/gdk_atoms.c b/gdk/gdk_atoms.c
--- a/gdk/gdk_atoms.c
+++ b/gdk/gdk_atoms.c
@@ -963,6 +963,9 @@ strHash(const char *s)
return res;
}
+/* if at least (2*SIZEOF_BUN), also store length (heaps are then
+ * incompatible) */
+#define EXTRALEN ((SIZEOF_BUN + GDK_VARALIGN - 1) & ~(GDK_VARALIGN - 1))
void
strCleanHash(Heap *h, int rebuild)
@@ -971,6 +974,50 @@ strCleanHash(Heap *h, int rebuild)
if (!GDK_ELIMDOUBLES(h)) {
/* flush hash table for security */
memset(h->base, 0, GDK_STRHASHSIZE);
+ } else {
+ /* rebuild hash table for double elimination
+ *
+ * If appending strings to the BAT was aborted, if the
+ * heap was memory mapped, the hash in the string heap
+ * may well be incorrect. Therefore we don't trust it
+ * when we read in a string heap and we rebuild the
+ * complete table (it is small, so this won't take any
+ * time at all). */
+ size_t pad, pos;
+ const size_t extralen = h->hashash ? EXTRALEN : 0;
+ stridx_t *bucket;
+ BUN off, strhash;
+ const char *s;
+
+ memset(h->base, 0, GDK_STRHASHSIZE);
+ pos = GDK_STRHASHSIZE;
+ while (pos < h->free) {
+ pad = GDK_VARALIGN - (pos & (GDK_VARALIGN - 1));
+ if (pad < sizeof(stridx_t))
+ pad += GDK_VARALIGN;
+ pos += pad + extralen;
+ s = h->base + pos;
+ if (h->hashash)
+ strhash = ((const BUN *) s)[-1];
+ else
+ GDK_STRHASH(s, strhash);
+ off = strhash & GDK_STRHASHMASK;
+ bucket = ((stridx_t *) h->base) + off;
+ *bucket = pos - extralen - sizeof(stridx_t);
+ pos += GDK_STRLEN(s);
+ }
+#ifndef NDEBUG
+ pos = GDK_STRHASHSIZE;
+ while (pos < h->free) {
+ pad = GDK_VARALIGN - (pos & (GDK_VARALIGN - 1));
+ if (pad < sizeof(stridx_t))
+ pad += GDK_VARALIGN;
+ pos += pad + extralen;
+ s = h->base + pos;
+ assert(strLocate(h, s) != 0);
+ pos += GDK_STRLEN(s);
+ }
+#endif
}
}
@@ -979,15 +1026,11 @@ strCleanHash(Heap *h, int rebuild)
* the location of a string in the heap if it exists. Otherwise it
* returns zero.
*/
-/* if at least (2*SIZEOF_BUN), also store length (heaps are then
- * incompatible) */
-#define EXTRALEN ((SIZEOF_BUN + GDK_VARALIGN - 1) & ~(GDK_VARALIGN - 1))
-
var_t
strLocate(Heap *h, const char *v)
{
stridx_t *ref, *next;
- size_t extralen = h->hashash ? EXTRALEN : 0;
+ const size_t extralen = h->hashash ? EXTRALEN : 0;
/* search hash-table, if double-elimination is still in place */
BUN off;
@@ -1012,7 +1055,7 @@ strPut(Heap *h, var_t *dst, const char *
size_t elimbase = GDK_ELIMBASE(h->free);
size_t pad = GDK_VARALIGN - (h->free & (GDK_VARALIGN - 1));
size_t pos, len = GDK_STRLEN(v);
- size_t extralen = h->hashash ? EXTRALEN : 0;
+ const size_t extralen = h->hashash ? EXTRALEN : 0;
stridx_t *bucket, *ref, *next;
BUN off, strhash;
@@ -1021,24 +1064,30 @@ strPut(Heap *h, var_t *dst, const char *
off &= GDK_STRHASHMASK;
bucket = ((stridx_t *) h->base) + off;
- /* search hash-table, if double-elimination is still in place */
- if (elimbase == 0) { /* small string heap (<64KB) -- fully double
eliminated */
- for (ref = bucket; *ref; ref = next) { /* search the linked
list */
+ /* if double-elimination is still in place, search hash-table */
+ if (elimbase == 0) {
+ /* small string heap (<64KB) -- fully double eliminated */
+ for (ref = bucket; *ref; ref = next) {
+ /* search the linked list */
next = (stridx_t *) (h->base + *ref);
- if (GDK_STRCMP(v, (str) (next + 1) + extralen) == 0) {
/* found */
+ if (GDK_STRCMP(v, (str) (next + 1) + extralen) == 0) {
+ /* found */
pos = sizeof(stridx_t) + *ref + extralen;
return *dst = (var_t) (pos >> GDK_VARSHIFT);
}
}
/* is there room for the next pointer in the padding space? */
- if (pad < sizeof(stridx_t))
- pad += GDK_VARALIGN; /* if not, pad more */
+ if (pad < sizeof(stridx_t)) {
+ /* if not, pad more */
+ pad += GDK_VARALIGN;
+ }
} else if (*bucket) {
/* large string heap (>=64KB) --
* opportunistic/probabilistic double elimination */
pos = elimbase + *bucket + extralen;
if (GDK_STRCMP(v, h->base + pos) == 0) {
- return *dst = (var_t) (pos >> GDK_VARSHIFT); /*
already in heap; do not insert! */
+ /* already in heap; do not insert! */
+ return *dst = (var_t) (pos >> GDK_VARSHIFT);
}
#if SIZEOF_VAR_T >= SIZEOF_VOID_P /* in fact SIZEOF_VAR_T == SIZEOF_VOID_P */
if (extralen == 0)
diff --git a/gdk/gdk_bbp.c b/gdk/gdk_bbp.c
--- a/gdk/gdk_bbp.c
+++ b/gdk/gdk_bbp.c
@@ -3165,8 +3165,14 @@ do_backup(const char *srcdir, const char
int ret = 0;
/* direct mmap is unprotected (readonly usage, or has WAL
- * protection) */
- if (h->storage != STORE_MMAP) {
+ * protection); however, if we're backing up for subcommit
+ * and a backup already exists in the main backup directory
+ * (see GDKupgradevarheap), move the file */
+ if (subcommit && file_exists(BAKDIR, nme, extbase)) {
+ assert(h->storage == STORE_MMAP);
+ if (file_move(BAKDIR, SUBDIR, nme, extbase))
+ return -1;
+ } else if (h->storage != STORE_MMAP) {
/* STORE_PRIV saves into X.new files. Two cases could
* happen. The first is when a valid X.new exists
* because of an access change or a previous
diff --git a/gdk/gdk_heap.c b/gdk/gdk_heap.c
--- a/gdk/gdk_heap.c
+++ b/gdk/gdk_heap.c
@@ -390,6 +390,20 @@ HEAPshrink(Heap *h, size_t size)
return -1;
}
+/* returns 1 if the file exists */
+static int
+file_exists(const char *dir, const char *name, const char *ext)
+{
+ long_str path;
+ struct stat st;
+ int ret;
+
+ GDKfilepath(path, dir, name, ext);
+ ret = stat(path, &st);
+ IODEBUG THRprintf(GDKstdout, "#stat(%s) = %d\n", path, ret);
+ return (ret == 0);
+}
+
int
GDKupgradevarheap(COLrec *c, var_t v, int copyall)
{
@@ -403,6 +417,7 @@ GDKupgradevarheap(COLrec *c, var_t v, in
#endif
size_t i, n;
size_t savefree;
+ const char *filename;
assert(c->heap.parentid == 0);
assert(width != 0);
@@ -414,11 +429,65 @@ GDKupgradevarheap(COLrec *c, var_t v, in
}
assert(c->width < width);
assert(c->shift < shift);
+
/* if copyall is set, we need to convert the whole heap, since
* we may be in the middle of an insert loop that adjusts the
* free value at the end; otherwise only copy the area
* indicated by the "free" pointer */
n = (copyall ? c->heap.size : c->heap.free) >> c->shift;
+
+ /* for memory mapped files, create a backup copy before widening
+ *
+ * this solves a problem that we don't control what's in the
+ * actual file until the next commit happens, so a crash might
+ * otherwise leave the file (and the database) in an
+ * inconsistent state
+ *
+ * also see do_backup in gdk_bbp.c */
+ filename = strrchr(c->heap.filename, DIR_SEP);
+ if (filename == NULL)
+ filename = c->heap.filename;
+ else
+ filename++;
+ if (c->heap.storage == STORE_MMAP && !file_exists(BAKDIR, filename,
NULL)) {
+ int fd;
+ ssize_t ret = 0;
+ size_t size = n << c->shift;
+ const char *base = c->heap.base;
+
+ /* first save heap in file with extra .tmp extension */
+ if ((fd = GDKfdlocate(c->heap.filename, "wb", "tmp")) < 0)
+ return GDK_FAIL;
+ while (size > 0) {
+ ret = write(fd, base, (unsigned) MIN(1 << 30, size));
+ if (ret < 0)
+ size = 0;
+ size -= ret;
+ base += ret;
+ }
+ if (ret < 0 ||
+#if defined(NATIVE_WIN32)
+ _commit(fd) < 0 ||
+#elif defined(HAVE_FDATASYNC)
+ fdatasync(fd) < 0 ||
+#elif defined(HAVE_FSYNC)
+ fsync(fd) < 0 ||
+#endif
+ close(fd) < 0) {
+ /* something went wrong: abandon ship */
+ close(fd);
+ GDKunlink(BATDIR, c->heap.filename, "tmp");
+ return GDK_FAIL;
+ }
+ /* move tmp file to backup directory (without .tmp
+ * extension) */
+ if (GDKmove(BATDIR, c->heap.filename, "tmp", BAKDIR, filename,
NULL) < 0) {
+ /* backup failed */
+ GDKunlink(BATDIR, c->heap.filename, "tmp");
+ return GDK_FAIL;
+ }
+ }
+
savefree = c->heap.free;
if (copyall)
c->heap.free = c->heap.size;
diff --git a/monetdb5/mal/mal_factory.c b/monetdb5/mal/mal_factory.c
--- a/monetdb5/mal/mal_factory.c
+++ b/monetdb5/mal/mal_factory.c
@@ -122,7 +122,7 @@
* end random;
* @end example
*
- * The operators to built client aware factories are,
+ * The operators to build client aware factories are,
* @sc{factories.getCaller()}, which returns a client
* index, @sc{factories.getModule()} and @sc{factories.getFunction()},
* which returns the identity of scope enclosed.
diff --git a/monetdb5/mal/mal_instruction.h b/monetdb5/mal/mal_instruction.h
--- a/monetdb5/mal/mal_instruction.h
+++ b/monetdb5/mal/mal_instruction.h
@@ -137,7 +137,7 @@ typedef struct MALBLK {
ProfPtr profiler;
struct MALBLK *history; /* of optimizer actions */
short keephistory; /* do we need the history at
all */
- short dotfile; /* sent dot file to
stethoscope? */
+ short dotfile; /* send dot file to
stethoscope? */
str marker; /* history points are
marked for backtracking */
int maxarg; /* keep track on the
maximal arguments used */
ptr replica; /* for the replicator tests */
diff --git a/monetdb5/mal/mal_profiler.c b/monetdb5/mal/mal_profiler.c
--- a/monetdb5/mal/mal_profiler.c
+++ b/monetdb5/mal/mal_profiler.c
@@ -203,7 +203,7 @@ deactivateCounter(str name)
* Offline processing
* The offline processing structure is the easiest. We merely have to
* produce a correct tuple format for the front-end.
- * To avoid unnecessary locks we first built the event as a string
+ * To avoid unnecessary locks we first build the event as a string
* It uses a local logbuffer[LOGLEN] and logbase, logtop, loglen
*/
#define LOGLEN 8192
diff --git a/monetdb5/mal/mal_readline.c b/monetdb5/mal/mal_readline.c
--- a/monetdb5/mal/mal_readline.c
+++ b/monetdb5/mal/mal_readline.c
@@ -61,7 +61,7 @@
* using the commands @sc{manual.help}.
* Keyword based lookup is supported by the operation @sc{manual.search};
_______________________________________________
checkin-list mailing list
checkin-list@monetdb.org
https://www.monetdb.org/mailman/listinfo/checkin-list
