Changeset: 5519f87af1df for MonetDB
URL: https://dev.monetdb.org/hg/MonetDB/rev/5519f87af1df
Modified Files:
gdk/gdk.h
gdk/gdk_bat.c
gdk/gdk_batop.c
Branch: Jul2021
Log Message:
Use more bat iterators to access data.
diffs (truncated from 1012 to 300 lines):
diff --git a/gdk/gdk.h b/gdk/gdk.h
--- a/gdk/gdk.h
+++ b/gdk/gdk.h
@@ -800,25 +800,6 @@ typedef struct BAT {
#define tprops T.props
-typedef struct BATiter {
- BAT *b;
- Heap *h;
- void *base;
- Heap *vh;
- BUN count;
- uint16_t width;
- uint8_t shift;
- int8_t type;
- oid tseq;
- union {
- oid tvid;
- bool tmsk;
- };
-#ifndef NDEBUG
- bool locked;
-#endif
-} BATiter;
-
/* some access functions for the bitmask type */
static inline void
mskSet(BAT *b, BUN p)
@@ -890,6 +871,26 @@ gdk_export size_t HEAPmemsize(Heap *h);
gdk_export void HEAPdecref(Heap *h, bool remove);
gdk_export void HEAPincref(Heap *h);
+/* BAT iterator, also protects use of BAT heaps with reference counts */
+typedef struct BATiter {
+ BAT *b;
+ Heap *h;
+ void *base;
+ Heap *vh;
+ BUN count;
+ uint16_t width;
+ uint8_t shift;
+ int8_t type;
+ oid tseq;
+ union {
+ oid tvid;
+ bool tmsk;
+ };
+#ifndef NDEBUG
+ bool locked;
+#endif
+} BATiter;
+
static inline BATiter
bat_iterator(BAT *b)
{
@@ -1114,6 +1115,8 @@ typedef var_t stridx_t;
static inline oid
BUNtoid(BAT *b, BUN p)
{
+ oid o;
+
assert(ATOMtype(b->ttype) == TYPE_oid);
/* BATcount is the number of valid entries, so with
* exceptions, the last value can well be larger than
@@ -1122,13 +1125,18 @@ BUNtoid(BAT *b, BUN p)
assert(b->ttype == TYPE_void || b->tvheap == NULL);
if (is_oid_nil(b->tseqbase)) {
if (b->ttype == TYPE_void)
- return b->tseqbase;
- return ((const oid *) b->theap->base)[p + b->tbaseoff];
+ return b->tseqbase; /* i.e. oid_nil */
+ MT_lock_set(&b->theaplock);
+ o = ((const oid *) b->theap->base)[p + b->tbaseoff];
+ MT_lock_unset(&b->theaplock);
+ return o;
}
- oid o = b->tseqbase + p;
+ o = b->tseqbase + p;
if (b->ttype == TYPE_oid || b->tvheap == NULL) {
return o;
}
+ /* b->tvheap != NULL, so we know there will be no parallel
+ * modifications (so no locking) */
assert(!mask_cand(b));
/* exceptions only allowed on transient BATs */
assert(b->batRole == TRANSIENT);
@@ -2000,10 +2008,16 @@ Tpos(BATiter *bi, BUN p)
return (void*)&bi->tvid;
}
+static inline bool
+Tmskval(BATiter *bi, BUN p)
+{
+ return ((uint32_t *) bi->h->base)[p / 32] & (1U << (p % 32));
+}
+
static inline void *
Tmsk(BATiter *bi, BUN p)
{
- bi->tmsk = ((uint32_t *) bi->h->base)[p / 32] & (1U << (p % 32));
+ bi->tmsk = Tmskval(bi, p);
return &bi->tmsk;
}
diff --git a/gdk/gdk_bat.c b/gdk/gdk_bat.c
--- a/gdk/gdk_bat.c
+++ b/gdk/gdk_bat.c
@@ -793,12 +793,11 @@ BAT *
COLcopy(BAT *b, int tt, bool writable, role_t role)
{
BUN bunstocopy = BUN_NONE;
- BUN cnt;
BAT *bn = NULL;
+ BATiter bi;
BATcheck(b, NULL);
assert(tt != TYPE_bat);
- cnt = b->batCount;
/* maybe a bit ugly to change the requested bat type?? */
if (b->ttype == TYPE_void && !writable)
@@ -809,6 +808,8 @@ COLcopy(BAT *b, int tt, bool writable, r
return NULL;
}
+ bi = bat_iterator(b);
+
/* first try case (1); create a view, possibly with different
* atom-types */
if (!writable &&
@@ -818,39 +819,41 @@ COLcopy(BAT *b, int tt, bool writable, r
(!VIEWtparent(b) ||
BBP_cache(VIEWtparent(b))->batRestricted == BAT_READ)) {
bn = VIEWcreate(b->hseqbase, b);
- if (bn == NULL)
+ if (bn == NULL) {
+ bat_iterator_end(&bi);
return NULL;
+ }
if (tt != bn->ttype) {
bn->ttype = tt;
bn->tvarsized = ATOMvarsized(tt);
- bn->tseqbase = ATOMtype(tt) == TYPE_oid ? b->tseqbase :
oid_nil;
+ bn->tseqbase = ATOMtype(tt) == TYPE_oid ? bi.tseq :
oid_nil;
}
} else {
/* check whether we need case (4); BUN-by-BUN copy (by
* setting bunstocopy != BUN_NONE) */
- if (ATOMsize(tt) != ATOMsize(b->ttype)) {
+ if (ATOMsize(tt) != ATOMsize(bi.type)) {
/* oops, void materialization */
- bunstocopy = cnt;
+ bunstocopy = bi.count;
} else if (BATatoms[tt].atomFix) {
/* oops, we need to fix/unfix atoms */
- bunstocopy = cnt;
- } else if (isVIEW(b)) {
+ bunstocopy = bi.count;
+ } else if (bi.h && bi.h->parentid != b->batCacheid) {
/* extra checks needed for views */
- bat tp = VIEWtparent(b);
-
- if (tp != 0 && BATcapacity(BBP_cache(tp)) > cnt + cnt)
+ if (BATcapacity(BBP_cache(bi.h->parentid)) > bi.count +
bi.count)
/* reduced slice view: do not copy too
* much garbage */
- bunstocopy = cnt;
+ bunstocopy = bi.count;
}
bn = COLnew(b->hseqbase, tt, MAX(1, bunstocopy == BUN_NONE ? 0
: bunstocopy), role);
- if (bn == NULL)
+ if (bn == NULL) {
+ bat_iterator_end(&bi);
return NULL;
+ }
if (bn->tvarsized && bn->ttype && bunstocopy == BUN_NONE) {
- bn->tshift = b->tshift;
- bn->twidth = b->twidth;
+ bn->tshift = bi.shift;
+ bn->twidth = bi.width;
if (HEAPextend(bn->theap, BATcapacity(bn) <<
bn->tshift, true) != GDK_SUCCEED)
goto bunins_failed;
}
@@ -865,11 +868,11 @@ COLcopy(BAT *b, int tt, bool writable, r
Heap bthp, thp;
bthp = (Heap) {
- .farmid = BBPselectfarm(role, b->ttype,
offheap),
+ .farmid = BBPselectfarm(role, bi.type, offheap),
.parentid = bn->batCacheid,
};
thp = (Heap) {
- .farmid = BBPselectfarm(role, b->ttype,
varheap),
+ .farmid = BBPselectfarm(role, bi.type, varheap),
.parentid = bn->batCacheid,
};
settailname(&bthp, BBP_physical(bn->batCacheid),
@@ -877,11 +880,12 @@ COLcopy(BAT *b, int tt, bool writable, r
strconcat_len(thp.filename, sizeof(thp.filename),
BBP_physical(bn->batCacheid),
".theap", NULL);
- if ((b->ttype && HEAPcopy(&bthp, b->theap, b->tbaseoff
<< b->tshift) != GDK_SUCCEED) ||
- (bn->tvheap && HEAPcopy(&thp, b->tvheap, 0) !=
GDK_SUCCEED)) {
+ if ((bi.type && HEAPcopy(&bthp, bi.h, (size_t) ((char
*) bi.base - bi.h->base)) != GDK_SUCCEED) ||
+ (bn->tvheap && HEAPcopy(&thp, bi.vh, 0) !=
GDK_SUCCEED)) {
HEAPfree(&thp, true);
HEAPfree(&bthp, true);
BBPreclaim(bn);
+ bat_iterator_end(&bi);
return NULL;
}
/* succeeded; replace dummy small heaps by the
@@ -900,7 +904,6 @@ COLcopy(BAT *b, int tt, bool writable, r
} else if (BATatoms[tt].atomFix || tt != TYPE_void ||
ATOMextern(tt)) {
/* case (4): one-by-one BUN insert (really slow) */
BUN p, q, r = 0;
- BATiter bi = bat_iterator(b);
BATloop(b, p, q) {
const void *t = BUNtail(bi, p);
@@ -911,12 +914,11 @@ COLcopy(BAT *b, int tt, bool writable, r
}
r++;
}
- bat_iterator_end(&bi);
bn->theap->dirty |= bunstocopy > 0;
- } else if (tt != TYPE_void && b->ttype == TYPE_void) {
+ } else if (tt != TYPE_void && bi.type == TYPE_void) {
/* case (4): optimized for unary void
* materialization */
- oid cur = b->tseqbase, *dst = (oid *) Tloc(bn, 0);
+ oid cur = bi.tseq, *dst = (oid *) Tloc(bn, 0);
oid inc = !is_oid_nil(cur);
bn->theap->free = bunstocopy * sizeof(oid);
@@ -925,27 +927,27 @@ COLcopy(BAT *b, int tt, bool writable, r
*dst++ = cur;
cur += inc;
}
- } else if (ATOMstorage(b->ttype) == TYPE_msk) {
+ } else if (ATOMstorage(bi.type) == TYPE_msk) {
/* convert number of bits to number of bytes,
* and round the latter up to a multiple of
* 4 (copy in units of 4 bytes) */
bn->theap->free = (bunstocopy + 7) / 8;
bn->theap->free = (bn->theap->free + 3) & ~(size_t)3;
bn->theap->dirty |= bunstocopy > 0;
- memcpy(Tloc(bn, 0), Tloc(b, 0), bn->theap->free);
+ memcpy(Tloc(bn, 0), bi.base, bn->theap->free);
} else {
/* case (4): optimized for simple array copy */
bn->theap->free = bunstocopy * Tsize(bn);
bn->theap->dirty |= bunstocopy > 0;
- memcpy(Tloc(bn, 0), Tloc(b, 0), bn->theap->free);
+ memcpy(Tloc(bn, 0), bi.base, bn->theap->free);
}
/* copy all properties (size+other) from the source bat */
- BATsetcount(bn, cnt);
+ BATsetcount(bn, bi.count);
}
/* set properties (note that types may have changed in the copy) */
- if (ATOMtype(tt) == ATOMtype(b->ttype)) {
+ if (ATOMtype(tt) == ATOMtype(bi.type)) {
if (ATOMtype(tt) == TYPE_oid) {
- BATtseqbase(bn, b->tseqbase);
+ BATtseqbase(bn, bi.tseq);
} else {
BATtseqbase(bn, oid_nil);
}
@@ -1003,57 +1005,14 @@ COLcopy(BAT *b, int tt, bool writable, r
bn->batRestricted = BAT_READ;
TRC_DEBUG(ALGO, ALGOBATFMT " -> " ALGOBATFMT "\n",
ALGOBATPAR(b), ALGOBATPAR(bn));
+ bat_iterator_end(&bi);
return bn;
bunins_failed:
+ bat_iterator_end(&bi);
BBPreclaim(bn);
return NULL;
}
-#ifdef HAVE_HGE
-#define un_move_sz16(src, dst, sz) \
- if (sz == 16) { \
- * (hge *) dst = * (hge *) src; \
- } else
-#else
-#define un_move_sz16(src, dst, sz) \
- if (sz == 16) { \
- * (uuid *) dst = * (uuid *) src; \
- } else
-#endif
-
-#define un_move(src, dst, sz) \
- do { \
- un_move_sz16(src,dst,sz) \
- if (sz == 8) { \
_______________________________________________
checkin-list mailing list
checkin-list@monetdb.org
https://www.monetdb.org/mailman/listinfo/checkin-list
