Changeset: 8811b0f7e1c7 for MonetDB
URL: https://dev.monetdb.org/hg/MonetDB/rev/8811b0f7e1c7
Modified Files:
gdk/gdk_bbp.c
gdk/gdk_join.c
gdk/gdk_private.h
monetdb5/modules/mal/tablet.c
sql/storage/objectset.c
sql/storage/store.c
Branch: default
Log Message:
Merge with Jun2023 branch.
diffs (truncated from 417 to 300 lines):
diff --git a/gdk/gdk_bbp.c b/gdk/gdk_bbp.c
--- a/gdk/gdk_bbp.c
+++ b/gdk/gdk_bbp.c
@@ -3844,10 +3844,10 @@ BBPsync(int cnt, bat *restrict subcommit
assert(sizes == NULL || bi.width == 0 || (bi.type ==
TYPE_msk ? ((size + 31) / 32) * 4 : size << bi.shift) <= bi.hfree);
if (size > bi.count) /* includes sizes==NULL */
size = bi.count;
+ MT_lock_set(&bi.b->theaplock);
bi.b->batInserted = size;
if (bi.b->ttype >= 0 && ATOMvarsized(bi.b->ttype)) {
/* see epilogue() for other part of this */
- MT_lock_set(&bi.b->theaplock);
/* remember the tail we're saving */
if (BATsetprop_nolock(bi.b, (enum prop_t) 20,
TYPE_ptr, &bi.h) == NULL) {
GDKerror("setprop failed\n");
@@ -3857,8 +3857,8 @@ BBPsync(int cnt, bat *restrict subcommit
bi.b->oldtail = (Heap *) 1;
HEAPincref(bi.h);
}
- MT_lock_unset(&bi.b->theaplock);
}
+ MT_lock_unset(&bi.b->theaplock);
if (ret == GDK_SUCCEED && b && size != 0) {
/* wait for BBPSAVING so that we
* can set it, wait for
diff --git a/gdk/gdk_cand.c b/gdk/gdk_cand.c
--- a/gdk/gdk_cand.c
+++ b/gdk/gdk_cand.c
@@ -405,6 +405,14 @@ void
canditer_init(struct canditer *ci, BAT *b, BAT *s)
{
assert(ci != NULL);
+ BUN batcount = 0;
+ oid hseq = 0;
+ if (b) {
+ MT_lock_set(&b->theaplock);
+ batcount = BATcount(b);
+ hseq = b->hseqbase;
+ MT_lock_unset(&b->theaplock);
+ }
if (s == NULL) {
if (b == NULL) {
@@ -417,9 +425,9 @@ canditer_init(struct canditer *ci, BAT *
/* every row is a candidate */
*ci = (struct canditer) {
.tpe = cand_dense,
- .seq = b->hseqbase,
- .hseq = b->hseqbase,
- .ncand = BATcount(b),
+ .seq = hseq,
+ .hseq = hseq,
+ .ncand = batcount,
};
return;
}
@@ -432,7 +440,7 @@ canditer_init(struct canditer *ci, BAT *
BUN cnt = BATcount(s);
- if (cnt == 0 || (b != NULL && BATcount(b) == 0)) {
+ if (cnt == 0 || (b != NULL && batcount == 0)) {
/* candidate list for empty BAT or empty candidate list */
*ci = (struct canditer) {
.tpe = cand_dense,
@@ -488,7 +496,7 @@ canditer_init(struct canditer *ci, BAT *
switch (ci->tpe) {
case cand_materialized:
if (b != NULL) {
- BUN p = binsearchcand(ci->oids, cnt - 1U, b->hseqbase);
+ BUN p = binsearchcand(ci->oids, cnt - 1U, hseq);
/* p == cnt means candidate list is completely
* before b */
ci->offset = p;
@@ -496,7 +504,7 @@ canditer_init(struct canditer *ci, BAT *
cnt -= p;
if (cnt > 0) {
cnt = binsearchcand(ci->oids, cnt - 1U,
- b->hseqbase + BATcount(b));
+ hseq + batcount);
/* cnt == 0 means candidate list is
* completely after b */
}
@@ -532,8 +540,8 @@ canditer_init(struct canditer *ci, BAT *
ci->oids[ci->nvals - 1U] == ci->seq + cnt + ci->nvals -
1U)
ci->nvals--;
if (b != NULL) {
- if (ci->seq + cnt + ci->nvals <= b->hseqbase ||
- ci->seq >= b->hseqbase + BATcount(b)) {
+ if (ci->seq + cnt + ci->nvals <= hseq ||
+ ci->seq >= hseq + batcount) {
/* candidate list does not overlap with b */
*ci = (struct canditer) {
.tpe = cand_dense,
@@ -546,33 +554,33 @@ canditer_init(struct canditer *ci, BAT *
if (b == NULL)
break;
BUN p;
- p = binsearchcand(ci->oids, ci->nvals - 1U,
b->hseqbase);
+ p = binsearchcand(ci->oids, ci->nvals - 1U, hseq);
if (p == ci->nvals) {
/* all exceptions before start of b */
- ci->offset = b->hseqbase - ci->seq - ci->nvals;
- cnt = ci->seq + cnt + ci->nvals - b->hseqbase;
- ci->seq = b->hseqbase;
+ ci->offset = hseq - ci->seq - ci->nvals;
+ cnt = ci->seq + cnt + ci->nvals - hseq;
+ ci->seq = hseq;
ci->nvals = 0;
ci->tpe = cand_dense;
ci->oids = NULL;
break;
}
- assert(b->hseqbase > ci->seq || p == 0);
- if (b->hseqbase > ci->seq) {
+ assert(hseq > ci->seq || p == 0);
+ if (hseq > ci->seq) {
/* skip candidates, possibly including
* exceptions */
ci->oids += p;
ci->nvals -= p;
- p = b->hseqbase - ci->seq - p;
+ p = hseq - ci->seq - p;
cnt -= p;
ci->offset += p;
- ci->seq = b->hseqbase;
+ ci->seq = hseq;
}
- if (ci->seq + cnt + ci->nvals > b->hseqbase +
BATcount(b)) {
+ if (ci->seq + cnt + ci->nvals > hseq + batcount) {
p = binsearchcand(ci->oids, ci->nvals - 1U,
- b->hseqbase + BATcount(b));
+ hseq + batcount);
ci->nvals = p;
- cnt = b->hseqbase + BATcount(b) - ci->seq -
ci->nvals;
+ cnt = hseq + batcount - ci->seq - ci->nvals;
}
while (ci->nvals > 0 && ci->oids[0] == ci->seq) {
ci->nvals--;
@@ -590,8 +598,8 @@ canditer_init(struct canditer *ci, BAT *
/* fall through */
case cand_dense:
if (b != NULL) {
- if (ci->seq + cnt <= b->hseqbase ||
- ci->seq >= b->hseqbase + BATcount(b)) {
+ if (ci->seq + cnt <= hseq ||
+ ci->seq >= hseq + batcount) {
/* no overlap */
*ci = (struct canditer) {
.tpe = cand_dense,
@@ -599,20 +607,20 @@ canditer_init(struct canditer *ci, BAT *
};
return;
}
- if (b->hseqbase > ci->seq) {
- cnt -= b->hseqbase - ci->seq;
- ci->offset += b->hseqbase - ci->seq;
- ci->seq = b->hseqbase;
+ if (hseq > ci->seq) {
+ cnt -= hseq - ci->seq;
+ ci->offset += hseq - ci->seq;
+ ci->seq = hseq;
}
- if (ci->seq + cnt > b->hseqbase + BATcount(b))
- cnt = b->hseqbase + BATcount(b) - ci->seq;
+ if (ci->seq + cnt > hseq + batcount)
+ cnt = hseq + batcount - ci->seq;
}
break;
case cand_mask:
assert(s->tseqbase != oid_nil);
if (b != NULL) {
- if (ci->seq + cnt <= b->hseqbase ||
- ci->seq >= b->hseqbase + BATcount(b)) {
+ if (ci->seq + cnt <= hseq ||
+ ci->seq >= hseq + batcount) {
/* no overlap */
*ci = (struct canditer) {
.tpe = cand_dense,
@@ -620,15 +628,15 @@ canditer_init(struct canditer *ci, BAT *
};
return;
}
- if (b->hseqbase > ci->seq) {
- cnt = b->hseqbase - ci->seq;
+ if (hseq > ci->seq) {
+ cnt = hseq - ci->seq;
ci->mask += cnt / 32U;
ci->firstbit = (uint8_t) (cnt % 32U);
cnt = BATcount(s) - cnt;
- ci->seq = b->hseqbase;
+ ci->seq = hseq;
}
- if (ci->seq + cnt > b->hseqbase + BATcount(b)) {
- cnt = b->hseqbase + BATcount(b) - ci->seq;
+ if (ci->seq + cnt > hseq + batcount) {
+ cnt = hseq + batcount - ci->seq;
}
ci->nvals = (ci->firstbit + cnt + 31U) / 32U;
}
diff --git a/gdk/gdk_join.c b/gdk/gdk_join.c
--- a/gdk/gdk_join.c
+++ b/gdk/gdk_join.c
@@ -3288,22 +3288,26 @@ guess_uniques(BAT *b, struct canditer *c
BUN cnt1, cnt2;
BAT *s1;
- if (b->tkey)
+ MT_lock_set(&b->theaplock);
+ bool key = b->tkey;
+ double unique_est = b->tunique_est;
+ BUN batcount = BATcount(b);
+ MT_lock_unset(&b->theaplock);
+ if (key)
return (double) ci->ncand;
if (ci->s == NULL ||
- (ci->tpe == cand_dense && ci->ncand == BATcount(b))) {
- MT_lock_set(&b->theaplock);
- double unique_est = b->tunique_est;
- MT_lock_unset(&b->theaplock);
+ (ci->tpe == cand_dense && ci->ncand == batcount)) {
if (unique_est != 0) {
TRC_DEBUG(ALGO, "b=" ALGOBATFMT " use cached value\n",
ALGOBATPAR(b));
return unique_est;
}
- s1 = BATsample_with_seed(b, 1000, (uint64_t) GDKusec() *
(uint64_t) b->batCacheid);
+ s1 = BATcreatesample(b->hseqbase, batcount, 1000,
+ (uint64_t) GDKusec() * (uint64_t)
b->batCacheid);
} else {
- BAT *s2 = BATsample_with_seed(ci->s, 1000, (uint64_t) GDKusec()
* (uint64_t) b->batCacheid);
+ BAT *s2 = BATcreatesample(ci->s->hseqbase, ci->ncand, 1000,
+ (uint64_t) GDKusec() * (uint64_t)
b->batCacheid);
if (s2 == NULL)
return -1;
s1 = BATproject(s2, ci->s);
@@ -3323,13 +3327,13 @@ guess_uniques(BAT *b, struct canditer *c
double B = cnt1 - n1 * A;
B += A * ci->ncand;
+ MT_lock_set(&b->theaplock);
if (ci->s == NULL ||
- (ci->tpe == cand_dense && ci->ncand == BATcount(b))) {
- MT_lock_set(&b->theaplock);
+ (ci->tpe == cand_dense && ci->ncand == BATcount(b) && ci->ncand ==
batcount)) {
if (b->tunique_est == 0)
b->tunique_est = B;
- MT_lock_unset(&b->theaplock);
}
+ MT_lock_unset(&b->theaplock);
return B;
}
diff --git a/gdk/gdk_private.h b/gdk/gdk_private.h
--- a/gdk/gdk_private.h
+++ b/gdk/gdk_private.h
@@ -58,6 +58,8 @@ gdk_return BATcheckmodes(BAT *b, bool pe
__attribute__((__visibility__("hidden")));
BAT *BATcreatedesc(oid hseq, int tt, bool heapnames, role_t role, uint16_t
width)
__attribute__((__visibility__("hidden")));
+BAT *BATcreatesample(oid hseq, BUN cnt, BUN n, uint64_t seed)
+ __attribute__((__visibility__("hidden")));
void BATdelete(BAT *b)
__attribute__((__visibility__("hidden")));
void BATdestroy(BAT *b)
diff --git a/gdk/gdk_sample.c b/gdk/gdk_sample.c
--- a/gdk/gdk_sample.c
+++ b/gdk/gdk_sample.c
@@ -98,26 +98,24 @@ OIDTreeToBATAntiset(struct oidtreenode *
}
static BAT *
-do_batsample(BAT *b, BUN n, random_state_engine rse, MT_Lock *lock)
+do_batsample(oid hseq, BUN cnt, BUN n, random_state_engine rse, MT_Lock *lock)
{
BAT *bn;
- BUN cnt, slen;
+ BUN slen;
BUN rescnt;
struct oidtreenode *tree = NULL;
- BATcheck(b, NULL);
ERRORcheck(n > BUN_MAX, "sample size larger than BUN_MAX\n", NULL);
- cnt = BATcount(b);
/* empty sample size */
if (n == 0) {
bn = BATdense(0, 0, 0);
} else if (cnt <= n) {
/* sample size is larger than the input BAT, return
* all oids */
- bn = BATdense(0, b->hseqbase, cnt);
+ bn = BATdense(0, hseq, cnt);
} else {
- oid minoid = b->hseqbase;
- oid maxoid = b->hseqbase + cnt;
+ oid minoid = hseq;
+ oid maxoid = hseq + cnt;
/* if someone samples more than half of our tree, we
_______________________________________________
checkin-list mailing list -- checkin-list@monetdb.org
To unsubscribe send an email to checkin-list-le...@monetdb.org
