Movatterモバイル変換

[0]ホーム
URL:

Skip to content

Navigation Menu

Sign in
Appearance settings

Search code, repositories, users, issues, pull requests...

Provide feedback

We read every piece of feedback, and take your input very seriously.

Saved searches

Use saved searches to filter your results more quickly

 Sign up
Appearance settings

Commit76281aa

Browse files
committed
Avoid a couple of zero-divide scenarios in the planner.
cost_subplan() supposed that the given subplan must have plan_rows > 0,which as far as I can tell was true until recent refactoring of thecode in createplan.c; but now that code allows the Result for a provablyempty subquery to have plan_rows = 0. Rather than undo that change,put in a clamp to prevent zero divide.get_cheapest_fractional_path() likewise supposed that best_path->rows > 0.This assumption has been wrong for longer. It's actually harmless givenIEEE float math, because a positive value divided by zero gives +Infinityand compare_fractional_path_costs() will do the right thing with that.Still, best not to assume that.final_cost_nestloop() also seems to have some risks in this area, soborrow the clamping logic already present in the mergejoin cost functions.Lastly, remove unnecessary clamp_row_est() in planner.c's calls toget_number_of_groups(). The only thing that function does with path_rowsis pass it to estimate_num_groups() which already has an internal clamp,so we don't need the extra call; and if we did, the callers are arguablythe wrong place for it anyway.First two items reported by Piotr Stefaniak, the others are productsof my nosing around for similar problems. No back-patch since there'sno evidence that problems arise in the back branches.
1 parent676265e commit76281aa

File tree

2 files changed

+17
-12
lines changed

2 files changed

+17
-12
lines changed

‎src/backend/optimizer/path/costsize.c

Lines changed: 12 additions & 5 deletions
Original file line numberDiff line numberDiff line change
@@ -45,9 +45,10 @@
4545
*(total_cost - startup_cost) * tuples_to_fetch / path->rows;
4646
* Note that a base relation's rows count (and, by extension, plan_rows for
4747
* plan nodes below the LIMIT node) are set without regard to any LIMIT, so
48-
* that this equation works properly. (Also, these routines guarantee not to
49-
* set the rows count to zero, so there will be no zero divide.) The LIMIT is
50-
* applied as a top-level plan node.
48+
* that this equation works properly. (Note: while path->rows is never zero
49+
* for ordinary relations, it is zero for paths for provably-empty relations,
50+
* so beware of division-by-zero.) The LIMIT is applied as a top-level
51+
* plan node.
5152
*
5253
* For largely historical reasons, most of the routines in this module use
5354
* the passed result Path only to store their results (rows, startup_cost and
@@ -1991,6 +1992,12 @@ final_cost_nestloop(PlannerInfo *root, NestPath *path,
19911992
QualCostrestrict_qual_cost;
19921993
doublentuples;
19931994

1995+
/* Protect some assumptions below that rowcounts aren't zero or NaN */
1996+
if (outer_path_rows <=0||isnan(outer_path_rows))
1997+
outer_path_rows=1;
1998+
if (inner_path_rows <=0||isnan(inner_path_rows))
1999+
inner_path_rows=1;
2000+
19942001
/* Mark the path with the correct row estimate */
19952002
if (path->path.param_info)
19962003
path->path.rows=path->path.param_info->ppi_rows;
@@ -3025,8 +3032,8 @@ cost_subplan(PlannerInfo *root, SubPlan *subplan, Plan *plan)
30253032

30263033
if (subplan->subLinkType==EXISTS_SUBLINK)
30273034
{
3028-
/* we only need to fetch 1 tuple */
3029-
sp_cost.per_tuple+=plan_run_cost /plan->plan_rows;
3035+
/* we only need to fetch 1 tuple; clamp to avoid zero divide */
3036+
sp_cost.per_tuple+=plan_run_cost /clamp_row_est(plan->plan_rows);
30303037
}
30313038
elseif (subplan->subLinkType==ALL_SUBLINK||
30323039
subplan->subLinkType==ANY_SUBLINK)

‎src/backend/optimizer/plan/planner.c

Lines changed: 5 additions & 7 deletions
Original file line numberDiff line numberDiff line change
@@ -3346,12 +3346,10 @@ create_grouping_paths(PlannerInfo *root,
33463346
}
33473347

33483348
/*
3349-
* Estimate number of groups. Note: if cheapest_path is a dummy, it will
3350-
* have zero rowcount estimate, which we don't want to use for fear of
3351-
* divide-by-zero. Hence clamp.
3349+
* Estimate number of groups.
33523350
*/
33533351
dNumGroups=get_number_of_groups(root,
3354-
clamp_row_est(cheapest_path->rows),
3352+
cheapest_path->rows,
33553353
rollup_lists,
33563354
rollup_groupclauses);
33573355

@@ -3415,7 +3413,7 @@ create_grouping_paths(PlannerInfo *root,
34153413

34163414
/* Estimate number of partial groups. */
34173415
dNumPartialGroups=get_number_of_groups(root,
3418-
clamp_row_est(cheapest_partial_path->rows),
3416+
cheapest_partial_path->rows,
34193417
NIL,
34203418
NIL);
34213419

@@ -4840,8 +4838,8 @@ get_cheapest_fractional_path(RelOptInfo *rel, double tuple_fraction)
48404838
if (tuple_fraction <=0.0)
48414839
returnbest_path;
48424840

4843-
/* Convert absolute # of tuples to a fraction; no need to clamp */
4844-
if (tuple_fraction >=1.0)
4841+
/* Convert absolute # of tuples to a fraction; no need to clampto 0..1*/
4842+
if (tuple_fraction >=1.0&&best_path->rows>0)
48454843
tuple_fraction /=best_path->rows;
48464844

48474845
foreach(l,rel->pathlist)

0 commit comments

Comments
 (0)
[8]ページ先頭
©2009-2025 Movatter.jp