Konstantin Bereznyakov created HIVE-29597:
---------------------------------------------
Summary: Vectorization: Scratch column reuse sometimes produces
wrong results
Key: HIVE-29597
URL: https://issues.apache.org/jira/browse/HIVE-29597
Project: Hive
Issue Type: Bug
Reporter: Konstantin Bereznyakov
## Description
The vectorizer can assign the same physical `LongColumnVector` slot to two
different logical columns: an intermediate scratch result (e.g. `cast(STRING AS
INT)`) and a small-side broadcast value column carried by a
`VectorMapJoinOuter*Operator`. The intermediate writes a typed value (an
`int`); the broadcast then writes a typed value of a different logical type
(e.g. a `boolean`) to the same slot for matched rows.
For *unmatched* rows the broadcast write is skipped — only `isNull[i]` is
flipped — and the slot retains the stale intermediate value. Any downstream
operator that reads `vector[i]` without first checking `isNull[i]` then
propagates a value of the wrong logical type into the query result. The bug is
silent.
Empirically reproduced on Apache Hive 4.x master at HEAD `6f2355e22c`. The
repro below uses the projection-mode `ColOrCol` path; analogous projection-mode
operators that read `vector[i]` directly (`ColAndCol`, long-column arithmetic,
equality/comparison expressions, `VectorUDFAdaptor`-routed UDFs) are exposed by
the same root cause but are not separately verified by this filing.
## When does the bug fire?
All of the following must hold:
1. Vectorized execution is on (default).
2. A vectorized OUTER-join MapJoin runs (`VectorMapJoinOuter*Operator`). INNER
/ LEFT SEMI / ANTI variants do not propagate stale slot data through unmatched
rows.
3. The vectorizer aliases a scratch column with a `smallTableValueMapping`
target — typical when a `LongColumnVector`-backed type (e.g. `int`) is computed
for the join-key expression and a `LongColumnVector`-backed broadcast value
(e.g. `boolean`) is also produced from the small side.
4. Some output row is unmatched on the join keys, going through
`generateOuterNulls`.
5. A downstream operator reads `vector[i]` of the affected slot without
consulting `isNull[i]`.
6. The reader's contract is violated by the dirty value (e.g. a strict `== 1`
check fails on a non-{0,1} long).
Removing any one condition masks the bug.
## Reproduction
```sql
SET hive.auto.convert.join=true; -- default true
SET hive.auto.convert.join.noconditionaltask=true; -- default true
CREATE TABLE t (k STRING, v STRING) STORED AS ORC;
INSERT INTO t VALUES
('A','1'),('A','2'),('A','3'),
('B','2'),('B','3'),
('C','3'),
('D','1'),('D','3');
WITH
probe AS (
SELECT k, v, (CAST(v AS INT) > 0) AS p_bool
FROM t WHERE CAST(v AS INT) >= 3
),
small_side AS (
SELECT k, v, (CAST(v AS INT) > 9999) AS s_bool
FROM t
),
classified AS (
SELECT p.k, p.v, CAST((s.s_bool OR p.p_bool) AS INT) AS observed_value
FROM probe p
LEFT JOIN small_side s
ON p.k = s.k
AND CAST(p.v AS INT) - 1 = CAST(s.v AS INT)
),
diagnosed AS (
-- Aggregate barrier: prevents Calcite from inlining the inner expression
-- back into the outer WHERE and simplifying the bug surface away.
SELECT k, v, MAX(observed_value) AS observed_value
FROM classified
GROUP BY k, v
)
SELECT k, v,
observed_value AS observed_value_returned_by_select,
1 AS required_value_per_sql_semantics
FROM diagnosed
WHERE observed_value = 0
ORDER BY k, v;
```
**Expected**: zero rows. Every probe row's `(s_bool OR p_bool)` is `TRUE` per
SQL semantics (matched: `FALSE OR TRUE`; unmatched: `NULL OR TRUE`), so
`cast(... AS INT) = 1` for every row, and no row can satisfy `WHERE
observed_value = 0`.
**Actual**:
```
C 3 0 1
D 3 0 1
```
Two rows are returned. The filter `WHERE observed_value = 0` accepts them
because their stored `observed_value` is 0. The `1 AS
required_value_per_sql_semantics` projected alongside makes the contradiction
visible per row: SQL semantics demand value 1; the database returned 0.
`C` and `D` are exactly the unmatched probe rows.
## Workaround
```sql
SET hive.vectorized.execution.enabled=false;
```
Same query then returns zero rows.
## Mechanism (pointer to the slot reuse)
`EXPLAIN VECTORIZATION DETAIL` of the repro shows physical column **`4`**
allocated for two purposes:
```
bigTableKeyExpressions:
LongColSubtractLongScalar(col 4:int, val 1)(children:
CastStringToLong(col 1:string) -> 4:int) -> 6:int
className: VectorMapJoinOuterMultiKeyOperator
smallTableValueMapping: 4:boolean
```
For matched rows the broadcast writes a `boolean` (0/1) into `vector[col
4][i]`, overwriting the cast intermediate. For unmatched rows
`generateOuterNulls`
(`ql/.../mapjoin/VectorMapJoinOuterGenerateResultOperator.java`) flips
`isNull[col 4][i] = true` but leaves `vector[col 4][i]` holding the stale `int`
from `CastStringToLong`. A subsequent projection of `cast(s_bool OR p_bool AS
INT)` reads the slot through `ColOrCol` (`ql/.../expressions/ColOrCol.java`)
followed by `IfExprLongScalarLongScalar`, and the dirty value (`stale_int | 1`)
fails the latter's strict `== 1` check, producing 0.
## Suggested fix layers
Listed root-first. Layers higher in this list cover broader classes of
downstream consumers; layers lower in the list narrow the symptom for one
specific chain.
1. **Vectorizer slot allocation**: don't reuse a scratch column for a
`smallTableValueMapping` target. Preserves the typed-slot invariant the rest of
the pipeline relies on; protects every downstream operator at once.
2. **`generateOuterNulls`**: when flipping `isNull[i]=true`, also write a
type-clean sentinel (e.g. `vector[i] = 0` or `LongColumnVector.NULL_VALUE`).
Defensive cleanup at the boundary; protects every downstream operator at once.
3. **Projection-mode operators that read `vector[i]` directly**: harden each
(e.g. `ColOrCol` masks `vector[i] & 1L` before OR'ing). Per-operator fix; many
operators to audit.
4. **Symptom-narrow fix in `IfExprLongScalarLongScalar`**: tolerate non-{0,1}
truthy values via `!= 0`. Hides this specific cast-to-int symptom; leaves
arithmetic, equality, and other dirty-input readers unaffected.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
