execute prepared statement with binary protocol will panic when table schema changed

[lcwangchao]

Bug Report

Please answer these questions before submitting your issue. Thanks!

1. Minimal reproduce step (Required)

import (
	"context"
	"database/sql"
	"testing"

	_ "github.com/go-sql-driver/mysql"
	"github.com/stretchr/testify/require"
)

func TestPrepareExecute(t *testing.T) {
	ctx := context.TODO()
	db, err := sql.Open("mysql", "root@tcp(127.0.0.1:4000)/test")
	require.NoError(t, err)

	conn, err := db.Conn(ctx)
	require.NoError(t, err)
	_, err = conn.ExecContext(ctx, "drop table if exists t1")
	require.NoError(t, err)
	_, err = conn.ExecContext(ctx, "create table t1 (id int primary key, v int)")
	require.NoError(t, err)
	_, err = conn.ExecContext(ctx, "insert into t1 values(1, 10)")
	require.NoError(t, err)

	stmt, err := conn.PrepareContext(context.TODO(), "select * from t1")
	require.NoError(t, err)

	require.Equal(t, []int{1, 10}, executePreparedStmtResult(t, stmt, 2))

	// drop one column
	_, err = conn.ExecContext(ctx, "alter table t1 drop column v")
	require.NoError(t, err)

	// the below will fail when get next row
	require.Equal(t, []int{1}, executePreparedStmtResult(t, stmt, 1))
}

func executePreparedStmtResult(t *testing.T, stmt *sql.Stmt, columns int) []int {
	rows, err := stmt.Query()
	require.NoError(t, err)
	if ok := rows.Next(); !ok {
		// fail here for the third execute
		require.Fail(t, rows.Err().Error())
	}
	row := make([]int, columns)
	scan := make([]interface{}, columns)
	for i := range row {
		scan[i] = &row[i]
	}

	require.NoError(t, rows.Scan(scan...))
	require.NoError(t, rows.Close())
	return row
}

2. What did you expect to see? (Required)

no error

3. What did you see instead (Required)

=== RUN   TestPrepareExecute
    staleread_test.go:91: 
        	Error Trace:	staleread_test.go:91
        	            				staleread_test.go:84
        	Error:      	Error 1105: runtime error: index out of range [1] with length 1
        	Test:       	TestPrepareExecute
--- FAIL: TestPrepareExecute (0.30s)

FAIL

Got a panic in tidb side:

[2022/03/28 17:45:17.492 +08:00] [ERROR] [conn.go:1031] ["connection running loop panic"] [conn=4089] [lastSQL="select * from t1"] [err="runtime error: index out of range [1] with length 1"] [stack="goroutine 380115 [running]:
github.com/pingcap/tidb/server.(*clientConn).Run.func1()
	/Users/wangchao/Code/pingcap/tidb/server/conn.go:1029 +0x88
panic({0x10738fac0, 0x14001f98510})
	/usr/local/go/src/runtime/panic.go:844 +0x26c
github.com/pingcap/tidb/server.(*clientConn).writeResultset.func1()
	/Users/wangchao/Code/pingcap/tidb/server/conn.go:2119 +0x3d4
panic({0x10738fac0, 0x14001f98510})
	/usr/local/go/src/runtime/panic.go:844 +0x26c
github.com/pingcap/tidb/util/chunk.Row.IsNull({0x140096b60a0, 0x0}, 0x1)
	/Users/wangchao/Code/pingcap/tidb/util/chunk/row.go:206 +0x74
github.com/pingcap/tidb/server.dumpBinaryRow({0x14001c26000, 0xa, 0x400}, {0x140051f81e0, 0x2, 0x2}, {0x140096b60a0, 0x0}, 0x140056a0780)
	/Users/wangchao/Code/pingcap/tidb/server/util.go:248 +0x1b0
github.com/pingcap/tidb/server.(*clientConn).writeChunks(0x140020ef400, {0x1075a5f10, 0x14003af06c0}, {0x1075ae598, 0x140096b6050}, 0x1, 0x0)
	/Users/wangchao/Code/pingcap/tidb/server/conn.go:2208 +0x390
github.com/pingcap/tidb/server.(*clientConn).writeResultset(0x140020ef400, {0x1075a5f10, 0x14003af06c0}, {0x1075ae598, 0x140096b6050}, 0x1, 0x0, 0x0)
	/Users/wangchao/Code/pingcap/tidb/server/conn.go:2136 +0x23c
github.com/pingcap/tidb/server.(*clientConn).executePreparedStmtAndWriteResult(0x140020ef400, {0x1075a5f10, 0x14003af06c0}, {0x1075bc220, 0x1400d01eaf0}, {0x1090cd0b8, 0x0, 0x0}, 0x0)
	/Users/wangchao/Code/pingcap/tidb/server/conn_stmt.go:263 +0x410
github.com/pingcap/tidb/server.(*clientConn).handleStmtExecute(0x140020ef400, {0x1075a5f10, 0x14003af06c0}, {0x14004c56205, 0x9, 0x9})
	/Users/wangchao/Code/pingcap/tidb/server/conn_stmt.go:209 +0x7c4
github.com/pingcap/tidb/server.(*clientConn).dispatch(0x140020ef400, {0x1075a5e68, 0x140056a0380}, {0x14004c56205, 0x9, 0x9})
	/Users/wangchao/Code/pingcap/tidb/server/conn.go:1367 +0xfd8
github.com/pingcap/tidb/server.(*clientConn).Run(0x140020ef400, {0x1075a5f10, 0x1401408c0c0})
	/Users/wangchao/Code/pingcap/tidb/server/conn.go:1095 +0x7b4
github.com/pingcap/tidb/server.(*Server).onConn(0x140077d00d0, 0x140020ef400)
	/Users/wangchao/Code/pingcap/tidb/server/server.go:551 +0xa64
created by github.com/pingcap/tidb/server.(*Server).startNetworkListener
	/Users/wangchao/Code/pingcap/tidb/server/server.go:448 +0x7a4
"]

4. What is your TiDB version? (Required)

master

[lcwangchao]

Seems introduced by this PR: #12388 , and the below code:

tidb/server/driver_tidb.go

Lines 347 to 370 in 4ff3ec2

	func (trs *tidbResultSet) Columns() []*ColumnInfo {
	if trs.columns != nil {
	return trs.columns
	}
	// for prepare statement, try to get cached columnInfo array
	if trs.preparedStmt != nil {
	ps := trs.preparedStmt
	if colInfos, ok := ps.ColumnInfos.([]*ColumnInfo); ok {
	trs.columns = colInfos
	}
	}
	if trs.columns == nil {
	fields := trs.recordSet.Fields()
	for _, v := range fields {
	trs.columns = append(trs.columns, convertColumnInfo(v))
	}
	if trs.preparedStmt != nil {
	// if ColumnInfo struct has allocated object,
	// here maybe we need deep copy ColumnInfo to do caching
	trs.preparedStmt.ColumnInfos = trs.columns
	}
	}
	return trs.columns
	}

The columns will be cached in prepared statement, but if the table's schema changed, the cache will not be cleared

[kennedy8312]

/type regression