2009-11-16
| ||
12:51 | • Ticket [6b00e0a3] The great OP_If bug status still Fixed with 1 other change (artifact: a16d9fdf user: drh) | |
2009-10-30
| ||
14:05 | • Fixed ticket [6b00e0a3]. (artifact: a67b11ba user: drh) | |
13:57 | • New ticket [6b00e0a3]. (artifact: 7684faf2 user: drh) | |
Ticket Hash: | 6b00e0a34c8abd71093ea35df3f8ce9a8012aa0d | |||
Title: | The great OP_If bug | |||
Status: | Fixed | Type: | Code_Defect | |
Severity: | Critical | Priority: | Immediate | |
Subsystem: | Code_Generator | Resolution: | Fixed | |
Last Modified: | 2009-11-16 12:51:04 | |||
Version Found In: | 3.6.16 | |||
Description: | ||||
The code generator for EXISTS and SELECT expressions was storing the
number of the register that holds the result in the Expr.iColumn field.
However, Expr.iColumn is a 16-bit signed integer. Hence, for very complex
statements that use more than 32768 registers prior to running the EXISTS
or SELECT, the register number will overflow and possibly result in using
a negative number for an array index and causing a subsequent malfunction
or crash.
Solutions to this problem include:
drh added on 2009-10-30 14:05:23: Fixed by check-in [7253f8fad1] on the trunk. |