This simple patch to catalog/pg_type.c fixes a buffer overrun.
It was detected by Electric Fence and triggered by statements like:
SELECT * into table t from pg_database;
The system would crash on a memmove call in DataFile() with arguments like this:
memmove(0x0, 0x0, 0);
Here's the patch.
> NameData name;
< values[i++] = PointerGetDatum(typeName); /* 1 */
> values[i++] = NameGetDatum(&name); /* 1 */
After applying the patch the problem goes away.
I'll try to make all the buffer overruns detected by Electric Fence
during the regression test on my Linux system go away -:).
Debugging is a nice way to learn about a system isn't it?
With regards from Maurice.
pgsql-hackers by date
|Next:||From: Michal Mosiewicz||Date: 1998-03-29 14:17:29|
|Subject: Re: [HACKERS] Optimizer fails?|
|Previous:||From: Bruce Momjian||Date: 1998-03-29 04:53:56|
|Subject: Re: (PM) Re: [HACKERS] Ways to crash the backend|