Re: Fix redundant memset after palloc0 in heap_form_minimal_tuple()

От:

Jeff Davis <pgsql@j-davis.com>

Дата:

4 ноября 2025 г. в 20:51:18

On Tue, 2025-11-04 at 17:05 +0800, Chao Li wrote:
> The current code does both palloc0() and memset(0)
> in heap_form_minimal_tuple():

Thank you, fixed.

> To fix the problem, my first impression was to delete the memset().
> But looking at a0942f4, I found a couple of other places that do the
> same pattern: palloc(len+extra) then memset(0, extra), so I think the
> correct fix should be changing the palloc0 to palloc.

The palloc0() is important because heap_fill_tuple() skips over
alignment bytes. So I just removed the memset().

Regards,
	Jeff Davis

Fix redundant memset after palloc0 in heap_form_minimal_tuple()

От:

Chao Li <li.evan.chao@gmail.com>

Дата:

4 ноября 2025 г. в 12:05:29

Hi Hackers,

While studying the other patch, I happened to notice this problem in heap_form_minimal_tuple().

The current code does both palloc0() and memset(0) in heap_form_minimal_tuple():

```

/*
* Allocate and zero the space needed.
*/
mem = palloc0(len + extra);
memset(mem, 0, extra);
tuple = (MinimalTuple) (mem + extra);

```

That looks like an oversight of a0942f4.

To fix the problem, my first impression was to delete the memset(). But looking at a0942f4, I found a couple of other places that do the same pattern: palloc(len+extra) then memset(0, extra), so I think the correct fix should be changing the palloc0 to palloc.

Best regards,

Chao Li (Evan)

---------------------

HighGo Software Co., Ltd.
https://www.highgo.com/