Re: Re: PROPOSAL: make PostgreSQL sanitizers-friendly (and prevent information disclosure)

From: Aleksander Alekseev <a(dot)alekseev(at)postgrespro(dot)ru>
To: Chapman Flack <chap(at)anastigmatix(dot)net>
Cc: pgsql-hackers <pgsql-hackers(at)postgresql(dot)org>
Subject: Re: Re: PROPOSAL: make PostgreSQL sanitizers-friendly (and prevent information disclosure)
Date: 2016-03-21 15:13:58
Message-ID: 20160321181358.21a5517c@fujitsu
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

> I'm new here so someone more experienced would have to weigh in,
> but I would wonder a couple of things:
>
> a. whether a braced struct assignment is supported in every
> C compiler that PostgreSQL still intends to support
>
> b. whether such a struct assignment is guaranteed to initialize
> padding spaces as well as declared fields (in all supported
> C versions/compilers).
>
> It's possible that memset() would be more convincing.

Frankly I'm not sure regarding all supported C versions/compilers. But
it seems to be a valid ANSI C. Here is a test program:

```
#include <stdio.h>

typedef struct {
int i;
char c;
long l;
short s;
} MyStruct;

int main()
{
int i, sum = 0;
char *c;
MyStruct s = {0};

s.i = 11;
s.c = 22;
s.l = 33;
s.s = 44;

c = (char*)&s;
for(i = 0; i < sizeof(s); i++) {
sum += *c;
c++;
}

printf("Sum: %d\n", sum);

return 0;
}
```

I compiled it with various versions of GCC and CLang with different
optimization flags:

clang38 -O3 -ansi -g t.c -o t
gcc -O0 -ansi -g t.c -o t

In all cases running a program under debugger shows that structure is
properly initialized:

(gdb) b main
Breakpoint 1 at 0x4007ae: file t.c, line 12.
(gdb) r
Starting program: /usr/home/eax/temp/t

Breakpoint 1, main () at t.c:12
12 int i, sum = 0;
(gdb) p memset(&s, 0xEA, sizeof(MyStruct))
$1 = -5376
(gdb) x/24xb &s
0x7fffffffeb00: 0xea 0xea 0xea 0xea 0xea 0xea 0xea 0xea
0x7fffffffeb08: 0xea 0xea 0xea 0xea 0xea 0xea 0xea 0xea
0x7fffffffeb10: 0xea 0xea 0xea 0xea 0xea 0xea 0xea 0xea
(gdb) n
14 MyStruct s = {0};
(gdb)
16 s.i = 11;
(gdb) x/24xb &s
0x7fffffffeb00: 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00
0x7fffffffeb08: 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00
0x7fffffffeb10: 0x00 0x00 0x00 0x00 0x00 0x00 0x00 0x00
(gdb) quit

Naturally we could use memset() as well. But I personally find it a bit
less readable. And in theory it doesn't prevent some _very_ "smart" C
compiler from not cleaning the whole structure anyway.

--
Best regards,
Aleksander Alekseev
http://eax.me/

In response to

Browse pgsql-hackers by date

  From Date Subject
Next Message David G. Johnston 2016-03-21 15:17:33 Re: [GENERAL] Request - repeat value of \pset title during \watch interations
Previous Message Craig Ringer 2016-03-21 15:11:37 Re: Proposal: Generic WAL logical messages