linux-kernel - Re: linux-kernel janitorial RFP: Mark static arrays as const

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Wed, 3 Mar 2021 10:41:58 +0100
From:   Rasmus Villemoes <rasmus.villemoes@...vas.dk>
To:     Joe Perches <joe@...ches.com>,
        kernelnewbies <kernelnewbies@...nelnewbies.org>,
        kernel-janitors <kernel-janitors@...r.kernel.org>,
        cocci <cocci@...teme.lip6.fr>
Cc:     LKML <linux-kernel@...r.kernel.org>
Subject: Re: linux-kernel janitorial RFP: Mark static arrays as const

On 02/03/2021 18.42, Joe Perches wrote:
> Here is a possible opportunity to reduce data usage in the kernel.
> 
> $ git grep -P -n '^static\s+(?!const|struct)(?:\w+\s+){1,3}\w+\s*\[\s*\]' drivers/ | \
>   grep -v __initdata | \
>   wc -l
> 3250
> 
> Meaning there are ~3000 declarations of arrays with what appears to be
> file static const content that are not marked const.
> 
> So there are many static arrays that could be marked const to move the
> compiled object code from data to text minimizing the total amount of
> exposed r/w data.

You can add const if you like, but it will rarely change the generated
code. gcc is already smart enough to take a static array whose contents
are provably never modified within the TU and put it in .rodata:

static int x = 7;
static int y[2] = {13, 19};

static int p(int a, const int *foo)
{
	return a + *foo;
}
int q(int a)
{
	int b = p(a, &x);
	return p(b, &y[b & 1]);
}
$ nm c.o
0000000000000000 T q
0000000000000000 r y
$ size c.o
   text    data     bss     dec     hex filename
    111       0       0     111      6f c.o

So x gets optimized away completely, but y isn't so easy to get rid of -
nevertheless, it's never modified and the address doesn't escape the TU,
so gcc treats is as if it was declared const.

I think you'll see the same if you try adding the const on a few of your
real-life examples. (Of course, the control flow may be so convoluted
that gcc isn't able to infer the constness, so I'm not saying it will
never make a difference - only that you shouldn't expect too much.)

Rasmus