linux-kernel - Re: [PATCH v2] Remove uninitialized

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]

Message-ID: <508DD443.9050505@gmail.com>
Date:	Mon, 29 Oct 2012 11:56:35 +1100
From:	Ryan Mallon <rmallon@...il.com>
To:	Ingo Molnar <mingo@...nel.org>
CC:	Andrew Morton <akpm@...ux-foundation.org>,
	Linus Torvalds <torvalds@...ux-foundation.org>,
	Corey Minyard <cminyard@...sta.com>, minyard@....org,
	Linux Kernel <linux-kernel@...r.kernel.org>,
	OpenIPMI Developers <openipmi-developer@...ts.sourceforge.net>
Subject: Re: [PATCH v2] Remove uninitialized_var()

On 28/10/12 21:20, Ingo Molnar wrote:
> 
> * Andrew Morton <akpm@...ux-foundation.org> wrote:
> 
>> On Sat, 27 Oct 2012 15:12:03 +0200 Ingo Molnar <mingo@...nel.org> wrote:
>>
>>> There's 3 types of conversions done:
>>>
>>>    uninitialized_var(x)        =>  x = 0       /* for scalar types */
>>>    uninitialized_var(x)        =>  x = NULL    /* for pointers */
>>>    uninitialized_var(x)        =>  x = { }     /* for structures, unions */
>>
>> It's regrettable that we lose information.  
>> uninitialized_var() says "this isn't needed - it's just there 
>> for gcc".  The reader can of course work out the reason with 
>> careful code inspection, but that's a lot more time consuming.
>>
>> We could go add "/* keep gcc quiet */" to every site, or add 
>> self-documenting macros for the above.
> 
> Ok, agreed - I've added /* GCC */ to every site to make people 
> think about it.
> 
> I left it a bit mystic because in some cases this macro was 
> mis-used not to suppress GCC being wrong, but to hide GCC being 
> *right*: for example unused variable warnings in cases like:
> 
>    int uninitialized_var(var);
> 
>    #ifdef XYZ
>    var = ...;
>    ...
>    #endif
> 
> which (ab-)use was no doubt actively dangerous beyond being 
> ugly. One such example is in arch/x86/mm/numa.c. (These cases 
> now turn into clear (and always harmless) compiler warnings, as 
> they should.)

Shouldn't those cases be using:

	int var __maybe_unused;

To clarify that the variable is not used in some configurations. Or
moving the variable declaration inside the #ifdef block if possible. The
alternative:

	#ifdef XYZ
	int var;
	#endif
	
	...

	#ifdef XYZ
	var = ...;
	#endif

Does get pretty clunky.

> 
> This further strengthens the argument that we should remove this 
> whole thing.
> 
> Thanks,
> 
> 	Ingo
> 
> ---
> Signed-off-by: Ingo Molnar <mingo@...nel.org>

<snip>

>  185 files changed, 281 insertions(+), 291 deletions(-)

It might be interesting to know how many instances of uninitialized_var
are no longer required because of code change around them. Possibly
redefining uninitialized_var as an empty macro and then checking how
many don't cause errors would identify if any can just be removed
outright rather than converting them to an assignment.

> diff --git a/arch/arm/mach-sa1100/assabet.c b/arch/arm/mach-sa1100/assabet.c
> index 6a7ad3c..5ec34d6 100644
> --- a/arch/arm/mach-sa1100/assabet.c
> +++ b/arch/arm/mach-sa1100/assabet.c
> @@ -388,7 +388,7 @@ static void __init map_sa1100_gpio_regs( void )
>   */
>  static void __init get_assabet_scr(void)
>  {
> -	unsigned long uninitialized_var(scr), i;
> +	unsigned long scr = 0 /* GCC */, i;
>  
>  	GPDR |= 0x3fc;			/* Configure GPIO 9:2 as outputs */
>  	GPSR = 0x3fc;			/* Write 0xFF to GPIO 9:2 */
> diff --git a/arch/ia64/kernel/process.c b/arch/ia64/kernel/process.c
> index 35e106f..2c2da64 100644
> --- a/arch/ia64/kernel/process.c
> +++ b/arch/ia64/kernel/process.c
> @@ -493,7 +493,7 @@ static void
>  do_copy_task_regs (struct task_struct *task, struct unw_frame_info *info, void *arg)
>  {
>  	unsigned long mask, sp, nat_bits = 0, ar_rnat, urbs_end, cfm;
> -	unsigned long uninitialized_var(ip);	/* GCC be quiet */
> +	unsigned long ip = 0 /* GCC */;	/* GCC be quiet */

Doubled up comment. Same in a few other places.

>  	elf_greg_t *dst = arg;
>  	struct pt_regs *pt;
>  	char nat;
> diff --git a/arch/ia64/mm/discontig.c b/arch/ia64/mm/discontig.c
> index c641333..75cc8da 100644
> --- a/arch/ia64/mm/discontig.c
> +++ b/arch/ia64/mm/discontig.c
> @@ -185,7 +185,7 @@ static void *per_cpu_node_setup(void *cpu_data, int node)
>  void __init setup_per_cpu_areas(void)
>  {
>  	struct pcpu_alloc_info *ai;
> -	struct pcpu_group_info *uninitialized_var(gi);
> +	struct pcpu_group_info *gi = NULL /* GCC */;

The placement of the comment here is a bit ugly, can you change it so it
appears after the semicolon when it is the last initialiser?

~Ryan

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/