linux-kernel - Re: [PATCH] mm/page_alloc: Avoid KERN_CONT uses in warn

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:   Tue, 07 Nov 2017 07:34:25 -0800
From:   Joe Perches <joe@...ches.com>
To:     Michal Hocko <mhocko@...nel.org>
Cc:     linux-kernel@...r.kernel.org, linux-mm@...ck.org
Subject: Re: [PATCH] mm/page_alloc: Avoid KERN_CONT uses in warn_alloc

On Tue, 2017-11-07 at 13:50 +0100, Michal Hocko wrote:
> On Mon 06-11-17 10:02:56, Joe Perches wrote:
> > KERN_CONT/pr_cont uses should be avoided where possible.
> > Use single pr_warn calls instead.
[]
> > diff --git a/mm/page_alloc.c b/mm/page_alloc.c
[]
> > @@ -3275,19 +3275,17 @@ void warn_alloc(gfp_t gfp_mask, nodemask_t *nodemask, const char *fmt, ...)
> >  	if ((gfp_mask & __GFP_NOWARN) || !__ratelimit(&nopage_rs))
> >  		return;
> >  
> > -	pr_warn("%s: ", current->comm);
> > -
> >  	va_start(args, fmt);
> >  	vaf.fmt = fmt;
> >  	vaf.va = &args;
> > -	pr_cont("%pV", &vaf);
> > -	va_end(args);
> > -
> > -	pr_cont(", mode:%#x(%pGg), nodemask=", gfp_mask, &gfp_mask);
> >  	if (nodemask)
> > -		pr_cont("%*pbl\n", nodemask_pr_args(nodemask));
> > +		pr_warn("%s: %pV, mode:%#x(%pGg), nodemask=%*pbl\n",
> > +			current->comm, &vaf, gfp_mask, &gfp_mask,
> > +			nodemask_pr_args(nodemask));
> >  	else
> > -		pr_cont("(null)\n");
> > +		pr_warn("%s: %pV, mode:%#x(%pGg), nodemask=(null)\n",
> > +			current->comm, &vaf, gfp_mask, &gfp_mask);
> > +	va_end(args);
> >  
> >  	cpuset_print_current_mems_allowed();
> 
> I do not like the duplication. It just calls for inconsistencies over
> time. Can we instead make %*pbl consume NULL nodemask instead?
> Something like the following pseudo patch + the if/else removed.
> If this would be possible we could simplify other code as well I think
> (at least oom code has to special case NULL nodemask).
> 
> What do you think?

I think it would be fine to have a single pr_warn.

> ---
> diff --git a/include/linux/nodemask.h b/include/linux/nodemask.h
[]
> @@ -104,7 +104,7 @@ extern nodemask_t _unused_nodemask_arg_;
>   *
>   * Can be used to provide arguments for '%*pb[l]' when printing a nodemask.
>   */
> -#define nodemask_pr_args(maskp)		MAX_NUMNODES, (maskp)->bits
> +#define nodemask_pr_args(maskp)		MAX_NUMNODES, (maskp) ? (maskp)->bits : NULL
>  
>  /*
>   * The inline keyword gives the compiler room to decide to inline, or
> diff --git a/lib/vsprintf.c b/lib/vsprintf.c
[]
> @@ -902,6 +902,9 @@ char *bitmap_list_string(char *buf, char *end, unsigned long *bitmap,
>  	int cur, rbot, rtop;
>  	bool first = true;
>  
> +	if (!bitmap)
> +		return buf;

I believe this is not necessary as any NULL pointer argument
passed to lib/vsprintf.c:pointer() (any %p<foo>) emits
"[2 or 10 spaces](null)" on 32bit or 64 bit systems.

I believe, but have not tested, that using a specific width
as an argument to %*pb[l] will constrain the number of
spaces before the '(null)' output in any NULL pointer use.

So how about a #define like

/*
 * nodemask_pr_args is only used with a "%*pb[l]" format for a nodemask.
 * A NULL nodemask uses 6 to emit "(null)" without leading spaces.
 */
#define nodemask_pr_args(maskp)			\
	(maskp) ? MAX_NUMNODES : 6,		\
	(maskp) ? (maskp)->bits : NULL