linux-kernel - Re: [PATCH] jump_label: jump

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: <20111201211350.GD2443@redhat.com>
Date:	Thu, 1 Dec 2011 16:13:50 -0500
From:	Jason Baron <jbaron@...hat.com>
To:	Peter Zijlstra <a.p.zijlstra@...llo.nl>
Cc:	KAMEZAWA Hiroyuki <kamezawa.hiroyu@...fujitsu.com>,
	"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
	Jeremy Fitzhardinge <jeremy.fitzhardinge@...rix.com>,
	rostedt@...dmis.org
Subject: Re: [PATCH] jump_label: jump_label for boot options.

On Thu, Dec 01, 2011 at 06:39:45PM +0100, Peter Zijlstra wrote:
> On Thu, 2011-12-01 at 11:50 -0500, Jason Baron wrote:
> 
> > I think its just a matter of reversing the true and false returns.
> > That is, instead of:
> 
> that's the same as !static_branch()
> 
> > jump_label_inc/dec(), don't need to be changed, they just mean reverse
> > the branch on 0, 1 transitions. Although using the same key in both
> > static_branch_true, and static_branch_false, might be confusing. Maybe
> > we rename jump_label_inc/dec to static_branch_reverse_inc()/dec()?
> 
> Right, that's the problem really. That makes it impossible to make the
> control code generic.
> 
> What I'd want is something that doesn't out-of-line the branch, is
> possibly enabled by default, but has the same inc/dec behaviour, not the
> reversed.
> 

I think what you have below should work modulo the no out-of-line
branches and the following change:

> -			if (neg)
> +			if (neg) {
>  				sysctl_sched_features &= ~(1UL << i);
> -			else
> +#ifdef HAVE_JUMP_LABEL
> +				if (!jump_label_enabled(&sched_feat_keys[i]))
> +					jump_label_inc(&sched_feat_keys[i]);
> +#endif

I think here its:
			if (jump_label_enabled())
				jump_label_dec();


> +			} else {
>  				sysctl_sched_features |= (1UL << i);
> +#ifdef HAVE_JUMP_LABEL
> +				if (jump_label_enabled(&sched_feat_keys[i]))
> +					jump_label_dec(&sched_feat_keys[i]);
> +#endif
> +			}

Same here:
			if (!jump_label_enabled())
				jump_label_inc()




The inc/dec behavior we have now, in fact will only mess up in the case
where we define 'static_branch_true()'. Because then, in that case the
jump_label_inc() will cause a jump to the false branch. So as long as we
don't introduce 'static_branch_true()' and do an early setting of those
branches which are true via __init code as you have here, I think things are
correct.

Thanks,

-Jason
--
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/