Re: [PATCH v3 1/1] x86: allow to enable/disable modify_ldt at run time

[Ingo Molnar <mingo@kernel.org>]

* Willy Tarreau <w@1wt.eu> wrote:

> > > +	if (!sysctl_modify_ldt) {
> > > +		printk_ratelimited(KERN_INFO
> > > +			"Denied a call to modify_ldt() from %s[%d] (uid: %d)."
> > > +			" Adjust the modify_ldt sysctl if this was not an"
> > 
> > Would it really be so difficult to write this as:
> > 
> >   Set "sys.kernel.modify_ldt = 1" in /etc/sysctl.conf if this was not an exploit attempt.
> 
> It's just a matter of taste. Normally I consider the kernel distro-agnostic so I 
> don't like to suggest one way to adjust sysctls nor to reference config files. 
> Here we're in a case where only standard distro users may hit the issue, and 
> users of embedded distros will not face this message or will easily translate it 
> into their respective configuration scheme. So OK for this one.

So it's a side issue, but it's not a matter of taste at all: why should we end up 
hurting 99% of Linux users (that use regular distros), just to make it slightly 
more 'correct' for the weird 1% 'embedded distro' case that decided to put sysctl 
configuration elsewhere?

Users of 'embedded' distros won't normally see kernel messages, and even if they 
do, the message is crystal clear even to them...

Such messages should be as helpful to the regular case as possible. The weird 
cases will be OK too: and it does not help to make a message unhelpful for _both_ 
cases.

Thanks,

	Ingo