Re: [PATCH 4/4] tracing/uprobes: Fix the usage of uprobe_buffer_enable() in probe_event_enable()

[Oleg Nesterov <oleg@redhat.com>]

On 06/30, Srikar Dronamraju wrote:
>
> > The usage of uprobe_buffer_enable() added by dcad1a20 is very wrong,
> >
> > 1. uprobe_buffer_enable() and uprobe_buffer_disable() are not balanced,
> >    _enable() should be called only if !enabled.
> >
> > 2. If uprobe_buffer_enable() fails probe_event_enable() should clear
> >    tp.flags and free event_file_link.
> >
> > 3. If uprobe_register() fails it should do uprobe_buffer_disable().
> >
> > Signed-off-by: Oleg Nesterov <oleg@redhat.com>
>
> Acked-by: Srikar Dronamraju <srikar@linux.vnet.ibm.com>

Thanks!

> (one nit .. )
>
> > +	ret = uprobe_buffer_enable();
> > +	if (ret)
> > +		goto err_flags;
> > +
> >  	tu->consumer.filter = filter;
> >  	ret = uprobe_register(tu->inode, tu->offset, &tu->consumer);
> > -	if (ret) {
> > -		if (file) {
> > -			list_del(&link->list);
> > -			kfree(link);
> > -			tu->tp.flags &= ~TP_FLAG_TRACE;
> > -		} else
> > -			tu->tp.flags &= ~TP_FLAG_PROFILE;
> > -	}
> > +	if (ret)
> > +		goto err_buffer;
> >
> > +	return 0;
> > +
> > + err_buffer:
> > +	uprobe_buffer_disable();
> > +
>
> How about avoiding err_buffer label?
> +	if (!ret)
> +		return 0;
>
> +	uprobe_buffer_disable();
> +

Well, I do not really mind. But to me it looks more consistent this way,
if-something-fail-goto-err_label.

IOW, I think that the code should either not use err-labels, or always
use them like above.

Besides, perhaps we will add "if (file) uprobe_apply()" after _register()
to mix perf/ftrace, then we will need to change this "if (!ret)" code again.

Oleg.