Re: [PATCH 1/3] Kprobes: Make kprobe modules more portable

[Ananth N Mavinakayanahalli <ananth@in.ibm.com>]

On Wed, Aug 09, 2006 at 05:10:39PM +0100, Christoph Hellwig wrote:
> On Wed, Aug 09, 2006 at 11:01:45AM -0500, David Smith wrote:
> > > +	if (p->symbol_name) {
> > > +		if (p->addr)
> > > +			return -EINVAL;
> > > +		p->addr = kprobe_lookup_name(p->symbol_name) + p->offset;
> > > +	}
> > 
> > What if kprobe_lookup_name() fails
> 
> for that case we need the check in your snipplet below.

The next check:

        if ((!kernel_text_address((unsigned long) p->addr)) ||
                in_kprobes_functions((unsigned long) p->addr))
                return -EINVAL;

will catch this case anyway (unless some arch has kernel text starting
at vaddr 0)

Ananth

> 
> > or if CONFIG_KALLSYMS isn't set?
> 
> I think we should just disallow that case.  having kprobes without kallsyms
> is rather pointless.  I'll send a patch to add the dependency to the Kconfig
> files.
> 
> > Perhaps this needs something like:
> > 
> > 	if (p->symbol_name) {
> > 		if (p->addr)
> > 			return -EINVAL;
> > 		p->addr = kprobe_lookup_name(p->symbol_name) + p->offset;
> > 		if (p->addr == p->offset)
> > 			return -EINVAL;
> > 	}