On Mon, Dec 16, 2013 at 08:50:35AM -0800, Joe Perches wrote:
> On Mon, 2013-12-16 at 11:04 -0500, Neil Horman wrote:
> > On Mon, Dec 16, 2013 at 04:45:05PM +0100, Daniel Borkmann wrote:
> > > On 12/16/2013 04:21 PM, Joe Perches wrote:
> > > >On Mon, 2013-12-16 at 16:13 +0100, Daniel Borkmann wrote:
> > > >>On 12/16/2013 04:03 PM, Joe Perches wrote:
> > > >>>On Mon, 2013-12-16 at 09:44 -0500, Neil Horman wrote:
> > > >>>>During a recent discussion regarding some sctp socket options, it was noted that
> > > >>>>we have several points at which we issue log warnings that can be flooded at an
> > > >>>>unbounded rate by any user. Fix this by converting all the pr_warns in the
> > > >>>>sctp_setsockopt path to be pr_warn_ratelimited.
> > > >>>
> > > >>>trivial note:
> > > >>[...]
> > > >>>>@@ -5311,8 +5311,8 @@ static int sctp_getsockopt_maxburst(struct sock *sk, int len,
> > > >>>[]
> > > >>>>+ pr_warn_ratelimited("Use of int in max_burst socket option deprecated\n");
> > > >>>>+ pr_warn_ratelimited("Use struct sctp_assoc_value instead\n");
> > > >>>
> > > >>>Perhaps a dedicated "deprecated" warning function
> > > >>>to centralize these?
> > > >>>
> > > >>>void _sctp_warn_deprecated(const char *func, const char *from, const char *to);
> > > >>>{
> > > >>> etc.
> > > >>>}
> > > >>>#define sctp_warn_deprecated(from, to) \
> > > >>> _sctp_warn_deprecated(__func__, from, to)
> > > >>
> > > >>If so, then this should better get even more "centralized" ... as e.g.
> > > >>pr_warn_deprecated() [which internally is ratelimited]. I don't see the
> > > >>point why only SCTP should have this special-cased.
> > > >
> > > >Sure, if it's useful outside of sctp, but I didn't
> > > >notice any other uses like it.
> > >
> > > If we have a generic API for that, they might come, sure.
> > I agree with Daniel. If we're going to make this common, theres no reason to
> > not make it common for all uses. Searching the kernel for uses of printk/pr_*
> > and the string "deprecated" shows lots of potential use sites.
>
> Does adding a couple of functions like:
>
> void pr_warn_deprecated(const char *old, const char *new)
> {
> static DEFINE_RATELIMIT_STATE(_rs,
> DEFAULT_RATELIMIT_INTERVAL,
> DEFAULT_RATELIMIT_BURST);
>
> if (!__ratelimit(&_rs))
> return;
>
> if (new)
> printk(KERN_WARNING "%pf: Use of \"%s\" is deprecated - use \"%s\" instead\n",
> __builtin_return_address(1), old, new);
> else
> printk(KERN_WARNING "%pf: Use of \"%s\" is deprecated\n",
> __builtin_return_address(1), old);
> }
>
> suit? Other suggestions?
>
i'm just doing this:
#define pr_warn_deprecated(fmt, ...) \
pr_warn_ratelimited("Deprecated: " fmt, ##__VA_ARGS__)
That will work for every form, giving consistency to the location of a single
word for ease of searching. I don't really see the need to institutionalize
"use <blank> instead", since there may be no drop in replacement for something
that is deprecated.
Neil
--
To unsubscribe from this list: send the line "unsubscribe linux-sctp" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
