On Mon, 19 Mar 2012 20:02:54 PDT, Joe Perches said:
> On Mon, 2012-03-19 at 22:58 -0400, Ted Ts'o wrote:
>
> [good stuff]
>
> > Of course, the body of the message needs to be standardized too. But
> > that's orthogonal to the problem of passing the kernel structure which
> > identifies the object which the log message is about. That part is
> > completely and utterly necessary if you're going to standardize the
> > *first* part of the printable dmesg log, which contains the structured
> > information.
>
> All of what you say is true, but you're neglecting
> the ability to scrape or notify a particular set of
> per subsystem messages that pr_<level> could easily
> provide that a bare printk could not.
OK. Say I'm a scraper. How do I distinguish between:
pr_info("foo");
printk(KERN_INFO "foo");
Oh my. seems that both result in exactly the same thing ending up in the
dmesg buffer, so a scraper can't tell the difference because there isn't any...
pr_<level> isn't currently providing any ability to scrap or notify that can't
be done equally well by a printk. Now if you fix things so there's actual
advantage to or_<level>, then I'm sure Ted will be willing to listen.
Attachment:
pgpZGDDA2L157.pgp
Description: PGP signature
