Re: [PATCH -next] acpi utmisc: use WARN_ON() instead of warn_on_slowpath()

[Andrew Morton <akpm@linux-foundation.org>]

On Wed, 02 Jul 2008 23:07:25 +0200
Andi Kleen <andi@firstfloor.org> wrote:

> Andrew Morton wrote:
> > On Wed, 02 Jul 2008 22:51:52 +0200
> > Andi Kleen <andi@firstfloor.org> wrote:
> > 
> >>> If you really want to do it this way (and it sounds reasonable) then
> >>> can we please do it in a less-than-totally-hacky-and-broken way?
> >>>
> >>>
> >>>
> >>> For example, define a new, always-available helper function in (say)
> >>> kernel/panic.c along the lines of
> >>>
> >>> void emit_warning_message(?)(const char *msg, int line)
> >>
> >> Ok we can just rename warn_on_slow_path() which is already there to that new name.
> > 
> > Sure, that'd work.  But we should pull it outside any ifdefs so that it
> > is always available (IMO).
> 
> For !CONFIG_BUG it's reasonable to not have the calls. These people
> who define it must be fighting for every byte ... So i think
> the stub inline in my previous patch is fine.

It should all be moved to include/linux/bug.h.  Or, perhaps,
include/linux/kernel.h.

> What name do you want?

umm, warn_with_context(), something like that?

I suppose it really should take a printk control string too.  Later.