Re: linux-next: Tree for August 14 (sysfs/acpi errors)

From: Len Brown <lenb@kernel.org>
To: Rusty Russell <rusty@rustcorp.com.au>
Cc: Andi Kleen <ak@linux.intel.com>, Greg KH <greg@kroah.com>,
	Kay Sievers <kay.sievers@vrfy.org>,
	Randy Dunlap <randy.dunlap@oracle.com>,
	Stephen Rothwell <sfr@canb.auug.org.au>,
	linux-next@vger.kernel.org, LKML <linux-kernel@vger.kernel.org>,
	linux-acpi@vger.kernel.org
Subject: Re: linux-next: Tree for August 14 (sysfs/acpi errors)
Date: Wed, 24 Sep 2008 03:59:41 -0400 (EDT)	[thread overview]
Message-ID: <alpine.LFD.1.10.0809240353520.20067@localhost.localdomain> (raw)
In-Reply-To: <200808171513.13611.rusty@rustcorp.com.au>

[-- Attachment #1: Type: TEXT/PLAIN, Size: 1818 bytes --]

On Sun, 17 Aug 2008, Rusty Russell wrote:

> On Sunday 17 August 2008 12:30:34 Andi Kleen wrote:
> > Greg KH wrote:
> > > On Sat, Aug 16, 2008 at 05:48:26AM +0200, Andi Kleen wrote:
> > >>> They have been module options, not prefixed kernel parameters so far,
> > >>> and the prefix was just the module name.
> > >>> So it just strikes back, that acpi uses generic names for the modules,
> > >>> there would have been no problem if "power" would be called
> > >>> "acpi_power" and the options would just be "acpi.acpica_version" and
> > >>> "acpi_power.nocheck".

I think we (all) agree that we should keep the acpi gunk together,
rather than creating namespace sprawl to satisfy the current sysfs code.

> > >>> But well, there are driver modules just called "option", so acpi is not
> > >>> that bad. :)
> > >>>
> > >>>> I think the generic params code should be fixed to handle this.
> > >>>
> > >>> We could try to look up existing directories to use instead of
> > >>> expecting that we need to create and own them. I guess,
> > >>
> > >> sysfs does this anyways, doesn't it. We would just need to teach it
> > >> to not BUG() in this case, perhaps with a special entry point.
> > >> Also a BUG() in general seems a little harsh for this, surely a WARN_ON
> > >> should be enough.
> > >
> > > It is a WARN() call, not a BUG().
> >
> > Ok. Can we remove it? Or add a new entry point that allows to disable it?
> >
> > I don't think relying on link order like Rusty proposes is a good long term
> > solution.
>
> To be clear, I agree with Andi.  If this is for current kernel I'd just fix 
> link order, for longer term we need something cleverer.

This problem first shows up with the addition of the
acpi.power_nocheck modparam, which is staged for 2.6.28.
Is 2.6.28 "current" or longer term?

thanks,
-Len