RE: [PATCH] Suspend and Resume Support for Intel IOMMU

[David Woodhouse <dwmw2@infradead.org>]

On Wed, 2009-02-18 at 09:28 -0800, Yu, Fenghua wrote:
> >> On Wed, 18 Feb 2009, Fenghua Yu wrote:
> >> >
> >> > +static int vtd_enabled;
> >> ..
> >> > +static int iommu_resume(struct sys_device *dev)
> >> > +{
> >> ..
> >> > +	if (!vtd_enabled)
> >> > +		return 0;
> >> 
> >> Hmm. Why do this?
> >> 
> >> Wouldn't it be much more logical to only register the 
> >> iommu_sysdev if the thing is enabled, rather than having an 
> >> odd flag that gets tested at runtime?
> >
> >The sysdev_class_register() could be done straight in 
> >intel_iommu_init(), because that gets called by pci_iommu_init() 
> >which is an fs_initcall() - so all the sysdev facilities should 
> >be up and running already.
> 
> That was my original concern. I'll remove the vtd_enabled flag and
> register device_iommu in intel_iommu_init() in an updated patch.

For some reason most of this thread is missing from my mailbox. Did you
ever submit this 'updated patch' that you threatened...?

-- 
David Woodhouse                            Open Source Technology Centre
David.Woodhouse@intel.com                              Intel Corporation