Re: [PATCH v2] hwmon: Add driver for AMD family 15h processor power information

[Andreas Herrmann <herrmann.der.user@googlemail.com>]

On Wed, Apr 06, 2011 at 09:14:07AM +0200, Clemens Ladisch wrote:
> Guenter Roeck wrote:
> > On Tue, Apr 05, 2011 at 10:45:36AM -0400, Andreas Herrmann wrote:
> > > +static int __devinit f15h_power_is_internal_node0(struct pci_dev *f4)
> > > +{
> > > +	u32 val;
> > > +	struct pci_dev *f3;
> > > +
> > > +	f3 = pci_get_slot(f4->bus, PCI_DEVFN(PCI_SLOT(f4->devfn), 3));
> > > +	if (!f3) {
> > > +		dev_err(&f4->dev, "no function 3 available on this slot\n");
> > > +		return 0;
> > 
> > It is a common practice to return a negative value on errors. Why not here ?
> 
> Apparently, this function returns a boolean value.  Using "bool"/"true"/
> "false" would have made this more obvious.

Agreed, using bool as return type would have been the better choice.
I'll adapt it.

> > Also, is this really an error which asks for an error message, or just a CPU
> > or system which does not support the attribute ?
> 
> AFAICT all F15h CPUs are _known_ to have all these PCI functions; the
> error should never occur in practice.

Except under virtualization. But common practice is to not pass any
such devices into a guest. So either all CPU northbridge functions are
there or no function at all. I'll adapt the code to your suggestion.

> What I do in the k10temp driver in this situation is to trust the CPU to
> be there, omitting the pci_get_slot/pci_dev_put and just replacing this:
> 
> > > +	pci_read_config_dword(f3, REG_NORTHBRIDGE_CAP, &val);
> > > +	pci_dev_put(f3);
> 
> with the equivalent of:
> 
>   pci_bus_read_config_dword(f4->bus, PCI_DEVFN(PCI_SLOT(f4->devfn), 3),
>                             REG_NORTHBRIDGE_CAP, &val);
> 
> 
> BTW: The family 15h CPUs have the same temperature sensor registers
> (D18F3xA4 and D18F3x64) as the earlier families, haven't they?

Yes and I think Andre has just sent a patch to add support for 15h to
k10temp.


Thanks,

Andreas