Re: wm(4) misreporting PCI bus properties

[Eduardo Horvath <eeh%NetBSD.org@localhost>]

On Wed, 15 Dec 2010, Matthias Scheler wrote:

> On Wed, Dec 15, 2010 at 05:14:41PM +0000, Eduardo Horvath wrote:
> > > my ULTRA60 (running NetBSD/sparc64 5.99.41) reports its Intel Gigabit
> > > ethernet card like this:
> > > 
> > > wm0 at pci1 dev 1 function 0: Intel i82540EM 1000BASE-T Ethernet, rev. 2
> > > wm0: interrupting at ivec 0
> > > wm0: 32-bit 66MHz PCI bus
> > > wm0: 64 word (6 address bits) MicroWire EEPROM
> > > wm0: Ethernet address 00:07:e9:**:**:**
> > > makphy0 at wm0 phy 1: Marvell 88E1011 Gigabit PHY, rev. 3
> > > 
> > > What puzzles me is why it claims that it is using a "32-bit PCI bus".
> > > The 66MHz PCI slot in the ULTRA60 definitely supports 64-bit. Is that
> > > a problem with wm(4), with the PCI support for the ULTRA60's chipset
> > > or something else?
> > 
> > Most likely the wm0 card is not 64-bit PCI and does not support 64-bit 
> > address cycles.
> 
> *aargh*
> 
> You are right.
> 
> It turns out I that forgot to install the discarded 64-bit PCI card I got
> from work a while ago. Sorry for the noise. :-(

OTOH, since SPARC uses an IOMMU, I'm not sure 64-bit is ever used.

Eduardo