Disable USB2 in AMD8111 and remove ide hardcode io addr

Eric W. Biederman ebiederman at lnxi.com
Wed May 5 09:18:01 CEST 2004


ron minnich <rminnich at lanl.gov> writes:

> On Tue, 4 May 2004, YhLu wrote:
> 
> > I'm going to add several lines in amd8111_enable of amd8111.c to disable
> > USB2 in amd8111. 
> > 
> >         /* disable usb2 in amd 8111 because it does not work awlays*/
> >         byte = pci_read_config8(lpc_dev, 0x47);
> >         byte |= (1<<7);
> >         pci_write_config8(lpc_dev, 0x47, byte);
> 
> this is the wrong way to do this, please don't do it just yet. 
> 
> The right way to do it is in the config file. 
> 
> I will look at your code and we'll send an example.

The code should already exist to do this.

> > Also will remove several lines in amd8111_ide.c
> > 
> >         /* The AMD768 has a bug where the BM DMA address must be
> >          * 256 byte aligned while it is only 16 bytes long.
> >          * Hard code this to a valid address below 0x1000
> >          * where automatic port address assignment starts.
> >          * FIXME: I assume the 8111 does the same thing. We should
> >          * clarify. stepan at suse.de
> >          */
> >         pci_write_config32(dev, 0x20, 0xf01);
> > 
> >         pci_write_config32(dev, 0x48, 0x205e5e5e);
> >         word = 0x06a;
> >         pci_write_config16(dev, 0x4c, word);
> 
> why?

We should provide the necessary information to the resource allocator to
do this.  The extra alignment probably isn't needed but it should not be hard
to implement.

> > 
> > and add 
> > 
> >         byte = 0x20 ; // Latency: 64-->32
> >         pci_write_config8(dev, 0xd, byte);
> 
> which is needed because ...
> 
> Let's talk some more.

Latency should certainly be a general PCI thing unless there is a good
alternative.

Eric



More information about the coreboot mailing list