[coreboot] On the credibility of 'getpir'.

Mats Erik Andersson mats.andersson at gisladisker.se
Thu Sep 18 00:46:48 CEST 2008

On Wed, 17 Sep 2008 14:36:36 -0700
Ron Minnich <rminnich at gmail.com> wrote
> On Wed, Sep 17, 2008 at 2:34 PM, Mats Erik Andersson
> <mats.andersson at gisladisker.se> wrote:
> >
> > Hello,
> >
> > I am beginning to doubt the credibility of 'getpir'.
> I don't think it is a getpir issue. The issue is that, increasingly,
> the PIR tables in the BIOS are completely wrong. I see this more and
> more all the time, but it was
> a problem as long as 8 years ago.
> It's a hard problem to resolve.
> And, of course, the MP and ACPI tables have similar errors. I love PCs :-)
> ron

Hello Ron,

I am so overjoyed by my experiment after the initial letter,
that I must point this out to the list. Since two weeks now
there has been a message that Linux could not find any routed
irq-signal for the usb_uhci part with every Coreboot image
that I so far have checked for my msi/ms6147. Based on the remark
I made in the letter, I now corrected by hand the relevant
entry in irq_tables.c to read

   {0x00,(0x07<<3)|0x2, {{0x60, 0xdeb8}, {0x61, 0xdeb8},
                         {0x62, 0xdeb8}, {0x63, 0xdeb8}},
          0x0, 0x0}, /* USB-UHCI */

Here, Getpir had all '0x00' instead of every desired linkage '0x6#'.

Building and testing the new Coreboot is an almost complete success:
the usb host controller is now getting an irq assigned and I can
attach and mount a flash memory to the usb port!

The only shortcoming is that my specifications in i82371eb/Config.lb
as to io and irq for this device 07:2 are completely ignored.
Could some kind soul tell me how and where I can achieve such
assignments at my will, and not having to trust some defaults
that Coreboot somehow and somewhere pulls out of the hat.

Being pleased with todays progress with USB, I send my regards

Mats Erik Andersson

More information about the coreboot mailing list