Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 3 Sep 2008 13:00:52 -0400
From:      "Don Bowman" <don@sandvine.com>
To:        "Ivan Voras" <ivoras@freebsd.org>, <freebsd-hardware@freebsd.org>
Subject:   RE: FreeBSD-AMD64 on Xeon MP
Message-ID:  <EB618291F3454E4DA10D152B9045C017016B923D@exchange-2.sandvine.com>
In-Reply-To: <g9lgk8$c2e$1@ger.gmane.org>
References:  <EB618291F3454E4DA10D152B9045C017016B91EA@exchange-2.sandvine.com> <g9lgk8$c2e$1@ger.gmane.org>

next in thread | previous in thread | raw e-mail | index | archive | help
>From Ivan Voras
> Don Bowman wrote:
>=20
> > I tried putting a call to uhci_dumpregs() there, it hangs as well.
So
> > my assumption Is that a read to 0x3080 is hanging.
> >
> > Anyone have any suggestions?
>=20
> If the same thing that worked for me works for you (booting Linux
> before soft-rebooting into FreeBSD) than it would look like something
> isn't initialized right.

I am focusing in on SMM emulation.
http://www.ussg.indiana.edu/hypermail/linux/kernel/0407.3/0688.html
suggests something to me.

What if I am executing this on a processor other than the boot
processor,
and the SMM code doesn't expect this?

I'm not sure where I would try this patch in freebsd, but it seems
promising.

Ie I read 0xe090. This trips SMM for legacy(?). SMM is entered, does=20
something, corrupts, leaves, and then a few instructions later I crash.

The suggested workaround for this (different machine and OS) is to
disable
the legacy mode. FreeBSD seems to do this already I think, calling:
pci_write_config(self, PCI_LEGSUP, PCI_LEGSUP_USBPIRQDEN, 2); in
uhci_pci_attach()





Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?EB618291F3454E4DA10D152B9045C017016B923D>