Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 14 Nov 2008 22:11:19 +0100
From:      Marius Strobl <marius@alchemy.franken.de>
To:        Mark Linimon <linimon@lonesome.com>
Cc:        freebsd-sparc64@freebsd.org
Subject:   Re: Free Ultra2 in Silicon Valley, USA
Message-ID:  <20081114211118.GG64456@alchemy.franken.de>
In-Reply-To: <20081114161933.GA24688@soaustin.net>
References:  <183638.12752.qm@web56802.mail.re3.yahoo.com> <20081031131827.GA9613@soaustin.net> <20081103223042.GB8256@alchemy.franken.de> <20081104115722.GA28394@soaustin.net> <20081104221003.GE31338@alchemy.franken.de> <20081105184746.GA26875@soaustin.net> <20081112201029.GE64456@alchemy.franken.de> <oqy6zoisrr.fsf@castrovalva.Ivy.NET> <20081112213543.GA71577@alchemy.franken.de> <20081114161933.GA24688@soaustin.net>

next in thread | previous in thread | raw e-mail | index | archive | help
On Fri, Nov 14, 2008 at 10:19:33AM -0600, Mark Linimon wrote:
> It turns out the T1-200s are much happier to boot if you actually
> include the gem(4) driver in the kernel.
> 
> Sigh.
> 
> Anyways, thanks for the help investigating.
> 

Ah, the problem then likely is that the GEMs are left initialized
and running by the firmware; at some point, probably when some
packet is received, the GEM DMAs something to a mapping the
IOMMU no longer knows about since the kernel has taken it over
and thus triggers a DMA error interrupt.
If this happens when netbooting then it's probably time to
fix libstand to no longer open and close the network device
for every file access so we can remove the hack form the
loader which just keeps the device open forever. On the other
hand, it's probably beneficial in general to not remove the
driver for the device one wants to netboot with :)

Marius




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20081114211118.GG64456>