Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 10 Dec 2007 21:37:12 -0500
From:      Travis Mikalson <bofh@terranova.net>
To:        =?ISO-8859-1?Q?S=F8ren_Schmidt?= <sos@deepcore.dk>,  freebsd-current@freebsd.org
Subject:   Re: Any successful installs on a Broadcom HT1000 chipset?
Message-ID:  <475DF7D8.2020203@terranova.net>
In-Reply-To: <475CEA07.1050500@deepcore.dk>
References:  <73807.10710.qm@web63912.mail.re1.yahoo.com>	<200711280842.09340.jhb@freebsd.org> <474D726A.8080807@deepcore.dk>	<200711280938.38545.jhb@freebsd.org>	<474E5B69.7070406@yandex.ru> <474E65D6.4040403@deepcore.dk>	<474E69AE.7000105@yandex.ru> <475978E1.2090507@deepcore.dk>	<475C6C3E.6070004@deepcore.dk> <475CC426.3060808@terranova.net> <475CCAA0.5040900@terranova.net> <475CEA07.1050500@deepcore.dk>

next in thread | previous in thread | raw e-mail | index | archive | help

Søren Schmidt wrote:
> Yep, I'm not convinced there are "generic" DMA problems with that 
> chipset, but the ATA part definitively has trouble, I'm (slowly) working 
> my way through the different results here to try pinpoint the problem 
> more precisely.
> I'm also going to try a Marvell ctlr later today, more news later...

Ah, very good, I was going to ask if you needed a Marvell PCI-X controller.

You're by far the expert compared to me in this arena, but I was having 
the exact same issue (including the memory corruption) when booting from 
the Marvell PCI-X SATA  controller... smells like a somewhat generic 
HT1000 barfing-on-DMA problem to my inexperienced nose :)

>> I also see this isn't first and only chipset to have the exact same 
>> dma max_iosize limit imposed :)
> Right, the usual need for this limit is that the 64K size means that the 
> count reg is set to zero, and some HW designers just didn't get that right.
> 
> In this case its different as it does not always fail, but I havn't 
> found the combo that makes it fail yet. However the workaround seems to 
> be quite solid, but there might be a better / more correct way to solve 
> it still.

Perhaps something could be devised involving a kernel option that works 
around the errata? Wouldn't be the first time. options 
HT1000_WORKAROUND... would suck if it involved making an ifdef for every 
ata driver :(

-- 
TerraNovaNet Internet Services - Key Largo, FL
Voice: (305)453-4011 x101   Fax: (305)451-5991
http://www.terranova.net/
----------------------------------------------
Life's not fair, but the root password helps.



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?475DF7D8.2020203>