Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 7 Jan 2000 09:46:41 -0800 (PST)
From:      Matthew Jacob <mjacob@feral.com>
To:        Andrew Gallatin <gallatin@cs.duke.edu>
Cc:        De la Cruz Lugo Eric <eric@iteso.mx>, freebsd-alpha@FreeBSD.ORG, sos@FreeBSD.ORG
Subject:   Re: HARDWARE.TXT specific for Alpha?
Message-ID:  <Pine.BSF.4.10.10001070941310.8867-100000@beppo.feral.com>
In-Reply-To: <14454.1900.901358.626546@grasshopper.cs.duke.edu>

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

> The problem is that we need to be able to DMA to/from any address is
> RAM.  On most platforms, the default direct-mapped DMA window is 1GB
> in size, so we cannot exceed 1GB.  It is possible to alter the number,
> size & placement of such windows so as to increase the amount of RAM
> one can DMA to/from.  I have done this for the 21264s which is why
> they can have 2GB.
> 
> Another solution would be to set up temporary scatter/gather dma
> mappings for all DMA & not just ISA DMA.  This would alow memory sizes
> greater than 4GB.  This would require changes to many device drivers
> and all drivers would need to support this mechanism to be able to use 
> it. I don't see this happening any time soon.

It's easy to enforce this, btw- just remove the alpha_XXX_dmamap so that
drivers have to use the busdma map routines instead of the compatibility
define of vtophys. This is why I mentioned this to Soren about the
so-called 'new' ATA driver (for which he got unbelievingly annoying in
return about).

This isn't going to happen any time soon as a default- there are too many
drivers even in NetBSD which is more stringent about this than FreeBSD
that still depend on this. However, we'll probably do something like add
an option that says "use large memory" that will force the issue.


-matt



To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-alpha" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.BSF.4.10.10001070941310.8867-100000>