Skip site navigation (1)Skip section navigation (2)
Date:      Sun, 30 Mar 2003 15:11:13 -0500
From:      Jake Burkholder <jake@locore.ca>
To:        Mike Silbersack <silby@silby.com>
Cc:        cvs-all@FreeBSD.org
Subject:   Re: HEADS UP Re: cvs commit: src/sys/conf options.i386 src/sys/i386/i386 bios.c locore.s machdep.c mpboot.s pmap.c vm86bios.s vm_machdep.c src/sys/i386/include _types.h bus_at386.h param.h pmap.
Message-ID:  <20030330201113.GA32298@locore.ca>
In-Reply-To: <20030330012410.I3292@odysseus.silby.com>
References:  <200303300524.h2U5Ora7061852@repoman.freebsd.org> <20030330061301.GC21973@locore.ca> <20030330070723.GE21973@locore.ca> <20030330012410.I3292@odysseus.silby.com>

next in thread | previous in thread | raw e-mail | index | archive | help
Apparently, On Sun, Mar 30, 2003 at 01:31:18AM -0600,
	Mike Silbersack said words to the effect of;

> 
> On Sun, 30 Mar 2003, Jake Burkholder wrote:
> 
> > To clarify that the ram above 4G is used for, it just goes into the general
> > page pool.  I don't intend to implement a means for user process's to access
> > more then their ~2.5G address space through a sliding window as has been done
> > on other systems, but this should be quite easy to do should someone be so
> > inclined.  To give an example, on a 6G system you see things like this:
> 
> Cool, that's much better than the situation was for large ram machines
> before. :)
> 
> Do these changes allow something like a 3G KVA space without shrinking
> processes address spaces?

No, it doesn't make the virtual address space any bigger, it just allows
more physical memory.  This is a bit of a problem because the tunables that
are based on physical memory size don't scale well past 4G of ram, its easy
to end up with may too many vnodes.

> 
> Also, I'm assuming that PAE can boot on machines with < 4 Gig of ram; can

Yes, you can use PAE on small memory machines.  This will at least give you
compiler warnings about truncating physical addresses in most cases.

> it also be coaxed into acting in such a manner than busdma is _required_,
> so that a 256MB i386 box can be used to see if a driver is busdma
> compliant?

Not really.  The best way is to buy a sparc :).  I suppose that you could
create your dma tags such that busdma thinks it needs to bounce, this would
at least test that you've got the right bus_dmamap_syncs. ie set lowaddr to
below the highest physical address in your machine.

> 
> In any case, very cool.

Thanks!

Jake



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20030330201113.GA32298>