[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-devel] [PATCH 11/11] x86: support up to 16Tb

> From: Jan Beulich [mailto:JBeulich@xxxxxxxx]
> Subject: [Xen-devel] [PATCH 11/11] x86: support up to 16Tb
> Since TMEM doesn't currently cope with the full 1:1 map not always
> being visible, it gets forcefully disabled in that case.
> Signed-off-by: Jan Beulich <jbeulich@xxxxxxxx>

I agree this is the correct short-term (and maybe mid-term)
answer.  Anyone who can afford to fill their machine with
more than 5TiB of RAM is likely not very interested in
memory overcommit technologies :-) at least for the next
year or three.  Cloud providers may be an exception but
I'd imagine most of those are buying small- to mid-range
machines to optimize cost/performance, rather than
behemoths that expand to 5TiB+ which are highly performant
but often not cost-effective.

Longer term, zcache in Linux (which is a tmem-based technology)
successfully uses kmap/kunmap to run on 32-bit Linux OS's
so I'd imagine a similar technique could be used in Xen?

In any case, thanks Jan for remembering to handle tmem.

One nit below...

Acked-by: Dan Magenheimer <dan.magenheimer@xxxxxxxxxx>

> +        if ( opt_tmem )
> +        {
> +           printk(XENLOG_WARNING "Forcing TMEM off\n");
> +           opt_tmem = 0;
> +        }
> +    }

Maybe a bit more descriptive? I.e. "TMEM physical RAM limit
exceeded, disabling TMEM".

Xen-devel mailing list



Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.