The ol' Linux-on-Linux

[Samuel Rydh]

On Thu, Jun 09, 2005 at 01:14:44AM -0700, Dan Potter wrote:
> On Jun 8, 2005, at 11:34 AM, Dan Potter wrote:
> 
> >With a few minutes' worth of research this morning I have a fully  
> >functional Debian sid system booting up inside MOL now. It's still  
> >using the 2.4.26 kernel, but I'm hoping to work on that tonight.

Great!

> The LOL patches are now applied to a 2.4.31 kernel which I'm using  
> for testing. Using that and an rsync'd copy of MOL from this morning  
> on a 2.6.11 host kernel, I'm able to get quite good performance for  
> both CPU and disk access. The time to untar a 2.6.8 kernel source  
> tree onto a native partition is close enough to the native host speed  
> to practically be a rounding error. Very cool!

Well, there is a actually quite a lot of room for performance improvements.
The supplies drivers are for instance quite primitive (much more so
than the ones supplied for OSX).

> The only thing that seems to be sluggish is network performance. I  
> tried both the sheep and tun drivers, and both seem to top out at  
> about 16Mbit/sec on a 100Mbit network. The tun driver also had an rx  
> buffer full error while trying out a large transfer. With native  
> machines on the same network I can get about 60Mbit/sec. Any idea  
> what's going on there or how it might be improved?

I would guess this is due to the linux sider drivers. For instance,
the block device driver is completely synchronous (the interface
intended for boot loaders is used) and any operation will
completely block interrupts.

> Samuel, even though this stuff still isn't 100% "primetime", do you  
> have any problems with me posting it up for others to try out as well?

No, not at all.

/Samuel