From: Jeff Dike <jdike@ka...> - 2000-12-11 05:53:58
> I got one customer to switch to it; he stayed for only 24 hours before
> requesting to be sent back to his old service, due to lag and
You're bolder than I would have been. At this point, I would not recommend
putting paying customers inside a UML :-)
> That's why I made that modification to the virtual ethernet code, in
> the hopes that it would have made his experience better.
I think that there are worse problems than the ethernet driver. I'm going to
make some changes in how system call interception is done and to the block
driver when I start taking advantage of 2.4/2.5 features, and then you'll
start seeing reasonable performance.
From: Jeff Dike <jdike@ka...> - 2000-12-11 17:24:52
> Not only that, but I can make diffs of my work against a clean UML
> tree and the resulting patch applies cleanly to the normal kernel
> tree. I'm actually surprised this works so well, and I'm reading your
> patch now to find out why.
The reason why is that I changed almost nothing in the generic kernel.
There's a hook or two in the block and network layers, but that's about it.
So, if you stay away from the arch layer, your diffs are going to be in
generic-kernel-land and they will apply cleanly to any other generic kernel.
> It's a safe guess that UML wouldn't be a lot of use for debugging the
> Linux thread mechanism itself
Threads are under the arch interface, UML would be useless in debugging the
threads of a different arch, just like an Alpha is useless in debugging the
> Today I'm moving on to running Tux2 on UML.
Cool. I was wondering if you had done that yet. I would also like to make
Tux2 part of my releases whenever you think it's ready.
> So I'd like it very much if we could put our heads together and come
> up with a way of loading getting gdb to load the module symbols
> automatically. I know it's hard :-).
It might not be. gdb is programmable enough that a .gdbinit could probably
define enough sugar to make loading and reloading modules fairly painless.
From: Jeff Dike <jdike@ka...> - 2000-12-11 21:26:41
> > On Mon, Dec 11, 2000 at 05:49:41PM +0000, James Stevenson wrote:
> > Hi
> > do you know if uml will be accepted into the main kernel
> > tree ?
> I'm sure it will not for 2.4. For 2.5/2.6, who knows ...
Not for 2.4.0. Alan has said he'd take it, though, so I'm aiming at an early