[lkml]   [2000]   [Mar]   [24]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
Messages in this thread
SubjectRe: VM modules in kernel?
"Mike A. Harris" wrote:
> On Fri, 24 Mar 2000, Alan Cox wrote:
> >Date: Fri, 24 Mar 2000 17:40:48 +0000 (GMT)
> >From: Alan Cox <>
> >To:
> >Cc: Linux Kernel mailing list <>
> >Subject: Re: VM modules in kernel?
> >
> >> I was thinking that faster VM speeds may be possible if the
> >> kernel can be tweaked more freely due to the GPL nature of the
> >> Plex86/bochs projects now.
> >
> >Remember something here. IBM tuned the hardware to this, and to an extent
> >they tuned the software on top of VM. They have a lot of cards to play that
> >Motorola m68K chips did but x86 does not.
> I'm not sure I understand what you're saying here Alan... My
> guess is you're taking VM to mean the kernel virtual memory
> system, when I'm meaning it to be "virtual machine". So my
> sentence really should read 'faster virtual machine speeds'. I
> remember a thread earlier this year on IBM tweaking the virtual
> memory system so I think that is what you're talking about..

Think Alan is refering to VM/ESA. It an OS that creates virtual machines.
In these virtual machines you can run almost any OS (including Linux) that
is coded for a S390. To make things fast enough IBM created microcode and
special instructions to support a virtual supervisor mode. ie

VM hypervisor (real supervisor mode)
guest os (virtual supervisor mode)

> If this is the case, then it is my mistake for using an unclear
> abbreviation. ;o)
> >Im hoping transmeta can manage to add a virtualised 386 mode to their chip.
> >Here's hoping they are listening ;)

I hope so too. This would be a very very nice thing to have. Of all the IBM
system I have used over the years VM/ESA is the nicest. I miss being able to
test new OS version in a virtual machine... I have worked more nights and
weeks since I switched to the Unix than I have while using IBM's VM.

> >In the Linux case we have another weapon (other than fear suprise and the
> >cuddly penguin) which is that we don't give two hoots[1] if the kernel running
> >on the hypervisor is not a normal x86 kernel. If it provides the APIs and
> >it runs x86 user space its fine.
> Hmm.. You've lost me a bit here.. Let me guess what you're
> getting at..
> The 'hypervisor' is a new term to me.. Are you refering to the
> Transmeta CPU and it's native mode? If I read you correctly
> here, then we could have a native VLIW kernel running on the
> Crusoe, which opens an x86 API to userland? Correct me if I'm
> wrong.

hypervisor, software and hardware that lets you run multiple virtual
machines (CPU(s), memory, devices (real and virtual)) under its control
all on the same real box.

> >Consider the user mode kernel as a more interesting starting point.


Ed Tomlinson <>

To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to
Please read the FAQ at

 \ /
  Last update: 2005-03-22 13:57    [W:0.023 / U:0.260 seconds]
©2003-2018 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site