How many lines of Xen source code

How many lines of Xen source code

[Xin Zhao]

Does someone have a stastics of xen source code?

Thereotically, Xen can be a trust computing base because it is simpler than standard operating
system. But do we have some concrete data to support this claim?

Also, Dom0 is pretty much a standard operating system. If it should be
considered as TCB, we might have a lot harder time to argue why it is
simpler. One way is to argue that services and applications running in
Dom0 can be reduced to minimum. But how to count the source code at this
point?

Thanks,
-x

Re: How many lines of Xen source code

[Anthony Liguori]

Xin Zhao wrote:

>Does someone have a stastics of xen source code?
>  
>
SLOC    Directory       SLOC-by-Language (Sorted)
30945   arch            ansic=28618,asm=2121,sh=206
21382   include         ansic=21382
6629    common          ansic=6629
3268    drivers         ansic=3268
0       top_dir         (none)


Totals grouped by language (dominant language first):
ansic:        59897 (96.26%)
asm:           2121 (3.41%)
sh:             206 (0.33%)

generated using David A. Wheeler's 'SLOCCount'

Regards,
Anthony Liguori

Re: How many lines of Xen source code

[Xin Zhao]

Many thanks! Your information is very helpful! :)

-x

On Fri, 1 Apr 2005, Anthony Liguori wrote:

> Xin Zhao wrote:
>
> >Does someone have a stastics of xen source code?
> >
> >
> SLOC    Directory       SLOC-by-Language (Sorted)
> 30945   arch            ansic=28618,asm=2121,sh=206
> 21382   include         ansic=21382
> 6629    common          ansic=6629
> 3268    drivers         ansic=3268
> 0       top_dir         (none)
>
>
> Totals grouped by language (dominant language first):
> ansic:        59897 (96.26%)
> asm:           2121 (3.41%)
> sh:             206 (0.33%)
>
> generated using David A. Wheeler's 'SLOCCount'
>
> Regards,
> Anthony Liguori
>

RE: How many lines of Xen source code

[Neugebauer, Rolf]

You are right, at the moment both Xen and Dom0 are part of the TCB, but
there are a few activities going on here.

There is some effort in reducing the size of xen by moving some of the
platform init code out of Xen. This should significantly reduce the
source code size of xen as we can get rid of most of the ACPI and PCI
related code. This is primarily a portability related issue, essentially
being able to leverage better standard linux platform init code, but
also has the side effect of making Xen simpler.

Within a security context we are also actively looking at reducing the
TCB, basically deconstructing Dom0. Fine grained MAC at it's current
operation is part of the picture as is delegating some of it's
functionality to other domains, in particular the IO functions it
currently performs on behalf of other domains, but also things like
domain building. 

The issue of reducing the TCB will probably play a major part in a
meeting prior to the xen summit next week. We plan to report back to the
list.

Rolf



> -----Original Message-----
> From: xen-devel-bounces@lists.xensource.com [mailto:xen-devel-
> bounces@lists.xensource.com] On Behalf Of Xin Zhao
> Sent: 01 April 2005 20:58
> To: xen-devel@lists.xensource.com
> Subject: [Xen-devel] How many lines of Xen source code
> 
> Does someone have a stastics of xen source code?
> 
> Thereotically, Xen can be a trust computing base because it is simpler
> than standard operating
> system. But do we have some concrete data to support this claim?
> 
> Also, Dom0 is pretty much a standard operating system. If it should be
> considered as TCB, we might have a lot harder time to argue why it is
> simpler. One way is to argue that services and applications running in
> Dom0 can be reduced to minimum. But how to count the source code at
this
> point?
> 
> Thanks,
> -x
> 
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xensource.com
> http://lists.xensource.com/xen-devel

Re: How many lines of Xen source code

[Jacob Gorm Hansen]

Neugebauer, Rolf wrote:

> Within a security context we are also actively looking at reducing the
> TCB, basically deconstructing Dom0. Fine grained MAC at it's current
> operation is part of the picture as is delegating some of it's
> functionality to other domains, in particular the IO functions it
> currently performs on behalf of other domains, but also things like
> domain building. 
> 
> The issue of reducing the TCB will probably play a major part in a
> meeting prior to the xen summit next week. We plan to report back to the
> list.

If any Xen people are going to USENIX NSDI in Boston in May, perhaps it 
would be a good idea to arrange a Xen BOF session there?

Jacob