Xen users - Nov 2006 - How long is a kernel supported?

Hi all!

I have a question regarding the release cycle of Xen. I''ve compiled an
Xen-able kernel with Xen 3.0.3_0 what means the "Makefile" downloaded
linux-kernel 2.6.16.29.

Ok, that is no problem. I''m flexible in choosing a certain kernel. BUT
I''m not flexible in changing the kernel version in frequent intervals.
That leads me to some questions:

A. How long will be Xen 3.0.3_0 supported?

B. How long will I be able to compile Xen-able linux-kernels ver.
2.6.16.xy with Xen 3.0.3_0?

C. Will I be able to compile Xen-able linux-kernels ver. 2.6.16.xy with
other versions than Xen 3.0.3_0?

Thanks for answering!

Goran


_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

I also also copy this message to xen-devel, because developers should
say something about this.

On 11/19/06, Goran <xamiw@arcor.de> wrote:
> [...]
> A. How long will be Xen 3.0.3_0 supported?
If "support" means for you something like bugfix or security releases
that keep most of the API and configuration, I have not seen such
thing in Xen 3. Each release is significantly different in terms of
kernel-version and -configuration (xend and vm configs keep _mostly_
the same), even if the developers decide to go only minor steps in the
version numbers, which I really don''t understand because the common
conception of minor version changes is that only bad bugs and security
problems get fixed, everything else keeps the same.
I''ve not seen any security release in these terms and my questions
about these things on xen-devel remained unanswerded.

Maybe you need to buy the Enterprise version and buy support to get
information about these things and to get security updates (hmm, but
if they give security updates to the kernel to customers, they have to
make them public, also.)

On the other hand, I''ve not heard from xen security problems until
now, but that''s a bad reason to have no security information for users
at all.
> B. How long will I be able to compile Xen-able linux-kernels ver.
> 2.6.16.xy with Xen 3.0.3_0?
I don''t understand this question exactly. As far as I understand it:
I''ve not seen a Xen Release where only the Kernel was updated.
Trying to use an other Kernel means, you have to see yourself to get
the most current Xen patches into the Kernel, and you probably have to
work with testing/unstable Xen and Xen kernel Versions.
>
> C. Will I be able to compile Xen-able linux-kernels ver. 2.6.16.xy with
> other versions than Xen 3.0.3_0?
The developers can tell about the future, in the past, later versions
of Xen where incompatible to older Kernels AFAIK. And the Xen Linux
Patches where for exact the Linux version that comes with the sources.

It would be really nice to hear comments from the developers here.
Maybe they hear and reply one day.


Henning

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel

On 21/11/06 12:50, "Henning Sprang" <henning_sprang@gmx.de>
wrote:
>> C. Will I be able to compile Xen-able linux-kernels ver. 2.6.16.xy with
>> other versions than Xen 3.0.3_0?
> 
> The developers can tell about the future, in the past, later versions
> of Xen where incompatible to older Kernels AFAIK. And the Xen Linux
> Patches where for exact the Linux version that comes with the sources.
> 
> It would be really nice to hear comments from the developers here.
> Maybe they hear and reply one day.
We support older XenLinux trees built against more recent Xen, but not vice
versa. We support older XenLinux binaries running on newer Xen, *and* newer
XenLinux binaries running as far back as Xen 3.0.2.

 -- Keir



_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel

Observation:
------------
When connecting two miniOs (using a shared ring), Xen (not a domain)
crashes when the miniOs''s exits..

Xen crashes and produces the following: 
(XEN) Xen call trace:
(XEN)    [<ff11d20d>] __bug+0x29/0x45
(XEN)    [<ff107cb3>] gnttab_release_mappings+0xcb/0x2e5
(XEN)    [<ff1046dd>] domain_kill+0x29/0x62
(XEN)    [<ff10349a>] do_domctl+0x6d6/0xfbc
(XEN)    [<ff165755>] hypercall+0x95/0xb5
(XEN)
(XEN)
(XEN) ****************************************
(XEN) Panic on CPU 1:
(XEN) BUG at grant_table.c:1122
(XEN) ****************************************


The cause:
----------
Xen tries to release the grant table mappings by accessing a remote
domain grant table. 
But the remote domain seems to be non-existent and consequently Xen
fails:
find_domain_by_id (in gnttab_release_mappings) returns NULL.


Analysis:
---------
This situation described above should never happen: if I understand
correctly, a domain should not be completely destroyed until there are
no more references to it.
See: put_domain(d) // sched.h
Which is defined as follows:
If ( atomic_dec_and_test( &(_d_->refcnt) ) domain_destroy(_d)

It does however happen when a domain crashes.

Note that there are two ways to "finish" with a domain (domain.c):
1.	domain_kill (which calls domain_destroy) - releases all
resources in a gracefull 
      manner.
2.	__domain_crash (which calls domain_shutdown) - which seems to
kill the domain 
      without proper releasing of resources that reference to it.. 
     (this function is called on extreme cases)


Our scenario:
-------------

We are running two miniOs with the same profile:
Open a ring (share a page with a grant ref and map a page from a remote
domain)
Write
Read
Close the ring (dealloc, unmap*)
do_exit()



Timeline - > 
MiniOs 1:  ..........         calls do_exit() -> 
                                     .. domain_kill() -> 
                                            .. gnttab_release_mapping()
-> 
                                                    .. BUG()

MiniOs 2:    crashes**

                 
*When we unmap we use Xen''s hypercall to unmap a grant reference 
and the gnttab_unmap_grant_ref structure.
Note that we have a bug and do NOT set unmap_op.dev_bus_addr to 0 as we
should.
Xen''s API (in public/grant_table.h) explicitly describes that it should
be 0 or 
the grant reference will be treated as valid device mapping. 

** Because of the bug descrived in * we cause the domain to crash.
We observe:
(XEN) grant_table.c:394: Bad frame number doesn''t match gntref
(XEN) mm.c:760: Attempt to implicitly unmap a granted PTE 
(XEN) domain_crash called from mm.c:761 



Summary:
-----------

1. Setting unmap_op.dev_bus_addr removes the BUG and all is well.
2. But crashing Xen - even with our error - doesn''t seem to be a
healthy
choice.



:) 
Micha.

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel

On 21/11/06 14:09, "Moffie, Micha" <micha.moffie@hp.com> wrote:
> 1. Setting unmap_op.dev_bus_addr removes the BUG and all is well.
> 2. But crashing Xen - even with our error - doesn''t seem to be a
healthy
> choice.
Does the attached patch fix this issue? It''s against unstable tip --
may
apply to your tree with patch fuzz.

 -- Keir



_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel

Yes, fixed. 

> +        domain_crash(ld); /* naughty... */
By why do you crash the local domain if the reference can not be
ungranted?
Is there no way to clean up? And if ld == 0?

Thanks,
:)
Micha.

-----Original Message-----
From: Keir Fraser [mailto:keir@xensource.com] 
Sent: 21 November 2006 14:33
To: Moffie, Micha; xen-devel
Subject: Re: [Xen-devel] Xen Crashes when releasing gnttab mappings - of
a crashed domain. 

On 21/11/06 14:09, "Moffie, Micha" <micha.moffie@hp.com> wrote:
> 1. Setting unmap_op.dev_bus_addr removes the BUG and all is well.
> 2. But crashing Xen - even with our error - doesn''t seem to be a 
> healthy choice.
Does the attached patch fix this issue? It''s against unstable tip --
may
apply to your tree with patch fuzz.

 -- Keir


_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel

On 21/11/06 15:52, "Moffie, Micha" <micha.moffie@hp.com> wrote:
> By why do you crash the local domain if the reference can not be
> ungranted?
> Is there no way to clean up? And if ld == 0?
This situation signals that the mapping domain cleared the granted mapping
by simply zapping the PTE, rather than going thru the explicit grant-unmap
operation. The effect of this is undefined (although of course it should not
crash Xen!) and for sanity''s sake I prefer to crash the offender
whenever I
can detect that this has happened.

 -- Keir


_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xensource.com
http://lists.xensource.com/xen-devel

Am Dienstag, den 21.11.2006, 13:50 +0100 schrieb Henning
Sprang:
> [...]
> On 11/19/06, Goran <xamiw@arcor.de> wrote:
> > [...]
> > A. How long will be Xen 3.0.3_0 supported?
> 
> If "support" means for you something like bugfix or security
releases
> that keep most of the API and configuration, I have not seen such
> thing in Xen 3.
> [...]
Yes, that''s what I mean. I just want to check the repository every
night
and if a change is found to recompile xen-hypervisor, xen-tools,
xen-kernel, xen-modules pack it into deb-packages and put it on my
private debian ftp-server for all my xen-machines. BUT where''s the
repository?! Is it http://xenbits.xensource.com/xen-3.0.3-testing.hg ?
> On the other hand, I''ve not heard from xen security problems until
> now, but that''s a bad reason to have no security information for
users
> at all.
> 
> > B. How long will I be able to compile Xen-able linux-kernels ver.
> > 2.6.16.xy with Xen 3.0.3_0?
> 
> I don''t understand this question exactly. As far as I understand
it:
> I''ve not seen a Xen Release where only the Kernel was updated.
> [...]
Ahh...
> Trying to use an other Kernel means, you have to see yourself to get
> the most current Xen patches into the Kernel, and you probably have to
> work with testing/unstable Xen and Xen kernel Versions.
> [...]
Right, but if Xen wants to get into the Linux kernel Xen must be
available in every kernelversion after entering the kernel (e.g
2.6.16.ab, 2.6.17.cd and so on should contains Xen). I know, Xen is
still not available in the official kernel but why can''t I download
just
a patch for my desired kernel? Xen-hypervisor and Xen-tools should be
independent by kernelversion XOR available for every kernelversion. 
> >
> > C. Will I be able to compile Xen-able linux-kernels ver. 2.6.16.xy
with
> > other versions than Xen 3.0.3_0?
> 
> The developers can tell about the future, in the past, later versions
> of Xen where incompatible to older Kernels AFAIK. And the Xen Linux
> Patches where for exact the Linux version that comes with the sources.
> 
> It would be really nice to hear comments from the developers here.
> Maybe they hear and reply one day.
> 
> 
> Henning
Yeah, nice. A developer answered...

Am Dienstag, den 21.11.2006, 13:46 +0000 schrieb Keir Fraser:
> [...]
> We support older XenLinux trees built against more recent Xen, but not vice
> versa. We support older XenLinux binaries running on newer Xen, *and* newer
> XenLinux binaries running as far back as Xen 3.0.2.
> 
>  -- Keir
That''s good to hear but one question. Where is the repository of the
stable Xen-versions? E.g. 3.0.2? I just can find
http://xenbits.xensource.com/xen-3.0.3-testing.hg .

Goran


_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users

On 21/11/06 18:05, "Goran" <xamiw@arcor.de> wrote:
>> We support older XenLinux trees built against more recent Xen, but not
vice
>> versa. We support older XenLinux binaries running on newer Xen, *and*
newer
>> XenLinux binaries running as far back as Xen 3.0.2.
>> 
>>  -- Keir
> That''s good to hear but one question. Where is the repository of
the stable
> Xen-versions? E.g. 3.0.2? I just can find
> http://xenbits.xensource.com/xen-3.0.3-testing.hg .
http://xenbits.xensource.com/xen-3.0-testing.hg. Look at ''hg
tags''. We moved
to a new repo-per-release scheme for 3.0.3.

 -- Keir


_______________________________________________
Xen-users mailing list
Xen-users@lists.xensource.com
http://lists.xensource.com/xen-users