preparing for 4.3.1

preparing for 4.3.1

[Jan Beulich]

Aiming at a release later in October (before Xen Summit I would
hope), I'd like to cut RC1 next week.

Please indicate any bug fixes that so far may have been missed
in the backports already done.

Jan

Re: preparing for 4.3.1

[Andrew Cooper]

On 27/09/13 11:07, Jan Beulich wrote:
> Aiming at a release later in October (before Xen Summit I would
> hope), I'd like to cut RC1 next week.
>
> Please indicate any bug fixes that so far may have been missed
> in the backports already done.
>
> Jan

In terms of bugs, it would be very nice for my migration stream
regression fix to get in, although I suspect that the answer will be
"its too close to the cutoff and still not in unstable yet".  FWIW,
given the lack of support for remus in libxl in 4.3, the backport is
substantially smaller than what needs to go into unstable.  I guess this
call goes to either of the Ian's

The first of my per_cpu() fixes (posted yesterday) really needs
backporting, although there has been curious silence to series so far. 
The second of the series would also be a very good idea.  (Given the
damage it is causing to XenServer trunk using Xen 4.3 and Linux 3.10 for
dom0, I am amazed we are the first people to find this, although it is
likely an interaction with our stats collecting daemon)

I know these are not strictly bugfixes but the 5 rombios fixes (top five
commits to tools/firmware/rombios/rombios.c) should be considered.  They
combine to a noticable reduction in time taken to boot HVM domains
(windows in particular which loads all of its core data with int $0x13),
as well as a substantial reduction in the spike qemu cpu usage in dom0. 
FWIW, Recent XenServers have shipped with these patches backported to
4.1, and it was only my lack of repeated prodding which prevented it
going into 4.3 in the first place.

There is nothing else obvious in the XenServer patch queue.

~Andrew

Re: preparing for 4.3.1

[Ian Campbell]

On Fri, 2013-09-27 at 11:35 +0100, Andrew Cooper wrote:
> On 27/09/13 11:07, Jan Beulich wrote:
> > Aiming at a release later in October (before Xen Summit I would
> > hope), I'd like to cut RC1 next week.
> >
> > Please indicate any bug fixes that so far may have been missed
> > in the backports already done.
> >
> > Jan
> 
> In terms of bugs, it would be very nice for my migration stream
> regression fix to get in, although I suspect that the answer will be
> "its too close to the cutoff and still not in unstable yet".  FWIW,
> given the lack of support for remus in libxl in 4.3, the backport is
> substantially smaller than what needs to go into unstable.  I guess this
> call goes to either of the Ian's

Has the backported version been posted somewhere? It's not got very much
to do with the mainline patch IIRC.

Re: preparing for 4.3.1

[Andrew Cooper]

On 27/09/13 11:41, Ian Campbell wrote:
> On Fri, 2013-09-27 at 11:35 +0100, Andrew Cooper wrote:
>> On 27/09/13 11:07, Jan Beulich wrote:
>>> Aiming at a release later in October (before Xen Summit I would
>>> hope), I'd like to cut RC1 next week.
>>>
>>> Please indicate any bug fixes that so far may have been missed
>>> in the backports already done.
>>>
>>> Jan
>> In terms of bugs, it would be very nice for my migration stream
>> regression fix to get in, although I suspect that the answer will be
>> "its too close to the cutoff and still not in unstable yet".  FWIW,
>> given the lack of support for remus in libxl in 4.3, the backport is
>> substantially smaller than what needs to go into unstable.  I guess this
>> call goes to either of the Ian's
> Has the backported version been posted somewhere? It's not got very much
> to do with the mainline patch IIRC.
>
>

If the unstable patch gets accepted (even if only in principle), I will
happily construct the backport for 4.3.  It will basically contain the
Xend and libxc bits, with the libxl callsights hardcoded with a constant.

~Andrew

Re: preparing for 4.3.1

[Dario Faggioli]

[-- Attachment #1.1: Type: text/plain, Size: 1042 bytes --]

On ven, 2013-09-27 at 11:07 +0100, Jan Beulich wrote:
> Aiming at a release later in October (before Xen Summit I would
> hope), I'd like to cut RC1 next week.
> 
> Please indicate any bug fixes that so far may have been missed
> in the backports already done.
> 

These two are both bugfixes, although the second one affects debug
builds only:

 * "xl: fix `xl cpupool-list' behavior in case no pool name is provided"
   git cherry-pick 3998afdbf99959582dcd9f9f4df5a6fe7ce4ded8

 * "sched_credit: filter node-affinity mask against online cpus"
   git cherry-pick 5e5a44b6c942d6ea47f15d6f1ed02b03e0d69445

The first one affects 4.2 too, and I already requested backport (and I
think IanJ said its on his list). The second only affects 4.3.

Regards,
Dario

-- 
<<This happens because I choose it to happen!>> (Raistlin Majere)
-----------------------------------------------------------------
Dario Faggioli, Ph.D, http://about.me/dario.faggioli
Senior Software Engineer, Citrix Systems R&D Ltd., Cambridge (UK)


[-- Attachment #1.2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

[-- Attachment #2: Type: text/plain, Size: 126 bytes --]

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

Re: preparing for 4.3.1

[Jan Beulich]

>>> On 27.09.13 at 12:35, Andrew Cooper <andrew.cooper3@citrix.com> wrote:
> In terms of bugs, it would be very nice for my migration stream
> regression fix to get in, although I suspect that the answer will be
> "its too close to the cutoff and still not in unstable yet".  FWIW,
> given the lack of support for remus in libxl in 4.3, the backport is
> substantially smaller than what needs to go into unstable.  I guess this
> call goes to either of the Ian's

Correct.

> The first of my per_cpu() fixes (posted yesterday) really needs
> backporting, although there has been curious silence to series so far. 
> The second of the series would also be a very good idea.  (Given the
> damage it is causing to XenServer trunk using Xen 4.3 and Linux 3.10 for
> dom0, I am amazed we are the first people to find this, although it is
> likely an interaction with our stats collecting daemon)

I can only approve the x86 part of it, and George (had you
Cc-ed him) could only approve the scheduler part of it. As in
any event I'd like to see Keir's opinion (in form of an ack or
whatever) on this sort of a change (including the second one),
it'll be unavoidable to wait for his reply. But yes, once in these
two would be backporting candidates even without you asking
for it.

Additionally, for both items you brought up above: The call for
backport requests is really mainly for stuff already in unstable
(and generally also already past the push gate). Therefore I
take all of the above only as a general hint towards a request
to come...

> I know these are not strictly bugfixes but the 5 rombios fixes (top five
> commits to tools/firmware/rombios/rombios.c) should be considered.  They
> combine to a noticable reduction in time taken to boot HVM domains
> (windows in particular which loads all of its core data with int $0x13),
> as well as a substantial reduction in the spike qemu cpu usage in dom0. 
> FWIW, Recent XenServers have shipped with these patches backported to
> 4.1, and it was only my lack of repeated prodding which prevented it
> going into 4.3 in the first place.

Again this is the tool stack maintainers' call.

> There is nothing else obvious in the XenServer patch queue.

Good to know.

Jan

Re: preparing for 4.3.1

[Jan Beulich]

>>> On 27.09.13 at 13:17, Dario Faggioli <dario.faggioli@citrix.com> wrote:
> These two are both bugfixes, although the second one affects debug
> builds only:
> 
>  * "xl: fix `xl cpupool-list' behavior in case no pool name is provided"
>    git cherry-pick 3998afdbf99959582dcd9f9f4df5a6fe7ce4ded8

IanC - if you feel confident in this being a reasonable backporting
candidate, I'd like to make use of your offer to help out in place
of IanJ.

>  * "sched_credit: filter node-affinity mask against online cpus"
>    git cherry-pick 5e5a44b6c942d6ea47f15d6f1ed02b03e0d69445

This one went in already (you could have checked).

> The first one affects 4.2 too, and I already requested backport (and I
> think IanJ said its on his list). The second only affects 4.3.

With 4.2.3 just having gone out, we can worry about 4.2 later.

Jan

Re: preparing for 4.3.1

[Dario Faggioli]

[-- Attachment #1.1: Type: text/plain, Size: 670 bytes --]

On ven, 2013-09-27 at 12:49 +0100, Jan Beulich wrote:
> >  * "sched_credit: filter node-affinity mask against online cpus"
> >    git cherry-pick 5e5a44b6c942d6ea47f15d6f1ed02b03e0d69445
> 
> This one went in already (you could have checked).
> 
I actually did check, but I did that in 'staging-4.3' (i.e., my local
branch) rather than in 'origin/staging-4.3'. :-(

Sorry for the noise.

Dario

-- 
<<This happens because I choose it to happen!>> (Raistlin Majere)
-----------------------------------------------------------------
Dario Faggioli, Ph.D, http://about.me/dario.faggioli
Senior Software Engineer, Citrix Systems R&D Ltd., Cambridge (UK)


[-- Attachment #1.2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

[-- Attachment #2: Type: text/plain, Size: 126 bytes --]

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

Re: preparing for 4.3.1

[Pasi Kärkkäinen]

On Fri, Sep 27, 2013 at 11:07:24AM +0100, Jan Beulich wrote:
> Aiming at a release later in October (before Xen Summit I would
> hope), I'd like to cut RC1 next week.
> 
> Please indicate any bug fixes that so far may have been missed
> in the backports already done.
> 

I'd like to see the following commit from qemu-xen-unstable (traditional)
backported to qemu-xen-4.3-testing (well it should apply as-is):

"passthrough: Correctly expose PCH ISA bridge for IGD passthrough":
http://xenbits.xen.org/gitweb/?p=qemu-xen-unstable.git;a=commit;h=73c4ad5a3d065267b878c8ae31ff0b4833f093e4

It's needed for Intel IGD GPU passthrough.

Thanks,

-- Pasi

Re: preparing for 4.3.1

[Jan Beulich]

>>> On 28.09.13 at 13:50, Pasi Kärkkäinen<pasik@iki.fi> wrote:
> On Fri, Sep 27, 2013 at 11:07:24AM +0100, Jan Beulich wrote:
>> Aiming at a release later in October (before Xen Summit I would
>> hope), I'd like to cut RC1 next week.
>> 
>> Please indicate any bug fixes that so far may have been missed
>> in the backports already done.
>> 
> 
> I'd like to see the following commit from qemu-xen-unstable (traditional)
> backported to qemu-xen-4.3-testing (well it should apply as-is):
> 
> "passthrough: Correctly expose PCH ISA bridge for IGD passthrough":
> http://xenbits.xen.org/gitweb/?p=qemu-xen-unstable.git;a=commit;h=73c4ad5a3d06 
> 5267b878c8ae31ff0b4833f093e4
> 
> It's needed for Intel IGD GPU passthrough.

Deciding whether this is suitable as well as applying if so will have
to wait for IanJ's return (i.e. is not expected to make RC1).

Jan

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

Re: preparing for 4.3.1

[AL13N]

Op vrijdag 27 september 2013 11:07:24 schreef Jan Beulich:
> Aiming at a release later in October (before Xen Summit I would
> hope), I'd like to cut RC1 next week.
> 
> Please indicate any bug fixes that so far may have been missed
> in the backports already done.

any ETA on this rc1? i'd like to have 4.3.1 on Mageia, but i kindof need an 
rc1 before version freeze next week...

Re: preparing for 4.3.1

[Andrew Cooper]

On 08/10/2013 21:42, AL13N wrote:
> Op vrijdag 27 september 2013 11:07:24 schreef Jan Beulich:
>> Aiming at a release later in October (before Xen Summit I would
>> hope), I'd like to cut RC1 next week.
>>
>> Please indicate any bug fixes that so far may have been missed
>> in the backports already done.
> any ETA on this rc1? i'd like to have 4.3.1 on Mageia, but i kindof need an 
> rc1 before version freeze next week...

rc1 has been tagged in the tree for 7 days now.

http://xenbits.xen.org/gitweb/?p=xen.git;a=shortlog;h=refs/heads/staging-4.3

~Andrew

Re: preparing for 4.3.1

[M A Young]

On Tue, 8 Oct 2013, Andrew Cooper wrote:

> On 08/10/2013 21:42, AL13N wrote:
>> Op vrijdag 27 september 2013 11:07:24 schreef Jan Beulich:
>>> Aiming at a release later in October (before Xen Summit I would
>>> hope), I'd like to cut RC1 next week.
>>>
>>> Please indicate any bug fixes that so far may have been missed
>>> in the backports already done.
>> any ETA on this rc1? i'd like to have 4.3.1 on Mageia, but i kindof need an
>> rc1 before version freeze next week...
>
> rc1 has been tagged in the tree for 7 days now.
>
> http://xenbits.xen.org/gitweb/?p=xen.git;a=shortlog;h=refs/heads/staging-4.3

You might get the full version in time for your freeze. According to 
http://xenbits.xen.org/xsa/ there are 5 security updates due for release 
tomorrow, which might be all the full version is waiting for.

 	Michael Young

Re: preparing for 4.3.1

[AL13N]

Op dinsdag 8 oktober 2013 22:23:32 schreef M A Young:
> On Tue, 8 Oct 2013, Andrew Cooper wrote:
> > On 08/10/2013 21:42, AL13N wrote:
> >> Op vrijdag 27 september 2013 11:07:24 schreef Jan Beulich:
> >>> Aiming at a release later in October (before Xen Summit I would
> >>> hope), I'd like to cut RC1 next week.
> >>> 
> >>> Please indicate any bug fixes that so far may have been missed
> >>> in the backports already done.
> >> 
> >> any ETA on this rc1? i'd like to have 4.3.1 on Mageia, but i kindof need
> >> an
> >> rc1 before version freeze next week...
> > 
> > rc1 has been tagged in the tree for 7 days now.
> > 
> > http://xenbits.xen.org/gitweb/?p=xen.git;a=shortlog;h=refs/heads/staging-4
> > .3
> You might get the full version in time for your freeze. According to
> http://xenbits.xen.org/xsa/ there are 5 security updates due for release
> tomorrow, which might be all the full version is waiting for.
> 
>  	Michael Young

ah, i didn't think 4.3.1 would have these patches in...

but if it's been unchanged the last 7 days, does this mean the 3 or so patches 
spoken of, in this thread, aren't included?

Re: preparing for 4.3.1

[M A Young]

On Wed, 9 Oct 2013, AL13N wrote:

> Op dinsdag 8 oktober 2013 22:23:32 schreef M A Young:
>> On Tue, 8 Oct 2013, Andrew Cooper wrote:
>>> On 08/10/2013 21:42, AL13N wrote:
>>>>
>>>> any ETA on this rc1? i'd like to have 4.3.1 on Mageia, but i kindof need
>>>> an
>>>> rc1 before version freeze next week...
>>>
>>> rc1 has been tagged in the tree for 7 days now.
>>>
>>> http://xenbits.xen.org/gitweb/?p=xen.git;a=shortlog;h=refs/heads/staging-4
>>> .3
>> You might get the full version in time for your freeze. According to
>> http://xenbits.xen.org/xsa/ there are 5 security updates due for release
>> tomorrow, which might be all the full version is waiting for.
>>
>>  	Michael Young
>
> ah, i didn't think 4.3.1 would have these patches in...
>
> but if it's been unchanged the last 7 days, does this mean the 3 or so patches
> spoken of, in this thread, aren't included?

I am not saying anything definite, merely speculating. It makes sense to 
wait a few days for a security patches, and they may be setting the 
timetable but there could still be other patches to come.

 	Michael Young

Re: preparing for 4.3.1

[Jan Beulich]

>>> On 09.10.13 at 07:50, AL13N <alien@rmail.be> wrote:
> Op dinsdag 8 oktober 2013 22:23:32 schreef M A Young:
>> On Tue, 8 Oct 2013, Andrew Cooper wrote:
>> > On 08/10/2013 21:42, AL13N wrote:
>> >> Op vrijdag 27 september 2013 11:07:24 schreef Jan Beulich:
>> >>> Aiming at a release later in October (before Xen Summit I would
>> >>> hope), I'd like to cut RC1 next week.
>> >>> 
>> >>> Please indicate any bug fixes that so far may have been missed
>> >>> in the backports already done.
>> >> 
>> >> any ETA on this rc1? i'd like to have 4.3.1 on Mageia, but i kindof need
>> >> an
>> >> rc1 before version freeze next week...
>> > 
>> > rc1 has been tagged in the tree for 7 days now.
>> > 
>> > http://xenbits.xen.org/gitweb/?p=xen.git;a=shortlog;h=refs/heads/staging-4 
>> > .3
>> You might get the full version in time for your freeze. According to
>> http://xenbits.xen.org/xsa/ there are 5 security updates due for release
>> tomorrow, which might be all the full version is waiting for.
>> 
>>  	Michael Young
> 
> ah, i didn't think 4.3.1 would have these patches in...
> 
> but if it's been unchanged the last 7 days, does this mean the 3 or so 
> patches 
> spoken of, in this thread, aren't included?

I think I said quite clearly in the RC1 announcement that RC2
will need to wait for IanJ's return, so that the tools side can be
taken care of (where I think we had a couple of additional
backport requests).

Jan

Re: preparing for 4.3.1

[AL13N]

>>>> On 09.10.13 at 07:50, AL13N <alien@rmail.be> wrote:
>> Op dinsdag 8 oktober 2013 22:23:32 schreef M A Young:
>>> On Tue, 8 Oct 2013, Andrew Cooper wrote:
>>> > On 08/10/2013 21:42, AL13N wrote:
>>> >> Op vrijdag 27 september 2013 11:07:24 schreef Jan Beulich:
>>> >>> Aiming at a release later in October (before Xen Summit I would
>>> >>> hope), I'd like to cut RC1 next week.
>>> >>>
>>> >>> Please indicate any bug fixes that so far may have been missed
>>> >>> in the backports already done.
>>> >>
>>> >> any ETA on this rc1? i'd like to have 4.3.1 on Mageia, but i kindof
>>> need
>>> >> an
>>> >> rc1 before version freeze next week...
>>> >
>>> > rc1 has been tagged in the tree for 7 days now.
>>> >
>>> > http://xenbits.xen.org/gitweb/?p=xen.git;a=shortlog;h=refs/heads/staging-4
>>> > .3
>>> You might get the full version in time for your freeze. According to
>>> http://xenbits.xen.org/xsa/ there are 5 security updates due for
>>> release
>>> tomorrow, which might be all the full version is waiting for.
>>>
>>>  	Michael Young
>>
>> ah, i didn't think 4.3.1 would have these patches in...
>>
>> but if it's been unchanged the last 7 days, does this mean the 3 or so
>> patches
>> spoken of, in this thread, aren't included?
>
> I think I said quite clearly in the RC1 announcement that RC2
> will need to wait for IanJ's return, so that the tools side can be
> taken care of (where I think we had a couple of additional
> backport requests).

ah, ic, by this you mean that the rc1 is a tag and is not actually
released as a tarball... ( or it is, and i've missed it? )

Re: preparing for 4.3.1

[Jan Beulich]

>>> On 09.10.13 at 13:19, "AL13N" <alien@rmail.be> wrote:
>>>>> On 09.10.13 at 07:50, AL13N <alien@rmail.be> wrote:
>>> Op dinsdag 8 oktober 2013 22:23:32 schreef M A Young:
>>>> On Tue, 8 Oct 2013, Andrew Cooper wrote:
>>>> > On 08/10/2013 21:42, AL13N wrote:
>>>> >> Op vrijdag 27 september 2013 11:07:24 schreef Jan Beulich:
>>>> >>> Aiming at a release later in October (before Xen Summit I would
>>>> >>> hope), I'd like to cut RC1 next week.
>>>> >>>
>>>> >>> Please indicate any bug fixes that so far may have been missed
>>>> >>> in the backports already done.
>>>> >>
>>>> >> any ETA on this rc1? i'd like to have 4.3.1 on Mageia, but i kindof
>>>> need
>>>> >> an
>>>> >> rc1 before version freeze next week...
>>>> >
>>>> > rc1 has been tagged in the tree for 7 days now.
>>>> >
>>>> > http://xenbits.xen.org/gitweb/?p=xen.git;a=shortlog;h=refs/heads/staging-4 
>>>> > .3
>>>> You might get the full version in time for your freeze. According to
>>>> http://xenbits.xen.org/xsa/ there are 5 security updates due for
>>>> release
>>>> tomorrow, which might be all the full version is waiting for.
>>>>
>>>>  	Michael Young
>>>
>>> ah, i didn't think 4.3.1 would have these patches in...
>>>
>>> but if it's been unchanged the last 7 days, does this mean the 3 or so
>>> patches
>>> spoken of, in this thread, aren't included?
>>
>> I think I said quite clearly in the RC1 announcement that RC2
>> will need to wait for IanJ's return, so that the tools side can be
>> taken care of (where I think we had a couple of additional
>> backport requests).
> 
> ah, ic, by this you mean that the rc1 is a tag and is not actually
> released as a tarball... ( or it is, and i've missed it? )

No, we generally don't produce tarballs for RCs of stable
releases (as opposed to major ones). There will of course be
a tarball for the actual release of 4.3.1.

Jan

Re: preparing for 4.3.1

[Ian Jackson]

Jan Beulich writes ("Re: [Xen-devel] preparing for 4.3.1"):
> On 28.09.13 at 13:50, Pasi Kärkkäinen<pasik@iki.fi> wrote:
> > On Fri, Sep 27, 2013 at 11:07:24AM +0100, Jan Beulich wrote:
> >> Aiming at a release later in October (before Xen Summit I would
> >> hope), I'd like to cut RC1 next week.
> >> 
> >> Please indicate any bug fixes that so far may have been missed
> >> in the backports already done.
> >> 
> > 
> > I'd like to see the following commit from qemu-xen-unstable (traditional)
> > backported to qemu-xen-4.3-testing (well it should apply as-is):
> > 
> > "passthrough: Correctly expose PCH ISA bridge for IGD passthrough":
> > http://xenbits.xen.org/gitweb/?p=qemu-xen-unstable.git;a=commit;h=73c4ad5a3d06 
> > 5267b878c8ae31ff0b4833f093e4
> > 
> > It's needed for Intel IGD GPU passthrough.
> 
> Deciding whether this is suitable as well as applying if so will have
> to wait for IanJ's return (i.e. is not expected to make RC1).

(I'm now back.)  This is now in 4.3.

Ian.

Re: preparing for 4.3.1

[Pasi Kärkkäinen]

On Fri, Oct 11, 2013 at 07:07:14PM +0100, Ian Jackson wrote:
> Jan Beulich writes ("Re: [Xen-devel] preparing for 4.3.1"):
> > On 28.09.13 at 13:50, Pasi Kärkkäinen<pasik@iki.fi> wrote:
> > > On Fri, Sep 27, 2013 at 11:07:24AM +0100, Jan Beulich wrote:
> > >> Aiming at a release later in October (before Xen Summit I would
> > >> hope), I'd like to cut RC1 next week.
> > >> 
> > >> Please indicate any bug fixes that so far may have been missed
> > >> in the backports already done.
> > >> 
> > > 
> > > I'd like to see the following commit from qemu-xen-unstable (traditional)
> > > backported to qemu-xen-4.3-testing (well it should apply as-is):
> > > 
> > > "passthrough: Correctly expose PCH ISA bridge for IGD passthrough":
> > > http://xenbits.xen.org/gitweb/?p=qemu-xen-unstable.git;a=commit;h=73c4ad5a3d06 
> > > 5267b878c8ae31ff0b4833f093e4
> > > 
> > > It's needed for Intel IGD GPU passthrough.
> > 
> > Deciding whether this is suitable as well as applying if so will have
> > to wait for IanJ's return (i.e. is not expected to make RC1).
> 
> (I'm now back.)  This is now in 4.3.
>

Thanks a lot Ian!

-- Pasi