preparing for 4.3.2 and 4.2.4

preparing for 4.3.2 and 4.2.4

[Jan Beulich]

Aiming at a release in late January or early February, I'd like to cut
RC1s later this or early next week.

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

Thanks, Jan

Re: preparing for 4.3.2 and 4.2.4

[Andrew Cooper]

On 13/01/14 15:14, Jan Beulich wrote:
> Aiming at a release in late January or early February, I'd like to cut
> RC1s later this or early next week.
>
> Please indicate any bug fixes that so far may have been missed
> in the backports already done.
>
> Thanks, Jan

Looking through the XenServer patch queue on top of stable-4.3

All of these are suggestions for "might be useful to take", rather than
"strictly bugfixes"

Hypervisor:

* x86/crash: Indicate how well nmi_shootdown_cpus() managed to do
xen-unstable.hg-27635.45bf542dd584
git: f12c1f0b09205cdf18a2c4a615fdc3e7357ce704

* x86/ats: Fix parsing of 'ats' command line option
xen-unstable.hg-27831.bd84d8277c21
git: 7b5af1df122092243a3697409d5a5ad3b9944da4

* x86/VT-x: Disable MSR intercept for SHADOW_GS_BASE
xen-unstable.hg-27960.48cd67917186
git: a82e98d473fd212316ea5aa078a7588324b020e5

* kexec: prevent deadlock on reentry to the crash path
xen-unstable.hg-28067.065befe6d07e
git: 470f58c159410b280627c2ea7798ea12ad93bd7c

Tools:

* libxl: Allow 4 MB of video RAM for Cirrus graphics on traditional QEM
xen-unstable.hg-27834.02b33d7e56f6
git: 13d13a45d0591fc195666ea20ddf8781a0367e88


~Andrew

Re: preparing for 4.3.2 and 4.2.4

[Ian Jackson]

Jan Beulich writes ("preparing for 4.3.2 and 4.2.4"):
> Please indicate any bug fixes that so far may have been missed
> in the backports already done.

All I'm aware of is

  9.1.14

(that's the date they went into staging)

  2d03be65d5c50053fec4a5fa1d691972e5d953c9
  libxl: Auto-assign NIC devids in initiate_domain_create

  1671cdeac7da663fb2963f3e587fa279dcd0238b
  tools/libxc: Correct read_exact() error messages

But Ian C probably has a nearly-disjoint view of bugfixes which have
gone into staging and could do with backporting.

Ian.

Re: preparing for 4.3.2 and 4.2.4

[Ian Campbell]

On Mon, 2014-01-13 at 16:49 +0000, Ian Jackson wrote:
> Jan Beulich writes ("preparing for 4.3.2 and 4.2.4"):
> > Please indicate any bug fixes that so far may have been missed
> > in the backports already done.
> 
> All I'm aware of is
> 
>   9.1.14
> 
> (that's the date they went into staging)
> 
>   2d03be65d5c50053fec4a5fa1d691972e5d953c9
>   libxl: Auto-assign NIC devids in initiate_domain_create
> 
>   1671cdeac7da663fb2963f3e587fa279dcd0238b
>   tools/libxc: Correct read_exact() error messages
> 
> But Ian C probably has a nearly-disjoint view of bugfixes which have
> gone into staging and could do with backporting.

I'm afraid I've not really been paying attention to things at that
level, sorry :-/

Ian

Re: preparing for 4.3.2 and 4.2.4

[Don Slutz]

On 01/13/14 10:14, Jan Beulich wrote:
> Aiming at a release in late January or early February, I'd like to cut
> RC1s later this or early next week.
>
> Please indicate any bug fixes that so far may have been missed
> in the backports already done.
>
> Thanks, Jan
>
>
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xen.org
> http://lists.xen.org/xen-devel


I do not see:


commit 3dbab7a8bf4bef1bb2967cb3a8c7ed2146482ab3


in any 4.2 or 4.3 tree.  As Andrew Cooper said:


 > This is a hypervisor reference counting error on a toolstack hypercall
 > path.  Irrespective of any of the other patches in this series, I think
 > this should be included ASAP (although probably subject to review from a
 > third person), which will fix the hypervisor crashes from gdbsx usage.

    -Don Slutz

Re: preparing for 4.3.2 and 4.2.4

[Don Slutz]

On 01/13/14 10:14, Jan Beulich wrote:
> Aiming at a release in late January or early February, I'd like to cut
> RC1s later this or early next week.
>
> Please indicate any bug fixes that so far may have been missed
> in the backports already done.
>
> Thanks, Jan
>
>
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xen.org
> http://lists.xen.org/xen-devel


I do not see:


commit 3dbab7a8bf4bef1bb2967cb3a8c7ed2146482ab3


in any 4.2 or 4.3 tree.  As Andrew Cooper said:


 > This is a hypervisor reference counting error on a toolstack hypercall
 > path.  Irrespective of any of the other patches in this series, I think
 > this should be included ASAP (although probably subject to review from a
 > third person), which will fix the hypervisor crashes from gdbsx usage.

    -Don Slutz

Re: preparing for 4.3.2 and 4.2.4

[Jan Beulich]

>>> On 13.01.14 at 19:38, Don Slutz <dslutz@verizon.com> wrote:
> On 01/13/14 10:14, Jan Beulich wrote:
>> Aiming at a release in late January or early February, I'd like to cut
>> RC1s later this or early next week.
>>
>> Please indicate any bug fixes that so far may have been missed
>> in the backports already done.
>>
>> Thanks, Jan
>>
>>
>> _______________________________________________
>> Xen-devel mailing list
>> Xen-devel@lists.xen.org 
>> http://lists.xen.org/xen-devel 
> 
> 
> I do not see:
> 
> 
> commit 3dbab7a8bf4bef1bb2967cb3a8c7ed2146482ab3
> 
> 
> in any 4.2 or 4.3 tree.  As Andrew Cooper said:
> 
> 
>  > This is a hypervisor reference counting error on a toolstack hypercall
>  > path.  Irrespective of any of the other patches in this series, I think
>  > this should be included ASAP (although probably subject to review from a
>  > third person), which will fix the hypervisor crashes from gdbsx usage.

Considering that this is only debugging code, I did not view this as
important enough to backport without explicit request.
Furthermore, it hadn't passed the push gate yet when I collected
the most recent batch of backports.

I just now added it to my list of pending backports.

As an aside: It would have helped (read: saved me some time) if you
had also reproduced the title of the commit instead of just the commit
ID. See Andrew's response to the original mail of this thread for an
example (the precise form doesn't matter, the conveyed information
does).

Jan

Re: preparing for 4.3.2 and 4.2.4

[Jan Beulich]

>>> On 13.01.14 at 16:31, Andrew Cooper <andrew.cooper3@citrix.com> wrote:
> On 13/01/14 15:14, Jan Beulich wrote:
>> Aiming at a release in late January or early February, I'd like to cut
>> RC1s later this or early next week.
>>
>> Please indicate any bug fixes that so far may have been missed
>> in the backports already done.
>>
>> Thanks, Jan
> 
> Looking through the XenServer patch queue on top of stable-4.3
> 
> All of these are suggestions for "might be useful to take", rather than
> "strictly bugfixes"
> 
> Hypervisor:
> 
> * x86/crash: Indicate how well nmi_shootdown_cpus() managed to do
> xen-unstable.hg-27635.45bf542dd584
> git: f12c1f0b09205cdf18a2c4a615fdc3e7357ce704
> 
> * x86/ats: Fix parsing of 'ats' command line option
> xen-unstable.hg-27831.bd84d8277c21
> git: 7b5af1df122092243a3697409d5a5ad3b9944da4
> 
> * x86/VT-x: Disable MSR intercept for SHADOW_GS_BASE
> xen-unstable.hg-27960.48cd67917186
> git: a82e98d473fd212316ea5aa078a7588324b020e5
> 
> * kexec: prevent deadlock on reentry to the crash path
> xen-unstable.hg-28067.065befe6d07e
> git: 470f58c159410b280627c2ea7798ea12ad93bd7c

I applied all of them to 4.3 and all but the first one to 4.2 (the
first one would require a backport to be done first).

Jan

> Tools:
> 
> * libxl: Allow 4 MB of video RAM for Cirrus graphics on traditional QEM
> xen-unstable.hg-27834.02b33d7e56f6
> git: 13d13a45d0591fc195666ea20ddf8781a0367e88
> 
> 
> ~Andrew