Re: [PATCH v11 11/11] (lib)xl: soft reset support

[Vitaly Kuznetsov <vkuznets@redhat.com>]

Wei Liu <wei.liu2@citrix.com> writes:

> Sorry for the delay.
>
> FYI all other patches of this series were applied by Jan. You only need
> to resend this one.

Cool, I will.

>
> See below for a few comments.
>
> On Fri, Sep 04, 2015 at 03:39:51PM +0200, Vitaly Kuznetsov wrote:
>> Use existing create/restore path to perform 'soft reset' for HVM
>> domains. Tear everything down, e.g. destroy domain's device model,
>> remove the domain from xenstore, save toolstack record and start
>> over.
>> 
>> Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
>> ---
>> Changes since v10:
>> - Adapt to 'migration v2' changes.
>> - Use LIBXL_DEVICE_MODEL_SAVE_FILE as Qemu save file (and rename it to
>>   LIBXL_DEVICE_MODEL_RESTORE_FILE later) to support stubdom case (as
>>   we connect consoles to both files on create.
>> - Fix coding style, minor description change in xl.cfg.pod.5 [Wei Liu]
>> 
>> Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
>> ---
>>  docs/man/xl.cfg.pod.5        |   8 +-
>>  tools/libxl/libxl.c          |  22 ++++-
>>  tools/libxl/libxl.h          |  15 ++++
>>  tools/libxl/libxl_create.c   | 192 ++++++++++++++++++++++++++++++++++++++-----
>>  tools/libxl/libxl_internal.h |   4 +
>>  tools/libxl/libxl_types.idl  |   2 +
>>  tools/libxl/xl.h             |   1 +
>>  tools/libxl/xl_cmdimpl.c     |  25 +++++-
>>  8 files changed, 242 insertions(+), 27 deletions(-)
>> 
>> diff --git a/docs/man/xl.cfg.pod.5 b/docs/man/xl.cfg.pod.5
>> index c6345b8..d8c4186 100644
>> --- a/docs/man/xl.cfg.pod.5
>> +++ b/docs/man/xl.cfg.pod.5
>> @@ -349,6 +349,12 @@ destroy the domain.
>>  write a "coredump" of the domain to F</var/lib/xen/dump/NAME> and then
>>  restart the domain.
>>  
>> +=item B<soft-reset>
>> +
>> +Reset all Xen specific interfaces for the Xen-aware HVM domain allowing
>> +it to reestablish these interfaces and continue executing the domain. PV
>> +guest is not supported.
>> +
>
> And "non-Xen-aware HVM will crash" ?

Sorry, I should have replied to that suggestion earlier. Non-Xen-aware
HVM guest can't really trigger this action and (in theory) is capable of
doing kexec without any assistance.

> If there is no definite answer to
> guest state maybe just saying "PV guest and non-Xen-aware HVM guests are
> not supported" ?

This sounds correct.

>
> It's important to let user know about the consequence because libxl
> doesn't actually stop you from soft-resetting a HVM guest that is not
> Xen-aware.

The question is who (and when/how) is going to trigger this action? In
case someone does that while HVM domain (doesn't really matter if it is
Xen-aware or not) does not expect this action it will crash.

*In theory* nothing bad is going to happen to a non-Xen-aware HVM guest
if someone else will trigger this action for it (e.g. on 'reset'
signal), it will just get an assistance it doesn't need.

[...]

-- 
  Vitaly