Single system binary & Dynamic machine model (KVM developers conference call 2022-12-13)

Single system binary & Dynamic machine model (KVM developers conference call 2022-12-13)

[Philippe Mathieu-Daudé]

Hi,

In the last years we had few discussions on "simplifying" QEMU (system 
emulation / virtualization), in particular for the "management layer".

Some of us are interested in having QEMU able to dynamically create
machine models. Mark Burton's current approach is via a Python script
which generates QMP commands. This is just another case of "management
layer".

Various problems have been raised regarding the current limitations of
QEMU's APIs. We'd like to remember / get a broader idea on these limits
and look at some ideas / proposals which have been discussed / posted
on this list.

Feel free to complete your thoughts on this public etherpad:
https://etherpad.opendev.org/p/qemu-emulation-bof%402022-12-13

Topic I remember which can be good starters:

- Current limitations of QAPI (& QMP) model (Markus Armbruster)

- Adding a new qemu-runtime-$TARGET / QMP-only binary without today's
   limitations (Daniel P. Berrangé & Paolo Bonzini)

- Problem with x-exit-preconfig, reworking MachinePhaseInit state
   machine (Paolo Bonzini)

Markus / Daniel / Paolo expressed their ideas on the list (the
historical threads are referenced in the etherpad) so reading the
relevant threads before the call will help to get in the topic.
These people don't have to be in the call, but if they can attend
that would be very nice :)

The call will be Tuesday, December 13 at 3pm CET on this Bluejeans link:
http://bluejeans.com/quintela

You can subscribe to the 'KVM developers conference call' calendar here:
https://calendar.google.com/calendar/u/0/embed?src=eged7cki05lmu1tngvkl3thids@group.calendar.google.com

Regards,

Phil.

Re: Single system binary & Dynamic machine model (KVM developers conference call 2022-12-13)

[Philippe Mathieu-Daudé]

On 12/12/22 00:41, Philippe Mathieu-Daudé wrote:
> Hi,
> 
> In the last years we had few discussions on "simplifying" QEMU (system 
> emulation / virtualization), in particular for the "management layer".
> 
> Some of us are interested in having QEMU able to dynamically create
> machine models. Mark Burton's current approach is via a Python script
> which generates QMP commands. This is just another case of "management
> layer".
> 
> Various problems have been raised regarding the current limitations of
> QEMU's APIs. We'd like to remember / get a broader idea on these limits
> and look at some ideas / proposals which have been discussed / posted
> on this list.
> 
> Feel free to complete your thoughts on this public etherpad:
> https://etherpad.opendev.org/p/qemu-emulation-bof%402022-12-13
> 
> Topic I remember which can be good starters:
> 
> - Current limitations of QAPI (& QMP) model (Markus Armbruster)
> 
> - Adding a new qemu-runtime-$TARGET / QMP-only binary without today's
>    limitations (Daniel P. Berrangé & Paolo Bonzini)
> 
> - Problem with x-exit-preconfig, reworking MachinePhaseInit state
>    machine (Paolo Bonzini)
> 
> Markus / Daniel / Paolo expressed their ideas on the list (the
> historical threads are referenced in the etherpad) so reading the
> relevant threads before the call will help to get in the topic.
> These people don't have to be in the call, but if they can attend
> that would be very nice :)
> 
> The call will be Tuesday, December 13 at 3pm CET on this Bluejeans link:
> http://bluejeans.com/quintela
We moved the call here:

https://teams.microsoft.com/_#/pre-join-calling/19:meeting_MjU2NDhkOGYtOGY5ZC00MzRjLWJjNjgtYWI2ODQxZGExMWNj@thread.v2

Re: Single system binary & Dynamic machine model (KVM developers conference call 2022-12-13)

[Stefan Hajnoczi]

On Tue, 13 Dec 2022 at 09:08, Philippe Mathieu-Daudé <philmd@linaro.org> wrote:
>
> On 12/12/22 00:41, Philippe Mathieu-Daudé wrote:
> > Hi,
> >
> > In the last years we had few discussions on "simplifying" QEMU (system
> > emulation / virtualization), in particular for the "management layer".
> >
> > Some of us are interested in having QEMU able to dynamically create
> > machine models. Mark Burton's current approach is via a Python script
> > which generates QMP commands. This is just another case of "management
> > layer".
> >
> > Various problems have been raised regarding the current limitations of
> > QEMU's APIs. We'd like to remember / get a broader idea on these limits
> > and look at some ideas / proposals which have been discussed / posted
> > on this list.
> >
> > Feel free to complete your thoughts on this public etherpad:
> > https://etherpad.opendev.org/p/qemu-emulation-bof%402022-12-13
> >
> > Topic I remember which can be good starters:
> >
> > - Current limitations of QAPI (& QMP) model (Markus Armbruster)
> >
> > - Adding a new qemu-runtime-$TARGET / QMP-only binary without today's
> >    limitations (Daniel P. Berrangé & Paolo Bonzini)
> >
> > - Problem with x-exit-preconfig, reworking MachinePhaseInit state
> >    machine (Paolo Bonzini)
> >
> > Markus / Daniel / Paolo expressed their ideas on the list (the
> > historical threads are referenced in the etherpad) so reading the
> > relevant threads before the call will help to get in the topic.
> > These people don't have to be in the call, but if they can attend
> > that would be very nice :)
> >
> > The call will be Tuesday, December 13 at 3pm CET on this Bluejeans link:
> > http://bluejeans.com/quintela
> We moved the call here:
>
> https://teams.microsoft.com/_#/pre-join-calling/19:meeting_MjU2NDhkOGYtOGY5ZC00MzRjLWJjNjgtYWI2ODQxZGExMWNj@thread.v2

Please use a video conferencing system that allows people to join
without logging in and works across browsers. I gave up on Teams.

Stefan

Re: Single system binary & Dynamic machine model (KVM developers conference call 2022-12-13)

[Edgar E. Iglesias]

On Tue, Dec 13, 2022 at 09:17:40AM -0500, Stefan Hajnoczi wrote:
> On Tue, 13 Dec 2022 at 09:08, Philippe Mathieu-Daudé <philmd@linaro.org> wrote:
> >
> > On 12/12/22 00:41, Philippe Mathieu-Daudé wrote:
> > > Hi,
> > >
> > > In the last years we had few discussions on "simplifying" QEMU (system
> > > emulation / virtualization), in particular for the "management layer".
> > >
> > > Some of us are interested in having QEMU able to dynamically create
> > > machine models. Mark Burton's current approach is via a Python script
> > > which generates QMP commands. This is just another case of "management
> > > layer".
> > >
> > > Various problems have been raised regarding the current limitations of
> > > QEMU's APIs. We'd like to remember / get a broader idea on these limits
> > > and look at some ideas / proposals which have been discussed / posted
> > > on this list.
> > >
> > > Feel free to complete your thoughts on this public etherpad:
> > > https://etherpad.opendev.org/p/qemu-emulation-bof%402022-12-13
> > >
> > > Topic I remember which can be good starters:
> > >
> > > - Current limitations of QAPI (& QMP) model (Markus Armbruster)
> > >
> > > - Adding a new qemu-runtime-$TARGET / QMP-only binary without today's
> > >    limitations (Daniel P. Berrangé & Paolo Bonzini)
> > >
> > > - Problem with x-exit-preconfig, reworking MachinePhaseInit state
> > >    machine (Paolo Bonzini)
> > >
> > > Markus / Daniel / Paolo expressed their ideas on the list (the
> > > historical threads are referenced in the etherpad) so reading the
> > > relevant threads before the call will help to get in the topic.
> > > These people don't have to be in the call, but if they can attend
> > > that would be very nice :)
> > >
> > > The call will be Tuesday, December 13 at 3pm CET on this Bluejeans link:
> > > http://bluejeans.com/quintela
> > We moved the call here:
> >
> > https://teams.microsoft.com/_#/pre-join-calling/19:meeting_MjU2NDhkOGYtOGY5ZC00MzRjLWJjNjgtYWI2ODQxZGExMWNj@thread.v2
> 
> Please use a video conferencing system that allows people to join
> without logging in and works across browsers. I gave up on Teams.
>

I'm also unable to join, sorry!

Best regards,
Edgar

Re: Single system binary & Dynamic machine model (KVM developers conference call 2022-12-13)

[Mark Burton]

[-- Attachment #1: Type: text/plain, Size: 2411 bytes --]



On 13/12/2022, 15:17, "Stefan Hajnoczi" <stefanha@gmail.com> wrote:

WARNING: This email originated from outside of Qualcomm. Please be wary of any links or attachments, and do not enable macros.

On Tue, 13 Dec 2022 at 09:08, Philippe Mathieu-Daudé <philmd@linaro.org> wrote:
>
> On 12/12/22 00:41, Philippe Mathieu-Daudé wrote:
> > Hi,
> >
> > In the last years we had few discussions on "simplifying" QEMU (system
> > emulation / virtualization), in particular for the "management layer".
> >
> > Some of us are interested in having QEMU able to dynamically create
> > machine models. Mark Burton's current approach is via a Python script
> > which generates QMP commands. This is just another case of "management
> > layer".
> >
> > Various problems have been raised regarding the current limitations of
> > QEMU's APIs. We'd like to remember / get a broader idea on these limits
> > and look at some ideas / proposals which have been discussed / posted
> > on this list.
> >
> > Feel free to complete your thoughts on this public etherpad:
> > https://etherpad.opendev.org/p/qemu-emulation-bof%402022-12-13
> >
> > Topic I remember which can be good starters:
> >
> > - Current limitations of QAPI (& QMP) model (Markus Armbruster)
> >
> > - Adding a new qemu-runtime-$TARGET / QMP-only binary without today's
> >    limitations (Daniel P. Berrangé & Paolo Bonzini)
> >
> > - Problem with x-exit-preconfig, reworking MachinePhaseInit state
> >    machine (Paolo Bonzini)
> >
> > Markus / Daniel / Paolo expressed their ideas on the list (the
> > historical threads are referenced in the etherpad) so reading the
> > relevant threads before the call will help to get in the topic.
> > These people don't have to be in the call, but if they can attend
> > that would be very nice :)
> >
> > The call will be Tuesday, December 13 at 3pm CET on this Bluejeans link:
> > http://bluejeans.com/quintela
> We moved the call here:
>
> https://teams.microsoft.com/_#/pre-join-calling/19:meeting_MjU2NDhkOGYtOGY5ZC00MzRjLWJjNjgtYWI2ODQxZGExMWNj@thread.v2

Please use a video conferencing system that allows people to join
without logging in and works across browsers. I gave up on Teams.

The ’latest’ teams works within the browser, doesn’t require a login/lobby, and kind of seems to work. It’s nobodies preference, but - it’s “ok”.
Cheers
Mark.


Stefan

[-- Attachment #2: Type: text/html, Size: 5306 bytes --]

Re: Single system binary & Dynamic machine model (KVM developers conference call 2022-12-13)

[Peter Maydell]

On Tue, 13 Dec 2022 at 14:17, Stefan Hajnoczi <stefanha@gmail.com> wrote:
>
> On Tue, 13 Dec 2022 at 09:08, Philippe Mathieu-Daudé <philmd@linaro.org> wrote:
> >
> > On 12/12/22 00:41, Philippe Mathieu-Daudé wrote:
> > > The call will be Tuesday, December 13 at 3pm CET on this Bluejeans link:
> > > http://bluejeans.com/quintela
> > We moved the call here:
> >
> > https://teams.microsoft.com/_#/pre-join-calling/19:meeting_MjU2NDhkOGYtOGY5ZC00MzRjLWJjNjgtYWI2ODQxZGExMWNj@thread.v2
>
> Please use a video conferencing system that allows people to join
> without logging in and works across browsers. I gave up on Teams.

Unfortunately this was a last minute thing, because only Juan
has the code to make the bluejeans meeting work, and he didn't
turn up in the meeting :-(

-- PMM

Re: Single system binary & Dynamic machine model (KVM developers conference call 2022-12-13)

[Igor Mammedov]

On Tue, 13 Dec 2022 15:08:53 +0100
Philippe Mathieu-Daudé <philmd@linaro.org> wrote:

> On 12/12/22 00:41, Philippe Mathieu-Daudé wrote:
> > Hi,
> > 
> > In the last years we had few discussions on "simplifying" QEMU (system 
> > emulation / virtualization), in particular for the "management layer".
> > 
> > Some of us are interested in having QEMU able to dynamically create
> > machine models. Mark Burton's current approach is via a Python script
> > which generates QMP commands. This is just another case of "management
> > layer".
> > 
> > Various problems have been raised regarding the current limitations of
> > QEMU's APIs. We'd like to remember / get a broader idea on these limits
> > and look at some ideas / proposals which have been discussed / posted
> > on this list.
> > 
> > Feel free to complete your thoughts on this public etherpad:
> > https://etherpad.opendev.org/p/qemu-emulation-bof%402022-12-13
> > 
> > Topic I remember which can be good starters:
> > 
> > - Current limitations of QAPI (& QMP) model (Markus Armbruster)
> > 
> > - Adding a new qemu-runtime-$TARGET / QMP-only binary without today's
> >    limitations (Daniel P. Berrangé & Paolo Bonzini)
> > 
> > - Problem with x-exit-preconfig, reworking MachinePhaseInit state
> >    machine (Paolo Bonzini)
> > 
> > Markus / Daniel / Paolo expressed their ideas on the list (the
> > historical threads are referenced in the etherpad) so reading the
> > relevant threads before the call will help to get in the topic.
> > These people don't have to be in the call, but if they can attend
> > that would be very nice :)
> > 
> > The call will be Tuesday, December 13 at 3pm CET on this Bluejeans link:
> > http://bluejeans.com/quintela  
> We moved the call here:
> 
> https://teams.microsoft.com/_#/pre-join-calling/19:meeting_MjU2NDhkOGYtOGY5ZC00MzRjLWJjNjgtYWI2ODQxZGExMWNj@thread.v2

this one, required to login after which it throws out some license error whatever.

Re: Single system binary & Dynamic machine model (KVM developers conference call 2022-12-13)

[Cédric Le Goater]

On 12/13/22 15:17, Stefan Hajnoczi wrote:
> On Tue, 13 Dec 2022 at 09:08, Philippe Mathieu-Daudé <philmd@linaro.org> wrote:
>>
>> On 12/12/22 00:41, Philippe Mathieu-Daudé wrote:
>>> Hi,
>>>
>>> In the last years we had few discussions on "simplifying" QEMU (system
>>> emulation / virtualization), in particular for the "management layer".
>>>
>>> Some of us are interested in having QEMU able to dynamically create
>>> machine models. Mark Burton's current approach is via a Python script
>>> which generates QMP commands. This is just another case of "management
>>> layer".
>>>
>>> Various problems have been raised regarding the current limitations of
>>> QEMU's APIs. We'd like to remember / get a broader idea on these limits
>>> and look at some ideas / proposals which have been discussed / posted
>>> on this list.
>>>
>>> Feel free to complete your thoughts on this public etherpad:
>>> https://etherpad.opendev.org/p/qemu-emulation-bof%402022-12-13
>>>
>>> Topic I remember which can be good starters:
>>>
>>> - Current limitations of QAPI (& QMP) model (Markus Armbruster)
>>>
>>> - Adding a new qemu-runtime-$TARGET / QMP-only binary without today's
>>>     limitations (Daniel P. Berrangé & Paolo Bonzini)
>>>
>>> - Problem with x-exit-preconfig, reworking MachinePhaseInit state
>>>     machine (Paolo Bonzini)
>>>
>>> Markus / Daniel / Paolo expressed their ideas on the list (the
>>> historical threads are referenced in the etherpad) so reading the
>>> relevant threads before the call will help to get in the topic.
>>> These people don't have to be in the call, but if they can attend
>>> that would be very nice :)
>>>
>>> The call will be Tuesday, December 13 at 3pm CET on this Bluejeans link:
>>> http://bluejeans.com/quintela
>> We moved the call here:
>>
>> https://teams.microsoft.com/_#/pre-join-calling/19:meeting_MjU2NDhkOGYtOGY5ZC00MzRjLWJjNjgtYWI2ODQxZGExMWNj@thread.v2
> 
> Please use a video conferencing system that allows people to join
> without logging in and works across browsers. I gave up on Teams.

I also gave up after 27 or so clicks, redirections, emails checks, etc.
Too complex. There are much simpler alternatives.

Sorry.

C.

Re: Single system binary & Dynamic machine model (KVM developers conference call 2022-12-13)

[Marc-André Lureau]

[-- Attachment #1: Type: text/plain, Size: 2522 bytes --]

Hi

On Tue, Dec 13, 2022 at 6:44 PM Cédric Le Goater <clg@kaod.org> wrote:

> On 12/13/22 15:17, Stefan Hajnoczi wrote:
> > On Tue, 13 Dec 2022 at 09:08, Philippe Mathieu-Daudé <philmd@linaro.org>
> wrote:
> >>
> >> On 12/12/22 00:41, Philippe Mathieu-Daudé wrote:
> >>> Hi,
> >>>
> >>> In the last years we had few discussions on "simplifying" QEMU (system
> >>> emulation / virtualization), in particular for the "management layer".
> >>>
> >>> Some of us are interested in having QEMU able to dynamically create
> >>> machine models. Mark Burton's current approach is via a Python script
> >>> which generates QMP commands. This is just another case of "management
> >>> layer".
> >>>
> >>> Various problems have been raised regarding the current limitations of
> >>> QEMU's APIs. We'd like to remember / get a broader idea on these limits
> >>> and look at some ideas / proposals which have been discussed / posted
> >>> on this list.
> >>>
> >>> Feel free to complete your thoughts on this public etherpad:
> >>> https://etherpad.opendev.org/p/qemu-emulation-bof%402022-12-13
> >>>
> >>> Topic I remember which can be good starters:
> >>>
> >>> - Current limitations of QAPI (& QMP) model (Markus Armbruster)
> >>>
> >>> - Adding a new qemu-runtime-$TARGET / QMP-only binary without today's
> >>>     limitations (Daniel P. Berrangé & Paolo Bonzini)
> >>>
> >>> - Problem with x-exit-preconfig, reworking MachinePhaseInit state
> >>>     machine (Paolo Bonzini)
> >>>
> >>> Markus / Daniel / Paolo expressed their ideas on the list (the
> >>> historical threads are referenced in the etherpad) so reading the
> >>> relevant threads before the call will help to get in the topic.
> >>> These people don't have to be in the call, but if they can attend
> >>> that would be very nice :)
> >>>
> >>> The call will be Tuesday, December 13 at 3pm CET on this Bluejeans
> link:
> >>> http://bluejeans.com/quintela
> >> We moved the call here:
> >>
> >>
> https://teams.microsoft.com/_#/pre-join-calling/19:meeting_MjU2NDhkOGYtOGY5ZC00MzRjLWJjNjgtYWI2ODQxZGExMWNj@thread.v2
> >
> > Please use a video conferencing system that allows people to join
> > without logging in and works across browsers. I gave up on Teams.
>
> I also gave up after 27 or so clicks, redirections, emails checks, etc.
> Too complex. There are much simpler alternatives.
>
>
Sorry, similar situation here:

"Hmm. Your browser version isn't supported."

firefox-107.0.1-1.fc37.x86_64

[-- Attachment #2: Type: text/html, Size: 3864 bytes --]

Re: Single system binary & Dynamic machine model (KVM developers conference call 2022-12-13)

[Juan Quintela]

Philippe Mathieu-Daudé <philmd@linaro.org> wrote:
> On 12/12/22 00:41, Philippe Mathieu-Daudé wrote:
>> Hi,
>> In the last years we had few discussions on "simplifying" QEMU
>> (system emulation / virtualization), in particular for the
>> "management layer".
>> Some of us are interested in having QEMU able to dynamically create
>> machine models. Mark Burton's current approach is via a Python script
>> which generates QMP commands. This is just another case of "management
>> layer".
>> Various problems have been raised regarding the current limitations
>> of
>> QEMU's APIs. We'd like to remember / get a broader idea on these limits
>> and look at some ideas / proposals which have been discussed / posted
>> on this list.
>> Feel free to complete your thoughts on this public etherpad:
>> https://etherpad.opendev.org/p/qemu-emulation-bof%402022-12-13
>> Topic I remember which can be good starters:
>> - Current limitations of QAPI (& QMP) model (Markus Armbruster)
>> - Adding a new qemu-runtime-$TARGET / QMP-only binary without
>> today's
>>    limitations (Daniel P. Berrangé & Paolo Bonzini)
>> - Problem with x-exit-preconfig, reworking MachinePhaseInit state
>>    machine (Paolo Bonzini)
>> Markus / Daniel / Paolo expressed their ideas on the list (the
>> historical threads are referenced in the etherpad) so reading the
>> relevant threads before the call will help to get in the topic.
>> These people don't have to be in the call, but if they can attend
>> that would be very nice :)
>> The call will be Tuesday, December 13 at 3pm CET on this Bluejeans
>> link:
>> http://bluejeans.com/quintela
> We moved the call here:
>
> https://teams.microsoft.com/_#/pre-join-calling/19:meeting_MjU2NDhkOGYtOGY5ZC00MzRjLWJjNjgtYWI2ODQxZGExMWNj@thread.v2

Hi

First of all I am sorry that my calendar/alarm system failed me.

Once thold that, I have been unable to go from that link to a meet room:
- on android phone with chrome
- on android phone with firefox
- on linux host with firefox (and teams installed)
- on linux host with chrome (and teams installed)

As everybody says, it got going for indirection to indirection and at
the end (if lucky) I ended on the teams app without a room name/id.

Later, Juan.

Re: Single system binary & Dynamic machine model (KVM developers conference call 2022-12-13)

[Mark Burton]

[-- Attachment #1: Type: text/plain, Size: 2844 bytes --]

(BTW, really happy that we use another approach for next time - we only used it because we didn’t have another quick choice when the meeting didn’t open….)

Cheers
Mark


On 13/12/2022, 15:52, "Marc-André Lureau" <marcandre.lureau@redhat.com> wrote:


WARNING: This email originated from outside of Qualcomm. Please be wary of any links or attachments, and do not enable macros.
Hi

On Tue, Dec 13, 2022 at 6:44 PM Cédric Le Goater <clg@kaod.org<mailto:clg@kaod.org>> wrote:
On 12/13/22 15:17, Stefan Hajnoczi wrote:
> On Tue, 13 Dec 2022 at 09:08, Philippe Mathieu-Daudé <philmd@linaro.org<mailto:philmd@linaro.org>> wrote:
>>
>> On 12/12/22 00:41, Philippe Mathieu-Daudé wrote:
>>> Hi,
>>>
>>> In the last years we had few discussions on "simplifying" QEMU (system
>>> emulation / virtualization), in particular for the "management layer".
>>>
>>> Some of us are interested in having QEMU able to dynamically create
>>> machine models. Mark Burton's current approach is via a Python script
>>> which generates QMP commands. This is just another case of "management
>>> layer".
>>>
>>> Various problems have been raised regarding the current limitations of
>>> QEMU's APIs. We'd like to remember / get a broader idea on these limits
>>> and look at some ideas / proposals which have been discussed / posted
>>> on this list.
>>>
>>> Feel free to complete your thoughts on this public etherpad:
>>> https://etherpad.opendev.org/p/qemu-emulation-bof%402022-12-13
>>>
>>> Topic I remember which can be good starters:
>>>
>>> - Current limitations of QAPI (& QMP) model (Markus Armbruster)
>>>
>>> - Adding a new qemu-runtime-$TARGET / QMP-only binary without today's
>>>     limitations (Daniel P. Berrangé & Paolo Bonzini)
>>>
>>> - Problem with x-exit-preconfig, reworking MachinePhaseInit state
>>>     machine (Paolo Bonzini)
>>>
>>> Markus / Daniel / Paolo expressed their ideas on the list (the
>>> historical threads are referenced in the etherpad) so reading the
>>> relevant threads before the call will help to get in the topic.
>>> These people don't have to be in the call, but if they can attend
>>> that would be very nice :)
>>>
>>> The call will be Tuesday, December 13 at 3pm CET on this Bluejeans link:
>>> http://bluejeans.com/quintela
>> We moved the call here:
>>
>> https://teams.microsoft.com/_#/pre-join-calling/19:meeting_MjU2NDhkOGYtOGY5ZC00MzRjLWJjNjgtYWI2ODQxZGExMWNj@thread.v2
>
> Please use a video conferencing system that allows people to join
> without logging in and works across browsers. I gave up on Teams.

I also gave up after 27 or so clicks, redirections, emails checks, etc.
Too complex. There are much simpler alternatives.

Sorry, similar situation here:

"Hmm. Your browser version isn't supported."

firefox-107.0.1-1.fc37.x86_64


[-- Attachment #2: Type: text/html, Size: 7680 bytes --]

Re: Single system binary & Dynamic machine model (KVM developers conference call 2022-12-13)

[Alex Bennée]

Juan Quintela <quintela@redhat.com> writes:

> Philippe Mathieu-Daudé <philmd@linaro.org> wrote:
>> On 12/12/22 00:41, Philippe Mathieu-Daudé wrote:
>>> Hi,
>>> In the last years we had few discussions on "simplifying" QEMU
>>> (system emulation / virtualization), in particular for the
>>> "management layer".
>>> Some of us are interested in having QEMU able to dynamically create
>>> machine models. Mark Burton's current approach is via a Python script
>>> which generates QMP commands. This is just another case of "management
>>> layer".
>>> Various problems have been raised regarding the current limitations
>>> of
>>> QEMU's APIs. We'd like to remember / get a broader idea on these limits
>>> and look at some ideas / proposals which have been discussed / posted
>>> on this list.
>>> Feel free to complete your thoughts on this public etherpad:
>>> https://etherpad.opendev.org/p/qemu-emulation-bof%402022-12-13
>>> Topic I remember which can be good starters:
>>> - Current limitations of QAPI (& QMP) model (Markus Armbruster)
>>> - Adding a new qemu-runtime-$TARGET / QMP-only binary without
>>> today's
>>>    limitations (Daniel P. Berrangé & Paolo Bonzini)
>>> - Problem with x-exit-preconfig, reworking MachinePhaseInit state
>>>    machine (Paolo Bonzini)
>>> Markus / Daniel / Paolo expressed their ideas on the list (the
>>> historical threads are referenced in the etherpad) so reading the
>>> relevant threads before the call will help to get in the topic.
>>> These people don't have to be in the call, but if they can attend
>>> that would be very nice :)
>>> The call will be Tuesday, December 13 at 3pm CET on this Bluejeans
>>> link:
>>> http://bluejeans.com/quintela
>> We moved the call here:
>>
>> https://teams.microsoft.com/_#/pre-join-calling/19:meeting_MjU2NDhkOGYtOGY5ZC00MzRjLWJjNjgtYWI2ODQxZGExMWNj@thread.v2
>
> Hi
>
> First of all I am sorry that my calendar/alarm system failed me.

I tried starting as a moderator myself but it needed a login. Is the
pass code personal to you or the meeting? Would it be worth adding the
pass code to the meeting details for the future.

> Once thold that, I have been unable to go from that link to a meet room:
> - on android phone with chrome
> - on android phone with firefox
> - on linux host with firefox (and teams installed)
> - on linux host with chrome (and teams installed)
>
> As everybody says, it got going for indirection to indirection and at
> the end (if lucky) I ended on the teams app without a room name/id.

Mine worked OK with Linux Chrome in web only mode. As a fallback someone
did post:

  https://play.workadventu.re/@/qemu/main/hq

a while back. More traditionally:

  https://meet.jit.si/

allows you to create meetings without sign-up.

>
> Later, Juan.


-- 
Alex Bennée

Re: Single system binary & Dynamic machine model (KVM developers conference call 2022-12-13)

[Felipe Franciosi]

> On 13 Dec 2022, at 14:08, Philippe Mathieu-Daudé <philmd@linaro.org> wrote:
> 
> We moved the call here:
> 
> https://teams.microsoft.com/_#/pre-join-calling/19:meeting_MjU2NDhkOGYtOGY5ZC00MzRjLWJjNjgtYWI2ODQxZGExMWNj@thread.v2

Just wondering: what's wrong with Google Hangout?

If people want to try Zoom, I'd be happy to look into setting it up.

F.

Re: Single system binary & Dynamic machine model (KVM developers conference call 2022-12-13)

[Mark Burton]

[-- Attachment #1: Type: text/plain, Size: 743 bytes --]

Happy with any choice so long as the meeting can be opened (either by any of us, or by a ‘larger’ number of people) - it’s not fair it relies on one person.
Cheers
Mark.


On 13/12/2022, 17:35, "Felipe Franciosi" <felipe@nutanix.com> wrote:

WARNING: This email originated from outside of Qualcomm. Please be wary of any links or attachments, and do not enable macros.

> On 13 Dec 2022, at 14:08, Philippe Mathieu-Daudé <philmd@linaro.org> wrote:
>
> We moved the call here:
>
> https://teams.microsoft.com/_#/pre-join-calling/19:meeting_MjU2NDhkOGYtOGY5ZC00MzRjLWJjNjgtYWI2ODQxZGExMWNj@thread.v2

Just wondering: what's wrong with Google Hangout?

If people want to try Zoom, I'd be happy to look into setting it up.

F.

[-- Attachment #2: Type: text/html, Size: 2956 bytes --]

Re: Single system binary & Dynamic machine model (KVM developers conference call 2022-12-13)

[Peter Maydell]

On Tue, 13 Dec 2022 at 16:34, Felipe Franciosi <felipe@nutanix.com> wrote:
>
>
> > On 13 Dec 2022, at 14:08, Philippe Mathieu-Daudé <philmd@linaro.org> wrote:
> >
> > We moved the call here:
> >
> > https://teams.microsoft.com/_#/pre-join-calling/19:meeting_MjU2NDhkOGYtOGY5ZC00MzRjLWJjNjgtYWI2ODQxZGExMWNj@thread.v2
>
> Just wondering: what's wrong with Google Hangout?

Nothing. We were just a dozen people sat in the chat window of a
BlueJeans meeting we couldn't get into, and somebody volunteered
"I can quickly set up a meeting in the tool I happen to use",
so we used that rather than spending another 10 minutes debating
which videoconferencing setup to use.

thanks
-- PMM