Re: [Qemu-devel] [PATCH slirp 4/5] slirp: add state saving/loading

2019-02-09 Thread Samuel Thibault
Marc-André Lureau, le sam. 09 févr. 2019 14:16:53 +0100, a ecrit:
> Hi
> 
> On Sat, Feb 9, 2019 at 2:13 PM Samuel Thibault  
> wrote:
> >
> > Marc-André Lureau, le sam. 09 févr. 2019 14:09:02 +0100, a ecrit:
> > > On Sat, Feb 9, 2019 at 12:40 PM Samuel Thibault  
> > > wrote:
> > > >
> > > > marcandre.lur...@redhat.com, le ven. 08 févr. 2019 19:11:21 +0100, a 
> > > > ecrit:
> > > > > From: Marc-André Lureau 
> > > > >
> > > > > Based on qemu vmstate serialization code. At this point it should
> > > > > produce the same result. However, with future state versions, slirp
> > > > > should be free to change its format, by bumping the reported state
> > > > > version.
> > > >
> > > > Mmm, then qemu needs to be taught to call slirp_state_save/load/version?
> > >
> > > Yes, see "[Qemu-devel] [PATCH 1/1] RFC: net/slirp: link with libslirp"
> >
> > Ah, I did have a look there but somehow missed it because burried within
> > file removals :)
> >
> 
> Indeed, the original patch series was splited in 2, but I thought as
> an RFC it was easier to send a single patch. Bad choice I suppose.

No, it's fine enough, it's just the file ordering in the patch which
posed problem :)

Samuel



Re: [Qemu-devel] [PATCH slirp 4/5] slirp: add state saving/loading

2019-02-09 Thread Marc-André Lureau
Hi

On Sat, Feb 9, 2019 at 2:13 PM Samuel Thibault  wrote:
>
> Marc-André Lureau, le sam. 09 févr. 2019 14:09:02 +0100, a ecrit:
> > On Sat, Feb 9, 2019 at 12:40 PM Samuel Thibault  
> > wrote:
> > >
> > > marcandre.lur...@redhat.com, le ven. 08 févr. 2019 19:11:21 +0100, a 
> > > ecrit:
> > > > From: Marc-André Lureau 
> > > >
> > > > Based on qemu vmstate serialization code. At this point it should
> > > > produce the same result. However, with future state versions, slirp
> > > > should be free to change its format, by bumping the reported state
> > > > version.
> > >
> > > Mmm, then qemu needs to be taught to call slirp_state_save/load/version?
> >
> > Yes, see "[Qemu-devel] [PATCH 1/1] RFC: net/slirp: link with libslirp"
>
> Ah, I did have a look there but somehow missed it because burried within
> file removals :)
>

Indeed, the original patch series was splited in 2, but I thought as
an RFC it was easier to send a single patch. Bad choice I suppose.



-- 
Marc-André Lureau



Re: [Qemu-devel] [PATCH slirp 4/5] slirp: add state saving/loading

2019-02-09 Thread Samuel Thibault
Marc-André Lureau, le sam. 09 févr. 2019 14:09:02 +0100, a ecrit:
> On Sat, Feb 9, 2019 at 12:40 PM Samuel Thibault  
> wrote:
> >
> > marcandre.lur...@redhat.com, le ven. 08 févr. 2019 19:11:21 +0100, a ecrit:
> > > From: Marc-André Lureau 
> > >
> > > Based on qemu vmstate serialization code. At this point it should
> > > produce the same result. However, with future state versions, slirp
> > > should be free to change its format, by bumping the reported state
> > > version.
> >
> > Mmm, then qemu needs to be taught to call slirp_state_save/load/version?
> 
> Yes, see "[Qemu-devel] [PATCH 1/1] RFC: net/slirp: link with libslirp"

Ah, I did have a look there but somehow missed it because burried within
file removals :)

Samuel



Re: [Qemu-devel] [PATCH slirp 4/5] slirp: add state saving/loading

2019-02-09 Thread Marc-André Lureau
Hi

On Sat, Feb 9, 2019 at 12:40 PM Samuel Thibault  wrote:
>
> marcandre.lur...@redhat.com, le ven. 08 févr. 2019 19:11:21 +0100, a ecrit:
> > From: Marc-André Lureau 
> >
> > Based on qemu vmstate serialization code. At this point it should
> > produce the same result. However, with future state versions, slirp
> > should be free to change its format, by bumping the reported state
> > version.
>
> Mmm, then qemu needs to be taught to call slirp_state_save/load/version?

Yes, see "[Qemu-devel] [PATCH 1/1] RFC: net/slirp: link with libslirp"



Re: [Qemu-devel] [PATCH slirp 4/5] slirp: add state saving/loading

2019-02-09 Thread Samuel Thibault
marcandre.lur...@redhat.com, le ven. 08 févr. 2019 19:11:21 +0100, a ecrit:
> From: Marc-André Lureau 
> 
> Based on qemu vmstate serialization code. At this point it should
> produce the same result. However, with future state versions, slirp
> should be free to change its format, by bumping the reported state
> version.

Mmm, then qemu needs to be taught to call slirp_state_save/load/version?

Samuel