[Virtio-fs] Changes in -dev branch

[Virtio-fs] Changes in -dev branch

[Dr. David Alan Gilbert]

Hi,
   I've just pushed a new version of the virtio-fs-dev branch in
 https://gitlab.com/virtio-fs/qemu
 
some things to note:

  a) The old '-o vhost_user_socket' syntax has now been replaced
     by the syntax matching qemu's standardised vhost-user interface
     which is;

     ./virtiofsd --socket-path=/path/whereever

  b) It's got about 64 of the main patches reviewed by Daniel Berrangé
     - about another 40 that need reviewing; please leap in
     and review any patches on the qemu-devel thread for any patches
     that are missing reviews.

  c) The set that I've just pushed has fixes for all the things
     Daniel spotted; so there's a bunch of mostly minor changes.

  d) It's also got patches that were received on list since
     mid December.

Dave

--
Dr. David Alan Gilbert / dgilbert@redhat.com / Manchester, UK

Re: [Virtio-fs] Changes in -dev branch

[Boeuf, Sebastien]

Hi Dave,

On Fri, 2020-01-10 at 19:03 +0000, Dr. David Alan Gilbert wrote:
> Hi,
>    I've just pushed a new version of the virtio-fs-dev branch in
>  https://gitlab.com/virtio-fs/qemu
>  
> some things to note:
> 
>   a) The old '-o vhost_user_socket' syntax has now been replaced
>      by the syntax matching qemu's standardised vhost-user interface
>      which is;
> 
>      ./virtiofsd --socket-path=/path/whereever

I was expecting that "./virtiofsd --socket-path /path/whereever" would
work too, but it doesn't. Is this the expected behavior?

Just asking cause it feels weird that we must add "=" here.

Thanks,
Sebastien

> 
>   b) It's got about 64 of the main patches reviewed by Daniel
> Berrangé
>      - about another 40 that need reviewing; please leap in
>      and review any patches on the qemu-devel thread for any patches
>      that are missing reviews.
> 
>   c) The set that I've just pushed has fixes for all the things
>      Daniel spotted; so there's a bunch of mostly minor changes.
> 
>   d) It's also got patches that were received on list since
>      mid December.
> 
> Dave
> 
> --
> Dr. David Alan Gilbert / dgilbert@redhat.com / Manchester, UK
> 
> _______________________________________________
> Virtio-fs mailing list
> Virtio-fs@redhat.com
> https://www.redhat.com/mailman/listinfo/virtio-fs
> 
---------------------------------------------------------------------
Intel Corporation SAS (French simplified joint stock company)
Registered headquarters: "Les Montalets"- 2, rue de Paris, 
92196 Meudon Cedex, France
Registration Number:  302 456 199 R.C.S. NANTERRE
Capital: 4,572,000 Euros

This e-mail and any attachments may contain confidential material for
the sole use of the intended recipient(s). Any review or distribution
by others is strictly prohibited. If you are not the intended
recipient, please contact the sender and delete all copies.

Re: [Virtio-fs] Changes in -dev branch

[Dr. David Alan Gilbert]

* Boeuf, Sebastien (sebastien.boeuf@intel.com) wrote:
> Hi Dave,
> 
> On Fri, 2020-01-10 at 19:03 +0000, Dr. David Alan Gilbert wrote:
> > Hi,
> >    I've just pushed a new version of the virtio-fs-dev branch in
> >  https://gitlab.com/virtio-fs/qemu
> >  
> > some things to note:
> > 
> >   a) The old '-o vhost_user_socket' syntax has now been replaced
> >      by the syntax matching qemu's standardised vhost-user interface
> >      which is;
> > 
> >      ./virtiofsd --socket-path=/path/whereever
> 
> I was expecting that "./virtiofsd --socket-path /path/whereever" would
> work too, but it doesn't. Is this the expected behavior?
> 
> Just asking cause it feels weird that we must add "=" here.

--socket-path= is the stadnard defined in qemu's
docs/interop/vhost-user.rst and fuse_lowlevel also seems
to standardise on options that take parameters using =.

Dave

> Thanks,
> Sebastien
> 
> > 
> >   b) It's got about 64 of the main patches reviewed by Daniel
> > Berrangé
> >      - about another 40 that need reviewing; please leap in
> >      and review any patches on the qemu-devel thread for any patches
> >      that are missing reviews.
> > 
> >   c) The set that I've just pushed has fixes for all the things
> >      Daniel spotted; so there's a bunch of mostly minor changes.
> > 
> >   d) It's also got patches that were received on list since
> >      mid December.
> > 
> > Dave
> > 
> > --
> > Dr. David Alan Gilbert / dgilbert@redhat.com / Manchester, UK
> > 
> > _______________________________________________
> > Virtio-fs mailing list
> > Virtio-fs@redhat.com
> > https://www.redhat.com/mailman/listinfo/virtio-fs
> > 
> ---------------------------------------------------------------------
> Intel Corporation SAS (French simplified joint stock company)
> Registered headquarters: "Les Montalets"- 2, rue de Paris, 
> 92196 Meudon Cedex, France
> Registration Number:  302 456 199 R.C.S. NANTERRE
> Capital: 4,572,000 Euros
> 
> This e-mail and any attachments may contain confidential material for
> the sole use of the intended recipient(s). Any review or distribution
> by others is strictly prohibited. If you are not the intended
> recipient, please contact the sender and delete all copies.
--
Dr. David Alan Gilbert / dgilbert@redhat.com / Manchester, UK

Re: [Virtio-fs] Changes in -dev branch

[Boeuf, Sebastien]

On Tue, 2020-01-14 at 12:25 +0000, Dr. David Alan Gilbert wrote:
> * Boeuf, Sebastien (sebastien.boeuf@intel.com) wrote:
> > Hi Dave,
> > 
> > On Fri, 2020-01-10 at 19:03 +0000, Dr. David Alan Gilbert wrote:
> > > Hi,
> > >    I've just pushed a new version of the virtio-fs-dev branch in
> > >  https://gitlab.com/virtio-fs/qemu
> > >  
> > > some things to note:
> > > 
> > >   a) The old '-o vhost_user_socket' syntax has now been replaced
> > >      by the syntax matching qemu's standardised vhost-user
> > > interface
> > >      which is;
> > > 
> > >      ./virtiofsd --socket-path=/path/whereever
> > 
> > I was expecting that "./virtiofsd --socket-path /path/whereever"
> > would
> > work too, but it doesn't. Is this the expected behavior?
> > 
> > Just asking cause it feels weird that we must add "=" here.
> 
> --socket-path= is the stadnard defined in qemu's
> docs/interop/vhost-user.rst and fuse_lowlevel also seems
> to standardise on options that take parameters using =.

Fair enough :)

Thanks,
Sebastien

> 
> Dave
> 
> > Thanks,
> > Sebastien
> > 
> > >   b) It's got about 64 of the main patches reviewed by Daniel
> > > Berrangé
> > >      - about another 40 that need reviewing; please leap in
> > >      and review any patches on the qemu-devel thread for any
> > > patches
> > >      that are missing reviews.
> > > 
> > >   c) The set that I've just pushed has fixes for all the things
> > >      Daniel spotted; so there's a bunch of mostly minor changes.
> > > 
> > >   d) It's also got patches that were received on list since
> > >      mid December.
> > > 
> > > Dave
> > > 
> > > --
> > > Dr. David Alan Gilbert / dgilbert@redhat.com / Manchester, UK
> > > 
> > > _______________________________________________
> > > Virtio-fs mailing list
> > > Virtio-fs@redhat.com
> > > https://www.redhat.com/mailman/listinfo/virtio-fs
> > > 
> > -----------------------------------------------------------------
> > ----
> > Intel Corporation SAS (French simplified joint stock company)
> > Registered headquarters: "Les Montalets"- 2, rue de Paris, 
> > 92196 Meudon Cedex, France
> > Registration Number:  302 456 199 R.C.S. NANTERRE
> > Capital: 4,572,000 Euros
> > 
> > This e-mail and any attachments may contain confidential material
> > for
> > the sole use of the intended recipient(s). Any review or
> > distribution
> > by others is strictly prohibited. If you are not the intended
> > recipient, please contact the sender and delete all copies.
> --
> Dr. David Alan Gilbert / dgilbert@redhat.com / Manchester, UK
> 
---------------------------------------------------------------------
Intel Corporation SAS (French simplified joint stock company)
Registered headquarters: "Les Montalets"- 2, rue de Paris, 
92196 Meudon Cedex, France
Registration Number:  302 456 199 R.C.S. NANTERRE
Capital: 4,572,000 Euros

This e-mail and any attachments may contain confidential material for
the sole use of the intended recipient(s). Any review or distribution
by others is strictly prohibited. If you are not the intended
recipient, please contact the sender and delete all copies.

[Virtio-fs] Changes in -dev branch

[Dr. David Alan Gilbert]

Hi,
   I've just pushed a new version of the virtio-fs-dev branch in
  https://gitlab.com/virtio-fs/qemu

This is based off a very recent qemu tree, and we've got some new
virtiofs features in (thanks to all contributors):

  a) A new sandbox option that uses chroot rather than pivot_root
    and namespacing

  b) A mechanism that lets xattr names be translated between the
    host and guest.

  c) an explicit configure option (--enable-virtiofsd) to force
   compilation of virtiofsd - the build system will then tell
   you if you're missing build dependencies rather than quietly
   disabling it.

Dave

-- 
Dr. David Alan Gilbert / dgilbert@redhat.com / Manchester, UK

[Virtio-fs] Changes in -dev branch

[Dr. David Alan Gilbert]

Hi,
   I've just pushed a new version of the virtio-fs-dev branch in
  https://gitlab.com/virtio-fs/qemu

There's no major changes from the virtiofs world in there, but it is
based on a bleeding edge qemu build, which has some major changes in
it's build system.

Note that the virtiofsd binary ends up in ./tools/virtiofsd/virtiofsd
rather than the top level where the existing build system puts it.

Dave

-- 
Dr. David Alan Gilbert / dgilbert@redhat.com / Manchester, UK

[Virtio-fs] Changes in -dev branch

[Dr. David Alan Gilbert]

Hi,
  I've just pushed a new version of the virtio-fs-dev branch in
https://gitlab.com/virtio-fs/qemu

some things to note:

  a) It's now rebased on upstream qemu 4.2-rc5
  b) The virtiofsd is now in tools/virtiofsd rather than contrib
     (after a long discussion on qemu-devel about where people
      want it)
  c) I've reformatted all the files to match qemu's coding style
     guidelines; mostly.  This is a mostly automatic process
     with some hand tweaking.  There's the possibility some stuff
     is screwed up so I need to check on it.
  d) It now uses libcap-ng rather libcap; that's the way qemu is
     moving, allthough the change is a bit hairy.
  e) If you're missing either libseccomp or libcapng during build
     it will omit virtiofsd rather than failing the build.

Dave


--
Dr. David Alan Gilbert / dgilbert@redhat.com / Manchester, UK

Re: [Virtio-fs] Changes in -dev branch

[Boeuf, Sebastien]

Hi Dave,

I've just checked the codebase and I can see it includes the libvhost-
user fixes "libvhost-user: fix SLAVE_SEND_FD handling" and "libvhost-
user: introduce and use vu_has_protocol_feature()", that's great!

Unfortunately, through the virtio-fs development (based on previous
libvhost-user code), a new bug has been introduced here: 
https://gitlab.com/virtio-fs/qemu/blob/virtio-fs-dev/contrib/libvhost-user/libvhost-user.c#L2558

Instead, it needs to follow the new code format (which fixes the bug)
like here: 
https://gitlab.com/virtio-fs/qemu/blob/virtio-fs-dev/contrib/libvhost-user/libvhost-user.c#L1136

Do you think that's something we can quickly fix on your branch?

I can push a patch if you need me to :)

Thanks,
Sebastien

On Wed, 2019-12-11 at 12:16 +0000, Dr. David Alan Gilbert wrote:
> Hi,
>   I've just pushed a new version of the virtio-fs-dev branch in
> https://gitlab.com/virtio-fs/qemu
> 
> some things to note:
> 
>   a) It's now rebased on upstream qemu 4.2-rc5
>   b) The virtiofsd is now in tools/virtiofsd rather than contrib
>      (after a long discussion on qemu-devel about where people
>       want it)
>   c) I've reformatted all the files to match qemu's coding style
>      guidelines; mostly.  This is a mostly automatic process
>      with some hand tweaking.  There's the possibility some stuff
>      is screwed up so I need to check on it.
>   d) It now uses libcap-ng rather libcap; that's the way qemu is
>      moving, allthough the change is a bit hairy.
>   e) If you're missing either libseccomp or libcapng during build
>      it will omit virtiofsd rather than failing the build.
> 
> Dave
> 
> 
> --
> Dr. David Alan Gilbert / dgilbert@redhat.com / Manchester, UK
> 
> _______________________________________________
> Virtio-fs mailing list
> Virtio-fs@redhat.com
> https://www.redhat.com/mailman/listinfo/virtio-fs
> 
---------------------------------------------------------------------
Intel Corporation SAS (French simplified joint stock company)
Registered headquarters: "Les Montalets"- 2, rue de Paris, 
92196 Meudon Cedex, France
Registration Number:  302 456 199 R.C.S. NANTERRE
Capital: 4,572,000 Euros

This e-mail and any attachments may contain confidential material for
the sole use of the intended recipient(s). Any review or distribution
by others is strictly prohibited. If you are not the intended
recipient, please contact the sender and delete all copies.

Re: [Virtio-fs] Changes in -dev branch

[Dr. David Alan Gilbert]

* Boeuf, Sebastien (sebastien.boeuf@intel.com) wrote:
> Hi Dave,
> 
> I've just checked the codebase and I can see it includes the libvhost-
> user fixes "libvhost-user: fix SLAVE_SEND_FD handling" and "libvhost-
> user: introduce and use vu_has_protocol_feature()", that's great!
> 
> Unfortunately, through the virtio-fs development (based on previous
> libvhost-user code), a new bug has been introduced here: 
> https://gitlab.com/virtio-fs/qemu/blob/virtio-fs-dev/contrib/libvhost-user/libvhost-user.c#L2558
> 
> Instead, it needs to follow the new code format (which fixes the bug)
> like here: 
> https://gitlab.com/virtio-fs/qemu/blob/virtio-fs-dev/contrib/libvhost-user/libvhost-user.c#L1136

Oops!

> Do you think that's something we can quickly fix on your branch?
> 
> I can push a patch if you need me to :)

Pushed; can you just check that's OK?

Dave

> Thanks,
> Sebastien
> 
> On Wed, 2019-12-11 at 12:16 +0000, Dr. David Alan Gilbert wrote:
> > Hi,
> >   I've just pushed a new version of the virtio-fs-dev branch in
> > https://gitlab.com/virtio-fs/qemu
> > 
> > some things to note:
> > 
> >   a) It's now rebased on upstream qemu 4.2-rc5
> >   b) The virtiofsd is now in tools/virtiofsd rather than contrib
> >      (after a long discussion on qemu-devel about where people
> >       want it)
> >   c) I've reformatted all the files to match qemu's coding style
> >      guidelines; mostly.  This is a mostly automatic process
> >      with some hand tweaking.  There's the possibility some stuff
> >      is screwed up so I need to check on it.
> >   d) It now uses libcap-ng rather libcap; that's the way qemu is
> >      moving, allthough the change is a bit hairy.
> >   e) If you're missing either libseccomp or libcapng during build
> >      it will omit virtiofsd rather than failing the build.
> > 
> > Dave
> > 
> > 
> > --
> > Dr. David Alan Gilbert / dgilbert@redhat.com / Manchester, UK
> > 
> > _______________________________________________
> > Virtio-fs mailing list
> > Virtio-fs@redhat.com
> > https://www.redhat.com/mailman/listinfo/virtio-fs
> > 
> ---------------------------------------------------------------------
> Intel Corporation SAS (French simplified joint stock company)
> Registered headquarters: "Les Montalets"- 2, rue de Paris, 
> 92196 Meudon Cedex, France
> Registration Number:  302 456 199 R.C.S. NANTERRE
> Capital: 4,572,000 Euros
> 
> This e-mail and any attachments may contain confidential material for
> the sole use of the intended recipient(s). Any review or distribution
> by others is strictly prohibited. If you are not the intended
> recipient, please contact the sender and delete all copies.
--
Dr. David Alan Gilbert / dgilbert@redhat.com / Manchester, UK

Re: [Virtio-fs] Changes in -dev branch

[Boeuf, Sebastien]

On Wed, 2019-12-11 at 14:00 +0000, Dr. David Alan Gilbert wrote:
> * Boeuf, Sebastien (sebastien.boeuf@intel.com) wrote:
> > Hi Dave,
> > 
> > I've just checked the codebase and I can see it includes the
> > libvhost-
> > user fixes "libvhost-user: fix SLAVE_SEND_FD handling" and
> > "libvhost-
> > user: introduce and use vu_has_protocol_feature()", that's great!
> > 
> > Unfortunately, through the virtio-fs development (based on previous
> > libvhost-user code), a new bug has been introduced here: 
> > https://gitlab.com/virtio-fs/qemu/blob/virtio-fs-dev/contrib/libvhost-user/libvhost-user.c#L2558
> > 
> > Instead, it needs to follow the new code format (which fixes the
> > bug)
> > like here: 
> > https://gitlab.com/virtio-fs/qemu/blob/virtio-fs-dev/contrib/libvhost-user/libvhost-user.c#L1136
> 
> Oops!
> 
> > Do you think that's something we can quickly fix on your branch?
> > 
> > I can push a patch if you need me to :)
> 
> Pushed; can you just check that's OK?

Just checked, it looks good :)

I need to verify it, but I think the branch can now be directly used
with Cloud-Hypervisor. I'll let you know if I run into any kind of
trouble.

Thanks,
Sebastien

> 
> Dave
> 
> > Thanks,
> > Sebastien
> > 
> > On Wed, 2019-12-11 at 12:16 +0000, Dr. David Alan Gilbert wrote:
> > > Hi,
> > >   I've just pushed a new version of the virtio-fs-dev branch in
> > > https://gitlab.com/virtio-fs/qemu
> > > 
> > > some things to note:
> > > 
> > >   a) It's now rebased on upstream qemu 4.2-rc5
> > >   b) The virtiofsd is now in tools/virtiofsd rather than contrib
> > >      (after a long discussion on qemu-devel about where people
> > >       want it)
> > >   c) I've reformatted all the files to match qemu's coding style
> > >      guidelines; mostly.  This is a mostly automatic process
> > >      with some hand tweaking.  There's the possibility some stuff
> > >      is screwed up so I need to check on it.
> > >   d) It now uses libcap-ng rather libcap; that's the way qemu is
> > >      moving, allthough the change is a bit hairy.
> > >   e) If you're missing either libseccomp or libcapng during build
> > >      it will omit virtiofsd rather than failing the build.
> > > 
> > > Dave
> > > 
> > > 
> > > --
> > > Dr. David Alan Gilbert / dgilbert@redhat.com / Manchester, UK
> > > 
> > > _______________________________________________
> > > Virtio-fs mailing list
> > > Virtio-fs@redhat.com
> > > https://www.redhat.com/mailman/listinfo/virtio-fs
> > > 
> > -----------------------------------------------------------------
> > ----
> > Intel Corporation SAS (French simplified joint stock company)
> > Registered headquarters: "Les Montalets"- 2, rue de Paris, 
> > 92196 Meudon Cedex, France
> > Registration Number:  302 456 199 R.C.S. NANTERRE
> > Capital: 4,572,000 Euros
> > 
> > This e-mail and any attachments may contain confidential material
> > for
> > the sole use of the intended recipient(s). Any review or
> > distribution
> > by others is strictly prohibited. If you are not the intended
> > recipient, please contact the sender and delete all copies.
> --
> Dr. David Alan Gilbert / dgilbert@redhat.com / Manchester, UK
> 
---------------------------------------------------------------------
Intel Corporation SAS (French simplified joint stock company)
Registered headquarters: "Les Montalets"- 2, rue de Paris, 
92196 Meudon Cedex, France
Registration Number:  302 456 199 R.C.S. NANTERRE
Capital: 4,572,000 Euros

This e-mail and any attachments may contain confidential material for
the sole use of the intended recipient(s). Any review or distribution
by others is strictly prohibited. If you are not the intended
recipient, please contact the sender and delete all copies.