Re: virtiofsd: Where should it live?

[Michael S. Tsirkin]

On Mon, Dec 02, 2019 at 04:44:23PM +0000, Dr. David Alan Gilbert wrote:
> * Michael S. Tsirkin (address@hidden) wrote:
> > On Mon, Nov 25, 2019 at 06:50:21PM +0000, Dr. David Alan Gilbert wrote:
> > > Hi,
> > >   There's been quite a bit of discussion about where virtiofsd, our
> > > implemenation of a virtiofs daemon, should live.  I'd like to get
> > > this settled now, because I'd like to tidy it up for the next
> > > qemu cycle.
> > > 
> > > For reference it's based on qemu's livhost-user+chunks of libfuse.
> > > It can't live in libfuse because we change enough of the library
> > > to break their ABI.
> > 
> > Generally there could be some ifdefs that allow one to
> > build libfuse-host or whatever from the same source.
> > I am guessing the big reason this doesn't fly is that
> > libfuse is not actively developed anymore.
> 
> libfuse is certainly taking patches; so it's not dead.
> However, the changes for the transport are quite invasive,
> and it doesn't feel right to impose them on it.
> We've pushed up small fixes/changes etc - but not things
> that are big intrusive lumps for our use.

Maybe they will want these patches then ....  The big question would be
around security, e.g.  what if you rebase, how do you know they didn't
introduce what is a security hole for virtiofsd ...  But then, that
question remains even if you keep a separate tree.

> > Given that, the main remaining part is libvhost-user,
> > and it's less work to use than to duplicate that.
> > That kind of dictates being in qemu.
> > 
> > >  It's C, and we've got ~100 patches - which
> > > we can split into about 3 chunks.
> > > 
> > > Some suggestions so far:
> > >   a) In contrib
> > >      This is my current working assumption; the main objection is it's
> > >      a bit big and pulls in a chunk of libfuse.
> > >   b) In a submodule
> > > 
> > >   c) Just separate
> > > 
> > > Your suggestions/ideas please.  My preference is (a).
> > > 
> > > Dave
> > 
> > 
> > My preference is close to a, and maybe to avoid confusion we should have
> > a new top-level directory for "separate daemons qemu invokes, and need
> > to be built together with qemu". libvhost-user would have to move there,
> > too. "modules"?
> 
> "modules" feels too close to "plugins" to my mind.
> 
> Dave

daemons?

> > 
> > > 
> > > --
> > > Dr. David Alan Gilbert / address@hidden / Manchester, UK
> > > 
> > 
> --
> Dr. David Alan Gilbert / address@hidden / Manchester, UK