Re: [Qemu-devel] [PATCH 1/1] 9pfs: fix and simplify dependencies

[Cornelia Huck]

On Tue, 8 Aug 2017 16:15:59 +0200
Greg Kurz <address@hidden> wrote:

> On Tue, 8 Aug 2017 14:50:00 +0200
> Greg Kurz <address@hidden> wrote:
> 
> > On Tue,  8 Aug 2017 14:17:01 +0200
> > Cornelia Huck <address@hidden> wrote:
> >   
> > > We want to compile the 9pfs code iff virtfs is enabled and a virtio
> > > transport is available (not only pci!).
> > > 
> > > Let's make the generated CONFIG_VIRTFS statement depend on either
> > > CONFIG_VIRTIO_PCI or (new) CONFIG_VIRTIO_CCW. The Makefiles for fsdev/
> > > and hw/9pfs/ can then depend on CONFIG_VIRTFS directly.
> > > 
> > > Suggested-by: Thomas Huth <address@hidden>
> > > Signed-off-by: Cornelia Huck <address@hidden>
> > > ---    
> > 
> > It looks good to me.
> >   
> 
> Oops, reading Thomas's other mail, I now realize that with this patch we
> end up with the following in config-host.mak:
> 
> CONFIG_VIRTFS=$(call lor, $(CONFIG_VIRTIO_PCI),$(CONFIG_VIRTIO_CCW))
> 
> which looks wrong even if it may "work"... but we also get the
> following in config-host.h:
> 
> #define CONFIG_VIRTFS $(call lor, $(CONFIG_VIRTIO_PCI),$(CONFIG_VIRTIO_CCW))
> 
> which is clearly wrong.

Eek. Oh well, it would have been nice...

> 
> Maybe post the changes from your other mail instead ?
> 
> Message-ID: <address@hidden>
> 
> The patch isn't that ugly, given the ugliness of the current makefiles :)

It is slightly differently ugly :)

> 
> And we can come up with some cleanup patch later.

Nod.

> 
> > Acked-by: Greg Kurz <address@hidden>
> >   
> 
> NAcked-by: in fact :)

Heh :)

I'll post the other patch properly.