Re: [Qemu-devel] [PATCH v9 4/5] Adding packet abstraction for VMWARE network devices

[Dmitry Fleytman]

On Thu, Jan 17, 2013 at 12:24 PM, Stefan Hajnoczi <address@hidden> wrote:

On Wed, Jan 16, 2013 at 04:33:48PM +0100, Paolo Bonzini wrote:
> Il 16/01/2013 15:48, Stefan Hajnoczi ha scritto:
> >> >  hw/vmxnet_rx_pkt.c | 187 ++++++++++++++++++
> >> >  hw/vmxnet_rx_pkt.h | 173 ++++++++++++++++
> >> >  hw/vmxnet_tx_pkt.c | 567 +++++++++++++++++++++++++++++++++++++++++++++++++++++
> >> >  hw/vmxnet_tx_pkt.h | 148 ++++++++++++++
> >> >  4 files changed, 1075 insertions(+)
> >> >  create mode 100644 hw/vmxnet_rx_pkt.c
> >> >  create mode 100644 hw/vmxnet_rx_pkt.h
> >> >  create mode 100644 hw/vmxnet_tx_pkt.c
> >> >  create mode 100644 hw/vmxnet_tx_pkt.h
> > There are other VMware-specific hw/ files.  Please create hw/vmware/ and
> > put source files in there without the prefix.  Check hw/pci/ or other
> > subdirectories for how to setup Makefile.objs, it's pretty simple.
>
> I don't think we have enough plans for hw/ structure to create
> directories yet.  What we have so far is basically hw/<arch> and
> hw/<bus>, so the logic would be to have hw/net (also mimicking Linux's
> drivers/net directory).  It's premature to create hw/vmware, we risk
> getting a spaghetti directory structure.

Okay, let's leave the hw/ sub-directory for a future hw/-wide cleanup
series that moves all device emulation code.

> eth.c/eth.h sound more interesting.  Could they be used by host-side
> code (e.g. SLIRP or a TAP backend)?  If so, they belong in net/ and
> include/net.  If not, they belong in hw/ (waiting for hw/net to be created).

Yes, they can be used by host (backend) code.

Stefan

eth.* moved to "include/net" and "net"

--

Dmitry Fleytman
Technology Expert and Consultant,

Daynix Computing Ltd.

Cell: +972-54-2819481

Skype: dmitry.fleytman