RE: [PATCH 0/5] hw/net/igb: emulated network device with SR-IOV

[Sriram Yagnaraman]

> -----Original Message-----
> From: Akihiko Odaki <akihiko.odaki@gmail.com>
> Sent: Thursday, 29 December 2022 03:43
> To: Jason Wang <jasowang@redhat.com>; Sriram Yagnaraman
> <sriram.yagnaraman@est.tech>
> Cc: qemu-devel@nongnu.org; Dmitry Fleytman
> <dmitry.fleytman@gmail.com>; Michael S . Tsirkin <mst@redhat.com>;
> Marcel Apfelbaum <marcel.apfelbaum@gmail.com>; Yan Vugenfirer
> <yan@daynix.com>
> Subject: Re: [PATCH 0/5] hw/net/igb: emulated network device with SR-IOV
> 
> On 2022/12/23 12:23, Jason Wang wrote:
> > On Thu, Dec 22, 2022 at 7:43 PM Sriram Yagnaraman
> > <sriram.yagnaraman@est.tech> wrote:
> >>
> >> A new attempt at adding support for Intel 82576 Gigabit Ethernet
> >> adapter with SR-IOV support.
> >>
> >> Start qemu with the following parameters.
> >>     qemu-system-x86_64 -enable-kvm -M q35 \
> >>     ...
> >>     -device pcie-root-port,slot=3,id=pcie_port.3 \
> >>     -netdev
> tap,id=net3,script=no,downscript=/tmp/rmtap,ifname=xcbr3_t2,queues=1 \
> >>     -device igb,bus=pcie_port.3,netdev=net3,mac=00:00:00:01:03:02
> >>
> >> Load IGB/IGBVF modules if needed.
> >> modprobe igb
> >> modprobe igbvf
> >>
> >> Create VFs via /sys
> >> ls /sys/bus/pci/devices/0000:01:00.0/
> >> echo 2 > /sys/bus/pci/devices/0000:01:00.0/sriov_numvfs
> >>
> >> Sriram Yagnaraman (5):
> >>    pcie: add helper function to get number of VFs
> >>    hw/net/net_tx_pkt: helper function to get l2 hdr
> >>    hw/net/igb: register definitions
> >>    hw/net/igb: emulated intel IGB (82576EB) network device
> >>    hw/net/igb: build support for igb/igbvf devices
> >
> > Haven't reviewed this series but I think we'd have two more things:
> >
> > 1) update the MAINTAINERS
> > 2) a qtest for igb (having a qtest for sr-iov would be even better)
> >
> > Thanks
> >
> >>
> >>   hw/i386/Kconfig             |    1 +
> >>   hw/net/Kconfig              |    5 +
> >>   hw/net/e1000_regs.h         |  357 +++-
> >>   hw/net/e1000x_common.c      |   13 +
> >>   hw/net/e1000x_common.h      |   28 +
> >>   hw/net/igb.c                |  627 ++++++
> >>   hw/net/igb.h                |  184 ++
> >>   hw/net/igb_core.c           | 3782 +++++++++++++++++++++++++++++++++++
> >>   hw/net/igb_core.h           |  216 ++
> >>   hw/net/igbvf.c              |  262 +++
> >>   hw/net/meson.build          |    2 +
> >>   hw/net/net_tx_pkt.c         |   17 +-
> >>   hw/net/net_tx_pkt.h         |    8 +
> >>   hw/net/trace-events         |  190 ++
> >>   hw/pci/pcie_sriov.c         |    6 +
> >>   include/hw/pci/pcie_sriov.h |    5 +
> >>   16 files changed, 5671 insertions(+), 32 deletions(-)
> >>   create mode 100644 hw/net/igb.c
> >>   create mode 100644 hw/net/igb.h
> >>   create mode 100644 hw/net/igb_core.c
> >>   create mode 100644 hw/net/igb_core.h
> >>   create mode 100644 hw/net/igbvf.c
> >>
> >> --
> >> 2.34.1
> >>
> >
> >
> 
> Hi Sriram and Jason,
> 
> Coincidentally we has also been working on igb emulation. Our tree is
> available at:
> https://github.com/daynix/qemu/tree/akihikodaki/igb_sriov_rebase
> 
> Briefly looking at Sriram's series, it seems to have better feature coverage, 
> but
> there are some changes present only in our tree as well.
> For instance, our tree includes a qtest though it does not test SR-IOV
> functionality.
> 
> I'd like to suggest you to review and merge Sriram's series first, and I can
> rebase our tree to extract changes present only in our tree, including qtest. 
> Or
> it can be the other way around; I think I can finish my series in weeks so I 
> may
> submit it once it is ready, and you can apply changes in Sriram's series on 
> top
> it. I'd like to hear which is more convenient for you, or another idea if you
> have.
> 
Great to know there are other people working on this at the same time. 
I just sent a v2 version with a qtest very similar to e1000e-test.

I see that your tree has good documentation and a very nice idea to translate
between PF and VF addresses. Would be good to get those changes as well.

> Regards,
> Akihiko Odaki