ast2600-evb qemu and usb-net
Joel Stanley
joel at jms.id.au
Tue Aug 10 12:31:00 AEST 2021
On Mon, 9 Aug 2021 at 04:56, Deepak Kodihalli
<deepak.kodihalli.83 at gmail.com> wrote:
>
> Hi All,
>
> I am trying to launch the ast2600-evb machine on QEMU by passing a a
> usb-net device:
> qemu-system-arm -machine ast2600-evb -drive file=<bmc
> image>,if=mtd,bus=0,unit=0,snapshot=on -nographic -nic
> user,hostname=qemu -device usb-net,netdev=net0 -netdev
> socket,id=net0,listen=:1234
>
> I downloaded the QEMU binary from
> https://github.com/openbmc/docs/blob/master/development/dev-environment.md#download-and-start-qemu-session.
This is unrelated to your question, but that qemu is built from an
outdated branch. I would recommend using mainline qemu instead.
We also have a qemu tree that Cedric maintains at
https://github.com/legoater/qemu. The aspeed-6.1 (and soon aspeed-6.2)
have changes that we have not yet merged to mainline.
> I get the following error when I run the command above:
> qemu-system-arm: -device usb-net,netdev=net0: Warning: speed mismatch
> trying to attach usb device "QEMU USB Network Interface" (full speed)
> to bus "usb-bus.1", port "1" (high speed)
As the error message says, you're attempting to attach a full speed
(USB 1.1) device to a high speed (USB 2.0) bus.
This is where I get a bit confused by USB. I thought that the
controller supported both full and high speed (that's what the data
sheet says).
The chip also has a USB 1.1 controller (UCHI) hasn't been enabled for
the ast2600. There's also no qemu model hooked up.
> The intent is to have two such QEMU instances talking over emulated
> usb-net (the other instance would do a connect on the netdev socket).
If you're hooking up two instances over network, is there any reason
you can't use the existing network devices? It would allow you to test
your userspace without requiring extra modelling.
>
> I'm guessing this is the lack of usb support/initialization in
> hw/arm/aspeed* in the QEMU tree. Has anyone else tried this with
> ast2600-evb?
>
> Thanks,
> Deepak
More information about the openbmc
mailing list