Unable to boot OpenBMC image in QEMU for AST2500

[Stephen Beckwith]

Mr. Jeffery,
   Thank you for your insights and comments.
1) RE: git clone - well, for whatever reason, this worked today. . .
(internet gremlins, github trolls, who knows).
2) RE: building:  I reviewed some documentation again, and pieced together
that I had built the wrong image (again).  So, as I now have a "fresh"
clone, I set about with some different steps towards building an
ast2500-evb image.
 -  I verified that the qemu-system-arm, listed in the documentation
provided by Mr. Geissler, is indeed a 4.1.0 version, and it does contain
the necessary machine support:
sbeckwit at ubuntu:/usr/bin$ qemu-system-arm -machine help | grep ast
ast2500-evb          Aspeed AST2500 EVB (ARM1176)
ast2600-evb          Aspeed AST2600 EVB (Cortex A7)
3) after 168 minutes, I the "baking" was done and checking the
deploy/images directory, I found the ast2500-evb files: kernel, u-boot,
initramfs, mtd image, etc.  Most Excellent!
4) I then, as you suggested, reviewed the documentation (again) and crafted
some minimal parameters to running QEMU.  And Behold:
Phosphor OpenBMC (Phosphor OpenBMC Project Reference Distro) 0.1.0
evb-ast2500 ttyS4

evb-ast2500 login: root
Password:
root at evb-ast2500:~# uname -a
Linux evb-ast2500 5.2.0-c01e98d #1 Wed Sep 25 16:54:08 UTC 2019 armv6l
GNU/Linux
 - I got in to this booted image under QEMU.   (not the blinking light we
Embedded FW Engineers like to see on our boards when something works, but
this will do just fine).
5) RE: runqemu script - My Python is rusty, but I'll take a gander at it,
since I will have some time here, as we are only working on an initial PoC
for this.  The next attempt will be to try and see if we can boot this on
our reference platform (AST2500 based).
 - I maybe getting back to you should I have an update on this script.

Thank you for your assistance and support.

Regards,
Stephen Beckwith



On Wed, Sep 25, 2019 at 11:52 AM Andrew Jeffery <andrew at aj.id.au> wrote:

>
>
> On Wed, 25 Sep 2019, at 22:52, Stephen Beckwith wrote:
> > 2) re: qemu machine - the qemu machine being used is the one that was
> > built as part fo the opembmc project. If this qemu machine won’t work,
> > then why is it built as part of the project?
>
> Because no-one has looked at removing it from the poky build
> configurations that openbmc inherits. Regardless, removing it isn't the
> right way to deal with the issue.
>
> > Why are there 13 copies of
> > this machine scattered through out the filesystem, consuming ~ 564MB of
> > space, if it’s the “wrong” machine?
>
> This is a question best asked of the upstream yocto project.
>
> >  - so where do I obtain the “BMC-specific machine” ?
>
> By running `qemu-system-arm -M ? | grep bmc`. It will give you output like:
>
> $ qemu-system-arm -M ? | grep bmc
> palmetto-bmc         OpenPOWER Palmetto BMC (ARM926EJ-S)
> rainier-bmc          Aspeed AST2600 EVB (Cortex A7)
> romulus-bmc          OpenPOWER Romulus BMC (ARM1176)
> swift-bmc            OpenPOWER Swift BMC (ARM1176)
> tacoma-bmc           Aspeed AST2600 EVB (Cortex A7)
> witherspoon-bmc      OpenPOWER Witherspoon BMC (ARM1176)
>
> You need to pick an appropriate one to run your image:
>
> ARM926EJ-S: AST2400
> ARM1176: AST2500
> Cortex A7: AST2600
>
> The remaining differences between listed machines are what peripherals
> are attached to e.g. the I2C buses.
>
> >
> > 3) if OpenBMC doesn’t properly integrate into the rumqemu scripts,
> > then: A) why is this script here?
>
> It is provided by the upstream yocto project.
>
> > B) why hasn’t anyone taken the
> > time/effort to correct the script so that it will integrate properly
> > with OpenBMC?
>
> Because no-one sees it as something worth spending time on? I've
> wanted to get it done since writing the initial QEMU support for the
> ASPEED chips, but it lives at the bottom of my todo list. I've asked
> other people asking questions like you to do it as well, but it hasn't
> been done yet by them either, so I can only assume it lives at the
> bottom of their todo lists too. I'd love it if you contributed a patch
> to fix it :)
>
> Andrew
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ozlabs.org/pipermail/openbmc/attachments/20190925/a5bae89e/attachment.htm>