Re: [Qemu-devel] [Qemu-ppc] Determining interest in PPC e500spin, yield, and openpic patches

[David Gibson]

On Mon, Jun 13, 2016 at 06:08:30PM -0500, address@hidden wrote:
> We've used older versions of QEMU for several years as a virtual
> target for our OS.  Many thanks to the community for providing this
> platform.

Hi,

Thanks for your interest.

> We've been working to get our OS running under QEMU 2.x and have
> identified a few bugs in QEMU, have made some enhancements, and are
> still tracking down some other curious behaviors.  I'm looking for
> some guidance as to how, and whether, you'd like patches for the
> following.

Always interested in bug fixes.  The e500 support probably doesn't get
a whole lot of attention these days, but you're proof that there are
at least a few people who care.

> 1. There is a defect in ppce500_spin.c:spin_kick() that creates an
>    incorrectly sized TLB entry.  This was reported as bug
>    https://bugs.launchpad.net/qemu/+bug/1587535  I can provide a
>    patch if desired.

Absolutely.

> 2. We have implemented the PPC "yield" instruction.  I can provide a
>    patch if desired.

Sounds good.

> 3. We're working on support for openpic timers.  We're not finished,
>    but it would be helpful to know if a patch is desired or if we
>    should expect to maintain the changes independently.

I don't see a reason we wouldn't want it, unless it's horribly
invasive.  By all means post patches, and I'll review as best I can.

> 4. We're currently tracking down why in our e500 (both unicore and
>    multi-core) PPC QEMU 2.5 guest (x86 host), that with interrupts
>    disabled, after enabling the decrementer and issuing a "wait"
>    instruction QEMU continues to "busy loop", consuming an entire host
>    CPU doing apparently nothing.  As expected, issuing a "wait" prior
>    to enabling the decrementer leaves the host process idle.  We found
>    the bug in the PPC "wait" instruction implementation that was
>    independently reported and resolved last week, but that did not fix
>    the problem.  We also have our OS running on the g3beige and using
>    MSR.POW which causes the host to "sleep", but we are having no joy
>    with e500 and "wait".  Any pointers would be appreciated.  When we
>    find something we'll report back.

Ok.


So, patches for ppc related code should be sent to myself and Alex
Graf <address@hidden>, co-maintainers for the target-ppc and related
code.  You should CC both address@hidden and
address@hidden

I have a (frequently rebased) git tree where I stage ppc patches at:
    https://github.com/dgibson/qemu/tree/ppc-for-2.7

-- 
David Gibson                    | I'll have my music baroque, and my code
david AT gibson.dropbear.id.au  | minimalist, thank you.  NOT _the_ _other_
                                | _way_ _around_!
http://www.ozlabs.org/~dgibson

signature.asc
Description: PGP signature