Re: [Qemu-ppc] [Qemu-devel] [PATCH v3 0/2] 40p: fix PCI interrupt routing

[Mark Cave-Ayland]

On 17/09/2018 04:54, David Gibson wrote:

> Mark,
> 
> I think we have all the necessary acks to go ahead with this.
> However, I'm afraid I've lost track of the various prereq patches that
> were necessary here.  Can you resend with all the necessary pieces
> rebased against ppc-for-3.1 and the appropriate acked-bys included?

Apologies for the delay on this one, but once I rebased onto your ppc
for-3.1 branch I noticed a regression whereby the 40p sandalfoot image
would fail to boot to userspace - it would hang towards the end of the
boot with:

Bad inittab entry:
Bad inittab entry:
No more tasks for init -- sleeping forever

In the end I bisected it down to this patch:

commit d12a22c5c7ccb6cb7e8438b7e6e1011c6b712ae2
Author: Roman Kapl <address@hidden>
Date:   Tue Sep 11 13:34:51 2018 +0200

    target/ppc: add external PID support

Playing with this a bit more, it appears that the bug is only triggered
if I configure QEMU with --enable-debug which is my development default.

Roman, can you reproduce this locally at all? My setup is nothing
special, just Debian Stretch on amd64.


ATB,

Mark.