Re: [Qemu-devel] [PULL 03/36] hw/pci: delay bus_master_enable_region initialization

[Marcel Apfelbaum]

On 07/11/2016 09:41 PM, Mark Cave-Ayland wrote:
> On 11/07/16 16:18, Marcel Apfelbaum wrote:
>
> > On 07/11/2016 05:42 PM, Leon Alrae wrote:
> > > Hi,
> > >
> > > This commit causes regressions in my MIPS tests. QEMU segfaults when
> > > booting Linux on Malta board;
> >
> > Hi Leon,
> >
> > Is a good thing you caught it in the pull request, thanks!
> > Is a pity we don't have a way to run sanity tests on all archs (beyond
> > make check)
> >
> >   this can be easily reproduced with
> > > Aurelien's Debian images:
> > >
> > > wget
> > > https://people.debian.org/~aurel32/qemu/mipsel/vmlinux-3.2.0-4-5kc-malta
> > > wget
> > > https://people.debian.org/~aurel32/qemu/mipsel/debian_wheezy_mipsel_standard.qcow2
> > >
> > > qemu-system-mips64el -M malta -kernel vmlinux-3.2.0-4-5kc-malta -hda
> > > debian_wheezy_mipsel_standard.qcow2 -append "root=/dev/sda1
> > > console=ttyS0" -nographic
> >
> > I'll reproduce and handle it, thanks!
> > Marcel
>
> FWIW this looks like the exact same backtrace I reported with
> qemu-system-sparc64 so hopefully the solution should fix both problems :)

Hi Mark and Leon,

I've reproduced both failures and I've found the root cause.
The PCI root bus is not realized as part of the machine initialization.

Until now the only consequence was the machine couldn't be migrated
(because the vmstate is initialized as part of 'realize').
However, from now on the devices cannot do DMA requests if the bus is not 
'realized'.

It took some time because I had to go over all the archs and check this issue.
I have the patches ready, I am running some more tests and post them soon.

Thanks for the patience,
Marcel

> ATB,
>
> Mark.