Re: [PATCH 5/6] spapr: Don't use CPU_FOREACH() in 'info pic'

2019-10-27 Thread David Gibson
On Thu, Oct 24, 2019 at 11:28:45AM +0200, Greg Kurz wrote: > On Thu, 24 Oct 2019 14:02:31 +1100 > David Gibson wrote: > > > On Wed, Oct 23, 2019 at 04:52:21PM +0200, Greg Kurz wrote: > > > Now that presenter objects are parented to the interrupt controller, stop > > > relying on CPU_FOREACH()

Re: [PATCH 5/6] spapr: Don't use CPU_FOREACH() in 'info pic'

2019-10-24 Thread Greg Kurz
On Thu, 24 Oct 2019 14:02:31 +1100 David Gibson wrote: > On Wed, Oct 23, 2019 at 04:52:21PM +0200, Greg Kurz wrote: > > Now that presenter objects are parented to the interrupt controller, stop > > relying on CPU_FOREACH() which can race with CPU hotplug and crash QEMU. > > > > Signed-off-by:

Re: [PATCH 5/6] spapr: Don't use CPU_FOREACH() in 'info pic'

2019-10-23 Thread David Gibson
On Wed, Oct 23, 2019 at 04:52:21PM +0200, Greg Kurz wrote: > Now that presenter objects are parented to the interrupt controller, stop > relying on CPU_FOREACH() which can race with CPU hotplug and crash QEMU. > > Signed-off-by: Greg Kurz So.. we might be able to go further than this. Having

[PATCH 5/6] spapr: Don't use CPU_FOREACH() in 'info pic'

2019-10-23 Thread Greg Kurz
Now that presenter objects are parented to the interrupt controller, stop relying on CPU_FOREACH() which can race with CPU hotplug and crash QEMU. Signed-off-by: Greg Kurz --- hw/intc/spapr_xive.c |8 +--- hw/intc/xics.c| 12 hw/intc/xics_spapr.c |8 +---