Re: [PATCH for-6.0 4/9] spapr: Set compat mode in spapr_reset_vcpu()

[Greg Kurz]

On Fri, 27 Nov 2020 15:59:16 +1100
David Gibson <david@gibson.dropbear.id.au> wrote:

[...]
> > 
> > This can't happen AFAICT.
> > 
> > (qemu) qom-get /machine/unattached/device[1] core-id
> > 0
> > (qemu) device_del /machine/unattached/device[1]
> > Error: Boot CPU core may not be unplugged
> > 
> > commit 62be8b044adf47327ebefdefb25f28a40316ebd0
> > Author: Bharata B Rao <bharata@linux.vnet.ibm.com>
> > Date:   Wed Jul 27 10:44:42 2016 +0530
> > 
> >     spapr: Prevent boot CPU core removal
> 
> Oh yeah, I'd forgotten we did this.
> 

Anyway, both you and Igor noted that this change is fragile, which
is true. So maybe we should just go on with the current behavior.

As mentioned in the changelog, one of the motivation to do this was
to get rid of the error path in spapr_core_plug() like other patches
in this series do for the other plug handlers. And I guess I should
do it like these other patches do : come up with a check in
spapr_core_pre_plug() that guarantees that ppc_set_compat() shouldn't
fail in spapr_core_plug() and can be passed &error_abort.

So I'll try to do just that. You can forget this patch.

pgpNxomoHot0A.pgp
Description: OpenPGP digital signature