Re: [PATCH 4/4] s390x: Beautify machine reset

[David Hildenbrand]

On 22.11.19 13:22, Janosch Frank wrote:
> On 11/22/19 1:10 PM, Cornelia Huck wrote:
> > On Fri, 22 Nov 2019 12:47:44 +0100
> > David Hildenbrand <address@hidden> wrote:
> >
> > > On 22.11.19 12:46, Janosch Frank wrote:
> > > > On 11/22/19 11:59 AM, David Hildenbrand wrote:
> > > > > On 22.11.19 08:52, Janosch Frank wrote:
> > > > > > * Add comments that tell you which diag308 subcode caused the reset
> > > > > > * Sort by diag308 reset subcode
> > > > > >
> > > > > > Signed-off-by: Janosch Frank <address@hidden>
> > > > > > ---
> > > > > >     hw/s390x/s390-virtio-ccw.c | 20 ++++++++++----------
> > > > > >     1 file changed, 10 insertions(+), 10 deletions(-)
> > > > > >
> > > > > > diff --git a/hw/s390x/s390-virtio-ccw.c b/hw/s390x/s390-virtio-ccw.c
> > > > > > index c1d1440272..88f7758721 100644
> > > > > > --- a/hw/s390x/s390-virtio-ccw.c
> > > > > > +++ b/hw/s390x/s390-virtio-ccw.c
> > > > > > @@ -330,15 +330,7 @@ static void s390_machine_reset(MachineState *machine)
> > > > > >         s390_cmma_reset();
> > > > > >     
> > > > > >         switch (reset_type) {
> > > > > > -    case S390_RESET_EXTERNAL:
> > > > > > -    case S390_RESET_REIPL:
> > > > > > -        qemu_devices_reset();
> > > > > > -        s390_crypto_reset();
> > > > > > -
> > > > > > -        /* configure and start the ipl CPU only */
> > > > > > -        run_on_cpu(cs, s390_do_cpu_ipl, RUN_ON_CPU_NULL);
> > > > > > -        break;
> > > > > > -    case S390_RESET_MODIFIED_CLEAR:
> > > > > > +    case S390_RESET_MODIFIED_CLEAR: /* Subcode 0 */
> > > > >
> > > > > IMHO "Subcode X" isn't of much help here. We're out of diag handling.
> > > > >
> > > > > I'd suggest to just document the subcodes along with the definitions, if
> > > > > really needed, and drop this patch, at least I don't quite see the value
> > > > > of moving code around here... or is the code shuffling of any value on
> > > > > your prot virt patches?
> > > > >   
> > > >
> > > > It keeps me from consulting the POP every time I need to change things
> > > > in the machine resets. This is basically a 1:1 mapping of diag 308
> > > > subcodes to machine resets, so why don't we want to make that obvious
> > > > and order them by the subcodes?
> > > >    
> > >
> > > Because it is not a 1:1 mapping: S390_RESET_EXTERNAL
> >
> > Tack the explanation onto the definitions of S390_RESET_, then?
> > Probably still quicker than consulting the POP :)
>
> Does it really bother you that much, that I add some explanations to the
> things we're doing. The external reset also gets a comment so Conni
> won't need that much coffee anymore to understand the code :-)

I'm really sorry, but I fail to see how "Subcode 0" is *any* better than 
S390_RESET_MODIFIED_CLEAR (and avoids consulting the PoP) and why the 
order should matter at all here to make it easier to understand.

I don't NACK this, I just find it *completely* useless :)

--

Thanks,

David / dhildenb