Re: [Qemu-devel] [qemu-s390x] [PATCH v4 1/3] s390x/sclp: proper support of larger send and receive masks

[Claudio Imbrenda <imbrenda@linux.vnet.ibm.com>]

On Tue, 6 Mar 2018 16:21:18 +0100
Cornelia Huck <cohuck@redhat.com> wrote:

> On Fri, 2 Mar 2018 10:09:16 +0100
> Christian Borntraeger <borntraeger@de.ibm.com> wrote:
> 
> > On 02/23/2018 06:42 PM, Claudio Imbrenda wrote:  
> > > Until 67915de9f0383ccf4a ("s390x/event-facility: variable-length
> > > event masks") we only supported sclp event masks with a size of
> > > exactly 4 bytes, even though the architecture allows the guests
> > > to set up sclp event masks from 1 to 1021 bytes in length.
> > > After that patch, the behaviour was almost compliant, but some
> > > issues were still remaining, in particular regarding the handling
> > > of selective reads and migration.
> > > 
> > > When setting the sclp event mask, a mask size is also specified.
> > > Until now we only considered the size in order to decide which
> > > bits to save in the internal state. On the other hand, when a
> > > guest performs a selective read, it sends a mask, but it does not
> > > specify a size; the implied size is the size of the last mask
> > > that has been set.
> > > 
> > > Specifying bits in the mask of selective read that are not
> > > available in the internal mask should return an error, and bits
> > > past the end of the mask should obviously be ignored. This can
> > > only be achieved by keeping track of the lenght of the mask.
> > > 
> > > The mask length is thus now part of the internal state that needs
> > > to be migrated.
> > > 
> > > This patch fixes the handling of selective reads, whose size will
> > > now match the length of the event mask, as per architecture.
> > > 
> > > While the default behaviour is to be compliant with the
> > > architecture, when using older machine models the old broken
> > > behaviour is selected (allowing only masks of size exactly 4), in
> > > order to be able to migrate toward older versions.
> > > 
> > > Fixes: 67915de9f0383ccf4a ("s390x/event-facility: variable-length
> > > event masks") Signed-off-by: Claudio Imbrenda
> > > <imbrenda@linux.vnet.ibm.com>    
> > 
> > Looks sane and seems to match the docs.
> > 
> > Reviewed-by: Christian Borntraeger <borntraeger@de.ibm.com>
> > 
> > One question below:
> >   
> > > ---
> > >  hw/s390x/event-facility.c  | 81
> > > ++++++++++++++++++++++++++++++++++++++--------
> > > hw/s390x/s390-virtio-ccw.c |  8 ++++- 2 files changed, 75
> > > insertions(+), 14 deletions(-)    
> > [...9  
> > >  static void init_event_facility(Object *obj)
> > >  {
> > >      SCLPEventFacility *event_facility = EVENT_FACILITY(obj);
> > >      DeviceState *sdev = DEVICE(obj);
> > >      Object *new;
> > > 
> > > +    event_facility->mask_length = 4;    
> > 
> > Shouldnt we start with 0 here (as no mask is set)
> > 
> > and  
> > > +    event_facility->allow_all_mask_sizes = true;
> > > +    object_property_add_bool(obj, "allow_all_mask_sizes",
> > > +                             sclp_event_get_allow_all_mask_sizes,
> > > +
> > > sclp_event_set_allow_all_mask_sizes, NULL); /* Spawn a new bus
> > > for SCLP events */ qbus_create_inplace(&event_facility->sbus,
> > > sizeof(event_facility->sbus), TYPE_SCLP_EVENTS_BUS, sdev, NULL);
> > > diff --git a/hw/s390x/s390-virtio-ccw.c
> > > b/hw/s390x/s390-virtio-ccw.c index 8b3053f..e9309fd 100644
> > > --- a/hw/s390x/s390-virtio-ccw.c
> > > +++ b/hw/s390x/s390-virtio-ccw.c
> > > @@ -29,6 +29,7 @@
> > >  #include "s390-pci-bus.h"
> > >  #include "hw/s390x/storage-keys.h"
> > >  #include "hw/s390x/storage-attributes.h"
> > > +#include "hw/s390x/event-facility.h"
> > >  #include "hw/compat.h"
> > >  #include "ipl.h"
> > >  #include "hw/s390x/s390-virtio-ccw.h"
> > > @@ -671,7 +672,12 @@ bool css_migration_enabled(void)
> > >      type_init(ccw_machine_register_##suffix)
> > > 
> > >  #define CCW_COMPAT_2_11 \
> > > -        HW_COMPAT_2_11
> > > +        HW_COMPAT_2_11 \
> > > +        {\
> > > +            .driver   = TYPE_SCLP_EVENT_FACILITY,\
> > > +            .property = "allow_all_mask_sizes",\
> > > +            .value    = "off",\
> > > +        },    
> > 
> > then set the mask_len here
> > 
> > or is this overkill? In the end it should not matter as the active
> > mask is 0 so the length does not matter.  
> 
> I think so (does not matter). I'll apply the patch as-is, unless
> someone has complaints.

the size is 4 to keep backwards compatibility. The guest will set a
mask very early anyway, and that will be the new size.
The only time this could be a problem is if the guest performs a
selective read, but in that case it's also not an issue, as all the
bits in the internal mask are 0, so the selective read will fail no
matter what.
Also selective reads are not supposed to be issued before the mask is
set anyway.

so yes, as you said, it does not matter

> > 
> >   
> > > 
> > >  #define CCW_COMPAT_2_10 \
> > >          HW_COMPAT_2_10
> > >     
> >   
>