Re: [PATCH v4 3/3] memory_hotplug.c: send DEVICE_UNPLUG_ERROR in acpi_memory_hotplug_write()

[Daniel Henrique Barboza]

On 7/10/21 3:57 AM, Markus Armbruster wrote:
> Igor Mammedov <imammedo@redhat.com> writes:
>
> > On Fri, 09 Jul 2021 13:25:43 +0200
> > Markus Armbruster <armbru@redhat.com> wrote:
> >
> > > Igor Mammedov <imammedo@redhat.com> writes:
> > >
> > > > On Thu, 08 Jul 2021 15:08:57 +0200
> > > > Markus Armbruster <armbru@redhat.com> wrote:
> > > >   
> > > > > Daniel Henrique Barboza <danielhb413@gmail.com> writes:
> > > > >    
> > > > > > MEM_UNPLUG_ERROR is deprecated since the introduction of
> > > > > > DEVICE_UNPLUG_ERROR. Keep emitting both while the deprecation of
> > > > > > MEM_UNPLUG_ERROR is pending.
> > > > > >
> > > > > > CC: Michael S. Tsirkin <mst@redhat.com>
> > > > > > CC: Igor Mammedov <imammedo@redhat.com>
> > > > > > Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
> > > > > > Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
> > > > > > ---
> > > > > >   hw/acpi/memory_hotplug.c | 13 +++++++++++--
> > > > > >   1 file changed, 11 insertions(+), 2 deletions(-)
> > > > > >
> > > > > > diff --git a/hw/acpi/memory_hotplug.c b/hw/acpi/memory_hotplug.c
> > > > > > index af37889423..fb9f4d2de7 100644
> > > > > > --- a/hw/acpi/memory_hotplug.c
> > > > > > +++ b/hw/acpi/memory_hotplug.c
> > > > > > @@ -8,6 +8,7 @@
> > > > > >   #include "qapi/error.h"
> > > > > >   #include "qapi/qapi-events-acpi.h"
> > > > > >   #include "qapi/qapi-events-machine.h"
> > > > > > +#include "qapi/qapi-events-qdev.h"
> > > > > >   
> > > > > >   #define MEMORY_SLOTS_NUMBER          "MDNR"
> > > > > >   #define MEMORY_HOTPLUG_IO_REGION     "HPMR"
> > > > > > @@ -177,9 +178,17 @@ static void acpi_memory_hotplug_write(void *opaque, hwaddr 
> > > > > > addr, uint64_t data,
> > > > > >               /* call pc-dimm unplug cb */
> > > > > >               hotplug_handler_unplug(hotplug_ctrl, dev, &local_err);
> > > > > >               if (local_err) {
> > > > > > +                const char *error_pretty = error_get_pretty(local_err);
> > > > > > +
> > > > > >                   trace_mhp_acpi_pc_dimm_delete_failed(mem_st->selector);
> > > > > > -                qapi_event_send_mem_unplug_error(dev->id,
> > > > > > -                                                 error_get_pretty(local_err));
> > > > > > +
> > > > > > +                /*
> > > > > > +                 * Send both MEM_UNPLUG_ERROR and DEVICE_UNPLUG_ERROR
> > > > > > +                 * while the deprecation of MEM_UNPLUG_ERROR is
> > > > > > +                 * pending.
> > > > > > +                 */
> > > > > > +                qapi_event_send_mem_unplug_error(dev->id, error_pretty);
> > > > > > +                qapi_event_send_device_unplug_error(dev->id, error_pretty);
> > > > > >                   error_free(local_err);
> > > > > >                   break;
> > > > > >               }
> > > > >
> > > > > Same question as for PATCH 2: can dev->id be null?
> > > > only theoretically (if memory device were created directly without
> > > > using device_add), which as far as I know is not the case as all
> > > > memory devices are created using -device/device_add so far.
> > > >
> > > > ( for device_add case see qdev_device_add->qdev_set_id where
> > > >    'id' is set to user provided or to generated "device[%d]" value)
> > >
> > > Something is set to a generated value, but it's not dev->id :)
> > >
> > >      void qdev_set_id(DeviceState *dev, const char *id)
> > >
> > > @id is the value of id=...  It may be null.
> > >
> > > dev->id still is null here.
> > >
> > >      {
> > >          if (id) {
> > >              dev->id = id;
> > >          }
> > >
> > > dev->id is now the value of id=...  It may be null.
> > >
> > >          if (dev->id) {
> > >              object_property_add_child(qdev_get_peripheral(), dev->id,
> > >                                        OBJECT(dev));
> > >
> > > If the user specified id=..., add @dev as child of /peripheral.  The
> > > child's name is the (non-null) value of id=...
> > >
> > >          } else {
> > >              static int anon_count;
> > >              gchar *name = g_strdup_printf("device[%d]", anon_count++);
> > >              object_property_add_child(qdev_get_peripheral_anon(), name,
> > >                                        OBJECT(dev));
> > >              g_free(name);
> > >
> > > Else, add @dev as child of /peripheral-anon.  The child's name is made
> > > up.
> > >
> > >
> > >          }
> > >      }
> > >
> > > dev->id is still the value of id=..., i.e. it may be null.
> > yep, I was wrong and confused it child name in QOM tree.
> >
> > > Sure dereferencing dev->id in acpi_memory_hotplug_write() is safe?
> >
> > it aren't safe since guest may trigger this error when
> > memory-device is created without id.
>
> Thanks!
>
> Daniel, the issue predates your series, but your series adds instances.
> We need a patch fixing the existing instances before your series, and
> fix up your series.  Can you take care of that?


Sure. I'll add a patch to handle the dev->id == NULL case before calling
qapi_event_send_mem_unplug_error() in acpi_memory_hotplug_write().



Daniel

>