qemu/hw/s390x/css-bridge.c

167 lines
4.8 KiB
C
Raw Normal View History

/*
* css bridge implementation
*
* Copyright 2012,2016 IBM Corp.
* Author(s): Cornelia Huck <cornelia.huck@de.ibm.com>
* Pierre Morel <pmorel@linux.vnet.ibm.com>
*
* This work is licensed under the terms of the GNU GPL, version 2 or (at
* your option) any later version. See the COPYING file in the top-level
* directory.
*/
#include "qemu/osdep.h"
#include "qapi/error.h"
#include "hw/hotplug.h"
#include "hw/sysbus.h"
#include "qemu/bitops.h"
#include "hw/s390x/css.h"
#include "ccw-device.h"
#include "hw/s390x/css-bridge.h"
#include "cpu.h"
/*
* Invoke device-specific unplug handler, disable the subchannel
* (including sending a channel report to the guest) and remove the
* device from the virtual css bus.
*/
static void ccw_device_unplug(HotplugHandler *hotplug_dev,
DeviceState *dev, Error **errp)
{
CcwDevice *ccw_dev = CCW_DEVICE(dev);
CCWDeviceClass *k = CCW_DEVICE_GET_CLASS(ccw_dev);
SubchDev *sch = ccw_dev->sch;
Error *err = NULL;
if (k->unplug) {
k->unplug(hotplug_dev, dev, &err);
if (err) {
error_propagate(errp, err);
return;
}
}
/*
* We should arrive here only for device_del, since we don't support
* direct hot(un)plug of channels.
*/
assert(sch != NULL);
/* Subchannel is now disabled and no longer valid. */
sch->curr_status.pmcw.flags &= ~(PMCW_FLAGS_MASK_ENA |
PMCW_FLAGS_MASK_DNV);
css_generate_sch_crws(sch->cssid, sch->ssid, sch->schid, 1, 0);
object_unparent(OBJECT(dev));
}
static void virtual_css_bus_reset(BusState *qbus)
{
/* This should actually be modelled via the generic css */
css_reset();
}
static char *virtual_css_bus_get_dev_path(DeviceState *dev)
{
CcwDevice *ccw_dev = CCW_DEVICE(dev);
SubchDev *sch = ccw_dev->sch;
VirtualCssBridge *bridge =
VIRTUAL_CSS_BRIDGE(qdev_get_parent_bus(dev)->parent);
/*
* We can't provide a dev path for backward compatibility on
* older machines, as it is visible in the migration stream.
*/
return bridge->css_dev_path ?
g_strdup_printf("/%02x.%1x.%04x", sch->cssid, sch->ssid, sch->devno) :
NULL;
}
static void virtual_css_bus_class_init(ObjectClass *klass, void *data)
{
BusClass *k = BUS_CLASS(klass);
k->reset = virtual_css_bus_reset;
k->get_dev_path = virtual_css_bus_get_dev_path;
}
static const TypeInfo virtual_css_bus_info = {
.name = TYPE_VIRTUAL_CSS_BUS,
.parent = TYPE_BUS,
.instance_size = sizeof(VirtualCssBus),
.class_init = virtual_css_bus_class_init,
};
VirtualCssBus *virtual_css_bus_init(void)
{
VirtualCssBus *cbus;
BusState *bus;
DeviceState *dev;
/* Create bridge device */
dev = qdev_create(NULL, TYPE_VIRTUAL_CSS_BRIDGE);
object_property_add_child(qdev_get_machine(), TYPE_VIRTUAL_CSS_BRIDGE,
OBJECT(dev), NULL);
qdev_init_nofail(dev);
/* Create bus on bridge device */
bus = qbus_create(TYPE_VIRTUAL_CSS_BUS, dev, "virtual-css");
cbus = VIRTUAL_CSS_BUS(bus);
cbus->squash_mcss = s390_get_squash_mcss();
/* Enable hotplugging */
qbus_set_hotplug_handler(bus, dev, &error_abort);
css_register_io_adapters(CSS_IO_ADAPTER_VIRTIO, true, false,
0, &error_abort);
return cbus;
}
/***************** Virtual-css Bus Bridge Device ********************/
static Property virtual_css_bridge_properties[] = {
DEFINE_PROP_BOOL("css_dev_path", VirtualCssBridge, css_dev_path,
true),
DEFINE_PROP_END_OF_LIST(),
};
s390x/css: unrestrict cssids The default css 0xfe is currently restricted to virtual subchannel devices. The hope when the decision was made was, that non-virtual subchannel devices will come around when guest can exploit multiple channel subsystems. Since the guests generally don't do, the pain of the partitioned (cssid) namespace outweighs the gain. Let us remove the corresponding restrictions (virtual devices can be put only in 0xfe and non-virtual devices in any css except the 0xfe -- while s390-squash-mcss then remaps everything to cssid 0). At the same time, change our schema for generating css bus ids to put both virtual and non-virtual devices into the default css (spilling over into other css images, if needed). The intention is to deprecate s390-squash-mcss. With this change devices without a specified devno won't end up hidden to guests not supporting multiple channel subsystems, unless this can not be avoided (default css full). Let us also advertise the changes to the management software (so it can tell are cssids unrestricted or restricted). The adverse effect of getting rid of the restriction on migration should not be too severe. Vfio-ccw devices are not live-migratable yet, and for virtual devices using the extra freedom would only make sense with the aforementioned guest support in place. The auto-generated bus ids are affected by both changes. We hope to not encounter any auto-generated bus ids in production as Libvirt is always explicit about the bus id. Since 8ed179c937 ("s390x/css: catch section mismatch on load", 2017-05-18) the worst that can happen because the same device ended up having a different bus id is a cleanly failed migration. I find it hard to reason about the impact of changed auto-generated bus ids on migration for command line users as I don't know which rules is such an user supposed to follow. Another pain-point is down- or upgrade of QEMU for command line users. The old way and the new way of doing vfio-ccw are mutually incompatible. Libvirt is only going to support the new way, so for libvirt users, the possible problems at QEMU downgrade are the following. If a domain contains virtual devices placed into a css different than 0xfe the domain will refuse to start with a QEMU not having this patch. Putting devices into a css different that 0xfe however won't make much sense in the near future (guest support). Libvirt will refuse to do vfio-ccw with a QEMU not having this patch. This is business as usual. Signed-off-by: Halil Pasic <pasic@linux.vnet.ibm.com> Acked-by: Christian Borntraeger <borntraeger@de.ibm.com> Reviewed-by: Dong Jia Shi <bjsdjshi@linux.vnet.ibm.com> Message-Id: <20171206144438.28908-2-pasic@linux.vnet.ibm.com> Signed-off-by: Cornelia Huck <cohuck@redhat.com>
2017-12-06 17:44:37 +03:00
static bool prop_get_true(Object *obj, Error **errp)
{
return true;
}
static void virtual_css_bridge_class_init(ObjectClass *klass, void *data)
{
HotplugHandlerClass *hc = HOTPLUG_HANDLER_CLASS(klass);
DeviceClass *dc = DEVICE_CLASS(klass);
hc->unplug = ccw_device_unplug;
set_bit(DEVICE_CATEGORY_BRIDGE, dc->categories);
dc->props = virtual_css_bridge_properties;
s390x/css: unrestrict cssids The default css 0xfe is currently restricted to virtual subchannel devices. The hope when the decision was made was, that non-virtual subchannel devices will come around when guest can exploit multiple channel subsystems. Since the guests generally don't do, the pain of the partitioned (cssid) namespace outweighs the gain. Let us remove the corresponding restrictions (virtual devices can be put only in 0xfe and non-virtual devices in any css except the 0xfe -- while s390-squash-mcss then remaps everything to cssid 0). At the same time, change our schema for generating css bus ids to put both virtual and non-virtual devices into the default css (spilling over into other css images, if needed). The intention is to deprecate s390-squash-mcss. With this change devices without a specified devno won't end up hidden to guests not supporting multiple channel subsystems, unless this can not be avoided (default css full). Let us also advertise the changes to the management software (so it can tell are cssids unrestricted or restricted). The adverse effect of getting rid of the restriction on migration should not be too severe. Vfio-ccw devices are not live-migratable yet, and for virtual devices using the extra freedom would only make sense with the aforementioned guest support in place. The auto-generated bus ids are affected by both changes. We hope to not encounter any auto-generated bus ids in production as Libvirt is always explicit about the bus id. Since 8ed179c937 ("s390x/css: catch section mismatch on load", 2017-05-18) the worst that can happen because the same device ended up having a different bus id is a cleanly failed migration. I find it hard to reason about the impact of changed auto-generated bus ids on migration for command line users as I don't know which rules is such an user supposed to follow. Another pain-point is down- or upgrade of QEMU for command line users. The old way and the new way of doing vfio-ccw are mutually incompatible. Libvirt is only going to support the new way, so for libvirt users, the possible problems at QEMU downgrade are the following. If a domain contains virtual devices placed into a css different than 0xfe the domain will refuse to start with a QEMU not having this patch. Putting devices into a css different that 0xfe however won't make much sense in the near future (guest support). Libvirt will refuse to do vfio-ccw with a QEMU not having this patch. This is business as usual. Signed-off-by: Halil Pasic <pasic@linux.vnet.ibm.com> Acked-by: Christian Borntraeger <borntraeger@de.ibm.com> Reviewed-by: Dong Jia Shi <bjsdjshi@linux.vnet.ibm.com> Message-Id: <20171206144438.28908-2-pasic@linux.vnet.ibm.com> Signed-off-by: Cornelia Huck <cohuck@redhat.com>
2017-12-06 17:44:37 +03:00
object_class_property_add_bool(klass, "cssid-unrestricted",
prop_get_true, NULL, NULL);
object_class_property_set_description(klass, "cssid-unrestricted",
"A css device can use any cssid, regardless whether virtual"
" or not (read only, always true)",
NULL);
}
static const TypeInfo virtual_css_bridge_info = {
.name = TYPE_VIRTUAL_CSS_BRIDGE,
.parent = TYPE_SYS_BUS_DEVICE,
.instance_size = sizeof(VirtualCssBridge),
.class_init = virtual_css_bridge_class_init,
.interfaces = (InterfaceInfo[]) {
{ TYPE_HOTPLUG_HANDLER },
{ }
}
};
static void virtual_css_register(void)
{
type_register_static(&virtual_css_bridge_info);
type_register_static(&virtual_css_bus_info);
}
type_init(virtual_css_register)