Let's add our virtio-mem-ccw proxy device and wire it up. We should
be supporting everything (e.g., device unplug, "dynamic-memslots") that
we already support for the virtio-pci variant.
With a Linux guest that supports virtio-mem (and has automatic memory
onlining properly configured, the following example will work:
1. Start a VM with 4G initial memory and a virtio-mem device with a maximum
capacity of 16GB:
qemu/build/qemu-system-s390x \
--enable-kvm \
-m 4G,maxmem=20G \
-nographic \
-smp 8 \
-hda Fedora-Server-KVM-40-1.14.s390x.qcow2 \
-chardev socket,id=monitor,path=/var/tmp/monitor,server,nowait \
-mon chardev=monitor,mode=readline \
-object memory-backend-ram,id=mem0,size=16G,reserve=off \
-device virtio-mem-ccw,id=vmem0,memdev=mem0,dynamic-memslots=on
2. Query the current size of virtio-mem device:
(qemu) info memory-devices
Memory device [virtio-mem]: "vmem0"
memaddr: 0x100000000
node: 0
requested-size: 0
size: 0
max-size: 17179869184
block-size: 1048576
memdev: /objects/mem0
3. Request to grow it to 8GB (hotplug 8GB):
(qemu) qom-set vmem0 requested-size 8G
(qemu) info memory-devices
Memory device [virtio-mem]: "vmem0"
memaddr: 0x100000000
node: 0
requested-size: 8589934592
size: 8589934592
max-size: 17179869184
block-size: 1048576
memdev: /objects/mem0
4. Request to grow to 16GB (hotplug another 8GB):
(qemu) qom-set vmem0 requested-size 16G
(qemu) info memory-devices
Memory device [virtio-mem]: "vmem0"
memaddr: 0x100000000
node: 0
requested-size: 17179869184
size: 17179869184
max-size: 17179869184
block-size: 1048576
memdev: /objects/mem0
5. Try to hotunplug all memory again, shrinking to 0GB:
(qemu) qom-set vmem0 requested-size 0G
(qemu) info memory-devices
Memory device [virtio-mem]: "vmem0"
memaddr: 0x100000000
node: 0
requested-size: 0
size: 0
max-size: 17179869184
block-size: 1048576
memdev: /objects/mem0
6. If it worked, unplug the device
(qemu) device_del vmem0
(qemu) info memory-devices
(qemu) object_del mem0
7. Hotplug a new device with a smaller capacity and directly size it to 1GB
(qemu) object_add memory-backend-ram,id=mem0,size=8G,reserve=off
(qemu) device_add virtio-mem-ccw,id=vmem0,memdev=mem0,\
dynamic-memslots=on,requested-size=1G
(qemu) info memory-devices
Memory device [virtio-mem]: "vmem0"
memaddr: 0x100000000
node: 0
requested-size: 1073741824
size: 1073741824
max-size: 8589934592
block-size: 1048576
memdev: /objects/mem0
Trying to use a virtio-mem device backed by hugetlb into a !hugetlb VM
correctly results in the error:
... Memory device uses a bigger page size than initial memory
Note that the virtio-mem driver in Linux will supports 1 MiB (pageblock)
granularity.
Note that we won't wire up virtio-mem-pci (should currently be
impossible due to lack of support for MSI-X), but we'll add a safety net
to reject plugging them.
Signed-off-by: David Hildenbrand <da...@redhat.com>
---
MAINTAINERS | 2 +
hw/s390x/Kconfig | 1 +
hw/s390x/meson.build | 1 +
hw/s390x/s390-virtio-ccw.c | 45 +++++++-
hw/s390x/virtio-ccw-mem.c | 226 +++++++++++++++++++++++++++++++++++++
hw/s390x/virtio-ccw-mem.h | 34 ++++++
hw/virtio/virtio-mem.c | 4 +-
7 files changed, 311 insertions(+), 2 deletions(-)
create mode 100644 hw/s390x/virtio-ccw-mem.c
create mode 100644 hw/s390x/virtio-ccw-mem.h
diff --git a/MAINTAINERS b/MAINTAINERS
index 53ed2c5f0f..f8e0b6c8e3 100644
--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -2413,6 +2413,8 @@ W: https://virtio-mem.gitlab.io/
F: hw/virtio/virtio-mem.c
F: hw/virtio/virtio-mem-pci.h
F: hw/virtio/virtio-mem-pci.c
+F: hw/s390x/virtio-ccw-mem.c
+F: hw/s390x/virtio-ccw-mem.h
F: include/hw/virtio/virtio-mem.h
virtio-snd
diff --git a/hw/s390x/Kconfig b/hw/s390x/Kconfig
index 3bbf4ae56e..5d57daff77 100644
--- a/hw/s390x/Kconfig
+++ b/hw/s390x/Kconfig
@@ -15,3 +15,4 @@ config S390_CCW_VIRTIO
select SCLPCONSOLE
select VIRTIO_CCW
select MSI_NONBROKEN
+ select VIRTIO_MEM_SUPPORTED
diff --git a/hw/s390x/meson.build b/hw/s390x/meson.build
index 4df40da855..a8434e7918 100644
--- a/hw/s390x/meson.build
+++ b/hw/s390x/meson.build
@@ -49,6 +49,7 @@ virtio_ss.add(when: 'CONFIG_VHOST_SCSI', if_true:
files('vhost-scsi-ccw.c'))
virtio_ss.add(when: 'CONFIG_VHOST_VSOCK', if_true: files('vhost-vsock-ccw.c'))
virtio_ss.add(when: 'CONFIG_VHOST_USER_FS', if_true:
files('vhost-user-fs-ccw.c'))
virtio_ss.add(when: 'CONFIG_VIRTIO_MD', if_true: files('virtio-ccw-md.c'))
+virtio_ss.add(when: 'CONFIG_VIRTIO_MEM', if_true: files('virtio-ccw-mem.c'))
s390x_ss.add_all(when: 'CONFIG_VIRTIO_CCW', if_true: virtio_ss)
hw_arch += {'s390x': s390x_ss}
diff --git a/hw/s390x/s390-virtio-ccw.c b/hw/s390x/s390-virtio-ccw.c
index 0a9d25620d..8d4a23db7d 100644
--- a/hw/s390x/s390-virtio-ccw.c
+++ b/hw/s390x/s390-virtio-ccw.c
@@ -45,6 +45,8 @@
#include "migration/blocker.h"
#include "qapi/visitor.h"
#include "hw/s390x/cpu-topology.h"
+#include "hw/virtio/virtio-md-pci.h"
+#include "hw/s390x/virtio-ccw-md.h"
#include CONFIG_DEVICES
static Error *pv_mig_blocker;
@@ -529,11 +531,37 @@ static void s390_machine_reset(MachineState *machine,
ShutdownCause reason)
s390_ipl_clear_reset_request();
}
+static void s390_machine_device_pre_plug(HotplugHandler *hotplug_dev,
+ DeviceState *dev, Error **errp)
+{
+ if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MD_CCW)) {
+ virtio_ccw_md_pre_plug(VIRTIO_MD_CCW(dev), MACHINE(hotplug_dev), errp);
+ } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MD_PCI)) {
+ error_setg(errp,
+ "PCI-attached virtio based memory device is not supported");
+ }
+}
+
static void s390_machine_device_plug(HotplugHandler *hotplug_dev,
DeviceState *dev, Error **errp)
{
if (object_dynamic_cast(OBJECT(dev), TYPE_CPU)) {
s390_cpu_plug(hotplug_dev, dev, errp);
+ } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MD_CCW)) {
+ /*
+ * At this point, the device is realized and set all memdevs mapped, so
+ * qemu_maxrampagesize() will pick up the page sizes of these memdevs
+ * as well. Before we plug the device and expose any RAM memory regions
+ * to the system, make sure we don't exceed the previously set max page
+ * size. While only relevant for KVM, there is not really any use case
+ * for this with TCG, so we'll unconditionally reject it.
+ */
+ if (qemu_maxrampagesize() != s390_get_max_pagesize()) {
+ error_setg(errp, "Memory device uses a bigger page size than"
+ " initial memory");
+ return;
+ }
+ virtio_ccw_md_plug(VIRTIO_MD_CCW(dev), MACHINE(hotplug_dev), errp);
}
}
@@ -543,6 +571,17 @@ static void
s390_machine_device_unplug_request(HotplugHandler *hotplug_dev,
if (object_dynamic_cast(OBJECT(dev), TYPE_CPU)) {
error_setg(errp, "CPU hot unplug not supported on this machine");
return;
+ } else if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MD_CCW)) {
+ virtio_ccw_md_unplug_request(VIRTIO_MD_CCW(dev), MACHINE(hotplug_dev),
+ errp);
+ }
+}
+
+static void s390_machine_device_unplug(HotplugHandler *hotplug_dev,
+ DeviceState *dev, Error **errp)
+{
+ if (object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MD_CCW)) {
+ virtio_ccw_md_unplug(VIRTIO_MD_CCW(dev), MACHINE(hotplug_dev), errp);
}
}
@@ -592,7 +631,9 @@ static const CPUArchIdList
*s390_possible_cpu_arch_ids(MachineState *ms)
static HotplugHandler *s390_get_hotplug_handler(MachineState *machine,
DeviceState *dev)
{
- if (object_dynamic_cast(OBJECT(dev), TYPE_CPU)) {
+ if (object_dynamic_cast(OBJECT(dev), TYPE_CPU) ||
+ object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MD_CCW) ||
+ object_dynamic_cast(OBJECT(dev), TYPE_VIRTIO_MD_PCI)) {
return HOTPLUG_HANDLER(machine);
}
return NULL;
@@ -768,8 +809,10 @@ static void ccw_machine_class_init(ObjectClass *oc, void
*data)
mc->possible_cpu_arch_ids = s390_possible_cpu_arch_ids;
/* it is overridden with 'host' cpu *in kvm_arch_init* */
mc->default_cpu_type = S390_CPU_TYPE_NAME("qemu");
+ hc->pre_plug = s390_machine_device_pre_plug;
hc->plug = s390_machine_device_plug;
hc->unplug_request = s390_machine_device_unplug_request;
+ hc->unplug = s390_machine_device_unplug;
nc->nmi_monitor_handler = s390_nmi;
mc->default_ram_id = "s390.ram";
mc->default_nic = "virtio-net-ccw";
diff --git a/hw/s390x/virtio-ccw-mem.c b/hw/s390x/virtio-ccw-mem.c
new file mode 100644
index 0000000000..bee0d560cb
--- /dev/null
+++ b/hw/s390x/virtio-ccw-mem.c
@@ -0,0 +1,226 @@
+/*
+ * virtio-mem CCW implementation
+ *
+ * Copyright (C) 2024 Red Hat, Inc.
+ *
+ * Authors:
+ * David Hildenbrand <da...@redhat.com>
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2.
+ * See the COPYING file in the top-level directory.
+ */
+
+#include "qemu/osdep.h"
+#include "hw/qdev-properties.h"
+#include "qapi/error.h"
+#include "qemu/module.h"
+#include "virtio-ccw-mem.h"
+#include "hw/mem/memory-device.h"
+#include "qapi/qapi-events-machine.h"
+#include "qapi/qapi-events-misc.h"
+
+static void virtio_ccw_mem_realize(VirtioCcwDevice *ccw_dev, Error **errp)
+{
+ VirtIOMEMCcw *dev = VIRTIO_MEM_CCW(ccw_dev);
+ DeviceState *vdev = DEVICE(&dev->vdev);
+
+ qdev_realize(vdev, BUS(&ccw_dev->bus), errp);
+}
+
+static void virtio_ccw_mem_set_addr(MemoryDeviceState *md, uint64_t addr,
+ Error **errp)
+{
+ object_property_set_uint(OBJECT(md), VIRTIO_MEM_ADDR_PROP, addr, errp);
+}
+
+static uint64_t virtio_ccw_mem_get_addr(const MemoryDeviceState *md)
+{
+ return object_property_get_uint(OBJECT(md), VIRTIO_MEM_ADDR_PROP,
+ &error_abort);
+}
+
+static MemoryRegion *virtio_ccw_mem_get_memory_region(MemoryDeviceState *md,
+ Error **errp)
+{
+ VirtIOMEMCcw *dev = VIRTIO_MEM_CCW(md);
+ VirtIOMEM *vmem = &dev->vdev;
+ VirtIOMEMClass *vmc = VIRTIO_MEM_GET_CLASS(vmem);
+
+ return vmc->get_memory_region(vmem, errp);
+}
+
+static void virtio_ccw_mem_decide_memslots(MemoryDeviceState *md,
+ unsigned int limit)
+{
+ VirtIOMEMCcw *dev = VIRTIO_MEM_CCW(md);
+ VirtIOMEM *vmem = VIRTIO_MEM(&dev->vdev);
+ VirtIOMEMClass *vmc = VIRTIO_MEM_GET_CLASS(vmem);
+
+ vmc->decide_memslots(vmem, limit);
+}
+
+static unsigned int virtio_ccw_mem_get_memslots(MemoryDeviceState *md)
+{
+ VirtIOMEMCcw *dev = VIRTIO_MEM_CCW(md);
+ VirtIOMEM *vmem = VIRTIO_MEM(&dev->vdev);
+ VirtIOMEMClass *vmc = VIRTIO_MEM_GET_CLASS(vmem);
+
+ return vmc->get_memslots(vmem);
+}
+
+static uint64_t virtio_ccw_mem_get_plugged_size(const MemoryDeviceState *md,
+ Error **errp)
+{
+ return object_property_get_uint(OBJECT(md), VIRTIO_MEM_SIZE_PROP,
+ errp);
+}
+
+static void virtio_ccw_mem_fill_device_info(const MemoryDeviceState *md,
+ MemoryDeviceInfo *info)
+{
+ VirtioMEMDeviceInfo *vi = g_new0(VirtioMEMDeviceInfo, 1);
+ VirtIOMEMCcw *dev = VIRTIO_MEM_CCW(md);
+ VirtIOMEM *vmem = &dev->vdev;
+ VirtIOMEMClass *vpc = VIRTIO_MEM_GET_CLASS(vmem);
+ DeviceState *vdev = DEVICE(md);
+
+ if (vdev->id) {
+ vi->id = g_strdup(vdev->id);
+ }
+
+ /* let the real device handle everything else */
+ vpc->fill_device_info(vmem, vi);
+
+ info->u.virtio_mem.data = vi;
+ info->type = MEMORY_DEVICE_INFO_KIND_VIRTIO_MEM;
+}
+
+static uint64_t virtio_ccw_mem_get_min_alignment(const MemoryDeviceState *md)
+{
+ return object_property_get_uint(OBJECT(md), VIRTIO_MEM_BLOCK_SIZE_PROP,
+ &error_abort);
+}
+
+static void virtio_ccw_mem_size_change_notify(Notifier *notifier, void *data)
+{
+ VirtIOMEMCcw *dev = container_of(notifier, VirtIOMEMCcw,
+ size_change_notifier);
+ DeviceState *vdev = DEVICE(dev);
+ char *qom_path = object_get_canonical_path(OBJECT(dev));
+ const uint64_t * const size_p = data;
+
+ qapi_event_send_memory_device_size_change(vdev->id, *size_p, qom_path);
+ g_free(qom_path);
+}
+
+static void virtio_ccw_mem_unplug_request_check(VirtIOMDCcw *vmd, Error **errp)
+{
+ VirtIOMEMCcw *dev = VIRTIO_MEM_CCW(vmd);
+ VirtIOMEM *vmem = &dev->vdev;
+ VirtIOMEMClass *vpc = VIRTIO_MEM_GET_CLASS(vmem);
+
+ vpc->unplug_request_check(vmem, errp);
+}
+
+static void virtio_ccw_mem_get_requested_size(Object *obj, Visitor *v,
+ const char *name, void *opaque,
+ Error **errp)
+{
+ VirtIOMEMCcw *dev = VIRTIO_MEM_CCW(obj);
+
+ object_property_get(OBJECT(&dev->vdev), name, v, errp);
+}
+
+static void virtio_ccw_mem_set_requested_size(Object *obj, Visitor *v,
+ const char *name, void *opaque,
+ Error **errp)
+{
+ VirtIOMEMCcw *dev = VIRTIO_MEM_CCW(obj);
+ DeviceState *vdev = DEVICE(obj);
+
+ /*
+ * If we passed virtio_ccw_mem_unplug_request_check(), making sure that
+ * the requested size is 0, don't allow modifying the requested size
+ * anymore, otherwise the VM might end up hotplugging memory before
+ * handling the unplug request.
+ */
+ if (vdev->pending_deleted_event) {
+ error_setg(errp, "'%s' cannot be changed if the device is in the"
+ " process of unplug", name);
+ return;
+ }
+
+ object_property_set(OBJECT(&dev->vdev), name, v, errp);
+}
+
+static Property virtio_ccw_mem_properties[] = {
+ DEFINE_PROP_BIT("ioeventfd", VirtioCcwDevice, flags,
+ VIRTIO_CCW_FLAG_USE_IOEVENTFD_BIT, true),
+ DEFINE_PROP_UINT32("max_revision", VirtioCcwDevice, max_rev,
+ VIRTIO_CCW_MAX_REV),
+ DEFINE_PROP_END_OF_LIST(),
+};
+
+static void virtio_ccw_mem_class_init(ObjectClass *klass, void *data)
+{
+ DeviceClass *dc = DEVICE_CLASS(klass);
+ VirtIOCCWDeviceClass *k = VIRTIO_CCW_DEVICE_CLASS(klass);
+ MemoryDeviceClass *mdc = MEMORY_DEVICE_CLASS(klass);
+ VirtIOMDCcwClass *vmdc = VIRTIO_MD_CCW_CLASS(klass);
+
+ k->realize = virtio_ccw_mem_realize;
+ set_bit(DEVICE_CATEGORY_MISC, dc->categories);
+ device_class_set_props(dc, virtio_ccw_mem_properties);
+
+ mdc->get_addr = virtio_ccw_mem_get_addr;
+ mdc->set_addr = virtio_ccw_mem_set_addr;
+ mdc->get_plugged_size = virtio_ccw_mem_get_plugged_size;
+ mdc->get_memory_region = virtio_ccw_mem_get_memory_region;
+ mdc->decide_memslots = virtio_ccw_mem_decide_memslots;
+ mdc->get_memslots = virtio_ccw_mem_get_memslots;
+ mdc->fill_device_info = virtio_ccw_mem_fill_device_info;
+ mdc->get_min_alignment = virtio_ccw_mem_get_min_alignment;
+
+ vmdc->unplug_request_check = virtio_ccw_mem_unplug_request_check;
+}
+
+static void virtio_ccw_mem_instance_init(Object *obj)
+{
+ VirtIOMEMCcw *dev = VIRTIO_MEM_CCW(obj);
+ VirtIOMEMClass *vmc;
+ VirtIOMEM *vmem;
+
+ virtio_instance_init_common(obj, &dev->vdev, sizeof(dev->vdev),
+ TYPE_VIRTIO_MEM);
+
+ dev->size_change_notifier.notify = virtio_ccw_mem_size_change_notify;
+ vmem = &dev->vdev;
+ vmc = VIRTIO_MEM_GET_CLASS(vmem);
+ /*
+ * We never remove the notifier again, as we expect both devices to
+ * disappear at the same time.
+ */
+ vmc->add_size_change_notifier(vmem, &dev->size_change_notifier);
+
+ object_property_add_alias(obj, VIRTIO_MEM_BLOCK_SIZE_PROP,
+ OBJECT(&dev->vdev), VIRTIO_MEM_BLOCK_SIZE_PROP);
+ object_property_add_alias(obj, VIRTIO_MEM_SIZE_PROP, OBJECT(&dev->vdev),
+ VIRTIO_MEM_SIZE_PROP);
+ object_property_add(obj, VIRTIO_MEM_REQUESTED_SIZE_PROP, "size",
+ virtio_ccw_mem_get_requested_size,
+ virtio_ccw_mem_set_requested_size, NULL, NULL);
+}
+
+static const TypeInfo virtio_ccw_mem = {
+ .name = TYPE_VIRTIO_MEM_CCW,
+ .parent = TYPE_VIRTIO_MD_CCW,
+ .instance_size = sizeof(VirtIOMEMCcw),
+ .instance_init = virtio_ccw_mem_instance_init,
+ .class_init = virtio_ccw_mem_class_init,
+};
+
+static void virtio_ccw_mem_register_types(void)
+{
+ type_register_static(&virtio_ccw_mem);
+}
+type_init(virtio_ccw_mem_register_types)
diff --git a/hw/s390x/virtio-ccw-mem.h b/hw/s390x/virtio-ccw-mem.h
new file mode 100644
index 0000000000..730cd9fcd7
--- /dev/null
+++ b/hw/s390x/virtio-ccw-mem.h
@@ -0,0 +1,34 @@
+/*
+ * Virtio MEM CCW device
+ *
+ * Copyright (C) 2024 Red Hat, Inc.
+ *
+ * Authors:
+ * David Hildenbrand <da...@redhat.com>
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2.
+ * See the COPYING file in the top-level directory.
+ */
+
+#ifndef HW_S390X_VIRTIO_CCW_MEM_H
+#define HW_S390X_VIRTIO_CCW_MEM_H
+
+#include "virtio-ccw-md.h"
+#include "hw/virtio/virtio-mem.h"
+#include "qom/object.h"
+
+typedef struct VirtIOMEMCcw VirtIOMEMCcw;
+
+/*
+ * virtio-mem-ccw: This extends VirtIOMDCcw
+ */
+#define TYPE_VIRTIO_MEM_CCW "virtio-mem-ccw"
+DECLARE_INSTANCE_CHECKER(VirtIOMEMCcw, VIRTIO_MEM_CCW, TYPE_VIRTIO_MEM_CCW)
+
+struct VirtIOMEMCcw {
+ VirtIOMDCcw parent_obj;
+ VirtIOMEM vdev;
+ Notifier size_change_notifier;
+};
+
+#endif /* QEMU_VIRTIO_MEM_PCI_H */
diff --git a/hw/virtio/virtio-mem.c b/hw/virtio/virtio-mem.c
index ef64bf1b4a..988101783f 100644
--- a/hw/virtio/virtio-mem.c
+++ b/hw/virtio/virtio-mem.c
@@ -61,6 +61,8 @@ static uint32_t virtio_mem_default_thp_size(void)
} else if (qemu_real_host_page_size() == 64 * KiB) {
default_thp_size = 512 * MiB;
}
+#elif defined(__s390x__)
+ default_thp_size = 1 * MiB;
#endif
return default_thp_size;
@@ -161,7 +163,7 @@ static bool virtio_mem_has_shared_zeropage(RAMBlock *rb)
* necessary (as the section size can change). But it's more likely that the
* section size will rather get smaller and not bigger over time.
*/
-#if defined(TARGET_X86_64) || defined(TARGET_I386)
+#if defined(TARGET_X86_64) || defined(TARGET_I386) || defined(TARGET_S390X)
#define VIRTIO_MEM_USABLE_EXTENT (2 * (128 * MiB))
#elif defined(TARGET_ARM)
#define VIRTIO_MEM_USABLE_EXTENT (2 * (512 * MiB))
--
2.46.0
