Add virtio-balloon support

This adds a VirtIO based balloon driver.  It uses madvise() to actually balloon
the memory when possible.

Until 2.6.27, KVM forced memory pinning so we must disable ballooning unless the
kernel actually supports it when using KVM.  It's always safe when using TCG.

Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>



git-svn-id: svn://svn.savannah.nongnu.org/qemu/trunk@5874 c046a42c-6fe2-441c-8c8c-71466251a162

7 files changed, 261 insertions, 3 deletions

diff --git a/Makefile.target b/Makefile.target
index 362799a..671d72a 100644
--- a/Makefile.target
+++ b/Makefile.target
@@ -665,7 +665,7 @@ OBJS+= fdc.o mc146818rtc.o serial.o i8259.o i8254.o pcspk.o pc.o
 OBJS+= cirrus_vga.o apic.o parallel.o acpi.o piix_pci.o
 OBJS+= usb-uhci.o vmmouse.o vmport.o vmware_vga.o
 # virtio support
-OBJS+= virtio.o virtio-blk.o
+OBJS+= virtio.o virtio-blk.o virtio-balloon.o
 CPPFLAGS += -DHAS_AUDIO -DHAS_AUDIO_CHOICE
 endif
 ifeq ($(TARGET_BASE_ARCH), ppc)
@@ -684,7 +684,7 @@ OBJS+= unin_pci.o ppc_chrp.o
 # PowerPC 4xx boards
 OBJS+= pflash_cfi02.o ppc4xx_devs.o ppc4xx_pci.o ppc405_uc.o ppc405_boards.o
 # virtio support
-OBJS+= virtio.o virtio-blk.o
+OBJS+= virtio.o virtio-blk.o virtio-balloon.o
 endif
 ifeq ($(TARGET_BASE_ARCH), mips)
 OBJS+= mips_r4k.o mips_jazz.o mips_malta.o mips_mipssim.o
diff --git a/hw/pc.c b/hw/pc.c
index 25dedc8..20827f2 100644
--- a/hw/pc.c
+++ b/hw/pc.c
@@ -34,6 +34,7 @@
 #include "console.h"
 #include "fw_cfg.h"
 #include "virtio-blk.h"
+#include "virtio-balloon.h"
 
 /* output Bochs bios info messages */
 //#define DEBUG_BIOS
@@ -1105,6 +1106,10 @@ static void pc_init1(ram_addr_t ram_size, int vga_ram_size,
             unit_id++;
         }
     }
+
+    /* Add virtio balloon device */
+    if (pci_enabled)
+        virtio_balloon_init(pci_bus);
 }
 
 static void pc_init_pci(ram_addr_t ram_size, int vga_ram_size,
diff --git a/hw/virtio-balloon.c b/hw/virtio-balloon.c
new file mode 100644
index 0000000..e668436
--- /dev/null
+++ b/hw/virtio-balloon.c
@@ -0,0 +1,194 @@
+/*
+ * Virtio Block Device
+ *
+ * Copyright IBM, Corp. 2008
+ *
+ * Authors:
+ *  Anthony Liguori   <aliguori@us.ibm.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-common.h"
+#include "virtio.h"
+#include "pc.h"
+#include "sysemu.h"
+#include "cpu.h"
+#include "balloon.h"
+#include "virtio-balloon.h"
+#include "kvm.h"
+
+#if defined(__linux__)
+#include <sys/mman.h>
+#endif
+
+typedef struct VirtIOBalloon
+{
+    VirtIODevice vdev;
+    VirtQueue *ivq, *dvq;
+    uint32_t num_pages;
+    uint32_t actual;
+} VirtIOBalloon;
+
+static VirtIOBalloon *to_virtio_balloon(VirtIODevice *vdev)
+{
+    return (VirtIOBalloon *)vdev;
+}
+
+static void balloon_page(void *addr, int deflate)
+{
+#if defined(__linux__)
+    if (!kvm_enabled() || kvm_has_sync_mmu())
+        madvise(addr, TARGET_PAGE_SIZE,
+                deflate ? MADV_WILLNEED : MADV_DONTNEED);
+#endif
+}
+
+/* FIXME: once we do a virtio refactoring, this will get subsumed into common
+ * code */
+static size_t memcpy_from_iovector(void *data, size_t offset, size_t size,
+                                   struct iovec *iov, int iovlen)
+{
+    int i;
+    uint8_t *ptr = data;
+    size_t iov_off = 0;
+    size_t data_off = 0;
+
+    for (i = 0; i < iovlen && size; i++) {
+        if (offset < (iov_off + iov[i].iov_len)) {
+            size_t len = MIN((iov_off + iov[i].iov_len) - offset , size);
+
+            memcpy(ptr + data_off, iov[i].iov_base + (offset - iov_off), len);
+
+            data_off += len;
+            offset += len;
+            size -= len;
+        }
+
+        iov_off += iov[i].iov_len;
+    }
+
+    return data_off;
+}
+
+static void virtio_balloon_handle_output(VirtIODevice *vdev, VirtQueue *vq)
+{
+    VirtIOBalloon *s = to_virtio_balloon(vdev);
+    VirtQueueElement elem;
+
+    while (virtqueue_pop(vq, &elem)) {
+        size_t offset = 0;
+        uint32_t pfn;
+
+        while (memcpy_from_iovector(&pfn, offset, 4,
+                                    elem.out_sg, elem.out_num) == 4) {
+            ram_addr_t pa;
+            ram_addr_t addr;
+
+            pa = (ram_addr_t)ldl_p(&pfn) << VIRTIO_BALLOON_PFN_SHIFT;
+            offset += 4;
+
+            addr = cpu_get_physical_page_desc(pa);
+            if ((addr & ~TARGET_PAGE_MASK) != IO_MEM_RAM)
+                continue;
+
+            balloon_page(phys_ram_base + addr, !!(vq == s->dvq));
+        }
+
+        virtqueue_push(vq, &elem, offset);
+        virtio_notify(vdev, vq);
+    }
+}
+
+static void virtio_balloon_get_config(VirtIODevice *vdev, uint8_t *config_data)
+{
+    VirtIOBalloon *dev = to_virtio_balloon(vdev);
+    struct virtio_balloon_config config;
+
+    config.num_pages = cpu_to_le32(dev->num_pages);
+    config.actual = cpu_to_le32(dev->actual);
+
+    memcpy(config_data, &config, 8);
+}
+
+static void virtio_balloon_set_config(VirtIODevice *vdev,
+                                      const uint8_t *config_data)
+{
+    VirtIOBalloon *dev = to_virtio_balloon(vdev);
+    struct virtio_balloon_config config;
+    memcpy(&config, config_data, 8);
+    dev->actual = config.actual;
+}
+
+static uint32_t virtio_balloon_get_features(VirtIODevice *vdev)
+{
+    return 0;
+}
+
+static ram_addr_t virtio_balloon_to_target(void *opaque, ram_addr_t target)
+{
+    VirtIOBalloon *dev = opaque;
+
+    if (target > ram_size)
+        target = ram_size;
+
+    if (target) {
+        dev->num_pages = (ram_size - target) >> VIRTIO_BALLOON_PFN_SHIFT;
+        virtio_notify_config(&dev->vdev);
+    }
+
+    return ram_size - (dev->actual << VIRTIO_BALLOON_PFN_SHIFT);
+}
+
+static void virtio_balloon_save(QEMUFile *f, void *opaque)
+{
+    VirtIOBalloon *s = opaque;
+
+    virtio_save(&s->vdev, f);
+
+    qemu_put_be32(f, s->num_pages);
+    qemu_put_be32(f, s->actual);
+}
+
+static int virtio_balloon_load(QEMUFile *f, void *opaque, int version_id)
+{
+    VirtIOBalloon *s = opaque;
+
+    if (version_id != 1)
+        return -EINVAL;
+
+    virtio_load(&s->vdev, f);
+
+    s->num_pages = qemu_get_be32(f);
+    s->actual = qemu_get_be32(f);
+
+    return 0;
+}
+
+void *virtio_balloon_init(PCIBus *bus)
+{
+    VirtIOBalloon *s;
+
+    s = (VirtIOBalloon *)virtio_init_pci(bus, "virtio-balloon",
+                                         6900, 0x1002,
+                                         0, VIRTIO_ID_BALLOON,
+                                         0x05, 0x00, 0x00,
+                                         8, sizeof(VirtIOBalloon));
+    if (s == NULL)
+        return NULL;
+
+    s->vdev.get_config = virtio_balloon_get_config;
+    s->vdev.set_config = virtio_balloon_set_config;
+    s->vdev.get_features = virtio_balloon_get_features;
+
+    s->ivq = virtio_add_queue(&s->vdev, 128, virtio_balloon_handle_output);
+    s->dvq = virtio_add_queue(&s->vdev, 128, virtio_balloon_handle_output);
+
+    qemu_add_balloon_handler(virtio_balloon_to_target, s);
+
+    register_savevm("virtio-balloon", -1, 1, virtio_balloon_save, virtio_balloon_load, s);
+
+    return &s->vdev;
+}
diff --git a/hw/virtio-balloon.h b/hw/virtio-balloon.h
new file mode 100644
index 0000000..c71f970
--- /dev/null
+++ b/hw/virtio-balloon.h
@@ -0,0 +1,42 @@
+/*
+ * Virtio Support
+ *
+ * Copyright IBM, Corp. 2007-2008
+ *
+ * Authors:
+ *  Anthony Liguori   <aliguori@us.ibm.com>
+ *  Rusty Russell     <rusty@rustcorp.com.au>
+ *
+ * This work is licensed under the terms of the GNU GPL, version 2.  See
+ * the COPYING file in the top-level directory.
+ *
+ */
+
+#ifndef _QEMU_VIRTIO_BALLOON_H
+#define _QEMU_VIRTIO_BALLOON_H
+
+#include "virtio.h"
+#include "pci.h"
+
+/* from Linux's linux/virtio_balloon.h */
+
+/* The ID for virtio_balloon */
+#define VIRTIO_ID_BALLOON 5
+
+/* The feature bitmap for virtio balloon */
+#define VIRTIO_BALLOON_F_MUST_TELL_HOST 0 /* Tell before reclaiming pages */
+
+/* Size of a PFN in the balloon interface. */
+#define VIRTIO_BALLOON_PFN_SHIFT 12
+
+struct virtio_balloon_config
+{
+    /* Number of pages host wants Guest to give up. */
+    uint32_t num_pages;
+    /* Number of pages we've actually got in balloon. */
+    uint32_t actual;
+};
+
+void *virtio_balloon_init(PCIBus *bus);
+
+#endif
diff --git a/kvm-all.c b/kvm-all.c
index 39f826b..69ca46b 100644
--- a/kvm-all.c
+++ b/kvm-all.c
@@ -549,3 +549,15 @@ int kvm_vcpu_ioctl(CPUState *env, int type, ...)
 
     return ret;
 }
+
+int kvm_has_sync_mmu(void)
+{
+    KVMState *s = kvm_state;
+
+#ifdef KVM_CAP_SYNC_MMU
+    if (kvm_ioctl(s, KVM_CHECK_EXTENSION, KVM_CAP_SYNC_MMU) > 0)
+        return 1;
+#endif
+
+    return 0;
+}
diff --git a/kvm.h b/kvm.h
index fb952a3..ec0083f 100644
--- a/kvm.h
+++ b/kvm.h
@@ -42,6 +42,9 @@ void kvm_physical_sync_dirty_bitmap(target_phys_addr_t start_addr, target_phys_a
 
 int kvm_log_start(target_phys_addr_t phys_addr, target_phys_addr_t len);
 int kvm_log_stop(target_phys_addr_t phys_addr, target_phys_addr_t len);
+
+int kvm_has_sync_mmu(void);
+
 /* internal API */
 
 struct KVMState;
diff --git a/monitor.c b/monitor.c
index a928c3e..f142a87 100644
--- a/monitor.c
+++ b/monitor.c
@@ -1402,7 +1402,9 @@ static void do_info_balloon(void)
     ram_addr_t actual;
 
     actual = qemu_balloon_status();
-    if (actual == 0)
+    if (kvm_enabled() && !kvm_has_sync_mmu())
+        term_printf("Using KVM without synchronous MMU, ballooning disabled\n");
+    else if (actual == 0)
         term_printf("Ballooning not activated in VM\n");
     else
         term_printf("balloon: actual=%d\n", (int)(actual >> 20));