VFIO update 2018-08-17

- Enhance balloon inhibitor for multiple users and use around vfio
    device assignment (Alex Williamson)
 -----BEGIN PGP SIGNATURE-----
 Version: GnuPG v2.0.14 (GNU/Linux)
 
 iQIcBAABAgAGBQJbdvs5AAoJECObm247sIsioA4QAKWqKWnqOu4AwIdADzOEKTv+
 jSG+M6S6uEBbhk6hC+scnIlyA6MwiTmh8uJYDOTPy/4S9kv4+Bj/LykoJxk1T3UI
 jREP+EHMUe13YfG//5yIb0HLIBEI7VJeBteyksboZ8KTO4kMt4RCnk1lpfARxIjB
 ouMIngGq9ltOWAann9tcjGaPXfKjB9LSSYyhQMlqSzl7LckOsGAgmdFU7uurlTrq
 a5kB2oRNP7CSbF/fDTsWd8nsCGy357Cou42KpsTRQN+TeLmqExQXguCQLszBup3H
 VYwS0FzpR1ix7jOuafHLcBh9VgSEFYWf71oznMoPDiv2vVjKGsDG3Tzu5Xqz0vsI
 ynKXCXNbXZhpazZzw8ThXoLs8mWLWSrtqvjQItDmrl0kboY6ZwOZc8m+PQmIqtHQ
 RDcVlCT0TecwYXBZ5G930JpcEsUFBKfZrla+v2W0ON6HVA6eVBweSata5r08XvHK
 P2rbHgO6NSDZTMZUOborh/sbINb61PNW7h7inigHzpDJxjseIz/vO+mVjfsCZMed
 dSu/0VYCbA0cS2IK21zZFsR4LNlL4yzFzhONdBqbZD7zhiZL25IcZV/TVzHJE+zS
 su7uDSLNihJYOXAXO7s0Ne2vO5lZzpYykVFOAmQQicxcYZz+u8LxXY+NlTjKJhEv
 x3fobaflymrG7xkRy3xq
 =JQ6k
 -----END PGP SIGNATURE-----

Merge remote-tracking branch 'remotes/awilliam/tags/vfio-update-20180817.0' into staging

VFIO update 2018-08-17

 - Enhance balloon inhibitor for multiple users and use around vfio
   device assignment (Alex Williamson)

# gpg: Signature made Fri 17 Aug 2018 17:43:37 BST
# gpg:                using RSA key 239B9B6E3BB08B22
# gpg: Good signature from "Alex Williamson <alex.williamson@redhat.com>"
# gpg:                 aka "Alex Williamson <alex@shazbot.org>"
# gpg:                 aka "Alex Williamson <alwillia@redhat.com>"
# gpg:                 aka "Alex Williamson <alex.l.williamson@gmail.com>"
# Primary key fingerprint: 42F6 C04E 540B D1A9 9E7B  8A90 239B 9B6E 3BB0 8B22

* remotes/awilliam/tags/vfio-update-20180817.0:
  vfio/ccw/pci: Allow devices to opt-in for ballooning
  vfio: Inhibit ballooning based on group attachment to a container
  kvm: Use inhibit to prevent ballooning without synchronous mmu
  balloon: Allow multiple inhibit users

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

accel/kvm/kvm-all.c

@@ -39,6 +39,7 @@
 #include "trace.h"
 #include "hw/irq.h"
 #include "sysemu/sev.h"
+#include "sysemu/balloon.h"
 
 #include "hw/boards.h"
 
@@ -1698,6 +1699,9 @@ static int kvm_init(MachineState *ms)
     s->many_ioeventfds = kvm_check_many_ioeventfds();
 
     s->sync_mmu = !!kvm_vm_check_extension(kvm_state, KVM_CAP_SYNC_MMU);
+    if (!s->sync_mmu) {
+        qemu_balloon_inhibit(true);
+    }
 
     return 0;
 

balloon.c

@@ -26,6 +26,7 @@
 
 #include "qemu/osdep.h"
 #include "qemu-common.h"
+#include "qemu/atomic.h"
 #include "exec/cpu-common.h"
 #include "sysemu/kvm.h"
 #include "sysemu/balloon.h"
@@ -37,16 +38,22 @@
 static QEMUBalloonEvent *balloon_event_fn;
 static QEMUBalloonStatus *balloon_stat_fn;
 static void *balloon_opaque;
-static bool balloon_inhibited;
+static int balloon_inhibit_count;
 
 bool qemu_balloon_is_inhibited(void)
 {
-    return balloon_inhibited;
+    return atomic_read(&balloon_inhibit_count) > 0;
 }
 
 void qemu_balloon_inhibit(bool state)
 {
-    balloon_inhibited = state;
+    if (state) {
+        atomic_inc(&balloon_inhibit_count);
+    } else {
+        atomic_dec(&balloon_inhibit_count);
+    }
+
+    assert(atomic_read(&balloon_inhibit_count) >= 0);
 }
 
 static bool have_balloon(Error **errp)

hw/vfio/ccw.c

@@ -349,6 +349,15 @@ static void vfio_ccw_get_device(VFIOGroup *group, VFIOCCWDevice *vcdev,
         }
     }
 
+    /*
+     * All vfio-ccw devices are believed to operate in a way compatible with
+     * memory ballooning, ie. pages pinned in the host are in the current
+     * working set of the guest driver and therefore never overlap with pages
+     * available to the guest balloon driver.  This needs to be set before
+     * vfio_get_device() for vfio common to handle the balloon inhibitor.
+     */
+    vcdev->vdev.balloon_allowed = true;
+
     if (vfio_get_device(group, vcdev->cdev.mdevid, &vcdev->vdev, errp)) {
         goto out_err;
     }

hw/vfio/common.c

@@ -32,6 +32,7 @@
 #include "hw/hw.h"
 #include "qemu/error-report.h"
 #include "qemu/range.h"
+#include "sysemu/balloon.h"
 #include "sysemu/kvm.h"
 #include "trace.h"
 #include "qapi/error.h"
@@ -1044,6 +1045,33 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
 
     space = vfio_get_address_space(as);
 
+    /*
+     * VFIO is currently incompatible with memory ballooning insofar as the
+     * madvise to purge (zap) the page from QEMU's address space does not
+     * interact with the memory API and therefore leaves stale virtual to
+     * physical mappings in the IOMMU if the page was previously pinned.  We
+     * therefore add a balloon inhibit for each group added to a container,
+     * whether the container is used individually or shared.  This provides
+     * us with options to allow devices within a group to opt-in and allow
+     * ballooning, so long as it is done consistently for a group (for instance
+     * if the device is an mdev device where it is known that the host vendor
+     * driver will never pin pages outside of the working set of the guest
+     * driver, which would thus not be ballooning candidates).
+     *
+     * The first opportunity to induce pinning occurs here where we attempt to
+     * attach the group to existing containers within the AddressSpace.  If any
+     * pages are already zapped from the virtual address space, such as from a
+     * previous ballooning opt-in, new pinning will cause valid mappings to be
+     * re-established.  Likewise, when the overall MemoryListener for a new
+     * container is registered, a replay of mappings within the AddressSpace
+     * will occur, re-establishing any previously zapped pages as well.
+     *
+     * NB. Balloon inhibiting does not currently block operation of the
+     * balloon driver or revoke previously pinned pages, it only prevents
+     * calling madvise to modify the virtual mapping of ballooned pages.
+     */
+    qemu_balloon_inhibit(true);
+
     QLIST_FOREACH(container, &space->containers, next) {
         if (!ioctl(group->fd, VFIO_GROUP_SET_CONTAINER, &container->fd)) {
             group->container = container;
@@ -1232,6 +1260,7 @@ close_fd_exit:
     close(fd);
 
 put_space_exit:
+    qemu_balloon_inhibit(false);
     vfio_put_address_space(space);
 
     return ret;
@@ -1352,6 +1381,9 @@ void vfio_put_group(VFIOGroup *group)
         return;
     }
 
+    if (!group->balloon_allowed) {
+        qemu_balloon_inhibit(false);
+    }
     vfio_kvm_device_del_group(group);
     vfio_disconnect_container(group);
     QLIST_REMOVE(group, next);
@@ -1387,6 +1419,25 @@ int vfio_get_device(VFIOGroup *group, const char *name,
         return ret;
     }
 
+    /*
+     * Clear the balloon inhibitor for this group if the driver knows the
+     * device operates compatibly with ballooning.  Setting must be consistent
+     * per group, but since compatibility is really only possible with mdev
+     * currently, we expect singleton groups.
+     */
+    if (vbasedev->balloon_allowed != group->balloon_allowed) {
+        if (!QLIST_EMPTY(&group->device_list)) {
+            error_setg(errp,
+                       "Inconsistent device balloon setting within group");
+            return -1;
+        }
+
+        if (!group->balloon_allowed) {
+            group->balloon_allowed = true;
+            qemu_balloon_inhibit(false);
+        }
+    }
+
     vbasedev->fd = fd;
     vbasedev->group = group;
     QLIST_INSERT_HEAD(&group->device_list, vbasedev, next);

hw/vfio/pci.c

@@ -2804,12 +2804,13 @@ static void vfio_realize(PCIDevice *pdev, Error **errp)
     VFIOPCIDevice *vdev = DO_UPCAST(VFIOPCIDevice, pdev, pdev);
     VFIODevice *vbasedev_iter;
     VFIOGroup *group;
-    char *tmp, group_path[PATH_MAX], *group_name;
+    char *tmp, *subsys, group_path[PATH_MAX], *group_name;
     Error *err = NULL;
     ssize_t len;
     struct stat st;
     int groupid;
     int i, ret;
+    bool is_mdev;
 
     if (!vdev->vbasedev.sysfsdev) {
         if (!(~vdev->host.domain || ~vdev->host.bus ||
@@ -2869,6 +2870,27 @@ static void vfio_realize(PCIDevice *pdev, Error **errp)
         }
     }
 
+    /*
+     * Mediated devices *might* operate compatibly with memory ballooning, but
+     * we cannot know for certain, it depends on whether the mdev vendor driver
+     * stays in sync with the active working set of the guest driver.  Prevent
+     * the x-balloon-allowed option unless this is minimally an mdev device.
+     */
+    tmp = g_strdup_printf("%s/subsystem", vdev->vbasedev.sysfsdev);
+    subsys = realpath(tmp, NULL);
+    g_free(tmp);
+    is_mdev = (strcmp(subsys, "/sys/bus/mdev") == 0);
+    free(subsys);
+
+    trace_vfio_mdev(vdev->vbasedev.name, is_mdev);
+
+    if (vdev->vbasedev.balloon_allowed && !is_mdev) {
+        error_setg(errp, "x-balloon-allowed only potentially compatible "
+                   "with mdev devices");
+        vfio_put_group(group);
+        goto error;
+    }
+
     ret = vfio_get_device(group, vdev->vbasedev.name, &vdev->vbasedev, errp);
     if (ret) {
         vfio_put_group(group);
@@ -3170,6 +3192,8 @@ static Property vfio_pci_dev_properties[] = {
     DEFINE_PROP_BIT("x-igd-opregion", VFIOPCIDevice, features,
                     VFIO_FEATURE_ENABLE_IGD_OPREGION_BIT, false),
     DEFINE_PROP_BOOL("x-no-mmap", VFIOPCIDevice, vbasedev.no_mmap, false),
+    DEFINE_PROP_BOOL("x-balloon-allowed", VFIOPCIDevice,
+                     vbasedev.balloon_allowed, false),
     DEFINE_PROP_BOOL("x-no-kvm-intx", VFIOPCIDevice, no_kvm_intx, false),
     DEFINE_PROP_BOOL("x-no-kvm-msi", VFIOPCIDevice, no_kvm_msi, false),
     DEFINE_PROP_BOOL("x-no-kvm-msix", VFIOPCIDevice, no_kvm_msix, false),

hw/vfio/trace-events

@@ -39,6 +39,7 @@ vfio_pci_hot_reset_result(const char *name, const char *result) "%s hot reset: %
 vfio_populate_device_config(const char *name, unsigned long size, unsigned long offset, unsigned long flags) "Device %s config:\n  size: 0x%lx, offset: 0x%lx, flags: 0x%lx"
 vfio_populate_device_get_irq_info_failure(void) "VFIO_DEVICE_GET_IRQ_INFO failure: %m"
 vfio_realize(const char *name, int group_id) " (%s) group %d"
+vfio_mdev(const char *name, bool is_mdev) " (%s) is_mdev %d"
 vfio_add_ext_cap_dropped(const char *name, uint16_t cap, uint16_t offset) "%s 0x%x@0x%x"
 vfio_pci_reset(const char *name) " (%s)"
 vfio_pci_reset_flr(const char *name) "%s FLR/VFIO_DEVICE_RESET"

hw/virtio/virtio-balloon.c

@@ -21,7 +21,6 @@
 #include "hw/mem/pc-dimm.h"
 #include "sysemu/balloon.h"
 #include "hw/virtio/virtio-balloon.h"
-#include "sysemu/kvm.h"
 #include "exec/address-spaces.h"
 #include "qapi/error.h"
 #include "qapi/qapi-events-misc.h"
@@ -36,8 +35,7 @@
 
 static void balloon_page(void *addr, int deflate)
 {
-    if (!qemu_balloon_is_inhibited() && (!kvm_enabled() ||
+    if (!qemu_balloon_is_inhibited()) {
-                                         kvm_has_sync_mmu())) {
         qemu_madvise(addr, BALLOON_PAGE_SIZE,
                 deflate ? QEMU_MADV_WILLNEED : QEMU_MADV_DONTNEED);
     }

include/hw/vfio/vfio-common.h

@@ -112,6 +112,7 @@ typedef struct VFIODevice {
     bool reset_works;
     bool needs_reset;
     bool no_mmap;
+    bool balloon_allowed;
     VFIODeviceOps *ops;
     unsigned int num_irqs;
     unsigned int num_regions;
@@ -131,6 +132,7 @@ typedef struct VFIOGroup {
     QLIST_HEAD(, VFIODevice) device_list;
     QLIST_ENTRY(VFIOGroup) next;
     QLIST_ENTRY(VFIOGroup) container_next;
+    bool balloon_allowed;
 } VFIOGroup;
 
 typedef struct VFIODMABuf {