vfio: Disable only uncoordinated discards for VFIO_TYPE1 iommus

We support coordinated discarding of RAM using the RamDiscardManager for
the VFIO_TYPE1 iommus. Let's unlock support for coordinated discards,
keeping uncoordinated discards (e.g., via virtio-balloon) disabled if
possible.

This unlocks virtio-mem + vfio on x86-64. Note that vfio used via "nvme://"
by the block layer has to be implemented/unlocked separately. For now,
virtio-mem only supports x86-64; we don't restrict RamDiscardManager to
x86-64, though: arm64 and s390x are supposed to work as well, and we'll
test once unlocking virtio-mem support. The spapr IOMMUs will need special
care, to be tackled later, e.g.., once supporting virtio-mem.

Note: The block size of a virtio-mem device has to be set to sane sizes,
depending on the maximum hotplug size - to not run out of vfio mappings.
The default virtio-mem block size is usually in the range of a couple of
MBs. The maximum number of mapping is 64k, shared with other users.
Assume you want to hotplug 256GB using virtio-mem - the block size would
have to be set to at least 8 MiB (resulting in 32768 separate mappings).

Acked-by: Alex Williamson <alex.williamson@redhat.com>
Reviewed-by: Alex Williamson <alex.williamson@redhat.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Alex Williamson <alex.williamson@redhat.com>
Cc: Dr. David Alan Gilbert <dgilbert@redhat.com>
Cc: Igor Mammedov <imammedo@redhat.com>
Cc: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
Cc: Peter Xu <peterx@redhat.com>
Cc: Auger Eric <eric.auger@redhat.com>
Cc: Wei Yang <richard.weiyang@linux.alibaba.com>
Cc: teawater <teawaterz@linux.alibaba.com>
Cc: Marek Kedzierski <mkedzier@redhat.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20210413095531.25603-14-david@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>

1 files changed, 53 insertions, 12 deletions

diff --git a/hw/vfio/common.c b/hw/vfio/common.c
index 8a9bbf2..3f0d111 100644
--- a/hw/vfio/common.c
+++ b/hw/vfio/common.c
@@ -135,6 +135,29 @@ static const char *index_to_str(VFIODevice *vbasedev, int index)
     }
 }
 
+static int vfio_ram_block_discard_disable(VFIOContainer *container, bool state)
+{
+    switch (container->iommu_type) {
+    case VFIO_TYPE1v2_IOMMU:
+    case VFIO_TYPE1_IOMMU:
+        /*
+         * We support coordinated discarding of RAM via the RamDiscardManager.
+         */
+        return ram_block_uncoordinated_discard_disable(state);
+    default:
+        /*
+         * VFIO_SPAPR_TCE_IOMMU most probably works just fine with
+         * RamDiscardManager, however, it is completely untested.
+         *
+         * VFIO_SPAPR_TCE_v2_IOMMU with "DMA memory preregistering" does
+         * completely the opposite of managing mapping/pinning dynamically as
+         * required by RamDiscardManager. We would have to special-case sections
+         * with a RamDiscardManager.
+         */
+        return ram_block_discard_disable(state);
+    }
+}
+
 int vfio_set_irq_signaling(VFIODevice *vbasedev, int index, int subindex,
                            int action, int fd, Error **errp)
 {
@@ -1977,15 +2000,25 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
      * new memory, it will not yet set ram_block_discard_set_required() and
      * therefore, neither stops us here or deals with the sudden memory
      * consumption of inflated memory.
+     *
+     * We do support discarding of memory coordinated via the RamDiscardManager
+     * with some IOMMU types. vfio_ram_block_discard_disable() handles the
+     * details once we know which type of IOMMU we are using.
      */
-    ret = ram_block_discard_disable(true);
-    if (ret) {
-        error_setg_errno(errp, -ret, "Cannot set discarding of RAM broken");
-        return ret;
-    }
 
     QLIST_FOREACH(container, &space->containers, next) {
         if (!ioctl(group->fd, VFIO_GROUP_SET_CONTAINER, &container->fd)) {
+            ret = vfio_ram_block_discard_disable(container, true);
+            if (ret) {
+                error_setg_errno(errp, -ret,
+                                 "Cannot set discarding of RAM broken");
+                if (ioctl(group->fd, VFIO_GROUP_UNSET_CONTAINER,
+                          &container->fd)) {
+                    error_report("vfio: error disconnecting group %d from"
+                                 " container", group->groupid);
+                }
+                return ret;
+            }
             group->container = container;
             QLIST_INSERT_HEAD(&container->group_list, group, container_next);
             vfio_kvm_device_add_group(group);
@@ -2023,6 +2056,12 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
         goto free_container_exit;
     }
 
+    ret = vfio_ram_block_discard_disable(container, true);
+    if (ret) {
+        error_setg_errno(errp, -ret, "Cannot set discarding of RAM broken");
+        goto free_container_exit;
+    }
+
     switch (container->iommu_type) {
     case VFIO_TYPE1v2_IOMMU:
     case VFIO_TYPE1_IOMMU:
@@ -2070,7 +2109,7 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
             if (ret) {
                 error_setg_errno(errp, errno, "failed to enable container");
                 ret = -errno;
-                goto free_container_exit;
+                goto enable_discards_exit;
             }
         } else {
             container->prereg_listener = vfio_prereg_listener;
@@ -2082,7 +2121,7 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
                 ret = -1;
                 error_propagate_prepend(errp, container->error,
                     "RAM memory listener initialization failed: ");
-                goto free_container_exit;
+                goto enable_discards_exit;
             }
         }
 
@@ -2095,7 +2134,7 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
             if (v2) {
                 memory_listener_unregister(&container->prereg_listener);
             }
-            goto free_container_exit;
+            goto enable_discards_exit;
         }
 
         if (v2) {
@@ -2110,7 +2149,7 @@ static int vfio_connect_container(VFIOGroup *group, AddressSpace *as,
             if (ret) {
                 error_setg_errno(errp, -ret,
                                  "failed to remove existing window");
-                goto free_container_exit;
+                goto enable_discards_exit;
             }
         } else {
             /* The default table uses 4K pages */
@@ -2151,6 +2190,9 @@ listener_release_exit:
     vfio_kvm_device_del_group(group);
     vfio_listener_release(container);
 
+enable_discards_exit:
+    vfio_ram_block_discard_disable(container, false);
+
 free_container_exit:
     g_free(container);
 
@@ -2158,7 +2200,6 @@ close_fd_exit:
     close(fd);
 
 put_space_exit:
-    ram_block_discard_disable(false);
     vfio_put_address_space(space);
 
     return ret;
@@ -2280,7 +2321,7 @@ void vfio_put_group(VFIOGroup *group)
     }
 
     if (!group->ram_block_discard_allowed) {
-        ram_block_discard_disable(false);
+        vfio_ram_block_discard_disable(group->container, false);
     }
     vfio_kvm_device_del_group(group);
     vfio_disconnect_container(group);
@@ -2334,7 +2375,7 @@ int vfio_get_device(VFIOGroup *group, const char *name,
 
         if (!group->ram_block_discard_allowed) {
             group->ram_block_discard_allowed = true;
-            ram_block_discard_disable(false);
+            vfio_ram_block_discard_disable(group->container, false);
         }
     }