]> git.baikalelectronics.ru Git - kernel.git/commitdiff
virtio-mem: disallow mapping virtio-mem memory via /dev/mem
authorDavid Hildenbrand <david@redhat.com>
Tue, 9 Nov 2021 02:35:53 +0000 (18:35 -0800)
committerLinus Torvalds <torvalds@linux-foundation.org>
Tue, 9 Nov 2021 18:02:52 +0000 (10:02 -0800)
We don't want user space to be able to map virtio-mem device memory
directly (e.g., via /dev/mem) in order to have guarantees that in a sane
setup we'll never accidentially access unplugged memory within the
device-managed region of a virtio-mem device, just as required by the
virtio-spec.

As soon as the virtio-mem driver is loaded, the device region is visible
in /proc/iomem via the parent device region.  From that point on user
space is aware of the device region and we want to disallow mapping
anything inside that region (where we will dynamically (un)plug memory)
until the driver has been unloaded cleanly and e.g., another driver might
take over.

By creating our parent IORESOURCE_SYSTEM_RAM resource with
IORESOURCE_EXCLUSIVE, we will disallow any /dev/mem access to our device
region until the driver was unloaded cleanly and removed the parent
region.  This will work even though only some memory blocks are actually
currently added to Linux and appear as busy in the resource tree.

So access to the region from user space is only possible
a) if we don't load the virtio-mem driver.
b) after unloading the virtio-mem driver cleanly.

Don't build virtio-mem if access to /dev/mem cannot be restricticted -- if
we have CONFIG_DEVMEM=y but CONFIG_STRICT_DEVMEM is not set.

Link: https://lkml.kernel.org/r/20210920142856.17758-4-david@redhat.com
Signed-off-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Dan Williams <dan.j.williams@intel.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Cc: Andy Shevchenko <andy.shevchenko@gmail.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Hanjun Guo <guohanjun@huawei.com>
Cc: Jason Wang <jasowang@redhat.com>
Cc: "Rafael J. Wysocki" <rafael.j.wysocki@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
drivers/virtio/Kconfig
drivers/virtio/virtio_mem.c

index 3654def9915c596130bd840f46a52ff0159aaffc..44a082639439a79e529f9d930077f1a4b2bf3422 100644 (file)
@@ -101,6 +101,7 @@ config VIRTIO_MEM
        depends on MEMORY_HOTPLUG
        depends on MEMORY_HOTREMOVE
        depends on CONTIG_ALLOC
+       depends on EXCLUSIVE_SYSTEM_RAM
        help
         This driver provides access to virtio-mem paravirtualized memory
         devices, allowing to hotplug and hotunplug memory.
index 3573b78f6b050533ca6e766f984553a3437a8d06..0da0af251c73d0bf5142254f328d7846191e0cf3 100644 (file)
@@ -2672,8 +2672,10 @@ static int virtio_mem_create_resource(struct virtio_mem *vm)
        if (!name)
                return -ENOMEM;
 
+       /* Disallow mapping device memory via /dev/mem completely. */
        vm->parent_resource = __request_mem_region(vm->addr, vm->region_size,
-                                                  name, IORESOURCE_SYSTEM_RAM);
+                                                  name, IORESOURCE_SYSTEM_RAM |
+                                                  IORESOURCE_EXCLUSIVE);
        if (!vm->parent_resource) {
                kfree(name);
                dev_warn(&vm->vdev->dev, "could not reserve device region\n");