[PATCH v3 0/2] fix two virtio queues memleak

[PATCH v3 0/2] fix two virtio queues memleak

[Pan Nengyuan]

This series fix two vqs leak:
1. Do delete vqs on the error path in virtio_blk_device_realize().
2. Do delete vqs in virtio_iommu_device_unrealize() to fix another leaks.

v2->v1:
- Fix incorrect free in virtio_blk_device_realize, it will cause a uaf.

v3->v2:
- Aslo clean 's->as_by_busptr' hash table in virtio_iommu_device_unrealize.(Suggested by Stefano Garzarella) 

Pan Nengyuan (2):
  virtio-blk: delete vqs on the error path in realize()
  virtio-iommu: delete vqs in unrealize to fix memleak

 hw/block/virtio-blk.c    | 3 +++
 hw/virtio/virtio-iommu.c | 3 +++
 2 files changed, 6 insertions(+)

-- 
2.18.2

[PATCH v3 1/2] virtio-blk: delete vqs on the error path in realize()

[Pan Nengyuan]

virtio_vqs forgot to free on the error path in realize(). Fix that.

The asan stack:
Direct leak of 14336 byte(s) in 1 object(s) allocated from:
    #0 0x7f58b93fd970 in __interceptor_calloc (/lib64/libasan.so.5+0xef970)
    #1 0x7f58b858249d in g_malloc0 (/lib64/libglib-2.0.so.0+0x5249d)
    #2 0x5562cc627f49 in virtio_add_queue /mnt/sdb/qemu/hw/virtio/virtio.c:2413
    #3 0x5562cc4b524a in virtio_blk_device_realize /mnt/sdb/qemu/hw/block/virtio-blk.c:1202
    #4 0x5562cc613050 in virtio_device_realize /mnt/sdb/qemu/hw/virtio/virtio.c:3615
    #5 0x5562ccb7a568 in device_set_realized /mnt/sdb/qemu/hw/core/qdev.c:891
    #6 0x5562cd39cd45 in property_set_bool /mnt/sdb/qemu/qom/object.c:2238

Reported-by: Euler Robot <euler.robot@huawei.com>
Signed-off-by: Pan Nengyuan <pannengyuan@huawei.com>
Reviewed-by: Stefano Garzarella <sgarzare@redhat.com>
---
v2->v1:
- Fix incorrect free in v1, it will cause a uaf.
---
Cc: Stefan Hajnoczi <stefanha@redhat.com>
Cc: Kevin Wolf <kwolf@redhat.com>
Cc: Max Reitz <mreitz@redhat.com>
Cc: qemu-block@nongnu.org
---
 hw/block/virtio-blk.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/hw/block/virtio-blk.c b/hw/block/virtio-blk.c
index 142863a3b2..97ba8a2187 100644
--- a/hw/block/virtio-blk.c
+++ b/hw/block/virtio-blk.c
@@ -1204,6 +1204,9 @@ static void virtio_blk_device_realize(DeviceState *dev, Error **errp)
     virtio_blk_data_plane_create(vdev, conf, &s->dataplane, &err);
     if (err != NULL) {
         error_propagate(errp, err);
+        for (i = 0; i < conf->num_queues; i++) {
+            virtio_del_queue(vdev, i);
+        }
         virtio_cleanup(vdev);
         return;
     }
-- 
2.18.2

[PATCH v3 2/2] virtio-iommu: delete vqs in unrealize to fix memleak

[Pan Nengyuan]

req_vq/event_vq forgot to free in unrealize. Fix that.
And aslo do clean 's->as_by_busptr' hash table in unrealize to fix another leak.

Signed-off-by: Pan Nengyuan <pannengyuan@huawei.com>
Acked-by: Eric Auger <eric.auger@redhat.com>
---
Cc: Eric Auger <eric.auger@redhat.com>
---
v3->v1/v2:
- Aslo clean 's->as_by_busptr' hash table in unrealize.(Suggested by Stefano Garzarella)
---
 hw/virtio/virtio-iommu.c | 3 +++
 1 file changed, 3 insertions(+)

diff --git a/hw/virtio/virtio-iommu.c b/hw/virtio/virtio-iommu.c
index 4cee8083bc..694698fa0f 100644
--- a/hw/virtio/virtio-iommu.c
+++ b/hw/virtio/virtio-iommu.c
@@ -696,7 +696,10 @@ static void virtio_iommu_device_unrealize(DeviceState *dev, Error **errp)
     g_tree_destroy(s->domains);
     g_tree_destroy(s->endpoints);
 
+    virtio_delete_queue(s->req_vq);
+    virtio_delete_queue(s->event_vq);
     virtio_cleanup(vdev);
+    g_hash_table_destroy(s->as_by_busptr);
 }
 
 static void virtio_iommu_device_reset(VirtIODevice *vdev)
-- 
2.18.2

Re: [PATCH v3 2/2] virtio-iommu: delete vqs in unrealize to fix memleak

[Stefano Garzarella]

On Fri, Mar 27, 2020 at 05:56:42PM +0800, Pan Nengyuan wrote:
> req_vq/event_vq forgot to free in unrealize. Fix that.
> And aslo do clean 's->as_by_busptr' hash table in unrealize to fix another leak.

s/aslo/also

Maybe we can also update the subject in something like this:
"virtio-iommu: avoid memleak in the unrealize"

> 
> Signed-off-by: Pan Nengyuan <pannengyuan@huawei.com>
> Acked-by: Eric Auger <eric.auger@redhat.com>
> ---
> Cc: Eric Auger <eric.auger@redhat.com>
> ---
> v3->v1/v2:
> - Aslo clean 's->as_by_busptr' hash table in unrealize.(Suggested by Stefano Garzarella)
> ---
>  hw/virtio/virtio-iommu.c | 3 +++
>  1 file changed, 3 insertions(+)
> 
> diff --git a/hw/virtio/virtio-iommu.c b/hw/virtio/virtio-iommu.c
> index 4cee8083bc..694698fa0f 100644
> --- a/hw/virtio/virtio-iommu.c
> +++ b/hw/virtio/virtio-iommu.c
> @@ -696,7 +696,10 @@ static void virtio_iommu_device_unrealize(DeviceState *dev, Error **errp)
>      g_tree_destroy(s->domains);
>      g_tree_destroy(s->endpoints);
>  
> +    virtio_delete_queue(s->req_vq);
> +    virtio_delete_queue(s->event_vq);
>      virtio_cleanup(vdev);
> +    g_hash_table_destroy(s->as_by_busptr);

If you need to respin, you could move g_hash_table_destroy()
at the beggining of unrealize(), just to follow a reverse order of
realize().

Thanks,
Stefano

>  }
>  
>  static void virtio_iommu_device_reset(VirtIODevice *vdev)
> -- 
> 2.18.2
> 
> 

Re: [PATCH v3 2/2] virtio-iommu: delete vqs in unrealize to fix memleak

[Pan Nengyuan]

On 3/28/2020 12:26 AM, Stefano Garzarella wrote:
> On Fri, Mar 27, 2020 at 05:56:42PM +0800, Pan Nengyuan wrote:
>> req_vq/event_vq forgot to free in unrealize. Fix that.
>> And aslo do clean 's->as_by_busptr' hash table in unrealize to fix another leak.
> 
> s/aslo/also
> 
> Maybe we can also update the subject in something like this:
> "virtio-iommu: avoid memleak in the unrealize"

OK.

> 
>>
>> Signed-off-by: Pan Nengyuan <pannengyuan@huawei.com>
>> Acked-by: Eric Auger <eric.auger@redhat.com>
>> ---
>> Cc: Eric Auger <eric.auger@redhat.com>
>> ---
>> v3->v1/v2:
>> - Aslo clean 's->as_by_busptr' hash table in unrealize.(Suggested by Stefano Garzarella)
>> ---
>>  hw/virtio/virtio-iommu.c | 3 +++
>>  1 file changed, 3 insertions(+)
>>
>> diff --git a/hw/virtio/virtio-iommu.c b/hw/virtio/virtio-iommu.c
>> index 4cee8083bc..694698fa0f 100644
>> --- a/hw/virtio/virtio-iommu.c
>> +++ b/hw/virtio/virtio-iommu.c
>> @@ -696,7 +696,10 @@ static void virtio_iommu_device_unrealize(DeviceState *dev, Error **errp)
>>      g_tree_destroy(s->domains);
>>      g_tree_destroy(s->endpoints);
>>  
>> +    virtio_delete_queue(s->req_vq);
>> +    virtio_delete_queue(s->event_vq);
>>      virtio_cleanup(vdev);
>> +    g_hash_table_destroy(s->as_by_busptr);
> 
> If you need to respin, you could move g_hash_table_destroy()
> at the beggining of unrealize(), just to follow a reverse order of
> realize().

OK.

Thanks.

> 
> Thanks,
> Stefano
> 
>>  }
>>  
>>  static void virtio_iommu_device_reset(VirtIODevice *vdev)
>> -- 
>> 2.18.2
>>
>>
>