linux-kernel - Re: [PATCH v3 09/13] vhost/vdpa: remove vhost_vdpa_config

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Fri, 5 Feb 2021 15:17:07 +0100
From:   Stefano Garzarella <sgarzare@...hat.com>
To:     "Michael S. Tsirkin" <mst@...hat.com>
Cc:     Jason Wang <jasowang@...hat.com>,
        virtualization@...ts.linux-foundation.org,
        Xie Yongji <xieyongji@...edance.com>, kvm@...r.kernel.org,
        Laurent Vivier <lvivier@...hat.com>,
        Stefan Hajnoczi <stefanha@...hat.com>,
        Max Gurtovoy <mgurtovoy@...dia.com>,
        linux-kernel@...r.kernel.org
Subject: Re: [PATCH v3 09/13] vhost/vdpa: remove vhost_vdpa_config_validate()

On Fri, Feb 05, 2021 at 08:32:37AM -0500, Michael S. Tsirkin wrote:
>On Fri, Feb 05, 2021 at 10:16:51AM +0100, Stefano Garzarella wrote:
>> On Fri, Feb 05, 2021 at 11:27:32AM +0800, Jason Wang wrote:
>> >
>> > On 2021/2/5 上午1:22, Stefano Garzarella wrote:
>> > > get_config() and set_config() callbacks in the 'struct vdpa_config_ops'
>> > > usually already validated the inputs. Also now they can return an error,
>> > > so we don't need to validate them here anymore.
>> > >
>> > > Let's use the return value of these callbacks and return it in case of
>> > > error in vhost_vdpa_get_config() and vhost_vdpa_set_config().
>> > >
>> > > Originally-by: Xie Yongji <xieyongji@...edance.com>
>> > > Signed-off-by: Stefano Garzarella <sgarzare@...hat.com>
>> > > ---
>> > >  drivers/vhost/vdpa.c | 41 +++++++++++++----------------------------
>> > >  1 file changed, 13 insertions(+), 28 deletions(-)
>> > >
>> > > diff --git a/drivers/vhost/vdpa.c b/drivers/vhost/vdpa.c
>> > > index ef688c8c0e0e..d61e779000a8 100644
>> > > --- a/drivers/vhost/vdpa.c
>> > > +++ b/drivers/vhost/vdpa.c
>> > > @@ -185,51 +185,35 @@ static long vhost_vdpa_set_status(struct vhost_vdpa *v, u8 __user *statusp)
>> > >  	return 0;
>> > >  }
>> > > -static int vhost_vdpa_config_validate(struct vhost_vdpa *v,
>> > > -				      struct vhost_vdpa_config *c)
>> > > -{
>> > > -	long size = 0;
>> > > -
>> > > -	switch (v->virtio_id) {
>> > > -	case VIRTIO_ID_NET:
>> > > -		size = sizeof(struct virtio_net_config);
>> > > -		break;
>> > > -	}
>> > > -
>> > > -	if (c->len == 0)
>> > > -		return -EINVAL;
>> > > -
>> > > -	if (c->len > size - c->off)
>> > > -		return -E2BIG;
>> > > -
>> > > -	return 0;
>> > > -}
>> > > -
>> > >  static long vhost_vdpa_get_config(struct vhost_vdpa *v,
>> > >  				  struct vhost_vdpa_config __user *c)
>> > >  {
>> > >  	struct vdpa_device *vdpa = v->vdpa;
>> > >  	struct vhost_vdpa_config config;
>> > >  	unsigned long size = offsetof(struct vhost_vdpa_config, buf);
>> > > +	long ret;
>> > >  	u8 *buf;
>> > >  	if (copy_from_user(&config, c, size))
>> > >  		return -EFAULT;
>> > > -	if (vhost_vdpa_config_validate(v, &config))
>> > > +	if (config.len == 0)
>> > >  		return -EINVAL;
>> > >  	buf = kvzalloc(config.len, GFP_KERNEL);
>> >
>> >
>> > Then it means usersapce can allocate a very large memory.
>>
>> Good point.
>>
>> >
>> > Rethink about this, we should limit the size here (e.g PAGE_SIZE) or
>> > fetch the config size first (either through a config ops as you
>> > suggested or a variable in the vdpa device that is initialized during
>> > device creation).
>>
>> Maybe PAGE_SIZE is okay as a limit.
>>
>> If instead we want to fetch the config size, then better a config ops in my
>> opinion, to avoid adding a new parameter to __vdpa_alloc_device().
>>
>> I vote for PAGE_SIZE, but it isn't a strong opinion.
>>
>> What do you and @Michael suggest?
>>
>> Thanks,
>> Stefano
>
>Devices know what the config size is. Just have them provide it.
>

Okay, I'll add get_config_size() callback in vdpa_config_ops and I'll 
leave vhost_vdpa_config_validate() that will use that callback instead 
of 'virtio_id' to get the config size from the device.

At this point I think I can remove the "vdpa: add return value to 
get_config/set_config callbacks" patch and leave void return to 
get_config/set_config callbacks.

Does this make sense?

Thanks,
Stefano