linux-kernel - Re: [PATCH] virtio: Return correct errno for function init

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]

Message-Id: <20160706111832.5d4d6f43.cornelia.huck@de.ibm.com>
Date:	Wed, 6 Jul 2016 11:18:32 +0200
From:	Cornelia Huck <cornelia.huck@...ibm.com>
To:	Minfei Huang <mnghuan@...il.com>, mst@...hat.com
Cc:	Minfei Huang <minfei.hmf@...baba-inc.com>,
	linux-kernel@...r.kernel.org,
	virtualization@...ts.linux-foundation.org
Subject: Re: [PATCH] virtio: Return correct errno for function init_vq's
 failure

On Mon, 27 Jun 2016 10:09:18 +0800
Minfei Huang <mnghuan@...il.com> wrote:

> The error number -ENOENT or 0 will be returned, if we can not allocate
> more memory in function init_vq. If host can support multiple virtual
> queues, and we fails to allocate necessary memory structures for vq,
> kernel may crash due to incorrect returning.
> 
> To fix it, kernel will return correct value in init_vq.
> 
> Signed-off-by: Minfei Huang <mnghuan@...il.com>
> Signed-off-by: Minfei Huang <minfei.hmf@...baba-inc.com>
> ---
>  drivers/block/virtio_blk.c | 5 ++---
>  1 file changed, 2 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/block/virtio_blk.c b/drivers/block/virtio_blk.c
> index 42758b5..40ecb2b 100644
> --- a/drivers/block/virtio_blk.c
> +++ b/drivers/block/virtio_blk.c
> @@ -393,11 +393,10 @@ static int init_vq(struct virtio_blk *vblk)
>  	if (err)
>  		num_vqs = 1;
> 
> +	err = -ENOMEM;
>  	vblk->vqs = kmalloc(sizeof(*vblk->vqs) * num_vqs, GFP_KERNEL);
> -	if (!vblk->vqs) {
> -		err = -ENOMEM;
> +	if (!vblk->vqs)
>  		goto out;
> -	}
> 
>  	names = kmalloc(sizeof(*names) * num_vqs, GFP_KERNEL);
>  	if (!names)

The error handling in this function looks horrible.

When mq was introduced, init_vq started mixing up several things:
- The mq feature is not available - which is not an error, and
therefore should not have any influence on the return code.
- One of the several memory allocations failed - only ->vqs gets
special treatment, however.
- The ->find_vqs callback failed.

Your patch fixes the code, but it is still very convoluted due to the
temporary arrays.

May it be worthwile to introduce a helper for setting up the virtqueues
where all virtqueues are essentially the same and just get a
consecutive number? Michael?