netdev - Re: [RFC PATCH] vhost-vdpa: mark vhost device invalid to reflect vdpa device unregistration

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:   Mon, 5 Jul 2021 11:48:36 +0800
From:   Jason Wang <jasowang@...hat.com>
To:     gautam.dawar@...inx.com
Cc:     martinh@...inx.com, hanand@...inx.com, gdawar@...inx.com,
        "Michael S. Tsirkin" <mst@...hat.com>, kvm@...r.kernel.org,
        virtualization@...ts.linux-foundation.org, netdev@...r.kernel.org,
        linux-kernel@...r.kernel.org
Subject: Re: [RFC PATCH] vhost-vdpa: mark vhost device invalid to reflect vdpa
 device unregistration


在 2021/7/5 上午4:52, gautam.dawar@...inx.com 写道:
>   	vma->vm_page_prot = pgprot_noncached(vma->vm_page_prot);
> @@ -1091,11 +1122,13 @@ static void vhost_vdpa_remove(struct vdpa_device *vdpa)
>   		opened = atomic_cmpxchg(&v->opened, 0, 1);
>   		if (!opened)
>   			break;
> -		wait_for_completion_timeout(&v->completion,
> -					    msecs_to_jiffies(1000));
> -		dev_warn_once(&v->dev,
> -			      "%s waiting for/dev/%s to be closed\n",
> -			      __func__, dev_name(&v->dev));
> +		if (!wait_for_completion_timeout(&v->completion,
> +					    msecs_to_jiffies(1000))) {
> +			dev_warn(&v->dev,
> +				 "%s/dev/%s in use, continue..\n",
> +				 __func__, dev_name(&v->dev));
> +			break;
> +		}
>   	} while (1);
>   
>   	put_device(&v->dev);
> +	v->dev_invalid = true;


Besides the mapping handling mentioned by Michael. I think this can lead 
use-after-free. put_device may release the memory.

Another fundamental issue, vDPA is the parent of vhost-vDPA device. I'm 
not sure the device core can allow the parent to go away first.

Thanks