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] [day] [month] [year] [list]
Message-ID: <20250430135735.GU3339421@horms.kernel.org>
Date: Wed, 30 Apr 2025 14:57:35 +0100
From: Simon Horman <horms@...nel.org>
To: Jakub Kicinski <kuba@...nel.org>
Cc: davem@...emloft.net, netdev@...r.kernel.org, edumazet@...gle.com,
	pabeni@...hat.com, andrew+netdev@...n.ch, mst@...hat.com,
	jasowang@...hat.com, xuanzhuo@...ux.alibaba.com,
	eperezma@...hat.com, hawk@...nel.org, john.fastabend@...il.com,
	virtualization@...ts.linux.dev, minhquangbui99@...il.com
Subject: Re: [PATCH net] virtio-net: free xsk_buffs on error in
 virtnet_xsk_pool_enable()

On Tue, Apr 29, 2025 at 09:43:23AM -0700, Jakub Kicinski wrote:
> The selftests added to our CI by Bui Quang Minh recently reveals
> that there is a mem leak on the error path of virtnet_xsk_pool_enable():
> 
> unreferenced object 0xffff88800a68a000 (size 2048):
>   comm "xdp_helper", pid 318, jiffies 4294692778
>   hex dump (first 32 bytes):
>     00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
>     00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
>   backtrace (crc 0):
>     __kvmalloc_node_noprof+0x402/0x570
>     virtnet_xsk_pool_enable+0x293/0x6a0 (drivers/net/virtio_net.c:5882)
>     xp_assign_dev+0x369/0x670 (net/xdp/xsk_buff_pool.c:226)
>     xsk_bind+0x6a5/0x1ae0
>     __sys_bind+0x15e/0x230
>     __x64_sys_bind+0x72/0xb0
>     do_syscall_64+0xc1/0x1d0
>     entry_SYSCALL_64_after_hwframe+0x77/0x7f
> 
> Fixes: e9f3962441c0 ("virtio_net: xsk: rx: support fill with xsk buffer")
> Signed-off-by: Jakub Kicinski <kuba@...nel.org>
> ---
> CC: mst@...hat.com
> CC: jasowang@...hat.com
> CC: xuanzhuo@...ux.alibaba.com
> CC: eperezma@...hat.com
> CC: hawk@...nel.org
> CC: john.fastabend@...il.com
> CC: virtualization@...ts.linux.dev
> CC: minhquangbui99@...il.com
> ---
>  drivers/net/virtio_net.c | 4 +++-
>  1 file changed, 3 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> index 848fab51dfa1..a3d4e666c2a0 100644
> --- a/drivers/net/virtio_net.c
> +++ b/drivers/net/virtio_net.c
> @@ -5886,7 +5886,7 @@ static int virtnet_xsk_pool_enable(struct net_device *dev,
>  	hdr_dma = virtqueue_dma_map_single_attrs(sq->vq, &xsk_hdr, vi->hdr_len,
>  						 DMA_TO_DEVICE, 0);
>  	if (virtqueue_dma_mapping_error(sq->vq, hdr_dma))
> -		return -ENOMEM;

Hi Jakub,

I think you need the following here:

		err = -ENOMEM;

Else err is uninitialised when jumping to the err_free_buffs label.

Flagged by clang 20.1.3 [-Wsometimes-uninitialized], and Smatch.

> +		goto err_free_buffs;
>  
>  	err = xsk_pool_dma_map(pool, dma_dev, 0);
>  	if (err)
> @@ -5914,6 +5914,8 @@ static int virtnet_xsk_pool_enable(struct net_device *dev,
>  err_xsk_map:
>  	virtqueue_dma_unmap_single_attrs(rq->vq, hdr_dma, vi->hdr_len,
>  					 DMA_TO_DEVICE, 0);
> +err_free_buffs:
> +	kvfree(rq->xsk_buffs);
>  	return err;
>  }
>  
> -- 
> 2.49.0
> 

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ