| 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
| ||
|
Message-ID: <620c6905-d528-1992-a08e-b22b21871f7e@redhat.com>
Date: Fri, 4 Jun 2021 14:00:09 +0800
From: Jason Wang <jasowang@...hat.com>
To: Xuan Zhuo <xuanzhuo@...ux.alibaba.com>
Cc: "Michael S. Tsirkin" <mst@...hat.com>,
"David S. Miller" <davem@...emloft.net>,
Jakub Kicinski <kuba@...nel.org>,
virtualization@...ts.linux-foundation.org,
Corentin Noël <corentin.noel@...labora.com>,
netdev@...r.kernel.org
Subject: Re: [PATCH net] virtio-net: fix for skb_over_panic inside big mode
在 2021/6/4 上午11:05, Xuan Zhuo 写道:
> On Fri, 4 Jun 2021 11:00:25 +0800, Jason Wang <jasowang@...hat.com> wrote:
>> 在 2021/6/4 上午10:30, Xuan Zhuo 写道:
>>> On Fri, 4 Jun 2021 10:28:41 +0800, Jason Wang <jasowang@...hat.com> wrote:
>>>> 在 2021/6/4 上午1:09, Xuan Zhuo 写道:
>>>>> In virtio-net's large packet mode, there is a hole in the space behind
>>>>> buf.
>>>> before the buf actually or behind the vnet header?
>>>>
>>>>
>>>>> hdr_padded_len - hdr_len
>>>>>
>>>>> We must take this into account when calculating tailroom.
>>>>>
>>>>> [ 44.544385] skb_put.cold (net/core/skbuff.c:5254 (discriminator 1) net/core/skbuff.c:5252 (discriminator 1))
>>>>> [ 44.544864] page_to_skb (drivers/net/virtio_net.c:485) [ 44.545361] receive_buf (drivers/net/virtio_net.c:849 drivers/net/virtio_net.c:1131)
>>>>> [ 44.545870] ? netif_receive_skb_list_internal (net/core/dev.c:5714)
>>>>> [ 44.546628] ? dev_gro_receive (net/core/dev.c:6103)
>>>>> [ 44.547135] ? napi_complete_done (./include/linux/list.h:35 net/core/dev.c:5867 net/core/dev.c:5862 net/core/dev.c:6565)
>>>>> [ 44.547672] virtnet_poll (drivers/net/virtio_net.c:1427 drivers/net/virtio_net.c:1525)
>>>>> [ 44.548251] __napi_poll (net/core/dev.c:6985)
>>>>> [ 44.548744] net_rx_action (net/core/dev.c:7054 net/core/dev.c:7139)
>>>>> [ 44.549264] __do_softirq (./arch/x86/include/asm/jump_label.h:19 ./include/linux/jump_label.h:200 ./include/trace/events/irq.h:142 kernel/softirq.c:560)
>>>>> [ 44.549762] irq_exit_rcu (kernel/softirq.c:433 kernel/softirq.c:637 kernel/softirq.c:649)
>>>>> [ 44.551384] common_interrupt (arch/x86/kernel/irq.c:240 (discriminator 13))
>>>>> [ 44.551991] ? asm_common_interrupt (./arch/x86/include/asm/idtentry.h:638)
>>>>> [ 44.552654] asm_common_interrupt (./arch/x86/include/asm/idtentry.h:638)
>>>>>
>>>>> Fixes: fb32856b16ad ("virtio-net: page_to_skb() use build_skb when there's sufficient tailroom")
>>>>> Signed-off-by: Xuan Zhuo <xuanzhuo@...ux.alibaba.com>
>>>>> Reported-by: Corentin Noël <corentin.noel@...labora.com>
>>>>> Tested-by: Corentin Noël <corentin.noel@...labora.com>
>>>>> ---
>>>>> drivers/net/virtio_net.c | 2 +-
>>>>> 1 file changed, 1 insertion(+), 1 deletion(-)
>>>>>
>>>>> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
>>>>> index fa407eb8b457..78a01c71a17c 100644
>>>>> --- a/drivers/net/virtio_net.c
>>>>> +++ b/drivers/net/virtio_net.c
>>>>> @@ -406,7 +406,7 @@ static struct sk_buff *page_to_skb(struct virtnet_info *vi,
>>>>> * add_recvbuf_mergeable() + get_mergeable_buf_len()
>>>>> */
>>>>> truesize = headroom ? PAGE_SIZE : truesize;
>>>>> - tailroom = truesize - len - headroom;
>>>>> + tailroom = truesize - len - headroom - (hdr_padded_len - hdr_len);
>>>> The patch looks correct and I saw it has been merged.
>>>>
>>>> But I prefer to do that in receive_big() instead of here.
>>>>
>>>> Thanks
>>> How?
>>>
>>> change truesize or headroom?
>>>
>>> I didn't find a good way. Do you have a good way?
>>
>> Something like the following? The API is designed to let the caller to
>> pass a correct headroom instead of figure it out by itself.
>>
>> struct sk_buff *skb =
>> page_to_skb(vi, rq, page, 0, len, PAGE_SIZE, true, 0,
>> hdr_padded_len - hdr_len);
>>
>> Thanks
>
> This line may be affected.
>
> buf = p - headroom;
>
> In my opinion, this changes the semantics of the original headroom. The meaning
> of headroom in big mode and merge mode has become different. The more confusing
> problem is that the parameters of page_to_skb() are getting more and more
> chaotic. So I wrote the previous patch. Of course, I understand your concern.
> This patch may bring Here are more questions, although I did a lot of tests.
>
> "virtio-net: Refactor the code related to page_to_skb"
>
> But I hope that our code development direction is as close to what this patch
> realizes. I hope that the meaning of the parameters can be more clear.
So I don't object to this method, but as I replied, it's better to do
some benchmark to see if it introduces any regression
>
> Do you think this is ok?
Looks ok, but if we decide to go with your approach, it can be squashed
into that patch.
Thanks
>
> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> index 78a01c71a17c..6d62bb45a188 100644
> --- a/drivers/net/virtio_net.c
> +++ b/drivers/net/virtio_net.c
> @@ -380,34 +380,20 @@ static struct sk_buff *page_to_skb(struct virtnet_info *vi,
> struct page *page, unsigned int offset,
> unsigned int len, unsigned int truesize,
> bool hdr_valid, unsigned int metasize,
> - unsigned int headroom)
> + int tailroom, char *buf,
> + unsigned int hdr_padded_len)
> {
> struct sk_buff *skb;
> struct virtio_net_hdr_mrg_rxbuf *hdr;
> - unsigned int copy, hdr_len, hdr_padded_len;
> + unsigned int copy, hdr_len;
> struct page *page_to_free = NULL;
> - int tailroom, shinfo_size;
> - char *p, *hdr_p, *buf;
> + int shinfo_size;
> + char *p, *hdr_p;
>
> p = page_address(page) + offset;
> hdr_p = p;
>
> hdr_len = vi->hdr_len;
> - if (vi->mergeable_rx_bufs)
> - hdr_padded_len = sizeof(*hdr);
> - else
> - hdr_padded_len = sizeof(struct padded_vnet_hdr);
> -
> - /* If headroom is not 0, there is an offset between the beginning of the
> - * data and the allocated space, otherwise the data and the allocated
> - * space are aligned.
> - *
> - * Buffers with headroom use PAGE_SIZE as alloc size, see
> - * add_recvbuf_mergeable() + get_mergeable_buf_len()
> - */
> - truesize = headroom ? PAGE_SIZE : truesize;
> - tailroom = truesize - len - headroom - (hdr_padded_len - hdr_len);
> - buf = p - headroom;
>
> len -= hdr_len;
> offset += hdr_padded_len;
> @@ -492,6 +478,51 @@ static struct sk_buff *page_to_skb(struct virtnet_info *vi,
> return skb;
> }
>
> +static struct sk_buff *merge_page_to_skb(struct virtnet_info *vi,
> + struct receive_queue *rq,
> + struct page *page, unsigned int offset,
> + unsigned int len, unsigned int truesize,
> + bool hdr_valid, unsigned int metasize,
> + unsigned int headroom)
> +{
> + int tailroom;
> + char *buf;
> +
> + /* If headroom is not 0, there is an offset between the beginning of the
> + * data and the allocated space, otherwise the data and the allocated
> + * space are aligned.
> + *
> + * Buffers with headroom use PAGE_SIZE as alloc size, see
> + * add_recvbuf_mergeable() + get_mergeable_buf_len()
> + */
> + truesize = headroom ? PAGE_SIZE : truesize;
> + tailroom = truesize - len - headroom;
> + buf = page_address(page) + offset - headroom;
> +
> + page_to_skb(vi, rq, page, offset, len, truesize, hdr_valid, metasize,
> + tailroom, buf, sizeof(struct virtio_net_hdr_mrg_rxbuf))
> +
> +}
> +
> +static struct sk_buff *big_page_to_skb(struct virtnet_info *vi,
> + struct receive_queue *rq,
> + struct page *page, unsigned int offset,
> + unsigned int len, unsigned int truesize,
> + bool hdr_valid, unsigned int metasize,
> + unsigned int headroom)
> +{
> + char *p = page_address(page);
> + int hold;
> + int tailroom;
> +
> + hold = sizeof(struct padded_vnet_hdr) - vi->hdr_len;
> +
> + tailroom = truesize - len - headroom - hold;
> +
> + page_to_skb(vi, rq, page, offset, len, truesize, hdr_valid, metasize,
> + tailroom, p, sizeof(struct padded_vnet_hdr));
> +}
> +
> static int __virtnet_xdp_xmit_one(struct virtnet_info *vi,
> struct send_queue *sq,
> struct xdp_frame *xdpf)
>
>
>>
>>> Thanks.
>>>
>>>>
>>>>> buf = p - headroom;
>>>>>
>>>>> len -= hdr_len;
Powered by blists - more mailing lists