Patchwork [net-next] core: adjust checks for calling skb_copy_bits in skb_try_coalesce

login
register
mail settings
Submitter roy.qing.li@gmail.com
Date Sept. 20, 2012, 3:19 a.m.
Message ID <1348111182-6827-1-git-send-email-roy.qing.li@gmail.com>
Download mbox | patch
Permalink /patch/185302/
State Rejected
Delegated to: David Miller
Headers show

Comments

roy.qing.li@gmail.com - Sept. 20, 2012, 3:19 a.m.
From: Li RongQing <roy.qing.li@gmail.com>

Ensure that frags and frags_list of dst skb are empty if need to call
skb_copy_bits, or else it will break the data sequence.

Cc: Eric Dumazet <edumazet@google.com>
Signed-off-by: Li RongQing <roy.qing.li@gmail.com>
---
 net/core/skbuff.c |    8 ++++----
 1 files changed, 4 insertions(+), 4 deletions(-)
Eric Dumazet - Sept. 20, 2012, 5:20 a.m.
On Thu, 2012-09-20 at 11:19 +0800, roy.qing.li@gmail.com wrote:
> From: Li RongQing <roy.qing.li@gmail.com>
> 
> Ensure that frags and frags_list of dst skb are empty if need to call
> skb_copy_bits, or else it will break the data sequence.
> 
> Cc: Eric Dumazet <edumazet@google.com>
> Signed-off-by: Li RongQing <roy.qing.li@gmail.com>
> ---
>  net/core/skbuff.c |    8 ++++----
>  1 files changed, 4 insertions(+), 4 deletions(-)
> 
> diff --git a/net/core/skbuff.c b/net/core/skbuff.c
> index fe00d12..f0b9446 100644
> --- a/net/core/skbuff.c
> +++ b/net/core/skbuff.c
> @@ -3455,15 +3455,15 @@ bool skb_try_coalesce(struct sk_buff *to, struct sk_buff *from,
>  	if (skb_cloned(to))
>  		return false;
>  
> -	if (len <= skb_tailroom(to)) {
> +	if (skb_has_frag_list(to) || skb_has_frag_list(from))
> +		return false;
> +
> +	if (len <= skb_tailroom(to) && !skb_shinfo(to)->nr_frags) {
>  		BUG_ON(skb_copy_bits(from, 0, skb_put(to, len), len));
>  		*delta_truesize = 0;
>  		return true;
>  	}
>  
> -	if (skb_has_frag_list(to) || skb_has_frag_list(from))
> -		return false;
> -
>  	if (skb_headlen(from) != 0) {
>  		struct page *page;
>  		unsigned int offset;

This is not needed at all.



--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
roy.qing.li@gmail.com - Sept. 20, 2012, 5:40 a.m.
>>               unsigned int offset;
>
> This is not needed at all.
>
>

I think the below modification maybe needed,
if (len <= skb_tailroom(to) && !skb_shinfo(to)->nr_frags) {
..
}

First skb A is added to skb TO frags, since the len is larger
than skb_tailroot(TO), but second len of skb B is less than
skb_tailroot(To)  which will call skb_copy_bits.

Of cause, this kinds of cases maybe only exist on my mind.

-Roy
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Eric Dumazet - Sept. 20, 2012, 5:53 a.m.
On Thu, 2012-09-20 at 13:40 +0800, RongQing Li wrote:
> >>               unsigned int offset;
> >
> > This is not needed at all.
> >
> >
> 
> I think the below modification maybe needed,
> if (len <= skb_tailroom(to) && !skb_shinfo(to)->nr_frags) {
> ..
> }
> 
> First skb A is added to skb TO frags, since the len is larger
> than skb_tailroot(TO), but second len of skb B is less than
> skb_tailroot(To)  which will call skb_copy_bits.
> 
> Of cause, this kinds of cases maybe only exist on my mind.
> 

Did you read skb_tailroom(to) definition by any chance ?

static inline int skb_tailroom(const struct sk_buff *skb)
{
        return skb_is_nonlinear(skb) ? 0 : skb->end - skb->tail;
}

Current code is fine, because if @to is linear, its ... linear.



--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
roy.qing.li@gmail.com - Sept. 20, 2012, 5:57 a.m.
> Did you read skb_tailroom(to) definition by any chance ?
>
> static inline int skb_tailroom(const struct sk_buff *skb)
> {
>         return skb_is_nonlinear(skb) ? 0 : skb->end - skb->tail;
> }
>
> Current code is fine, because if @to is linear, its ... linear.
>
>
>

I am wrong, thanks.
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Eric Dumazet - Sept. 20, 2012, 7 a.m.
On Thu, 2012-09-20 at 13:57 +0800, RongQing Li wrote:

> 
> I am wrong, thanks.

Thanks for taking a look at this stuff ;)


--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Patch

diff --git a/net/core/skbuff.c b/net/core/skbuff.c
index fe00d12..f0b9446 100644
--- a/net/core/skbuff.c
+++ b/net/core/skbuff.c
@@ -3455,15 +3455,15 @@  bool skb_try_coalesce(struct sk_buff *to, struct sk_buff *from,
 	if (skb_cloned(to))
 		return false;
 
-	if (len <= skb_tailroom(to)) {
+	if (skb_has_frag_list(to) || skb_has_frag_list(from))
+		return false;
+
+	if (len <= skb_tailroom(to) && !skb_shinfo(to)->nr_frags) {
 		BUG_ON(skb_copy_bits(from, 0, skb_put(to, len), len));
 		*delta_truesize = 0;
 		return true;
 	}
 
-	if (skb_has_frag_list(to) || skb_has_frag_list(from))
-		return false;
-
 	if (skb_headlen(from) != 0) {
 		struct page *page;
 		unsigned int offset;