* [Patch net] net: make pskb_trim_rcsum_slow() robust
@ 2018-10-30 0:35 Cong Wang
2018-10-30 2:14 ` Eric Dumazet
2018-10-31 19:36 ` David Miller
0 siblings, 2 replies; 8+ messages in thread
From: Cong Wang @ 2018-10-30 0:35 UTC (permalink / raw)
To: netdev; +Cc: Cong Wang, Eric Dumazet
Most callers of pskb_trim_rcsum() simply drops the skb when
it fails, however, ip_check_defrag() still continues to pass
the skb up to stack. In that case, we should restore its previous
csum if __pskb_trim() fails.
Found this during code review.
Fixes: 88078d98d1bb ("net: pskb_trim_rcsum() and CHECKSUM_COMPLETE are friends")
Cc: Eric Dumazet <edumazet@google.com>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
---
net/core/skbuff.c | 8 +++++++-
1 file changed, 7 insertions(+), 1 deletion(-)
diff --git a/net/core/skbuff.c b/net/core/skbuff.c
index 946de0e24c87..5decd6e6d2b6 100644
--- a/net/core/skbuff.c
+++ b/net/core/skbuff.c
@@ -1843,6 +1843,9 @@ EXPORT_SYMBOL(___pskb_trim);
*/
int pskb_trim_rcsum_slow(struct sk_buff *skb, unsigned int len)
{
+ __wsum old_csum = skb->csum;
+ int ret;
+
if (skb->ip_summed == CHECKSUM_COMPLETE) {
int delta = skb->len - len;
@@ -1850,7 +1853,10 @@ int pskb_trim_rcsum_slow(struct sk_buff *skb, unsigned int len)
skb_checksum(skb, len, delta, 0),
len);
}
- return __pskb_trim(skb, len);
+ ret = __pskb_trim(skb, len);
+ if (unlikely(ret))
+ skb->csum = old_csum;
+ return ret;
}
EXPORT_SYMBOL(pskb_trim_rcsum_slow);
--
2.16.4
^ permalink raw reply related [flat|nested] 8+ messages in thread
* Re: [Patch net] net: make pskb_trim_rcsum_slow() robust
2018-10-30 0:35 [Patch net] net: make pskb_trim_rcsum_slow() robust Cong Wang
@ 2018-10-30 2:14 ` Eric Dumazet
2018-10-30 2:21 ` Cong Wang
2018-10-31 19:36 ` David Miller
1 sibling, 1 reply; 8+ messages in thread
From: Eric Dumazet @ 2018-10-30 2:14 UTC (permalink / raw)
To: Cong Wang, netdev; +Cc: Eric Dumazet
On 10/29/2018 05:35 PM, Cong Wang wrote:
> Most callers of pskb_trim_rcsum() simply drops the skb when
> it fails, however, ip_check_defrag() still continues to pass
> the skb up to stack. In that case, we should restore its previous
> csum if __pskb_trim() fails.
>
> Found this during code review.
>
> Fixes: 88078d98d1bb ("net: pskb_trim_rcsum() and CHECKSUM_COMPLETE are friends")
> Cc: Eric Dumazet <edumazet@google.com>
> Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
> ---
> net/core/skbuff.c | 8 +++++++-
> 1 file changed, 7 insertions(+), 1 deletion(-)
>
> diff --git a/net/core/skbuff.c b/net/core/skbuff.c
> index 946de0e24c87..5decd6e6d2b6 100644
> --- a/net/core/skbuff.c
> +++ b/net/core/skbuff.c
> @@ -1843,6 +1843,9 @@ EXPORT_SYMBOL(___pskb_trim);
> */
> int pskb_trim_rcsum_slow(struct sk_buff *skb, unsigned int len)
> {
> + __wsum old_csum = skb->csum;
> + int ret;
> +
> if (skb->ip_summed == CHECKSUM_COMPLETE) {
> int delta = skb->len - len;
>
> @@ -1850,7 +1853,10 @@ int pskb_trim_rcsum_slow(struct sk_buff *skb, unsigned int len)
> skb_checksum(skb, len, delta, 0),
> len);
> }
> - return __pskb_trim(skb, len);
> + ret = __pskb_trim(skb, len);
> + if (unlikely(ret))
> + skb->csum = old_csum;
Would not it be simpler to set ip_summed to CHECKSUM_NONE (no need to save old_csum) ?
> + return ret;
> }
> EXPORT_SYMBOL(pskb_trim_rcsum_slow);
>
>
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: [Patch net] net: make pskb_trim_rcsum_slow() robust
2018-10-30 2:14 ` Eric Dumazet
@ 2018-10-30 2:21 ` Cong Wang
2018-10-30 2:25 ` Eric Dumazet
0 siblings, 1 reply; 8+ messages in thread
From: Cong Wang @ 2018-10-30 2:21 UTC (permalink / raw)
To: Eric Dumazet; +Cc: Linux Kernel Network Developers, Eric Dumazet
On Mon, Oct 29, 2018 at 7:14 PM Eric Dumazet <eric.dumazet@gmail.com> wrote:
>
> Would not it be simpler to set ip_summed to CHECKSUM_NONE (no need to save old_csum) ?
For !CHECKSUM_COMPLETE, ip_summed should be untouched, right?
If you mean only setting to CHECKSUM_NONE for CHECKSUM_COMPLETE case,
the end result may not be simpler.
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: [Patch net] net: make pskb_trim_rcsum_slow() robust
2018-10-30 2:21 ` Cong Wang
@ 2018-10-30 2:25 ` Eric Dumazet
2018-10-30 2:41 ` Cong Wang
0 siblings, 1 reply; 8+ messages in thread
From: Eric Dumazet @ 2018-10-30 2:25 UTC (permalink / raw)
To: Cong Wang, Eric Dumazet; +Cc: Linux Kernel Network Developers, Eric Dumazet
On 10/29/2018 07:21 PM, Cong Wang wrote:
> On Mon, Oct 29, 2018 at 7:14 PM Eric Dumazet <eric.dumazet@gmail.com> wrote:
>>
>> Would not it be simpler to set ip_summed to CHECKSUM_NONE (no need to save old_csum) ?
>
> For !CHECKSUM_COMPLETE, ip_summed should be untouched, right?
>
> If you mean only setting to CHECKSUM_NONE for CHECKSUM_COMPLETE case,
> the end result may not be simpler.
I meant to reinstate what was there before my patch in this error case
if (skb->ip_summed == CHECKSUM_COMPLETE)
skb->ip_summed = CHECKSUM_NONE;
That would only be run in error (quite unlikely) path, instead of saving old_csum in all cases.
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: [Patch net] net: make pskb_trim_rcsum_slow() robust
2018-10-30 2:25 ` Eric Dumazet
@ 2018-10-30 2:41 ` Cong Wang
2018-10-30 3:08 ` Eric Dumazet
0 siblings, 1 reply; 8+ messages in thread
From: Cong Wang @ 2018-10-30 2:41 UTC (permalink / raw)
To: Eric Dumazet; +Cc: Linux Kernel Network Developers, Eric Dumazet
On Mon, Oct 29, 2018 at 7:25 PM Eric Dumazet <eric.dumazet@gmail.com> wrote:
>
>
>
> On 10/29/2018 07:21 PM, Cong Wang wrote:
> > On Mon, Oct 29, 2018 at 7:14 PM Eric Dumazet <eric.dumazet@gmail.com> wrote:
> >>
> >> Would not it be simpler to set ip_summed to CHECKSUM_NONE (no need to save old_csum) ?
> >
> > For !CHECKSUM_COMPLETE, ip_summed should be untouched, right?
> >
> > If you mean only setting to CHECKSUM_NONE for CHECKSUM_COMPLETE case,
> > the end result may not be simpler.
>
> I meant to reinstate what was there before my patch in this error case
>
> if (skb->ip_summed == CHECKSUM_COMPLETE)
> skb->ip_summed = CHECKSUM_NONE;
>
> That would only be run in error (quite unlikely) path, instead of saving old_csum in all cases.
I know your point, however, I am not sure that is a desired behavior.
On failure, I think the whole skb should be restored to its previous state
before entering this function, changing it to CHECKSUM_NONE on failure
is inconsistent with success case.
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: [Patch net] net: make pskb_trim_rcsum_slow() robust
2018-10-30 2:41 ` Cong Wang
@ 2018-10-30 3:08 ` Eric Dumazet
2018-10-30 18:57 ` Cong Wang
0 siblings, 1 reply; 8+ messages in thread
From: Eric Dumazet @ 2018-10-30 3:08 UTC (permalink / raw)
To: Cong Wang, Eric Dumazet; +Cc: Linux Kernel Network Developers, Eric Dumazet
On 10/29/2018 07:41 PM, Cong Wang wrote:
> On Mon, Oct 29, 2018 at 7:25 PM Eric Dumazet <eric.dumazet@gmail.com> wrote:
>>
>>
>>
>> On 10/29/2018 07:21 PM, Cong Wang wrote:
>>> On Mon, Oct 29, 2018 at 7:14 PM Eric Dumazet <eric.dumazet@gmail.com> wrote:
>>>>
>>>> Would not it be simpler to set ip_summed to CHECKSUM_NONE (no need to save old_csum) ?
>>>
>>> For !CHECKSUM_COMPLETE, ip_summed should be untouched, right?
>>>
>>> If you mean only setting to CHECKSUM_NONE for CHECKSUM_COMPLETE case,
>>> the end result may not be simpler.
>>
>> I meant to reinstate what was there before my patch in this error case
>>
>> if (skb->ip_summed == CHECKSUM_COMPLETE)
>> skb->ip_summed = CHECKSUM_NONE;
>>
>> That would only be run in error (quite unlikely) path, instead of saving old_csum in all cases.
>
> I know your point, however, I am not sure that is a desired behavior.
>
> On failure, I think the whole skb should be restored to its previous state
> before entering this function, changing it to CHECKSUM_NONE on failure
> is inconsistent with success case.
>
Before my patch, we were changing skb->ip_summed to CHECKSUM_NONE,
so why suddenly we need to be consistent ?
In any case, ip_check_defrag() should really drop this skb, as for other allocation
failures (like skb_share_check()), if really we want consistency.
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: [Patch net] net: make pskb_trim_rcsum_slow() robust
2018-10-30 3:08 ` Eric Dumazet
@ 2018-10-30 18:57 ` Cong Wang
0 siblings, 0 replies; 8+ messages in thread
From: Cong Wang @ 2018-10-30 18:57 UTC (permalink / raw)
To: Eric Dumazet; +Cc: Linux Kernel Network Developers, Eric Dumazet
On Mon, Oct 29, 2018 at 8:08 PM Eric Dumazet <eric.dumazet@gmail.com> wrote:
>
>
>
> On 10/29/2018 07:41 PM, Cong Wang wrote:
> > On Mon, Oct 29, 2018 at 7:25 PM Eric Dumazet <eric.dumazet@gmail.com> wrote:
> >>
> >>
> >>
> >> On 10/29/2018 07:21 PM, Cong Wang wrote:
> >>> On Mon, Oct 29, 2018 at 7:14 PM Eric Dumazet <eric.dumazet@gmail.com> wrote:
> >>>>
> >>>> Would not it be simpler to set ip_summed to CHECKSUM_NONE (no need to save old_csum) ?
> >>>
> >>> For !CHECKSUM_COMPLETE, ip_summed should be untouched, right?
> >>>
> >>> If you mean only setting to CHECKSUM_NONE for CHECKSUM_COMPLETE case,
> >>> the end result may not be simpler.
> >>
> >> I meant to reinstate what was there before my patch in this error case
> >>
> >> if (skb->ip_summed == CHECKSUM_COMPLETE)
> >> skb->ip_summed = CHECKSUM_NONE;
> >>
> >> That would only be run in error (quite unlikely) path, instead of saving old_csum in all cases.
> >
> > I know your point, however, I am not sure that is a desired behavior.
> >
> > On failure, I think the whole skb should be restored to its previous state
> > before entering this function, changing it to CHECKSUM_NONE on failure
> > is inconsistent with success case.
> >
>
> Before my patch, we were changing skb->ip_summed to CHECKSUM_NONE,
> so why suddenly we need to be consistent ?
That is because setting it to CHECKSUM_NONE _was_ how the success
case works and nothing _was_ needed for failure case.
You changed how we handle checksum for success case, it is why we need
to change for the failure case too.
>
> In any case, ip_check_defrag() should really drop this skb, as for other allocation
> failures (like skb_share_check()), if really we want consistency.
I have the same feeling, just not brave enough to change the logic of
ip_check_defrag() where pskb_may_pull() failure is treated in a same way.
^ permalink raw reply [flat|nested] 8+ messages in thread
* Re: [Patch net] net: make pskb_trim_rcsum_slow() robust
2018-10-30 0:35 [Patch net] net: make pskb_trim_rcsum_slow() robust Cong Wang
2018-10-30 2:14 ` Eric Dumazet
@ 2018-10-31 19:36 ` David Miller
1 sibling, 0 replies; 8+ messages in thread
From: David Miller @ 2018-10-31 19:36 UTC (permalink / raw)
To: xiyou.wangcong; +Cc: netdev, edumazet
From: Cong Wang <xiyou.wangcong@gmail.com>
Date: Mon, 29 Oct 2018 17:35:15 -0700
> Most callers of pskb_trim_rcsum() simply drops the skb when
> it fails, however, ip_check_defrag() still continues to pass
> the skb up to stack. In that case, we should restore its previous
> csum if __pskb_trim() fails.
>
> Found this during code review.
>
> Fixes: 88078d98d1bb ("net: pskb_trim_rcsum() and CHECKSUM_COMPLETE are friends")
> Cc: Eric Dumazet <edumazet@google.com>
> Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
I kind of agree with Eric that we should make all callers, including
ip_check_defrag(), fail just as with any memory allocation failure.
^ permalink raw reply [flat|nested] 8+ messages in thread
end of thread, other threads:[~2018-11-01 4:35 UTC | newest]
Thread overview: 8+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2018-10-30 0:35 [Patch net] net: make pskb_trim_rcsum_slow() robust Cong Wang
2018-10-30 2:14 ` Eric Dumazet
2018-10-30 2:21 ` Cong Wang
2018-10-30 2:25 ` Eric Dumazet
2018-10-30 2:41 ` Cong Wang
2018-10-30 3:08 ` Eric Dumazet
2018-10-30 18:57 ` Cong Wang
2018-10-31 19:36 ` David Miller
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.