netdev - Re: [PATCH 1/2][net-next] skbuff: move netlink_large_alloc_large

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: Fri, 3 Nov 2023 11:50:10 +0800
From: Yunsheng Lin <linyunsheng@...wei.com>
To: "Li,Rongqing" <lirongqing@...du.com>, "netdev@...r.kernel.org"
	<netdev@...r.kernel.org>
Subject: Re: [PATCH 1/2][net-next] skbuff: move
 netlink_large_alloc_large_skb() to skbuff.c

On 2023/11/2 20:09, Li,Rongqing wrote:
>> -----Original Message-----
>> From: Yunsheng Lin <linyunsheng@...wei.com>
>> Sent: Thursday, November 2, 2023 7:02 PM
>> To: Li,Rongqing <lirongqing@...du.com>; netdev@...r.kernel.org
>> Subject: Re: [PATCH 1/2][net-next] skbuff: move netlink_large_alloc_large_skb()
>> to skbuff.c
>>
>> On 2023/11/2 14:28, Li RongQing wrote:
>>> move netlink_alloc_large_skb and netlink_skb_destructor to skbuff.c
>>> and rename them more generic, so they can be used elsewhere large
>>> non-contiguous physical memory is needed
>>>
>>> Signed-off-by: Li RongQing <lirongqing@...du.com>
>>> ---
>>>  include/linux/skbuff.h   |  3 +++
>>>  net/core/skbuff.c        | 40
>> ++++++++++++++++++++++++++++++++++++++++
>>>  net/netlink/af_netlink.c | 41
>>> ++---------------------------------------
>>>  3 files changed, 45 insertions(+), 39 deletions(-)
>>>
>>> diff --git a/include/linux/skbuff.h b/include/linux/skbuff.h index
>>> 4174c4b..774a401 100644
>>> --- a/include/linux/skbuff.h
>>> +++ b/include/linux/skbuff.h
>>> @@ -5063,5 +5063,8 @@ static inline void skb_mark_for_recycle(struct
>>> sk_buff *skb)  ssize_t skb_splice_from_iter(struct sk_buff *skb, struct iov_iter
>> *iter,
>>>  			     ssize_t maxsize, gfp_t gfp);
>>>
>>> +
>>> +void large_skb_destructor(struct sk_buff *skb); struct sk_buff
>>> +*alloc_large_skb(unsigned int size, int broadcast);
>>>  #endif	/* __KERNEL__ */
>>>  #endif	/* _LINUX_SKBUFF_H */
>>> diff --git a/net/core/skbuff.c b/net/core/skbuff.c index
>>> 4570705..20ffcd5 100644
>>> --- a/net/core/skbuff.c
>>> +++ b/net/core/skbuff.c
>>> @@ -6917,3 +6917,43 @@ ssize_t skb_splice_from_iter(struct sk_buff *skb,
>> struct iov_iter *iter,
>>>  	return spliced ?: ret;
>>>  }
>>>  EXPORT_SYMBOL(skb_splice_from_iter);
>>> +
>>> +void large_skb_destructor(struct sk_buff *skb) {
>>> +	if (is_vmalloc_addr(skb->head)) {
>>> +		if (!skb->cloned ||
>>> +		    !atomic_dec_return(&(skb_shinfo(skb)->dataref)))
>>> +			vfree(skb->head);
>>> +
>>> +		skb->head = NULL;
>>
>> There seems to be an assumption that skb returned from
>> netlink_alloc_large_skb() is not expecting the frag page for shinfo->frags*, as the
>> above NULL setting will bypass most of the handling in skb_release_data(),then
>> how can we ensure that the user is not breaking the assumption if we make it
>> more generic?
>>
> 
> How about to add WARN_ON(skb_shinfo(skb)-> nr_frags) to find this condition
> 

There is some other handling other than skb_shinfo(skb)-> nr_frags, such as
zcopy, fraglist and pp_recycle handling, I am not sure if adding those check
in the normal datapatch is worth it if netlink_alloc_large_skb() is only used
in the nlmsg operations, which is less performance senstive.

If there are other nlmsg operations that needs it too? if not, maybe we limit
netlink_alloc_large_skb() in nlmsg if we can assume all nlmsg APIs dosen't
break the above assumptionm, introducing something like vnlmsg_new() or only
change nlmsg_new() to use netlink_alloc_large_skb(), so that all nlmsg users can
make use of it.

If there is more user making use of netlink_alloc_large_skb() in the future, we
can make it usable outside of nlmsg.

> 
> -Li RongQing
>>
>