netdev - Re: [PATCH 04/12 v2 RFC] skbuff: Push status and refcounts into sock_zerocopy

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]

Message-ID: <f21da0d8-fdad-f98a-62c6-6ab8a84b8e75@gmail.com>
Date:   Tue, 22 Dec 2020 09:49:19 -0700
From:   David Ahern <dsahern@...il.com>
To:     Willem de Bruijn <willemdebruijn.kernel@...il.com>,
        Jonathan Lemon <jonathan.lemon@...il.com>
Cc:     Network Development <netdev@...r.kernel.org>,
        Eric Dumazet <edumazet@...gle.com>,
        Kernel Team <kernel-team@...com>
Subject: Re: [PATCH 04/12 v2 RFC] skbuff: Push status and refcounts into
 sock_zerocopy_callback

On 12/22/20 7:43 AM, Willem de Bruijn wrote:
> 
>>  void sock_zerocopy_put(struct ubuf_info *uarg)
>>  {
>> -       if (uarg && refcount_dec_and_test(&uarg->refcnt))
>> +       if (uarg)
>>                 uarg->callback(uarg, uarg->zerocopy);
>>  }
>>  EXPORT_SYMBOL_GPL(sock_zerocopy_put);
> 
> This does increase the number of indirect function calls. Which are
> not cheap post spectre.
> 
> In the common case for msg_zerocopy we only have two clones, one sent
> out and one on the retransmit queue. So I guess the cost will be
> acceptable.
> 


sock_zerocopy_callback seems to be the only one at the moment - or the
most dominant if I goofed my search. Could use the INDIRECT_CALL macros
for it.