Re: [PATCH] tcp: splice as many packets as possible at once

[Eric Dumazet <dada1@cosmosbay.com>]

Evgeniy Polyakov a écrit :
> Hi Eric.
> 
> On Sat, Jan 10, 2009 at 08:40:05AM +0100, Eric Dumazet (dada1@cosmosbay.com) wrote:
>>> Not to interrupt the discussion, but for the clarification, that
>>> release_sock/lock_sock is used to process the backlog accumulated while
>>> socket was locked. And while dropping additional pair before the final
>>> release is ok, but moving this itself should be thought of twice.
>>>
>> Hum... I just caught the release_sock(sk)/lock_sock(sk) done in skb_splice_bits()
>>
>> So :
>>
>> 1) the release_sock/lock_sock done in tcp_splice_read() is not necessary
>> to process backlog. Its already done in skb_splice_bits()
> 
> Yes, in the tcp_splice_read() they are added to remove a deadlock.

Could you elaborate ? A deadlock only if !SPLICE_F_NONBLOCK ?

> 
>> 2) If we loop in tcp_read_sock() calling skb_splice_bits() several times
>> then we should perform the following tests inside this loop ?
>>
>> if (sk->sk_err || sk->sk_state == TCP_CLOSE || (sk->sk_shutdown & RCV_SHUTDOWN) ||
>>    signal_pending(current)) break;
>>
>> And removie them from tcp_splice_read() ?
> 
> It could be done, but for what reason? To detect disconnected socket early?
> Does it worth the changes?
> 

I was thinking about the case your thread is doing a splice() from tcp socket to
 a pipe, while another thread is doing the splice from this pipe to something else.

Once patched, tcp_read_sock() could loop a long time...