Re: [PATCH] tcp: do not promote SPLICE_F_NONBLOCK to socket O_NONBLOCK

[Evgeniy Polyakov <johnpol@2ka.mipt.ru>]

Hi Octavian.

On Thu, Jul 17, 2008 at 04:33:49PM +0300, Octavian Purdila (opurdila@ixiacom.com) wrote:
>     This patch changes tcp_splice_read to the behavior implied by man 2
>     splice:
>     
>          SPLICE_F_NONBLOCK - Do not block on I/O. This makes the splice
>          pipe operations non-blocking, but splice() may nevertheless block
>          because the file descriptors that are spliced to/from may block
>          (unless they have the O_NONBLOCK flag set).
>     
>     This approach also provides a simple solution to the splice
>     transfer size problem. Say we have the following common sequence:
>     
>          splice(socket, pipe);
>          splice(pipe, file);
>     
>     Unless we specify SPLICE_F_NONBLOCK, we can't use arbitrarily large
>     transfer sizes with the 1st splice since otherwise we will deadlock
>     due to pipe being full.  But if we use SPLICE_F_NONBLOCK, the current
>     implementation will make the underlying socket non-blocking and thus
>     will force us use poll or other async I/O notification mechanism.

Existing behaviour was selected to be able to have a progress if socket
does not have enough data to fill the pipe. With your change if socket
is not opened with non-blocking mode reading will block not matter if
SPLICE_F_NONBLOCK is set or not. This is a quite serious break of the
overall idea behind SPLICE_F_NONBLOCK.

Socket will not be marked as non-blocking if SPLICE_F_NONBLOCK is
specified, only splicing will used non-blocking reading, any read via
recv() will use existing socket flags.

-- 
	Evgeniy Polyakov