net: introduce SO_MAX_PACING_RATE
authorEric Dumazet <edumazet@google.com>
Tue, 24 Sep 2013 15:20:52 +0000 (08:20 -0700)
committerDavid S. Miller <davem@davemloft.net>
Sat, 28 Sep 2013 22:35:41 +0000 (15:35 -0700)
commit62748f32d501f5d3712a7c372bbb92abc7c62bc7
tree847a6e4b66aaf012809ff8656476743e738f1a2b
parent4aa0a03f519812f48ac48d046bc451e97649ec82
net: introduce SO_MAX_PACING_RATE

As mentioned in commit afe4fd062416b ("pkt_sched: fq: Fair Queue packet
scheduler"), this patch adds a new socket option.

SO_MAX_PACING_RATE offers the application the ability to cap the
rate computed by transport layer. Value is in bytes per second.

u32 val = 1000000;
setsockopt(sockfd, SOL_SOCKET, SO_MAX_PACING_RATE, &val, sizeof(val));

To be effectively paced, a flow must use FQ packet scheduler.

Note that a packet scheduler takes into account the headers for its
computations. The effective payload rate depends on MSS and retransmits
if any.

I chose to make this pacing rate a SOL_SOCKET option instead of a
TCP one because this can be used by other protocols.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Steinar H. Gunderson <sesse@google.com>
Cc: Michael Kerrisk <mtk.manpages@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 files changed:
arch/alpha/include/uapi/asm/socket.h
arch/avr32/include/uapi/asm/socket.h
arch/cris/include/uapi/asm/socket.h
arch/frv/include/uapi/asm/socket.h
arch/h8300/include/uapi/asm/socket.h
arch/ia64/include/uapi/asm/socket.h
arch/m32r/include/uapi/asm/socket.h
arch/mips/include/uapi/asm/socket.h
arch/mn10300/include/uapi/asm/socket.h
arch/parisc/include/uapi/asm/socket.h
arch/powerpc/include/uapi/asm/socket.h
arch/s390/include/uapi/asm/socket.h
arch/sparc/include/uapi/asm/socket.h
arch/xtensa/include/uapi/asm/socket.h
include/net/sock.h
include/uapi/asm-generic/socket.h
net/core/sock.c
net/ipv4/tcp_input.c