]> err.no Git - linux-2.6/log
linux-2.6
17 years ago[IP]: Introduce ip_hdrlen()
Arnaldo Carvalho de Melo [Mon, 12 Mar 2007 23:09:15 +0000 (20:09 -0300)]
[IP]: Introduce ip_hdrlen()

For the common sequence "skb->nh.iph->ihl * 4", removing a good number of open
coded skb->nh.iph uses, now to go after the rest...

Just out of curiosity, here are the idioms found to get the same result:

skb->nh.iph->ihl << 2
skb->nh.iph->ihl<<2
skb->nh.iph->ihl * 4
skb->nh.iph->ihl*4
(skb->nh.iph)->ihl * sizeof(u32)

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[SK_BUFF] ipmr: Missed one conversion to skb_network_header()
Arnaldo Carvalho de Melo [Mon, 12 Mar 2007 23:05:39 +0000 (20:05 -0300)]
[SK_BUFF] ipmr: Missed one conversion to skb_network_header()

We can't access skb->nh.raw directly anymore, it will become an offset.

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[NET]: show bound packet types
Stephen Hemminger [Mon, 12 Mar 2007 21:35:37 +0000 (14:35 -0700)]
[NET]: show bound packet types

Show what protocols are bound to what packet types in /proc/net/ptype
Uses kallsyms to decode function pointers if possible.
Example:
Type Device      Function
ALL  eth1     packet_rcv_spkt+0x0
0800          ip_rcv+0x0
0806          arp_rcv+0x0
86dd          :ipv6:ipv6_rcv+0x0

Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[NET]: make seq_operations const
Stephen Hemminger [Mon, 12 Mar 2007 21:34:29 +0000 (14:34 -0700)]
[NET]: make seq_operations const

The seq_file operations stuff can be marked constant to
get it out of dirty cache.

Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[NET]: network dev read_mostly
Stephen Hemminger [Mon, 12 Mar 2007 21:33:50 +0000 (14:33 -0700)]
[NET]: network dev read_mostly

For Eric, mark packet type and network device watermarks
as read mostly.

Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[SK_BUFF]: Introduce skb_set_network_header
Arnaldo Carvalho de Melo [Mon, 12 Mar 2007 01:39:41 +0000 (22:39 -0300)]
[SK_BUFF]: Introduce skb_set_network_header

For the cases where the network header is being set to a offset from skb->data.

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[SK_BUFF] ipmr: Another skb_push related conversion to skb_reset_network_header
Arnaldo Carvalho de Melo [Mon, 12 Mar 2007 01:38:29 +0000 (22:38 -0300)]
[SK_BUFF] ipmr: Another skb_push related conversion to skb_reset_network_header

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[SK_BUFF]: Introduce skb_network_header()
Arnaldo Carvalho de Melo [Wed, 11 Apr 2007 03:50:43 +0000 (20:50 -0700)]
[SK_BUFF]: Introduce skb_network_header()

For the places where we need a pointer to the network header, it is still legal
to touch skb->nh.raw directly if just adding to, subtracting from or setting it
to another layer header.

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[SK_BUFF]: Introduce skb_network_offset()
Arnaldo Carvalho de Melo [Sun, 11 Mar 2007 01:16:10 +0000 (22:16 -0300)]
[SK_BUFF]: Introduce skb_network_offset()

For the quite common 'skb->nh.raw - skb->data' sequence.

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[SK_BUFF] bonding: Set skb->nh.raw relative to skb->mac.raw
Arnaldo Carvalho de Melo [Sat, 10 Mar 2007 23:09:45 +0000 (20:09 -0300)]
[SK_BUFF] bonding: Set skb->nh.raw relative to skb->mac.raw

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[SK_BUFF] xfrm4: use skb_reset_network_header
Arnaldo Carvalho de Melo [Sat, 10 Mar 2007 22:59:16 +0000 (19:59 -0300)]
[SK_BUFF] xfrm4: use skb_reset_network_header

Setting it to skb->h.raw, which is valid, in the (to become) old pointer based
world order and in the new world of offset based layer headers.

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[SK_BUFF] ipv6: More skb_reset_network_header conversions related to skb_pull
Arnaldo Carvalho de Melo [Sat, 10 Mar 2007 22:57:15 +0000 (19:57 -0300)]
[SK_BUFF] ipv6: More skb_reset_network_header conversions related to skb_pull

Now related to this form:

skb->nh.ipv6h = (struct ipv6hdr *)skb_put(skb, length);

That, as the others, is done when skb->tail is still equal to skb->data, making
the conversion to skb_reset_network_header possible.

Also one more case equivalent to skb->nh.raw = skb->data, of this form:

iph = (struct ipv6hdr *)skb->data;
<SNIP>
skb->nh.ipv6h = iph;

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[SK_BUFF]: Use skb_reset_network_header after skb_push
Arnaldo Carvalho de Melo [Sat, 10 Mar 2007 22:40:39 +0000 (19:40 -0300)]
[SK_BUFF]: Use skb_reset_network_header after skb_push

Some more cases where skb->nh.iph was being set that were converted
to using skb_reset_network_header.

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[SK_BUFF] ipconfig: Another conversion to skb_reset_network_header related to skb_put
Arnaldo Carvalho de Melo [Sat, 10 Mar 2007 22:27:27 +0000 (19:27 -0300)]
[SK_BUFF] ipconfig: Another conversion to skb_reset_network_header related to skb_put

boot_pkt->iph is the first member, that is at skb->data, so just use
skb_reset_network_header().

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[SK_BUFF]: Some more skb_put cases converted to skb_reset_network_header
Arnaldo Carvalho de Melo [Sat, 10 Mar 2007 22:15:25 +0000 (19:15 -0300)]
[SK_BUFF]: Some more skb_put cases converted to skb_reset_network_header

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[SK_BUFF]: Some more simple skb_reset_network_header conversions
Arnaldo Carvalho de Melo [Sat, 10 Mar 2007 22:04:55 +0000 (19:04 -0300)]
[SK_BUFF]: Some more simple skb_reset_network_header conversions

This time of the type:

 skb->nh.iph = (struct iphdr *)skb->data;

That is completely equivalent to:

 skb->nh.raw = skb->data;

Wonder why people love casts... :-)

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[SK_BUFF]: Use skb_reset_network_header where the return of __pskb_pull was being...
Arnaldo Carvalho de Melo [Sat, 10 Mar 2007 21:42:03 +0000 (18:42 -0300)]
[SK_BUFF]: Use skb_reset_network_header where the return of __pskb_pull was being used

It returns skb->data, so we can just use skb_reset_network_header after it.

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[SK_BUFF]: Use skb_reset_network_header where the skb_pull return was being used
Arnaldo Carvalho de Melo [Sat, 10 Mar 2007 21:40:59 +0000 (18:40 -0300)]
[SK_BUFF]: Use skb_reset_network_header where the skb_pull return was being used

But only in the cases where its a newly allocated skb, i.e. one where skb->tail
is equal to skb->data, or just after skb_reserve, where this requirement is
maintained.

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[SK_BUFF]: Use skb_reset_network_header in skb_push cases
Arnaldo Carvalho de Melo [Wed, 11 Apr 2007 03:46:21 +0000 (20:46 -0700)]
[SK_BUFF]: Use skb_reset_network_header in skb_push cases

skb_push updates and returns skb->data, so we can just call
skb_reset_network_header after the call to skb_push.

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[SK_BUFF]: Introduce skb_reset_network_header(skb)
Arnaldo Carvalho de Melo [Wed, 11 Apr 2007 03:45:18 +0000 (20:45 -0700)]
[SK_BUFF]: Introduce skb_reset_network_header(skb)

For the common, open coded 'skb->nh.raw = skb->data' operation, so that we can
later turn skb->nh.raw into a offset, reducing the size of struct sk_buff in
64bit land while possibly keeping it as a pointer on 32bit.

This one touches just the most simple case, next will handle the slightly more
"complex" cases.

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[IPV6]: Use skb->nh.ipv6h instead of casting skb->nh.raw
Arnaldo Carvalho de Melo [Sat, 10 Mar 2007 19:21:45 +0000 (16:21 -0300)]
[IPV6]: Use skb->nh.ipv6h instead of casting skb->nh.raw

nh.ipv6h is there exactly for this reason! Use it while it exists ;-)

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[BONDING]: Introduce arp_pkt()
Arnaldo Carvalho de Melo [Sat, 10 Mar 2007 19:07:19 +0000 (16:07 -0300)]
[BONDING]: Introduce arp_pkt()

For consistency with all the other skb->nh.raw accessors.

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[PPPOE]: Introduce pppoe_hdr()
Arnaldo Carvalho de Melo [Sat, 10 Mar 2007 18:56:08 +0000 (15:56 -0300)]
[PPPOE]: Introduce pppoe_hdr()

For consistency with all the other skb->nh.raw accessors.

Also do some really obvious simplifications in pppoe_recvmsg, well the
kfree_skb one is not so obvious, but free() and kfree() have the same behaviour
(hint :-) ).

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[LLC]: Kill llc_set_pdu_hdr
Arnaldo Carvalho de Melo [Sat, 10 Mar 2007 18:34:36 +0000 (15:34 -0300)]
[LLC]: Kill llc_set_pdu_hdr

We'll have skb_reset_network_header soon.

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[SK_BUFF]: Introduce skb_mac_header()
Arnaldo Carvalho de Melo [Mon, 19 Mar 2007 22:33:04 +0000 (15:33 -0700)]
[SK_BUFF]: Introduce skb_mac_header()

For the places where we need a pointer to the mac header, it is still legal to
touch skb->mac.raw directly if just adding to, subtracting from or setting it
to another layer header.

This one also converts some more cases to skb_reset_mac_header() that my
regex missed as it had no spaces before nor after '=', ugh.

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[TCP]: Use skb_set_mac_header in tcp_collapse
Arnaldo Carvalho de Melo [Sat, 10 Mar 2007 15:48:37 +0000 (12:48 -0300)]
[TCP]: Use skb_set_mac_header in tcp_collapse

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[TCP]: Do the layer header setting in tcp_collapse relative to skb->data
Arnaldo Carvalho de Melo [Sat, 10 Mar 2007 15:47:22 +0000 (12:47 -0300)]
[TCP]: Do the layer header setting in tcp_collapse relative to skb->data

That is equal to skb->head before skb_reserve, to help in the layer header
changes.

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[SK_BUFF] xfrm: Use skb_set_mac_header in the memmove cases
Arnaldo Carvalho de Melo [Sat, 10 Mar 2007 15:40:27 +0000 (12:40 -0300)]
[SK_BUFF] xfrm: Use skb_set_mac_header in the memmove cases

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[SK_BUFF]: Introduce skb_set_mac_header()
Arnaldo Carvalho de Melo [Sat, 10 Mar 2007 15:30:58 +0000 (12:30 -0300)]
[SK_BUFF]: Introduce skb_set_mac_header()

For the cases where we want to set skb->mac.raw to an offset from skb->data.

Simple cases first, the memmove ones and specially pktgen will be left for later.

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[LLC]: Use skb_reset_mac_header in llc_mac_hdr_init
Arnaldo Carvalho de Melo [Sat, 10 Mar 2007 15:17:29 +0000 (12:17 -0300)]
[LLC]: Use skb_reset_mac_header in llc_mac_hdr_init

skb_push updates and returns skb->data, so we can just call
skb_reset_mac_header after the call to skb_push.

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[LLC]: Use skb_reset_mac_header in llc_alloc_frame
Arnaldo Carvalho de Melo [Sat, 10 Mar 2007 15:14:56 +0000 (12:14 -0300)]
[LLC]: Use skb_reset_mac_header in llc_alloc_frame

skb->head is equal to skb->data after alloc_skb, so reset the mac header while
this is true, i.e. before skb_reserve.

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[SK_BUFF]: Introduce skb_reset_mac_header(skb)
Arnaldo Carvalho de Melo [Mon, 19 Mar 2007 22:30:44 +0000 (15:30 -0700)]
[SK_BUFF]: Introduce skb_reset_mac_header(skb)

For the common, open coded 'skb->mac.raw = skb->data' operation, so that we can
later turn skb->mac.raw into a offset, reducing the size of struct sk_buff in
64bit land while possibly keeping it as a pointer on 32bit.

This one touches just the most simple case, next will handle the slightly more
"complex" cases.

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[ETH]: Make eth_type_trans set skb->dev like the other *_type_trans
Arnaldo Carvalho de Melo [Thu, 26 Apr 2007 00:40:23 +0000 (17:40 -0700)]
[ETH]: Make eth_type_trans set skb->dev like the other *_type_trans

One less thing for drivers writers to worry about.

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[AOE]: Introduce aoe_hdr()
Arnaldo Carvalho de Melo [Sat, 10 Mar 2007 14:20:07 +0000 (11:20 -0300)]
[AOE]: Introduce aoe_hdr()

For consistency with other skb->mac.raw users.

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[QETH]: Use eth_hdr()
Arnaldo Carvalho de Melo [Sat, 10 Mar 2007 14:13:59 +0000 (11:13 -0300)]
[QETH]: Use eth_hdr()

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[HIPPI/FDDI]: Make {hippi,fddi}_type_trans set skb->dev
Arnaldo Carvalho de Melo [Sat, 10 Mar 2007 13:57:13 +0000 (10:57 -0300)]
[HIPPI/FDDI]: Make {hippi,fddi}_type_trans set skb->dev

Now all the _type_trans routines are consistent in this regard.

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[TR]: Make tr_type_trans set skb->dev
Arnaldo Carvalho de Melo [Mon, 19 Mar 2007 22:29:16 +0000 (15:29 -0700)]
[TR]: Make tr_type_trans set skb->dev

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[TR]: Use tr_hdr() were appropriate
Arnaldo Carvalho de Melo [Mon, 19 Mar 2007 22:27:07 +0000 (15:27 -0700)]
[TR]: Use tr_hdr() were appropriate

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[SOCKET]: Export __sock_recv_timestamp
Arnaldo Carvalho de Melo [Sat, 10 Mar 2007 03:39:35 +0000 (00:39 -0300)]
[SOCKET]: Export __sock_recv_timestamp

Kernel: arch/x86_64/boot/bzImage is ready  (#2)
  MODPOST 1816 modules
WARNING: "__sock_recv_timestamp" [net/sctp/sctp.ko] undefined!
WARNING: "__sock_recv_timestamp" [net/packet/af_packet.ko] undefined!
WARNING: "__sock_recv_timestamp" [net/key/af_key.ko] undefined!
WARNING: "__sock_recv_timestamp" [net/ipv6/ipv6.ko] undefined!
WARNING: "__sock_recv_timestamp" [net/atm/atm.ko] undefined!
make[2]: *** [__modpost] Error 1
make[1]: *** [modules] Error 2
make: *** [_all] Error 2

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[NET]: Adding SO_TIMESTAMPNS / SCM_TIMESTAMPNS support
Eric Dumazet [Mon, 26 Mar 2007 05:14:49 +0000 (22:14 -0700)]
[NET]: Adding SO_TIMESTAMPNS / SCM_TIMESTAMPNS support

Now that network timestamps use ktime_t infrastructure, we can add a new
SOL_SOCKET sockopt  SO_TIMESTAMPNS.

This command is similar to SO_TIMESTAMP, but permits transmission of
a 'timespec struct' instead of a 'timeval struct' control message.
(nanosecond resolution instead of microsecond)

Control message is labelled SCM_TIMESTAMPNS instead of SCM_TIMESTAMP

A socket cannot mix SO_TIMESTAMP and SO_TIMESTAMPNS : the two modes are
mutually exclusive.

sock_recv_timestamp() became too big to be fully inlined so I added a
__sock_recv_timestamp() helper function.

Signed-off-by: Eric Dumazet <dada1@cosmosbay.com>
CC: linux-arch@vger.kernel.org
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[UDP]: Use __skb_pull since we have checked it won't fail with pskb_may_pull
Arnaldo Carvalho de Melo [Fri, 9 Mar 2007 21:51:54 +0000 (13:51 -0800)]
[UDP]: Use __skb_pull since we have checked it won't fail with pskb_may_pull

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[NET]: New sysctls should use __read_mostly tags
Eric Dumazet [Fri, 9 Mar 2007 06:36:37 +0000 (22:36 -0800)]
[NET]: New sysctls should use __read_mostly tags

net_msg_warn should be placed in the read_mostly section, to avoid
performance problems on SMP

Signed-off-by: Eric Dumazet <dada1@cosmosbay.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[IPV6]: Ensure to truncate result and return full length for sticky options.
YOSHIFUJI Hideaki [Fri, 9 Mar 2007 04:48:23 +0000 (20:48 -0800)]
[IPV6]: Ensure to truncate result and return full length for sticky options.

Bug noticed by Chris Wright <chrisw@sous-sol.org>.

Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[IPV6]: Return correct result for sticky options.
YOSHIFUJI Hideaki [Mon, 19 Mar 2007 00:35:57 +0000 (17:35 -0700)]
[IPV6]: Return correct result for sticky options.

We returned incorrect result with IPV6_RTHDRDSTOPTS, IPV6_RTHDR and
IPV6_DSTOPTS.

Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[UDP]: deinline
Stephen Hemminger [Fri, 9 Mar 2007 04:46:41 +0000 (20:46 -0800)]
[UDP]: deinline

A couple of functions are exported or used indirectly
so it is pointless to mark them as inline.

Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[NET]: deinline some functions
Stephen Hemminger [Fri, 9 Mar 2007 04:46:03 +0000 (20:46 -0800)]
[NET]: deinline some functions

Several functions are marked inline or forced inline, but it
would be better to let the compiler decide.

Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[TCP]: whitespace cleanup
Stephen Hemminger [Fri, 9 Mar 2007 04:45:19 +0000 (20:45 -0800)]
[TCP]: whitespace cleanup

Add whitespace around keywords.

Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[IPV4]: cleanup
Stephen Hemminger [Fri, 9 Mar 2007 04:44:43 +0000 (20:44 -0800)]
[IPV4]: cleanup

Add whitespace around keywords.

Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[WIRELESS]: use ARRAY_SIZE()
Stephen Hemminger [Fri, 9 Mar 2007 04:43:49 +0000 (20:43 -0800)]
[WIRELESS]: use ARRAY_SIZE()

Use ARRAY_SIZE() macro now.

Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[NET] core: whitespace cleanup
Stephen Hemminger [Wed, 11 Apr 2007 03:10:33 +0000 (20:10 -0700)]
[NET] core: whitespace cleanup

Fix whitespace around keywords. Fix indentation especially of switch
statements.

Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[UDP]: ipv6 style cleanup
Stephen Hemminger [Fri, 9 Mar 2007 04:42:35 +0000 (20:42 -0800)]
[UDP]: ipv6 style cleanup

Fix whitespace around keywords. Eliminate unnecessary ()'s on return
statements.

Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[UDP]: ipv4 whitespace cleanup
Stephen Hemminger [Fri, 9 Mar 2007 04:41:55 +0000 (20:41 -0800)]
[UDP]: ipv4 whitespace cleanup

Fix whitespace around keywords.

Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[NET]: Replace CONFIG_NET_DEBUG with sysctl.
Stephen Hemminger [Fri, 9 Mar 2007 04:41:08 +0000 (20:41 -0800)]
[NET]: Replace CONFIG_NET_DEBUG with sysctl.

Covert network warning messages from a compile time to runtime choice.
Removes kernel config option and replaces it with new /proc/sys/net/core/warnings.

Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[NET]: Introduce SIOCGSTAMPNS ioctl to get timestamps with nanosec resolution
Eric Dumazet [Mon, 19 Mar 2007 00:33:16 +0000 (17:33 -0700)]
[NET]: Introduce SIOCGSTAMPNS ioctl to get timestamps with nanosec resolution

Now network timestamps use ktime_t infrastructure, we can add a new
ioctl() SIOCGSTAMPNS command to get timestamps in 'struct timespec'.
User programs can thus access to nanosecond resolution.

Signed-off-by: Eric Dumazet <dada1@cosmosbay.com>
CC: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[TCP/DCCP/RANDOM]: Remove unused exports.
Adrian Bunk [Thu, 8 Mar 2007 03:33:52 +0000 (19:33 -0800)]
[TCP/DCCP/RANDOM]: Remove unused exports.

This patch removes the following not or no longer used exports:
- drivers/char/random.c: secure_tcp_sequence_number
- net/dccp/options.c: sysctl_dccp_feat_sequence_window
- net/netlink/af_netlink.c: netlink_set_err

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[TCP]: Abstract out all write queue operations.
David S. Miller [Wed, 7 Mar 2007 20:12:44 +0000 (12:12 -0800)]
[TCP]: Abstract out all write queue operations.

This allows the write queue implementation to be changed,
for example, to one which allows fast interval searching.

Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[NET] TIPC: Use htons() where appropriate.
YOSHIFUJI Hideaki [Wed, 7 Mar 2007 05:21:31 +0000 (14:21 +0900)]
[NET] TIPC: Use htons() where appropriate.

Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[NET] SCHED: Use htons() where appropriate.
YOSHIFUJI Hideaki [Wed, 7 Mar 2007 05:21:20 +0000 (14:21 +0900)]
[NET] SCHED: Use htons() where appropriate.

Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[NET] NETFILTER: Use htonl() where appropriate.
YOSHIFUJI Hideaki [Wed, 7 Mar 2007 05:21:00 +0000 (14:21 +0900)]
[NET] NETFILTER: Use htonl() where appropriate.

Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[NET] IPV4: Use hton{s,l}() where appropriate.
YOSHIFUJI Hideaki [Wed, 7 Mar 2007 05:19:10 +0000 (14:19 +0900)]
[NET] IPV4: Use hton{s,l}() where appropriate.

Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[NET] IEEE80211: Use htons() where appropriate.
YOSHIFUJI Hideaki [Wed, 7 Mar 2007 05:19:05 +0000 (14:19 +0900)]
[NET] IEEE80211: Use htons() where appropriate.

Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[NET] ETHERNET: Use htons() where appropriate.
YOSHIFUJI Hideaki [Wed, 7 Mar 2007 05:19:03 +0000 (14:19 +0900)]
[NET] ETHERNET: Use htons() where appropriate.

Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[NET] CORE: Use htons() where appropriate.
YOSHIFUJI Hideaki [Mon, 26 Mar 2007 03:13:04 +0000 (20:13 -0700)]
[NET] CORE: Use htons() where appropriate.

Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[NET] BLUETOOTH: Use cpu_to_le{16,32}() where appropriate.
YOSHIFUJI Hideaki [Mon, 26 Mar 2007 03:12:50 +0000 (20:12 -0700)]
[NET] BLUETOOTH: Use cpu_to_le{16,32}() where appropriate.

Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[NET] ATM: Use htons() where appropriate.
YOSHIFUJI Hideaki [Mon, 26 Mar 2007 03:12:32 +0000 (20:12 -0700)]
[NET] ATM: Use htons() where appropriate.

Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[NET] 8021Q: Use htons() where appropriate.
YOSHIFUJI Hideaki [Mon, 26 Mar 2007 03:12:18 +0000 (20:12 -0700)]
[NET] 8021Q: Use htons() where appropriate.

Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[NET] 802: Use hton{s,l}() where appropriate.
YOSHIFUJI Hideaki [Mon, 26 Mar 2007 03:11:55 +0000 (20:11 -0700)]
[NET] 802: Use hton{s,l}() where appropriate.

Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[UDP]: Clean up UDP-Lite receive checksum
Herbert Xu [Mon, 26 Mar 2007 03:10:56 +0000 (20:10 -0700)]
[UDP]: Clean up UDP-Lite receive checksum

This patch eliminates some duplicate code for the verification of
receive checksums between UDP-Lite and UDP.  It does this by
introducing __skb_checksum_complete_head which is identical to
__skb_checksum_complete_head apart from the fact that it takes
a length parameter rather than computing the first skb->len bytes.

As a result UDP-Lite will be able to use hardware checksum offload
for packets which do not use partial coverage checksums.  It also
means that UDP-Lite loopback no longer does unnecessary checksum
verification.

If any NICs start support UDP-Lite this would also start working
automatically.

This patch removes the assumption that msg_flags has MSG_TRUNC clear
upon entry in recvmsg.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[UDP6]: Restore sk_filter optimisation
Herbert Xu [Wed, 7 Mar 2007 04:29:58 +0000 (20:29 -0800)]
[UDP6]: Restore sk_filter optimisation

This reverts the changeset

    [IPV6]: UDPv6 checksum.

    We always need to check UDPv6 checksum because it is mandatory.

The sk_filter optimisation has nothing to do whether we verify the
checksum.  It simply postpones it to the point when the user calls
recv or poll.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[IPV4]: Optimize inet_getpeer()
Eric Dumazet [Wed, 7 Mar 2007 04:23:10 +0000 (20:23 -0800)]
[IPV4]: Optimize inet_getpeer()

1) Some sysctl vars are declared __read_mostly

2) We can avoid updating stack[] when doing an AVL lookup only.

    lookup() macro is extended to receive a second parameter, that may be NULL
in case of a pure lookup (no need to save the AVL path). This removes
unnecessary instructions, because compiler knows if this _stack parameter is
NULL or not.

    text size of net/ipv4/inetpeer.o is 2063 bytes instead of 2107 on x86_64

Signed-off-by: Eric Dumazet <dada1@cosmosbay.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[TCP] TCP Yeah: cleanup
Stephen Hemminger [Wed, 7 Mar 2007 04:21:20 +0000 (20:21 -0800)]
[TCP] TCP Yeah: cleanup

Eliminate need for full 6/4/64 divide to compute queue.
Variable maxqueue was really a constant.
Fix indentation.

Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[TCP] tcp_cubic: faster cube root
Stephen Hemminger [Mon, 26 Mar 2007 03:21:15 +0000 (20:21 -0700)]
[TCP] tcp_cubic: faster cube root

The Newton-Raphson method is quadratically convergent so
only a small fixed number of steps are necessary.
Therefore it is faster to unroll the loop. Since div64_64 is no longer
inline it won't cause code explosion.

Also fixes a bug that can occur if x^2 was bigger than 32 bits.

Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[ATM] ENI: Convert to struct timeval to ktime_t.
YOSHIFUJI Hideaki [Wed, 7 Mar 2007 04:19:26 +0000 (20:19 -0800)]
[ATM] ENI: Convert to struct timeval to ktime_t.

Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[NETLINK]: Limit NLMSG_GOODSIZE to 8K.
David S. Miller [Mon, 26 Mar 2007 03:27:59 +0000 (20:27 -0700)]
[NETLINK]: Limit NLMSG_GOODSIZE to 8K.

Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[IPV6] ADDRCONF: Fix possible inet6_ifaddr leakage with CONFIG_OPTIMISTIC_DAD.
YOSHIFUJI Hideaki [Wed, 28 Feb 2007 14:13:20 +0000 (23:13 +0900)]
[IPV6] ADDRCONF: Fix possible inet6_ifaddr leakage with CONFIG_OPTIMISTIC_DAD.

The inet6_ifaddr for source address of RS is leaked if the address
is not an optimistic address.

Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[IPV6] ADDRCONF: Optimistic Duplicate Address Detection (RFC 4429) Support.
Neil Horman [Thu, 26 Apr 2007 00:08:10 +0000 (17:08 -0700)]
[IPV6] ADDRCONF: Optimistic Duplicate Address Detection (RFC 4429) Support.

Nominally an autoconfigured IPv6 address is added to an interface in the
Tentative state (as per RFC 2462).  Addresses in this state remain in this
state while the Duplicate Address Detection process operates on them to
determine their uniqueness on the network.  During this period, these
tentative addresses may not be used for communication, increasing the time
before a node may be able to communicate on a network.  Using Optimistic
Duplicate Address Detection, autoconfigured addresses may be used
immediately for communication on the network, as long as certain rules are
followed to avoid conflicts with other nodes during the Duplicate Address
Detection process.

Signed-off-by: Neil Horman <nhorman@tuxdriver.com>
Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[IPV6] IP6TUNNEL: Enable to control the handled inner protocol.
Yasuyuki Kozakai [Thu, 30 Nov 2006 05:43:28 +0000 (14:43 +0900)]
[IPV6] IP6TUNNEL: Enable to control the handled inner protocol.

ip6_tunnel before supporting IPv4/IPv6 tunnel allows only IPPROTO_IPV6
in configurations from userland. This allows userland to set IPPROTO_IPIP
and 0(wildcard). ip6_tunnel only handles allowed inner protocols.

Signed-off-by: Yasuyuki Kozakai <yasuyuki.kozakai@toshiba.co.jp>
Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[IPV6] IP6TUNNEL: Rename functions ip6ip6_* to ip6_tnl_*.
Yasuyuki Kozakai [Fri, 9 Feb 2007 15:30:33 +0000 (00:30 +0900)]
[IPV6] IP6TUNNEL: Rename functions ip6ip6_* to ip6_tnl_*.

Signed-off-by: Yasuyuki Kozakai <yasuyuki.kozakai@toshiba.co.jp>
Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[IPV6] IP6TUNNEL: Add support to IPv4 over IPv6 tunnel.
Yasuyuki Kozakai [Wed, 14 Feb 2007 15:43:16 +0000 (00:43 +0900)]
[IPV6] IP6TUNNEL: Add support to IPv4 over IPv6 tunnel.

Some notes
- Protocol number IPPROTO_IPIP is used for IPv4 over IPv6 packets.
- If IP6_TNL_F_USE_ORIG_TCLASS is set, TOS in IPv4 header is copied to
  Traffic Class in outer IPv6 header on xmit.
- IP6_TNL_F_USE_ORIG_FLOWLABEL is ignored on xmit of IPv4 packets, because
  IPv4 header does not have flow label.
- Kernel sends ICMP error if IPv4 packet is too big on xmit, even if
  DF flag is not set.

Signed-off-by: Yasuyuki Kozakai <yasuyuki.kozakai@toshiba.co.jp>
Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[IPV6] IP6TUNNEL: Split out generic routine in ip6ip6_xmit().
Yasuyuki Kozakai [Sun, 5 Nov 2006 13:56:45 +0000 (22:56 +0900)]
[IPV6] IP6TUNNEL: Split out generic routine in ip6ip6_xmit().

This enables to add IPv4/IPv6 specific handling later,

Signed-off-by: Yasuyuki Kozakai <yasuyuki.kozakai@toshiba.co.jp>
Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[IPV6] IP6TUNNEL: Split out generic routine in ip6ip6_rcv().
Yasuyuki Kozakai [Fri, 3 Nov 2006 00:39:14 +0000 (09:39 +0900)]
[IPV6] IP6TUNNEL: Split out generic routine in ip6ip6_rcv().

This enables to add IPv4/IPv6 specific handling later,

Signed-off-by: Yasuyuki Kozakai <yasuyuki.kozakai@toshiba.co.jp>
Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[IPV6] IP6TUNNEL: Split out generic routine in ip6ip6_err().
Yasuyuki Kozakai [Tue, 31 Oct 2006 14:11:25 +0000 (23:11 +0900)]
[IPV6] IP6TUNNEL: Split out generic routine in ip6ip6_err().

This enables to add IPv4/IPv6 specific error handling later,

Signed-off-by: Yasuyuki Kozakai <yasuyuki.kozakai@toshiba.co.jp>
Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[IPV6]: Decentralize EXPORT_SYMBOLs.
YOSHIFUJI Hideaki [Thu, 22 Feb 2007 13:05:40 +0000 (22:05 +0900)]
[IPV6]: Decentralize EXPORT_SYMBOLs.

Signed-off-by: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org>
17 years ago[NETLINK]: Mirror UDP MSG_TRUNC semantics.
David S. Miller [Wed, 7 Mar 2007 01:02:35 +0000 (17:02 -0800)]
[NETLINK]: Mirror UDP MSG_TRUNC semantics.

If the user passes MSG_TRUNC in via msg_flags, return
the full packet size not the truncated size.

Idea from Herbert Xu and Thomas Graf.

Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[NET]: convert network timestamps to ktime_t
Eric Dumazet [Thu, 19 Apr 2007 23:16:32 +0000 (16:16 -0700)]
[NET]: convert network timestamps to ktime_t

We currently use a special structure (struct skb_timeval) and plain
'struct timeval' to store packet timestamps in sk_buffs and struct
sock.

This has some drawbacks :
- Fixed resolution of micro second.
- Waste of space on 64bit platforms where sizeof(struct timeval)=16

I suggest using ktime_t that is a nice abstraction of high resolution
time services, currently capable of nanosecond resolution.

As sizeof(ktime_t) is 8 bytes, using ktime_t in 'struct sock' permits
a 8 byte shrink of this structure on 64bit architectures. Some other
structures also benefit from this size reduction (struct ipq in
ipv4/ip_fragment.c, struct frag_queue in ipv6/reassembly.c, ...)

Once this ktime infrastructure adopted, we can more easily provide
nanosecond resolution on top of it. (ioctl SIOCGSTAMPNS and/or
SO_TIMESTAMPNS/SCM_TIMESTAMPNS)

Note : this patch includes a bug correction in
compat_sock_get_timestamp() where a "err = 0;" was missing (so this
syscall returned -ENOENT instead of 0)

Signed-off-by: Eric Dumazet <dada1@cosmosbay.com>
CC: Stephen Hemminger <shemminger@linux-foundation.org>
CC: John find <linux.kernel@free.fr>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[NET]: div64_64 consolidate (rev3)
Stephen Hemminger [Mon, 26 Mar 2007 02:54:23 +0000 (19:54 -0700)]
[NET]: div64_64 consolidate (rev3)

Here is the current version of the 64 bit divide common code.

Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[NET]: Convert xtime.tv_sec to get_seconds()
James Morris [Mon, 5 Mar 2007 00:12:44 +0000 (16:12 -0800)]
[NET]: Convert xtime.tv_sec to get_seconds()

Where appropriate, convert references to xtime.tv_sec to the
get_seconds() helper function.

Signed-off-by: James Morris <jmorris@namei.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[PKTGEN]: fix device name handling
Stephen Hemminger [Mon, 5 Mar 2007 00:11:51 +0000 (16:11 -0800)]
[PKTGEN]: fix device name handling

Since devices can change name and other wierdness, don't hold onto
a copy of device name, instead use pointer to output device.

Fix a couple of leaks in error handling path as well.

Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: Robert Olsson <robert.olsson@its.uu.se>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[PKTGEN]: don't use __constant_htonl()
Stephen Hemminger [Mon, 5 Mar 2007 00:08:08 +0000 (16:08 -0800)]
[PKTGEN]: don't use __constant_htonl()

The existing htonl() macro is smart enough to do the same code as
using __constant_htonl() and it looks cleaner.

Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: Robert Olsson <robert.olsson@its.uu.se>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[PKTGEN]: use random32
Stephen Hemminger [Mon, 5 Mar 2007 00:07:28 +0000 (16:07 -0800)]
[PKTGEN]: use random32

Can use random32() now.

Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: Robert Olsson <robert.olsson@its.uu.se>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[PKTGEN]: use pr_debug
Stephen Hemminger [Mon, 5 Mar 2007 00:06:47 +0000 (16:06 -0800)]
[PKTGEN]: use pr_debug

Remove private debug macro and replace with standard version

Signed-off-by: Stephen Hemminger <shemminger@linux-foundation.org>
Signed-off-by: Robert Olsson <robert.olsson@its.uu.se>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[NET]: Keep sk_backlog near sk_lock
Eric Dumazet [Mon, 5 Mar 2007 00:05:44 +0000 (16:05 -0800)]
[NET]: Keep sk_backlog near sk_lock

sk_backlog is a critical field of struct sock. (known famous words)

It is (ab)used in hot paths, in particular in release_sock(), tcp_recvmsg(),
tcp_v4_rcv(), sk_receive_skb().

It really makes sense to place it next to sk_lock, because sk_backlog is only
used after sk_lock locked (and thus memory cache line in L1 cache). This
should reduce cache misses and sk_lock acquisition time.

(In theory, we could only move the head pointer near sk_lock, and leaving tail
far away, because 'tail' is normally not so hot, but keep it simple :) )

Signed-off-by: Eric Dumazet <dada1@cosmosbay.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[TCP]: FRTO undo response falls back to ratehalving one if ECEd
Ilpo Järvinen [Fri, 2 Mar 2007 21:34:19 +0000 (13:34 -0800)]
[TCP]: FRTO undo response falls back to ratehalving one if ECEd

Undoing ssthresh is disabled in fastretrans_alert whenever
FLAG_ECE is set by clearing prior_ssthresh. The clearing does
not protect FRTO because FRTO operates before fastretrans_alert.
Moving the clearing of prior_ssthresh earlier seems to be a
suboptimal solution to the FRTO case because then FLAG_ECE will
cause a second ssthresh reduction in try_to_open (the first
occurred when FRTO was entered). So instead, FRTO falls back
immediately to the rate halving response, which switches TCP to
CA_CWR state preventing the latter reduction of ssthresh.

If the first ECE arrived before the ACK after which FRTO is able
to decide RTO as spurious, prior_ssthresh is already cleared.
Thus no undoing for ssthresh occurs. Besides, FLAG_ECE should be
set also in the following ACKs resulting in rate halving response
that sees TCP is already in CA_CWR, which again prevents an extra
ssthresh reduction on that round-trip.

If the first ECE arrived before RTO, ssthresh has already been
adapted and prior_ssthresh remains cleared on entry because TCP
is in CA_CWR (the same applies also to a case where FRTO is
entered more than once and ECE comes in the middle).

High_seq must not be touched after tcp_enter_cwr because CWR
round-trip calculation depends on it.

I believe that after this patch, FRTO should be ECN-safe and
even able to take advantage of synergy benefits.

Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[TCP]: Complete icsk-to-local-variable change (in tcp_enter_cwr)
Ilpo Järvinen [Fri, 2 Mar 2007 21:27:25 +0000 (13:27 -0800)]
[TCP]: Complete icsk-to-local-variable change (in tcp_enter_cwr)

A local variable for icsk was created but this change was
missing. Spotted by Jarek Poplawski.

Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[TCP] Sysctl documentation: tcp_frto_response
Ilpo Järvinen [Tue, 27 Feb 2007 18:10:55 +0000 (10:10 -0800)]
[TCP] Sysctl documentation: tcp_frto_response

In addition, fixed minor things in tcp_frto sysctl.

Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[TCP]: Add two new spurious RTO responses to FRTO
Ilpo Järvinen [Tue, 27 Feb 2007 18:09:49 +0000 (10:09 -0800)]
[TCP]: Add two new spurious RTO responses to FRTO

New sysctl tcp_frto_response is added to select amongst these
responses:
- Rate halving based; reuses CA_CWR state (default)
- Very conservative; used to be the only one available (=1)
- Undo cwr; undoes ssthresh and cwnd reductions (=2)

The response with rate halving requires a new parameter to
tcp_enter_cwr because FRTO has already reduced ssthresh and
doing a second reduction there has to be prevented. In addition,
to keep things nice on 80 cols screen, a local variable was
added.

Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[TCP]: Correct reordering detection change (no FRTO case)
Ilpo Järvinen [Sat, 24 Feb 2007 00:22:06 +0000 (16:22 -0800)]
[TCP]: Correct reordering detection change (no FRTO case)

The reordering detection must work also when FRTO has not been
used at all which was the original intention of mine, just the
expression of the idea was flawed.

Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[TCP]: Make snd_cwnd_clamp a u32.
David S. Miller [Fri, 23 Feb 2007 06:52:59 +0000 (22:52 -0800)]
[TCP]: Make snd_cwnd_clamp a u32.

Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[TCP]: Keep copied_seq, rcv_wup and rcv_next together.
Eric Dumazet [Thu, 22 Feb 2007 11:20:44 +0000 (03:20 -0800)]
[TCP]: Keep copied_seq, rcv_wup and rcv_next together.

I noticed in oprofile study a cache miss in tcp_rcv_established() to read
copied_seq.

ffffffff80400a80 <tcp_rcv_established>: /* tcp_rcv_established total: 4034293  
2.0400 */

 55493  0.0281 :ffffffff80400bc9:   mov    0x4c8(%r12),%eax copied_seq
543103  0.2746 :ffffffff80400bd1:   cmp    0x3e0(%r12),%eax   rcv_nxt    

if (tp->copied_seq == tp->rcv_nxt &&
        len - tcp_header_len <= tp->ucopy.len) {

In this function, the cache line 0x4c0 -> 0x500 is used only for this
reading 'copied_seq' field.

rcv_wup and copied_seq should be next to rcv_nxt field, to lower number of
active cache lines in hot paths. (tcp_rcv_established(), tcp_poll(), ...)

As you suggested, I changed tcp_create_openreq_child() so that these fields
are changed together, to avoid adding a new store buffer stall.

Patch is 64bit friendly (no new hole because of alignment constraints)

Signed-off-by: Eric Dumazet <dada1@cosmosbay.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
17 years ago[TCP]: struct *sock argument renamed: sp -> sk
Ilpo Järvinen [Thu, 22 Feb 2007 09:13:58 +0000 (01:13 -0800)]
[TCP]: struct *sock argument renamed: sp -> sk

In general, TCP code uses "sk" for struct sock pointer.

Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@helsinki.fi>
Signed-off-by: David S. Miller <davem@davemloft.net>