- added typedef unsigned int __nocast gfp_t;
- replaced __nocast uses for gfp flags with gfp_t - it gives exactly
the same warnings as far as sparse is concerned, doesn't change
generated code (from gcc point of view we replaced unsigned int with
typedef) and documents what's going on far better.
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
Arnaldo and I agreed it could be applied now, because I have other
pending patches depending on this one (Thank you Arnaldo)
(The other important patch moves skc_refcnt in a separate cache line,
so that the SMP/NUMA performance doesnt suffer from cache line ping pongs)
1) First some performance data :
--------------------------------
tcp_v4_rcv() wastes a *lot* of time in __inet_lookup_established()
The most time critical code is :
sk_for_each(sk, node, &head->chain) {
if (INET_MATCH(sk, acookie, saddr, daddr, ports, dif))
goto hit; /* You sunk my battleship! */
}
The sk_for_each() does use prefetch() hints but only the begining of
"struct sock" is prefetched.
As INET_MATCH first comparison uses inet_sk(__sk)->daddr, wich is far
away from the begining of "struct sock", it has to bring into CPU
cache cold cache line. Each iteration has to use at least 2 cache
lines.
This can be problematic if some chains are very long.
2) The goal
-----------
The idea I had is to change things so that INET_MATCH() may return
FALSE in 99% of cases only using the data already in the CPU cache,
using one cache line per iteration.
3) Description of the patch
---------------------------
Adds a new 'unsigned int skc_hash' field in 'struct sock_common',
filling a 32 bits hole on 64 bits platform.
struct sock_common {
unsigned short skc_family;
volatile unsigned char skc_state;
unsigned char skc_reuse;
int skc_bound_dev_if;
struct hlist_node skc_node;
struct hlist_node skc_bind_node;
atomic_t skc_refcnt;
+ unsigned int skc_hash;
struct proto *skc_prot;
};
Store in this 32 bits field the full hash, not masked by (ehash_size -
1) Using this full hash as the first comparison done in INET_MATCH
permits us immediatly skip the element without touching a second cache
line in case of a miss.
Suppress the sk_hashent/tw_hashent fields since skc_hash (aliased to
sk_hash and tw_hash) already contains the slot number if we mask with
(ehash_size - 1)
File include/net/inet_hashtables.h
64 bits platforms :
#define INET_MATCH(__sk, __hash, __cookie, __saddr, __daddr, __ports, __dif)\
(((__sk)->sk_hash == (__hash))
((*((__u64 *)&(inet_sk(__sk)->daddr)))== (__cookie)) && \
((*((__u32 *)&(inet_sk(__sk)->dport))) == (__ports)) && \
(!((__sk)->sk_bound_dev_if) || ((__sk)->sk_bound_dev_if == (__dif))))
32bits platforms:
#define TCP_IPV4_MATCH(__sk, __hash, __cookie, __saddr, __daddr, __ports, __dif)\
(((__sk)->sk_hash == (__hash)) && \
(inet_sk(__sk)->daddr == (__saddr)) && \
(inet_sk(__sk)->rcv_saddr == (__daddr)) && \
(!((__sk)->sk_bound_dev_if) || ((__sk)->sk_bound_dev_if == (__dif))))
- Adds a prefetch(head->chain.first) in
__inet_lookup_established()/__tcp_v4_check_established() and
__inet6_lookup_established()/__tcp_v6_check_established() and
__dccp_v4_check_established() to bring into cache the first element of the
list, before the {read|write}_lock(&head->lock);
Signed-off-by: Eric Dumazet <dada1@cosmosbay.com>
Acked-by: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
Allocation for the optnames is similar to the DCCP options, with a
range for rx and tx half connection CCIDs.
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Moving the TFRC sender and receiver variables to separate structs, so
that we can copy these structs to userspace thru getsockopt,
dccp_diag, etc.
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Isolating it, that will be used when we introduce a CCID2 (TCP-Like)
implementation.
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
As discussed in the dccp@vger mailing list:
Now applications have to use setsockopt(DCCP_SOCKOPT_SERVICE, service[s]),
prior to calling listen() and connect().
An array of unsigned ints can be passed meaning that the listening sock accepts
connection requests for several services.
With this we can ditch struct sockaddr_dccp and use only sockaddr_in (and
sockaddr_in6 in the future).
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Moving the setting of DCCP_SKB_CB(skb)->dccpd_reset_code to the places
where events happen that trigger sending a RESET packet.
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Eliciting a SYNCACK in response, we were handling SYNC packets
only in the DCCP_OPEN state, in dccp_rcv_established.
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
It is possible to receive more than one CLOSEREQ packet if the
CLOSE packet sent in response is somehow lost, change the state
to DCCP_CLOSING only on the first CLOSEREQ packet received.
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Also use some BUG_ON where appropriate and use LIMIT_NETDEBUG for the unlikely
cases where we, at this stage, want to know about, that in my tests hasn't
appeared in the radar.
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
To match more closely what is described in RFC 3448.
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: Ian McDonald <iam4@cs.waikato.ac.nz>
To start the timestamps with 0.0ms, easing the integer maths in the CCIDs, this
probably will be reworked to use the to be introduced struct timeval_offset
infrastructure out of skb_get_timestamp, etc.
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
The initialization of ccid3hcrx_rtt to 5ms is just a bandaid, I'll continue
auditing the CCID3 HC rx codebase to fix this properly, probably I'll add a
feedback timer as suggested in the CCID3 draft.
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
We can get this value in an TIMESTAMP_ECHO and/or in an ELAPSED_TIME option, if
receiving both give precendence to the biggest one.
In my tests they are very close if not equal at all times, so we may well think
about removing the code in CCID3 that inserts this option and leaving this to
the core, and perhaps even use just TIMESTAMP_ECHO including the elapsed time.
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
This makes the send rate calculations behave way more closely to what
is specified, with the jitter previously seen on x and x_recv
disappearing completely on non lossy setups.
This resembles the tcp_data_snd_check code, that possibly we'll end up
using in DCCP as well, perhaps moving this code to
inet_connection_sock.
For now I'm doing the simplest implementation tho.
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
So that applications can set dccp_sock->dccps_pkt_size, that in turn
is used in the CCID3 half connection init routines to set
ccid3hc[tr]x_s and use it in its rate calculations.
Signed-off-by: Arnaldo Carvalho de Melo <acme@mandriva.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Al Viro
Eric Dumazet
Arnaldo Carvalho de Melo