Gerrit Renker [Mon, 17 Dec 2007 12:25:06 +0000 (10:25 -0200)]
[CCID3]: Remove two irrelevant states in TX feedback handling
* the NO_SENT state is only triggered in bidirectional mode,
costing unnecessary processing.
* the TERM (terminating) state is irrelevant.
Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk> Signed-off-by: Ian McDonald <ian.mcdonald@jandi.co.nz> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Gerrit Renker [Mon, 17 Dec 2007 12:07:44 +0000 (10:07 -0200)]
[CCID3]: Use a function to update p_inv, and p is never used
This patch
1) concentrates previously scattered computation of p_inv into one function;
2) removes the `p' element of the CCID3 RX sock (it is redundant);
3) makes the tfrc_rx_info structure standalone, only used on demand.
Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk> Signed-off-by: Ian McDonald <ian.mcdonald@jandi.co.nz> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Vlad Yasevich [Sun, 16 Dec 2007 22:06:41 +0000 (14:06 -0800)]
[SCTP]: Use crc32c library for checksum calculations.
The crc32c library used an identical table and algorithm
as SCTP. Switch to using the library instead of carrying
our own table. Using crypto layer proved to have too
much overhead compared to using the library directly.
Signed-off-by: Vlad Yasevich <vladislav.yasevich@hp.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Herbert Xu [Sun, 16 Dec 2007 22:04:02 +0000 (14:04 -0800)]
[PACKET]: Fix /proc/net/packet crash due to bogus private pointer
The seq_open_net patch changed the meaning of seq->private.
Unfortunately it missed two spots in AF_PACKET, which still
used the old way of dereferencing seq->private, thus causing
weird and wonderful crashes when reading /proc/net/packet.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
Pavel Emelyanov [Sun, 16 Dec 2007 21:32:48 +0000 (13:32 -0800)]
[IPV4]: Switch users of ipv4_devconf(_all) to use the pernet one
These are scattered over the code, but almost all the
"critical" places already have the proper struct net
at hand except for snmp proc showing function and routing
rtnl handler.
Signed-off-by: Pavel Emelyanov <xemul@openvz.org> Signed-off-by: David S. Miller <davem@davemloft.net>
Pavel Emelyanov [Sun, 16 Dec 2007 21:31:14 +0000 (13:31 -0800)]
[IPV4]: Store the net pointer on devinet's ctl tables
Some handers and strategies of devinet sysctl tables need
to know the net to propagate the ctl change to all the
net devices.
I use the (currently unused) extra2 pointer on the tables
to get it.
Holding the reference on the struct net is not possible,
because otherwise we'll get a net->ctl_table->net circular
dependency. But since the ctl tables are unregistered during
the net destruction, this is safe to get it w/o additional
protection.
Signed-off-by: Pavel Emelyanov <xemul@openvz.org> Signed-off-by: David S. Miller <davem@davemloft.net>
Pavel Emelyanov [Sun, 16 Dec 2007 21:30:39 +0000 (13:30 -0800)]
[IPV4]: Pass the net pointer to the arp_req_set_proxy()
This one will need to set the IPV4_DEVCONF_ALL(PROXY_ARP), but
there's no ways to get the net right in place, so we have to
pull one from the inet_ioctl's struct sock.
Signed-off-by: Pavel Emelyanov <xemul@openvz.org> Signed-off-by: David S. Miller <davem@davemloft.net>
Pavel Emelyanov [Sun, 16 Dec 2007 21:30:07 +0000 (13:30 -0800)]
[IPV4]: Make __devinet_sysctl_register return an error
Currently, this function is void, so failures in creating
sysctls for new/renamed devices are not reported to anywhere.
Fixing this is another complex (needed?) task, but this
return value is needed during the namespaces creation to
handle the case, when we failed to create "all" and "default"
entries.
Signed-off-by: Pavel Emelyanov <xemul@openvz.org> Signed-off-by: David S. Miller <davem@davemloft.net>
Pavel Emelyanov [Fri, 14 Dec 2007 19:38:04 +0000 (11:38 -0800)]
[XFRM]: Fix potential race vs xfrm_state(only)_find and xfrm_hash_resize.
The _find calls calculate the hash value using the
xfrm_state_hmask, without the xfrm_state_lock. But the
value of this mask can change in the _resize call under
the state_lock, so we risk to fail in finding the desired
entry in hash.
I think, that the hash value is better to calculate
under the state lock.
Signed-off-by: Pavel Emelyanov <xemul@openvz.org> Signed-off-by: David S. Miller <davem@davemloft.net>
[PPP] synchronous tty: convert dead_sem to completion
PPP synchronous tty channel driver: convert the semaphore dead_sem to a
completion
Signed-off-by: Matthias Kaehlcke <matthias.kaehlcke@gmail.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: David S. Miller <davem@davemloft.net>
Herbert Xu [Fri, 14 Dec 2007 19:25:26 +0000 (11:25 -0800)]
[UDP]: Move udp_stats_in6 into net/ipv4/udp.c
Now that external users may increment the counters directly, we need
to ensure that udp_stats_in6 is always available. Otherwise we'd
either have to requrie the external users to be built as modules or
ipv6 to be built-in.
This isn't too bad because udp_stats_in6 is just a pair of pointers
plus an EXPORT, e.g., just 40 (16 + 24) bytes on x86-64.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
Gerrit Renker [Fri, 14 Dec 2007 01:37:55 +0000 (23:37 -0200)]
[DCCP]: Introducing CCMPS
This introduces a CCMPS field for setting a CCID-specific upper bound on the application payload
size, as is defined in RFC 4340, section 14.
Only the TX CCID is considered in setting this limit, since the RX CCID generates comparatively
small (DCCP-Ack) feedback packets. The CCMPS field includes network and transport layer header
lengths. The only current CCMPS customer is CCID4 (via RFC 4828).
A wrapper is used to allow querying the CCMPS even at times where the CCID modules may not have
been fully negotiated yet.
In dccp_sync_mss() the variable `mss_now' has been renamed into `cur_mps', to reflect that we are
dealing with an MPS, but not an MSS.
Since the DCCP code closely follows the TCP code, the identifiers `dccp_sync_mss' and
`dccps_mss_cache' have been kept, as they have direct TCP counterparts.
Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk> Signed-off-by: Ian McDonald <ian.mcdonald@jandi.co.nz> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
To allow spaces in the name, the slab name string has been changed to
refer to the numeric CCID identifier, using the same format as before.
Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk> Signed-off-by: Ian McDonald <ian.mcdonald@jandi.co.nz> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Gerrit Renker [Fri, 14 Dec 2007 01:31:14 +0000 (23:31 -0200)]
[DCCP]: Documentation for CCID operations
This adds documentation for the ccid_operations structure.
Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk> Signed-off-by: Ian McDonald <ian.mcdonald@jandi.co.nz> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Denis V. Lunev [Thu, 13 Dec 2007 17:47:57 +0000 (09:47 -0800)]
[IPV4]: Thresholds in fib_trie.c are used as consts, so make them const.
There are several thresholds for trie fib hash management. They are used
in the code as a constants. Make them constants from the compiler point of
view.
Signed-off-by: Denis V. Lunev <den@openvz.org> Signed-off-by: David S. Miller <davem@davemloft.net>
Michal Schmidt [Thu, 13 Dec 2007 17:47:00 +0000 (09:47 -0800)]
[IPV6] sit: Rebinding of SIT tunnels to other interfaces
This is similar to the change already done for IPIP tunnels.
Once created, a SIT tunnel can't be bound to another device.
To reproduce:
# create a tunnel:
ip tunnel add tunneltest0 mode sit remote 10.0.0.1 dev eth0
# try to change the bounding device from eth0 to eth1:
ip tunnel change tunneltest0 dev eth1
# show the result:
ip tunnel show tunneltest0
tunneltest0: ipv6/ip remote 10.0.0.1 local any dev eth0 ttl inherit
Notice the bound device has not changed from eth0 to eth1.
This patch fixes it. When changing the binding, it also recalculates the
MTU according to the new bound device's MTU.
Signed-off-by: Michal Schmidt <mschmidt@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Michal Schmidt [Thu, 13 Dec 2007 17:46:32 +0000 (09:46 -0800)]
[IP_GRE]: Rebinding of GRE tunnels to other interfaces
This is similar to the change already done for IPIP tunnels.
Once created, a GRE tunnel can't be bound to another device.
To reproduce:
# create a tunnel:
ip tunnel add tunneltest0 mode gre remote 10.0.0.1 dev eth0
# try to change the bounding device from eth0 to eth1:
ip tunnel change tunneltest0 dev eth1
# show the result:
ip tunnel show tunneltest0
tunneltest0: gre/ip remote 10.0.0.1 local any dev eth0 ttl inherit
Notice the bound device has not changed from eth0 to eth1.
This patch fixes it. When changing the binding, it also recalculates the
MTU according to the new bound device's MTU.
Signed-off-by: Michal Schmidt <mschmidt@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Herbert Xu [Thu, 13 Dec 2007 17:30:59 +0000 (09:30 -0800)]
[IPSEC]: Fix zero return value in xfrm_lookup on error
Further testing shows that my ICMP relookup patch can cause xfrm_lookup
to return zero on error which isn't very nice since it leads to the caller
dying on null pointer dereference. The bug is due to not setting err
to ENOENT just before we leave xfrm_lookup in case of no policy.
This patch moves the err setting to where it should be.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
Gerrit Renker [Thu, 13 Dec 2007 14:48:19 +0000 (12:48 -0200)]
[DCCP]: Ignore feature negotiation on Data packets
This implements [RFC 4340, p. 32]: "any feature negotiation options received
on DCCP-Data packets MUST be ignored".
Also added a FIXME for further processing, since the code currently (wrongly)
classifies empty Confirm options as invalid - this needs to be resolved in
a separate patch.
Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk> Signed-off-by: Ian McDonald <ian.mcdonald@jandi.co.nz> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Gerrit Renker [Thu, 13 Dec 2007 14:41:46 +0000 (12:41 -0200)]
[DCCP]: Make code assumptions explicit
This removes several `XXX' references which indicate a missing support
for non-1-byte feature values: this is unnecessary, as all currently known
(standardised) SP feature values are 1-byte quantities.
Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk> Signed-off-by: Ian McDonald <ian.mcdonald@jandi.co.nz> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Gerrit Renker [Thu, 13 Dec 2007 14:40:40 +0000 (12:40 -0200)]
[DCCP]: Remove unused and redundant validation functions
This removes two inlines which were both called in a single function only:
1) dccp_feat_change() is always called with either DCCPO_CHANGE_L or DCCPO_CHANGE_R as argument
* from dccp_set_socktopt_change() via do_dccp_setsockopt() with DCCP_SOCKOPT_CHANGE_R/L
* from __dccp_feat_init() via dccp_feat_init() also with DCCP_SOCKOPT_CHANGE_R/L.
Hence the dccp_feat_is_valid_type() is completely unnecessary and always returns true.
2) Due to (1), the length test reduces to 'len >= 4', which in turn makes
dccp_feat_is_valid_length() unnecessary.
Furthermore, the inline function dccp_feat_is_reserved() was unfolded,
since only called in a single place.
Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk> Signed-off-by: Ian McDonald <ian.mcdonald@jandi.co.nz> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Gerrit Renker [Thu, 13 Dec 2007 14:38:11 +0000 (12:38 -0200)]
[DCCP]: Support inserting options during the 3-way handshake
This provides a separate routine to insert options during the initial handshake.
The main purpose is to conduct feature negotiation, for the moment the only user
is the timestamp echo needed for the (CCID3) handshake RTT sample.
Padding of options has been put into a small separate routine, to be shared among
the two functions. This could also be used as a generic routine to finish inserting
options.
Also removed an `XXX' comment since its content was obvious.
Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk> Signed-off-by: Ian McDonald <ian.mcdonald@jandi.co.nz> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Gerrit Renker [Thu, 13 Dec 2007 14:37:19 +0000 (12:37 -0200)]
[DCCP]: Handle timestamps on Request/Response exchange separately
In DCCP, timestamps can occur on packets anytime, CCID3 uses a timestamp(/echo) on the Request/Response
exchange. This patch addresses the following situation:
* timestamps are recorded on the listening socket;
* Responses are sent from dccp_request_sockets;
* suppose two connections reach the listening socket with very small time in between:
* the first timestamp value gets overwritten by the second connection request.
This is not really good, so this patch separates timestamps into
* those which are received by the server during the initial handshake (on dccp_request_sock);
* those which are received by the client or the client after connection establishment.
As before, a timestamp of 0 is regarded as indicating that no (meaningful) timestamp has been
received (in addition, a warning message is printed if hosts send 0-valued timestamps).
The timestamp-echoing now works as follows:
* when a timestamp is present on the initial Request, it is placed into dreq, due to the
call to dccp_parse_options in dccp_v{4,6}_conn_request;
* when a timestamp is present on the Ack leading from RESPOND => OPEN, it is copied over
from the request_sock into the child cocket in dccp_create_openreq_child;
* timestamps received on an (established) dccp_sock are treated as before.
Since Elapsed Time is measured in hundredths of milliseconds (13.2), the new dccp_timestamp()
function is used, as it is expected that the time between receiving the timestamp and
sending the timestamp echo will be very small against the wrap-around time. As a byproduct,
this allows smaller timestamping-time fields.
Furthermore, inserting the Timestamp Echo option has been taken out of the block starting with
'!dccp_packet_without_ack()', since Timestamp Echo can be carried on any packet (5.8 and 13.3).
Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk> Acked-by: Ian McDonald <ian.mcdonald@jandi.co.nz> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Gerrit Renker [Thu, 13 Dec 2007 14:31:26 +0000 (12:31 -0200)]
[DCCP]: Add (missing) option parsing to request_sock processing
This adds option-parsing code to processing of Acks in the listening state
on request_socks on the server, serving two purposes
(i) resolves a FIXME (removed);
(ii) paves the way for feature-negotiation during connection-setup.
There is an intended subtlety here with regard to dccp_check_req:
Parsing options happens only after testing whether the received packet is
a retransmitted Request. Otherwise, if the Request contained (a possibly
large number of) feature-negotiation options, recomputing state would have to
happen each time a retransmitted Request arrives, which opens the door to an
easy DoS attack. Since in a genuine retransmission the options should not be
different from the original, reusing the already computed state seems better.
The other point is - if there are timestamp options on the Request, they will
not be answered; which means that in the presence of retransmission (likely
due to loss and/or other problems), the use of Request/Response RTT sampling
is suspended, so that startup problems here do not propagate.
Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk> Signed-off-by: Ian McDonald <ian.mcdonald@jandi.co.nz> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Gerrit Renker [Thu, 13 Dec 2007 14:29:24 +0000 (12:29 -0200)]
[DCCP]: Allow to parse options on Request Sockets
The option parsing code currently only parses on full sk's. This causes a problem for
options sent during the initial handshake (in particular timestamps and feature-negotiation
options). Therefore, this patch extends the option parsing code with an additional argument
for request_socks: if it is non-NULL, options are parsed on the request socket, otherwise
the normal path (parsing on the sk) is used.
Subsequent patches, which implement feature negotiation during connection setup, make use
of this facility.
Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk> Signed-off-by: Ian McDonald <ian.mcdonald@jandi.co.nz> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Gerrit Renker [Thu, 13 Dec 2007 14:27:14 +0000 (12:27 -0200)]
[DCCP]: Collapse repeated `len' statements into one
This replaces 4 individual assignments for `len' with a single
one, placed where the control flow of those 4 leads to.
Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk> Signed-off-by: Ian McDonald <ian.mcdonald@jandi.co.nz> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Gerrit Renker [Thu, 13 Dec 2007 14:25:01 +0000 (12:25 -0200)]
[DCCP]: Support for server holding timewait state
This adds a socket option and signalling support for the case where the server
holds timewait state on closing the connection, as described in RFC 4340, 8.3.
Since holding timewait state at the server is the non-usual case, it is enabled
via a socket option. Documentation for this socket option has been added.
The setsockopt statement has been made resilient against different possible cases
of expressing boolean `true' values using a suggestion by Ian McDonald.
Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk> Signed-off-by: Ian McDonald <ian.mcdonald@jandi.co.nz> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Gerrit Renker [Thu, 13 Dec 2007 14:16:23 +0000 (12:16 -0200)]
[DCCP]: Use maximum-RTO backoff from DCCP spec
This removes another Fixme, using the TCP maximum RTO rather than the value
specified by the DCCP specification. Across the sections in RFC 4340, 64
seconds is consistently suggested as maximum RTO backoff value; and this is
the value which is now used.
I have checked both termination cases for retransmissions of Close/CloseReq:
with the default value 15 of `retries2', and an initial icsk_retransmit = 0,
it takes about 614 seconds to declare a non-responding peer as dead, after
which the final terminating Reset is sent. With the TCP maximum RTO value of
120 seconds it takes (as might be expected) almost twice as long, about 23
minutes.
Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk> Signed-off-by: Ian McDonald <ian.mcdonald@jandi.co.nz> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Gerrit Renker [Thu, 13 Dec 2007 14:02:43 +0000 (12:02 -0200)]
[DCCP]: Shift the retransmit timer for active-close into output.c
When performing active close, RFC 4340, 8.3. requires to retransmit the
Close/CloseReq with a backoff-retransmit timer starting at intially 2 RTTs.
This patch shifts the existing code for active-close retransmit timer
into output.c, so that the retransmit timer is started when the first
Close/CloseReq is sent. Previously, the timer was started when, after
releasing the socket in dccp_close(), the actively-closing side had not yet
reached the CLOSED/TIMEWAIT state.
The patch further reduces the initial timeout from 3 seconds to the required
2 RTTs, where - in absence of a known RTT - the fallback value specified in
RFC 4340, 3.4 is used.
Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk> Signed-off-by: Ian McDonald <ian.mcdonald@jandi.co.nz> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Daniel Lezcano [Thu, 13 Dec 2007 13:34:58 +0000 (05:34 -0800)]
[IPV6]: fix section mismatch warnings
Removed useless and buggy __exit section in the different
ipv6 subsystems. Otherwise they will be called inside an
init section during rollbacking in case of an error in the
protocol initialization.
Signed-off-by: Daniel Lezcano <dlezcano@fr.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Gerrit Renker [Thu, 13 Dec 2007 13:28:43 +0000 (11:28 -0200)]
[DCCP]: Perform SHUT_RD and SHUT_WR on receiving close
This patch performs two changes:
1) Close the write-end in addition to the read-end when a fin-like segment
(Close or CloseReq) is received by DCCP. This accounts for the fact that DCCP,
in contrast to TCP, does not have a half-close. RFC 4340 says in this respect
that when a fin-like segment has been sent there is no guarantee at all that
any further data will be processed.
Thus this patch performs SHUT_WR in addition to the SHUT_RD when a fin-like
segment is encountered.
2) Minor change: I noted that code appears twice in different places and think it
makes sense to put this into a self-contained function (dccp_enqueue()).
Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk> Signed-off-by: Ian McDonald <ian.mcdonald@jandi.co.nz> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Michael Chan [Fri, 21 Dec 2007 03:56:37 +0000 (19:56 -0800)]
[BNX2]: Introduce new bnx2_napi structure.
Introduce a bnx2_napi structure that will hold a napi_struct and
other fields to handle NAPI polling for the napi_struct. Various tx
and rx indexes and status block pointers will be moved from the main
bnx2 structure to this bnx2_napi structure.
Most NAPI path functions are modified to be passed this bnx2_napi
struct pointer.
Signed-off-by: Michael Chan <mchan@broadcom.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Michael Chan [Fri, 21 Dec 2007 03:56:09 +0000 (19:56 -0800)]
[BNX2]: Restructure IRQ datastructures.
Add a table to keep track of multiple IRQs and restructure the IRQ
request and free functions so that they can be easily expanded to
handle multiple IRQs.
Signed-off-by: Michael Chan <mchan@broadcom.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Michael Chan [Wed, 12 Dec 2007 19:19:57 +0000 (11:19 -0800)]
[BNX2]: Enable S/G for jumbo RX.
If the MTU requires more than 1 page for the SKB, enable the page ring
and calculate the size of the page ring. This will guarantee order-0
allocation regardless of the MTU size.
Fixup loopback test packet size so that we don't deal with the pages
during loopback test.
Signed-off-by: Michael Chan <mchan@broadcom.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Michael Chan [Wed, 12 Dec 2007 19:17:43 +0000 (11:17 -0800)]
[BNX2]: Restructure RX ring init. code.
Factor out the common functions that will be used to initialize the
normal RX rings and the page rings.
Change the copybreak constant RX_COPY_THRESH to 128. This same
constant will be used for the max. size of the linear SKB when pages
are used. Copybreak will be turned off when pages are used.
Signed-off-by: Michael Chan <mchan@broadcom.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Michal Schmidt [Wed, 12 Dec 2007 19:01:43 +0000 (11:01 -0800)]
[IPIP]: Allow rebinding the tunnel to another interface
Once created, an IP tunnel can't be bound to another device.
(reported as https://bugzilla.redhat.com/show_bug.cgi?id=419671)
To reproduce:
# create a tunnel:
ip tunnel add tunneltest0 mode ipip remote 10.0.0.1 dev eth0
# try to change the bounding device from eth0 to eth1:
ip tunnel change tunneltest0 dev eth1
# show the result:
ip tunnel show tunneltest0
tunneltest0: ip/ip remote 10.0.0.1 local any dev eth0 ttl inherit
Notice the bound device has not changed from eth0 to eth1.
This patch fixes it. When changing the binding, it also recalculates the
MTU according to the new bound device's MTU.
If the change is acceptable, I'll do the same for GRE and SIT tunnels.
Signed-off-by: Michal Schmidt <mschmidt@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Herbert Xu [Wed, 12 Dec 2007 18:44:43 +0000 (10:44 -0800)]
[IPSEC]: Add ICMP host relookup support
RFC 4301 requires us to relookup ICMP traffic that does not match any
policies using the reverse of its payload. This patch implements this
for ICMP traffic that originates from or terminates on localhost.
This is activated on outbound with the new policy flag XFRM_POLICY_ICMP,
and on inbound by the new state flag XFRM_STATE_ICMP.
On inbound the policy check is now performed by the ICMP protocol so
that it can repeat the policy check where necessary.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
Herbert Xu [Wed, 12 Dec 2007 18:44:16 +0000 (10:44 -0800)]
[IPSEC]: Added xfrm_decode_session_reverse and xfrmX_policy_check_reverse
RFC 4301 requires us to relookup ICMP traffic that does not match any
policies using the reverse of its payload. This patch adds the functions
xfrm_decode_session_reverse and xfrmX_policy_check_reverse so we can get
the reverse flow to perform such a lookup.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au> Signed-off-by: David S. Miller <davem@davemloft.net>
Gerrit Renker [Wed, 12 Dec 2007 16:23:08 +0000 (14:23 -0200)]
[TFRC]: Remove previous loss intervals implementation
Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk> Signed-off-by: Ian McDonald <ian.mcdonald@jandi.co.nz> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Gerrit Renker [Wed, 12 Dec 2007 16:06:14 +0000 (14:06 -0200)]
[CCID3]: Interface CCID3 code with newer Loss Intervals Database
This hooks up the TFRC Loss Interval database with CCID 3 packet reception.
In addition, it makes the CCID-specific computation of the first loss
interval (which requires access to all the guts of CCID3) local to ccid3.c.
The patch also fixes an omission in the DCCP code, that of a default /
fallback RTT value (defined in section 3.4 of RFC 4340 as 0.2 sec); while
at it, the upper bound of 4 seconds for an RTT sample has been reduced to
match the initial TCP RTO value of 3 seconds from[RFC 1122, 4.2.3.1].
Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk> Signed-off-by: Ian McDonald <ian.mcdonald@jandi.co.nz> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Gerrit Renker [Wed, 12 Dec 2007 16:03:01 +0000 (14:03 -0200)]
[TFRC]: CCID3 (and CCID4) needs to access these inlines
This moves two inlines back to packet_history.h: these are not private
to packet_history.c, but are needed by CCID3/4 to detect whether a new
loss is indicated, or whether a loss is already pending.
Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Further, documentation of ccid3_hc_rx_sock (capitalisation) is made consistent.
Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Gerrit Renker [Wed, 12 Dec 2007 15:50:51 +0000 (13:50 -0200)]
[TFRC]: Ringbuffer to track loss interval history
A ringbuffer-based implementation of loss interval history is easier to
maintain, allocate, and update.
The `swap' routine to keep the RX history sorted is due to and was written
by Arnaldo Carvalho de Melo, simplifying an earlier macro-based variant.
Details:
* access to the Loss Interval Records via macro wrappers (with safety checks);
* simplified, on-demand allocation of entries (no extra memory consumption on
lossless links); cache allocation is local to the module / exported as service;
* provision of RFC-compliant algorithm to re-compute average loss interval;
* provision of comprehensive, new loss detection algorithm
- support for all cases of loss, including re-ordered/duplicate packets;
- waiting for NDUPACK=3 packets to fill the hole;
- updating loss records when a late-arriving packet fills a hole.
Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk> Signed-off-by: Ian McDonald <ian.mcdonald@jandi.co.nz> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Gerrit Renker [Wed, 12 Dec 2007 14:28:40 +0000 (12:28 -0200)]
[TFRC]: Loss interval code needs the macros/inlines that were moved
This moves the inlines (which were previously declared as macros) back into
packet_history.h since the loss detection code needs to be able to read entries
from the RX history in order to create the relevant loss entries: it needs at
least tfrc_rx_hist_loss_prev() and tfrc_rx_hist_last_rcv(), which in turn
require the definition of the other inlines (macros).
Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Gerrit Renker [Wed, 12 Dec 2007 14:24:49 +0000 (12:24 -0200)]
[TFRC]: Put RX/TX initialisation into tfrc.c
This separates RX/TX initialisation and puts all packet history / loss intervals
initialisation into tfrc.c.
The organisation is uniform: slab declaration -> {rx,tx}_init() -> {rx,tx}_exit()
Signed-off-by: Gerrit Renker <gerrit@erg.abdn.ac.uk> Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Denis V. Lunev [Tue, 11 Dec 2007 12:19:17 +0000 (04:19 -0800)]
[NETNS]: struct net content re-work (v3)
Recently David Miller and Herbert Xu pointed out that struct net becomes
overbloated and un-maintainable. There are two solutions:
- provide a pointer to a network subsystem definition from struct net.
This costs an additional dereferrence
- place sub-system definition into the structure itself. This will speedup
run-time access at the cost of recompilation time
The second approach looks better for us. Other sub-systems will follow.
Signed-off-by: Denis V. Lunev <den@openvz.org> Acked-by: Daniel Lezcano <dlezcano@fr.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Daniel Lezcano [Tue, 11 Dec 2007 10:25:35 +0000 (02:25 -0800)]
[IPV6]: make the protocol initialization to return an error code
This patchset makes the different protocols to return an error code, so
the af_inet6 module can check the initialization was correct or not.
The raw6 was taken into account to be consistent with the rest of the
protocols, but the registration is at the same place.
Because the raw6 has its own init function, the proto and the ops structure
can be moved inside the raw6.c file.
Signed-off-by: Daniel Lezcano <dlezcano@fr.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Daniel Lezcano [Tue, 11 Dec 2007 10:25:01 +0000 (02:25 -0800)]
[IPV6]: make inet6_register_protosw to return an error code
This patch makes the inet6_register_protosw to return an error code.
The different protocols can be aware the registration was successful or
not and can pass the error to the initial caller, af_inet6.
Signed-off-by: Daniel Lezcano <dlezcano@fr.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Daniel Lezcano [Tue, 11 Dec 2007 10:23:54 +0000 (02:23 -0800)]
[IPV6]: make extended headers to return an error at initialization
This patch factorize the code for the differents init functions for rthdr,
nodata, destopt in a single function exthdrs_init.
This function returns an error so the af_inet6 module can check correctly
the initialization.
Signed-off-by: Daniel Lezcano <dlezcano@fr.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Daniel Lezcano [Tue, 11 Dec 2007 10:23:18 +0000 (02:23 -0800)]
[IPV6]: make flowlabel to return an error
This patch makes the flowlab subsystem to return an error code and makes
some cleanup with procfs ifdefs.
The af_inet6 will use the flowlabel init return code to check the initialization
was correct.
Signed-off-by: Daniel Lezcano <dlezcano@fr.ibm.com> Signed-off-by: David S. Miller <davem@davemloft.net>
Pavel Emelyanov [Tue, 11 Dec 2007 10:17:40 +0000 (02:17 -0800)]
[IPV4]: Cleanup sysctl manipulations in devinet.c
This includes:
* moving neigh_sysctl_(un)register calls inside
devinet_sysctl_(un)register ones, as they are always
called in pairs;
* making __devinet_sysctl_unregister() to unregister
the ipv4_devconf struct, while original devinet_sysctl_unregister()
works with the in_device to handle both - devconf and
neigh sysctls;
* make stubs for CONFIG_SYSCTL=n case to get rid of
in-code ifdefs.
Signed-off-by: Pavel Emelyanov <xemul@openvz.org> Signed-off-by: David S. Miller <davem@davemloft.net>
projects / mirror_ubuntu-focal-kernel.git / log