18 Jul, 2009
1 commit
-
Pure style cleanups.
Signed-off-by: Eric Dumazet
Signed-off-by: David S. Miller
17 Jul, 2009
3 commits
-
Conflicts:
drivers/net/wireless/orinoco/main.c -
Commit e912b1142be8f1e2c71c71001dc992c6e5eb2ec1
(net: sk_prot_alloc() should not blindly overwrite memory)
took care of not zeroing whole new socket at allocation time.sock_copy() is another spot where we should be very careful.
We should not set refcnt to a non null value, until
we are sure other fields are correctly setup, or
a lockless reader could catch this socket by mistake,
while not fully (re)initialized.This patch puts sk_node & sk_refcnt to the very beginning
of struct sock to ease sock_copy() & sk_prot_alloc() job.We add appropriate smp_wmb() before sk_refcnt initializations
to match our RCU requirements (changes to sock keys should
be committed to memory before sk_refcnt setting)Signed-off-by: Eric Dumazet
Signed-off-by: David S. Miller
16 Jul, 2009
4 commits
-
When a slab cache uses SLAB_DESTROY_BY_RCU, we must be careful when allocating
objects, since slab allocator could give a freed object still used by lockless
readers.In particular, nf_conntrack RCU lookups rely on ct->tuplehash[xxx].hnnode.next
being always valid (ie containing a valid 'nulls' value, or a valid pointer to next
object in hash chain.)kmem_cache_zalloc() setups object with NULL values, but a NULL value is not valid
for ct->tuplehash[xxx].hnnode.next.Fix is to call kmem_cache_alloc() and do the zeroing ourself.
As spotted by Patrick, we also need to make sure lookup keys are committed to
memory before setting refcount to 1, or a lockless reader could get a reference
on the old version of the object. Its key re-check could then pass the barrier.Signed-off-by: Eric Dumazet
Signed-off-by: Patrick McHardy -
The first argument is the address family, the second one the hook
number.Signed-off-by: Patrick McHardy
-
Add appropriate MODULE_ALIAS() to facilitate autoloading of can protocol drivers
Signed-off-by: Lothar Wassmann
Acked-by: Oliver Hartkopp
Signed-off-by: David S. Miller -
Fix a use after free bug in can protocol drivers
The release functions of the can protocol drivers lack a call to
sock_orphan() which leads to referencing freed memory under certain
circumstances.This patch fixes a bug reported here:
https://lists.berlios.de/pipermail/socketcan-users/2009-July/000985.htmlSigned-off-by: Lothar Wassmann
Acked-by: Oliver Hartkopp
Signed-off-by: David S. Miller
15 Jul, 2009
4 commits
-
Wireless extensions have the unfortunate problem that events
are multicast netlink messages, and are not independent of
pointer size. Thus, currently 32-bit tasks on 64-bit platforms
cannot properly receive events and fail with all kinds of
strange problems, for instance wpa_supplicant never notices
disassociations, due to the way the 64-bit event looks (to a
32-bit process), the fact that the address is all zeroes is
lost, it thinks instead it is 00:00:00:00:01:00.The same problem existed with the ioctls, until David Miller
fixed those some time ago in an heroic effort.A different problem caused by this is that we cannot send the
ASSOCREQIE/ASSOCRESPIE events because sending them causes a
32-bit wpa_supplicant on a 64-bit system to overwrite its
internal information, which is worse than it not getting the
information at all -- so we currently resort to sending a
custom string event that it then parses. This, however, has a
severe size limitation we are frequently hitting with modern
access points; this limitation would can be lifted after this
patch by sending the correct binary, not custom, event.A similar problem apparently happens for some other netlink
users on x86_64 with 32-bit tasks due to the alignment for
64-bit quantities.In order to fix these problems, I have implemented a way to
send compat messages to tasks. When sending an event, we send
the non-compat event data together with a compat event data in
skb_shinfo(main_skb)->frag_list. Then, when the event is read
from the socket, the netlink code makes sure to pass out only
the skb that is compatible with the task. This approach was
suggested by David Miller, my original approach required
always sending two skbs but that had various small problems.To determine whether compat is needed or not, I have used the
MSG_CMSG_COMPAT flag, and adjusted the call path for recv and
recvfrom to include it, even if those calls do not have a cmsg
parameter.I have not solved one small part of the problem, and I don't
think it is necessary to: if a 32-bit application uses read()
rather than any form of recvmsg() it will still get the wrong
(64-bit) event. However, neither do applications actually do
this, nor would it be a regression.Signed-off-by: Johannes Berg
Signed-off-by: David S. Miller -
The current function for sending events first allocates the
event stream buffer, and then an skb to copy the event stream
into. This can be done in one go. Also, the current function
leaks kernel data to userspace in a 4 uninitialised bytes,
initialise those explicitly. Finally also add a few useful
comments, as opposed to the current comments.Signed-off-by: Johannes Berg
Signed-off-by: David S. Miller -
This makes wireless extensions netns aware. The
tasklet sending the events is converted to a work
struct so that we can rtnl_lock() in it.Signed-off-by: Johannes Berg
Signed-off-by: David S. Miller -
Fixes two bugs:
- ToS/DiffServ inheritance was unintentionally activated when using impair fixed ToS values
- ECN bit was lost during ToS/DiffServ inheritanceSigned-off-by: Andreas Jaggi
Signed-off-by: David S. Miller
14 Jul, 2009
1 commit
-
Rename lookup_neigh_params to lookup_neigh_parms as the struct is named
neigh_parms and all other functions dealing with the struct carry
neigh_parms in their names.Signed-off-by: Tobias Klauser
Signed-off-by: David S. Miller
13 Jul, 2009
11 commits
-
Remove redundant sched/ in net/Makefile.
sched/ is contained in previous:
obj-$(CONFIG_NET) += ethernet/ 802/ sched/ netlink/,
so the later
obj-$(CONFIG_NET_SCHED) += sched/
isn't necessary.Signed-off-by: Changli Gao
----
Makefile | 1 -
1 file changed, 1 deletion(-)
Signed-off-by: David S. Miller -
- validate and forward GSO UDP/IPv6 packets from untrusted sources.
- do software UFO if the outgoing device doesn't support UFO.Signed-off-by: Sridhar Samudrala
Acked-by: Herbert Xu
Signed-off-by: David S. Miller -
- move ipv6_select_ident() inline function to ipv6.h and remove the unused
skb argumentSigned-off-by: Sridhar Samudrala
Acked-by: Herbert Xu
Signed-off-by: David S. Miller -
- fix gso_size setting for ipv6 fragment to be a multiple of 8 bytes.
Signed-off-by: Sridhar Samudrala
Acked-by: Herbert Xu
Signed-off-by: David S. Miller -
- add HW checksum support for outgoing large UDP/IPv6 packets destined for
a UFO enabled device.Signed-off-by: Sridhar Samudrala
Acked-by: Herbert Xu
Signed-off-by: David S. Miller -
- validate and forward GSO UDP/IPv4 packets from untrusted sources.
- do software UFO if the outgoing device doesn't support UFO.Signed-off-by: Sridhar Samudrala
Acked-by: Herbert Xu
Signed-off-by: David S. Miller -
The function get_net_ns_by_pid(), to get a network
namespace from a pid_t, will be required in cfg80211
as well. Therefore, let's move it to net_namespace.c
and export it. We can't make it a static inline in
the !NETNS case because it needs to verify that the
given pid even exists (and return -ESRCH).Signed-off-by: Johannes Berg
Signed-off-by: David S. Miller -
This makes generic netlink network namespace aware. No
generic netlink families except for the controller family
are made namespace aware, they need to be checked one by
one and then set the family->netnsok member to true.A new function genlmsg_multicast_netns() is introduced to
allow sending a multicast message in a given namespace,
for example when it applies to an object that lives in
that namespace, a new function genlmsg_multicast_allns()
to send a message to all network namespaces (for objects
that do not have an associated netns).The function genlmsg_multicast() is changed to multicast
the message in just init_net, which is currently correct
for all generic netlink families since they only work in
init_net right now. Some will later want to work in all
net namespaces because they do not care about the netns
at all -- those will have to be converted to use one of
the new functions genlmsg_multicast_allns() or
genlmsg_multicast_netns() whenever they are made netns
aware in some way.After this patch families can easily decide whether or
not they should be available in all net namespaces. Many
genl families us it for objects not related to networking
and should therefore be available in all namespaces, but
that will have to be done on a per family basis.Note that this doesn't touch on the checkpoint/restart
problem where network namespaces could be used, genl
families and multicast groups are numbered globally and
I see no easy way of changing that, especially since it
must be possible to multicast to all network namespaces
for those families that do not care about netns.Signed-off-by: Johannes Berg
Signed-off-by: David S. Miller -
All we need to take care of is using proper RCU list
add/del primitives and inserting a synchronize_rcu()
at one place to make sure the exit notifiers are run
after everybody has stopped iterating the list.Signed-off-by: Johannes Berg
Signed-off-by: David S. Miller -
For the network namespace work in generic netlink I need
to be able to call this function under rcu_read_lock(),
otherwise the locking becomes a nightmare and more locks
would be needed. Instead, just embed a struct rcu_head
(actually a struct listeners_rcu_head that also carries
the pointer to the memory block) into the listeners
memory so we can use call_rcu() instead of synchronising
and then freeing. No rcu_barrier() is needed since this
code cannot be modular.Signed-off-by: Johannes Berg
Signed-off-by: David S. Miller -
I added those myself in commits b4ff4f04 and 84659eb5,
but I see no reason now why they should be exported,
only generic netlink uses them which cannot be modular.Signed-off-by: Johannes Berg
Signed-off-by: David S. Miller
12 Jul, 2009
3 commits
-
The sit module makes use of skb->dst in it's xmit function, so since
93f154b594fe47 ("net: release dst entry in dev_hard_start_xmit()") sit
tunnels are broken, because the flag IFF_XMIT_DST_RELEASE is not
unset.This patch unsets that flag for sit devices to fix this
regression.Signed-off-by: Sascha Hlusiak
Acked-by: Eric Dumazet
Signed-off-by: David S. Miller -
After commit 2b85a34e911bf483c27cfdd124aeb1605145dc80
(net: No more expensive sock_hold()/sock_put() on each tx)
we do not take any more references on sk->sk_refcnt on outgoing packets.I forgot to delete two __sock_put() from ip_push_pending_frames()
and ip6_push_pending_frames().Reported-by: Emil S Tantilov
Signed-off-by: Eric Dumazet
Tested-by: Emil S Tantilov
Signed-off-by: David S. Miller -
Some sockets use SLAB_DESTROY_BY_RCU, and our RCU code correctness
depends on sk->sk_nulls_node.next being always valid. A NULL
value is not allowed as it might fault a lockless reader.Current sk_prot_alloc() implementation doesnt respect this hypothesis,
calling kmem_cache_alloc() with __GFP_ZERO. Just call memset() around
the forbidden field.Signed-off-by: Eric Dumazet
Signed-off-by: David S. Miller
11 Jul, 2009
13 commits
-
In order to force drivers to advertise their interface
types, don't just disallow creating new interfaces with
unadvertised types but also disallow setting them UP.
Additionally, add some validation on the operations the
drivers support.Signed-off-by: Johannes Berg
Signed-off-by: John W. Linville -
We've named the registered devices 'drv' sometimes,
thinking of "driver", which is not what it is, it's
the internal representation of a wiphy, i.e. a
device. Let's clean up the naming once and and use
'rdev' aka 'registered device' everywhere.Signed-off-by: Johannes Berg
Signed-off-by: John W. Linville -
Over time, a lot of locking issues have crept into
the smarts of cfg80211, so e.g. scan completion can
race against a new scan, IBSS join can race against
leaving an IBSS, etc.Introduce a new per-interface lock that protects
most of the per-interface data that we need to keep
track of, and sprinkle assertions about that lock
everywhere. Some things now need to be offloaded to
work structs so that we don't require being able to
sleep in functions the drivers call. The exception
to that are the MLME callbacks (rx_auth etc.) that
currently only mac80211 calls because it was easier
to do that there instead of in cfg80211, and future
drivers implementing those calls will, if they ever
exist, probably need to use a similar scheme like
mac80211 anyway...In order to be able to handle _deauth and _disassoc
properly, introduce a cookie passed to it that will
determine locking requirements.Signed-off-by: Johannes Berg
Signed-off-by: John W. Linville -
sparse warns about a number of things, and one of them
(use_mfp shadowed variable) actually is a bug, fix all
of them.Signed-off-by: Johannes Berg
Signed-off-by: John W. Linville -
Currently we call that cfg80211_put_dev(), but that is
misleading. With the new convention of using 'rdev' for
registered_device variables, also call that function
cfg80211_unlock_rdev().Signed-off-by: Johannes Berg
Signed-off-by: John W. Linville -
The original code in mac80211 could send a deauth
frame under certain circumstances even if nothing
had ever requested an authentication. This has been
fixed with the rework there, so cfg80211 can now
warn again about spurious events to catch possible
future drivers or mac80211 regressions.Signed-off-by: Johannes Berg
Signed-off-by: John W. Linville -
After the mac80211 mlme cleanup, we can require that
the MLME functions in cfg80211 can sleep. This will
simplify future work in cfg80211 a lot.Signed-off-by: Johannes Berg
Signed-off-by: John W. Linville -
We shouldn't be looking at the ssid_len for non-IBSS,
and for IBSS we should also return an error on trying
to leave an IBSS while not in or joining an IBSS.This fixes an issue where we wouldn't disconnect() on
an interface being taken down since there's no SSID
configured this way.Signed-off-by: Johannes Berg
Signed-off-by: John W. Linville -
The new key work for cfg80211 will only give us the WEP
key for shared auth to do that authentication, and not
via the regular key settings, so we need to be able to
encrypt a single frame in software, and that without a
key struct. Thus, refactor the WEP code to not require
a key structure but use the key, len and idx directly.Signed-off-by: Johannes Berg
Signed-off-by: John W. Linville -
Sit tight. This shakes up the world as you know
it. Let go of your spaghetti tongs, they will no
longer be required, the horrible statemachine in
net/mac80211/mlme.c is no more...With the cfg80211 SME mac80211 now has much less
to keep track of, but, on the other hand, for FT
it needs to be able to keep track of at least one
authentication being in progress while associated.
So convert from a single state machine to having
small ones for all the different things we need to
do. For real FT it will still need work wrt. PS,
but this should be a good step.Signed-off-by: Johannes Berg
Signed-off-by: John W. Linville -
The ap_capab and last_probe struct members are unused.
Signed-off-by: Johannes Berg
Signed-off-by: John W. Linville -
Since we don't really know that well in the kernel,
let's let the SME control whether it wants to use
reassociation or not, by allowing it to give the
previous BSSID in the associate() parameters.Signed-off-by: Johannes Berg
Signed-off-by: John W. Linville -
We've designed the /dev/rfkill API in a way that we
can increase the event struct by adding members at
the end, should it become necessary. To validate the
events, userspace and the kernel need to have the
proper event size to check for -- when reading from
the other end they need to verify that it's at least
version 1 of the event API, with the current struct
size, so define a constant for that and make the
code a little more 'future proof'.Not that I expect that we'll have to change the event
size any time soon, but it's better to write the code
in a way that lends itself to extending.Due to the current size of the event struct, the code
is currently equivalent, but should the event struct
ever need to be increased the new code might not need
changing.Signed-off-by: Johannes Berg
Signed-off-by: John W. Linville