Modified the hash function to return just hash separating from the
modulo operation that can be performed by the caller. This is to
make way for the tlb mode to use the same hashing policies that
are used in the 802.3ad and Xor mode.
Change-Id: I276609e87e0ca213c4d1b17b79c5e0b0f3d0dd6f
Signed-off-by: Mahesh Bandewar <maheshb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Johannes noted this is not needed, all of the fragment
accessors don't need CONFIG_NET_NS. This goes test compiled with
CONFIG_BT_6LOWPAN=y and a disabled CONFIG_NET_NS.
CC: Alexander Smirnov <alex.bluesman.smirnov@gmail.com>
Cc: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>
Cc: linux-zigbee-devel@lists.sourceforge.net
Cc: David S. Miller" <davem@davemloft.net>
Cc: netdev@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Cc: Johannes Berg <johannes@sipsolutions.net>
Signed-off-by: Luis R. Rodriguez <mcgrof@suse.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Jeff Kirsher says:
====================
Intel Wired LAN Driver Updates
This series contains updates to ixgbe, ixgbevf, e1000e, igb and i40e.
Jacob converts the ixgbe low_water into an array which allows the
algorithm to output different values for different TCs and we can
distinguish between them. Removes vlan_filter_disable() and
vlan_filter_enable() in ixgbe so that we can do the work directly in
set_rx_mode(). Changes the setting of multicast filters only when
the interface is not in promiscuous mode for multicast packets in
ixgbe. Improves MAC filter handling by adding mac_table API based
on work done for igb, which includes functions to add/delete MAC
filters.
Mark changes register reads in ixgbe to an out-of-line function since
register reads are slow.
Emil provides a ixgbevf patch to update the driver description since
it supports more than just 82599 parts now.
David provides several cleanup patches for e1000e which resolve some
checkpatch issues as well as changing occurrences of returning 0 or 1 in
bool functions to returning true false or true.
Carolyn provides several cleanup patches for igb which fix checkpatch
warnings.
Mitch provides a fix for i40evf where the driver would correctly allow
the virtual function link state to be controlled by 'ip set link', but
would not report it correctly back. This is fixed by filling out
the appropriate field in the VF info struct.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
This was caught when using a spatch (aka. coccinelle) script
written by Joe Perches.
Cc: Joe Perches <joe@perches.com>
Signed-off-by: Shruti Kanetkar <Shruti@Freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Alexey Charkov says:
====================
net: via-rhine: add support for on-chip Rhine controllers
This series introduces platform bus (OpenFirmware) binding for
via-rhine, as used in various ARM-based Systems-on-Chip by
VIA/WonderMedia.
This has been tested in OF configuration by myself on a WM8950-based VIA
APC Rock development board and on a WM8850-based netbook, and in PCI
configuration by Roger.
Please note that the initial version of these patches was signed off by
Roger, but some time has passed since then, so I'm not including his
sign-off until explicit notice.
Changes since v1:
- Fixed indentation of function arguments
- Switched to 'dev_is_pci' instead of string comparison on bus name
- Dropped 'rhine,revision' DT attribute, put the revision into OF match
table instead
- Included actual device tree nodes where applicable
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
This should make the driver usable with VIA/WonderMedia ARM-based
Systems-on-Chip integrated Rhine III adapters. Note that these
are always in MMIO mode, and don't have any known EEPROM.
Signed-off-by: Alexey Charkov <alchark@gmail.com>
Acked-by: Rob Herring <robh@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Use more generic data structures instead of struct pci_dev wherever
possible in preparation for OF bus binding
Signed-off-by: Alexey Charkov <alchark@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Remove legacy PCI DMA wrappers and instead use generic DMA functions
directly in preparation for OF bus binding
Signed-off-by: Alexey Charkov <alchark@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The recent patch that moved broadcasts to process context added
a couple of bugs on the error path where we may dereference NULL
or leak an skb. This patch fixes them.
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
Nicolas Dichtel says:
====================
gre: allow to switch netns during encap/decap
The goal of this serie is to add x-netns support for the module ip_gre and
ip6_gre, ie the encapsulation addresses and the network device are not owned by
the same namespace.
Example to configure an ipv4 gre tunnel:
modprobe ip_gre
ip netns add netns1
ip netns exec netns1 ip link set lo up
ip link add gre1 type gre remote 10.16.0.121 local 10.16.0.249 ikey 10 okey 10 csum
ip link set gre1 netns netns1
ip netns exec netns1 ip link set gre1 up
ip netns exec netns1 ip addr add dev gre1 192.168.0.249 remote 192.168.0.121
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch allows to switch the netns when packet is encapsulated or
decapsulated. In other word, the encapsulated packet is received in a netns,
where the lookup is done to find the tunnel. Once the tunnel is found, the
packet is decapsulated and injecting into the corresponding interface which
stands to another netns.
When one of the two netns is removed, the tunnel is destroyed.
Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch allows to switch the netns when packet is encapsulated or
decapsulated. In other word, the encapsulated packet is received in a netns,
where the lookup is done to find the tunnel. Once the tunnel is found, the
packet is decapsulated and injecting into the corresponding interface which
stands to another netns.
When one of the two netns is removed, the tunnel is destroyed.
Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The union contains only one member now, so we use the variables in it directly.
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Reviewed-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Removed recv_pkt_list and lock, and updated related code, so that
the locking overhead is reduced especially when multiple channels
are in use.
The recv_pkt_list isn't actually necessary because the packets are
processed sequentially in each channel. It has been replaced by a
local variable, and the related lock for this list is also removed.
The is_data_pkt field is not used in receive path, so its assignment
is cleaned up.
Signed-off-by: Haiyang Zhang <haiyangz@microsoft.com>
Reviewed-by: K. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Although the driver would correctly allow the VF link state to be
controlled by 'ip set link', it would not report it correctly back.
Fix this by filling out the appropriate field in the vf info struct.
Change-ID: I58d8e356438190e1ee9660b424301af6f416cdbe
Signed-off-by: Mitch Williams <mitch.a.williams@intel.com>
Signed-off-by: Catherine Sullivan <catherine.sullivan@intel.com>
Tested-by: Sibai Li <sibai.li@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
This patch fixes WARNING:LEADING_SPACE, WARNING:SPACING, ERROR:SPACING,
WARNING:SPACE_BEFORE_TAB and ERROR_CODE_INDENT from checkpatch file check.
Signed-off-by: Carolyn Wyborny <carolyn.wyborny@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
This patch fixes WARNING:BRACES and ERROR:OPEN_BRACE from
checkpatch file check.
Signed-off-by: Carolyn Wyborny <carolyn.wyborny@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
This patch fixes WARNING:PREFER_PR_LEVEL and WARNING:SPLIT_STRING
from checkpatch file check.
Signed-off-by: Carolyn Wyborny <carolyn.wyborny@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: Dave Ertman <davidx.m.ertman@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Fixing "WARNING:SPACING: Unnecessary space before function pointer arguments"
Signed-off-by: Dave Ertman <davidx.m.ertman@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Fixing "WARNING:SPACING: networking uses a blank line after declarations"
Signed-off-by: Dave Ertman <davidx.m.ertman@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Changing occurrences of returning 0 and 1 from bool functions to false and
true, respectively
Signed-off-by: Dave Ertman <davidx.m.ertman@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
This patch removes 82599 from the description of the ixgbevf module
since the VF driver is supported on other parts as well.
Signed-off-by: Emil Tantilov <emil.s.tantilov@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Add mac_table API based on work done for igb, which includes functions
to add and delete mac filters. This simplifies code for various entities
that use MAC filters such as VMDQ, SR-IOV, MACVLAN, and such.
Reported-by: Mitch Williams <mitch.a.williams@intel.com>
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
In line with changes done by Alex Duyck regarding unicast filters, we
now only set multicast filters when the interface is not in promiscuous
mode for multicast packets. This also has an impact on the RAR usage
such that SR-IOV has some RARs reserved for its own usage.
Reported-by: Alex Duyck <alexander.h.duyck@intel.com>
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Previously these functions handled stripping setup as well, but this has
already been removed from these functions. Rather than encapsulating
this into a function, we can just do the work directly in set_rx_mode.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Register reads are slow, so don't inline them.
Size before:
text data bss dec hex filename
226337 8280 552 235169 396a1 ixgbe.ko
Size after:
text data bss dec hex filename
194578 8280 552 203410 31a92 ixgbe.ko
for about a 14% reduction in text size.
Signed-off-by: Mark Rustad <mark.d.rustad@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Since fc.high_water is an array, we should treat low_water as an array
also. This allows the algorithm to output different values for different
TCs, and then we can distinguish between them. In addition, this patch
changes one path that didn't honor the return value from ixgbe_setup_fc.
Reported-by: Aaron Salter <aaron.k.salter@intel.com>
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Phil Schmitt <phillip.j.schmitt@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Jeff Kirsher says:
====================
Intel Wired LAN Driver Updates
This series contains updates to i40e and i40evf.
Greg provides two patches for i40e, the first adds the netdev ops to support
the addition of static FDB entries in the physical function (PF) MAC/VLAN
filter table so that the virtual functions (VFs) can communicate with
bridged virtual Ethernet ports such as those provided by the virtio
driver. The second is to fix an issue where the assignment of a port
VLAN after it is already up and running requires the VF driver to be
reloaded, so print a message warning the host administrator about the
need to reload the VF driver. In addition, knock the VF offline so that
it does not continue to receive traffic not on the port VLAN assigned to
it.
Jesse provides a patch for i40e and i40evf to unhide and enable the
PREFENA field in the receive host memory cache (RX-HMC) for best
performance.
Mitch provides a i40e patch to implement the net device op for Tx
bandwidth setting.
Catherine removes a firmware workaround that is no longer needed with
the latest firmware for i40e. She also provides some minor cleanups
as well bumps the driver versions.
Anjali provides a fix for i40e displaying IPv4 flow director filters
which needed additional information to be communicated up above in
order for it to be displayed correctly.
Shannon adds tracking of the NVM busy state so that the driver won't
allow a new NVM update command until a completion event is received
from the current update. Updates the admin queue API to reflect
recent changes in the firmware. Also rearranges the "if netdev" logic
to prepare for handling non-netdev VSIs. Lastly rework the fdir
setup and tear down to use the newly created i40e_vsi_open() and
i40e_vsi_close(), which also fixes a memory leak of the FDIR queue
buffer info structs across a reset.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Richard Guy Briggs says:
====================
audit: implement multicast socket for journald
This is a patch set Eric Paris and I have been working on to add a restricted
capability read-only netlink multicast socket to kernel audit to enable
userspace clients such as systemd/journald to receive audit logs, in addition
to the bidirectional auditd userspace client.
Currently, auditd has the CAP_AUDIT_CONTROL and CAP_AUDIT_WRITE capabilities
(but uses CAP_NET_ADMIN). The CAP_AUDIT_READ capability will be added for use
by read-only AUDIT_NLGRP_READLOG multicast group clients to the kaudit
subsystem. This will remove the dependence on CAP_NET_ADMIN for the multicast
read-only socket.
Patches 1-3 provide a way for per-protocol bind functions to
signal an error and to be able to clean up after themselves.
The first netfilter cleanup patch has already been accepted by a netfilter
maintainer, though I don't see it upstream yet, so it is included for
completeness.
The second patch adds the per-protocol bind function return code to signal to
the netlink code that no further processing should be done and to undo the work
already done.
V1: This rev fixes a bug introduced by flattening the code in the last posting.
*V2: This rev moves the per-protocol bind call above the socket exposure call
and refactors out the unbind procedure.
The third provides a way per protocol to undo bind actions on DROP.
Patches 4-6 implement the audit multicast socket with capability checking.
The fourth patch adds the bind function capability check to multicast join
requests for audit.
The fifth patch adds the audit log read multicast group. An assumption has
been made that systemd/journald reside in the initial network namespace. This
could be changed to check the actual network namespace of systemd/journald
should this assumption no longer be true since audit now supports all network
namespaces. This version of the patch now directly sends the broadcast when
the packet is ready rather than waiting until it passes the queue.
The sixth checks if any clients actually exist before sending.
Since the net tree is busier than the audit tree, conflicts are more likely and
the audit patches depend on the net patches, it is proposed to have the net
tree carry this entire patchset for 3.16. Are the net maintainers ok with this?
https://bugzilla.redhat.com/show_bug.cgi?id=887992
First posted: https://www.redhat.com/archives/linux-audit/2013-January/msg00008.htmlhttps://lkml.org/lkml/2013/1/27/279
Please find source for a test program at:
http://people.redhat.com/rbriggs/audit-multicast-listen/
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Test first to see if there are any userspace multicast listeners bound to the
socket before starting the multicast send work.
Signed-off-by: Richard Guy Briggs <rgb@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add a netlink multicast socket with one group to kaudit for "best-effort"
delivery to read-only userspace clients such as systemd, in addition to the
existing bidirectional unicast auditd userspace client.
Currently, auditd is intended to use the CAP_AUDIT_CONTROL and CAP_AUDIT_WRITE
capabilities, but actually uses CAP_NET_ADMIN. The CAP_AUDIT_READ capability
is added for use by read-only AUDIT_NLGRP_READLOG netlink multicast group
clients to the kaudit subsystem.
This will safely give access to services such as systemd to consume audit logs
while ensuring write access remains restricted for integrity.
Signed-off-by: Richard Guy Briggs <rgb@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Register a netlink per-protocol bind fuction for audit to check userspace
process capabilities before allowing a multicast group connection.
Signed-off-by: Richard Guy Briggs <rgb@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Call the per-protocol unbind function rather than bind function on
NETLINK_DROP_MEMBERSHIP in netlink_setsockopt().
Signed-off-by: Richard Guy Briggs <rgb@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Have the netlink per-protocol optional bind function return an int error code
rather than void to signal a failure.
This will enable netlink protocols to perform extra checks including
capabilities and permissions verifications when updating memberships in
multicast groups.
In netlink_bind() and netlink_setsockopt() the call to the per-protocol bind
function was moved above the multicast group update to prevent any access to
the multicast socket groups before checking with the per-protocol bind
function. This will enable the per-protocol bind function to be used to check
permissions which could be denied before making them available, and to avoid
the messy job of undoing the addition should the per-protocol bind function
fail.
The netfilter subsystem seems to be the only one currently using the
per-protocol bind function.
Signed-off-by: Richard Guy Briggs <rgb@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Remove duplicity and simplify code flow by moving the rcu_read_unlock() above
the condition and let the flow control exit naturally at the end of the
function.
Signed-off-by: Richard Guy Briggs <rgb@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Added a new ancillary load (bpf call in eBPF parlance) that produces
a 32-bit random number. We are implementing it as an ancillary load
(instead of an ISA opcode) because (a) it is simpler, (b) allows easy
JITing, and (c) seems more in line with generic ISAs that do not have
"get a random number" as a instruction, but as an OS call.
The main use for this ancillary load is to perform random packet sampling.
Signed-off-by: Chema Gonzalez <chema@google.com>
Acked-by: Alexei Starovoitov <ast@plumgrid.com>
Acked-by: Daniel Borkmann <dborkman@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
if allocating memory for vlan_pcpu_stats failed, the device can not be operated
Signed-off-by: Li RongQing <roy.qing.li@gmail.com>
Cc: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This be2net patch implements the get/set_rxfh() ethtool hooks.
RSS_CONFIG device command is invoked to set hashkey and indirection table.
It also uses an initial random value for RSS hash key instead of a
hard-coded value as hard-coded values for a hash-key are usually
considered a security risk.
Signed-off-by: Venkat Duvvuru <VenkatKumar.Duvvuru@Emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This ethtool patch primarily copies the ioctl command data structures
from/to the User space and invokes the driver hook.
Signed-off-by: Venkat Duvvuru <VenkatKumar.Duvvuru@Emulex.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In commit 0e280af026 ("tcp: introduce TCPSpuriousRtxHostQueues SNMP
counter") we added a logic to detect when a packet was retransmitted
while the prior clone was still in a qdisc or driver queue.
We are now confident we can do better, and catch the problem before
we fragment a TSO packet before retransmit, or in TLP path.
This patch fully exploits the logic by simply canceling the spurious
retransmit.
Original packet is in a queue and will eventually leave the host.
This helps to avoid network collapses when some events make the RTO
estimations very wrong, particularly when dealing with huge number of
sockets with synchronized blast.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Userspace applications can use IFA_F_MANAGETEMPADDR with RTM_NEWADDR
already to indicate that the kernel should take care of temporary
address management.
This patch adds related functionality to RTM_DELADDR. By setting
IFA_F_MANAGETEMPADDR a userspace application can indicate that the kernel
should delete all related temporary addresses as well.
A corresponding patch for the "ip addr del" command has been applied to
iproute2 already.
Signed-off-by: Heiner Kallweit <heiner.kallweit@web.de>
Reviewed-by: Jiri Pirko <jiri@resnulli.us>
Acked-by: Hannes Frederic Sowa <hannes@stressinduktion.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Ying Xue says:
====================
purge tipc_net_lock
Now tipc routing hierarchy comprises the structures 'node', 'link'and
'bearer'. The whole hierarchy is protected by a big read/write lock,
tipc_net_lock, to ensure that nothing is added or removed while code
is accessing any of these structures. Obviously the locking policy
makes node, link and bearer components closely bound together so that
their relationship becomes unnecessarily complex. In the worst case,
such locking policy not only has a negative influence on performance,
but also it's prone to lead to deadlock occasionally.
In order o decouple the complex relationship between bearer and node
as well as link, the locking policy is adjusted as follows:
- Bearer level
RTNL lock is used on update side, and RCU is used on read side.
Meanwhile, all bearer instances including broadcast bearer are
saved into bearer_list array.
- Node and link level
All node instances are saved into two tipc_node_list and node_htable
lists. The two lists are protected by node_list_lock on write side,
and they are guarded with RCU lock on read side. All members in node
structure including link instances are protected by node spin lock.
- The relationship between bearer and node
When link accesses bearer, it first needs to find the bearer with
its bearer identity from the bearer_list array. When bearer accesses
node, it can iterate the node_htable hash list with the node address
to find the corresponding node.
In the new locking policy, every component has its private locking
solution and the relationship between bearer and node is very simple,
that is, they can find each other with node address or bearer identity
from node_htable hash list or bearer_list array.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Commit a21a584d67 (tipc: fix neighbor
detection problem after hw address change) introduces a race condition
involving tipc_disc_delete() and tipc_disc_add/remove_dest that can
cause TIPC to dereference the pointer to the bearer discovery request
structure after it has been freed since a stray pointer is left in the
bearer structure.
In order to fix the issue, the process of resetting the discovery
request handler is optimized: the discovery request handler and request
buffer are just reset instead of being freed, allocated and initialized.
As the request point is always valid and the request's lock is taken
while the request handler is reset, the race doesn't happen any more.
Reported-by: Erik Hugne <erik.hugne@ericsson.com>
Signed-off-by: Ying Xue <ying.xue@windriver.com>
Reviewed-by: Erik Hugne <erik.hugne@ericsson.com>
Tested-by: Erik Hugne <erik.hugne@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The node map variable - 'nodes' in bearer structure is only used by
bclink. When bclink accesses it, bc_lock is held. But when change it,
for instance, in tipc_bearer_add_dest() or tipc_bearer_remove_dest()
the bc_lock is not taken at all. To avoid any inconsistent data, we
should always grab bc_lock while accessing node map variable.
Signed-off-by: Ying Xue <ying.xue@windriver.com>
Reviewed-by: Jon Maloy <jon.maloy@ericsson.com>
Reviewed-by: Erik Hugne <erik.hugne@ericsson.com>
Tested-by: Erik Hugne <erik.hugne@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
As bearer pointer is known in tipc_l2_device_event(), it's unnecessary
to search it again in tipc_disable_bearer(). If tipc_disable_bearer()
is replaced with bearer_disable() in tipc_l2_device_event(), this will
help us save a bit time when bearer is disabled.
Signed-off-by: Ying Xue <ying.xue@windriver.com>
Reviewed-by: Jon Maloy <jon.maloy@ericsson.com>
Reviewed-by: Erik Hugne <erik.hugne@ericsson.com>
Tested-by: Erik Hugne <erik.hugne@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The 'media_ptr' pointer in bearer structure which points to network
device, is protected by RCU. So, before netdevice is released,
synchronize_net() should be involved to prevent no any user of
the netdevice on read side from accessing it after it is freed.
Signed-off-by: Ying Xue <ying.xue@windriver.com>
Reviewed-by: Jon Maloy <jon.maloy@ericsson.com>
Reviewed-by: Erik Hugne <erik.hugne@ericsson.com>
Tested-by: Erik Hugne <erik.hugne@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Now tipc routing hierarchy comprises the structures 'node', 'link'and
'bearer'. The whole hierarchy is protected by a big read/write lock,
tipc_net_lock, to ensure that nothing is added or removed while code
is accessing any of these structures. Obviously the locking policy
makes node, link and bearer components closely bound together so that
their relationship becomes unnecessarily complex. In the worst case,
such locking policy not only has a negative influence on performance,
but also it's prone to lead to deadlock occasionally.
In order o decouple the complex relationship between bearer and node
as well as link, the locking policy is adjusted as follows:
- Bearer level
RTNL lock is used on update side, and RCU is used on read side.
Meanwhile, all bearer instances including broadcast bearer are
saved into bearer_list array.
- Node and link level
All node instances are saved into two tipc_node_list and node_htable
lists. The two lists are protected by node_list_lock on write side,
and they are guarded with RCU lock on read side. All members in node
structure including link instances are protected by node spin lock.
- The relationship between bearer and node
When link accesses bearer, it first needs to find the bearer with
its bearer identity from the bearer_list array. When bearer accesses
node, it can iterate the node_htable hash list with the node
address to find the corresponding node.
In the new locking policy, every component has its private locking
solution and the relationship between bearer and node is very simple,
that is, they can find each other with node address or bearer identity
from node_htable hash list or bearer_list array.
Until now above all changes have been done, so tipc_net_lock can be
removed safely.
Signed-off-by: Ying Xue <ying.xue@windriver.com>
Reviewed-by: Jon Maloy <jon.maloy@ericsson.com>
Reviewed-by: Erik Hugne <erik.hugne@ericsson.com>
Tested-by: Erik Hugne <erik.hugne@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Now the media_ptr pointer is protected with tipc_net_lock write lock
on write side; tipc_net_lock read lock is used to read side. As the
part of effort of eliminating tipc_net_lock, we decide to adjust the
locking policy of media_ptr pointer protection: on write side, RTNL
lock is use while on read side RCU read lock is applied.
Signed-off-by: Ying Xue <ying.xue@windriver.com>
Reviewed-by: Jon Maloy <jon.maloy@ericsson.com>
Reviewed-by: Erik Hugne <erik.hugne@ericsson.com>
Tested-by: Erik Hugne <erik.hugne@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Mahesh Bandewar
Luis R. Rodriguez
David S. Miller
Shruti Kanetkar
Alexey Charkov
Herbert Xu
Nicolas Dichtel
Haiyang Zhang
Mitch Williams
Carolyn Wyborny
David Ertman
Emil Tantilov
Jacob Keller
Mark Rustad
Richard Guy Briggs
Chema Gonzalez
Li RongQing
Venkata Duvvuru
Ying Xue
Eric Dumazet
Heiner Kallweit