- 25 Sep, 2015 32 commits
-
-
Eric Dumazet authored
Like tcp_make_synack() the only time we might change the socket is when calling sock_wmalloc(), which is using atomic operation to update sk->sk_wmem_alloc Also use MAX_DCCP_HEADER as both IPv4/IPv6 use this value for max_header. Signed-off-by:
Eric Dumazet <edumazet@google.com> Signed-off-by:
David S. Miller <davem@davemloft.net>
-
Eric Dumazet authored
This documents fact that listener lock might not be held at the time SYNACK are sent. Signed-off-by:
-
Eric Dumazet authored
This is to document that socket lock might not be held at this point. skb_set_owner_w() and ipv6_local_error() are using proper atomic ops or spinlocks, so we promote the socket to non const when calling them. netfilter hooks should never assume socket lock is held, we also promote the socket to non const. Signed-off-by:
-
Eric Dumazet authored
listener socket is not locked when tcp_make_synack() is called. We better make sure no field is written. There is one exception : Since SYNACK packets are attached to the listener at this moment (or SYN_RECV child in case of Fast Open), sock_wmalloc() needs to update sk->sk_wmem_alloc, but this is done using atomic operations so this is safe. Signed-off-by:
-
Eric Dumazet authored
SYNACK packets might be sent without holding socket lock. For DCTCP/ECN sake, we should call INET_ECN_xmit() while socket lock is owned, and only when we init/change congestion control. This also fixies a bug if congestion module is changed from dctcp to another one on a listener : we now clear ECN bits properly. Signed-off-by:
-
Eric Dumazet authored
We do not use the socket in this function. Signed-off-by:
-
Eric Dumazet authored
This function is used to build and send SYNACK packets, possibly on behalf of unlocked listener socket. Make sure we did not miss a write by making this socket const. We no longer can use ip_select_ident() and have to either set iph->id to 0 or directly call __ip_select_ident() Signed-off-by:
-
Eric Dumazet authored
When TCP new listener is done, these functions will be called without socket lock being held. Make sure they don't change anything. Signed-off-by:
-
Eric Dumazet authored
ip_dont_fragment() can accept const socket and dst Signed-off-by:
-
Eric Dumazet authored
socket is not modified, make it const so that callers can do the same if they need. Signed-off-by:
-
Eric Dumazet authored
ip6_dst_lookup_flow() and ip6_dst_lookup_tail() do not touch socket, lets add a const qualifier. This will permit the same change in inet6_csk_route_req() Signed-off-by:
-
Eric Dumazet authored
This is used by TCP listener core, and listener socket shall not be modified by inet_csk_route_req(). Signed-off-by:
-
Eric Dumazet authored
Very soon, TCP stack might call inet_csk_route_req(), which calls inet_csk_route_req() with an unlocked listener socket, so we need to make sure ip_route_output_flow() is not trying to change any field from its socket argument. Signed-off-by:
-
Eric Dumazet authored
Soon, listener socket wont be locked when tcp_openreq_init_rwin() is called. We need to read socket fields once, as their value could change under us. Signed-off-by:
-
Eric Dumazet authored
Soon, listener socket spinlock will no longer be held, add const arguments to tcp_v[46]_init_req() to make clear these functions can not mess socket fields. Signed-off-by:
-
Michal Kubeček authored
The __netdev_find_adj() helper does not use its first argument, only the device to find and list to walk through. Signed-off-by:
Michal Kubecek <mkubecek@suse.cz> Signed-off-by:
-
David S. Miller authored
Stephen Hemminger says: ==================== l2tp: module autoloading With L2TP it was necessary to manually load modules which is a nuisance and not required with other tunneling protocols. This set of patches adds the aliases and module load hook to get rid of the necessity of modprobing. ==================== Acked-By:
James Chapman <jchapman@katalix.com> Signed-off-by:
-
stephen hemminger authored
No longer need explicit modprobe's and update to use ip instead of deprecated ifconfig command. Signed-off-by:
Stephen Hemminger <stephen@networkplumber.org> Signed-off-by:
-
stephen hemminger authored
When creating a IP encapsulated tunnel the necessary l2tp module should be loaded. It already works for UDP encapsulation, it just doesn't work for direct IP encap. Signed-off-by:
-
stephen hemminger authored
It should not be necessary to do explicit module loading when configuring L2TP. Modules should be loaded as needed instead (as is done already with netlink and other tunnel types). This patch adds a new module alias type and code to load the sub module on demand. Signed-off-by:
-
Florian Fainelli authored
Provide a device_type information for slave network devices created by DSA, this is useful for user-space application to easily locate/search for devices of a specific kind. Signed-off-by:
Florian Fainelli <f.fainelli@gmail.com> Reviewed-by:
Vivien Didelot <vivien.didelot@savoirfairelinux.com> Signed-off-by:
-
David S. Miller authored
Jiri Pirko says: ==================== switchdev: transaction item queue and cleanup ==================== Acked-by:
Scott Feldman <sfeldma@gmail.com> Signed-off-by:
-
Jiri Pirko authored
Now, since we have only 2 values for transaction phase, just use bool. Signed-off-by:
Jiri Pirko <jiri@mellanox.com> Signed-off-by:
-
Jiri Pirko authored
The enum is going to disappear, use the helpers instead. Signed-off-by:
-
Jiri Pirko authored
No longer used by drivers, as transaction queue with item destructors takes care of abort phase internally in switchdev code. So kill it. Signed-off-by:
-
Jiri Pirko authored
Shouldn't have been there in the first place. Now it is unused, kill it. Signed-off-by:
-
Jiri Pirko authored
Benefit from previously introduced transaction item queue infrastructure and remove rocker specific transaction memory management. Signed-off-by:
-
Jiri Pirko authored
There will be needed to have switchdev_trans available down in the call chain, so propagate it instead of trans phase enum. This enum will be removed anyway. Also, use prepare/commit phase check helpers to get information about current phase of transaction. Signed-off-by:
-
Jiri Pirko authored
Add helpers which should be used int attr_set/obj_add switchdev ops to check the phase of transaction. Signed-off-by:
-
Jiri Pirko authored
Before it disappears completely, move transaction phase enum under transaction structure and make attr/obj structures a bit cleaner. Signed-off-by:
-
Jiri Pirko authored
Now, the memory allocation in prepare/commit state is done separatelly in each driver (rocker). Introduce the similar mechanism in generic switchdev code, in form of queue. That can be used not only for memory allocations, but also for different items. Abort item destruction is handled as well. Signed-off-by:
-
Jiri Pirko authored
This is temporary, name "trans" will be used for something else and "trans_ph" will eventually disappear. Signed-off-by:
-
- 24 Sep, 2015 8 commits
-
-
git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/next-queueDavid S. Miller authored
Jeff Kirsher says: ==================== Intel Wired LAN Driver Updates 2015-09-23 This series contains updates to ixgbe only. Mark provides all the changes in this series, first clears the destination location for I2C data initially so that the received data will not be corrupted by previous attempts. Then reduced the pauses/delays in the PHY detection when no SFP is present by reducing the number of retires, once an SFP is detected, the "normal" number of retries in PHY detection will be used. Added support for X55EM_x SFP+ dual-speed, and fixed 1G and 10G link stability for X550EM_x by configuring the CS4227 correctly by moving code to ixgbe_setup_mac_link_sfp_x550em(). Added functionality to reset CS4227, since on some platforms the CS4227 does not initialize properly. Next reduces the SFP polling rate, due to when an SFP is not present, the I2C timeouts that result are very costly. So prevent the SFP polling from being done more than once every two seconds. Added support for I2C bus MUX. Fixed the setting of RDRXCTL register which should fall through X540 and 82599, not 82598. In addition, added small packet padding support in X550 by setting RDRXCTL.PSP when the driver is in SRIOV mode. Fixed a known hardware issue where the PCI transactions pending bit sticks high when there are pending transactions, so workaround the issue by wait and then continue with our reset flow. Added a new device ID for X550EM device with SFPs. Provided a fix with the DCA setup, which was suggested by Alex Duyck <aduyck@mirantis.com>, by making it so that we always set the relaxed ordering bits related to the DCA registers even if DCA is not enbaled. Then moves the configuration out of the ixgbe_down() and into ixgbe_configure() before enabling the transmit and receive rings. This ensures that DCA is configured correctly before starting the processing of packets. Fixed VM-to-VM loopback mode which requires that FCRTH be set, but the datasheets did not specify what the value should be. It has now been determined that the correct value should be RXPBSIZE - (24*1024). ==================== Signed-off-by: -
Eric Dumazet authored
Neal suggested to move sk_txhash init into tcp_create_openreq_child(), called both from IPv4 and IPv6. This opportunity was missed in commit 58d607d3 ("tcp: provide skb->hash to synack packets") Signed-off-by:
Neal Cardwell <ncardwell@google.com> Signed-off-by:
-
Tobias Klauser authored
Use of_property_read_u32 instead of of_get_property with return value checks and endianness conversion. Signed-off-by:
Tobias Klauser <tklauser@distanz.ch> Reviewed-by:
Sören Brinkmann <soren.brinkmann@xilinx.com> Signed-off-by:
-
Tobias Klauser authored
Use of_property_read_u32 to read the "clock-frequency" property instead of using of_get_property with return value checks. Signed-off-by:
-
Jiri Benc authored
Since commit 12fd84f4 ("ipv6: Remove unused neigh argument for icmp6_dst_alloc() and its callers."), the neigh parameter of ndisc_send_na and ndisc_send_ns is unused. CC: YOSHIFUJI Hideaki <yoshfuji@linux-ipv6.org> Signed-off-by:
Jiri Benc <jbenc@redhat.com> Signed-off-by:
-
Jiri Benc authored
The genl_notify function has too many arguments for no real reason - all callers use genl_info to get them anyway. Just pass the genl_info down to genl_notify. Signed-off-by:
-
Mark Rustad authored
With the addition of X550em_x SFP+ support, the driver is now functionally equivalent to what will be the 4.2.1 driver when released, so change the version to match. Signed-off-by:
Mark Rustad <mark.d.rustad@intel.com> Tested-by:
Krishneil Singh <krishneil.k.singh@intel.com> Signed-off-by:
Jeff Kirsher <jeffrey.t.kirsher@intel.com>
-
Mark Rustad authored
The X540 thermal interrupt (IXGBE_EIMS_TS) is not an SDP, so it doesn't need to be enabled in ixgbe_setup_gpie(). In fact the value is simply not for the GPIE register at all. Signed-off-by:
-
