- 18 Feb, 2016 23 commits
-
-
David S. Miller authored
Jiri Pirko says: ==================== rocker: do world split This patchset allows new rocker worlds to be easily added in future. Two new worlds are now under development: P4 and eBPF. The main part of the patchset is the OF-DPA carve-out. It resuts in OF-DPA specific file. Clean cut. Note this patchset is based on my original attempt in October 2015. I had to rebase, included all suggestions and did lot of small changes. Main change to go with all-port-one-world approach. Port world is set according to what is setup in HW. Not possible to change worlds from driver. v1->v2: patch 12/13: - split port_init into pre-init and init ==================== Signed-off-by:David S. Miller <davem@davemloft.net>
-
Jiri Pirko authored
Suggested-by:
Scott Feldman <sfeldma@gmail.com> Signed-off-by:
Jiri Pirko <jiri@resnulli.us> Signed-off-by:
-
Jiri Pirko authored
Carve out OF-DPA would specific code from the common file to the world file. This change required struct rocker and struct rocker_port split into world specific struct ofdpa and struct ofdpa_port. Along with this the world specific functions and defines were renamed from prefix "rocker_" to "ofdpa_". Signed-off-by:
Jiri Pirko <jiri@mellanox.com> Signed-off-by:
-
Jiri Pirko authored
No need to push down rocker flags just to check if this is nowait or not. Let the caller handle that. Signed-off-by:
-
Jiri Pirko authored
The only purpose of passing this parameter is to check for prepare phase. The only reason for a failure in that state is if TLVs don't fit into descriptor. That is highly unlikely and if that happens, it is a driver bug. So remove this parameter from rocker_cmd_exec, and check for prepare phase in caller. Signed-off-by:
-
Jiri Pirko authored
This avoids need to alloc/free wait structure for every command call. Signed-off-by:
-
Jiri Pirko authored
Be consistent with the rest of the setting functions, and pass "learning" as a bool function parameter. Signed-off-by:
-
Jiri Pirko authored
This is another step on the way to per-world clean cut. Introduce world ops hooks which each world can implement in world-specific way. Also introduce world infrastructure along with OF-DPA world stub. Signed-off-by:
-
Jiri Pirko authored
And take some other related thing along. They are going to be pushed into of-dpa part anyway. Signed-off-by:
-
Jiri Pirko authored
Introduce a helper to ask HW for the port mode (world). Signed-off-by:
-
Jiri Pirko authored
Carve out TLV processing helpers into separate files. Signed-off-by:
-
Jiri Pirko authored
Since "rocker.c" is going to be split into multiple files, start with renaming original "rocker.c" file to "rocker_main.c". Multiple code parts are going to be cut from "rocker_main.c" later on. Fix couple of checkpatch issues on the way. Signed-off-by:
-
Jiri Pirko authored
Since "rocker.h" file is going to be used for different purpose, rename the hardware-specific header to "rocker_hw.h". Signed-off-by:
-
Jiri Pirko authored
No need to pass rocker_port around to alloc/free rocker functions, since they now use switchdev_trans for memory management storage. With the param removal, shorten the name of the functions since they now has nothing to do with rocker port. Signed-off-by:
-
David S. Miller authored
Iyappan Subramanian says: ==================== Add support for Classifier and RSS This patch set enables, (i) Classifier engine that is used for parsing through the packet and extracting a search string that is then used to search a database to find associative data. (ii) Receive Side Scaling (RSS) that does dynamic load balancing of the CPUs by controlling the number of messages enqueued per CPU though the help of Toeplitz Hash function of 4-tuple of source TCP/UDP port, destination TCP/UDP port, source IPV4 address and destination IPV4 address. (iii) Multi queue, to make advantage of RSS v3: Address review comments from v2 - reordered local variables from longest to shortlest line v2: Address review comments from v1 - fix kbuild warning - add default coalescing v1: - Initial version ==================== Signed-off-by:Iyappan Subramanian <isubramanian@apm.com> Signed-off-by:
-
Iyappan Subramanian authored
Signed-off-by:
Khuong Dinh <kdinh@apm.com> Signed-off-by:
Tanmay Inamdar <tinamdar@apm.com> Tested-by:
Toan Le <toanle@apm.com> Signed-off-by:
-
Iyappan Subramanian authored
Signed-off-by:
-
Iyappan Subramanian authored
Signed-off-by:
-
Iyappan Subramanian authored
Signed-off-by:
-
Zhang Shengju authored
Since function vlan_proc_rem_dev() will only return 0, it's better to return void instead of int. Signed-off-by:
Zhang Shengju <zhangshengju@cmss.chinamobile.com> Signed-off-by:
-
John Fastabend authored
By packing the structure we can remove a few holes as Jamal suggests. before: struct tc_cls_u32_knode { struct tcf_exts * exts; /* 0 8 */ u8 fshift; /* 8 1 */ /* XXX 3 bytes hole, try to pack */ u32 handle; /* 12 4 */ u32 val; /* 16 4 */ u32 mask; /* 20 4 */ u32 link_handle; /* 24 4 */ /* XXX 4 bytes hole, try to pack */ struct tc_u32_sel * sel; /* 32 8 */ /* size: 40, cachelines: 1, members: 7 */ /* sum members: 33, holes: 2, sum holes: 7 */ /* last cacheline: 40 bytes */ }; after: struct tc_cls_u32_knode { struct tcf_exts * exts; /* 0 8 */ struct tc_u32_sel * sel; /* 8 8 */ u32 handle; /* 16 4 */ u32 val; /* 20 4 */ u32 mask; /* 24 4 */ u32 link_handle; /* 28 4 */ u8 fshift; /* 32 1 */ /* size: 40, cachelines: 1, members: 7 */ /* padding: 7 */ /* last cacheline: 40 bytes */ }; Suggested-by:Jamal Hadi Salim <jhs@mojatatu.com> Signed-off-by:
John Fastabend <john.r.fastabend@intel.com> Signed-off-by:
-
John Fastabend authored
Fixes: 9d35cf06 ("net: ixgbe: add minimal parser details for ixgbe") Reported-by:
Mark Rustad <mark.d.rustad@intel.com> Signed-off-by:
Jeff Kirsher <jeffrey.t.kirsher@intel.com> Signed-off-by:
-
John Fastabend authored
I incorrectly used __u32 types where we should be using u32 types when I added the ixgbe_model.h file. Fixes: 9d35cf06 ("net: ixgbe: add minimal parser details for ixgbe") Suggested-by:
-
- 17 Feb, 2016 17 commits
-
-
David S. Miller authored
Ben Hutchings says: ==================== Remove the inet_lro library The old inet_lro library has been deprecated ever since GRO was introduced, but there are still a few drivers using it. Convert them to GRO and remove it. ==================== Signed-off-by: -
Ben Hutchings authored
There are no longer any in-tree drivers that use it. Signed-off-by:
Ben Hutchings <ben@decadent.org.uk> Signed-off-by:
-
Ben Hutchings authored
GRO is simpler to use than the old inet_lro library, and is compatible with forwarding and bridging configurations. Compile-tested only. Signed-off-by:
-
Ben Hutchings authored
GRO is simpler to use than the old inet_lro library, and is compatible with forwarding and bridging configurations. Compile-tested only. Signed-off-by:
-
One Thousand Gnomes authored
The timeout is a long, we return it truncated if it is huge. Basically harmless as the only caller does a boolean check, but tidy it up anyway. (64bit build tested this time. Thank you 0day) Signed-off-by:
Alan Cox <alan@linux.intel.com> Signed-off-by:
-
Yuval Mintz authored
This patch moves the qed* driver into utilizing the 8.7.3.0 FW. This new FW is required for a lot of new SW features, including: - Vlan filtering offload - Encapsulation offload support - HW ingress aggregations As well as paving the way for the possibility of adding storage protocols in the future. V2: - Fix kbuild test robot error/warnings. Signed-off-by:
Yuval Mintz <Yuval.Mintz@qlogic.com> Signed-off-by:
Sudarsana Reddy Kalluru <Sudarsana.Kalluru@qlogic.com> Signed-off-by:
Manish Chopra <manish.chopra@qlogic.com> Signed-off-by:
-
David S. Miller authored
Xin Long says: ==================== sctp: some cleanups for sctp There are some unused function and redundant rcu locks should be removed. ==================== Signed-off-by: -
Xin Long authored
Since commit 8b570dc9 ("sctp: only drop the reference on the datamsg after sending a msg") used sctp_datamsg_put in sctp_sendmsg, instead of sctp_datamsg_free, this function has no use in sctp. So we will remove it. Signed-off-by:
Xin Long <lucien.xin@gmail.com> Signed-off-by:
-
Xin Long authored
sctp_seq_dump_remote_addrs is only called by sctp_assocs_seq_show() and it has been protected by rcu_read_lock that is from rhashtable_walk_start(). So we will remove this one. Signed-off-by:
-
Xin Long authored
__sctp_lookup_association() is only invoked by sctp_v4_err() and sctp_rcv(), both which run on the rx BH, and it has been protected by rcu_read_lock [see ip_local_deliver_finish() / ipv6_rcv()]. So we can move it to sctp_lookup_association, only let sctp_lookup_association use rcu_read_lock. Signed-off-by:
-
Rosen, Rami authored
commit 3ed80a62 (cgroup: drop module support) made including module.h redundant in the net cgroup controllers, netclassid_cgroup.c and netprio_cgroup.c. This patch removes them. Signed-off-by:
Rami Rosen <rami.rosen@intel.com> Acked-by:
Tejun Heo <tj@kernel.org> Signed-off-by:
-
David S. Miller authored
Sergei Shtylyov says: ==================== Kill useless *switch* defaults in the Renesas Ethernet drivers Here's a set of 2 patches against DaveM's 'net-next.git' repo removing often used *default* cases doing nothing. [1/2] ravb: kill useless *switch* defaults [2/2] sh_eth: kill useless *switch* defaults ==================== Signed-off-by: -
Sergei Shtylyov authored
The driver often has the *default* cases doing nothing in the *switch* statements with the integer expressions -- remove them. Signed-off-by:
Sergei Shtylyov <sergei.shtylyov@cogentembedded.com> Reviewed-by:
Simon Horman <horms+renesas@verge.net.au> Signed-off-by:
-
Sergei Shtylyov authored
The driver has the *default* case doing nothing in the *switch* statement with an integer expression -- remove it. Signed-off-by:
-
git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/next-queueDavid S. Miller authored
Jeff Kirsher says: ==================== 40GbE Intel Wired LAN Driver Updates 2016-02-17 This series contains updates to i40e/i40evf only. Jesse cleans up a duplicate declaration in probe. Then fixes a bug where the driver was using a offset based off a DMA handle while mapping and unmapping using sync_single_range_for_[cpu|device], when it should be using DMA handle (returned from alloc_coherent) and the offset of the memory to be sync'd. Fixed an issue where sync_vsi_filter() was allocating memory in a way that could sleep (GFP_KERNEL) which was causing a problem when called by the team driver under rcu_read_lock(). Shannon adds a check to ensure we do not attempt to do TSO when skb->ip_summed is not set to CHECKSUM_PARTIAL. Kiran add functions related to port mirroring features such as add/delete mirror rule. Jacob assigns the i40e_pf structure directly instead of using a large memcpy, to avoid a sparse warning and lets the compiler optimize the copy since it know the size of the structure in advance. Anjali enables GENEVE capability for XL710/X710 devices with firmware API version higher than 1.4. Added flag for automatic rule eviction feature for X722, which is disabled by default. ==================== Signed-off-by: -
David S. Miller authored
Tom Lendacky says: ==================== amd-xgbe: AMD XGBE driver updates 2016-02-12 The following updates and fixes are included in this driver update series: - Disable VLAN filtering in promiscuous mode - Change from using napi_complete to napi_complete_done - Use __napi_schedule_irqoff when running in interrupt context - Verify ethtool speed setting is valid for the selected speedset - Enable PFC based on the pfc_en setting - Fix the mapping of priorities to traffic classes - Do traffic class setup when DCB nl callbacks are invoked - Check Rx queue fifos before stopping Rx queue DMA - Switch from disable_irq to masking interrupts for auto-negotiation This patch series is based on net-next. Changes from v1: - Removed #ifndef and #define of CRCPOLY_LE as part of the patch to disable VLAN filtering in promiscuous mode - Reworked changes to xgbe_setup_tc to resolve conflicts with commit 16e5cc64 (net: rework setup_tc ndo op to consume general tc operand) ==================== Signed-off-by:
-
Lendacky, Thomas authored
Currently the auto-negotiation interrupt handling disables the irq instead of masking off the interrupts. This was done because the phy library was originally used to read and write the PCS registers, which could not be performed in interrupt context. Now that the phy library is no longer used to read and write the PCS registers the interrupts can be masked off in the interrupt service routine eliminating the need to call disable_irq/enable_irq. This also requires changing the protection mutex to a spinlock. Signed-off-by:
Tom Lendacky <thomas.lendacky@amd.com> Signed-off-by:
-
