- 16 Jan, 2018 16 commits
-
-
David S. Miller authored
Jiri Pirko says: ==================== devlink: Add support for resource abstraction Arkadi says: Many of the ASIC's internal resources are limited and are shared between several hardware procedures. For example, unified hash-based memory can be used for many lookup purposes, like FDB and LPM. In many cases the user can provide a partitioning scheme for such a resource in order to perform fine tuning for his application. In such cases performing driver reload is needed for the changes to take place, thus this patchset also adds support for hot reload. Such an abstraction can be coupled with devlink's dpipe interface, which models the ASIC's pipeline as a graph of match/action tables. By modeling the hardware resource object, and by coupling it to several dpipe tables, further visibility can be achieved in order to debug ASIC-wide issues. The proposed interface will provide the user the ability to understand the limitations of the hardware, and receive notification regarding its occupancy. Furthermore, monitoring the resource occupancy can be done in real-time and can be useful in many cases. --- v2->v3 - Mix/Max/Gran attributes. - Add resource consumption per table. - Change basic resource unit to 'entry'. - ABI documentation. v1->v2 - Add resource size attribute. - Fix split bug. ==================== Signed-off-by:David S. Miller <davem@davemloft.net>
-
Arkadi Sharshevsky authored
Add resources ABI documentation. Signed-off-by:
Arkadi Sharhsevsky <arkadis@mellanox.com> Signed-off-by:
Jiri Pirko <jiri@mellanox.com> Signed-off-by:
-
Arkadi Sharshevsky authored
Add support for hot reload. First, all the driver/core resources are released but the PCI and devlink instances, then reset is performed through the PCI interface. Finally the driver performs initialization. In case of reload failure the driver is left in a partially initialized state. Special care is taken during the driver removal in order to properly handle this state. Signed-off-by:
-
Arkadi Sharshevsky authored
Up until now the KVD partition was static. This patch introduces the ability to get the resource sizes via devlink. In case the resource is not available the default configuration is used. Signed-off-by:
-
Arkadi Sharshevsky authored
Add support for getting the kvdl occupancy through the resource interface. Signed-off-by:
-
Arkadi Sharshevsky authored
Connect current dpipe tables to resources. The tables are connected in the following fashion: 1. IPv4 host -> KVD hash single 2. IPv6 host -> KVD hash double 3. Adjacency -> KVD linear Signed-off-by:
-
Arkadi Sharshevsky authored
Register the KVD resources with devlink. The KVD is a memory resource which is subdivided into three partitions which are the linear, hash single and hash double. Signed-off-by:
-
Arkadi Sharshevsky authored
This is a preparation stage before introducing hot reload. During the reload process the ASIC should be resetted by accessing the PCI BAR due to unavailability of the mailbox/emad interfaces. Signed-off-by:
-
Arkadi Sharshevsky authored
The hardware processes which are modeled via dpipe commonly use some internal hardware resources. Such relation can improve the understanding of hardware limitations. The number of resource's unit consumed per table's entry are also provided for each table. Signed-off-by:
-
Arkadi Sharshevsky authored
Add support for performing driver hot reload. Signed-off-by:
-
Arkadi Sharshevsky authored
Add support for hardware resource abstraction over devlink. Each resource is identified via id, furthermore it contains information regarding its size and its related sub resources. Each resource can also provide its current occupancy. In some cases the sizes of some resources can be changed, yet for those changes to take place a hot driver reload may be needed. The reload capability will be introduced in the next patch. Signed-off-by:
-
Arkadi Sharshevsky authored
This is a preparation before introducing resources and hot reload support. Currently there are two global lock where one protects all devlink access, and the second one protects devlink port access. This patch adds per devlink instance lock which protects the internal members which are the sb/dpipe/ resource/ports. By introducing this lock the global devlink port lock can be discarded. Signed-off-by:
-
Heiner Kallweit authored
Make use of the new helpers for paged register access. Signed-off-by:
Heiner Kallweit <hkallweit1@gmail.com> Signed-off-by:
-
David S. Miller authored
Heiner Kallweit says: ==================== phy: add helpers for setting/clearing bits in PHY registers Based on the recent introduction of phy_modify add helpers for setting and clearing bits in PHY registers. First user is phylib. ==================== Signed-off-by: -
Heiner Kallweit authored
Use new helpers phy_set_bits / phy_clear_bits in phylib. Signed-off-by:
-
Heiner Kallweit authored
Based on the recent introduction of phy_modify add helpers for setting and clearing bits in PHY registers. Signed-off-by:
-
- 15 Jan, 2018 17 commits
-
-
David S. Miller authored
Merge tag 'linux-can-next-for-4.16-20180105' of ssh://gitolite.kernel.org/pub/scm/linux/kernel/git/mkl/linux-can-next Marc Kleine-Budde says: ==================== pull-request: can-next 2017-12-01,Re: pull-request: can-next this is a pull request of 7 patches for net-next/master. All patches are by me. Patch 6 is for the "can_raw" protocol and add error checking to the bind() function. All other patches clean up the coding style and remove unused parameters in various CAN drivers and infrastructure. ==================== Signed-off-by:
-
David S. Miller authored
Sergei Shtylyov says: ==================== sh_eth: simplify TSU initialization Here's a set of 2 patches against DaveM's 'net-next.git' repo. With those, I'm somewhat simplifying the TSU init code in the driver probe() method... ==================== Signed-off-by: -
Sergei Shtylyov authored
The dual-port Ether configurations always have a shared TSU to e.g. pass the packets between those ports. With the TSU init. code gathered under the single *if*, we now can only get the port # from 'platform_device::id' only when we actually need it (and not recalculate it each time)... Signed-off-by:
Sergei Shtylyov <sergei.shtylyov@cogentembedded.com> Reviewed-by:
Geert Uytterhoeven <geert+renesas@glider.be> Signed-off-by:
-
Sergei Shtylyov authored
The sh_eth_cpu_data::chip_reset() method always resets using ARSTR and this register is always located at the start of the TSU register region. Therefore, we can only call this method if we know TSU is there and thus simplify the probing code a bit... Signed-off-by:
-
David S. Miller authored
Merge tag 'wireless-drivers-next-for-davem-2018-01-13' of git://git.kernel.org/pub/scm/linux/kernel/git/kvalo/wireless-drivers-next Kalle Valo says: ==================== wireless-drivers-next patches for 4.16 Here are patches which have been accumulating over the holidays and after the New Year. Business as usual and nothing special really standing out. But what's noteworthy here is that Larry Finger is stepping down as the rtlwifi maintainer. He has been maintaining rtlwifi since it was applied back in 2010 in commit 0c817338 ("rtl8192ce: Add new driver") and it has been no easy role trying to juggle between the vendor, demanding upstream community and users. So big thank you to Larry for all his efforts! ath10k * more preparation work for wcn3990 support * add memory dump to firmware coredump files wil6210 * support scheduled scan * support 40-bit DMA addresses qtnfmac * support MAC address based access control * support for radar detection and Channel Availibility Check (CAC) mwifiex * firmware coredump for usb devices rtlwifi * Larry Finger steps down as the maintainer and Ping-Ke Shih becomes the new maintainer * add debugfs interfaces to dump register and btcoex status, and also write registers and h2c ==================== Signed-off-by:
-
Linus Walleij authored
The Gemini ethernet has been around for years as an out-of-tree patch used with the NAS boxen and routers built on StorLink SL3512 and SL3516, later Storm Semiconductor, later Cortina Systems. These ASICs are still being deployed and brand new off-the-shelf systems using it can easily be acquired. The full name of the IP block is "Net Engine and Gigabit Ethernet MAC" commonly just called "GMAC". The hardware block contains a common TCP Offload Enginer (TOE) that can be used by both MACs. The current driver does not use it. Cc: Tobias Waldvogel <tobias.waldvogel@gmail.com> Signed-off-by:
Michał Mirosław <mirq-linux@rere.qmqm.pl> Signed-off-by:
Linus Walleij <linus.walleij@linaro.org> Signed-off-by:
-
Linus Walleij authored
This adds the device tree bindings for the Gemini ethernet controller. It is pretty straight-forward, using standard bindings and modelling the two child ports as child devices under the parent ethernet controller device. Cc: devicetree@vger.kernel.org Cc: Tobias Waldvogel <tobias.waldvogel@gmail.com> Cc: Michał Mirosław <mirq-linux@rere.qmqm.pl> Reviewed-by:
Rob Herring <robh@kernel.org> Signed-off-by:
-
Ido Schimmel authored
Emil reported the following compiler errors: net/ipv6/route.c: In function `rt6_sync_up`: net/ipv6/route.c:3586: error: unknown field `nh_flags` specified in initializer net/ipv6/route.c:3586: warning: missing braces around initializer net/ipv6/route.c:3586: warning: (near initialization for `arg.<anonymous>`) net/ipv6/route.c: In function `rt6_sync_down_dev`: net/ipv6/route.c:3695: error: unknown field `event` specified in initializer net/ipv6/route.c:3695: warning: missing braces around initializer net/ipv6/route.c:3695: warning: (near initialization for `arg.<anonymous>`) Problem is with the named initializers for the anonymous union members. Fix this by adding curly braces around the initialization. Fixes: 4c981e28 ("ipv6: Prepare to handle multiple netdev events") Signed-off-by:
Ido Schimmel <idosch@mellanox.com> Reported-by:
Emil S Tantilov <emils.tantilov@gmail.com> Tested-by:
-
Jon Maloy authored
In commit 232d07b7 ("tipc: improve groupcast scope handling") we inadvertently broke non-group multicast transmission when changing the parameter 'domain' to 'scope' in the function tipc_nametbl_lookup_dst_nodes(). We missed to make the corresponding change in the calling function, with the result that the lookup always fails. A closer anaysis reveals that this parameter is not needed at all. Non-group multicast is hard coded to use CLUSTER_SCOPE, and in the current implementation this will be delivered to all matching destinations except those which are published with NODE_SCOPE on other nodes. Since such publications never will be visible on the sending node anyway, it makes no sense to discriminate by scope at all. We now remove this parameter altogether. Signed-off-by:
Jon Maloy <jon.maloy@ericsson.com> Signed-off-by:
-
Kirill Tkhai authored
Since net could be obtained from RCU lists, and there is a race with net destruction, the patch converts net::count to refcount_t. This provides sanity checks for the cases of incrementing counter of already dead net, when maybe_get_net() has to used instead of get_net(). Drivers: allyesconfig and allmodconfig are OK. Suggested-by:
Eric Dumazet <eric.dumazet@gmail.com> Signed-off-by:
Kirill Tkhai <ktkhai@virtuozzo.com> Reviewed-by:
Eric Dumazet <edumazet@google.com> Signed-off-by:
-
Marcelo Ricardo Leitner authored
Signed-off-by:
Marcelo Ricardo Leitner <marcelo.leitner@gmail.com> Acked-by:
Neil Horman <nhorman@tuxdriver.com> Reviewed-by:
Xin Long <lucien.xin@gmail.com> Signed-off-by:
-
Heiner Kallweit authored
I see two issues with parameter new_link: 1. It's not needed. See also phy_interrupt(), works w/o this parameter. phy_mac_interrupt sets the state to PHY_CHANGELINK and triggers the state machine which then calls phy_read_status. And phy_read_status updates the link state. 2. phy_mac_interrupt is used in interrupt context and getting the link state may sleep (at least when having to access the PHY registers via MDIO bus). So let's remove it. Signed-off-by:
Florian Fainelli <f.fainelli@gmail.com> Tested-by:
-
Jon Maloy authored
In commit d12d2e12 "tipc: send out join messages as soon as new member is discovered") we added a call to the function tipc_group_join() without considering the case that the preceding tipc_sk_publish() might have failed, and the group item already deleted. We fix this by returning from tipc_sk_join() directly after the failed tipc_sk_publish. Reported-by: syzbot+e3eeae78ea88b8d6d858@syzkaller.appspotmail.com Signed-off-by:
-
David S. Miller authored
Phil Reid says: ==================== net: dsa: lan9303: check error value from devm_gpiod_get_optional() Errors need to be prograted back from probe. Note: I have only compile tested the code as I don't have the hardware. Egil Hjelmeland <privat@egil-hjelmeland.no> has tested it but I haven't added at Test-by: wasn't in the standard form. Not sure if that's ok or not. Changes from v1: - rebased on net-next ==================== Signed-off-by: -
Phil Reid authored
devm_gpiod_get_optional() can return an error in addition to a NULL ptr. Check for error and propagate that to the probe function. Check return value in probe. This will now handle EPROBE_DEFER for the reset gpio. Signed-off-by:
Phil Reid <preid@electromag.com.au> Signed-off-by:
-
Phil Reid authored
lan9303_handle_reset never returns anything other than success. So there's not need for it to return an error code. Signed-off-by:
-
Andrew Lunn authored
__phy_modify would return the old value of the register before it was modified. Thus on success, it does not return 0, but a positive value. Thus functions using phy_modify, which is a wrapper around __phy_modify, can start returning > 0 on success, rather than 0. As a result, breakage has been noticed in various places, where 0 was assumed. Code inspection does not find any current location where the return of the old value is currently used. So have __phy_modify return 0 on success. When there is a real need for the old value, either a new accessor can be added, or an additional parameter passed. Fixes: fea23fb5 ("net: phy: convert read-modify-write to phy_modify()") Fixes: 2b74e5be ("net: phy: add phy_modify() accessor") Reported-by:
Geert Uytterhoeven <geert@linux-m68k.org> Tested-by:
Andrew Lunn <andrew@lunn.ch> Tested-by:
Niklas Cassel <niklas.cassel@axis.com> Signed-off-by:
-
- 14 Jan, 2018 7 commits
-
-
git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/next-queueDavid S. Miller authored
Jeff Kirsher says: ==================== 10GbE Intel Wired LAN Driver Updates 2018-01-12 This series contains updates to ixgbe, fm10k and net core. Alex updates the driver to remove a duplicate MAC address check and verifies that we have not run out of resources to configure a MAC rule in our filter table. Also do not assume that dev->num_tc was populated and configured with the driver, since it can be configured via mqprio without any hardware coordination. Fixed the recording of stats for MACVLAN in ixgbe and fm10k instead of recording the receive queue on MACVLAN offloaded frames. When handling a MACVLAN offload, we should be stopping/starting traffic on our own queues instead of the upper devices transmit queues. Fixed possible race conditions with the MACVLAN cleanup with the interface cleanup on shutdown. With the recent fixes to ixgbe, we can cap the number of queues regardless of accel_priv being in use or not, since the actual number of queues are being reported via real_num_tx_queues. Tony fixes up the kernel documentation for ixgbe and ixgbevf to resolve warnings when W=1 is used. ==================== Signed-off-by: -
David S. Miller authored
Jiri Pirko says: ==================== mlxsw: Offload PRIO qdisc Nogah says: Add an offload support for PRIO qdisc for mlxsw driver. PRIO qdisc is being offloaded by using ndo_setup_tc. It has three commands, to set or tune the qdisc, to remove it and to get its stats. Like RED offloading, offloading this qdisc is not enforced on the driver and determining its offload state is done in the dump action, when the stats are being updated. In the driver, offloading of PRIO is supported as root qdisc only. It supports only priorities 0-7 (the range that is used by the current static mapping of DSCP to skb prio and by 1:1 PCP values mapping) and up to 8 bands. Patches 1-2 offload DSCP to priority mapping in the mlxsw_sp driver. Patch 3 adds offload support for PRIO qdisc. Patches 4-5 Add PRIO offload support in the mlxsw_sp driver. --- v1->v2: - Patch 1/5: - Rewrite patch msg - Patch 3/5: - Send all the qstats in the replace command (and not just backlog) - Patch 5/5: - Align with the changes from 3/5 - Move backlog to the generic qdisc stats struct - Delete extra newline ==================== Signed-off-by: -
Nogah Frankel authored
Support basic stats for PRIO qdisc, which includes tx packets and bytes count, drops count and backlog size. The rest of the stats are irrelevant for this qdisc offload. Since backlog is not only incremental but reflecting momentary value, in case of a qdisc that stops being offloaded but is not destroyed, backlog value needs to be updated about the un-offloading. For that reason an unoffload function is being added to the ops struct. Signed-off-by:
Nogah Frankel <nogahf@mellanox.com> Reviewed-by:
Yuval Mintz <yuvalm@mellanox.com> Signed-off-by:
-
Nogah Frankel authored
Add support for offloading PRIO qdisc as root qdisc. The support is for up to 8 bands. Routed packets priority is determined by the DSCP field with the default translations. Bridged packets priority is determined by the PCP field, if exist, otherwise it is set to 0. Since both options have only priorities 0-7, higher priorities mapping are being ignored. Signed-off-by:
-
Nogah Frankel authored
Add the ability to offload PRIO qdisc by using ndo_setup_tc. There are three commands for PRIO offloading: * TC_PRIO_REPLACE: handles set and tune * TC_PRIO_DESTROY: handles qdisc destroy * TC_PRIO_STATS: updates the qdiscs counters (given as reference) Like RED qdisc, the indication of whether PRIO is being offloaded is being set and updated as part of the dump function. It is so because the driver could decide to offload or not based on the qdisc parent, which could change without notifying the qdisc. Signed-off-by:
-
Yuval Mintz authored
When routing ip packets, the kernel is setting the SKB's priority based on the tos field of the packet. Imitate this behavior in the mlxsw router, having the internal switch priority of a routed packet determined according to its DS field. Signed-off-by:
-
Yuval Mintz authored
Add rdpm definition - router DSCP to priority mapping register. This register will be utilized later to align the default mapping between packet DSCP and switch-priority to the kernel's mapping between packet priority and skb priority. This is the first non-bit indexed register where the entries are arranged in descending order, i.e., entry at offset 0 matches configuration for dscp[63]. As a result, the item's step is converted into a signed variable to support descending arrays [where step would be negative]. Signed-off-by:
-
