OSDN Git Service

tomoyo/tomoyo-test1.git
7 years agoMerge branch 'qed-Enhance-storage-APIs'
David S. Miller [Fri, 2 Jun 2017 14:33:04 +0000 (10:33 -0400)]
Merge branch 'qed-Enhance-storage-APIs'

Yuval Mintz says:

====================
qed: Enhance storage APIs

This series is intended to add additional information and features
to the API between qed and its storage protocol drivers [qedi, qedf].

Patch #2 adds some information stored on device such as wwpn & wwnn
to allow qedf utilize it; #1 fixes an issue with the reading of those
values [which were unused until now].

Patch #3 would allow the protocol drivers access to images on persistent
storage which is a prerequirement for adding boot from SAN support.

Patch #4 adds infrastrucutre to a future feature for qedi.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoqed: Add support for changing iSCSI mac
Mintz, Yuval [Fri, 2 Jun 2017 05:58:33 +0000 (08:58 +0300)]
qed: Add support for changing iSCSI mac

Enhance API between qedi and qed, allowing qedi to inform device's
firmware when the iSCSI mac is to be changed.

Signed-off-by: Yuval Mintz <Yuval.Mintz@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoqed: Support NVM-image reading API
Mintz, Yuval [Fri, 2 Jun 2017 05:58:32 +0000 (08:58 +0300)]
qed: Support NVM-image reading API

Storage drivers require images from the nvram in boot-from-SAN
scenarios. This provides the necessary API between qed and the
protocol drivers to perform such reads.

Signed-off-by: Yuval Mintz <Yuval.Mintz@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoqed: Share additional information with qedf
Mintz, Yuval [Fri, 2 Jun 2017 05:58:31 +0000 (08:58 +0300)]
qed: Share additional information with qedf

Share several new tidbits with qedf:
 - wwpn & wwnn
 - Absolute pf-id [this one is actually meant for qedi as well]
 - Number of available CQs

While we're at it, now that qedf will be aware of the available CQs
we can add some validation on the inputs it provides.

Signed-off-by: Yuval Mintz <Yuval.Mintz@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoqed: Correct order of wwnn and wwpn
Mintz, Yuval [Fri, 2 Jun 2017 05:58:30 +0000 (08:58 +0300)]
qed: Correct order of wwnn and wwpn

Driver reads values via HSI splitting this 8-byte into 2 32-bit
values and builds a single u64 field - but it does so by shifting
the lower field instead of the higher.
Luckily, we still don't use these fields for anything - but we're about
to start.

Signed-off-by: Yuval Mintz <Yuval.Mintz@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoteam: add macro MODULE_ALIAS_TEAM_MODE for team mode alias
Zhang Shengju [Thu, 1 Jun 2017 07:37:02 +0000 (15:37 +0800)]
team: add macro MODULE_ALIAS_TEAM_MODE for team mode alias

Add a new macro MODULE_ALIAS_TEAM_MODE to unify and simplify the
declaration of team mode alias.

Signed-off-by: Zhang Shengju <zhangshengju@cmss.chinamobile.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoMerge branch 'dsa-tagger-simplification'
David S. Miller [Thu, 1 Jun 2017 21:34:57 +0000 (17:34 -0400)]
Merge branch 'dsa-tagger-simplification'

Vivien Didelot says:

====================
net: dsa: tagger simplification

This series clarifies the hot path, removes the labels in tagging
implementations, and frees the original SKB in the xmit caller.

Changes in v3:
  - drop removal of usused rcv arguments because they will be used later
  - include the new ksz tagging implementation
  - add reviewers' tags

Changes in v2:
  - do not remove tagger function copies
  - document hot path requirements
  - make netdev_uses_dsa simpler
  - add reviewers' tags
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: dsa: factor skb freeing on xmit
Vivien Didelot [Thu, 1 Jun 2017 20:07:15 +0000 (16:07 -0400)]
net: dsa: factor skb freeing on xmit

As of a86d8becc3f0 ("net: dsa: Factor bottom tag receive functions"),
the rcv caller frees the original SKB in case or error.

Be symmetric with that and make the xmit caller do the same.

At the same time, fix the checkpatch NULL comparison check:

        CHECK: Comparison to NULL could be written "!nskb"
    #208: FILE: net/dsa/tag_trailer.c:35:
    + if (nskb == NULL)

Signed-off-by: Vivien Didelot <vivien.didelot@savoirfairelinux.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: dsa: remove out_drop label in taggers rcv
Vivien Didelot [Thu, 1 Jun 2017 20:07:14 +0000 (16:07 -0400)]
net: dsa: remove out_drop label in taggers rcv

Many rcv functions from net/dsa/tag_*.c have a useless out_drop goto
label which simply returns NULL. Kill it in favor of the obvious.

Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Vivien Didelot <vivien.didelot@savoirfairelinux.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: dsa: remove dsa_uses_tagged_protocol
Vivien Didelot [Thu, 1 Jun 2017 20:07:13 +0000 (16:07 -0400)]
net: dsa: remove dsa_uses_tagged_protocol

Since dev->dsa_ptr is a pointer to a dsa_switch_tree, there is no need
to have another inline helper just to check rcv.

Remove dsa_uses_tagged_protocol and check dsa_ptr && dsa_ptr->rcv
together at the same time.

Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Vivien Didelot <vivien.didelot@savoirfairelinux.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: dsa: do not cast dst
Vivien Didelot [Thu, 1 Jun 2017 20:07:12 +0000 (16:07 -0400)]
net: dsa: do not cast dst

dsa_ptr is not a void pointer anymore since Nov 2011, as of cf50dcc24f82
("dsa: Change dsa_uses_{dsa, trailer}_tags() into inline functions"),
but an explicit dsa_switch_tree pointer, thus remove the (void *) cast.

Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Vivien Didelot <vivien.didelot@savoirfairelinux.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: dsa: comment hot path requirements
Vivien Didelot [Thu, 1 Jun 2017 20:07:11 +0000 (16:07 -0400)]
net: dsa: comment hot path requirements

The DSA layer uses inline helpers and copy of the tagging functions for
faster access in hot path. Add comments to detail that.

Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Vivien Didelot <vivien.didelot@savoirfairelinux.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agomlxsw: spectrum: Add bridge dependency for spectrum
Ido Schimmel [Wed, 31 May 2017 18:33:42 +0000 (21:33 +0300)]
mlxsw: spectrum: Add bridge dependency for spectrum

When BRIDGE is a loadable module, MLXSW_SPECTRUM mustn't be built-in:

drivers/built-in.o: In function `mlxsw_sp_bridge_device_create':
drivers/net/ethernet/mellanox/mlxsw/spectrum_switchdev.c:145: undefined reference to `br_vlan_enabled'
drivers/net/ethernet/mellanox/mlxsw/spectrum_switchdev.c:158: undefined reference to `br_multicast_enabled'
drivers/built-in.o: In function `mlxsw_sp_dev_rif_type':
drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c:2972: undefined reference to `br_vlan_enabled'
drivers/built-in.o: In function `mlxsw_sp_inetaddr_vlan_event':
drivers/net/ethernet/mellanox/mlxsw/spectrum_router.c:3310: undefined reference to `br_vlan_enabled'

Add Kconfig dependency to enforce usable configurations.

Fixes: c57529e1d5d8 ("mlxsw: spectrum: Replace vPorts with Port-VLAN")
Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Reported-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
Tested-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
Cc: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoMerge branch 'stmmac-add-dwmac-sun8i-ethernet-driver'
David S. Miller [Thu, 1 Jun 2017 18:53:10 +0000 (14:53 -0400)]
Merge branch 'stmmac-add-dwmac-sun8i-ethernet-driver'

Corentin Labbe says:

====================
net-next: stmmac: add dwmac-sun8i ethernet driver

This patch series add the driver for dwmac-sun8i which handle the Ethernet MAC
present on Allwinner H3/H5/A83T/A64 SoCs.

This driver is the continuation of the sun8i-emac driver.
During the development, it appeared that in fact the hardware was a modified
version of some dwmac.
So the driver is now written as a glue driver for stmmac.

It supports 10/100/1000 Mbit/s speed with half/full duplex.
It can use an internal PHY (MII 10/100) or an external PHY
via RGMII/RMII.

This patch series enable the driver only for the H3/A64/H5 SoC since A83T
doesn't have the necessary clocks present in mainline.

The driver have been tested on the following boards:
- H3 Orange PI PC, BananaPI-M2+
- A64 Pine64, BananaPi-M64
- A83T BananaPI-M3

The first two patchs are some mandatory changes for letting dwmac-sun8i be used.
The following three patchs add the driver and its documentation.
The remaining are DT patch enabling it.

Regards
Corentin Labbe

Changes since v5:
- Added DT patch for NanoPi neo
- Use the new adjust_link variables (speedxxx/speedmask)
- Made the timeout of readl_poll_timeout from 10 to 100ms
- Fix sun8i_unpower_phy that could be called twice
- Replace phy by phy-handle in doc/dwmac-sun8i.txt

Changes since v4:
- Re-ordered by alphabetical order some DT nodes
- Simplified power/unpower_phy functions by testing the use of internal_phy
- Added a patch for adding dwmac-sun8i to arm64 defconfig
- Fix a typo in sun50i-a64-system-controller (wrongly used sun8i)
- Reworked uc/mc filter address setting

Changes since v3:
- Renamed tx-delay/rx-delay to tx-delay-ps/rx-delay-ps
- fix syscon compatible example
- Changed parameter type for setup() function
- Dropped some DT patchs for boards which I could not test further

Changes since v2:
- corrected order of syscon compatible
- added compatible = "ethernet-phy-ieee802.3-c22 to PHY
- added set_mac function

Changes since v1:
- added TX/RX delay units
- splitted syscon documentation in its own patch
- regulator is now disabled after clk_prepare_enable(gmac->tx_clk) error
- Fixed a memory leak on mac_device_info
- Use now generic pin config for all DT stuff
- CONFIG_DWMAC_SUN8I is now set to y in defconfigs
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoarm64: defconfig: Enable dwmac-sun8i driver on defconfig
LABBE Corentin [Wed, 31 May 2017 07:18:52 +0000 (09:18 +0200)]
arm64: defconfig: Enable dwmac-sun8i driver on defconfig

Enable the dwmac-sun8i ethernet driver as a module in the ARM64 defconfig.

Signed-off-by: Corentin Labbe <clabbe.montjoie@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoarm: multi_v7: Enable dwmac-sun8i driver on multi_v7_defconfig
LABBE Corentin [Wed, 31 May 2017 07:18:51 +0000 (09:18 +0200)]
arm: multi_v7: Enable dwmac-sun8i driver on multi_v7_defconfig

Enable the dwmac-sun8i driver in the multi_v7 default configuration

Signed-off-by: Corentin Labbe <clabbe.montjoie@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoarm: sunxi: Enable dwmac-sun8i driver on sunxi_defconfig
LABBE Corentin [Wed, 31 May 2017 07:18:50 +0000 (09:18 +0200)]
arm: sunxi: Enable dwmac-sun8i driver on sunxi_defconfig

Enable the dwmac-sun8i driver in the sunxi default configuration

Signed-off-by: Corentin Labbe <clabbe.montjoie@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoarm64: allwinner: bananapi-m64: Enable dwmac-sun8i
LABBE Corentin [Wed, 31 May 2017 07:18:49 +0000 (09:18 +0200)]
arm64: allwinner: bananapi-m64: Enable dwmac-sun8i

The dwmac-sun8i  hardware is present on the BananaPi M64.
It uses an external PHY rtl8211e via RGMII.

Signed-off-by: Corentin Labbe <clabbe.montjoie@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoarm64: allwinner: pine64-plus: Enable dwmac-sun8i
LABBE Corentin [Wed, 31 May 2017 07:18:48 +0000 (09:18 +0200)]
arm64: allwinner: pine64-plus: Enable dwmac-sun8i

The dwmac-sun8i hardware is present on the pine64 plus.
It uses an external PHY rtl8211e via RGMII.

Signed-off-by: Corentin Labbe <clabbe.montjoie@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoarm64: allwinner: pine64: Enable dwmac-sun8i
LABBE Corentin [Wed, 31 May 2017 07:18:47 +0000 (09:18 +0200)]
arm64: allwinner: pine64: Enable dwmac-sun8i

The dwmac-sun8i hardware is present on the pine64
It uses an external PHY via RMII.

Signed-off-by: Corentin Labbe <clabbe.montjoie@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoarm64: allwinner: sun50i-a64: add dwmac-sun8i Ethernet driver
LABBE Corentin [Wed, 31 May 2017 07:18:46 +0000 (09:18 +0200)]
arm64: allwinner: sun50i-a64: add dwmac-sun8i Ethernet driver

The dwmac-sun8i is an Ethernet MAC that supports 10/100/1000 Mbit
connections. It is very similar to the device found in the Allwinner
H3, but lacks the internal 100 Mbit PHY and its associated control
bits.
This adds the necessary bits to the Allwinner A64 SoC .dtsi, but keeps
it disabled at this level.

Signed-off-by: Corentin Labbe <clabbe.montjoie@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoarm64: allwinner: sun50i-a64: Add dt node for the syscon control module
LABBE Corentin [Wed, 31 May 2017 07:18:45 +0000 (09:18 +0200)]
arm64: allwinner: sun50i-a64: Add dt node for the syscon control module

This patch add the dt node for the syscon register present on the
Allwinner A64.

Only two register are present in this syscon and the only one useful is
the one dedicated to EMAC clock.

Signed-off-by: Corentin Labbe <clabbe.montjoie@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoarm: sun8i: nanopi-neo: Enable dwmac-sun8i
LABBE Corentin [Wed, 31 May 2017 07:18:44 +0000 (09:18 +0200)]
arm: sun8i: nanopi-neo: Enable dwmac-sun8i

The dwmac-sun8i hardware is present on the NanoPi Neo.
It uses the internal PHY.
This patch create the needed emac node.

Signed-off-by: Corentin Labbe <clabbe.montjoie@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoarm: sun8i: orangepi-pc-plus: Set EMAC activity LEDs to active high
LABBE Corentin [Wed, 31 May 2017 07:18:43 +0000 (09:18 +0200)]
arm: sun8i: orangepi-pc-plus: Set EMAC activity LEDs to active high

On the Orange Pi PC Plus, the polarity of the LEDs on the RJ45 Ethernet
port were changed from active low to active high.

Signed-off-by: Chen-Yu Tsai <wens@csie.org>
Signed-off-by: Corentin Labbe <clabbe.montjoie@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoarm: sun8i: orangepi-2: Enable dwmac-sun8i
LABBE Corentin [Wed, 31 May 2017 07:18:42 +0000 (09:18 +0200)]
arm: sun8i: orangepi-2: Enable dwmac-sun8i

The dwmac-sun8i hardware is present on the Orange PI 2.
It uses the internal PHY.

This patch create the needed emac node.

Signed-off-by: Corentin Labbe <clabbe.montjoie@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoarm: sun8i: orangepi-one: Enable dwmac-sun8i
LABBE Corentin [Wed, 31 May 2017 07:18:41 +0000 (09:18 +0200)]
arm: sun8i: orangepi-one: Enable dwmac-sun8i

The dwmac-sun8i hardware is present on the Orange PI One.
It uses the internal PHY.

This patch create the needed emac node.

Signed-off-by: Corentin Labbe <clabbe.montjoie@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoarm: sun8i: orangepi-zero: Enable dwmac-sun8i
LABBE Corentin [Wed, 31 May 2017 07:18:40 +0000 (09:18 +0200)]
arm: sun8i: orangepi-zero: Enable dwmac-sun8i

The dwmac-sun8i hardware is present on the Orange PI Zero.
It uses the internal PHY.

This patch create the needed emac node.

Signed-off-by: Corentin Labbe <clabbe.montjoie@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoarm: sun8i: orangepi-pc: Enable dwmac-sun8i
LABBE Corentin [Wed, 31 May 2017 07:18:39 +0000 (09:18 +0200)]
arm: sun8i: orangepi-pc: Enable dwmac-sun8i

The dwmac-sun8i hardware is present on the Orange PI PC.
It uses the internal PHY.

This patch create the needed emac node.

Signed-off-by: Corentin Labbe <clabbe.montjoie@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoarm: sun8i: sunxi-h3-h5: add dwmac-sun8i ethernet driver
LABBE Corentin [Wed, 31 May 2017 07:18:38 +0000 (09:18 +0200)]
arm: sun8i: sunxi-h3-h5: add dwmac-sun8i ethernet driver

The dwmac-sun8i is an ethernet MAC hardware that support 10/100/1000
speed.

This patch enable the dwmac-sun8i on Allwinner H3/H5 SoC Device-tree.
SoC H3/H5 have an internal PHY, so optionals syscon and ephy are set.

Signed-off-by: Corentin Labbe <clabbe.montjoie@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoarm: sun8i: sunxi-h3-h5: Add dt node for the syscon control module
LABBE Corentin [Wed, 31 May 2017 07:18:37 +0000 (09:18 +0200)]
arm: sun8i: sunxi-h3-h5: Add dt node for the syscon control module

This patch add the dt node for the syscon register present on the
Allwinner H3/H5

Only two register are present in this syscon and the only one useful is
the one dedicated to EMAC clock..

Signed-off-by: Corentin Labbe <clabbe.montjoie@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet-next: stmmac: Add dwmac-sun8i
LABBE Corentin [Wed, 31 May 2017 07:18:36 +0000 (09:18 +0200)]
net-next: stmmac: Add dwmac-sun8i

The dwmac-sun8i is a heavy hacked version of stmmac hardware by
allwinner.
In fact the only common part is the descriptor management and the first
register function.

Signed-off-by: Corentin Labbe <clabbe.montjoie@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agodt-bindings: syscon: Add DT bindings documentation for Allwinner syscon
LABBE Corentin [Wed, 31 May 2017 07:18:35 +0000 (09:18 +0200)]
dt-bindings: syscon: Add DT bindings documentation for Allwinner syscon

This patch adds documentation for Device-Tree bindings for the
syscon present in allwinner devices.

Signed-off-by: Corentin Labbe <clabbe.montjoie@gmail.com>
Acked-by: Rob Herring <robh@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agodt-bindings: net-next: Add DT bindings documentation for Allwinner dwmac-sun8i
LABBE Corentin [Wed, 31 May 2017 07:18:34 +0000 (09:18 +0200)]
dt-bindings: net-next: Add DT bindings documentation for Allwinner dwmac-sun8i

This patch adds documentation for Device-Tree bindings for the
Allwinner dwmac-sun8i driver.

Signed-off-by: Corentin Labbe <clabbe.montjoie@gmail.com>
Acked-by: Rob Herring <robh@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet-next: stmmac: add optional setup function
LABBE Corentin [Wed, 31 May 2017 07:18:33 +0000 (09:18 +0200)]
net-next: stmmac: add optional setup function

Instead of adding more ifthen logic for adding a new mac_device_info
setup function, it is easier to add a function pointer to the function
needed.

Signed-off-by: Corentin Labbe <clabbe.montjoie@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet-next: stmmac: export stmmac_set_mac_addr/stmmac_get_mac_addr
LABBE Corentin [Wed, 31 May 2017 07:18:32 +0000 (09:18 +0200)]
net-next: stmmac: export stmmac_set_mac_addr/stmmac_get_mac_addr

Thoses symbol will be needed for the dwmac-sun8i ethernet driver.
For letting it to be build as module, they need to be exported.

Signed-off-by: Corentin Labbe <clabbe.montjoie@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agopowerpc: use asm-generic/socket.h as much as possible
Stephen Rothwell [Wed, 31 May 2017 05:43:37 +0000 (15:43 +1000)]
powerpc: use asm-generic/socket.h as much as possible

asm-generic/socket.h already has an exception for the differences that
powerpc needs, so just include it after defining the differences.

Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agomlxsw: spectrum: Implement the ethtool flash_device callback
Yotam Gigi [Thu, 1 Jun 2017 13:26:46 +0000 (16:26 +0300)]
mlxsw: spectrum: Implement the ethtool flash_device callback

Add callback to the ethtool flash_device op. This callback uses the mlxfw
module to flash the new firmware file to the device.

As the firmware flash process takes about 20 seconds and ethtool takes the
rtnl lock during the flash_device callback, release the rtnl lock at the
beginning of the flash process and take it again before leaving the
callback. This way, the rtnl is not held during the process. To make sure
the device does not get deleted during the flash process, take a reference
to it before releasing the rtnl lock.

Signed-off-by: Yotam Gigi <yotamg@mellanox.com>
Reviewed-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoMerge branch 'qed-Status-block-changes'
David S. Miller [Thu, 1 Jun 2017 16:17:21 +0000 (12:17 -0400)]
Merge branch 'qed-Status-block-changes'

Yuval Mintz says:

====================
qed: Status block changes

The device maintains a CAM mapping of the internal status blocks
and the various PF/VF MSI-x vector mappings.
During initialization, the driver reads the HW memory and constructs
a shadow SW implementation which it would later use for manipulation
of interrupts. E.g., when enabling VFs and setting their MSI-x tables.

The driver currently has some very strict assumptions on the order the
entries are placed in the CAM. Specifically, it assumes that all entries
belonging to a PF would be consecutive and in-order in the CAM, and that
the VF entries would then follow. But there's no actual HW constraint
enforcing this assumption [although management firmware does set it
accordingly to same assumption initially].

Since the CAM is re-configurable, there are now SW flows employeed
by other OSes that might cause the assumption to be invalid.
Such flows allow the PF to forfeit some of it's available interrupts
in favor of its VFs or vice versa.
While those are not employeed today by qed, we want to relax the
assumptions as much as we can -
both to allow functionality after PDA as well as allowing future
compatibility where the driver would be loaded after a newer one has
'dirtied' the CAM configuration.

In addition to patches meant for the above relaxation, the series
also contains various cleanups & refactoring for interrupt logic
[most of which is !semantic].
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoqed: No need to reset SBs on IOV init
Mintz, Yuval [Thu, 1 Jun 2017 12:29:11 +0000 (15:29 +0300)]
qed: No need to reset SBs on IOV init

Since we're resetting the IGU CAM each time we initialize the PF
device, there's no need to reset the VF SBs again when initializing
IOV.

Signed-off-by: Yuval Mintz <Yuval.Mintz@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoqed: Reset IGU CAM to default on init
Mintz, Yuval [Thu, 1 Jun 2017 12:29:10 +0000 (15:29 +0300)]
qed: Reset IGU CAM to default on init

The IGU CAM contains an assocaition between hardware SBs
and interrupt lines, and it can be dynamically configured
to allow more interrupts in one entity over another, specifically
for Re-distibution of SBs between a PF and its child VFs.

While we don't yet use this functionality, there are other
clients that do and as such its possible the information
passed from management firmware during initialization in
regard to the possible number of SBs doesn't accurately reflect
the current HW configuration.

The following changes are going to apply to the driver init sequence:

 a. PF is going to re-configure all entries belonging to itself and
    its child VFs in IGU CAM based on the management firmware info
    regarding the number of SBs that are supposed to exist there.

 b. PF is going to stop using the SB resource [management firmware
    provided information] for anything but the initialization.
    Instead, it would use the live-time counters it maintains for
    the numbers.

Signed-off-by: Yuval Mintz <Yuval.Mintz@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoqed: Hold a single array for SBs
Mintz, Yuval [Thu, 1 Jun 2017 12:29:09 +0000 (15:29 +0300)]
qed: Hold a single array for SBs

A PF today holds 2 different arrays - one holding information
about the HW configuration and one holding information about
the SBs that are used by the protocol drivers.
These arrays aren't really connected - e.g., protocol driver
initializing a given SB would not mark the same SB as occupied
in the HW shadow array.

Move into a single array [at least for PFs] - hold the mapping
of the driver-protocol SBs on the HW entry which they configure.

Signed-off-by: Yuval Mintz <Yuval.Mintz@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoqed: Provide auxiliary for getting free VF SB
Mintz, Yuval [Thu, 1 Jun 2017 12:29:08 +0000 (15:29 +0300)]
qed: Provide auxiliary for getting free VF SB

IOV code is very intrusive in its manipulation of the status block
database.
Add a new auxiliary function to allow the PF to find an available unused
status block to configure for a specific VF's MSI-x vector.

Signed-off-by: Yuval Mintz <Yuval.Mintz@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoqed: Remove assumption on SB order in IGU
Mintz, Yuval [Thu, 1 Jun 2017 12:29:07 +0000 (15:29 +0300)]
qed: Remove assumption on SB order in IGU

Current code assumes there's a known layout for SBs in the IGU,
where all the SBs of a single entity would be laid in consecutive
order of vectors.

While the assumption is still kept by management firmware, we already
have the necessary information to eliminate it, so no reason to keep
it in code.

Signed-off-by: Yuval Mintz <Yuval.Mintz@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoqed: Encapsulate interrupt counters in struct
Mintz, Yuval [Thu, 1 Jun 2017 12:29:06 +0000 (15:29 +0300)]
qed: Encapsulate interrupt counters in struct

We already have an API struct that contains interrupt-related
numbers. Use it to encapsulate all information relating to the
status of SBs as (used|free).

Signed-off-by: Yuval Mintz <Yuval.Mintz@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoqed: Add aux. function translating sb_id -> igu_sb_id
Mintz, Yuval [Thu, 1 Jun 2017 12:29:05 +0000 (15:29 +0300)]
qed: Add aux. function translating sb_id -> igu_sb_id

An additional step for relaxing the IGU order assumption, we now add
an auxiliary function that can be used for finding the HW status block
that's associated with a given MSI-x vector.

Signed-off-by: Yuval Mintz <Yuval.Mintz@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoqed: Distinguish between sb_id and igu_sb_id
Mintz, Yuval [Thu, 1 Jun 2017 12:29:04 +0000 (15:29 +0300)]
qed: Distinguish between sb_id and igu_sb_id

In qed code, sb_id means 2 different things:
  - An interrupt vector [usually when received as a parameter from
    a protocol driver, but not only] that's associated with a status
    block.

  - An index to a status block entity existing in HW.

This patch renames the references to the HW entity, adding an 'igu_'
prefix to allow an easier distinction.

Signed-off-by: Yuval Mintz <Yuval.Mintz@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoqed: IGU read revised
Mintz, Yuval [Thu, 1 Jun 2017 12:29:03 +0000 (15:29 +0300)]
qed: IGU read revised

As a first step for relaxing various assumptions done by driver
about the IGU mapping, the driver is now going to read the entire
IGU into a shadow copy, and mark in its database each status block
that's relevant for it.

Signed-off-by: Yuval Mintz <Yuval.Mintz@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoqed: Minor refactoring in interrupt code
Mintz, Yuval [Thu, 1 Jun 2017 12:29:02 +0000 (15:29 +0300)]
qed: Minor refactoring in interrupt code

Separate the portions controlling interrupt enablement form those
controlling the ability of HW to generate attentions.

Signed-off-by: Yuval Mintz <Yuval.Mintz@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoqed: Make qed_int_cau_conf_pi() static
Mintz, Yuval [Thu, 1 Jun 2017 12:29:01 +0000 (15:29 +0300)]
qed: Make qed_int_cau_conf_pi() static

Signed-off-by: Yuval Mintz <Yuval.Mintz@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: dsa: make function ksz_rcv static
Colin Ian King [Thu, 1 Jun 2017 12:22:42 +0000 (13:22 +0100)]
net: dsa: make function ksz_rcv static

function ksz_rcv can be made static as it does not need to be
in global scope. Reformat arguments to make it checkpatch warning
free too.

Cleans up sparse warning: "symbol 'ksz_rcv' was not declared. Should
it be static?"

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Reviewed-by: Woojung Huh <Woojung.Huh@microchip.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoppp: remove unnecessary bh disable in xmit path
Gao Feng [Thu, 1 Jun 2017 09:58:39 +0000 (17:58 +0800)]
ppp: remove unnecessary bh disable in xmit path

Since the commit 55454a565836 ("ppp: avoid dealock on recursive xmit"),
the PPP xmit path is protected by wrapper functions which disable the
bh already. So it is unnecessary to disable the bh again in the real
xmit path.

Signed-off-by: Gao Feng <gfree.wind@vip.163.com>
Acked-by: Guillaume Nault <g.nault@alphalink.fr>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoipv4: route: restore skb_dst_set in inet_rtm_getroute
Roopa Prabhu [Thu, 1 Jun 2017 05:53:25 +0000 (22:53 -0700)]
ipv4: route: restore skb_dst_set in inet_rtm_getroute

recent updates to inet_rtm_getroute dropped skb_dst_set in
inet_rtm_getroute. This patch restores it because it is
needed to release the dst correctly.

Fixes: 3765d35ed8b9 ("net: ipv4: Convert inet_rtm_getroute to rcu versions of route lookup")
Reported-by: John Fastabend <john.fastabend@gmail.com>
Signed-off-by: David Ahern <dsahern@gmail.com>
Signed-off-by: Roopa Prabhu <roopa@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agobpf: Take advantage of stack_depth tracking in sparc64 JIT
David S. Miller [Thu, 1 Jun 2017 02:35:00 +0000 (19:35 -0700)]
bpf: Take advantage of stack_depth tracking in sparc64 JIT

Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoMerge branch 'dsa-add-Microchip-KSZ9477-DSA-driver'
David S. Miller [Thu, 1 Jun 2017 00:56:40 +0000 (20:56 -0400)]
Merge branch 'dsa-add-Microchip-KSZ9477-DSA-driver'

Woojung Huh says:

====================
dsa: add Microchip KSZ9477 DSA driver

This series of patches is for Microchip KSZ9477 DSA driver.
KSZ9477 is 7 ports GigE switch with numerous advanced features.
5 ports are 10/100/1000 Mbps internal PHYs and 2 ports have
Interfaces to SGMII, RGMII, MII or RMII.

This patch supports VLAN, MDB, FDB and port mirroring offloads.

Welcome reviews and comments from community.

Note: Tests are performed on internal development board.

V5
- add missing MODULE_LICENSE

V4
- update per review comments
- cosmetic changes
- net/dsa/tag_ksz.c
  * skb_put() & memset() are changed to skb_put_padto()
- drivers/net/dsa/microchip/ksz_common.
   * vlan access mutex is updated
   * mib_names[] is changed to static const

V3
- update per review comments
- cosmetic changes
- drivers/net/dsa/microchip/ksz_common.c
  * clean up ksz_switch_chips[]
  * consolidate checking loops into functions
  * update mutex for better locking
  * replace devm_kmalloc_array() to devm_kcalloc()
- MAINTAINERS
  * add missing net/dsa/tag_ksz.c

V2
- update per review comments
- several cosmetic changes
- net/dsa/tag_ksz.c
  * constants are changed to defines
  * remove skb_linearize() in ksz_rcv()
  * ksz_xmit()checks skb tailroom before allocate new skb
- drivers/net/phy/micrel.c
  * remove PHY_HAS_MAGICANEG from ksphy_driver[]
- drivers/net/dsa/microchip/ksz_common.c
  * add timeout to avoid endless loop
  * port initialization is move to ksz_port_enable() instead of  ksz_setup_ports()
- Documentation/devicetree/bindings/net/dsa/ksz.txt
  * fix typo and indentations
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agodsa: add maintainer of Microchip KSZ switches
Woojung Huh [Wed, 31 May 2017 20:19:29 +0000 (20:19 +0000)]
dsa: add maintainer of Microchip KSZ switches

Adding maintainer of Microchip KSZ switches.

Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Woojung Huh <Woojung.Huh@microchip.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: dsa: Add Microchip KSZ switches binding
Woojung Huh [Wed, 31 May 2017 20:19:23 +0000 (20:19 +0000)]
net: dsa: Add Microchip KSZ switches binding

A sample SPI configuration for Microchip KSZ switches.

Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Woojung Huh <Woojung.Huh@microchip.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agodsa: add DSA switch driver for Microchip KSZ9477
Woojung Huh [Wed, 31 May 2017 20:19:19 +0000 (20:19 +0000)]
dsa: add DSA switch driver for Microchip KSZ9477

The KSZ9477 is a fully integrated layer 2, managed, 7 ports GigE switch
with numerous advanced features. 5 ports incorporate 10/100/1000 Mbps PHYs.
The other 2 ports have interfaces that can be configured as SGMII, RGMII, MII
or RMII. Either of these may connect directly to a host processor or
to an external PHY. The SGMII port may interface to a fiber optic transceiver.

This driver currently supports vlan, fdb, mdb & mirror dsa switch operations.

Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Woojung Huh <Woojung.Huh@microchip.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agophy: micrel: add Microchip KSZ 9477 Switch PHY support
Woojung Huh [Wed, 31 May 2017 20:19:13 +0000 (20:19 +0000)]
phy: micrel: add Microchip KSZ 9477 Switch PHY support

Adding Microchip 9477 Phy included in KSZ9477 Switch.

Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Woojung Huh <Woojung.Huh@microchip.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agodsa: add support for Microchip KSZ tail tagging
Woojung Huh [Wed, 31 May 2017 20:19:06 +0000 (20:19 +0000)]
dsa: add support for Microchip KSZ tail tagging

Adding support for the Microchip KSZ switch family tail tagging.

Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Woojung Huh <Woojung.Huh@microchip.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoMerge branch 'bpf-stack-tracker'
David S. Miller [Wed, 31 May 2017 23:29:48 +0000 (19:29 -0400)]
Merge branch 'bpf-stack-tracker'

Alexei Starovoitov says:

====================
bpf: stack depth tracking

Introduce tracking of bpf program stack depth in the verifier and use that
info to reduce bpf program stack consumption in the interpreter and x64 JIT.
Other JITs can take advantage of it as well in the future.
Most of the programs consume very little stack, so it's good optimization
in general and it's the first step toward bpf to bpf function calls.

Also use internal opcode for bpf_tail_call() marking to make clear
that jmp|call|x opcode is not uapi and may be used for actual
indirect call opcode in the future.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agobpf: take advantage of stack_depth tracking in x64 JIT
Alexei Starovoitov [Tue, 30 May 2017 20:31:35 +0000 (13:31 -0700)]
bpf: take advantage of stack_depth tracking in x64 JIT

Take advantage of stack_depth tracking in x64 JIT.
Round up allocated stack by 8 bytes to make sure it stays aligned
for functions called from JITed bpf program.

Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agobpf: change x86 JITed program stack layout
Alexei Starovoitov [Tue, 30 May 2017 20:31:34 +0000 (13:31 -0700)]
bpf: change x86 JITed program stack layout

in order to JIT programs with different stack sizes we need to
make epilogue and exception path to be stack size independent,
hence move auxiliary stack space from the bottom of the stack
to the top of the stack.
Nice side effect is that JITed function prologue becomes shorter
due to imm8 offset encoding vs imm32.

Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agobpf: use different interpreter depending on required stack size
Alexei Starovoitov [Tue, 30 May 2017 20:31:33 +0000 (13:31 -0700)]
bpf: use different interpreter depending on required stack size

16 __bpf_prog_run() interpreters for various stack sizes add .text
but not a lot comparing to run-time stack savings

   text    data     bss     dec     hex filename
  26350   10328     624   37302    91b6 kernel/bpf/core.o.before_split
  25777   10328     624   36729    8f79 kernel/bpf/core.o.after_split
  26970   10328     624   37922    9422 kernel/bpf/core.o.now

Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agobpf: fix stack_depth usage by test_bpf.ko
Alexei Starovoitov [Tue, 30 May 2017 20:31:32 +0000 (13:31 -0700)]
bpf: fix stack_depth usage by test_bpf.ko

test_bpf.ko doesn't call verifier before selecting interpreter or JITing,
hence the tests need to manually specify the amount of stack they consume.

Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agobpf: track stack depth of classic bpf programs
Alexei Starovoitov [Tue, 30 May 2017 20:31:31 +0000 (13:31 -0700)]
bpf: track stack depth of classic bpf programs

To track stack depth of classic bpf programs we only need
to analyze ST|STX instructions, since check_load_and_stores()
verifies that programs can load from stack only after write.

We also need to change the way cBPF stack slots map to eBPF stack,
since typical classic programs are using slots 0 and 1, so they
need to map to stack offsets -4 and -8 respectively in order
to take advantage of small stack interpreter and JITs.

Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agobpf: reconcile bpf_tail_call and stack_depth
Alexei Starovoitov [Tue, 30 May 2017 20:31:30 +0000 (13:31 -0700)]
bpf: reconcile bpf_tail_call and stack_depth

The next set of patches will take advantage of stack_depth tracking,
so make sure that the program that does bpf_tail_call() has
stack depth large enough for the callee.
We could have tracked the stack depth of the prog_array owner program
and only allow insertion of the programs with stack depth less
than the owner, but it will break existing applications.
Some of them have trivial root bpf program that only does
multiple bpf_tail_calls and at init time the prog array is empty.
In the future we may add a flag to do such tracking optionally,
but for now play simple and safe.

Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agobpf: teach verifier to track stack depth
Alexei Starovoitov [Tue, 30 May 2017 20:31:29 +0000 (13:31 -0700)]
bpf: teach verifier to track stack depth

teach verifier to track bpf program stack depth

Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agobpf: split bpf core interpreter
Alexei Starovoitov [Tue, 30 May 2017 20:31:28 +0000 (13:31 -0700)]
bpf: split bpf core interpreter

split __bpf_prog_run() interpreter into stack allocation and execution parts.
The code section shrinks which helps interpreter performance in some cases.
   text    data     bss     dec     hex filename
  26350   10328     624   37302    91b6 kernel/bpf/core.o.before
  25777   10328     624   36729    8f79 kernel/bpf/core.o.after

Very short programs got slower (due to extra function call):
Before:
test_bpf: #89 ALU64_ADD_K: 1 + 2 = 3 jited:0 7 PASS
test_bpf: #90 ALU64_ADD_K: 3 + 0 = 3 jited:0 8 PASS
test_bpf: #91 ALU64_ADD_K: 1 + 2147483646 = 2147483647 jited:0 7 PASS
test_bpf: #92 ALU64_ADD_K: 4294967294 + 2 = 4294967296 jited:0 11 PASS
test_bpf: #93 ALU64_ADD_K: 2147483646 + -2147483647 = -1 jited:0 7 PASS
After:
test_bpf: #89 ALU64_ADD_K: 1 + 2 = 3 jited:0 11 PASS
test_bpf: #90 ALU64_ADD_K: 3 + 0 = 3 jited:0 11 PASS
test_bpf: #91 ALU64_ADD_K: 1 + 2147483646 = 2147483647 jited:0 11 PASS
test_bpf: #92 ALU64_ADD_K: 4294967294 + 2 = 4294967296 jited:0 14 PASS
test_bpf: #93 ALU64_ADD_K: 2147483646 + -2147483647 = -1 jited:0 10 PASS

Longer programs got faster:
Before:
test_bpf: #266 BPF_MAXINSNS: Ctx heavy transformations jited:0 20286 20513 PASS
test_bpf: #267 BPF_MAXINSNS: Call heavy transformations jited:0 31853 31768 PASS
test_bpf: #268 BPF_MAXINSNS: Jump heavy test jited:0 9815 PASS
test_bpf: #269 BPF_MAXINSNS: Very long jump backwards jited:0 6 PASS
test_bpf: #270 BPF_MAXINSNS: Edge hopping nuthouse jited:0 13959 PASS
test_bpf: #271 BPF_MAXINSNS: Jump, gap, jump, ... jited:0 210 PASS
test_bpf: #272 BPF_MAXINSNS: ld_abs+get_processor_id jited:0 21724 PASS
test_bpf: #273 BPF_MAXINSNS: ld_abs+vlan_push/pop jited:0 19118 PASS
After:
test_bpf: #266 BPF_MAXINSNS: Ctx heavy transformations jited:0 19008 18827 PASS
test_bpf: #267 BPF_MAXINSNS: Call heavy transformations jited:0 29238 28450 PASS
test_bpf: #268 BPF_MAXINSNS: Jump heavy test jited:0 9485 PASS
test_bpf: #269 BPF_MAXINSNS: Very long jump backwards jited:0 12 PASS
test_bpf: #270 BPF_MAXINSNS: Edge hopping nuthouse jited:0 13257 PASS
test_bpf: #271 BPF_MAXINSNS: Jump, gap, jump, ... jited:0 213 PASS
test_bpf: #272 BPF_MAXINSNS: ld_abs+get_processor_id jited:0 19389 PASS
test_bpf: #273 BPF_MAXINSNS: ld_abs+vlan_push/pop jited:0 19583 PASS

For real world production programs the difference is noise.

This patch is first step towards reducing interpreter stack consumption.

Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agobpf: free up BPF_JMP | BPF_CALL | BPF_X opcode
Alexei Starovoitov [Tue, 30 May 2017 20:31:27 +0000 (13:31 -0700)]
bpf: free up BPF_JMP | BPF_CALL | BPF_X opcode

free up BPF_JMP | BPF_CALL | BPF_X opcode to be used by actual
indirect call by register and use kernel internal opcode to
mark call instruction into bpf_tail_call() helper.

Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoMerge branch 'nfp-move-BPF-offload-code-into-app'
David S. Miller [Wed, 31 May 2017 21:58:14 +0000 (17:58 -0400)]
Merge branch 'nfp-move-BPF-offload-code-into-app'

Jakub Kicinski says:

====================
nfp: move BPF offload code into app

This series moves the eBPF offload code out of netdev/vNIC handling and
starts building the nfp_app.  Port init is moved into the apps as well
because various apps associate vNICs, representors with ports differently.

First patch adds a helper for updating tc stats which has been waiting
in my tree to be included in any moderately related series.

Next series will bring communicating with FW using control messages,
then representors, BPF maps, tc flower... :)
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonfp: fix memory leak on FW load error
Jakub Kicinski [Wed, 31 May 2017 15:06:51 +0000 (08:06 -0700)]
nfp: fix memory leak on FW load error

Free management FW info when app FW load failed.

Fixes: eefbde7e1002 ("nfp: add hwmon support")
Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonfp: move basic eBPF stats to app-specific code
Jakub Kicinski [Wed, 31 May 2017 15:06:50 +0000 (08:06 -0700)]
nfp: move basic eBPF stats to app-specific code

Allow apps to associate private data with vNICs and move
BPF-specific fields of nfp_net to such structure.

Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonfp: move bpf offload code to the BPF app
Jakub Kicinski [Wed, 31 May 2017 15:06:49 +0000 (08:06 -0700)]
nfp: move bpf offload code to the BPF app

Move bulk of the eBPF offload code out of common vNIC code into
app-specific callbacks.

Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonfp: move eBPF offload files to BPF app directory
Jakub Kicinski [Wed, 31 May 2017 15:06:48 +0000 (08:06 -0700)]
nfp: move eBPF offload files to BPF app directory

Pure move of eBPF offload files to BPF app directory,
only change the names and relative header location.
nfp_asm.h stays in the main dir and it doesn't really
have to include nfp_bpf.h.

Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonfp: report app name in ethtool -i
Jakub Kicinski [Wed, 31 May 2017 15:06:47 +0000 (08:06 -0700)]
nfp: report app name in ethtool -i

Let the app print its name in ethtool -i output.

Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonfp: move port init to apps
Jakub Kicinski [Wed, 31 May 2017 15:06:46 +0000 (08:06 -0700)]
nfp: move port init to apps

Start fleshing out the apps by turning the vNIC init code to
a per-app callback.  The two initial apps we have are NIC and
eBPF.

Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonfp: turn reading PCIe RTsym parameters into a helper
Jakub Kicinski [Wed, 31 May 2017 15:06:45 +0000 (08:06 -0700)]
nfp: turn reading PCIe RTsym parameters into a helper

Turn the function to read number of ports into a generic helper.
While at it make sure we propagate all errors other than -ENOENT.

Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonfp: add missing fall through statements
Jakub Kicinski [Wed, 31 May 2017 15:06:44 +0000 (08:06 -0700)]
nfp: add missing fall through statements

GCC 7 checks for fall through comments, add the two missing ones.

Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agosched: add helper for updating statistics on all actions
Jakub Kicinski [Wed, 31 May 2017 15:06:43 +0000 (08:06 -0700)]
sched: add helper for updating statistics on all actions

Forgetting to disable preemption around tcf_action_stats_update()
seems to be a common mistake.  Add a helper function for updating
stats on all actions of a filter.

Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoMerge branch '10GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/next...
David S. Miller [Wed, 31 May 2017 21:55:10 +0000 (17:55 -0400)]
Merge branch '10GbE' of git://git./linux/kernel/git/jkirsher/next-queue

Jeff Kirsher says:

====================
10GbE Intel Wired LAN Driver Updates 2017-05-31

This series contains updates to ixgbe and ixgbevf only.

Scott enables support for TSO & GSO for MPLS encapsulated packets for both
ixgbe and ixgbevf.

Liwei Song fixes an issue where seqcount/seqlock in ixgbe_get_stats64()
are not initialized in time, so move the initialization into probe routine
after the transmit and receive rings are initialized.

Paul cleans up led_[on|off] for X550EM_X, since the firmware configures
the PHY & MAC and we have no PHY access so LED on/off is not supported
with this device.

Emil provides several fixes, starting with enabling RSS on VF to VF
traffic on the same PF.  Fixed PHY identification, where the previous
method was unreliable, so use a different register to ensure proper
identification.  Cleaned up the logic which could cause us to
skip the link configuration, this skipping over the link configuration
was leaving SFP+ PHY's in an unstable state, so always call
setup_mac_link().  Added RS1 (rate select 1) support for ixgbe.  Lastly,
fixed incorrect logic in the setting up of SFP+ link speed.

Mark fixes the thermal sensor event logic, where it was being executed
when there really was no thermal event.  So simplify the logic to only
execute when there is a thermal event.

Tony adds additional error checks and reporting when setting a VF MAC
address to ensure that the MAC filter was successfully added.  Also
fixed possible truncation warnings, as well as implicit fallthrough
warnings.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoMerge branch '40GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/next...
David S. Miller [Wed, 31 May 2017 21:49:37 +0000 (17:49 -0400)]
Merge branch '40GbE' of git://git./linux/kernel/git/jkirsher/next-queue

Jeff Kirsher says:

====================
40GbE Intel Wired LAN Driver Updates 2017-05-31

This series contains updates to i40e and i40evf only.

Jesse provides a couple of fixes, starting with cleaning up duplicate
lines of code.  Fixed a missing line which enables RSS as a negotiated
feature.  Since the VF does not have any way of reporting FCoE enabled,
so just force the code to always report FCoE as disabled.

Jake provides several fixes and changes, starting with fixing a race
condition in i40e.  The hardware has a limitation on transmit PTP packets,
which requires us to limit the driver to timestamping a single packet at
once.  This is done using a state bitlock which enforces that only one
timestamp request is honored at a time, unfortunately this suffers from
a race condition.  Fixed a corner case where we failed to cleanup the
bit lock after a failed transmit, and resulted in a state bit being
locked forever.  Added a new statistic which tracks when a transmit
timestamp request is skipped/ignored, since the driver can only handle
one transmit timestamp request at a time.

Christophe Jaillet fixes a NULL pointer dereference if kzalloc fails.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonetvsc: Add #include's for csum_* function declarations
Michael Kelley [Tue, 30 May 2017 18:36:56 +0000 (11:36 -0700)]
netvsc: Add #include's for csum_* function declarations

Add direct #include statements for declarations of csum_tcpudp_magic()
and csum_ipv6_magic().  While the needed #include's are picked up
indirectly for the x86 architecture, they aren't on other
architectures, resulting in compile errors.

Signed-off-by: Michael Kelley <mikelley@microsoft.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agortnetlink: use the new rtnl_get_event() interface
Vlad Yasevich [Tue, 30 May 2017 14:07:02 +0000 (10:07 -0400)]
rtnetlink: use the new rtnl_get_event() interface

Small clean-up to rtmsg_ifinfo() to use the rtnl_get_event()
interface instead of using 'internal' values directly.

Signed-off-by: Vladislav Yasevich <vyasevic@redhat.com>
Acked-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agosamples/bpf: bpf_load.c order of prog_fd[] should correspond with ELF order
Jesper Dangaard Brouer [Tue, 30 May 2017 12:37:51 +0000 (14:37 +0200)]
samples/bpf: bpf_load.c order of prog_fd[] should correspond with ELF order

An eBPF ELF file generated with LLVM can contain several program
section, which can be used for bpf tail calls.  The bpf prog file
descriptors are accessible via array prog_fd[].

At-least XDP samples assume ordering, and uses prog_fd[0] is the main
XDP program to attach.  The actual order of array prog_fd[] depend on
whether or not a bpf program section is referencing any maps or not.
Not using a map result in being loaded/processed after all other
prog section.  Thus, this can lead to some very strange and hard to
debug situation, as the user can only see a FD and cannot correlated
that with the ELF section name.

The fix is rather simple, and even removes duplicate memcmp code.
Simply load program sections as the last step, instead of
load_and_attach while processing the relocation section.

When working with tail calls, it become even more essential that the
order of prog_fd[] is consistant, like the current dependency of the
map_fd[] order.

Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet/mlxfw: select CONFIG_XZ_DEC
Arnd Bergmann [Tue, 30 May 2017 09:26:14 +0000 (11:26 +0200)]
net/mlxfw: select CONFIG_XZ_DEC

The new mlxfw code fails to build without the xz library:

drivers/net/ethernet/mellanox/mlxfw/mlxfw_mfa2.o: In function `mlxfw_mfa2_xz_dec_run':
:(.text.mlxfw_mfa2_xz_dec_run+0x8): undefined reference to `xz_dec_run'
drivers/net/ethernet/mellanox/mlxfw/mlxfw_mfa2.o: In function `mlxfw_mfa2_file_component_get':
:(.text.mlxfw_mfa2_file_component_get+0x218): undefined reference to `xz_dec_init'
:(.text.mlxfw_mfa2_file_component_get+0x2c0): undefined reference to `xz_dec_end'

This adds a Kconfig 'select' statement for it, which is also what
the other user of that library has.

Fixes: 410ed13cae39 ("Add the mlxfw module for Mellanox firmware flash process")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Acked-by: Yotam Gigi <yotamg@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: dsa: remove dev arg of dsa_register_switch
Vivien Didelot [Fri, 26 May 2017 22:12:51 +0000 (18:12 -0400)]
net: dsa: remove dev arg of dsa_register_switch

The current dsa_register_switch function takes a useless struct device
pointer argument, which always equals ds->dev.

Drivers either call it with ds->dev, or with the same device pointer
passed to dsa_switch_alloc, which ends up being assigned to ds->dev.

This patch removes the second argument of the dsa_register_switch and
_dsa_register_switch functions.

Signed-off-by: Vivien Didelot <vivien.didelot@savoirfairelinux.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoMerge branch 'dsa-mv88e6xxx-PHY-nitpicks'
David S. Miller [Wed, 31 May 2017 16:35:01 +0000 (12:35 -0400)]
Merge branch 'dsa-mv88e6xxx-PHY-nitpicks'

Vivien Didelot says:

====================
net: dsa: mv88e6xxx: PHY nitpicks

This patchset isolates more PPU code into phy.c and makes distinction
between PHY Registers read and write implementations vs. generic PHY
routines.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: dsa: mv88e6xxx: rename PHY PPU functions
Vivien Didelot [Fri, 26 May 2017 22:03:07 +0000 (18:03 -0400)]
net: dsa: mv88e6xxx: rename PHY PPU functions

Respect the implicit naming convention used in all register sets
specific files, by renaming the mv88e6xxx_ppu_* functions with the
mv88e6xxx_phy_* prefix.

This is simply a s/xxx_ppu/xxx_phy_ppu/ substitution.

Signed-off-by: Vivien Didelot <vivien.didelot@savoirfairelinux.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: dsa: mv88e6xxx: rename PHY PPU accessors
Vivien Didelot [Fri, 26 May 2017 22:03:06 +0000 (18:03 -0400)]
net: dsa: mv88e6xxx: rename PHY PPU accessors

Make it clear that mv88e6xxx_phy_ppu_{read,write} are an implementation
of the .phy_{read,write} operations, by renaming them with the mv88e6185
prefix, since 88E6185 it is the reference switch model supported in an
upstream board (ZII Dev Rev B), which makes use of them.

Distinguish the signatures of implementation specific and generic PHY
functions in the phy.h header.

Signed-off-by: Vivien Didelot <vivien.didelot@savoirfairelinux.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: dsa: mv88e6xxx: provide a PHY setup helper
Vivien Didelot [Fri, 26 May 2017 22:03:05 +0000 (18:03 -0400)]
net: dsa: mv88e6xxx: provide a PHY setup helper

Similarly to the VTU, PVT and ATU setup, provide a mv88e6xxx_phy_setup
helper which wraps mv88e6xxx_ppu_enable, so that no more PPU-related
functions are exposed outside of phy.c.

Thus make mv88e6xxx_ppu_enable static.

Signed-off-by: Vivien Didelot <vivien.didelot@savoirfairelinux.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoixgbe: fix incorrect status check
Emil Tantilov [Thu, 18 May 2017 23:45:06 +0000 (16:45 -0700)]
ixgbe: fix incorrect status check

Check for ret_val instead of !ret_val to allow the rest of
the code to execute and configure the speed properly.

Signed-off-by: Emil Tantilov <emil.s.tantilov@intel.com>
Tested-by: Krishneil Singh <krishneil.k.singh@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
7 years agoixgbe: add missing configuration for rate select 1
Emil Tantilov [Wed, 17 May 2017 22:17:56 +0000 (15:17 -0700)]
ixgbe: add missing configuration for rate select 1

Add RS1 configuration to ixgbe_set_soft_rate_select_speed()

Signed-off-by: Emil Tantilov <emil.s.tantilov@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
7 years agoixgbe: always call setup_mac_link for multispeed fiber
Emil Tantilov [Wed, 17 May 2017 22:17:51 +0000 (15:17 -0700)]
ixgbe: always call setup_mac_link for multispeed fiber

Remove the logic which would previously skip the link configuration
in the case where we are already at the requested speed in
ixgbe_setup_mac_link_multispeed_fiber().

By exiting early we are skipping the link configuration and as such
the driver may not always configure the PHY correctly for SFP+.

Signed-off-by: Emil Tantilov <emil.s.tantilov@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
7 years agoixgbe: add write flush when configuring CS4223/7
Emil Tantilov [Wed, 17 May 2017 22:17:46 +0000 (15:17 -0700)]
ixgbe: add write flush when configuring CS4223/7

Make sure the writes are processed immediately. Without the flush it
is possible for operations on one port to spill over the other as the
resource is shared.

Signed-off-by: Emil Tantilov <emil.s.tantilov@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
7 years agoixgbe: correct CS4223/7 PHY identification
Emil Tantilov [Wed, 17 May 2017 22:17:41 +0000 (15:17 -0700)]
ixgbe: correct CS4223/7 PHY identification

Previous method was unreliable. Use a different register to
differentiate between the SKUs.

Signed-off-by: Emil Tantilov <emil.s.tantilov@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
7 years agoixgbevf: Resolve warnings for -Wimplicit-fallthrough
Tony Nguyen [Fri, 12 May 2017 18:38:10 +0000 (11:38 -0700)]
ixgbevf: Resolve warnings for -Wimplicit-fallthrough

Additions to gcc 7 now warn whenever a switch statement falls through
implicitly.  This patch adds explicit fall through comments to address the
following warnings:

drivers/net/ethernet/intel/ixgbevf/vf.c: In function ‘ixgbevf_get_reta_locked’:
drivers/net/ethernet/intel/ixgbevf/vf.c:336:6: warning: this statement may fall through [-Wimplicit-fallthrough=]
   if (hw->mac.type < ixgbe_mac_X550_vf)
      ^
drivers/net/ethernet/intel/ixgbevf/vf.c:338:2: note: here
  default:
  ^~~~~~~
drivers/net/ethernet/intel/ixgbevf/vf.c: In function ‘ixgbevf_get_rss_key_locked’:
drivers/net/ethernet/intel/ixgbevf/vf.c:402:6: warning: this statement may fall through [-Wimplicit-fallthrough=]
   if (hw->mac.type < ixgbe_mac_X550_vf)
      ^
drivers/net/ethernet/intel/ixgbevf/vf.c:404:2: note: here
  default:
  ^~~~~~~

Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
7 years agoixgbevf: Resolve truncation warning for q_vector->name
Tony Nguyen [Fri, 12 May 2017 18:38:09 +0000 (11:38 -0700)]
ixgbevf: Resolve truncation warning for q_vector->name

The following warning is now shown as a result of new checks added for
gcc 7:

drivers/net/ethernet/intel/ixgbevf/ixgbevf_main.c: In function ‘ixgbevf_open’:
drivers/net/ethernet/intel/ixgbevf/ixgbevf_main.c:1363:13: warning: ‘%d’ directive output may be truncated writing between 1 and 10 bytes into a region of size between 3 and 18 [-Wformat-truncation=]
      "%s-%s-%d", netdev->name, "TxRx", ri++);
             ^~
drivers/net/ethernet/intel/ixgbevf/ixgbevf_main.c:1363:6: note: directive argument in the range [0, 2147483647]
      "%s-%s-%d", netdev->name, "TxRx", ri++);
      ^~~~~~~~~~
drivers/net/ethernet/intel/ixgbevf/ixgbevf_main.c:1362:4: note: ‘snprintf’ output between 8 and 32 bytes into a destination of size 24
    snprintf(q_vector->name, sizeof(q_vector->name) - 1,
    ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
      "%s-%s-%d", netdev->name, "TxRx", ri++);
      ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Resolve this warning by making a couple of changes.
 - Don't reserve space for the null terminator.  Since snprintf adds the
   null terminator automatically, there is no need for us to reserve a byte
   for it.

 - Change a couple variables that can never be negative from int to
   unsigned int.

While we're making changes to the format string, move the constant strings
into the format string instead of providing them as specifiers.

Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
7 years agoixgbe: Resolve warnings for -Wimplicit-fallthrough
Tony Nguyen [Wed, 31 May 2017 11:43:47 +0000 (04:43 -0700)]
ixgbe: Resolve warnings for -Wimplicit-fallthrough

This patch adds/changes fall through comments to address new warnings
produced by gcc 7.

Fixed formatting on a couple of comments in the function.

Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
7 years agoixgbe: Resolve truncation warning for q_vector->name
Tony Nguyen [Fri, 12 May 2017 18:38:07 +0000 (11:38 -0700)]
ixgbe: Resolve truncation warning for q_vector->name

The following warning is now shown as a result of new checks added for
gcc 7:

drivers/net/ethernet/intel/ixgbe/ixgbe_main.c: In function ‘ixgbe_open’:
drivers/net/ethernet/intel/ixgbe/ixgbe_main.c:3118:13: warning: ‘%d’ directive output may be truncated writing between 1 and 10 bytes into a region of size between 3 and 18 [-Wformat-truncation=]
      "%s-%s-%d", netdev->name, "TxRx", ri++);
             ^~
drivers/net/ethernet/intel/ixgbe/ixgbe_main.c:3118:6: note: directive argument in the range [0, 2147483647]
      "%s-%s-%d", netdev->name, "TxRx", ri++);
      ^~~~~~~~~~
drivers/net/ethernet/intel/ixgbe/ixgbe_main.c:3117:4: note: ‘snprintf’ output between 8 and 32 bytes into a destination of size 24
    snprintf(q_vector->name, sizeof(q_vector->name) - 1,
    ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
      "%s-%s-%d", netdev->name, "TxRx", ri++);
      ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Resolve this warning by making a couple of changes.
 - Don't reserve space for the null terminator.  Since snprintf adds the
   null terminator automatically, there is no need for us to reserve a byte
   for it.

 - Change a couple variables that can never be negative from int to
   unsigned int.

While we're making changes to the format string, move the constant strings
into the format string instead of providing them as specifiers.

Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
7 years agoixgbe: Add error checking to setting VF MAC
Tony Nguyen [Fri, 28 Apr 2017 19:42:03 +0000 (12:42 -0700)]
ixgbe: Add error checking to setting VF MAC

Currently, when setting a VF MAC address there are no error checks to
ensure that the MAC filter was successfully added.  This patch adds
additional error checks, reporting, and propagation of errors.  It also
will not set the MAC address unless adding the MAC filter was successful.

With these changes, setting the mac address to zeros can no longer call
ixgbe_set_vf_mac() as adding a zero MAC address filter is not valid.
Instead directly delete the filter and, if successful, clear the MAC
address.

Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>