platform/kernel/linux-rpi.git
4 years agonet: phy: dp83867: Add speed optimization feature
Dan Murphy [Tue, 18 Feb 2020 14:11:30 +0000 (08:11 -0600)]
net: phy: dp83867: Add speed optimization feature

Set the speed optimization bit on the DP83867 PHY.
This feature can also be strapped on the 64 pin PHY devices
but the 48 pin devices do not have the strap pin available to enable
this feature in the hardware.  PHY team suggests to have this bit set.

With this bit set the PHY will auto negotiate and report the link
parameters in the PHYSTS register.  This register provides a single
location within the register set for quick access to commonly accessed
information.

In this case when auto negotiation is on the PHY core reads the bits
that have been configured or if auto negotiation is off the PHY core
reads the BMCR register and sets the phydev parameters accordingly.

This Giga bit PHY can throttle the speed to 100Mbps or 10Mbps to accomodate a
4-wire cable.  If this should occur the PHYSTS register contains the
current negotiated speed and duplex mode.

In overriding the genphy_read_status the dp83867_read_status will do a
genphy_read_status to setup the LP and pause bits.  And then the PHYSTS
register is read and the phydev speed and duplex mode settings are
updated.

Signed-off-by: Dan Murphy <dmurphy@ti.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoenetc: remove "depends on (ARCH_LAYERSCAPE || COMPILE_TEST)"
Vladimir Oltean [Fri, 21 Feb 2020 14:46:24 +0000 (16:46 +0200)]
enetc: remove "depends on (ARCH_LAYERSCAPE || COMPILE_TEST)"

ARCH_LAYERSCAPE isn't needed for this driver, it builds and
sends/receives traffic without this config option just fine.

Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Acked-by: Claudiu Manoil <claudiu.manoil@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agotc-testing: updated tdc tests for basic filter with u16 extended match rules
Roman Mashak [Fri, 21 Feb 2020 14:38:57 +0000 (09:38 -0500)]
tc-testing: updated tdc tests for basic filter with u16 extended match rules

Signed-off-by: Roman Mashak <mrv@mojatatu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agonet: page_pool: Add documentation on page_pool API
Ilias Apalodimas [Fri, 21 Feb 2020 09:15:19 +0000 (11:15 +0200)]
net: page_pool: Add documentation on page_pool API

Add documentation explaining the basic functionality and design
principles of the API

Signed-off-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>
Acked-by: Randy Dunlap <rdunlap@infradead.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoMerge branch '1GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/next...
David S. Miller [Fri, 21 Feb 2020 00:00:14 +0000 (16:00 -0800)]
Merge branch '1GbE' of git://git./linux/kernel/git/jkirsher/next-queue

Jeff Kirsher says:

====================
1GbE Intel Wired LAN Driver Updates 2020-02-19

This series contains updates to e1000e and igc drivers.

Ben Dooks adds a missing cpu_to_le64() in the e1000e transmit ring flush
function.

Jia-Ju Bai replaces a couple of udelay() with usleep_range() where we
could sleep while holding a spinlock in e1000e.

Chen Zhou make 2 functions static in igc,

Sasha finishes the legacy power management support in igc by adding
resume and schedule suspend requests.  Also added register dump
functionality in the igc driver.  Added device id support for the next
generation of i219 devices in e1000e.  Fixed a typo in the igc driver
that referenced a device that is not support in the driver.  Added the
missing PTP support when suspending now that igc has legacy power
management support.  Added PCIe error detection, slot reset and resume
capability in igc.  Added WoL support for igc as well.  Lastly, added a
code comment to distinguish between interrupt and flag definitions.

Vitaly adds device id support for Tiger Lake platforms, which has
another next generation of i219 device in e1000e.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoMerge branch '100GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/next...
David S. Miller [Thu, 20 Feb 2020 23:04:49 +0000 (15:04 -0800)]
Merge branch '100GbE' of git://git./linux/kernel/git/jkirsher/next-queue

Jeff Kirsher says:

====================
100GbE Intel Wired LAN Driver Updates 2020-02-19

This series contains updates to the ice driver only.

Avinash adds input validation for software DCB configurations received
via lldptool or pcap to ensure bad bandwidth inputs are not inputted
which could cause the loss of link.

Paul update the malicious driver detection event messages to rate limit
once per second and to include the total number of receive|transmit MDD
event count.

Dan updates how TCAM entries are managed to ensure when overriding
pre-existing TCAM entries, properly delete the existing entry and remove
it from the change/update list.

Brett ensures we clear the relevant values in the QRXFLXP_CNTXT register
for VF queues to ensure the receive queue data is not stale.

Avinash adds required DCBNL operations for configuring ETS in software
DCB CEE mode.  Also added code to detect if DCB is in IEEE or CEE mode
to properly report what mode we are in.

Dave fixes the driver to properly report the current maximum TC, not the
maximum allowed number of TCs.

Krzysztof adds support for AF_XDP feature in the ice driver.

Jake increases the maximum time that the driver will wait for a PR reset
to account for possibility of a slightly longer than expected PD reset.

Jesse fixes a number of strings which did not have line feeds, so add
line feeds so that messages do not rum together, creating a jumbled
mess.

Bruce adds support for additional E810 and E823 device ids.  Also
updated the product name change for E822 devices.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agonet: use netif_is_bridge_port() to check for IFF_BRIDGE_PORT
Julian Wiedmann [Thu, 20 Feb 2020 08:00:07 +0000 (09:00 +0100)]
net: use netif_is_bridge_port() to check for IFF_BRIDGE_PORT

Trivial cleanup, so that all bridge port-specific code can be found in
one go.

CC: Johannes Berg <johannes@sipsolutions.net>
CC: Roopa Prabhu <roopa@cumulusnetworks.com>
CC: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: Julian Wiedmann <jwi@linux.ibm.com>
Reviewed-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agonet: page_pool: API cleanup and comments
Ilias Apalodimas [Thu, 20 Feb 2020 07:41:55 +0000 (09:41 +0200)]
net: page_pool: API cleanup and comments

Functions starting with __ usually indicate those which are exported,
but should not be called directly. Update some of those declared in the
API and make it more readable.

page_pool_unmap_page() and page_pool_release_page() were doing
exactly the same thing calling __page_pool_clean_page().  Let's
rename __page_pool_clean_page() to page_pool_release_page() and
export it in order to show up on perf logs and get rid of
page_pool_unmap_page().

Finally rename __page_pool_put_page() to page_pool_put_page() since we
can now directly call it from drivers and rename the existing
page_pool_put_page() to page_pool_put_full_page() since they do the same
thing but the latter is trying to sync the full DMA area.

This patch also updates netsec, mvneta and stmmac drivers which use
those functions.

Suggested-by: Jonathan Lemon <jonathan.lemon@gmail.com>
Acked-by: Toke Høiland-Jørgensen <toke@redhat.com>
Acked-by: Jesper Dangaard Brouer <brouer@redhat.com>
Signed-off-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoMerge branch 'mlxsw-Preparation-for-RTNL-removal'
David S. Miller [Thu, 20 Feb 2020 18:04:34 +0000 (10:04 -0800)]
Merge branch 'mlxsw-Preparation-for-RTNL-removal'

Ido Schimmel says:

====================
mlxsw: Preparation for RTNL removal

The driver currently acquires RTNL in its route insertion path, which
contributes to very large control plane latencies. This patch set
prepares mlxsw for RTNL removal from its route insertion path in a
follow-up patch set.

Patches #1-#2 protect shared resources - KVDL and counter pool - with
their own locks. All allocations of these resources are currently
performed under RTNL, so no locks were required.

Patches #3-#7 ensure that updates to mirroring sessions only take place
in case there are active mirroring sessions. This allows us to avoid
taking RTNL when it is unnecessary, as updating of the mirroring
sessions must be performed under RTNL for the time being.

Patches #8-#10 replace the use of APIs that assume that RTNL is taken
with their RCU counterparts. Specifically, patches #8 and #9 replace
__in_dev_get_rtnl() with __in_dev_get_rcu() under RCU read-side critical
section. Patch #10 replaces __dev_get_by_index() with
dev_get_by_index_rcu().

Patches #11-#15 perform small adjustments in the code to make it easier
to later introduce a router lock instead of relying on RTNL.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agomlxsw: spectrum_nve: Make tunnel initialization symmetric
Ido Schimmel [Thu, 20 Feb 2020 07:08:00 +0000 (09:08 +0200)]
mlxsw: spectrum_nve: Make tunnel initialization symmetric

The device supports a single VTEP whose configuration is shared between
all VXLAN tunnels.

While the shared configuration is cleared upon the destruction of the
last tunnel - in mlxsw_sp_nve_tunnel_fini() - it is set in
mlxsw_sp_nve_fid_enable(), after calling mlxsw_sp_nve_tunnel_init().

Make tunnel initialization and destruction symmetric and set the
configuration in mlxsw_sp_nve_tunnel_init().

This will later allow us to protect the shared configuration with a
lock.

Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agomlxsw: spectrum: Export function to check if RIF exists
Ido Schimmel [Thu, 20 Feb 2020 07:07:59 +0000 (09:07 +0200)]
mlxsw: spectrum: Export function to check if RIF exists

After the previous patch, all the callers of mlxsw_sp_rif_find_by_dev()
outside of the routing code use it to understand if a RIF exists for the
passed netdev.

Therefore, export a function to check if a RIF exists and make
mlxsw_sp_rif_find_by_dev() internal to the routing code.

This will later allow us to more easily introduce the router lock which
will also protect the RIFs.

Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agomlxsw: spectrum: Prevent RIF access outside of routing code
Ido Schimmel [Thu, 20 Feb 2020 07:07:58 +0000 (09:07 +0200)]
mlxsw: spectrum: Prevent RIF access outside of routing code

There are currently 5 users of mlxsw_sp_rif_find_by_dev() outside of the
routing code. Only one call site actually needs to dereference the
router interface (RIF). The rest merely need to know if a RIF exists for
the provided netdev.

Convert this call site to query the needed information directly from the
routing code instead of dereferencing the RIF.

This will later allow us to replace mlxsw_sp_rif_find_by_dev() with a
function that checks if a RIF exist.

Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agomlxsw: spectrum_router: Prepare function for router lock introduction
Ido Schimmel [Thu, 20 Feb 2020 07:07:57 +0000 (09:07 +0200)]
mlxsw: spectrum_router: Prepare function for router lock introduction

The function de-associates the port-vlan from its router interface
(RIF). It is called both from the netdev notifier block and the inetaddr
notifier block that will soon hold the router lock.

Make sure that router code calls the internal version, as it will
already have the router lock held when the function is called.

Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agomlxsw: spectrum_router: Prepare function for router lock introduction
Ido Schimmel [Thu, 20 Feb 2020 07:07:56 +0000 (09:07 +0200)]
mlxsw: spectrum_router: Prepare function for router lock introduction

The function removes the FDB entry that directs the macvlan's MAC to the
router port. It is called from both the netdev notifier block and the
inetaddr notifier block that will soon hold the router lock.

Make sure that only the netdev notifier calls the exported version, so
that is will take the router lock, which will already be held by the
inetaddr notifier.

Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agomlxsw: spectrum_router: Do not assume RTNL is taken when resolving underlay device
Ido Schimmel [Thu, 20 Feb 2020 07:07:55 +0000 (09:07 +0200)]
mlxsw: spectrum_router: Do not assume RTNL is taken when resolving underlay device

The function that resolves the underlay device of the IPIP tunnel
assumes that RTNL is taken, but this will not be correct when RTNL is
removed from the route insertion path.

Convert the function to use dev_get_by_index_rcu() instead of
__dev_get_by_index() and make sure it is always called from an RCU
read-side critical section.

Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agomlxsw: spectrum_router: Do not assume RTNL is taken during RIF teardown
Ido Schimmel [Thu, 20 Feb 2020 07:07:54 +0000 (09:07 +0200)]
mlxsw: spectrum_router: Do not assume RTNL is taken during RIF teardown

IPv6 addresses are deleted in an atomic context, so the driver defers
the potential teardown of the associated router interface (RIF) to a
work item that takes RTNL.

The RIF is only destroyed if the associated netdev does not have any IP
addresses (both IPv4 and IPv6). The IPv4 device ('struct in_device') is
currently fetched via __in_dev_get_rtnl() which assumes RTNL is taken.

Since RTNL is going to be removed, convert it to use __in_dev_get_rcu()
from an RCU read-side critical section.

Note that the IPv6 device ('struct inet6_dev') is fetched via
__in6_dev_get(), which does not require RTNL.

Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agomlxsw: spectrum_router: Do not assume RTNL is taken during nexthop init
Ido Schimmel [Thu, 20 Feb 2020 07:07:53 +0000 (09:07 +0200)]
mlxsw: spectrum_router: Do not assume RTNL is taken during nexthop init

RTNL is going to be removed from route insertion path, so use
__in_dev_get_rcu() from an RCU read-side critical section instead of
__in_dev_get_rtnl() which assumes RTNL is taken.

Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agomlxsw: spectrum_span: Only update mirroring agents if present
Ido Schimmel [Thu, 20 Feb 2020 07:07:52 +0000 (09:07 +0200)]
mlxsw: spectrum_span: Only update mirroring agents if present

In order not to needlessly schedule the work item that updates the
mirroring agents, only schedule it if there are any mirroring agents
present.

This is done by adding an atomic counter that counts the active
mirroring agents.

It is incremented / decremented whenever a mirroring agent is created /
destroyed. It is read before scheduling the work item and in the
devlink-resource occupancy callback.

Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agomlxsw: spectrum: Convert callers to use new mirroring API
Ido Schimmel [Thu, 20 Feb 2020 07:07:51 +0000 (09:07 +0200)]
mlxsw: spectrum: Convert callers to use new mirroring API

Previous patch added a work item in the mirroring code that will take
care of updating the active mirroring agents in response to different
events.

Change the mirroring agents update function - mlxsw_sp_span_respin() -
to invoke this work item when called.

Therefore there is no need for callers to schedule a work item
themselves.

Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agomlxsw: spectrum_span: Prepare work item to update mirroring agents
Ido Schimmel [Thu, 20 Feb 2020 07:07:50 +0000 (09:07 +0200)]
mlxsw: spectrum_span: Prepare work item to update mirroring agents

The driver updates its mirroring agents whenever it receives a
notification about an event that can affect these. For example, the
addition of a route might require the driver to change the egress port
of an ERSPAN session.

Currently, RTNL needs to be held when these agents are updates, so the
driver either:

1. Calls directly into the mirroring code, in case RTNL is held

2. Schedules a work item that will take RTNL and call into the mirroring
code

Simplify this by having the mirroring code schedule the work item for
the update instead of requiring callers to schedule a work item
themselves.

The conversion of the callers will be done in the next patch to make
review easier.

This will later allow us to remove RTNL from different parts of the
driver. It will also allow us to only schedule the work item in case
there are active mirroring agents, which is information private to the
mirroring code.

Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agomlxsw: spectrum_span: Use struct_size() to simplify allocation
Ido Schimmel [Thu, 20 Feb 2020 07:07:49 +0000 (09:07 +0200)]
mlxsw: spectrum_span: Use struct_size() to simplify allocation

Allocate the main mirroring struct and the individual structs for the
different mirroring agents in a single allocation.

Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agomlxsw: spectrum_span: Do no expose mirroring agents to entire driver
Ido Schimmel [Thu, 20 Feb 2020 07:07:48 +0000 (09:07 +0200)]
mlxsw: spectrum_span: Do no expose mirroring agents to entire driver

The struct holding the different mirroring agents is currently allocated
as part of the main driver struct. This is unlike other driver modules.

Allocate the memory required to store the different mirroring agents as
part of the initialization of the mirroring module.

Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agomlxsw: spectrum: Protect counter pool with a lock
Ido Schimmel [Thu, 20 Feb 2020 07:07:47 +0000 (09:07 +0200)]
mlxsw: spectrum: Protect counter pool with a lock

The counter pool is a shared resource. It is used by both the ACL code
to allocate counters for actions and by the routing code to allocate
counters for adjacency entries (for example).

Currently, all allocations are protected by RTNL, but this is going to
change with the removal of RTNL from the routing code.

Therefore, protect counter allocations with a spin lock.

Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agomlxsw: spectrum_kvdl: Protect allocations with a lock
Ido Schimmel [Thu, 20 Feb 2020 07:07:46 +0000 (09:07 +0200)]
mlxsw: spectrum_kvdl: Protect allocations with a lock

The KVDL is used to store objects allocated throughout various places
in the driver. For example, both nexthops (adjacency entries) and ACL
actions are stored in the KVDL.

Currently, all allocations are protected by RTNL, but this is going to
change with the removal of RTNL from the routing code.

Therefore, protect KVDL allocations with a lock. A mutex is used since
the free operation can block in Spectrum-2.

Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agonet: remove unused macro from fib_trie.c
Li RongQing [Thu, 20 Feb 2020 06:50:19 +0000 (14:50 +0800)]
net: remove unused macro from fib_trie.c

TNODE_KMALLOC_MAX and VERSION are not used, so remove them

Signed-off-by: Li RongQing <lirongqing@baidu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agonet: neigh: remove unused NEIGH_SYSCTL_MS_JIFFIES_ENTRY
Li RongQing [Thu, 20 Feb 2020 06:49:02 +0000 (14:49 +0800)]
net: neigh: remove unused NEIGH_SYSCTL_MS_JIFFIES_ENTRY

this macro is never used, so remove it

Signed-off-by: Li RongQing <lirongqing@baidu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoigc: Add comment
Sasha Neftin [Mon, 3 Feb 2020 08:11:50 +0000 (10:11 +0200)]
igc: Add comment

Separate interrupt and flag definitions.
Made the code clear.

Signed-off-by: Sasha Neftin <sasha.neftin@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
4 years agoigc: Add WOL support
Sasha Neftin [Mon, 3 Feb 2020 07:55:20 +0000 (09:55 +0200)]
igc: Add WOL support

This patch adds a define and WOL support for an i225 parts.

Signed-off-by: Sasha Neftin <sasha.neftin@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
4 years agoigc: Add pcie error handler support
Sasha Neftin [Wed, 29 Jan 2020 14:30:07 +0000 (16:30 +0200)]
igc: Add pcie error handler support

Add pcie error detection, slot reset and resume capability

Signed-off-by: Sasha Neftin <sasha.neftin@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
4 years agoigc: Complete to commit Add basic skeleton for PTP
Sasha Neftin [Wed, 22 Jan 2020 09:21:13 +0000 (11:21 +0200)]
igc: Complete to commit Add basic skeleton for PTP

commit 5f2958052c58 ("igc: Add basic skeleton for PTP") added basic
support for PTP, what's missing is support for suspending.
Legacy power management has been added. Now we can add
the suspend method to the igc_shutdown.
By cleaning the runtime storage for timestamp this avoids a possible
invalid memory access when the system comes back from suspend state.

Signed-off-by: Sasha Neftin <sasha.neftin@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
4 years agoe1000e: Add support for Tiger Lake device
Vitaly Lifshits [Tue, 21 Jan 2020 23:46:28 +0000 (15:46 -0800)]
e1000e: Add support for Tiger Lake device

Added support for a device id that is a part of the Intel Tiger Lake
platform.

Signed-off-by: Vitaly Lifshits <vitaly.lifshits@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
4 years agoigc: Fix the typo in comment
Sasha Neftin [Mon, 20 Jan 2020 06:17:58 +0000 (08:17 +0200)]
igc: Fix the typo in comment

Fix the typo and comment to correspond to the i225 device

Signed-off-by: Sasha Neftin <sasha.neftin@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
4 years agoe1000e: Add support for Alder Lake
Sasha Neftin [Sun, 19 Jan 2020 11:57:13 +0000 (13:57 +0200)]
e1000e: Add support for Alder Lake

Add devices ID's for the next LOM generations that will be
available on the next Intel Client platform (Alder Lake)
This patch provides the initial support for these devices

Signed-off-by: Sasha Neftin <sasha.neftin@intel.com>
Reviewed-by: Paul Menzel <pmenzel@molgen.mpg.de>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
4 years agotc-testing: updated tdc tests for basic filter
Roman Mashak [Wed, 19 Feb 2020 21:37:56 +0000 (16:37 -0500)]
tc-testing: updated tdc tests for basic filter

Added tests for 'u32' extended match rules for u8 alignment.

Signed-off-by: Roman Mashak <mrv@mojatatu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoMerge branch 'net-phy-Better-support-for-BCM54810'
David S. Miller [Thu, 20 Feb 2020 00:36:48 +0000 (16:36 -0800)]
Merge branch 'net-phy-Better-support-for-BCM54810'

Florian Fainelli says:

====================
net: phy: Better support for BCM54810

This patch series updates the broadcom PHY driver to better support the
BCM54810 and allow it to make use of the exiting
bcm54xx_adjust_rxrefclk() as well as fix suspend/resume for it.

Changes in v2:

- added Reviewed-by tags from Andrew for patches #1 and #3
- expanded commit message in #2 to explain the change
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agonet: phy: broadcom: Wire suspend/resume for BCM54810
Florian Fainelli [Wed, 19 Feb 2020 20:00:49 +0000 (12:00 -0800)]
net: phy: broadcom: Wire suspend/resume for BCM54810

The BCM54810 PHY can use the standard BMCR Power down suspend, but needs
a custom resume routine which first clear the Power down bit, and then
re-initializes the PHY. While in low-power mode, the PHY only accepts
writes to the BMCR register. The datasheet clearly says it:

Reads or writes to any MII register other than MII Control register
(address 00h) while the device is in the standby power-down mode may
cause unpredictable results.

Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agonet: phy: broadcom: Have bcm54xx_adjust_rxrefclk() check for flags
Florian Fainelli [Wed, 19 Feb 2020 20:00:48 +0000 (12:00 -0800)]
net: phy: broadcom: Have bcm54xx_adjust_rxrefclk() check for flags

bcm54xx_adjust_rxrefclk() already checks for PHY_BRCM_AUTO_PWRDWN_ENABLE
and PHY_BRCM_DIS_TXCRXC_NOENRGY in order to set the appropriate bit. The
situation is a bit more complicated with the flag
PHY_BRCM_RX_REFCLK_UNUSED but essentially amounts to the same situation.

The default setting for the 125MHz clock is to be on for all PHYs and
we still treat BCM50610 and BCM50610M specifically with the polarity of
the bit reversed.

Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agonet: phy: broadcom: Allow BCM54810 to use bcm54xx_adjust_rxrefclk()
Florian Fainelli [Wed, 19 Feb 2020 20:00:47 +0000 (12:00 -0800)]
net: phy: broadcom: Allow BCM54810 to use bcm54xx_adjust_rxrefclk()

The function bcm54xx_adjust_rxrefclk() works correctly on the BCM54810
PHY, allow this device ID to proceed through.

Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agosfc: remove unused variable 'efx_default_channel_type'
YueHaibing [Wed, 19 Feb 2020 01:34:58 +0000 (09:34 +0800)]
sfc: remove unused variable 'efx_default_channel_type'

drivers/net/ethernet/sfc/efx.c:116:38: warning:
 efx_default_channel_type defined but not used [-Wunused-const-variable=]

commit 83975485077d ("sfc: move channel alloc/removal code")
left behind this, remove it.

Reported-by: Hulk Robot <hulkci@huawei.com>
Fixes: 83975485077d ("sfc: move channel alloc/removal code")
Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Acked-by: Martin Habets <mhabets@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoMerge branch 'hns3-next'
David S. Miller [Thu, 20 Feb 2020 00:25:12 +0000 (16:25 -0800)]
Merge branch 'hns3-next'

Huazhong Tan says:

====================
net: hns3: misc updates for -net-next

This series includes some misc updates for the HNS3
ethernet driver.

[patch 1] modifies an unsuitable print when setting dulex mode.
[patch 2] adds some debugfs info for TC and DWRR.
[patch 3] adds some debugfs info for loopback.
[patch 4] adds a missing help info for QS shaper in debugfs.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agonet: hns3: add missing help info for QS shaper in debugfs
Yonglong Liu [Wed, 19 Feb 2020 01:23:33 +0000 (09:23 +0800)]
net: hns3: add missing help info for QS shaper in debugfs

HNS3 driver can dump QS shaper configs via debugfs, but missing
help info in debugfs for this operation.

Signed-off-by: Yonglong Liu <liuyonglong@huawei.com>
Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agonet: hns3: add support for dump MAC ID and loopback status in debugfs
Yufeng Mo [Wed, 19 Feb 2020 01:23:32 +0000 (09:23 +0800)]
net: hns3: add support for dump MAC ID and loopback status in debugfs

The MAC ID and loopback status information are obtained from
the hardware, which will be helpful for debugging. This patch
adds support for these two items in debugfs.

Signed-off-by: Yufeng Mo <moyufeng@huawei.com>
Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agonet: hns3: add enabled TC numbers and DWRR weight info in debugfs
Yonglong Liu [Wed, 19 Feb 2020 01:23:31 +0000 (09:23 +0800)]
net: hns3: add enabled TC numbers and DWRR weight info in debugfs

The actual enabled TC numbers and the DWRR weight of each
TC may be helpful for debugging, so adds them into debugfs.

Signed-off-by: Yonglong Liu <liuyonglong@huawei.com>
Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agonet: hns3: modify an unsuitable print when setting unknown duplex to fibre
Guangbin Huang [Wed, 19 Feb 2020 01:23:30 +0000 (09:23 +0800)]
net: hns3: modify an unsuitable print when setting unknown duplex to fibre

Currently, if device is in link down status and user uses
'ethtool -s' command to set speed but not specify duplex
mode, the duplex mode passed from ethtool to driver is
unknown value(255), and the fibre port will identify this
value as half duplex mode and print "only copper port
supports half duplex!". This message is confusing.

So for fibre port, only the setting duplex is half, prints
error and returns.

Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: Huazhong Tan <tanhuazhong@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agomlxsw: Replace zero-length array with flexible-array member
Gustavo A. R. Silva [Tue, 18 Feb 2020 20:57:05 +0000 (14:57 -0600)]
mlxsw: Replace zero-length array with flexible-array member

The current codebase makes use of the zero-length array language
extension to the C90 standard, but the preferred mechanism to declare
variable-length types such as these ones is a flexible array member[1][2],
introduced in C99:

struct foo {
        int stuff;
        struct boo array[];
};

By making use of the mechanism above, we will get a compiler warning
in case the flexible array does not occur last in the structure, which
will help us prevent some kind of undefined behavior bugs from being
inadvertently introduced[3] to the codebase from now on.

Also, notice that, dynamic memory allocations won't be affected by
this change:

"Flexible array members have incomplete type, and so the sizeof operator
may not be applied. As a quirk of the original implementation of
zero-length arrays, sizeof evaluates to zero."[1]

This issue was found with the help of Coccinelle.

[1] https://gcc.gnu.org/onlinedocs/gcc/Zero-Length.html
[2] https://github.com/KSPP/linux/issues/21
[3] commit 76497732932f ("cxgb3/l2t: Fix undefined behaviour")

Signed-off-by: Gustavo A. R. Silva <gustavo@embeddedor.com>
Reviewed-by: Ido Schimmel <idosch@mellanox.com>
Tested-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agophy: avoid unnecessary link-up delay in polling mode
Petr Oros [Tue, 18 Feb 2020 09:35:55 +0000 (10:35 +0100)]
phy: avoid unnecessary link-up delay in polling mode

commit 93c0970493c71f ("net: phy: consider latched link-down status in
polling mode") removed double-read of latched link-state register for
polling mode from genphy_update_link(). This added extra ~1s delay into
sequence link down->up.
Following scenario:
 - After boot link goes up
 - phy_start() is called triggering an aneg restart, hence link goes
   down and link-down info is latched.
 - After aneg has finished link goes up. In phy_state_machine is checked
   link state but it is latched "link is down". The state machine is
   scheduled after one second and there is detected "link is up". This
   extra delay can be avoided when we keep link-state register double read
   in case when link was down previously.

With this solution we don't miss a link-down event in polling mode and
link-up is faster.

Details about this quirky behavior on Realtek phy:
Without patch:
T0:    aneg is started, link goes down, link-down status is latched
T0+3s: state machine runs, up-to-date link-down is read
T0+4s: state machine runs, aneg is finished (BMSR_ANEGCOMPLETE==1),
       here i read link-down (BMSR_LSTATUS==0),
T0+5s: state machine runs, aneg is finished (BMSR_ANEGCOMPLETE==1),
       up-to-date link-up is read (BMSR_LSTATUS==1),
       phydev->link goes up, state change PHY_NOLINK to PHY_RUNNING

With patch:
T0:    aneg is started, link goes down, link-down status is latched
T0+3s: state machine runs, up-to-date link-down is read
T0+4s: state machine runs, aneg is finished (BMSR_ANEGCOMPLETE==1),
       first BMSR read: BMSR_ANEGCOMPLETE==1 and BMSR_LSTATUS==0,
       second BMSR read: BMSR_ANEGCOMPLETE==1 and BMSR_LSTATUS==1,
       phydev->link goes up, state change PHY_NOLINK to PHY_RUNNING

Signed-off-by: Petr Oros <poros@redhat.com>
Reviewed-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoigc: Add dump options
Sasha Neftin [Tue, 14 Jan 2020 07:55:54 +0000 (09:55 +0200)]
igc: Add dump options

Placeholder for debugging functionality.
In this patch, we add some registers and rings summary dumps.

Signed-off-by: Sasha Neftin <sasha.neftin@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
4 years agoigc: Complete to commit Add legacy power management support
Sasha Neftin [Mon, 13 Jan 2020 13:33:59 +0000 (15:33 +0200)]
igc: Complete to commit Add legacy power management support

commit 9513d2a5dc7f ("igc: Add legacy power management support")
Add power management resume and schedule suspend requests.
Add power management get and put synchronization.

Signed-off-by: Sasha Neftin <sasha.neftin@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
4 years agoigc: make non-global functions static
Chen Zhou [Wed, 8 Jan 2020 13:39:59 +0000 (21:39 +0800)]
igc: make non-global functions static

Fix sparse warning:
drivers/net/ethernet/intel/igc/igc_ptp.c:512:6:
warning: symbol 'igc_ptp_tx_work' was not declared. Should it be static?
drivers/net/ethernet/intel/igc/igc_ptp.c:644:6:
warning: symbol 'igc_ptp_suspend' was not declared. Should it be static?

Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: Chen Zhou <chenzhou10@huawei.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
4 years agonet: intel: e1000e: fix possible sleep-in-atomic-context bugs in e1000e_get_hw_semaph...
Jia-Ju Bai [Wed, 18 Dec 2019 14:16:56 +0000 (22:16 +0800)]
net: intel: e1000e: fix possible sleep-in-atomic-context bugs in e1000e_get_hw_semaphore()

The driver may sleep while holding a spinlock.
The function call path (from bottom to top) in Linux 4.19 is:

drivers/net/ethernet/intel/e1000e/mac.c, 1366:
usleep_range in e1000e_get_hw_semaphore
drivers/net/ethernet/intel/e1000e/80003es2lan.c, 322:
e1000e_get_hw_semaphore in e1000_release_swfw_sync_80003es2lan
drivers/net/ethernet/intel/e1000e/80003es2lan.c, 197:
e1000_release_swfw_sync_80003es2lan in e1000_release_phy_80003es2lan
drivers/net/ethernet/intel/e1000e/netdev.c, 4883:
(FUNC_PTR) e1000_release_phy_80003es2lan in e1000e_update_phy_stats
drivers/net/ethernet/intel/e1000e/netdev.c, 4917:
e1000e_update_phy_stats in e1000e_update_stats
drivers/net/ethernet/intel/e1000e/netdev.c, 5945:
e1000e_update_stats in e1000e_get_stats64
drivers/net/ethernet/intel/e1000e/netdev.c, 5944:
spin_lock in e1000e_get_stats64

drivers/net/ethernet/intel/e1000e/mac.c, 1384:
usleep_range in e1000e_get_hw_semaphore
drivers/net/ethernet/intel/e1000e/80003es2lan.c, 322:
e1000e_get_hw_semaphore in e1000_release_swfw_sync_80003es2lan
drivers/net/ethernet/intel/e1000e/80003es2lan.c, 197:
e1000_release_swfw_sync_80003es2lan in e1000_release_phy_80003es2lan
drivers/net/ethernet/intel/e1000e/netdev.c, 4883:
(FUNC_PTR) e1000_release_phy_80003es2lan in e1000e_update_phy_stats
drivers/net/ethernet/intel/e1000e/netdev.c, 4917:
e1000e_update_phy_stats in e1000e_update_stats
drivers/net/ethernet/intel/e1000e/netdev.c, 5945:
e1000e_update_stats in e1000e_get_stats64
drivers/net/ethernet/intel/e1000e/netdev.c, 5944:
spin_lock in e1000e_get_stats64

(FUNC_PTR) means a function pointer is called.

To fix these bugs, usleep_range() is replaced with udelay().

These bugs are found by a static analysis tool STCheck written by myself.

Signed-off-by: Jia-Ju Bai <baijiaju1990@gmail.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
4 years agoe1000e: fix missing cpu_to_le64 on buffer_addr
Ben Dooks (Codethink) [Wed, 18 Dec 2019 09:53:08 +0000 (09:53 +0000)]
e1000e: fix missing cpu_to_le64 on buffer_addr

The following warning suggests there is a missing cpu_to_le64() in
the e1000_flush_tx_ring() function (it is also the behaviour
elsewhere in the driver to do cpu_to_le64() on the buffer_addr
when setting it)

drivers/net/ethernet/intel/e1000e/netdev.c:3813:30: warning: incorrect type in assignment (different base types)
drivers/net/ethernet/intel/e1000e/netdev.c:3813:30:    expected restricted __le64 [usertype] buffer_addr
drivers/net/ethernet/intel/e1000e/netdev.c:3813:30:    got unsigned long long [usertype] dma

Signed-off-by: Ben Dooks (Codethink) <ben.dooks@codethink.co.uk>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
4 years agoice: fix define for E822 backplane device
Bruce Allan [Thu, 13 Feb 2020 21:31:29 +0000 (13:31 -0800)]
ice: fix define for E822 backplane device

This product's name has changed; update the macro identifier accordingly.

Signed-off-by: Bruce Allan <bruce.w.allan@intel.com>
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>
4 years agoice: add support for E823 devices
Bruce Allan [Thu, 13 Feb 2020 21:31:28 +0000 (13:31 -0800)]
ice: add support for E823 devices

Add E823 device ids and convert conditional expressions to a more
appropriate switch statement.

Signed-off-by: Bruce Allan <bruce.w.allan@intel.com>
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>
4 years agoice: add additional E810 device id
Bruce Allan [Thu, 13 Feb 2020 21:31:27 +0000 (13:31 -0800)]
ice: add additional E810 device id

Add support for device id 0x159b.

Signed-off-by: Bruce Allan <bruce.w.allan@intel.com>
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>
4 years agoice: add backslash-n to strings
Jesse Brandeburg [Thu, 13 Feb 2020 21:31:26 +0000 (13:31 -0800)]
ice: add backslash-n to strings

There were several strings found without line feeds, fix
them by adding a line feed, as is typical.  Without this
lotsofmessagescanbejumbledtogether.

This patch has known checkpatch warnings from long lines
for the NL_* messages, because checkpatch doesn't know
how to ignore them.

Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
4 years agoice: increase PF reset wait timeout to 300 milliseconds
Jacob Keller [Thu, 13 Feb 2020 21:31:25 +0000 (13:31 -0800)]
ice: increase PF reset wait timeout to 300 milliseconds

Increase the maximum time that the driver will wait for a PF reset from
200 milliseconds to 300 milliseconds, to account for possibility of
a slightly longer than expected PF reset.

Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
4 years agoice: Support XDP UMEM wake up mechanism
Krzysztof Kazimierczak [Thu, 13 Feb 2020 21:31:22 +0000 (13:31 -0800)]
ice: Support XDP UMEM wake up mechanism

Add support for a new AF_XDP feature that has already been introduced in
upstreamed Intel NIC drivers. If a user space application signals that
it might sleep using the new bind flag XDP_USE_NEED_WAKEUP, the driver
will then set this flag if it has no more buffers on the NIC Rx ring and
yield to the application. For Tx, it will set the flag if it has no
outstanding Tx completion interrupts and return to the application.

Signed-off-by: Krzysztof Kazimierczak <krzysztof.kazimierczak@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
4 years agoice: SW DCB, report correct max TC value
Dave Ertman [Thu, 13 Feb 2020 21:31:21 +0000 (13:31 -0800)]
ice: SW DCB, report correct max TC value

lldpad is using the value reported in the DCB config for
max_tc as the max allowed number of TCs, not the current
max.  ICE driver was reporting it as current maximum TC.

Change DCB_NL function to report maximum TC allowed by
this device.

Signed-off-by: Dave Ertman <david.m.ertman@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
4 years agoice: Report correct DCB mode
Avinash Dayanand [Thu, 13 Feb 2020 21:31:20 +0000 (13:31 -0800)]
ice: Report correct DCB mode

Add code to detect if DCB is in IEEE or CEE mode. Without this the code
will always report as IEEE mode which is incorrect and confuses the
user.

Signed-off-by: Avinash Dayanand <avinash.dayanand@intel.com>
Signed-off-by: Scott Register <scottx.register@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
4 years agoice: Add DCBNL ops required to configure ETS in CEE for SW DCB
Avinash JD [Thu, 13 Feb 2020 21:31:19 +0000 (13:31 -0800)]
ice: Add DCBNL ops required to configure ETS in CEE for SW DCB

Couple of DCBNL ops are required for configuring ETS in SW DCB CEE mode. If
these functions are not added, it'll break the CEE functionality.

Signed-off-by: Avinash JD <avinash.dayanand@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
4 years agoice: Always clear the QRXFLXP_CNTXT register for VF Rx queues
Brett Creeley [Thu, 13 Feb 2020 21:31:18 +0000 (13:31 -0800)]
ice: Always clear the QRXFLXP_CNTXT register for VF Rx queues

Currently when the PF reduces its number of channels via ethtool and
then VFs are created there may be stale data for some of the Rx queues
belonging to VFs. This happens when a VF reuses an Rx queue that was
previously used by the PF. Specifically, the QRXFLXP_CNTXT register
will have incorrect values. Fix this by always clearing the relevant
values in the QRXFLXP_CNTXT register for VF queues.

Signed-off-by: Brett Creeley <brett.creeley@intel.com>
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>
4 years agoice: Fix for TCAM entry management
Dan Nowlin [Thu, 13 Feb 2020 21:31:17 +0000 (13:31 -0800)]
ice: Fix for TCAM entry management

Order intermediate VSIG list correct in order to correctly match existing
VSIG lists.

When overriding pre-existing TCAM entries, properly delete the existing
entry and remove it from the change/update list.

Signed-off-by: Dan Nowlin <dan.nowlin@intel.com>
Tested-by: Andrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
4 years agoice: update malicious driver detection event handling
Paul Greenwalt [Thu, 13 Feb 2020 21:31:16 +0000 (13:31 -0800)]
ice: update malicious driver detection event handling

Update the PF VFs MDD event message to rate limit once per second and
report the total number Rx|Tx event count. Add support to print pending
MDD events that occur during the rate limit. The use of net_ratelimit did
not allow for per VF Rx|Tx granularity.

Additional PF MDD log messages are guarded by netif_msg_[rx|tx]_err().

Since VF RX MDD events disable the queue, add ethtool private flag
mdd-auto-reset-vf to configure VF reset to re-enable the queue.

Disable anti-spoof detection interrupt to prevent spurious events
during a function reset.

To avoid race condition do not make PF MDD register reads conditional
on global MDD result.

Signed-off-by: Paul Greenwalt <paul.greenwalt@intel.com>
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>
4 years agoice: Validate config for SW DCB map
Avinash Dayanand [Thu, 13 Feb 2020 21:30:59 +0000 (13:30 -0800)]
ice: Validate config for SW DCB map

Validate the inputs for SW DCB config received either via lldptool or pcap
file. And don't apply DCB for bad bandwidth inputs. Without this patch, any
config having bad inputs will cause the loss of link making PF unusable
even after driver reload. Recoverable only via system reboot.

Signed-off-by: Avinash Dayanand <avinash.dayanand@intel.com>
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>
4 years agoMerge branch 'tcp_v6_gso_csum_prep'
David S. Miller [Wed, 19 Feb 2020 19:21:00 +0000 (11:21 -0800)]
Merge branch 'tcp_v6_gso_csum_prep'

Heiner Kallweit says:

====================
net: core: add helper tcp_v6_gso_csum_prep

Several network drivers for chips that support TSO6 share the same code
for preparing the TCP header, so let's factor it out to a helper.
A difference is that some drivers reset the payload_len whilst others
don't do this. This value is overwritten by TSO anyway, therefore
the new helper resets it in general.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agovmxnet3: use new helper tcp_v6_gso_csum_prep
Heiner Kallweit [Tue, 18 Feb 2020 20:13:58 +0000 (21:13 +0100)]
vmxnet3: use new helper tcp_v6_gso_csum_prep

Use new helper tcp_v6_gso_csum_prep in additional network drivers.

Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agor8152: use new helper tcp_v6_gso_csum_prep
Heiner Kallweit [Tue, 18 Feb 2020 20:12:52 +0000 (21:12 +0100)]
r8152: use new helper tcp_v6_gso_csum_prep

Use new helper tcp_v6_gso_csum_prep in additional network drivers.

Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Acked-by: Hayes Wang <hayeswang@realtek.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agohv_netvsc: use new helper tcp_v6_gso_csum_prep
Heiner Kallweit [Tue, 18 Feb 2020 20:11:43 +0000 (21:11 +0100)]
hv_netvsc: use new helper tcp_v6_gso_csum_prep

Use new helper tcp_v6_gso_csum_prep in additional network drivers.

Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agonet: socionext: use new helper tcp_v6_gso_csum_prep
Heiner Kallweit [Tue, 18 Feb 2020 20:09:17 +0000 (21:09 +0100)]
net: socionext: use new helper tcp_v6_gso_csum_prep

Use new helper tcp_v6_gso_csum_prep in additional network drivers.

Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Acked-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agonet: qcom/emac: use new helper tcp_v6_gso_csum_prep
Heiner Kallweit [Tue, 18 Feb 2020 20:08:21 +0000 (21:08 +0100)]
net: qcom/emac: use new helper tcp_v6_gso_csum_prep

Use new helper tcp_v6_gso_csum_prep in additional network drivers.

Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoionic: use new helper tcp_v6_gso_csum_prep
Heiner Kallweit [Tue, 18 Feb 2020 20:07:16 +0000 (21:07 +0100)]
ionic: use new helper tcp_v6_gso_csum_prep

Use new helper tcp_v6_gso_csum_prep in additional network drivers.

Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Acked-by: Shannon Nelson <snelson@pensando.io>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agojme: use new helper tcp_v6_gso_csum_prep
Heiner Kallweit [Tue, 18 Feb 2020 20:06:11 +0000 (21:06 +0100)]
jme: use new helper tcp_v6_gso_csum_prep

Use new helper tcp_v6_gso_csum_prep in additional network drivers.

Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoe1000(e): use new helper tcp_v6_gso_csum_prep
Heiner Kallweit [Tue, 18 Feb 2020 20:05:02 +0000 (21:05 +0100)]
e1000(e): use new helper tcp_v6_gso_csum_prep

Use new helper tcp_v6_gso_csum_prep in additional network drivers.

Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Reviewed-by: Alexander Duyck <alexander.h.duyck@linux.intel.com>
Acked-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoenic: use new helper tcp_v6_gso_csum_prep
Heiner Kallweit [Tue, 18 Feb 2020 20:02:26 +0000 (21:02 +0100)]
enic: use new helper tcp_v6_gso_csum_prep

Use new helper tcp_v6_gso_csum_prep in additional network drivers.

Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agobna: use new helper tcp_v6_gso_csum_prep
Heiner Kallweit [Tue, 18 Feb 2020 20:01:14 +0000 (21:01 +0100)]
bna: use new helper tcp_v6_gso_csum_prep

Use new helper tcp_v6_gso_csum_prep in additional network drivers.

Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agonet: atheros: use new helper tcp_v6_gso_csum_prep
Heiner Kallweit [Tue, 18 Feb 2020 19:59:55 +0000 (20:59 +0100)]
net: atheros: use new helper tcp_v6_gso_csum_prep

Use new helper tcp_v6_gso_csum_prep in additional network drivers.

Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agor8169: use new helper tcp_v6_gso_csum_prep
Heiner Kallweit [Tue, 18 Feb 2020 19:58:14 +0000 (20:58 +0100)]
r8169: use new helper tcp_v6_gso_csum_prep

Simplify the code by using the new helper tcp_v6_gso_csum_prep.

Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agonet: core: add helper tcp_v6_gso_csum_prep
Heiner Kallweit [Tue, 18 Feb 2020 19:56:41 +0000 (20:56 +0100)]
net: core: add helper tcp_v6_gso_csum_prep

Several network drivers for chips that support TSO6 share the same code
for preparing the TCP header, so let's factor it out to a helper.
A difference is that some drivers reset the payload_len whilst others
don't do this. This value is overwritten by TSO anyway, therefore
the new helper resets it in general.

Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Reviewed-by: Alexander Duyck <alexander.h.duyck@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agonet/ipv4/sysctl: show tcp_{allowed, available}_congestion_control in non-initial...
Christian Brauner [Wed, 19 Feb 2020 12:02:53 +0000 (13:02 +0100)]
net/ipv4/sysctl: show tcp_{allowed, available}_congestion_control in non-initial netns

It is currenty possible to switch the TCP congestion control algorithm
in non-initial network namespaces:

unshare -U --map-root --net --fork --pid --mount-proc
echo "reno" > /proc/sys/net/ipv4/tcp_congestion_control

works just fine. But currently non-initial network namespaces have no
way of kowing which congestion algorithms are available or allowed other
than through trial and error by writing the names of the algorithms into
the aforementioned file.
Since we already allow changing the congestion algorithm in non-initial
network namespaces by exposing the tcp_congestion_control file there is
no reason to not also expose the
tcp_{allowed,available}_congestion_control files to non-initial network
namespaces. After this change a container with a separate network
namespace will show:

root@f1:~# ls -al /proc/sys/net/ipv4/tcp_* | grep congestion
-rw-r--r-- 1 root root 0 Feb 19 11:54 /proc/sys/net/ipv4/tcp_allowed_congestion_control
-r--r--r-- 1 root root 0 Feb 19 11:54 /proc/sys/net/ipv4/tcp_available_congestion_control
-rw-r--r-- 1 root root 0 Feb 19 11:54 /proc/sys/net/ipv4/tcp_congestion_control

Link: https://github.com/lxc/lxc/issues/3267
Reported-by: Haw Loeung <haw.loeung@canonical.com>
Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agonet: mvneta: align xdp stats naming scheme to mlx5 driver
Lorenzo Bianconi [Wed, 19 Feb 2020 09:57:37 +0000 (10:57 +0100)]
net: mvneta: align xdp stats naming scheme to mlx5 driver

Introduce "rx" prefix in the name scheme for xdp counters
on rx path.
Differentiate between XDP_TX and ndo_xdp_xmit counters

Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Acked-by: Jesper Dangaard Brouer <brouer@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoMerge branch 'octeontx2-af-Cleanup-changes'
David S. Miller [Wed, 19 Feb 2020 18:51:12 +0000 (10:51 -0800)]
Merge branch 'octeontx2-af-Cleanup-changes'

Sunil Goutham says:

====================
octeontx2-af: Cleanup changes

These patches cleanup AF driver by removing unnecessary function
exports and cleanup repititive logic.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoocteontx2-af: Cleanup nixlf and blkaddr retrieval logic
Sunil Goutham [Wed, 19 Feb 2020 09:51:08 +0000 (15:21 +0530)]
octeontx2-af: Cleanup nixlf and blkaddr retrieval logic

Cleanedup repititive nixlf and blkaddr retrieving logic
is various mailbox handlers throughout the rvu_nix.c file.

Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoocteontx2-af: Cleanup CGX config permission checks
Sunil Goutham [Wed, 19 Feb 2020 09:51:07 +0000 (15:21 +0530)]
octeontx2-af: Cleanup CGX config permission checks

Most of the CGX register config is restricted to mapped RVU PFs,
this patch cleans up these permission checks spread across
the rvu_cgx.c file by moving the checks to a common fn().

Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoocteontx2-af: Remove unnecessary export symbols in CGX driver
Sunil Goutham [Wed, 19 Feb 2020 09:51:06 +0000 (15:21 +0530)]
octeontx2-af: Remove unnecessary export symbols in CGX driver

Since CGX driver and AF driver are built into a single module
the export symbols in CGX driver are not needed. This patch
gets rid of them.

Signed-off-by: Sunil Goutham <sgoutham@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agoMerge tag 'mlx5-updates-2020-01-24' of git://git.kernel.org/pub/scm/linux/kernel...
David S. Miller [Wed, 19 Feb 2020 18:44:25 +0000 (10:44 -0800)]
Merge tag 'mlx5-updates-2020-01-24' of git://git./linux/kernel/git/saeed/linux

Saeed Mahameed says:

====================
This series adds two moderate updates and some misc small patches to
mlx5 driver.

1) From Aya, Add the missing devlink health dump callbacks support for
both rx and tx health reporters.

First patch of the series is extending devlink API to set binary fmsg
data.

All others patches in the series are adding the mlx5 devlink health
callbacks support and the needed FW commands.

2) Also from Aya, Support for FEC modes based on 50G per lane links.
Part of this series, Aya adds one missing link mode define "FEC_LLRS"
to include/uapi/linux/ethtool.h.

3) From Joe, Use proper logging and tracing line terminations

4) From Christophe, Remove a useless 'drain_workqueue()'
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agonet/mlx5: Remove a useless 'drain_workqueue()' call in 'mlx5e_ipsec_cleanup()'
Christophe JAILLET [Tue, 11 Feb 2020 22:32:54 +0000 (14:32 -0800)]
net/mlx5: Remove a useless 'drain_workqueue()' call in 'mlx5e_ipsec_cleanup()'

'destroy_workqueue()' already calls 'drain_workqueue()', there is no need
to call it explicitly.

Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
4 years agonet/mlx5e: Add support for FEC modes based on 50G per lane links
Aya Levin [Tue, 11 Feb 2020 22:32:53 +0000 (14:32 -0800)]
net/mlx5e: Add support for FEC modes based on 50G per lane links

Introduce new FEC modes:
- RS-FEC-(544,514)
- LL_RS-FEC-(272,257+1)
Add support in ethtool for set and get callbacks for the new modes
above. While RS-FEC-(544,514) is mapped to exsiting RS FEC mode,
LL_RS-FEC-(272,257+1) is mapped to a new ethtool link mode: LL-RS.

Add support for FEC on 50G per lane link modes up to 400G. The new link
modes uses a u16 fields instead of u8 fields for the legacy link modes.

Signed-off-by: Aya Levin <ayal@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
4 years agoethtool: Add support for low latency RS FEC
Aya Levin [Tue, 11 Feb 2020 22:32:52 +0000 (14:32 -0800)]
ethtool: Add support for low latency RS FEC

Add support for low latency Reed Solomon FEC as LLRS.

The LL-FEC is defined by the 25G/50G ethernet consortium,
in the document titled "Low Latency Reed Solomon Forward Error Correction"

Signed-off-by: Aya Levin <ayal@mellanox.com>
Reviewed-by: Eran Ben Elisha <eranbe@mellanox.com>
CC: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
4 years agonet/mlxe5: Separate between FEC and current speed
Aya Levin [Tue, 11 Feb 2020 22:32:51 +0000 (14:32 -0800)]
net/mlxe5: Separate between FEC and current speed

FEC mode is per link type, not necessary per speed. This patch access
FEC register by link modes instead of speeds. This patch will allow
further enhacment of link modes supporting FEC with the same speed
(different lane type).

Signed-off-by: Aya Levin <ayal@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
4 years agonet/mlx5e: Advertise globaly supported FEC modes
Aya Levin [Tue, 11 Feb 2020 22:32:50 +0000 (14:32 -0800)]
net/mlx5e: Advertise globaly supported FEC modes

Ethtool advertise supported link modes on an interface. Per each FEC
mode, query if there is a link type which supports it. If so, add this
FEC mode to the supported FEC modes list. Prior to this patch, ethtool
advertised only the supported FEC modes on the current link type.
Add an explicit mapping between internal FEC modes and ethtool link mode
bits. With this change, adding new FEC modes in the downstream patch
would be easier.

Signed-off-by: Aya Levin <ayal@mellanox.com>
Reviewed-by: Eran Ben Elisha <eranbe@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
4 years agonet/mlx5e: Enforce setting of a single FEC mode
Aya Levin [Tue, 11 Feb 2020 22:32:49 +0000 (14:32 -0800)]
net/mlx5e: Enforce setting of a single FEC mode

Ethtool command allow setting of several FEC modes in a single set
command. The driver can only set a single FEC mode at a time. With this
patch driver will reply not-supported on setting several FEC modes.

Signed-off-by: Aya Levin <ayal@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
4 years agonet/mlx5e: Set FEC to auto when configured mode is not supported
Aya Levin [Tue, 11 Feb 2020 22:32:48 +0000 (14:32 -0800)]
net/mlx5e: Set FEC to auto when configured mode is not supported

When configuring FEC mode, driver tries to set it for all available
link types. If a link type doesn't support a FEC mode, set this link
type to auto (FW best effort). Prior to this patch, when a link type
didn't support a FEC mode is was set to no FEC.

Signed-off-by: Aya Levin <ayal@mellanox.com>
Reviewed-by: Eran Ben Elisha <eranbe@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
4 years agomlx5: Use proper logging and tracing line terminations
Joe Perches [Tue, 11 Feb 2020 22:32:47 +0000 (14:32 -0800)]
mlx5: Use proper logging and tracing line terminations

netdev_err should use newline termination but mlx5_health_report
is used in a trace output function devlink_health_report where
no newline should be used.

Remove the newlines from a couple formats and add a format string
of "%s\n" to the netdev_err call to not directly output the
logging string.

Also use snprintf to avoid any possible output string overrun.

Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
4 years agonet/mlx5e: Support dump callback in RX reporter
Aya Levin [Tue, 11 Feb 2020 22:32:46 +0000 (14:32 -0800)]
net/mlx5e: Support dump callback in RX reporter

Add support for SQ's FW dump on RX reporter's events. Use Resource dump
API to retrieve the relevant data: RX slice, RQ dump, RX buffer and
ICOSQ dump (depends on the error). Wrap it in formatted messages and
store the binary output in devlink core.

Example:
$ devlink health dump show pci/0000:00:0b.0 reporter rx
RX Slice:
   data:
     00 00 00 00 00 00 00 80 00 01 00 00 00 00 ad de
     22 01 00 00 00 00 ad de 00 00 00 00 00 00 00 00
     00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
     ff ff ff ff 01 00 00 00 00 00 00 00 00 00 00 00
     00 00 00 00 00 00 00 80 00 01 00 00 00 00 ad de
     22 01 00 00 00 00 ad de 00 00 00 00 00 00 00 00
     00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
     ff ff ff ff 01 00 00 00 00 00 00 00 00 00 00 00
     00 00 00 00 00 00 00 80 00 01 00 00 00 00 ad de
  RQs:
    RQ:
      rqn: 1512
      data:
        00 00 00 00 00 00 00 80 00 01 00 00 00 00 ad de
        22 01 00 00 00 00 ad de 00 00 00 00 00 00 00 00
        00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
        ff ff ff ff 01 00 00 00 00 00 00 00 00 00 00 00
        00 00 00 00 00 00 00 80 00 01 00 00 00 00 ad de
        22 01 00 00 00 00 ad de 00 00 00 00 00 00 00 00
        00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
        ff ff ff ff 01 00 00 00 00 00 00 00 00 00 00 00
        00 00 00 00 00 00 00 80 00 01 00 00 00 00 ad de
    RQ:
      rqn: 1517
      data:
        00 00 00 00 00 00 00 80 00 01 00 00 00 00 ad de
        22 01 00 00 00 00 ad de 00 00 00 00 00 00 00 00
        00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
        ff ff ff ff 01 00 00 00 00 00 00 00 00 00 00 00
        00 00 00 00 00 00 00 80 00 01 00 00 00 00 ad de
        22 01 00 00 00 00 ad de 00 00 00 00 00 00 00 00
        00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
        ff ff ff ff 01 00 00 00 00 00 00 00 00 00 00 00
        00 00 00 00 00 00 00 80 00 01 00 00 00 00 ad de

$ devlink health dump show pci/0000:00:0b.0 reporter rx -jp
{
    "RX Slice": {
     "data":[ 0,0,0,0,0,0,0,128,0,1,0,0,0,0,173,222,34,1,0,0,0,0,173,222,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,255,255,255,255,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,128,0,1,0,0,0,0,173,222,34,1,0,0,0,0,173,222,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,255,255,255,255,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,128,0,1,0,0,0,0,173,222,34,1,0,0,0,0,173,222,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,255,255,255,255,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,128,0,1,0,0,0,0,173,222,34,1,0,0,0,0,173,222,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,255,255,255,255,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,128,0,1,0,0,0,0,173,222,34,1,0,0,0,0,173,222,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,255,255,255,255,1,0,0,0,0,0,0,0,0,0,0,0,0,2,1,0,0,0,0,128,0,1,0,0,0,0,173,222]
    },
    "RQs": [ {
            "RQ": {
                "index": 1512,
                "data": [ 0,0,0,0,0,0,0,128,0,1,0,0,0,0,173,222,34,1,0,0,0,0,173,222,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,255,255,255,255,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,128,0,1,0,0,0,0,173,222,34,1,0,0,0,0,173,222,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,255,255,255,255,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,128,0,1,0,0,0,0,173,222,34,1,0,0,0,0,173,222,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,255,255,255,255,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,128,0,1,0,0,0,0,173,222,34,1,0,0,0,0,173,222,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,255,255,255,255,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,128,0,1,0,0,0,0,173,222,34,1,0,0,0,0,173,222,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,255,255,255,255,1,0,0,0,0,0,0,0,0,0,0,0,0,2,1,0,0,0,0,128,0,1,0,0,0,0,173,222]
            }
        },{
            "RQ": {
                "index": 1517,
                "data": [ 0,0,0,0,0,0,0,128,0,1,0,0,0,0,173,222,34,1,0,0,0,0,173,222,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,255,255,255,255,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,128,0,1,0,0,0,0,173,222,34,1,0,0,0,0,173,222,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,255,255,255,255,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,128,0,1,0,0,0,0,173,222,34,1,0,0,0,0,173,222,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,255,255,255,255,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,128,0,1,0,0,0,0,173,222,34,1,0,0,0,0,173,222,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,255,255,255,255,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,128,0,1,0,0,0,0,173,222,34,1,0,0,0,0,173,222,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,255,255,255,255,1,0,0,0,0,0,0,0,0,0,0,0,0,2,1,0,0,0,0,128,0,1,0,0,0,0,173]
            }
        } ]
}

Signed-off-by: Aya Levin <ayal@mellanox.com>
Reviewed-by: Moshe Shemesh <moshe@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
4 years agonet/mlx5e: Support dump callback in TX reporter
Aya Levin [Tue, 11 Feb 2020 22:32:45 +0000 (14:32 -0800)]
net/mlx5e: Support dump callback in TX reporter

Add support for SQ's FW dump on TX reporter's events. Use Resource dump
API to retrieve the relevant data: SX slice, SQ dump and SQ buffer. Wrap
it in formatted messages and store the binary output in devlink core.

Example:
$ devlink health dump show pci/0000:00:0b.0 reporter tx
SX Slice:
   data:
     00 00 00 00 00 00 00 80 00 01 00 00 00 00 ad de
     22 01 00 00 00 00 ad de 00 00 00 00 00 00 00 00
     00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
     ff ff ff ff 01 00 00 00 00 00 00 00 00 00 00 00
     00 02 01 00 00 00 00 80 00 01 00 00 00 00 ad de
     22 01 00 00 00 00 ad de 00 20 40 90 81 88 ff ff
     00 00 00 00 00 00 00 00 15 00 15 00 00 00 00 00
     ff ff ff ff 01 00 00 00 00 00 00 00 00 00 00 00
     00 00 00 00 00 00 00 80 81 ae 41 06 00 ea ff ff
  SQs:
    SQ:
      index: 1511
      data:
        00 00 00 00 00 00 00 80 00 01 00 00 00 00 ad de
        22 01 00 00 00 00 ad de 00 00 00 00 00 00 00 00
        00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
        ff ff ff ff 01 00 00 00 00 00 00 00 00 00 00 00
        00 02 01 00 00 00 00 80 00 01 00 00 00 00 ad de
        22 01 00 00 00 00 ad de 00 20 40 90 81 88 ff ff
        00 00 00 00 00 00 00 00 15 00 15 00 00 00 00 00
        ff ff ff ff 01 00 00 00 00 00 00 00 00 00 00 00
        00 00 00 00 00 00 00 80 81 ae 41 06 00 ea ff ff
    SQ:
      index: 1516
      data:
        00 00 00 00 00 00 00 80 00 01 00 00 00 00 ad de
        22 01 00 00 00 00 ad de 00 00 00 00 00 00 00 00
        00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
        ff ff ff ff 01 00 00 00 00 00 00 00 00 00 00 00
        00 02 01 00 00 00 00 80 00 01 00 00 00 00 ad de
        22 01 00 00 00 00 ad de 00 20 40 90 81 88 ff ff
        00 00 00 00 00 00 00 00 15 00 15 00 00 00 00 00
        ff ff ff ff 01 00 00 00 00 00 00 00 00 00 00 00
        00 00 00 00 00 00 00 80 81 ae 41 06 00 ea ff ff

$ devlink health dump show pci/0000:00:0b.0 reporter tx -jp
{
    "SX Slice": {
     "data": [ 0,0,0,0,0,0,0,128,0,1,0,0,0,0,173,222,34,1,0,0,0,0,173,222,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,255,255,255,255,1,0,0,0,0,0,0,0,0,0,0,0,0,2,1,0,0,0,0,128,0,1,0,0,0,0,173,222,34,1,0,0,0,0,173,222,0,32,64,144,129,136,255,255,0,0,0,0,0,0,0,0,21,0,21,0,0,0,0,0,255,255,255,255,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,128,129,174,65,6,0,234,255,255],
     },
    "SQs": [ {
            "SQ": {
                "index": 1511,
                "data": [ 0,0,0,0,0,0,0,128,0,1,0,0,0,0,173,222,34,1,0,0,0,0,173,222,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,255,255,255,255,1,0,0,0,0,0,0,0,0,0,0,0,0,2,1,0,0,0,0,128,0,1,0,0,0,0,173,222,34,1,0,0,0,0,173,222,0,32,64,144,129,136,255,255,0,0,0,0,0,0,0,0,21,0,21,0,0,0,0,0,255,255,255,255,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,128,129,174,65,6,0,234,255,255]
            }
        },{
            "SQ": {
                "index": 1516,
                "data": [ 0,0,0,0,0,0,0,128,0,1,0,0,0,0,173,222,34,1,0,0,0,0,173,222,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,255,255,255,255,1,0,0,0,0,0,0,0,0,0,0,0,0,2,1,0,0,0,0,128,0,1,0,0,0,0,173,222,34,1,0,0,0,0,173,222,0,32,64,144,129,136,255,255,0,0,0,0,0,0,0,0,21,0,21,0,0,0,0,0,255,255,255,255,1,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,128,129,174,65,6,0,234,255,255]
            }
        } ]
}

Signed-off-by: Aya Levin <ayal@mellanox.com>
Reviewed-by: Moshe Shemesh <moshe@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
4 years agonet/mlx5e: Gather reporters APIs together
Aya Levin [Tue, 11 Feb 2020 22:32:44 +0000 (14:32 -0800)]
net/mlx5e: Gather reporters APIs together

Assemble all the API's to ease insertion of dump callbacks in the
following patches in the set.

Signed-off-by: Aya Levin <ayal@mellanox.com>
Reviewed-by: Moshe Shemesh <moshe@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
4 years agonet/mlx5: Add support for resource dump
Aya Levin [Tue, 11 Feb 2020 22:32:43 +0000 (14:32 -0800)]
net/mlx5: Add support for resource dump

On driver load:
- Initialize resource dump data structure and memory access tools (mkey
  & pd).
- Read the resource dump's menu which contains the FW segment
  identifier. Each record is identified by the segment name (ASCII).

During the driver's course of life, users (like reporters) may request
dumps per segment. The user should create a command providing the
segment identifier (SW enumeration) and command keys. In return, the
user receives a command context. In order to receive the dump, the user
should supply the command context and a memory (aligned to a PAGE) on
which the dump content will be written. Since the dump may be larger
than the given memory, the user may resubmit the command until received
an indication of end-of-dump. It is the user's responsibility to destroy
the command.

Signed-off-by: Aya Levin <ayal@mellanox.com>
Reviewed-by: Moshe Shemesh <moshe@mellanox.com>
Acked-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
4 years agodevlink: Force enclosing array on binary fmsg data
Aya Levin [Tue, 11 Feb 2020 22:32:42 +0000 (14:32 -0800)]
devlink: Force enclosing array on binary fmsg data

Add a new API for start/end binary array brackets [] to force array
around binary data as required from JSON. With this restriction, re-open
API to set binary fmsg data.

Signed-off-by: Aya Levin <ayal@mellanox.com>
Reviewed-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
4 years agosfc: elide assignment of skb
Edward Cree [Tue, 18 Feb 2020 17:34:00 +0000 (17:34 +0000)]
sfc: elide assignment of skb

Instead of assigning skb = segments before the loop, just pass
 segments directly as the first argument to skb_list_walk_safe().

Signed-off-by: Edward Cree <ecree@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
4 years agonet: fec: Prevent unbind operation
Fabio Estevam [Tue, 18 Feb 2020 15:34:44 +0000 (12:34 -0300)]
net: fec: Prevent unbind operation

After performing an unbind/bind operation the network is no longer
functional on i.MX6 (which has a single FEC instance):

# echo 2188000.ethernet > /sys/bus/platform/drivers/fec/unbind
# echo 2188000.ethernet > /sys/bus/platform/drivers/fec/bind
[   10.756519] pps pps0: new PPS source ptp0
[   10.792626] libphy: fec_enet_mii_bus: probed
[   10.799330] fec 2188000.ethernet eth0: registered PHC device 1
# udhcpc -i eth0
udhcpc: started, v1.31.1
[   14.985211] fec 2188000.ethernet eth0: no PHY, assuming direct connection to switch
[   14.993140] libphy: PHY fixed-0:00 not found
[   14.997643] fec 2188000.ethernet eth0: could not attach to PHY

On SoCs with two FEC instances there are some cases where one FEC instance
depends on the other one being present. One such example is i.MX28, which
has the following FEC dependency as noted in the comments:

/*
 * The i.MX28 dual fec interfaces are not equal.
 * Here are the differences:
 *
 *  - fec0 supports MII & RMII modes while fec1 only supports RMII
 *  - fec0 acts as the 1588 time master while fec1 is slave
 *  - external phys can only be configured by fec0
 *
 * That is to say fec1 can not work independently. It only works
 * when fec0 is working. The reason behind this design is that the
 * second interface is added primarily for Switch mode.
 *
 * Because of the last point above, both phys are attached on fec0
 * mdio interface in board design, and need to be configured by
 * fec0 mii_bus.
 */

Prevent the unbind operation to avoid these issues.

Signed-off-by: Fabio Estevam <festevam@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>