platform/kernel/linux-starfive.git
2 years agonet: add CONFIG_DEBUG_NET
Eric Dumazet [Tue, 10 May 2022 03:57:38 +0000 (20:57 -0700)]
net: add CONFIG_DEBUG_NET

This config option enables network debugging checks.

This patch adds DEBUG_NET_WARN_ON_ONCE(cond)
Note that this is not a replacement for WARN_ON_ONCE(cond)
as (cond) is not evaluated if CONFIG_DEBUG_NET is not set.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: add include/net/net_debug.h
Eric Dumazet [Tue, 10 May 2022 03:57:37 +0000 (20:57 -0700)]
net: add include/net/net_debug.h

Remove from include/linux/netdevice.h helpers
that send debug/info/warnings to syslog.

We plan adding more helpers in following patches.

v2: added two includes, and 'struct net_device' forward declaration
    to avoid compile errors (kernel bots)

Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoMerge tag 'mlx5-updates-2022-05-09' of git://git.kernel.org/pub/scm/linux/kernel...
David S. Miller [Wed, 11 May 2022 11:12:27 +0000 (12:12 +0100)]
Merge tag 'mlx5-updates-2022-05-09' of git://git./linux/kernel/git/saeed/linux

Saeed  Mahameed says:

====================
mlx5-updates-2022-05-09

1) Gavin Li, adds exit route from waiting for FW init on device boot and
   increases FW init timeout on health recovery flow

2) Support 4 ports HCAs LAG mode

Mark Bloch Says:
================

This series adds to mlx5 drivers support for 4 ports HCAs.
Starting with ConnectX-7 HCAs with 4 ports are possible.

As most driver parts aren't affected by such configuration most driver
code is unchanged.

Specially the only affected areas are:
- Lag
- Devcom
- Merged E-Switch
- Single FDB E-Switch

Lag was chosen to be converted first. Creating hardware LAG when all 4
ports are added to the same bond device.

Devom, merge E-Switch and single FDB E-Switch, are marked as supporting
only 2 ports HCAs and future patches will add support for 4 ports HCAs.

In order to activate the hardware lag a user can execute the:

ip link add bond0 type bond
ip link set bond0 type bond miimon 100 mode 2
ip link set eth2 master bond0
ip link set eth3 master bond0
ip link set eth4 master bond0
ip link set eth5 master bond0

Where eth2, eth3, eth4 and eth5 are the PFs of the same HCA.

================

====================

Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoMerge branch 'net-phy-add-comments-for-lan8742-phy-support'
Jakub Kicinski [Wed, 11 May 2022 01:57:17 +0000 (18:57 -0700)]
Merge branch 'net-phy-add-comments-for-lan8742-phy-support'

Yuiko Oshino says:

====================
net: phy: add comments for LAN8742 phy support

Add comments for 0xfffffff2 phy ID mask for the LAN8742 and the LAN88xx, explaining that they can coexist and allow future hardware revisions.
Also add one missing tab in smsc.c.
====================

Link: https://lore.kernel.org/r/20220509185804.7147-1-yuiko.oshino@microchip.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agonet: phy: smsc: add comments for the LAN8742 phy ID mask.
Yuiko Oshino [Mon, 9 May 2022 18:58:04 +0000 (11:58 -0700)]
net: phy: smsc: add comments for the LAN8742 phy ID mask.

add comments for the LAN8742 phy ID mask in the previous patch.
add one missing tab in the LAN8742 phy ID line.

Signed-off-by: Yuiko Oshino <yuiko.oshino@microchip.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agonet: phy: microchip: add comments for the modified LAN88xx phy ID mask.
Yuiko Oshino [Mon, 9 May 2022 18:58:03 +0000 (11:58 -0700)]
net: phy: microchip: add comments for the modified LAN88xx phy ID mask.

add comments for the updated LAN88xx phy ID mask in the previous patch.

Signed-off-by: Yuiko Oshino <yuiko.oshino@microchip.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agodt-bindings: net: orion-mdio: Convert to JSON schema
Chris Packham [Thu, 5 May 2022 21:06:20 +0000 (09:06 +1200)]
dt-bindings: net: orion-mdio: Convert to JSON schema

Convert the marvell,orion-mdio binding to JSON schema.

Signed-off-by: Chris Packham <chris.packham@alliedtelesis.co.nz>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Link: https://lore.kernel.org/r/20220505210621.3637268-1-chris.packham@alliedtelesis.co.nz
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoMerge branch 'docs-document-some-aspects-of-struct-sk_buff'
Jakub Kicinski [Wed, 11 May 2022 00:48:40 +0000 (17:48 -0700)]
Merge branch 'docs-document-some-aspects-of-struct-sk_buff'

Jakub Kicinski says:

====================
docs: document some aspects of struct sk_buff

This small set creates a place to render sk_buff documentation,
documents one random thing (data-only skbs) and converts the big
checksum comment to kdoc.
====================

Link: https://lore.kernel.org/r/20220323233715.2104106-1-kuba@kernel.org/
Link: https://lore.kernel.org/r/20220324231312.2241166-1-kuba@kernel.org/
Link: https://lore.kernel.org/r/20220509160456.1058940-1-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoskbuff: render the checksum comment to documentation
Jakub Kicinski [Mon, 9 May 2022 16:04:56 +0000 (09:04 -0700)]
skbuff: render the checksum comment to documentation

Long time ago Tom added a giant comment to skbuff.h explaining
checksums. Now that we have a place in Documentation for skbuff
docs we should render it. Sprinkle some markup while at it.

Reviewed-by: David Ahern <dsahern@kernel.org>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoskbuff: rewrite the doc for data-only skbs
Jakub Kicinski [Mon, 9 May 2022 16:04:55 +0000 (09:04 -0700)]
skbuff: rewrite the doc for data-only skbs

The comment about shinfo->dataref split is really unhelpful,
at least to me. Rewrite it and render it to skb documentation.

Reviewed-by: David Ahern <dsahern@kernel.org>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoskbuff: add a basic intro doc
Jakub Kicinski [Mon, 9 May 2022 16:04:54 +0000 (09:04 -0700)]
skbuff: add a basic intro doc

Add basic skb documentation. It's mostly an intro to the subsequent
patches - it would looks strange if we documented advanced topics
without covering the basics in any way.

Reviewed-by: David Ahern <dsahern@kernel.org>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agonet: fix kdoc on __dev_queue_xmit()
Jakub Kicinski [Mon, 9 May 2022 17:04:12 +0000 (10:04 -0700)]
net: fix kdoc on __dev_queue_xmit()

Commit c526fd8f9f4f21 ("net: inline dev_queue_xmit()") exported
__dev_queue_xmit(), now it's being rendered in html docs, triggering:

Documentation/networking/kapi:92: net/core/dev.c:4101: WARNING: Missing matching underline for section title overline.

Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
Link: https://lore.kernel.org/linux-next/20220503073420.6d3f135d@canb.auug.org.au/
Fixes: c526fd8f9f4f21 ("net: inline dev_queue_xmit()")
Link: https://lore.kernel.org/r/20220509170412.1069190-1-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoMerge branch 'move-siena-into-a-separate-subdirectory'
Jakub Kicinski [Tue, 10 May 2022 22:38:17 +0000 (15:38 -0700)]
Merge branch 'move-siena-into-a-separate-subdirectory'

Martin Habets says:

====================
Move Siena into a separate subdirectory

The Siena NICs (SFN5000 and SFN6000 series) went EOL in November 2021.
Most of these adapters have been remove from our test labs, and testing
has been reduced to a minimum.

This patch series creates a separate kernel module for the Siena architecture,
analogous to what was done for Falcon some years ago.
This reduces our maintenance for the sfc.ko module, and allows us to
enhance the EF10 and EF100 drivers without the risk of breaking Siena NICs.

After this series further enhancements are needed to differentiate the
new kernel module from sfc.ko, and the Siena code can be removed from sfc.ko.
Thes will be posted as a small follow-up series.
The Siena module is not built by default, but can be enabled
using Kconfig option SFC_SIENA. This will create module sfc-siena.ko.

Patches

Patches 1-3 establish the code base for the Siena driver.
Patches 4-10 ensure the allyesconfig build succeeds.
Patch 11 adds the basic Siena module.

I do not expect patch 1 through 3 to be reviewed, they are FYI only.
No checkpatch issues were resolved as part of these, but they
were fixed in the subsequent patches.

Testing

Various build tests were done such as allyesconfig, W=1 and sparse.
The new sfc-siena.ko and sfc.ko modules were tested on a machine with both
these NICs in them, and several tests were run on both drivers.
====================

Link: https://lore.kernel.org/r/165211018297.5289.9658523545298485394.stgit@palantir17.mph.net
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agosfc: Add a basic Siena module
Martin Habets [Mon, 9 May 2022 15:33:23 +0000 (16:33 +0100)]
sfc: Add a basic Siena module

Make the (un)load message more specific to differentiate it from
the sfc.ko messages.

Signed-off-by: Martin Habets <habetsm.xilinx@gmail.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agosfc/siena: Inline functions in sriov.h to avoid conflicts with sfc
Martin Habets [Mon, 9 May 2022 15:33:10 +0000 (16:33 +0100)]
sfc/siena: Inline functions in sriov.h to avoid conflicts with sfc

The implementation of each is quite short. This means sriov.c is
not needed any more.

Signed-off-by: Martin Habets <habetsm.xilinx@gmail.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agosfc/siena: Rename functions in nic_common.h to avoid conflicts with sfc
Martin Habets [Mon, 9 May 2022 15:32:58 +0000 (16:32 +0100)]
sfc/siena: Rename functions in nic_common.h to avoid conflicts with sfc

For siena use efx_siena_ as the function prefix.
efx_nic_update_stats_atomic is only used in efx_common.c, so move
it there.
efx_nic_copy_stats is not used in Siena, so it is removed.

Signed-off-by: Martin Habets <habetsm.xilinx@gmail.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agosfc/siena: Rename functions in mcdi headers to avoid conflicts with sfc
Martin Habets [Mon, 9 May 2022 15:32:45 +0000 (16:32 +0100)]
sfc/siena: Rename functions in mcdi headers to avoid conflicts with sfc

For siena use efx_siena_ as the function prefix.
Several functions are not used in Siena, so they are removed.

Signed-off-by: Martin Habets <habetsm.xilinx@gmail.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agosfc/siena: Rename peripheral functions to avoid conflicts with sfc
Martin Habets [Mon, 9 May 2022 15:32:33 +0000 (16:32 +0100)]
sfc/siena: Rename peripheral functions to avoid conflicts with sfc

For siena use efx_siena_ as the function prefix.
This patch covers selftest.h, ptp.h, net_driver.h and ethtool_common.h.
efx_ethtool_fill_self_tests() can become static.
Some functions in ptp.c can also become static.
Rename loopback_mode in net_driver.h.

Signed-off-by: Martin Habets <habetsm.xilinx@gmail.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agosfc/siena: Rename RX/TX functions to avoid conflicts with sfc
Martin Habets [Mon, 9 May 2022 15:32:20 +0000 (16:32 +0100)]
sfc/siena: Rename RX/TX functions to avoid conflicts with sfc

For siena use efx_siena_ as the function prefix.
Several functions are not used in Siena, so they are removed.

Use a Siena specific variable name for module parameter
efx_separate_tx_channels.
Move efx_fini_tx_queue() to avoid a forward declaration of
efx_dequeue_buffer().

Signed-off-by: Martin Habets <habetsm.xilinx@gmail.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agosfc/siena: Rename functions in efx headers to avoid conflicts with sfc
Martin Habets [Mon, 9 May 2022 15:32:08 +0000 (16:32 +0100)]
sfc/siena: Rename functions in efx headers to avoid conflicts with sfc

When building with allyesconfig there are many identical
symbol names.
For siena use efx_siena_ as the function and variable prefix
to avoid build errors.

efx_mtd_remove_partition can become static as it is no longer called
from other files.
efx_ticks_to_usecs and efx_xmit_done_single are not used in Siena, so
they are removed.
Several functions are only used inside efx_channels.c for Siena so
they can become static.

Signed-off-by: Martin Habets <habetsm.xilinx@gmail.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agosfc/siena: Remove build references to missing functionality
Martin Habets [Mon, 9 May 2022 15:31:55 +0000 (16:31 +0100)]
sfc/siena: Remove build references to missing functionality

Functionality not supported or needed on Siena includes:
- Anything for EF100
- EF10 specifics such as register access, PIO and TSO offload.
Also only bind to Siena NICs.

Remove EF10 specifics from nic.h.
The functions that start with efx_farch_ will be removed from sfc.ko
with a subsequent patch.
Add the efx_ prefix to siena_prepare_flush() to make it consistent
with the other APIs.

Signed-off-by: Martin Habets <habetsm.xilinx@gmail.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agosfc: Copy shared files needed for Siena (part 2)
Martin Habets [Mon, 9 May 2022 15:31:43 +0000 (16:31 +0100)]
sfc: Copy shared files needed for Siena (part 2)

These are the files starting with m through w.
No changes are done, those will be done with subsequent commits.

Signed-off-by: Martin Habets <habetsm.xilinx@gmail.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agosfc: Copy shared files needed for Siena (part 1)
Martin Habets [Mon, 9 May 2022 15:31:31 +0000 (16:31 +0100)]
sfc: Copy shared files needed for Siena (part 1)

These are the files starting with b through i.
No changes are done, those will be done with subsequent commits.

Signed-off-by: Martin Habets <habetsm.xilinx@gmail.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agosfc: Move Siena specific files
Martin Habets [Mon, 9 May 2022 15:31:18 +0000 (16:31 +0100)]
sfc: Move Siena specific files

Files are only moved, no changes are made.

Signed-off-by: Martin Habets <habetsm.xilinx@gmail.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agonfp: flower: fix 'variable 'flow6' set but not used'
Louis Peens [Tue, 10 May 2022 07:48:45 +0000 (09:48 +0200)]
nfp: flower: fix 'variable 'flow6' set but not used'

Kernel test robot reported an issue after a recent patch about an
unused variable when CONFIG_IPV6 is disabled. Move the variable
declaration to be inside the #ifdef, and do a bit more cleanup. There
is no need to use a temporary ipv6 bool value, it is just checked once,
remove the extra variable and just do the check directly.

Fixes: 9d5447ed44b5 ("nfp: flower: fixup ipv6/ipv4 route lookup for neigh events")
Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Louis Peens <louis.peens@corigine.com>
Signed-off-by: Simon Horman <simon.horman@corigine.com>
Link: https://lore.kernel.org/r/20220510074845.41457-1-simon.horman@corigine.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agox25: remove redundant pointer dev
Colin Ian King [Sun, 8 May 2022 21:45:00 +0000 (22:45 +0100)]
x25: remove redundant pointer dev

Pointer dev is being assigned a value that is never used, the assignment
and the variable are redundant and can be removed. Also replace null check
with the preferred !ptr idiom.

Cleans up clang scan warning:
net/x25/x25_proc.c:94:26: warning: Although the value stored to 'dev' is
used in the enclosing expression, the value is never actually read
from 'dev' [deadcode.DeadStores]

Signed-off-by: Colin Ian King <colin.i.king@gmail.com>
Link: https://lore.kernel.org/r/20220508214500.60446-1-colin.i.king@gmail.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2 years agoMerge branch 'this-is-a-patch-series-for-ethernet-driver-of-sunplus-sp7021-soc'
Paolo Abeni [Tue, 10 May 2022 09:31:34 +0000 (11:31 +0200)]
Merge branch 'this-is-a-patch-series-for-ethernet-driver-of-sunplus-sp7021-soc'

Wells Lu says:

====================
This is a patch series for Ethernet driver of Sunplus SP7021 SoC.

Sunplus SP7021 is an ARM Cortex A7 (4 cores) based SoC. It integrates
many peripherals (ex: UART, I2C, SPI, SDIO, eMMC, USB, SD card and
etc.) into a single chip. It is designed for industrial control
applications.

Refer to:
https://sunplus.atlassian.net/wiki/spaces/doc/overview
https://tibbo.com/store/plus1.html
====================

Link: https://lore.kernel.org/r/1652004800-3212-1-git-send-email-wellslutw@gmail.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2 years agonet: ethernet: Add driver for Sunplus SP7021
Wells Lu [Sun, 8 May 2022 10:13:20 +0000 (18:13 +0800)]
net: ethernet: Add driver for Sunplus SP7021

Add driver for Sunplus SP7021 SoC.

Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Wells Lu <wellslutw@gmail.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2 years agodevicetree: bindings: net: Add bindings doc for Sunplus SP7021.
Wells Lu [Sun, 8 May 2022 10:13:19 +0000 (18:13 +0800)]
devicetree: bindings: net: Add bindings doc for Sunplus SP7021.

Add bindings documentation for Sunplus SP7021 SoC.

Reviewed-by: Rob Herring <robh@kernel.org>
Signed-off-by: Wells Lu <wellslutw@gmail.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2 years agoMerge branch 'ptp-support-hardware-clocks-with-additional-free-running-cycle-counter'
Paolo Abeni [Tue, 10 May 2022 07:48:11 +0000 (09:48 +0200)]
Merge branch 'ptp-support-hardware-clocks-with-additional-free-running-cycle-counter'

Gerhard Engleder says:

====================
ptp: Support hardware clocks with additional free running cycle counter

ptp vclocks require a clock with free running time for the timecounter.
Currently only a physical clock forced to free running is supported.
If vclocks are used, then the physical clock cannot be synchronized
anymore. The synchronized time is not available in hardware in this
case. As a result, timed transmission with TAPRIO hardware support
is not possible anymore.

If hardware would support a free running time additionally to the
physical clock, then the physical clock does not need to be forced to
free running. Thus, the physical clocks can still be synchronized while
vclocks are in use.

The physical clock could be used to synchronize the time domain of the
TSN network and trigger TAPRIO. In parallel vclocks can be used to
synchronize other time domains.

One year ago I thought for two time domains within a TSN network also
two physical clocks are required. This would lead to new kernel
interfaces for asking for the second clock, ... . But actually for a
time triggered system like TSN there can be only one time domain that
controls the system itself. All other time domains belong to other
layers, but not to the time triggered system itself. So other time
domains can be based on a free running counter if similar mechanisms
like 2 step synchroisation are used.

Synchronisation was tested with two time domains between two directly
connected hosts. Each host run two ptp4l instances, the first used the
physical clock and the second used the virtual clock. I used my FPGA
based network controller as network device. ptp4l was used in
combination with the virtual clock support patches from Miroslav
Lichvar.

v4:
- if_index of 0 is invalid (Jonathan Lemon)
- set if_index to 0 in the SOF_TIMESTAMPING_RAW_HARDWARE block (Jonathan
  Lemon)
- add helper function for netdev_get_tstamp() call (Jonathan Lemon)
- update SKBTX_ANY_TSTAMP (Paolo Abeni)
- use separate bits for new tx_flags (Richard Cochran)

v3:
- optimize ptp_convert_timestamp (Richard Cochran)
- call dev_get_by_napi_id() only if needed (Richard Cochran)
- use non-negated logical test (Richard Cochran)
- add comment for skipped output (Richard Cochran)
- add comment for SKBTX_HW_TSTAMP_USE_CYCLES masking (Richard Cochran)

v2:
- rename ptp_clock cycles to has_cycles (Richard Cochran)
- call it free running cycle counter (Richard Cochran)
- update struct skb_shared_hwtstamps kdoc (Richard Cochran)
- optimize timestamp address/cookie processing path (Richard Cochran,
  Vinicius Costa Gomes)

v1:
- complete rework based on suggestions (Richard Cochran)
====================

Link: https://lore.kernel.org/r/20220506200142.3329-1-gerhard@engleder-embedded.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2 years agotsnep: Add free running cycle counter support
Gerhard Engleder [Fri, 6 May 2022 20:01:42 +0000 (22:01 +0200)]
tsnep: Add free running cycle counter support

The TSN endpoint Ethernet MAC supports a free running counter
additionally to its clock. This free running counter can be read and
hardware timestamps are supported. As the name implies, this counter
cannot be set and its frequency cannot be adjusted.

Add free running cycle counter support based on this free running
counter to physical clock. This also requires hardware time stamps
based on that free running counter.

Signed-off-by: Gerhard Engleder <gerhard@engleder-embedded.com>
Acked-by: Jonathan Lemon <jonathan.lemon@gmail.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2 years agoptp: Speed up vclock lookup
Gerhard Engleder [Fri, 6 May 2022 20:01:41 +0000 (22:01 +0200)]
ptp: Speed up vclock lookup

ptp_convert_timestamp() is called in the RX path of network messages.
The current implementation takes ~5000ns on 1.2GHz A53. This is too much
for the hot path of packet processing.

Introduce hash table for fast vclock lookup in ptp_convert_timestamp().
The execution time of ptp_convert_timestamp() is reduced to ~700ns on
1.2GHz A53.

Signed-off-by: Gerhard Engleder <gerhard@engleder-embedded.com>
Acked-by: Richard Cochran <richardcochran@gmail.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2 years agoptp: Support late timestamp determination
Gerhard Engleder [Fri, 6 May 2022 20:01:40 +0000 (22:01 +0200)]
ptp: Support late timestamp determination

If a physical clock supports a free running cycle counter, then
timestamps shall be based on this time too. For TX it is known in
advance before the transmission if a timestamp based on the free running
cycle counter is needed. For RX it is impossible to know which timestamp
is needed before the packet is received and assigned to a socket.

Support late timestamp determination by a network device. Therefore, an
address/cookie is stored within the new netdev_data field of struct
skb_shared_hwtstamps. This address/cookie is provided to a new network
device function called ndo_get_tstamp(), which returns a timestamp based
on the normal/adjustable time or based on the free running cycle
counter. If function is not supported, then timestamp handling is not
changed.

This mechanism is intended for RX, but TX use is also possible.

Signed-off-by: Gerhard Engleder <gerhard@engleder-embedded.com>
Acked-by: Jonathan Lemon <jonathan.lemon@gmail.com>
Acked-by: Richard Cochran <richardcochran@gmail.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2 years agoptp: Pass hwtstamp to ptp_convert_timestamp()
Gerhard Engleder [Fri, 6 May 2022 20:01:39 +0000 (22:01 +0200)]
ptp: Pass hwtstamp to ptp_convert_timestamp()

ptp_convert_timestamp() converts only the timestamp hwtstamp, which is
a field of the argument with the type struct skb_shared_hwtstamps *. So
a pointer to the hwtstamp field of this structure is sufficient.

Rework ptp_convert_timestamp() to use an argument of type ktime_t *.
This allows to add additional timestamp manipulation stages before the
call of ptp_convert_timestamp().

Signed-off-by: Gerhard Engleder <gerhard@engleder-embedded.com>
Acked-by: Richard Cochran <richardcochran@gmail.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2 years agoptp: Request cycles for TX timestamp
Gerhard Engleder [Fri, 6 May 2022 20:01:38 +0000 (22:01 +0200)]
ptp: Request cycles for TX timestamp

The free running cycle counter of physical clocks called cycles shall be
used for hardware timestamps to enable synchronisation.

Introduce new flag SKBTX_HW_TSTAMP_USE_CYCLES, which signals driver to
provide a TX timestamp based on cycles if cycles are supported.

Signed-off-by: Gerhard Engleder <gerhard@engleder-embedded.com>
Acked-by: Richard Cochran <richardcochran@gmail.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2 years agoptp: Add cycles support for virtual clocks
Gerhard Engleder [Fri, 6 May 2022 20:01:37 +0000 (22:01 +0200)]
ptp: Add cycles support for virtual clocks

ptp vclocks require a free running time for their timecounter.
Currently only a physical clock forced to free running is supported.
If vclocks are used, then the physical clock cannot be synchronized
anymore. The synchronized time is not available in hardware in this
case. As a result, timed transmission with TAPRIO hardware support
is not possible anymore.

If hardware would support a free running time additionally to the
physical clock, then the physical clock does not need to be forced to
free running. Thus, the physical clocks can still be synchronized
while vclocks are in use.

The physical clock could be used to synchronize the time domain of the
TSN network and trigger TAPRIO. In parallel vclocks can be used to
synchronize other time domains.

Introduce support for a free running cycle counter called cycles to
physical clocks. Rework ptp vclocks to use this free running cycle
counter. Default implementation is based on time of physical clock.
Thus, behavior of ptp vclocks based on physical clocks without free
running cycle counter is identical to previous behavior.

Signed-off-by: Gerhard Engleder <gerhard@engleder-embedded.com>
Acked-by: Richard Cochran <richardcochran@gmail.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2 years agoeth: dpaa2-mac: remove a dead-code NULL check on fwnode parent
Jakub Kicinski [Fri, 6 May 2022 20:00:29 +0000 (13:00 -0700)]
eth: dpaa2-mac: remove a dead-code NULL check on fwnode parent

Since commit 4e30e98c4b4c ("dpaa2-mac: return -EPROBE_DEFER from dpaa2_mac_open in case the fwnode is not set")
@parent can't be NULL after the if. It's either the address
of the ->fwnode of @dpmacs or @fwnode in case of ACPI.

Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Link: https://lore.kernel.org/r/20220506200029.852310-1-kuba@kernel.org
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2 years agonet/mlx5: Lag, add debugfs to query hardware lag state
Mark Bloch [Tue, 15 Mar 2022 16:56:50 +0000 (16:56 +0000)]
net/mlx5: Lag, add debugfs to query hardware lag state

Lag state has become very complicated with many modes, flags, types and
port selections methods and future work will add additional features.

Add a debugfs to query the current lag state. A new directory named "lag"
will be created under the mlx5 debugfs directory. As the driver has
debugfs per pci function the location will be: <debugfs>/mlx5/<BDF>/lag

For example:
/sys/kernel/debug/mlx5/0000:08:00.0/lag

The following files are exposed:

- state: Returns "active" or "disabled". If "active" it means hardware
         lag is active.

- members: Returns the BDFs of all the members of lag object.

- type: Returns the type of the lag currently configured. Valid only
if hardware lag is active.
* "roce" - Members are bare metal PFs.
* "switchdev" - Members are in switchdev mode.
* "multipath" - ECMP offloads.

- port_sel_mode: Returns the egress port selection method, valid
 only if hardware lag is active.
 * "queue_affinity" - Egress port is selected by
   the QP/SQ affinity.
 * "hash" - Egress port is selected by hash done on
   each packet. Controlled by: xmit_hash_policy of the
   bond device.
- flags: Returns flags that are specific per lag @type. Valid only if
 hardware lag is active.
 * "shared_fdb" - "on" or "off", if "on" single FDB is used.

- mapping: Returns the mapping which is used to select egress port.
   Valid only if hardware lag is active.
   If @port_sel_mode is "hash" returns the active egress ports.
   The hash result will select only active ports.
   if @port_sel_mode is "queue_affinity" returns the mapping
   between the configured port affinity of the QP/SQ and actual
   egress port. For example:
   * 1:1 - Mapping means if the configured affinity is port 1
           traffic will egress via port 1.
   * 1:2 - Mapping means if the configured affinity is port 1
   traffic will egress via port 2. This can happen
   if port 1 is down or in active/backup mode and port 1
   is backup.

Signed-off-by: Mark Bloch <mbloch@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5: Lag, use buckets in hash mode
Mark Bloch [Wed, 2 Mar 2022 15:38:50 +0000 (15:38 +0000)]
net/mlx5: Lag, use buckets in hash mode

When in hardware lag and the NIC has more than 2 ports when one port
goes down need to distribute the traffic between the remaining
active ports.

For better spread in such cases instead of using 1-to-1 mapping and only
4 slots in the hash, use many.

Each port will have many slots that point to it. When a port goes down
go over all the slots that pointed to that port and spread them between
the remaining active ports. Once the port comes back restore the default
mapping.

We will have number_of_ports * MLX5_LAG_MAX_HASH_BUCKETS slots.
Each MLX5_LAG_MAX_HASH_BUCKETS belong to a different port.
The native mapping is such that:

port 1: The first MLX5_LAG_MAX_HASH_BUCKETS slots are: [1, 1, .., 1]
which means if a packet is hased into one of this slots it will hit the
wire via port 1.

port 2: The second MLX5_LAG_MAX_HASH_BUCKETS slots are: [2, 2, .., 2]
which means if a packet is hased into one of this slots it will hit the
wire via port2.

and this mapping is the same of the rest of the ports.
On a failover, lets say port 2 goes down (port 1, 3, 4 are still up).
the new mapping for port 2 will be:

port 2: The second MLX5_LAG_MAX_HASH_BUCKETS are: [1, 3, 1, 4, .., 4]
which means the mapping was changed from the native mapping to a mapping
that consists of only the active ports.

With this if a port goes down the traffic will be split between the
active ports randomly

Signed-off-by: Mark Bloch <mbloch@nvidia.com>
Reviewed-by: Maor Gottlieb <maorg@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5: Lag, refactor dmesg print
Mark Bloch [Wed, 2 Mar 2022 09:12:01 +0000 (09:12 +0000)]
net/mlx5: Lag, refactor dmesg print

Combine dmesg lag prints into a single function.

Signed-off-by: Mark Bloch <mbloch@nvidia.com>
Reviewed-by: Maor Gottlieb <maorg@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5: Support devices with more than 2 ports
Mark Bloch [Tue, 1 Mar 2022 17:34:58 +0000 (17:34 +0000)]
net/mlx5: Support devices with more than 2 ports

Increase the define MLX5_MAX_PORTS to 4 as the driver is ready
to support NICs with 4 ports.

Signed-off-by: Mark Bloch <mbloch@nvidia.com>
Reviewed-by: Maor Gottlieb <maorg@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5: Lag, use actual number of lag ports
Mark Bloch [Tue, 1 Mar 2022 17:20:35 +0000 (17:20 +0000)]
net/mlx5: Lag, use actual number of lag ports

Refactor the entire lag code to use ldev->ports instead of hard-coded
defines (like MLX5_MAX_PORTS) for its operations.

Signed-off-by: Mark Bloch <mbloch@nvidia.com>
Reviewed-by: Maor Gottlieb <maorg@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5: Lag, use hash when in roce lag on 4 ports
Mark Bloch [Tue, 1 Mar 2022 17:34:31 +0000 (17:34 +0000)]
net/mlx5: Lag, use hash when in roce lag on 4 ports

Downstream patches will add support for lag over 4 ports.
In that mode we will only use hash as the uplink selection method.
Using hash instead of queue affinity (before this patch) offers key
advantages like:

- Align ports selection method with the method used by the bond device

- Better packets distribution where a single queue can transmit from
  multiple ports (with queue affinity a queue is bound to a single port
  regardless of the packet being sent).

- In case of failover we traffic is split between multiple ports and not
  a single one like in queue affinity.

Going forward it was decided that queue affinity will be deprecated
as using hash provides a better user experience which means on 4 ports
HCAs hash will always be used.

Future work will add hash support for 2 ports HCAs as well.

Signed-off-by: Mark Bloch <mbloch@nvidia.com>
Reviewed-by: Maor Gottlieb <maorg@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5: Lag, support single FDB only on 2 ports
Mark Bloch [Tue, 1 Mar 2022 17:24:40 +0000 (17:24 +0000)]
net/mlx5: Lag, support single FDB only on 2 ports

E-Switch currently doesn't support more than 2 E-Switch managers
being aggregated under a single hardware lag. Have specific checks
to disallow creating lag when the code doesn't support it.

Signed-off-by: Mark Bloch <mbloch@nvidia.com>
Reviewed-by: Maor Gottlieb <maorg@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5: Lag, store number of ports inside lag object
Mark Bloch [Sun, 27 Feb 2022 13:45:59 +0000 (13:45 +0000)]
net/mlx5: Lag, store number of ports inside lag object

Store the number of lag ports inside the lag object. Lag object is a single
shared object managing the lag state of multiple mlx5 devices on the same
physical HCA.

Downstream patches will allow hardware lag to be created over devices with
more than 2 ports.

Signed-off-by: Mark Bloch <mbloch@nvidia.com>
Reviewed-by: Maor Gottlieb <maorg@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5: Lag, filter non compatible devices
Mark Bloch [Sun, 27 Feb 2022 12:40:39 +0000 (12:40 +0000)]
net/mlx5: Lag, filter non compatible devices

When search for a peer lag device we can filter based on that
device's capabilities.

Downstream patch will be less strict when filtering compatible devices
and remove the limitation where we require exact MLX5_MAX_PORTS and
change it to a range.

Signed-off-by: Mark Bloch <mbloch@nvidia.com>
Reviewed-by: Maor Gottlieb <maorg@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5: Lag, use lag lock
Mark Bloch [Wed, 15 Dec 2021 06:51:18 +0000 (06:51 +0000)]
net/mlx5: Lag, use lag lock

Use a lag specific lock instead of depending on external locks to
synchronise the lag creation/destruction.

With this, taking E-Switch mode lock is no longer needed for syncing
lag logic.

Cleanup any dead code that is left over and don't export functions that
aren't used outside the E-Switch core code.

Signed-off-by: Mark Bloch <mbloch@nvidia.com>
Reviewed-by: Maor Gottlieb <maorg@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5: Lag, move E-Switch prerequisite check into lag code
Mark Bloch [Tue, 1 Mar 2022 16:45:41 +0000 (16:45 +0000)]
net/mlx5: Lag, move E-Switch prerequisite check into lag code

There is no need to expose E-Switch function for something that can be
checked with already present API inside lag code.

Signed-off-by: Mark Bloch <mbloch@nvidia.com>
Reviewed-by: Maor Gottlieb <maorg@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5: devcom only supports 2 ports
Mark Bloch [Sun, 27 Feb 2022 12:23:34 +0000 (12:23 +0000)]
net/mlx5: devcom only supports 2 ports

Devcom API is intended to be used between 2 devices only add this
implied assumption into the code and check when it's no true.

Signed-off-by: Mark Bloch <mbloch@nvidia.com>
Reviewed-by: Maor Gottlieb <maorg@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5: Lag, expose number of lag ports
Mark Bloch [Tue, 1 Mar 2022 15:42:01 +0000 (15:42 +0000)]
net/mlx5: Lag, expose number of lag ports

Downstream patches will add support for hardware lag with
more than 2 ports. Add a way for users to query the number of lag ports.

Signed-off-by: Mark Bloch <mbloch@nvidia.com>
Reviewed-by: Maor Gottlieb <maorg@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5: Increase FW pre-init timeout for health recovery
Gavin Li [Sun, 27 Mar 2022 14:36:44 +0000 (17:36 +0300)]
net/mlx5: Increase FW pre-init timeout for health recovery

Currently, health recovery will reload driver to recover it from fatal
errors. During the driver's load process, it would wait for FW to set the
pre-init bit for up to 120 seconds, beyond this threshold it would abort
the load process. In some cases, such as a FW upgrade on the DPU, this
timeout period is insufficient, and the user has no way to recover the
host device.

To solve this issue, introduce a new FW pre-init timeout for health
recovery, which is set to 2 hours.

The timeout for devlink reload and probe will use the original one because
they are user triggered flows, and therefore should not have a
significantly long timeout, during which the user command would hang.

Signed-off-by: Gavin Li <gavinl@nvidia.com>
Reviewed-by: Moshe Shemesh <moshe@nvidia.com>
Reviewed-by: Shay Drory <shayd@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5: Add exit route when waiting for FW
Gavin Li [Sun, 27 Mar 2022 14:45:32 +0000 (17:45 +0300)]
net/mlx5: Add exit route when waiting for FW

Currently, removing a device needs to get the driver interface lock before
doing any cleanup. If the driver is waiting in a loop for FW init, there
is no way to cancel the wait, instead the device cleanup waits for the
loop to conclude and release the lock.

To allow immediate response to remove device commands, check the TEARDOWN
flag while waiting for FW init, and exit the loop if it has been set.

Signed-off-by: Gavin Li <gavinl@nvidia.com>
Reviewed-by: Moshe Shemesh <moshe@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agoMerge branch 'nfp-support-corigine-pcie-vendor-id'
Jakub Kicinski [Tue, 10 May 2022 01:20:42 +0000 (18:20 -0700)]
Merge branch 'nfp-support-corigine-pcie-vendor-id'

Simon Horman says:

====================
nfp: support Corigine PCIE vendor ID

Historically the nfp driver has supported NFP chips with Netronome's
PCIE vendor ID. This patch extends the driver to also support NFP
chips, which at this point are assumed to be otherwise identical from
a software perspective, that have Corigine's PCIE vendor ID (0x1da8).

This patchset begins by cleaning up strings to make them:
* Vendor neutral for the NFP chip
* Relate to Corigine for the driver itself

It then adds support to the driver for the Corigine's PCIE vendor ID
====================

Link: https://lore.kernel.org/r/20220508173816.476357-1-simon.horman@corigine.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agonfp: support Corigine PCIE vendor ID
Yu Xiao [Sun, 8 May 2022 17:38:16 +0000 (19:38 +0200)]
nfp: support Corigine PCIE vendor ID

Historically the nfp driver has supported NFP chips with Netronome's
PCIE vendor ID. This patch extends the driver to also support NFP
chips, which at this point are assumed to be otherwise identical from
a software perspective, that have Corigine's PCIE vendor ID (0x1da8).

Also, Rename the macro definitions PCI_DEVICE_ID_NERTONEOME_NFPXXXX
to PCI_DEVICE_ID_NFPXXXX, as they are now used in conjunction with two
PCIE vendor IDs.

Signed-off-by: Yu Xiao <yu.xiao@corigine.com>
Signed-off-by: Yinjun Zhang <yinjun.zhang@corigine.com>
Signed-off-by: Simon Horman <simon.horman@corigine.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agonfp: vendor neutral strings for chip and Corigne in strings for driver
Yu Xiao [Sun, 8 May 2022 17:38:15 +0000 (19:38 +0200)]
nfp: vendor neutral strings for chip and Corigne in strings for driver

Historically the nfp driver has supported NFP chips with Netronome's
PCIE vendor ID. In preparation for extending the to also support NFP
chips that have Corigine's PCIE vendor ID (0x1da8) make printk statements
relating to the chip vendor neutral.

An alternate approach is to set the string based on the PCI vendor ID.
In our judgement this proved to cumbersome so we have taken this simpler
approach.

Update strings relating to the driver to use Corigine, who have taken
over maintenance of the driver.

Signed-off-by: Yu Xiao <yu.xiao@corigine.com>
Signed-off-by: Yinjun Zhang <yinjun.zhang@corigine.com>
Signed-off-by: Simon Horman <simon.horman@corigine.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoMerge branch '100GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/next...
Jakub Kicinski [Tue, 10 May 2022 00:51:01 +0000 (17:51 -0700)]
Merge branch '100GbE' of git://git./linux/kernel/git/tnguy/next-queue

Tony Nguyen says:

====================
100GbE Intel Wired LAN Driver Updates 2022-05-06

Marcin Szycik says:

This patchset adds support for systemd defined naming scheme for port
representors, as well as re-enables displaying PCI bus-info in ethtool.

bus-info information has previously been removed from ethtool for port
representors, as a workaround for a bug in lshw tool, where the tool would
sometimes display wrong descriptions for port representors/PF. Now the bug
has been fixed in lshw tool [1].

Removing the workaround can be considered a regression (user might be
running an older, unpatched version of lshw) (see [2] for discussion).
However, calling SET_NETDEV_DEV also produces the same effect as removing
the workaround, i.e. lshw is able to access PCI bus-info (this time not
via ethtool, but in some other way) and the bug can occur.

Adding SET_NETDEV_DEV is important, as it greatly improves netdev naming -
- port representors are named based on PF name. Currently port representors
are named "ethX", which might be confusing, especially when spawning VFs on
multiple PFs. Furthermore, it's currently harder to determine to which PF
does a particular port representor belong, as bus-info is not shown in
ethtool.

Consider the following three cases:

Case 1: current code - driver workaround in place, no SET_NETDEV_DEV,
lshw with or without fix. Port representors are not displayed because they
don't have bus-info (the workaround), PFs are labelled correctly:

$ sudo ./lshw -c net -businfo
Bus info          Device      Class          Description
========================================================
pci@0000:02:00.0  ens6f0      network        Ethernet Controller E810-XXV for SFP <-- PF
pci@0000:02:00.1  ens6f1      network        Ethernet Controller E810-XXV for SFP
pci@0000:02:01.0  ens6f0v0    network        Ethernet Adaptive Virtual Function <-- VF
pci@0000:02:01.1  ens6f0v1    network        Ethernet Adaptive Virtual Function
...

Case 2: driver workaround in place, SET_NETDEV_DEV, no lshw fix. Port
representors have predictable names. lshw is able to get bus-info because
of SET_NETDEV_DEV and netdevs CAN be mislabelled:

$ sudo ./lshw -c net -businfo
Bus info          Device           Class          Description
=============================================================
pci@0000:02:00.0  ens6f0npf0vf60   network        Ethernet Controller E810-XXV for SFP <-- mislabeled port representor
pci@0000:02:00.1  ens6f1           network        Ethernet Controller E810-XXV for SFP
pci@0000:02:01.0  ens6f0v0         network        Ethernet Adaptive Virtual Function
pci@0000:02:01.1  ens6f0v1         network        Ethernet Adaptive Virtual Function
...
pci@0000:02:00.0  ens6f0npf0vf26   network        Ethernet interface
pci@0000:02:00.0  ens6f0           network        Ethernet interface <-- mislabeled PF
pci@0000:02:00.0  ens6f0npf0vf81   network        Ethernet interface
...
$ sudo ethtool -i ens6f0npf0vf60
driver: ice
...
bus-info:
...

Output of lshw would be the same with workaround removed; it does not
change the fact that lshw labels netdevs incorrectly, while at the same
time it prevents ethtool from displaying potentially useful data
(bus-info).

Case 3: workaround removed, SET_NETDEV_DEV, lshw fix:

$ sudo ./lshw -c net -businfo
Bus info          Device           Class          Description
=============================================================
pci@0000:02:00.0  ens6f0npf0vf73   network        Ethernet Controller E810-XXV for SFP
pci@0000:02:00.1  ens6f1           network        Ethernet Controller E810-XXV for SFP
pci@0000:02:01.0  ens6f0v0         network        Ethernet Adaptive Virtual Function
pci@0000:02:01.1  ens6f0v1         network        Ethernet Adaptive Virtual Function
...
pci@0000:02:00.0  ens6f0npf0vf5    network        Ethernet Controller E810-XXV for SFP
pci@0000:02:00.0  ens6f0           network        Ethernet Controller E810-XXV for SFP
pci@0000:02:00.0  ens6f0npf0vf60   network        Ethernet Controller E810-XXV for SFP
...
$ sudo ethtool -i ens6f0npf0vf73
driver: ice
...
bus-info: 0000:02:00.0
...

In this case poort representors have predictable names, netdevs are not
mislabelled in lshw, and bus-info is shown in ethtool.

[1] https://ezix.org/src/pkg/lshw/commit/9bf4e4c9c1
[2] https://patchwork.ozlabs.org/project/intel-wired-lan/patch/20220321144731.3935-1-marcin.szycik@linux.intel.com

* '100GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/next-queue:
  Revert "ice: Hide bus-info in ethtool for PRs in switchdev mode"
  ice: link representors to PCI device
====================

Link: https://lore.kernel.org/r/20220506180052.5256-1-anthony.l.nguyen@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoROSE: Remove unused code and clean up some inconsistent indenting
Jiapeng Chong [Sat, 7 May 2022 03:42:07 +0000 (11:42 +0800)]
ROSE: Remove unused code and clean up some inconsistent indenting

Eliminate the follow smatch warning:

net/rose/rose_route.c:1136 rose_node_show() warn: inconsistent
indenting.

Reported-by: Abaci Robot <abaci@linux.alibaba.com>
Signed-off-by: Jiapeng Chong <jiapeng.chong@linux.alibaba.com>
Link: https://lore.kernel.org/r/20220507034207.18651-1-jiapeng.chong@linux.alibaba.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoMerge branch 'hns3-next'
David S. Miller [Mon, 9 May 2022 13:30:38 +0000 (14:30 +0100)]
Merge branch 'hns3-next'

Guangbin Huang says:

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

This series includes some updates for the HNS3 ethernet driver.

Change logs:
V1 -> V2:
 - Fix some sparse warnings of patch 3# and 4#.
 - Add patch #6 to fix sparse warnings of incorrect type of argument.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: hns3: fix incorrect type of argument in declaration of function hclge_comm_get_r...
Guangbin Huang [Mon, 9 May 2022 07:55:32 +0000 (15:55 +0800)]
net: hns3: fix incorrect type of argument in declaration of function hclge_comm_get_rss_indir_tbl

The argument rss_ind_tbl_size is type u16 in function definition of
hclge_comm_get_rss_indir_tbl(), but it is set to type __le16 in function
declaration by mistake, so fix it.

Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: hns3: add query vf ring and vector map relation
Guangbin Huang [Mon, 9 May 2022 07:55:31 +0000 (15:55 +0800)]
net: hns3: add query vf ring and vector map relation

This patch adds a new mailbox opcode to query map relation between
vf ring and vector.

Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: hns3: add byte order conversion for VF to PF mailbox message
Jie Wang [Mon, 9 May 2022 07:55:30 +0000 (15:55 +0800)]
net: hns3: add byte order conversion for VF to PF mailbox message

This patch uses __le16/__32 to define mailbox data structures. Then byte
order conversion are added for mailbox messages from VF to PF.

Signed-off-by: Jie Wang <wangjie125@huawei.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: hns3: add byte order conversion for PF to VF mailbox message
Jie Wang [Mon, 9 May 2022 07:55:29 +0000 (15:55 +0800)]
net: hns3: add byte order conversion for PF to VF mailbox message

Currently, hns3 mailbox processing between PF and VF missed to convert
message byte order and use data type u16 instead of __le16 for mailbox
data process. These processes may cause problems between different
architectures.

So this patch uses __le16/__le32 data type to define mailbox data
structures. To be compatible with old hns3 driver, these structures use
one-byte alignment. Then byte order conversions are added to mailbox
messages from PF to VF.

Signed-off-by: Jie Wang <wangjie125@huawei.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: hns3: remove the affinity settings of vector0
Yufeng Mo [Mon, 9 May 2022 07:55:28 +0000 (15:55 +0800)]
net: hns3: remove the affinity settings of vector0

Vector0 is used for common interrupt control events and is
irrelevant to performance. Currently, the driver sets the
default affinity of vector0 to NUMA nodes, which is unnecessary.
Therefore, the default setting is removed, and the driver does
not set the affinity for vector0.

Signed-off-by: Yufeng Mo <moyufeng@huawei.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: hns3: fix access null pointer issue when set tx-buf-size as 0
Hao Chen [Mon, 9 May 2022 07:55:27 +0000 (15:55 +0800)]
net: hns3: fix access null pointer issue when set tx-buf-size as 0

When set tx-buf-size as 0 by ethtool, hns3_init_tx_spare_buffer()
will return directly and priv->ring->tx_spare->len is uninitialized,
then print function access priv->ring->tx_spare->len will cause
this issue.

When set tx-buf-size as 0 by ethtool, the print function will
print 0 directly and not access priv->ring->tx_spare->len.

Fixes: 2373b35c24ff ("net: hns3: add log for setting tx spare buf size")
Signed-off-by: Hao Chen <chenhao288@hisilicon.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoMerge branch 'lan8742-phy'
David S. Miller [Mon, 9 May 2022 11:20:57 +0000 (12:20 +0100)]
Merge branch 'lan8742-phy'

Yuiko Oshino says:

====================
net: phy: add LAN8742 phy support

add LAN8742 phy support
update LAN88xx phy ID and phy ID mask so that it can coexist with LAN8742

The current phy IDs on the available hardware.
    LAN8742 0x0007C130, 0x0007C131
    LAN88xx 0x0007C132

v3->v4:
- fixed the one tab missing issue in the smsc.c.

v2->v3:
-added comments about the 0xfffffff2 mask that is for the differentiation and the future revisions.

v1->v2:
-removed "REVIEW REQUEST3" from the PATCH 1/2.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: phy: smsc: add LAN8742 phy support.
Yuiko Oshino [Thu, 5 May 2022 18:12:52 +0000 (11:12 -0700)]
net: phy: smsc: add LAN8742 phy support.

The current phy IDs on the available hardware.
        LAN8742 0x0007C130, 0x0007C131

Signed-off-by: Yuiko Oshino <yuiko.oshino@microchip.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: phy: microchip: update LAN88xx phy ID and phy ID mask.
Yuiko Oshino [Thu, 5 May 2022 18:12:51 +0000 (11:12 -0700)]
net: phy: microchip: update LAN88xx phy ID and phy ID mask.

update LAN88xx phy ID and phy ID mask because the existing code conflicts with the LAN8742 phy.

The current phy IDs on the available hardware.
        LAN8742 0x0007C130, 0x0007C131
        LAN88xx 0x0007C132

Signed-off-by: Yuiko Oshino <yuiko.oshino@microchip.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoMerge branch 'add-ti-dp83td510-support'
David S. Miller [Mon, 9 May 2022 11:09:30 +0000 (12:09 +0100)]
Merge branch 'add-ti-dp83td510-support'

Oleksij Rempel says:

====================
add ti dp83td510 support

changes v4:
- dp83td510: remove unused variables
- s/base1/baset1
- s/genphy_c45_baset1_read_master_slave/genphy_c45_pma_baset1_read_master_slave

changes v3:
- export reusable code snippets and make use of it in the dp83td510
  driver

changes v2:
- rewrite the driver reduce usage of common code and to reduce amount of
  quirks.
- add genphy_c45_baset1_an_config_aneg fix
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: phy: dp83td510: Add support for the DP83TD510 Ethernet PHY
Oleksij Rempel [Fri, 6 May 2022 04:23:57 +0000 (06:23 +0200)]
net: phy: dp83td510: Add support for the DP83TD510 Ethernet PHY

The DP83TD510E is an ultra-low power Ethernet physical layer transceiver
that supports 10M single pair cable.

This driver was tested with NXP SJA1105, STMMAC and ASIX AX88772B USB Ethernet
controller.

Signed-off-by: Oleksij Rempel <o.rempel@pengutronix.de>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: phy: export genphy_c45_baset1_read_status()
Oleksij Rempel [Fri, 6 May 2022 04:23:56 +0000 (06:23 +0200)]
net: phy: export genphy_c45_baset1_read_status()

Export genphy_c45_baset1_read_status() to make it reusable by PHY drivers.

Signed-off-by: Oleksij Rempel <o.rempel@pengutronix.de>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: phy: genphy_c45_pma_baset1_read_master_slave: read actual configuration
Oleksij Rempel [Fri, 6 May 2022 04:23:55 +0000 (06:23 +0200)]
net: phy: genphy_c45_pma_baset1_read_master_slave: read actual configuration

Since MDIO_PMA_PMD_BT1_CTRL register shows actual configuration (and
forced state configuration is equal to the state), we should show
this configuration for ethtool.

Signed-off-by: Oleksij Rempel <o.rempel@pengutronix.de>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: phy: introduce genphy_c45_pma_baset1_read_master_slave()
Oleksij Rempel [Fri, 6 May 2022 04:23:54 +0000 (06:23 +0200)]
net: phy: introduce genphy_c45_pma_baset1_read_master_slave()

Move baset1 specific part of genphy_c45_read_pma() code to
separate function to make it reusable by PHY drivers.

Signed-off-by: Oleksij Rempel <o.rempel@pengutronix.de>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: phy: genphy_c45_pma_baset1_setup_master_slave: do no set unknown configuration
Oleksij Rempel [Fri, 6 May 2022 04:23:53 +0000 (06:23 +0200)]
net: phy: genphy_c45_pma_baset1_setup_master_slave: do no set unknown configuration

Do not change default master/slave forced configuration if no changes was
requested.

Signed-off-by: Oleksij Rempel <o.rempel@pengutronix.de>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: phy: introduce genphy_c45_pma_baset1_setup_master_slave()
Oleksij Rempel [Fri, 6 May 2022 04:23:52 +0000 (06:23 +0200)]
net: phy: introduce genphy_c45_pma_baset1_setup_master_slave()

Move baset1 specific part of genphy_c45_pma_setup_forced() code to
separate function to make it reusable by PHY drivers.

Signed-off-by: Oleksij Rempel <o.rempel@pengutronix.de>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: phy: genphy_c45_baset1_an_config_aneg: do no set unknown configuration
Oleksij Rempel [Fri, 6 May 2022 04:23:51 +0000 (06:23 +0200)]
net: phy: genphy_c45_baset1_an_config_aneg: do no set unknown configuration

Do not change default master/slave autoneg configuration if no
changes was requested.

Fixes: 3da8ffd8545f ("net: phy: Add 10BASE-T1L support in phy-c45")
Signed-off-by: Oleksij Rempel <o.rempel@pengutronix.de>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoMerge branch 'vxlan_fdb_delete-extack'
David S. Miller [Mon, 9 May 2022 10:58:20 +0000 (11:58 +0100)]
Merge branch 'vxlan_fdb_delete-extack'

Alaa Mohamed says:

====================
propagate extack to vxlan_fdb_delete

In order to propagate extack to vxlan_fdb_delete and vxlan_fdb_parse,
add extack to .ndo_fdb_del and edit all fdb del handelers.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: vxlan: Add extack support to vxlan_fdb_delete
Alaa Mohamed [Thu, 5 May 2022 15:09:58 +0000 (17:09 +0200)]
net: vxlan: Add extack support to vxlan_fdb_delete

This patch adds extack msg support to vxlan_fdb_delete and vxlan_fdb_parse.
extack is used to propagate meaningful error msgs to the user of vxlan
fdb netlink api

Signed-off-by: Alaa Mohamed <eng.alaamohamedsoliman.am@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agortnetlink: add extack support in fdb del handlers
Alaa Mohamed [Thu, 5 May 2022 15:09:57 +0000 (17:09 +0200)]
rtnetlink: add extack support in fdb del handlers

Add extack support to .ndo_fdb_del in netdevice.h and
all related methods.

Signed-off-by: Alaa Mohamed <eng.alaamohamedsoliman.am@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoMerge branch 'wwan-t7xx'
David S. Miller [Mon, 9 May 2022 09:51:59 +0000 (10:51 +0100)]
Merge branch 'wwan-t7xx'

Ricardo Martinez says:

====================
net: wwan: t7xx: PCIe driver for MediaTek M.2 modem

t7xx is the PCIe host device driver for Intel 5G 5000 M.2 solution which
is based on MediaTek's T700 modem to provide WWAN connectivity.
The driver uses the WWAN framework infrastructure to create the following
control ports and network interfaces:
* /dev/wwan0mbim0 - Interface conforming to the MBIM protocol.
  Applications like libmbim [1] or Modem Manager [2] from v1.16 onwards
  with [3][4] can use it to enable data communication towards WWAN.
* /dev/wwan0at0 - Interface that supports AT commands.
* wwan0 - Primary network interface for IP traffic.

The main blocks in t7xx driver are:
* PCIe layer - Implements probe, removal, and power management callbacks.
* Port-proxy - Provides a common interface to interact with different types
  of ports such as WWAN ports.
* Modem control & status monitor - Implements the entry point for modem
  initialization, reset and exit, as well as exception handling.
* CLDMA (Control Layer DMA) - Manages the HW used by the port layer to send
  control messages to the modem using MediaTek's CCCI (Cross-Core
  Communication Interface) protocol.
* DPMAIF (Data Plane Modem AP Interface) - Controls the HW that provides
  uplink and downlink queues for the data path. The data exchange takes
  place using circular buffers to share data buffer addresses and metadata
  to describe the packets.
* MHCCIF (Modem Host Cross-Core Interface) - Provides interrupt channels
  for bidirectional event notification such as handshake, exception, PM and
  port enumeration.

The compilation of the t7xx driver is enabled by the CONFIG_MTK_T7XX config
option which depends on CONFIG_WWAN.
This driver was originally developed by MediaTek. Intel adapted t7xx to
the WWAN framework, optimized and refactored the driver source code in close
collaboration with MediaTek. This will enable getting the t7xx driver on the
Approved Vendor List for interested OEM's and ODM's productization plans
with Intel 5G 5000 M.2 solution.

List of contributors:
Amir Hanania <amir.hanania@intel.com>
Andriy Shevchenko <andriy.shevchenko@linux.intel.com>
Chandrashekar Devegowda <chandrashekar.devegowda@intel.com>
Dinesh Sharma <dinesh.sharma@intel.com>
Eliot Lee <eliot.lee@intel.com>
Haijun Liu <haijun.liu@mediatek.com>
M Chetan Kumar <m.chetan.kumar@intel.com>
Mika Westerberg <mika.westerberg@linux.intel.com>
Moises Veleta <moises.veleta@intel.com>
Pierre-louis Bossart <pierre-louis.bossart@intel.com>
Chiranjeevi Rapolu <chiranjeevi.rapolu@intel.com>
Ricardo Martinez <ricardo.martinez@linux.intel.com>
Madhusmita Sahu <madhusmita.sahu@intel.com>
Muralidharan Sethuraman <muralidharan.sethuraman@intel.com>
Soumya Prakash Mishra <Soumya.Prakash.Mishra@intel.com>
Sreehari Kancharla <sreehari.kancharla@intel.com>
Suresh Nagaraj <suresh.nagaraj@intel.com>

[1] https://www.freedesktop.org/software/libmbim/
[2] https://www.freedesktop.org/software/ModemManager/
[3] https://gitlab.freedesktop.org/mobile-broadband/ModemManager/-/merge_requests/582
[4] https://gitlab.freedesktop.org/mobile-broadband/ModemManager/-/merge_requests/523

V8:
- Rebase skb_data_area_size() patch (02).

V7:
- Delete unused macros.
- Avoid duplicated calls to le32_to_cpu().
- Fix 'out of bounds' compilation error.
- Rename port_number to port_count.
- Remove '!!' when the destination variable is a boolean.
- Remove unneeded spinlock around rx_length_th.
- Remove common field from union inside dpmaif_drb struct.
- Use 'goto' for the exit flow in t7xx_pci_enable_sleep().
- Merge CLDMA tgpd and rgpd structs.
- Add comments to clarify skb consumption by ports.
- Introduce skb_data_area_size() helper.
- Declare the port config array as constant.
- Update CLDMA_JUMBO_BUFF_SZ definition when ccci_header
  is introduced by port-proxy patch.
- Update Reviewed-by tags.
- Simplify t7xx_dpmaif_tx_send_skb() and make t7xx_dpmaif_add_skb_to_ring()
  report the tx queue full state early.

V6:
- Remove unneeded initializations and bit masks.
- Remove t7xx_common.h file.
- Add comment to circular linking in GPD list.
- Use min instead of min_t.
- Use int for local indexes instead of short or char.
- Update the commit message in CLDMA patch about dependencies on core patch.
- Add space between contributor name and email address.
- Rename registers with double negatives
  e.g. DIS_ASPM_LOWPWR_CLR_0 -> ENABLE_ASPM_LOWPWR.
- Fix a race condition in pci sleep resource locking.
- Initialize interrupts with t7xx_pcie_mac_set_int() instead of 'clear'.
- Remove duplicate spin_lock_init(&md->exp_lock).
- Remove .ndo_select_queue callback due to singular TX queue.
- Remove call to deprecated netif_rx_any_context().
- Fix include guard name in t7xx_hif_dpmaif.h.
- Remove unused q_num parameter in DPMAIF functions.
- Do not serialize the drb_wr_idx write in t7xx_dpmaif_add_skb_to_ring()
  and the read from t7xx_txq_drb_wr_available().
- Fix potential leak in t7xx_dpmaif_add_skb_to_ring().
- Unionize:
    DRB structs: msg and pd.
    PIT structs: msg and pd.
- Replace list_head & spinlock with skb_buff_head in dpmaif_tx_queue.
- Remove rx_length_th check in TX WWAN port flow.
- Remove wwan_remove_port() from the critical section in WWAN port uninit.
- Use skb_end_pointer() to avoid conditional compilation.
- Simplify the loop in t7xx_port_ctrl_tx() by checking the buffer offset
  instead of calculating the number of required packets.
- Remove the code for unused channel PORT_CH_STATUS_RX.
- Remove bit flags from ports. Ports can check chan_enable instead of the
  PORT_F_RX_ALLOW_DROP flag.
- Use INVALID_SEQ_NUM to identify the first seq number.
- Rename port_static to port_conf and ports_private to ports.
- Implement t7xx_port_send_skb() and t7xx_port_send_ctl_skb() in a layered
  approach to reduce duplicated code and simplify the CCCI header handling.
- Move wwan_port_rx() call from port-proxy to WWAN port.
- Rename t7xx_port_recv_skb() to t7xx_port_enqueue_skb().
- Move control message parsing logic from port-proxy to control port,
  preserve the endianness when parsing the message and make port-proxy
  export a function to enable/disable ports.
- Use flexible arrays for:
    port-proxy ports.
    payload data in t7xx_fsm_event, port_msg, and mtk_runtime_feature.

v5:
- Update Intel's copyright years to 2021-2022.
- Remove circular dependency between DPMAIF HW (07) and HIF (08).
- Keep separate patches for CLDMA (02) and Core (03)
  but improve the code split by decoupling CLDMA from
  modem ops and cleaning up t7xx_common.h.
- Rename ID_CLDMA0/ID_CLDMA1 to CLDMA_ID_AP/CLDMA_ID_MD.
- Consistently use CLDMA's ring_lock to protect tr_ring.
- Free resources first and then print messages.
- Implement suggested name changes.
- Do not explicitly include dev_printk.h.
- Remove redundant dev_err()s.
- Fix possible memory leak during probe.
- Remove infrastructure for legacy interrupts.
- Remove unused macros and variables, including those that
  can be replaced with constants.
- Remove PCIE_MAC_MSIX_MSK_SET macro which is duplicated code.
- Refactor __t7xx_pci_pm_suspend() for clarity.
- Refactor t7xx_cldma_rx_ring_init() and t7xx_cldma_tx_ring_init().
- Do not use & for function callbacks.
- Declare a structure to access skb->cb[] data.
- Use skb_put_data instead of memcpy.
- No need to use kfree_sensitive.
- Use dev_kfree_skb() instead of kfree_skb().
- Refactor t7xx_prepare_device_rt_data() to remove potential leaks,
  avoid unneeded memset and keep rt_data and packet_size updates
  inside the same 'if' block.
- Set port's rx_length_th back to 0 during uninit.
- Remove unneeded 'blocking' parameter from t7xx_cldma_send_skb().
- Return -EIO in t7xx_cldma_send_skb() if the queue is inactive.
- Refactor t7xx_cldma_qs_are_active() to use pci_device_is_present().
- Simplify t7xx_cldma_stop_q() and rename it to t7xx_cldma_stop_all_qs().
- Fix potential leaks in t7xx_cldma_init().
- Improve error handling in fsm_append_event and fsm_routine_starting().
- Propagate return codes from fsm_append_cmd() and t7xx_fsm_append_event().
- Refactor fsm_wait_for_event() to avoid unnecessary sleep.
- Create the WWAN ports and net device only after the modem is in
  the ready state.
- Refactor t7xx_port_proxy_recv_skb() and port_recv_skb().
- Rename t7xx_port_check_rx_seq_num() as t7xx_port_next_rx_seq_num()
  and fix the seq_num logic to handle overflows.
- Declare seq_nums as u16 instead of short.
- Use unsigned int for local indexes.
- Use min_t instead of the ternary operator.
- Refactor the loop in t7xx_dpmaif_rx_data_collect() to avoid a dead
  condition check.
- Use a bitmap (bat_bitmap) instead of an array to keep track of
  the DRB status. Used in t7xx_dpmaif_avail_pkt_bat_cnt().
- Refactor t7xx_dpmaif_tx_send_skb() to protect tx_submit_skb_cnt
  with spinlock and remove the misleading tx_drb_available variable.
- Consolidate bit operations before endianness conversion.
- Use C bit fields in dpmaif_drb_skb struct which is not HW related.
- Add back the que_started check in t7xx_select_tx_queue().
- Create a helper function to get the DRB count.
- Simplify the use of 'usage' during t7xx_ccmni_close().
- Enforce CCMNI MTU selection with BUILD_BUG_ON() instead of a comment.
- Remove t7xx_ccmni_ctrl->capability parameter which remains constant.

v4:
- Implement list_prev_entry_circular() and list_next_entry_circular() macros.
- Remove inline from all c files.
- Define ioread32_poll_timeout_atomic() helper macro.
- Fix return code for WWAN port tx op.
- Allow AT commands fragmentation same as MBIM commands.
- Introduce t7xx_common.h file in the first patch.
- Rename functions and variables as suggested in v3.
- Reduce code duplication by creating fsm_wait_for_event() helper function.
- Remove unneeded dev_err in t7xx_fsm_clr_event().
- Remove unused variable last_state from struct t7xx_fsm_ctl.
- Remove unused variable txq_select_times from struct dpmaif_ctrl.
- Replace ETXTBSY with EBUSY.
- Refactor t7xx_dpmaif_rx_buf_alloc() to remove an unneeded allocation.
- Fix potential leak at t7xx_dpmaif_rx_frag_alloc().
- Simplify return value handling at t7xx_dpmaif_rx_start().
- Add a helper to handle the common part of CCCI header initialization.
- Make sure interrupts are enabled during PM resume.
- Add a parameter to t7xx_fsm_append_cmd() to tell if it is in interrupt context.

v3:
- Avoid unneeded ping-pong changes between patches.
- Use t7xx_ prefix in functions.
- Use t7xx_ prefix in generic structs where mtk_ or ccci prefix was used.
- Update Authors/Contributors header.
- Remove skb pools used for control path.
- Remove skb pools used for RX data path.
- Do not use dedicated TX queue for ACK-only packets.
- Remove __packed attribute from GPD structs.
- Remove the infrastructure for test and debug ports.
- Use the skb control buffer to store metadata.
- Get the IP packet type from RX PIT.
- Merge variable declaration and simple assignments.
- Use preferred coding patterns.
- Remove global variables.
- Declare HW facing structure members as little endian.
- Rename goto tags to describe what is going to be done.
- Do not use variable length arrays.
- Remove unneeded blank lines source code and kdoc headers.
- Use C99 initialization format for port-proxy ports.
- Clean up comments.
- Review included headers.
- Better use of 100 column limit.
- Remove unneeded mb() in CLDMA.
- Remove unneeded spin locks and atomics.
- Handle read_poll_timeout error.
- Use dev_err_ratelimited() where required.
- Fix resource leak when requesting IRQs.
- Use generic DEFAULT_TX_QUEUE_LEN instead custom macro.
- Use ETH_DATA_LEN instead of defining WWAN_DEFAULT_MTU.
- Use sizeof() instead of defines when the size of structures is required.
- Remove unneeded code from netdev:
    No need to configure HW address length
    No need to implement .ndo_change_mtu
    Remove random address generation
- Code simplifications by using kernel provided functions and macros such as:
    module_pci_driver
    PTR_ERR_OR_ZERO
    for_each_set_bit
    pci_device_is_present
    skb_queue_purge
    list_prev_entry
    __ffs64

v2:
- Replace pdev->driver->name with dev_driver_string(&pdev->dev).
- Replace random_ether_addr() with eth_random_addr().
- Update kernel-doc comment for enum data_policy.
- Indicate the driver is 'Supported' instead of 'Maintained'.
- Fix the Signed-of-by and Co-developed-by tags in the patches.
- Added authors and contributors in the top comment of the src files.
====================

2 years agonet: wwan: t7xx: Add maintainers and documentation
Ricardo Martinez [Fri, 6 May 2022 18:13:10 +0000 (11:13 -0700)]
net: wwan: t7xx: Add maintainers and documentation

Adds maintainers and documentation for MediaTek t7xx 5G WWAN modem
device driver.

Signed-off-by: Ricardo Martinez <ricardo.martinez@linux.intel.com>
Reviewed-by: Loic Poulain <loic.poulain@linaro.org>
Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
Reviewed-by: Sergey Ryazanov <ryazanov.s.a@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: wwan: t7xx: Device deep sleep lock/unlock
Haijun Liu [Fri, 6 May 2022 18:13:09 +0000 (11:13 -0700)]
net: wwan: t7xx: Device deep sleep lock/unlock

Introduce the mechanism to lock/unlock the device 'deep sleep' mode.
When the PCIe link state is L1.2 or L2, the host side still can keep
the device is in D0 state from the host side point of view. At the same
time, if the device's 'deep sleep' mode is unlocked, the device will
go to 'deep sleep' while it is still in D0 state on the host side.

Signed-off-by: Haijun Liu <haijun.liu@mediatek.com>
Signed-off-by: Chandrashekar Devegowda <chandrashekar.devegowda@intel.com>
Co-developed-by: Ricardo Martinez <ricardo.martinez@linux.intel.com>
Signed-off-by: Ricardo Martinez <ricardo.martinez@linux.intel.com>
Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: wwan: t7xx: Runtime PM
Haijun Liu [Fri, 6 May 2022 18:13:08 +0000 (11:13 -0700)]
net: wwan: t7xx: Runtime PM

Enables runtime power management callbacks including runtime_suspend
and runtime_resume. Autosuspend is used to prevent overhead by frequent
wake-ups.

Signed-off-by: Haijun Liu <haijun.liu@mediatek.com>
Signed-off-by: Chandrashekar Devegowda <chandrashekar.devegowda@intel.com>
Co-developed-by: Eliot Lee <eliot.lee@intel.com>
Signed-off-by: Eliot Lee <eliot.lee@intel.com>
Signed-off-by: Ricardo Martinez <ricardo.martinez@linux.intel.com>
Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: wwan: t7xx: Introduce power management
Haijun Liu [Fri, 6 May 2022 18:13:07 +0000 (11:13 -0700)]
net: wwan: t7xx: Introduce power management

Implements suspend, resumes, freeze, thaw, poweroff, and restore
`dev_pm_ops` callbacks.

From the host point of view, the t7xx driver is one entity. But, the
device has several modules that need to be addressed in different ways
during power management (PM) flows.
The driver uses the term 'PM entities' to refer to the 2 DPMA and
2 CLDMA HW blocks that need to be managed during PM flows.
When a dev_pm_ops function is called, the PM entities list is iterated
and the matching function is called for each entry in the list.

Signed-off-by: Haijun Liu <haijun.liu@mediatek.com>
Signed-off-by: Chandrashekar Devegowda <chandrashekar.devegowda@intel.com>
Co-developed-by: Ricardo Martinez <ricardo.martinez@linux.intel.com>
Signed-off-by: Ricardo Martinez <ricardo.martinez@linux.intel.com>
Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: wwan: t7xx: Add WWAN network interface
Haijun Liu [Fri, 6 May 2022 18:13:06 +0000 (11:13 -0700)]
net: wwan: t7xx: Add WWAN network interface

Creates the Cross Core Modem Network Interface (CCMNI) which implements
the wwan_ops for registration with the WWAN framework, CCMNI also
implements the net_device_ops functions used by the network device.
Network device operations include open, close, start transmission, TX
timeout and change MTU.

Signed-off-by: Haijun Liu <haijun.liu@mediatek.com>
Co-developed-by: Chandrashekar Devegowda <chandrashekar.devegowda@intel.com>
Signed-off-by: Chandrashekar Devegowda <chandrashekar.devegowda@intel.com>
Co-developed-by: Ricardo Martinez <ricardo.martinez@linux.intel.com>
Signed-off-by: Ricardo Martinez <ricardo.martinez@linux.intel.com>
Reviewed-by: Loic Poulain <loic.poulain@linaro.org>
Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
Reviewed-by: Sergey Ryazanov <ryazanov.s.a@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: wwan: t7xx: Add data path interface
Haijun Liu [Fri, 6 May 2022 18:13:05 +0000 (11:13 -0700)]
net: wwan: t7xx: Add data path interface

Data Path Modem AP Interface (DPMAIF) HIF layer provides methods
for initialization, ISR, control and event handling of TX/RX flows.

DPMAIF TX
Exposes the 'dmpaif_tx_send_skb' function which can be used by the
network device to transmit packets.
The uplink data management uses a Descriptor Ring Buffer (DRB).
First DRB entry is a message type that will be followed by 1 or more
normal DRB entries. Message type DRB will hold the skb information
and each normal DRB entry holds a pointer to the skb payload.

DPMAIF RX
The downlink buffer management uses Buffer Address Table (BAT) and
Packet Information Table (PIT) rings.
The BAT ring holds the address of skb data buffer for the HW to use,
while the PIT contains metadata about a whole network packet including
a reference to the BAT entry holding the data buffer address.
The driver reads the PIT and BAT entries written by the modem, when
reaching a threshold, the driver will reload the PIT and BAT rings.

Signed-off-by: Haijun Liu <haijun.liu@mediatek.com>
Signed-off-by: Chandrashekar Devegowda <chandrashekar.devegowda@intel.com>
Co-developed-by: Ricardo Martinez <ricardo.martinez@linux.intel.com>
Signed-off-by: Ricardo Martinez <ricardo.martinez@linux.intel.com>
Reviewed-by: Loic Poulain <loic.poulain@linaro.org>
Reviewed-by: Sergey Ryazanov <ryazanov.s.a@gmail.com>
Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: wwan: t7xx: Data path HW layer
Haijun Liu [Fri, 6 May 2022 18:13:04 +0000 (11:13 -0700)]
net: wwan: t7xx: Data path HW layer

Data Path Modem AP Interface (DPMAIF) HW layer provides HW abstraction
for the upper layer (DPMAIF HIF). It implements functions to do the HW
configuration, TX/RX control and interrupt handling.

Signed-off-by: Haijun Liu <haijun.liu@mediatek.com>
Signed-off-by: Chandrashekar Devegowda <chandrashekar.devegowda@intel.com>
Co-developed-by: Ricardo Martinez <ricardo.martinez@linux.intel.com>
Signed-off-by: Ricardo Martinez <ricardo.martinez@linux.intel.com>
Reviewed-by: Loic Poulain <loic.poulain@linaro.org>
Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
Reviewed-by: Sergey Ryazanov <ryazanov.s.a@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: wwan: t7xx: Add AT and MBIM WWAN ports
Chandrashekar Devegowda [Fri, 6 May 2022 18:13:03 +0000 (11:13 -0700)]
net: wwan: t7xx: Add AT and MBIM WWAN ports

Adds AT and MBIM ports to the port proxy infrastructure.
The initialization method is responsible for creating the corresponding
ports using the WWAN framework infrastructure. The implemented WWAN port
operations are start, stop, and TX.

Signed-off-by: Chandrashekar Devegowda <chandrashekar.devegowda@intel.com>
Co-developed-by: Ricardo Martinez <ricardo.martinez@linux.intel.com>
Signed-off-by: Ricardo Martinez <ricardo.martinez@linux.intel.com>
Reviewed-by: Loic Poulain <loic.poulain@linaro.org>
Reviewed-by: Sergey Ryazanov <ryazanov.s.a@gmail.com>
Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: wwan: t7xx: Add control port
Haijun Liu [Fri, 6 May 2022 18:13:02 +0000 (11:13 -0700)]
net: wwan: t7xx: Add control port

Control Port implements driver control messages such as modem-host
handshaking, controls port enumeration, and handles exception messages.

The handshaking process between the driver and the modem happens during
the init sequence. The process involves the exchange of a list of
supported runtime features to make sure that modem and host are ready
to provide proper feature lists including port enumeration. Further
features can be enabled and controlled in this handshaking process.

Signed-off-by: Haijun Liu <haijun.liu@mediatek.com>
Signed-off-by: Chandrashekar Devegowda <chandrashekar.devegowda@intel.com>
Co-developed-by: Ricardo Martinez <ricardo.martinez@linux.intel.com>
Signed-off-by: Ricardo Martinez <ricardo.martinez@linux.intel.com>
Reviewed-by: Loic Poulain <loic.poulain@linaro.org>
Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
Reviewed-by: Sergey Ryazanov <ryazanov.s.a@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: wwan: t7xx: Add port proxy infrastructure
Haijun Liu [Fri, 6 May 2022 18:13:01 +0000 (11:13 -0700)]
net: wwan: t7xx: Add port proxy infrastructure

Port-proxy provides a common interface to interact with different types
of ports. Ports export their configuration via `struct t7xx_port` and
operate as defined by `struct port_ops`.

Signed-off-by: Haijun Liu <haijun.liu@mediatek.com>
Co-developed-by: Chandrashekar Devegowda <chandrashekar.devegowda@intel.com>
Signed-off-by: Chandrashekar Devegowda <chandrashekar.devegowda@intel.com>
Co-developed-by: Ricardo Martinez <ricardo.martinez@linux.intel.com>
Signed-off-by: Ricardo Martinez <ricardo.martinez@linux.intel.com>
Reviewed-by: Loic Poulain <loic.poulain@linaro.org>
Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
Reviewed-by: Sergey Ryazanov <ryazanov.s.a@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: wwan: t7xx: Add core components
Haijun Liu [Fri, 6 May 2022 18:13:00 +0000 (11:13 -0700)]
net: wwan: t7xx: Add core components

Registers the t7xx device driver with the kernel. Setup all the core
components: PCIe layer, Modem Host Cross Core Interface (MHCCIF),
modem control operations, modem state machine, and build
infrastructure.

* PCIe layer code implements driver probe and removal.
* MHCCIF provides interrupt channels to communicate events
  such as handshake, PM and port enumeration.
* Modem control implements the entry point for modem init,
  reset and exit.
* The modem status monitor is a state machine used by modem control
  to complete initialization and stop. It is used also to propagate
  exception events reported by other components.

Signed-off-by: Haijun Liu <haijun.liu@mediatek.com>
Signed-off-by: Chandrashekar Devegowda <chandrashekar.devegowda@intel.com>
Co-developed-by: Ricardo Martinez <ricardo.martinez@linux.intel.com>
Signed-off-by: Ricardo Martinez <ricardo.martinez@linux.intel.com>
Reviewed-by: Loic Poulain <loic.poulain@linaro.org>
Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
Reviewed-by: Sergey Ryazanov <ryazanov.s.a@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: wwan: t7xx: Add control DMA interface
Haijun Liu [Fri, 6 May 2022 18:12:59 +0000 (11:12 -0700)]
net: wwan: t7xx: Add control DMA interface

Cross Layer DMA (CLDMA) Hardware interface (HIF) enables the control
path of Host-Modem data transfers. CLDMA HIF layer provides a common
interface to the Port Layer.

CLDMA manages 8 independent RX/TX physical channels with data flow
control in HW queues. CLDMA uses ring buffers of General Packet
Descriptors (GPD) for TX/RX. GPDs can represent multiple or single
data buffers (DB).

CLDMA HIF initializes GPD rings, registers ISR handlers for CLDMA
interrupts, and initializes CLDMA HW registers.

CLDMA TX flow:
1. Port Layer write
2. Get DB address
3. Configure GPD
4. Triggering processing via HW register write

CLDMA RX flow:
1. CLDMA HW sends a RX "done" to host
2. Driver starts thread to safely read GPD
3. DB is sent to Port layer
4. Create a new buffer for GPD ring

Note: This patch does not enable compilation since it has dependencies
such as t7xx_pcie_mac_clear_int()/t7xx_pcie_mac_set_int() and
struct t7xx_pci_dev which are added by the core patch.

Signed-off-by: Haijun Liu <haijun.liu@mediatek.com>
Signed-off-by: Chandrashekar Devegowda <chandrashekar.devegowda@intel.com>
Co-developed-by: Ricardo Martinez <ricardo.martinez@linux.intel.com>
Signed-off-by: Ricardo Martinez <ricardo.martinez@linux.intel.com>
Reviewed-by: Loic Poulain <loic.poulain@linaro.org>
Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
Reviewed-by: Sergey Ryazanov <ryazanov.s.a@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: skb: introduce skb_data_area_size()
Ricardo Martinez [Fri, 6 May 2022 18:12:58 +0000 (11:12 -0700)]
net: skb: introduce skb_data_area_size()

Helper to calculate the linear data space in the skb.

Signed-off-by: Ricardo Martinez <ricardo.martinez@linux.intel.com>
Reviewed-by: Sergey Ryazanov <ryazanov.s.a@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agolist: Add list_next_entry_circular() and list_prev_entry_circular()
Ricardo Martinez [Fri, 6 May 2022 18:12:57 +0000 (11:12 -0700)]
list: Add list_next_entry_circular() and list_prev_entry_circular()

Add macros to get the next or previous entries and wraparound if
needed. For example, calling list_next_entry_circular() on the last
element should return the first element in the list.

Signed-off-by: Ricardo Martinez <ricardo.martinez@linux.intel.com>
Reviewed-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoMerge tag 'batadv-next-pullrequest-20220508' of git://git.open-mesh.org/linux-merge
David S. Miller [Sun, 8 May 2022 16:10:59 +0000 (17:10 +0100)]
Merge tag 'batadv-next-pullrequest-20220508' of git://git.open-mesh.org/linux-merge

This cleanup patchset includes the following patches:

 - bump version strings, by Simon Wunderlich

 - remove unnecessary type castings, by Yu Zhe

Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoMerge branch 'mlxsw-dedicated-router-notification-block'
David S. Miller [Sun, 8 May 2022 10:46:21 +0000 (11:46 +0100)]
Merge branch 'mlxsw-dedicated-router-notification-block'

Ido Schimmel says:

====================
mlxsw: A dedicated notifier block for router code

Petr says:

Currently all netdevice events are handled in the centralized notifier
handler maintained by spectrum.c. Since a number of events are involving
router code, spectrum.c needs to dispatch them to spectrum_router.c. The
spectrum module therefore needs to know more about the router code than it
should have, and there is are several API points through which the two
modules communicate.

In this patchset, move bulk of the router-related event handling to the
router code. Some of the knowledge has to stay: spectrum.c cannot veto
events that the router supports, and vice versa. But beyond that, the two
can ignore each other's details, which leads to more focused and simpler
code.

As a side effect, this fixes L3 HW stats support on tunnel netdevices.

The patch set progresses as follows:

- In patch #1, change spectrum code to not bounce L3 enslavement, which the
  router code supports.

- In patch #2, add a new do-nothing notifier block to the router code.

- In patches #3-#6, move router-specific event handling to the router
  module. In patch #7, clean up a comment.

- In patch #8, use the advantage that all router event handling is in the
  router code and clean up taking router lock.

- mlxsw supports L3 HW stats on tunnels as of this patchset. Patches #9 and
  #10 therefore add a selftest for L3 HW stats support on tunnels.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoselftests: forwarding: Add a tunnel-based test for L3 HW stats
Petr Machata [Sun, 8 May 2022 08:08:23 +0000 (11:08 +0300)]
selftests: forwarding: Add a tunnel-based test for L3 HW stats

Add a selftest that uses an IPIP topology and tests that L3 HW stats
reflect the traffic in the tunnel.

Signed-off-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: Ido Schimmel <idosch@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoselftests: lib: Add a generic helper for obtaining HW stats
Petr Machata [Sun, 8 May 2022 08:08:22 +0000 (11:08 +0300)]
selftests: lib: Add a generic helper for obtaining HW stats

The function get_l3_stats() from the test hw_stats_l3.sh will be useful for
any test that wishes to work with L3 stats. Furthermore, it is easy to
generalize to other HW stats suites (for when such are added). Therefore,
move the code to lib.sh, rewrite it to have the same interface as the other
stats-collecting functions, and generalize to take the name of the HW stats
suite to collect as an argument.

Signed-off-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: Ido Schimmel <idosch@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agomlxsw: spectrum_router: Take router lock in router notifier handler
Petr Machata [Sun, 8 May 2022 08:08:21 +0000 (11:08 +0300)]
mlxsw: spectrum_router: Take router lock in router notifier handler

For notifications that the router needs to handle, router lock is taken.
Further, at least to determine whether an event is related to a tunnel
underlay, router lock also needs to be taken. Due to this, the router lock
is always taken for each unhandled event, and also for some handled events,
even if they are not related to underlay. Thus each event implies at least
one router lock, sometimes two.

Instead of deferring the locking to the leaf handlers, take the lock in the
router notifier handler always. This simplifies thinking about the locking
state, and in some cases saves one lock cycle.

Signed-off-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: Ido Schimmel <idosch@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agomlxsw: spectrum: Update a comment
Petr Machata [Sun, 8 May 2022 08:08:20 +0000 (11:08 +0300)]
mlxsw: spectrum: Update a comment

The position of netdevice notifier registration no longer depends on the
router initialization, because the event handler no longer dispatches to
the router code. Update the comment at the registration to that effect.

Signed-off-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: Ido Schimmel <idosch@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agomlxsw: spectrum: Move handling of tunnel events to router code
Petr Machata [Sun, 8 May 2022 08:08:19 +0000 (11:08 +0300)]
mlxsw: spectrum: Move handling of tunnel events to router code

The events related to IPIP tunnels are handled by the router code. Move the
handling from the central dispatcher in spectrum.c to the new notifier
handler in the router module.

Signed-off-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: Ido Schimmel <idosch@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>