Claudiu Manoil says:
====================
enetc: Add adaptive interrupt coalescing
Apart from some related cleanup patches, this set
introduces in a straightforward way the support needed
to enable and configure interrupt coalescing for ENETC.
Patch 5 introduces the support needed for configuring the
interrupt coalescing parameters and for switching between
moderated (int. coalescing) and per-packet interrupt modes.
When interrupt coalescing is enabled the Rx/Tx time
thresholds are configurable, packet thresholds are fixed.
To make this work reliably, patch 5 uses the traffic
pause procedure introduced in patch 2.
Patch 6 adds DIM (Dynamic Interrupt Moderation) to implement
adaptive coalescing based on time thresholds, for the Rx 'channel'.
On the Tx side a default optimal value is used instead, optimized for
TCP traffic over 1G and 2.5G links. This default 'optimal' value can
be overridden anytime via 'ethtool -C tx-usecs'.
netperf -t TCP_MAERTS measurements show a significant CPU load
reduction correlated w/ reduced interrupt rates. For the
measurement results refer to the comments in patch 6.
v2: Replaced Tx DIM with predefined optimal value, giving
better results. This was also suggested by Jakub (cc).
Switched order of patches 4 and 5, for better grouping.
v3: minor cleanup/improvements
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Use the generic dynamic interrupt moderation (dim)
framework to implement adaptive interrupt coalescing
on Rx. With the per-packet interrupt scheme, a high
interrupt rate has been noted for moderate traffic flows
leading to high CPU utilization. The 'dim' scheme
implemented by the current patch addresses this issue
improving CPU utilization while using minimal coalescing
time thresholds in order to preserve a good latency.
On the Tx side use an optimal time threshold value by
default. This value has been optimized for Tx TCP
streams at a rate of around 85kpps on a 1G link,
at which rate half of the Tx ring size (128) gets filled
in 1500 usecs. Scaling this down to 2.5G links yields
the current value of 600 usecs, which is conservative
and gives good enough results for 1G links too (see
next).
Below are some measurement results for before and after
this patch (and related dependencies) basically, for a
2 ARM Cortex-A72 @1.3Ghz CPUs system (32 KB L1 data cache),
using 60secs log netperf TCP stream tests @ 1Gbit link
(maximum throughput):
1) 1 Rx TCP flow, both Rx and Tx processed by the same NAPI
thread on the same CPU:
CPU utilization int rate (ints/sec)
Before: 50%-60% (over 50%) 92k
After: 13%-22% 3.5k-12k
Comment: Major CPU utilization improvement for a single flow
Rx TCP flow (i.e. netperf -t TCP_MAERTS) on a single
CPU. Usually settles under 16% for longer tests.
2) 4 Rx TCP flows + 4 Tx TCP flows (+ pings to check the latency):
Total CPU utilization Total int rate (ints/sec)
Before: ~80% (spikes to 90%) ~100k
After: 60% (more steady) ~4k
Comment: Important improvement for this load test, while the
ping test outcome does not show any notable
difference compared to before.
Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Enable programming of the interrupt coalescing registers
and allow manual configuration of the coalescing time
thresholds via ethtool. Packet thresholds have been fixed
to predetermined values as there's no point in making them
run-time configurable, also anticipating the dynamic interrupt
moderation (DIM) algorithm which uses fixed packet thresholds
as well. If the interface is up when the operation mode of
traffic interrupt events is changed by the user (i.e. switching
from default per-packet interrupts to coalesced interrupts),
the traffic needs to be paused in the process.
This patch also prepares the ground for introducing DIM on Rx.
Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
'struct enetc_bdr' is already '____cacheline_aligned_in_smp'.
Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Interrupt coalescing registers naming in the current revision
of the Ref Man (RM) is ICR, deprecating the ICIR name used
in earlier (draft) versions of the RM.
Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
A reliable traffic pause (and reconfiguration) procedure
is needed to be able to safely make h/w configuration
changes during run-time, like changing the mode in which the
interrupts are operating (i.e. with or without coalescing),
as opposed to making on-the-fly register updates that
may be subject to h/w or s/w concurrency issues.
To this end, the code responsible of the run-time device
configurations that basically starts resp. stops the traffic
flow through the device has been extracted from the
the enetc_open/_close procedures, to the separate standalone
enetc_start/_stop procedures. Traffic stop should be as
graceful as possible, it lets the executing napi threads to
to finish while the interrupts stay disabled. But since
the napi thread will try to re-enable interrupts by clearing
the device's unmask register, the enable_irq/ disable_irq
API has been used to avoid this potential concurrency issue
and make the traffic pause procedure more reliable.
Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
It's time to differentiate between Rx and Tx ring sizes.
Not only Tx rings are processed differently than Rx rings,
but their default number also differs - i.e. up to 8 Tx rings
per device (8 traffic classes) vs. 2 Rx rings (one per CPU).
So let's set Tx rings sizes to half the size of the Rx rings
for now, to be conservative.
The default ring sizes were decreased as well (to the next
lower power of 2), to reduce the memory footprint, buffering
etc., since the measurements I've made so far show that the
rings are very unlikely to get full.
This change also anticipates the introduction of the
dynamic interrupt moderation (dim) algorithm which operates
on maximum packet thresholds of 256 packets for Rx and 128
packets for Tx.
Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Use devm_gpiod_get_array() to simplify the error handling and exit
code path.
Signed-off-by: Jisheng Zhang <Jisheng.Zhang@synaptics.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Now that DSA supports MTU configuration, undo the effects of commit
8b1efc0f83 ("net: remove MTU limits on a few ether_setup callers") and
let DSA interfaces use the default min_mtu and max_mtu specified by
ether_setup(). This is more important for min_mtu: since DSA is
Ethernet, the minimum MTU is the same as of any other Ethernet
interface, and definitely not zero. For the max_mtu, we have a callback
through which drivers can override that, if they want to.
Signed-off-by: Vladimir Oltean <olteanv@gmail.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This switch has a single max frame size configuration register, so we
track the requested MTU for each port and apply the largest.
v2:
- Address review feedback from Vladimir Oltean
Signed-off-by: Jonathan McDowell <noodles@earth.li>
Acked-by: Vladimir Oltean <olteanv@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Because kfree_skb already checked NULL skb parameter,
so the additional checks are unnecessary, just remove them.
Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: Wang Hai <wanghai38@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add tdc to existing kselftest infrastructure so that it can be run with
existing kselftests. TDC now generates objects in objdir/kselftest
without cluttering main objdir, leaves source directory clean, and
installs correctly in kselftest_install, properly adding itself to
run_kselftest.sh script.
Add tc-testing as a target of selftests/Makefile. Create tdc.sh to run
tdc.py targets with correct arguments. To support single target from
selftest/Makefile, combine tc-testing/bpf/Makefile and
tc-testing/Makefile. Move action.c up a directory to tc-testing/.
Tested with:
make O=/tmp/{objdir} TARGETS="tc-testing" kselftest
cd /tmp/{objdir}
cd kselftest
cd tc-testing
./tdc.sh
make -C tools/testing/selftests/ TARGETS=tc-testing run_tests
make TARGETS="tc-testing" kselftest
cd tools/testing/selftests
./kselftest_install.sh /tmp/exampledir
My VM doesn't run all the kselftests so I commented out all except my
target and net/pmtu.sh then:
cd /tmp/exampledir && ./run_kselftest.sh
Co-developed-by: Davide Caratti <dcaratti@redhat.com>
Signed-off-by: Davide Caratti <dcaratti@redhat.com>
Signed-off-by: Briana Oursler <briana.oursler@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Enable tcp window scaling option in hw based on sysctl settings
and option in connection request.
v1->v2:
- Set window scale option based on option in connection request.
Signed-off-by: Vinay Kumar Yadav <vinay.yadav@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Mark Starovoytov says:
====================
net: atlantic: various features
This patchset adds more features for Atlantic NICs:
* media detect;
* additional per-queue stats;
* PTP stats;
* ipv6 support for TCP LSO and UDP GSO;
* 64-bit operations;
* A0 ntuple filters;
* MAC temperature (hwmon).
This work is a joint effort of Marvell developers.
v3:
* reworked patches related to stats:
. fixed u64_stats_update_* usage;
. use simple assignment in _get_stats / _fill_stats_data;
. made _get_sw_stats / _fill_stats_data return count as return value;
. split rx and tx per-queue stats;
v2: https://patchwork.ozlabs.org/cover/1329652/
* removed media detect feature (will be reworked and submitted later);
* removed irq counter from stats;
* use u64_stats_update_* to protect 64-bit stats;
* use io-64-nonatomic-lo-hi.h for readq/writeq fallbacks;
v1: https://patchwork.ozlabs.org/cover/1327894/
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch adds the possibility to obtain MAC temperature via hwmon.
On A1 there are two separate temperature sensors.
On A2 there's only one temperature sensor, which is used for reporting
both MAC and PHY temperature.
Signed-off-by: Mark Starovoytov <mstarovoitov@marvell.com>
Signed-off-by: Igor Russkikh <irusskikh@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch syncs up hw_atl_a0.c with an out-of-tree driver, where an
intermediate variable was introduced in a couple of functions to
improve the code readability a bit.
Signed-off-by: Nikita Danilov <ndanilov@marvell.com>
Signed-off-by: Mark Starovoytov <mstarovoitov@marvell.com>
Signed-off-by: Igor Russkikh <irusskikh@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch enables ipv6 support for TCP LSO and UDP GSO.
The code itself (aq_nic_map_skb) was ready for this after udp gso feature,
but corresponding NETIF_F_TSO6 wasn't enabled.
We now have tested both tcp and udp v6 GSO, and enabling them safely.
Signed-off-by: Igor Russkikh <irusskikh@marvell.com>
Signed-off-by: Mark Starovoytov <mstarovoitov@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch adds PTP rings statistics. Before that
these were missing from overall stats, hardening debugging
and analysis.
Signed-off-by: Pavel Belous <pbelous@marvell.com>
Signed-off-by: Mark Starovoytov <mstarovoitov@marvell.com>
Signed-off-by: Igor Russkikh <irusskikh@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch adds u64_stats_update_* usage to protect access to 64-bit stats,
where necessary.
This is necessary for per-ring stats, because they are updated by the
driver directly, so there is a possibility for a partial read.
Other stats require no additional protection, e.g.:
* all MACSec stats are fetched directly from HW (under semaphore);
* nic/ndev stats (aq_stats_s) are fetched directly from FW (under mutex).
Signed-off-by: Mark Starovoytov <mstarovoitov@marvell.com>
Signed-off-by: Igor Russkikh <irusskikh@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch splits rx and tx per-queue stats.
This change simplifies the follow-up introduction of PTP stats and
u64_stats_update_* usage.
Signed-off-by: Mark Starovoytov <mstarovoitov@marvell.com>
Signed-off-by: Igor Russkikh <irusskikh@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch changes aq_vec_get_sw_stats() to return count as a return
value (which was unused) instead of an out parameter.
Signed-off-by: Mark Starovoytov <mstarovoitov@marvell.com>
Signed-off-by: Igor Russkikh <irusskikh@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch replaces addition assignment operator with a simple assignment
in aq_vec_get_stats() and aq_vec_get_sw_stats(), because it is
sufficient in both cases and this change simplifies the introduction of
u64_stats_update_* in these functions.
Signed-off-by: Mark Starovoytov <mstarovoitov@marvell.com>
Signed-off-by: Igor Russkikh <irusskikh@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch moves FRAC_PER_NS to aq_hw.h so that it can be used in both
hw_atl (A1) and hw_atl2 (A2) in the future.
Signed-off-by: Mark Starovoytov <mstarovoitov@marvell.com>
Signed-off-by: Igor Russkikh <irusskikh@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Vladimir Oltean says:
====================
Extend testptp with PTP perout waveform
Demonstrate the usage of the newly introduced flags in the
PTP_PEROUT_REQUEST2 ioctl:
https://www.spinics.net/lists/netdev/msg669346.html
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Extend the example program for PTP ancillary functionality with the
ability to configure not only the periodic output's period (frequency),
but also the phase and duty cycle (pulse width) which were newly
introduced.
The ioctl level also needs to be updated to the new PTP_PEROUT_REQUEST2,
since the original PTP_PEROUT_REQUEST doesn't support this
functionality. For an in-tree testing program, not having explicit
backwards compatibility is fine, as it should always be tested with the
current kernel headers and sources.
Tested with an oscilloscope on the felix switch PHC:
echo '2 0' > /sys/class/ptp/ptp1/pins/switch_1588_dat0
./testptp -d /dev/ptp1 -p 1000000000 -w 100000000 -H 1000 -i 0
Signed-off-by: Vladimir Oltean <olteanv@gmail.com>
Acked-by: Richard Cochran <richardcochran@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Since 'perout' holds the nanosecond value of the signal's period, it
should be a 64-bit value. Current assumption is that it cannot be larger
than 1 second.
Signed-off-by: Vladimir Oltean <olteanv@gmail.com>
Acked-by: Richard Cochran <richardcochran@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Drivers using legacy PM have to manage PCI states and device's PM states
themselves. They also need to take care of configuration registers.
With improved and powerful support of generic PM, PCI Core takes care of
above mentioned, device-independent, jobs.
This driver makes use of PCI helper functions like
pci_save/restore_state(), pci_enable/disable_device(),
pci_set_power_state() and pci_set_master() to do required operations. In
generic mode, they are no longer needed.
Change function parameter in both .suspend() and .resume() to
"struct device*" type. Use to_pci_dev() and dev_get_drvdata() to get
"struct pci_dev*" variable and drv data.
Compile-tested only.
Signed-off-by: Vaibhav Gupta <vaibhavgupta40@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Alexander Lobakin says:
====================
qed, qede: add support for new operating modes
This series covers the support for the following:
- new port modes;
- loopback modes, previously missing;
- new speed/link modes;
- several FEC modes;
- multi-rate transceivers;
and also cleans up and optimizes several related parts of code.
v3 (from [2]):
- dropped custom link mode declaration; qed, qede and qedf switched to
Ethtool link modes and definitions (#0001, #0002, per Andrew Lunn's
suggestion);
- exchange more .text size to .initconst and .ro_after_init in qede
(#0003).
v2 (from [1]):
- added a patch (#0010) that drops discussed dead struct member;
- addressed checkpatch complaints on #0014 (former #0013);
- rebased on top of latest net-next;
- no other changes.
[1] https://lore.kernel.org/netdev/20200716115446.994-1-alobakin@marvell.com/
[2] https://lore.kernel.org/netdev/20200719201453.3648-1-alobakin@marvell.com/
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Add all necessary code (NVM parsing, MFW and Ethtool reports etc.) to
support extended speed and FEC modes.
These new modes are supported by the new boards revisions and newer
MFW versions.
Misc: correct port type for MEDIA_KR.
Signed-off-by: Alexander Lobakin <alobakin@marvell.com>
Signed-off-by: Igor Russkikh <irusskikh@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Simplify and lighten qed_set_link() by declaring static link modes maps
and populating them on module init. This way we save plenty of text size
at the low expense of __ro_after_init and __initconst data (the latter
will be purged after module init is done).
Misc: sanitize exit callback.
Signed-off-by: Alexander Lobakin <alobakin@marvell.com>
Signed-off-by: Igor Russkikh <irusskikh@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
These modes are relevant only for several boards, but may be reported by
MFW as well as the others.
Signed-off-by: Alexander Lobakin <alobakin@marvell.com>
Signed-off-by: Igor Russkikh <irusskikh@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
These ports ship on new boards revisions and are supported by newer
firmware versions.
Signed-off-by: Alexander Lobakin <alobakin@marvell.com>
Signed-off-by: Igor Russkikh <irusskikh@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Struct field qed_hw_info::port_mode isn't used anywhere in the code, so
can be safely removed to prevent possible dead code addition.
Also remove the enumeration QED_PORT_MODE orphaned after this deletion.
Signed-off-by: Alexander Lobakin <alobakin@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Reformat a few nvm_cfg* structures (and partly qed_dev) prior to adding
new fields and definitions.
Signed-off-by: Alexander Lobakin <alobakin@marvell.com>
Signed-off-by: Igor Russkikh <irusskikh@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add Ethtool callbacks for querying and setting FEC parameters if it's
supported by the underlying qed module and MFW version running on the
device.
Signed-off-by: Alexander Lobakin <alobakin@marvell.com>
Signed-off-by: Igor Russkikh <irusskikh@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Prior to adding new callbacks, format qede ethtool_ops structs to make
declarations more fancy and readable.
Signed-off-by: Alexander Lobakin <alobakin@marvell.com>
Signed-off-by: Igor Russkikh <irusskikh@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add all necessary routines for reading supported FEC modes from NVM and
querying FEC control to the MFW (if the running version supports it).
Signed-off-by: Alexander Lobakin <alobakin@marvell.com>
Signed-off-by: Igor Russkikh <irusskikh@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Prior to adding new fields and bitfields, reformat the related
structures according to the Linux style (spaces to tabs,
lowercase hex, indentation etc.).
Signed-off-by: Alexander Lobakin <alobakin@marvell.com>
Signed-off-by: Igor Russkikh <irusskikh@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Currently qed driver does not take into consideration transceiver's
capabilities when generating link partner's speed advertisement. This
leads to e.g. incorrect ethtool link info on 10GbaseT modules.
Use transceiver info not only for advertisement and support arrays, but
also for link partner's abilities to fix it.
Misc: fix a couple of comments nearby.
Signed-off-by: Alexander Lobakin <alobakin@marvell.com>
Signed-off-by: Igor Russkikh <irusskikh@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Set the corresponding advertised and supported link modes according
to the detected transceiver type and device capabilities.
Signed-off-by: Alexander Lobakin <alobakin@marvell.com>
Signed-off-by: Igor Russkikh <irusskikh@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Prior to adding new bitfields, reformat the existing ones from spaces
to tabs, and unify all hex values to lowercase.
Signed-off-by: Alexander Lobakin <alobakin@marvell.com>
Signed-off-by: Igor Russkikh <irusskikh@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Simplify and lighten qede_set_link_ksettings() by declaring static link
modes maps and populating them on module init. This way we save plenty
of text size at the low expense of __ro_after_init and __initconst data
(the latter will be purged after module init is done).
Signed-off-by: Alexander Lobakin <alobakin@marvell.com>
Signed-off-by: Igor Russkikh <irusskikh@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Currently qed driver already ran out of 32 bits to store link modes,
and this doesn't allow to add and support more speeds.
Convert custom link mode to generic Ethtool bitmap and definitions
(convenient Phylink shorthands are used for elegance and readability).
This allowed us to drop all conversions/mappings between the driver
and Ethtool.
This involves changes in qede and qedf as well, as they used definitions
from shared "qed_if.h".
Suggested-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Alexander Lobakin <alobakin@marvell.com>
Signed-off-by: Igor Russkikh <irusskikh@marvell.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add a new helper to find intersections between Ethtool link modes,
linkmode_intersects(), similar to the other linkmode helpers.
Signed-off-by: Alexander Lobakin <alobakin@marvell.com>
Signed-off-by: Igor Russkikh <irusskikh@marvell.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
