Due to a hardware issue, an access to MDIO registers
that is concurrent with other ENETC register accesses
may lead to the MDIO access being dropped or corrupted.
The workaround introduces locking for all register accesses
to the ENETC register space. To reduce performance impact,
a readers-writers locking scheme has been implemented.
The writer in this case is the MDIO access code (irrelevant
whether that MDIO access is a register read or write), and
the reader is any access code to non-MDIO ENETC registers.
Also, the datapath functions acquire the read lock fewer times
and use _hot accessors. All the rest of the code uses the _wa
accessors which lock every register access.
The commit introducing MDIO support is -
commit ebfcb23d62 ("enetc: Add ENETC PF level external MDIO support")
but due to subsequent refactoring this patch is applicable on
top of a later commit.
Fixes: 6517798dd3 ("enetc: Make MDIO accessors more generic and export to include/linux/fsl")
Signed-off-by: Alex Marginean <alexandru.marginean@nxp.com>
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com>
Link: https://lore.kernel.org/r/20201112182608.26177-1-claudiu.manoil@nxp.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Tx checksumming has been defeatured and completely removed
from the h/w reference manual. Made a little cleanup for the
TSE case as this is complementary code.
Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com>
Link: https://lore.kernel.org/r/20201103140213.3294-1-claudiu.manoil@nxp.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
This is a methodical transition of the driver from phylib
to phylink, following the guidelines from sfp-phylink.rst.
The MAC register configurations based on interface mode
were moved from the probing path to the mac_config() hook.
MAC enable and disable commands (enabling Rx and Tx paths
at MAC level) were also extracted and assigned to their
corresponding phylink hooks.
As part of the migration to phylink, the serdes configuration
from the driver was offloaded to the PCS_LYNX module,
introduced in commit 0da4c3d393 ("net: phy: add Lynx PCS module"),
the PCS_LYNX module being a mandatory component required to
make the enetc driver work with phylink.
Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com>
Reviewed-by: Ioana Ciornei <ioana.cionei@nxp.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The driver calls napi_schedule_irqoff() from a context where, in RT,
hardirqs are not disabled, since the IRQ handler is force-threaded.
In the call path of this function, __raise_softirq_irqoff() is modifying
its per-CPU mask of pending softirqs that must be processed, using
or_softirq_pending(). The or_softirq_pending() function is not atomic,
but since interrupts are supposed to be disabled, nobody should be
preempting it, and the operation should be safe.
Nonetheless, when running with hardirqs on, as in the PREEMPT_RT case,
it isn't safe, and the pending softirqs mask can get corrupted,
resulting in softirqs being lost and never processed.
To have common code that works with PREEMPT_RT and with mainline Linux,
we can use plain napi_schedule() instead. The difference is that
napi_schedule() (via __napi_schedule) also calls local_irq_save, which
disables hardirqs if they aren't already. But, since they already are
disabled in non-RT, this means that in practice we don't see any
measurable difference in throughput or latency with this patch.
Signed-off-by: Jiafei Pan <Jiafei.Pan@nxp.com>
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
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>
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>
The rings bitmap of an interrupt vector encodes
which of the device's rings were assigned to that
interrupt vector.
Hence the iteration range of the tx rings bitmap
(for_each_set_bit()) should be the total number of
Tx rings of that netdevice instead of the number of
rings assigned to the interrupt vector.
Since there are 2 cores, and one interrupt vector for
each core, the number of rings asigned to an interrupt
vector is half the number of available rings.
The impact of this error is that the upper half of the
tx rings could still generate interrupts during napi
polling.
Fixes: d4fd0404c1 ("enetc: Introduce basic PF and VF ENETC ethernet drivers")
Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Minor overlapping changes in xfrm_device.c, between the double
ESP trailing bug fix setting the XFRM_INIT flag and the changes
in net-next preparing for bonding encryption support.
Signed-off-by: David S. Miller <davem@davemloft.net>
VLAN tag insertion/extraction offload is correctly
activated at probe time but deactivation of this feature
(i.e. via ethtool) is broken. Toggling works only for
Tx/Rx ring 0 of a PF, and is ignored for the other rings,
including the VF rings.
To fix this, the existing VLAN offload toggling code
was extended to all the rings assigned to a netdevice,
instead of the default ring 0 (likely a leftover from the
early validation days of this feature). And the code was
moved to the common set_features() function to fix toggling
for the VF driver too.
Fixes: d4fd0404c1 ("enetc: Introduce basic PF and VF ENETC ethernet drivers")
Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Make use of the struct_size() helper instead of an open-coded version
in order to avoid any potential type mistakes.
This code was detected with the help of Coccinelle and, audited and
fixed manually.
Signed-off-by: Gustavo A. R. Silva <gustavoars@kernel.org>
Reviewed-by: Claudiu Manoil <claudiu.manoil@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch is to add tc flower offload for the enetc IEEE 802.1Qci(PSFP)
function. There are four main feature parts to implement the flow
policing and filtering for ingress flow with IEEE 802.1Qci features.
They are stream identify(this is defined in the P802.1cb exactly but
needed for 802.1Qci), stream filtering, stream gate and flow metering.
Each function block includes many entries by index to assign parameters.
So for one frame would be filtered by stream identify first, then
flow into stream filter block by the same handle between stream identify
and stream filtering. Then flow into stream gate control which assigned
by the stream filtering entry. And then policing by the gate and limited
by the max sdu in the filter block(optional). At last, policing by the
flow metering block, index choosing at the fitering block.
So you can see that each entry of block may link to many upper entries
since they can be assigned same index means more streams want to share
the same feature in the stream filtering or stream gate or flow
metering.
To implement such features, each stream filtered by source/destination
mac address, some stream maybe also plus the vlan id value would be
treated as one flow chain. This would be identified by the chain_index
which already in the tc filter concept. Driver would maintain this chain
and also with gate modules. The stream filter entry create by the gate
index and flow meter(optional) entry id and also one priority value.
Offloading only transfer the gate action and flow filtering parameters.
Driver would create (or search same gate id and flow meter id and
priority) one stream filter entry to set to the hardware. So stream
filtering do not need transfer by the action offloading.
This architecture is same with tc filter and actions relationship. tc
filter maintain the list for each flow feature by keys. And actions
maintain by the action list.
Below showing a example commands by tc:
> tc qdisc add dev eth0 ingress
> ip link set eth0 address 10:00:80:00:00:00
> tc filter add dev eth0 parent ffff: protocol ip chain 11 \
flower skip_sw dst_mac 10:00:80:00:00:00 \
action gate index 10 \
sched-entry open 200000000 1 8000000 \
sched-entry close 100000000 -1 -1
Command means to set the dst_mac 10:00:80:00:00:00 to index 11 of stream
identify module. Then setting the gate index 10 of stream gate module.
Keep the gate open for 200ms and limit the traffic volume to 8MB in this
sched-entry. Then direct the frames to the ingress queue 1.
Signed-off-by: Po Liu <Po.Liu@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch is to let ethtool enable/disable the tc flower offload
features. Hardware ENETC has the feature of PSFP which is for per-stream
policing. When enable the tc hw offloading feature, driver would enable
the IEEE 802.1Qci feature. It is only set the register enable bit for
this feature not enable for any entry of per stream filtering and stream
gate or stream identify but get how much capabilities for each feature.
Signed-off-by: Po Liu <Po.Liu@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Hardware timestamping support (PTP) on Rx requires extended
buffer descriptors, double the size of normal Rx descriptors.
On the current controller revision only the timestamping offload
requires extended Rx descriptors.
Since Rx timestamping can be turned on/off at runtime, make Rx ring
allocation configurable at runtime too. As a result, the static
config option FSL_ENETC_HW_TIMESTAMPING can be dropped and the
extended descriptors can be used only when Rx timestamping gets
activated.
The extension has the same size as the base descriptor, making
the descriptor iterators easy to update for the extended case.
Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Improve maintainability of the code iterating the Rx buffer
descriptors to prepare it to support iterating extended Rx BD
descriptors as well.
Don't increment by one the h/w descriptor pointers explicitly,
provide an iterator that takes care of the h/w details.
Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
ENETC implement time specific departure capability, which enables
the user to specify when a frame can be transmitted. When this
capability is enabled, the device will delay the transmission of
the frame so that it can be transmitted at the precisely specified time.
The delay departure time up to 0.5 seconds in the future. If the
departure time in the transmit BD has not yet been reached, based
on the current time, the packet will not be transmitted.
This driver was loaded by Qos driver ETF. User could load it by tc
commands. Here are the example commands:
tc qdisc add dev eth0 root handle 1: mqprio \
num_tc 8 map 0 1 2 3 4 5 6 7 hw 1
tc qdisc replace dev eth0 parent 1:8 etf \
clockid CLOCK_TAI delta 30000 offload
These example try to set queue mapping first and then set queue 7
with 30us ahead dequeue time.
Then user send test frame should set SO_TXTIME feature for socket.
There are also some limitations for this feature in hardware:
- Transmit checksum offloads and time specific departure operation
are mutually exclusive.
- Time Aware Shaper feature (Qbv) offload and time specific departure
operation are mutually exclusive.
Signed-off-by: Po Liu <Po.Liu@nxp.com>
Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Provide a software TX timestamp and add it to the ethtool query
interface.
skb_tx_timestamp() is also needed if one would like to use PHY
timestamping.
Signed-off-by: Michael Walle <michael@walle.cc>
Signed-off-by: David S. Miller <davem@davemloft.net>
The EEE support has not been enabled on ENETC, but it may connect
to a PHY which supports EEE and advertises EEE by default, while
its link partner also advertises EEE. If this happens, the PHY enters
low power mode when the traffic rate is low and causes packet loss.
This patch disables EEE advertisement by default for any PHY that
ENETC connects to, to prevent the above unwanted outcome.
Signed-off-by: Yangbo Lu <yangbo.lu@nxp.com>
Reviewed-by: Claudiu Manoil <claudiu.manoil@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The ENETC hardware support the Credit Based Shaper(CBS) which part
of the IEEE-802.1Qav. The CBS driver was loaded by the sch_cbs
interface when set in the QOS in the kernel.
Here is an example command to set 20Mbits bandwidth in 1Gbits port
for taffic class 7:
tc qdisc add dev eth0 root handle 1: mqprio \
num_tc 8 map 0 1 2 3 4 5 6 7 hw 1
tc qdisc replace dev eth0 parent 1:8 cbs \
locredit -1470 hicredit 30 \
sendslope -980000 idleslope 20000 offload 1
Signed-off-by: Po Liu <Po.Liu@nxp.com>
Reviewed-by: Claudiu Manoil <claudiu.manoil@nxp.com>
Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
While using ARCH=mips CROSS_COMPILE=mips-linux-gnu- command to compile,
make C=2 drivers/net/ethernet/freescale/enetc/enetc.o
one warning can be found:
drivers/net/ethernet/freescale/enetc/enetc.c:1439:5:
warning: symbol 'enetc_setup_tc_mqprio' was not declared.
Should it be static?
This patch make symbol enetc_setup_tc_mqprio static.
Fixes: 34c6adf197 ("enetc: Configure the Time-Aware Scheduler via tc-taprio offload")
Signed-off-by: Mao Wenan <maowenan@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
ENETC has a register PSPEED to indicate the link speed of hardware.
It is need to update accordingly. PSPEED field needs to be updated
with the port speed for QBV scheduling purposes. Or else there is
chance for gate slot not free by frame taking the MAC if PSPEED and
phy speed not match. So update PSPEED when link adjust. This is
implement by the adjust_link.
Signed-off-by: Po Liu <Po.Liu@nxp.com>
Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com>
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
ENETC supports in hardware for time-based egress shaping according
to IEEE 802.1Qbv. This patch implement the Qbv enablement by the
hardware offload method qdisc tc-taprio method.
Also update cbdr writeback to up level since control bd ring may
writeback data to control bd ring.
Signed-off-by: Po Liu <Po.Liu@nxp.com>
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Return -EOPNOTSUPP instead of -EINVAL if the requested ioctl is not
implemented.
Signed-off-by: Michael Walle <michael@walle.cc>
Signed-off-by: David S. Miller <davem@davemloft.net>
If there is an attached PHY try to handle the requested ioctl with its
handler, which allows the userspace to access PHY registers, for
example. This will make mii-diag and similar tools work.
Signed-off-by: Michael Walle <michael@walle.cc>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
In preparation for unifying the skb_frag and bio_vec, use the fine
accessors which already exist and use skb_frag_t instead of
struct skb_frag_struct.
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add support to configure multiple prioritized TX traffic
classes with mqprio.
Configure one BD ring per TC for the moment, one netdev
queue per TC.
Signed-off-by: Camelia Groza <camelia.groza@nxp.com>
Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Fix blow sparse warning introduced by a previous patch.
- restricted __le32 degrades to integer
- restricted __le16 degrades to integer
Fixes: d398231219 ("enetc: add hardware timestamping support")
Signed-off-by: Yangbo Lu <yangbo.lu@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch is to add hardware timestamping support
for ENETC. On Rx, timestamping is enabled for all
frames. On Tx, we only instruct the hardware to
timestamp the frames marked accordingly by the stack.
Because the RX BD ring dynamic allocation has not been
supported and it is too expensive to use extended RX BDs
if timestamping is not used, a Kconfig option is used to
enable extended RX BDs in order to support hardware
timestamping. This option will be removed once RX BD
ring dynamic allocation is implemented.
Signed-off-by: Yangbo Lu <yangbo.lu@nxp.com>
Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
For the unlikely case of TxBD extensions (i.e. ptp)
the driver tries to unmap the tx_swbd corresponding
to the extension, which is bogus as it has no buffer
attached.
Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Fixes: d4fd0404c1 ("enetc: Introduce basic PF and VF ENETC ethernet drivers")
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
A ternary match table is used for RFS. If multiple entries in the table
match, the entry with the lowest numerical values index is chosen as the
matching entry. Entries in the table are identified using an index
which takes a value from 0 to PRFSCAPR[NUM_RFS]-1 when accessed by the
PSI (PF).
Portions of the RFS table can be assigned to each SI by the PSI (PF)
driver in PSIaRFSCFGR. Assignments are cumulative, the entries assigned
to SIn start after those assigned to SIn-1. The total assignments to
all SIs must be equal to or less than the number available to the port
as found in PRFSCAPR.
For RSS, the Toeplitz hash function used requires two inputs, a 40B
random secret key that is supplied through the PRSSKR0-9 registers as well
as the relevant pieces of the packet header (n-tuple). The 6 LSB bits of
the hash function result will then be used as a pointer to obtain the tag
referenced in the 64 entry indirection table. The result will provide a
winning group which will be used to help route the received packet.
Signed-off-by: Alex Marginean <alexandru.marginean@nxp.com>
Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
ENETC is a multi-port virtualized Ethernet controller supporting GbE
designs and Time-Sensitive Networking (TSN) functionality.
ENETC is operating as an SR-IOV multi-PF capable Root Complex Integrated
Endpoint (RCIE). As such, it contains multiple physical (PF) and
virtual (VF) PCIe functions, discoverable by standard PCI Express.
Introduce basic PF and VF ENETC ethernet drivers. The PF has access to
the ENETC Port registers and resources and makes the required privileged
configurations for the underlying VF devices. Common functionality is
controlled through so called System Interface (SI) register blocks, PFs
and VFs own a SI each. Though SI register blocks are almost identical,
there are a few privileged SI level controls that are accessible only to
PFs, and so the distinction is made between PF SIs (PSI) and VF SIs (VSI).
As such, the bulk of the code, including datapath processing, basic h/w
offload support and generic pci related configuration, is shared between
the 2 drivers and is factored out in common source files (i.e. enetc.c).
Major functionalities included (for both drivers):
MSI-X support for Rx and Tx processing, assignment of Rx/Tx BD ring pairs
to MSI-X entries, multi-queue support, Rx S/G (Rx frame fragmentation) and
jumbo frame (up to 9600B) support, Rx paged allocation and reuse, Tx S/G
support (NETIF_F_SG), Rx and Tx checksum offload, PF MAC filtering and
initial control ring support, VLAN extraction/ insertion, PF Rx VLAN
CTAG filtering, VF mac address config support, VF VLAN isolation support,
etc.
Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
