Optimize Tx descriptor cleaning for XDP. Current approach doesn't
really scale and chokes when multiple flows are handled.
Introduce two ring fields, @next_dd and @next_rs that will keep track of
descriptor that should be looked at when the need for cleaning arise and
the descriptor that should have the RS bit set, respectively.
Note that at this point the threshold is a constant (32), but it is
something that we could make configurable.
First thing is to get away from setting RS bit on each descriptor. Let's
do this only once NTU is higher than the currently @next_rs value. In
such case, grab the tx_desc[next_rs], set the RS bit in descriptor and
advance the @next_rs by a 32.
Second thing is to clean the Tx ring only when there are less than 32
free entries. For that case, look up the tx_desc[next_dd] for a DD bit.
This bit is written back by HW to let the driver know that xmit was
successful. It will happen only for those descriptors that had RS bit
set. Clean only 32 descriptors and advance the DD bit.
Actual cleaning routine is moved from ice_napi_poll() down to the
ice_xmit_xdp_ring(). It is safe to do so as XDP ring will not get any
SKBs in there that would rely on interrupts for the cleaning. Nice side
effect is that for rare case of Tx fallback path (that next patch is
going to introduce) we don't have to trigger the SW irq to clean the
ring.
With those two concepts, ring is kept at being almost full, but it is
guaranteed that driver will be able to produce Tx descriptors.
This approach seems to work out well even though the Tx descriptors are
produced in one-by-one manner. Test was conducted with the ice HW
bombarded with packets from HW generator, configured to generate 30
flows.
Xdp2 sample yields the following results:
<snip>
proto 17: 79973066 pkt/s
proto 17: 80018911 pkt/s
proto 17: 80004654 pkt/s
proto 17: 79992395 pkt/s
proto 17: 79975162 pkt/s
proto 17: 79955054 pkt/s
proto 17: 79869168 pkt/s
proto 17: 79823947 pkt/s
proto 17: 79636971 pkt/s
</snip>
As that sample reports the Rx'ed frames, let's look at sar output.
It says that what we Rx'ed we do actually Tx, no noticeable drops.
Average: IFACE rxpck/s txpck/s rxkB/s txkB/s rxcmp/s txcmp/s rxmcst/s %ifutil
Average: ens4f1 79842324.00 79842310.40 4678261.17 4678260.38 0.00 0.00 0.00 38.32
with tx_busy staying calm.
When compared to a state before:
Average: IFACE rxpck/s txpck/s rxkB/s txkB/s rxcmp/s txcmp/s rxmcst/s %ifutil
Average: ens4f1 90919711.60 42233822.60 5327326.85 2474638.04 0.00 0.00 0.00 43.64
it can be observed that the amount of txpck/s is almost doubled, meaning
that the performance is improved by around 90%. All of this due to the
drops in the driver, previously the tx_busy stat was bumped at a 7mpps
rate.
Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Tested-by: George Kuruvinakunnel <george.kuruvinakunnel@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
With rings being split, it is now convenient to introduce a pointer to
XDP ring within the Rx ring. For XDP_TX workloads this means that
xdp_rings array access will be skipped, which was executed per each
processed frame.
Also, read the XDP prog once per NAPI and if prog is present, set up the
local xdp_ring pointer. Reading prog a single time was discussed in [1]
with some concern raised by Toke around dispatcher handling and having
the need for going through the RCU grace period in the ndo_bpf driver
callback, but ice currently is torning down NAPI instances regardless of
the prog presence on VSI.
Although the pointer to XDP ring introduced to Rx ring makes things a
lot slimmer/simpler, I still feel that single prog read per NAPI
lifetime is beneficial.
Further patch that will introduce the fallback path will also get a
profit from that as xdp_ring pointer will be set during the XDP rings
setup.
[1]: https://lore.kernel.org/bpf/87k0oseo6e.fsf@toke.dk/
Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Tested-by: George Kuruvinakunnel <george.kuruvinakunnel@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
There has been a long lasting issue of improper xdp_rings indexing for
XDP_TX and XDP_REDIRECT actions. Given that currently rx_ring->q_index
is mixed with smp_processor_id(), there could be a situation where Tx
descriptors are produced onto XDP Tx ring, but tail is never bumped -
for example pin a particular queue id to non-matching IRQ line.
Address this problem by ignoring the user ring count setting and always
initialize the xdp_rings array to be of num_possible_cpus() size. Then,
always use the smp_processor_id() as an index to xdp_rings array. This
provides serialization as at given time only a single softirq can run on
a particular CPU.
Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Tested-by: George Kuruvinakunnel <george.kuruvinakunnel@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
While it was convenient to have a generic ring structure that served
both Tx and Rx sides, next commits are going to introduce several
Tx-specific fields, so in order to avoid hurting the Rx side, let's
pull out the Tx ring onto new ice_tx_ring and ice_rx_ring structs.
Rx ring could be handled by the old ice_ring which would reduce the code
churn within this patch, but this would make things asymmetric.
Make the union out of the ring container within ice_q_vector so that it
is possible to iterate over newly introduced ice_tx_ring.
Remove the @size as it's only accessed from control path and it can be
calculated pretty easily.
Change definitions of ice_update_ring_stats and
ice_fetch_u64_stats_per_ring so that they are ring agnostic and can be
used for both Rx and Tx rings.
Sizes of Rx and Tx ring structs are 256 and 192 bytes, respectively. In
Rx ring xdp_rxq_info occupies its own cacheline, so it's the major
difference now.
Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Tested-by: Gurucharan G <gurucharanx.g@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
This field is dead and driver is not making any use of it. Simply remove
it.
Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Tested-by: Gurucharan G <gurucharanx.g@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
In the remove path, there is an attempt to free the aux_idx IDA whether
it was allocated or not. This can potentially cause a crash when
unloading the driver on systems that do not initialize support for RDMA.
But, this free cannot be gated by the status bit for RDMA, since it is
allocated if the driver detects support for RDMA at probe time, but the
driver can enter into a state where RDMA is not supported after the IDA
has been allocated at probe time and this would lead to a memory leak.
Initialize aux_idx to an invalid value and check for a valid value when
unloading to determine if an IDA free is necessary.
Fixes: d25a0fc41c ("ice: Initialize RDMA support")
Reported-by: Jun Miao <jun.miao@windriver.com>
Signed-off-by: Dave Ertman <david.m.ertman@intel.com>
Tested-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Implement ndo_setup_tc net device callback for TC HW offload on PF device.
ndo_setup_tc provides support for HW offloading various TC filters.
Add support for configuring the following filter with tc-flower:
- default L2 filters (src/dst mac addresses, ethertype, VLAN)
- variations of L3, L3+L4, L2+L3+L4 filters using advanced filters
(including ipv4 and ipv6 addresses).
Allow for adding/removing TC flows when PF device is configured in
eswitch switchdev mode. Two types of actions are supported at the
moment: FLOW_ACTION_DROP and FLOW_ACTION_REDIRECT.
Co-developed-by: Priyalee Kushwaha <priyalee.kushwaha@intel.com>
Signed-off-by: Priyalee Kushwaha <priyalee.kushwaha@intel.com>
Signed-off-by: Kiran Patil <kiran.patil@intel.com>
Signed-off-by: Wojciech Drewek <wojciech.drewek@intel.com>
Tested-by: Sandeep Penigalapati <sandeep.penigalapati@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
As resetting all VFs behaves mostly like creating new VFs also
eswitch infrastructure has to be recreated. The easiest way to
do that is to rebuild eswitch after resetting VFs.
Implement helper functions to start and stop all representors
queues. This is used to disable traffic on port representors.
In rebuild path:
- NAPI has to be disabled
- eswitch environment has to be set up
- new port representors have to be created, because the old
one had pointer to not existing VFs
- new control plane VSI ring should be remapped
- NAPI hast to be enabled
- rxdid has to be set to FLEX_NIC_2, because this descriptor id
support source_vsi, which is needed on control plane VSI queues
- port representors queues have to be started
Signed-off-by: Grzegorz Nitka <grzegorz.nitka@intel.com>
Tested-by: Sandeep Penigalapati <sandeep.penigalapati@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Only way to enable switchdev is to create VFs when the eswitch
mode is set to switchdev. Check if correct mode is set and
enable switchdev in function which creating VFs.
Disable switchdev when user change number of VFs to 0. Changing
eswitch mode back to legacy when VFs are created in switchdev
mode isn't allowed.
As switchdev takes care of managing filter rules, adding new
rules on VF is blocked.
In case of resetting VF driver has to update pointer in ice_repr
struct, because after reset VSI related things can change.
Co-developed-by: Wojciech Drewek <wojciech.drewek@intel.com>
Signed-off-by: Wojciech Drewek <wojciech.drewek@intel.com>
Signed-off-by: Grzegorz Nitka <grzegorz.nitka@intel.com>
Tested-by: Sandeep Penigalapati <sandeep.penigalapati@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
New type of VSI has to be defined for switchdev control plane
VSI. Number of allocated Tx and Rx queue has to be equal to
amount of VFs, because each port representor should have one
Tx and Rx queue.
Also to not increase number of used irqs too much, control plane
VSI uses only one q_vector and handle all queues in one irq.
To allow handling all queues in one irq , new function to clean
msix for eswitch was introduced. This function will schedule napi
for each representor instead of scheduling it only for one like in
normal clean irq function.
Only one additional msix has to be requested. Always try to request
it in ice_ena_msix_range function.
Signed-off-by: Grzegorz Nitka <grzegorz.nitka@intel.com>
Tested-by: Sandeep Penigalapati <sandeep.penigalapati@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Switchdev environment has to be set up when user create VFs
and eswitch mode is switchdev. Release is done when user
delete all VFs.
Data path in this implementation is based on control plane VSI.
This VSI is used to pass traffic from port representors to
corresponding VFs and vice versa. Default TX rule has to be
added to forward packet to control plane VSI. This will redirect
packets from VFs which don't match other rules to control plane
VSI.
On RX side default rule is added on uplink VSI to receive all
traffic that doesn't match other rules. When setting switchdev
environment all other rules from VFs should be removed. Packet to
VFs will be forwarded by control plane VSI.
As VF without any mac rules can't send any packet because of
antispoof mechanism, VSI antispoof should be turned off on each VFs.
To send packet from representor to correct VSI, destination VSI
field in TX descriptor will have to be filled. Allow that by
setting destination override bit in control plane VSI security config.
Packet from VFs will be received on control plane VSI. Driver
should decide to which netdev forward the packet. Decision is
made based on src_vsi field from descriptor. There is a target
netdev list in control plane VSI struct which choose netdev
based on src_vsi number.
Co-developed-by: Michal Swiatkowski <michal.swiatkowski@linux.intel.com>
Signed-off-by: Michal Swiatkowski <michal.swiatkowski@linux.intel.com>
Signed-off-by: Grzegorz Nitka <grzegorz.nitka@intel.com>
Tested-by: Sandeep Penigalapati <sandeep.penigalapati@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Keeping devlink port inside VSI data structure causes some issues.
Since VF VSI is released during reset that means that we have to
unregister devlink port and register it again every time reset is
triggered. With the new changes in devlink API it
might cause deadlock issues. After calling
devlink_port_register/devlink_port_unregister devlink API is going to
lock rtnl_mutex. It's an issue when VF reset is triggered in netlink
operation context (like setting VF MAC address or VLAN),
because rtnl_lock is already taken by netlink. Another call of
rtnl_lock from devlink API results in dead-lock.
By moving devlink port to PF/VF we avoid creating/destroying it
during reset. Since this patch, devlink ports are created during
ice_probe, destroyed during ice_remove for PF and created during
ice_repr_add, destroyed during ice_repr_rem for VF.
Signed-off-by: Wojciech Drewek <wojciech.drewek@intel.com>
Tested-by: Sandeep Penigalapati <sandeep.penigalapati@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Convert all Ethernet drivers from memcpy(... dev->addr_len)
to eth_hw_addr_set():
@@
expression dev, np;
@@
- memcpy(dev->dev_addr, np, dev->addr_len)
+ eth_hw_addr_set(dev, np)
In theory addr_len may not be ETH_ALEN, but we don't expect
non-Ethernet devices to live under this directory, and only
the following cases of setting addr_len exist:
- cxgb4 for mgmt device,
and the drivers which set it to ETH_ALEN: s2io, mlx4, vxge.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Convert Ethernet from ether_addr_copy() to eth_hw_addr_set():
@@
expression dev, np;
@@
- ether_addr_copy(dev->dev_addr, np)
+ eth_hw_addr_set(dev, np)
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
The messaging for unsupported module detection is different for
lenient mode and strict mode. Update the code to print the right
messaging for a given link mode.
Media topology conflict is not an error in lenient mode, so return
an error code only if not in lenient mode.
Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com>
Tested-by: Gurucharan G <gurucharanx.g@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
DSCP a.k.a L3 QoS is only supported on certain devices. To enforce this,
this patch introduces a bitmap of features and helper functions.
The feature bitmap is set based on device IDs on driver init. Currently,
DSCP is the only feature in this bitmap, but there will be more in the
future. In the DCB netlink flow, check if the feature bit is set before
exercising DSCP.
Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com>
Tested-by: Gurucharan G <gurucharanx.g@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Implement code to handle submission of APP TLV's
containing DSCP to TC mapping.
The first such mapping received on an interface
will cause that PF to switch to L3 DSCP QoS mode,
apply the default config for that mode, and apply
the received mapping.
Only one such mapping will be allowed per DSCP value,
and when the last DSCP mapping is deleted, the PF
will switch back into L2 VLAN QoS mode, applying the
appropriate default QoS settings.
L3 DSCP QoS mode will only be allowed in SW DCBx
mode, in other words, when the FW LLDP engine is
disabled. Commands that break this mutual exclusivity
will be blocked.
Co-developed-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com>
Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com>
Signed-off-by: Dave Ertman <david.m.ertman@intel.com>
Tested-by: Gurucharan G <gurucharanx.g@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Move devlink_registration routine to be the last command, when the
device is fully initialized.
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
PF pointer is always valid when PCI core calls its .shutdown() and
.remove() callbacks. There is no need to check it again.
Fixes: 837f08fdec ("ice: Add basic driver framework for Intel(R) E800 Series")
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
devlink_register() can't fail and always returns success, but all drivers
are obligated to check returned status anyway. This adds a lot of boilerplate
code to handle impossible flow.
Make devlink_register() void and simplify the drivers that use that
API call.
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
Acked-by: Simon Horman <simon.horman@corigine.com>
Acked-by: Vladimir Oltean <olteanv@gmail.com> # dsa
Reviewed-by: Jiri Pirko <jiri@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In some circumstances, such as with bridging, it's possible that the
stack will add the device's own MAC address to its unicast address list.
If, later, the stack deletes this address, the driver will receive a
request to remove this address.
The driver stores its current MAC address as part of the VSI MAC filter
list instead of separately. So, this causes a problem when the device's
MAC address is deleted unexpectedly, which results in traffic failure in
some cases.
The following configuration steps will reproduce the previously
mentioned problem:
> ip link set eth0 up
> ip link add dev br0 type bridge
> ip link set br0 up
> ip addr flush dev eth0
> ip link set eth0 master br0
> echo 1 > /sys/class/net/br0/bridge/vlan_filtering
> modprobe -r veth
> modprobe -r bridge
> ip addr add 192.168.1.100/24 dev eth0
The following ping command fails due to the netdev->dev_addr being
deleted when removing the bridge module.
> ping <link partner>
Fix this by making sure to not delete the netdev->dev_addr during MAC
address sync. After fixing this issue it was noticed that the
netdev_warn() in .set_mac was overly verbose, so make it at
netdev_dbg().
Also, there is a possibility of a race condition between .set_mac and
.set_rx_mode. Fix this by calling netif_addr_lock_bh() and
netif_addr_unlock_bh() on the device's netdev when the netdev->dev_addr
is going to be updated in .set_mac.
Fixes: e94d447866 ("ice: Implement filter sync, NDO operations and bump version")
Signed-off-by: Brett Creeley <brett.creeley@intel.com>
Tested-by: Liang Li <liali@redhat.com>
Tested-by: Gurucharan G <gurucharanx.g@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
The userspace utility "driverctl" can be used to change/override the
system's default driver choices. This is useful in some situations
(buggy driver, old driver missing a device ID, trying a workaround,
etc.) where the user needs to load a different driver.
However, this is also prone to user error, where a driver is mapped
to a device it's not designed to drive. For example, if the ice driver
is mapped to driver iavf devices, the ice driver crashes.
Add a check to return an error if the ice driver is being used to
probe a virtual function.
Fixes: 837f08fdec ("ice: Add basic driver framework for Intel(R) E800 Series")
Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com>
Tested-by: Gurucharan G <gurucharanx.g@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Most users of ndo_do_ioctl are ethernet drivers that implement
the MII commands SIOCGMIIPHY/SIOCGMIIREG/SIOCSMIIREG, or hardware
timestamping with SIOCSHWTSTAMP/SIOCGHWTSTAMP.
Separate these from the few drivers that use ndo_do_ioctl to
implement SIOCBOND, SIOCBR and SIOCWANDEV commands.
This is a purely cosmetic change intended to help readers find
their way through the implementation.
Cc: Doug Ledford <dledford@redhat.com>
Cc: Jason Gunthorpe <jgg@ziepe.ca>
Cc: Jay Vosburgh <j.vosburgh@gmail.com>
Cc: Veaceslav Falico <vfalico@gmail.com>
Cc: Andy Gospodarek <andy@greyhouse.net>
Cc: Andrew Lunn <andrew@lunn.ch>
Cc: Vivien Didelot <vivien.didelot@gmail.com>
Cc: Florian Fainelli <f.fainelli@gmail.com>
Cc: Vladimir Oltean <olteanv@gmail.com>
Cc: Leon Romanovsky <leon@kernel.org>
Cc: linux-rdma@vger.kernel.org
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Acked-by: Jason Gunthorpe <jgg@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The E810 device supports programmable pins for enabling both input and
output events related to the PTP hardware clock. This includes both
output signals with programmable period, as well as timestamping of
events on input pins.
Add support for enabling these using the CONFIG_PTP_1588_CLOCK
interface.
This allows programming the software defined pins to take advantage of
the hardware clock features.
Signed-off-by: Maciej Machnikowski <maciej.machnikowski@intel.com>
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
This patch is modeled after one by Scott Peterson for i40e.
Add tracepoints to the driver, via a new file ice_trace.h and some new
trace calls added in interesting places in the driver. Add some tracing
for DIMLIB to help debug interrupt moderation problems.
Performance should not be affected, and this can be very useful
for debugging and adding new trace events to paths in the future.
Note eBPF programs can attach to these events, as well as perf
can count them since we're attaching to the events subsystem
in the kernel.
Co-developed-by: Ben Shelton <benjamin.h.shelton@intel.com>
Signed-off-by: Ben Shelton <benjamin.h.shelton@intel.com>
Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Tested-by: Tony Brelinski <tonyx.brelinski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Trivial conflicts in net/can/isotp.c and
tools/testing/selftests/net/mptcp/mptcp_connect.sh
scaled_ppm_to_ppb() was moved from drivers/ptp/ptp_clock.c
to include/linux/ptp_clock_kernel.h in -next so re-apply
the fix there.
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Remove the local variable since it's only used once. Instead, use it
directly.
Signed-off-by: Paul M Stillwell Jr <paul.m.stillwell.jr@intel.com>
Tested-by: Tony Brelinski <tonyx.brelinski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
There are some places where the scope of a variable can
be reduced so do that.
Signed-off-by: Paul M Stillwell Jr <paul.m.stillwell.jr@intel.com>
Tested-by: Tony Brelinski <tonyx.brelinski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Add support for enabling Tx timestamp requests for outgoing packets on
E810 devices.
The ice hardware can support multiple outstanding Tx timestamp requests.
When sending a descriptor to hardware, a Tx timestamp request is made by
setting a request bit, and assigning an index that represents which Tx
timestamp index to store the timestamp in.
Hardware makes no effort to synchronize the index use, so it is up to
software to ensure that Tx timestamp indexes are not re-used before the
timestamp is reported back.
To do this, introduce a Tx timestamp tracker which will keep track of
currently in-use indexes.
In the hot path, if a packet has a timestamp request, an index will be
requested from the tracker. Unfortunately, this does require a lock as
the indexes are shared across all queues on a PHY. There are not enough
indexes to reliably assign only 1 to each queue.
For the E810 devices, the timestamp indexes are not shared across PHYs,
so each port can have its own tracking.
Once hardware captures a timestamp, an interrupt is fired. In this
interrupt, trigger a new work item that will figure out which timestamp
was completed, and report the timestamp back to the stack.
This function loops through the Tx timestamp indexes and checks whether
there is now a valid timestamp. If so, it clears the PHY timestamp
indication in the PHY memory, locks and removes the SKB and bit in the
tracker, then reports the timestamp to the stack.
It is possible in some cases that a timestamp request will be initiated
but never completed. This might occur if the packet is dropped by
software or hardware before it reaches the PHY.
Add a task to the periodic work function that will check whether
a timestamp request is more than a few seconds old. If so, the timestamp
index is cleared in the PHY, and the SKB is released.
Just as with Rx timestamps, the Tx timestamps are only 40 bits wide, and
use the same overall logic for extending to 64 bits of nanoseconds.
With this change, E810 devices should be able to perform basic PTP
functionality.
Future changes will extend the support to cover the E822-based devices.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Tony Brelinski <tonyx.brelinski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Add SIOCGHWTSTAMP and SIOCSHWTSTAMP ioctl handlers to respond to
requests to enable timestamping support. If the request is for enabling
Rx timestamps, set a bit in the Rx descriptors to indicate that receive
timestamps should be reported.
Hardware captures receive timestamps in the PHY which only captures part
of the timer, and reports only 40 bits into the Rx descriptor. The upper
32 bits represent the contents of GLTSYN_TIME_L at the point of packet
reception, while the lower 8 bits represent the upper 8 bits of
GLTSYN_TIME_0.
The networking and PTP stack expect 64 bit timestamps in nanoseconds. To
support this, implement some logic to extend the timestamps by using the
full PHC time.
If the Rx timestamp was captured prior to the PHC time, then the real
timestamp is
PHC - (lower_32_bits(PHC) - timestamp)
If the Rx timestamp was captured after the PHC time, then the real
timestamp is
PHC + (timestamp - lower_32_bits(PHC))
These calculations are correct as long as neither the PHC timestamp nor
the Rx timestamps are more than 2^32-1 nanseconds old. Further, we can
detect when the Rx timestamp is before or after the PHC as long as the
PHC timestamp is no more than 2^31-1 nanoseconds old.
In that case, we calculate the delta between the lower 32 bits of the
PHC and the Rx timestamp. If it's larger than 2^31-1 then the Rx
timestamp must have been captured in the past. If it's smaller, then the
Rx timestamp must have been captured after PHC time.
Add an ice_ptp_extend_32b_ts function that relies on a cached copy of
the PHC time and implements this algorithm to calculate the proper upper
32bits of the Rx timestamps.
Cache the PHC time periodically in all of the Rx rings. This enables
each Rx ring to simply call the extension function with a recent copy of
the PHC time. By ensuring that the PHC time is kept up to date
periodically, we ensure this algorithm doesn't use stale data and
produce incorrect results.
To cache the time, introduce a kworker and a kwork item to periodically
store the Rx time. It might seem like we should use the .do_aux_work
interface of the PTP clock. This doesn't work because all PFs must cache
this time, but only one PF owns the PTP clock device.
Thus, the ice driver will manage its own kthread instead of relying on
the PTP do_aux_work handler.
With this change, the driver can now report Rx timestamps on all
incoming packets.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Tony Brelinski <tonyx.brelinski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Add a new ice_ptp.c file for holding the basic PTP clock interface
functions. If the device supports PTP, call the new ice_ptp_init and
ice_ptp_release functions where appropriate.
If the function owns the hardware resource associated with the PTP
hardware clock, register with the PTP_1588_CLOCK infrastructure to
allocate a new clock object that represents the device hardware clock.
Implement basic functionality for reading and setting the clock time,
performing clock adjustments, and adjusting the clock frequency.
Future changes will introduce functionality for handling related
features including Tx and Rx timestamps.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Tony Brelinski <tonyx.brelinski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
In order to support certain device features, including enabling the PTP
hardware clock, the ice driver needs to control some registers on the
device PHY.
These registers are accessed by sending sideband messages. For some
hardware, these messages must be sent over the device admin queue, while
other hardware has a dedicated control queue for the sideband messages.
Add the neighbor device message structure for sending a message to the
neighboring device. Where supported, initialize the sideband control
queue and handle cleanup.
Add a wrapper function for sending sideband control queue messages that
read or write a neighboring device register.
Because some devices send sideband messages over the AdminQ, also
increase the length of the admin queue to allow more messages to be
queued up. This is important because the sideband messages add
additional pressure on the AQ usage.
This support will be used in following patches to enable support for
CONFIG_1588_PTP_CLOCK.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Tony Brelinski <tonyx.brelinski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
ice driver requires a programmable pipeline firmware package in order to
have a support for advanced features. Otherwise, driver falls back to so
called 'safe mode'. For that mode, ndo_bpf callback is not exposed and
when user tries to load XDP program, the following happens:
$ sudo ./xdp1 enp179s0f1
libbpf: Kernel error message: Underlying driver does not support XDP in native mode
link set xdp fd failed
which is sort of confusing, as there is a native XDP support, but not in
the current mode. Improve the user experience by providing the specific
ndo_bpf callback dedicated for safe mode which will make use of extack
to explicitly let the user know that the DDP package is missing and
that's the reason that the XDP can't be loaded onto interface currently.
Cc: Jamal Hadi Salim <jhs@mojatatu.com>
Fixes: efc2214b60 ("ice: Add support for XDP")
Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Tested-by: Kiran Bhandare <kiranx.bhandare@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Determine whether an unsupported power configuration is preventing link
establishment by storing and checking the link_cfg_err_byte. Print error
messages when module power levels are unsupported. Also add a new flag
bit to prevent spamming said error messages.
Co-developed-by: Jeb Cramer <jeb.j.cramer@intel.com>
Signed-off-by: Jeb Cramer <jeb.j.cramer@intel.com>
Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com>
Tested-by: Tony Brelinski <tonyx.brelinski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
After performing a flash update, a device EMP reset may occur. This
reset will cause the newly downloaded firmware to be initialized. When
this happens, the driver still reports the previous NVM version
information.
This is because the NVM versions are cached within the hw structure.
This can be confusing, as the new firmware is in fact running in this
case.
Handle this by calling ice_init_nvm when rebuilding the driver state.
This will update the flash version information and ensures that the
current values are displayed when reporting the NVM versions to the
stack.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Tony Brelinski <tonyx.brelinski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Requesting device firmware information while the device is busy cleaning
up after a reset can result in an unexpected failure:
This occurs because the command is attempting to access the device
AdminQ while it is down. Resolve this by having the command wait for
a while until the reset is complete. To do this, introduce
a reset_wait_queue and associated helper function "ice_wait_for_reset".
This helper will use the wait queue to sleep until the driver is done
rebuilding. Use of a wait queue is preferred because the potential sleep
duration can be several seconds.
To ensure that the thread wakes up properly, a new wake_up call is added
during all code paths which clear the reset state bits associated with
the driver rebuild flow.
Using this ensures that tools can request device information without
worrying about whether the driver is cleaning up from a reset.
Specifically, it is expected that a flash update could result in
a device reset, and it is better to delay the response for information
until the reset is complete rather than exit with an immediate failure.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Tony Brelinski <tonyx.brelinski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Register ice client auxiliary RDMA device on the auxiliary bus per
PCIe device function for the auxiliary driver (irdma) to attach to.
It allows to realize a single RDMA driver (irdma) capable of working with
multiple netdev drivers over multi-generation Intel HW supporting RDMA.
There is no load ordering dependencies between ice and irdma.
Signed-off-by: Dave Ertman <david.m.ertman@intel.com>
Signed-off-by: Shiraz Saleem <shiraz.saleem@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Add implementations for supporting iidc operations for device operation
such as allocation of resources and event notifications.
Signed-off-by: Dave Ertman <david.m.ertman@intel.com>
Signed-off-by: Shiraz Saleem <shiraz.saleem@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Probe the device's capabilities to see if it supports RDMA. If so, allocate
and reserve resources to support its operation; populate structures with
initial values.
Signed-off-by: Dave Ertman <david.m.ertman@intel.com>
Signed-off-by: Shiraz Saleem <shiraz.saleem@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Attempt to detect malicious VFs and, if suspected, log the information but
keep going to allow the user to take any desired actions.
Potentially malicious VFs are identified by checking if the VFs are
transmitting too many messages via the PF-VF mailbox which could cause an
overflow of this channel resulting in denial of service. This is done by
creating a snapshot or static capture of the mailbox buffer which can be
traversed and in which the messages sent by VFs are tracked.
Co-developed-by: Yashaswini Raghuram Prathivadi Bhayankaram <yashaswini.raghuram.prathivadi.bhayankaram@intel.com>
Signed-off-by: Yashaswini Raghuram Prathivadi Bhayankaram <yashaswini.raghuram.prathivadi.bhayankaram@intel.com>
Co-developed-by: Paul M Stillwell Jr <paul.m.stillwell.jr@intel.com>
Signed-off-by: Paul M Stillwell Jr <paul.m.stillwell.jr@intel.com>
Co-developed-by: Brett Creeley <brett.creeley@intel.com>
Signed-off-by: Brett Creeley <brett.creeley@intel.com>
Signed-off-by: Vignesh Sridhar <vignesh.sridhar@intel.com>
Tested-by: Konrad Jankowski <konrad0.jankowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
We were saving the return value from ice_vsi_manage_rss_lut(), but
the errors from that function are not critical so change it to
return void and remove the code that saved the value.
Signed-off-by: Paul M Stillwell Jr <paul.m.stillwell.jr@intel.com>
Tested-by: Tony Brelinski <tonyx.brelinski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
The driver previously printed it's PCI address in
the name field for the pci resource, which when displayed
via /proc/iomem, would print the same thing twice.
It's more useful for debugging to see the driver name, as
most other modules do.
Here's a diff of before and after this change:
99100000-991fffff : 0000:3b:00.1
9a000000-a04fffff : PCI Bus 0000:3b
9a000000-9bffffff : 0000:3b:00.1
- 9a000000-9bffffff : 0000:3b:00.1
+ 9a000000-9bffffff : ice
9c000000-9dffffff : 0000:3b:00.0
- 9c000000-9dffffff : 0000:3b:00.0
+ 9c000000-9dffffff : ice
9e000000-9effffff : 0000:3b:00.1
9f000000-9fffffff : 0000:3b:00.0
a0000000-a000ffff : 0000:3b:00.1
Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Tested-by: Tony Brelinski <tonyx.brelinski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
The ice driver has support for adaptive interrupt moderation, an
algorithm for tuning the interrupt rate dynamically. This algorithm
is based on various assumptions about ring size, socket buffer size,
link speed, SKB overhead, ethernet frame overhead and more.
The Linux kernel has support for a dynamic interrupt moderation
algorithm known as "dimlib". Replace the custom driver-specific
implementation of dynamic interrupt moderation with the kernel's
algorithm.
The Intel hardware has a different hardware implementation than the
originators of the dimlib code had to work with, which requires the
driver to use a slightly different set of inputs for the actual
moderation values, while getting all the advice from dimlib of
better/worse, shift left or right.
The change made for this implementation is to use a pair of values
for each of the 5 "slots" that the dimlib moderation expects, and
the driver will program those pairs when dimlib recommends a slot to
use. The currently implementation uses two tables, one for receive
and one for transmit, and the pairs of values in each slot set the
maximum delay of an interrupt and a maximum number of interrupts per
second (both expressed in microseconds).
There are two separate kinds of bugs fixed by using DIMLIB, one is
UDP single stream send was too slow, and the other is that 8K
ping-pong was going to the most aggressive moderation and has much
too high latency.
The overall result of using DIMLIB is that we meet or exceed our
performance expectations set based on the old algorithm.
Co-developed-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Tony Brelinski <tonyx.brelinski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Add two new VSI states, one to track if a netdev for the VSI has been
allocated and the other to track if the netdev has been registered.
Call unregister_netdev/free_netdev only when the corresponding state
bits are set.
Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com>
Tested-by: Tony Brelinski <tonyx.brelinski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Remove the leading underscores in enum ice_pf_state. This is not really
communicating anything and is unnecessary. No functional change.
Signed-off-by: Anirudh Venkataramanan <anirudh.venkataramanan@intel.com>
Tested-by: Tony Brelinski <tonyx.brelinski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Conflicts:
MAINTAINERS
- keep Chandrasekar
drivers/net/ethernet/mellanox/mlx5/core/en_main.c
- simple fix + trust the code re-added to param.c in -next is fine
include/linux/bpf.h
- trivial
include/linux/ethtool.h
- trivial, fix kdoc while at it
include/linux/skmsg.h
- move to relevant place in tcp.c, comment re-wrapped
net/core/skmsg.c
- add the sk = sk // sk = NULL around calls
net/tipc/crypto.c
- trivial
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
In ice_suspend(), ice_clear_interrupt_scheme() is called, and then
irq_free_descs() will be eventually called to free irq and its descriptor.
In ice_resume(), ice_init_interrupt_scheme() is called to allocate new
irqs. However, in ice_rebuild_arfs(), struct irq_glue and struct cpu_rmap
maybe cannot be freed, if the irqs that released in ice_suspend() were
reassigned to other devices, which makes irq descriptor's affinity_notify
lost.
So call ice_free_cpu_rx_rmap() before ice_clear_interrupt_scheme(), which
can make sure all irq_glue and cpu_rmap can be correctly released before
corresponding irq and descriptor are released.
Fix the following memory leak.
unreferenced object 0xffff95bd951afc00 (size 512):
comm "kworker/0:1", pid 134, jiffies 4294684283 (age 13051.958s)
hex dump (first 32 bytes):
18 00 00 00 18 00 18 00 70 fc 1a 95 bd 95 ff ff ........p.......
00 00 ff ff 01 00 ff ff 02 00 ff ff 03 00 ff ff ................
backtrace:
[<0000000072e4b914>] __kmalloc+0x336/0x540
[<0000000054642a87>] alloc_cpu_rmap+0x3b/0xb0
[<00000000f220deec>] ice_set_cpu_rx_rmap+0x6a/0x110 [ice]
[<000000002370a632>] ice_probe+0x941/0x1180 [ice]
[<00000000d692edba>] local_pci_probe+0x47/0xa0
[<00000000503934f0>] work_for_cpu_fn+0x1a/0x30
[<00000000555a9e4a>] process_one_work+0x1dd/0x410
[<000000002c4b414a>] worker_thread+0x221/0x3f0
[<00000000bb2b556b>] kthread+0x14c/0x170
[<00000000ad2cf1cd>] ret_from_fork+0x1f/0x30
unreferenced object 0xffff95bd81b0a2a0 (size 96):
comm "kworker/0:1", pid 134, jiffies 4294684283 (age 13051.958s)
hex dump (first 32 bytes):
38 00 00 00 01 00 00 00 e0 ff ff ff 0f 00 00 00 8...............
b0 a2 b0 81 bd 95 ff ff b0 a2 b0 81 bd 95 ff ff ................
backtrace:
[<00000000582dd5c5>] kmem_cache_alloc_trace+0x31f/0x4c0
[<000000002659850d>] irq_cpu_rmap_add+0x25/0xe0
[<00000000495a3055>] ice_set_cpu_rx_rmap+0xb4/0x110 [ice]
[<000000002370a632>] ice_probe+0x941/0x1180 [ice]
[<00000000d692edba>] local_pci_probe+0x47/0xa0
[<00000000503934f0>] work_for_cpu_fn+0x1a/0x30
[<00000000555a9e4a>] process_one_work+0x1dd/0x410
[<000000002c4b414a>] worker_thread+0x221/0x3f0
[<00000000bb2b556b>] kthread+0x14c/0x170
[<00000000ad2cf1cd>] ret_from_fork+0x1f/0x30
Fixes: 769c500dcc ("ice: Add advanced power mgmt for WoL")
Signed-off-by: Yongxin Liu <yongxin.liu@windriver.com>
Tested-by: Tony Brelinski <tonyx.brelinski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
