[ Upstream commit 86b0ca5b11 ]
Free Tx port timestamping metadata entries in the NAPI poll context and
consume metadata enties in the WQE xmit path. Do not free a Tx port
timestamping metadata entry in the WQE xmit path even in the error path to
avoid a race between two metadata entry producers.
Fixes: 3178308ad4 ("net/mlx5e: Make tx_port_ts logic resilient to out-of-order CQEs")
Signed-off-by: Rahul Rameshbabu <rrameshbabu@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
Link: https://lore.kernel.org/r/20240409190820.227554-10-tariqt@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
Just simply reordering the functions mlx5e_ptp_metadata_map_put and
mlx5e_ptpsq_track_metadata in the mlx5e_txwqe_complete context is not good
enough since both the compiler and CPU are free to reorder these two
functions. If reordering does occur, the issue that was supposedly fixed by
7e3f3ba97e ("net/mlx5e: Track xmit submission to PTP WQ after populating
metadata map") will be seen. This will lead to NULL pointer dereferences in
mlx5e_ptpsq_mark_ts_cqes_undelivered in the NAPI polling context due to the
tracking list being populated before the metadata map.
Fixes: 7e3f3ba97e ("net/mlx5e: Track xmit submission to PTP WQ after populating metadata map")
Signed-off-by: Rahul Rameshbabu <rrameshbabu@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
CC: Vadim Fedorenko <vadfed@meta.com>
Make CQ struct and methods independent of "priv", use more basic
arguments instead.
This will ease the transition to netdev with multiple mdevs.
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
Reviewed-by: Gal Pressman <gal@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
When SQ is a port timestamping SQ for PTP, do not access tx flags of skb
after free-ing the skb. Free the skb only after all references that depend
on it have been handled in the dropped WQE path.
Fixes: 3178308ad4 ("net/mlx5e: Make tx_port_ts logic resilient to out-of-order CQEs")
Signed-off-by: Rahul Rameshbabu <rrameshbabu@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Link: https://lore.kernel.org/r/20231114215846.5902-10-saeed@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Use a map structure for associating CQEs containing port timestamping
information with the appropriate skb. Track order of WQEs submitted using a
FIFO. Check if the corresponding port timestamping CQEs from the lookup
values in the FIFO are considered dropped due to time elapsed. Return the
lookup value to a freelist after consuming the skb. Reuse the freed lookup
in future WQE submission iterations.
The map structure uses an integer identifier for the key and returns an skb
corresponding to that identifier. Embed the integer identifier in the WQE
submitted to the WQ for the transmit path when the SQ is a PTP (port
timestamping) SQ. The embedded identifier can then be queried using a field
in the CQE of the corresponding port timestamping CQ. In the port
timestamping napi_poll context, the identifier is queried from the CQE
polled from CQ and used to lookup the corresponding skb from the WQE submit
path. The skb reference is removed from map and then embedded with the port
HW timestamp information from the CQE and eventually consumed.
The metadata freelist FIFO is an array containing integer identifiers that
can be pushed and popped in the FIFO. The purpose of this structure is
bookkeeping what identifier values can safely be used in a subsequent WQE
submission and should not contain identifiers that have still not been
reaped by processing a corresponding CQE completion on the port
timestamping CQ.
The ts_cqe_pending_list structure is a combination of an array and linked
list. The array is pre-populated with the nodes that will be added and
removed from the head of the linked list. Each node contains the unique
identifier value associated with the values submitted in the WQEs and
retrieved in the port timestamping CQEs. When a WQE is submitted, the node
in the array corresponding to the identifier popped from the metadata
freelist is added to the end of the CQE pending list and is marked as
"in-use". The node is removed from the linked list under two conditions.
The first condition is that the corresponding port timestamping CQE is
polled in the PTP napi_poll context. The second condition is that more than
a second has elapsed since the DMA timestamp value corresponding to the WQE
submission. When the first condition occurs, the "in-use" bit in the linked
list node is cleared, and the resources corresponding to the WQE submission
are then released. The second condition, however, indicates that the port
timestamping CQE will likely never be delivered. It's not impossible for
the device to post a CQE after an infinite amount of time though highly
improbable. In order to be resilient to this improbable case, resources
related to the corresponding WQE submission are still kept, the identifier
value is not returned to the freelist, and the "in-use" bit is cleared on
the node to indicate that it's no longer part of the linked list of "likely
to be delivered" port timestamping CQE identifiers. A count for the number
of port timestamping CQEs considered highly likely to never be delivered by
the device is maintained. This count gets decremented in the unlikely event
a port timestamping CQE considered unlikely to ever be delivered is polled
in the PTP napi_poll context.
Signed-off-by: Rahul Rameshbabu <rrameshbabu@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Check in the mlx5e_ptp_poll_ts_cq context if the ptp tx sq should be woken
up. Before change, the ptp tx sq may never wake up if the ptp tx ts skb
fifo is full when mlx5e_poll_tx_cq checks if the queue should be woken up.
Fixes: 1880bc4e4a ("net/mlx5e: Add TX port timestamp support")
Signed-off-by: Rahul Rameshbabu <rrameshbabu@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
The last usage was removed as part of
commit 40379a0084 ("net/mlx5_fpga: Drop INNOVA TLS support").
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
Reviewed-by: Gal Pressman <gal@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
The commit cited below started using the firmware capability for the
maximum TX WQE size. This commit adds an important check to verify that
the driver doesn't attempt to exceed this capability, and also restores
another check mistakenly removed in the cited commit (a WQE must not
exceed the page size).
Fixes: c27bd1718c ("net/mlx5e: Read max WQEBBs on the SQ from firmware")
Signed-off-by: Maxim Mikityanskiy <maximmi@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
When tx_port_ts is set, the driver diverts all UPD traffic over PTP port
to a dedicated PTP-SQ. The SKBs are cached until the wire-CQE arrives.
When the packet size is greater then MTU, the firmware might drop it and
the packet won't be transmitted to the wire, hence the wire-CQE won't
reach the driver. In this case the SKBs are accumulated in the SKB fifo.
Add room check to consider the PTP-SQ SKB fifo, when the SKB fifo is
full, driver stops the queue resulting in a TX timeout. Devlink
TX-reporter can recover from it.
Fixes: 1880bc4e4a ("net/mlx5e: Add TX port timestamp support")
Signed-off-by: Aya Levin <ayal@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Link: https://lore.kernel.org/r/20221026135153.154807-5-saeed@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
MACsec driver marks Tx packets for device offload using a dedicated
skb_metadata_dst which holds a 64 bits SCI number.
A previously set rule will match on this number so the correct SA is used
for the MACsec operation.
As device driver can only provide 32 bits of metadata to flow tables,
need to used a mapping from 64 bit to 32 bits marker or id,
which is can be achieved by provide a 32 bit unique flow id in the
control path, and used a hash table to map 64 bit to the unique id in the
datapath.
Signed-off-by: Lior Nahmanson <liorna@nvidia.com>
Reviewed-by: Raed Salem <raeds@nvidia.com>
Signed-off-by: Raed Salem <raeds@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Use skb_inner_tcp_all_headers() instead of skb_tcp_all_headers() when
transmitting an encapsulated packet in mlx5e_tx_get_gso_ihs().
Fixes: 504148fedb ("net: add skb_[inner_]tcp_all_headers helpers")
Cc: Eric Dumazet <edumazet@google.com>
Signed-off-by: Gal Pressman <gal@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
PTP TX port timestamp relies on receiving 2 CQEs for each outgoing
packet (WQE). The regular CQE has a less accurate timestamp than the
wire CQE. On link change, the wire CQE may get lost. Let the driver
detect and restore the relation between the CQEs, and re-sync after
timeout.
Add resiliency for this as follows: add id (producer counter)
into the WQE's metadata. This id will be received in the wire
CQE (in wqe_counter field). On handling the wire CQE, if there is no
match, replay the PTP application with the time-stamp from the regular
CQE and restore the sync between the CQEs and their SKBs. This patch
adds 2 ptp counters:
1) ptp_cq0_resync_event: number of times a mismatch was detected between
the regular CQE and the wire CQE.
2) ptp_cq0_resync_cqe: total amount of missing wire CQEs.
Signed-off-by: Aya Levin <ayal@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
TX doorbells may be postponed, because sometimes the driver knows that
another packet follows (for example, when xmit_more is true, or when a
MPWQE session is closed before transmitting a packet).
However, the DMA mapping may fail for the next packet, in which case a
new WQE is not posted, the doorbell isn't updated either, and the
transmission of the previous packet will be delayed indefinitely.
This commit fixes the described rare error flow by posting a NOP and
ringing the doorbell on errors to flush all the previous packets. The
MPWQE session is closed before that. DMA mapping in the MPWQE flow is
moved to the beginning of mlx5e_sq_xmit_mpwqe, because empty sessions
are not allowed. Stop room always has enough space for a NOP, because
the actual TX WQE is not posted.
Fixes: e586b3b0ba ("net/mlx5: Ethernet Datapath files")
Signed-off-by: Maxim Mikityanskiy <maximmi@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Most drivers use "skb_transport_offset(skb) + tcp_hdrlen(skb)"
to compute headers length for a TCP packet, but others
use more convoluted (but equivalent) ways.
Add skb_tcp_all_headers() and skb_inner_tcp_all_headers()
helpers to harmonize this a bit.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
mlx5 supports LSOv2.
IPv6 gro/tcp stacks insert a temporary Hop-by-Hop header
with JUMBO TLV for big packets.
We need to ignore/skip this HBH header when populating TX descriptor.
Note that ipv6_has_hopopt_jumbo() only recognizes very specific packet
layout, thus mlx5e_sq_xmit_wqe() is taking care of this layout only.
v7: adopt unsafe_memcpy() and MLX5_UNSAFE_MEMCPY_DISCLAIMER
v2: clear hopbyhop in mlx5e_tx_get_gso_ihs()
v4: fix compile error for CONFIG_MLX5_CORE_IPOIB=y
Signed-off-by: Coco Li <lixiaoyan@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Cc: Saeed Mahameed <saeedm@nvidia.com>
Cc: Leon Romanovsky <leon@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
As we continue to narrow the scope of what the FORTIFY memcpy() will
accept and build alternative APIs that give the compiler appropriate
visibility into more complex memcpy scenarios, there is a need for
"unfortified" memcpy use in rare cases where combinations of compiler
behaviors, source code layout, etc, result in cases where the stricter
memcpy checks need to be bypassed until appropriate solutions can be
developed (i.e. fix compiler bugs, code refactoring, new API, etc). The
intention is for this to be used only if there's no other reasonable
solution, for its use to include a justification that can be used
to assess future solutions, and for it to be temporary.
Example usage included, based on analysis and discussion from:
https://lore.kernel.org/netdev/CANn89iLS_2cshtuXPyNUGDPaic=sJiYfvTb_wNLgWrZRyBxZ_g@mail.gmail.com
Cc: Jakub Kicinski <kuba@kernel.org>
Cc: Eric Dumazet <edumazet@google.com>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Paolo Abeni <pabeni@redhat.com>
Cc: Coco Li <lixiaoyan@google.com>
Cc: Tariq Toukan <tariqt@nvidia.com>
Cc: Saeed Mahameed <saeedm@nvidia.com>
Cc: Leon Romanovsky <leon@kernel.org>
Cc: netdev@vger.kernel.org
Cc: linux-hardening@vger.kernel.org
Signed-off-by: Kees Cook <keescook@chromium.org>
Link: https://lore.kernel.org/r/20220511025301.3636666-1-keescook@chromium.org
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
This commit moves mlx5e_select_queue and all stuff related to
ndo_select_queue to en/selq.c to put all stuff working with selq into a
separate file.
Signed-off-by: Maxim Mikityanskiy <maximmi@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
This commit makes necessary changes to guarantee that txq2sq remains
stable while mlx5e_xmit is running. Proper synchronization is added for
HTB TX queues.
All updates to txq2sq are performed while the corresponding queue is
disabled (i.e. mlx5e_xmit doesn't run on that queue). smp_wmb after each
change guarantees that mlx5e_xmit can see the updated value after the
queue is enabled. Comments explaining this mechanism are added to
mlx5e_xmit.
When an HTB SQ can be deleted (after deleting an HTB node), synchronize
with RCU to wait for mlx5e_select_queue to finish and stop selecting
that queue, before we re-enable it to avoid TX timeout watchdog alarms.
Signed-off-by: Maxim Mikityanskiy <maximmi@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Calculate maximal count of MPW WQEBBs on SQ's creation and store it
there. Remove MLX5E_TX_MPW_MAX_NUM_DS and MLX5E_TX_MPW_MAX_WQEBBS.
Update mlx5e_tx_mpwqe_is_full() and mlx5e_xdp_mpqwe_is_full() .
Signed-off-by: Aya Levin <ayal@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
In preparation for FORTIFY_SOURCE performing compile-time and run-time
field bounds checking for memcpy(), memmove(), and memset(), avoid
intentionally writing across neighboring fields.
Use struct_group() in struct vlan_ethhdr around members h_dest and
h_source, so they can be referenced together. This will allow memcpy()
and sizeof() to more easily reason about sizes, improve readability,
and avoid future warnings about writing beyond the end of h_dest.
"pahole" shows no size nor member offset changes to struct vlan_ethhdr.
"objdump -d" shows no object code changes.
Fixes: 34802a42b3 ("net/mlx5e: Do not modify the TX SKB")
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Part of code that is related solely to IPsec is always compiled in the
driver code regardless if the IPsec functionality is enabled or disabled
in the driver code, this will add unnecessary branch in case IPsec is
disabled at Tx data path.
Move IPsec related code to IPsec related file such that in case of IPsec
is disabled and because of unlikely macro the compiler should be able to
optimize and omit the checksum IPsec code all together from Tx data path
Signed-off-by: Raed Salem <raeds@nvidia.com>
Reviewed-by: Emeel Hakim <ehakim@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Current Work Queue Entry (WQE) checksum (csum) flags in the ethernet
segment (eseg) in case of IPsec crypto offload datapath are not aligned
with PRM/HW expectations.
Currently the driver always sets the l3_inner_csum flag in case of IPsec
because of the wrong usage of skb->encapsulation as indicator for inner
IPsec header since skb->encapsulation is always ON for IPsec packets
since IPsec itself is an encapsulation protocol. The above forced a
failing attempts of calculating csum of non-existing segments (like in
the IP|ESP|TCP packet case which does not have an l3_inner) which led
to lots of packet drops hence the low throughput.
Fix by using xo->inner_ipproto as indicator for inner IPsec header
instead of skb->encapsulation in addition to setting the csum flags
as following:
* Tunnel Mode:
* Pkt: MAC IP ESP IP L4
* CSUM: l3_cs | l3_inner_cs | l4_inner_cs
*
* Transport Mode:
* Pkt: MAC IP ESP L4
* CSUM: l3_cs [ | l4_cs (checksum partial case)]
*
* Tunnel(VXLAN TCP/UDP) over Transport Mode
* Pkt: MAC IP ESP UDP VXLAN IP L4
* CSUM: l3_cs | l3_inner_cs | l4_inner_cs
Fixes: f1267798c9 ("net/mlx5: Fix checksum issue of VXLAN and IPsec crypto offload")
Signed-off-by: Emeel Hakim <ehakim@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.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>
Steering packets to PTP-SQ should be done only if the SKB has
SKBTX_HW_TSTAMP set in the tx_flags. While here, take the function into
a header and inline it.
Set the whole condition to select the PTP-SQ to unlikely.
Fixes: 24c22dd091 ("net/mlx5e: Add states to PTP channel")
Signed-off-by: Aya Levin <ayal@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Reviewed-by: Maxim Mikityanskiy <maximmi@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
After some commits, mlx5e_txwqe_build_eseg() lost its ability to return
boolean value and became effectively void.
Change its return type to void and remove unreachable branches.
Signed-off-by: Vladyslav Tarasiuk <vladyslavt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Add PTP TX state to PTP channel, which indicates the corresponding SQ is
available. Further patches in the set extend PTP channel to include RQ.
The PTP channel state will be used for separation and coexistence of RX
and TX PTP. Enhance conditions to verify the TX PTP state is set.
Signed-off-by: Aya Levin <ayal@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Following patches in the set add support for RX PTP. Rename PTP prefix
from %s/port_ptp/ptp/g to include RX PTP too.
In addition rename indication (used in statistics context) that PTP-SQ
was opened: %s/port_ptp_opened/tx_ptp_opened/g. This will simplify adding
indication that PTP-RQ was opened.
Signed-off-by: Aya Levin <ayal@nvidia.com>
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Commit e20f0dbf20 ("net/mlx5e: RX, Add a prefetch command for small
L1_CACHE_BYTES") switched to using net_prefetchw at all places in mlx5e.
In the same time frame, commit 5af75c747e ("net/mlx5e: Enhanced TX
MPWQE for SKBs") added one more usage of prefetchw. When these two
changes were merged, this new occurrence of prefetchw wasn't replaced
with net_prefetchw.
This commit fixes this last occurrence of prefetchw in
mlx5e_tx_mpwqe_session_start, making the same change that was done in
mlx5e_xdp_mpwqe_session_start.
Signed-off-by: Maxim Mikityanskiy <maximmi@mellanox.com>
Reviewed-by: Saeed Mahameed <saeedm@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Saeed Mahameed says:
====================
pull-request: mlx5-next 2021-02-16
The patches in this pr are already submitted and reviewed through the
netdev and rdma mailing lists.
The series includes mlx5 HW bits and definitions for mlx5 real time clock
translation and handling in the mlx5 driver clock module to enable and
support such mode [1]
[1] https://patchwork.kernel.org/project/netdevbpf/patch/20210212223042.449816-7-saeed@kernel.org/
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Device timestamp can be in real time mode (cycles to time translation is
offloaded into the Hardware). With real time mode, HW provides timestamp
which is already translated into nanoseconds.
With this mode, driver adjusts both the HW and timecounter (to keep
clock_info_page updated) using callbacks: adjfreq, adjtime and settime.
HW clock modifications are done via MTUTC access reg commands. Driver is
allowed to modify HW real time clock only if MCAM ptpcyc2realtime_modify
capability is set.
Add MTUTC set function to be used for configuring the HW real time
clock. Modify existing code to support both internal timer (with
conversion via timecounter_cyc2time() and real time (no conversions).
Align the signatures of the helpers converting from timestamp to
nanoseconds. With that, when allocating a queue assign the corresponding
callback with respect to the capability.
Adjust 1PPS timestamp calculation flows based on the timestamp mode.
Cyc2time offload brings two major advantages:
- Improve MTAE (Max Time Absolute Error) for HW TS by up to 160 ns over a
100% loaded CPU.
- Faster data-path timestamp to nanoseconds, as translation is
lock-less and done in HW.
On real time mode, timestamp format is 32 high bits of seconds and 32
low bits of nanoseconds. On some flows, driver shall convert this format
into nanoseconds wall-clock with REAL_TIME_TO_NS macro.
HW supports a single clock, and it is shared by all functions on a
device. In case real time clock is used, it is recommended to use
a single GM to all device's functions.
Signed-off-by: Eran Ben Elisha <eranbe@mellanox.com>
Signed-off-by: Aya Levin <ayal@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
This commit adds support for HTB offload in the mlx5e driver.
Performance:
NIC: Mellanox ConnectX-6 Dx
CPU: Intel(R) Xeon(R) CPU E5-2680 v3 @ 2.50GHz (24 cores with HT)
100 Gbit/s line rate, 500 UDP streams @ ~200 Mbit/s each
48 traffic classes, flower used for steering
No shaping (rate limits set to 4 Gbit/s per TC) - checking for max
throughput.
Baseline: 98.7 Gbps, 8.25 Mpps
HTB: 6.7 Gbps, 0.56 Mpps
HTB offload: 95.6 Gbps, 8.00 Mpps
Limitations:
1. 256 leaf nodes, 3 levels of depth.
2. Granularity for ceil is 1 Mbit/s. Rates are converted to weights, and
the bandwidth is split among the siblings according to these weights.
Other parameters for classes are not supported.
Ethtool statistics support for QoS SQs are also added. The counters are
called qos_txN_*, where N is the QoS queue number (starting from 0, the
numeration is separate from the normal SQs), and * is the counter name
(the counters are the same as for the normal SQs).
Signed-off-by: Maxim Mikityanskiy <maximmi@mellanox.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
All IPsec logic should be wrapped under the compile flag,
including its checksum logic.
Introduce an inline function in ipsec datapath header,
with a corresponding stub.
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
Reviewed-by: Raed Salem <raeds@nvidia.com>
Reviewed-by: Huy Nguyen <huyn@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
In case WQE includes inline header the vlan is inserted by driver even
if vlan offload is set. On geneve over vlan interface where software
parser is used the SWP offsets should be updated according to the added
vlan.
Fixes: e3cfc7e6b7 ("net/mlx5e: TX, Add geneve tunnel stateless offload support")
Signed-off-by: Moshe Shemesh <moshe@mellanox.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Transmitted packet timestamping accuracy can be improved when using
timestamp from the port, instead of packet CQE creation timestamp, as
it better reflects the actual time of a packet's transmit.
TX port timestamping is supported starting from ConnectX6-DX hardware.
Although at the original completion, only CQE timestamp can be attached,
we are able to get TX port timestamping via an additional completion over
a special CQ associated with the SQ (in addition to the regular CQ).
Driver to ignore the original packet completion timestamp, and report
back the timestamp of the special CQ completion. If the absolute timestamp
diff between the two completions is greater than 1 / 128 second, ignore
the TX port timestamp as it has a jitter which is too big.
No skb will be generate out of the extra completion.
Allocate additional CQ per ptpsq, to receive the TX port timestamp.
Driver to hold an skb FIFO in order to map between transmitted skb to
the two expected completions. When using ptpsq, hold double refcount on
the skb, to gaurantee it will not get released before both completions
arrive.
Expose dedicated counters of the ptp additional CQ and connect it to the
TX health reporter.
This patch improves TX Hardware timestamping offset to be less than 40ns
at a 100Gbps line rate, compared to 600ns before.
With that, making our HW compliant with G.8273.2 class C, and allow Linux
systems to be deployed in the 5G telco edge, where this standard is a
must.
Signed-off-by: Eran Ben Elisha <eranbe@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Add TX PTP port object support for better TX timestamping accuracy.
Currently, driver supports CQE based TX port timestamp. Device
also offers TX port timestamp, which has less jitter and better
reflects the actual time of a packet's transmit.
Define new driver layout called ptpsq, on which driver will create
SQs that will support TX port timestamp for their transmitted packets.
Driver to identify PTP TX skbs and steer them to these dedicated SQs
as part of the select queue ndo.
Driver to hold ptpsq per TC and report them at
netif_set_real_num_tx_queues().
Add support for all needed functionality in order to xmit and poll
completions received via ptpsq.
Add ptpsq to the TX reporter recover, diagnose and dump methods.
Creation of ptpsqs is disabled by default, and can be enabled via
tx_port_ts private flag.
This patch steer all timestamp related packets to a ptpsq, but it
does not open the port timestamp support for it. The support will
be added in the following patch.
Signed-off-by: Eran Ben Elisha <eranbe@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
The skb fifo push/pop API used pre-defined attributes within the
mlx5e_txqsq.
In order to share the skb fifo API with other non-SQ use cases,
change the API input to get newly defined mlx5e_skb_fifo struct.
Signed-off-by: Eran Ben Elisha <eranbe@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
In order to be able to create a CQ outside of a channel context, remove
cq->channel direct pointer. This requires adding a direct pointer to
channel statistics, netdevice, priv and to mlx5_core in order to support
CQs that are a part of mlx5e_channel.
In addition, parameters the were previously derived from the channel
like napi, NUMA node, channel stats and index are now assembled in
struct mlx5e_create_cq_param which is given to mlx5e_open_cq() instead
of channel pointer. Generalizing mlx5e_open_cq() allows opening CQ
outside of channel context which will be used in following patches in
the patch-set.
Signed-off-by: Aya Levin <ayal@nvidia.com>
Signed-off-by: Eran Ben Elisha <eranbe@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Checksum calculation cannot be done in SW for TX kTLS HW offloaded
packets.
Offload it to the device, disregard the declared state of the TX
csum offload feature.
Fixes: d2ead1f360 ("net/mlx5e: Add kTLS TX HW offload support")
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
Reviewed-by: Maxim Mikityanskiy <maximmi@mellanox.com>
Reviewed-by: Boris Pismenny <borisp@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The IP's checksum partial still requires L4 csum flag on Ethernet WQE.
Make the IPsec WAs only for the IP's non checksum partial case
(for example icmd packet)
Fixes: 5be019040c ("net/mlx5e: IPsec: Add Connect-X IPsec Tx data path offload")
Signed-off-by: Huy Nguyen <huyn@mellanox.com>
Reviewed-by: Raed Salem <raeds@nvidia.com>
Reviewed-by: Alaa Hleihel <alaa@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
This commit adds support for Enhanced TX MPWQE feature in the regular
(SKB) data path. A MPWQE (multi-packet work queue element) can serve
multiple packets, reducing the PCI bandwidth on control traffic.
Two new stats (tx*_mpwqe_blks and tx*_mpwqe_pkts) are added. The feature
is on by default and controlled by the skb_tx_mpwqe private flag.
In a MPWQE, eseg is shared among all packets, so eseg-based offloads
(IPSEC, GENEVE, checksum) run on a separate eseg that is compared to the
eseg of the current MPWQE session to decide if the new packet can be
added to the same session.
MPWQE is not compatible with certain offloads and features, such as TLS
offload, TSO, nonlinear SKBs. If such incompatible features are in use,
the driver gracefully falls back to non-MPWQE.
This change has no performance impact in TCP single stream test and
XDP_TX single stream test.
UDP pktgen, 64-byte packets, single stream, MPWQE off:
Packet rate: 16.96 Mpps (±0.12 Mpps) -> 17.01 Mpps (±0.20 Mpps)
Instructions per packet: 421 -> 429
Cycles per packet: 156 -> 161
Instructions per cycle: 2.70 -> 2.67
UDP pktgen, 64-byte packets, single stream, MPWQE on:
Packet rate: 16.96 Mpps (±0.12 Mpps) -> 20.94 Mpps (±0.33 Mpps)
Instructions per packet: 421 -> 329
Cycles per packet: 156 -> 123
Instructions per cycle: 2.70 -> 2.67
Enabling MPWQE can reduce PCI bandwidth:
PCI Gen2, pktgen at fixed rate of 36864000 pps on 24 CPU cores:
Inbound PCI utilization with MPWQE off: 80.3%
Inbound PCI utilization with MPWQE on: 59.0%
PCI Gen3, pktgen at fixed rate of 56064000 pps on 24 CPU cores:
Inbound PCI utilization with MPWQE off: 65.4%
Inbound PCI utilization with MPWQE on: 49.3%
Enabling MPWQE can also reduce CPU load, increasing the packet rate in
case of CPU bottleneck:
PCI Gen2, pktgen at full rate on 24 CPU cores:
Packet rate with MPWQE off: 37.5 Mpps
Packet rate with MPWQE on: 49.0 Mpps
PCI Gen3, pktgen at full rate on 24 CPU cores:
Packet rate with MPWQE off: 57.0 Mpps
Packet rate with MPWQE on: 66.8 Mpps
Burst size in all pktgen tests is 32.
CPU: Intel(R) Xeon(R) CPU E5-2680 v3 @ 2.50GHz (x86_64)
NIC: Mellanox ConnectX-6 Dx
GCC 10.2.0
Signed-off-by: Maxim Mikityanskiy <maximmi@mellanox.com>
Reviewed-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
mlx5e_txwqe_complete performs some actions that can be taken to separate
functions:
1. Update the flags needed for hardware timestamping.
2. Stop the TX queue if it's full.
Take these actions into separate functions to be reused by the MPWQE
code in the following commit and to maintain clear responsibilities of
functions.
Signed-off-by: Maxim Mikityanskiy <maximmi@mellanox.com>
Reviewed-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
TX MPWQE support for SKBs is coming in one of the following patches, and
a single MPWQE can send multiple SKBs. This commit prepares the TX path
code to handle such cases:
1. An additional FIFO for SKBs is added, just like the FIFO for DMA
chunks.
2. struct mlx5e_tx_wqe_info will contain num_fifo_pkts. If a given WQE
contains only one packet, num_fifo_pkts will be zero, and the SKB will
be stored in mlx5e_tx_wqe_info, as usual. If num_fifo_pkts > 0, the SKB
pointer will be NULL, and the SKBs will be stored in the FIFO.
This change has no performance impact in TCP single stream test and
XDP_TX single stream test.
When compiled with a recent GCC, this change shows no visible
performance impact on UDP pktgen (burst 32) single stream test either:
Packet rate: 16.95 Mpps (±0.15 Mpps) -> 16.96 Mpps (±0.12 Mpps)
Instructions per packet: 429 -> 421
Cycles per packet: 160 -> 156
Instructions per cycle: 2.69 -> 2.70
CPU: Intel(R) Xeon(R) CPU E5-2680 v3 @ 2.50GHz (x86_64)
NIC: Mellanox ConnectX-6 Dx
GCC 10.2.0
Signed-off-by: Maxim Mikityanskiy <maximmi@mellanox.com>
Reviewed-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Before this patch, mlx5e_ktls_tx_handle_resync_dump_comp checked for
resync_dump_frag_page. It happened for all WQEs without an SKB,
including padding WQEs, and required a function call. Normally, padding
WQEs happen more often than TLS resyncs. Take this check out of the
function and put it to an inline function to save a call on all padding
WQEs.
Signed-off-by: Maxim Mikityanskiy <maximmi@mellanox.com>
Reviewed-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
A constant for the number of DS in an empty WQE (i.e. a WQE without data
segments) is needed in multiple places (normal TX data path, MPWQE in
XDP), but currently we have a constant for XDP and an inline formula in
normal TX. This patch introduces a common constant.
Additionally, mlx5e_xdp_mpwqe_session_start is converted to use struct
assignment, because the code nearby is touched.
Signed-off-by: Maxim Mikityanskiy <maximmi@mellanox.com>
Reviewed-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
A huge function mlx5e_sq_xmit was split into several to achieve multiple
goals:
1. Reuse the code in IPoIB.
2. Better intergrate with TLS, IPSEC, GENEVE and checksum offloads. Now
it's possible to reserve space in the WQ before running eseg-based
offloads, so:
2.1. It's not needed to copy cseg and eseg after mlx5e_fill_sq_frag_edge
anymore.
2.2. mlx5e_txqsq_get_next_pi will be used instead of the legacy
mlx5e_fill_sq_frag_edge for better code maintainability and reuse.
3. Prepare for the upcoming TX MPWQE for SKBs. It will intervene after
mlx5e_sq_calc_wqe_attr to check if it's possible to use MPWQE, and the
code flow will split into two paths: MPWQE and non-MPWQE.
Two high-level functions are provided to send packets:
* mlx5e_xmit is called by the networking stack, runs offloads and sends
the packet. In one of the following patches, MPWQE support will be added
to this flow.
* mlx5e_sq_xmit_simple is called by the TLS offload, runs only the
checksum offload and sends the packet.
This change has no performance impact in TCP single stream test and
XDP_TX single stream test.
When compiled with a recent GCC, this change shows no visible
performance impact on UDP pktgen (burst 32) single stream test either:
Packet rate: 16.86 Mpps (±0.15 Mpps) -> 16.95 Mpps (±0.15 Mpps)
Instructions per packet: 434 -> 429
Cycles per packet: 158 -> 160
Instructions per cycle: 2.75 -> 2.69
CPU: Intel(R) Xeon(R) CPU E5-2680 v3 @ 2.50GHz (x86_64)
NIC: Mellanox ConnectX-6 Dx
GCC 10.2.0
Signed-off-by: Maxim Mikityanskiy <maximmi@mellanox.com>
Reviewed-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Move mlx5e_tx_wqe_inline_mode from en/txrx.h to en_tx.c as it's only
used there.
Signed-off-by: Maxim Mikityanskiy <maximmi@mellanox.com>
Reviewed-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Struct assignment guarantees that all fields of the structure are
initialized (those that are not mentioned are zeroed). It makes code
mode robust and reduces chances for unpredictable behavior when one
forgets to reset some field and it holds an old value from previous
iterations of using the structure.
Signed-off-by: Maxim Mikityanskiy <maximmi@mellanox.com>
Reviewed-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>