Cross-merge networking fixes after downstream PR (net-6.17-rc8).
Conflicts:
tools/testing/selftests/drivers/net/bonding/Makefile
87951b5664 selftests: bonding: add test for passive LACP mode
c2377f1763 selftests: bonding: add test for LACP actor port priority
Adjacent changes:
drivers/net/ethernet/cadence/macb.h
fca3dc859b net: macb: remove illusion about TBQPH/RBQPH being per-queue
89934dbf16 net: macb: Add TAPRIO traffic scheduling support
drivers/net/ethernet/cadence/macb_main.c
fca3dc859b net: macb: remove illusion about TBQPH/RBQPH being per-queue
89934dbf16 net: macb: Add TAPRIO traffic scheduling support
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
The mailbox receive path allocates coherent DMA memory with
dma_alloc_coherent(), but frees it with dmam_free_coherent().
This is incorrect since dmam_free_coherent() is only valid for
buffers allocated with dmam_alloc_coherent().
Fix the mismatch by using dma_free_coherent() instead of
dmam_free_coherent
Fixes: e54232da12 ("idpf: refactor idpf_recv_mb_msg")
Signed-off-by: Alok Tiwari <alok.a.tiwari@oracle.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Reviewed-by: Aleksandr Loktionov <aleksandr.loktionov@intel.com>
Reviewed-by: Jacob Keller <jacob.e.keller@intel.com>
Reviewed-by: Madhu Chittim <madhu.chittim@intel.com>
Link: https://patch.msgid.link/20250925180212.415093-1-alok.a.tiwari@oracle.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Now that AF_XDP functionality is fully implemented, advertise XSk XDP
feature and add .ndo_xsk_wakeup() callback to be able to use it with
this driver.
Co-developed-by: Michal Kubiak <michal.kubiak@intel.com>
Signed-off-by: Michal Kubiak <michal.kubiak@intel.com>
Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Tested-by: Ramu R <ramu.r@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Implement Rx packet processing specific to AF_XDP ZC using the libeth
XSk infra. Initialize queue registers before allocating buffers to
avoid redundant ifs when updating the queue tail.
Co-developed-by: Michal Kubiak <michal.kubiak@intel.com>
Signed-off-by: Michal Kubiak <michal.kubiak@intel.com>
Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Tested-by: Ramu R <ramu.r@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Implement the XSk transmit path using the libeth (libeth_xdp)
XSk infra.
When the NAPI poll is called, XSk Tx queues are polled first,
before regular Tx and Rx. They're generally faster to serve
and have higher priority comparing to regular traffic.
Co-developed-by: Michal Kubiak <michal.kubiak@intel.com>
Signed-off-by: Michal Kubiak <michal.kubiak@intel.com>
Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Tested-by: Ramu R <ramu.r@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Add functionality to setup an XSk buffer pool, including ability to
stop, reconfig and start only selected queues, not the whole device.
Pool DMA mapping is managed by libeth_xdp.
Signed-off-by: Michal Kubiak <michal.kubiak@intel.com>
Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Tested-by: Ramu R <ramu.r@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Implement VC functions dedicated to enabling, disabling and configuring
not all but only selected queues.
Also, refactor the existing implementation to make the code more
modular. Introduce new generic functions for sending VC messages
consisting of chunks, in order to isolate the sending algorithm
and its implementation for specific VC messages.
Finally, rewrite the function for mapping queues to q_vectors using the
new modular approach to avoid copying the code that implements the VC
message sending algorithm.
Signed-off-by: Michal Kubiak <michal.kubiak@intel.com>
Co-developed-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Tested-by: Ramu R <ramu.r@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Add HW timestamping statistics support - through implementing get_ts_stats.
Timestamp statistics include correctly timestamped packets, discarded,
skipped and flushed during PTP release.
Most of the stats are collected per vport, only requests skipped due to
lack of free latch index are collected per Tx queue.
Statistics can be obtained using kernel ethtool since version 6.10
with:
ethtool -I -T <interface>
The output will include:
Statistics:
tx_pkts: 15
tx_lost: 0
tx_err: 0
Signed-off-by: Milena Olech <milena.olech@intel.com>
Co-developed-by: Anton Nadezhdin <anton.nadezhdin@intel.com>
Signed-off-by: Anton Nadezhdin <anton.nadezhdin@intel.com>
Reviewed-by: Aleksandr Loktionov <aleksandr.loktionov@intel.com>
Reviewed-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Samuel Salin <Samuel.salin@intel.com>
Reviewed-by: Paul Menzel <pmenzel@molgen.mpg.de>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Add &xdp_metadata_ops with a callback to get RSS hash hint from the
descriptor. Declare the splitq 32-byte descriptor as 4 u64s to parse
them more efficiently when possible.
Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Tested-by: Ramu R <ramu.r@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Use libeth XDP infra to implement .ndo_xdp_xmit() in idpf.
The Tx callbacks are reused from XDP_TX code. XDP redirect target
feature is set/cleared depending on the XDP prog presence, as for now
we still don't allocate XDP Tx queues when there's no program.
Reviewed-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Tested-by: Ramu R <ramu.r@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Use libeth XDP infra to support running XDP program on Rx polling.
This includes all of the possible verdicts/actions.
XDP Tx queues are cleaned only in "lazy" mode when there are less than
1/4 free descriptors left on the ring. libeth helper macros to define
driver-specific XDP functions make sure the compiler could uninline
them when needed.
Use __LIBETH_WORD_ACCESS to parse descriptors more efficiently when
applicable. It really gives some good boosts and code size reduction
on x86_64:
XDP only: add/remove: 0/0 grow/shrink: 3/3 up/down: 5/-59 (-54)
with XSk: add/remove: 0/0 grow/shrink: 5/6 up/down: 23/-124 (-101)
with the most demanding workloads like XSk xmit differing in up to 5-8%.
Co-developed-by: Michal Kubiak <michal.kubiak@intel.com>
Signed-off-by: Michal Kubiak <michal.kubiak@intel.com>
Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Tested-by: Ramu R <ramu.r@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
In preparation of XDP support, move from having skb as the main frame
container during the Rx polling to &xdp_buff.
This allows to use generic and libeth helpers for building an XDP
buffer and changes the logics: now we try to allocate an skb only
when we processed all the descriptors related to the frame.
Store &libeth_xdp_stash instead of the skb pointer on the Rx queue.
It's only 8 bytes wider, but contains everything we may need.
Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Tested-by: Ramu R <ramu.r@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Implement loading/removing XDP program using .ndo_bpf callback
in the split queue mode. Reconfigure and restart the queues if needed
(!!old_prog != !!new_prog), otherwise, just update the pointers.
Signed-off-by: Michal Kubiak <michal.kubiak@intel.com>
Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Tested-by: Ramu R <ramu.r@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Extend basic structures of the driver (e.g. 'idpf_vport', 'idpf_*_queue',
'idpf_vport_user_config_data') by adding members necessary to support XDP.
Add extra XDP Tx queues needed to support XDP_TX and XDP_REDIRECT actions
without interfering with regular Tx traffic.
Also add functions dedicated to support XDP initialization for Rx and
Tx queues and call those functions from the existing algorithms of
queues configuration.
Signed-off-by: Michal Kubiak <michal.kubiak@intel.com>
Co-developed-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Tested-by: Ramu R <ramu.r@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Currently, queues are associated 1:1 with interrupt vectors as it's
assumed queues are always interrupt-driven. For XDP, we want to use
Tx queues without interrupts and only do "lazy" cleaning when the number
of free elements is <= threshold (closest pow-2 to 1/4 of the ring).
In order to use a queue without an interrupt, idpf still needs to have
a vector assigned to it to flush descriptors. This vector can be global
and only one for the whole vport to handle all its noirq queues.
Always request one excessive vector and configure it in non-interrupt
mode right away when creating vport, so that it can be used later by
queues when needed (not only XDP ones).
Co-developed-by: Michal Kubiak <michal.kubiak@intel.com>
Signed-off-by: Michal Kubiak <michal.kubiak@intel.com>
Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Tested-by: Ramu R <ramu.r@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
SW marker descriptors on completion queues are used only when a queue
is about to be destroyed. It's far from hotpath and handling it in the
hotpath NAPI poll makes no sense.
Instead, run a simple poller after a virtchnl message for destroying
the queue is sent and wait for the replies. If replies for all of the
queues are received, this means the synchronization is done correctly
and we can go forth with stopping the link.
Signed-off-by: Michal Kubiak <michal.kubiak@intel.com>
Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Tested-by: Ramu R <ramu.r@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
In the queue-based scheduling mode, Tx completion descriptor is 4 bytes
comparing to 8 bytes in flow-based.
Add definition for it and allocate the corresponding amount of memory
for the descriptors during the completion queue creation.
This does not include handling 4-byte completions during Tx polling, as
for now, the only user of QB will be XDP, which has its own routines.
Signed-off-by: Michal Kubiak <michal.kubiak@intel.com>
Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Tested-by: Ramu R <ramu.r@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Add the missing linking of NAPIs to netdev queues when enabling
interrupt vectors in order to support NAPI configuration and
interfaces requiring get_rx_queue()->napi to be set (like XSk
busy polling).
As currently, idpf_vport_{start,stop}() is called from several flows
with inconsistent RTNL locking, we need to synchronize them to avoid
runtime assertions. Notably:
* idpf_{open,stop}() -- regular NDOs, RTNL is always taken;
* idpf_initiate_soft_reset() -- usually called under RTNL;
* idpf_init_task -- called from the init work, needs RTNL;
* idpf_vport_dealloc -- called without RTNL taken, needs it.
Expand common idpf_vport_{start,stop}() to take an additional bool
telling whether we need to manually take the RTNL lock.
Suggested-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com> # helper
Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Tested-by: Ramu R <ramu.r@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Currently, the maximum number of queues available for one vport is 16.
This is hardcoded, but then the function calculating the optimal number
of queues takes min(16, num_online_cpus()).
In order to be able to allocate more queues, which will be then used for
XDP, stop hardcoding 16 and rely on what the device gives us[*]. Instead
of num_online_cpus(), which is considered suboptimal since at least 2013,
use netif_get_num_default_rss_queues() to still have free queues in the
pool.
[*] With the note:
Currently, idpf always allocates `IDPF_MAX_BUFQS_PER_RXQ_GRP` (== 2)
buffer queues for each Rx queue and one completion queue for each Tx for
best performance. But there was no check whether such number is available,
IOW the assumption was not backed by any "harmonizing" / actual checks.
Fix this while at it.
nr_cpu_ids number of Tx queues are needed only for lockless XDP sending,
the regular stack doesn't benefit from that anyhow.
On a 128-thread Xeon, this now gives me 32 regular Tx queues and leaves
224 free for XDP (128 of which will handle XDP_TX, .ndo_xdp_xmit(), and
XSk xmit when enabled).
Note 2:
Unfortunately, some CP/FW versions are not able to
reconfigure/enable/disable large amount of queues within the minimum
timeout (2 seconds). For now, fall back to the default timeout for
every operation until this is resolved.
Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Tested-by: Ramu R <ramu.r@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
No idea what the current barrier position was meant for. At that point,
nothing is read from the descriptor, only the pointer to the actual one
is fetched.
The correct barrier usage here is after the generation check, so that
only the first qword is read if the descriptor is not yet ready and we
need to stop polling. Debatable on coherent DMA as the Rx descriptor
size is <= cacheline size, but anyway, the current barrier position
only makes the codegen worse.
Fixes: 3a8845af66 ("idpf: add RX splitq napi poll support")
Reviewed-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Tested-by: Ramu R <ramu.r@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
On control planes that allow changing the MAC address of the interface,
the driver must provide a MAC type to avoid errors such as:
idpf 0000:0a:00.0: Transaction failed (op 535)
idpf 0000:0a:00.0: Received invalid MAC filter payload (op 535) (len 0)
idpf 0000:0a:00.0: Transaction failed (op 536)
These errors occur during driver load or when changing the MAC via:
ip link set <iface> address <mac>
Add logic to set the MAC type when sending ADD/DEL (opcodes 535/536) to
the control plane. Since only one primary MAC is supported per vport, the
driver only needs to send an ADD opcode when setting it. Remove the old
address by calling __idpf_del_mac_filter(), which skips the message and
just clears the entry from the internal list. This avoids an error on DEL
as it attempts to remove an address already cleared by the preceding ADD
opcode.
Fixes: ce1b75d063 ("idpf: add ptypes and MAC filter support")
Reported-by: Jian Liu <jianliu@redhat.com>
Signed-off-by: Emil Tantilov <emil.s.tantilov@intel.com>
Reviewed-by: Aleksandr Loktionov <aleksandr.loktionov@intel.com>
Reviewed-by: Paul Menzel <pmenzel@molgen.mpg.de>
Reviewed-by: Simon Horman <horms@kernel.org>
Tested-by: Samuel Salin <Samuel.salin@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Cross-merge networking fixes after downstream PR (net-6.17-rc4).
No conflicts.
Adjacent changes:
drivers/net/ethernet/intel/idpf/idpf_txrx.c
02614eee26 ("idpf: do not linearize big TSO packets")
6c4e684802 ("idpf: remove obsolete stashing code")
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
idpf has a limit on number of scatter-gather frags
that can be used per segment.
Currently, idpf_tx_start() checks if the limit is hit
and forces a linearization of the whole packet.
This requires high order allocations that can fail
under memory pressure. A full size BIG-TCP packet
would require order-7 alocation on x86_64 :/
We can move the check earlier from idpf_features_check()
for TSO packets, to force GSO in this case, removing the
cost of a big copy.
This means that a linearization will eventually happen
with sizes smaller than one MSS.
__idpf_chk_linearize() is renamed to idpf_chk_tso_segment()
and moved to idpf_lib.c
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Cc: Jacob Keller <jacob.e.keller@intel.com>
Cc: Madhu Chittim <madhu.chittim@intel.com>
Cc: Pavan Kumar Linga <pavan.kumar.linga@intel.com>
Cc: Willem de Bruijn <willemb@google.com>
Cc: Andrew Lunn <andrew+netdev@lunn.ch>
Reviewed-by: Joshua Hay <joshua.a.hay@intel.com>
Tested-by: Brian Vazquez <brianvv@google.com>
Acked-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Link: https://patch.msgid.link/20250818195934.757936-1-edumazet@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
With the new Tx buffer management scheme, there is no need for all of
the stashing mechanisms, the hash table, the reserve buffer stack, etc.
Remove all of that.
Signed-off-by: Joshua Hay <joshua.a.hay@intel.com>
Reviewed-by: Madhu Chittim <madhu.chittim@intel.com>
Reviewed-by: Aleksandr Loktionov <aleksandr.loktionov@intel.com>
Tested-by: Samuel Salin <Samuel.salin@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
The Tx refillq logic will cause packets to be silently dropped if there
are not enough buffer resources available to send a packet in flow
scheduling mode. Instead, determine how many buffers are needed along
with number of descriptors. Make sure there are enough of both resources
to send the packet, and stop the queue if not.
Fixes: 7292af042b ("idpf: fix a race in txq wakeup")
Signed-off-by: Joshua Hay <joshua.a.hay@intel.com>
Reviewed-by: Madhu Chittim <madhu.chittim@intel.com>
Tested-by: Samuel Salin <Samuel.salin@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Replace the TxQ buffer ring with one large pool/array of buffers (only
for flow scheduling). This eliminates the tag generation and makes it
impossible for a tag to be associated with more than one packet.
The completion tag passed to HW through the descriptor is the index into
the array. That same completion tag is posted back to the driver in the
completion descriptor, and used to index into the array to quickly
retrieve the buffer during cleaning. In this way, the tags are treated
as a fix sized resource. If all tags are in use, no more packets can be
sent on that particular queue (until some are freed up). The tag pool
size is 64K since the completion tag width is 16 bits.
For each packet, the driver pulls a free tag from the refillq to get the
next free buffer index. When cleaning is complete, the tag is posted
back to the refillq. A multi-frag packet spans multiple buffers in the
driver, therefore it uses multiple buffer indexes/tags from the pool.
Each frag pulls from the refillq to get the next free buffer index.
These are tracked in a next_buf field that replaces the completion tag
field in the buffer struct. This chains the buffers together so that the
packet can be cleaned from the starting completion tag taken from the
completion descriptor, then from the next_buf field for each subsequent
buffer.
In case of a dma_mapping_error occurs or the refillq runs out of free
buf_ids, the packet will execute the rollback error path. This unmaps
any buffers previously mapped for the packet. Since several free
buf_ids could have already been pulled from the refillq, we need to
restore its original state as well. Otherwise, the buf_ids/tags
will be leaked and not used again until the queue is reallocated.
Descriptor completions only advance the descriptor ring index to "clean"
the descriptors. The packet completions only clean the buffers
associated with the given packet completion tag and do not update the
descriptor ring index.
When operating in queue based scheduling mode, the array still acts as a
ring and will only have TxQ descriptor count entries. The tx_bufs are
still associated 1:1 with the descriptor ring entries and we can use the
conventional indexing mechanisms.
Fixes: c2d548cad1 ("idpf: add TX splitq napi poll support")
Signed-off-by: Luigi Rizzo <lrizzo@google.com>
Signed-off-by: Brian Vazquez <brianvv@google.com>
Signed-off-by: Joshua Hay <joshua.a.hay@intel.com>
Reviewed-by: Madhu Chittim <madhu.chittim@intel.com>
Reviewed-by: Aleksandr Loktionov <aleksandr.loktionov@intel.com>
Tested-by: Samuel Salin <Samuel.salin@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Move (and rename) the existing rollback logic to singleq.c since that
will be the only consumer. Create a simplified splitq specific rollback
function to loop through and unmap tx_bufs based on the completion tag.
This is critical before replacing the Tx buffer ring with the buffer
pool since the previous rollback indexing will not work to unmap the
chained buffers from the pool.
Cache the next_to_use index before any portion of the packet is put on
the descriptor ring. In case of an error, the rollback will bump tail to
the correct next_to_use value. Because the splitq path now supports
different types of context descriptors (and potentially multiple in the
future), this will take care of rolling back any and all context
descriptors encoded on the ring for the erroneous packet. The previous
rollback logic was broken for PTP packets since it would not account for
the PTP context descriptor.
Fixes: 1a49cf814f ("idpf: add Tx timestamp flows")
Signed-off-by: Joshua Hay <joshua.a.hay@intel.com>
Reviewed-by: Madhu Chittim <madhu.chittim@intel.com>
Tested-by: Samuel Salin <Samuel.salin@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Track the gap between next_to_use and the last RE index. Set RE again
if the gap is large enough to ensure RE bit is set frequently. This is
critical before removing the stashing mechanisms because the
opportunistic descriptor ring cleaning from the out-of-order completions
will go away. Previously the descriptors would be "cleaned" by both the
descriptor (RE) completion and the out-of-order completions. Without the
latter, we must ensure the RE bit is set more frequently. Otherwise,
it's theoretically possible for the descriptor ring next_to_clean to
never advance. The previous implementation was dependent on the start
of a packet falling on a 64th index in the descriptor ring, which is not
guaranteed with large packets.
Signed-off-by: Luigi Rizzo <lrizzo@google.com>
Signed-off-by: Brian Vazquez <brianvv@google.com>
Signed-off-by: Joshua Hay <joshua.a.hay@intel.com>
Reviewed-by: Madhu Chittim <madhu.chittim@intel.com>
Tested-by: Samuel Salin <Samuel.salin@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
In certain production environments, it is possible for completion tags
to collide, meaning N packets with the same completion tag are in flight
at the same time. In this environment, any given Tx queue is effectively
used to send both slower traffic and higher throughput traffic
simultaneously. This is the result of a customer's specific
configuration in the device pipeline, the details of which Intel cannot
provide. This configuration results in a small number of out-of-order
completions, i.e., a small number of packets in flight. The existing
guardrails in the driver only protect against a large number of packets
in flight. The slower flow completions are delayed which causes the
out-of-order completions. The fast flow will continue sending traffic
and generating tags. Because tags are generated on the fly, the fast
flow eventually uses the same tag for a packet that is still in flight
from the slower flow. The driver has no idea which packet it should
clean when it processes the completion with that tag, but it will look
for the packet on the buffer ring before the hash table. If the slower
flow packet completion is processed first, it will end up cleaning the
fast flow packet on the ring prematurely. This leaves the descriptor
ring in a bad state resulting in a crash or Tx timeout.
In summary, generating a tag when a packet is sent can lead to the same
tag being associated with multiple packets. This can lead to resource
leaks, crashes, and/or Tx timeouts.
Before we can replace the tag generation, we need a new mechanism for
the send path to know what tag to use next. The driver will allocate and
initialize a refillq for each TxQ with all of the possible free tag
values. During send, the driver grabs the next free tag from the refillq
from next_to_clean. While cleaning the packet, the clean routine posts
the tag back to the refillq's next_to_use to indicate that it is now
free to use.
This mechanism works exactly the same way as the existing Rx refill
queues, which post the cleaned buffer IDs back to the buffer queue to be
reposted to HW. Since we're using the refillqs for both Rx and Tx now,
genericize some of the existing refillq support.
Note: the refillqs will not be used yet. This is only demonstrating how
they will be used to pass free tags back to the send path.
Signed-off-by: Joshua Hay <joshua.a.hay@intel.com>
Reviewed-by: Madhu Chittim <madhu.chittim@intel.com>
Tested-by: Samuel Salin <Samuel.salin@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
To eliminate the use of struct page in page pool, the page pool users
should use netmem descriptor and APIs instead.
Make idpf access ->pp through netmem_desc instead of page.
Signed-off-by: Byungchul Park <byungchul@sk.com>
Link: https://patch.msgid.link/20250721021835.63939-10-byungchul@sk.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The IRQ coalescing config currently reside only inside struct
idpf_q_vector. However, all idpf_q_vector structs are de-allocated and
re-allocated during resets. This leads to user-set coalesce configuration
to be lost.
Add new fields to struct idpf_vport_user_config_data to save the user
settings and re-apply them after reset.
Reviewed-by: Madhu Chittim <madhu.chittim@intel.com>
Signed-off-by: Ahmed Zaki <ahmed.zaki@intel.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Reviewed-by: Willem de Bruijn <willemb@google.com>
Tested-by: Samuel Salin <Samuel.salin@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Add cross timestamp support through virtchnl mailbox messages and directly,
through PCIe BAR registers. Cross timestamping assumes that both system
time and device clock time values are cached simultaneously, what is
triggered by HW. Feature is enabled for both ARM and x86 archs.
Signed-off-by: Milena Olech <milena.olech@intel.com>
Reviewed-by: Karol Kolacinski <karol.kolacinski@intel.com>
Reviewed-by: Willem de Bruijn <willemb@google.com>
Tested-by: Samuel Salin <Samuel.salin@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Use the new virtchnl2 OP codes to communicate with the Control Plane to
add flow steering filters. We add the basic functionality for add/delete
with TCP/UDP IPv4 only. Support for other OP codes and protocols will be
added later.
Standard 'ethtool -N|--config-ntuple' should be used, for example:
# ethtool -N ens801f0d1 flow-type tcp4 src-ip 10.0.0.1 action 6
to route all IPv4/TCP traffic from IP 10.0.0.1 to queue 6.
Reviewed-by: Sridhar Samudrala <sridhar.samudrala@intel.com>
Signed-off-by: Ahmed Zaki <ahmed.zaki@intel.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Add opcodes and corresponding message structure to add and delete
flow steering rules. Flow steering enables configuration
of rules to take an action or subset of actions based on a match
criteria. Actions could be redirect to queue, redirect to queue
group, drop packet or mark.
Reviewed-by: Aleksandr Loktionov <aleksandr.loktionov@intel.com>
Reviewed-by: Sridhar Samudrala <sridhar.samudrala@intel.com>
Co-developed-by: Dinesh Kumar <dinesh.kumar@intel.com>
Signed-off-by: Dinesh Kumar <dinesh.kumar@intel.com>
Signed-off-by: Sudheer Mogilappagari <sudheer.mogilappagari@intel.com>
Signed-off-by: Ahmed Zaki <ahmed.zaki@intel.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
The "enum virtchnl2_cap_rss" will be used for negotiating flow
steering capabilities. Instead of adding a new enum, rename
virtchnl2_cap_rss to virtchnl2_flow_types. Also rename the enum's
constants.
Flow steering will use this enum in the next patches.
Reviewed-by: Sridhar Samudrala <sridhar.samudrala@intel.com>
Signed-off-by: Ahmed Zaki <ahmed.zaki@intel.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Tony Nguyen says:
====================
Add RDMA support for Intel IPU E2000 in idpf
Tatyana Nikolova says:
This idpf patch series is the second part of the staged submission for
introducing RDMA RoCEv2 support for the IPU E2000 line of products,
referred to as GEN3.
To support RDMA GEN3 devices, the idpf driver uses common definitions
of the IIDC interface and implements specific device functionality in
iidc_rdma_idpf.h.
The IPU model can host one or more logical network endpoints called
vPorts per PCI function that are flexibly associated with a physical
port or an internal communication port.
Other features as it pertains to GEN3 devices include:
* MMIO learning
* RDMA capability negotiation
* RDMA vectors discovery between idpf and control plane
These patches are split from the submission "Add RDMA support for Intel
IPU E2000 (GEN3)" [1]. The patches have been tested on a range of hosts
and platforms with a variety of general RDMA applications which include
standalone verbs (rping, perftest, etc.), storage and HPC applications.
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
[1] https://lore.kernel.org/all/20240724233917.704-1-tatyana.e.nikolova@intel.com/
This idpf patch series is the second part of the staged submission for
introducing RDMA RoCEv2 support for the IPU E2000 line of products,
referred to as GEN3.
To support RDMA GEN3 devices, the idpf driver uses common definitions
of the IIDC interface and implements specific device functionality in
iidc_rdma_idpf.h.
The IPU model can host one or more logical network endpoints called
vPorts per PCI function that are flexibly associated with a physical
port or an internal communication port.
Other features as it pertains to GEN3 devices include:
* MMIO learning
* RDMA capability negotiation
* RDMA vectors discovery between idpf and control plane
These patches are split from the submission "Add RDMA support for Intel
IPU E2000 (GEN3)" [1]. The patches have been tested on a range of hosts
and platforms with a variety of general RDMA applications which include
standalone verbs (rping, perftest, etc.), storage and HPC applications.
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
[1] https://lore.kernel.org/all/20240724233917.704-1-tatyana.e.nikolova@intel.com/
IWL reviews:
v3: https://lore.kernel.org/all/20250708210554.1662-1-tatyana.e.nikolova@intel.com/
v2: https://lore.kernel.org/all/20250612220002.1120-1-tatyana.e.nikolova@intel.com/
v1 (split from previous series):
https://lore.kernel.org/all/20250523170435.668-1-tatyana.e.nikolova@intel.com/
v3: https://lore.kernel.org/all/20250207194931.1569-1-tatyana.e.nikolova@intel.com/
RFC v2: https://lore.kernel.org/all/20240824031924.421-1-tatyana.e.nikolova@intel.com/
RFC: https://lore.kernel.org/all/20240724233917.704-1-tatyana.e.nikolova@intel.com/
* 'for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/linux:
idpf: implement get LAN MMIO memory regions
idpf: implement IDC vport aux driver MTU change handler
idpf: implement remaining IDC RDMA core callbacks and handlers
idpf: implement RDMA vport auxiliary dev create, init, and destroy
idpf: implement core RDMA auxiliary dev create, init, and destroy
idpf: use reserved RDMA vectors from control plane
====================
Link: https://patch.msgid.link/20250714181002.2865694-1-anthony.l.nguyen@intel.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
The RDMA driver needs to map its own MMIO regions for the sake of
performance, meaning the IDPF needs to avoid mapping portions of the BAR
space. However, to be HW agnostic, the IDPF cannot assume where
these are and must avoid mapping hard coded regions as much as possible.
The IDPF maps the bare minimum to load and communicate with the
control plane, i.e., the mailbox registers and the reset state
registers. Because of how and when mailbox register offsets are
initialized, it is easier to adjust the existing defines to be relative
to the mailbox region starting address. Use a specific mailbox register
write function that uses these relative offsets. The reset state
register addresses are calculated the same way as for other registers,
described below.
The IDPF then calls a new virtchnl op to fetch a list of MMIO regions
that it should map. The addresses for the registers in these regions are
calculated by determining what region the register resides in, adjusting
the offset to be relative to that region, and then adding the
register's offset to that region's mapped address.
If the new virtchnl op is not supported, the IDPF will fallback to
mapping the whole bar. However, it will still map them as separate
regions outside the mailbox and reset state registers. This way we can
use the same logic in both cases to access the MMIO space.
Reviewed-by: Madhu Chittim <madhu.chittim@intel.com>
Signed-off-by: Joshua Hay <joshua.a.hay@intel.com>
Signed-off-by: Tatyana Nikolova <tatyana.e.nikolova@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
The only event an RDMA vport aux driver cares about right now is an MTU
change on its underlying vport. Implement and plumb the handler to
signal the pre MTU change event and post MTU change events to the RDMA
vport aux driver.
Reviewed-by: Madhu Chittim <madhu.chittim@intel.com>
Signed-off-by: Joshua Hay <joshua.a.hay@intel.com>
Signed-off-by: Tatyana Nikolova <tatyana.e.nikolova@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Implement the idpf_idc_request_reset and idpf_idc_rdma_vc_send_sync
callbacks for the rdma core auxiliary driver to issue reset events to
the idpf and send (synchronous) virtchnl messages to the control plane
respectively.
Implement and plumb the reset handler for the opposite flow as well,
i.e. when the idpf is resetiing and needs to notify the rdma core
auxiliary driver.
Reviewed-by: Madhu Chittim <madhu.chittim@intel.com>
Signed-off-by: Joshua Hay <joshua.a.hay@intel.com>
Signed-off-by: Tatyana Nikolova <tatyana.e.nikolova@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Implement the functions to create, initialize, and destroy an RDMA vport
auxiliary device. The vport aux dev creation is dependent on the
core aux device to call idpf_idc_vport_dev_ctrl to signal that it is
ready for vport aux devices. Implement that core callback to either
create and initialize the vport aux dev or deinitialize.
RDMA vport aux dev creation is also dependent on the control plane to
tell us the vport is RDMA enabled. Add a flag in the create vport
message to signal individual vport RDMA capabilities.
Reviewed-by: Madhu Chittim <madhu.chittim@intel.com>
Signed-off-by: Joshua Hay <joshua.a.hay@intel.com>
Signed-off-by: Tatyana Nikolova <tatyana.e.nikolova@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Add the initial idpf_idc.c file with the functions to kick off the IDC
initialization, create and initialize a core RDMA auxiliary device, and
destroy said device.
The RDMA core has a dependency on the vports being created by the
control plane before it can be initialized. Therefore, once all the
vports are up after a hard reset (either during driver load a function
level reset), the core RDMA device info will be created. It is populated
with the function type (as distinguished by the IDC initialization
function pointer), the core idc_ops function points (just stubs for
now), the reserved RDMA MSIX table, and various other info the core RDMA
auxiliary driver will need. It is then plugged on to the bus.
During a function level reset or driver unload, the device will be
unplugged from the bus and destroyed.
Reviewed-by: Madhu Chittim <madhu.chittim@intel.com>
Signed-off-by: Joshua Hay <joshua.a.hay@intel.com>
Signed-off-by: Tatyana Nikolova <tatyana.e.nikolova@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Fetch the number of reserved RDMA vectors from the control plane.
Adjust the number of reserved LAN vectors if necessary. Adjust the
minimum number of vectors the OS should reserve to include RDMA; and
fail if the OS cannot reserve enough vectors for the minimum number of
LAN and RDMA vectors required. Create a separate msix table for the
reserved RDMA vectors, which will just get handed off to the RDMA core
device to do with what it will.
Reviewed-by: Madhu Chittim <madhu.chittim@intel.com>
Signed-off-by: Joshua Hay <joshua.a.hay@intel.com>
Signed-off-by: Tatyana Nikolova <tatyana.e.nikolova@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Back when the libeth Rx core was initially written, devmem was a draft
and netmem_ref didn't exist in the mainline. Now that it's here, make
libeth MP-agnostic before introducing any new code or any new library
users.
When it's known that the created PP/FQ is for header buffers, use faster
"unsafe" underscored netmem <--> virt accessors as netmem_is_net_iov()
is always false in that case, but consumes some cycles (bit test +
true branch).
Reviewed-by: Mina Almasry <almasrymina@google.com>
Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Mailbox operations are not possible while the driver is in reset.
Operations that require MBX exchange with the control plane will result
in long delays if executed while a reset is in progress:
ethtool -L <inf> combined 8& echo 1 > /sys/class/net/<inf>/device/reset
idpf 0000:83:00.0: HW reset detected
idpf 0000:83:00.0: Device HW Reset initiated
idpf 0000:83:00.0: Transaction timed-out (op:504 cookie:be00 vc_op:504 salt:be timeout:2000ms)
idpf 0000:83:00.0: Transaction timed-out (op:508 cookie:bf00 vc_op:508 salt:bf timeout:2000ms)
idpf 0000:83:00.0: Transaction timed-out (op:512 cookie:c000 vc_op:512 salt:c0 timeout:2000ms)
idpf 0000:83:00.0: Transaction timed-out (op:510 cookie:c100 vc_op:510 salt:c1 timeout:2000ms)
idpf 0000:83:00.0: Transaction timed-out (op:509 cookie:c200 vc_op:509 salt:c2 timeout:60000ms)
idpf 0000:83:00.0: Transaction timed-out (op:509 cookie:c300 vc_op:509 salt:c3 timeout:60000ms)
idpf 0000:83:00.0: Transaction timed-out (op:505 cookie:c400 vc_op:505 salt:c4 timeout:60000ms)
idpf 0000:83:00.0: Failed to configure queues for vport 0, -62
Disable mailbox communication in case of a reset, unless it's done during
a driver load, where the virtchnl operations are needed to configure the
device.
Fixes: 8077c72756 ("idpf: add controlq init and reset checks")
Co-developed-by: Joshua Hay <joshua.a.hay@intel.com>
Signed-off-by: Joshua Hay <joshua.a.hay@intel.com>
Signed-off-by: Emil Tantilov <emil.s.tantilov@intel.com>
Reviewed-by: Ahmed Zaki <ahmed.zaki@intel.com>
Reviewed-by: Aleksandr Loktionov <aleksandr.loktionov@intel.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Tested-by: Samuel Salin <Samuel.salin@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Add a helper function to correctly handle the lockless
synchronization when the sender needs to block. The paradigm is
if (no_resources()) {
stop_queue();
barrier();
if (!no_resources())
restart_queue();
}
netif_subqueue_maybe_stop already handles the paradigm correctly, but
the code split the check for resources in three parts, the first one
(descriptors) followed the protocol, but the other two (completions and
tx_buf) were only doing the first part and so race prone.
Luckily netif_subqueue_maybe_stop macro already allows you to use a
function to evaluate the start/stop conditions so the fix only requires
the right helper function to evaluate all the conditions at once.
The patch removes idpf_tx_maybe_stop_common since it's no longer needed
and instead adjusts separately the conditions for singleq and splitq.
Note that idpf_tx_buf_hw_update doesn't need to check for resources
since that will be covered in idpf_tx_splitq_frame.
To reproduce:
Reduce the threshold for pending completions to increase the chances of
hitting this pause by changing your kernel:
drivers/net/ethernet/intel/idpf/idpf_txrx.h
-#define IDPF_TX_COMPLQ_OVERFLOW_THRESH(txcq) ((txcq)->desc_count >> 1)
+#define IDPF_TX_COMPLQ_OVERFLOW_THRESH(txcq) ((txcq)->desc_count >> 4)
Use pktgen to force the host to push small pkts very aggressively:
./pktgen_sample02_multiqueue.sh -i eth1 -s 100 -6 -d $IP -m $MAC \
-p 10000-10000 -t 16 -n 0 -v -x -c 64
Fixes: 6818c4d5b3 ("idpf: add splitq start_xmit")
Reviewed-by: Jacob Keller <jacob.e.keller@intel.com>
Reviewed-by: Madhu Chittim <madhu.chittim@intel.com>
Signed-off-by: Josh Hay <joshua.a.hay@intel.com>
Signed-off-by: Brian Vazquez <brianvv@google.com>
Signed-off-by: Luigi Rizzo <lrizzo@google.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Tested-by: Samuel Salin <Samuel.salin@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Paolo Abeni