QUIC servers would like to use SO_TXTIME, without having CAP_NET_ADMIN,
to efficiently pace UDP packets.

As far as sch_fq is concerned, we need to add safety checks, so
that a buggy application does not fill the qdisc with packets
having delivery time far in the future.

This patch adds a configurable horizon (default: 10 seconds),
and a configurable policy when a packet is beyond the horizon
at enqueue() time:
- either drop the packet (default policy)
- or cap its delivery time to the horizon.

$ tc -s -d qd sh dev eth0
qdisc fq 8022: root refcnt 257 limit 10000p flow_limit 100p buckets 1024
 orphan_mask 1023 quantum 10Kb initial_quantum 51160b low_rate_threshold 550Kbit
 refill_delay 40.0ms timer_slack 10.000us horizon 10.000s
 Sent 1234215879 bytes 837099 pkt (dropped 21, overlimits 0 requeues 6)
 backlog 0b 0p requeues 6
  flows 1191 (inactive 1177 throttled 0)
  gc 0 highprio 0 throttled 692 latency 11.480us
  pkts_too_long 0 alloc_errors 0 horizon_drops 21 horizon_caps 0

v2: fixed an overflow on 32bit kernels in fq_init(), reported
    by kbuild test robot <lkp@intel.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Willem de Bruijn <willemb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Currently if the default qdisc setup/init fails, the device ends up with
qdisc "noop", which causes all TX packets to get dropped.

With the introduction of sysctl net/core/default_qdisc it is possible
to change the default qdisc to be more advanced, which opens for the
possibility that Qdisc_ops->init() can fail.

This patch detect these kind of failures, and choose to fallback to
qdisc "noqueue", which is so simple that its init call will not fail.
This allows the interface to continue functioning.

V2:
As this also captures memory failures, which are transient, the
device is not kept in IFF_NO_QUEUE state.  This allows the net_device
to retry to default qdisc assignment.
Signed-off-by: Jesper Dangaard Brouer <brouer@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Alex Elder says:

====================
net: ipa: I/O map SMEM and IMEM

This series adds the definition of two memory regions that must be
mapped for IPA to access through an SMMU.  It requires the SMMU to
be defined in the IPA node in the SoC's Device Tree file.

There is no change since version 1 to the content of the code in
these patches, *however* this time the first patch is an update to
the binding definition rather than an update to a DTS file.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

Arrange to use an item from SMEM memory for IPA.  SMEM item number
497 is designated to be used by the IPA.  Specify the item ID and
size of the region in platform configuration data.  Allocate and get
a pointer to this region from ipa_mem_init().  The memory must be
mapped for access through an SMMU.
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Define a region of IMEM memory available for use by IPA in the
platform configuration data.  Initialize it from ipa_mem_init().
The memory must be mapped for access through an SMMU.
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

The ipa_mem_data structure type was never actually used.  Instead,
the IPA memory regions were defined using the ipa_mem structure.

Redefine struct ipa_mem_data so it encapsulates the array of IPA-local
memory region descriptors along with the count of entries in that
array.  Pass just an ipa_mem structure pointer to ipa_mem_init().

Rename the ipa_mem_data[] array ipa_mem_local_data[] to emphasize
that the memory regions it defines are IPA-local memory.
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

The IPA accesses "IMEM" and main system memory through an SMMU, so
its DT node requires an iommus property to define range of stream IDs
it uses.
Signed-off-by: Alex Elder <elder@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>

Heiner Kallweit says:

====================
net: add helper eth_hw_addr_crc

Several drivers use the same code as basis for filter hashes. Therefore
let's factor it out to a helper. This way drivers don't have to access
struct netdev_hw_addr internals.

First user is r8169.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

Use new helper eth_hw_addr_crc to simplify the code.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Several drivers use the same code as basis for filter hashes. Therefore
let's factor it out to a helper. This way drivers don't have to access
struct netdev_hw_addr internals.
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

If there is no specific configuration of the felix switch in the device
tree, but only the default configuration (ie. given by the SoCs dtsi
file), the probe fails because no CPU port has been set. On the other
hand you cannot set a default CPU port because that depends on the
actual board using the switch.

[    2.701300] DSA: tree 0 has no CPU port
[    2.705167] mscc_felix 0000:00:00.5: Failed to register DSA switch: -22
[    2.711844] mscc_felix: probe of 0000:00:00.5 failed with error -22

Thus let the device tree disable this device entirely, like it is also
done with the enetc driver of the same SoC.
Signed-off-by: Michael Walle <michael@walle.cc>
Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Karsten Graul says:

====================
net/smc: add failover processing

This patch series adds the actual SMC-R link failover processing and
improved link group termination. There will be one more (very small)
series after this which will complete the SMC-R link failover support.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

During SMC-R link establishment the peers exchange the link_uid that
is used for debugging purposes. Save the peer link_uid in smc_link so it
can be retrieved by the smc_diag netlink interface.
Signed-off-by: Karsten Graul <kgraul@linux.ibm.com>
Reviewed-by: Ursula Braun <ubraun@linux.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The link_uid of an SMC-R link is exchanged between SMC peers and its
value can be used for debugging purposes. Create a unique link_uid
during link initialization and use it in communication with SMC-R peers.
Signed-off-by: Karsten Graul <kgraul@linux.ibm.com>
Reviewed-by: Ursula Braun <ubraun@linux.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Add helper smcr_lgr_link_deactivate_all() and eliminate duplicate code.
In smc_lgr_free(), clear the smc-r links before smc_lgr_free_bufs() is
called so buffers are already prepared for free. The usage of the soft
parameter in __smc_lgr_terminate() is no longer needed, smc_lgr_free()
can be called directly. smc_lgr_terminate_sched() and
smc_smcd_terminate() set lgr->freeing to indicate that the link group
will be freed soon to avoid unnecessary schedules of the free worker.
Signed-off-by: Karsten Graul <kgraul@linux.ibm.com>
Reviewed-by: Ursula Braun <ubraun@linux.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Allow to set the reason code for the link group termination, and set
meaningful values before termination processing is triggered. This
reason code is sent to the peer in the final delete link message.
When the LLC request or response layer receives a message type that was
not handled, drop a warning and terminate the link group.
Signed-off-by: Karsten Graul <kgraul@linux.ibm.com>
Reviewed-by: Ursula Braun <ubraun@linux.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

New connections must not be assigned to asymmetric links. Add asymmetric
link tagging using new link variable link_is_asym. The new helpers
smcr_lgr_set_type() and smcr_lgr_set_type_asym() are called to set the
state of the link group, and tag all links accordingly.
smcr_lgr_conn_assign_link() respects the link tagging and will not
assign new connections to links tagged as asymmetric link.
Signed-off-by: Karsten Graul <kgraul@linux.ibm.com>
Reviewed-by: Ursula Braun <ubraun@linux.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

For new connections, assign a link from the link group, using some
simple load balancing.
Signed-off-by: Karsten Graul <kgraul@linux.ibm.com>
Reviewed-by: Ursula Braun <ubraun@linux.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Add smc_llc_send_message_wait() which uses smc_wr_tx_send_wait() to send
an LLC message and waits for the message send to complete.
smc_llc_send_link_delete_all() calls the new function to send an
DELETE_LINK,ALL LLC message. The RFC states that the sender of this type
of message needs to wait for the completion event of the message
transmission and can terminate the link afterwards.
Signed-off-by: Karsten Graul <kgraul@linux.ibm.com>
Reviewed-by: Ursula Braun <ubraun@linux.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Introduce smc_wr_tx_send_wait() to send an IB message and wait for the
tx completion event of the message. This makes sure that the message is
no longer in-flight when the function returns.
Signed-off-by: Karsten Graul <kgraul@linux.ibm.com>
Reviewed-by: Ursula Braun <ubraun@linux.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Call smc_cdc_msg_validate() when a CDC message with the failover
validation bit enabled was received. Validate that the sequence number
sent with the message is one we already have received. If not, messages
were lost and the connection is terminated using a new abort_work.
Signed-off-by: Karsten Graul <kgraul@linux.ibm.com>
Reviewed-by: Ursula Braun <ubraun@linux.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

When a connection is switched to a new link then a link validation
message must be sent to the peer over the new link, containing the
sequence number of the last CDC message that was sent over the old link.
The peer will validate if this sequence number is the same or lower then
the number he received, and abort the connection if messages were lost.
Add smcr_cdc_msg_send_validation() to send the message validation
message and call it when a connection was switched in
smc_switch_cursor().
Signed-off-by: Karsten Graul <kgraul@linux.ibm.com>
Reviewed-by: Ursula Braun <ubraun@linux.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Add smc_switch_conns() to switch all connections from a link that is
going down. Find an other link to switch the connections to, and
switch each connection to the new link. smc_switch_cursor() updates the
cursors of a connection to the state of the last successfully sent CDC
message. When there is no link to switch to, terminate the link group.
Call smc_switch_conns() when a link is going down.
And with the possibility that links of connections can switch adapt CDC
and TX functions to detect and handle link switches.
Signed-off-by: Karsten Graul <kgraul@linux.ibm.com>
Reviewed-by: Ursula Braun <ubraun@linux.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

When a link goes down and all connections of this link need to be
switched to an other link then the producer cursor and the sequence of
the last successfully sent CDC message must be known. Add the two fields
to the SMC connection and update it in the tx completion handler.
And to allow matching of sequences in error cases reset the seqno to the
old value in smc_cdc_msg_send() when the actual send failed.
Signed-off-by: Karsten Graul <kgraul@linux.ibm.com>
Reviewed-by: Ursula Braun <ubraun@linux.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Michael Chan says:

====================
bnxt_en: Updates for net-next.

This patchset includes these main changes:

1. Firmware spec. update.
2. Context memory sizing improvements for the hardware TQM block.
3. ethtool chip reset improvements and fixes for correctness.
4. Improve L2 doorbell mapping by mapping only up to the size specified
by firmware.  This allows the RoCE driver to map the remaining doorbell
space for its purpose, such as write-combining.
5. Improve ethtool -S channel statistics by showing only relevant ring
counters for non-combined channels.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

Currently, ethtool -S shows all TX/RX ring counters whether the
channel is combined, RX, or TX.  The unused counters will always be
zero.  Improve it by showing only the relevant counters if the channel
is RX or TX.  If the channel is combined, the counters will be shown
exactly the same as before.

[ MChan: Lots of cleanups and simplifications on Rajesh's original
code]
Signed-off-by: Rajesh Ravi <rajesh.ravi@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

This will allow the RX and TX ring statistics to be separated if needed.
In the next patch, we'll be able to only display RX or TX statistcis if
the channel is RX only or TX only.
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

We currently have 3 software ring counters, rx_l4_csum_errors,
rx_buf_errors, and missed_irqs.  The 1st two are RX counters and the
last one is a common counter.  Organize them into 2 structures
bnxt_rx_sw_stats and bnxt_cmn_sw_stats.
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The purpose of this is to inform the RDMA driver the size of the doorbell
BAR that the L2 driver has mapped and the portion that is mapped
uncacheable.  The unchaeable portion is shared with the RoCE driver.
Any remaining unmapped doorbell BAR can be used by the RDMA driver for
its own purpose.  Currently, the entire L2 portion is mapped uncacheable.
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Read the L2 doorbell size from the firmware and only map the portion
of the doorbell BAR for L2 use.  This will leave the remaining doorbell
BAR available for the RoCE driver to use.  The RoCE driver can map
the remaining portion as write-combining to support the push feature.
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The driver provides completion ring or NQ doorbell offset for each
MSIX entry requested by the RDMA driver.  The NQ offset on 57500
chips is different than legacy chips.  Set it correctly based on
chip type for correctness.  The RDMA driver is ignoring this field
for the 57500 chips so it is not causing any problem.
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Define the 57500 chip doorbell offsets instead of using the magic
values in the C file.
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Kernel log messages for failed AP reset commands should be suppressed.
These are expected to fail on devices that do not have an AP.  Add
missing driver reload message after AP reset and log it in a common
way without duplication.
Signed-off-by: Edwin Peer <edwin.peer@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The ethtool ABI specifies that the reset operation should only clear
the flags that were actually reset. Setting the flags to zero after
a chip reset violates this because it does not include resetting the
application processor complex. Similarly, components that are not yet
defined are also not necessarily being reset.

The fact that chip reset does not cover the AP also means that it is
inappropriate to treat these two components exclusively of one another.
The ABI provides a mechanism to report a failure to reset independent
components via the returned bitmask, so it is also wrong to fail hard
if one of a set of independent resets is not possible.

It is incorrect to rely on the passed by reference flags in bnxt_reset(),
which are being updated as components are reset. The initially requested
value should be used instead so that hard errors do not propagate if any
earlier components could have been reset successfully.

Note, AP and chip resets are global in nature. Dedicated resets are
thus not currently supported.
Signed-off-by: Edwin Peer <edwin.peer@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The case statement in bnxt_firmware_reset() dangerously mixes types.
This patch separates the application processor and whole chip resets
from the rest such that the selection is performed on a pure type.
Signed-off-by: Edwin Peer <edwin.peer@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Extract bnxt_hwrm_firmware_reset() for performing firmware reset
operations. This new helper function will be used in a subsequent
patch to separate unrelated reset types out of bnxt_firmware_reset().
Signed-off-by: Edwin Peer <edwin.peer@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The firmware does not expect the CRC to be included in the length
passed from the driver.  The firmware always configures the chip
to strip out the CRC.
Signed-off-by: Vasundhara Volam <vasundhara-v.volam@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The current formulas to calculate the TQM slow path and fast path ring
context memory sizes are not quite correct.  TQM slow path entry is
array index 0 of ctx->tqm_mem[].  The other array entries are for fast
path.  Fix these sizes according to latest firmware spec. for 57500 and
newer chips.

Fixes: 3be8136c ("bnxt_en: Initialize context memory to the value specified by firmware.")
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Newer firmware spec. will specify the number of TQM rings to allocate
context memory for.  Use the firmware specified value and fall back
to the old value derived from bp->max_q if it is not available.
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Changes include additional statistics, ECN support, context memory
interface change for better TQM context memory sizing, firmware
health status definitions, etc.
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>