This adds support for the EF10 network controller architecture and the
SFC9100 family, starting with SFC9120 'Farmingdale', and bumps the
driver version to 4.0.

New features in the SFC9100 family include:

- Flexible allocation of internal resources to PCIe physical and virtual
  functions under firmware control
- RX event merging to reduce DMA writes at high packet rates
- Integrated RX timestamping
- PIO buffers for lower TX latency
- Firmware-driven data path that supports additional offload features
  and filter types
- Delivery of packets between functions and to multiple recipients,
  allowing firmware to implement a vswitch
- Multiple RX flow hash (RSS) contexts with their own hash keys and
  indirection tables
- 40G MAC (single port only)

...not all of which are enabled in this initial driver or the initial
firmware release.

Much of the new code is by Jon Cooper.
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>

This should be done during MCDI initialisation for any NIC.
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>

Also update comments and assertions in io.h:
- EF10 does not have a general BIU collector and does not have the
  bug affecting TIMER_COMMAND_REG[0] on Falcon/Siena
- The WPTR field moved within RX_DESC_UPD_REG and TX_DESC_UPD_REG.
  Adjust efx_writed_page() accordingly
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>

The TX path firmware for EF10 supports 'option descriptors' to control
offloads and various other features.  Add a flag and field for these
in struct efx_tx_buffer, and don't treat them as DMA descriptors on
completion.
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>

On EF10, the firmware will initiate a queue flush in certain
error cases.  We need to accept that flush events might appear
at any time after a queue has been initialised, not just when
we try to flush them.

We can handle Falcon-architecture in just the same way.
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>

RX DMA scatter is always enabled on EF10.  Adjust the common RX
completion handling to allow for this.

RX completion events on EF10 include the length used from a single
descriptor, not the cumulative length used.  Add a field to struct
efx_rx_queue to hold the cumulative length.

[bwh: Also fix a related comment]
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>

On EF10, event queue initialisation requires an MCDI request which
may return failure.
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>

Add the efx_filter_is_mc_recip() function to decide whether a filter
is for a multicast recipient and can coexist with other filters with
the same match values.  Update efx_filter_insert_filter() kernel-doc
to explain the conditions for this.
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>

This will allow use of MCDI from the data path, in particular for
accelerated RFS.
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>

We can set, get and compare-and-exchange without using atomic_t.
Change efx_mcdi_iface::state to the enum type we really wanted it to
be.
Suggested-by: David Miller <davem@davemloft.net>
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>

Preparation for asynchronous MCDI requests.
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>

Add support for power and current sensors, which need to be named
differently in sysfs.  Power sensors also require values to be scaled
between MCDI and sysfs, and have no minimum value.

Add definitions of the power, current, fan, and additional temperature
and voltage sensors found on SFA6902F, SFN7022F and SFN7122F.

(Includes a bug fix from Andrew Jackson.)
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>

We need to use extended requests to read and get metadata for sensors
numbered > 31.
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>

[bwh: Also name this new state, though we don't expect to see it in an event]
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>

Define a flag for struct efx_rx_buffer and efx_rx_packet() that
indicates packet length must be read from the prefix.  If this
is set, read the length in __efx_rx_packet() (when the prefix
should have arrived in cache).
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>

Add a counter for TX merged completion events.

This is implemented in the common TX path, because the NIC event
handlers only know how many descriptors were completed, not how many
packets.
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>

EF10 uses an entirely different RX prefix format from Falcon-arch.
Extend struct efx_nic_type to describe this.

[bwh: Also replace the magic numbers used for the Falcon-arch RX prefix]
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>

efx_reset_up() calls efx_nic_type::reconfigure_mac once directly,
then again through efx_start_all() -> efx_start_port() ->
efx->type->reconfigure_mac().

This first call is also made too early to work properly on EF10.
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>

The Huntington MC will reject all MCDI requests after an MC reboot until it sees
one with the NOT_EPOCH flag clear.  This flag is set by default for all requests,
and then cleared on the first request after we detect that an MC reboot has
occurred.

The old MCDI_STATUS_DELAY_COUNT gave a timeout of 10ms, which was not long enough
for the driver to detect that a reboot had occurred based on the warm boot count
while calling efx_mcdi_poll_reboot() from the loop in efx_mcdi_ev_death().
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>

Also, since we handle all DMA errors in the same way, merge
RESET_TYPE_(RX|TX)_DESC_FETCH into RESET_TYPE_DMA_ERROR.
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>

This field is ignored by Siena firmware but is significant to EF10 firmware.
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>

Various hardware statistics that are available for Siena are
unavailable or meaningless for Falcon.  Huntington adds further to the
NIC-type-specific statistics, as it has different MAC blocks from
Falcon/Siena.

All NIC types still provide most statistics by DMA, and use
little-endian byte order.

Therefore:
1. Add some general utility functions for reporting hardware statistics,
   efx_nic_describe_stats() and efx_nic_update_stats().
2. Add an efx_nic_type::describe_stats operation to get the number and
   names of statistics, implemented using efx_nic_describe_stats()
3. Change efx_nic_type::update_stats to store the core statistics
   (struct rtnl_link_stats64) or full statistics (array of u64) in a
   caller-provided buffer.  Use efx_nic_update_stats() to aid in the
   implementation.
4. Rename struct efx_ethtool_stat to struct efx_sw_stat_desc and
   EFX_ETHTOOL_NUM_STATS to EFX_ETHTOOL_SW_STAT_COUNT.
5. Remove efx_nic::mac_stats and struct efx_mac_stats.
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>

It's not really helpful to pretend ethtool string arrays are
structured.
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>

We only ever used the XMAC (10G link speed) in production.
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>

Merge the per-NIC-type MTD probe selection and struct efx_mtd_ops into
struct efx_nic_type.  Move the implementations into the appropriate
source files.

Several NVRAM functions are now only called from MTD operations which
are now implemented in the same file (falcon.c or mcdi.c).  There is no
need for them to be extern, or to be defined at all if CONFIG_SFC_MTD
is not enabled, so move them into the #ifdef CONFIG_SFC_MTD sections
in those files.

Most of the SPI-related definitions are also only used in falcon.c,
so move them there.  Put the remainder of spi.h into nic.h (which
previously included it).
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>

Currently we use struct efx_mtd to represent a physical NVRAM device
and struct efx_mtd_partition to represent a partition on that device.
But this only really makes sense for Falcon, as we don't know or care
whether MC-managed NVRAM partitions are on one or more physical
devices.  It complicates iteration and provides little benefit.
Therefore:

- Replace the pointer to efx_mtd in mtd_info::priv with a pointer to efx_nic
- Move the falcon_spi_device pointer into the union in struct efx_mtd_partition
- Move the device name to efx_mtd_partition::dev_type_name
- Move the efx_mtd_ops pointer to efx_nic::mtd_ops
- Make efx_nic::mtd_list a list of partitions
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>

Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>

On Falcon we implement MAC filtering requested by the stack using the
MAC wrapper's single unicast filter and multicast hash filter.  Siena
is very similar, though MAC configuration is mediated by the MC.

Since MCDI operations may sleep, reconfiguration is deferred from
ndo_set_rx_mode to a work item.  However, it still updates the private
variables describing the filter state synchronously.  Contrary to
comments, the later use of these variables is not protected using the
address lock, resulting in race conditions.

Move the state update to a new function
efx_farch_filter_sync_rx_mode() and make the Falcon-arch MAC
configuration functions call that, so that its use is consistently
serialised by the mac_lock.

Invert and rename the promiscuous flag to the more accurate
unicast_filter, and comment that both this and multicast_hash are
not used on EF10.
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>

There is only one user now, but we're about to add many more.
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>

MAC filters inserted on request from the stack (ndo_set_rx_mode)
should allow manual steering but not removal.  Currently we have a
special case for Siena's all-multicast and all-unicast MAC filters,
but on EF10 we need to allow for steering of precise MAC filters as
well.

The EFX_FILTER_FLAG_RX_STACK flag changes the behaviour of replacement
and removal requests:

- Replacement *of* a filter with this flag never clears the flag but
  does change steering and saved priority
- Replacement *by* a filter with this flag only sets the flag but does
  not change steering
- Removal with priority < EFX_FILTER_PRI_REQUIRED really resets RX
  steering and saved priority

This could support precise MAC filtering on Siena in future.

As a side-benefit, the default MAC filters are hidden from ethtool
until they are steered.
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>

Move the special case for removal of default filters from
efx_farch_filter_table_clear_entry() into a wrapper function,
efx_farch_filter_table_remove().  Move the existence and priority
checks into the latter and use it where appropriate.
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>

Aside from accelerated RFS, there is almost nothing that can be shared
between the filter table implementations for the Falcon architecture
and EF10.

Move the few shared functions into efx.c and rx.c and the rest into
farch.c.  Introduce efx_nic_type operations for the implementation and
inline wrapper functions that call these.
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>

Currently every call to efx_farch_filter_table_clear_entry() is
shortly followed by a conditional reset of the table limits.  The new
limits (0) are not pushed to hardware until the next filter insertion.
Move both the reset and the hardware reconfiguration into
efx_farch_filter_table_clear_entry(), and add an explanatory comment.

Also, make consistent use of the term 'search limit' for the maximum
number of probes the NIC must make when searching for a filter of a
particular type.
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>

Move the common state from struct efx_filter_state into struct efx_nic.
Rename struct efx_filter_state to efx_farch_filter_state and change
the type of efx_nic::filter_state to void *.
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>