efx_unregister_netdev() should not call efx_release_tx_buffers()
directly, as it is already done when closing the device:
efx_net_stop() -> efx_stop_all() -> efx_stop_datapath() ->
efx_fini_tx_queue() -> efx_release_tx_buffers().

(This was presumably a workaround for a race between efx_stop_all()
and the data path that has since been properly fixed.)
Signed-off-by: NBen Hutchings <bhutchings@solarflare.com>

rx_queue::enabled guards refill, so rename it to reflect that.  Clear
it at the start of the queue teardown process rather than waiting for
the RX queue to be flushed.
Signed-off-by: NBen Hutchings <bhutchings@solarflare.com>

We unconditionally acknowledge legacy interrupts just before disabling
them.  This workaround is needed on Falcon A1 but probably not on
later chips where the legacy interrupt mechanism is different.  It was
also originally done after the IRQ handler was removed, not before.
Restore the original behaviour for Falcon A1 only by doing this
acknowledgement in the efx_nic_type::fini operation.
Signed-off-by: NBen Hutchings <bhutchings@solarflare.com>

There are many problems with the current efx_stop_interrupts() and
efx_start_interrupts():

1. On Siena, it is unsafe to disable the master IRQ enable bit
(DRV_INT_EN_KER) while any IRQ sources are enabled.

2. On EF10 there is no master IRQ enable bit, so we cannot expect to
defer IRQs without tearing down event queues.  (Though I don't think
we will need to keep any event queues around while the device is down,
as we do for VFDI on Siena.)

3. synchronize_irq() only waits for a running IRQ handler to finish,
not for any propagation through IRQ controllers.  Therefore an IRQ may
still be received and handled after efx_stop_interrupts() returns.
IRQ handlers can then race with channel reallocation.

To fix this:

a. Introduce a software IRQ enable flag.  So long as this is clear,
IRQ handlers will only acknowledge IRQs and not touch the channel
structures.

b. Define a new struct efx_msi_context as the context for MSIs.  This
is never reallocated and is sufficient to find the software enable
flag and the channel structure.  It also includes the channel/IRQ
name, which was previously separated out as it must also not be
reallocated.

c. Split efx_{start,stop}_interrupts() into
efx_{,soft_}_{enable,disable}_interrupts().  The 'soft' functions
don't touch the hardware master enable flag (if it exists) and don't
reinitialise or tear down channels with the keep_eventq flag set.
Signed-off-by: NBen Hutchings <bhutchings@solarflare.com>

efx_process_channel_now() is unneeded since self-tests can rely on
normal NAPI polling.  Remove it and all calls to it.

efx_channel::work_pending and efx_channel_processed() are also
unneeded (the latter being the same as efx_nic_eventq_read_ack()).
Signed-off-by: NBen Hutchings <bhutchings@solarflare.com>

The EF10 architecture has a very different register layout from
previous controllers, so we'll use separate files for the two sets of
register definitions.  Use 'farch' as an abbreviation for
Falcon-architecture.
Signed-off-by: NBen Hutchings <bhutchings@solarflare.com>

On EF10, the firmware is in charge of allocating buffer table entries.
Change struct efx_special_buffer to use a struct efx_buffer member,
so that it can be used with efx_nic_{alloc,free}_buffer() in that
case.
Signed-off-by: NBen Hutchings <bhutchings@solarflare.com>

Most call sites for efx_nic_alloc_buffer() are part of the probe or
reconfiguration paths and can allocate with GFP_KERNEL.  A few others
should use GFP_NOIO (I think).  Only one is in atomic context and
must use the current GFP_ATOMIC.
Signed-off-by: NBen Hutchings <bhutchings@solarflare.com>

Move the lowest layer (transport) of the current MCDI code to
per-NIC-type operations.

Introduce a new structure and efx_nic member for MCDI-specific data.
Signed-off-by: NBen Hutchings <bhutchings@solarflare.com>

This should probably be done during MCDI initialisation for any NIC.
Change efx_mcdi_init() to return an error code.
Signed-off-by: NBen Hutchings <bhutchings@solarflare.com>

Collect together MCDI port functions from mcdi.c, mcdi_mac.c,
mcdi_phy.c and siena.c.  Rename the 'siena' functions accordingly.
Signed-off-by: NBen Hutchings <bhutchings@solarflare.com>

EF10 does not include a multicast hash filter, so this function is
specific to Siena.
Signed-off-by: NBen Hutchings <bhutchings@solarflare.com>

These implementations should work for EF10 too.  Rename them
accordingly.
Signed-off-by: NBen Hutchings <bhutchings@solarflare.com>

No need to keep open-coding the assignment of high and low dwords.
Signed-off-by: NBen Hutchings <bhutchings@solarflare.com>

We currently require that MCDI request and response lengths are
multiples of 4 bytes, because we will copy dwords in and out of shared
memory and we want to be sure we won't read or write out of bounds.
But all we really need to know is that there is sufficient padding for
that.  Also, we should ensure that buffers are dword-aligned, as on
some architectures misaligned access will result in data corruption or
a crash.

Change the buffer type to array-of-efx_dword_t and remove the
requirement that the lengths are multiples of 4.
Signed-off-by: NBen Hutchings <bhutchings@solarflare.com>

A few functions are using heap buffers; change them to use stack
buffers as we really don't need to resort to the heap for a 252
byte buffer in process context.

MC_CMD_MEMCPY is quite weird in that it can use inline data placed in
the request buffer after the array of records.  Thus there are two
variable-length arrays and we can't use the normal accessors for
the second.  So we have to use _MCDI_PTR() in efx_sriov_memcpy().
Signed-off-by: NBen Hutchings <bhutchings@solarflare.com>

The MCDI version 2 protocol supports larger payloads, but will
not be implemented on Siena.
Signed-off-by: NBen Hutchings <bhutchings@solarflare.com>

We need to access arrays of 16-bit words and 32-bit dwords in MCDI
buffers based on the MCDI protocol definitions.

We should also be able to read and write fields within structures,
without specifying an array index each time.  So add MCDI_FIELD()
and make MCDI_ARRAY_FIELD() use it.  Also add MCDI_SET_FIELD().

Split MCDI_ARRAY_PTR() into MCDI_ARRAY_STRUCT_PTR() and
_MCDI_ARRAY_PTR(), which are currently identical but will diverge in
later changes.
Signed-off-by: NBen Hutchings <bhutchings@solarflare.com>

Add _MCDI_DWORD() which yields an lvalue for the given dword field
and change MCDI_DWORD(), MCDI_SET_DWORD() and MCDI_QWORD() to use it.

Fold the rather trivial MCDI_PTR2() into MCDI_PTR() and _MCDI_DWORD().

Remove MCDI_SET_DWORD2() and MCDI_QWORD2().  MCDI_DWORD2() should also
go, but it still has one user which we'll get rid of later.
Signed-off-by: NBen Hutchings <bhutchings@solarflare.com>

MCDI_DECLARE_BUF declares a variable as an MCDI buffer of the
requested length, adding any necessary padding.
Signed-off-by: NBen Hutchings <bhutchings@solarflare.com>

In particular, fold in the whole of falcon_xmac.c.

Drop some entirely unused definitions.
Signed-off-by: NBen Hutchings <bhutchings@solarflare.com>

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

This needs to be done before we separate MCDI from siena_nic_data.
Signed-off-by: NBen Hutchings <bhutchings@solarflare.com>

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

commit 385904f8 ('sfc: Don't use
efx_filter_{build,hash,increment}() for default MAC filters') used the
wrong name to find the index of default RX MAC filters at insertion/
update time.  This could result in memory corruption and would in any
case silently fail to update the filter.
Signed-off-by: NBen Hutchings <bhutchings@solarflare.com>

Received packets are only scattered if this is enabled in both the
matching filter and the receiving queue.  This was not being done for
filters inserted for RFS, so any packet requiring more than a single
descriptor was dropped.
Signed-off-by: NBen Hutchings <bhutchings@solarflare.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Commit 2768935a ('sfc: reuse pages to avoid DMA mapping/unmapping
costs') did not fully take account of DMA scattering which was
introduced immediately before.  If a received packet is invalid and
must be discarded, we only drop a reference to the first buffer's
page, but we need to drop a reference for each buffer the packet
used.

I think this bug was missed partly because efx_recycle_rx_buffers()
was not renamed and so no longer does what its name says.  It does not
change the state of buffers, but only prepares the underlying pages
for recycling.  Rename it accordingly.
Signed-off-by: NBen Hutchings <bhutchings@solarflare.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The device::iommu_group field may be set even if no IOMMU is in use.
iommu_present() is still a better indicator, although it doesn't tell
us whether *our* device is affected.
Reported-by: NAlex Williamson <alex.williamson@redhat.com>
Signed-off-by: NBen Hutchings <bhutchings@solarflare.com>

The lifetime of an irq_cpu_rmap is odd: we have to allocate it before
installing IRQ handlers and free it before removing the IRQ handlers.
As a result of this asymmetry, it was omitted from some failure paths.

On another failure path, we could try to remove IRQ handlers we
had not yet installed.

Move the irq_cpu_rmap allocation and freeing alongside IRQ handler
installation and removal, in efx_nic_{init,fini}_interrupts().
Count the number of IRQ handlers successfully installed and only
remove those on the failure path.
Signed-off-by: NBen Hutchings <bhutchings@solarflare.com>

GRO can handle non-TCP packets and pass them up without coalescing,
but it has to do some extra work to parse the packet which we can
bypass using the hardware parse result.  (This condition yields a
false negative for TCP/IPv6 packets received by Falcon, but its
performance is already poor in that case due to lack of checksum
offload.)
Signed-off-by: NBen Hutchings <bhutchings@solarflare.com>

This will be useful for shortcutting some software packet parsing.
Signed-off-by: NBen Hutchings <bhutchings@solarflare.com>

As far as I know, the hardware doesn't support matching on both IP
fields and vlan tag, but it can at least match on the IP fields.
Signed-off-by: NAndy Lutomirski <luto@amacapital.net>
Signed-off-by: NBen Hutchings <bhutchings@solarflare.com>

The kernel can generate software receive timestamps and we should
report those for all ports regardless of hardware capabilities.
Signed-off-by: NBen Hutchings <bhutchings@solarflare.com>

This allows the SKB to hold the headers without reallocation more often.
Signed-off-by: NBen Hutchings <bhutchings@solarflare.com>

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

PCI legacy interrupts are level-triggered, and we cannot mask them up
on an isolated device.  Instead, disable the IRQ at the controller
until we have recovered.
Signed-off-by: NBen Hutchings <bhutchings@solarflare.com>

Driver probe currently results in:

WARNING: at drivers/base/core.c:576 device_create_file+0x57/0x7e()
Attribute phy_type: write permission without 'store'
Signed-off-by: NBen Hutchings <bhutchings@solarflare.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

We should not use net_device::dev_id to indicate the port number, as
this affects the way the local part of IPv6 addresses is normally
generated.

This field was intended for use where multiple devices may share a
single assigned MAC address and need to have different IPv6 addresses.
Siena's two ports each have their own MAC addresses.
Signed-off-by: NBen Hutchings <bhutchings@solarflare.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

So far, only net_device * could be passed along with netdevice notifier
event. This patch provides a possibility to pass custom structure
able to provide info that event listener needs to know.
Signed-off-by: NJiri Pirko <jiri@resnulli.us>

v2->v3: fix typo on simeth
	shortened dev_getter
	shortened notifier_info struct name
v1->v2: fix notifier_call parameter in call_netdevice_notifier()
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

efx_start_datapath() asserts that we can fit 2 RX scatter buffers plus
a software structure, each appropriately aligned, into a single page.
Where L1_CACHE_BYTES == 256 and PAGE_SIZE == 4096, which is the case
on s390, this assertion fails.

The current scatter buffer size is also not a multiple of 64 or 128,
which are more common cache line sizes.  If we can make both the start
and end of a scatter buffer cache-aligned, this will reduce the need
for read-modify-write operations on inter- processor links.

Fix the alignment by reducing EFX_RX_USR_BUF_SIZE to 2048 - 256 ==
1792.  (We could use 2048 - L1_CACHE_BYTES, but EFX_RX_USR_BUF_SIZE
also affects user-level networking where a larger amount of
housekeeping data may be needed.  Although this version of the driver
does not support user-level networking, I prefer to keep scattering
behaviour consistent with the out-of-tree version.)

This still doesn't fix the s390 build because like most architectures
it has NET_IP_ALIGN == 2.  When NET_IP_ALIGN != 0 we cannot achieve
cache line alignment at either the start or end of a scatter buffer,
so there is actually no point in padding the buffers to a multiple of
the cache line size.  All we need is 4-byte alignment of the network
header, so do that.

Adjust the assertions accordingly.
Reported-by: NGeert Uytterhoeven <geert@linux-m68k.org>
Reported-by: NHeiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: NBen Hutchings <bhutchings@solarflare.com>
Acked-by: NGeert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>