Similar to the helper function to copy the ethtool stats strings, add
and use a helper function for copying the ethtool stats into the
supplied buffer.

Just like before, we use a macro to avoid having to pass ARRAY_SIZE
manually, so as to reduce chance of bugs.

Some of the stats, especially queue stats, are a bit trickier, and will
be handled in future patches.
Signed-off-by: NJacob Keller <jacob.e.keller@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

Many of the ethtool statistics use the same basic logic for copying
strings into the supplied buffer. A set of stats are stored in a const
array of i40e_stats structures, and we apply these all together.

Simplify the stats code by introducing a helper function which can take
a stats array and copy the strings into the buffer, updating the buffer
pointer as we go.

We use a macro to implement i40e_add_stat_strings so that ARRAY_SIZE can
be used on the array passed in. This ensures that we always use the
matching size in __i40e_add_stat_strings.

More complex stats currently do not use i40e_stats arrays, usually due
to custom formatted strings, or because the stats are not laid out in
the expected way. These stats will be updated to use the helper function
in separate future patches.
Signed-off-by: NJacob Keller <jacob.e.keller@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

Function call to i40e_prep_for_reset() is duplicated in
i40e_shutdown routine and gets called before
i40e_enable_mc_magic_wake() which blocks it from being executed
correctly on system reboot or shutdown because adminq is already
disabled by first i40e_prep_for_reset() call.

Two register write calls are also duplicated.
Signed-off-by: NSergey Nemov <sergey.nemov@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

prog_attached of struct netdev_bpf should have been superseded
by simply setting prog_id long time ago, but we kept it around
to allow offloading drivers to communicate attachment mode (drv
vs hw).  Subsequently drivers were also allowed to report back
attachment flags (prog_flags), and since nowadays only programs
attached will XDP_FLAGS_HW_MODE can get offloaded, we can tell
the attachment mode from the flags driver reports.  Remove
prog_attached member.
Signed-off-by: NJakub Kicinski <jakub.kicinski@netronome.com>
Reviewed-by: NQuentin Monnet <quentin.monnet@netronome.com>
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>

The driver was combining the XDP_TX tail flush and XDP_REDIRECT
map flushing (xdp_do_flush_map).  This is suboptimal, these two
flush operations should be kept separate.

It looks like the mistake was copy-pasted from ixgbe.

Fixes: d9314c47 ("i40e: add support for XDP_REDIRECT")
Signed-off-by: NJesper Dangaard Brouer <brouer@redhat.com>
Acked-by: NBjörn Töpel <bjorn.topel@intel.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Pass the extact struct from a tc qdisc add to the block bind function and,
in turn, to the setup_tc ndo of binding device via the tc_block_offload
struct. Pass this back to any block callback registrations to allow
netlink logging of fails in the bind process.
Signed-off-by: NJohn Hurley <john.hurley@netronome.com>
Signed-off-by: NJakub Kicinski <jakub.kicinski@netronome.com>
Acked-by: NJiri Pirko <jiri@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

-EOPNOTSUPP is the error value that should be reported if a flower
command is not supported by a driver. Fix it in couple of Intel drivers.
Signed-off-by: NJiri Pirko <jiri@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

random_ether_addr is a #define for eth_random_addr which is
generally preferred in kernel code by ~3:1

Convert the uses of random_ether_addr to enable removing the #define

Miscellanea:

o Convert &vfmac[0] to equivalent vfmac and avoid unnecessary line wrap
Signed-off-by: NJoe Perches <joe@perches.com>
Acked-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Using skb_reserve(skb, I40E_SKB_PAD + (xdp->data - xdp->data_hard_start))
is clearly wrong since I40E_SKB_PAD already points to the offset where
the original xdp->data was sitting since xdp->data_hard_start is defined
as xdp->data - i40e_rx_offset(rx_ring) where latter offsets to I40E_SKB_PAD
when build skb is used.

However, also before cc5b114d ("bpf, i40e: add meta data support")
this seems broken since bpf_xdp_adjust_head() helper could have been used
to alter headroom and enlarge / shrink the frame and with that the assumption
that the xdp->data remains unchanged does not hold and would push a bogus
packet to upper stack.

ixgbe got this right in 92470808 ("ixgbe: add XDP support for pass and
drop actions"). In any case, fix it by removing the I40E_SKB_PAD from both
skb_reserve() and truesize calculation.

Fixes: cc5b114d ("bpf, i40e: add meta data support")
Fixes: 0c8493d9 ("i40e: add XDP support for pass and drop actions")
Reported-by: NKeith Busch <keith.busch@linux.intel.com>
Reported-by: NToshiaki Makita <makita.toshiaki@lab.ntt.co.jp>
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>
Cc: Björn Töpel <bjorn.topel@intel.com>
Cc: John Fastabend <john.fastabend@gmail.com>
Tested-by: NKeith Busch <keith.busch@linux.intel.com>
Acked-by: NJohn Fastabend <john.fastabend@gmail.com>
Acked-by: NAlexander Duyck <alexander.h.duyck@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Remove the ndo_xdp_flush call implementation i40e_xdp_flush
as no callers of ndo_xdp_flush are left.
Signed-off-by: NJesper Dangaard Brouer <brouer@redhat.com>
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>

Add support for XDP meta data when using build skb variant of
the i40e driver. Implementation is analogous to the existing
ixgbe and ixgbevf support for meta data from 366a88fe ("bpf,
ixgbe: add meta data support") and be833332 ("ixgbevf: Add
support for meta data"). With the build skb variant we get
192 bytes of extra headroom which can be used for encaps or
meta data.
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>
Acked-by: NJohn Fastabend <john.fastabend@gmail.com>
Tested-by: NJohn Fastabend <john.fastabend@gmail.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

When passed the XDP_XMIT_FLUSH flag i40e_xdp_xmit now performs the
same kind of ring tail update as in i40e_xdp_flush.  The advantage is
that all the necessary checks have been performed and xdp_ring can be
updated, instead of having to perform the exact same steps/checks in
i40e_xdp_flush
Signed-off-by: NJesper Dangaard Brouer <brouer@redhat.com>
Signed-off-by: NAlexei Starovoitov <ast@kernel.org>

This patch only change the API and reject any use of flags. This is an
intermediate step that allows us to implement the flush flag operation
later, for each individual driver in a separate patch.

The plan is to implement flush operation via XDP_XMIT_FLUSH flag
and then remove XDP_XMIT_FLAGS_NONE when done.
Signed-off-by: NJesper Dangaard Brouer <brouer@redhat.com>
Acked-by: NSong Liu <songliubraving@fb.com>
Signed-off-by: NAlexei Starovoitov <ast@kernel.org>

This patch change the API for ndo_xdp_xmit to support bulking
xdp_frames.

When kernel is compiled with CONFIG_RETPOLINE, XDP sees a huge slowdown.
Most of the slowdown is caused by DMA API indirect function calls, but
also the net_device->ndo_xdp_xmit() call.

Benchmarked patch with CONFIG_RETPOLINE, using xdp_redirect_map with
single flow/core test (CPU E5-1650 v4 @ 3.60GHz), showed
performance improved:
 for driver ixgbe: 6,042,682 pps -> 6,853,768 pps = +811,086 pps
 for driver i40e : 6,187,169 pps -> 6,724,519 pps = +537,350 pps

With frames avail as a bulk inside the driver ndo_xdp_xmit call,
further optimizations are possible, like bulk DMA-mapping for TX.

Testing without CONFIG_RETPOLINE show the same performance for
physical NIC drivers.

The virtual NIC driver tun sees a huge performance boost, as it can
avoid doing per frame producer locking, but instead amortize the
locking cost over the bulk.

V2: Fix compile errors reported by kbuild test robot <lkp@intel.com>
V4: Isolated ndo, driver changes and callers.
Signed-off-by: NJesper Dangaard Brouer <brouer@redhat.com>
Signed-off-by: NAlexei Starovoitov <ast@kernel.org>

Since we no longer use i as an array index for the data variable,
replace the use of 'j' with 'i' so that we match the general loop
variable name.
Signed-off-by: NJacob Keller <jacob.e.keller@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

Add documentation for the i40e_get_stats_count, i40e_get_stat_strings
and i40e_get_ethtool_stats explaining that the number and ordering of
statistics must remain constant for a given netdevice.
Signed-off-by: NJacob Keller <jacob.e.keller@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

A future patch is going to add a helper function i40e_add_ethtool_stats
that will help lower the amount of boiler plate code in the
i40e_get_ethtool_stats function.

This conversion will take place over many patches, and the helper
function will work by directly updating a reference to the data pointer.

Since this would not work combined with the current method of accessing
data like an array, update all the code that copies stats into the data
buffer to use direct updates to the pointer instead of array accesses.

This will prevent incorrect stat updates for patches in between the
conversion.

Similarly, when copying strings, we used a separate char *p pointer.
Instead, use the data pointer directly as it's already a (u8 *) type
which is the same size.
Signed-off-by: NJacob Keller <jacob.e.keller@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

We always prefix these stats with a fixed string, so just fold this
prefix into the stat string definition. This preparatory work will make
it easier to implement a helper function to copy stats and strings into
the supplied buffers in a future patch.
Signed-off-by: NJacob Keller <jacob.e.keller@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

We don't really want to use BUG_ON here since that would completely
crash the kernel, thus the reason we commented it out. We *can't* use
BUILD_BUG_ON because at least now (a) the sizes aren't constant (we are
fixing this) and (b) not all compilers are smart enough to understand
that "p - data" is a constant.

Instead, just use a WARN_ONCE so that the first time we end up with an
incorrect size we will dump a stack trace and a message, hopefully
highlighting the issues early in testing.
Signed-off-by: NJacob Keller <jacob.e.keller@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

Split the statistic strings and private flags strings into their own
separate functions to aid code readability.
Signed-off-by: NJacob Keller <jacob.e.keller@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

The ethtool API for obtaining device statistics is not intended to allow
runtime changes in the number of statistics reported. It may *appear*
this way, as there is an ability to request the number of stats using
ethtool_get_set_count(). However, it is expected that this must always
return the same value for invocations of the same device.

If we don't satisfy this contract, and allow the number of stats to
change during run time, we could cause invalid memory accesses or report
the stat strings incorrectly. This is because the API for obtaining
stats is to (1) get the size, (2) get the strings and finally (3) get
the stats. Since these are each separate ethtool op commands, it is not
possible to maintain consistency by holding the RTNL lock over the whole
operation. This results in the potential for a race condition to occur
where the size changed between any of the 3 calls.

Avoid this issue by requiring that we always return the same value for
a given device. We can check any values which remain constant for the
life of the device, but must not report different sizes depending on
runtime attributes.

This patch specifically fixes the queue statistics to always return
every queue even if it's not currently in use.
Signed-off-by: NJacob Keller <jacob.e.keller@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

The ethtool API for obtaining device statistics is not intended to allow
runtime changes in the number of statistics reported. It may *appear*
this way, as there is an ability to request the number of stats using
ethtool_get_set_count(). However, it is expected that this must always
return the same value for invocations of the same device.

If we don't satisfy this contract, and allow the number of stats to
change during run time, we could cause invalid memory accesses or report
the stat strings incorrectly. This is because the API for obtaining
stats is to (1) get the size, (2) get the strings and finally (3) get
the stats. Since these are each separate ethtool op commands, it is not
possible to maintain consistency by holding the RTNL lock over the whole
operation. This results in the potential for a race condition to occur
where the size changed between any of the 3 calls.

Avoid this issue by requiring that we always return the same value for
a given device. We can check any values which remain constant for the
life of the device, but must not report different sizes depending on
runtime attributes.

This patch specifically fixes the VEB statistics strings to always be
reported. Other issues will be fixed in future patches.
Signed-off-by: NJacob Keller <jacob.e.keller@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

Use the same logic to free the skb after clearing the Tx timestamp bit
lock in i40e_ptp_stop as we use in the other locations. It is not as
important here since we are not racing against a future Tx timestamp
request (as we are disabling PTP at this point). However it is good to
be consistent in how we approach the bit lock so that future callers
don't copy the old anti-pattern.
Signed-off-by: NJacob Keller <jacob.e.keller@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

This patch fixes a hardware reset support in VF driver.
It is needed because when a hardware reset is detected
adapter->state is in __I40EVF_RESETTING state before
i40evf_reset_task is called. Without this patch
unloading VF driver after a hardware reset ends
with a system crash.
Signed-off-by: NPaweł Jabłoński <pawel.jablonski@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

In commit bbc4e7d2 ("i40e: fix race condition with PTP_TX_IN_PROGRESS
bits") we modified the code which handles Tx timestamps so that we would
clear the progress bit as soon as possible.

A later commit 0bc0706b ("i40e: check for Tx timestamp timeouts during
watchdog") introduced similar code for detecting and handling cleanup of
a blocked Tx timestamp. This code did not use the same pattern for cleaning
up the skb.

Update this code to wait to free the skb until after the bit lock is
free, by first setting the ptp_tx_skb to NULL and clearing the lock.
Signed-off-by: NJacob Keller <jacob.e.keller@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

Fix up the English in the header comment for i40e_ptp_tx_hang.
Signed-off-by: NJacob Keller <jacob.e.keller@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

This patch adds the tx_busy stat to the ethtool stats. The tx_busy
stat tracks the number of times we return NETDEV_TX_BUSY to the stack
during transmit.
Signed-off-by: NHarshitha Ramamurthy <harshitha.ramamurthy@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

This patch adds a recalculation of number of MSI-X
vectors for VMDq in the case where we have less
vectors available than we would want to reserve for
VMDq.

It fixes the issue where we recalculate vectors left
and vectors wanted but we didn't take into account
the reduced number of queue pairs per VSI.
Signed-off-by: NPatryk Małek <patryk.malek@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

A future patch is going to refactor some of the ethtool statistic code.
To keep the patches easy to review, cleanup some of the indentation used
for macro definitions first.
Signed-off-by: NJacob Keller <jacob.e.keller@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

The pfc related priority stats are already handled separately as these
stats are actually arrays of length I40E_MAX_USER_PRIORITY. Thus,
including them within i40e_gstrings_stats will just duplicate data.

Worse, the sizeof will be incorrect, as it will be the total size of the
stat arrays, which in this case is 8 * sizeof(u64), so we will only copy
the stat contents as if they were a u32.

Since we already correctly handle these stats else where, remove them
from the i40e_gstrings_stats.
Signed-off-by: NJacob Keller <jacob.e.keller@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

Use a separate function to calculate the number of stats for
a particular device. This helps reduce the clutter in
i40e_get_sset_count().
Signed-off-by: NJacob Keller <jacob.e.keller@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

Fix warnings regarding restricted __be32 type usage by strictly
specifying the type of the ipv4 address being printed in the dev_err
statement.
Signed-off-by: NJacob Keller <jacob.e.keller@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

The expectation of the ops VIRTCHNL_OP_ENABLE_QUEUES and
VIRTCHNL_OP_DISABLE_QUEUES is that the queue map sent by
the VF is taken into account when enabling/disabling
queues in the VF VSI. This patch makes sure that happens.

By breaking out the individual queue set up functions so
that they can be called directly from the i40e_virtchnl_pf.c
file, only the queues as specified by the queue bit map that
accompanies the enable/disable queues ops will be handled.
Signed-off-by: NHarshitha Ramamurthy <harshitha.ramamurthy@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

When operating at 1GbE, the base incval for the PTP clock is so large
that multiplying it by numbers close to the max_adj can overflow the
u64.

Rather than attempting to limit the max_adj to a value small enough to
avoid overflow, instead calculate the incvalue adjustment based on the
40GbE incvalue, and then multiply that by the scaling factor for the
link speed.

This sacrifices a small amount of precision in the adjustment but we
avoid erratic behavior of the clock due to the overflow caused if ppb is
very near the maximum adjustment.
Signed-off-by: NJacob Keller <jacob.e.keller@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

This fixes at least 2 issues I have found with the UDP tunnel filter
configuration.

The first issue is the fact that the tunnels didn't have any sort of mutual
exclusion in place to prevent an update from racing with a user request to
add/remove a port. As such you could request to add and remove a port
before the port update code had a chance to respond which would result in a
very confusing result. To address it I have added 2 changes. First I added
the RTNL mutex wrapper around our updating of the pending, port, and
filter_index bits. Second I added logic so that we cannot use a port that
has a pending deletion since we need to free the space in hardware before
we can allow software to reuse it.

The second issue addressed is the fact that we were not recording the
actual filter index provided to us by the admin queue. As a result we were
deleting filters that were not associated with the actual filter we wanted
to delete. To fix that I added a filter_index member to the UDP port
tracking structure.
Signed-off-by: NAlexander Duyck <alexander.h.duyck@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

The advertising 10G LR mode should be possible to set
but in the function i40e_set_link_ksettings() check for this
is missed. This patch adds check for 10000baseLR_Full
flag for 10G modes.
Signed-off-by: NJakub Pawlak <jakub.pawlak@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

Previous method for reading LLDP config was based on hard-coded
offsets. It happened to work, because of structured architecture of
the NVM memory. In the new approach, known as FLAT, we need to
calculate the absolute address, instead of using relative values.
Needed defines for memory location were added.
Signed-off-by: NMariusz Stachura <mariusz.stachura@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

Recent versions of the Linux kernel now warn about incorrect parameter
definitions for function comments. Fix up several function comments to
correctly reflect the current function arguments. This cleans up the
warnings and helps ensure our documentation is accurate.
Signed-off-by: NJacob Keller <jacob.e.keller@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

After many years of having a ~30 line copyright and license header to our
source files, we are finally able to reduce that to one line with the
advent of the SPDX identifier.

Also caught a few files missing the SPDX license identifier, so fixed
them up.
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
Acked-by: NShannon Nelson <shannon.nelson@oracle.com>
Acked-by: NRichard Cochran <richardcochran@gmail.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Changing API ndo_xdp_xmit to take a struct xdp_frame instead of struct
xdp_buff.  This brings xdp_return_frame and ndp_xdp_xmit in sync.

This builds towards changing the API further to become a bulk API,
because xdp_buff is not a queue-able object while xdp_frame is.

V4: Adjust for commit 59655a5b ("tuntap: XDP_TX can use native XDP")
V7: Adjust for commit d9314c47 ("i40e: add support for XDP_REDIRECT")
Signed-off-by: NJesper Dangaard Brouer <brouer@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>