Improve the performance of the AF_XDP zero-copy Tx completion
path. When there are no XDP buffers being sent using XDP_TX or
XDP_REDIRECT, we do not have go through the SW ring to clean up any
entries since the AF_XDP path does not use these. In these cases, just
fast forward the next-to-use counter and skip going through the SW
ring. The limit on the maximum number of entries to complete is also
removed since the algorithm is now O(1). To simplify the code path, the
maximum number of entries to complete for the XDP path is therefore
also increased from 256 to 512 (the default number of Tx HW
descriptors). This should be fine since the completion in the XDP path
is faster than in the SKB path that has 256 as the maximum number.

This patch provides around 4% throughput improvement for the l2fwd
application in xdpsock on my machine.
Signed-off-by: NMagnus Karlsson <magnus.karlsson@intel.com>
Reviewed-by: NSridhar Samudrala <sridhar.samudrala@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NTony Nguyen <anthony.l.nguyen@intel.com>

Remove all the unused defines as they are just dead weight.
Signed-off-by: NJesse Brandeburg <jesse.brandeburg@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

Remove MEM_TYPE_ZERO_COPY in favor of the new MEM_TYPE_XSK_BUFF_POOL
APIs. The AF_XDP zero-copy rx_bi ring is now simply a struct xdp_buff
pointer.

v4->v5: Fixed "warning: Excess function parameter 'bi' description in
        'i40e_construct_skb_zc'". (Jakub)
Signed-off-by: NBjörn Töpel <bjorn.topel@intel.com>
Signed-off-by: NAlexei Starovoitov <ast@kernel.org>
Cc: intel-wired-lan@lists.osuosl.org
Link: https://lore.kernel.org/bpf/20200520192103.355233-9-bjorn.topel@gmail.com

Continuing the path to support MEM_TYPE_XSK_BUFF_POOL, the AF_XDP
zero-copy/sk_buff rx_bi rings are now separate. Functions to properly
allocate the different rings are added as well.

v3->v4: Made i40e_fd_handle_status() static. (kbuild test robot)
v4->v5: Fix kdoc for i40e_clean_programming_status(). (Jakub)
Signed-off-by: NBjörn Töpel <bjorn.topel@intel.com>
Signed-off-by: NAlexei Starovoitov <ast@kernel.org>
Cc: intel-wired-lan@lists.osuosl.org
Link: https://lore.kernel.org/bpf/20200520192103.355233-8-bjorn.topel@gmail.com

In preparation for unifying the skb_frag and bio_vec, use the fine
accessors which already exist and use skb_frag_t instead of
struct skb_frag_struct.
Signed-off-by: NMatthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This patch adds zero-copy Rx support for AF_XDP sockets. Instead of
allocating buffers of type MEM_TYPE_PAGE_SHARED, the Rx frames are
allocated as MEM_TYPE_ZERO_COPY when AF_XDP is enabled for a certain
queue.

All AF_XDP specific functions are added to a new file, i40e_xsk.c.

Note that when AF_XDP zero-copy is enabled, the XDP action XDP_PASS
will allocate a new buffer and copy the zero-copy frame prior passing
it to the kernel stack.
Signed-off-by: NBjörn Töpel <bjorn.topel@intel.com>
Signed-off-by: NAlexei Starovoitov <ast@kernel.org>

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>

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>

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>

Also convert driver i40e, which very recently got XDP_REDIRECT support
in commit d9314c47 ("i40e: add support for XDP_REDIRECT").

V7: This patch got added in V7 of this patchset.
Signed-off-by: NJesper Dangaard Brouer <brouer@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The driver now acts upon the XDP_REDIRECT return action. Two new ndos
are implemented, ndo_xdp_xmit and ndo_xdp_flush.

XDP_REDIRECT action enables XDP program to redirect frames to other
netdevs.
Signed-off-by: NBjörn Töpel <bjorn.topel@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

Add the SPDX identifiers to all the Intel wired LAN driver files, as
outlined in Documentation/process/license-rules.rst.
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
Tested-by: NAaron Brown <aaron.f.brown@intel.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The i40e_detect_recover_hung function uses the i40e_get_tx_pending
function to determine if there are packets stalled on the ring.
i40e_get_tx_pending calculates the pending packets using the head
writeback value and HW tail.  If the queue is stopped and we lose the
interrupt to update our next_to_clean then we a) won't get another
interrupt to clean because queue is stopped b) we won't catch the
problem with i40e_detect_recover_hung because the HW values look like
there's no packets waiting to be transmitted.  Using the SW values we
can catch the issue because next_to_clean will be out of sync with head
writeback.

This has the added benefit being less CPU intensive because we don't
need to reach into the hardware to get the values.
Signed-off-by: NAlan Brady <alan.brady@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

This patch replaces the existing mechanism for determining the correct
value to program for adaptive ITR with yet another new and more
complicated approach.

The basic idea from a 30K foot view is that this new approach will push the
Rx interrupt moderation up so that by default it starts in low latency and
is gradually pushed up into a higher latency setup as long as doing so
increases the number of packets processed, if the number of packets drops
to 4 to 1 per packet we will reset and just base our ITR on the size of the
packets being received. For Tx we leave it floating at a high interrupt
delay and do not pull it down unless we start processing more than 112
packets per interrupt. If we start exceeding that we will cut our interrupt
rates in half until we are back below 112.

The side effect of these patches are that we will be processing more
packets per interrupt. This is both a good and a bad thing as it means we
will not be blocking processing in the case of things like pktgen and XDP,
but we will also be consuming a bit more CPU in the cases of things such as
network throughput tests using netperf.

One delta from this versus the ixgbe version of the changes is that I have
made the interrupt moderation a bit more aggressive when we are in bulk
mode by moving our "goldilocks zone" up from 48 to 96 to 56 to 112. The
main motivation behind moving this is to address the fact that we need to
update less frequently, and have more fine grained control due to the
separate Tx and Rx ITR times.
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>

This patch is mostly prep-work for replacing the current approach to
programming the dynamic aka adaptive ITR. Specifically here what we are
doing is splitting the Tx and Rx ITR each into two separate values.

The first value current_itr represents the current value of the register.

The second value target_itr represents the desired value of the register.

The general plan by doing this is to allow for deferring the update of the
ITR value under certain circumstances. For now we will work with what we
have, but in the future I hope to change the behavior so that we always
only update one ITR at a time using some simple logic to determine which
ITR requires an update.
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>

Instead of using the register value for the defines when setting up the
ring ITR we can just use the actual values and avoid the use of shifts and
macros to translate between the values we have and the values we want.

This helps to make the code more readable as we can quickly translate from
one value to the other.
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 rings are already split out into Tx and Rx rings so it doesn't make
sense to have any single ring store both a Tx and Rx itr_setting value.
Since that is the case drop the pair in favor of storing just a single ITR
value.
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>

When compared to ixgbe and other previous Intel drivers the i40e and i40evf
drivers actually reserve 2 additional descriptors in maybe_stop_tx for
cache line alignment. We need to update DESC_NEEDED to reflect this as
otherwise we are more likely to return TX_BUSY which will cause issues with
things like xmit_more.
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>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

In VFs, there is a known issue which can cause writebacks
to not occur when interrupts are disabled and there are
less than 4 descriptors resulting in TX timeout. Timeout
can also occur due to lost interrupt.

The current implementation for detecting and recovering
from hung queues in the PF is problematic because it actually
actively encourages lost interrupts.  By triggering a SW
interrupt, interrupts are forced on.  If we are already in
napi_poll and an interrupt fires, napi_poll will not be
rescheduled and the interrupt is effectively lost; thereby
potentially *causing* hung queues.

This patch checks whether packets are being processed between
every watchdog cycle and determine potential hung queue and
fires triggers SW interrupt only for that particular queue.
Signed-off-by: NSudheer Mogilappagari <sudheer.mogilappagari@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

The i40e driver has a special "FDIR" RX-ring (I40E_VSI_FDIR) which is
a sideband channel for configuring/updating the flow director tables.
This (i40e_vsi_)type does not invoke XDP-ebpf code.

As suggested by Björn (V2): Instead of marking this I40E_VSI_FDIR RX-ring
a special case, reverse the logic and only select RX-rings of type
I40E_VSI_MAIN to register xdp_rxq_info's for.

Driver hook points for xdp_rxq_info:
 * reg  : i40e_setup_rx_descriptors (via i40e_vsi_setup_rx_resources)
 * unreg: i40e_free_rx_resources    (via i40e_vsi_free_rx_resources)

Tested on actual hardware with samples/bpf program.

V2: Fixed bug in i40e_set_ringparam (memset zero) + match on I40E_VSI_MAIN.
V4: Update patch desc that got out-of-sync with code.

Cc: intel-wired-lan@lists.osuosl.org
Cc: Björn Töpel <bjorn.topel@intel.com>
Cc: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Cc: Paul Menzel <pmenzel@molgen.mpg.de>
Signed-off-by: NJesper Dangaard Brouer <brouer@redhat.com>
Reviewed-by: NPaul Menzel <pmenzel@molgen.mpg.de>
Acked-by: NJohn Fastabend <john.fastabend@gmail.com>
Signed-off-by: NAlexei Starovoitov <ast@kernel.org>

This patch sets up the infrastructure for offloading TCs and
queue configurations to the hardware by creating HW channels(VSI).
A new channel is created for each of the traffic class
configuration offloaded via mqprio framework except for the first TC
(TC0). TC0 for the main VSI is also reconfigured as per user provided
queue parameters. Queue counts that are not power-of-2 are handled by
reconfiguring RSS by reprogramming LUTs using the queue count value.
This patch also handles configuring the TX rings for the channels,
setting up the RX queue map for channel.

Also, the channels so created are removed and all the queue
configuration is set to default when the qdisc is detached from the
root of the device.
Signed-off-by: NAmritha Nambiar <amritha.nambiar@intel.com>
Signed-off-by: NKiran Patil <kiran.patil@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

Double the number of descriptors we'll bundle into one tail bump when
receiving. Empirical testing has shown that we reduce CPU utilization
and don't appear to reduce throughput or packet rate. 32 seems to be the
sweet spot, as it's half the default polling budget, so we'd essentially
reduce from 4 tail writes when polling down to 2. Increasing this up to
64 appears to have negative impacts as it may become possible that we
don't bump the tail each time we get polled, which could cause a long
delay between returning descriptors to the hardware.
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 ITR register expects to be programmed in units of 2 microseconds.
Because of this, all of the drivers I40E_ITR_* constants are in terms of
this 2 microsecond register.

Unfortunately, the rx_itr_default value is expected to be programmed in
microseconds.

Effectively the driver defaults to an ITR value of half the expected
value (in terms of minimum microseconds between interrupts).

Fix this by changing the default values to be calculated using
ITR_REG_TO_USEC macro which indicates that we're converting from the
register units into microseconds.
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>

When using set_bit and friends, we should be using actual
bitmaps, and fix all the locations where we might access
it.
Signed-off-by: NJesse Brandeburg <jesse.brandeburg@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

The dynamic ITR algorithm depends on a calculation of usecs which
assumes that the interrupts have been firing constantly at the interrupt
throttle rate. This is not guaranteed because we could have a low packet
rate, or have been polling in software.

We'll estimate whether this is the case by using jiffies to determine if
we've been too long. If the time difference of jiffies is larger we are
guaranteed to have an incorrect calculation. If the time difference of
jiffies is smaller we might have been polling some but the difference
shouldn't affect the calculation too much.

This ensures that we don't get stuck in BULK latency during certain rare
situations where we receive bursts of packets that force us into NAPI
polling.
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>

Since commit c56625d5 ("i40e/i40evf: change dynamic interrupt
thresholds") a new higher latency ITR setting called I40E_ULTRA_LATENCY
was added with a cryptic comment about how it was meant for adjusting Rx
more aggressively when streaming small packets.

This mode was attempting to calculate packets per second and then kick
in when we have a huge number of small packets.

Unfortunately, the ULTRA setting was kicking in for workloads it wasn't
intended for including single-thread UDP_STREAM workloads.

This wasn't caught for a variety of reasons. First, the ip_defrag
routines were improved somewhat which makes the UDP_STREAM test still
reasonable at 10GbE, even when dropped down to 8k interrupts a second.
Additionally, some other obvious workloads appear to work fine, such
as TCP_STREAM.

The number 40k doesn't make sense for a number of reasons. First, we
absolutely can do more than 40k packets per second. Second, we calculate
the value inline in an integer, which sometimes can overflow resulting
in using incorrect values.

If we fix this overflow it makes it even more likely that we'll enter
ULTRA mode which is the opposite of what we want.

The ULTRA mode was added originally as a way to reduce CPU utilization
during a small packet workload where we weren't keeping up anyways. It
should never have been kicking in during these other workloads.

Given the issues outlined above, let's remove the ULTRA latency mode. If
necessary, a better solution to the CPU utilization issue for small
packet workloads will be added 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>

The number of flags found in pf->flags has grown quite large, and there
are a lot of different types of flags. Most of the flags are simply
hardware features which are enabled on some firmware or some MAC types.
Other flags are dynamic run-time flags which enable or disable certain
features of the driver.

Separate these two types of flags into pf->hw_features and pf->flags.
The hw_features list will contain a set of features which are enabled at
init time. This will not contain toggles or otherwise dynamically
changing features. These flags should not need atomic protections, as
they will be set once during init and then be essentially read only.

Everything else will remain in the flags variable. These flags may be
modified at any time during run time. A future patch may wish to convert
these flags into set_bit/clear_bit/test_bit or similar approach to
ensure atomic correctness.

The I40E_FLAG_MFP_ENABLED flag may be a good fit for hw_features but
currently is used by ethtool in the private flags settings, and thus has
been left as part of flags.

Additionally, I40E_FLAG_DCB_CAPABLE may be a good fit for the
hw_features but this patch has not tried to untangle it yet.
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>

Now that the kernel supports double VLAN tags, we should at least play
nice. Adjust the max packet size to account for two VLAN tags, not just
one.
Signed-off-by: NMitch Williams <mitch.a.williams@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

This patch adds proper XDP_TX action support. For each Tx ring, an
additional XDP Tx ring is allocated and setup. This version does the
DMA mapping in the fast-path, which will penalize performance for
IOMMU enabled systems. Further, debugfs support is not wired up for
the XDP Tx rings.
Signed-off-by: NBjörn Töpel <bjorn.topel@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

This commit adds basic XDP support for i40e derived NICs. All XDP
actions will end up in XDP_DROP.
Signed-off-by: NBjörn Töpel <bjorn.topel@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

This patch adds padding to the start of frames to make room for headroom
for us to eventually start using build_skb.  Right now we guarantee at
least NET_SKB_PAD + NET_IP_ALIGN, however we allocate more space if more is
available.  For example on x86 the headroom should be 192 bytes.

On systems that have too large of a cache line size to support storing 1.5K
padding and shared info we default to using 3K buffers and reserve
everything that isn't used for skb_shared_info or the data buffer for
headroom.

Change-ID: I33c641c9a1ea10cf7cc484c2d20985368d2d709a
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>

There are situations where adding padding to the front and back of an Rx
buffer will require that we add additional padding.  Specifically if
NET_IP_ALIGN is non-zero, or the MTU size is larger than 7.5K we would need
to use 2K buffers which leaves us with no room for the padding.

To preemptively address these cases I am adding support for 3K buffers to
the Rx path so that we can provide the additional padding needed in the
event of NET_IP_ALIGN being non-zero or a cache line being greater than 64.

Change-ID: I938bc1ba611285428df39a613cd66f98e60b55c7
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>

This patch greatly reduces the unneeded complexity in the
i40e_detect_recover_hung_queue code path.  The previous implementation
set a 'hung bit' which would then get cleared while polling.  If the
detection routine was called a second time with the bit already set, we
would issue a software interrupt.  This patch makes it such that if
interrupts are disabled and we have pending TX descriptors, we trigger a
software interrupt since in, the worst case, queues are already clean
and we have an extra interrupt.

Additionally this patch removes the workaround for lost interrupts as
calling napi_reschedule in this context can cause software interrupts to
fire on the wrong CPU.

Change-ID: Iae108582a3ceb6229ed1d22e4ed6e69cf97aad8d
Signed-off-by: NAlan Brady <alan.brady@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

This patch changes the way we handle the maximum frame size for the Rx
path.  Previously we were rounding up to 2K for a 1500 MTU and then brining
the max frame size down to MTU plus a fixed amount.  With this patch
applied what we now do is limit the maximum frame to 1.5K minus the value
for NET_IP_ALIGN for standard MTU, and for any MTU greater than 1500 we
allow up to the maximum frame size.  This makes the behavior more
consistent with the other drivers such as igb which had similar logic.  In
addition it reduces the test matrix for MTU since we only have two max
frame sizes that are handled for Rx now.

Change-ID: I23a9d3c857e7df04b0ef28c64df63e659c013f3f
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>

Looking over the code for FCoE it looks like the Rx path has been broken at
least since the last major Rx refactor almost a year ago.  It seems like
FCoE isn't supported for any of the Fortville/Fortpark hardware so there
isn't much point in carrying the code around, especially if it is broken
and untested.

Change-ID: I892de8fa551cb129ce2361e738ff82ce55fa229e
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>

Update the driver code so that we do bulk updates of the page reference
count instead of just incrementing it by one reference at a time.  The
advantage to doing this is that we cut down on atomic operations and
this in turn should give us a slight improvement in cycles per packet.
In addition if we eventually move this over to using build_skb the gains
will be more noticeable.

I also found and fixed a store forwarding stall from where we were
assigning "*new_buff = *old_buff".  By breaking it up into individual
copies we can avoid this and as a result the performance is slightly
improved.

Change-ID: I1d3880dece4133eca3c32423b04a5467321ccc52
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>

This patch adds support for DMA_ATTR_SKIP_CPU_SYNC and
DMA_ATTR_WEAK_ORDERING. By enabling both of these for the Rx path we
are able to see performance improvements on architectures that implement
either one due to the fact that page mapping and unmapping only has to
sync what is actually being used instead of the entire buffer. In addition
by enabling the weak ordering attribute enables a performance improvement
for architectures that can associate a memory ordering with a DMA buffer
such as Sparc.

Change-ID: If176824e8231c5b24b8a5d55b339a6026738fc75
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>

This patch reduces the size of struct i40e_rx_buffer by one pointer,
and makes the i40e driver a little more consistent with the igb driver
in terms of packets that span buffers.

We do this by moving the skb field from struct i40e_rx_buffer to
struct i40e_ring. We pass the skb we already have (or NULL if we
don't) to i40e_fetch_rx_buffer(), which skips the skb allocation if we
already have one for this packet.

Change-ID: I4ad48a531844494ba0c5d8e1a62209a057f661b0
Signed-off-by: NScott Peterson <scott.d.peterson@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>