We used to queue tx packets in sk_receive_queue, this is less
efficient since it requires spinlocks to synchronize between producer
and consumer.

This patch tries to address this by:

- switch from sk_receive_queue to a skb_array, and resize it when
  tx_queue_len was changed.
- introduce a new proto_ops peek_len which was used for peeking the
  skb length.
- implement a tun version of peek_len for vhost_net to use and convert
  vhost_net to use peek_len if possible.

Pktgen test shows about 15.3% improvement on guest receiving pps for small
buffers:

Before: ~1300000pps
After : ~1500000pps
Signed-off-by: NJason Wang <jasowang@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Point the ethtool .get_link() callback to the standard
ethtool_op_get_link() implementation.
Signed-off-by: NJakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

nfp_net_write_mac_addr() always writes to the BAR the current
device address taken from netdev struct.  The address given
as parameter is actually ignored.  Since all callers pass
netdev->dev_addr simply remove the parameter.

While at it improve the function's kdoc a bit.
Signed-off-by: NJakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Spell abbreviation of control as ctrl not crtl.
Signed-off-by: NJakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

"num_vec" needs to be signed for the error handling to work.

Fixes: e261768e ('be2net: support asymmetric rx/tx queue counts')
Signed-off-by: NDan Carpenter <dan.carpenter@oracle.com>
Acked-by: NSathya Perla <sathya.perla@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The ethernet core has 3 IRQs. Using the IRQ grouping registers we are able
to separate TX and RX IRQs, which allows us to service them on separate
cores. This patch splits the IRQ handler into 2 separate functions, one for
TX and another for RX. The TX housekeeping is split out into its own NAPI
handler.
Signed-off-by: NJohn Crispin <john@phrozen.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The code that enables and disables IRQs is missing proper locking. After
adding the IRQ grouping patch and routing the RX and TX IRQs to different
cores we experienced IRQ stalls. Fix this by adding proper locking.
We use a dedicated lock to reduce the latency if the IRQ code.
Signed-off-by: NJohn Crispin <john@phrozen.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The code currently uses variables to store and never modify the bit masks
of interrupts. This is legacy code from an early version of the driver
that supported MIPS based SoCs where the IRQ bits depended on the actual
SoC. As the bits are the same for all ARM based SoCs using this driver we
can remove the intermediate variables.
Signed-off-by: NJohn Crispin <john@phrozen.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The driver was originally written for MIPS based SoC. These required the
IRQ mask register to be read after writing it to ensure that the content
was actually applied. As this version only works on ARM based SoCs, we can
safely remove the 2 reads as they are no longer required.
Signed-off-by: NJohn Crispin <john@phrozen.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

On some platforms, syncing a buffer for DMA is expensive. Rather than
sync the whole 2K receive buffer, only synchronise the length of the
frame, which will typically be the MTU, or a much smaller TCP ACK.

For an IMX6Q, this gives around 6% increased TCP receive performance,
which is cache operations bound and reduces CPU load for TCP transmit.
Signed-off-by: NAndrew Lunn <andrew@lunn.ch>
Tested-by: NAaron Brown <aaron.f.brown@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

When setting spoofing, both VLAN and MAC need to be set together.
This change resolves an issue where MAC-VLANs on the VF fail to pass
traffic due to spoofed packets.
Signed-off-by: NEmil Tantilov <emil.s.tantilov@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

Update ixgbe_ethtool_get_ts_info() to show that x550 supports hardware
timestamping of all packets.
Reported-by: NGuy Harris <guy@alum.mit.edu>
Signed-off-by: NTony Nguyen <anthony.l.nguyen@intel.com>
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>

alloc_workqueue replaces deprecated create_workqueue().

A dedicated workqueue has been used since the workitem (viz
fm10k_service_task, which manages and runs other subtasks) is involved in
normal device operation and requires forward progress under memory
pressure.

create_workqueue has been replaced with alloc_workqueue with max_active
as 0 since there is no need for throttling the number of active work
items.

Since network devices may be used in memory reclaim path,
WQ_MEM_RECLAIM has been set to guarantee forward progress.

flush_workqueue is unnecessary since destroy_workqueue() itself calls
drain_workqueue() which flushes repeatedly till the workqueue
becomes empty. Hence the call to flush_workqueue() has been dropped.
Signed-off-by: NBhaktipriya Shridhar <bhaktipriya96@gmail.com>
Acked-by: NTejun Heo <tj@kernel.org>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

Properly stop the extra workqueue items and ensure that we resume
cleanly. This is better than using igb_ptp_init and igb_ptp_stop since
these functions destroy the PHC device, which will cause other problems
if we do so. Since igb_ptp_reset now re-schedules the work-queue item we
don't need an equivalent igb_ptp_resume in the resume workflow.
Signed-off-by: NJacob Keller <jacob.e.keller@intel.com>
Tested-by: NAaron Brown <aaron.f.brown@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

Make igb_ptp_stop take advantage of this new function to reduce code
duplication.
Signed-off-by: NJacob Keller <jacob.e.keller@intel.com>
Tested-by: NAaron Brown <aaron.f.brown@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

Modify igb_ptp_init to take advantage of igb_ptp_reset, and remove
duplicated work that was occurring in both igb_ptp_reset and
igb_ptp_init.

In total, resetting the TSAUXC register, and resetting the system time
both happen in igb_ptp_reset already. igb_ptp_reset now also takes care
of starting the delayed work item for overflow checks, as well.
Signed-off-by: NJacob Keller <jacob.e.keller@intel.com>
Tested-by: NAaron Brown <aaron.f.brown@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

Don't continue to use complex MAC type checks for handling various cases
where we have overflow check code. Make this code more obvious by
introducing a flag which is enabled for hardware that needs these
checks.
Signed-off-by: NJacob Keller <jacob.e.keller@intel.com>
Tested-by: NAaron Brown <aaron.f.brown@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

Upcoming patches will introduce new PTP specific flags. To avoid
cluttering the normal flags variable, introduce PTP specific "ptp_flags"
variable for this purpose, and move IGB_FLAG_PTP to become
IGB_PTP_ENABLED.
Signed-off-by: NJacob Keller <jacob.e.keller@intel.com>
Tested-by: NAaron Brown <aaron.f.brown@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

For u32 classifier filters, avoid overwriting existing filter
in a hardware location without removing it first, to clean up
inconsistencies due to duplicate values for filter location.

Verified with the following filters:

Create child hash tables:
	handle 1: u32 divisor 1
	handle 2: u32 divisor 1

Link to the child hash table from parent hash table:
	handle 800:0:11 u32 ht 800: link 1: \
	offset at 0 mask 0f00 shift 6 plus 0 eat \
	match ip protocol 6 ff match ip dst 15.0.0.1/32

	handle 800:0:12 u32 ht 800: link 2: \
	offset at 0 mask 0f00 shift 6 plus 0 eat \
	match ip protocol 17 ff match ip dst 16.0.0.1/32

Add filter into child hash table:
	handle 1:0:3 u32 ht 1: \
	match tcp src 22 ffff action drop

Add another filter to the same location:
	handle 2:0:3 u32 ht 2: \
	match tcp src 33 ffff action drop
Signed-off-by: NAmritha Nambiar <amritha.nambiar@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

On deleting filters which are links to a child hash table, the filters
in the child hash table must be cleared from the hardware if there
is no link between the parent and child hash table.

Verified with the following filters:

Create a child hash table:
	handle 1: u32 divisor 1

Link to the child hash table from parent hash table:
	handle 800:0:10 u32 ht 800: link 1: \
	offset at 0 mask 0f00 shift 6 plus 0 eat \
	match ip protocol 6 ff match ip dst 15.0.0.1/32

Add filters into child hash table:
	handle 1:0:2 u32 ht 1: \
	match tcp src 22 ffff action drop
        handle 1:0:3 u32 ht 1: \
        match tcp src 33 ffff action drop

Delete link filter from parent hash table:
	handle 800:0:10 u32
Signed-off-by: NAmritha Nambiar <amritha.nambiar@intel.com>
Acked-by: NSridhar Samudrala <sridhar.samudrala@intel.com>
Tested-by: NAndrew Bowers <andrewx.bowers@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

Users report that under VMWare, er32(TIMINCA) returns zero.
This causes division by zero at init time as follows:

 ==>       incvalue = er32(TIMINCA) & E1000_TIMINCA_INCVALUE_MASK;
           for (i = 0; i < E1000_MAX_82574_SYSTIM_REREADS; i++) {
                   /* latch SYSTIMH on read of SYSTIML */
                   systim_next = (cycle_t)er32(SYSTIML);
                   systim_next |= (cycle_t)er32(SYSTIMH) << 32;

                   time_delta = systim_next - systim;
                   temp = time_delta;
 ====>             rem = do_div(temp, incvalue);

This change makes kernel survive this, and users report that
NIC does work after this change.

Since on real hardware incvalue is never zero, this should not affect
real hardware use case.
Signed-off-by: NDenys Vlasenko <dvlasenk@redhat.com>
Tested-by: NAaron Brown <aaron.f.brown@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

The index calculated when looping through the indir array passed to
fm10k_write_reta was incorrectly calculated as the first part i needs to
be multiplied by 4.

Fixes: 0cfea7a65738 ("fm10k: fix possible null pointer deref after kcalloc", 2016-04-13)
Signed-off-by: NJacob Keller <jacob.e.keller@intel.com>
Tested-by: NKrishneil Singh <Krishneil.k.singh@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

While reviewing the i40e driver changes to support page based receive I
realized that I had overlooked the fact that the fm10k hardware required a
512 byte alignment for Rx buffers.  This patch is meant to address that by
changing the alignment for Rx buffers to 512 bytes instead of allowing it
to be L1 cache aligned.
Signed-off-by: NAlexander Duyck <aduyck@mirantis.com>
Tested-by: NKrishneil Singh <Krishneil.k.singh@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

The FM10K_MAX_DATA_PER_TXD is really just using a bitshift as a power of
2 operation in an efficient manner. We shouldn't represent this as a BIT()
because that obscures the intention of the operation.
Signed-off-by: NJacob Keller <jacob.e.keller@intel.com>
Tested-by: NKrishneil Singh <Krishneil.k.singh@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>

There are two generics functions phy_ethtool_{get|set}_link_ksettings,
so we can use them instead of defining the same code in the driver.
Signed-off-by: NPhilippe Reynes <tremyfr@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The private structure contain a pointer to phydev, but the structure
net_device already contain such pointer. So we can remove the pointer
phydev in the private structure, and update the driver to use the
one contained in struct net_device.
Signed-off-by: NPhilippe Reynes <tremyfr@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

SPQ doorbell is currently protected with the compilation barrier. Under the
stress scenarios, we may get into a state where (due to the weak ordering)
several ramrod doorbells were written to the BAR with an out-of-order
producer values. Need to change the barrier type to a write barrier to make
sure that the write buffer is flushed after each doorbell.
Signed-off-by: NSudarsana Reddy Kalluru <sudarsana.kalluru@qlogic.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Static checker warnings:
drivers/net/ethernet/qlogic/qede/qede_ethtool.c:435 qede_get_coalesce()
warn: passing casted pointer '&coal->rx_coalesce_usecs' to
'edev->ops->common->get_coalesce()' 32 vs 16.

The u32 pointer is being typecasted to u16 which may fail for big-endian
platforms.

Fixes: d552fa84 ("qede: Add support for coalescing config read/update.")
Reported-by: NDan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: NSudarsana Reddy Kalluru <sudarsana.kalluru@qlogic.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Static checker warnings:
drivers/net/ethernet/qlogic/qed/qed_int.c:2450 qed_init_cau_sb_entry()
warn: always true condition '(cdev->rx_coalesce_usecs <= 255) =>
(0-255 <= 255)'
drivers/net/ethernet/qlogic/qed/qed_int.c:2511 qed_int_cau_conf_sb()
warn: always true condition '(p_hwfn->cdev->rx_coalesce_usecs <= 255)
=> (0-255 <= 255)'
..

The data types for rx/tx_coalesce_usecs should be u16.

Fixes: commit 722003ac ("qed: Add support for coalescing config read/update.")
Reported-by: NDan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: NSudarsana Reddy Kalluru <sudarsana.kalluru@qlogic.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

I've got a bug report about an e1000e interface, where a VLAN interface is
set up on top of it:

$ ip link add link ens1f0 name ens1f0.99 type vlan id 99
$ ip link set ens1f0 up
$ ip link set ens1f0.99 up
$ ip addr add 192.168.99.92 dev ens1f0.99

At this point, I can ping another host on vlan 99, ip 192.168.99.91.
However, if I do the following:

$ ethtool -K ens1f0 rxvlan off

Then no traffic passes on ens1f0.99. It comes back if I toggle rxvlan on
again. I'm not sure if this is actually intended behavior, or if there's a
lack of software VLAN stripping fallback, or what, but things continue to
work if I simply don't call e1000e_vlan_strip_disable() if there are
active VLANs (plagiarizing a function from the e1000 driver here) on the
interface.

Also slipped a related-ish fix to the kerneldoc text for
e1000e_vlan_strip_disable here...
Signed-off-by: NJarod Wilson <jarod@redhat.com>
Tested-by: NAaron Brown <aaron.f.brown@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

There are two generics functions phy_ethtool_{get|set}_link_ksettings,
so we can use them instead of defining the same code in the driver.
Signed-off-by: NPhilippe Reynes <tremyfr@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The private structure contain a pointer to phydev, but the structure
net_device already contain such pointer. So we can remove the pointer
phydev in the private structure, and update the driver to use the
one contained in struct net_device.
Signed-off-by: NPhilippe Reynes <tremyfr@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

There is a static checker warning here "warn: mask and shift to zero"
and the code sets "ring" to zero every time.  From looking at how
QLCNIC_FETCH_RING_ID() is used in qlcnic_83xx_process_rcv_ring() the
qlcnic_83xx_hndl() should be removed.

Fixes: 4be41e92 ('qlcnic: 83xx data path routines')
Signed-off-by: NDan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Since we will remove items off the list using list_del() we need
to use a safe version of the list_for_each() macro aptly named
list_for_each_safe().
Signed-off-by: NWei Yongjun <yongjun_wei@trendmicro.com.cn>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

TL4 calculation for a given SQ of secondary Qsets is incorrect
and goes out of bounds and also for some SQ's TL4 chosen will
transmit data via a different BGX interface and not same as
primary Qset's interface.

This patch fixes this issue.
Signed-off-by: NSunil Goutham <sgoutham@cavium.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Check for SMU RX local/remote faults along with SPU LINK
status. Otherwise at times link is UP at our end but DOWN
at link partner's side. Also due to an issue in BGX it's
rarely seen that initialization doesn't happen properly
and SMU RX reports faults with everything fine at SPU.
This patch tries to reinitialize LMAC to fix it.

Also fixed LMAC disable sequence to properly bring down link.
Signed-off-by: NSunil Goutham <sgoutham@cavium.com>
Signed-off-by: NTao Wang <tao.wang@cavium.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

In some configurations, gcc produces a warning for correct code
in this driver:

drivers/net/dsa/b53/b53_mmap.c: In function 'b53_mmap_read64':
drivers/net/dsa/b53/b53_mmap.c:107:10: error: 'hi' may be used uninitialized in this function [-Werror=maybe-uninitialized]
  *val = ((u64)hi << 32) | lo;
          ^~~~~~~
drivers/net/dsa/b53/b53_mmap.c: In function 'b53_mmap_read48':
drivers/net/dsa/b53/b53_mmap.c:91:11: error: 'hi' may be used uninitialized in this function [-Werror=maybe-uninitialized]
   *val = ((u64)hi << 32) | lo;
           ^~~~~~~
drivers/net/dsa/b53/b53_mmap.c:83:11: error: 'hi' may be used uninitialized in this function [-Werror=maybe-uninitialized]
   *val = ((u64)hi << 16) | lo;

I have seen the warning before and at the time thought I had fixed
it with 55e7f6ab ("dsa: b53: fix big-endian register access"),
however it now came back in a different randconfig build that happens
to have different inlining decisions in the compiler.

The mistake that gcc makes here is that it thinks the second call to
readl() might fail because the address 'reg + 4' is not a multiple
of four despite having knowing that 'reg' itself is a multiple of four.

By open-coding the two reads without the redundant alignment check,
we can avoid the warning and produce slightly better object code, but
get slightly longer source code instead.
Signed-off-by: NArnd Bergmann <arnd@arndb.de>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Categorize and reorganize ethtool statistics counters by renaming to
"rx_*" and "tx_*" and removing redundant and duplicated counters, this
way they are easier to grasp and more user friendly.
Signed-off-by: NGal Pressman <galp@mellanox.com>
Signed-off-by: NSaeed Mahameed <saeedm@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Number of PFC counters used to count only number of priorities with PFC
enabled, but each priority has more than one counter, hence the need to
multiply it by the number of PFC counters per priority.

Fixes: cf678570 ('net/mlx5e: Add per priority group to PPort counters')
Signed-off-by: NGal Pressman <galp@mellanox.com>
Signed-off-by: NSaeed Mahameed <saeedm@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Do not allow the same vxlan udp port to be added to the device more than
once.

Fixes: b3f63c3d ("net/mlx5e: Add netdev support for VXLAN tunneling")
Signed-off-by: NMatthew Finlay <matt@mellanox.com>
Signed-off-by: NSaeed Mahameed <saeedm@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>