The Rx BSY error interrupt indicates that a frame was
received and discarded due to lack of buffers, so it's
a rx ring overflow condition and has nothing to do with
with bad rx packets.  Use the right counter.

BSY conditions happen when the SoC is under performance
stress.  Doing *more* work in stress situations by trying
to schedule NAPI is not a good idea as the stressed system
becomes still more stressed.  The Rx interrupt is already
at work making sure the NAPI is scheduled.
So calling gfar_receive() here does not help.  This issue
was present since day 1.
Signed-off-by: NClaudiu Manoil <claudiu.manoil@freescale.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Under one unusual circumstance it's possible to wrongly set
FILREN without enabling PRSDEP as well in the RCTRL register,
against the hardware specifications.  With the default config
this does not happen because the default Rx offloads (Rx csum
and Rx VLAN) properly enable PRSDEP.  But if anyone disables
all these offloads (via ethtool), we get a wrong configuration
were the Rx flow classification and hashing, and other Filer
based features (e.g. wake-on-filer interrupt) won't work.
This patch fixes the issue.
Also, account for Rx FCB insertion which happens every time
PRSDEP is set.
Signed-off-by: NClaudiu Manoil <claudiu.manoil@freescale.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

RQFCR_AND is duplicated.
Add missing space as well.
Signed-off-by: NClaudiu Manoil <claudiu.manoil@freescale.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Add support for ThunderX pass2 CPI and MPI configuration changes.
MPI_ALG is not enabled i.e MCAM parsing is disabled.
Signed-off-by: NThanneeru Srinivasulu <tsrinivasulu@caviumnetworks.com>
Signed-off-by: NSunil Goutham <sgoutham@cavium.com>
Signed-off-by: NDavid Daney <david.daney@cavium.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The test for pass-1 silicon was incorrect, it should be for all
revisions less than 8.  Also the revision is already present in the
pci_dev, so there is no need to read and keep a private copy.

Remove rev_id and code to read it from struct nicpf.  Create new
static inline function pass1_silicon() to be used to testing the
silicon version.  Use pass1_silicon() for revision checks, this will
be more widely used in follow on patches.
Signed-off-by: NDavid Daney <david.daney@cavium.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Signed-off-by: NSunil Goutham <sgoutham@cavium.com>
Signed-off-by: NDavid Daney <david.daney@cavium.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

In some silicon revisions, the soft reset clobbers PCI config space,
so quit doing the reset.
Signed-off-by: NSunil Goutham <sgoutham@cavium.com>
Signed-off-by: NDavid Daney <david.daney@cavium.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The code currently uses the lightweight dma_wmb barrier before updating
the current descriptor count. Under heavy load, the Tx cleanup routine
was seeing the updated current descriptor count before the updated
descriptor information. As a result, the Tx descriptor was being cleaned
up before it was used because it was not "owned" by the hardware yet,
resulting in a Tx queue hang.

Using the wmb barrier insures that the descriptor is updated before the
descriptor counter preventing the Tx queue hang. For extra insurance,
the Tx cleanup routine is changed to grab the current decriptor count on
entry and uses that initial value in the processing loop rather than
trying to chase the current value.
Signed-off-by: NTom Lendacky <thomas.lendacky@amd.com>
Tested-by: NChristoffer Dall <christoffer.dall@linaro.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The driver was printing a message about not being able
to assign VMDq because of a lack of MSI-X vectors.

This was because a line was missing that initialized a variable,
simply a merge error.
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 code was setting up stats that were not being initialized.
This caused several counters to be displayed incorrectly, due
to indexing beyond the array of strings when printing stats.
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>

Fix sun4i-emac not releasing the following resources:
-iomapped memory not released on probe-failure nor on remove
-clock not getting disabled on probe-failure nor on remove
-sram not being released on remove

And while at it also add error checking to the clk_prepare_enable call
done on probe.
Signed-off-by: NHans de Goede <hdegoede@redhat.com>
Acked-by: NMaxime Ripard <maxime.ripard@free-electrons.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

CONFIG_NET_VENDOR_CAVIUM is only used to hide/show config options and to
include subdirectories in the build, so it doesn't make sense to make it
tristate.
Signed-off-by: NAndreas Schwab <schwab@suse.de>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

To prevent a race between the TX DMA engine and the CPU the writing of the
first transmit descriptor must be deferred until all following descriptors
have been updated. The network card may otherwise start transmitting before
all packet descriptors are set up correctly, which leads to data corruption
or an aborted transmit operation.

This deferral is already done in the non-TSO TX path, implement it also in
the TSO TX path.
Signed-off-by: NPhilipp Kirchhofer <philipp@familie-kirchhofer.de>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The TX DMA engine requires that buffers with a size of 8 bytes or smaller
must be 64 bit aligned. This requirement may be violated when doing TSO,
as in this case larger skb frags can be broken up and transmitted in small
parts with then inappropriate alignment.

Fix this by checking for proper alignment before handing a buffer to the
DMA engine. If the data is misaligned realign it by copying it into the
TSO header data area.
Signed-off-by: NPhilipp Kirchhofer <philipp@familie-kirchhofer.de>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

add the ability to parse "phy-handle". This
is needed for phys, which have a DT node, and
need to parse DT properties.
Signed-off-by: NHeiko Schocher <hs@denx.de>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

If a gigabit ethernet PHY is connected to a fast ethernet MAC,
then it can detect 1000 support from the partner but not use it.

This results in a forced speed of 1000 and RX/TX failure.

Check for 1000BASE-T support and then check the advertisement
configuration before setting the MAC speed to 1000mbit.
Signed-off-by: NSimon Arlott <simon@fire.lp0.eu>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Link interrupts are enabled in init_umac(), which is too early for us to
process them since we do not yet have a valid PHY device pointer. On
BCM7425 chips for instance, we will crash calling phy_mac_interrupt()
because phydev is NULL.

Fix this by moving the link interrupts enabling in
bcmgenet_netif_start(), under a specific function:
bcmgenet_link_intr_enable() and while at it, update the comments
surrounding the code.

Fixes: 6cc8e6d4 ("net: bcmgenet: Delay PHY initialization to bcmgenet_open()")
Signed-off-by: NFlorian Fainelli <f.fainelli@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Because eth_type_trans() consumes ethernet header worth of bytes, a call
to read TCI from end of packet using rhine_rx_vlan_tag() no longer works
as it's reading from an invalid offset.

Tested to be working on PCEngines Alix board.

Fixes: 810f19bc ("via-rhine: add consistent memory barrier in vlan receive code.")
Signed-off-by: NAndrej Ota <andrej@ota.si>
Acked-by: NFrancois Romieu <romieu@fr.zoreil.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Under certain conditions EMAD responses can be returned from the device
even before setting trans_active. This will cause the EMAD Rx listener
to drop the EMAD response - as there are no active transactions - and
timeouts will be generated.

Fix this by setting trans_active before transmitting the EMAD skb.

Fixes: 4ec14b76 ("mlxsw: Add interface to access registers and process events")
Signed-off-by: NIdo Schimmel <idosch@mellanox.com>
Signed-off-by: NJiri Pirko <jiri@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The commit 435452aa ("Prevent VFs from enabling VLAN promiscuous mode")
fixed the PF driver to not include the VLAN promisc capability while
provisioning the interface for a VF. But the fix did not remove this
capability from the profile descriptor of the VF. This causes the VF
driver to request this capability when it tries to create it's interface
at probe time.  This could potentailly cause the VF probe to fail if the
FW enforces strict checking of the flags based on what was provisoned
by the PF.  This strict checking is not being done by FW currently but
will be fixed in a future version. This patch fixes this issue by updating
the VF's profile descriptor so that they match the interface capability
flags provisioned by the PF.

Fixes: 435452aa ("Prevent VFs from enabling VLAN promiscuous mode")
Signed-off-by: NKalesh AP <kalesh.purayil@avagotech.com>
Signed-off-by: NSathya Perla <sathya.perla@avagotech.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The FW requires the pf_num field in the cmd hdr to be set for it to return
the specific function's descriptors in the GET_PROFILE_CONFIG cmd. If not
set, the FW returns the descriptors of all the functions on the device.
If the first descriptor is not what is being queried for, the driver will
read wrong data. This patch fixes this issue by using the GET_CNTL_ATTRIB
cmd to query the real pci_func_num of a function and then uses it in the
GET_PROFILE_CONFIG cmd.
Signed-off-by: NSomnath Kotur <somnath.kotur@emulex.com>
Signed-off-by: NSathya Perla <sathya.perla@avagotech.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

On BE3 chips in SRIOV configs, the TX path stalls when a packet less
than 32B is received from the host. A workaround to pad such packets
already exists for the Skyhawk and Lancer chips. Use the same workaround
for BE3 chips too.
Signed-off-by: NSuresh Reddy <suresh.reddy@avagotech.com>
Signed-off-by: NSathya Perla <sathya.perla@avagotech.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The mcc/mbox lock is not being released when be_cmd_copy() returns
an error.
Signed-off-by: NSuresh Reddy <suresh.reddy@avagotech.com>
Signed-off-by: NSathya Perla <sathya.perla@avagotech.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

In the BE3 FW image, unlike Skyhawk's, the "asic_type_rev" field doesn't
track the asic_rev of chip it is compatible with. When asic_type_rev
is 0 the image is compatible only with pre-BE3-R chips (asic_rev < 0x10).
Fix the current compatibility check to take care of this.
We hit this issue when we try to flash old BE3 images (used prior to the
release of BE3-R) on pre-BE3-R adapters.

Fixes: a6e6ff6e ("be2net: simplify UFI compatibility checking")
Signed-off-by: NKalesh AP <kalesh.purayil@avagotech.com>
Signed-off-by: NSathya Perla <sathya.perla@avagotech.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

commit afae5ad7
  "net/fsl_pq_mdio: streamline probing of MDIO nodes"

added support for different types of MDIO devices:
1) Gianfar MDIO nodes that only map the MII registers
2) Gianfar MDIO nodes that map the full MDIO register set
3) eTSEC2 MDIO nodes (which map the full MDIO register set)
4) QE MDIO nodes (which map only the MII registers)

However, the implementation for types 1 and 4 would mistakenly assume
a mapping of the full MDIO register set, thereby computing the address
for the TBI register starting from the containing structure.
The TBI register would therefore be accessed at a wrong (much bigger)
address, not giving the expected result at all.
This patch restores the correct behavior we had prior to the above one.

The consequences of this bug are apparent when trying to access a PHY
with the same address as the value contained in the initial value of
the TBI register (normally 0); in that case you'll get answers from the
internal TBI device (even though MDIO/MDC pins are actually *also*
toggling on the physical bus!).
Beware that you also need to add a fake tbi node to your device tree
with an unused address.

Notice how this fix is related to commit
22066949
  "powerpc: Add TBI PHY node to first MDIO bus"

which fixed the behavior in kernel 3.3, which was later broken by the
above commit on kernel 3.7.
Signed-off-by: NGerlando Falauto <gerlando.falauto@keymile.com>
Cc: Timur Tabi <timur@tabi.org>
Cc: David S. Miller <davem@davemloft.net>
Cc: Kumar Gala <galak@kernel.crashing.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

When configuring the MDIO subsystem it is also necessary to configure
the TBI register. Make sure the TBI is contained within the mapped
register range in order to:
a) make sure the address is computed correctly
b) make users aware that we're actually accessing that register

In case of error, print a message but continue anyway.
Signed-off-by: NGerlando Falauto <gerlando.falauto@keymile.com>
Cc: Timur Tabi <timur@tabi.org>
Cc: David S. Miller <davem@davemloft.net>
Cc: Kumar Gala <galak@kernel.crashing.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Configuring 4-tuple RSS hsahing for UDP
[E.g., by using `ethtool -N <interface> rx-flow-hash udp4 sdfn']
on a 57710/57711 adapter would cause it to assert as HW does not
support such a configuration.
Signed-off-by: NYuval Mintz <Yuval.Mintz@qlogic.com>
Signed-off-by: NAriel Elior <Ariel.Elior@qlogic.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

When calculating the shift needed in order to access a bit array element
in a byte, we should multiply the index by the element size and not
assume it is fixed at 2-bits.

Fixes: 93c1edb2 ("mlxsw: Introduce Mellanox switch driver core")
Signed-off-by: NIdo Schimmel <idosch@mellanox.com>
Signed-off-by: NJiri Pirko <jiri@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Fixes: 31557f0f ("mlxsw: Introduce Mellanox SwitchX-2 ASIC support")
Signed-off-by: NElad Raz <eladr@mellanox.com>
Signed-off-by: NJiri Pirko <jiri@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

When the device was set to promiscuous mode, we didn't disable
VLAN filtering, which is wrong behaviour, fix that.

Now when the device is set to promiscuous mode RX packets
sent over any VLAN (or no VLAN tag at all) will be accepted.
Signed-off-by: NAchiad Shochat <achiad@mellanox.com>
Signed-off-by: NOr Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

We used the wrong register name for querying the PVLC register

Fixes: a124d13e ('net/mlx5_core: Add more query port helpers')
Signed-off-by: NJiri Pirko <jiri@mellanox.com>
Signed-off-by: NOr Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Test interrupts fails if not all completion vectors called
request_irq. This case happens if only mlx4_en is loaded and
we have more completion vectors than rx rings.

Fixes: c66fa19c ('net/mlx4: Add EQ pool')
Signed-off-by: NCarol L Soto <clsoto@linux.vnet.ibm.com>
Acked-by: NMatan Barak <matanb@mellanox.com>
Signed-off-by: NOr Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

If we get MAX_MSIX interrupts would like to have each receive ring
with his own msix interrupt line. Do not need the shared_ports
variable at mlx4_enable_msix

Fixes: 9293267a ('net/mlx4_core: Capping number of requested MSIXs to MAX_MSIX')
Signed-off-by: NCarol L Soto <clsoto@linux.vnet.ibm.com>
Acked-by: NMatan Barak <matanb@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The recently added mlxsw driver produces warnings in ARM
allmodconfig:

drivers/net/ethernet/mellanox/mlxsw/pci.c: In function 'mlxsw_pci_cmd_exec':
drivers/net/ethernet/mellanox/mlxsw/pci.c:1585:59: warning: right shift count >= width of type [-Wshift-count-overflow]
linux/byteorder/big_endian.h:38:51: note: in definition of macro '__cpu_to_be32'
drivers/net/ethernet/mellanox/mlxsw/pci.c:76:2: note: in expansion of macro 'iowrite32be'

This uses upper_32_bits() to extract the bits while avoiding that warning.
Signed-off-by: NArnd Bergmann <arnd@arndb.de>
Acked-by: NJiri Pirko <jiri@mellanox.com>
Fixes: eda6500a "mlxsw: Add PCI bus implementation"
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The standard way to check if the AQ is enabled is to look at the
count field. So we should only set this field after we have
successfully allocated memory. To do otherwise is to incite
panic among the populace.
Signed-off-by: NMitch Williams <mitch.a.williams@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The kasprintf function can return NULL if the allocation fails. Check for
successful allocation before attempting to use the returned buffer.
Signed-off-by: NTom Lendacky <thomas.lendacky@amd.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The driver still was not offloading TSO on GRE tunnels because
it forgot to set the GSO_GRE flag, causing lots of retransmits.

This fixes generic GRE traffic (like a tunnel added like below)
whereas before it would get 1Gb/s or less, now on a 10G adapter
it gets 8.7Gb/s.

ip ad ad 11.1.0.2/24 dev ens2f0
ip l set ens2f0 up
ip link add gre2 type gretap remote 11.1.0.1 local 11.1.0.2 dev ens2f0
ip l set gre2 up
ip ad ad 192.168.124.2/24 dev gre2
ping 192.168.124.1
netperf -H 192.168.124.1
Signed-off-by: NJesse Brandeburg <jesse.brandeburg@intel.com>
Signed-off-by: NJeff Kirsher <jeffrey.t.kirsher@intel.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

If common clock framework is configured, the driver generates warnings,
which are fixed by this change:

    WARNING: CPU: 0 PID: 1 at linux/drivers/clk/clk.c:727 clk_core_enable+0x2c/0xa4()
    Modules linked in:
    CPU: 0 PID: 1 Comm: swapper Not tainted 4.3.0-rc2+ #141
    Hardware name: LPC32XX SoC (Flattened Device Tree)
    Backtrace:
    [<>] (dump_backtrace) from [<>] (show_stack+0x18/0x1c)
    [<>] (show_stack) from [<>] (dump_stack+0x20/0x28)
    [<>] (dump_stack) from [<>] (warn_slowpath_common+0x90/0xb8)
    [<>] (warn_slowpath_common) from [<>] (warn_slowpath_null+0x24/0x2c)
    [<>] (warn_slowpath_null) from [<>] (clk_core_enable+0x2c/0xa4)
    [<>] (clk_core_enable) from [<>] (clk_enable+0x24/0x38)
    [<>] (clk_enable) from [<>] (lpc_eth_drv_probe+0xfc/0x99c)
    [<>] (lpc_eth_drv_probe) from [<>] (platform_drv_probe+0x50/0xa0)
    [<>] (platform_drv_probe) from [<>] (driver_probe_device+0x18c/0x408)
    [<>] (driver_probe_device) from [<>] (__driver_attach+0x70/0x94)
    [<>] (__driver_attach) from [<>] (bus_for_each_dev+0x74/0x98)
    [<>] (bus_for_each_dev) from [<>] (driver_attach+0x20/0x28)
    [<>] (driver_attach) from [<>] (bus_add_driver+0x11c/0x248)
    [<>] (bus_add_driver) from [<>] (driver_register+0xa4/0xe8)
    [<>] (driver_register) from [<>] (__platform_driver_register+0x50/0x64)
    [<>] (__platform_driver_register) from [<>] (lpc_eth_driver_init+0x18/0x20)
    [<>] (lpc_eth_driver_init) from [<>] (do_one_initcall+0x11c/0x1dc)
    [<>] (do_one_initcall) from [<>] (kernel_init_freeable+0x10c/0x1d4)
    [<>] (kernel_init_freeable) from [<>] (kernel_init+0x10/0xec)
    [<>] (kernel_init) from [<>] (ret_from_fork+0x14/0x24)
Signed-off-by: NVladimir Zapolskiy <vz@mleia.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Added kfree() to avoid the memory leak when debugfs_create_dir() fails.
Signed-off-by: NGeliang Tang <geliangtang@163.com>
Acked-by: NTom Lendacky <thomas.lendacky@amd.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Both new_steering_entry() and existing_steering_entry() return values
based on their success or failure, but currently they fall through
silently.  This can make troubleshooting difficult, as we were unable
to tell which one of these two functions returned errors or
specifically what code was returned.  This patch remedies that
situation by passing the return codes to err, which is returned by
mlx4_qp_attach_common() itself.

This also addresses a leak in the call to mlx4_bitmap_free() as well.
Signed-off-by: NRobb Manes <rmanes@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>