TX ring QP wasn't released at mlx4_en_destroy_tx_ring. Instead, the code
used the deprecated base_tx_qpn field. Move TX QP release to
mlx4_en_destroy_tx_ring and remove the base_tx_qpn field.

Fixes: ddae0349 ('net/mlx4: Change QP allocation scheme')
Signed-off-by: NEran Ben Elisha <eranbe@mellanox.com>
Signed-off-by: NOr Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Under SRIOV, the port rx/tx bytes/packets statistics should by read
from the HW instead of using the PF netdevice SW accounting. This is
needed in order to get the full port statistics and not just the PF
own ones
Signed-off-by: NEran Ben Elisha <eranbe@mellanox.com>
Signed-off-by: NOr Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

NUM_ALL_STATS was not updated with the new four entries, instead
NUM_FLOW_STATS was updated, fix it. that caused off-by-four for all
counters below pf_*_*.

Fixes: b42de4d0 ('net/mlx4_en: Show PF own statistics via ethtool')
Signed-off-by: NEran Ben Elisha <eranbe@mellanox.com>
Signed-off-by: NOr Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Implement the ndo to gather VF statistics through the PF.

All counters related to this VF are stored in a per slave
list, run over the slave's list and collect all statistics.
Signed-off-by: NEran Ben Elisha <eranbe@mellanox.com>
Signed-off-by: NHadar Hen Zion <hadarh@mellanox.com>
Signed-off-by: NOr Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Allow the user to observe the PF own statistics using ethtool with pf_
prefixed counter names.

Those counters are the PF statistics out of the overall port statistics.
Every PF QP is attached to a counter and the summary of those counters
is the PF statistics.
Signed-off-by: NEran Ben Elisha <eranbe@mellanox.com>
Signed-off-by: NHadar Hen Zion <hadarh@mellanox.com>
Signed-off-by: NOr Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This is an infrastructure step for querying VF and PF counters.

This code was in the IB driver, move it to the mlx4 core driver
so it will be accessible for more use cases.
Signed-off-by: NEran Ben Elisha <eranbe@mellanox.com>
Signed-off-by: NHadar Hen Zion <hadarh@mellanox.com>
Signed-off-by: NOr Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Default counter per port will be allocated at the mlx4 core driver load.

Every QP opened by the Ethernet driver will be attached to the port's default
counter.  This is an infrastructure step to collect VF statistics from the PF.
Signed-off-by: NEran Ben Elisha <eranbe@mellanox.com>
Signed-off-by: NHadar Hen Zion <hadarh@mellanox.com>
Signed-off-by: NOr Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Counter will get its port attribute within the resource tracker when
the first QP attached to it is modified to RTR. If a QP is counter-less,
an attempt to create a new counter with assigned port will be made.
Signed-off-by: NEran Ben Elisha <eranbe@mellanox.com>
Signed-off-by: NHadar Hen Zion <hadarh@mellanox.com>
Signed-off-by: NOr Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Each physical function has a guarantee of two counters per port, one
for a default counter and one for the IB driver.

Each virtual function has a guarantee of one counter per port.
All other counters are free and can be obtained on demand.

This is a preparation step for supporting a get_vf_stats ndo call,
so we can promise a counter for every VF in order to collect their
statistics from the PF context.
Signed-off-by: NEran Ben Elisha <eranbe@mellanox.com>
Signed-off-by: NHadar Hen Zion <hadarh@mellanox.com>
Signed-off-by: NOr Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Since virtual functions get their counters indices allocation from the PF,
allocate counters indices bitmap only in case the function isn't virtual.

Also, check that the device has counters to allocate before creating the
indices bitmap table.
Signed-off-by: NEran Ben Elisha <eranbe@mellanox.com>
Signed-off-by: NHadar Hen Zion <hadarh@mellanox.com>
Signed-off-by: NOr Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Reserve the last valid counter index for "sink" counter, when a
new counter cannot be allocated, the driver will use this counter.

In order to avoid allocating this counter on any other flow, fix the
indices bitmap allocation range, and reserve the sink counter index.

Add macro for the sink counter index and replace all appearences of the
index with the macro.
Signed-off-by: NEran Ben Elisha <eranbe@mellanox.com>
Signed-off-by: NHadar Hen Zion <hadarh@mellanox.com>
Signed-off-by: NOr Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Add resetting the counter data to the free counter flow, so the counter's
data won't be accessible anymore if querying the counter. Also, on next
counter allocation (to another VM for example), it will be fresh and clear.
Signed-off-by: NEran Ben Elisha <eranbe@mellanox.com>
Signed-off-by: NHadar Hen Zion <hadarh@mellanox.com>
Signed-off-by: NOr Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

If counters are not supported by the device. The indices bitmap table is not
allocated during initialization. Add the symmetrical check before cleaning
the counters bitmap table or freeing a counter.
Signed-off-by: NEran Ben Elisha <eranbe@mellanox.com>
Signed-off-by: NHadar Hen Zion <hadarh@mellanox.com>
Signed-off-by: NOr Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Due to firmware bug, under VPI configuration when port1 = IB and
port2 = Eth, Granular QoS per VF isn't working properly. More over,
the whole QP0/QP1 Para-Virtualization in the mlx4 IB driver is
broken on that config.

Hence, we must disable Granular QoS per VF under that configuration
till a fix is introduced. Once that happens, a new device capability
will be used to mark the feature support on that specific configuration.
Reported-by: NDoug Ledford <dledford@redhat.com>
Signed-off-by: NOr Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

In order to read the HCA's cycle counter efficiently in
user space, we need to map the HCA's register.
This is done through mmap call.
Signed-off-by: NMatan Barak <matanb@mellanox.com>
Signed-off-by: NOr Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: NDoug Ledford <dledford@redhat.com>

Use kernel.h macro definition.

Thanks to Julia Lawall for Coccinelle scripting support.
Signed-off-by: NFabian Frederick <fabf@skynet.be>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Use kernel.h macro definition.

Thanks to Julia Lawall for Coccinelle scripting support.
Signed-off-by: NFabian Frederick <fabf@skynet.be>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

fix typo in mlx4_set_vf_mac
Acked-by: NOr Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: NCarol L Soto <clsoto@linux.vnet.ibm.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

If mlx4_enable_sriov is called by adapter without this
feature MLX4_DEV_CAP_FLAG2_SYS_EQS then during this path the function alloc
icm is called twice without freeing the structures from the first time.
Acked-by: NOr Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: NCarol L Soto <clsoto@linux.vnet.ibm.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

If user loads mlx4_core with num_vfs greater than
supported then variable dev->dev_vfs is freed 2 times after unloading the
driver.
Acked-by: NOr Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: NCarol L Soto <clsoto@linux.vnet.ibm.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

When CONFIG_RFS_ACCEL or SMP aren't set, we fail to build, fix it.

Also, avoid build warning as of unused function on that setup.

Fixes: c66fa19c ('net/mlx4: Add EQ pool')
Reported-by: NMichael Ellerman <mpe@ellerman.id.au>
Signed-off-by: NMatan Barak <matanb@mellanox.com>
Signed-off-by: NOr Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

When freeing a CQ, we need to make sure there are no
asynchronous events (on the ASYNC EQ) that could
relate to this CQ before freeing it.

This is done by introducing synchronize_irq.
Signed-off-by: NMatan Barak <matanb@mellanox.com>
Signed-off-by: NIdo Shamay <idos@mellanox.com>
Signed-off-by: NOr Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Now that EQs management is in the sole responsibility of mlx4_core,
the IRQ affinity hints configuration should be in its hands as well.
request_irq is called only once by the first consumer (maybe mlx4_ib),
so mlx4_en passes the affinity mask too late. We also need to request
vectors according to the cores we want to run on.

mlx4_core distribution of IRQs to cores is straight forward,
EQ(i)->IRQ will set affinity hint to core i.
Consumers need to request EQ vectors, according to their cores
considerations (NUMA).
Signed-off-by: NIdo Shamay <idos@mellanox.com>
Signed-off-by: NMatan Barak <matanb@mellanox.com>
Signed-off-by: NOr Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Previously, mlx4_en allocated EQs and used them exclusively.
This affected RoCE performance, as applications which are
events sensitive were limited to use only the legacy EQs.

Change that by introducing an EQ pool. This pool is managed
by mlx4_core. EQs are assigned to ports (when there are limited
number of EQs, multiple ports could be assigned to the same EQs).

An exception to this rule is the ASYNC EQ which handles various events.

Legacy EQs are completely removed as all EQs could be shared.

When a consumer (mlx4_ib/mlx4_en) requests an EQ, it asks for
EQ serving on a specific port. The core driver calculates which
EQ should be assigned to that request.

Because IRQs are shared between IB and Ethernet modules, their
names only include the PCI device BDF address.
Signed-off-by: NMatan Barak <matanb@mellanox.com>
Signed-off-by: NIdo Shamay <idos@mellanox.com>
Signed-off-by: NOr Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

In SRIOV, when simple (i.e - Ethernet L2 only) flow steering rules are
created, always create them at MLX4_DOMAIN_NIC priority (instead of
the real priority the function created them at). This is done in order
to let multiple functions add broadcast/multicast rules without
affecting other functions, which is necessary for DPDK in SRIOV.
Signed-off-by: NMatan Barak <matanb@mellanox.com>
Signed-off-by: NOr Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

da91309e (cpumask: Utility function to set n'th cpu...) created a
genuinely weird function.  I never saw it before, it went through DaveM.
(He only does this to make us other maintainers feel better about our own
mistakes.)

cpumask_set_cpu_local_first's purpose is say "I need to spread things
across N online cpus, choose the ones on this numa node first"; you call
it in a loop.

It can fail.  One of the two callers ignores this, the other aborts and
fails the device open.

It can fail in two ways: allocating the off-stack cpumask, or through a
convoluted codepath which AFAICT can only occur if cpu_online_mask
changes.  Which shouldn't happen, because if cpu_online_mask can change
while you call this, it could return a now-offline cpu anyway.

It contains a nonsensical test "!cpumask_of_node(numa_node)".  This was
drawn to my attention by Geert, who said this causes a warning on Sparc.
It sets a single bit in a cpumask instead of returning a cpu number,
because that's what the callers want.

It could be made more efficient by passing the previous cpu rather than
an index, but that would be more invasive to the callers.

Fixes: da91309e
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au> (then rebased)
Tested-by: NAmir Vadai <amirv@mellanox.com>
Acked-by: NAmir Vadai <amirv@mellanox.com>
Acked-by: NDavid S. Miller <davem@davemloft.net>

The test in mlx4_load_one() to remove MLX4_FLAG_MSI_X expects mlx4_NOP() to
fail with -EBUSY. It is also necessary to avoid the reset since the device
is not fully reinitialized before calling mlx4_start_hca() a second time.

Note that this will also affect mlx4_test_interrupts(), the only other user
of MLX4_CMD_NOP.

Fixes: f5aef5aa ("net/mlx4_core: Activate reset flow upon fatal command cases")
Signed-off-by: NBenjamin Poirier <bpoirier@suse.de>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Remove the limitation that disallows configuring single ported VFs
in the presence of IB ports, after addressing the issues that
prevented that to work.

SMI (QP0) requests/responses are still not supported for single
ported IB VFs.
Signed-off-by: NOr Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: NJack Morgenstein <jackm@dev.mellanox.co.il>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

It's legal for drivers to provide the QP port through the
QPC schedule-queue field on the reset-to-init QP state change.

Add adjusting of the schedule queue port in the SRIOV wrapper
for that operation too.
Signed-off-by: NOr Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: NJack Morgenstein <jackm@dev.mellanox.co.il>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Some VF drivers flow set the schedule queue in the QP context but
without setting none of OPTPAR_SCHED_QUEUE or OPTPAR_PRIMARY_ADDR_PATH.

To allow for such non-modified drivers to function as single ported
IB VFs, we must adjust the schedule queue port whenever being set,
e.g as currently done for single ported Eth VFs.
Signed-off-by: NOr Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: NJack Morgenstein <jackm@dev.mellanox.co.il>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

As part of enabling single ported VFs over IB ports we need to handle
some of the flows for generting EQ events for VFs which don't come
into play under Eth ports.

This mainly includes port management events derived from changes of the
phyiscal port (lid change, client re-register, down/up, etc), VF pkey table
changes and VF guid changes initiated by the IB driver.

(1) make sure that events are generated only for VFs sitting on
    the relevant physical port (under the ALL_SLAVES flow).

(2) before generating the event, convert from physical (one or two)
    to VF port (always equals one).
Signed-off-by: NOr Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: NJack Morgenstein <jackm@dev.mellanox.co.il>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Single port VFs always provide port = 1 (even if the actual physical
port used is port 2). As such, we need to convert the port provided
by the VF to the physical port before calling into the firmware.

It turns out that the Linux mlx4 VF RoCE driver maintains a copy of
the GID table and hence this change became critical only for single
ported IB VFs, but it could be needed for other RoCE VF drivers too.

Fixes: 449fc488 ('net/mlx4: Adapt code for N-Port VF')
Signed-off-by: NOr Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: NJack Morgenstein <jackm@dev.mellanox.co.il>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

With a cross-compiler based on gcc-4.9, I see warnings like the following:

  drivers/net/ethernet/mellanox/mlx4/resource_tracker.c: In function 'mlx4_SW2HW_CQ_wrapper':
  drivers/net/ethernet/mellanox/mlx4/resource_tracker.c:3048:10: error: 'cq' may be used uninitialized in this function [-Werror=maybe-uninitialized]
    cq->mtt = mtt;

I think the warning is spurious because we only use cq when
cq_res_start_move_to() returns zero, and it always initializes *cq in that
case.  The srq case is similar.  But maybe gcc isn't smart enough to figure
that out.

Initialize cq and srq explicitly to avoid the warnings.
Signed-off-by: NBjorn Helgaas <bhelgaas@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The Firmware uses dynamic EQs allocation based on number of VFs and
max EQs that can be allocated. As a result, VF can have EQ numbers
that are larger than 256.

According to the firmware spec, the max value is limited to be 1024
(10 bits), adapt the relevant code accordingly. This bug was impossible
to hit prior to commit 7ae0e400 ("net/mlx4_core: Flexible (asymmetric)
allocation of EQs and MSI-X vectors for PF/VFs") which actually enables
large number of EQs for VFs.
Signed-off-by: NYishai Hadas <yishaih@mellanox.com>
Signed-off-by: NOr Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This caused the en_stats_adder helper to accumulate a field which is
not related to the counter, fix that.

Fixes: a3333b35 ('net/mlx4_en: Moderate ethtool callback to show [..]')
Signed-off-by: NEran Ben Elisha <eranbe@mellanox.com>
Signed-off-by: NOr Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

When system is out of memory, refilling of RX buffers fails while
the driver continue to pass the received packets to the kernel stack.
At some point, when all RX buffers deplete, driver may fall into a
sleep, and not recover when memory for new RX buffers is once again
availible. This is because hardware does not have valid descriptors,
so no interrupt will be generated for the driver to return to work
in napi context. Fix it by schedule the napi poll function from
stats_task delayed workqueue, as long as the allocations fail.
Signed-off-by: NIdo Shamay <idos@mellanox.com>
Signed-off-by: NAmir Vadai <amirv@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Addresses the following kernel logs seen during boot:

Kernel unaligned access at TPC[100ee150] mlx4_QUERY_HCA+0x80/0x248 [mlx4_core]
Kernel unaligned access at TPC[100f071c] mlx4_QUERY_ADAPTER+0x100/0x12c [mlx4_core]
Kernel unaligned access at TPC[100f071c] mlx4_QUERY_ADAPTER+0x100/0x12c [mlx4_core]
Kernel unaligned access at TPC[100f071c] mlx4_QUERY_ADAPTER+0x100/0x12c [mlx4_core]
Kernel unaligned access at TPC[100f071c] mlx4_QUERY_ADAPTER+0x100/0x12c [mlx4_core]
Signed-off-by: NDavid Ahern <david.ahern@oracle.com>
Acked-by: NOr Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Signed-off-by: NBenjamin Poirier <bpoirier@suse.de>
Fixes: 9e311e77 ("net/mlx4_en: Use affinity hint")
Acked-by: NAmir Vadai <amirv@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

By default, the number of tx queues is limited by the number of online cpus
in mlx4_en_get_profile(). However, this limit no longer holds after the
ethtool .set_channels method has been called. In that situation, the driver
may access invalid bits of certain cpumask variables when queue_index >=
nr_cpu_ids.
Signed-off-by: NBenjamin Poirier <bpoirier@suse.de>
Acked-by: NIdo Shamay <idos@mellanox.com>
Fixes: d03a68f8 ("net/mlx4_en: Configure the XPS queue mapping on driver load")
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

mlx4_en_check_rxfh_func() was checking for hardware support before
setting a known RSS hash function, but didn't do any check before
setting unknown RSS hash function. Need to make it fail on such values.
In this occasion, moved the actual setting of the new value from the
check function into mlx4_en_set_rxfh().

Fixes: 947cbb0a ("net/mlx4_en: Support for configurable RSS hash function")
Signed-off-by: NAmir Vadai <amirv@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>