As diagnosed by Song Liu, ndo_poll_controller() can
be very dangerous on loaded hosts, since the cpu
calling ndo_poll_controller() might steal all NAPI
contexts (for all RX/TX queues of the NIC). This capture
can last for unlimited amount of time, since one
cpu is generally not able to drain all the queues under load.

bnxt uses NAPI for TX completions, so we better let core
networking stack call the napi->poll() to avoid the capture.
Signed-off-by: NEric Dumazet <edumazet@google.com>
Cc: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The recent commit to always forward the VF MAC address to the PF for
approval may not work if the PF driver or the firmware is older.  This
will cause the VF driver to fail during probe:

  bnxt_en 0000:00:03.0 (unnamed net_device) (uninitialized): hwrm req_type 0xf seq id 0x5 error 0xffff
  bnxt_en 0000:00:03.0 (unnamed net_device) (uninitialized): VF MAC address 00:00:17:02:05:d0 not approved by the PF
  bnxt_en 0000:00:03.0: Unable to initialize mac address.
  bnxt_en: probe of 0000:00:03.0 failed with error -99

We fix it by treating the error as fatal only if the VF MAC address is
locally generated by the VF.

Fixes: 707e7e96 ("bnxt_en: Always forward VF MAC address to the PF.")
Reported-by: NSeth Forshee <seth.forshee@canonical.com>
Reported-by: NSiwei Liu <loseweigh@gmail.com>
Signed-off-by: NMichael Chan <michael.chan@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Currently, the driver adjusts the bp->hw_resc.max_cp_rings by the number
of MSIX vectors used by RDMA.  There is one code path in open that needs
to check the true max_cp_rings including any used by RDMA.  This code
is now checking for the reduced max_cp_rings which will fail when the
number of cp rings is very small.

To fix this in a clean way, we don't adjust max_cp_rings anymore.
Instead, we add a helper bnxt_get_max_func_cp_rings_for_en() to get the
reduced max_cp_rings when appropriate.

Fixes: ec86f14e ("bnxt_en: Add ULP calls to stop and restart IRQs.")
Signed-off-by: NMichael Chan <michael.chan@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Remove unused bnxt_subtract_ulp_resources().  Change
bnxt_get_max_func_irqs() to static since it is only locally used.
Signed-off-by: NMichael Chan <michael.chan@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

When the driver detects that resources have changed during open, it
should reset the rx and tx rings to 0.  This will properly setup the
init sequence to initialize the default rings again.  We also need
to signal the RDMA driver to stop and clear its interrupts.  We then
call the RoCE driver to restart if a new set of default rings is
successfully reserved.

Fixes: 25e1acd6 ("bnxt_en: Notify firmware about IF state changes.")
Signed-off-by: NMichael Chan <michael.chan@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.

Addresses-Coverity-ID: 114878 ("Missing break in switch")
Signed-off-by: NGustavo A. R. Silva <gustavo@embeddedor.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The CNP CoS queue is reserved for internal RDMA Congestion Notification
Packets (CNP) and should not be used for a TC.  Modify the CoS queue
discovery code to skip over the CNP CoS queue and to reduce
bp->max_tc accordingly.  However, if RDMA is disabled in NVRAM, the
the CNP CoS queue can be used for a TC.
Signed-off-by: NMichael Chan <michael.chan@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Export temperature sensor reading via hwmon sysfs.
Signed-off-by: NVasundhara Volam <vasundhara-v.volam@broadcom.com>
Signed-off-by: NMichael Chan <michael.chan@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Use latest firmware API to notify firmware about IF state changes.
Firmware has the option to clean up resources during IF down and
to require the driver to reserve resources again during IF up.
Signed-off-by: NMichael Chan <michael.chan@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The flags field is almost getting full.  Move firmware capability flags
to a new fw_cap field to better organize these firmware flags.
Signed-off-by: NMichael Chan <michael.chan@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The BNXT_FLAG_NEW_RM flag is checked a lot in the code to determine if
the new resource manager is in effect.  Define a macro to perform
this check.
Signed-off-by: NMichael Chan <michael.chan@broadocm.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Set the default hash mode flag in HWRM_VNIC_RSS_CFG to signal to the
firmware that the driver is compliant with the latest spec.  With
that, the firmware can return expanded RSS profile IDs that the driver
checks to setup the proper gso_type for GRO-HW packets.  But instead
of checking for the new profile IDs, we check the IP_TYPE flag
in TPA_START which is more straight forward than checking a list of
profile IDs.
Signed-off-by: NMichael Chan <michael.chan@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The new mode is "minimal-static" to be used when resources are more
limited to support a large number of VFs, for example  The PF driver
will provision guaranteed minimum resources of 0.  Each VF has no
guranteed resources until it tries to reserve resources during device
open.
Signed-off-by: NMichael Chan <michael.chan@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

During hotplug, the driver's open function can be called almost
immediately after power on reset.  The PHY may not be ready and the
firmware may return failure when the driver tries to update PHY
settings.  Add retry logic fired from the driver's timer to retry
the operation for 5 seconds.
Signed-off-by: NMichael Chan <michael.chan@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Add code to detect firmware support for external loopback and the extra
test entry for external loopback.
Signed-off-by: NMichael Chan <michael.chan@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

hwrm_dbg_resp_addr and hwrm_dbg_resp_dma_addr are never used
and can be removed.
Signed-off-by: NYueHaibing <yuehaibing@huawei.com>
Acked-by: NMichael Chan <michael.chan@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Fix bug in the error code path when bnxt_request_irq() returns failure.
bnxt_disable_napi() should not be called in this error path because
NAPI has not been enabled yet.

Fixes: c0c050c5 ("bnxt_en: New Broadcom ethernet driver.")
Signed-off-by: NVikas Gupta <vikas.gupta@broadcom.com>
Signed-off-by: NMichael Chan <michael.chan@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Calling bnxt_set_max_func_irqs() to modify the max IRQ count requested or
freed by the RDMA driver is flawed.  The max IRQ count is checked when
re-initializing the IRQ vectors and this can happen multiple times
during ifup or ethtool -L.  If the max IRQ is reduced and the RDMA
driver is operational, we may not initailize IRQs correctly.  This
problem shows up on VFs with very small number of MSIX.

There is no other logic that relies on the IRQ count excluding the ones
used by RDMA.  So we fix it by just removing the call to subtract or
add the IRQs used by RDMA.

Fixes: a588e458 ("bnxt_en: Add interface to support RDMA driver.")
Signed-off-by: NMichael Chan <michael.chan@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Currently, the driver assumes IFF_BROADCAST is always set and always sets
the broadcast filter.  Modify the code to set or clear the broadcast
filter according to the IFF_BROADCAST flag.
Signed-off-by: NMichael Chan <michael.chan@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The current code returns -ENOMEM and does not bother to set the output
parameters to 0 when no rings are available.  Some callers, such as
bnxt_get_channels() will display garbage ring numbers when that happens.
Fix it by always setting the output parameters.

Fixes: 6e6c5a57 ("bnxt_en: Modify bnxt_get_max_rings() to support shared or non shared rings.")
Signed-off-by: NMichael Chan <michael.chan@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

If there aren't enough RX rings available, the driver will attempt to
use a single RX ring without the aggregation ring.  If that also
fails, the BNXT_FLAG_AGG_RINGS flag is cleared but the other ring
parameters are not set consistently to reflect that.  If more RX
rings become available at the next open, the RX rings will be in
an inconsistent state and may crash when freeing the RX rings.

Fix it by restoring the BNXT_FLAG_AGG_RINGS if not enough RX rings are
available to run without aggregation rings.

Fixes: bdbd1eb5 ("bnxt_en: Handle no aggregation ring gracefully.")
Signed-off-by: NMichael Chan <michael.chan@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

In preparation to enabling -Wimplicit-fallthrough, mark switch cases
where we are expecting to fall through.
Signed-off-by: NGustavo A. R. Silva <gustavo@embeddedor.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>

Previously the driver used pcie_get_minimum_link() to warn when the NIC
is in a slot that can't supply as much bandwidth as the NIC could use.

pcie_get_minimum_link() can be misleading because it finds the slowest link
and the narrowest link (which may be different links) without considering
the total bandwidth of each link.  For a path with a 16 GT/s x1 link and a
2.5 GT/s x16 link, it returns 2.5 GT/s x1, which corresponds to 250 MB/s of
bandwidth, not the true available bandwidth of about 1969 MB/s for a
16 GT/s x1 link.

Use pcie_print_link_status() to report PCIe link speed and possible
limitations instead of implementing this in the driver itself.  This finds
the slowest link in the path to the device by computing the total bandwidth
of each link and compares that with the capabilities of the device.

The dmesg change is:

  - PCIe: Speed %s Width x%d
  + %u.%03u Gb/s available PCIe bandwidth (%s x%d link)
Signed-off-by: NBjorn Helgaas <bhelgaas@google.com>

The current code already forwards the VF MAC address to the PF, except
in one case.  If the VF driver gets a valid MAC address from the firmware
during probe time, it will not forward the MAC address to the PF,
incorrectly assuming that the PF already knows the MAC address.  This
causes "ip link show" to show zero VF MAC addresses for this case.

This assumption is not correct.  Newer firmware remembers the VF MAC
address last used by the VF and provides it to the VF driver during
probe.  So we need to always forward the VF MAC address to the PF.

The forwarded MAC address may now be the PF assigned MAC address and so we
need to make sure we approve it for this case.
Signed-off-by: NMichael Chan <michael.chan@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Only non-NPAR PFs need to actively check and manage unsupported link
speeds.  NPAR functions and VFs do not control the link speed and
should skip the unsupported speed detection logic, to avoid warning
messages from firmware rejecting the unsupported firmware calls.
Signed-off-by: NMichael Chan <michael.chan@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

A recent change to reduce delay granularity waiting for firmware
reponse has caused a regression.  With a tighter delay loop,
the driver may see the beginning part of the response faster.
The original 5 usec delay to wait for the rest of the message
is not long enough and some messages are detected as invalid.

Increase the maximum wait time from 5 usec to 20 usec.  Also, fix
the debug message that shows the total delay time for the response
when the message times out.  With the new logic, the delay time
is not fixed per iteration of the loop, so we define a macro to
show the total delay time.

Fixes: 9751e8e7 ("bnxt_en: reduce timeout on initial HWRM calls")
Signed-off-by: NMichael Chan <michael.chan@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Add logic to reserve default rings at driver open time if none was
reserved during probe time.  This will happen when the PF driver did
not provision minimum rings to the VF, due to more limited resources.

Driver open will only succeed if some minimum rings can be reserved.
Signed-off-by: NMichael Chan <michael.chan@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

For completeness and correctness, the VF driver needs to reserve these
RSS and L2 contexts.
Signed-off-by: NMichael Chan <michael.chan@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

When rings are more limited and the PF has not provisioned minimum
guaranteed rings to the VF, do not reserve rings during driver probe.
Wait till device open before reserving rings when they will be used.
Device open will succeed if some minimum rings can be successfully
reserved and allocated.
Signed-off-by: NMichael Chan <michael.chan@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This adds debugfs support for bnxt_en with the purpose of allowing users
to examine the current DIM profile in use for each receive queue.  This
was instrumental in debugging issues found with DIM and ensuring that
the profiles we expect to use are the profiles being used.
Signed-off-by: NAndy Gospodarek <gospo@broadcom.com>
Signed-off-by: NMichael Chan <michael.chan@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Testing with DIM enabled on older kernels indicated that firmware calls
were slower than expected.  More detailed analysis indicated that the
default 25us delay was higher than necessary.  Reducing the time spend in
usleep_range() for the first several calls would reduce the overall
latency of firmware calls on newer Intel processors.
Signed-off-by: NAndy Gospodarek <gospo@broadcom.com>
Signed-off-by: NMichael Chan <michael.chan@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This keeps the RING_IDLE flag set in hardware for higher coalesce
settings by default and improved latency.
Signed-off-by: NAndy Gospodarek <gospo@broadcom.com>
Signed-off-by: NMichael Chan <michael.chan@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Replace switch statements printing different messages for every ring type
with a common message.
Signed-off-by: NMichael Chan <michael.chan@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Older firmware will reject this call and cause an error message to
be printed by the VF driver.
Signed-off-by: NMichael Chan <michael.chan@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The current driver maps MQPRIO traffic classes directly 1:1 to the
internal hardware queues (TC0 maps to hardware queue 0, etc).  This
direct mapping requires the internal hardware queues to be reconfigured
from lossless to lossy and vice versa when necessary.  This
involves reconfiguring internal buffer thresholds which is
disruptive and not always reliable.

Implement a new scheme to map TCs to internal hardware queues by
matching up their PFC requirements.  This will eliminate the need
to reconfigure a hardware queue internal buffers at run time.  After
remapping, the NIC is closed and opened for the new TC to hardware
queues to take effect.

This patch only adds the basic mapping logic.
Signed-off-by: NMichael Chan <michael.chan@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

When open fails during ethtool -L ring change, for example, the driver
may crash at bnxt_free_irq() because bp->bnapi is NULL.

If we fail to allocate all the new rings, bnxt_open_nic() will free
all the memory including bp->bnapi.  Subsequent call to bnxt_close_nic()
will try to dereference bp->bnapi in bnxt_free_irq().

Fix it by checking for !bp->bnapi in bnxt_free_irq().

Fixes: e5811b8c ("bnxt_en: Add IRQ remapping logic.")
Signed-off-by: NMichael Chan <michael.chan@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

With recent changes to reserve both L2 and RDMA rings, we need to include
the RDMA rings in bnxt_check_rings().  Otherwise we will under-estimate
the rings we need during ethtool -L and may lead to failure.

Fixes: fbcfc8e4 ("bnxt_en: Reserve completion rings and MSIX for bnxt_re RDMA driver.")
Signed-off-by: NMichael Chan <michael.chan@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

When the driver needs to re-initailize the IRQ vectors, we make the
new ulp_irq_stop() call to tell the RDMA driver to disable and free
the IRQ vectors.  After IRQ vectors have been re-initailized, we
make the ulp_irq_restart() call to tell the RDMA driver that
IRQs can be restarted.
Signed-off-by: NMichael Chan <michael.chan@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Add additional logic to reserve completion rings for the bnxt_re driver
when it requests MSIX vectors.  The function bnxt_cp_rings_in_use()
will return the total number of completion rings used by both drivers
that need to be reserved.  If the network interface in up, we will
close and open the NIC to reserve the new set of completion rings and
re-initialize the vectors.
Signed-off-by: NMichael Chan <michael.chan@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>