When the device tunneling offloads mode is vxlan do the following
- call SET_PORT with the relevant setting
- add DMFS steering vxlan rule for the device self and multicast mac addresses
of the form: {<ETH, outer-mac> <VXLAN, ANY vnid> <ETH, ANY mac>} --> RSS QP
- set relevant QPC fields in RSS context and RX ring QPs
- in TX flow, set WQE fields to generate HW checksum, and handle gso skbs
which are marked for encapsulation such that the HW will segment them properly.
- in RX flow, read HW offloaded checksum for encapsulated packets from the CQE
- advertize hw_enc_features and NETIF_F_GSO_UDP_TUNNEL to the networking stack
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add the low-level device commands and definitions used for TCP/IP HW offloads
of tunneled/vxlan traffic which are supported by the ConnectX3-pro NIC.
This is done through the following elements:
- read tunneling device caps in QUERY_DEV_CAP
- add helper function to do SET_PORT for tunneling
- add DMFS VXLAN steering rule definitions
- add CQE and WQE checksum offload field definitions
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch adds support so that the rps_flow_tables (RFS) can be
programmed using the tun flows which are already set up to track flows
for the purposes of queue selection.
On the receive path (corresponding to select_queue and tun_net_xmit) the
rxhash is saved in the flow_entry. The original code only does flow
lookup in select_queue, so this patch adds a flow lookup in tun_net_xmit
if num_queues == 1 (select_queue is not called from
dev_queue_xmit->netdev_pick_tx in that case).
The flow is recorded (processing CPU) in tun_flow_update (TX path), and
reset when flow is deleted.
Signed-off-by: Tom Herbert <therbert@google.com>
Signed-off-by: Zhi Yong Wu <wuzhy@linux.vnet.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Found by klocwork analysis.
mvm could be NULL which may cause a NULL dereference
in a theoretical call flow
rs_fill_lq_cmd(mvm = NULL, ...)
rs_build_rates_table
rs_fill_rates_for_column
ucode_rate_from_rs_rate
IWL_ERR(mvm,...)
No real reason for passing NULL to rs_fill_lq_cmd so fix that.
Reported-by: Eytan Lifshitz <eytan.lifshitz@intel.com>
Signed-off-by: Eyal Shapira <eyal@wizery.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Currently, the MAC context tsf_id assignment and the master/slave
relations are determined only when a new vif is added, as part
of the MAC context resource allocation. However, at this stage, the
beacon interval is not known, and thus could not be taken into account
in the master-slave algorithm.
To fix this, recalculate the MAC context tsf_id assignment,
just before the MAC context is activated, i.e., just before
a station VMAC is configured to associated and before an AP
VMAC is started.
Signed-off-by: Ilan Peer <ilan.peer@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Change the parameters for calculating an AP TBTT to 64/36 instead of
80/20, to increase the interval between a station vif and an AP
vif TBTT events.
Signed-off-by: Ilan Peer <ilan.peer@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
The access to the CSR_RESET reg should be done as a complete
DWORD and not by setting a bit. This is the right way to reset
the device.
Signed-off-by: Eran Harary <eran.harary@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Instead of always calling ieee80211_beacon_loss() on every missed
beacons notification, call this function only if the number of
consecutive missed beacons from last rx is higher than a predefined
threshold.
Signed-off-by: Ilan Peer <ilan.peer@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
If the channel min-width changes, we can update the PHY ctx, even if
it has multiple references.
Signed-off-by: Arik Nemtsov <arik@wizery.com>
Reviewed-by: Eliad Peller <eliad@wizery.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
This allows to format it at will using external tools.
Since different teams want it in different formats, dump
the raw data and everyone can play with the data the way
they want.
While at it - make this code slightly more robust by making
the required verification on the offsets / length in the
write handler.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Don't check if mvm->fw->cs is NULL since it can't be.
cs is an array member of iwl_fw, it can't be NULL.
Use memset(ptr, 0, sizeof(*ptr)); instead of
memset(ptr, 0, sizeof(struct ptr_type));
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
The firmware needs to be stopped quickly (100ms) after the
RFKILL interrupt fired. Failing to do so would allow the
firmware to access the radio registers which would lead to
a hardware error.
Before this change, we would kill the firmware only when
mac80211 stops the device which can take a fair amount of
time. Take a shortcut by stopping the device right away
in the interrupt.
This is not relevant if the current firmware is INIT
firmware since that firmware can run while in RFKILL.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Calling stop_device when start_fw wasn't called would issue:
Stopping tx queues that aren't allocated...
Also allow the op_mode to call stop_device and then to
disable the Tx queues - in that case just silently ignore
the disabling on the Tx queues, since the PRPH registers
aren't reachable any more.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
This is useless and introduces a dependency between rfkill
and stop_device - the op_mode can't call stop_device from
the rfkill notification since it would lead to an endless
recursion.
Next patches will need to do so.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Under very specific circumstances, the firmware might
ignore a host command. This was debugged and we ended up
seeing that the power management hardware was faulty.
In order to workaround this issue, we keep the NIC awake
as long as we have host commands in flight. This will avoid
to put the hardware into buggy condition.
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
In case of invalid section_id, the function returns after
it aleready allocated memory. Fixed by change the order of actions.
Signed-off-by: Eytan Lifshitz <eytan.lifshitz@intel.com>
Reviewed-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
The AC / fifo mapping was wrong - BE packets landed in VO
FIFO. The iwl_mvm_tx_fifo enumeration isn't in the same
order as ieee80211_ac_numbers enumeration.
Since the firmware relies on fifo / ac mapping - this led
to wrong behavior. E.g. the firmware sends beacon with the
same QoS parameters as VO, and it actually took the
parameters of BK. There are probably more severe issues.
Reviewed-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
The bond 3ad and TLB/ALB has the same check path, so combine them.
Signed-off-by: Ding Tianhong <dingtianhong@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch corrects a problem in stmmac_ptp.c, functions
stmmac_adjust_time and stmmac_adjust_freq where the incorrect spinlocks
were released. This patch also addresses a problem in stmmac_main,
function stmmac_init_ptp where the capability detection for
advanced timestamping was masked by message masking.
This patch was touch tested using linuxptp, and runs without the previously
observed instabilities. More extensive testing is ongoing.
Vince
Signed-off-by: Vince Bridgers <vbridgers2013@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
With the way PHY IRQ handler is implemented (all real handling being pushed to
the workqueue and returning IRQ_HANDLED all the time PHY is active), we cannot
really claim that PHY IRQ can be shared when calling request_irq().
Signed-off-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com>
Acked-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The sizes chosen for the metadata and grant_copy_op arrays on the guest
receive size are wrong;
- The meta array is needlessly twice the ring size, when we only ever
consume a single array element per RX ring slot
- The grant_copy_op array is way too small. It's sized based on a bogus
assumption: that at most two copy ops will be used per ring slot. This
may have been true at some point in the past but it's clear from looking
at start_new_rx_buffer() that a new ring slot is only consumed if a frag
would overflow the current slot (plus some other conditions) so the actual
limit is MAX_SKB_FRAGS grant_copy_ops per ring slot.
This patch fixes those two sizing issues and, because grant_copy_ops grows
so much, it pulls it out into a separate chunk of vmalloc()ed memory.
Signed-off-by: Paul Durrant <paul.durrant@citrix.com>
Acked-by: Wei Liu <wei.liu2@citrix.com>
Cc: Ian Campbell <ian.campbell@citrix.com>
Cc: David Vrabel <david.vrabel@citrix.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
We should not assume that the PHY reset is always active low.
Retrieve this information from the device tree instead, so that the PHY reset
can work on both cases.
Reported-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Fabio Estevam <fabio.estevam@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch updates the generic iommu backend code to use the
it_page_shift field to determine the iommu page size instead of
using hardcoded values.
Signed-off-by: Alistair Popple <alistair@popple.id.au>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
The powerpc iommu uses a hardcoded page size of 4K. This patch changes
the name of the IOMMU_PAGE_* macros to reflect the hardcoded values. A
future patch will use the existing names to support dynamic page
sizes.
Signed-off-by: Alistair Popple <alistair@popple.id.au>
Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
netxen_process_lro() contains two bounds checks. One for the ring number
against the number of rings, and one for the Rx buffer ID against the
array of receive buffers.
Both of these have off-by-one errors, using > instead of >=. The correct
versions are used in netxen_process_rcv(), they're just wrong in
netxen_process_lro().
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
The current driver assumes that an skb fragment can only be upto jumbo
size. Presumably this was a fast-path optimization. This assumption is
no longer true as fragments can be upto 32k.
v2: Remove unnecessary parantheses per Eric Dumazet.
Cc: stable@vger.kernel.org
Signed-off-by: Nithin Nayak Sujir <nsujir@broadcom.com>
Signed-off-by: Michael Chan <mchan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
All other net drivers with PTP support enable it unconditionally.
Make tile_net consistent with this.
Signed-off-by: Ben Hutchings <bhutchings@solarflare.com>
Acked-by: Chris Metcalf <cmetcalf@tilera.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c: In function 'bnx2x_drv_info_ether_stat':
drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c:3302:46: error: dereferencing pointer to incomplete type
make[6]: *** [drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.o] Error 1
Reported-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The recent conversion to skb_set_hash() was incorrect. 'skb' is
uninitialized at this point.
Fixes: 0b68070347 ('net: vmxnet3 calls skb_set_hash')
Signed-off-by: Michal Schmidt <mschmidt@redhat.com>
Acked-by: Tom Herbert <therbert@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
There are inconsistencies wrt. feature propagation/inheritance between
macvlan and the underlying interface.
When a feature is turned off on the real device before a macvlan is
created on top, these will remain enabled on the macvlan device, whereas
turning off the feature on the lower device after macvlan creation the
kernel will propagate the changes to the macvlan.
The second issue is that, when propagating changes from underlying device
to the macvlan interface, macvlan can erronously lose its NETIF_F_LLTX flag,
as features are anded with the underlying device.
However, LLTX should be kept since it has no dependencies on physical
hardware (LLTX is set on macvlan creation regardless of the lower
device properties, see 8ffab51b3d
(macvlan: lockless tx path).
The LLTX flag is now forced regardless of user settings in absence of
layer2 hw acceleration (a6cc0cfa72,
net: Add layer 2 hardware acceleration operations for macvlan devices).
Use netdev_increment_features to rebuild the feature set on capability
changes on either the lower device or on the macvlan interface.
As pointed out by Ben Hutchings, use netdev_update_features on
NETDEV_FEAT_CHANGE event (it calls macvlan_fix_features/netdev_features_change
if needed).
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch adds the necessary support for configuring (and removing) multicast
filters to VFs.
Signed-off-by: Yuval Mintz <yuvalmin@broadcom.com>
Signed-off-by: Ariel Elior <ariele@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
It's possible for VMs with older versions of bnx2x to run over a hypervisor
with latest driver. If a VF in such a VM does not support RSS configuration,
the PF driver in the hypervisor will print an error message to system logs.
This changes the error message into a debug message, as this is very likely
a false alarm for an older VF (i.e., VF manages to work properly; PF simply
cannot configure the additional queues for it).
Signed-off-by: Yuval Mintz <yuvalmin@broadcom.com>
Signed-off-by: Ariel Elior <ariele@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
There's a known issue that using `ethtool -e' flips the endianity of the
written data, i.e., using `ethtool -E' to dump eeprom image and than using
`ethtool -e' to re-write that same image will result in an image where
the data has the opposite endianity.
Sadly, this cannot be fixed as there are already various tools deployed
based on the endianity of the eeprom read/write.
Instead, a comment is added to the code to help explain why this is un-fixable.
Signed-off-by: Yuval Mintz <yuvalmin@broadcom.com>
Signed-off-by: Ariel Elior <ariele@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This function adds several OS calls required to fully enable PCIe AER support -
pci_enable_pcie_error_reporting() and pci_cleanup_aer_uncorrect_error_status().
Signed-off-by: Yuval Mintz <yuvalmin@broadcom.com>
Signed-off-by: Ariel Elior <ariele@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This adds the ability for bnx2x to load after UNDI is used in the
preboot environment on a multi-function interface which is not the first
interface of a given device.
Notice a side-effect is that the order by which the functions are probed and
thus interfaces appear might change, as this patch utilizes the EPROBE_DEFER
return value (and mechanism).
Signed-off-by: Yuval Mintz <yuvalmin@broadcom.com>
Signed-off-by: Ariel Elior <ariele@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
