Use of_platform_depopulate() in cpsw_remove() instead of
of_device_unregister(), because CSPW child devices will not be
recreated otherwise on next insmod. of_platform_depopulate() is
correct way now as it will ensure that all steps done in
of_platform_populate() are reverted, including cleaning up of
OF_POPULATED flag.
Signed-off-by: NGrygorii Strashko <grygorii.strashko@ti.com>
Reviewed-by: NMugunthan V N <mugunthanvnm@ti.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The L3 error will be generated and system will crash during unloading
of CPSW driver if CPSW is used as module and ethX devices are down.
This happens because CPSW can be power off by PM runtime now when ethX
devices are down.

Hence, ensure that CPSW powered up by PM runtime before performing any
deinitialization actions which require CPSW registers access. In case
of PM runtime error just leave cpsw_remove() as we can't do anything
anymore.
Signed-off-by: NGrygorii Strashko <grygorii.strashko@ti.com>
Reviewed-by: NMugunthan V N <mugunthanvnm@ti.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Fix deadlock in cpdma_ctlr_destroy() which is triggered now on
cpsw module removal:
 cpsw_remove()
 - cpdma_ctlr_destroy()
   - spin_lock_irqsave(&ctlr->lock, flags)
   - cpdma_ctlr_stop()
     - spin_lock_irqsave(&ctlr->lock, flags);
   - cpdma_chan_destroy()
     - spin_lock_irqsave(&ctlr->lock, flags);

The issue has not been observed before because CPDMA channels have
been destroyed manually by CPSW until commit d941ebe8 ("net:
ethernet: ti: cpsw: use destroy ctlr to destroy channels") was merged.
Signed-off-by: NGrygorii Strashko <grygorii.strashko@ti.com>
Reviewed-by: NMugunthan V N <mugunthanvnm@ti.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The commit 637d3e99 ("cxgb4: Discard the packet if the length is
greater than mtu") introduced a regression in the VLAN interface
performance when Tx VLAN offload is disabled.

Check if skb is tagged, regardless of whether it is hardware accelerated
or not. Presently we were checking only for hardware acclereated one,
which caused performance to drop to ~0.17Mbps on a 10GbE adapter for
VLAN interface, when tx vlan offload is turned off using ethtool.
The ethernet head length calculation was going wrong in this case, and
driver ended up dropping packets.

Fixes: 637d3e99 ("cxgb4: Discard the packet if the length is greater than mtu")
Signed-off-by: NHariprasad Shenai <hariprasad@chelsio.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

There is a error message within devm_ioremap_resource
already, so remove the dev_err call to avoid redundant
error message.
Signed-off-by: NWei Yongjun <weiyj.lk@gmail.com>
Acked-By: NIyappan Subramanian <isubramanian@apm.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Each PF/VF has a limited number of vlan filters for
configuration purposes; This information is passed to qede
and is used to prevent over-usage - once a vlan is to be
configured and no filter credit is available, the driver
would switch into working in vlan-promisc mode.

Problem is the credit pool is shared by both PFs and VFs,
and currently PFs aren't deducting the filters that are
reserved for their VFs from their quota, which may lead
to some vlan filters failing unknowingly due to lack of credit.
Signed-off-by: NYuval Mintz <Yuval.Mintz@qlogic.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Driver uses reverse logic when checking if minimum
bandwidth configuration applied, causing it to
configure the guarantee only on the first hw-function.

Fixes: a0d26d5a ("qed*: Don't reset statistics on inner reload")
Signed-off-by: NYuval Mintz <Yuval.Mintz@qlogic.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Adding the necessary logic to prevet statistics reset
on inner-reload introduced a bug, and now statistics
are reset only when re-probing the driver.

Fixes: a0d26d5a ("qed*: Don't reset statistics on inner reload")
Signed-off-by: NYuval Mintz <Yuval.Mintz@qlogic.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Before requesting the firmware to start Rx queues,
driver goes and sets the queue producer in the device to 0.
But while the producer is 32-bit, the driver currently clears 64 bits,
effectively zeroing an additional CID's producer as well.
Signed-off-by: NYuval Mintz <Yuval.Mintz@qlogic.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Driver has reverse logic for checking the result of the
spoof-checking configuration. As a result, it would log that
the configuration failed [even though it succeeded], and will
no longer do anything when requested to remove the configuration,
as it's accounting of the feature will be incorrect.

Fixes: 6ddc7608 ("qed*: IOV support spoof-checking")
Signed-off-by: NYuval Mintz <Yuval.Mintz@qlogic.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

As part of ndo_vlan_rx_kill_vid() implementation,
qede is requesting firmware to remove the vlan filter.
This currently happens even if the vlan wasn't previously
added [In case device ran out of vlan credits].

For PFs this doesn't cause any issues as the firmware
would simply ignore the removal request. But for VFs their
parent PF is holding an accounting of the configured vlans,
and such a request would cause the PF to fail the VF's
removal request.

Simply fix this for both PFs & VFs and don't remove filters
that were not previously added.

Fixes: 7c1bfcad ("qede: Add vlan filtering offload support")
Signed-off-by: NYuval Mintz <Yuval.Mintz@qlogic.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

drivers/net/ethernet/apm/xgene/xgene_enet_hw.c: In function 'xgene_enet_phy_connect':
drivers/net/ethernet/apm/xgene/xgene_enet_hw.c:759:22: warning: unused variable 'adev' [-Wunused-variable]

Fixes: 8089a96f ("drivers: net: xgene: Add backward compatibility")
Reported-by: NStephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The be2net driver performs fw temperature queries on be_worker() routine,
which is executed each second for each be_adapter. There is a frequency
threshold to avoid fw query to happens at each call to be_worker();
instead, currently a fw query occurs once in 64 runs of the procedure.

Nevertheless, this fw temperature query is invoked only for adapters which
interface is up, so we can see I/O errors on read of hwmon counters from
userspace (from tools like lm-sensors) in case we have adapters' functions
which interface is down.

This patch moves the fw query code to be invoked even if interface is down.
No functional changes were introduced.
Signed-off-by: NGuilherme G. Piccoli <gpiccoli@linux.vnet.ibm.com>
Acked-by: NSathya Perla <sathya.perla@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The workqueue health->wq was used as per device private health thread.
This was done to perform delayed work.

The workqueue has a single workitem(&health->work) and
hence doesn't require ordering. It is involved in handling the health of
the device and is not being used on a memory reclaim path.
Hence, the singlethreaded workqueue has been replaced with the use of
system_wq.

Work item has been flushed in mlx5_health_cleanup() to ensure that
there are no pending tasks while disconnecting the driver.
Signed-off-by: NBhaktipriya Shridhar <bhaktipriya96@gmail.com>
Acked-by: NLeon Romanovsky <leonro@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

macsec_decrypt() is not called when validation is disabled and so
macsec_skb_cb(skb)->rx_sa is not set; but it is used later in
macsec_post_decrypt(), ensure that it's always initialized.

Fixes: c09440f7 ("macsec: introduce IEEE 802.1AE driver")
Signed-off-by: NBeniamino Galvani <bgalvani@redhat.com>
Acked-by: NSabrina Dubroca <sd@queasysnail.net>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Changed SGMII 1G get_settings to use phy_ethtool_gset.
Changed SGMII 1G set_settings to use phy_ethtool_sset.
Signed-off-by: NIyappan Subramanian <isubramanian@apm.com>
Tested-by: NFushen Chen <fchen@apm.com>
Tested-by: NToan Le <toanle@apm.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This patch reuses the mdio read/write and phy_register functions
and removed the local definitions.
Signed-off-by: NIyappan Subramanian <isubramanian@apm.com>
Tested-by: NFushen Chen <fchen@apm.com>
Tested-by: NToan Le <toanle@apm.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This patch enables MDIO driver by,

- Selecting MDIO_XGENE
- Changed open and close to use phy_start and phy_stop
- Changed to use mac_ops->tx(rx)_enable and tx(rx)_disable
Signed-off-by: NIyappan Subramanian <isubramanian@apm.com>
Tested-by: NFushen Chen <fchen@apm.com>
Tested-by: NToan Le <toanle@apm.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This patch adds xgene_enet_check_phy_hanlde() function that checks whether
MDIO driver is probed successfully and sets pdata->mdio_driver to true.
If MDIO driver is not probed, ethernet driver falls back to backward
compatibility mode.

Since enum xgene_enet_cmd is used by MDIO driver, removing this from
ethernet driver.
Signed-off-by: NIyappan Subramanian <isubramanian@apm.com>
Tested-by: NFushen Chen <fchen@apm.com>
Tested-by: NToan Le <toanle@apm.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Currently, SGMII based 1G rely on the hardware registers for link state
and sometimes it's not reliable.  To get most accurate link state, this
interface has to use the MDIO bus to poll the PHY.

In X-Gene SoC, MDIO bus is shared across RGMII and SGMII based 1G
interfaces, so adding this driver to manage MDIO bus.  This driver
registers the mdio bus and registers the PHYs connected to it.
Signed-off-by: NIyappan Subramanian <isubramanian@apm.com>
Tested-by: NFushen Chen <fchen@apm.com>
Tested-by: NToan Le <toanle@apm.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This patch fixes clock reset sequence.

- Added clock reset sequence for ACPI
- Added delay in clock reset sequence to make sure pulse is generated
- Added clk_unprepare_disable() in port shutdown to make sure
  clock increment/decrement counts are matching
- Removed MII_MGMT_CONFIG programming, since it is not required
- Fixed programming XGENET_CONFIG_REG to enable SGMII mode
Signed-off-by: NIyappan Subramanian <isubramanian@apm.com>
Tested-by: NFushen Chen <fchen@apm.com>
Tested-by: NToan Le <toanle@apm.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

When the driver is configured as kernel module and when it gets
unloaded and reloaded, kernel crash was observed.  This patch
addresses the software cleanup by doing the following,

- Moved register_netdev call after hardware is ready
- Since ndev is not ready, added set_irq_name to set irq name
- Since ndev is not ready, changed mdio_bus->parent to pdev->dev
- Replaced netif_start(stop)_queue by netif_tx_start(stop)_queues
- Removed napi_del call since it's called by free_netdev
- Added dev_close call, within remove
- Added shutdown callback
- Changed to use dmam_ APIs
Signed-off-by: NIyappan Subramanian <isubramanian@apm.com>
Tested-by: NFushen Chen <fchen@apm.com>
Tested-by: NToan Le <toanle@apm.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

When the driver is configured as kernel module and when it gets
unloaded and reloaded, kernel crash was observed.  This patch
address the hardware resource cleanups by doing the following,

- Added mac_ops->clear() to do prefetch buffer clean up
- Fixed delete freepool buffers logic
- Reordered mac_enable and mac_disable
- Added Tx completion ring free
- Moved down delete_desc_rings after ring cleanup
Signed-off-by: NIyappan Subramanian <isubramanian@apm.com>
Tested-by: NFushen Chen <fchen@apm.com>
Tested-by: NToan Le <toanle@apm.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Since mac_init is too heavy to be called when the link changes,
moved the speed_set configuration to a new function and added
mac_ops->set_speed function pointer.  This function will be
called from adjust_link callback.

Added cases for 10/100 support for SGMII based 1G interface.
Signed-off-by: NIyappan Subramanian <isubramanian@apm.com>
Tested-by: NFushen Chen <fchen@apm.com>
Tested-by: NToan Le <toanle@apm.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The earlier patch c3aaa06d (ipvlan: scrub skb before routing
in L3 mode.) did this but only for TX path in L3 mode. This
patch extends it for both the modes for TX/RX path.
Signed-off-by: NMahesh Bandewar <maheshb@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Add 5741X/5731X NPAR device IDs and dual media SFP/10GBase-T device IDs.
Signed-off-by: NMichael Chan <michael.chan@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

If there are not enough resources to enable ntuple filtering,
log a warning message.

v2: Use single message and add missing newline.
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>

Include the destination MAC address in the ntuple filter structure.  The
current code assumes that the destination MAC address is always the MAC
address of the NIC.  This may not be true if there are macvlans, for
example.  Add destination MAC address checking and configure the filter
correctly using the correct index for the destination MAC address.
Signed-off-by: NMichael Chan <michael.chan@broadcom.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Slowpath completion handling is incorrectly changing
SPQ_RING_SIZE bits instead of a single one.

Fixes: 76a9a364 ("qed: fix handling of concurrent ramrods")
Signed-off-by: NManish Chopra <manish.chopra@qlogic.com>
Signed-off-by: NYuval Mintz <Yuval.Mintz@qlogic.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

alloc_workqueue replaces deprecated create_singlethread_workqueue().

A dedicated workqueue has been used since the workitems are being used
on a packet tx/rx path. Hence, WQ_MEM_RECLAIM has been set to guarantee
forward progress under memory pressure.

An ordered workqueue has been used since workitems &cfhsi->wake_up_work
and &cfhsi->wake_down_work cannot be run concurrently.

Calls to flush_workqueue() before destroy_workqueue() have been dropped
since destroy_workqueue() itself calls drain_workqueue() which flushes
repeatedly till the workqueue becomes empty.
Signed-off-by: NBhaktipriya Shridhar <bhaktipriya96@gmail.com>
Acked-by: NTejun Heo <tj@kernel.org>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Verify that the device state is registered before un-registering it.
This check is required to prevent an OOPS on flows that do
re-registration of the device and its previous state was
unregistered.

Fixes: 225c7b1f ("IB/mlx4: Add a driver Mellanox ConnectX InfiniBand adapters")
Signed-off-by: NAlex Vesker <valex@mellanox.com>
Signed-off-by: NTariq Toukan <tariqt@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

When CONFIG_NET_CLS_ACT isn't set 'struct tcf_exts' has no member named
'actions' and we therefore must not access it. Otherwise compilation
fails.

Fix this by introducing a new macro similar to tc_no_actions(), which
always returns 'false' if CONFIG_NET_CLS_ACT isn't set.

Fixes: 763b4b70 ("mlxsw: spectrum: Add support in matchall mirror TC offloading")
Reported-by: Nkbuild test robot <fengguang.wu@intel.com>
Signed-off-by: NJiri Pirko <jiri@mellanox.com>
Signed-off-by: NIdo Schimmel <idosch@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Such a big dump of register values is hardly useful on a production
system.

Another downside of the now removed functions is that calling
emac_dump_regs resulted in at least 87 calls to dev_info while holding a
spinlock and having irqs off which is a big source of latency.
Signed-off-by: NUwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Add support for query the minimum inline mode from the Firmware.
It is required for correct TX steering according to L3/L4 packet
headers.

Each send queue (SQ) has inline mode that defines the minimal required
headers that needs to be copied into the SQ WQE.
The driver asks the Firmware for the wqe_inline_mode device capability
value.  In case the device capability defined as "vport context" the
driver must check the reported min inline mode from the vport context
before creating its SQs.
Signed-off-by: NHadar Hen Zion <hadarh@mellanox.com>
Signed-off-by: NSaeed Mahameed <saeedm@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Each send queue (SQ) has inline mode that defines the minimal required
inline headers in the SQ WQE.
Before sending each packet check that the minimum required headers
on the WQE are copied.
Signed-off-by: NHadar Hen Zion <hadarh@mellanox.com>
Signed-off-by: NSaeed Mahameed <saeedm@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Added a condition to avoid bonding devices with same MAC registering
as VF.
Signed-off-by: NHaiyang Zhang <haiyangz@microsoft.com>
Reviewed-by: NK. Y. Srinivasan <kys@microsoft.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Remove including <linux/version.h> that don't need it.
Signed-off-by: NWei Yongjun <weiyj.lk@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Test the cipher suite initialization in case ICV length has a value
different than its default. If this test fails, creation of a new macsec
link will also fail. This avoids situations where further security
associations can't be added due to failures of crypto_aead_setauthsize(),
caused by unsupported user-provided values of the ICV length.
Signed-off-by: NDavide Caratti <dcaratti@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

preserve the return value of AEAD functions that are called when a SA is
created, to avoid inappropriate display of "RTNETLINK answers: Cannot
allocate memory" message.
Signed-off-by: NDavide Caratti <dcaratti@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

IEEE 802.1AE-2006 standard recommends that the ICV element in a MACsec
frame should not exceed 16 octets: add MACSEC_STD_ICV_LEN in uapi
definitions accordingly, and avoid accepting configurations where the ICV
length exceeds the standard value. Leave definition of MACSEC_MAX_ICV_LEN
unchanged for backwards compatibility with userspace programs.

Fixes: dece8d2b ("uapi: add MACsec bits")
Signed-off-by: NDavide Caratti <dcaratti@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>