Jeff Kirsher says:
====================
Intel Wired LAN Driver Updates 2015-06-26
This series contains fixes for igb, e1000e and i40evf.
Todd disables IPv6 extension header processing due to a hardware errata
and bumps the driver version.
Yanir provides six fixes for e1000e. First is a fix for a locking issue
where we were not always taking the pci_bus_sem semaphore all the time
when calling pci_disable_link_state_locked(), so fix the code to only call
pci_disable_link_state_locked() when the semaphore has been acquired,
otherwise call pci_disable_link_state(). A previous fix for i219 where
the hardware prevented ULP entry caused EEE in Sx not the be enabled, so
modify the code flow that allows both ULP and EEE in Sx. Fix an issue
when running 10/100 full duplex on i219 where CRC errors were occurring
by increasing the IPG from 8 to 0xC as per the hardware developers.
Fix a data corruption issue found on some platforms by increasing the
minimum gap between the PHY FIFO read and write pointers. Fix i219,
which does not require the K1 workaround for LPT devices.
Mitch provides a i40evf fix for a panic when changing MTU. Down was
requesting queue disables, but then exited immediately without waiting
for the queues to actually be disabled. This could allow any function
called after i40evf_down() to run immediately, including i40evf_up(),
and causes a memory leak. Fixed the issue by removing the whole
reinit_locked function which allows for the driver to handle the state
changes by requesting reset from the periodic timer. The second fix
resolves an issue where RSS was being configured as though it is using
the maximum number of queue. This can cause the device to drop a lot
of receive traffic, as the packets get assigned to non-functional queues.
This is resolved by only configuring RSS with the number of active queues.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
As some C45 10G PHYs(e.g. Cortina CS4315/CS4340 PHY) have
zero Devices In package, current driver can't get correct
devices_in_package value by non-zero Devices In package.
so let's probe more with zero Devices In package to support
more C45 PHYs.
Signed-off-by: Shengzhou Liu <Shengzhou.Liu@freescale.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Not all silicon implementations of the Freescale FEC hardware module
have the RACC (Receive Accelerator Function) register, so we should not
be trying to access it on those that don't. Currently none of the ColdFire
based parts with a FEC have it.
Support for RACC was introduced by commit 4c09eed9 ("net: fec: Enable imx6
enet checksum acceleration"). A fix was introduced in commit d1391930
("net: fec: Fix build for MCF5272") that disables its use on the ColdFire
M5272 part, but it doesn't fix the general case of other ColdFire parts.
To fix we create a quirk flag, FEC_QUIRK_HAS_RACC, and check it before
working with the RACC register.
Signed-off-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
When limiting phy link speed using "max-speed" to 100mbps or less on a
giga bit phy, phy never completes auto negotiation and phy state
machine is held in PHY_AN. Fixing this issue by comparing the giga
bit advertise though phydev->supported doesn't have it but phy has
BMSR_ESTATEN set. So that auto negotiation is restarted as old and
new advertise are different and link comes up fine.
Signed-off-by: Mugunthan V N <mugunthanvnm@ti.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
If CONFIG_ACPI=n:
drivers/net/ethernet/apm/xgene/xgene_enet_main.c: In function ‘xgene_enet_get_resources’:
drivers/net/ethernet/apm/xgene/xgene_enet_main.c:951: warning: ‘ret’ may be used uninitialized in this function
If the driver is bound to a legacy platform device, ret will contain
arbitrary data. If it is non-zero, it will be returned to the caller as
an error code.
Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
The function netif_set_real_num_tx_queues() will return -EINVAL if
the second parameter < 1, so call this function with the second
parameter set to 0 is meaningless.
Signed-off-by: Liang Li <liang.z.li@intel.com>
Reviewed-by: David Vrabel <david.vrabel@citrix.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The driver will only configure as many queues as there are available
CPUs, up the maximum number of queues. However, it always configures
RSS as though it is using the maximum number of queues. This can cause
the device to drop a lot of RX traffic, as the packets get assigned to
nonfunctional queues.
Fix this by only configuring RSS with the number of active queues.
Signed-off-by: Mitch Williams <mitch.a.williams@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Down was requesting queue disables, but then exited immediately
without waiting for the queues to actually disable. This could
allow any function called after i40evf_down to run immediately,
including i40evf_up, and causes a memory leak.
Removing the whole reinit_locked function is the best way
to go about this, and allows for the driver to handle the
state changes by requesting reset from the periodic timer.
Also, add a couple WARN_ONs in slow path to help us recognize
if we re-introduce this issue or missed any cases.
Signed-off-by: Mitch Williams <mitch.a.williams@intel.com>
Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
In SPT hardware does not require this driver workaround.
Removed the conditional that caused K1 workaround execution on SPT.
Signed-off-by: Yanir Lubetkin <yanirx.lubetkin@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Due to clocking changes in the Skylake platform, there was i219
data corruption. To work around this, HW team reported the need
to increase the minimum gap between the PHY FIFO read and write pointers.
Signed-off-by: Yanir Lubetkin <yanirx.lubetkin@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
In SPT/i219, there were CRC errors in speed 10/100 full duplex.
The solution given by the HW team is to increase the IPG from 8 to 0xC
Signed-off-by: Yanir Lubetkin <yanirx.lubetkin@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
In i219, there is a hardware bug that prevented ULP entry.
A side effect of the original software fix for this was that EEE in
Sx couldn't be enabled.
This patch implements a modified flow that allows both ULP and EEE in Sx.
Signed-off-by: Yanir Lubetkin <yanirx.lubetkin@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
On power up, the MAC - PHY interface needs to be set to PCIe, even if
cable is disconnected. In ME systems, the ME handles this on exit from
Sx state. In non-ME, the driver handles it. Added a check for non-ME
system to the driver code that handles that.
Signed-off-by: Yanir Lubetkin <yanirx.lubetkin@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
e1000e_disable_aspm called pci_disable_link_state_locked which requires
pci_bus_sem to be held, but is also called from places where this semaphore
was not previously acquired. This patch implements two flavors of
disable_aspm, one that acquires the lock, and the other (_locked) which
should be called when the semaphore is already acquired.
Signed-off-by: Yanir Lubetkin <yanirx.lubetkin@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Bump version of igb to igb-5.2.18
Signed-off-by: Todd Fujinaka <todd.fujinaka@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Disable IPv6 extension header processing as per hardware errata.
Also fix copyright date.
Signed-off-by: Todd Fujinaka <todd.fujinaka@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
Due to FW constraints, driver must make sure that transmitted SKBs will
not be too fragmented, or in the case that they are - that each 'window'
of fragments passed to the FW would contain at least an mss worth of data.
For encapsultaed packets the calculation is wrong, since it ignores the
inner headers in the calculation of the headers' length.
This could lead to a FW assertion in case of a too-fragmented encapsulated
packet.
Signed-off-by: Yuval Mintz <Yuval.Mintz@qlogic.com>
Signed-off-by: Ariel Elior <Ariel.Elior@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
During an error flow when trying to access the nvram the driver doesn't
release the hw lock it acquired.
Signed-off-by: Yuval Mintz <Yuval.Mintz@qlogic.com>
Signed-off-by: Ariel Elior <Ariel.Elior@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Since driver statistics flow access MACs and those might reset during
link re-configurations, when we're about to change link properties we
have to make sure that statistics are not operational.
Statisics would be re-enabled [i.e., gathering of statistics would
re-commence] once physical link is achieved again.
Since driver employs a link-flap avoidance scheme, there are scenarios
where driver will receive no indication that the new link is up, and
as a result the statistics would not be re-enabled.
Preventing LFA from working in such cases would guarantee that we'll
always receive such indications and thus will fix statistics gathering.
Signed-off-by: Ariel Elior <Ariel.Elior@qlogic.com>
Signed-off-by: Yuval Mintz <Yuval.Mintz@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
20g-capable devices are not configured properly for self-test, using
10g as their speed which cause the link indication to remain down and
fail the internal loopback test.
Signed-off-by: Yuval Mintz <Yuval.Mintz@qlogic.com>
Signed-off-by: Ariel Elior <Ariel.Elior@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
There's a bug in today's driver where VF requests to add/remove MAC filters
always reach the Hypervisor as add requests.
This prevents the VF from changing its MAC address, as it cannot remove the
previously configured MAC and runs out of MAC credits.
Signed-off-by: Shahed Shaikh <Shahed.Shaikh@qlogic.com>
Signed-off-by: Yuval Mintz <Yuval.Mintz@qlogic.com>
Signed-off-by: Ariel Elior <Ariel.Elior@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The scratchpad is a shared block between all functions of a given device.
Due to HW limitations, we can't properly close its parity notifications
to all functions on legal flows.
E.g., it's possible that while taking a register dump from one function
a parity error would be triggered on other functions.
Today driver doesn't consider this parity as a 'real' parity unless its
being accompanied by additional indications [which would happen in a real
parity scenario]; But it does print notifications for such events in the
system logs.
This eliminates such prints - in case of real parities driver would have
additional indications; But if this is the only signal user will not even
see a parity being logged in the system.
Signed-off-by: Manish Chopra <Manish.Chopra@qlogic.com>
Signed-off-by: Yuval Mintz <Yuval.Mintz@qlogic.com>
Signed-off-by: Ariel Elior <Ariel.Elior@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Each time a flow finishes reads from the classification shadow
configuration in the driver, that flow would check for pending commands
and pass them to FW if possible.
In case there's already a completion pending command, I.e., a ramrod
that has been sent to the FW and is yet to be completed while said flow
tries to configure the pending command we would get a false error message
in logs [and panic if SOE was used for driver compilation] since the
command could not have been completed.
This prevents said print [and panic]; The pending command will be sent by
the time the completion of the current sent command would arrive.
Signed-off-by: Yuval Mintz <Yuval.Mintz@qlogic.com>
Signed-off-by: Ariel Elior <Ariel.Elior@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
ethtool shows KR supported/advertised speeds incorrectly as baseT
in cases the board is in fact KR-base.
Signed-off-by: Yaniv Rosner <Yaniv.Rosner@qlogic.com>
Signed-off-by: Yuval Mintz <Yuval.Mintz@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This fixes several issues relating to asymmetric configuration:
1. When user requests to disable TX, the local-device needs to
advertise both PAUSE and ASM_DIR, but to avoid transmitting pause
frames. In the 578xx, it would ignore the TX disable.
2. When user advertises RX-only, ASM_DIR was advertised instead of
PAUSE/ASM_DIR.
3. When changing mode, the advertised PAUSE/ASM_DIR was not cleared
before setting new one, so disabling RX or TX had no impact on the
'advertised' as appeared in the 'ethtool -a' output.
Signed-off-by: Yaniv Rosner <Yaniv.Rosner@qlogic.com>
Signed-off-by: Yuval Mintz <Yuval.Mintz@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
We use spinlock to access a single flag. We can avoid spin_locks by using
atomic variable and atomic_cmpxchg(). Use atomic_cmpxchg to set the flag
for idle to poll. And a simple atomic_set to unlock (set idle from poll).
In napi poll, if gro is enabled, we call napi_gro_receive() to deliver the
packets. Before we call napi_complete(), i.e while re-polling, if low
latency busy poll is called, we use netif_receive_skb() to deliver the packets.
At this point if there are some skb's held in GRO, busy poll could deliver the
packets out of order. So we call napi_gro_flush() to flush skbs before we
move the napi poll to idle.
Signed-off-by: Govindarajulu Varadarajan <_govind@gmx.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Vitesse VSC8641 is compatible with Vitesse 82xx
Signed-off-by: Shaohui Xie <Shaohui.Xie@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
CONFIG_GIANFAR is not depended on FSL_SOC, it
can be built on non-PPC platforms.
Signed-off-by: Alison Wang <alison.wang@freescale.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
There was a missing assignment so the "if (ret)" on the next line is
never true.
Fixes: f21fb3ed36 ('Add support of Cavium Liquidio ethernet adapters')
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
We need to unlock before returning here.
Fixes: a0d2f20650 ('Renesas Ethernet AVB PTP clock driver')
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Acked-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When in HA mode, the driver exposes an IB (RoCE) device instance with only
one port. Under SRIOV, the existing implementation doesn't go well with
the PF RoCE driver's role of Special QPs Para-Virtualization, etc.
As such, disable HA for the mlx4 PF RoCE device in SRIOV mode.
Fixes: a575009030 ('IB/mlx4: Add port aggregation support')
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The check_csum() function relied on hwtstamp_rx_filter to know if rxvlan
offload is disabled. This is wrong since rxvlan offload can be switched
on/off regardless of hwtstamp_rx_filter.
Also moved check_csum to query CQE information to identify VLAN packets
and removed the check of IP packets, since it has been validated before.
Fixes: f8c6455bb0 ('net/mlx4_en: Extend checksum offloading by CHECKSUM COMPLETE')
Signed-off-by: Ido Shamay <idos@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Indication of a single completed packet, marked by txbbs_skipped
being bigger then zero, in not enough in order to wake up a
stopped TX queue. The completed packet may contain a single TXBB,
while next packet to be sent (after the wake up) may have multiple
TXBBs (LSO/TSO packets for example), causing overflow in queue followed
by WQE corruption and TX queue timeout.
Instead, wake the stopped queue only when there's enough room for the
worst case (maximum sized WQE) packet that we should need to handle after
the queue is opened again.
Also created an helper routine - mlx4_en_is_tx_ring_full, which checks
if the current TX ring is full or not. It provides better code readability
and removes code duplication.
Signed-off-by: Ido Shamay <idos@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
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: ddae0349fd ('net/mlx4: Change QP allocation scheme')
Signed-off-by: Eran Ben Elisha <eranbe@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Pull networking updates from David Miller:
1) Add TX fast path in mac80211, from Johannes Berg.
2) Add TSO/GRO support to ibmveth, from Thomas Falcon
3) Move away from cached routes in ipv6, just like ipv4, from Martin
KaFai Lau.
4) Lots of new rhashtable tests, from Thomas Graf.
5) Run ingress qdisc lockless, from Alexei Starovoitov.
6) Allow servers to fetch TCP packet headers for SYN packets of new
connections, for fingerprinting. From Eric Dumazet.
7) Add mode parameter to pktgen, for testing receive. From Alexei
Starovoitov.
8) Cache access optimizations via simplifications of build_skb(), from
Alexander Duyck.
9) Move page frag allocator under mm/, also from Alexander.
10) Add xmit_more support to hv_netvsc, from KY Srinivasan.
11) Add a counter guard in case we try to perform endless reclassify
loops in the packet scheduler.
12) Extern flow dissector to be programmable and use it in new "Flower"
classifier. From Jiri Pirko.
13) AF_PACKET fanout rollover fixes, performance improvements, and new
statistics. From Willem de Bruijn.
14) Add netdev driver for GENEVE tunnels, from John W Linville.
15) Add ingress netfilter hooks and filtering, from Pablo Neira Ayuso.
16) Fix handling of epoll edge triggers in TCP, from Eric Dumazet.
17) Add an ECN retry fallback for the initial TCP handshake, from Daniel
Borkmann.
18) Add tail call support to BPF, from Alexei Starovoitov.
19) Add several pktgen helper scripts, from Jesper Dangaard Brouer.
20) Add zerocopy support to AF_UNIX, from Hannes Frederic Sowa.
21) Favor even port numbers for allocation to connect() requests, and
odd port numbers for bind(0), in an effort to help avoid
ip_local_port_range exhaustion. From Eric Dumazet.
22) Add Cavium ThunderX driver, from Sunil Goutham.
23) Allow bpf programs to access skb_iif and dev->ifindex SKB metadata,
from Alexei Starovoitov.
24) Add support for T6 chips in cxgb4vf driver, from Hariprasad Shenai.
25) Double TCP Small Queues default to 256K to accomodate situations
like the XEN driver and wireless aggregation. From Wei Liu.
26) Add more entropy inputs to flow dissector, from Tom Herbert.
27) Add CDG congestion control algorithm to TCP, from Kenneth Klette
Jonassen.
28) Convert ipset over to RCU locking, from Jozsef Kadlecsik.
29) Track and act upon link status of ipv4 route nexthops, from Andy
Gospodarek.
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next: (1670 commits)
bridge: vlan: flush the dynamically learned entries on port vlan delete
bridge: multicast: add a comment to br_port_state_selection about blocking state
net: inet_diag: export IPV6_V6ONLY sockopt
stmmac: troubleshoot unexpected bits in des0 & des1
net: ipv4 sysctl option to ignore routes when nexthop link is down
net: track link-status of ipv4 nexthops
net: switchdev: ignore unsupported bridge flags
net: Cavium: Fix MAC address setting in shutdown state
drivers: net: xgene: fix for ACPI support without ACPI
ip: report the original address of ICMP messages
net/mlx5e: Prefetch skb data on RX
net/mlx5e: Pop cq outside mlx5e_get_cqe
net/mlx5e: Remove mlx5e_cq.sqrq back-pointer
net/mlx5e: Remove extra spaces
net/mlx5e: Avoid TX CQE generation if more xmit packets expected
net/mlx5e: Avoid redundant dev_kfree_skb() upon NOP completion
net/mlx5e: Remove re-assignment of wq type in mlx5e_enable_rq()
net/mlx5e: Use skb_shinfo(skb)->gso_segs rather than counting them
net/mlx5e: Static mapping of netdev priv resources to/from netdev TX queues
net/mlx4_en: Use HW counters for rx/tx bytes/packets in PF device
...
Conflicts:
drivers/net/ethernet/mellanox/mlx4/main.c
net/packet/af_packet.c
Both conflicts were cases of simple overlapping changes.
Signed-off-by: David S. Miller <davem@davemloft.net>
Current implementation of descriptor init procedure only takes
care about setting/clearing ownership flag in "des0"/"des1"
fields while it is perfectly possible to get unexpected bits
set because of the following factors:
[1] On driver probe underlying memory allocated with
dma_alloc_coherent() might not be zeroed and so
it will be filled with garbage.
[2] During driver operation some bits could be set by SD/MMC
controller (for example error flags etc).
And unexpected and/or randomly set flags in "des0"/"des1"
fields may lead to unpredictable behavior of GMAC DMA block.
This change addresses both items above with:
[1] Use of dma_zalloc_coherent() instead of simple
dma_alloc_coherent() to make sure allocated memory is
zeroed. That shouldn't affect performance because
this allocation only happens once on driver probe.
[2] Do explicit zeroing of both "des0" and "des1" fields
of all buffer descriptors during initialization of
DMA transfer.
And while at it fixed identation of dma_free_coherent()
counterpart as well.
Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
Cc: Giuseppe Cavallaro <peppe.cavallaro@st.com>
Cc: arc-linux-dev@synopsys.com
Cc: linux-kernel@vger.kernel.org
Cc: stable@vger.kernel.org
Cc: David Miller <davem@davemloft.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
This bug pops up with NetworkManager on Fedora 21. NetworkManager tends to
stop the interface (nicvf_stop() is called) before changing settings. In
stopped state MAC cannot be sent to a PF. However, when the interface is
restarted (nicvf_open() is called), we ping the PF using NIC_MBOX_MSG_READY
message, and the PF replies back with old MAC address, overriding what we
had after MAC setting from userspace. As a result, we cannot set MAC
address using NetworkManager.
This patch introduces special tracking of MAC change in stopped state so
that the correct new MAC address is sent to a PF when interface is reopen.
Signed-off-by: Pavel Fedin <p.fedin@samsung.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Prefetch the 1st cache line used by the buffer pointed by
the skb linear data.
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Separate between mlx5e_get_cqe() and mlx5_cqwq_pop(), this helps for
better code readability and better CQ buffer management.
Signed-off-by: Achiad Shochat <achiad@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Use container_of() instead.
Signed-off-by: Achiad Shochat <achiad@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Coding Style fix, remove extra spaces.
Signed-off-by: Achiad Shochat <achiad@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In order to save PCI BW consumed by TX CQEs and to reduce the amount of
CPU cache misses caused by TX CQE reading, we request TX CQE generation
only when skb->xmit_more=0.
As a consequence of the above, a single TX CQE may now indicate the
transmission completion of multiple TX SKBs.
This also handles a problem introduced in commit b1b8105ebf41 "net/mlx5e:
Support NETIF_F_SG" where we didn't ask for NOP completions while the
driver didn't have the proper code to handle this case.
Fixes: b1b8105ebf41 ('net/mlx5e: Support NETIF_F_SG')
Signed-off-by: Achiad Shochat <achiad@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
NOP completion SKBs are always NULL.
Signed-off-by: Achiad Shochat <achiad@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
It is already assigned at mlx5e_build_rq_param()
Signed-off-by: Achiad Shochat <achiad@mellanox.com>
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Instead of counting number of gso fragments, we can use
skb_shinfo(skb)->gso_segs.
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
To save per-packet calculations, we use the following static mappings:
1) priv {channel, tc} to netdev txq (used @mlx5e_selec_queue())
2) netdev txq to priv sq (used @mlx5e_xmit())
Thanks to these static mappings, no more need for a separate implementation
of ndo_start_xmit when multiple TCs are configured.
We believe the performance improvement of such separation would be negligible, if any.
The previous way of dynamically calculating the above mappings required
allocating more TX queues than actually used (@alloc_etherdev_mqs()),
which is now no longer needed.
Signed-off-by: Saeed Mahameed <saeedm@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David 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: Eran Ben Elisha <eranbe@mellanox.com>
Signed-off-by: Or Gerlitz <ogerlitz@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
