The current default ITR for Tx is overly restrictive. Using a simple
netperf TCP_STREAM test, we top out at about 10Gb/s for a single thread
when running using 1500 byte frames. By reducing the ITR value to 25usec
(up to 40K interrupts a second from 10K), we are able to achieve 36Gb/s
for a single thread TCP stream test.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Krishneil Singh <Krishneil.k.singh@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Reviewed-by: Bruce Allan <bruce.w.allan@intel.com>
Tested-by: Krishneil Singh <Krishneil.k.singh@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

The existing adaptive ITR algorithm is overly restrictive. It throttles
incorrectly for various traffic rates, and does not produce good
performance. The algorithm now allows for more interrupts per second,
and does some calculation to help improve for smaller packet loads. In
addition, take into account the new itr_scale from the hardware which
indicates how much to scale due to PCIe link speed.
Reported-by: Matthew Vick <matthew.vick@intel.com>
Reported-by: Alex Duyck <alexander.duyck@gmail.com>
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Krishneil Singh <krishneil.k.singh@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

Define a macro for identifying when the itr value is dynamic or
adaptive. The concept was taken from i40e. This helps make clear what
the check is, and reduces the line length to something more reasonable
in a few places.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Reviewed-by: Bruce Allan <bruce.w.allan@intel.com>
Tested-by: Krishneil Singh <krishneil.k.singh@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

The Intel Ethernet Switch FM10000 Host Interface interrupt throttle
timers are based on the PCIe link speed. Because of this, the value
being programmed into the ITR registers must be scaled accordingly.

For the PF, this is as simple as reading the PCIe link speed and storing
the result. However, in the case of SR-IOV, the VF's interrupt throttle
timers are based on the link speed of the PF. However, the VF is unable
to get the link speed information from its configuration space, so the
PF must inform it of what scale to use.

Rather than pass this scale via mailbox message, take advantage of
unused bits in the TDLEN register to pass the scale. It is the
responsibility of the PF to program this for the VF while setting up the
VF queues and the responsibility of the VF to get the information
accordingly. This is preferable because it allows the VF to set up the
interrupts properly during initialization and matches how the MAC
address is passed in the TDBAL/TDBAH registers.

Since we're modifying fm10k_type.h, we may as well also update the
copyright year.
Reported-by: Matthew Vick <matthew.vick@intel.com>
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Reviewed-by: Bruce Allan <bruce.w.allan@intel.com>
Tested-by: Krishneil Singh <krishneil.k.singh@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

Originally this statistic was renamed because the method of dropping was
called "drop_oversized_messages", but this logic has changed much, and
this counter does actually represent messages which we failed to
transmit for a number of reasons. Rename the counter back to tx_dropped
since this is when it will increment, and it is less confusing.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Krishneil Singh <Krishneil.k.singh@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

A previous bug was uncovered by addition of a debug stat to indicate the
actual number of DWORDS we pulled from the mbmem. It turned out this was
not the same as the tx_dwords counter. While the previous bug fix should
have corrected this in all cases, add some debug stats that count the
number of DWORDs pushed or pulled from the mbmem. A future debugger may
take advantage of this statistic for debugging purposes. Since we're
modifying fm10k_mbx.h, update the copyright year as well.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Reviewed-by: Bruce Allan <bruce.w.allan@intel.com>
Tested-by: Krishneil Singh <Krishneil.k.singh@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Tested-by: Krishneil Singh <Krishneil.k.singh@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

Since the resultant data type of the mac_update.mac_upper field is u16,
it does not make sense to typecast u8 variables to u32 first. Since
we're modifying fm10k_pf.c, also update the copyright year.
Reported-by: Matthew Vick <matthew.vick@intel.com>
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Reviewed-by: Bruce Allan <bruce.w.allan@intel.com>
Tested-by: Krishneil Singh <Krishneil.k.singh@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

The init_hw function may fail, and in the case of VFs, it might change
the number of maximum queues available. Thus, for every flow which
checks init_hw, we need to ensure that we clear the queue scheme before,
and initialize it after. The fm10k_io_slot_reset path will end up
triggering a reset so fm10k_reinit needs this change. The
fm10k_io_error_detected and fm10k_io_resume also need to properly clear
and reinitialize the queue scheme.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Reviewed-by: Bruce Allan <bruce.w.allan@intel.com>
Tested-by: Krishneil Singh <Krishneil.k.singh@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

A recent change modified init_hw in some flows the function may fail on
VF devices. For example, if a VF doesn't yet own its own queues.
However, many callers of init_hw didn't bother to check the error code.
Other callers checked but only displayed diagnostic messages without
actually handling the consequences.

Fix this by (a) always returning and preventing the netdevice from going
up, and (b) printing the diagnostic in every flow for consistency. This
should resolve an issue where VF drivers would attempt to come up
before the PF has finished assigning queues.

In addition, change the dmesg output to explicitly show the actual
function that failed, instead of combining reset_hw and init_hw into a
single check, to help for future debugging.

Fixes: 1d568b0f6424 ("fm10k: do not assume VF always has 1 queue")
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Reviewed-by: Bruce Allan <bruce.w.allan@intel.com>
Tested-by: Krishneil Singh <Krishneil.k.singh@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

VF drivers must detect how many queues are available. Previously, the
driver assumed that each VF has at minimum 1 queue. This assumption is
incorrect, since it is possible that the PF has not yet assigned the
queues to the VF by the time the VF checks. To resolve this, we added a
check first to ensure that the first queue is infact owned by the VF at
init_hw_vf time. However, the code flow did not reset hw->mac.max_queues
to 0. In some cases, such as during reinit flows, we call init_hw_vf
without clearing the previous value of hw->mac.max_queues. Due to this,
when init_hw_vf errors out, if its error code is not properly handled
the VF driver may still believe it has queues which no longer belong to
it. Fix this by clearing the hw->mac.max_queues on exit due to errors.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Reviewed-by: Bruce Allan <bruce.w.allan@intel.com>
Tested-by: Krishneil Singh <Krishneil.k.singh@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

Don't change netdev hw_features later in fm10k_probe, instead set all
values inside fm10k_alloc_netdev. To do so, we need to know the MAC type
(whether it is PF or VF) in order to determine what to do. This helps
ensure that all logic regarding features is co-located.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Reviewed-by: Bruce Allan <bruce.w.allan@intel.com>
Tested-by: Krishneil Singh <krishneil.k.singh@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

Sergei Shtylyov says:

====================
Renesas: read MAC address registers only once

Here's 2 patches against DaveM's 'net-next.git' repo. Here we optimize
the MAC address register reading (left over from a bootloader).

[1/2] ravb: read MAC address registers only once
[2/2] sh_eth: read MAC address registers only once
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

The code reading the MAHR/MALR registers in read_mac_address() is terribly
ineffective -- it reads MAHR 4 times and MALR 2 times, while it's enough to
read each register only once.  Use the local variables to achieve that,
somewhat beautifying the code while at it...
Signed-off-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The code reading the MAHR/MALR registers in ravb_read_mac_address() is
terribly ineffective -- it reads  MAHR 4 times and MALR 2 times, while
it's enough to read each register only once. Use the local variables to
achieve that, somewhat beautifying the code while at it...
Signed-off-by: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Michal Schmidt says:

====================
bnx2x: fewer error messages, simplification

This removes one redundant error message in bnx2x and changes another one to
WARN_ONCE. The third patch is a small simplification in ethtool stats.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

The 'flags' field in bnx2x_stats_arr[] serves only one purpose - to tell
us if the statistic is a per-port stat and thus should not be shown for
virtual functions. It's strange that the field can have three different
values. A boolean will do just fine.

Also remove IS_FUNC_STAT(). It was used only once and it's in fact just
a negation of IS_PORT_STAT().
Signed-off-by: Michal Schmidt <mschmidt@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

It's supposed to be impossible for TPA to give us anything else
than IPv4 or IPv6 here. But in case there is a way to reach this error
by some strange received frames, we don't want to flood the kernel log.
WARN_ONCE is better for this purpose.
Signed-off-by: Michal Schmidt <mschmidt@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

alloc_pages() already prints a warning when it fails. No need to emit
another message. Certainly not at KERN_ERR level, because it is no big
deal if this GFP_ATOMIC allocation fails occasionally.
Signed-off-by: Michal Schmidt <mschmidt@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

As suggested by Eric, these helpers should have const dev param.
Suggested-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

A change in MCFW behaviour means that the net driver must update its record
of the warm_boot_count by reading it from the ER_DZ_BIU_MC_SFT_STATUS
register.

On v4.6.x MCFW the global boot count was incremented when some functions
needed to be reset to enable multicast chaining, so all functions saw the
same value.  In that case, the driver needed to increment its
warm_boot_count when other functions were reset, to avoid noticing it later
and then trying to reset itself to recover unnecessarily.

With v4.7+ MCFW, the boot count in firmware doesn't change as that is
unnecessary since the PFs that have been reset will each receive an MC
reboot notification.  In that case, the driver re-reads the unchanged
value.
Signed-off-by: Bert Kenward <bkenward@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The simple_strtol function is obsolete.
This patch replace it by kstrtoint.
This will simplify code, since some error case not handled by
simple_strtol are handled by kstrtoint.
Signed-off-by: LABBE Corentin <clabbe.montjoie@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Andrew Lunn says:

====================
Allow BATMAN to use hdlc-eth interfaces

BATMAN works over Ethernet like interfaces. hdlc-eth provides the need
requirements. However, hdlc devices are often created as raw hdlc
devices, which batman cannot use, and are then be transmuted into
other types using sethdlc(1). Have the HDLC code emit
NETDEV_*_TYPE_CHANGE events when the type changes, and have BATMAN
react on these events.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

A network interface can change type. It may change from a type which
batman does not support, e.g. hdlc, to one it does, e.g. hdlc-eth.
When an interface changes type, it sends two notifications. Handle
these notifications.
Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>

An interface changing type may not have IPv6 addresses. Don't
call the address configuration type change in this case.
Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>

An HDLC device can change type when the protocol driver is changed.
Calling the notifier change allows potential users of the interface
know about this planned change, and even block it. After the change
has occurred, send a second notification to users can evaluate the new
device type etc.
Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>

The current code first unregisters the device, and then detaches the
protocol from it. This should be performed the other way around, since
the detach may try to use state which has been freed by the
unregister. Swap the order, so that we first detach and then remove the
netdev.
Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>

Bjørn Mork says:

====================
net: qmi_wwan: MDM9x30 support

We add new device IDs all the time, often without any testing on
actual hardware. This is usually OK as long as the device is similar
to already supported devices, using the same chipset and firmware
basis.  But the Sierra Wireless MC7455 is an example of a new chipset
generation. Adding it based on assumed similarity with its ancestors
proved too optimistic.

This series adds the missing bits and pieces necessary to support LTE
Advanced modems based on the Qualcomm MDM9x30 chipset. A big thanks to
Sierra Wireless for providing MC7455 samples for testing

The most important change is the "raw-ip" support. The series also
adds a necessary control request, removes an unsupported device ID,
and adds a driver specific entry in MAINTAINERS.

A few random notes about "raw-ip":

"I rather have these all running in raw IP mode. The 802.3 framing is
utterly stupid." - Marcel Holtmann in Jan 2012 [1]

Marcel was right.  I should have listened to him. What more can I say?

The 802.3 framing has provided a steady supply of firmware bugs for
many years. We've added driver workarounds for many of these, but
there are still known bugs where the workaround is so yucky that we
have refused to apply it. But all that is over now.  The latest
generation Qualcomm chips no longer supports 802.3 framing at all.

I had two open questions regarding the "raw-ip" userspace API:

1) Should we continue faking an ethernet device, even if we don't use
   the L2 headers on the USB link anymore?

   There was a vote in favour of the "headerless" device. This is the
   honest representation of the hardware/firmware interface.

2) What input should the driver base its framing on?

   Snooping or directly manipulating QMI is considered out of the
   question. We delegated all QMI handling to userspace from the
   beginning.

   We have so far required userspace to configure the firmware for
   "802.3" framing, or fail if that proved impossible.  This
   requirement is now changed.  Userspace must now inform the driver
   if it negotiates "raw-ip" framing.  Two alternative interfaces were
   proposed:
    - ethtool private driver flag, or
    - sysfs file

   The NetworkManager/ModemManager developers were in favour of the
   sysfs alternative.

These questions (or any other you migh have :) are of course still
open.  This patch set presents the solutions I currently prefer,
considering the above.

All comments are appreciated, even simple '+1' ones.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: David S. Miller <davem@davemloft.net>

Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: David S. Miller <davem@davemloft.net>

QMI wwan devices have traditionally emulated ethernet devices
by default. But they have always had the capability of operating
without any L2 header at all, transmitting and receiving "raw"
IP packets over the USB link.  This firmware feature used to be
configurable through the QMI management protocol.

Traditionally there was no way to verify the firmware mode
without attempting to change it.  And the firmware would often
disallow changes anyway, i.e. due to a session already being
established.  In some cases, this could be a hidden firmware
internal session, completely outside host control.  For these
reasons, sticking with the "well known" default mode was safest.

But newer generations of QMI hardware and firmware have moved
towards defaulting to "raw IP" mode instead, followed by an
increasing number of bugs in the already buggy "802.3" firmware
implementation. At the same time, the QMI management protocol
gained the ability to detect the current mode.  This has enabled
the userspace QMI management application to verify the current
firmware mode without trying to modify it.

Following this development, the latest QMI hardware and firmware
(the MDM9x30 generation) has dropped support for "802.3" mode
entirely. Support for "raw IP" framing in the driver is therefore
necessary for these devices, and to a certain degree to work
around problems with the previous generation,

This patch adds support for "raw IP" framing for QMI devices,
changing the netdev from an ethernet device to an ARPHRD_NONE
p-t-p device when "raw IP" framing is enabled.

The firmware setup is fully delegated to the QMI userspace
management application, through simple tunneling of the QMI
protocol. The driver will therefore not know which mode has been
"negotiated" between firmware and userspace. Allowing userspace
to inform the driver of the result through a sysfs switch is
considered a better alternative than to change the well established
clean delegation of firmware management to userspace.
Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: David S. Miller <davem@davemloft.net>

Assume the minidriver has taken care of all L2 header parsing
if it sets skb->protocol.  This allows the minidriver to
support non-ethernet L2 headers, and even operate without
any L2 header at all.
Signed-off-by: Bjørn Mork <bjorn@mork.no>
Acked-by: Oliver Neukum <oneukum@suse.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

This turned out to be a bootloader device ID.  No need for
that in this driver.  It will only provide a single serial
function.
Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: David S. Miller <davem@davemloft.net>

MDM9x30 based modems appear to go into a deeper sleep when
suspended without "Remote Wakeup" enabled.  The QMI interface
will not respond unless a "set DTR" control request is sent
on resume. The effect is similar to a QMI_CTL SYNC request,
resetting (some of) the firmware state.

We allow userspace sessions to span multiple character device
open/close sequences.  This means that userspace can depend
on firmware state while both the netdev and the character
device are closed.  We have disabled "needs_remote_wakeup" at
this point to allow devices without remote wakeup support to
be auto-suspended.

To make sure the MDM9x30 keeps firmware state, we need to
keep "needs_remote_wakeup" always set. We also need to
issue a "set DTR" request to enable the QMI interface.
Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: David S. Miller <davem@davemloft.net>

Salil Mehta says:

====================
net:hns: Add support of Hip06 SoC to the Hislicon Network Subsystem

This PATCH V7 addresses the TAB formatting comments by
Sergei Shtylyov. Missing TABs at some other palces have
also been corrected.

PATCH V6:
This addresses the review comments provided by
David Miller over the existing use of ENABLE/DISABLE
hash defines with the code. These hash defines are doing
a similar job as implicit type bool would do. So these are
kind of duplicate and are redundant.

PATCH V5:
This PATCH addresses the review comments by Yuval Mintz
 <Yuval.Mintz@qlogic.com>. This rework of comments are basically
 related to:
 1) styling of the code,
 2) RSS default Key initiailization related code
 3) redundant code removal

PATCH V4:
This addresses the review comment provided by
Sergei Shtylyov. The changelog of every patch has also
been modified.

PATCH V3:
 Addresses the review comment floated by David Miller

PATCH V2:
1) Bug Fixes and Clean-up: Internally identified
2) Addresses internal review comments by Kenneth Lee and
   by Huang Daode
3) Addresses the review comment from "Yisen.Zhuang(Zhuangyuzeng)"
4) Adds fix from Fengguang Wu for an error generated from
   "kbuild test robot" from Intel
5) Ethtool support for TSO set option from Lisheng

PATCH V1:
Adds initial support of Hip06 SoC with below changes:
This patch-set adds support of new Hisilicon Hip06 SoC to the existing
(already part of net-next) HNS ethernet driver for Hip05 SoC. Hip06 is
a multi-core SoC and is a derivative of Hip05 SoC with lots of new
hardware featres supported like RSS, TSO, hardware VLAN assist etc.

The changes in the driver are mainly due to following:
 1) changes in the DMA descriptor provided by the Hip06 ethernet
    hardware. These changes need to co-exist with already present
    Hip05 DMA descriptor and its operating functions. The decision
    to choose the correct type of DMA descriptor is taken dynamically
    depending upon the version of the hardware (i.e. V1/hip05 or
    V2/hip06, see already existing hisilicon-hns-nic.txt binding file
    for the detailed description version and naming).
 2) To support new features added to the Hip06 ethernet hardware:
    a. RSS (Receive Side Scaling)
    b. TSO (TCP Segment Offload)
    c. Hardware VLAN support (currently we are initializing hardware
       to not assist in stripping the vlan tag at hardware level.
       Proper support of this feature and ethtool would come after
       these patches have been accepted)

Kindly note that, this patchset has been based on latest net-next.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

This patch adds the initializzation code to disable the hardware
vlan support for VLAN Tag stripping by default for now.

Proper support of "hardware VLAN assitance" feature would
soon come in the next coming patches.
Signed-off-by: Salil Mehta <salil.mehta@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

This patch adds the support of ethtool TSO option to support
Hip06 SoC to HNS
Signed-off-by: Salil Mehta <salil.mehta@huawei.com>
Signed-off-by: lisheng <lisheng011@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

This patch adds the support of "TSO (TCP Segment Offload)" feature
provided by the Hip06 ethernet hardware to the HNS ethernet
driver.

Enabling this feature would help offload the TCP Segmentation
process to the Hip06 ethernet hardware. This eventually would help
in saving precious cpu cycles.
Signed-off-by: Salil Mehta <salil.mehta@huawei.com>
Signed-off-by: lisheng <lisheng011@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

This patch adds the support of "RSS (Receive Side Scaling)" feature
provided by the Hip06 ethernet hardware to the HNS ethernet
driver.

This feature helps in distributing the different flows (mapped as
hash by hardware using Toeplitz Hash) to different Queues asssociated
with the processor cores. The mapping of flow-hash values to the
different queues is stored in indirection table (which is per Packet-
parse-Engine/PPE). This patch also provides the changes to re-program
the (flow-hash<->Qid) mapping using the ethtool.
Signed-off-by: Salil Mehta <salil.mehta@huawei.com>
Reviewed-by: Kenneth Lee <liguozhu@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>