This is a small supplement for commit e7428e95
("virtio-net: put virtio-net header inline with data"). TCP packages have
enough room to put virtio-net header in, but UDP packages do not. By
setting dev->needed_headroom for virtio-net device, UDP packages could have
enough room.

For UDP packages, sk_buff is alloced in fun __ip_append_data. The size is
"alloclen + hh_len + 15", and "hh_len = LL_RESERVED_SPACE(rt-dst.dev);".
The Macro is defined as follows:
#define LL_RESERVED_SPACE(dev) \
     ((((dev)->hard_header_len+(dev)->needed_headroom)\
     &~(HH_DATA_MOD - 1)) + HH_DATA_MOD)
By default, for UDP packages, after skb is allocated, only 16 bytes
reserved. And 2 bytes remained after mac header is set. That is not enough
to put virtio-net header in. If we set dev->needed_headroom to 12 or 10
(according to mergeable_rx_bufs is on or off ), more room can be reserved.
Then there is enough room for UDP packages to put the header in.

test result list as below:
guest and host: suse11sp3, netperf, intel 2.4GHz
+-------+---------+---------+---------+---------+
|       |   old             |   new             |
+-------+---------+---------+---------+---------+
| UDP   |  Gbit/s | pps     |  Gbit/s | pps     |
| 64    |  0.57   | 692232  |  0.61   | 742420  |
| 256   |  1.60   | 686860  |  1.71   | 733331  |
| 512   |  2.92   | 674576  |  3.07   | 710446  |
| 1024  |  4.99   | 598977  |  5.17   | 620821  |
| 1460  |  5.68   | 483757  |  7.16   | 610519  |
| 4096  |  6.98   | 637468  |  7.21   | 658471  |
+-------+---------+---------+---------+---------+
Signed-off-by: Zhang Jie <zhangjie14@huawei.com>
Acked-by: Rusty Russell <rusty@rustcorp.com.au>
Acked-by: Jason Wang <jasowang@redhat.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Commit 1bb8dce5 ("tipc: fix memory
leak during module removal") introduced a memory leak issue: when
name table is stopped, it's forgotten that publication instances are
freed properly. Additionally the useless "continue" statement in
tipc_nametbl_stop() is removed as well.
Reported-by: Jason <huzhijiang@gmail.com>
Signed-off-by: Ying Xue <ying.xue@windriver.com>
Acked-by: Erik Hugne <erik.hugne@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

This replaces 6 identical code snippets with a call to a new
static inline function.
Signed-off-by: Lorenzo Colitti <lorenzo@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Use the helper function to retrieve the driver private context instead of
using (void *)(ds + 1).
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

DSA drivers have a trick which consists in allocating "priv_size" more
bytes to account for the DSA driver private context. Add a helper
function to access that private context instead of open-coding it in
drivers with (void *)(ds + 1).
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Jeff Kirsher says:

====================
Intel Wired LAN Driver Updates

This series contains updates to i40e and i40evf.

Jacob provides a i40e patch to remove ptp_tx_work() because it conflicts
with the interrupt-based handler and will cause spurious warning messages
to be displayed in the kernel log.  Since the hardware can properly trigger
an interrupt, we do not need to poll for an available timestamp in a work
queue any more.

Greg provides two patches for i40e, first to retain the MAC filter list
when changing a port VLAN.  Second fixes an issue where the function to
set the VF MAC address was not taking the port VLAN filter into account
when setting/clearing/resetting the VF's host administered MAC address.

Mitch provides three patches for i40evf, first adds support for getting
and setting RSS hashing options, RSS LUT entries, and getting the number
of device channels from ethtool.  Fixes a panic that woudl occur in the VF
if the PF driver failed or was removed from the host kernel.

Shannon adds driver version string to the driver version command so that
can be passed down through the firmware to low-level NC-SI functions.

Neerav provides several patches, first redistributes queue vectors after
DCB reconfiguration because the total number of queues that are configured
for a VSI may change based on a change in the total number of TCs enabled
as a result of a change in the DCB configuration based on DCBX.  Second,
allows the driver probe to continue when querying DCB config fails.
Lastly, provides changes to retrieve and store missing link config
information that was not being cached in the struct i40e_link_status.

Kevin provides a couple of patches, first to change a int variable type
to u16 to avoid a typecheck failure.  Second to update a function
parameters to remove an argument which is no longer used.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

Commit a89778d8 ("tipc: add support
for link state subscriptions") introduced below possible deadlock
scenario:

       CPU0                          CPU1
T0:   tipc_publish()                 link_timeout()
T1:   tipc_nametbl_publish()         [grab node lock]*
T2:   [grab nametbl write lock]*     link_state_event()
T3:   named_cluster_distribute()     link_activate()
T4:   [grab node lock]*              tipc_node_link_up()
T5:                                  tipc_nametbl_publish()
T6:                                  [grab nametble write lock]*

The opposite order of holding nametbl write lock and node lock on
above two different paths may result in a deadlock. If we move the
the delivery of named messages via link out of name nametbl lock,
the reverse order of holding locks will be eliminated, as a result,
the deadlock will be killed as well.
Signed-off-by: Ying Xue <ying.xue@windriver.com>
Reviewed-by: Erik Hugne <erik.hugne@ericsson.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Commit 78acb1f9 ("tipc: add
ioctl to fetch link names") introduced a buffer overflow bug where
specially crafted ioctl requests could cause out-of-bounds indexing
of the node->links array. This was caused by an incorrect check vs
MAX_BEARERS, and the static code checker complaint is:
net/tipc/node.c:459 tipc_node_get_linkname() error: buffer overflow 'node->links' 2 <= 2
Signed-off-by: Erik Hugne <erik.hugne@ericsson.com>
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

PTR_RET is deprecated. Do not recommend its usage anymore.
Use PTR_ERR_OR_ZERO instead.
Signed-off-by: Frank Blaschka <frank.blaschka@de.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

PTR_RET is deprecated. Do not recommend its usage anymore.
Use PTR_ERR_OR_ZERO instead.
Signed-off-by: Frank Blaschka <frank.blaschka@de.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The existing options for priority queueing within OSA devices were based on
the now partially defunct TOS field. This patch adds two new options as
follows:
   - prio_queueing_skb: bases priority queueing on skb-priority
   - prio_queueing_vlan: uses the priority code point in the 802.1Q header
Signed-off-by: Stefan Raspl <raspl@linux.vnet.ibm.com>
Signed-off-by: Frank Blaschka <frank.blaschka@de.ibm.com>
Reviewed-by: Ursula Braun <ursula.braun@de.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Make the current priority queueing logic apply to IPv6 traffic.
Signed-off-by: Stefan Raspl <raspl@linux.vnet.ibm.com>
Signed-off-by: Frank Blaschka <frank.blaschka@de.ibm.com>
Reviewed-by: Ursula Braun <ursula.braun@de.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

PTR_RET is deprecated. Do not recommend its usage anymore.
Use PTR_ERR_OR_ZERO instead.
Signed-off-by: Duan Jiong <duanj.fnst@cn.fujitsu.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: Frank Blaschka <frank.blaschka@de.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

PTR_RET is deprecated. Do not recommend its usage anymore.
Use PTR_ERR_OR_ZERO instead.
Signed-off-by: Duan Jiong <duanj.fnst@cn.fujitsu.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: Frank Blaschka <frank.blaschka@de.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The default queue for queueing modes prio_queueing_prec and
prio_queueing_tos is supposed to be QETH_DEFAULT_QUEUE. However, neither
mode will reset the default default_out_queue value when enabled, leaving
a previously set value activate as the default.
E.g. enabling one of the fixed queueing modes, e.g. no_prio_queueing:0,
and successively switching to any of the prio_queueing_* modes will
leave default_out_queue at the previous (wrong) value 0.
Signed-off-by: Stefan Raspl <raspl@linux.vnet.ibm.com>
Signed-off-by: Frank Blaschka <frank.blaschka@de.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

PTP Hardware Clock drivers specify a maximum frequency adjustment that
their clocks can accommodate. Normally, user space programs will want to
respect the advertised limits. However, no kernel or driver code checks
that the dialed frequency offset is within the bounds, and out of range
values can lead to surprising results.

This patch fixes the issue by rejecting bad values.
Signed-off-by: Richard Cochran <richardcochran@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Signed-off-by: Hisao Tanabe <xtanabe@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The commit 3de0b592 ("ethtool: Support for configurable RSS hash
key") introduced a new function ethtool_copy_validate_indir() with
full iteration of the loop to validate the ring indices, which could
be an overkill. To minimize the impact, we ought to exit the loop as
soon as the invalid index occurs for the very first time. The
remaining loop simply doesn't serve any more purpose.
Signed-off-by: Jean Sacren <sakiwit@gmail.com>
Cc: Venkata Duvvuru <VenkatKumar.Duvvuru@Emulex.Com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Bump i40e to 0.3.46 and i40evf to 0.9.23.

Change-ID: Ia604ae6d513d9aaa8bfdac79665d9a3a72507df7
Signed-off-by: Catherine Sullivan <catherine.sullivan@intel.com>
Tested-by: Kavindya Deegala <kavindya.s.deegala@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

Some information returned via "Get Link Status" command was not
being cached in the struct i40e_link_status.
Add this so the driver can utilize this information as needed.

Change-ID: If084a0ae2a63b16b97572196b993742af2c67772
Signed-off-by: Neerav Parikh <neerav.parikh@intel.com>
Acked-by: Greg Rose <gregory.v.rose@intel.com>
Tested-by: Kavindya Deegala <kavindya.s.deegala@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

Change the formal parameters to remove argument which is no longer used.

Change-ID: I493b56eba74a1f396fa1b7d24c0a4acbe536b5bf
Signed-off-by: Kevin Scott <kevin.c.scott@intel.com>
Acked-by: Shannon Nelson <shannon.nelson@intel.com>
Tested-by: Kavindya Deegala <kavindya.s.deegala@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

Ethtool -k/-K can handle this so we do not need it in debugfs.

Change-ID: I9df692f10a60b71805f8f48d3b87c7da3820b2aa
Signed-off-by: Anjali Singhai Jain <anjali.singhai@intel.com>
Signed-off-by: Catherine Sullivan <catherine.sullivan@intel.com>
Tested-by: Kavindya Deegala <kavindya.s.deegala@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

Although NTUPLE feature control is disabled by default, do not
allow user to be able to change it.

Change-ID: I4c3f95fb6dbc56b1b6cc47116ea1f72de02bc99c
Signed-off-by: Anjali Singhai Jain <anjali.singhai@intel.com>
Signed-off-by: Catherine Sullivan <catherine.sullivan@intel.com>
Tested-by: Kavindya Deegala <kavindya.s.deegala@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

Change the variable type to avoid compiler warning about int to u16
possible data truncation.

Change-ID: I5eb3b578c86513c9625ca32d2f0b57cc01d7dc98
Signed-off-by: Kevin Scott <kevin.c.scott@intel.com>
Acked-by: Greg Rose <gregory.v.rose@intel.com>
Acked-by: Shannon Nelson <shannon.nelson@intel.com>
Tested-by: Kavindya Deegala <kavindya.s.deegala@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

If the PF driver fails or is removed from the host, the VF driver will
fill up its log with this message.

Change-ID: I67045f987f7c0d444d21ded403adc509343cdb8f
Signed-off-by: Mitch Williams <mitch.a.williams@intel.com>
Signed-off-by: Catherine Sullivan <catherine.sullivan@intel.com>
Tested-by: Sibai Li <sibai.li@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

Fix a panic that would occur in the VF if the PF driver failed or was
removed from the host kernel. In this case, the VF driver calls
i40evf_close(), but this function does nothing because the driver is in
the resetting state. Because of this, the driver doesn't free its irqs
and causes a kernel panic when it tries to disable MSI-X.

Change-ID: If95644a89e554b4d7be0dca1b6add26f63047129
Signed-off-by: Mitch Williams <mitch.a.williams@intel.com>
Signed-off-by: Catherine Sullivan <catherine.sullivan@intel.com>
Tested-by: Sibai Li <sibai.li@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

In case of any AQ command to query port's DCB configuration fails
during driver's probe time; the probe fails and returns an error.

This patch prevents this issue by continuing the driver probe even
when an error is returned.

Also, added an error message to dump the AQ error status to show what
error caused the failure to get the DCB configuration from firmware.

Change-ID: Ifd5663512588bca684069bb7d4fb586dd72221af
Signed-off-by: Neerav Parikh <Neerav.Parikh@intel.com>
Signed-off-by: Catherine Sullivan <catherine.sullivan@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

The queue vectors are originally distributed among the total
number of queues that are configured for a given VSI. The number
of queues that are configured for a VSI depends on the number of
TCs that are enabled for the VSI at VSI setup/creation time.

But, the total number of queues that are configured for a VSI may
change based on change in the total number of TCs enabled as a
result of change in DCB configuration based on DCBX.

Hence, it is required for the queue vectors to be redistributed
when the total queues that are configured for a given VSI changes.

Without redistributing the vectors the queues that are configured
for the newly enabled TCs may not be able to do any Tx/Rx.

Change-ID: I4e780903db019e6199c7ce43627cef22c916acfe
Signed-off-by: Neerav Parikh <Neerav.Parikh@intel.com>
Signed-off-by: Catherine Sullivan <catherine.sullivan@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

The driver version string was added to this struct to be passed
down through the firmware to low-level NC-SI functions. We tell
the firmware about the length of the ASCII string not counting
any terminating null.

Change-ID: I09ac98ff9b869e8661c55fc6a5c98808fc280c91
Signed-off-by: Shannon Nelson <shannon.nelson@intel.com>
Acked-by: Anjali Singhai <anjali.singhai@intel.com>
Tested-by: Kavindya Deegala <kavindya.s.deegala@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

The function to set the VF MAC address was not taking the port VLAN
filter into account when setting/clearing/resetting the VF's host
administered MAC address.  Be sure to use the port VLAN for VF MAC
filtering configurations.

Change-ID: I12595331981c79529738f041dcdbdb667ef8f5e6
Signed-off-by: Greg Rose <gregory.v.rose@intel.com>
Signed-off-by: Catherine Sullivan <catherine.sullivan@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

Add support for getting and setting RSS hashing options, RSS LUT
entries, and getting the number of device channels from ethtool. Because
the LUT is so small for the VFs, we just read and write the device
registers directly instead of maintaining a shadow copy.
Signed-off-by: Mitch Williams <mitch.a.williams@intel.com>
Tested-by: Sibai Li <sibai.li@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

We were already taking care to retain the MAC filter list when deleting a
port VLAN.  Take some additional care to retain the MAC filter list when
changing a port VLAN.

Change-ID: Iacf9599ea24ecb4dca8e419aacaf4b58ca361a9c
Signed-off-by: Greg Rose <gregory.v.rose@intel.com>
Signed-off-by: Catherine Sullivan <catherine.sullivan@intel.com>
Tested-by: Sibai Li <sibai.li@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

This patch removes the unnecessary ptp_tx_work item. It conflicts with
the interrupt-based handler, and will cause spurious warning messages to
be displayed in the kernel log. Since the hardware can properly trigger
an interrupt, we do not need to poll for an available timestamp in a
work queue any more.
Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Acked-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Tested-by: Kavindya Deegala <kavindya.s.deegala@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>

When we change the list of action on a given filter, currently we don't
change it to empty. This is a bug, we should allow to change to whatever
users given.

Cc: Jamal Hadi Salim <jhs@mojatatu.com>
Cc: David S. Miller <davem@davemloft.net>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: Cong Wang <cwang@twopensource.com>
Signed-off-by: Jamal Hadi Salim <jhs@mojatatu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

When actions are attached to a filter, they are a part of the filter
itself, so when changing a filter we should allow to overwrite the actions
inside as well.

In my specific case, when I tried to _append_ a new action to an existing
filter which already has an action, I got EEXIST since kernel refused
to overwrite the existing one in kernel.

This patch checks if we are changing the filter checking NLM_F_CREATE flag
(Sigh, filters don't use NLM_F_REPLACE...) and then passes the boolean down
to actions. This fixes the problem above.

Cc: Jamal Hadi Salim <jhs@mojatatu.com>
Cc: David S. Miller <davem@davemloft.net>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: Cong Wang <cwang@twopensource.com>
Signed-off-by: Jamal Hadi Salim <jhs@mojatatu.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Jitendra Kalsaria says:

====================
qlcnic driver updates

This patch series containes following changes -
* Optimize MAC learning code.
* Export board temperature using hwmon-sysfs interface.
* Add support for configuring Tx interrupt coalescing parameters from VF.
* Logging error messages in error path.
* Restrict extended vNIC support in legacy interrupt mode and log appropriate
  message.

Please apply this series to net-next.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

Signed-off-by: Jitendra Kalsaria <jitendra.kalsaria@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

o When the driver loads in legacy interrupt mode, only vNICs
  with PCI function number 0-7 are supported.
Signed-off-by: Sucheta Chakraborty <sucheta.chakraborty@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Signed-off-by: Sucheta Chakraborty <sucheta.chakraborty@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

o Tx interrupt coalescing parameters can now be set from VF.
o Added validation code in PF to validate the parameters.
Signed-off-by: Sucheta Chakraborty <sucheta.chakraborty@qlogic.com>
Signed-off-by: David S. Miller <davem@davemloft.net>