openwrt/staging/blogic.git
7 years agor8152: move the default coalesce setting for RTL8153
hayeswang [Fri, 9 Jun 2017 09:11:44 +0000 (17:11 +0800)]
r8152: move the default coalesce setting for RTL8153

Only RTL8153 could set coalesce, so move the default setting for
rtl8152_probe() to r8153_init().

Signed-off-by: Hayes Wang <hayeswang@realtek.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agor8152: adjust U2P3 for RTL8153
hayeswang [Fri, 9 Jun 2017 09:11:43 +0000 (17:11 +0800)]
r8152: adjust U2P3 for RTL8153

Use another way to keep disabling the U2P3 for both RTL_VER_03 and
RTL_VER_04.

Move enabling U2P3 from r8153_init() to r8153_hw_phy_cfg(). The
engineer ask the setting should be done after PHY settings.

Disable U2P3 first in rtl8153_up().

Signed-off-by: Hayes Wang <hayeswang@realtek.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agor8152: adjust rtl8153_runtime_enable function
hayeswang [Fri, 9 Jun 2017 09:11:42 +0000 (17:11 +0800)]
r8152: adjust rtl8153_runtime_enable function

Adjust the order of rtl8153_runtime_enable() according to the
suggestion from the engineer.

Signed-off-by: Hayes Wang <hayeswang@realtek.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agor8152: move the setting of rx aggregation
hayeswang [Fri, 9 Jun 2017 09:11:41 +0000 (17:11 +0800)]
r8152: move the setting of rx aggregation

Move the setting from r8153_first_init() to r8153_init(). It only needs to
be set once.

Signed-off-by: Hayes Wang <hayeswang@realtek.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agor8152: adjust the settings about MAC clock speed down for RTL8153
hayeswang [Fri, 9 Jun 2017 09:11:40 +0000 (17:11 +0800)]
r8152: adjust the settings about MAC clock speed down for RTL8153

The MAC clock speed down could be enabled if the U1/U2 is disabled.

Signed-off-by: Hayes Wang <hayeswang@realtek.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agor8152: adjust lpm settings for RTL8153
hayeswang [Fri, 9 Jun 2017 09:11:39 +0000 (17:11 +0800)]
r8152: adjust lpm settings for RTL8153

Enable lpm after r8153_init() and remove other enable/disable lpm.

Signed-off-by: Hayes Wang <hayeswang@realtek.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agor8152: add r8153_phy_status function
hayeswang [Fri, 9 Jun 2017 09:11:38 +0000 (17:11 +0800)]
r8152: add r8153_phy_status function

Use r8153_phy_status() to check phy status of RTL8153.

Signed-off-by: Hayes Wang <hayeswang@realtek.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agobpf: Fix test_obj_id.c for llvm 5.0
Martin KaFai Lau [Fri, 9 Jun 2017 05:30:17 +0000 (22:30 -0700)]
bpf: Fix test_obj_id.c for llvm 5.0

llvm 5.0 does not like the section name and the function name
to be the same:

clang -I. -I./include/uapi -I../../../include/uapi \
-I../../../../samples/bpf/ \
-Wno-compare-distinct-pointer-types \
-O2 -target bpf -c \
linux/tools/testing/selftests/bpf/test_obj_id.c -o \
linux/tools/testing/selftests/bpf/test_obj_id.o
fatal error: error in backend: 'test_prog_id' label emitted multiple times to
assembly file
clang-5.0: error: clang frontend command failed with exit code 70 (use -v to
see invocation)
clang version 5.0.0 (trunk 304326) (llvm/trunk 304329)

This patch makes changes to the section name and the function name.

Fixes: 95b9afd3987f ("bpf: Test for bpf ID")
Reported-by: Alexei Starovoitov <ast@fb.com>
Reported-by: Yonghong Song <yhs@fb.com>
Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Yonghong Song <yhs@fb.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agobpf: Fix test_bpf_obj_id() when the bpf_jit_enable sysctl is diabled
Martin KaFai Lau [Fri, 9 Jun 2017 05:30:16 +0000 (22:30 -0700)]
bpf: Fix test_bpf_obj_id() when the bpf_jit_enable sysctl is diabled

test_bpf_obj_id() should not expect a non zero jited_prog_len
to be returned by bpf_obj_get_info_by_fd() when
net.core.bpf_jit_enable is 0.

The patch checks for net.core.bpf_jit_enable and
has different expectation on jited_prog_len.

This patch also removes the pwd.h header which I forgot
to remove after making changes.

Fixes: 95b9afd3987f ("bpf: Test for bpf ID")
Reported-by: Yonghong Song <yhs@fb.com>
Signed-off-by: Martin KaFai Lau <kafai@fb.com>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Yonghong Song <yhs@fb.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoipv6: Initial skb->dev and skb->protocol in ip6_output
Chenbo Feng [Fri, 9 Jun 2017 19:06:07 +0000 (12:06 -0700)]
ipv6: Initial skb->dev and skb->protocol in ip6_output

Move the initialization of skb->dev and skb->protocol from
ip6_finish_output2 to ip6_output. This can make the skb->dev and
skb->protocol information avalaible to the CGROUP eBPF filter.

Signed-off-by: Chenbo Feng <fengc@google.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agocxgb4: handle interrupt raised when FW crashes
Rahul Lakkireddy [Fri, 9 Jun 2017 05:42:35 +0000 (11:12 +0530)]
cxgb4: handle interrupt raised when FW crashes

Handle TIMER0INT when FW crashes. Check for PCIE_FW[FW_EVAL]
and if it says "Device FW Crashed", then treat it as fatal.
Else, non-fatal.

Signed-off-by: Rahul Lakkireddy <rahul.lakkireddy@chelsio.com>
Signed-off-by: Ganesh Goudar <ganeshgr@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoMerge branch 'nfp-FW-app-build-name-reporting'
David S. Miller [Fri, 9 Jun 2017 16:52:09 +0000 (12:52 -0400)]
Merge branch 'nfp-FW-app-build-name-reporting'

Jakub Kicinski says:

====================
nfp: FW app build name reporting

This series adds reporting FW build name in ethtool -i.  Most
of the patches are restructuring where information caching is
done.  There is also a minor error path fix.

These are last few patches finishing the basic nfp_app support.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonfp: report application FW build name in ethtool -i
Jakub Kicinski [Fri, 9 Jun 2017 03:56:14 +0000 (20:56 -0700)]
nfp: report application FW build name in ethtool -i

Make sure application FW build name is NULL-terminated and
print it as a part of ethtool's firmware version string.

Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonfp: keep MIP object around
Jakub Kicinski [Fri, 9 Jun 2017 03:56:13 +0000 (20:56 -0700)]
nfp: keep MIP object around

Microcode Information Page contains some useful information, like
application firmware build name.  Keep it around, similar to RTSym
and HWInfo.

Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonfp: remove automatic caching of HWInfo
Jakub Kicinski [Fri, 9 Jun 2017 03:56:12 +0000 (20:56 -0700)]
nfp: remove automatic caching of HWInfo

Make callers take care of managing life time of HWInfo.

Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonfp: remove automatic caching of RTsym table
Jakub Kicinski [Fri, 9 Jun 2017 03:56:11 +0000 (20:56 -0700)]
nfp: remove automatic caching of RTsym table

The fact that RTsym table is cached inside nfp_cpp handle is
a relic of old times when nfpcore was a library module.  All
the nfp_cpp "caches" are awkward to deal with because of
concurrency and prone to keeping stale information.  Make
the run time symbol table be an object read out from the device
and managed by whoever requested it.  Since the driver loads
FW at ->probe() and never reloads, we can hold onto the table
for ever.

Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonfp: make sure to cancel port refresh on the error path
Jakub Kicinski [Fri, 9 Jun 2017 03:56:10 +0000 (20:56 -0700)]
nfp: make sure to cancel port refresh on the error path

If very last stages of netdev registering and init fail some
other netdevs and devlink ports may have been visible to user
space before we torn them back down.  In this case there is a
slight chance user may have triggered port refresh.  We need
to make sure the async work is cancelled.

We have to cancel after releasing pf->lock, so we will always
try to cancel, regardless of which part of probe has failed.

Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoliquidio: disallow enabling firmware debug from a VF
Derek Chickles [Fri, 9 Jun 2017 02:20:36 +0000 (19:20 -0700)]
liquidio: disallow enabling firmware debug from a VF

Disallow enabling firmware debug from a VF.  Only PF is allowed to do that.

Signed-off-by: Derek Chickles <derek.chickles@cavium.com>
Signed-off-by: Felix Manlunas <felix.manlunas@cavium.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agogeneve: add missing rx stats accounting
Girish Moodalbail [Fri, 9 Jun 2017 00:07:48 +0000 (17:07 -0700)]
geneve: add missing rx stats accounting

There are few places on the receive path where packet drops and packet
errors were not accounted for. This patch fixes that issue.

Signed-off-by: Girish Moodalbail <girish.moodalbail@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoIpvlan should return an error when an address is already in use.
Krister Johansen [Thu, 8 Jun 2017 20:12:14 +0000 (13:12 -0700)]
Ipvlan should return an error when an address is already in use.

The ipvlan code already knows how to detect when a duplicate address is
about to be assigned to an ipvlan device.  However, that failure is not
propogated outward and leads to a silent failure.

Introduce a validation step at ip address creation time and allow device
drivers to register to validate the incoming ip addresses.  The ipvlan
code is the first consumer.  If it detects an address in use, we can
return an error to the user before beginning to commit the new ifa in
the networking code.

This can be especially useful if it is necessary to provision many
ipvlans in containers.  The provisioning software (or operator) can use
this to detect situations where an ip address is unexpectedly in use.

Signed-off-by: Krister Johansen <kjlx@templeofstupid.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoMerge branch 'netvsc-small-cleanups'
David S. Miller [Fri, 9 Jun 2017 16:15:06 +0000 (12:15 -0400)]
Merge branch 'netvsc-small-cleanups'

Stephen Hemminger says:

====================
netvsc: small cleanups

These are all small optimizations found during development of later features.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonetvsc: fold in get_outbound_net_device
stephen hemminger [Thu, 8 Jun 2017 23:21:23 +0000 (16:21 -0700)]
netvsc: fold in get_outbound_net_device

No longer need common code to find get_outbound_net_device.

Signed-off-by: Stephen Hemminger <sthemmin@microsoft.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonetvsc: pass net_device to netvsc_init_buf and netvsc_connect_vsp
stephen hemminger [Thu, 8 Jun 2017 23:21:22 +0000 (16:21 -0700)]
netvsc: pass net_device to netvsc_init_buf and netvsc_connect_vsp

Don't need to find netvsc_device structure, caller already had it.
Also rearrange declarations.

Signed-off-by: Stephen Hemminger <sthemmin@microsoft.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonetvsc: mark error cases as unlikely
stephen hemminger [Thu, 8 Jun 2017 23:21:21 +0000 (16:21 -0700)]
netvsc: mark error cases as unlikely

Mark if() statements used for error handling only as unlikely()

Signed-off-by: Stephen Hemminger <sthemmin@microsoft.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonetvsc: use typed pointer for internal state
stephen hemminger [Thu, 8 Jun 2017 23:21:20 +0000 (16:21 -0700)]
netvsc: use typed pointer for internal state

The element netvsc_device:extension is always a pointer to RNDIS
information.

Signed-off-by: Stephen Hemminger <sthemmin@microsoft.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonetvsc: use hv_get_bytes_to_read
stephen hemminger [Thu, 8 Jun 2017 23:21:19 +0000 (16:21 -0700)]
netvsc: use hv_get_bytes_to_read

Don't need need to look at write space in netvsc_close.

Signed-off-by: Stephen Hemminger <sthemmin@microsoft.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonetvsc: optimize calculation of number of slots
stephen hemminger [Thu, 8 Jun 2017 23:21:18 +0000 (16:21 -0700)]
netvsc: optimize calculation of number of slots

Speed up transmit check for fragmented packets by using existing
macros to compute number of pages, and eliminate loop since
skb fragments each take a page. Number of slots is also unsigned.

Signed-off-by: Stephen Hemminger <sthemmin@microsoft.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoMerge branch 'dsa-mv88e6xxx-ops-cosmetics'
David S. Miller [Fri, 9 Jun 2017 16:11:00 +0000 (12:11 -0400)]
Merge branch 'dsa-mv88e6xxx-ops-cosmetics'

Vivien Didelot says:

====================
net: dsa: mv88e6xxx: ops cosmetics

This patchset brings no functional changes. It is a first step in a
bigger cosmetics change to the driver. It simplifies print messages and
polishes data types and chip operations.

The next patchs will only prefix and document the port registers macros.

Changes in v2:
  - KISS and simply use dev_* since chip->ds may not be initialized
  - add reviewers tags
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: dsa: mv88e6xxx: prefix PHY macros
Vivien Didelot [Thu, 8 Jun 2017 22:34:14 +0000 (18:34 -0400)]
net: dsa: mv88e6xxx: prefix PHY macros

Prefix the PHY_* macros with a Marvell specific MV88E6XXX_ prefix.
There is no functional changes.

Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Vivien Didelot <vivien.didelot@savoirfairelinux.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: dsa: mv88e6xxx: rework jumbo size operation
Vivien Didelot [Thu, 8 Jun 2017 22:34:13 +0000 (18:34 -0400)]
net: dsa: mv88e6xxx: rework jumbo size operation

Marvell chips have a Jumbo Mode to set the maximum frame size (MTU).

The mv88e6xxx_ops structure is meant to contain generic functionalities,
no driver logic. Change port_jumbo_config to port_set_jumbo_size setting
the mode from a given maximum size value.

There is no functional changes since we still use 10240 bytes.

At the same time, correctly clear all Jumbo Mode bits before writing.

Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Vivien Didelot <vivien.didelot@savoirfairelinux.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: dsa: mv88e6xxx: rework pause limit operation
Vivien Didelot [Thu, 8 Jun 2017 22:34:12 +0000 (18:34 -0400)]
net: dsa: mv88e6xxx: rework pause limit operation

All Marvell chips supporting Pause frames limiting use 1-byte value for
input and output.

Old chips have both bytes adjacent in a 16-bit register. New ones have
an indirect table using 8-bit data.

The mv88e6xxx library functions (such as in port.c) must not contain
driver logic, but only generic helpers. This patch changes the
port_pause_config operation for port_pause_limit taking two u8 arguments
for input and output limits. There is no functional changes.

Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Vivien Didelot <vivien.didelot@savoirfairelinux.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: dsa: mv88e6xxx: do not prefix ops with g1
Vivien Didelot [Thu, 8 Jun 2017 22:34:11 +0000 (18:34 -0400)]
net: dsa: mv88e6xxx: do not prefix ops with g1

The mv88e6xxx_ops describe functionalities, regardless their locations
(which can be Global1, Global2, or whatever register set.)

Rename the g1_set_cpu_port and g1_set_egress_port ops to set_cpu_port
and set_egress_port. No functional changes.

Signed-off-by: Vivien Didelot <vivien.didelot@savoirfairelinux.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: dsa: mv88e6xxx: use bridge state values
Vivien Didelot [Thu, 8 Jun 2017 22:34:10 +0000 (18:34 -0400)]
net: dsa: mv88e6xxx: use bridge state values

Reuse the BR_STATE_* values to abstract a port STP state value.

This provides shorter names and better control over the DSA switch
operation call.

Signed-off-by: Vivien Didelot <vivien.didelot@savoirfairelinux.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: dsa: mv88e6xxx: add egress mode enumeration
Vivien Didelot [Thu, 8 Jun 2017 22:34:09 +0000 (18:34 -0400)]
net: dsa: mv88e6xxx: add egress mode enumeration

As for the frame mode, add a mv88e6xxx_egress_mode enumeration instead
of a 16-bit register mask.

Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Vivien Didelot <vivien.didelot@savoirfairelinux.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: dsa: mv888e6xxx: do not use netdev printing
Vivien Didelot [Thu, 8 Jun 2017 22:34:08 +0000 (18:34 -0400)]
net: dsa: mv888e6xxx: do not use netdev printing

The mv888e6xxx driver accesses a port's netdev mostly for printing.

This is bad for 2 reasons: DSA and CPU ports do not have a netdev
pointer; it doesn't give us a correct picture of why a DSA driver might
need to access a port's netdev.

Instead simply use dev_* printing functions with chip->dev (or ds->dev
depending on the scope, both guaranteed to exist), with a p%d prefix for
the target port.

Signed-off-by: Vivien Didelot <vivien.didelot@savoirfairelinux.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoskbuff: only inherit relevant tx_flags
Willem de Bruijn [Thu, 8 Jun 2017 15:35:03 +0000 (11:35 -0400)]
skbuff: only inherit relevant tx_flags

When inheriting tx_flags from one skbuff to another, always apply a
mask to avoid overwriting unrelated other bits in the field.

The two SKBTX_SHARED_FRAG cases clears all other bits. In practice,
tx_flags are zero at this point now. But this is fragile. Timestamp
flags are set, for instance, if in tcp_gso_segment, after this clear
in skb_segment.

The SKBTX_ANY_TSTAMP mask in __skb_tstamp_tx ensures that new
skbs do not accidentally inherit flags such as SKBTX_SHARED_FRAG.

Signed-off-by: Willem de Bruijn <willemb@google.com>
Acked-by: Soheil Hassas Yeganeh <soheil@google.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoMerge branch 'net-add-const-to-mii_phy_ops-structures'
David S. Miller [Thu, 8 Jun 2017 19:32:48 +0000 (15:32 -0400)]
Merge branch 'net-add-const-to-mii_phy_ops-structures'

Bhumika Goyal says:

====================
drivers: net: add const to mii_phy_ops structures

The object references of mii_phy_ops structures are only stored
in the ops field of a mii_phy_def structure. This ops field is of type
const. So, mii_phy_ops structures having similar properties can be
declared as const.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agodrivers: net: emac: add const to mii_phy_ops structures
Bhumika Goyal [Thu, 8 Jun 2017 06:00:58 +0000 (11:30 +0530)]
drivers: net: emac: add const to mii_phy_ops structures

The object references of mii_phy_ops structures are only stored
in the ops field of a mii_phy_def structure. This ops field is of type
const. So, mii_phy_ops structures having similar properties can be
declared as const.

Signed-off-by: Bhumika Goyal <bhumirks@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agodrivers/net/sungem: add const to mii_phy_ops structures
Bhumika Goyal [Thu, 8 Jun 2017 06:00:57 +0000 (11:30 +0530)]
drivers/net/sungem: add const to mii_phy_ops structures

The object references of mii_phy_ops structures are only stored
in the ops field of a mii_phy_def structure. This ops field is of type
const. So, mii_phy_ops structures having similar properties can be
declared as const.

Signed-off-by: Bhumika Goyal <bhumirks@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoMerge branch '1GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/jkirsher/next...
David S. Miller [Thu, 8 Jun 2017 18:41:19 +0000 (14:41 -0400)]
Merge branch '1GbE' of git://git./linux/kernel/git/jkirsher/next-queue

Jeff Kirsher says:

====================
1GbE Intel Wired LAN Driver Updates 2017-06-07

This series contains a fix for e1000e and igb.

Colin Ian King fixes sparse warnings in igb by making functions static.

Chris Wilson provides a fix for a previous commit which is causing an
issue during suspend "e1000e_pm_suspend()", where we need to run
e1000e_pm_thaw() if __e1000_shutdown() is unsuccessful.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agocxgb4: fix to bring link down after adapter crash
Rahul Lakkireddy [Thu, 8 Jun 2017 05:22:11 +0000 (10:52 +0530)]
cxgb4: fix to bring link down after adapter crash

Use PORT_REG for T4 and T5_PORT_REG for > T4 to write to correct
register to bring down link during shutdown after adapter crash.

Signed-off-by: Rahul Lakkireddy <rahul.lakkireddy@chelsio.com>
Signed-off-by: Ganesh Goudar <ganeshgr@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: ipmr: add getlink support
Nikolay Aleksandrov [Wed, 7 Jun 2017 15:02:32 +0000 (18:02 +0300)]
net: ipmr: add getlink support

Currently there's no way to dump the VIF table for an ipmr table other
than the default (via proc). This is a major issue when debugging ipmr
issues and in general it is good to know which interfaces are
configured. This patch adds support for RTM_GETLINK for the ipmr family
so we can dump the VIF table and the ipmr table's current config for
each table. We're protected by rtnl so no need to acquire RCU or
mrt_lock.

Signed-off-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoMerge branch 'mlxsw-Remove-compatibility-with-old-firmware'
David S. Miller [Thu, 8 Jun 2017 18:33:42 +0000 (14:33 -0400)]
Merge branch 'mlxsw-Remove-compatibility-with-old-firmware'

Jiri Pirko says:

====================
mlxsw: Remove compatibility with old firmware

Up until recently we couldn't enforce a minimal firmware version, which
forced us to be compatible with old firmware versions. This patchset
removes this code and simplifies the driver.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agomlxsw: spectrum: Pass port argument to module mapping functions
Ido Schimmel [Thu, 8 Jun 2017 06:47:45 +0000 (08:47 +0200)]
mlxsw: spectrum: Pass port argument to module mapping functions

Previous patch made it unnecessary to map ports to modules before we
allocate their struct. We can now therefore pass the port struct to
these functions, thereby making them consistent with other functions
that operate on ports.

Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agomlxsw: spectrum: Simplify port split flow
Ido Schimmel [Thu, 8 Jun 2017 06:47:44 +0000 (08:47 +0200)]
mlxsw: spectrum: Simplify port split flow

In commit be94535f9531 ("mlxsw: spectrum: Make split flow match firmware
requirements") we had to modify the port split flow to overcome quirks
in the device's firmware. This resulted in asymmetrical code with
regards to port creation and removal.

The problem in the firmware is long gone and since we can now enforce a
minimal firmware version, we can simplify the code and make it symmetric
again.

Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agomlxsw: spectrum_router: Mark only first LPM tree as reserved
Ido Schimmel [Thu, 8 Jun 2017 06:47:43 +0000 (08:47 +0200)]
mlxsw: spectrum_router: Mark only first LPM tree as reserved

In new firmware versions (that we can now enforce via
request_firmware()), only the first LPM tree is reserved and not the
first two as in older versions.

Signed-off-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoMerge branch 'net-Remove-support-from-bridge-bypass-for-mlxsw-rocker-drivers'
David S. Miller [Thu, 8 Jun 2017 18:16:31 +0000 (14:16 -0400)]
Merge branch 'net-Remove-support-from-bridge-bypass-for-mlxsw-rocker-drivers'

Jiri Pirko says:

===================
net: Remove support from bridge bypass for mlxsw/rocker drivers

Currently setting bridge port attributes and adding FDBs are done via
setting the SELF flag which implies unconsistent offloading model. This
patch-set fixes this behavior by making the bridge and drivers which are
using it to be totally in sync.

This implies several changes:
- Offloading bridge flags from the bridge code.
- Sending notification about FDB add/del to the software bridge in a
  similiar way it is done for the hardware externally learned FDBs.

By making the offloading model more consistent a cleanup is done in
the drivers supporting it. This is done in order to remove un-needed
logic related to dump operation which is redundant.

First add missing functionality to bridge, then clean up the mlxsw/rocker
drivers.

v1->v2
- Move bridge-switchdev related stuff to br_switchdev.c as suggested by Nik
===================

Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agorocker: Remove support bridge bypass FDB
Arkadi Sharshevsky [Thu, 8 Jun 2017 06:44:28 +0000 (08:44 +0200)]
rocker: Remove support bridge bypass FDB

The FDB add/delete are now done through the notification chain. The FDBs
are synced with the bridge and there is no need for extra dumping.

Signed-off-by: Arkadi Sharshevsky <arkadis@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agorocker: Remove support for bypass bridge port attributes/vlan set
Arkadi Sharshevsky [Thu, 8 Jun 2017 06:44:27 +0000 (08:44 +0200)]
rocker: Remove support for bypass bridge port attributes/vlan set

The bridge port attributes/vlan for mlxsw devices should be set only
from bridge code. The vlans are synced totally with the bridge so
there is no need to special dump support.

Signed-off-by: Arkadi Sharshevsky <arkadis@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agorocker: Add support for learning FDB through notification
Arkadi Sharshevsky [Thu, 8 Jun 2017 06:44:26 +0000 (08:44 +0200)]
rocker: Add support for learning FDB through notification

Add support for learning FDB through notification. The driver defers
the hardware update via ordered work queue.

Signed-off-by: Arkadi Sharshevsky <arkadis@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agorocker: Change world_ops API and implementation to be switchdev independant
Arkadi Sharshevsky [Thu, 8 Jun 2017 06:44:25 +0000 (08:44 +0200)]
rocker: Change world_ops API and implementation to be switchdev independant

Currently the switchdev_trans struct is embedded in the world_ops API.
In order to add support for adding FDB via a notfication chain the API should
be switchdev independent.

Signed-off-by: Arkadi Sharshevsky <arkadis@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agorocker: Add support for querying supported bridge flags
Arkadi Sharshevsky [Thu, 8 Jun 2017 06:44:24 +0000 (08:44 +0200)]
rocker: Add support for querying supported bridge flags

Add support for querying supported bridge flags.

Signed-off-by: Arkadi Sharshevsky <arkadis@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agorocker: Remove support for bridge FDB learning sync
Arkadi Sharshevsky [Thu, 8 Jun 2017 06:44:23 +0000 (08:44 +0200)]
rocker: Remove support for bridge FDB learning sync

Currently the rocker driver supports an option for disabling syncing
the hardware learned FDBs with the software bridge. This behavior
breaks the bridge offload model and thus it is removed.

Signed-off-by: Arkadi Sharshevsky <arkadis@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: Remove support for bridge bypass ndos from stacked devices
Arkadi Sharshevsky [Thu, 8 Jun 2017 06:44:22 +0000 (08:44 +0200)]
net: Remove support for bridge bypass ndos from stacked devices

Remove support for bridge bypass ndos from stacked devices. At this point
no driver which supports stack device behavior offload supports operation
with SELF flag. The case for upper device is already taken care of in both
of the following cases:

1. FDB add/del - driver should check at the notification cb if the
                 stacked device contains his ports.

2. Port attribute - calls switchdev code directly which checks
                    for case of stack device.

Signed-off-by: Arkadi Sharshevsky <arkadis@mellanox.com>
Reviewed-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agomlxsw: spectrum: Remove support for bridge bypass FDB add/del
Arkadi Sharshevsky [Thu, 8 Jun 2017 06:44:21 +0000 (08:44 +0200)]
mlxsw: spectrum: Remove support for bridge bypass FDB add/del

The FDB add/del are now done through the notification chain. The FDBs
are synced with the bridge and there is no need for extra dumping.

Signed-off-by: Arkadi Sharshevsky <arkadis@mellanox.com>
Reviewed-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agomlxsw: spectrum_switchdev: Add support for learning FDB through notification
Arkadi Sharshevsky [Thu, 8 Jun 2017 06:44:20 +0000 (08:44 +0200)]
mlxsw: spectrum_switchdev: Add support for learning FDB through notification

Add support for learning FDB through notification. The driver defers
the hardware update via ordered work queue. Support for stacked devices
is also provided. In case of a successful FDB add a notification is
sent back to bridge.

Signed-off-by: Arkadi Sharshevsky <arkadis@mellanox.com>
Reviewed-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agomlxsw: spectrum_switchdev: Change switchdev notifier API
Arkadi Sharshevsky [Thu, 8 Jun 2017 06:44:19 +0000 (08:44 +0200)]
mlxsw: spectrum_switchdev: Change switchdev notifier API

The current API for sending switchdev notifications implies only FDB
add/del. In order to support notification about successful FDB offload
the API is changed.

Signed-off-by: Arkadi Sharshevsky <arkadis@mellanox.com>
Reviewed-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agomlxsw: spectrum: Remove support for bypass bridge port attributes/vlan set
Arkadi Sharshevsky [Thu, 8 Jun 2017 06:44:18 +0000 (08:44 +0200)]
mlxsw: spectrum: Remove support for bypass bridge port attributes/vlan set

The bridge port attributes/vlan for mlxsw devices should be set only
from bridge code. The vlans are synced totally with the bridge so
there is no need to special dump support.

Signed-off-by: Arkadi Sharshevsky <arkadis@mellanox.com>
Reviewed-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agomlxsw: spectrum_switchdev: Add support for querying supported bridge flags
Arkadi Sharshevsky [Thu, 8 Jun 2017 06:44:17 +0000 (08:44 +0200)]
mlxsw: spectrum_switchdev: Add support for querying supported bridge flags

Add support for querying supported bridge flags.

Signed-off-by: Arkadi Sharshevsky <arkadis@mellanox.com>
Reviewed-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agomlxsw: spectrum: Remove support for bridge FDB learning sync
Arkadi Sharshevsky [Thu, 8 Jun 2017 06:44:16 +0000 (08:44 +0200)]
mlxsw: spectrum: Remove support for bridge FDB learning sync

Currently the mlxsw driver supports an option for disabling syncing
the hardware learned FDBs with the software bridge. This behavior
breaks the bridge offload model and thus it is removed.

Signed-off-by: Arkadi Sharshevsky <arkadis@mellanox.com>
Reviewed-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: bridge: Receive notification about successful FDB offload
Arkadi Sharshevsky [Thu, 8 Jun 2017 06:44:15 +0000 (08:44 +0200)]
net: bridge: Receive notification about successful FDB offload

When a new static FDB is added to the bridge a notification is sent to
the driver for offload. In case of successful offload the driver should
notify the bridge back, which in turn should mark the FDB as offloaded.

Currently, externally learned is equivalent for being offloaded which is
not correct due to the fact that FDBs which are added from user-space are
also marked as externally learned. In order to specify if an FDB was
successfully offloaded a new flag is introduced.

Signed-off-by: Arkadi Sharshevsky <arkadis@mellanox.com>
Reviewed-by: Ido Schimmel <idosch@mellanox.com>
Reviewed-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: bridge: Add support for notifying devices about FDB add/del
Arkadi Sharshevsky [Thu, 8 Jun 2017 06:44:14 +0000 (08:44 +0200)]
net: bridge: Add support for notifying devices about FDB add/del

Currently the bridge doesn't notify the underlying devices about new
FDBs learned. The FDB sync is placed on the switchdev notifier chain
because devices may potentially learn FDB that are not directly related
to their ports, for example:

1. Mixed SW/HW bridge - FDBs that point to the ASICs external devices
                        should be offloaded as CPU traps in order to
perform forwarding in slow path.
2. EVPN - Externally learned FDBs for the vtep device.

Notification is sent only about static FDB add/del. This is done due
to fact that currently this is the only scenario supported by switch
drivers.

Signed-off-by: Arkadi Sharshevsky <arkadis@mellanox.com>
Reviewed-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Reviewed-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
Reviewed-by: Ivan Vecera <ivecera@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: switchdev: Change notifier chain to be atomic
Arkadi Sharshevsky [Thu, 8 Jun 2017 06:44:13 +0000 (08:44 +0200)]
net: switchdev: Change notifier chain to be atomic

In order to use the switchdev notifier chain for FDB sync with the
device it has to be changed to atomic. The is done because the bridge
can learn new FDBs in atomic context.

Signed-off-by: Arkadi Sharshevsky <arkadis@mellanox.com>
Reviewed-by: Ido Schimmel <idosch@mellanox.com>
Reviewed-by: Ivan Vecera <ivecera@redhat.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Reviewed-by: Ivan Vecera <ivecera@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: bridge: Add support for calling FDB external learning under rcu
Arkadi Sharshevsky [Thu, 8 Jun 2017 06:44:12 +0000 (08:44 +0200)]
net: bridge: Add support for calling FDB external learning under rcu

This is done as a preparation to moving the switchdev notifier chain
to be atomic. The FDB external learning should be called under rtnl
or rcu.

Signed-off-by: Arkadi Sharshevsky <arkadis@mellanox.com>
Reviewed-by: Ido Schimmel <idosch@mellanox.com>
Reviewed-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: bridge: Add support for offloading port attributes
Arkadi Sharshevsky [Thu, 8 Jun 2017 06:44:11 +0000 (08:44 +0200)]
net: bridge: Add support for offloading port attributes

Currently the flood, learning and learning_sync port attributes are
offloaded by setting the SELF flag. Add support for offloading the
flood and learning attribute through the bridge code. In case of
setting an unsupported flag on a offloded port the operation will
fail.

The learning_sync attribute doesn't have any software representation
and cannot be offloaded through the bridge code.

Signed-off-by: Arkadi Sharshevsky <arkadis@mellanox.com>
Reviewed-by: Ido Schimmel <idosch@mellanox.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Reviewed-by: Nikolay Aleksandrov <nikolay@cumulusnetworks.com>
Reviewed-by: Ivan Vecera <ivecera@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: switchdev: Add support for querying supported bridge flags by hardware
Arkadi Sharshevsky [Thu, 8 Jun 2017 06:44:10 +0000 (08:44 +0200)]
net: switchdev: Add support for querying supported bridge flags by hardware

This is done as a preparation stage before setting the bridge port flags
from the bridge code. Currently the device can be queried for the bridge
flags state, but the querier cannot distinguish if the flag is disabled
or if it is not supported at all. Thus, add new attr and a bit-mask which
include information regarding the support on a per-flag basis.

Drivers that support bridge offload but not support bridge flags should
return zeroed bitmask.

Signed-off-by: Arkadi Sharshevsky <arkadis@mellanox.com>
Reviewed-by: Ido Schimmel <idosch@mellanox.com>
Reviewed-by: Ivan Vecera <ivecera@redhat.com>
Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Reviewed-by: Ivan Vecera <ivecera@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoMerge branch 'dsa-add-cross-chip-VLAN-support'
David S. Miller [Thu, 8 Jun 2017 15:43:33 +0000 (11:43 -0400)]
Merge branch 'dsa-add-cross-chip-VLAN-support'

Vivien Didelot says:

====================
net: dsa: add cross-chip VLAN support

The current code in DSA does not support cross-chip VLAN. This means
that in a multi-chip environment such as this one (similar to ZII Rev B)

         [CPU].................... (mdio)
    (eth0) |   :       :          :
          _|_____    _______    _______
         [__sw0__]--[__sw1__]--[__sw2__]
          |  |  |    |  |  |    |  |  |
          v  v  v    v  v  v    v  v  v
          p1 p2 p3   p4 p5 p6   p7 p8 p9

adding a VLAN to p9 won't be enough to reach the CPU, until at least one
port of sw0 and sw1 join the VLAN as well and become aware of the VID.

This patchset makes the DSA core program the VLAN on the CPU and DSA
links itself, which brings seamlessly cross-chip VLAN support to DSA.

With this series applied*, the hardware VLAN tables of a 3-switch setup
look like this after adding a VLAN to only one port of the end switch:

    # cat /sys/class/net/br0/bridge/default_pvid
    42
    # cat /sys/kernel/debug/mv88e6xxx/sw{0,1,2}/vtu
    # ip link set up master br0 dev lan6
    # cat /sys/kernel/debug/mv88e6xxx/sw{0,1,2}/vtu
     VID  FID  SID  0  1  2  3  4  5  6
      42    1    0  x  x  x  x  x  =  =
     VID  FID  SID  0  1  2  3  4  5  6
      42    1    0  x  x  x  x  x  =  =
     VID  FID  SID  0  1  2  3  4  5  6  7  8  9
      42    1    0  u  x  x  x  x  x  x  x  x  =

('x' is excluded, 'u' is untagged, '=' is unmodified DSA and CPU ports.)

Completely removing a VLAN entry (which is currently the responsibility
of drivers anyway) is not supported yet since it requires some caching.

(*) the output is shown from this out-of-tree debugfs patch:
https://github.com/vivien/linux/commit/7b61a684b9d6b6a499135a587c7f62a1fddceb8b.patch

Changes in v2:
  - canonical incrementation (port++ instead of ++port)
  - check CPU and DSA ports before purging a VLAN
  - add Reviewed-by tags
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: dsa: mv88e6xxx: do not skip ports on VLAN del
Vivien Didelot [Wed, 7 Jun 2017 22:12:17 +0000 (18:12 -0400)]
net: dsa: mv88e6xxx: do not skip ports on VLAN del

The mv88e6xxx driver currently tries to be smart and remove by itself a
VLAN entry from the VTU when the driven switch sees no user ports as
members of the VLAN.

This is bad in a multi-chip switch fabric, since a chip in between
others may have no bridge port members, but still needs to be aware of
the VID in order to correctly pass frames in the data path.

Now that the DSA core explicitly manages DSA and CPU ports, do not skip
them when checking remaining VLAN members.

Signed-off-by: Vivien Didelot <vivien.didelot@savoirfairelinux.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: dsa: mv88e6xxx: exclude all ports in new VLAN
Vivien Didelot [Wed, 7 Jun 2017 22:12:16 +0000 (18:12 -0400)]
net: dsa: mv88e6xxx: exclude all ports in new VLAN

Now that the DSA core adds the CPU and DSA ports itself to the new VLAN
entry, there is no need to include them as members of this VLAN when
initializing a new VTU entry.

As of now, initialize a new VTU entry with all ports excluded.

Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Vivien Didelot <vivien.didelot@savoirfairelinux.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: dsa: add CPU and DSA ports as VLAN members
Vivien Didelot [Wed, 7 Jun 2017 22:12:15 +0000 (18:12 -0400)]
net: dsa: add CPU and DSA ports as VLAN members

In a multi-chip switch fabric, it is currently the responsibility of the
driver to add the CPU or DSA (interconnecting chips together) ports as
members of a new VLAN entry. This makes the drivers more complicated.

We want the DSA drivers to be stupid and the DSA core being the one
responsible for caring about the abstracted switch logic and topology.

Make the DSA core program the CPU and DSA ports as part of the VLAN.

This makes all chips of the data path to be aware of VIDs spanning the
the whole fabric and thus, seamlessly add support for cross-chip VLAN.

Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Vivien Didelot <vivien.didelot@savoirfairelinux.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: dsa: check VLAN capability of every switch
Vivien Didelot [Wed, 7 Jun 2017 22:12:14 +0000 (18:12 -0400)]
net: dsa: check VLAN capability of every switch

Now that the VLAN object is propagated to every switch chip of the
switch fabric, we can easily ensure that they all support the required
VLAN operations before modifying an entry on a single switch.

To achieve that, remove the condition skipping other target switches,
and add a bitmap of VLAN members, eventually containing the target port,
if we are programming the switch target.

This will allow us to easily add other VLAN members, such as the DSA or
CPU ports (to introduce cross-chip VLAN support) or the other port
members if we want to reduce hardware accesses later.

Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Vivien Didelot <vivien.didelot@savoirfairelinux.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: dsa: mv88e6xxx: define membership on VLAN add
Vivien Didelot [Wed, 7 Jun 2017 22:12:13 +0000 (18:12 -0400)]
net: dsa: mv88e6xxx: define membership on VLAN add

Define the target port membership of the VLAN entry in
mv88e6xxx_port_vlan_add where ds is scoped.

Allow the DSA core to call later the port_vlan_add operation for CPU or
DSA ports, by using the Unmodified membership for these ports, as in the
current behavior.

Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Vivien Didelot <vivien.didelot@savoirfairelinux.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoMerge tag 'rxrpc-rewrite-20170607-v2' of git://git.kernel.org/pub/scm/linux/kernel...
David S. Miller [Thu, 8 Jun 2017 15:41:41 +0000 (11:41 -0400)]
Merge tag 'rxrpc-rewrite-20170607-v2' of git://git./linux/kernel/git/dhowells/linux-fs

David Howells says:

====================
rxrpc: Tx length parameter

Here's a set of patches that allows someone initiating a client call with
AF_RXRPC to indicate upfront the total amount of data that will be
transmitted.  This will allow AF_RXRPC to encrypt directly from source
buffer to packet rather than having to copy into the buffer and only
encrypt when it's full (the encrypted portion of the packet starts with a
length and so we can't encrypt until we know what the length will be).

The three patches are:

 (1) Provide a means of finding out what control message types are actually
     supported.  EINVAL is reported if an unsupported cmsg type is seen, so
     we don't want to set the new cmsg unless we know it will be accepted.

 (2) Consolidate some stuff into a struct to reduce the parameter count on
     the function that parses the cmsg buffer.

 (3) Introduce the RXRPC_TX_LENGTH cmsg.  This can be provided on the first
     sendmsg() that contributes data to a client call request or a service
     call reply.  If provided, the user must provide exactly that amount of
     data or an error will be incurred.

Changes in version 2:

 (*) struct rxrpc_send_params::tx_total_len should be s64 not u64.  Thanks to
     Julia Lawall for reporting this.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoMerge branch 'qrtr-features'
David S. Miller [Thu, 8 Jun 2017 15:34:58 +0000 (11:34 -0400)]
Merge branch 'qrtr-features'

Bjorn Andersson says:

====================
Missing QRTR features

The QMUX specification covers packet routing as well as service life cycle and
discovery. The current implementation of qrtr supports the prior part, but in
order to fully implement service management on-top a few more parts are needed.

The first patch in the series serves the purpose of reducing duplication in
patch two and three.

The second and third patch adds two qrtr-level notifications required by the
specification, in order to notify local and remote service controllers about
dying clients.

The last patch serves the purpose of notifying local clients about the presence
of a local service register, allowing them to register services as well as
querying for remote registered services.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: qrtr: Inform open sockets about new controller
Bjorn Andersson [Wed, 7 Jun 2017 21:07:39 +0000 (14:07 -0700)]
net: qrtr: Inform open sockets about new controller

As the higher level communication only deals with "services" the
a service directory is required to keep track of local and remote
services. In order for qrtr clients to be informed about when the
service directory implementation is available some event needs to be
passed to them.

Rather than introducing support for broadcasting such a message in-band
to all open local sockets we flag each socket with ENETRESET, as there
are no other expected operations that would benefit from having support
from locally broadcasting messages.

Cc: Courtney Cavin <ccavin@gmail.com>
Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: qrtr: Broadcast DEL_CLIENT message when endpoint is closed
Bjorn Andersson [Wed, 7 Jun 2017 21:07:38 +0000 (14:07 -0700)]
net: qrtr: Broadcast DEL_CLIENT message when endpoint is closed

Per the QMUXv2 protocol specificiation a DEL_CLIENT message should be
broadcasted when an endpoint is disconnected.

The protocol specification does suggest that the router can keep track
of which nodes the endpoint has been communicating with to not wake up
sleeping remotes unecessarily, but implementation of this suggestion is
left for the future.

Cc: Courtney Cavin <ccavin@gmail.com>
Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: qrtr: Inject BYE on remote termination
Bjorn Andersson [Wed, 7 Jun 2017 21:07:37 +0000 (14:07 -0700)]
net: qrtr: Inject BYE on remote termination

Per the QMUX protocol specification a terminating node can send a BYE
control message to signal that the link is going down, upon receiving
this all information about remote services should be discarded and local
clients should be notified.

In the event that the link was brought down abruptly the router is
supposed to act like a BYE message has arrived. As there is no harm in
receiving an extra BYE from the remote this patch implements the latter
by injecting a BYE when the link to the remote is unregistered.

The name service will receive the BYE and can implement the notification
to the local clients.

Cc: Courtney Cavin <ccavin@gmail.com>
Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: qrtr: Refactor packet allocation
Bjorn Andersson [Wed, 7 Jun 2017 21:07:36 +0000 (14:07 -0700)]
net: qrtr: Refactor packet allocation

Extract the allocation and filling in the control message header fields
to a separate function in order to reuse this in subsequent patches.

Cc: Courtney Cavin <ccavin@gmail.com>
Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agomISDN: remove unnecessary variable assignments
Gustavo A. R. Silva [Wed, 7 Jun 2017 20:15:01 +0000 (15:15 -0500)]
mISDN: remove unnecessary variable assignments

Remove unnecessary variable assignments.

Addresses-Coverity-ID: 1226917
Signed-off-by: Gustavo A. R. Silva <garsilva@embeddedor.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agotcp: add TCPMemoryPressuresChrono counter
Eric Dumazet [Wed, 7 Jun 2017 20:29:12 +0000 (13:29 -0700)]
tcp: add TCPMemoryPressuresChrono counter

DRAM supply shortage and poor memory pressure tracking in TCP
stack makes any change in SO_SNDBUF/SO_RCVBUF (or equivalent autotuning
limits) and tcp_mem[] quite hazardous.

TCPMemoryPressures SNMP counter is an indication of tcp_mem sysctl
limits being hit, but only tracking number of transitions.

If TCP stack behavior under stress was perfect :
1) It would maintain memory usage close to the limit.
2) Memory pressure state would be entered for short times.

We certainly prefer 100 events lasting 10ms compared to one event
lasting 200 seconds.

This patch adds a new SNMP counter tracking cumulative duration of
memory pressure events, given in ms units.

$ cat /proc/sys/net/ipv4/tcp_mem
3088    4117    6176
$ grep TCP /proc/net/sockstat
TCP: inuse 180 orphan 0 tw 2 alloc 234 mem 4140
$ nstat -n ; sleep 10 ; nstat |grep Pressure
TcpExtTCPMemoryPressures        1700
TcpExtTCPMemoryPressuresChrono  5209

v2: Used EXPORT_SYMBOL_GPL() instead of EXPORT_SYMBOL() as David
instructed.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoMerge branch 'tcp-Namespaceify-3-sysctls'
David S. Miller [Thu, 8 Jun 2017 14:53:30 +0000 (10:53 -0400)]
Merge branch 'tcp-Namespaceify-3-sysctls'

Eric Dumazet says:

====================
tcp: Namespaceify 3 sysctls

Move tcp_sack, tcp_window_scaling and tcp_timestamps
sysctls to network namespaces.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agotcp: Namespaceify sysctl_tcp_timestamps
Eric Dumazet [Wed, 7 Jun 2017 17:34:39 +0000 (10:34 -0700)]
tcp: Namespaceify sysctl_tcp_timestamps

Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agotcp: Namespaceify sysctl_tcp_window_scaling
Eric Dumazet [Wed, 7 Jun 2017 17:34:38 +0000 (10:34 -0700)]
tcp: Namespaceify sysctl_tcp_window_scaling

Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agotcp: Namespaceify sysctl_tcp_sack
Eric Dumazet [Wed, 7 Jun 2017 17:34:37 +0000 (10:34 -0700)]
tcp: Namespaceify sysctl_tcp_sack

Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agotcp: add a struct net parameter to tcp_parse_options()
Eric Dumazet [Wed, 7 Jun 2017 17:34:36 +0000 (10:34 -0700)]
tcp: add a struct net parameter to tcp_parse_options()

We want to move some TCP sysctls to net namespaces in the future.

tcp_window_scaling, tcp_sack and tcp_timestamps being fetched
from tcp_parse_options(), we need to pass an extra parameter.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: propagate tc filter chain index down the ndo_setup_tc call
Jiri Pirko [Tue, 6 Jun 2017 15:00:16 +0000 (17:00 +0200)]
net: propagate tc filter chain index down the ndo_setup_tc call

We need to push the chain index down to the drivers, so they have the
information to which chain the rule belongs. For now, no driver supports
multichain offload, so only chain 0 is supported. This is needed to
prevent chain squashes during offload for now. Later this will be used
to implement multichain offload.

Signed-off-by: Jiri Pirko <jiri@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoe1000e: Undo e1000e_pm_freeze if __e1000_shutdown fails
Chris Wilson [Wed, 31 May 2017 15:50:43 +0000 (18:50 +0300)]
e1000e: Undo e1000e_pm_freeze if __e1000_shutdown fails

An error during suspend (e100e_pm_suspend),

[  429.994338] ACPI : EC: event blocked
[  429.994633] e1000e: EEE TX LPI TIMER: 00000011
[  430.955451] pci_pm_suspend(): e1000e_pm_suspend+0x0/0x30 [e1000e] returns -2
[  430.955454] dpm_run_callback(): pci_pm_suspend+0x0/0x140 returns -2
[  430.955458] PM: Device 0000:00:19.0 failed to suspend async: error -2
[  430.955581] PM: Some devices failed to suspend, or early wake event detected
[  430.957709] ACPI : EC: event unblocked

lead to complete failure:

[  432.585002] ------------[ cut here ]------------
[  432.585013] WARNING: CPU: 3 PID: 8372 at kernel/irq/manage.c:1478 __free_irq+0x9f/0x280
[  432.585015] Trying to free already-free IRQ 20
[  432.585016] Modules linked in: cdc_ncm usbnet x86_pkg_temp_thermal intel_powerclamp coretemp mii crct10dif_pclmul crc32_pclmul ghash_clmulni_intel snd_hda_codec_hdmi snd_hda_codec_realtek snd_hda_codec_generic snd_hda_intel snd_hda_codec snd_hwdep lpc_ich snd_hda_core snd_pcm mei_me mei sdhci_pci sdhci i915 mmc_core e1000e ptp pps_core prime_numbers
[  432.585042] CPU: 3 PID: 8372 Comm: kworker/u16:40 Tainted: G     U          4.10.0-rc8-CI-Patchwork_3870+ #1
[  432.585044] Hardware name: LENOVO 2356GCG/2356GCG, BIOS G7ET31WW (1.13 ) 07/02/2012
[  432.585050] Workqueue: events_unbound async_run_entry_fn
[  432.585051] Call Trace:
[  432.585058]  dump_stack+0x67/0x92
[  432.585062]  __warn+0xc6/0xe0
[  432.585065]  warn_slowpath_fmt+0x4a/0x50
[  432.585070]  ? _raw_spin_lock_irqsave+0x49/0x60
[  432.585072]  __free_irq+0x9f/0x280
[  432.585075]  free_irq+0x34/0x80
[  432.585089]  e1000_free_irq+0x65/0x70 [e1000e]
[  432.585098]  e1000e_pm_freeze+0x7a/0xb0 [e1000e]
[  432.585106]  e1000e_pm_suspend+0x21/0x30 [e1000e]
[  432.585113]  pci_pm_suspend+0x71/0x140
[  432.585118]  dpm_run_callback+0x6f/0x330
[  432.585122]  ? pci_pm_freeze+0xe0/0xe0
[  432.585125]  __device_suspend+0xea/0x330
[  432.585128]  async_suspend+0x1a/0x90
[  432.585132]  async_run_entry_fn+0x34/0x160
[  432.585137]  process_one_work+0x1f4/0x6d0
[  432.585140]  ? process_one_work+0x16e/0x6d0
[  432.585143]  worker_thread+0x49/0x4a0
[  432.585145]  kthread+0x107/0x140
[  432.585148]  ? process_one_work+0x6d0/0x6d0
[  432.585150]  ? kthread_create_on_node+0x40/0x40
[  432.585154]  ret_from_fork+0x2e/0x40
[  432.585156] ---[ end trace 6712df7f8c4b9124 ]---

The unwind failures stems from commit 2800209994f8 ("e1000e: Refactor PM
flows"), but it may be a later patch that introduced the non-recoverable
behaviour.

Fixes: 2800209994f8 ("e1000e: Refactor PM flows")
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=99847
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
7 years agoigb: make a few local functions static
Colin Ian King [Thu, 27 Apr 2017 17:59:11 +0000 (18:59 +0100)]
igb: make a few local functions static

Clean up a few sparse warnings, these following
functions can be made static:

drivers/net/ethernet/intel/igb/igb_main.c: warning: symbol
  'igb_add_mac_filter' was not declared. Should it be static?
drivers/net/ethernet/intel/igb/igb_main.c: warning: symbol
  'igb_del_mac_filter' was not declared. Should it be static?
drivers/net/ethernet/intel/igb/igb_main.c: warning: symbol
  'igb_set_vf_mac_filter' was not declared. Should it be static?

Signed-off-by: Colin Ian King <colin.king@canonical.com>
Tested-by: Aaron Brown <aaron.f.brown@intel.com>
Signed-off-by: Jeff Kirsher <jeffrey.t.kirsher@intel.com>
7 years agoMerge branch 'mlx4-drivers-version-update'
David S. Miller [Wed, 7 Jun 2017 19:33:02 +0000 (15:33 -0400)]
Merge branch 'mlx4-drivers-version-update'

Tariq Toukan says:

====================
mlx4 drivers: version update

This patchset contains version updates for the MLX4 drivers:
Core, EN, and IB.

Just like we've done in mlx5, we modify the outdated driver
version (reported in ethtool for example).
This better reflects the current driver state, and removes the
redundant date string.
We are not going to change this frequently or even use it.

I include the IB patch in this series as it has similar subject
and content.
It does not cause any kind of conflict with Doug's tree.
The rdma mailing list is CCed.
Please let me know if I need to submit this differently.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoIB/mlx4: Bump driver version
Tariq Toukan [Wed, 7 Jun 2017 13:26:15 +0000 (16:26 +0300)]
IB/mlx4: Bump driver version

Remove date and bump version for mlx4_ib driver.

Signed-off-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet/mlx4_en: Bump driver version
Tariq Toukan [Wed, 7 Jun 2017 13:26:14 +0000 (16:26 +0300)]
net/mlx4_en: Bump driver version

Remove date and bump version for mlx4_en driver.

Signed-off-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet/mlx4_core: Bump driver version
Tariq Toukan [Wed, 7 Jun 2017 13:26:13 +0000 (16:26 +0300)]
net/mlx4_core: Bump driver version

Remove date and bump version for mlx4_core driver.

Signed-off-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonet: dsa: mv88e6xxx: Have 6161/6123 use EDSA tags
Andrew Lunn [Wed, 7 Jun 2017 13:06:19 +0000 (15:06 +0200)]
net: dsa: mv88e6xxx: Have 6161/6123 use EDSA tags

The mv88e6161 and mv88e6123 are capable of using EDSA tags when
passing frames from the host to the switch and back.

Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Reviewed-by: Vivien Didelot <vivien.didelot@savoirfairelinux.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agovxlan: use a more suitable function when assigning NULL
Mark Bloch [Wed, 7 Jun 2017 11:36:58 +0000 (14:36 +0300)]
vxlan: use a more suitable function when assigning NULL

When stopping the vxlan interface we detach it from the socket.
Use RCU_INIT_POINTER() and not rcu_assign_pointer() to do so.

Suggested-by: Stephen Hemminger <stephen@networkplumber.org>
Signed-off-by: Mark Bloch <markb@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agocxgb4: Fix tids count for ipv6 offload connection
Ganesh Goudar [Wed, 7 Jun 2017 09:34:51 +0000 (15:04 +0530)]
cxgb4: Fix tids count for ipv6 offload connection

the adapter consumes two tids for every ipv6 offload
connection be it active or passive, calculate tid usage
count accordingly.

Also change the signatures of relevant functions to get
the address family.

Signed-off-by: Rizwan Ansari <rizwana@chelsio.com>
Signed-off-by: Varun Prakash <varun@chelsio.com>
Signed-off-by: Ganesh Goudar <ganeshgr@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agoMerge branch 'nfp-ctrl-vNIC'
David S. Miller [Wed, 7 Jun 2017 16:51:42 +0000 (12:51 -0400)]
Merge branch 'nfp-ctrl-vNIC'

Jakub Kicinski says:

====================
nfp: ctrl vNIC

This series adds the ability to use one vNIC as a control channel
for passing messages to and from the application firmware.  The
implementation restructures the existing netdev vNIC code to be able
to deal with nfp_nets with netdev pointer set to NULL.  Control vNICs
are not visible to userspace (other than for dumping ring state), and
since they don't have netdevs we use a tasklet for RX and simple skb
list for TX queuing.

Due to special status of the control vNIC we have to reshuffle the
init code a bit to make sure control vNIC will be fully brought up
(and therefore communication with app FW can happen) before any netdev
or port is visible to user space.

FW will designate which vNIC is supposed to be used as control one
by setting _pf%u_net_ctrl_bar symbol.  Some FWs depend on metadata
being prepended to control message, some prefer to look at queue ID
to decide that something is a control message.  Our implementation
can cater to both.

First two users of this code will be eBPF maps and flower offloads.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonfp: advertise support for NFD ABI 0.5
Jakub Kicinski [Tue, 6 Jun 2017 00:01:57 +0000 (17:01 -0700)]
nfp: advertise support for NFD ABI 0.5

NFD ABI 0.5 is equivalent to NFD ABI 3.0 but requires that the
driver checks the APP id symbol and makes sure it can support
given app.  Most advanced apps will likely require control vNIC
(ability to exchange control messages between the driver and
app FW).  Detailed app version checking and capability exchange
is left to app-specific code.

Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonfp: create control vNICs and wire up rx/tx
Jakub Kicinski [Tue, 6 Jun 2017 00:01:56 +0000 (17:01 -0700)]
nfp: create control vNICs and wire up rx/tx

When driver encounters an nfp_app which has a control message handler
defined, allocate a control vNIC.  This control channel will be used
to exchange data with the application FW such as flow table programming,
statistics and global datapath control.

Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonfp: allow non-equal distribution of IRQs
Jakub Kicinski [Tue, 6 Jun 2017 00:01:55 +0000 (17:01 -0700)]
nfp: allow non-equal distribution of IRQs

Thus far the code assumed all vNICs will request similar number of IRQs.
This will be no longer true with control vNICs (where 1 IRQ will suffice).

Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
7 years agonfp: slice the netdev spawning function
Jakub Kicinski [Tue, 6 Jun 2017 00:01:54 +0000 (17:01 -0700)]
nfp: slice the netdev spawning function

We want to be able to create a special vNIC for control messages.
This vNIC should be created before any netdev is registered to allow
nfp_app logic to exchange messages with the FW app before any netdev
is visible to user space.  Unfortunately we can't enable IRQs until
we know how many vNICs we will need to spawn.

Divide the function which spawns netdevs for vNICs into three parts:
 - vNIC/memory allocation;
 - IRQ allocation;
 - netdev init and register.

This will help us insert the initialization of the control channel
after IRQ allocation but before netdev init and register.

Signed-off-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>