linux-2.6-microblaze.git
2 years agonet/mlx5: CT: Set flow source hint from provided tuple device
Paul Blakey [Mon, 3 Jan 2022 11:44:52 +0000 (13:44 +0200)]
net/mlx5: CT: Set flow source hint from provided tuple device

Get originating device from tuple offload metadata match ingress_ifindex,
and set flow_source hint to either LOCAL for vf/sf reps, UPLINK for
uplink/wire/tunnel devices/bond, or ANY (as before this patch)
for all others.

This allows lower layer (software steering or firmware) to insert the tuple
rule only in one table (either rx or tx) instead of two (rx and tx).

Signed-off-by: Paul Blakey <paulb@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: openvswitch: Fill act ct extension
Paul Blakey [Mon, 3 Jan 2022 11:44:51 +0000 (13:44 +0200)]
net: openvswitch: Fill act ct extension

To give drivers the originating device information for optimized
connection tracking offload, fill in act ct extension with
ifindex from skb.

Signed-off-by: Paul Blakey <paulb@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet/sched: act_ct: Fill offloading tuple iifidx
Paul Blakey [Mon, 3 Jan 2022 11:44:50 +0000 (13:44 +0200)]
net/sched: act_ct: Fill offloading tuple iifidx

Driver offloading ct tuples can use the information of which devices
received the packets that created the offloaded connections, to
more efficiently offload them only to the relevant device.

Add new act_ct nf conntrack extension, which is used to store the skb
devices before offloading the connection, and then fill in the tuple
iifindex so drivers can get the device via metadata dissector match.

Signed-off-by: Oz Shlomo <ozsh@nvidia.com>
Signed-off-by: Paul Blakey <paulb@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoMerge tag 'batadv-next-pullrequest-20220103' of git://git.open-mesh.org/linux-merge
Jakub Kicinski [Tue, 4 Jan 2022 03:52:15 +0000 (19:52 -0800)]
Merge tag 'batadv-next-pullrequest-20220103' of git://git.open-mesh.org/linux-merge

Simon Wunderlich says:

====================
This cleanup patchset includes the following patches:

 - bump version strings, by Simon Wunderlich
 - allow netlink usage in unprivileged containers, by Linus Lüssing
 - remove unneeded variable, by Minghao Chi

* tag 'batadv-next-pullrequest-20220103' of git://git.open-mesh.org/linux-merge:
  batman-adv: remove unneeded variable in batadv_nc_init
  batman-adv: allow netlink usage in unprivileged containers
  batman-adv: Start new development cycle
====================

Link: https://lore.kernel.org/r/20220103171722.1126109-1-sw@simonwunderlich.de
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agonet: mdio: Demote probed message to debug print
Florian Fainelli [Mon, 3 Jan 2022 19:40:24 +0000 (11:40 -0800)]
net: mdio: Demote probed message to debug print

On systems with large numbers of MDIO bus/muxes the message indicating
that a given MDIO bus has been successfully probed is repeated for as
many buses we have, which can eat up substantial boot time for no
reason, demote to a debug print.

Reported-by: Maxime Bizon <mbizon@freebox.fr>
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Link: https://lore.kernel.org/r/20220103194024.2620-1-f.fainelli@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agonet: vxge: Use dma_set_mask_and_coherent() and simplify code
Christophe JAILLET [Sun, 2 Jan 2022 21:07:05 +0000 (22:07 +0100)]
net: vxge: Use dma_set_mask_and_coherent() and simplify code

Use dma_set_mask_and_coherent() instead of unrolling it with some
dma_set_mask()+dma_set_coherent_mask().

Moreover, as stated in [1], dma_set_mask() with a 64-bit mask will never
fail if dev->dma_mask is non-NULL.
So, if it fails, the 32 bits case will also fail for the same reason.

That said, 'high_dma' can only be 1 after a successful
dma_set_mask_and_coherent().

Simplify code and remove some dead code accordingly, including a now
useless parameter to vxge_device_register().

[1]: https://lkml.org/lkml/2021/6/7/398

Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoethernet: s2io: Use dma_set_mask_and_coherent() and simplify code
Christophe JAILLET [Sun, 2 Jan 2022 20:27:39 +0000 (21:27 +0100)]
ethernet: s2io: Use dma_set_mask_and_coherent() and simplify code

Use dma_set_mask_and_coherent() instead of unrolling it with some
dma_set_mask()+dma_set_coherent_mask().

Moreover, as stated in [1], dma_set_mask() with a 64-bit mask will never
fail if dev->dma_mask is non-NULL.
So, if it fails, the 32 bits case will also fail for the same reason.

That said, 'dma_flag' can only be 'true' after a successful
dma_set_mask_and_coherent().

Simplify code and remove some dead code accordingly, including the now
useless 'high_dma_flag' field in 'struct s2io_nic'.

[1]: https://lkml.org/lkml/2021/6/7/398

Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: vertexcom: default to disabled on kbuild
Saeed Mahameed [Sun, 2 Jan 2022 22:11:26 +0000 (14:11 -0800)]
net: vertexcom: default to disabled on kbuild

Sorry for being rude but new vendors/drivers are supposed to be disabled
by default, otherwise we will have to manually keep track of all vendors
we are not interested in building.

Fixes: 2f207cbf0dd4 ("net: vertexcom: Add MSE102x SPI support")
CC: Stefan Wahren <stefan.wahren@i2se.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoMerge branch 'lynx-pcs-interface-cleanup'
David S. Miller [Sun, 2 Jan 2022 18:48:47 +0000 (18:48 +0000)]
Merge branch 'lynx-pcs-interface-cleanup'

Colin Foster says:

====================
lynx pcs interface cleanup

The current Felix driver (and Seville) rely directly on the lynx_pcs
device. There are other possible PCS interfaces that can be used with
this hardware, so this should be abstracted from felix. The generic
phylink_pcs is used instead.

While going through the code, there were some opportunities to change
some misleading variable names. Those are included in this patch set.

v1->v2
    * compile-time fixes for freescale parts
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: pcs: lynx: use a common naming scheme for all lynx_pcs variables
Colin Foster [Wed, 29 Dec 2021 05:03:10 +0000 (21:03 -0800)]
net: pcs: lynx: use a common naming scheme for all lynx_pcs variables

pcs-lynx.c used lynx_pcs and lynx as a variable name within the same file.
This standardizes all internal variables to just "lynx"

Signed-off-by: Colin Foster <colin.foster@in-advantage.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: ethernet: enetc: name change for clarity from pcs to mdio_device
Colin Foster [Wed, 29 Dec 2021 05:03:09 +0000 (21:03 -0800)]
net: ethernet: enetc: name change for clarity from pcs to mdio_device

A simple variable update from "pcs" to "mdio_device" for the mdio device
will make things a little cleaner.

Signed-off-by: Colin Foster <colin.foster@in-advantage.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: dsa: seville: name change for clarity from pcs to mdio_device
Colin Foster [Wed, 29 Dec 2021 05:03:08 +0000 (21:03 -0800)]
net: dsa: seville: name change for clarity from pcs to mdio_device

A simple variable update from "pcs" to "mdio_device" for the mdio device
will make things a little cleaner.

Signed-off-by: Colin Foster <colin.foster@in-advantage.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: dsa: felix: name change for clarity from pcs to mdio_device
Colin Foster [Wed, 29 Dec 2021 05:03:07 +0000 (21:03 -0800)]
net: dsa: felix: name change for clarity from pcs to mdio_device

Simple rename of a variable to make things more logical.

Signed-off-by: Colin Foster <colin.foster@in-advantage.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: phy: lynx: refactor Lynx PCS module to use generic phylink_pcs
Colin Foster [Wed, 29 Dec 2021 05:03:06 +0000 (21:03 -0800)]
net: phy: lynx: refactor Lynx PCS module to use generic phylink_pcs

Remove references to lynx_pcs structures so drivers like the Felix DSA
can reference alternate PCS drivers.

Signed-off-by: Colin Foster <colin.foster@in-advantage.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoenic: Remove usage of the deprecated "pci-dma-compat.h" API
Christophe JAILLET [Sun, 2 Jan 2022 10:23:39 +0000 (11:23 +0100)]
enic: Remove usage of the deprecated "pci-dma-compat.h" API

In [1], Christoph Hellwig has proposed to remove the wrappers in
include/linux/pci-dma-compat.h.

Some reasons why this API should be removed have been given by Julia
Lawall in [2].

A coccinelle script has been used to perform the needed transformation
Only relevant parts are given below.

@@
expression e1, e2;
@@
-    pci_dma_mapping_error(e1, e2)
+    dma_mapping_error(&e1->dev, e2)

[1]: https://lore.kernel.org/kernel-janitors/20200421081257.GA131897@infradead.org/
[2]: https://lore.kernel.org/kernel-janitors/alpine.DEB.2.22.394.2007120902170.2424@hadrien/

Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoqed: Use dma_set_mask_and_coherent() and simplify code
Christophe JAILLET [Sun, 2 Jan 2022 09:20:05 +0000 (10:20 +0100)]
qed: Use dma_set_mask_and_coherent() and simplify code

Use dma_set_mask_and_coherent() instead of unrolling it with some
dma_set_mask()+dma_set_coherent_mask().

Moreover, as stated in [1], dma_set_mask() with a 64-bit mask will never
fail if dev->dma_mask is non-NULL.
So, if it fails, the 32 bits case will also fail for the same reason.

Simplify code and remove some dead code accordingly.

Now that qed_set_coherency_mask() is mostly a single call to
dma_set_mask_and_coherent(), fold it in its only caller.

[1]: https://lkml.org/lkml/2021/6/7/398

Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agochelsio: cxgb: Use dma_set_mask_and_coherent() and simplify code
Christophe JAILLET [Sat, 1 Jan 2022 21:15:29 +0000 (22:15 +0100)]
chelsio: cxgb: Use dma_set_mask_and_coherent() and simplify code

Use dma_set_mask_and_coherent() instead of unrolling it with some
dma_set_mask()+dma_set_coherent_mask().

Moreover, as stated in [1], dma_set_mask() with a 64-bit mask will never
fail if dev->dma_mask is non-NULL.
So, if it fails, the 32 bits case will also fail for the same reason.

That said, 'pci_using_dac' can only be 1 after a successful
dma_set_mask_and_coherent().

Simplify code and remove some dead code accordingly.

[1]: https://lkml.org/lkml/2021/6/7/398

Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agosun/cassini: Use dma_set_mask_and_coherent() and simplify code
Christophe JAILLET [Sat, 1 Jan 2022 19:04:45 +0000 (20:04 +0100)]
sun/cassini: Use dma_set_mask_and_coherent() and simplify code

Use dma_set_mask_and_coherent() instead of unrolling it with some
dma_set_mask()+dma_set_coherent_mask().

Moreover, as stated in [1], dma_set_mask() with a 64-bit mask will never
fail if dev->dma_mask is non-NULL.
So, if it fails, the 32 bits case will also fail for the same reason.

That said, 'pci_using_dac' can only be 1 after a successful
dma_set_mask_and_coherent().

Simplify code and remove some dead code accordingly.

[1]: https://lkml.org/lkml/2021/6/7/398

Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet/smc: add comments for smc_link_{usable|sendable}
Dust Li [Fri, 31 Dec 2021 06:08:53 +0000 (14:08 +0800)]
net/smc: add comments for smc_link_{usable|sendable}

Add comments for both smc_link_sendable() and smc_link_usable()
to help better distinguish and use them.

No function changes.

Signed-off-by: Dust Li <dust.li@linux.alibaba.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agotehuti: Use dma_set_mask_and_coherent() and simplify code
Christophe JAILLET [Sat, 1 Jan 2022 14:48:25 +0000 (15:48 +0100)]
tehuti: Use dma_set_mask_and_coherent() and simplify code

Use dma_set_mask_and_coherent() instead of unrolling it with some
dma_set_mask()+dma_set_coherent_mask().

Moreover, as stated in [1], dma_set_mask_and_coherent() with a 64-bit mask
will never fail if dev->dma_mask is non-NULL.
So, if it fails, the 32 bits case will also fail for the same reason.

That said, 'pci_using_dac' can only be 1 after a successful
dma_set_mask_and_coherent().

Simplify code and remove some dead code accordingly.

[1]: https://lkml.org/lkml/2021/6/7/398
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoenic: Use dma_set_mask_and_coherent()
Christophe JAILLET [Sat, 1 Jan 2022 14:02:45 +0000 (15:02 +0100)]
enic: Use dma_set_mask_and_coherent()

Use dma_set_mask_and_coherent() instead of unrolling it with some
dma_set_mask()+dma_set_coherent_mask().

This simplifies code and removes some dead code (dma_set_coherent_mask()
can not fail after a successful dma_set_mask())

Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: socket.c: style fix
Hamish MacDonald [Sat, 1 Jan 2022 12:38:25 +0000 (22:38 +1000)]
net: socket.c: style fix

Removed spaces and added a tab that was causing an error on checkpatch

Signed-off-by: Hamish MacDonald <elusivenode@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoipv6: ioam: Support for Queue depth data field
Justin Iurman [Thu, 30 Dec 2021 17:10:04 +0000 (18:10 +0100)]
ipv6: ioam: Support for Queue depth data field

v3:
 - Report 'backlog' (bytes) instead of 'qlen' (number of packets)

v2:
 - Fix sparse warning (use rcu_dereference)

This patch adds support for the queue depth in IOAM trace data fields.

The draft [1] says the following:

   The "queue depth" field is a 4-octet unsigned integer field.  This
   field indicates the current length of the egress interface queue of
   the interface from where the packet is forwarded out.  The queue
   depth is expressed as the current amount of memory buffers used by
   the queue (a packet could consume one or more memory buffers,
   depending on its size).

An existing function (i.e., qdisc_qstats_qlen_backlog) is used to
retrieve the current queue length without reinventing the wheel.

Note: it was tested and qlen is increasing when an artificial delay is
added on the egress with tc.

  [1] https://datatracker.ietf.org/doc/html/draft-ietf-ippm-ioam-data#section-5.4.2.7

Signed-off-by: Justin Iurman <justin.iurman@uliege.be>
Reviewed-by: Ido Schimmel <idosch@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet/smc: remove redundant re-assignment of pointer link
Colin Ian King [Thu, 30 Dec 2021 15:39:00 +0000 (15:39 +0000)]
net/smc: remove redundant re-assignment of pointer link

The pointer link is being re-assigned the same value that it was
initialized with in the previous declaration statement. The
re-assignment is redundant and can be removed.

Fixes: 387707fdf486 ("net/smc: convert static link ID to dynamic references")
Signed-off-by: Colin Ian King <colin.i.king@gmail.com>
Reviewed-by: Tony Lu <tonylu@linux.alibaba.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet/smc: Introduce TCP ULP support
Tony Lu [Tue, 28 Dec 2021 13:44:36 +0000 (21:44 +0800)]
net/smc: Introduce TCP ULP support

This implements TCP ULP for SMC, helps applications to replace TCP with
SMC protocol in place. And we use it to implement transparent
replacement.

This replaces original TCP sockets with SMC, reuse TCP as clcsock when
calling setsockopt with TCP_ULP option, and without any overhead.

To replace TCP sockets with SMC, there are two approaches:

- use setsockopt() syscall with TCP_ULP option, if error, it would
  fallback to TCP.

- use BPF prog with types BPF_CGROUP_INET_SOCK_CREATE or others to
  replace transparently. BPF hooks some points in create socket, bind
  and others, users can inject their BPF logics without modifying their
  applications, and choose which connections should be replaced with SMC
  by calling setsockopt() in BPF prog, based on rules, such as TCP tuples,
  PID, cgroup, etc...

  BPF doesn't support calling setsockopt with TCP_ULP now, I will send the
  patches after this accepted.

Signed-off-by: Tony Lu <tonylu@linux.alibaba.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoMerge branch 'smc-RDMA-net-namespace'
David S. Miller [Sun, 2 Jan 2022 12:07:39 +0000 (12:07 +0000)]
Merge branch 'smc-RDMA-net-namespace'

Tony Lu says:

====================
RDMA device net namespace support for SMC

This patch set introduces net namespace support for linkgroups.

Path 1 is the main approach to implement net ns support.

Path 2 - 4 are the additional modifications to let us know the netns.
Also, I will submit changes of smc-tools to github later.

Currently, smc doesn't support net namespace isolation. The ibdevs
registered to smc are shared for all linkgroups and connections. When
running applications in different net namespaces, such as container
environment, applications should only use the ibdevs that belongs to the
same net namespace.

This adds a new field, net, in smc linkgroup struct. During first
contact, it checks and find the linkgroup has same net namespace, if
not, it is going to create and initialized the net field with first
link's ibdev net namespace. When finding the rdma devices, it also checks
the sk net device's and ibdev's net namespaces. After net namespace
destroyed, the net device and ibdev move to root net namespace,
linkgroups won't be matched, and wait for lgr free.

If rdma net namespace exclusive mode is not enabled, it behaves as
before.

Steps to enable and test net namespaces:

1. enable RDMA device net namespace exclusive support
rdma system set netns exclusive # default is shared

2. create new net namespace, move and initialize them
ip netns add test1
rdma dev set mlx5_1 netns test1
ip link set dev eth2 netns test1
ip netns exec test1 ip link set eth2 up
ip netns exec test1 ip addr add ${HOST_IP}/26 dev eth2

3. setup server and client, connect N <-> M
ip netns exec test1 smc_run sockperf server --tcp # server
ip netns exec test1 smc_run sockperf pp --tcp -i ${SERVER_IP} # client

4. netns isolated linkgroups (2 * 2 mesh) with their own linkgroups
  - server
LG-ID    LG-Role  LG-Type  VLAN  #Conns  PNET-ID
00000100 SERV     SINGLE      0       0
00000200 SERV     SINGLE      0       0
00000300 SERV     SINGLE      0       0
00000400 SERV     SINGLE      0       0

  - client
LG-ID    LG-Role  LG-Type  VLAN  #Conns  PNET-ID
00000100 CLNT     SINGLE      0       0
00000200 CLNT     SINGLE      0       0
00000300 CLNT     SINGLE      0       0
00000400 CLNT     SINGLE      0       0
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet/smc: Add net namespace for tracepoints
Tony Lu [Tue, 28 Dec 2021 13:06:12 +0000 (21:06 +0800)]
net/smc: Add net namespace for tracepoints

This prints net namespace ID, helps us to distinguish different net
namespaces when using tracepoints.

Signed-off-by: Tony Lu <tonylu@linux.alibaba.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet/smc: Print net namespace in log
Tony Lu [Tue, 28 Dec 2021 13:06:11 +0000 (21:06 +0800)]
net/smc: Print net namespace in log

This adds net namespace ID to the kernel log, net_cookie is unique in
the whole system. It is useful in container environment.

Signed-off-by: Tony Lu <tonylu@linux.alibaba.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet/smc: Add netlink net namespace support
Tony Lu [Tue, 28 Dec 2021 13:06:10 +0000 (21:06 +0800)]
net/smc: Add netlink net namespace support

This adds net namespace ID to diag of linkgroup, helps us to distinguish
different namespaces, and net_cookie is unique in the whole system.

Signed-off-by: Tony Lu <tonylu@linux.alibaba.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet/smc: Introduce net namespace support for linkgroup
Tony Lu [Tue, 28 Dec 2021 13:06:09 +0000 (21:06 +0800)]
net/smc: Introduce net namespace support for linkgroup

Currently, rdma device supports exclusive net namespace isolation,
however linkgroup doesn't know and support ibdev net namespace.
Applications in the containers don't want to share the nics if we
enabled rdma exclusive mode. Every net namespaces should have their own
linkgroups.

This patch introduce a new field net for linkgroup, which is standing
for the ibdev net namespace in the linkgroup. The net in linkgroup is
initialized with the net namespace of link's ibdev. It compares the net
of linkgroup and sock or ibdev before choose it, if no matched, create
new one in current net namespace. If rdma net namespace exclusive mode
is not enabled, it behaves as before.

Signed-off-by: Tony Lu <tonylu@linux.alibaba.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf-next
David S. Miller [Fri, 31 Dec 2021 14:35:40 +0000 (14:35 +0000)]
Merge git://git./linux/kernel/git/bpf/bpf-next

Alexei Starovoitov says:

====================
pull-request: bpf-next 2021-12-30

The following pull-request contains BPF updates for your *net-next* tree.

We've added 72 non-merge commits during the last 20 day(s) which contain
a total of 223 files changed, 3510 insertions(+), 1591 deletions(-).

The main changes are:

1) Automatic setrlimit in libbpf when bpf is memcg's in the kernel, from Andrii.

2) Beautify and de-verbose verifier logs, from Christy.

3) Composable verifier types, from Hao.

4) bpf_strncmp helper, from Hou.

5) bpf.h header dependency cleanup, from Jakub.

6) get_func_[arg|ret|arg_cnt] helpers, from Jiri.

7) Sleepable local storage, from KP.

8) Extend kfunc with PTR_TO_CTX, PTR_TO_MEM argument support, from Kumar.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoMerge tag 'mlx5-updates-2021-12-28' of git://git.kernel.org/pub/scm/linux/kernel...
David S. Miller [Fri, 31 Dec 2021 14:29:31 +0000 (14:29 +0000)]
Merge tag 'mlx5-updates-2021-12-28' of git://git./linux/kernel/git/saeed/linux

Saeed Mahameed says:

====================
mlx5 Software steering, New features and optimizations

This patch series brings various SW steering features, optimizations and
debug-ability focused improvements.

 1) Expose debugfs for dumping the SW steering resources
 2) Removing unused fields
 3) support for matching on new fields
 4) steering optimization for RX/TX-only rules
 5) Make Software steering the default steering mechanism when
    available, applies only to Switchdev mode FDB

From Yevgeny Kliteynik and Muhammad Sammar:

 - Patch 1 fixes an error flow in creating matchers
 - Patch 2 fix lower case macro prefix "mlx5_" to "MLX5_"
 - Patch 3 removes unused struct member in mlx5dr_matcher
 - Patch 4 renames list field in matcher struct to list_node to reflect the
   fact that is field is for list node that is stored on another struct's lists
 - Patch 5 adds checking for valid Flex parser ID value
 - Patch 6 adds the missing reserved fields to dr_match_param and aligns it to
   the format that is defined by HW spec
 - Patch 7 adds support for dumping SW steering (SMFS) resources using debugfs
   in CSV format: domain and its tables, matchers and rules
 - Patch 8 adds support for a new destination type - UPLINK
 - Patch 9 adds WARN_ON_ONCE on refcount checks in SW steering object destructors
 - Patches 10, 11, 12 add misc5 flow table match parameters and add support for
   matching on tunnel headers 0 and 1
 - Patch 13 adds support for matching on geneve_tlv_option_0_exist field
 - Patch 14 implements performance optimization for for empty or RX/TX-only
   matchers by splitting RX and TX matchers handling: matcher connection in the
   matchers chain is split into two separate lists (RX only and TX only), which
   solves a usecase of many RX or TX only rules that create a long chain of
   RX/TX-only paths w/o the actual rules
 - Patch 15 ignores modify TTL if device doesn't support it instead of
   adding and unsupported action
 - Patch 16 sets SMFS as a default steering mode
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoMerge branch 'hnsd3-next'
David S. Miller [Fri, 31 Dec 2021 14:25:47 +0000 (14:25 +0000)]
Merge branch 'hnsd3-next'

Guangbin Huang says:

====================
net: hns3: refactor cmdq functions in PF/VF

Currently, hns3 PF and VF module have two sets of cmdq APIs to provide
cmdq message interaction functions. Most of these APIs are the same. The
only differences are the function variables and names with pf and vf
suffixes. These two sets of cmdq APIs are redundent and add extra bug fix
work.

This series refactor the cmdq APIs in hns3 PF and VF by implementing one
set of common cmdq APIs for PF and VF reuse and deleting the old APIs.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: hns3: delete the hclge_cmd.c and hclgevf_cmd.c
Jie Wang [Fri, 31 Dec 2021 10:22:43 +0000 (18:22 +0800)]
net: hns3: delete the hclge_cmd.c and hclgevf_cmd.c

currently most cmdq APIs are unified in hclge_comm_cmd.c. Newly developed
cmdq APIs should also be placed in hclge_comm_cmd.c. So there is no need to
keep hclge_cmd.c and hclgevf_cmd.c.

This patch moves the hclge(vf)_cmd_send to hclge(vf)_main.c and deletes
the source files and makefile scripts.

Signed-off-by: Jie Wang <wangjie125@huawei.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: hns3: refactor VF cmdq init and uninit APIs with new common APIs
Jie Wang [Fri, 31 Dec 2021 10:22:42 +0000 (18:22 +0800)]
net: hns3: refactor VF cmdq init and uninit APIs with new common APIs

This patch uses common cmdq init and uninit APIs to replace the old APIs in
VF cmdq module init and uninit module. Then the old VF init and uninit
APIs is deleted.

Signed-off-by: Jie Wang <wangjie125@huawei.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: hns3: refactor PF cmdq init and uninit APIs with new common APIs
Jie Wang [Fri, 31 Dec 2021 10:22:41 +0000 (18:22 +0800)]
net: hns3: refactor PF cmdq init and uninit APIs with new common APIs

This patch uses common cmdq init and uninit APIs to replace the old APIs in
PF cmdq module init and uninit modules. Then the old PF init and uninit
APIs is deleted.

Signed-off-by: Jie Wang <wangjie125@huawei.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: hns3: create common cmdq init and uninit APIs
Jie Wang [Fri, 31 Dec 2021 10:22:40 +0000 (18:22 +0800)]
net: hns3: create common cmdq init and uninit APIs

The PF and VF cmdq init and uninit APIs are also almost same espect the
suffixes of API names.

This patch creates common cmdq init and uninit APIs needed by PF and VF
cmdq modules. The next patch will use the new unified APIs to replace init
and uninit APIs in PF module.

Signed-off-by: Jie Wang <wangjie125@huawei.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: hns3: refactor VF cmdq resource APIs with new common APIs
Jie Wang [Fri, 31 Dec 2021 10:22:39 +0000 (18:22 +0800)]
net: hns3: refactor VF cmdq resource APIs with new common APIs

This patch uses common cmdq resource allocate/free/query APIs to replace
the old APIs in VF cmdq module and deletes the old cmdq resource APIs.
Still we kept hclgevf_cmd_setup_basic_desc name as a seam API to avoid too
many meaningless replacement.

Signed-off-by: Jie Wang <wangjie125@huawei.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: hns3: refactor PF cmdq resource APIs with new common APIs
Jie Wang [Fri, 31 Dec 2021 10:22:38 +0000 (18:22 +0800)]
net: hns3: refactor PF cmdq resource APIs with new common APIs

This patch uses common cmdq resource allocate/free/query APIs to replace
the old APIs in PF cmdq module and deletes the old cmdq resource APIs.
Still we kept hclge_cmd_setup_basic_desc name as a seam API to avoid too
many meaningless replacement.

Signed-off-by: Jie Wang <wangjie125@huawei.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: hns3: create common cmdq resource allocate/free/query APIs
Jie Wang [Fri, 31 Dec 2021 10:22:37 +0000 (18:22 +0800)]
net: hns3: create common cmdq resource allocate/free/query APIs

The PF and VF cmdq module resource allocate/free/query APIs are almost the
same espect the suffixes of API names. These same implementations bring
double development and bugfix work.

This patch creates common cmdq resource allocate/free/query APIs called by
PF and VF cmdq init/uninit APIs. The next patch will use the new unified
APIs to replace init/uninit APIs.

Signed-off-by: Jie Wang <wangjie125@huawei.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: hns3: refactor hclgevf_cmd_send with new hclge_comm_cmd_send API
Jie Wang [Fri, 31 Dec 2021 10:22:36 +0000 (18:22 +0800)]
net: hns3: refactor hclgevf_cmd_send with new hclge_comm_cmd_send API

This patch firstly uses new hardware description struct hclge_comm_hw as
child member of hclgevf_hw and deletes the old hardware description child
members. All the hclgevf_hw variables used in VF module is modified
according to the new hclgevf_hw.

Secondly hclgevf_cmd_send is refactored to use hclge_comm_cmd_send APIs.
The old functions called by hclgevf_cmd_send are all deleted. Still we kept
hclgevf_cmd_send to avoid too many meaningless modifications.

Signed-off-by: Jie Wang <wangjie125@huawei.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: hns3: refactor hclge_cmd_send with new hclge_comm_cmd_send API
Jie Wang [Fri, 31 Dec 2021 10:22:35 +0000 (18:22 +0800)]
net: hns3: refactor hclge_cmd_send with new hclge_comm_cmd_send API

This patch firstly uses new hardware description struct hclge_comm_hw as
child member of hclge_hw and deletes the original child memebers of
hclge_hw. All the hclge_hw variables used in PF module is modified
according to the new hclge_hw.

Secondly hclge_cmd_send is refactored to use hclge_comm_cmd_send APIs. The
old functions called by hclge_cmd_send are deleted and hclge_cmd_send is
kept to avoid too many meaningless modifications.

Signed-off-by: Jie Wang <wangjie125@huawei.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: hns3: create new set of unified hclge_comm_cmd_send APIs
Jie Wang [Fri, 31 Dec 2021 10:22:34 +0000 (18:22 +0800)]
net: hns3: create new set of unified hclge_comm_cmd_send APIs

This patch create new set of unified hclge_comm_cmd_send APIs for PF and VF
cmdq module. Subfunctions called by hclge_comm_cmd_send are also created
include cmdq result check, cmdq return code conversion and ring space
opertaion APIs.

These new common cmdq APIs will be used to replace the old PF and VF cmdq
APIs in next patches.

Signed-off-by: Jie Wang <wangjie125@huawei.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: hns3: use struct hclge_desc to replace hclgevf_desc in VF cmdq module
Jie Wang [Fri, 31 Dec 2021 10:22:33 +0000 (18:22 +0800)]
net: hns3: use struct hclge_desc to replace hclgevf_desc in VF cmdq module

This patch use new common struct hclge_desc to replace struct hclgevf_desc
in VF cmdq module and then delete the old struct hclgevf_desc.

Signed-off-by: Jie Wang <wangjie125@huawei.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: hns3: create new cmdq hardware description structure hclge_comm_hw
Jie Wang [Fri, 31 Dec 2021 10:22:32 +0000 (18:22 +0800)]
net: hns3: create new cmdq hardware description structure hclge_comm_hw

Currently PF and VF cmdq APIs use struct hclge(vf)_hw to describe cmdq
hardware information needed by hclge(vf)_cmd_send. There are a little
differences between its child struct hclge_cmq_ring and hclgevf_cmq_ring.
It is redundent to use two sets of structures to support same functions.

So this patch creates new set of common cmdq hardware description
structures(hclge_comm_hw) to unify PF and VF cmdq functions. The struct
hclge_desc is still kept to avoid too many meaningless replacement.

These new structures will be used to unify hclge(vf)_hw structures in PF
and VF cmdq APIs in next patches.

Signed-off-by: Jie Wang <wangjie125@huawei.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: hns3: refactor hns3 makefile to support hns3_common module
Jie Wang [Fri, 31 Dec 2021 10:22:31 +0000 (18:22 +0800)]
net: hns3: refactor hns3 makefile to support hns3_common module

Currently we plan to refactor PF and VF cmdq module. A new file folder
hns3_common will be created to store new common APIs used by PF and VF
cmdq module. Thus the PF and VF compilation process will both depends on
the hns3_common. This may cause parallel building problems if we add a new
makefile building unit.

So this patch combined the PF and VF makefile scripts to the top level
makefile to support the new hns3_common which will be created in the next
patch.

Signed-off-by: Jie Wang <wangjie125@huawei.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet/mlx5: Set SMFS as a default steering mode if device supports it
Yevgeny Kliteynik [Sun, 28 Nov 2021 19:48:47 +0000 (21:48 +0200)]
net/mlx5: Set SMFS as a default steering mode if device supports it

Set SMFS (SW-managed flow steering) as a default steering mode
instead of DMFS (device-managed flow steering)

In SMFS, the driver writes the STEs (Steering Table Entries) directly
to the device's ICM, which allows for a higher rule insertion rate
than through using FW command interface, as it is done in DMFS.

SMFS/DMFS steering modes can be configured through devlink param
'flow_steering_mode'. The possible values are 'smfs' or 'dmfs'.
The desired 'flow_steering_mode' param value should be set before
enabling switchdev mode.

Example:

  # devlink dev param set pci/0000:05:00.0 name flow_steering_mode smfs
  # devlink dev eswitch set pci/0000:05:00.0 mode switchdev

Signed-off-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
2 years agonet/mlx5: DR, Ignore modify TTL if device doesn't support it
Yevgeny Kliteynik [Tue, 23 Nov 2021 00:11:12 +0000 (02:11 +0200)]
net/mlx5: DR, Ignore modify TTL if device doesn't support it

When modifying TTL, packet's csum has to be recalculated.
Due to HW issue in ConnectX-5, csum recalculation for modify TTL
is supported through a work-around that is specifically enabled
by configuration.
If the work-around isn't enabled, ignore the modify TTL action
rather than adding an unsupported action.

Signed-off-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
2 years agonet/mlx5: DR, Improve steering for empty or RX/TX-only matchers
Yevgeny Kliteynik [Wed, 13 Oct 2021 23:34:00 +0000 (02:34 +0300)]
net/mlx5: DR, Improve steering for empty or RX/TX-only matchers

Every matcher has RX and TX paths. When a new matcher is created, its RX
and TX start/end anchors are connected to the respective RX and TX anchors
of the previous and next matchers.
This creates a potential performance issue: when a certain rule is added
to a matcher, in many cases it is RX or TX only rule, which may create a
long chain of RX/TX-only paths w/o the actual rules.

This patch aims to handle this issue.

RX and TX matchers are now handled separately: matcher connection in the
matchers chain is split into two separate lists: RX only and TX only.
when a new matcher is created, it is initially created 'detached' - its
RX/TX members are not inserted into the table's matcher list.
When an actual rule is added, only its appropriate RX or TX nic matchers
are then added to the table's nic matchers list and inserted into its
place in the chain of matchers.
I.e., if the rule that is being added is an RX-only rule, only the RX
part of the matcher will be connected to the chain, while TX part of the
matcher remains detached and doesn't prolong the TX chain of the matchers.

Same goes for rule deletion: when the last RX/TX rule of the nic matcher
is destroyed, the nic matcher is removed from its list.

Signed-off-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
2 years agonet/mlx5: DR, Add support for matching on geneve_tlv_option_0_exist field
Yevgeny Kliteynik [Mon, 8 Nov 2021 00:42:50 +0000 (02:42 +0200)]
net/mlx5: DR, Add support for matching on geneve_tlv_option_0_exist field

Match on geneve_tlv_option_0_exist field on devices that support STEv1.

Signed-off-by: Muhammad Sammar <muhammads@nvidia.com>
Signed-off-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
2 years agonet/mlx5: DR, Support matching on tunnel headers 0 and 1
Muhammad Sammar [Sun, 5 Sep 2021 14:07:49 +0000 (17:07 +0300)]
net/mlx5: DR, Support matching on tunnel headers 0 and 1

Tunnel headers are generic encapsulation headers, applies for all
tunneling protocols identified by the device native parser or by the
programmable parser, this support will enable raw matching headers 0 and 1.

Signed-off-by: Muhammad Sammar <muhammads@nvidia.com>
Signed-off-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
2 years agonet/mlx5: DR, Add misc5 to match_param structs
Muhammad Sammar [Sun, 5 Sep 2021 11:54:59 +0000 (14:54 +0300)]
net/mlx5: DR, Add misc5 to match_param structs

Add misc5 match params to enable matching tunnel headers.

Signed-off-by: Muhammad Sammar <muhammads@nvidia.com>
2 years agonet/mlx5: Add misc5 flow table match parameters
Muhammad Sammar [Sun, 5 Sep 2021 12:16:21 +0000 (15:16 +0300)]
net/mlx5: Add misc5 flow table match parameters

Add support for misc5 match parameter as per HW spec, this will allow
matching on tunnel_header fields.

Signed-off-by: Muhammad Sammar <muhammads@nvidia.com>
Signed-off-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
2 years agonet/mlx5: DR, Warn on failure to destroy objects due to refcount
Yevgeny Kliteynik [Wed, 29 Sep 2021 12:36:32 +0000 (15:36 +0300)]
net/mlx5: DR, Warn on failure to destroy objects due to refcount

Add WARN_ON_ONCE on refcount checks in SW steering object destructors

Signed-off-by: Paul Blakey <paulb@nvidia.com>
Signed-off-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
2 years agonet/mlx5: DR, Add support for UPLINK destination type
Yevgeny Kliteynik [Sun, 24 Oct 2021 18:36:30 +0000 (21:36 +0300)]
net/mlx5: DR, Add support for UPLINK destination type

Add support for a new destination type - UPLINK.

Signed-off-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
2 years agonet/mlx5: DR, Add support for dumping steering info
Muhammad Sammar [Wed, 9 Dec 2020 11:40:50 +0000 (13:40 +0200)]
net/mlx5: DR, Add support for dumping steering info

Extend mlx5 debugfs support to present Software Steering resources:
dr_domain including it's tables, matchers and rules.
The interface is read-only. While dump is being presented, new steering
rules cannot be inserted/deleted.

The steering information is dumped in the CSV form with the following
format:

    <object_type>,<object_ID>, <object_info>,...,<object_info>

This data can be read at the following path:

    /sys/kernel/debug/mlx5/<BDF>/steering/fdb/<domain_handle>

Example:

    # cat /sys/kernel/debug/mlx5/0000:82:00.0/steering/fdb/dmn_000018644
    3100,0x55caa4621c50,0xee802,4,65533
    3101,0x55caa4621c50,0xe0100008

Changes in V2:
 - Reduce temp hex buffer size and avoid unnecessary memset
 - Use bin2hex() instead of DIY loop
 - Don't check debugfs functions return values

Signed-off-by: Muhammad Sammar <muhammads@nvidia.com>
Signed-off-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
2 years agonet/mlx5: DR, Add missing reserved fields to dr_match_param
Muhammad Sammar [Mon, 5 Jul 2021 12:37:56 +0000 (15:37 +0300)]
net/mlx5: DR, Add missing reserved fields to dr_match_param

Add the reserved fields to dr_match_param and arrange
as mlx5_ifc_dr_match_param_bits.

Signed-off-by: Muhammad Sammar <muhammads@nvidia.com>
2 years agonet/mlx5: DR, Add check for flex parser ID value
Yevgeny Kliteynik [Wed, 10 Nov 2021 22:22:29 +0000 (00:22 +0200)]
net/mlx5: DR, Add check for flex parser ID value

Allow only legal values for flex parser ID - values from 0 to 7.
For other values skip the parser, and as a result the matcher creation
will fail for using invalid flex parser ID.

Signed-off-by: Hamdan Igbaria <hamdani@nvidia.com>
Signed-off-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
2 years agonet/mlx5: DR, Rename list field in matcher struct to list_node
Yevgeny Kliteynik [Tue, 16 Nov 2021 15:57:57 +0000 (17:57 +0200)]
net/mlx5: DR, Rename list field in matcher struct to list_node

In dr_types structs, some list fields are list heads, and some
are just list nodes that are stored on the other structs' lists.
Rename the appropriate list field to reflect this distinction.

Signed-off-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
2 years agonet/mlx5: DR, Remove unused struct member in matcher
Yevgeny Kliteynik [Wed, 10 Nov 2021 22:52:59 +0000 (00:52 +0200)]
net/mlx5: DR, Remove unused struct member in matcher

Signed-off-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
2 years agonet/mlx5: DR, Fix lower case macro prefix "mlx5_" to "MLX5_"
Yevgeny Kliteynik [Thu, 18 Nov 2021 00:32:37 +0000 (02:32 +0200)]
net/mlx5: DR, Fix lower case macro prefix "mlx5_" to "MLX5_"

Macros prefix should be capital letters - fix the prefix in
mlx5_FLEX_PARSER_MPLS_OVER_UDP_ENABLED.

Signed-off-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
2 years agonet/mlx5: DR, Fix error flow in creating matcher
Yevgeny Kliteynik [Tue, 14 Dec 2021 10:56:18 +0000 (12:56 +0200)]
net/mlx5: DR, Fix error flow in creating matcher

The error code of nic matcher init functions wasn't checked.
This patch improves the matcher init function and fix error flow bug:
the handling of match parameter is moved into a separate function
and error flow is simplified.

Signed-off-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
2 years agobpf: Fix typo in a comment in bpf lpm_trie.
Leon Huayra [Wed, 29 Dec 2021 14:44:22 +0000 (22:44 +0800)]
bpf: Fix typo in a comment in bpf lpm_trie.

Fix typo in a comment in trie_update_elem().

Signed-off-by: Leon Huayra <hffilwlqm@gmail.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20211229144422.70339-1-hffilwlqm@gmail.com
2 years agonet: dsa: bcm_sf2: refactor LED regs access
Rafał Miłecki [Wed, 29 Dec 2021 17:16:42 +0000 (18:16 +0100)]
net: dsa: bcm_sf2: refactor LED regs access

1. Define more regs. Some switches (e.g. BCM4908) have up to 6 regs.
2. Add helper for handling non-lineral port <-> reg mappings.
3. Add support for 12 B LED reg blocks on BCM4908 (different layout)

Complete support for LEDs setup will be implemented once Linux receives
a proper design & implementation for "hardware" LEDs.

Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
Acked-by: Florian Fainelli <f.fainelli@gmail.com>
Link: https://lore.kernel.org/r/20211229171642.22942-1-zajec5@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agonet: remove references to CONFIG_IRDA in network header files
Lukas Bulwahn [Wed, 29 Dec 2021 11:36:20 +0000 (12:36 +0100)]
net: remove references to CONFIG_IRDA in network header files

Commit d64c2a76123f ("staging: irda: remove the irda network stack and
drivers") removes the config IRDA.

Remove the remaining references to this non-existing config in the network
header files.

Signed-off-by: Lukas Bulwahn <lukas.bulwahn@gmail.com>
Link: https://lore.kernel.org/r/20211229113620.19368-1-lukas.bulwahn@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agonfc: st21nfca: remove redundant assignment to variable i
Colin Ian King [Thu, 30 Dec 2021 16:12:30 +0000 (16:12 +0000)]
nfc: st21nfca: remove redundant assignment to variable i

Variable i is being assigned a value that is never read, the
assignment is redundant and can be removed. Cleans up clang-scan
build warning:

drivers/nfc/st21nfca/i2c.c:319:4: warning: Value stored to 'i'
is never read [deadcode.DeadStores]
                        i = 0;

Signed-off-by: Colin Ian King <colin.i.king@gmail.com>
Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@canonical.com>
Link: https://lore.kernel.org/r/20211230161230.428457-1-colin.i.king@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agobpf, docs: Move the packet access instructions last in instruction-set.rst
Christoph Hellwig [Thu, 23 Dec 2021 10:19:06 +0000 (11:19 +0100)]
bpf, docs: Move the packet access instructions last in instruction-set.rst

The packet access instructions are a convoluted leftover from classic
BPF.  Move them last past the much more important atomic operations,
and improve the rendering of the code example.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20211223101906.977624-5-hch@lst.de
2 years agobpf, docs: Generate nicer tables for instruction encodings
Christoph Hellwig [Thu, 23 Dec 2021 10:19:05 +0000 (11:19 +0100)]
bpf, docs: Generate nicer tables for instruction encodings

Use RST tables that are nicely readable both in plain ascii as well as
in html to render the instruction encodings, and add a few subheadings
to better structure the text.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20211223101906.977624-4-hch@lst.de
2 years agobpf, docs: Split the comparism to classic BPF from instruction-set.rst
Christoph Hellwig [Thu, 23 Dec 2021 10:19:04 +0000 (11:19 +0100)]
bpf, docs: Split the comparism to classic BPF from instruction-set.rst

Split the introductory that explain eBPF vs classic BPF and how it maps
to hardware from the instruction set specification into a standalone
document.  This duplicates a little bit of information but gives us a
useful reference for the eBPF instrution set that is not encumbered by
classic BPF.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20211223101906.977624-3-hch@lst.de
2 years agobpf, docs: Fix verifier references
Christoph Hellwig [Thu, 23 Dec 2021 10:19:03 +0000 (11:19 +0100)]
bpf, docs: Fix verifier references

Use normal RST file reference instead of linkage copied from the old filter.rst
document that does not actually work when using HTML output.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20211223101906.977624-2-hch@lst.de
2 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net
Jakub Kicinski [Thu, 30 Dec 2021 20:12:12 +0000 (12:12 -0800)]
Merge git://git./linux/kernel/git/netdev/net

drivers/net/ethernet/mellanox/mlx5/core/en_tc.c
  commit 077cdda764c7 ("net/mlx5e: TC, Fix memory leak with rules with internal port")
  commit 31108d142f36 ("net/mlx5: Fix some error handling paths in 'mlx5e_tc_add_fdb_flow()'")
  commit 4390c6edc0fb ("net/mlx5: Fix some error handling paths in 'mlx5e_tc_add_fdb_flow()'")
  https://lore.kernel.org/all/20211229065352.30178-1-saeed@kernel.org/

net/smc/smc_wr.c
  commit 49dc9013e34b ("net/smc: Use the bitmap API when applicable")
  commit 349d43127dac ("net/smc: fix kernel panic caused by race of smc_sock")
  bitmap_zero()/memset() is removed by the fix

Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoMerge tag 'net-5.16-rc8' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net
Linus Torvalds [Thu, 30 Dec 2021 19:12:12 +0000 (11:12 -0800)]
Merge tag 'net-5.16-rc8' of git://git./linux/kernel/git/netdev/net

Pull networking fixes from Jakub Kicinski:
 "Including fixes from.. Santa?

  No regressions on our radar at this point. The igc problem fixed here
  was the last one I was tracking but it was broken in previous
  releases, anyway. Mostly driver fixes and a couple of largish SMC
  fixes.

  Current release - regressions:

   - xsk: initialise xskb free_list_node, fixup for a -rc7 fix

  Current release - new code bugs:

   - mlx5: handful of minor fixes:

   - use first online CPU instead of hard coded CPU

   - fix some error handling paths in 'mlx5e_tc_add_fdb_flow()'

   - fix skb memory leak when TC classifier action offloads are disabled

   - fix memory leak with rules with internal OvS port

  Previous releases - regressions:

   - igc: do not enable crosstimestamping for i225-V models

  Previous releases - always broken:

   - udp: use datalen to cap ipv6 udp max gso segments

   - fix use-after-free in tw_timer_handler due to early free of stats

   - smc: fix kernel panic caused by race of smc_sock

   - smc: don't send CDC/LLC message if link not ready, avoid timeouts

   - sctp: use call_rcu to free endpoint, avoid UAF in sock diag

   - bridge: mcast: add and enforce query interval minimum

   - usb: pegasus: do not drop long Ethernet frames

   - mlx5e: fix ICOSQ recovery flow for XSK

   - nfc: uapi: use kernel size_t to fix user-space builds"

* tag 'net-5.16-rc8' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net: (47 commits)
  fsl/fman: Fix missing put_device() call in fman_port_probe
  selftests: net: using ping6 for IPv6 in udpgro_fwd.sh
  Documentation: fix outdated interpretation of ip_no_pmtu_disc
  net/ncsi: check for error return from call to nla_put_u32
  net: bridge: mcast: fix br_multicast_ctx_vlan_global_disabled helper
  net: fix use-after-free in tw_timer_handler
  selftests: net: Fix a typo in udpgro_fwd.sh
  selftests/net: udpgso_bench_tx: fix dst ip argument
  net: bridge: mcast: add and enforce startup query interval minimum
  net: bridge: mcast: add and enforce query interval minimum
  ipv6: raw: check passed optlen before reading
  xsk: Initialise xskb free_list_node
  net/mlx5e: Fix wrong features assignment in case of error
  net/mlx5e: TC, Fix memory leak with rules with internal port
  ionic: Initialize the 'lif->dbid_inuse' bitmap
  igc: Fix TX timestamp support for non-MSI-X platforms
  igc: Do not enable crosstimestamping for i225-V models
  net/smc: fix kernel panic caused by race of smc_sock
  net/smc: don't send CDC/LLC message if link not ready
  NFC: st21nfca: Fix memory leak in device probe and remove
  ...

2 years agoMerge tag 'char-misc-5.16' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh...
Linus Torvalds [Thu, 30 Dec 2021 17:52:32 +0000 (09:52 -0800)]
Merge tag 'char-misc-5.16' of git://git./linux/kernel/git/gregkh/char-misc

Pull char/misc fixes from Greg KH:
 "Here are two misc driver fixes for 5.16-final:

   - binder accounting fix to resolve reported problem

   - nitro_enclaves fix for mmap assert warning output

  Both of these have been for over a week with no reported issues"

* tag 'char-misc-5.16' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/char-misc:
  nitro_enclaves: Use get_user_pages_unlocked() call to handle mmap assert
  binder: fix async_free_space accounting for empty parcels

2 years agoMerge tag 'usb-5.16' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb
Linus Torvalds [Thu, 30 Dec 2021 17:49:54 +0000 (09:49 -0800)]
Merge tag 'usb-5.16' of git://git./linux/kernel/git/gregkh/usb

Pull USB fixes from Greg KH:
 "Here are some small USB driver fixes for 5.16 to resolve some reported
  problems:

   - mtu3 driver fixes

   - typec ucsi driver fix

   - xhci driver quirk added

   - usb gadget f_fs fix for reported crash

  All of these have been in linux-next for a while with no reported
  problems"

* tag 'usb-5.16' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb:
  usb: typec: ucsi: Only check the contract if there is a connection
  xhci: Fresco FL1100 controller should not have BROKEN_MSI quirk set.
  usb: mtu3: set interval of FS intr and isoc endpoint
  usb: mtu3: fix list_head check warning
  usb: mtu3: add memory barrier before set GPD's HWO
  usb: mtu3: fix interval value for intr and isoc
  usb: gadget: f_fs: Clear ffs_eventfd in ffs_data_clear.

2 years agofsl/fman: Fix missing put_device() call in fman_port_probe
Miaoqian Lin [Thu, 30 Dec 2021 12:26:27 +0000 (12:26 +0000)]
fsl/fman: Fix missing put_device() call in fman_port_probe

The reference taken by 'of_find_device_by_node()' must be released when
not needed anymore.
Add the corresponding 'put_device()' in the and error handling paths.

Fixes: 18a6c85fcc78 ("fsl/fman: Add FMan Port Support")
Signed-off-by: Miaoqian Lin <linmq006@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet/smc: Use the bitmap API when applicable
Christophe JAILLET [Thu, 30 Dec 2021 10:40:40 +0000 (11:40 +0100)]
net/smc: Use the bitmap API when applicable

Using the bitmap API is less verbose than hand writing them.
It also improves the semantic.

Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoselftests: net: using ping6 for IPv6 in udpgro_fwd.sh
Jianguo Wu [Thu, 30 Dec 2021 10:40:29 +0000 (18:40 +0800)]
selftests: net: using ping6 for IPv6 in udpgro_fwd.sh

udpgro_fwd.sh output following message:
  ping: 2001:db8:1::100: Address family for hostname not supported

Using ping6 when pinging IPv6 addresses.

Fixes: a062260a9d5f ("selftests: net: add UDP GRO forwarding self-tests")
Signed-off-by: Jianguo Wu <wujianguo@chinatelecom.cn>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoethtool: Remove redundant ret assignments
luo penghao [Thu, 30 Dec 2021 06:26:44 +0000 (06:26 +0000)]
ethtool: Remove redundant ret assignments

The assignment here will be overwritten, so it should be deleted

The clang_analyzer complains as follows:

net/ethtool/netlink.c:

Value stored to 'ret' is never read

Reported-by: Zeal Robot <zealci@zte.com.cn>
Signed-off-by: luo penghao <luo.penghao@zte.com.cn>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoDocumentation: fix outdated interpretation of ip_no_pmtu_disc
xu xin [Thu, 30 Dec 2021 03:28:56 +0000 (03:28 +0000)]
Documentation: fix outdated interpretation of ip_no_pmtu_disc

The updating way of pmtu has changed, but documentation is still in the
old way. So this patch updates the interpretation of ip_no_pmtu_disc and
min_pmtu.

See commit 28d35bcdd3925 ("net: ipv4: don't let PMTU updates increase
route MTU")

Reported-by: Zeal Robot <zealci@zte.com.cn>
Signed-off-by: xu xin <xu.xin16@zte.com.cn>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: lantiq_etop: remove unnecessary space in cast
Aleksander Jan Bajkowski [Wed, 29 Dec 2021 23:52:06 +0000 (00:52 +0100)]
net: lantiq_etop: remove unnecessary space in cast

As reported by checkpatch.pl, no space is necessary after a cast.

Signed-off-by: Aleksander Jan Bajkowski <olek2@wp.pl>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: lantiq_etop: make alignment match open parenthesis
Aleksander Jan Bajkowski [Wed, 29 Dec 2021 23:39:52 +0000 (00:39 +0100)]
net: lantiq_etop: make alignment match open parenthesis

checkpatch.pl complains as the following:

Alignment should match open parenthesis

Signed-off-by: Aleksander Jan Bajkowski <olek2@wp.pl>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: lantiq_etop: remove multiple assignments
Aleksander Jan Bajkowski [Wed, 29 Dec 2021 23:36:26 +0000 (00:36 +0100)]
net: lantiq_etop: remove multiple assignments

Documentation/process/coding-style.rst says (in line 88)
"Don't put multiple assignments on a single line either."

This patch fixes the coding style issue reported by checkpatch.pl.

Signed-off-by: Aleksander Jan Bajkowski <olek2@wp.pl>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: lantiq_etop: avoid precedence issues
Aleksander Jan Bajkowski [Wed, 29 Dec 2021 23:27:25 +0000 (00:27 +0100)]
net: lantiq_etop: avoid precedence issues

Add () around macro argument to avoid precedence issues

Signed-off-by: Aleksander Jan Bajkowski <olek2@wp.pl>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: lantiq_etop: replace strlcpy with strscpy
Aleksander Jan Bajkowski [Wed, 29 Dec 2021 23:22:57 +0000 (00:22 +0100)]
net: lantiq_etop: replace strlcpy with strscpy

strlcpy is marked as deprecated in Documentation/process/deprecated.rst,
and there is no functional difference when the caller expects truncation
(when not checking the return value). strscpy is relatively better as it
also avoids scanning the whole source string.

This silences the related checkpatch warnings from:
commit 5dbdb2d87c29 ("checkpatch: prefer strscpy to strlcpy")

Signed-off-by: Aleksander Jan Bajkowski <olek2@wp.pl>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoice: Add flow director support for channel mode
Kiran Patil [Wed, 29 Dec 2021 18:54:33 +0000 (10:54 -0800)]
ice: Add flow director support for channel mode

Add support to enable flow-director filter when multiple TCs are
configured. Flow director filter can be configured using ethtool
(--config-ntuple option). When multiple TCs are configured, each
TC is mapped to an unique HW VSI. So VSI corresponding to queue
used in filter is identified and flow director context is updated
with correct VSI while configuring ntuple filter in HW.

Signed-off-by: Kiran Patil <kiran.patil@intel.com>
Signed-off-by: Amritha Nambiar <amritha.nambiar@intel.com>
Signed-off-by: Sudheer Mogilappagari <sudheer.mogilappagari@intel.com>
Tested-by: Bharathi Sreenivas <bharathi.sreenivas@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoMerge branch '1GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/next-
David S. Miller [Thu, 30 Dec 2021 13:13:25 +0000 (13:13 +0000)]
Merge branch '1GbE' of git://git./linux/kernel/git/tnguy/next-
queue

Tony Nguyen says:

====================
1GbE Intel Wired LAN Driver Updates 2021-12-29

Ruud Bos says:

The igb driver provides support for PEROUT and EXTTS pin functions that
allow adapter external use of timing signals. At Hottinger Bruel & Kjaer we
are using the PEROUT function to feed a PTP corrected 1pps signal into an
FPGA as cross system synchronized time source.

Support for the PEROUT and EXTTS SDP functions is currently limited to
i210/i211 based adapters. This patch series enables these functions also
for 82580/i354/i350 based ones. Because the time registers of these
adapters do not have the nice split in second rollovers as the i210 has,
the implementation is slightly more complex compared to the i210
implementation.

The PEROUT function has been successfully tested on an i350 based ethernet
adapter. Using the following user space code excerpt, the driver outputs a
PTP corrected 1pps signal on the SDP0 pin of an i350:

    struct ptp_pin_desc desc;
    memset(&desc, 0, sizeof(desc));
    desc.index = 0;
    desc.func = PTP_PF_PEROUT;
    desc.chan = 0;
    if (ioctl(fd, PTP_PIN_SETFUNC, &desc) == 0) {
        struct timespec ts;
        if (clock_gettime(clkid, &ts) == 0) {
            struct ptp_perout_request rq;
            memset(&rq, 0, sizeof(rq));
            rq.index = 0;
            rq.start.sec = ts.tv_sec + 1;
            rq.start.nsec = 500000000;
            rq.period.sec  = 1;
            rq.period.nsec = 0;
            if (ioctl(fd, PTP_PEROUT_REQUEST, &rq) == 0) {
                /* 1pps signal is now available on SDP0 */
            }
        }
    }

The added EXTTS function has not been tested. However, looking at the data
sheets, the layout of the registers involved match the i210 exactly except
for the time registers mentioned before. Hence the almost identical
implementation.

---
Note: I made changes to fix RCT and checkpatch messages regarding
unnecessary parenthesis.
====================

Acked-by: Richard Cochran <richardcochran@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoMerge branch 'prestera-router-driver'
David S. Miller [Thu, 30 Dec 2021 13:12:42 +0000 (13:12 +0000)]
Merge branch 'prestera-router-driver'

Yevhen Orlov says:

====================
prestera: add basic router driver support

Add initial router support for Marvell Prestera driver.
Subscribe on inetaddr notifications. TRAP packets, that has to be routed
(if packet has router's destination MAC address).

Add features:
 - Support ip address adding on port.
   e.g.: "ip address add PORT 1.1.1.1/24"

Limitations:
 - Only regular port supported. Vlan will be added soon.
 - It is routing through CPU. Offloading will be added in
   next patches.

Co-developed-by: Taras Chornyi <tchornyi@marvell.com>
Signed-off-by: Taras Chornyi <tchornyi@marvell.com>
Co-developed-by: Oleksandr Mazur <oleksandr.mazur@plvision.eu>
Signed-off-by: Oleksandr Mazur <oleksandr.mazur@plvision.eu>
Signed-off-by: Yevhen Orlov <yevhen.orlov@plvision.eu>
Changes for v2:
* Remove useless assignment
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: marvell: prestera: Implement initial inetaddr notifiers
Yevhen Orlov [Mon, 27 Dec 2021 21:52:31 +0000 (23:52 +0200)]
net: marvell: prestera: Implement initial inetaddr notifiers

Add inetaddr notifiers to support add/del IPv4 address on switchdev
port. We create TRAP on first address, added on port and delete TRAP,
when last address removed.
Currently, driver supports only regular port to became routed.
Other port type support will be added later

Co-developed-by: Taras Chornyi <tchornyi@marvell.com>
Signed-off-by: Taras Chornyi <tchornyi@marvell.com>
Co-developed-by: Oleksandr Mazur <oleksandr.mazur@plvision.eu>
Signed-off-by: Oleksandr Mazur <oleksandr.mazur@plvision.eu>
Signed-off-by: Yevhen Orlov <yevhen.orlov@plvision.eu>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: marvell: prestera: Register inetaddr stub notifiers
Yevhen Orlov [Mon, 27 Dec 2021 21:52:30 +0000 (23:52 +0200)]
net: marvell: prestera: Register inetaddr stub notifiers

Initial implementation of notification handlers. For now this is just
stub.
So that we can move forward and add prestera_router_hw's objects
manipulations.

We support several addresses on interface. We just have nothing to do for
second address, because rif is already enabled on this interface, after
first one.

Co-developed-by: Taras Chornyi <tchornyi@marvell.com>
Signed-off-by: Taras Chornyi <tchornyi@marvell.com>
Co-developed-by: Oleksandr Mazur <oleksandr.mazur@plvision.eu>
Signed-off-by: Oleksandr Mazur <oleksandr.mazur@plvision.eu>
Signed-off-by: Yevhen Orlov <yevhen.orlov@plvision.eu>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: marvell: prestera: add hardware router objects accounting
Yevhen Orlov [Mon, 27 Dec 2021 21:52:29 +0000 (23:52 +0200)]
net: marvell: prestera: add hardware router objects accounting

Add prestera_router_hw.c. This file contains functions, which track HW
objects relations and links. This include implicity creation of objects,
that needed by requested one and implicity removing of objects, which
reference counter is became zero.

We need this layer, because kernel callbacks not always mapped to
creation of single HW object. So let it be two different layers - one
for subscribing and parsing kernel structures, and another
(prestera_router_hw.c) for HW objects relations tracking.

There is two types of objects on router_hw layer:
 - Explicit objects (rif_entry) : created by higher layer.
 - Implicit objects (vr) : created on demand by explicit objects.

Co-developed-by: Taras Chornyi <tchornyi@marvell.com>
Signed-off-by: Taras Chornyi <tchornyi@marvell.com>
Co-developed-by: Oleksandr Mazur <oleksandr.mazur@plvision.eu>
Signed-off-by: Oleksandr Mazur <oleksandr.mazur@plvision.eu>
Signed-off-by: Yevhen Orlov <yevhen.orlov@plvision.eu>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: marvell: prestera: Add prestera router infra
Yevhen Orlov [Mon, 27 Dec 2021 21:52:28 +0000 (23:52 +0200)]
net: marvell: prestera: Add prestera router infra

Add prestera_router.c, which contains code to subscribe/unsubscribe on
kernel notifiers for router. This handle kernel notifications,
parse structures to make key to manipulate prestera_router_hw's objects.

Also prestera_router is container for router's objects database.

Co-developed-by: Taras Chornyi <tchornyi@marvell.com>
Signed-off-by: Taras Chornyi <tchornyi@marvell.com>
Co-developed-by: Oleksandr Mazur <oleksandr.mazur@plvision.eu>
Signed-off-by: Oleksandr Mazur <oleksandr.mazur@plvision.eu>
Signed-off-by: Yevhen Orlov <yevhen.orlov@plvision.eu>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: marvell: prestera: Add router interface ABI
Yevhen Orlov [Mon, 27 Dec 2021 21:52:27 +0000 (23:52 +0200)]
net: marvell: prestera: Add router interface ABI

Add functions to enable routing on port, which is not in vlan.
Also we can enable routing on vlan.
prestera_hw_rif_create() take index of allocated virtual router.

Co-developed-by: Taras Chornyi <tchornyi@marvell.com>
Signed-off-by: Taras Chornyi <tchornyi@marvell.com>
Co-developed-by: Oleksandr Mazur <oleksandr.mazur@plvision.eu>
Signed-off-by: Oleksandr Mazur <oleksandr.mazur@plvision.eu>
Signed-off-by: Yevhen Orlov <yevhen.orlov@plvision.eu>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: marvell: prestera: add virtual router ABI
Yevhen Orlov [Mon, 27 Dec 2021 21:52:26 +0000 (23:52 +0200)]
net: marvell: prestera: add virtual router ABI

Add functions and structures to allocate virtual router.
prestera_hw_vr_create() return index of allocated VR so that we can move
forward and also add another objects (e.g. router interface),
which has link to VR.

Co-developed-by: Taras Chornyi <tchornyi@marvell.com>
Signed-off-by: Taras Chornyi <tchornyi@marvell.com>
Co-developed-by: Oleksandr Mazur <oleksandr.mazur@plvision.eu>
Signed-off-by: Oleksandr Mazur <oleksandr.mazur@plvision.eu>
Signed-off-by: Yevhen Orlov <yevhen.orlov@plvision.eu>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoMerge branch 'lighten uapi/bpf.h rebuilds'
Alexei Starovoitov [Thu, 30 Dec 2021 04:03:05 +0000 (20:03 -0800)]
Merge branch 'lighten uapi/bpf.h rebuilds'

Jakub Kicinski says:

====================

Last change in the bpf headers - disentangling BPF uapi from
netdevice.h. Both linux/bpf.h and uapi/bpf.h changes should
now rebuild ~1k objects down from the original ~18k. There's
probably more that can be done but it's diminishing returns.

Split into two patches for ease of review.
====================

Signed-off-by: Alexei Starovoitov <ast@kernel.org>
2 years agobpf: Invert the dependency between bpf-netns.h and netns/bpf.h
Jakub Kicinski [Thu, 30 Dec 2021 01:27:42 +0000 (17:27 -0800)]
bpf: Invert the dependency between bpf-netns.h and netns/bpf.h

netns/bpf.h gets included by netdevice.h (thru net_namespace.h)
which in turn gets included in a lot of places. We should keep
netns/bpf.h as light-weight as possible.

bpf-netns.h seems to contain more implementation details than
deserves to be included in a netns header. It needs to pull in
uapi/bpf.h to get various enum types.

Move enum netns_bpf_attach_type to netns/bpf.h and invert the
dependency. This makes netns/bpf.h fit the mold of a struct
definition header more clearly, and drops the number of objects
rebuilt when uapi/bpf.h is touched from 7.7k to 1.1k.

Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20211230012742.770642-3-kuba@kernel.org
2 years agonet: Add includes masked by netdevice.h including uapi/bpf.h
Jakub Kicinski [Thu, 30 Dec 2021 01:27:41 +0000 (17:27 -0800)]
net: Add includes masked by netdevice.h including uapi/bpf.h

Add missing includes unmasked by the subsequent change.

Mostly network drivers missing an include for XDP_PACKET_HEADROOM.

Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20211230012742.770642-2-kuba@kernel.org
2 years agoMerge tag 'mlx5-fixes-2021-12-28' of git://git.kernel.org/pub/scm/linux/kernel/git...
Jakub Kicinski [Thu, 30 Dec 2021 02:19:01 +0000 (18:19 -0800)]
Merge tag 'mlx5-fixes-2021-12-28' of git://git./linux/kernel/git/saeed/linux

Saeed Mahameed says:

====================
mlx5 fixes 2021-12-28

This series provides bug fixes to mlx5 driver.

* tag 'mlx5-fixes-2021-12-28' of git://git.kernel.org/pub/scm/linux/kernel/git/saeed/linux:
  net/mlx5e: Fix wrong features assignment in case of error
  net/mlx5e: TC, Fix memory leak with rules with internal port
====================

Link: https://lore.kernel.org/r/20211229065352.30178-1-saeed@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoMerge branch 'Sleepable local storage'
Alexei Starovoitov [Thu, 30 Dec 2021 01:54:41 +0000 (17:54 -0800)]
Merge branch 'Sleepable local storage'

KP Singh says:

====================

Local storage is currently unusable in sleepable helpers. One of the
important use cases of local_storage is to attach security (or
performance) contextual information to kernel objects in LSM / tracing
programs to be used later in the life-cyle of the object.

Sometimes this context can only be gathered from sleepable programs
(because it needs accesing __user pointers or helpers like
bpf_ima_inode_hash). Allowing local storage to be used from sleepable
programs allows such context to be managed with the benefits of
local_storage.

# v2 -> v3

* Fixed some RCU issues pointed by Martin
* Added Martin's ack

# v1 -> v2

* Generalize RCU checks (will send a separate patch for updating
  non local storage code where this can be used).
* Add missing RCU lock checks from v1
====================

Signed-off-by: Alexei Starovoitov <ast@kernel.org>
2 years agobpf/selftests: Update local storage selftest for sleepable programs
KP Singh [Fri, 24 Dec 2021 15:29:16 +0000 (15:29 +0000)]
bpf/selftests: Update local storage selftest for sleepable programs

Remove the spin lock logic and update the selftests to use sleepable
programs to use a mix of sleepable and non-sleepable programs. It's more
useful to test the sleepable programs since the tests don't really need
spinlocks.

Signed-off-by: KP Singh <kpsingh@kernel.org>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Martin KaFai Lau <kafai@fb.com>
Link: https://lore.kernel.org/bpf/20211224152916.1550677-3-kpsingh@kernel.org
2 years agobpf: Allow bpf_local_storage to be used by sleepable programs
KP Singh [Fri, 24 Dec 2021 15:29:15 +0000 (15:29 +0000)]
bpf: Allow bpf_local_storage to be used by sleepable programs

Other maps like hashmaps are already available to sleepable programs.
Sleepable BPF programs run under trace RCU. Allow task, sk and inode
storage to be used from sleepable programs. This allows sleepable and
non-sleepable programs to provide shareable annotations on kernel
objects.

Sleepable programs run in trace RCU where as non-sleepable programs run
in a normal RCU critical section i.e.  __bpf_prog_enter{_sleepable}
and __bpf_prog_exit{_sleepable}) (rcu_read_lock or rcu_read_lock_trace).

In order to make the local storage maps accessible to both sleepable
and non-sleepable programs, one needs to call both
call_rcu_tasks_trace and call_rcu to wait for both trace and classical
RCU grace periods to expire before freeing memory.

Paul's work on call_rcu_tasks_trace allows us to have per CPU queueing
for call_rcu_tasks_trace. This behaviour can be achieved by setting
rcupdate.rcu_task_enqueue_lim=<num_cpus> boot parameter.

In light of these new performance changes and to keep the local storage
code simple, avoid adding a new flag for sleepable maps / local storage
to select the RCU synchronization (trace / classical).

Also, update the dereferencing of the pointers to use
rcu_derference_check (with either the trace or normal RCU locks held)
with a common bpf_rcu_lock_held helper method.

Signed-off-by: KP Singh <kpsingh@kernel.org>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Martin KaFai Lau <kafai@fb.com>
Link: https://lore.kernel.org/bpf/20211224152916.1550677-2-kpsingh@kernel.org