]> git.ipfire.org Git - thirdparty/iproute2.git/log
thirdparty/iproute2.git
3 years agouapi: update vdpa.h
Stephen Hemminger [Sat, 12 Mar 2022 03:16:25 +0000 (19:16 -0800)] 
uapi: update vdpa.h

Update header from upstream.

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
3 years agoipaddress: remove 'label' compatibility with Linux-2.0 net aliases
Maxime de Roucy [Mon, 7 Mar 2022 11:40:57 +0000 (12:40 +0100)] 
ipaddress: remove 'label' compatibility with Linux-2.0 net aliases

As Linux-2.0 is getting old and systemd allows non Linux-2.0 compatible
aliases to be set, I think iproute2 should be able to manage such
aliases.

Signed-off-by: Maxime de Roucy <maxime.deroucy@gmail.com>
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
3 years agolib/fs: fix memory leak in get_task_name()
Andrea Claudi [Tue, 8 Mar 2022 17:04:56 +0000 (18:04 +0100)] 
lib/fs: fix memory leak in get_task_name()

asprintf() allocates memory which is not freed on the error path of
get_task_name(), thus potentially leading to memory leaks.
%m specifier on fscanf allocates memory, too, which needs to be freed by
the caller.

This reworks get_task_name() to avoid memory allocation.
- Pass a buffer and its length to the function, similarly to what
  get_command_name() does, thus avoiding to allocate memory for
  the string to be returned;
- Use snprintf() instead of asprintf();
- Use fgets() instead of fscanf() to limit string length.

Fixes: 81bfd01a4c9e ("lib: move get_task_name() from rdma")
Signed-off-by: Andrea Claudi <aclaudi@redhat.com>
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
3 years agouapi: update magic.h
Stephen Hemminger [Fri, 11 Mar 2022 01:23:23 +0000 (17:23 -0800)] 
uapi: update magic.h

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
3 years agordma: Fix the logic to print unsigned int.
Shangyan Zhou [Sun, 6 Mar 2022 06:56:06 +0000 (14:56 +0800)] 
rdma: Fix the logic to print unsigned int.

Use the corresponding function and fmt string to print unsigned int32
and int64.

Signed-off-by: Shangyan Zhou <sy.zhou@hotmail.com>
Reviewed-by: Leon Romanovsky <leonro@nvidia.com>
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
3 years agoRevert "rdma: Fix res_print_uint() and add res_print_u64()"
Stephen Hemminger [Mon, 7 Mar 2022 19:27:07 +0000 (11:27 -0800)] 
Revert "rdma: Fix res_print_uint() and add res_print_u64()"

This reverts commit 9d0badecea4c5e85345577984a328f38c75685c3.

3 years agordma: Fix res_print_uint() and add res_print_u64()
Shangyan Zhou [Fri, 4 Mar 2022 12:46:37 +0000 (20:46 +0800)] 
rdma: Fix res_print_uint() and add res_print_u64()

Use the corresponding function and fmt string to print unsigned int32
and int64.

Signed-off-by: Shangyan Zhou <sy.zhou@hotmail.com>
Reviewed-by: Leon Romanovsky <leonro@nvidia.com>
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
3 years agouapi: update to xfrm.h
Stephen Hemminger [Fri, 4 Mar 2022 22:26:09 +0000 (14:26 -0800)] 
uapi: update to xfrm.h

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
3 years agobridge: Remove vlan listing from `bridge link`
Benjamin Poirier [Mon, 14 Feb 2022 01:44:46 +0000 (10:44 +0900)] 
bridge: Remove vlan listing from `bridge link`

vlan listing from `bridge link -d` was broken by commit f32e4977dcb0 ("bridge:
add json support for link command"). print_vlan_info() expects to be passed a
IFLA_AF_SPEC attribute (as is done in print_vlan()) but that commit changed
the call in link.c to pass a IFLA_BRIDGE_VLAN_INFO attribute instead. As a
result, a struct bridge_vlan_info is mistakenly parsed as a struct rtattr and
print_vlan_info() usually exits early in this callpath.

The output style of print_vlan_info() (one line per vlan) is different from
the output style of `bridge link` (multiple attributes per line). The json
output is also unsuitable for `bridge link`. Since vlan listing is available
from `bridge vlan`, remove it from `bridge link` instead of trying to change
print_vlan_info().

Note that previously, bridge master devices would be included in the output
when specifying '-d' (and only in that case) but they are no longer
included because there is no detailed information to show for master
devices if we are not printing a vlan listing:
$ bridge link
4: vxlan0: <BROADCAST,MULTICAST> mtu 1500 master br0 state disabled priority 32 cost 100
$ bridge -d link
3: br0: <BROADCAST,MULTICAST> mtu 1500 master br0
4: vxlan0: <BROADCAST,MULTICAST> mtu 1500 master br0 state disabled priority 32 cost 100
    hairpin off guard off root_block off fastleave off learning on flood on mcast_flood on mcast_to_unicast off neigh_suppress off vlan_tunnel on isolated off
$ ./bridge/bridge -d link
4: vxlan0: <BROADCAST,MULTICAST> mtu 1500 master br0 state disabled priority 32 cost 100
    hairpin off guard off root_block off fastleave off learning on flood on mcast_flood on mcast_to_unicast off neigh_suppress off vlan_tunnel on isolated off

Fixes: f32e4977dcb0 ("bridge: add json support for link command")
Signed-off-by: Benjamin Poirier <bpoirier@nvidia.com>
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
3 years agobridge: Fix error string typo
Benjamin Poirier [Mon, 14 Feb 2022 01:44:45 +0000 (10:44 +0900)] 
bridge: Fix error string typo

Fixes: fab9a18a2e52 ("bridge: request vlans along with link information")
Signed-off-by: Benjamin Poirier <bpoirier@nvidia.com>
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
3 years agolnstat: fix strdup leak in -w argument parsing
Maxim Petrov [Tue, 15 Feb 2022 20:53:47 +0000 (23:53 +0300)] 
lnstat: fix strdup leak in -w argument parsing

'tmp' string is used for safe tokenizing, but it is not required after
getting all the widths in -w option. As 'tmp' string is obtained by strdup
call, the caller has to deallocate it to avoid memory leak.

Signed-off-by: Maxim Petrov <mmrmaximuzz@gmail.com>
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
3 years agoiplink_can: print_usage: typo fix, add missing spaces
Vincent Mailhol [Sat, 12 Feb 2022 13:27:27 +0000 (22:27 +0900)] 
iplink_can: print_usage: typo fix, add missing spaces

The can help menu misses three spaces for the TDCV, TDCO and TDCF
parameters, making the closing curly bracket unaligned.

For reference, before this patch:

| $ ip link help can
| Usage: ip link set DEVICE type can
|  [ bitrate BITRATE [ sample-point SAMPLE-POINT] ] |
|  [ tq TQ prop-seg PROP_SEG phase-seg1 PHASE-SEG1
|     phase-seg2 PHASE-SEG2 [ sjw SJW ] ]
|
|  [ dbitrate BITRATE [ dsample-point SAMPLE-POINT] ] |
|  [ dtq TQ dprop-seg PROP_SEG dphase-seg1 PHASE-SEG1
|     dphase-seg2 PHASE-SEG2 [ dsjw SJW ] ]
|  [ tdcv TDCV tdco TDCO tdcf TDCF ]
|
|  [ loopback { on | off } ]
|  [ listen-only { on | off } ]
|  [ triple-sampling { on | off } ]
|  [ one-shot { on | off } ]
|  [ berr-reporting { on | off } ]
|  [ fd { on | off } ]
|  [ fd-non-iso { on | off } ]
|  [ presume-ack { on | off } ]
|  [ cc-len8-dlc { on | off } ]
|  [ tdc-mode { auto | manual | off } ]
|
|  [ restart-ms TIME-MS ]
|  [ restart ]
|
|  [ termination { 0..65535 } ]
|
|  Where: BITRATE := { NUMBER in bps }
|    SAMPLE-POINT := { 0.000..0.999 }
|    TQ := { NUMBER in ns }
|    PROP-SEG := { NUMBER in tq }
|    PHASE-SEG1 := { NUMBER in tq }
|    PHASE-SEG2 := { NUMBER in tq }
|    SJW := { NUMBER in tq }
|    TDCV := { NUMBER in tc}
|    TDCO := { NUMBER in tc}
|    TDCF := { NUMBER in tc}
|    RESTART-MS := { 0 | NUMBER in ms }

... and after this patch:

| $ ip link help can
| Usage: ip link set DEVICE type can
|  [ bitrate BITRATE [ sample-point SAMPLE-POINT] ] |
|  [ tq TQ prop-seg PROP_SEG phase-seg1 PHASE-SEG1
|     phase-seg2 PHASE-SEG2 [ sjw SJW ] ]
|
|  [ dbitrate BITRATE [ dsample-point SAMPLE-POINT] ] |
|  [ dtq TQ dprop-seg PROP_SEG dphase-seg1 PHASE-SEG1
|     dphase-seg2 PHASE-SEG2 [ dsjw SJW ] ]
|  [ tdcv TDCV tdco TDCO tdcf TDCF ]
|
|  [ loopback { on | off } ]
|  [ listen-only { on | off } ]
|  [ triple-sampling { on | off } ]
|  [ one-shot { on | off } ]
|  [ berr-reporting { on | off } ]
|  [ fd { on | off } ]
|  [ fd-non-iso { on | off } ]
|  [ presume-ack { on | off } ]
|  [ cc-len8-dlc { on | off } ]
|  [ tdc-mode { auto | manual | off } ]
|
|  [ restart-ms TIME-MS ]
|  [ restart ]
|
|  [ termination { 0..65535 } ]
|
|  Where: BITRATE := { NUMBER in bps }
|    SAMPLE-POINT := { 0.000..0.999 }
|    TQ := { NUMBER in ns }
|    PROP-SEG := { NUMBER in tq }
|    PHASE-SEG1 := { NUMBER in tq }
|    PHASE-SEG2 := { NUMBER in tq }
|    SJW := { NUMBER in tq }
|    TDCV := { NUMBER in tc }
|    TDCO := { NUMBER in tc }
|    TDCF := { NUMBER in tc }
|    RESTART-MS := { 0 | NUMBER in ms }

Fixes: 0c263d7c36ff ("iplink_can: add new CAN FD bittiming parameters:
Transmitter Delay Compensat ion (TDC)")
Signed-off-by: Vincent Mailhol <mailhol.vincent@wanadoo.fr>
Acked-by: Marc Kleine-Budde <mkl@pengutronix.de>
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
3 years agodcb: Fix error reporting when accessing "dcb app"
Petr Machata [Wed, 9 Feb 2022 14:41:40 +0000 (15:41 +0100)] 
dcb: Fix error reporting when accessing "dcb app"

Currently dcb decodes the response from "dcb app add" and "del" by
interpreting the returned attribute as u8. But the value stored there is
actually a negative errno value.

Additionally, "dcb app" currently shows two sets of messages, one in
dcb_set_attribute_attr_cb() where the issue is detected, and another as a
result of error return from that function.

The current state is as follows:

# dcb app add dev swp36 dscp-prio 20:2
Error when attempting to set attribute: Unknown error 239
Attribute write: No such file or directory

Fix the "unknown error" issue by correctly decoding the attribute as i8 and
negating it. Furthermore, set errno to that value, and let the top-level
"attribute write" error message show the correct message.

Initialize errno to 0 before the dcb_talk() dispatch, and make the error
print conditional on errno != 0. This way the few error messages that are
worth describing in the place where they are detected will not cause the
second error message to be printed.

The fixed reporting looks like this:

# dcb app add dev swp36 dscp-prio 20:2
Attribute write: File exists

Reported-by: Maksym Yaremchuk <maksymy@nvidia.com>
Signed-off-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
3 years agotc: fix duplicate fall-through
Stephen Hemminger [Mon, 14 Feb 2022 21:07:49 +0000 (13:07 -0800)] 
tc: fix duplicate fall-through

Cleanup from last patch

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
3 years agolibnetlink: fix socket leak in rtnl_open_byproto()
Maxim Petrov [Tue, 8 Feb 2022 17:20:45 +0000 (20:20 +0300)] 
libnetlink: fix socket leak in rtnl_open_byproto()

rtnl_open_byproto() does not close the opened socket in case of
errors, and the socket is returned to the caller in the `fd` field of
the struct. However, none of the callers care about the socket, so
close it in the function immediately to avoid any potential resource
leaks.

Signed-off-by: Maxim Petrov <mmrmaximuzz@gmail.com>
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
3 years agotc_util: Fix parsing action control with space and slash
Roi Dayan [Thu, 3 Feb 2022 12:20:46 +0000 (14:20 +0200)] 
tc_util: Fix parsing action control with space and slash

For action police there is an conform-exceed action control
which can be for example "jump 2 / pipe".
The current parsing loop is doing one more iteration than necessary
and results in ok var being 3.

Example filter:

tc filter add dev enp8s0f0_0 ingress protocol ip prio 2 flower \
    verbose action police rate 100mbit burst 12m \
    conform-exceed jump 1 / pipe mirred egress redirect dev enp8s0f0_1 action drop

Before this change the command will fail.
Trying to add another "pipe" before mirred as a workaround for the stopping the loop
in ok var 3 resulting in result2 not being saved and wrong filter.

... conform-exceed jump 1 / pipe pipe mirred ...

Example dump of the action part:
... action order 1:  police 0x1 rate 100Mbit burst 12Mb mtu 2Kb action jump 1 overhead 0b  ...

Fix the behavior by removing redundant case 2 handling, either argc is over or breaking.

Example dump of the action part with the fix:
... action order 1:  police 0x1 rate 100Mbit burst 12Mb mtu 2Kb action jump 1/pipe overhead 0b ...

Signed-off-by: Roi Dayan <roid@nvidia.com>
Reviewed-by: Maor Dickman <maord@nvidia.com>
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
3 years agoiprule: Allow option dsfield in 'ip rule show'
Guillaume Nault [Mon, 31 Jan 2022 11:02:13 +0000 (12:02 +0100)] 
iprule: Allow option dsfield in 'ip rule show'

When the dsfield option was added to ip rule, it only worked for add
and delete operations. For consistency, allow it when dumping rules
too.

Fixes: dec01609dc62 ("iproute2: Add dsfield as alias for tos for ip rules")
Signed-off-by: Guillaume Nault <gnault@redhat.com>
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
3 years agoss: use freecon() instead of free() when appropriate
Andrea Claudi [Fri, 28 Jan 2022 10:53:58 +0000 (11:53 +0100)] 
ss: use freecon() instead of free() when appropriate

According to SELinux API, when resources are allocated using getpidcon()
of getfilecon(), they should be freed using freecon().

This commit makes ss use freecon() where appropriate, defining a stub
function executing a free() useful when iproute2 is compiled without
SELinux support.

Signed-off-by: Andrea Claudi <aclaudi@redhat.com>
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
3 years agoman: Fix a typo in the flag documentation of ip address
Thomas Niederberger [Fri, 28 Jan 2022 17:21:46 +0000 (09:21 -0800)] 
man: Fix a typo in the flag documentation of ip address

Signed-off-by: Thomas Niederberger <thomas@niederb.ch>
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
3 years agodcb: app: Add missing "dcb app show dev X default-prio"
Petr Machata [Thu, 20 Jan 2022 08:57:54 +0000 (09:57 +0100)] 
dcb: app: Add missing "dcb app show dev X default-prio"

All the actual code exists, but we neglect to recognize "default-prio" as a
CLI key for selection of what to show.

Reported-by: Maksym Yaremchuk <maksymy@nvidia.com>
Signed-off-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
3 years agouapi: update kernel headers from 5.17-rc1
Stephen Hemminger [Mon, 24 Jan 2022 17:55:39 +0000 (09:55 -0800)] 
uapi: update kernel headers from 5.17-rc1

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
3 years agotc/action: print error to stderr
Stephen Hemminger [Thu, 20 Jan 2022 20:45:49 +0000 (12:45 -0800)] 
tc/action: print error to stderr

Error messages should go to stderr even if using JSON.

Fixes: 2704bd625583 ("tc: jsonify actions core")
Cc: jiri@mellanox.com
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
3 years agomptcp: add id check for deleting address
Geliang Tang [Sat, 15 Jan 2022 16:04:33 +0000 (00:04 +0800)] 
mptcp: add id check for deleting address

This patch added the id check for deleting address in mptcp_parse_opt().
The ADDRESS argument is invalid for the non-zero id address, only needed
for the id 0 address.

 # ip mptcp endpoint delete id 1
 # ip mptcp endpoint delete id 0 10.0.1.1

Signed-off-by: Geliang Tang <geliang.tang@suse.com>
Acked-by: Mat Martineau <mathew.j.martineau@linux.intel.com>
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
3 years agodcb: Rewrite array-formatting code to not cause warnings with Clang
Petr Machata [Tue, 18 Jan 2022 11:09:30 +0000 (12:09 +0100)] 
dcb: Rewrite array-formatting code to not cause warnings with Clang

Some installation of Clang are unhappy about the use of a hand-rolled
formatting strings, and emit warnings such as this one:

dcb.c:334:31: warning: format string is not a string literal
[-Wformat-nonliteral]

Rewrite the impacted code so that it always uses literal format strings.

Reported-by: Stephen Hemminger <stephen@networkplumber.org>
Signed-off-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
3 years agof_flower: fix checkpatch warnings
Stephen Hemminger [Tue, 18 Jan 2022 05:02:50 +0000 (21:02 -0800)] 
f_flower: fix checkpatch warnings

Fix minor whitespace and other easy to fix complaints from
checkpatch.

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
3 years agonetem: fix checkpatch warnings
Stephen Hemminger [Tue, 18 Jan 2022 04:59:26 +0000 (20:59 -0800)] 
netem: fix checkpatch warnings

netem is old and pre-dates checkpatch, so fix it up.

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
3 years agolib: fix ax25.h include for musl
Sam James [Thu, 13 Jan 2022 08:14:13 +0000 (08:14 +0000)] 
lib: fix ax25.h include for musl

ax25.h isn't guaranteed to be avilable in netax25/*;
it's dependent on our choice of libc (it's not available
on musl at least) [0].

Let's use the version from linux-headers.

[0] https://sourceware.org/glibc/wiki/Synchronizing_Headers
Bug: https://bugs.gentoo.org/831102

Signed-off-by: Sam James <sam@gentoo.org>
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
3 years agouapi: add missing virtio headers
Stephen Hemminger [Sat, 15 Jan 2022 18:39:43 +0000 (10:39 -0800)] 
uapi: add missing virtio headers

These headers are referred to by virtio_net but were
not in uapi directory.

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
3 years agouapi: add missing rose and ax25 files
Stephen Hemminger [Sat, 15 Jan 2022 18:38:58 +0000 (10:38 -0800)] 
uapi: add missing rose and ax25 files

All iproute2 headers should come from sanitized kernel headers
not the local system.

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
3 years agoq_cake: allow changing to diffserv3
Kevin Bracey [Thu, 6 Jan 2022 11:16:04 +0000 (13:16 +0200)] 
q_cake: allow changing to diffserv3

A diffserv3 option (enum value 0) was never sent to the kernel, so it
was not possible to use "tc qdisc change" to select it.

This also meant that were also relying on the kernel's default being
diffserv3 when adding. If the default were to change, we wouldn't have
been able to request diffserv3 explicitly.

Signed-off-by: Kevin Bracey <kevin@bracey.fi>
Acked-by: Toke Høiland-Jørgensen <toke@toke.dk>
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
3 years agoiplink_can: add ctrlmode_{supported,_static} to the "--details --json" output
Vincent Mailhol [Sun, 9 Jan 2022 15:30:40 +0000 (00:30 +0900)] 
iplink_can: add ctrlmode_{supported,_static} to the "--details --json" output

This patch is the userland counterpart of [1]. Indeed, [1] enables the
can netlink interface to report the CAN controller capabilities.

Previously, only the options which were switched on were reported
(i.e. can_priv::ctrlmode). Here, we add two additional pieces of
information to the json report:

  - ctrlmode_supported: the options that can be modified by netlink

  - ctrlmode_static: option which are statically enabled by the driver
    (i.e. can not be turned off)

For your information, we borrowed the naming convention from struct
can_priv [2].

Contrary to the ctrlmode, the ctrlmode_{supported,_static} are only
reported in the json context. The reason is that this newly added
information can quickly become very verbose and we do not want to
overload the default output. You can think of the "ip --details link
show canX" output as the verbose mode and the "ip --details --json
link show canX" output as the *very* verbose mode.

*Example:*

This is how the output would look like for a dummy driver which would
have:

  - CAN_CTRLMODE_LOOPBACK, CAN_CTRLMODE_LISTENONLY,
    CAN_CTRLMODE_3_SAMPLES, CAN_CTRLMODE_FD, CAN_CTRLMODE_CC_LEN8_DLC
    and TDC-AUTO supported by the driver

  - CAN_CTRLMODE_CC_LEN8_DLC turned on by the user

  - CAN_CTRLMODE_FD_NON_ISO statically enabled by the driver

| $ ip link set can0 type can cc-len8-dlc on
| $ ip --details --json --pretty link show can0
| [ {
|         "ifindex": 1,
|         "ifname": "can0",
|         "flags": [ "NOARP","ECHO" ],
|         "mtu": 16,
|         "qdisc": "noop",
|         "operstate": "DOWN",
|         "linkmode": "DEFAULT",
|         "group": "default",
|         "txqlen": 10,
|         "link_type": "can",
|         "promiscuity": 0,
|         "min_mtu": 0,
|         "max_mtu": 0,
|         "linkinfo": {
|             "info_kind": "can",
|             "info_data": {
|                 "ctrlmode": [ "FD-NON-ISO","CC-LEN8-DLC" ],
|                 "ctrlmode_supported": [ "LOOPBACK","LISTEN-ONLY","TRIPLE-SAMPLING","FD","CC-LEN8-DLC","TDC-AUTO" ],
|                 "ctrlmode_static": [ "FD-NON-ISO" ],
|                 "state": "STOPPED",
|                 "restart_ms": 0,
|                 "bittiming_const": {
|                     "name": "DUMMY_CAN_DEV",
|                     "tseg1": {
|                         "min": 2,
|                         "max": 256
|                     },
|                     "tseg2": {
|                         "min": 2,
|                         "max": 128
|                     },
|                     "sjw": {
|                         "min": 1,
|                         "max": 128
|                     },
|                     "brp": {
|                         "min": 1,
|                         "max": 512
|                     },
|                     "brp_inc": 1
|                 },
|                 "data_bittiming_const": {
|                     "name": "DUMMY_CAN_DEV",
|                     "tseg1": {
|                         "min": 2,
|                         "max": 32
|                     },
|                     "tseg2": {
|                         "min": 1,
|                         "max": 16
|                     },
|                     "sjw": {
|                         "min": 1,
|                         "max": 8
|                     },
|                     "brp": {
|                         "min": 1,
|                         "max": 32
|                     },
|                     "brp_inc": 1,
|                     "tdc": {
|                         "tdco": {
|                             "min": 0,
|                             "max": 127
|                         },
|                         "tdcf": {
|                             "min": 0,
|                             "max": 127
|                         }
|                     }
|                 },
|                 "clock": 80000000
|             }
|         },
|         "num_tx_queues": 1,
|         "num_rx_queues": 1,
|         "gso_max_size": 65536,
|         "gso_max_segs": 65535,
|         "parentbus": "usb",
|         "parentdev": "1-10:1.1"
|     } ]

As mentioned above, the default output remains unchanged:

| $ ip --details link show can0
| 1: can0: <NOARP,ECHO> mtu 16 qdisc noop state DOWN mode DEFAULT group default qlen 10
|     link/can  promiscuity 0 minmtu 0 maxmtu 0
|     can <FD-NON-ISO,CC-LEN8-DLC> state STOPPED restart-ms 0
|    DUMMY_CAN_DEV: tseg1 2..256 tseg2 2..128 sjw 1..128 brp 1..512 brp_inc 1
|    DUMMY_CAN_DEV: dtseg1 2..32 dtseg2 1..16 dsjw 1..8 dbrp 1..32 dbrp_inc 1
|    tdco 0..127 tdcf 0..127
|    clock 80000000 numtxqueues 1 numrxqueues 1 gso_max_size 65536 gso_max_segs 65535 parentbus usb parentdev 1-10:1.1

[1] commit 383f0993fc77 ("can: netlink: report the CAN controller mode
    supported flags")
    https://lore.kernel.org/linux-can/20220105144402.1174191-16-mkl@pengutronix.de/T/#u

[2] https://elixir.bootlin.com/linux/v5.14/source/include/linux/can/dev.h#61

Signed-off-by: Vincent Mailhol <mailhol.vincent@wanadoo.fr>
Signed-off-by: David Ahern <dsahern@kernel.org>
3 years agoUpdate kernel headers
David Ahern [Wed, 12 Jan 2022 17:31:20 +0000 (10:31 -0700)] 
Update kernel headers

Update kernel headers to commit:
    fe8152b38d3a ("Merge tag 'devprop-5.17-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm")

Signed-off-by: David Ahern <dsahern@kernel.org>
3 years agoMerge branch 'rdma-clang-compile' into next
David Ahern [Tue, 11 Jan 2022 16:20:02 +0000 (09:20 -0700)] 
Merge branch 'rdma-clang-compile' into next

Signed-off-by: David Ahern <dsahern@kernel.org>
3 years agordma: Don't allocate sparse array
Leon Romanovsky [Sun, 9 Jan 2022 18:41:39 +0000 (20:41 +0200)] 
rdma: Don't allocate sparse array

The addition of driver QP type with index 0xFF caused to the following
clang compilation error:

res.c:152:10: warning: result of comparison of constant 256 with expression of type 'uint8_t' (aka 'unsigned char') is always true [-Wtautological-constant-out-of-range-compare]
        if (idx < ARRAY_SIZE(qp_types_str) && qp_types_str[idx])
            ~~~ ^ ~~~~~~~~~~~~~~~~~~~~~~~~

Instead of allocating very sparse array, simply create separate check
for the driver QP type.

Fixes: 39307384cea7 ("rdma: Add driver QP type string")
Reported-by: Stephen Hemminger <stephen@networkplumber.org>
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
Signed-off-by: David Ahern <dsahern@kernel.org>
3 years agordma: Limit copy data by the destination size
Leon Romanovsky [Sun, 9 Jan 2022 18:41:38 +0000 (20:41 +0200)] 
rdma: Limit copy data by the destination size

The strncat() function will copy upto n bytes supplied as third
argument. The n bytes shouldn't be no more than destination and
not the source.

This change fixes the following clang compilation warnings:

res-srq.c:75:25: warning: size argument in 'strncat' call appears to be size of the source [-Wstrncat-size]
                        strncat(qp_str, tmp, sizeof(tmp) - 1);
                                             ^~~~~~~~~~~~~~~
res-srq.c:99:23: warning: size argument in 'strncat' call appears to be size of the source [-Wstrncat-size]
        strncat(qp_str, tmp, sizeof(tmp) - 1);
                             ^~~~~~~~~~~~~~~
res-srq.c:142:25: warning: size argument in 'strncat' call appears to be size of the source [-Wstrncat-size]
                        strncat(qp_str, tmp, sizeof(tmp) - 1);
                                             ^~~~~~~~~~~~~~~

Fixes: 9b272e138d23 ("rdma: Add SRQ resource tracking information")
Reported-by: Stephen Hemminger <stephen@networkplumber.org>
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
Signed-off-by: David Ahern <dsahern@kernel.org>
3 years agoMerge branch 'main' into next
David Ahern [Tue, 11 Jan 2022 16:16:35 +0000 (09:16 -0700)] 
Merge branch 'main' into next

Signed-off-by: David Ahern <dsahern@kernel.org>
3 years agov5.16.0 v5.16.0
Stephen Hemminger [Mon, 10 Jan 2022 23:24:04 +0000 (15:24 -0800)] 
v5.16.0

3 years agotestsuite: Fix tc/vlan.t test
Andrea Claudi [Mon, 3 Jan 2022 18:00:22 +0000 (19:00 +0100)] 
testsuite: Fix tc/vlan.t test

Following commit 8323b20f1d76 ("net/sched: act_vlan: No dump for unset
priority"), the kernel no longer dump vlan priority if not explicitly
set before.

When modifying a vlan, tc/vlan.t test expects to find priority set to 0
without setting it explicitly. Thus, after 8323b20f1d76 this test fails.

Fix this simply removing the check on priority.

Signed-off-by: Andrea Claudi <aclaudi@redhat.com>
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
3 years agoMerge branch 'vdpa-query-set-config' into next
David Ahern [Sat, 18 Dec 2021 21:10:08 +0000 (14:10 -0700)] 
Merge branch 'vdpa-query-set-config' into next

Parav Pandit  says:

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

This series implements querying and setting of the mac address and mtu
device config fields of the vdpa device of type net.

An example of query and set as below.

$ vdpa dev add name bar mgmtdev vdpasim_net mac 00:11:22:33:44:55 mtu 9000

$ vdpa dev config show
bar: mac 00:11:22:33:44:55 link up link_announce false mtu 9000

$ vdpa dev config show -jp
{
    "config": {
        "bar": {
            "mac": "00:11:22:33:44:55",
            "link ": "up",
            "link_announce ": false,
            "mtu": 1500,
        }
    }
}

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

Signed-off-by: David Ahern <dsahern@kernel.org>
3 years agovdpa: Enable user to set mtu of the vdpa device
Parav Pandit [Fri, 17 Dec 2021 08:08:27 +0000 (10:08 +0200)] 
vdpa: Enable user to set mtu of the vdpa device

Implement mtu setting for vdpa device.

$ vdpa mgmtdev show
vdpasim_net:
  supported_classes net

Add the device with mac address and mtu:
$ vdpa dev add name bar mgmtdev vdpasim_net mac 00:11:22:33:44:55 mtu 9000

In above command only mac address or only mtu can also be set.

View the config after setting:
$ vdpa dev config show
bar: mac 00:11:22:33:44:55 link up link_announce false mtu 9000

Signed-off-by: Parav Pandit <parav@nvidia.com>
Signed-off-by: David Ahern <dsahern@kernel.org>
3 years agovdpa: Enable user to set mac address of vdpa device
Parav Pandit [Fri, 17 Dec 2021 08:08:26 +0000 (10:08 +0200)] 
vdpa: Enable user to set mac address of vdpa device

vdpa: Enable user to set mtu of the vdpa device

Implement mtu setting for vdpa device.

$ vdpa mgmtdev show
vdpasim_net:
  supported_classes net

Add the device with specified mac address:
$ vdpa dev add name bar mgmtdev vdpasim_net mac 00:11:22:33:44:55

View the config after setting:
$ vdpa dev config show
bar: mac 00:11:22:33:44:55 link up link_announce false mtu 1500

Signed-off-by: Parav Pandit <parav@nvidia.com>
Signed-off-by: David Ahern <dsahern@kernel.org>
3 years agovdpa: Enable user to query vdpa device config layout
Parav Pandit [Fri, 17 Dec 2021 08:08:25 +0000 (10:08 +0200)] 
vdpa: Enable user to query vdpa device config layout

Query the device configuration layout whenever kernel supports it.

An example of configuration layout of vdpa device of type network:

$ vdpa dev add name bar mgmtdev vdpasim_net

$ vdpa dev config show
bar: mac 00:35:09:19:48:05 link up link_announce false mtu 1500

$ vdpa dev config show -jp
{
    "config": {
        "bar": {
            "mac": "00:35:09:19:48:05",
            "link ": "up",
            "link_announce ": false,
            "mtu": 1500,
        }
    }
}

Signed-off-by: Parav Pandit <parav@nvidia.com>
Signed-off-by: David Ahern <dsahern@kernel.org>
3 years agovdpa: Update kernel headers
Parav Pandit [Fri, 17 Dec 2021 08:08:24 +0000 (10:08 +0200)] 
vdpa: Update kernel headers

Update kernel headers to commit:
ad69dd0bf26b ("vdpa: Introduce query of device config layout")

Signed-off-by: Parav Pandit <parav@nvidia.com>
Signed-off-by: David Ahern <dsahern@kernel.org>
3 years agoUpdate kernel headers and import virtio_net
David Ahern [Sat, 18 Dec 2021 21:00:29 +0000 (14:00 -0700)] 
Update kernel headers and import virtio_net

Update kernel headers to commit:
    f85b244ee395 ("xdp: move the if dev statements to the first")

and import virtio_net.h for vdpa.

Signed-off-by: David Ahern <dsahern@kernel.org>
3 years agomptcp: add support for changing the backup flag
Davide Caratti [Thu, 16 Dec 2021 14:29:59 +0000 (15:29 +0100)] 
mptcp: add support for changing the backup flag

Linux supports 'MPTCP_PM_CMD_SET_FLAGS' since v5.12, and this control has
recently been extended to allow setting flags for a given endpoint id.
Although there is no use for changing 'signal' or 'subflow' flags, it can
be helpful to set/clear the backup bit on existing endpoints: add the 'ip
mptcp endpoint change <...>' command for this purpose.

Link: https://github.com/multipath-tcp/mptcp_net-next/issues/158
Acked-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Signed-off-by: Davide Caratti <dcaratti@redhat.com>
Signed-off-by: David Ahern <dsahern@kernel.org>
3 years agouapi: update to mptcp.h
Stephen Hemminger [Fri, 17 Dec 2021 23:24:54 +0000 (15:24 -0800)] 
uapi: update to mptcp.h

From 5.16-rc5

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
3 years agolib/bpf: fix verbose flag when using libbpf
Paul Chaignon [Thu, 16 Dec 2021 15:33:36 +0000 (16:33 +0100)] 
lib/bpf: fix verbose flag when using libbpf

Since commit 6d61a2b55799 ("lib: add libbpf support"), passing the
verbose flag to tc filter doesn't dump the verifier logs anymore in case
of successful loading.

This commit fixes it by setting the log_level attribute before loading.
To that end, we need to call bpf_object__load_xattr directly instead of
relying on bpf_object__load.

Fixes: 6d61a2b55799 ("lib: add libbpf support")
Signed-off-by: Paul Chaignon <paul@isovalent.com>
Acked-by: Hangbin Liu <haliu@redhat.com>
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
3 years agotc: Add support for ce_threshold_value/mask in fq_codel
Toke Høiland-Jørgensen [Wed, 8 Dec 2021 12:45:17 +0000 (13:45 +0100)] 
tc: Add support for ce_threshold_value/mask in fq_codel

Commit dfcb63ce1de6 ("fq_codel: generalise ce_threshold marking for subset
of traffic") added support in fq_codel for setting a value and mask that
will be applied to the diffserv/ECN byte to turn on the ce_threshold
feature for a subset of traffic.

This adds support to iproute for setting these values. The parameter is
called ce_threshold_selector and takes a value followed by a
slash-separated mask. Some examples:

 # apply ce_threshold to ECT(1) traffic
 tc qdisc replace dev eth0 root fq_codel ce_threshold 1ms ce_threshold_selector 0x1/0x3

 # apply ce_threshold to ECN-capable traffic marked as diffserv AF22
 tc qdisc replace dev eth0 root fq_codel ce_threshold 1ms ce_threshold_selector 0x50/0xfc

Signed-off-by: Toke Høiland-Jørgensen <toke@redhat.com>
Signed-off-by: David Ahern <dsahern@kernel.org>
3 years agotc: flower: Fix buffer overflow on large labels
Paul Blakey [Sun, 5 Dec 2021 13:20:25 +0000 (15:20 +0200)] 
tc: flower: Fix buffer overflow on large labels

Buffer is 64bytes, but label printing can take 66bytes printing
in hex, and will overflow when setting the string delimiter ('\0').

Fix that by increasing the print buffer size.

Example of overflowing ct_label:
ct_label 11111111111111111111111111111111/11111111111111111111111111111111

Fixes: 2fffb1c03056 ("tc: flower: Add matching on conntrack info")
Signed-off-by: Paul Blakey <paulb@nvidia.com>
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
3 years agouapi: update to if_ether.h
Stephen Hemminger [Fri, 3 Dec 2021 20:20:02 +0000 (12:20 -0800)] 
uapi: update to if_ether.h

Merged from 5.16-rc3

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
3 years agobond: add arp_missed_max option
Hangbin Liu [Tue, 30 Nov 2021 04:29:48 +0000 (12:29 +0800)] 
bond: add arp_missed_max option

Bond arp_missed_max is the maximum number of arp_interval monitor cycle
for missed ARP replies. If this number is exceeded, link is reported as
down.

Signed-off-by: Hangbin Liu <liuhangbin@gmail.com>
Signed-off-by: David Ahern <dsahern@kernel.org>
3 years agomptcp: add support for fullmesh flag
Paolo Abeni [Fri, 26 Nov 2021 10:35:44 +0000 (11:35 +0100)] 
mptcp: add support for fullmesh flag

The link kernel supports this endpoint flag since v5.15, let's
expose it to user-space. It allows creation on fullmesh topolgy
via MPTCP subflow.

Additionally update the related man-page, clarifying the behavior
of related options.

Acked-by: Mat Martineau <mathew.j.martineau@linux.intel.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: David Ahern <dsahern@kernel.org>
3 years agoUpdate kernel headers
David Ahern [Fri, 3 Dec 2021 17:17:51 +0000 (10:17 -0700)] 
Update kernel headers

Update kernel headers to commit:
    bb14bfc7eb92 ("net: lan966x: fix a IS_ERR() vs NULL check in lan966x_create_targets()")

Signed-off-by: David Ahern <dsahern@kernel.org>
3 years agoip/ipnexthop: fix unsigned overflow in parse_nh_group_type_res()
Maxim Petrov [Wed, 17 Nov 2021 19:11:24 +0000 (22:11 +0300)] 
ip/ipnexthop: fix unsigned overflow in parse_nh_group_type_res()

0UL has type 'unsigned long' which is likely to be 64bit on modern machines. At
the same time, the '{idle,unbalanced}_timer' variables are declared as u32, so
these variables cannot be greater than '~0UL / 100' when 'unsigned long' is 64
bits. In such condition it is still possible to pass the check but get the
overflow later when the timers are multiplied by 100 in 'addattr32'.

Fix the possible overflow by changing '~0UL' to 'UINT32_MAX'.

Fixes: 91676718228b ("nexthop: Add support for resilient nexthop groups")
Signed-off-by: Maxim Petrov <mmrmaximuzz@gmail.com>
Reviewed-by: Ido Schimmel <idosch@nvidia.com>
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
3 years agolib/bpf_legacy: remove always-true check
Maxim Petrov [Tue, 16 Nov 2021 19:32:26 +0000 (22:32 +0300)] 
lib/bpf_legacy: remove always-true check

The 'name' field of the 'struct bpf_prog_info' is a plain C array. Thus, the
logical condition in bpf_dump_prog_info() is useless as the array address is
always true, so just remove it.

Signed-off-by: Maxim Petrov <mmrmaximuzz@gmail.com>
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
3 years agordma: update uapi headers
Stephen Hemminger [Thu, 18 Nov 2021 18:00:19 +0000 (10:00 -0800)] 
rdma: update uapi headers

Update the RDMA uapi headers from 5.16.0-rc1

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
3 years agovdpa: align uapi headers
Stephen Hemminger [Thu, 18 Nov 2021 17:56:57 +0000 (09:56 -0800)] 
vdpa: align uapi headers

Update vdpa headers based on 5.16.0-rc1 and remove redundant
copy.

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
3 years agovdpa: Remove duplicate vdpa UAPI header file
Parav Pandit [Sat, 6 Nov 2021 06:41:52 +0000 (08:41 +0200)] 
vdpa: Remove duplicate vdpa UAPI header file

vdpa header file is already present in the tree at
vdpa/include/uapi/linux/vdpa.h and used by vdpa/vdpa.c.

As we discussed in thread [1] vdpa header comes from a different
tree, similar to rdma subsystem. Hence remove the duplicate vdpa
UAPI header file.

[1] https://www.spinics.net/lists/netdev/msg748458.html

Fixes: b5a6ed9cc9fc ("uapi: add missing virtio related headers")
Signed-off-by: Parav Pandit <parav@nvidia.com>
Signed-off-by: David Ahern <dsahern@gmail.com>
3 years agolnstat: fix buffer overflow in header output
[200~jiangheng [Wed, 17 Nov 2021 21:41:10 +0000 (13:41 -0800)] 
lnstat: fix buffer overflow in header output

Running lnstat will cause core dump from reading past end of array.

Segmentation fault (core dumped)

The maximum  value of th.num_lines is HDR_LINES(10),  h should not be equal to th.num_lines, array th.hdr may be out of bounds.

Signed-off-by jiangheng <jiangheng12@huawei.com>
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
3 years agotc/m_vlan: fix print_vlan() conditional on TCA_VLAN_ACT_PUSH_ETH
Maxim Petrov [Wed, 17 Nov 2021 18:05:33 +0000 (21:05 +0300)] 
tc/m_vlan: fix print_vlan() conditional on TCA_VLAN_ACT_PUSH_ETH

Fix the wild bracket in the if clause leading to the error in the condition.

Fixes: d61167dd88b4 ("m_vlan: add pop_eth and push_eth actions")
Signed-off-by: Maxim Petrov <mmrmaximuzz@gmail.com>
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
3 years agomptcp: fix JSON output when dumping endpoints by id
Davide Caratti [Thu, 11 Nov 2021 09:52:13 +0000 (10:52 +0100)] 
mptcp: fix JSON output when dumping endpoints by id

iproute ignores '-j' command line argument when dumping endpoints by id:

 [dcaratti@dcaratti iproute2]$ ./ip/ip -j mptcp endpoint show
 [{"address":"1.2.3.4","id":42,"signal":true,"backup":true}]
 [dcaratti@dcaratti iproute2]$ ./ip/ip -j mptcp endpoint show id 42
 1.2.3.4 id 42 signal backup

fix mptcp_addr_show() to use the proper JSON helpers.

Fixes: 7e0767cd862b ("add support for mptcp netlink interface")
Signed-off-by: Davide Caratti <dcaratti@redhat.com>
Acked-by: Andrea Claudi <aclaudi@redhat.com>
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
3 years agoman: tc-u32: Fix page to match new firstfrag behavior
Anssi Hannula [Thu, 4 Nov 2021 14:42:05 +0000 (16:42 +0200)] 
man: tc-u32: Fix page to match new firstfrag behavior

Commit 690b11f4a6b8 ("tc: u32: Fix firstfrag filter.") applied in 2012
changed the "ip firstfrag" selector to not match non-fragmented packets
anymore.

However, the documentation added in f15a23966fff ("tc: add a man page
for u32 filter") in 2015 includes an example that relies on the previous
behavior (non-fragmented packet counted as first fragment).

Due to this, the example does not work correctly and does not actually
classify regular SSH packets.

Modify the example to use a raw u16 selector on the fragment offset to
make it work, and also make the firstfrag description more clear about
the current behavior.

Fixes: f15a23966fff ("tc: add a man page for u32 filter")
Signed-off-by: Anssi Hannula <anssi.hannula@bitwise.fi>
Cc: Phil Sutter <phil@nwl.cc>
Cc: Hiroaki SHIMODA <shimoda.hiroaki@gmail.com>
Acked-by: Phil Sutter <phil@nwl.cc>
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
3 years agoFix some typos detected by Lintian in manpages
Luca Boccassi [Fri, 5 Nov 2021 23:36:02 +0000 (23:36 +0000)] 
Fix some typos detected by Lintian in manpages

Signed-off-by: Luca Boccassi <bluca@debian.org>
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
3 years agouapi: update vdpa.h
Stephen Hemminger [Tue, 9 Nov 2021 18:40:40 +0000 (10:40 -0800)] 
uapi: update vdpa.h

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
3 years agoMerge branch 'main' into next
David Ahern [Thu, 4 Nov 2021 15:45:31 +0000 (09:45 -0600)] 
Merge branch 'main' into next
Signed-off-by: David Ahern <dsahern@kernel.org>
3 years agoMerge branch 'can-tdc-plus-cleanups' into next
David Ahern [Thu, 4 Nov 2021 15:44:56 +0000 (09:44 -0600)] 
Merge branch 'can-tdc-plus-cleanups' into next

Vincent Mailhol  says:

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

The main purpose is to add commandline support for Transmitter Delay
Compensation (TDC) in iproute. Other issues found during the
development of this feature also get addressed.

This patch series contains four patches which respectively:

  1. Correct the bittiming ranges in the print_usage function and add
  the units to give more clarity: some parameters are in milliseconds,
  some in nano seconds, some in time quantum and the newly TDC
  parameters introduced in this series would be in clock period.

  2. Do some code refactoring on function print_ctrlmode().

  3. factorize the many print_*(PRINT_JSON, ...) and fprintf
  occurrences in a single print_*(PRINT_ANY, ...) call and fix the
  signedness while doing that.

  4. report the value of the bitrate prescalers (brp and dbrp).

  5. adds command line support for the TDC in iproute and goes together
  with below series in the kernel:
  https://lore.kernel.org/linux-can/20210814091750.73931-1-mailhol.vincent@wanadoo.fr/T/#t

** Changelog **

>From RFC v5 to v6:
  * Dropped the RFC tag because the related patch series on the kernel
    side were pulled into net-next.
  * Remove the changes in include/uapi/linux/can/netlink.h because
    these should be pulled separately.
  * Add another patch (the second of this series) to do some cleanup
    on function print_ctrlmode().
  * Minor fixes in the patch comments (grammar, rephrasing).

>From RFC v4 to RFC v5:
  * Add the unit (bps, tq, ns or ms) in print_usage()
  * Rewrote void can_print_timing_min_max() to better factorize the
    code.
  * Rewrote the commit message of the two last patches (those related
    to TDC) to either add clarification of fix inacurracies.

>From v3 to RFC v4:
  * Reflect the changes made on the kernel side.

>From RFC v2 to v3:
  * Dropped the RFC tag. Now that the kernel patch reach the testing
    branch, I am finaly ready.
  * Regression fix: configuring a link with only nominal bittiming
    returned -EOPNOTSUPP
  * Added two more patches to the series:
      - iplink_can: fix configuration ranges in print_usage()
      - iplink_can: print brp and dbrp bittiming variables
  * Other small fixes on formatting.

>From RFC v1 to RFC v2:
  * Add an additional patch to the series to fix the issues reported
    by Stephen Hemminger
    Ref: https://lore.kernel.org/linux-can/20210506112007.1666738-1-mailhol.vincent@wanadoo.fr/T/#t

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

Signed-off-by: David Ahern <dsahern@kernel.org>
3 years agoiplink_can: add new CAN FD bittiming parameters: Transmitter Delay Compensation ...
Vincent Mailhol [Wed, 3 Nov 2021 16:44:28 +0000 (01:44 +0900)] 
iplink_can: add new CAN FD bittiming parameters: Transmitter Delay Compensation (TDC)

At high bit rates, the propagation delay from the TX pin to the RX pin
of the transceiver causes measurement errors: the sample point on the
RX pin might occur on the previous bit.

This issue is addressed in ISO 11898-1 section 11.3.3 "Transmitter
delay compensation" (TDC).

This patch brings command line support to nine TDC parameters which
were recently added to the kernel's CAN netlink interface in order to
implement TDC:
  - IFLA_CAN_TDC_TDCV_MIN: Transmitter Delay Compensation Value
    minimum value
  - IFLA_CAN_TDC_TDCV_MAX: Transmitter Delay Compensation Value
    maximum value
  - IFLA_CAN_TDC_TDCO_MIN: Transmitter Delay Compensation Offset
    minimum value
  - IFLA_CAN_TDC_TDCO_MAX: Transmitter Delay Compensation Offset
    maximum value
  - IFLA_CAN_TDC_TDCF_MIN: Transmitter Delay Compensation Filter
    window minimum value
  - IFLA_CAN_TDC_TDCF_MAX: Transmitter Delay Compensation Filter
    window maximum value
  - IFLA_CAN_TDC_TDCV: Transmitter Delay Compensation Value
  - IFLA_CAN_TDC_TDCO: Transmitter Delay Compensation Offset
  - IFLA_CAN_TDC_TDCF: Transmitter Delay Compensation Filter window

All those new parameters are nested together into the attribute
IFLA_CAN_TDC.

The TDC parameters extend the FD parameters. As such, the TDC
parameters must be specified together the "fd on" flag.

When "fd on" flag is provided, a tdc-mode parameter allows to specify
how to operate.  Valid options for tdc-mode are:

  * auto: the transmitter dynamically measures TDCV for each of the
    transmitted frames. As such, TDCV can not be manually provided. In
    this mode, the user must specify TDCO and may also specify TDCF if
    supported.

  * manual: use a static TDCV provided by the user. In this mode, the
    user must specify both TDCV and TDCO and may also specify TDCF if
    supported.

  * off: TDC is explicitly disabled.

  * tdc-mode parameter omitted (default mode): the kernel decides
    whether TDC should be enabled or not and if so, it calculates the
    TDC values. TDC parameters are an expert option and the average
    user is not expected to provide those, thus the presence of this
    "default mode".

If the fd flag is omitted, all the FD values (including TDC values)
remain unchanged.

If "fd off" flag is specified, all FD values (including TDC values)
are zeroed.

TDCV is always reported in manual mode. In auto mode, TDCV is reported
only if the value is available. Especially, the TDCV might not be
available if the controller has no feature to report it or if the
value in not yet available (i.e. no data sent yet and measurement did
not occur).

TDCF is reported only if tdcf_max is not zero (i.e. if supported by
the controller).

For reference, here are a few samples of how the output looks like:

| $ ip link set can0 type can bitrate 1000000 dbitrate 8000000 fd on tdco 7 tdcf 8 tdc-mode auto

| $ ip --details link show can0
| 1:  can0: <NOARP,ECHO> mtu 72 qdisc noop state DOWN mode DEFAULT group default qlen 10
|     link/can  promiscuity 0 minmtu 0 maxmtu 0
|     can <FD,TDC-AUTO> state STOPPED (berr-counter tx 0 rx 0) restart-ms 0
|    bitrate 1000000 sample-point 0.750
|    tq 12 prop-seg 29 phase-seg1 30 phase-seg2 20 sjw 1 brp 1
|    ES582.1/ES584.1: tseg1 2..256 tseg2 2..128 sjw 1..128 brp 1..512 brp_inc 1
|    dbitrate 8000000 dsample-point 0.700
|    dtq 12 dprop-seg 3 dphase-seg1 3 dphase-seg2 3 dsjw 1 dbrp 1
|    tdco 7 tdcf 8
|    ES582.1/ES584.1: dtseg1 2..32 dtseg2 1..16 dsjw 1..8 dbrp 1..32 dbrp_inc 1
|    tdco 0..127 tdcf 0..127
|    clock 80000000 numtxqueues 1 numrxqueues 1 gso_max_size 65536 gso_max_segs 65535

| $ ip --details --json --pretty link show can0
| [ {
|         "ifindex": 1,
|         "ifname": "can0",
|         "flags": [ "NOARP","ECHO" ],
|         "mtu": 72,
|         "qdisc": "noop",
|         "operstate": "DOWN",
|         "linkmode": "DEFAULT",
|         "group": "default",
|         "txqlen": 10,
|         "link_type": "can",
|         "promiscuity": 0,
|         "min_mtu": 0,
|         "max_mtu": 0,
|         "linkinfo": {
|             "info_kind": "can",
|             "info_data": {
|                 "ctrlmode": [ "FD","TDC-AUTO" ],
|                 "state": "STOPPED",
|                 "berr_counter": {
|                     "tx": 0,
|                     "rx": 0
|                 },
|                 "restart_ms": 0,
|                 "bittiming": {
|                     "bitrate": 1000000,
|                     "sample_point": "0.750",
|                     "tq": 12,
|                     "prop_seg": 29,
|                     "phase_seg1": 30,
|                     "phase_seg2": 20,
|                     "sjw": 1,
|                     "brp": 1
|                 },
|                 "bittiming_const": {
|                     "name": "ES582.1/ES584.1",
|                     "tseg1": {
|                         "min": 2,
|                         "max": 256
|                     },
|                     "tseg2": {
|                         "min": 2,
|                         "max": 128
|                     },
|                     "sjw": {
|                         "min": 1,
|                         "max": 128
|                     },
|                     "brp": {
|                         "min": 1,
|                         "max": 512
|                     },
|                     "brp_inc": 1
|                 },
|                 "data_bittiming": {
|                     "bitrate": 8000000,
|                     "sample_point": "0.700",
|                     "tq": 12,
|                     "prop_seg": 3,
|                     "phase_seg1": 3,
|                     "phase_seg2": 3,
|                     "sjw": 1,
|                     "brp": 1,
|                     "tdc": {
|                         "tdco": 7,
|                         "tdcf": 8
|                     }
|                 },
|                 "data_bittiming_const": {
|                     "name": "ES582.1/ES584.1",
|                     "tseg1": {
|                         "min": 2,
|                         "max": 32
|                     },
|                     "tseg2": {
|                         "min": 1,
|                         "max": 16
|                     },
|                     "sjw": {
|                         "min": 1,
|                         "max": 8
|                     },
|                     "brp": {
|                         "min": 1,
|                         "max": 32
|                     },
|                     "brp_inc": 1,
|                     "tdc": {
|                         "tdco": {
|                             "min": 0,
|                             "max": 127
|                         },
|                         "tdcf": {
|                             "min": 0,
|                             "max": 127
|                         }
|                     }
|                 },
|                 "clock": 80000000
|             }
|         },
|         "num_tx_queues": 1,
|         "num_rx_queues": 1,
|         "gso_max_size": 65536,
|         "gso_max_segs": 65535
|     } ]

Signed-off-by: Vincent Mailhol <mailhol.vincent@wanadoo.fr>
Signed-off-by: David Ahern <dsahern@kernel.org>
3 years agoiplink_can: print brp and dbrp bittiming variables
Vincent Mailhol [Wed, 3 Nov 2021 16:44:27 +0000 (01:44 +0900)] 
iplink_can: print brp and dbrp bittiming variables

Report the value of the bit-rate prescaler (brp) for both the nominal
and the data bittiming.

Currently, only the constant brp values (brp_{min,max,inc}) are being
reported. Also, brp is the only member of struct can_bittiming not
being reported.

Noticeably, brp could be calculated by hand from the other bittiming
parameters with below formula:

        brp = clock * tq / 1000000000

with clock in hertz and tq in nano second (thus the need of a 1
billion factor to convert it back to second).

But because above formula is not so trivial to remember and is
subjected to rounding errors, it makes sense to directly output
{d,}bpr.

Signed-off-by: Vincent Mailhol <mailhol.vincent@wanadoo.fr>
Signed-off-by: David Ahern <dsahern@kernel.org>
3 years agoiplink_can: use PRINT_ANY to factorize code and fix signedness
Vincent Mailhol [Wed, 3 Nov 2021 16:44:26 +0000 (01:44 +0900)] 
iplink_can: use PRINT_ANY to factorize code and fix signedness

Current implementation heavily relies on some "if (is_json_context())"
switches to decide the context and then does some print_*(PRINT_JSON,
...) when in json context and some fprintf(...) else.

Furthermore, current implementation uses either print_int() or the
conversion specifier %d to print unsigned integers.

This patch factorizes each pairs of print_*(PRINT_JSON, ...) and
fprintf() into a single print_*(PRINT_ANY, ...) call. While doing this
replacement, it uses proper unsigned function print_uint() as well as
the conversion specifier %u when the parameter is an unsigned integer.

Signed-off-by: Vincent Mailhol <mailhol.vincent@wanadoo.fr>
Signed-off-by: David Ahern <dsahern@kernel.org>
3 years agoiplink_can: code refactoring of print_ctrlmode()
Vincent Mailhol [Wed, 3 Nov 2021 16:44:25 +0000 (01:44 +0900)] 
iplink_can: code refactoring of print_ctrlmode()

This patch only does cleanup and do not introduce any functional
changes.

We do some code refactoring of print_ctrlmode() in prevision of the
upcoming patch:

  - remove the first argument of print_ctrlmode(). It is a pointer to
    FILE and is never used.

  - add a new function argument: enum output_type t in order to
    specify the output type (i.e. PRINT_{FP,JSON,ANY}).

  - add a new function argument: const char *key in order to specify
    the name of the json array (e.g. "ctrlmode").

  - replace the _PF() macro with the print_flag() function to increase
    readability.

  - directly return if none of the flags are set (previously, this
    check was done before calling the function).

Signed-off-by: Vincent Mailhol <mailhol.vincent@wanadoo.fr>
Signed-off-by: David Ahern <dsahern@kernel.org>
3 years agoiplink_can: fix configuration ranges in print_usage() and add unit
Vincent Mailhol [Wed, 3 Nov 2021 16:44:24 +0000 (01:44 +0900)] 
iplink_can: fix configuration ranges in print_usage() and add unit

The configuration ranges in print_usage() are taken from "Table 8 -
Time segments' minimum configuration ranges" in section 11.3.1.2
"Configuration of the bit time parameters" of ISO 11898-1.

The standard clearly specifies that "implementations may allow time
segments that exceed the minimum required configuration ranges
specified in Table 8".

Because no maximum ranges are given in the standard, all given ranges
{ a..b } are simply replaced with { NUMBER }.

The actual ranges are specific to each device and can be confirmed
doing:

$ ip --details link show can0
1: can0: <NOARP,ECHO> mtu 16 qdisc noop state DOWN mode DEFAULT group default qlen 10
    link/can  promiscuity 0 minmtu 0 maxmtu 0
    can state STOPPED restart-ms 0
  ES582.1/ES584.1: tseg1 2..256 tseg2 2..128 sjw 1..128 brp 1..512 brp-inc 1
  ES582.1/ES584.1: dtseg1 2..32 dtseg2 1..16 dsjw 1..8 dbrp 1..32 dbrp-inc 1
  clock 80000000 numtxqueues 1 numrxqueues 1 gso_max_size 65536 gso_max_segs 65535

Finally, the unit (bps, tq, ns or ms) are given. The rationale to add
the units is that the TDC parameters (that will be introduced in the
upcoming patches) are measured in a different unit than the other
bittiming parameters: clock period (a.k.a. minimum time quantum)
instead of time quantum. Adding the units disambiguates things.

For reference, before the change:
$ ip link set can0 type can help
Usage: ip link set DEVICE type can
[ bitrate BITRATE [ sample-point SAMPLE-POINT] ] |
[ tq TQ prop-seg PROP_SEG phase-seg1 PHASE-SEG1
    phase-seg2 PHASE-SEG2 [ sjw SJW ] ]

[ dbitrate BITRATE [ dsample-point SAMPLE-POINT] ] |
[ dtq TQ dprop-seg PROP_SEG dphase-seg1 PHASE-SEG1
    dphase-seg2 PHASE-SEG2 [ dsjw SJW ] ]

[ loopback { on | off } ]
[ listen-only { on | off } ]
[ triple-sampling { on | off } ]
[ one-shot { on | off } ]
[ berr-reporting { on | off } ]
[ fd { on | off } ]
[ fd-non-iso { on | off } ]
[ presume-ack { on | off } ]

[ restart-ms TIME-MS ]
[ restart ]

[ termination { 0..65535 } ]

Where: BITRATE := { 1..1000000 }
  SAMPLE-POINT := { 0.000..0.999 }
  TQ := { NUMBER }
  PROP-SEG := { 1..8 }
  PHASE-SEG1 := { 1..8 }
  PHASE-SEG2 := { 1..8 }
  SJW := { 1..4 }
  RESTART-MS := { 0 | NUMBER }

...and after it:
$ ip link set can0 type can help
Usage: ip link set DEVICE type can
[ bitrate BITRATE [ sample-point SAMPLE-POINT] ] |
[ tq TQ prop-seg PROP_SEG phase-seg1 PHASE-SEG1
    phase-seg2 PHASE-SEG2 [ sjw SJW ] ]

[ dbitrate BITRATE [ dsample-point SAMPLE-POINT] ] |
[ dtq TQ dprop-seg PROP_SEG dphase-seg1 PHASE-SEG1
    dphase-seg2 PHASE-SEG2 [ dsjw SJW ] ]

[ loopback { on | off } ]
[ listen-only { on | off } ]
[ triple-sampling { on | off } ]
[ one-shot { on | off } ]
[ berr-reporting { on | off } ]
[ fd { on | off } ]
[ fd-non-iso { on | off } ]
[ presume-ack { on | off } ]
[ cc-len8-dlc { on | off } ]

[ restart-ms TIME-MS ]
[ restart ]

[ termination { 0..65535 } ]

Where: BITRATE := { NUMBER in bps }
  SAMPLE-POINT := { 0.000..0.999 }
  TQ := { NUMBER in ns }
  PROP-SEG := { NUMBER in tq }
  PHASE-SEG1 := { NUMBER in tq }
  PHASE-SEG2 := { NUMBER in tq }
  SJW := { NUMBER in tq }
  RESTART-MS := { 0 | NUMBER in ms }

Signed-off-by: Vincent Mailhol <mailhol.vincent@wanadoo.fr>
Signed-off-by: David Ahern <dsahern@kernel.org>
3 years agoip: add AMT support
Taehee Yoo [Sat, 30 Oct 2021 14:08:58 +0000 (14:08 +0000)] 
ip: add AMT support

Add basic support for Automatic Multicast Tunneling (AMT) network devices.

Signed-off-by: Taehee Yoo <ap420073@gmail.com>
3 years agoImport amt.h
David Ahern [Wed, 3 Nov 2021 19:23:38 +0000 (13:23 -0600)] 
Import amt.h

Impor amt.h uapi from last kernel sync point

Signed-off-by: David Ahern <dsahern@kernel.org>
3 years agoUpdate kernel headers
David Ahern [Wed, 3 Nov 2021 19:22:15 +0000 (13:22 -0600)] 
Update kernel headers

Update kernel headers to commit:
    cc0356d6a02e ("Merge tag 'x86_core_for_v5.16_rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip")

Signed-off-by: David Ahern <dsahern@kernel.org>
3 years agodevlink: Fix cmd_dev_param_set() to check configuration mode
Moshe Shemesh [Sun, 31 Oct 2021 06:48:47 +0000 (08:48 +0200)] 
devlink: Fix cmd_dev_param_set() to check configuration mode

This patch is fixing a bug, when param set user command includes
configuration mode which is not supported, the tool may not respond
with error if the requested value is 0. In such case
cmd_dev_param_set_cb() won't find the requested configuration mode and
returns ctx->value as initialized (equal 0). Then cmd_dev_param_set()
may find that requested value equals current value and returns success.

Fixing the bug by adding a flag cmode_found which is set only if
cmd_dev_param_set_cb() finds the requested configuration mode.

Fixes: 13925ae9eb38 ("devlink: Add param command support")
Signed-off-by: Moshe Shemesh <moshe@nvidia.com>
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
3 years agov5.15.0 v5.15.0
Stephen Hemminger [Mon, 1 Nov 2021 23:41:02 +0000 (16:41 -0700)] 
v5.15.0

3 years agordma: Fix SRQ resource tracking information json
Neta Ostrovsky [Mon, 25 Oct 2021 10:37:07 +0000 (13:37 +0300)] 
rdma: Fix SRQ resource tracking information json

Fix the json output for the QPs that are associated with the SRQ -
The qpn are now displayed in a json array.

Sample output before the fix:
$ rdma res show srq lqpn 126-141 -j -p
[ {
        "ifindex":0,
"ifname":"ibp8s0f0",
"srqn":4,
"type":"BASIC",
"lqpn":["126-128,130-140"],
"pdn":9,
"pid":3581,
"comm":"ibv_srq_pingpon"
    },{
"ifindex":0,
"ifname":"ibp8s0f0",
"srqn":5,
"type":"BASIC",
"lqpn":["141"],
"pdn":10,
"pid":3584,
"comm":"ibv_srq_pingpon"
    } ]

Sample output after the fix:
$ rdma res show srq lqpn 126-141 -j -p
[ {
        "ifindex":0,
"ifname":"ibp8s0f0",
"srqn":4,
"type":"BASIC",
"lqpn":["126-128","130-140"],
"pdn":9,
"pid":3581,
"comm":"ibv_srq_pingpon"
    },{
"ifindex":0,
"ifname":"ibp8s0f0",
"srqn":5,
"type":"BASIC",
"lqpn":["141"],
"pdn":10,
"pid":3584,
"comm":"ibv_srq_pingpon"
    } ]

Fixes: 9b272e138d23 ("rdma: Add SRQ resource tracking information")
Signed-off-by: Neta Ostrovsky <netao@nvidia.com>
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
3 years agoman: devlink-port: fix pfnum for devlink port add
Antoine Tenart [Mon, 18 Oct 2021 08:52:20 +0000 (10:52 +0200)] 
man: devlink-port: fix pfnum for devlink port add

When configuring a devlink PCI port, the pfnumber can be specified
using 'pfnum' and not 'pcipf' as stated in the man page. Fix this.

Signed-off-by: Antoine Tenart <atenart@kernel.org>
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
3 years agoMerge branch 'managed-neighbor' into next
David Ahern [Thu, 28 Oct 2021 15:00:26 +0000 (09:00 -0600)] 
Merge branch 'managed-neighbor' into next

Daniel Borkmann  says:

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

iproute2 patches to add support for managed neighbor entries as per recent
net-next commits:

  2ed08b5ead3c ("Merge branch 'Managed-Neighbor-Entries'")
  c47fedba94bc ("Merge branch 'minor-managed-neighbor-follow-ups'")

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

Signed-off-by: David Ahern <dsahern@kernel.org>
3 years agoip, neigh: Add NTF_EXT_MANAGED support
Daniel Borkmann [Mon, 25 Oct 2021 15:47:28 +0000 (17:47 +0200)] 
ip, neigh: Add NTF_EXT_MANAGED support

Currently, ip neigh does not support the NTF_EXT_MANAGED flag. Add cmdline
support.

Usage example:

  # ./ip/ip n replace 192.168.178.30 dev enp5s0 managed extern_learn
  # ./ip/ip n
  192.168.178.30 dev enp5s0 lladdr f4:8c:50:5e:71:9a managed extern_learn REACHABLE
  [...]

Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David Ahern <dsahern@kernel.org>
3 years agoip, neigh: Add missing NTF_USE support
Daniel Borkmann [Mon, 25 Oct 2021 15:47:27 +0000 (17:47 +0200)] 
ip, neigh: Add missing NTF_USE support

Currently, ip neigh does not support the NTF_USE flag. Similar to other flags
such as extern_learn, add cmdline support. The flag dump support is explicitly
missing here, since the kernel does not propagate the flag back to user space.

Usage example:

  # ./ip/ip n replace 192.168.178.30 dev enp5s0 use extern_learn
  # ./ip/ip n
  192.168.178.30 dev enp5s0 lladdr f4:8c:50:5e:71:9a extern_learn REACHABLE
  [...]

Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David Ahern <dsahern@kernel.org>
3 years agoip, neigh: Fix up spacing in netlink dump
Daniel Borkmann [Mon, 25 Oct 2021 15:47:26 +0000 (17:47 +0200)] 
ip, neigh: Fix up spacing in netlink dump

Fix up spacing to consistently add a single ' ' after an attribute has
been printed. Currently, it is a bit of a mix of before and after which
can lead to double spacing to be printed.

Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: David Ahern <dsahern@kernel.org>
3 years agoxfrm: enable to manage default policies
Nicolas Dichtel [Mon, 25 Oct 2021 08:17:06 +0000 (10:17 +0200)] 
xfrm: enable to manage default policies

Two new commands to manage default policies:
 - ip xfrm policy setdefault
 - ip xfrm policy getdefault

And the corresponding part in 'ip xfrm monitor'.

Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Signed-off-by: David Ahern <dsahern@kernel.org>
3 years agoMerge branch 'rdma-optional-stats' into next
David Ahern [Sat, 16 Oct 2021 18:52:02 +0000 (12:52 -0600)] 
Merge branch 'rdma-optional-stats' into next

Mark Zhang  says:

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

This is supplementary part of kernel series [1], which provides an
extension to the rdma statistics tool that allows to set or list
optional counters dynamically, using netlink.

Thanks

[1] https://www.spinics.net/lists/linux-rdma/msg106283.html

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

Signed-off-by: David Ahern <dsahern@kernel.org>
3 years agouapi: pickup fix for xfrm ABI breakage
Stephen Hemminger [Sat, 16 Oct 2021 00:40:30 +0000 (17:40 -0700)] 
uapi: pickup fix for xfrm ABI breakage

See kernel
Commit 844f7eaaed9 ("include/uapi/linux/xfrm.h: Fix XFRM_MSG_MAPPING ABI breakage")

Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>
3 years agoiplink: enable to specify index when changing netns
Nicolas Dichtel [Tue, 12 Oct 2021 09:34:05 +0000 (11:34 +0200)] 
iplink: enable to specify index when changing netns

When an interface is moved to another netns, it's possible to specify a
new ifindex. Let's add this support.

Link: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=eeb85a14ee34
Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Signed-off-by: David Ahern <dsahern@kernel.org>
3 years agoMerge branch 'config-libdir' into next
David Ahern [Fri, 15 Oct 2021 23:59:33 +0000 (17:59 -0600)] 
Merge branch 'config-libdir' into next

Andrea Claudi  says:

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

This series add support for the libdir parameter in iproute2 configure
script. The idea is to make use of the fact that packaging systems may
assume that 'configure' comes from autotools allowing a syntax similar
to the autotools one, and using it to tell iproute2 where the distro
expects to find its lib files.

Patches 1-2 fix a parsing issue on current configure options, that may
trigger an endless loop when no value is provided with some options;

Patch 3 fixes a parsing issue bailing out when more than one value is
provided for a single option;

Patch 4 simplifies options parsing, moving semantic checks out of the
while loop processing options;

Patch 5 introduces support for the --opt=value style on current options,
for uniformity;

Patch 6 adds the --prefix option, that may be used by some packaging
systems when calling the configure script;

Patch 7 finally adds the --libdir option, and also drops the static
LIBDIR var from the Makefile.

Changelog:
----------
v4 -> v5
  - bail out when multiple values are provided with a single option
  - simplify option parsing and reduce code duplication, as suggested
    by Phil Sutter
  - remove a nasty eval on libdir option processing

v3 -> v4
  - fix parsing issue on '--include_dir' and '--libbpf_dir'
  - split '--opt value' and '--opt=value' use cases, avoid code
    duplication moving semantic checks on value to dedicated functions

v2 -> v3
  - fix parsing error on prefix and libdir options.

v1 -> v2
  - consolidate '--opt value' and '--opt=value' use cases, as suggested
    by David Ahern.
  - added patch 2 to manage the --prefix option, used by the Debian
    packaging system, as reported by Luca Boccassi, and use it when
    setting lib directory.

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

Signed-off-by: David Ahern <dsahern@kernel.org>
3 years agoconfigure: add the --libdir option
Andrea Claudi [Thu, 14 Oct 2021 08:50:55 +0000 (10:50 +0200)] 
configure: add the --libdir option

This commit allows users/packagers to choose a lib directory to store
iproute2 lib files.

At the moment iproute2 ship lib files in /usr/lib and offers no way to
modify this setting. However, according to the FHS, distros may choose
"one or more variants of the /lib directory on systems which support
more than one binary format" (e.g. /usr/lib64 on Fedora).

As Luca states in commit a3272b93725a ("configure: restore backward
compatibility"), packaging systems may assume that 'configure' is from
autotools, and try to pass it some parameters.

Allowing the '--libdir=/path/to/libdir' syntax, we can use this to our
advantage, and let the lib directory to be chosen by the distro
packaging system.

Note that LIBDIR uses "\${prefix}/lib" as default value because autoconf
allows this to be expanded to the --prefix value at configure runtime.
"\${prefix}" is replaced with the PREFIX value in check_lib_dir().

Signed-off-by: Andrea Claudi <aclaudi@redhat.com>
Acked-by: Phil Sutter <phil@nwl.cc>
Signed-off-by: David Ahern <dsahern@kernel.org>
3 years agoconfigure: add the --prefix option
Andrea Claudi [Thu, 14 Oct 2021 08:50:54 +0000 (10:50 +0200)] 
configure: add the --prefix option

This commit add the '--prefix' option to the iproute2 configure script.

This mimics the '--prefix' option that autotools configure provides, and
will be used later to allow users or packagers to set the lib directory.

Signed-off-by: Andrea Claudi <aclaudi@redhat.com>
Acked-by: Phil Sutter <phil@nwl.cc>
Signed-off-by: David Ahern <dsahern@kernel.org>
3 years agoconfigure: support --param=value style
Andrea Claudi [Thu, 14 Oct 2021 08:50:53 +0000 (10:50 +0200)] 
configure: support --param=value style

This commit makes it possible to specify values for configure params
using the common autotools configure syntax '--param=value'.

Signed-off-by: Andrea Claudi <aclaudi@redhat.com>
Acked-by: Phil Sutter <phil@nwl.cc>
Signed-off-by: David Ahern <dsahern@kernel.org>
3 years agoconfigure: simplify options parsing
Andrea Claudi [Thu, 14 Oct 2021 08:50:52 +0000 (10:50 +0200)] 
configure: simplify options parsing

This commit simplifies options parsing moving all the code not related to
parsing out of the case statement.

- The conditional shift after the assignments is moved right after the
  case, reducing code duplication.
- The semantic checks on the LIBBPF_FORCE value is moved after the loop
  like we already did for INCLUDE and LIBBPF_DIR.
- Finally, the loop condition is changed to check remaining arguments, thus
  making it possible to get rid of the null string case break.

As a bonus, now the help message states that on or off should follow
--libbpf_force

Signed-off-by: Andrea Claudi <aclaudi@redhat.com>
Acked-by: Phil Sutter <phil@nwl.cc>
Signed-off-by: David Ahern <dsahern@kernel.org>
3 years agoconfigure: fix parsing issue with more than one value per option
Andrea Claudi [Thu, 14 Oct 2021 08:50:51 +0000 (10:50 +0200)] 
configure: fix parsing issue with more than one value per option

With commit a9c3d70d902a ("configure: add options ability") users are no
more able to provide wrong command lines like:

$ ./configure --include_dir foo bar

The script simply bails out when user provides more than one value for a
single option. However, in doing so, it breaks backward compatibility with
some packaging system, which expects unknown options to be ignored.

Commit a3272b93725a ("configure: restore backward compatibility") fix this
issue, but makes it possible again for users to provide wrong command lines
such as the one above.

This fixes the issue simply ignoring autoconf-like options such as
'--opt=value'.

Fixes: a3272b93725a ("configure: restore backward compatibility")
Signed-off-by: Andrea Claudi <aclaudi@redhat.com>
Acked-by: Phil Sutter <phil@nwl.cc>
Signed-off-by: David Ahern <dsahern@kernel.org>
3 years agoconfigure: fix parsing issue on libbpf_dir option
Andrea Claudi [Thu, 14 Oct 2021 08:50:50 +0000 (10:50 +0200)] 
configure: fix parsing issue on libbpf_dir option

configure is stuck in an endless loop if '--libbpf_dir' option is used
without a value:

$ ./configure --libbpf_dir
./configure: line 515: shift: 2: shift count out of range
./configure: line 515: shift: 2: shift count out of range
[...]

Fix it splitting 'shift 2' into two consecutive shifts, and making the
second one conditional to the number of remaining arguments.

A check is also provided after the while loop to verify the libbpf dir
exists; also, as LIBBPF_DIR does not have a default value, configure bails
out if the user does not specify a value after --libbpf_dir, thus avoiding
to produce an erroneous configuration.

Fixes: 7ae2585b865a ("configure: convert LIBBPF environment variables to command-line options")
Signed-off-by: Andrea Claudi <aclaudi@redhat.com>
Acked-by: Phil Sutter <phil@nwl.cc>
Signed-off-by: David Ahern <dsahern@kernel.org>
3 years agoconfigure: fix parsing issue on include_dir option
Andrea Claudi [Thu, 14 Oct 2021 08:50:49 +0000 (10:50 +0200)] 
configure: fix parsing issue on include_dir option

configure is stuck in an endless loop if '--include_dir' option is used
without a value:

$ ./configure --include_dir
./configure: line 506: shift: 2: shift count out of range
./configure: line 506: shift: 2: shift count out of range
[...]

Fix it splitting 'shift 2' into two consecutive shifts, and making the
second one conditional to the number of remaining arguments.

A check is also provided after the while loop to verify the include dir
exists; this avoid to produce an erroneous configuration.

Fixes: a9c3d70d902a ("configure: add options ability")
Signed-off-by: Andrea Claudi <aclaudi@redhat.com>
Acked-by: Phil Sutter <phil@nwl.cc>
Signed-off-by: David Ahern <dsahern@kernel.org>
3 years agordma: Add optional-counters set/unset support
Neta Ostrovsky [Thu, 14 Oct 2021 07:53:58 +0000 (10:53 +0300)] 
rdma: Add optional-counters set/unset support

This patch provides an extension to the rdma statistics tool
that allows to set/unset optional counters set dynamically,
using new netlink commands.
Note that the optional counter statistic implementation is
driver-specific and may impact the performance.

Examples:
To enable a set of optional counters on link rocep8s0f0/1:
    $ sudo rdma statistic set link rocep8s0f0/1 optional-counters cc_rx_ce_pkts,cc_rx_cnp_pkts
To disable all optional counters on link rocep8s0f0/1:
    $ sudo rdma statistic unset link rocep8s0f0/1 optional-counters

Signed-off-by: Neta Ostrovsky <netao@nvidia.com>
Reviewed-by: Leon Romanovsky <leonro@nvidia.com>
Signed-off-by: Mark Zhang <markzhang@nvidia.com>
Signed-off-by: David Ahern <dsahern@kernel.org>
3 years agordma: Add stat "mode" support
Neta Ostrovsky [Thu, 14 Oct 2021 07:53:57 +0000 (10:53 +0300)] 
rdma: Add stat "mode" support

This patch introduces the "mode" command, which presents the enabled or
supported (when the "supported" argument is available) optional
counters.

An optional counter is a vendor-specific counter that may be
dynamically enabled/disabled. This enhancement of hwcounters allows
exposing of counters which are for example mutual exclusive and cannot
be enabled at the same time, counters that might degrades performance,
optional debug counters, etc.

Examples:
To present currently enabled optional counters on link rocep8s0f0/1:
    $ rdma statistic mode link rocep8s0f0/1
    link rocep8s0f0/1 optional-counters cc_rx_ce_pkts

To present supported optional counters on link rocep8s0f0/1:
    $ rdma statistic mode supported link rocep8s0f0/1
    link rocep8s0f0/1 supported optional-counters cc_rx_ce_pkts,cc_rx_cnp_pkts,cc_tx_cnp_pkts

Signed-off-by: Neta Ostrovsky <netao@nvidia.com>
Reviewed-by: Leon Romanovsky <leonro@nvidia.com>
Signed-off-by: Mark Zhang <markzhang@nvidia.com>
Signed-off-by: David Ahern <dsahern@kernel.org>
3 years agordma: Update uapi headers
Neta Ostrovsky [Thu, 14 Oct 2021 07:53:56 +0000 (10:53 +0300)] 
rdma: Update uapi headers

Update rdma_netlink.h file upto kernel commit 7301d0a9834c
("RDMA/nldev: Add support to get status of all counters")

Signed-off-by: Neta Ostrovsky <netao@nvidia.com>
Reviewed-by: Leon Romanovsky <leonro@nvidia.com>
Signed-off-by: Mark Zhang <markzhang@nvidia.com>
Signed-off-by: David Ahern <dsahern@kernel.org>
3 years agoUpdate kernel headers
David Ahern [Fri, 15 Oct 2021 23:49:19 +0000 (17:49 -0600)] 
Update kernel headers

Update kernel headers to commit:
    295711fa8fec ("Merge branch 'dpaa2-irq-coalescing'")

Signed-off-by: David Ahern <dsahern@kernel.org>
3 years agomptcp: cleanup include section.
Stephen Hemminger [Tue, 12 Oct 2021 15:39:05 +0000 (17:39 +0200)] 
mptcp: cleanup include section.

David reported ipmptcp breaks hard the build when updating the
relevant kernel headers.

We should be more careful in the header section, explicitly
including all the required dependencies respecting the usual order
between systems and local headers.

Stephen Hemminger <stephen@networkplumber.org>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: David Ahern <dsahern@kernel.org>
3 years agolib/bpf: fix map-in-map creation without prepopulation
Paul Chaignon [Wed, 13 Oct 2021 14:39:27 +0000 (16:39 +0200)] 
lib/bpf: fix map-in-map creation without prepopulation

When creating map-in-maps, the outer map can be prepopulated using the
inner_idx field of inner maps. That field defines the index of the inner
map in the outer map. It is ignored if set to -1.

Commit 6d61a2b55799 ("lib: add libbpf support") however started using
that field to identify inner maps. While iterating over all maps looking
for inner maps, maps with inner_idx set to -1 are erroneously skipped.
As a result, trying to create a map-in-map with prepopulation disabled
fails because the inner_id of the outer map is not correctly set.

This bug can be observed with strace -ebpf (notice the zero inner_map_fd
for the outer map creation):

    bpf(BPF_MAP_CREATE, {map_type=BPF_MAP_TYPE_ARRAY, key_size=4, value_size=130996, max_entries=1, map_flags=0, inner_map_fd=0, map_name="maglev_inner", map_ifindex=0, btf_fd=0, btf_key_type_id=0, btf_value_type_id=0}, 128) = 32
    bpf(BPF_MAP_CREATE, {map_type=BPF_MAP_TYPE_HASH_OF_MAPS, key_size=2, value_size=4, max_entries=65536, map_flags=BPF_F_NO_PREALLOC, inner_map_fd=0, map_name="maglev_outer", map_ifindex=0, btf_fd=0, btf_key_type_id=0, btf_value_type_id=0}, 128) = -1 EINVAL (Invalid argument)

Fixes: 6d61a2b55799 ("lib: add libbpf support")
Signed-off-by: Paul Chaignon <paul@isovalent.com>
Signed-off-by: Stephen Hemminger <stephen@networkplumber.org>