This commit adds AF_XDP support to libbpf. The main reason for this is
to facilitate writing applications that use AF_XDP by offering
higher-level APIs that hide many of the details of the AF_XDP
uapi. This is in the same vein as libbpf facilitates XDP adoption by
offering easy-to-use higher level interfaces of XDP
functionality. Hopefully this will facilitate adoption of AF_XDP, make
applications using it simpler and smaller, and finally also make it
possible for applications to benefit from optimizations in the AF_XDP
user space access code. Previously, people just copied and pasted the
code from the sample application into their application, which is not
desirable.

The interface is composed of two parts:

* Low-level access interface to the four rings and the packet
* High-level control plane interface for creating and setting
  up umems and af_xdp sockets as well as a simple XDP program.
Tested-by: NBjörn Töpel <bjorn.topel@intel.com>
Signed-off-by: NMagnus Karlsson <magnus.karlsson@intel.com>
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>

Simple test that I used to reproduce the issue in the previous commit:
Do BPF_PROG_TEST_RUN with max iterations, each program is 4096 simple
move instructions. File alarm in 0.1 second and check that
bpf_prog_test_run is interrupted (i.e. test doesn't hang).

Note: reposting this for bpf-next to avoid linux-next conflict. In this
version I test both BPF_PROG_TYPE_SOCKET_FILTER (which uses generic
bpf_test_run implementation) and BPF_PROG_TYPE_FLOW_DISSECTOR (which has
it own loop with preempt handling in bpf_prog_test_run_flow_dissector).
Signed-off-by: NStanislav Fomichev <sdf@google.com>
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>

Syzbot found out that running BPF_PROG_TEST_RUN with repeat=0xffffffff
makes process unkillable. The problem is that when CONFIG_PREEMPT is
enabled, we never see need_resched() return true. This is due to the
fact that preempt_enable() (which we do in bpf_test_run_one on each
iteration) now handles resched if it's needed.

Let's disable preemption for the whole run, not per test. In this case
we can properly see whether resched is needed.
Let's also properly return -EINTR to the userspace in case of a signal
interrupt.

This is a follow up for a recently fixed issue in bpf_test_run, see
commit df1a2cb7 ("bpf/test_run: fix unkillable
BPF_PROG_TEST_RUN").
Reported-by: Nsyzbot <syzkaller@googlegroups.com>
Signed-off-by: NStanislav Fomichev <sdf@google.com>
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>

When running BPF test suite the following splat occurs:

[  415.930950] test_bpf: #0 TAX jited:0
[  415.931067] BUG: assuming atomic context at lib/test_bpf.c:6674
[  415.946169] in_atomic(): 0, irqs_disabled(): 0, pid: 11556, name: modprobe
[  415.953176] INFO: lockdep is turned off.
[  415.957207] CPU: 1 PID: 11556 Comm: modprobe Tainted: G        W         5.0.0-rc7-next-20190220 #1
[  415.966328] Hardware name: HiKey Development Board (DT)
[  415.971592] Call trace:
[  415.974069]  dump_backtrace+0x0/0x160
[  415.977761]  show_stack+0x24/0x30
[  415.981104]  dump_stack+0xc8/0x114
[  415.984534]  __cant_sleep+0xf0/0x108
[  415.988145]  test_bpf_init+0x5e0/0x1000 [test_bpf]
[  415.992971]  do_one_initcall+0x90/0x428
[  415.996837]  do_init_module+0x60/0x1e4
[  416.000614]  load_module+0x1de0/0x1f50
[  416.004391]  __se_sys_finit_module+0xc8/0xe0
[  416.008691]  __arm64_sys_finit_module+0x24/0x30
[  416.013255]  el0_svc_common+0x78/0x130
[  416.017031]  el0_svc_handler+0x38/0x78
[  416.020806]  el0_svc+0x8/0xc

Rework so that preemption is disabled when we loop over function
'BPF_PROG_RUN(...)'.

Fixes: 568f1967 ("bpf: check that BPF programs run with preemption disabled")
Suggested-by: NArnd Bergmann <arnd@arndb.de>
Signed-off-by: NAnders Roxell <anders.roxell@linaro.org>
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>

The xdp_redirect and xdp_redirect_map sample programs both load a dummy
program onto the egress interfaces. However, the unload code checks these
programs against the wrong id number, and thus refuses to unload them. Fix
the comparison to avoid this.

Fixes: 3b7a8ec2 ("samples/bpf: Check the prog id before exiting")
Signed-off-by: NToke Høiland-Jørgensen <toke@redhat.com>
Acked-by: NMaciej Fijalkowski <maciej.fijalkowski@intel.com>
Acked-by: NMartin KaFai Lau <kafai@fb.com>
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>

All BPF programs must be called with preemption disabled.

Fixes: 568f1967 ("bpf: check that BPF programs run with preemption disabled")
Reported-by: syzbot+8bf19ee2aa580de7a2a7@syzkaller.appspotmail.com
Signed-off-by: NAlexei Starovoitov <ast@kernel.org>
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>

The skb->queue_mapping already have read access, via __sk_buff->queue_mapping.

This patch allow BPF tc qdisc clsact write access to the queue_mapping via
tc_cls_act_is_valid_access.  Also handle that the value NO_QUEUE_MAPPING
is not allowed.

It is already possible to change this via TC filter action skbedit
tc-skbedit(8).  Due to the lack of TC examples, lets show one:

  # tc qdisc  add  dev ixgbe1 clsact
  # tc filter add  dev ixgbe1 ingress matchall action skbedit queue_mapping 5
  # tc filter list dev ixgbe1 ingress

The most common mistake is that XPS (Transmit Packet Steering) takes
precedence over setting skb->queue_mapping. XPS is configured per DEVICE
via /sys/class/net/DEVICE/queues/tx-*/xps_cpus via a CPU hex mask. To
disable set mask=00.

The purpose of changing skb->queue_mapping is to influence the selection of
the net_device "txq" (struct netdev_queue), which influence selection of
the qdisc "root_lock" (via txq->qdisc->q.lock) and txq->_xmit_lock. When
using the MQ qdisc the txq->qdisc points to different qdiscs and associated
locks, and HARD_TX_LOCK (txq->_xmit_lock), allowing for CPU scalability.

Due to lack of TC examples, lets show howto attach clsact BPF programs:

 # tc qdisc  add  dev ixgbe2 clsact
 # tc filter add  dev ixgbe2 egress bpf da obj XXX_kern.o sec tc_qmap2cpu
 # tc filter list dev ixgbe2 egress
Signed-off-by: NJesper Dangaard Brouer <brouer@redhat.com>
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>

Introduce cant_sleep() macro for annotation of functions that
cannot sleep.

Use it in BPF_PROG_RUN to catch execution of BPF programs in
preemptable context.
Suggested-by: NJann Horn <jannh@google.com>
Signed-off-by: NPeter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: NAlexei Starovoitov <ast@kernel.org>
Acked-by: NSong Liu <songliubraving@fb.com>
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>

bpftool has support for attach types "stream_verdict" and
"stream_parser" but the documentation was referring to them as
"skb_verdict" and "skb_parse". The inconsistency comes from commit
b7d3826c ("bpf: bpftool, add support for attaching programs to
maps").

This patch changes the documentation to match the implementation:
- "bpftool prog help"
- man pages
- bash completion
Signed-off-by: NAlban Crequy <alban@kinvolk.io>
Reviewed-by: NQuentin Monnet <quentin.monnet@netronome.com>
Signed-off-by: NDaniel Borkmann <daniel@iogearbox.net>

Fixes gcc '-Wunused-but-set-variable' warning:

drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c: In function 'bnx2x_get_hwinfo':
drivers/net/ethernet/broadcom/bnx2x/bnx2x_main.c:11940:10: warning:
 variable 'mfw_vn' set but not used [-Wunused-but-set-variable]

It's never used since introduction.
Signed-off-by: NYueHaibing <yuehaibing@huawei.com>
Acked-by: NSudarsana Reddy Kalluru <skalluru@marvell.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Heiner Kallweit says:

====================
net: phy: add helpers for handling C45 10GBT AN register values

Similar to the existing helpers for the Clause 22 registers add helpers
to deal with converting Clause 45 advertisement registers to / from
link mode bitmaps.

Note that these helpers are defined in linux/mdio.h, not like the
Clause 22 helpers in linux/mii.h. Reason is that the Clause 45 register
constants are defined in uapi/linux/mdio.h. And uapi/linux/mdio.h
includes linux/mii.h before defining the C45 register constants.

v2:
- Remove few helpers which aren't used by this series. They will
  follow together with the users.
====================
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Use mii_10gbt_stat_mod_linkmode_lpa_t() in genphy_c45_read_lpa() to
simplify the code.
Signed-off-by: NHeiner Kallweit <hkallweit1@gmail.com>
Reviewed-by: NFlorian Fainelli <f.fainelli@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Similar to the existing helpers for the Clause 22 registers add helper
mii_10gbt_stat_mod_linkmode_lpa_t.

Note that this helper is defined in linux/mdio.h, not like the
Clause 22 helpers in linux/mii.h. Reason is that the Clause 45 register
constants are defined in uapi/linux/mdio.h. And uapi/linux/mdio.h
includes linux/mii.h before defining the C45 register constants.

v2:
- remove helpers that don't have users in this series
Signed-off-by: NHeiner Kallweit <hkallweit1@gmail.com>
Reviewed-by: NFlorian Fainelli <f.fainelli@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Fix following warning:

drivers/net/ethernet/cavium/liquidio/cn23xx_pf_device.c:1453:35: warning: Using plain integer as NULL pointer
drivers/net/ethernet/cavium/liquidio/lio_main.c:2910:23: warning: Using plain integer as NULL pointer
Signed-off-by: NYueHaibing <yuehaibing@huawei.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

writex() has implicit barriers, that's what makes it different from
writex_relaxed(). Therefore these calls to mmiowb() can be removed.

This patch was recently reverted due to a dependency with another
problematic patch. But because it didn't contribute to the problem
it was rebased and can be resubmitted.
Signed-off-by: NHeiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

RPS not work for DSA devices since the 'skb_get_hash'
will always get the invalid hash for dsa tagged packets.

"[PATCH] tag_mtk: add flow_dissect callback to the ops struct"
introduced the flow_dissect callback to get the right hash for
MTK tagged packet. Tag_dsa and tag_edsa  also need to implement
the callback.
Signed-off-by: NRundong Ge <rdong.ge@gmail.com>
Reviewed-by: NAndrew Lunn <andrew@lunn.ch>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The callback function of call_rcu() just calls a kfree(), so we
can use kfree_rcu() instead of call_rcu() + callback function.
Signed-off-by: NWei Yongjun <weiyongjun1@huawei.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Fix coccinelle warning:

./drivers/net/phy/mdio_bus.c:51:5-12: ERROR: PTR_ERR applied after initialization to constant on line 44
./drivers/net/phy/mdio_bus.c:52:5-12: ERROR: PTR_ERR applied after initialization to constant on line 44

fix this by using IS_ERR before PTR_ERR

Fixes: bafbdd52 ("phylib: Add device reset GPIO support")
Signed-off-by: NYueHaibing <yuehaibing@huawei.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The IP2ME packet trap is triggered by packets hitting local routes.
After evaluating current defaults used by the driver it was decided to
reduce the amount of traffic generated by this trap to 1Kpps and
increase the burst size. This is inline with similarly deployed systems.
Signed-off-by: NShalom Toledo <shalomt@mellanox.com>
Signed-off-by: NIdo Schimmel <idosch@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This file does not use hweight*() at all, and the definition is
surrounded by #if 0 ... #endif.
Signed-off-by: NMasahiro Yamada <yamada.masahiro@socionext.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Pablo Neira Ayuso says:

====================
Netfilter/IPVS updates for net-next

The following patchset contains Netfilter/IPVS updates for you net-next
tree:

1) Missing NFTA_RULE_POSITION_ID netlink attribute validation,
   from Phil Sutter.

2) Restrict matching on tunnel metadata to rx/tx path, from wenxu.

3) Avoid indirect calls for IPV6=y, from Florian Westphal.

4) Add two indirections to prepare merger of IPV4 and IPV6 nat
   modules, from Florian Westphal.

5) Broken indentation in ctnetlink, from Colin Ian King.

6) Patches to use struct_size() from netfilter and IPVS,
   from Gustavo A. R. Silva.

7) Display kernel splat only once in case of racing to confirm
   conntrack from bridge plus nfqueue setups, from Chieh-Min Wang.

8) Skip checksum validation for layer 4 protocols that don't need it,
   patch from Alin Nastac.

9) Sparse warning due to symbol that should be static in CLUSTERIP,
   from Wei Yongjun.

10) Add new toggle to disable SDP payload translation when media
    endpoint is reachable though the same interface as the signalling
    peer, from Alin Nastac.
====================
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Fixes the following sparse warning:

drivers/net/ethernet/hisilicon/hns3/hns3pf/hclge_main.c:2431:5: warning:
 symbol 'hclge_set_all_vf_rst' was not declared. Should it be static?

Fixes: aa5c4f17 ("net: hns3: add reset handling for VF when doing PF reset")
Signed-off-by: NWei Yongjun <weiyongjun1@huawei.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Remove duplicated include.
Signed-off-by: NYueHaibing <yuehaibing@huawei.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The function meth_init_tx_ring() is called from meth_tx_timeout(),
in which spin_lock is held, so we should use GFP_ATOMIC instead.

Fixes: 8d4c28fb ("meth: pass struct device to DMA API functions")
Signed-off-by: NWei Yongjun <weiyongjun1@huawei.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

This patch sets an error message in extack when the number of qdisc
handles exceeds the maximum. Also the error-code ENOSPC is more
appropriate than ENOMEM in this situation.
Signed-off-by: NIvan Vecera <ivecera@redhat.com>
Reported-by: NLi Shuang <shuali@redhat.com>
Reviewed-by: NStefano Brivio <sbrivio@redhat.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

The PHY core expects PHY drivers not to set Pause and Asym_Pause bits,
unless the driver only wants to specify one of them due to HW
limitation. In the case of the Marvell10g driver, we don't need to set
them.
Signed-off-by: NMaxime Chevallier <maxime.chevallier@bootlin.com>
Suggested-by: NAndrew Lunn <andrew@lunn.ch>
Reviewed-by: NAndrew Lunn <andrew@lunn.ch>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Fixes gcc '-Wunused-but-set-variable' warning:

drivers/net/dsa/bcm_sf2_cfp.c: In function 'bcm_sf2_cfp_ipv6_rule_set':
drivers/net/dsa/bcm_sf2_cfp.c:606:40: warning:
 variable 'v6_m_spec' set but not used [-Wunused-but-set-variable]
drivers/net/dsa/bcm_sf2_cfp.c:606:30: warning:
 variable 'v6_spec' set but not used [-Wunused-but-set-variable]

It not used any more after commit e4f7ef54 ("dsa: bcm_sf2: use flow_rule
infrastructure")
Signed-off-by: NYueHaibing <yuehaibing@huawei.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Check mask fields of tcp and ip flags when setting the corresponding mask
flag used in hardware.

Fixes: 8f256622 ("flow_offload: add flow_rule and flow_match")
Signed-off-by: NPieter Jansen van Vuuren <pieter.jansenvanvuuren@netronome.com>
Reviewed-by: NJohn Hurley <john.hurley@netronome.com>
Reviewed-by: NJakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Jakub Kicinski says:

====================
devlink: add the ability to update device flash

This series is the second step to allow trouble shooting and recovering
devices in bad state without the use of netdevs as handles.  We can
already query FW versions over devlink, now we add the ability to update
the FW.  This will allow drivers to implement some from of "limp-mode"
where the device can't really be used for networking and hence has no
netdev, but we can interrogate it over devlink and fix the broken FW.

Small but nice advantage of devlink is that it only holds the devlink
instance lock during flashing, unlike ethtool which holds rtnl_lock().
====================
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Devlink now allows updating device flash.  Implement this
callback.

Compared to ethtool update we no longer have to release
the networking locks - devlink doesn't take them.
Signed-off-by: NJakub Kicinski <jakub.kicinski@netronome.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

If driver does not support ethtool flash update operation
call into devlink.
Signed-off-by: NJakub Kicinski <jakub.kicinski@netronome.com>
Acked-by: NJiri Pirko <jiri@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Add devlink flash update command. Advanced NICs have firmware
stored in flash and often cryptographically secured. Updating
that flash is handled by management firmware. Ethtool has a
flash update command which served us well, however, it has two
shortcomings:
 - it takes rtnl_lock unnecessarily - really flash update has
   nothing to do with networking, so using a networking device
   as a handle is suboptimal, which leads us to the second one:
 - it requires a functioning netdev - in case device enters an
   error state and can't spawn a netdev (e.g. communication
   with the device fails) there is no netdev to use as a handle
   for flashing.

Devlink already has the ability to report the firmware versions,
now with the ability to update the firmware/flash we will be
able to recover devices in bad state.

To enable updates of sub-components of the FW allow passing
component name.  This name should correspond to one of the
versions reported in devlink info.

v1: - replace target id with component name (Jiri).
Signed-off-by: NJakub Kicinski <jakub.kicinski@netronome.com>
Acked-by: NJiri Pirko <jiri@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Heiner Kallweit says:

====================
net: phy: improve and use phy_resolve_aneg_linkmode

Improve phy_resolve_aneg_linkmode and use it in genphy_read_status.
====================
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Now that we have phy_resolve_aneg_linkmode() we can make
genphy_read_status() much simpler.
Signed-off-by: NHeiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

We have the settings array of modes which is sorted based on aneg
priority. Instead of checking each mode manually let's simply iterate
over the sorted settings.
Signed-off-by: NHeiner Kallweit <hkallweit1@gmail.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Check that filter is not NULL before passing it to tcf_walker->fn()
callback in cls_cgroup_walk(). This can happen when cls_cgroup_change()
failed to set first filter.

Fixes: ed76f5ed ("net: sched: protect filter_chain list with filter_chain_lock mutex")
Signed-off-by: NVlad Buslov <vladbu@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Check that filter is not NULL before passing it to tcf_walker->fn()
callback. This can happen when mall_change() failed to offload filter to
hardware.

Fixes: ed76f5ed ("net: sched: protect filter_chain list with filter_chain_lock mutex")
Reported-by: NIdo Schimmel <idosch@mellanox.com>
Tested-by: NIdo Schimmel <idosch@mellanox.com>
Signed-off-by: NVlad Buslov <vladbu@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Some classifiers set arg->stop in their implementation of tp->walk() API
when empty. Most of classifiers do not adhere to that convention. Do not
set arg->stop in route4_walk() to unify tp->walk() behavior among
classifier implementations.

Fixes: ed76f5ed ("net: sched: protect filter_chain list with filter_chain_lock mutex")
Signed-off-by: NVlad Buslov <vladbu@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Some classifiers set arg->stop in their implementation of tp->walk() API
when empty. Most of classifiers do not adhere to that convention. Do not
set arg->stop in fw_walk() to unify tp->walk() behavior among classifier
implementations.

Fixes: ed76f5ed ("net: sched: protect filter_chain list with filter_chain_lock mutex")
Signed-off-by: NVlad Buslov <vladbu@mellanox.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Use existing skb_put_data() and skb_trim() instead of open-coding them,
with the skb_put_data() first so that logically, `skb` still contains the
data to be copied in its data..tail area when skb_put_data() reads it.
This change on its own is a cleanup, and it is also necessary for potential
future integration of skbuffs with things like KASAN.
Signed-off-by: NJann Horn <jannh@google.com>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>