When tipc is loaded while many processes try to create a TIPC socket,
a crash occurs:
 PANIC: Unable to handle kernel paging request at virtual
 address "dfff20000000021d"
 pc : tipc_sk_create+0x374/0x1180 [tipc]
 lr : tipc_sk_create+0x374/0x1180 [tipc]
   Exception class = DABT (current EL), IL = 32 bits
 Call trace:
  tipc_sk_create+0x374/0x1180 [tipc]
  __sock_create+0x1cc/0x408
  __sys_socket+0xec/0x1f0
  __arm64_sys_socket+0x74/0xa8
 ...

This is due to race between sock_create and unfinished
register_pernet_device. tipc_sk_insert tries to do
"net_generic(net, tipc_net_id)".
but tipc_net_id is not initialized yet.

So switch the order of the two to close the race.

This can be reproduced with multiple processes doing socket(AF_TIPC, ...)
and one process doing module removal.

Fixes: a62fbcce ("tipc: make subscriber server support net namespace")
Signed-off-by: Junwei Hu <hujunwei4@huawei.com>
Reported-by: Wang Wang <wangwang2@huawei.com>
Reviewed-by: Xiaogang Wang <wangxiaogang3@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

At ipv6 route dismantle, fib6_drop_pcpu_from() is responsible
for finding all percpu routes and set their ->from pointer
to NULL, so that fib6_ref can reach its expected value (1).

The problem right now is that other cpus can still catch the
route being deleted, since there is no rcu grace period
between the route deletion and call to fib6_drop_pcpu_from()

This can leak the fib6 and associated resources, since no
notifier will take care of removing the last reference(s).

I decided to add another boolean (fib6_destroying) instead
of reusing/renaming exception_bucket_flushed to ease stable backports,
and properly document the memory barriers used to implement this fix.

This patch has been co-developped with Wei Wang.

Fixes: 93531c67 ("net/ipv6: separate handling of FIB entries from dst based routes")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reported-by: syzbot <syzkaller@googlegroups.com>
Cc: Wei Wang <weiwan@google.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Martin Lau <kafai@fb.com>
Acked-by: Wei Wang <weiwan@google.com>
Acked-by: Martin KaFai Lau <kafai@fb.com>
Reviewed-by: David Ahern <dsahern@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Zerocopy skbs without completion notification were added for packet
sockets with PACKET_TX_RING user buffers. Those signal completion
through the TP_STATUS_USER bit in the ring. Zerocopy annotation was
added only to avoid premature notification after clone or orphan, by
triggering a copy on these paths for these packets.

The mechanism had to define a special "no-uarg" mode because packet
sockets already use skb_uarg(skb) == skb_shinfo(skb)->destructor_arg
for a different pointer.

Before deferencing skb_uarg(skb), verify that it is a real pointer.

Fixes: 5cd8d46e ("packet: copy user buffers before orphan or clone")
Signed-off-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Added support for Telit LE910Cx 0x1260 and 0x1261 compositions.
Signed-off-by: Daniele Palmas <dnlplm@gmail.com>
Acked-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: David S. Miller <davem@davemloft.net>

XGMII interface mode no longer works on AQR107 after the recent changes,
adding back support.

Fixes: 570c8a7d ("net: phy: aquantia: check for supported interface modes in config_init")
Signed-off-by: Madalin Bucur <madalin.bucur@nxp.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

If bpfilter is not available return ENOPROTOOPT to fallback to netfilter.

Function request_module() returns both errors and userspace exit codes.
Just ignore them. Rechecking bpfilter_ops is enough.

Fixes: d2ba09c1 ("net: add skeleton of bpfilter kernel module")
Signed-off-by: Konstantin Khlebnikov <khlebnikov@yandex-team.ru>
Signed-off-by: David S. Miller <davem@davemloft.net>

Currently, hvsock does not implement any delayed or background close
logic. Whenever the hvsock socket is closed, a FIN is sent to the peer, and
the last reference to the socket is dropped, which leads to a call to
.destruct where the socket can hang indefinitely waiting for the peer to
close it's side. The can cause the user application to hang in the close()
call.

This change implements proper STREAM(TCP) closing handshake mechanism by
sending the FIN to the peer and the waiting for the peer's FIN to arrive
for a given timeout. On timeout, it will try to terminate the connection
(i.e. a RST). This is in-line with other socket providers such as virtio.

This change does not address the hang in the vmbus_hvsock_device_unregister
where it waits indefinitely for the host to rescind the channel. That
should be taken up as a separate fix.
Signed-off-by: Sunil Muthuswamy <sunilmut@microsoft.com>
Reviewed-by: Dexuan Cui <decui@microsoft.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Remove the MEMDUMP_DEV case in ia_ioctl to avoid copy
pointers to user space.
Signed-off-by: Fuqian Huang <huangfq.daxian@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Edward Cree says:

====================
flow_offload: fix CVLAN support

When the flow_offload infrastructure was added, CVLAN matches weren't
 plumbed through, and flow_rule_match_vlan() was incorrectly called in
 the mlx5 driver when populating CVLAN match information.  This series
 adds flow_rule_match_cvlan(), and uses it in the mlx5 code.
Both patches should also go to 5.1 stable.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

When flow_rule_match_XYZ() functions were first introduced,
flow_rule_match_cvlan() for inner vlan is missing.

In mlx5_core driver, to get inner vlan key and mask, flow_rule_match_vlan()
is just called, which is wrong because it obtains outer vlan information by
FLOW_DISSECTOR_KEY_VLAN.

This commit fixes this by changing to call flow_rule_match_cvlan() after
it's added.

Fixes: 8f256622 ("flow_offload: add flow_rule and flow_match structures and use them")
Signed-off-by: Jianbo Liu <jianbol@mellanox.com>
Signed-off-by: Edward Cree <ecree@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Plumb it through from the flow_dissector.
Signed-off-by: Edward Cree <ecree@solarflare.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Herbert Xu says:

====================
rhashtable: Fix sparse warnings

This patch series fixes all the sparse warnings.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

As cmpxchg is a non-RCU mechanism it will cause sparse warnings
when we use it for RCU.  This patch adds explicit casts to silence
those warnings.  This should probably be moved to RCU itself in
future.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>

The opaque type rhash_lock_head should not be marked with __rcu
because it can never be dereferenced.  We should apply the RCU
marking when we turn it into a pointer which can be dereferenced.

This patch does exactly that.  This fixes a number of sparse
warnings as well as getting rid of some unnecessary RCU checking.
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>

Daniel Borkmann says:

====================
pull-request: bpf 2019-05-16

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

The main changes are:

1) Fix a use after free in __dev_map_entry_free(), from Eric.

2) Several sockmap related bug fixes: a splat in strparser if
   it was never initialized, remove duplicate ingress msg list
   purging which can race, fix msg->sg.size accounting upon
   skb to msg conversion, and last but not least fix a timeout
   bug in tcp_bpf_wait_data(), from John.

3) Fix LRU map to avoid messing with eviction heuristics upon
   syscall lookup, e.g. map walks from user space side will
   then lead to eviction of just recently created entries on
   updates as it would mark all map entries, from Daniel.

4) Don't bail out when libbpf feature probing fails. Also
   various smaller fixes to flow_dissector test, from Stanislav.

5) Fix missing brackets for BTF_INT_OFFSET() in UAPI, from Gary.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>

The tcp_bpf_wait_data() routine needs to check timeo != 0 before
calling sk_wait_event() otherwise we may see unexpected stalls
on receiver.

Arika did all the leg work here I just formatted, posted and ran
a few tests.

Fixes: 604326b4 ("bpf, sockmap: convert to generic sk_msg interface")
Reported-by: Arika Chen <eaglesora@gmail.com>
Suggested-by: Arika Chen <eaglesora@gmail.com>
Signed-off-by: John Fastabend <john.fastabend@gmail.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>

In case we are not running in a namespace (which we don't do by default),
let's try to detach the bpf program that we use for eth_get_headlen tests.

Fixes: 0905beec ("selftests/bpf: run flow dissector tests in skb-less mode")
Signed-off-by: Stanislav Fomichev <sdf@google.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>

Otherwise, in case of an error, everything gets mushed together.

Fixes: a5cb3346 ("selftests/bpf: make flow dissector tests more extensible")
Signed-off-by: Stanislav Fomichev <sdf@google.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>

Otherwise libbpf is unusable from unprivileged process with
kernel.kernel.unprivileged_bpf_disabled=1.
All I get is EPERM from the probes, even if I just want to
open an ELF object and look at what progs/maps it has.

Instead of dying on probes, let's just pr_debug the error and
try to continue.
Signed-off-by: Stanislav Fomichev <sdf@google.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>

It is illegal to change arbitrary fields in skb_shared_info if the
skb is cloned.

Before calling skb_zcopy_clear() we need to ensure this rule,
therefore we need to move the test from sk_stream_alloc_skb()
to sk_wmem_free_skb()

Fixes: 4f661542 ("tcp: fix zerocopy and notsent_lowat issues")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Diagnosed-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Just hook get_link to standard ethtool_op_get_link,
nothing special needed at this point.
Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The fact that the Tx SG flag is fixed to 'on' is only
an oversight. Non-SG mode is also supported. Fix this
by allowing to turn SG off.
Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

For the unlikely case of TxBD extensions (i.e. ptp)
the driver tries to unmap the tx_swbd corresponding
to the extension, which is bogus as it has no buffer
attached.
Signed-off-by: Claudiu Manoil <claudiu.manoil@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Add rcu locks when accessing netdev when processing route request
and tunnel keep alive messages received from hardware.

Fixes: 8e6a9046 ("nfp: flower vxlan neighbour offload")
Fixes: 856f5b13 ("nfp: flower vxlan neighbour keep-alive")
Signed-off-by: Pieter Jansen van Vuuren <pieter.jansenvanvuuren@netronome.com>
Reviewed-by: Jakub Kicinski <jakub.kicinski@netronome.com>
Reviewed-by: John Hurley <john.hurley@netronome.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

BUG: unable to handle kernel paging request at ffffffffa018f000
PGD 3270067 P4D 3270067 PUD 3271063 PMD 2307eb067 PTE 0
Oops: 0000 [#1] PREEMPT SMP
CPU: 0 PID: 4138 Comm: modprobe Not tainted 5.1.0-rc7+ #1
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS
rel-1.9.3-0-ge2fc41e-prebuilt.qemu-project.org 04/01/2014
RIP: 0010:ppp_register_compressor+0x3e/0xd0 [ppp_generic]
Code: 98 4a 3f e2 48 8b 15 c1 67 00 00 41 8b 0c 24 48 81 fa 40 f0 19 a0
75 0e eb 35 48 8b 12 48 81 fa 40 f0 19 a0 74
RSP: 0018:ffffc90000d93c68 EFLAGS: 00010287
RAX: ffffffffa018f000 RBX: ffffffffa01a3000 RCX: 000000000000001a
RDX: ffff888230c750a0 RSI: 0000000000000000 RDI: ffffffffa019f000
RBP: ffffc90000d93c80 R08: 0000000000000001 R09: 0000000000000000
R10: 0000000000000000 R11: 0000000000000000 R12: ffffffffa0194080
R13: ffff88822ee1a700 R14: 0000000000000000 R15: ffffc90000d93e78
FS:  00007f2339557540(0000) GS:ffff888237a00000(0000)
knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: ffffffffa018f000 CR3: 000000022bde4000 CR4: 00000000000006f0
Call Trace:
 ? 0xffffffffa01a3000
 deflate_init+0x11/0x1000 [ppp_deflate]
 ? 0xffffffffa01a3000
 do_one_initcall+0x6c/0x3cc
 ? kmem_cache_alloc_trace+0x248/0x3b0
 do_init_module+0x5b/0x1f1
 load_module+0x1db1/0x2690
 ? m_show+0x1d0/0x1d0
 __do_sys_finit_module+0xc5/0xd0
 __x64_sys_finit_module+0x15/0x20
 do_syscall_64+0x6b/0x1d0
 entry_SYSCALL_64_after_hwframe+0x49/0xbe

If ppp_deflate fails to register in deflate_init,
module initialization failed out, however
ppp_deflate_draft may has been regiestred and not
unregistered before return.
Then the seconed modprobe will trigger crash like this.
Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Acked-by: Guillaume Nault <gnault@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Errors are negative numbers. Using %u shows them as very large positive
numbers such as 4294967277 that don't make sense. Use the %d format
instead, and get a much nicer -19.
Signed-off-by: Luca Ceresoli <luca@lucaceresoli.net>
Fixes: b48e0bab ("net: macb: Migrate to devm clock interface")
Fixes: 93b31f48 ("net/macb: unify clock management")
Fixes: 421d9df0 ("net/macb: merge at91_ether driver into macb driver")
Fixes: aead88bd ("net: ethernet: macb: Add support for rx_clk")
Fixes: f5473d1d ("net: macb: Support clock management for tsu_clk")
Acked-by: Nicolas Ferre <nicolas.ferre@microchip.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>

Currently, nla_put_iflink() doesn't put the IFLA_LINK attribute when
iflink == ifindex.

In some cases, a device can be created in a different netns with the
same ifindex as its parent. That device will not dump its IFLA_LINK
attribute, which can confuse some userspace software that expects it.
For example, if the last ifindex created in init_net and foo are both
8, these commands will trigger the issue:

    ip link add parent type dummy                   # ifindex 9
    ip link add link parent netns foo type macvlan  # ifindex 9 in ns foo

So, in case a device puts the IFLA_LINK_NETNSID attribute in a dump,
always put the IFLA_LINK attribute as well.

Thanks to Dan Winship for analyzing the original OpenShift bug down to
the missing netlink attribute.

v2: change Fixes tag, it's been here forever, as Nicolas Dichtel said
    add Nicolas' ack
v3: change Fixes tag
    fix subject typo, spotted by Edward Cree
Analyzed-by: Dan Winship <danw@redhat.com>
Fixes: d8a5ec67 ("[NET]: netlink support for moving devices between network namespaces.")
Signed-off-by: Sabrina Dubroca <sd@queasysnail.net>
Acked-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

The error print within mlx4_flow_steer_promisc_add() should
be a info print.

Fixes: 592e49dd ('net/mlx4: Implement promiscuous mode with device managed flow-steering')
Signed-off-by: Yunjian Wang <wangyunjian@huawei.com>
Reviewed-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Samuel clearly hasn't been working on this in many years and
patches getting to the wireless list are just being ignored
entirely now. Mark the subsystem as orphan to reflect the
current state and revert back to the netdev list so at least
some fixes can be picked up by Dave.
Signed-off-by: Johannes Berg <johannes@sipsolutions.net>
Acked-by: Samuel Ortiz <sameo@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Jiri reported that with a kernel built with CONFIG_FIXED_PHY=y,
CONFIG_NET_DSA=m and CONFIG_NET_DSA_LOOP=m, we would not get to a
functional state where the mock-up driver is registered. Turns out that
we are not descending into drivers/net/dsa/ unconditionally, and we
won't be able to link-in dsa_loop_bdinfo.o which does the actual mock-up
mdio device registration.
Reported-by: Jiri Pirko <jiri@resnulli.us>
Fixes: 40013ff2 ("net: dsa: Fix functional dsa-loop dependency on FIXED_PHY")
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Reviewed-by: Vivien Didelot <vivien.didelot@gmail.com>
Tested-by: Jiri Pirko <jiri@resnulli.us>
Signed-off-by: David S. Miller <davem@davemloft.net>

Commit c7d13c8f ("tcp: properly track retry time on
passive Fast Open") sets the start of SYNACK retransmission
time on passive Fast Open in "retrans_stamp". However the
timestamp is not reset upon the handshake has completed. As a
result, future data packet retransmission may not update it in
tcp_retransmit_skb(). This may lead to socket aborting earlier
unexpectedly by retransmits_timed_out() since retrans_stamp remains
the SYNACK rtx time.

This bug only manifests on passive TFO sender that a) suffered
SYNACK timeout and then b) stalls on very first loss recovery. Any
successful loss recovery would reset the timestamp to avoid this
issue.

Fixes: c7d13c8f ("tcp: properly track retry time on passive Fast Open")
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

Pull virtio updates from Michael Tsirkin:

 - enable packed ring support for s390

 - several fixes

* tag 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mst/vhost:
  virtio/s390: enable packed ring
  virtio/s390: DMA support for virtio-ccw
  virtio/s390: use vring_create_virtqueue
  virtio/virtio_ring: do some comment fixes
  vhost-scsi: remove incorrect memory barrier
  tools/virtio/ringtest: Remove bogus definition of BUG_ON()
  virtio_ring: Fix potential mem leak in virtqueue_add_indirect_packed

Commit f1b5618e ("vfs: Add a sample program for the new mount API")
added sample programs that get built during the kernel build, but then
cause 'git status' to worry about whether the resulting binaries should
be managed by git.

Tell git not to worry, and to ignore the sample binaries.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Pull more parisc updates from Helge Deller:
 "Two small enhancements, which I didn't included in the last pull
  request because I wanted to keep them a few more days in for-next
  before sending upstream:

   - Replace the ldcw barrier instruction by a nop instruction in the
     CAS code on uniprocessor machines.

   - Map variables read-only after init (enable ro_after_init feature)"

* 'parisc-5.2-2' of git://git.kernel.org/pub/scm/linux/kernel/git/deller/parisc-linux:
  parisc: Use __ro_after_init in init.c
  parisc: Use __ro_after_init in unwind.c
  parisc: Use __ro_after_init in time.c
  parisc: Use __ro_after_init in processor.c
  parisc: Use __ro_after_init in process.c
  parisc: Use __ro_after_init in perf_images.h
  parisc: Use __ro_after_init in pci.c
  parisc: Use __ro_after_init in inventory.c
  parisc: Use __ro_after_init in head.S
  parisc: Use __ro_after_init in firmware.c
  parisc: Use __ro_after_init in drivers.c
  parisc: Use __ro_after_init in cache.c
  parisc: Enable the ro_after_init feature
  parisc: Drop LDCW barrier in CAS code when running UP

Pull kgdb updates from Daniel Thompson:
 "Mostly cleanups but there are also a couple of fixes for out-of-bounds
  accesses (including a potential write to the byte before a static
  buffer).

  The main changes are:

   - Fixes to those out-of-bounds access (empty string to configure test
     module could write the byte before a buffer, high cpu counts could
     read outside of per-cpu structures).

   - Improvements to string handling problems picked up by new compiler
     warnings and other static checks. Most are fixing benign issues
     that can't be tickled without code changes but still reduce the wtf
     factor a little.

   - Tidy up the terminal output"

* tag 'kgdb-5.2-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/danielt/linux:
  kdb: Fix bound check compiler warning
  kdb: do a sanity check on the cpu in kdb_per_cpu()
  kdb: Get rid of broken attempt to print CCVERSION in kdb summary
  misc: kgdbts: fix out-of-bounds access in function param_set_kgdbts_var
  kdb: kdb_support: replace strcpy() by strscpy()
  gdbstub: Replace strcpy() by strscpy()
  gdbstub: mark expected switch fall-throughs

Pull modules updates from Jessica Yu:

 - Use a separate table to store symbol types instead of hijacking
   fields in struct Elf_Sym

 - Trivial code cleanups

* tag 'modules-for-v5.2' of git://git.kernel.org/pub/scm/linux/kernel/git/jeyu/linux:
  module: add stubs for within_module functions
  kallsyms: store type information in its own array
  vmlinux.lds.h: drop unused __vermagic

Daniel Borkmann says:

====================
This set fixes LRU map eviction in combination with map lookups out
of system call side from user space. Main patch is the second one and
test cases are adapted and added in the last one. Thanks!
====================
Acked-by: Andrii Nakryiko <andriin@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>

The test_lru_map is relying on marking the LRU map entry via regular
BPF map lookup from system call side. This is basically for simplicity
reasons. Given we fixed marking entries in that case, the test needs
to be fixed as well. Here we add a small drop-in replacement to retain
existing behavior for the tests by marking out of the BPF program and
transferring the retrieved value out via temporary map. This also adds
new test cases to track the new behavior where two elements are marked,
one via system call side and one via program side, where the next update
then evicts the key looked up only from system call side.

  # ./test_lru_map
  nr_cpus:8

  test_lru_sanity0 (map_type:9 map_flags:0x0): Pass
  test_lru_sanity1 (map_type:9 map_flags:0x0): Pass
  test_lru_sanity2 (map_type:9 map_flags:0x0): Pass
  test_lru_sanity3 (map_type:9 map_flags:0x0): Pass
  test_lru_sanity4 (map_type:9 map_flags:0x0): Pass
  test_lru_sanity5 (map_type:9 map_flags:0x0): Pass
  test_lru_sanity7 (map_type:9 map_flags:0x0): Pass
  test_lru_sanity8 (map_type:9 map_flags:0x0): Pass

  test_lru_sanity0 (map_type:10 map_flags:0x0): Pass
  test_lru_sanity1 (map_type:10 map_flags:0x0): Pass
  test_lru_sanity2 (map_type:10 map_flags:0x0): Pass
  test_lru_sanity3 (map_type:10 map_flags:0x0): Pass
  test_lru_sanity4 (map_type:10 map_flags:0x0): Pass
  test_lru_sanity5 (map_type:10 map_flags:0x0): Pass
  test_lru_sanity7 (map_type:10 map_flags:0x0): Pass
  test_lru_sanity8 (map_type:10 map_flags:0x0): Pass

  test_lru_sanity0 (map_type:9 map_flags:0x2): Pass
  test_lru_sanity4 (map_type:9 map_flags:0x2): Pass
  test_lru_sanity6 (map_type:9 map_flags:0x2): Pass
  test_lru_sanity7 (map_type:9 map_flags:0x2): Pass
  test_lru_sanity8 (map_type:9 map_flags:0x2): Pass

  test_lru_sanity0 (map_type:10 map_flags:0x2): Pass
  test_lru_sanity4 (map_type:10 map_flags:0x2): Pass
  test_lru_sanity6 (map_type:10 map_flags:0x2): Pass
  test_lru_sanity7 (map_type:10 map_flags:0x2): Pass
  test_lru_sanity8 (map_type:10 map_flags:0x2): Pass
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Martin KaFai Lau <kafai@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>

One of the biggest issues we face right now with picking LRU map over
regular hash table is that a map walk out of user space, for example,
to just dump the existing entries or to remove certain ones, will
completely mess up LRU eviction heuristics and wrong entries such
as just created ones will get evicted instead. The reason for this
is that we mark an entry as "in use" via bpf_lru_node_set_ref() from
system call lookup side as well. Thus upon walk, all entries are
being marked, so information of actual least recently used ones
are "lost".

In case of Cilium where it can be used (besides others) as a BPF
based connection tracker, this current behavior causes disruption
upon control plane changes that need to walk the map from user space
to evict certain entries. Discussion result from bpfconf [0] was that
we should simply just remove marking from system call side as no
good use case could be found where it's actually needed there.
Therefore this patch removes marking for regular LRU and per-CPU
flavor. If there ever should be a need in future, the behavior could
be selected via map creation flag, but due to mentioned reason we
avoid this here.

  [0] http://vger.kernel.org/bpfconf.html

Fixes: 29ba732a ("bpf: Add BPF_MAP_TYPE_LRU_HASH")
Fixes: 8f844938 ("bpf: Add BPF_MAP_TYPE_LRU_PERCPU_HASH")
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Martin KaFai Lau <kafai@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>

Add a callback map_lookup_elem_sys_only() that map implementations
could use over map_lookup_elem() from system call side in case the
map implementation needs to handle the latter differently than from
the BPF data path. If map_lookup_elem_sys_only() is set, this will
be preferred pick for map lookups out of user space. This hook is
used in a follow-up fix for LRU map, but once development window
opens, we can convert other map types from map_lookup_elem() (here,
the one called upon BPF_MAP_LOOKUP_ELEM cmd is meant) over to use
the callback to simplify and clean up the latter.
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Martin KaFai Lau <kafai@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>