platform/kernel/linux-starfive.git
3 years agonet: dsa: reference count the MDB entries at the cross-chip notifier level
Vladimir Oltean [Tue, 29 Jun 2021 14:06:50 +0000 (17:06 +0300)]
net: dsa: reference count the MDB entries at the cross-chip notifier level

Ever since the cross-chip notifiers were introduced, the design was
meant to be simplistic and just get the job done without worrying too
much about dangling resources left behind.

For example, somebody installs an MDB entry on sw0p0 in this daisy chain
topology. It gets installed using ds->ops->port_mdb_add() on sw0p0,
sw1p4 and sw2p4.

                                                    |
           sw0p0     sw0p1     sw0p2     sw0p3     sw0p4
        [  user ] [  user ] [  user ] [  dsa  ] [  cpu  ]
        [   x   ] [       ] [       ] [       ] [       ]
                                          |
                                          +---------+
                                                    |
           sw1p0     sw1p1     sw1p2     sw1p3     sw1p4
        [  user ] [  user ] [  user ] [  dsa  ] [  dsa  ]
        [       ] [       ] [       ] [       ] [   x   ]
                                          |
                                          +---------+
                                                    |
           sw2p0     sw2p1     sw2p2     sw2p3     sw2p4
        [  user ] [  user ] [  user ] [  user ] [  dsa  ]
        [       ] [       ] [       ] [       ] [   x   ]

Then the same person deletes that MDB entry. The cross-chip notifier for
deletion only matches sw0p0:

                                                    |
           sw0p0     sw0p1     sw0p2     sw0p3     sw0p4
        [  user ] [  user ] [  user ] [  dsa  ] [  cpu  ]
        [   x   ] [       ] [       ] [       ] [       ]
                                          |
                                          +---------+
                                                    |
           sw1p0     sw1p1     sw1p2     sw1p3     sw1p4
        [  user ] [  user ] [  user ] [  dsa  ] [  dsa  ]
        [       ] [       ] [       ] [       ] [       ]
                                          |
                                          +---------+
                                                    |
           sw2p0     sw2p1     sw2p2     sw2p3     sw2p4
        [  user ] [  user ] [  user ] [  user ] [  dsa  ]
        [       ] [       ] [       ] [       ] [       ]

Why?

Because the DSA links are 'trunk' ports, if we just go ahead and delete
the MDB from sw1p4 and sw2p4 directly, we might delete those multicast
entries when they are still needed. Just consider the fact that somebody
does:

- add a multicast MAC address towards sw0p0 [ via the cross-chip
  notifiers it gets installed on the DSA links too ]
- add the same multicast MAC address towards sw0p1 (another port of that
  same switch)
- delete the same multicast MAC address from sw0p0.

At this point, if we deleted the MAC address from the DSA links, it
would be flooded, even though there is still an entry on switch 0 which
needs it not to.

So that is why deletions only match the targeted source port and nothing
on DSA links. Of course, dangling resources means that the hardware
tables will eventually run out given enough additions/removals, but hey,
at least it's simple.

But there is a bigger concern which needs to be addressed, and that is
our support for SWITCHDEV_OBJ_ID_HOST_MDB. DSA simply translates such an
object into a dsa_port_host_mdb_add() which ends up as ds->ops->port_mdb_add()
on the upstream port, and a similar thing happens on deletion:
dsa_port_host_mdb_del() will trigger ds->ops->port_mdb_del() on the
upstream port.

When there are 2 VLAN-unaware bridges spanning the same switch (which is
a use case DSA proudly supports), each bridge will install its own
SWITCHDEV_OBJ_ID_HOST_MDB entries. But upon deletion, DSA goes ahead and
emits a DSA_NOTIFIER_MDB_DEL for dp->cpu_dp, which is shared between the
user ports enslaved to br0 and the user ports enslaved to br1. Not good.
The host-trapped multicast addresses installed by br1 will be deleted
when any state changes in br0 (IGMP timers expire, or ports leave, etc).

To avoid this, we could of course go the route of the zero-sum game and
delete the DSA_NOTIFIER_MDB_DEL call for dp->cpu_dp. But the better
design is to just admit that on shared ports like DSA links and CPU
ports, we should be reference counting calls, even if this consumes some
dynamic memory which DSA has traditionally avoided. On the flip side,
the hardware tables of switches are limited in size, so it would be good
if the OS managed them properly instead of having them eventually
overflow.

To address the memory usage concern, we only apply the refcounting of
MDB entries on ports that are really shared (CPU ports and DSA links)
and not on user ports. In a typical single-switch setup, this means only
the CPU port (and the host MDB entries are not that many, really).

The name of the newly introduced data structures (dsa_mac_addr) is
chosen in such a way that will be reusable for host FDB entries (next
patch).

With this change, we can finally have the same matching logic for the
MDB additions and deletions, as well as for their host-trapped variants.

Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: dsa: introduce a separate cross-chip notifier type for host MDBs
Vladimir Oltean [Tue, 29 Jun 2021 14:06:49 +0000 (17:06 +0300)]
net: dsa: introduce a separate cross-chip notifier type for host MDBs

Commit abd49535c380 ("net: dsa: execute dsa_switch_mdb_add only for
routing port in cross-chip topologies") does a surprisingly good job
even for the SWITCHDEV_OBJ_ID_HOST_MDB use case, where DSA simply
translates a switchdev object received on dp into a cross-chip notifier
for dp->cpu_dp.

To visualize how that works, imagine the daisy chain topology below and
consider a SWITCHDEV_OBJ_ID_HOST_MDB object emitted on sw2p0. How does
the cross-chip notifier know to match on all the right ports (sw0p4, the
dedicated CPU port, sw1p4, an upstream DSA link, and sw2p4, another
upstream DSA link)?

                                                |
       sw0p0     sw0p1     sw0p2     sw0p3     sw0p4
    [  user ] [  user ] [  user ] [  dsa  ] [  cpu  ]
    [       ] [       ] [       ] [       ] [   x   ]
                                      |
                                      +---------+
                                                |
       sw1p0     sw1p1     sw1p2     sw1p3     sw1p4
    [  user ] [  user ] [  user ] [  dsa  ] [  dsa  ]
    [       ] [       ] [       ] [       ] [   x   ]
                                      |
                                      +---------+
                                                |
       sw2p0     sw2p1     sw2p2     sw2p3     sw2p4
    [  user ] [  user ] [  user ] [  user ] [  dsa  ]
    [       ] [       ] [       ] [       ] [   x   ]

The answer is simple: the dedicated CPU port of sw2p0 is sw0p4, and
dsa_routing_port returns the upstream port for all switches.

That is fine, but there are other topologies where this does not work as
well. There are trees with "H" topologies in the wild, where there are 2
or more switches with DSA links between them, but every switch has its
dedicated CPU port. For these topologies, it seems stupid for the neighbor
switches to install an MDB entry on the routing port, since these
multicast addresses are fundamentally different than the usual ones we
support (and that is the justification for this patch, to introduce the
concept of a termination plane multicast MAC address, as opposed to a
forwarding plane multicast MAC address).

For example, when a SWITCHDEV_OBJ_ID_HOST_MDB would get added to sw0p0,
without this patch, it would get treated as a regular port MDB on sw0p2
and it would match on the ports below (including the sw1p3 routing port).

                         |                                  |
    sw0p0     sw0p1     sw0p2     sw0p3          sw1p3     sw1p2     sw1p1     sw1p0
 [  user ] [  user ] [  cpu  ] [  dsa  ]      [  dsa  ] [  cpu  ] [  user ] [  user ]
 [       ] [       ] [   x   ] [       ] ---- [   x   ] [       ] [       ] [       ]

With the patch, the host MDB notifier on sw0p0 matches only on the local
switch, which is what we want for a termination plane address.

                         |                                  |
    sw0p0     sw0p1     sw0p2     sw0p3          sw1p3     sw1p2     sw1p1     sw1p0
 [  user ] [  user ] [  cpu  ] [  dsa  ]      [  dsa  ] [  cpu  ] [  user ] [  user ]
 [       ] [       ] [   x   ] [       ] ---- [       ] [       ] [       ] [       ]

Name this new matching function "dsa_switch_host_address_match" since we
will be reusing it soon for host FDB entries as well.

Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: dsa: introduce dsa_is_upstream_port and dsa_switch_is_upstream_of
Vladimir Oltean [Tue, 29 Jun 2021 14:06:48 +0000 (17:06 +0300)]
net: dsa: introduce dsa_is_upstream_port and dsa_switch_is_upstream_of

In preparation for the new cross-chip notifiers for host addresses,
let's introduce some more topology helpers which we are going to use to
discern switches that are in our path towards the dedicated CPU port
from switches that aren't.

Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: dsa: delete dsa_legacy_fdb_add and dsa_legacy_fdb_del
Vladimir Oltean [Tue, 29 Jun 2021 14:06:47 +0000 (17:06 +0300)]
net: dsa: delete dsa_legacy_fdb_add and dsa_legacy_fdb_del

We want to add reference counting for FDB entries in cross-chip
topologies, and in order for that to have any chance of working and not
be unbalanced (leading to entries which are never deleted), we need to
ensure that higher layers are sane, because if they aren't, it's garbage
in, garbage out.

For example, if we add a bridge FDB entry twice, the bridge properly
errors out:

$ bridge fdb add dev swp0 00:01:02:03:04:07 master static
$ bridge fdb add dev swp0 00:01:02:03:04:07 master static
RTNETLINK answers: File exists

However, the same thing cannot be said about the bridge bypass
operations:

$ bridge fdb add dev swp0 00:01:02:03:04:07
$ bridge fdb add dev swp0 00:01:02:03:04:07
$ bridge fdb add dev swp0 00:01:02:03:04:07
$ bridge fdb add dev swp0 00:01:02:03:04:07
$ echo $?
0

But one 'bridge fdb del' is enough to remove the entry, no matter how
many times it was added.

The bridge bypass operations are impossible to maintain in these
circumstances and lack of support for reference counting the cross-chip
notifiers is holding us back from making further progress, so just drop
support for them. The only way left for users to install static bridge
FDB entries is the proper one, using the "master static" flags.

With this change, rtnl_fdb_add() falls back to calling
ndo_dflt_fdb_add() which uses the duplicate-exclusive variant of
dev_uc_add(): dev_uc_add_excl(). Because DSA does not (yet) declare
IFF_UNICAST_FLT, this results in us going to promiscuous mode:

$ bridge fdb add dev swp0 00:01:02:03:04:05
[   28.206743] device swp0 entered promiscuous mode
$ bridge fdb add dev swp0 00:01:02:03:04:05
RTNETLINK answers: File exists

So even if it does not completely fail, there is at least some indication
that it is behaving differently from before, and closer to user space
expectations, I would argue (the lack of a "local|static" specifier
defaults to "local", or "host-only", so dev_uc_add() is a reasonable
default implementation). If the generic implementation of .ndo_fdb_add
provided by Vlad Yasevich is a proof of anything, it only proves that
the implementation provided by DSA was always wrong, by not looking at
"ndm->ndm_state & NUD_NOARP" (the "static" flag which means that the FDB
entry points outwards) and "ndm->ndm_state & NUD_PERMANENT" (the "local"
flag which means that the FDB entry points towards the host). It all
used to mean the same thing to DSA.

Update the documentation so that the users are not confused about what's
going on.

Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: bridge: allow br_fdb_replay to be called for the bridge device
Vladimir Oltean [Tue, 29 Jun 2021 14:06:46 +0000 (17:06 +0300)]
net: bridge: allow br_fdb_replay to be called for the bridge device

When a port joins a bridge which already has local FDB entries pointing
to the bridge device itself, we would like to offload those, so allow
the "dev" argument to be equal to the bridge too. The code already does
what we need in that case.

Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: bridge: switchdev: send FDB notifications for host addresses
Tobias Waldekranz [Tue, 29 Jun 2021 14:06:45 +0000 (17:06 +0300)]
net: bridge: switchdev: send FDB notifications for host addresses

Treat addresses added to the bridge itself in the same way as regular
ports and send out a notification so that drivers may sync it down to
the hardware FDB.

Signed-off-by: Tobias Waldekranz <tobias@waldekranz.com>
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: bridge: use READ_ONCE() and WRITE_ONCE() compiler barriers for fdb->dst
Vladimir Oltean [Tue, 29 Jun 2021 14:06:44 +0000 (17:06 +0300)]
net: bridge: use READ_ONCE() and WRITE_ONCE() compiler barriers for fdb->dst

Annotate the writer side of fdb->dst:

- fdb_create()
- br_fdb_update()
- fdb_add_entry()
- br_fdb_external_learn_add()

with WRITE_ONCE() and the reader side:

- br_fdb_test_addr()
- br_fdb_update()
- fdb_fill_info()
- fdb_add_entry()
- fdb_delete_by_addr_and_port()
- br_fdb_external_learn_add()
- br_switchdev_fdb_notify()

with compiler barriers such that the readers do not attempt to reload
fdb->dst multiple times, leading to potentially different destination
ports when the fdb entry is updated concurrently.

This is especially important in read-side sections where fdb->dst is
used more than once, but let's convert all accesses for the sake of
uniformity.

Suggested-by: Nikolay Aleksandrov <nikolay@nvidia.com>
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agoMerge branch 'do_once_lite'
David S. Miller [Mon, 28 Jun 2021 22:54:57 +0000 (15:54 -0700)]
Merge branch 'do_once_lite'

Tanner Love says:

====================
net: update netdev_rx_csum_fault() print dump only once

First patch implements DO_ONCE_LITE to abstract uses of the ".data.once"
trick. It is defined in its own, new header file  -- rather than
alongside the existing DO_ONCE in include/linux/once.h -- because
include/linux/once.h includes include/linux/jump_label.h, and this
causes the build to break for some architectures if
include/linux/once.h is included in include/linux/printk.h or
include/asm-generic/bug.h.

Second patch uses DO_ONCE_LITE in netdev_rx_csum_fault to print dump
only once.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: update netdev_rx_csum_fault() print dump only once
Tanner Love [Mon, 28 Jun 2021 13:50:07 +0000 (09:50 -0400)]
net: update netdev_rx_csum_fault() print dump only once

Printing this stack dump multiple times does not provide additional
useful information, and consumes time in the data path. Printing once
is sufficient.

Changes
  v2: Format indentation properly

Signed-off-by: Tanner Love <tannerlove@google.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Acked-by: Mahesh Bandewar <maheshb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agoonce: implement DO_ONCE_LITE for non-fast-path "do once" functionality
Tanner Love [Mon, 28 Jun 2021 13:50:06 +0000 (09:50 -0400)]
once: implement DO_ONCE_LITE for non-fast-path "do once" functionality

Certain uses of "do once" functionality reside outside of fast path,
and so do not require jump label patching via static keys, making
existing DO_ONCE undesirable in such cases.

Replace uses of __section(".data.once") with DO_ONCE_LITE(_IF)?

This patch changes the return values of xfs_printk_once, printk_once,
and printk_deferred_once. Before, they returned whether the print was
performed, but now, they always return true. This is okay because the
return values of the following macros are entirely ignored throughout
the kernel:
- xfs_printk_once
- xfs_warn_once
- xfs_notice_once
- xfs_info_once
- printk_once
- pr_emerg_once
- pr_alert_once
- pr_crit_once
- pr_err_once
- pr_warn_once
- pr_notice_once
- pr_info_once
- pr_devel_once
- pr_debug_once
- printk_deferred_once
- orc_warn

Changes
v3:
  - Expand commit message to explain why changing return values of
    xfs_printk_once, printk_once, printk_deferred_once is benign
v2:
  - Fix i386 build warnings

Signed-off-by: Tanner Love <tannerlove@google.com>
Acked-by: Eric Dumazet <edumazet@google.com>
Acked-by: Mahesh Bandewar <maheshb@google.com>
Acked-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: sparx5: Do not use mac_addr uninitialized in mchp_sparx5_probe()
Nathan Chancellor [Sun, 27 Jun 2021 18:45:43 +0000 (11:45 -0700)]
net: sparx5: Do not use mac_addr uninitialized in mchp_sparx5_probe()

Clang warns:

drivers/net/ethernet/microchip/sparx5/sparx5_main.c:760:29: warning:
variable 'mac_addr' is uninitialized when used here [-Wuninitialized]
        if (of_get_mac_address(np, mac_addr)) {
                                   ^~~~~~~~
drivers/net/ethernet/microchip/sparx5/sparx5_main.c:669:14: note:
initialize the variable 'mac_addr' to silence this warning
        u8 *mac_addr;
                    ^
                     = NULL
1 warning generated.

mac_addr is only used to store the value retrieved from
of_get_mac_address(), which is then copied into the base_mac member of
the sparx5 struct using ether_addr_copy(). It is easier to just use the
base_mac address directly, which avoids the warning and the extra copy.

Fixes: 3cfa11bac9bb ("net: sparx5: add the basic sparx5 driver")
Link: https://github.com/ClangBuiltLinux/linux/issues/1413
Signed-off-by: Nathan Chancellor <nathan@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: dsa: sja1105: fix dynamic access to L2 Address Lookup table for SJA1110
Vladimir Oltean [Sun, 27 Jun 2021 14:27:08 +0000 (17:27 +0300)]
net: dsa: sja1105: fix dynamic access to L2 Address Lookup table for SJA1110

The SJA1105P/Q/R/S and SJA1110 may have the same layout for the command
to read/write/search for L2 Address Lookup entries, but as explained in
the comments at the beginning of the sja1105_dynamic_config.c file, the
command portion of the buffer is at the end, and we need to obtain a
pointer to it by adding the length of the entry to the buffer.

Alas, the length of an L2 Address Lookup entry is larger in SJA1110 than
it is for SJA1105P/Q/R/S, so we need to create a common helper to access
the command buffer, and this receives as argument the length of the
entry buffer.

Fixes: 3e77e59bf8cf ("net: dsa: sja1105: add support for the SJA1110 switch family")
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: bridge: mrp: Update the Test frames for MRA
Horatiu Vultur [Sat, 26 Jun 2021 20:18:04 +0000 (22:18 +0200)]
net: bridge: mrp: Update the Test frames for MRA

According to the standard IEC 62439-2, in case the node behaves as MRA
and needs to send Test frames on ring ports, then these Test frames need
to have an Option TLV and a Sub-Option TLV which has the type AUTO_MGR.

Signed-off-by: Horatiu Vultur <horatiu.vultur@microchip.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agoMerge tag 'for-net-next-2021-06-28' of git://git.kernel.org/pub/scm/linux/kernel...
David S. Miller [Mon, 28 Jun 2021 22:35:50 +0000 (15:35 -0700)]
Merge tag 'for-net-next-2021-06-28' of git://git./linux/kernel/git/bluetooth/bluetooth-next

Luiz Augusto von Dentz says:

====================
bluetooth-next pull request for net-next:

 - Add support for QCA_ROME device (0cf3:e500) and RTL8822CE
 - Update management interface revision to 21
 - Use of incluse language
 - Proper handling of HCI_LE_Advertising_Set_Terminated event
 - Recovery handing of HCI ncmd=0
 - Various memory fixes
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf-next
David S. Miller [Mon, 28 Jun 2021 22:28:03 +0000 (15:28 -0700)]
Merge git://git./linux/kernel/git/bpf/bpf-next

Daniel Borkmann says:

====================
pull-request: bpf-next 2021-06-28

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

We've added 37 non-merge commits during the last 12 day(s) which contain
a total of 56 files changed, 394 insertions(+), 380 deletions(-).

The main changes are:

1) XDP driver RCU cleanups, from Toke Høiland-Jørgensen and Paul E. McKenney.

2) Fix bpf_skb_change_proto() IPv4/v6 GSO handling, from Maciej Żenczykowski.

3) Fix false positive kmemleak report for BPF ringbuf alloc, from Rustam Kovhaev.

4) Fix x86 JIT's extable offset calculation for PROBE_LDX NULL, from Ravi Bangoria.

5) Enable libbpf fallback probing with tracing under RHEL7, from Jonathan Edwards.

6) Clean up x86 JIT to remove unused cnt tracking from EMIT macro, from Jiri Olsa.

7) Netlink cleanups for libbpf to please Coverity, from Kumar Kartikeya Dwivedi.

8) Allow to retrieve ancestor cgroup id in tracing programs, from Namhyung Kim.

9) Fix lirc BPF program query to use user-provided prog_cnt, from Sean Young.

10) Add initial libbpf doc including generated kdoc for its API, from Grant Seltzer.

11) Make xdp_rxq_info_unreg_mem_model() more robust, from Jakub Kicinski.

12) Fix up bpfilter startup log-level to info level, from Gary Lin.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agoipv6: ICMPV6: add response to ICMPV6 RFC 8335 PROBE messages
Andreas Roeseler [Sat, 26 Jun 2021 14:07:46 +0000 (09:07 -0500)]
ipv6: ICMPV6: add response to ICMPV6 RFC 8335 PROBE messages

This patch builds off of commit 2b246b2569cd2ac6ff700d0dce56b8bae29b1842
and adds functionality to respond to ICMPV6 PROBE requests.

Add icmp_build_probe function to construct PROBE requests for both
ICMPV4 and ICMPV6.

Modify icmpv6_rcv to detect ICMPV6 PROBE messages and call the
icmpv6_echo_reply handler.

Modify icmpv6_echo_reply to build a PROBE response message based on the
queried interface.

This patch has been tested using a branch of the iputils git repo which can
be found here: https://github.com/Juniper-Clinic-2020/iputils/tree/probe-request

Signed-off-by: Andreas Roeseler <andreas.a.roeseler@gmail.com>
Reviewed-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: sparx5: fix error return code in sparx5_register_notifier_blocks()
Yang Yingliang [Sat, 26 Jun 2021 04:44:20 +0000 (12:44 +0800)]
net: sparx5: fix error return code in sparx5_register_notifier_blocks()

Fix to return a negative error code from the error handling
case instead of 0, as done elsewhere in this function.

Fixes: d6fce5141929 ("net: sparx5: add switching support")
Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: Yang Yingliang <yangyingliang@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: sparx5: fix return value check in sparx5_create_targets()
Yang Yingliang [Sat, 26 Jun 2021 04:44:19 +0000 (12:44 +0800)]
net: sparx5: fix return value check in sparx5_create_targets()

In case of error, the function devm_ioremap() returns NULL pointer
not ERR_PTR(). The IS_ERR() test in the return value check should
be replaced with NULL test.

Fixes: 3cfa11bac9bb ("net: sparx5: add the basic sparx5 driver")
Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: Yang Yingliang <yangyingliang@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: sparx5: check return value after calling platform_get_resource()
Yang Yingliang [Sat, 26 Jun 2021 04:44:18 +0000 (12:44 +0800)]
net: sparx5: check return value after calling platform_get_resource()

It will cause null-ptr-deref if platform_get_resource() returns NULL,
we need check the return value.

Fixes: 3cfa11bac9bb ("net: sparx5: add the basic sparx5 driver")
Signed-off-by: Yang Yingliang <yangyingliang@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agoMerge tag 'mlx5-updates-2021-06-26' of git://git.kernel.org/pub/scm/linux/kernel...
David S. Miller [Mon, 28 Jun 2021 21:15:07 +0000 (14:15 -0700)]
Merge tag 'mlx5-updates-2021-06-26' of git://git./linux/kernel/git/saeed/linux

Saeed Mahameed says:

====================
mlx5-updates-2021-06-26

This series provides small updates to mlx5 driver.

1) Increase hairpin buffer size

2) Improve peroformance in SF allocation

3) Add IPsec support to uplink representor

4) Add stats for number of deleted kTLS TX offloaded connections

5) Add support for flow sampler in SW steering
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agoMerge branch 'bridge-replay-helpers'
David S. Miller [Mon, 28 Jun 2021 21:09:04 +0000 (14:09 -0700)]
Merge branch 'bridge-replay-helpers'

Vladimir Oltean says:

====================
Cleanup for the bridge replay helpers

This patch series brings some improvements to the logic added to the
bridge and DSA to handle LAG interfaces sandwiched between a bridge and
a DSA switch port.

        br0
        /  \
       /    \
     bond0  swp2
     /  \
    /    \
  swp0  swp1

In particular, it ensures that the switchdev object additions and
deletions are well balanced per physical port. This is important for
future work in the area of offloading local bridge FDB entries to
hardware in the context of DSA requesting a replay of those entries at
bridge join time (this will be submitted in a future patch series).
Due to some difficulty ensuring that the deletion of local FDB entries
pointing towards the bridge device itself is notified to switchdev in
time (before the switchdev port disconnects from the bridge), this is
potentially still not the final form in which the replay helpers will
exist. I'm thinking about moving from the pull mode (in which DSA
requests the replay) to a push mode (in which the bridge initiates the
replay). Nonetheless, these preliminary changes are needed either way.

The patch series also addresses some feedback from Nikolai which is long
overdue by now (sorry).

Switchdev driver maintainers were deliberately omitted due to the
trivial nature of the driver changes (just a function prototype).

Changes in v2:
- fix build issue in patch 4 (function prototype mismatch)
- move switchdev object unsync to the NETDEV_PRECHANGEUPPER code path
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: dsa: replay a deletion of switchdev objects for ports leaving a bridged LAG
Vladimir Oltean [Sun, 27 Jun 2021 11:54:29 +0000 (14:54 +0300)]
net: dsa: replay a deletion of switchdev objects for ports leaving a bridged LAG

When a DSA switch port leaves a bonding interface that is under a
bridge, there might be dangling switchdev objects on that port left
behind, because the bridge is not aware that its lower interface (the
bond) changed state in any way.

Call the bridge replay helpers with adding=false before changing
dp->bridge_dev to NULL, because we need to simulate to
dsa_slave_port_obj_del() that these notifications were emitted by the
bridge.

We add this hook to the NETDEV_PRECHANGEUPPER event handler, because
we are calling into switchdev (and the __switchdev_handle_port_obj_del
fanout helpers expect the upper/lower adjacency lists to still be valid)
and PRECHANGEUPPER is the last moment in time when they still are.

Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: dsa: refactor the prechangeupper sanity checks into a dedicated function
Vladimir Oltean [Sun, 27 Jun 2021 11:54:28 +0000 (14:54 +0300)]
net: dsa: refactor the prechangeupper sanity checks into a dedicated function

We need to add more logic to the DSA NETDEV_PRECHANGEUPPER event
handler, more exactly we need to request an unsync of switchdev objects.
In order to fit more code, refactor the existing logic into a helper.

Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: bridge: allow the switchdev replay functions to be called for deletion
Vladimir Oltean [Sun, 27 Jun 2021 11:54:27 +0000 (14:54 +0300)]
net: bridge: allow the switchdev replay functions to be called for deletion

When a switchdev port leaves a LAG that is a bridge port, the switchdev
objects and port attributes offloaded to that port are not removed:

ip link add br0 type bridge
ip link add bond0 type bond mode 802.3ad
ip link set swp0 master bond0
ip link set bond0 master br0
bridge vlan add dev bond0 vid 100
ip link set swp0 nomaster

VLAN 100 will remain installed on swp0 despite it going into standalone
mode, because as far as the bridge is concerned, nothing ever happened
to its bridge port.

Let's extend the bridge vlan, fdb and mdb replay functions to take a
'bool adding' argument, and make DSA and ocelot call the replay
functions with 'adding' as false from the switchdev unsync path, for the
switch port that leaves the bridge.

Note that this patch in itself does not salvage anything, because in the
current pull mode of operation, DSA still needs to call the replay
helpers with adding=false. This will be done in another patch.

Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: bridge: constify variables in the replay helpers
Vladimir Oltean [Sun, 27 Jun 2021 11:54:26 +0000 (14:54 +0300)]
net: bridge: constify variables in the replay helpers

Some of the arguments and local variables for the newly added switchdev
replay helpers can be const, so let's make them so.

Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: bridge: ignore switchdev events for LAG ports which didn't request replay
Vladimir Oltean [Sun, 27 Jun 2021 11:54:25 +0000 (14:54 +0300)]
net: bridge: ignore switchdev events for LAG ports which didn't request replay

There is a slight inconvenience in the switchdev replay helpers added
recently, and this is when:

ip link add br0 type bridge
ip link add bond0 type bond
ip link set bond0 master br0
bridge vlan add dev bond0 vid 100
ip link set swp0 master bond0
ip link set swp1 master bond0

Since the underlying driver (currently only DSA) asks for a replay of
VLANs when swp0 and swp1 join the LAG because it is bridged, what will
happen is that DSA will try to react twice on the VLAN event for swp0.
This is not really a huge problem right now, because most drivers accept
duplicates since the bridge itself does, but it will become a problem
when we add support for replaying switchdev object deletions.

Let's fix this by adding a blank void *ctx in the replay helpers, which
will be passed on by the bridge in the switchdev notifications. If the
context is NULL, everything is the same as before. But if the context is
populated with a valid pointer, the underlying switchdev driver
(currently DSA) can use the pointer to 'see through' the bridge port
(which in the example above is bond0) and 'know' that the event is only
for a particular physical port offloading that bridge port, and not for
all of them.

Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: switchdev: add a context void pointer to struct switchdev_notifier_info
Vladimir Oltean [Sun, 27 Jun 2021 11:54:24 +0000 (14:54 +0300)]
net: switchdev: add a context void pointer to struct switchdev_notifier_info

In the case where the driver asks for a replay of a certain type of
event (port object or attribute) for a bridge port that is a LAG, it may
do so because this port has just joined the LAG.

But there might already be other switchdev ports in that LAG, and it is
preferable that those preexisting switchdev ports do not act upon the
replayed event.

The solution is to add a context to switchdev events, which is NULL most
of the time (when the bridge layer initiates the call) but which can be
set to a value controlled by the switchdev driver when a replay is
requested. The driver can then check the context to figure out if all
ports within the LAG should act upon the switchdev event, or just the
ones that match the context.

We have to modify all switchdev_handle_* helper functions as well as the
prototypes in the drivers that use these helpers too, because these
helpers hide the underlying struct switchdev_notifier_info from us and
there is no way to retrieve the context otherwise.

The context structure will be populated and used in later patches.

Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: ocelot: delete call to br_fdb_replay
Vladimir Oltean [Sun, 27 Jun 2021 11:54:23 +0000 (14:54 +0300)]
net: ocelot: delete call to br_fdb_replay

Not using this driver, I did not realize it doesn't react to
SWITCHDEV_FDB_{ADD,DEL}_TO_DEVICE notifications, but it implements just
the bridge bypass operations (.ndo_fdb_{add,del}). So the call to
br_fdb_replay just produces notifications that are ignored, delete it
for now.

Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: bridge: include the is_local bit in br_fdb_replay
Vladimir Oltean [Sun, 27 Jun 2021 11:54:22 +0000 (14:54 +0300)]
net: bridge: include the is_local bit in br_fdb_replay

Since commit 2c4eca3ef716 ("net: bridge: switchdev: include local flag
in FDB notifications"), the bridge emits SWITCHDEV_FDB_ADD_TO_DEVICE
events with the is_local flag populated (but we ignore it nonetheless).

We would like DSA to start treating this bit, but it is still not
populated by the replay helper, so add it there too.

Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agoxdp: Move the rxq_info.mem clearing to unreg_mem_model()
Jakub Kicinski [Fri, 25 Jun 2021 22:16:12 +0000 (15:16 -0700)]
xdp: Move the rxq_info.mem clearing to unreg_mem_model()

xdp_rxq_info_unreg() implicitly calls xdp_rxq_info_unreg_mem_model().
This may well be confusing to the driver authors, and lead to double free
if they call xdp_rxq_info_unreg_mem_model() before xdp_rxq_info_unreg()
(when mem model type == MEM_TYPE_PAGE_POOL).

In fact error path of mvpp2_rxq_init() seems to currently do exactly that.

The double free will result in refcount underflow in page_pool_destroy().
Make the interface a little more programmer friendly by clearing type and
id so that xdp_rxq_info_unreg_mem_model() can be called multiple times.

Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/bpf/20210625221612.2637086-1-kuba@kernel.org
3 years agoMerge branch 'bnxt_en-ptp'
David S. Miller [Mon, 28 Jun 2021 20:41:06 +0000 (13:41 -0700)]
Merge branch 'bnxt_en-ptp'

Michael Chan says:

====================
bnxt_en: Add hardware PTP timestamping support on 575XX devices

Add PTP RX and TX hardware timestamp support on 575XX devices.  These
devices use the two-step method to implement the IEEE-1588 timestamping
support.

v2: Add spinlock to serialize access to the timecounter.
    Use .do_aux_work() for the periodic timer reading and to get the TX
    timestamp from the firmware.
    Propagate error code from ptp_clock_register().
    Make the 64-bit timer access safe on 32-bit CPUs.
    Read PHC using direct register access.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agobnxt_en: Enable hardware PTP support
Michael Chan [Sun, 27 Jun 2021 17:19:50 +0000 (13:19 -0400)]
bnxt_en: Enable hardware PTP support

Call bnxt_ptp_init() to initialize and register with the clock driver
to enable PTP support.  Call bnxt_ptp_free() to unregister and clean
up during shutdown.

Reviewed-by: Edwin Peer <edwin.peer@broadcom.com>
Reviewed-by: Pavan Chebbi <pavan.chebbi@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agobnxt_en: Transmit and retrieve packet timestamps
Pavan Chebbi [Sun, 27 Jun 2021 17:19:49 +0000 (13:19 -0400)]
bnxt_en: Transmit and retrieve packet timestamps

Setup the TXBD to enable TX timestamp if requested.  At TX packet DMA
completion, if we requested TX timestamp on that packet, we defer to
.do_aux_work() to obtain the TX timestamp from the firmware before we
free the TX SKB.

v2: Use .do_aux_work() to get the TX timestamp from firmware.

Reviewed-by: Edwin Peer <edwin.peer@broadcom.com>
Signed-off-by: Pavan Chebbi <pavan.chebbi@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agobnxt_en: Get the RX packet timestamp
Pavan Chebbi [Sun, 27 Jun 2021 17:19:48 +0000 (13:19 -0400)]
bnxt_en: Get the RX packet timestamp

If the RX packet is timestamped by the hardware, the RX completion
record will contain the lower 32-bit of the timestamp.  This needs
to be combined with the upper 16-bit of the periodic timestamp that
we get from the timer.  The previous snapshot in ptp->old_timer is
used to make sure that the snapshot is not ahead of the RX timestamp
and we adjust for wrap-around if needed.

v2: Make ptp->old_time read access safe on 32-bit CPUs.

Reviewed-by: Edwin Peer <edwin.peer@broadcom.com>
Signed-off-by: Pavan Chebbi <pavan.chebbi@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agobnxt_en: Get the full 48-bit hardware timestamp periodically
Pavan Chebbi [Sun, 27 Jun 2021 17:19:47 +0000 (13:19 -0400)]
bnxt_en: Get the full 48-bit hardware timestamp periodically

From the bnxt_timer(), read the 48-bit hardware running clock
periodically and store it in ptp->current_time.  The previous snapshot
of the clock will be stored in ptp->old_time.  The old_time snapshot
will be used in the next patches to compute the RX packet timestamps.

v2: Use .do_aux_work() to read the timer periodically.

Reviewed-by: Edwin Peer <edwin.peer@broadcom.com>
Signed-off-by: Pavan Chebbi <pavan.chebbi@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agobnxt_en: Add PTP clock APIs, ioctls, and ethtool methods
Michael Chan [Sun, 27 Jun 2021 17:19:46 +0000 (13:19 -0400)]
bnxt_en: Add PTP clock APIs, ioctls, and ethtool methods

Add the clock APIs to set/get/adjust the hw clock, and the related
ioctls and ethtool methods.

v2: Propagate error code from ptp_clock_register().
    Add spinlock to serialize access to the timecounter.  The
    timecounter is accessed in process context and the RX datapath.
    Read the PHC using direct registers.

Reviewed-by: Edwin Peer <edwin.peer@broadcom.com>
Signed-off-by: Pavan Chebbi <pavan.chebbi@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agobnxt_en: Get PTP hardware capability from firmware
Michael Chan [Sun, 27 Jun 2021 17:19:45 +0000 (13:19 -0400)]
bnxt_en: Get PTP hardware capability from firmware

Store PTP hardware info in a structure if hardware and firmware support PTP.

Reviewed-by: Edwin Peer <edwin.peer@broadcom.com>
Reviewed-by: Pavan Chebbi <pavan.chebbi@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agobnxt_en: Update firmware interface to 1.10.2.47
Michael Chan [Sun, 27 Jun 2021 17:19:44 +0000 (13:19 -0400)]
bnxt_en: Update firmware interface to 1.10.2.47

Adding the PTP related firmware interface is the main change.

There is also a name change for admin_mtu, requiring code fixup.

Reviewed-by: Pavan Chebbi <pavan.chebbi@broadcom.com>
Signed-off-by: Edwin Peer <edwin.peer@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agoMerge branch 'hns3-next'
David S. Miller [Mon, 28 Jun 2021 20:34:58 +0000 (13:34 -0700)]
Merge branch 'hns3-next'

Guangbin Huang says:

====================
net: hns3: add new debugfs commands

This series adds three new debugfs commands for the HNS3 ethernet driver.

change log:
V1 -> V2:
1. remove patch "net: hns3: add support for link diagnosis info in debugfs"
   and use ethtool extended link state to implement similar function
   according to Jakub Kicinski's opinion.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: hns3: add support for dumping MAC umv counter in debugfs
Jian Shen [Sat, 26 Jun 2021 01:00:17 +0000 (09:00 +0800)]
net: hns3: add support for dumping MAC umv counter in debugfs

This patch adds support of dumping MAC umv counter in debugfs,
which will be helpful for debugging.

The display style is below:
$ cat umv_info
num_alloc_vport  : 2
max_umv_size     : 256
wanted_umv_size  : 256
priv_umv_size    : 85
share_umv_size   : 86
vport(0) used_umv_num : 1
vport(1) used_umv_num : 1

Signed-off-by: Jian Shen <shenjian15@huawei.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: hns3: add support for FD counter in debugfs
Jian Shen [Sat, 26 Jun 2021 01:00:16 +0000 (09:00 +0800)]
net: hns3: add support for FD counter in debugfs

Previously, the flow director counter is not enabled. To improve the
maintainability for chechking whether flow director hit or not, enable
flow director counter for each function, and add debugfs query inerface
to query the counters for each function.

The debugfs command is below:
cat fd_counter
func_id    hit_times
pf         0
vf0        0
vf1        0

Signed-off-by: Jian Shen <shenjian15@huawei.com>
Signed-off-by: Guangbin Huang <huangguangbin2@huawei.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agoMerge branch 'tipc-next'
David S. Miller [Mon, 28 Jun 2021 20:31:57 +0000 (13:31 -0700)]
Merge branch 'tipc-next'

Menglong Dong says:

====================
net: tipc: fix FB_MTU eat two pages and do some code cleanup

In the first patch, FB_MTU is redefined to make sure data size will not
exceed PAGE_SIZE. Besides, I removed the alignment for buf_size in
tipc_buf_acquire, because skb_alloc_fclone will do the alignment job.

In the second patch, I removed align() in msg.c and replace it with
ALIGN().

Changes since V5:
- remove blank line after Fixes in commit log in the first patch

Changes since V4:
- remove ONE_PAGE_SKB_SZ and replace it with one_page_mtu in the first
  patch.
- fix some code style problems for the second patch.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: tipc: replace align() with ALIGN in msg.c
Menglong Dong [Mon, 28 Jun 2021 06:37:45 +0000 (23:37 -0700)]
net: tipc: replace align() with ALIGN in msg.c

The function align() which is defined in msg.c is redundant, replace it
with ALIGN() and introduce a BUF_ALIGN().

Signed-off-by: Menglong Dong <dong.menglong@zte.com.cn>
Acked-by: Jon Maloy <jmaloy@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: tipc: fix FB_MTU eat two pages
Menglong Dong [Mon, 28 Jun 2021 06:37:44 +0000 (23:37 -0700)]
net: tipc: fix FB_MTU eat two pages

FB_MTU is used in 'tipc_msg_build()' to alloc smaller skb when memory
allocation fails, which can avoid unnecessary sending failures.

The value of FB_MTU now is 3744, and the data size will be:

  (3744 + SKB_DATA_ALIGN(sizeof(struct skb_shared_info)) + \
    SKB_DATA_ALIGN(BUF_HEADROOM + BUF_TAILROOM + 3))

which is larger than one page(4096), and two pages will be allocated.

To avoid it, replace '3744' with a calculation:

  (PAGE_SIZE - SKB_DATA_ALIGN(BUF_OVERHEAD) - \
    SKB_DATA_ALIGN(sizeof(struct skb_shared_info)))

What's more, alloc_skb_fclone() will call SKB_DATA_ALIGN for data size,
and it's not necessary to make alignment for buf_size in
tipc_buf_acquire(). So, just remove it.

Fixes: 4c94cc2d3d57 ("tipc: fall back to smaller MTU if allocation of local send skb fails")
Signed-off-by: Menglong Dong <dong.menglong@zte.com.cn>
Acked-by: Jon Maloy <jmaloy@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agoMerge branch 'master' of git://git.kernel.org/pub/scm/linux/kernel/git
David S. Miller [Mon, 28 Jun 2021 20:17:16 +0000 (13:17 -0700)]
Merge branch 'master' of git://git./linux/kernel/git
/klassert/ipsec-next

Steffen Klassert says:

====================
pull request (net-next): ipsec-next 2021-06-28

1) Remove an unneeded error assignment in esp4_gro_receive().
   From Yang Li.

2) Add a new byseq state hashtable to find acquire states faster.
   From Sabrina Dubroca.

3) Remove some unnecessary variables in pfkey_create().
   From zuoqilin.

4) Remove the unused description from xfrm_type struct.
   From Florian Westphal.

5) Fix a spelling mistake in the comment of xfrm_state_ok().
   From gushengxian.

6) Replace hdr_off indirections by a small helper function.
   From Florian Westphal.

7) Remove xfrm4_output_finish and xfrm6_output_finish declarations,
   they are not used anymore.From Antony Antony.

8) Remove xfrm replay indirections.
   From Florian Westphal.

Please pull or let me know if there are problems.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agoMerge tag 'mac80211-next-for-net-next-2021-06-25' of git://git.kernel.org/pub/scm...
David S. Miller [Mon, 28 Jun 2021 20:06:12 +0000 (13:06 -0700)]
Merge tag 'mac80211-next-for-net-next-2021-06-25' of git://git./linux/kernel/git/jberg/mac80211-next

Johannes berg says:

====================
Lots of changes:
 * aggregation handling improvements for some drivers
 * hidden AP discovery on 6 GHz and other HE 6 GHz
   improvements
 * minstrel improvements for no-ack frames
 * deferred rate control for TXQs to improve reaction
   times
 * virtual time-based airtime scheduler
 * along with various little cleanups/fixups
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agomptcp: fix 'masking a bool' warning
Matthieu Baerts [Fri, 25 Jun 2021 21:25:22 +0000 (14:25 -0700)]
mptcp: fix 'masking a bool' warning

Dan Carpenter reported an issue introduced in
commit fde56eea01f9 ("mptcp: refine mptcp_cleanup_rbuf") where a new
boolean (ack_pending) is masked with 0x9.

This is not the intention to ignore values by using a boolean. This
variable should not have a 'bool' type: we should keep the 'u8' to allow
this comparison.

Fixes: fde56eea01f9 ("mptcp: refine mptcp_cleanup_rbuf")
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Signed-off-by: Mat Martineau <mathew.j.martineau@linux.intel.com>
Acked-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agoMerge branch 'reset-mac'
David S. Miller [Mon, 28 Jun 2021 19:44:18 +0000 (12:44 -0700)]
Merge branch 'reset-mac'

Guillaume Nault says:

====================
net: reset MAC header consistently across L3 virtual devices

Some virtual L3 devices, like vxlan-gpe and gre (in collect_md mode),
reset the MAC header pointer after they parsed the outer headers. This
accurately reflects the fact that the decapsulated packet is pure L3
packet, as that makes the MAC header 0 bytes long (the MAC and network
header pointers are equal).

However, many L3 devices only adjust the network header after
decapsulation and leave the MAC header pointer to its original value.
This can confuse other parts of the networking stack, like TC, which
then considers the outer headers as one big MAC header.

This patch series makes the following L3 tunnels behave like VXLAN-GPE:
bareudp, ipip, sit, gre, ip6gre, ip6tnl, gtp.

The case of gre is a bit special. It already resets the MAC header
pointer in collect_md mode, so only the classical mode needs to be
adjusted. However, gre also has a special case that expects the MAC
header pointer to keep pointing to the outer header even after
decapsulation. Therefore, patch 4 keeps an exception for this case.

Ideally, we'd centralise the call to skb_reset_mac_header() in
ip_tunnel_rcv(), to avoid manual calls in ipip (patch 2),
sit (patch 3) and gre (patch 4). That's unfortunately not feasible
currently, because of the gre special case discussed above that
precludes us from resetting the MAC header unconditionally.

The original motivation is to redirect bareudp packets to Ethernet
devices (as described in patch 1). The rest of this series aims at
bringing consistency across all L3 devices (apart from gre's special
case unfortunately).

Note: the gtp patch results from pure code inspection and has been
compiled tested only.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agogtp: reset mac_header after decap
Guillaume Nault [Fri, 25 Jun 2021 13:33:23 +0000 (15:33 +0200)]
gtp: reset mac_header after decap

For consistency with other L3 tunnel devices, reset the mac_header
pointer after decapsulation. This makes the mac_header 0 bytes long,
thus making it clear that this skb has no mac_header.

Compile tested only.

Signed-off-by: Guillaume Nault <gnault@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agoip6_tunnel: allow redirecting ip6gre and ipxip6 packets to eth devices
Guillaume Nault [Fri, 25 Jun 2021 13:33:17 +0000 (15:33 +0200)]
ip6_tunnel: allow redirecting ip6gre and ipxip6 packets to eth devices

Reset the mac_header pointer even when the tunnel transports only L3
data (in the ARPHRD_ETHER case, this is already done by eth_type_trans).
This prevents other parts of the stack from mistakenly accessing the
outer header after the packet has been decapsulated.

In practice, this allows to push an Ethernet header to ipip6, ip6ip6,
mplsip6 or ip6gre packets and redirect them to an Ethernet device:

  $ tc filter add dev ip6tnl0 ingress matchall       \
      action vlan push_eth dst_mac 00:00:5e:00:53:01 \
                           src_mac 00:00:5e:00:53:00 \
      action mirred egress redirect dev eth0

Without this patch, push_eth refuses to add an ethernet header because
the skb appears to already have a MAC header.

Signed-off-by: Guillaume Nault <gnault@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agogre: let mac_header point to outer header only when necessary
Guillaume Nault [Fri, 25 Jun 2021 13:33:13 +0000 (15:33 +0200)]
gre: let mac_header point to outer header only when necessary

Commit e271c7b4420d ("gre: do not keep the GRE header around in collect
medata mode") did reset the mac_header for the collect_md case. Let's
extend this behaviour to classical gre devices as well.

ipgre_header_parse() seems to be the only case that requires mac_header
to point to the outer header. We can detect this case accurately by
checking ->header_ops. For all other cases, we can reset mac_header.

This allows to push an Ethernet header to ipgre packets and redirect
them to an Ethernet device:

  $ tc filter add dev gre0 ingress matchall          \
      action vlan push_eth dst_mac 00:00:5e:00:53:01 \
                           src_mac 00:00:5e:00:53:00 \
      action mirred egress redirect dev eth0

Before this patch, this worked only for collect_md gre devices.
Now this works for regular gre devices as well. Only the special case
of gre devices that use ipgre_header_ops isn't supported.

Signed-off-by: Guillaume Nault <gnault@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agosit: allow redirecting ip6ip, ipip and mplsip packets to eth devices
Guillaume Nault [Fri, 25 Jun 2021 13:33:08 +0000 (15:33 +0200)]
sit: allow redirecting ip6ip, ipip and mplsip packets to eth devices

Even though sit transports L3 data (IPv6, IPv4 or MPLS) packets, it
needs to reset the mac_header pointer, so that other parts of the stack
don't mistakenly access the outer header after the packet has been
decapsulated. There are two rx handlers to modify: ipip6_rcv() for the
ip6ip mode and sit_tunnel_rcv() which is used to re-implement the ipip
and mplsip modes of ipip.ko.

This allows to push an Ethernet header to sit packets and redirect
them to an Ethernet device:

  $ tc filter add dev sit0 ingress matchall          \
      action vlan push_eth dst_mac 00:00:5e:00:53:01 \
                           src_mac 00:00:5e:00:53:00 \
      action mirred egress redirect dev eth0

Without this patch, push_eth refuses to add an ethernet header because
the skb appears to already have a MAC header.

Signed-off-by: Guillaume Nault <gnault@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agoipip: allow redirecting ipip and mplsip packets to eth devices
Guillaume Nault [Fri, 25 Jun 2021 13:33:04 +0000 (15:33 +0200)]
ipip: allow redirecting ipip and mplsip packets to eth devices

Even though ipip transports IPv4 or MPLS packets, it needs to reset the
mac_header pointer, so that other parts of the stack don't mistakenly
access the outer header after the packet has been decapsulated.

This allows to push an Ethernet header to ipip or mplsip packets and
redirect them to an Ethernet device:

  $ tc filter add dev ipip0 ingress matchall         \
      action vlan push_eth dst_mac 00:00:5e:00:53:01 \
                           src_mac 00:00:5e:00:53:00 \
      action mirred egress redirect dev eth0

Without this patch, push_eth refuses to add an ethernet header because
the skb appears to already have a MAC header.

Signed-off-by: Guillaume Nault <gnault@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agobareudp: allow redirecting bareudp packets to eth devices
Guillaume Nault [Fri, 25 Jun 2021 13:33:01 +0000 (15:33 +0200)]
bareudp: allow redirecting bareudp packets to eth devices

Even though bareudp transports L3 data (typically IP or MPLS), it needs
to reset the mac_header pointer, so that other parts of the stack don't
mistakenly access the outer header after the packet has been
decapsulated.

This allows to push an Ethernet header to bareudp packets and redirect
them to an Ethernet device:

  $ tc filter add dev bareudp0 ingress matchall      \
      action vlan push_eth dst_mac 00:00:5e:00:53:01 \
                           src_mac 00:00:5e:00:53:00 \
      action mirred egress redirect dev eth0

Without this patch, push_eth refuses to add an ethernet header because
the skb appears to already have a MAC header.

Signed-off-by: Guillaume Nault <gnault@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agobpf: Fix false positive kmemleak report in bpf_ringbuf_area_alloc()
Rustam Kovhaev [Sat, 26 Jun 2021 18:11:56 +0000 (11:11 -0700)]
bpf: Fix false positive kmemleak report in bpf_ringbuf_area_alloc()

kmemleak scans struct page, but it does not scan the page content. If we
allocate some memory with kmalloc(), then allocate page with alloc_page(),
and if we put kmalloc pointer somewhere inside that page, kmemleak will
report kmalloc pointer as a false positive.

We can instruct kmemleak to scan the memory area by calling kmemleak_alloc()
and kmemleak_free(), but part of struct bpf_ringbuf is mmaped to user space,
and if struct bpf_ringbuf changes we would have to revisit and review size
argument in kmemleak_alloc(), because we do not want kmemleak to scan the
user space memory. Let's simplify things and use kmemleak_not_leak() here.

For posterity, also adding additional prior analysis from Andrii:

  I think either kmemleak or syzbot are misreporting this. I've added a
  bunch of printks around all allocations performed by BPF ringbuf. [...]
  On repro side I get these two warnings:

  [vmuser@archvm bpf]$ sudo ./repro
  BUG: memory leak
  unreferenced object 0xffff88810d538c00 (size 64):
    comm "repro", pid 2140, jiffies 4294692933 (age 14.540s)
    hex dump (first 32 bytes):
      00 af 19 04 00 ea ff ff c0 ae 19 04 00 ea ff ff  ................
      80 ae 19 04 00 ea ff ff c0 29 2e 04 00 ea ff ff  .........)......
    backtrace:
      [<0000000077bfbfbd>] __bpf_map_area_alloc+0x31/0xc0
      [<00000000587fa522>] ringbuf_map_alloc.cold.4+0x48/0x218
      [<0000000044d49e96>] __do_sys_bpf+0x359/0x1d90
      [<00000000f601d565>] do_syscall_64+0x2d/0x40
      [<0000000043d3112a>] entry_SYSCALL_64_after_hwframe+0x44/0xae

  BUG: memory leak
  unreferenced object 0xffff88810d538c80 (size 64):
    comm "repro", pid 2143, jiffies 4294699025 (age 8.448s)
    hex dump (first 32 bytes):
      80 aa 19 04 00 ea ff ff 00 ab 19 04 00 ea ff ff  ................
      c0 ab 19 04 00 ea ff ff 80 44 28 04 00 ea ff ff  .........D(.....
    backtrace:
      [<0000000077bfbfbd>] __bpf_map_area_alloc+0x31/0xc0
      [<00000000587fa522>] ringbuf_map_alloc.cold.4+0x48/0x218
      [<0000000044d49e96>] __do_sys_bpf+0x359/0x1d90
      [<00000000f601d565>] do_syscall_64+0x2d/0x40
      [<0000000043d3112a>] entry_SYSCALL_64_after_hwframe+0x44/0xae

  Note that both reported leaks (ffff88810d538c80 and ffff88810d538c00)
  correspond to pages array bpf_ringbuf is allocating and tracking properly
  internally. Note also that syzbot repro doesn't close FD of created BPF
  ringbufs, and even when ./repro itself exits with error, there are still
  two forked processes hanging around in my system. So clearly ringbuf maps
  are alive at that point. So reporting any memory leak looks weird at that
  point, because that memory is being used by active referenced BPF ringbuf.

  It's also a question why repro doesn't clean up its forks. But if I do a
  `pkill repro`, I do see that all the allocated memory is /properly/ cleaned
  up [and the] "leaks" are deallocated properly.

  BTW, if I add close() right after bpf() syscall in syzbot repro, I see that
  everything is immediately deallocated, like designed. And no memory leak
  is reported. So I don't think the problem is anywhere in bpf_ringbuf code,
  rather in the leak detection and/or repro itself.

Reported-by: syzbot+5d895828587f49e7fe9b@syzkaller.appspotmail.com
Signed-off-by: Rustam Kovhaev <rkovhaev@gmail.com>
[ Daniel: also included analysis from Andrii to the commit log ]
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Tested-by: syzbot+5d895828587f49e7fe9b@syzkaller.appspotmail.com
Cc: Dmitry Vyukov <dvyukov@google.com>
Cc: Andrii Nakryiko <andrii@kernel.org>
Link: https://lore.kernel.org/bpf/CAEf4BzYk+dqs+jwu6VKXP-RttcTEGFe+ySTGWT9CRNkagDiJVA@mail.gmail.com
Link: https://lore.kernel.org/lkml/YNTAqiE7CWJhOK2M@nuc10
Link: https://lore.kernel.org/lkml/20210615101515.GC26027@arm.com
Link: https://syzkaller.appspot.com/bug?extid=5d895828587f49e7fe9b
Link: https://lore.kernel.org/bpf/20210626181156.1873604-1-rkovhaev@gmail.com
3 years agobpf: Allow bpf_get_current_ancestor_cgroup_id for tracing
Namhyung Kim [Sun, 27 Jun 2021 15:36:27 +0000 (08:36 -0700)]
bpf: Allow bpf_get_current_ancestor_cgroup_id for tracing

Allow the helper to be called from tracing programs. This is needed to
handle cgroup hiererachies in the program.

Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/bpf/20210627153627.824198-1-namhyung@kernel.org
3 years agobpf, x86: Fix extable offset calculation
Ravi Bangoria [Tue, 22 Jun 2021 11:00:26 +0000 (16:30 +0530)]
bpf, x86: Fix extable offset calculation

Commit 4c5de127598e1 ("bpf: Emit explicit NULL pointer checks for PROBE_LDX
instructions.") is emitting a couple of instructions before the actual load.
Consider those additional instructions while calculating extable offset.

Fixes: 4c5de127598e1 ("bpf: Emit explicit NULL pointer checks for PROBE_LDX instructions.")
Signed-off-by: Ravi Bangoria <ravi.bangoria@linux.ibm.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/bpf/20210622110026.1157847-1-ravi.bangoria@linux.ibm.com
3 years agonet/mlx5e: Add IPsec support to uplink representor
Raed Salem [Fri, 19 Jun 2020 14:42:07 +0000 (09:42 -0500)]
net/mlx5e: Add IPsec support to uplink representor

Add the xfrm xdo and ipsec_init/cleanup to uplink representor to
support IPsec in SRIOV switchdev mode.

Signed-off-by: Raed Salem <raeds@nvidia.com>
Signed-off-by: Huy Nguyen <huyn@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
3 years agonet/mlx5e: kTLS, Add stats for number of deleted kTLS TX offloaded connections
Tariq Toukan [Mon, 7 Sep 2020 12:11:06 +0000 (15:11 +0300)]
net/mlx5e: kTLS, Add stats for number of deleted kTLS TX offloaded connections

Expose ethtool SW counter for the number of kTLS device-offloaded
TX connections that are finished and deleted.

Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
3 years agonet/mlx5: SF, Improve performance in SF allocation
Eli Cohen [Mon, 26 Apr 2021 06:58:55 +0000 (09:58 +0300)]
net/mlx5: SF, Improve performance in SF allocation

Avoid second traversal on the SF table by recording the first free entry
and using it in case the looked up entry was not found in the table.

Signed-off-by: Eli Cohen <elic@nvidia.com>
Signed-off-by: Parav Pandit <parav@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
3 years agonet/mlx5: Increase hairpin buffer size
Ariel Levkovich [Wed, 2 Dec 2020 19:46:04 +0000 (21:46 +0200)]
net/mlx5: Increase hairpin buffer size

The max packet size a hairpin queue is able to handle
is determined by the total hairpin buffer size divided
by 4.

Currently the buffer size is set to 32KB which makes
the max packet size to be 8KB and doesn't support
jumbo frames of size 9KB.

This change increases the buffer size to 64KB to increase
the max frame size and support 9KB frames.

Signed-off-by: Ariel Levkovich <lariel@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
3 years agonet/mlx5: DR, Add support for flow sampler offload
Yevgeny Kliteynik [Mon, 19 Apr 2021 13:23:49 +0000 (16:23 +0300)]
net/mlx5: DR, Add support for flow sampler offload

Add SW steering support for sFlow / flow sampler action.

Signed-off-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
3 years agonet/mlx5: Compare sampler flow destination ID in fs_core
Yevgeny Kliteynik [Mon, 19 Apr 2021 13:32:16 +0000 (16:32 +0300)]
net/mlx5: Compare sampler flow destination ID in fs_core

When comparing sampler flow destinations,
in fs_core, consider sampler ID as well.

Signed-off-by: Yevgeny Kliteynik <kliteyn@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
3 years agoBluetooth: hci_uart: Remove redundant assignment to fw_ptr
Nigel Christian [Wed, 16 Jun 2021 19:45:02 +0000 (15:45 -0400)]
Bluetooth: hci_uart: Remove redundant assignment to fw_ptr

The variable fw_ptr is assigned a value that is not read and the same value
is assigned in the patch goto. The assignment is redundant and can be
removed.

Addresses-Coverity: ("Unused value")
Signed-off-by: Nigel Christian <nigel.l.christian@gmail.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
3 years agoBluetooth: Fix handling of HCI_LE_Advertising_Set_Terminated event
Luiz Augusto von Dentz [Wed, 23 Jun 2021 03:59:02 +0000 (20:59 -0700)]
Bluetooth: Fix handling of HCI_LE_Advertising_Set_Terminated event

Error status of this event means that it has ended due reasons other
than a connection:

 'If advertising has terminated as a result of the advertising duration
 elapsing, the Status parameter shall be set to the error code
 Advertising Timeout (0x3C).'

 'If advertising has terminated because the
 Max_Extended_Advertising_Events was reached, the Status parameter
 shall be set to the error code Limit Reached (0x43).'

Fixes: acf0aeae431a0 ("Bluetooth: Handle ADv set terminated event")
Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
3 years agoBluetooth: Increment management interface revision
Marcel Holtmann [Tue, 15 Jun 2021 19:23:35 +0000 (21:23 +0200)]
Bluetooth: Increment management interface revision

Increment the mgmt revision due to recent changes.

Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
3 years agoBluetooth: btmrvl: remove redundant continue statement
Colin Ian King [Wed, 16 Jun 2021 13:07:57 +0000 (14:07 +0100)]
Bluetooth: btmrvl: remove redundant continue statement

The continue statement in the for-loop has no effect,
remove it.

Addresses-Coverity: ("Continue has no effect")
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
3 years agoBluetooth: Fix Set Extended (Scan Response) Data
Luiz Augusto von Dentz [Wed, 9 Jun 2021 18:09:27 +0000 (11:09 -0700)]
Bluetooth: Fix Set Extended (Scan Response) Data

These command do have variable length and the length can go up to 251,
so this changes the struct to not use a fixed size and then when
creating the PDU only the actual length of the data send to the
controller.

Fixes: a0fb3726ba551 ("Bluetooth: Use Set ext adv/scan rsp data if controller supports")
Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
3 years agoBluetooth: use inclusive language when filtering devices
Archie Pusaka [Fri, 4 Jun 2021 08:26:27 +0000 (16:26 +0800)]
Bluetooth: use inclusive language when filtering devices

This patch replaces some non-inclusive terms based on the appropriate
language mapping table compiled by the Bluetooth SIG:
https://specificationrefs.bluetooth.com/language-mapping/Appropriate_Language_Mapping_Table.pdf

Specifically, these terms are replaced:
blacklist -> reject list
whitelist -> accept list

Signed-off-by: Archie Pusaka <apusaka@chromium.org>
Reviewed-by: Miao-chen Chou <mcchou@chromium.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
3 years agoBluetooth: use inclusive language when tracking connections
Archie Pusaka [Fri, 4 Jun 2021 08:26:26 +0000 (16:26 +0800)]
Bluetooth: use inclusive language when tracking connections

This patch replaces some non-inclusive terms based on the appropriate
language mapping table compiled by the Bluetooth SIG:
https://specificationrefs.bluetooth.com/language-mapping/Appropriate_Language_Mapping_Table.pdf

Specifically, these terms are replaced:
master -> central
slave  -> peripheral

Signed-off-by: Archie Pusaka <apusaka@chromium.org>
Reviewed-by: Miao-chen Chou <mcchou@chromium.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
3 years agoBluetooth: use inclusive language in HCI role comments
Archie Pusaka [Fri, 4 Jun 2021 08:26:25 +0000 (16:26 +0800)]
Bluetooth: use inclusive language in HCI role comments

This patch replaces some non-inclusive terms based on the appropriate
language mapping table compiled by the Bluetooth SIG:
https://specificationrefs.bluetooth.com/language-mapping/Appropriate_Language_Mapping_Table.pdf

Specifically, these terms are replaced:
master -> initiator (for smp) or central (everything else)
slave  -> responder (for smp) or peripheral (everything else)

The #define preprocessor terms are unchanged for now to not disturb
dependent APIs.

Signed-off-by: Archie Pusaka <apusaka@chromium.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
3 years agoBluetooth: smp: Use the correct print format
Kai Ye [Thu, 3 Jun 2021 07:41:05 +0000 (15:41 +0800)]
Bluetooth: smp: Use the correct print format

According to Documentation/core-api/printk-formats.rst,
Use the correct print format. Printing an unsigned int value should use %u
instead of %d. Otherwise printk() might end up displaying negative numbers.

Signed-off-by: Kai Ye <yekai13@huawei.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
3 years agoBluetooth: sco: Use the correct print format
Kai Ye [Thu, 3 Jun 2021 07:41:04 +0000 (15:41 +0800)]
Bluetooth: sco: Use the correct print format

According to Documentation/core-api/printk-formats.rst,
Use the correct print format. Printing an unsigned int value should use %u
instead of %d. Otherwise printk() might end up displaying negative numbers.

Signed-off-by: Kai Ye <yekai13@huawei.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
3 years agoBluetooth: msft: Use the correct print format
Kai Ye [Thu, 3 Jun 2021 07:41:03 +0000 (15:41 +0800)]
Bluetooth: msft: Use the correct print format

According to Documentation/core-api/printk-formats.rst,
Use the correct print format. Printing an unsigned int value should use %u
instead of %d. Otherwise printk() might end up displaying negative numbers.

Signed-off-by: Kai Ye <yekai13@huawei.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
3 years agoBluetooth: mgmt: Use the correct print format
Kai Ye [Thu, 3 Jun 2021 07:41:02 +0000 (15:41 +0800)]
Bluetooth: mgmt: Use the correct print format

According to Documentation/core-api/printk-formats.rst,
Use the correct print format. Printing an unsigned int value should use %u
instead of %d. Otherwise printk() might end up displaying negative numbers.

Signed-off-by: Kai Ye <yekai13@huawei.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
3 years agoBluetooth: amp: Use the correct print format
Kai Ye [Thu, 3 Jun 2021 07:41:00 +0000 (15:41 +0800)]
Bluetooth: amp: Use the correct print format

According to Documentation/core-api/printk-formats.rst,
Use the correct print format. Printing an unsigned int value should use %u
instead of %d. Otherwise printk() might end up displaying negative numbers.

Signed-off-by: Kai Ye <yekai13@huawei.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
3 years agoBluetooth: a2mp: Use the correct print format
Kai Ye [Thu, 3 Jun 2021 07:40:59 +0000 (15:40 +0800)]
Bluetooth: a2mp: Use the correct print format

According to Documentation/core-api/printk-formats.rst,
Use the correct print format. Printing an unsigned int value should use %u
instead of %d. Otherwise printk() might end up displaying negative numbers.

Signed-off-by: Kai Ye <yekai13@huawei.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
3 years agoBluetooth: 6lowpan: Use the correct print format
Kai Ye [Thu, 3 Jun 2021 07:40:58 +0000 (15:40 +0800)]
Bluetooth: 6lowpan: Use the correct print format

According to Documentation/core-api/printk-formats.rst,
Use the correct print format. Printing an unsigned int value should use %u
instead of %d. Otherwise printk() might end up displaying negative numbers.

Signed-off-by: Kai Ye <yekai13@huawei.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
3 years agoBluetooth: hidp: Use the correct print format
Kai Ye [Thu, 3 Jun 2021 07:40:56 +0000 (15:40 +0800)]
Bluetooth: hidp: Use the correct print format

According to Documentation/core-api/printk-formats.rst,
Use the correct print format. Printing an unsigned int value should use %u
instead of %d. Otherwise printk() might end up displaying negative numbers.

Signed-off-by: Kai Ye <yekai13@huawei.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
3 years agoBluetooth: cmtp: Use the correct print format
Kai Ye [Thu, 3 Jun 2021 07:40:55 +0000 (15:40 +0800)]
Bluetooth: cmtp: Use the correct print format

According to Documentation/core-api/printk-formats.rst,
Use the correct print format. Printing an unsigned int value should use %u
instead of %d. Otherwise printk() might end up displaying negative numbers.

Signed-off-by: Kai Ye <yekai13@huawei.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
3 years agoBluetooth: bnep: Use the correct print format
Kai Ye [Thu, 3 Jun 2021 07:40:54 +0000 (15:40 +0800)]
Bluetooth: bnep: Use the correct print format

According to Documentation/core-api/printk-formats.rst,
Use the correct print format. Printing an unsigned int value should use %u
instead of %d. Otherwise printk() might end up displaying negative numbers.

Signed-off-by: Kai Ye <yekai13@huawei.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
3 years agoBluetooth: btusb: fix bt fiwmare downloading failure issue for qca btsoc.
Tim Jiang [Tue, 1 Jun 2021 09:57:10 +0000 (17:57 +0800)]
Bluetooth: btusb: fix bt fiwmare downloading failure issue for qca btsoc.

This is btsoc timing issue, after host start to downloading bt firmware,
ep2 need time to switch from function acl to function dfu, so host add
20ms delay as workaround.

Signed-off-by: Tim Jiang <tjiang@codeaurora.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
3 years agoBluetooth: btusb: use default nvm if boardID is 0 for wcn6855.
Tim Jiang [Tue, 1 Jun 2021 09:57:25 +0000 (17:57 +0800)]
Bluetooth: btusb: use default nvm if boardID is 0 for wcn6855.

if boardID is 0, will use the default nvm file without surfix.

Signed-off-by: Tim Jiang <tjiang@codeaurora.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
3 years agoBluetooth: btmtkuart: using pm_runtime_resume_and_get instead of pm_runtime_get_sync
Zhang Qilong [Mon, 31 May 2021 14:24:49 +0000 (22:24 +0800)]
Bluetooth: btmtkuart: using pm_runtime_resume_and_get instead of pm_runtime_get_sync

Using pm_runtime_resume_and_get is more appropriate
for simplifing code

Signed-off-by: Zhang Qilong <zhangqilong3@huawei.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
3 years agoBluetooth: use inclusive language in comments
Archie Pusaka [Mon, 31 May 2021 08:37:28 +0000 (16:37 +0800)]
Bluetooth: use inclusive language in comments

This patch replaces some non-inclusive terms based on the appropriate
language mapping table compiled by the Bluetooth SIG:
https://specificationrefs.bluetooth.com/language-mapping/Appropriate_Language_Mapping_Table.pdf

Specifically, these terms are replaced:
slave       -> peripheral
blacklisted -> blocked

Signed-off-by: Archie Pusaka <apusaka@chromium.org>
Reviewed-by: Miao-chen Chou <mcchou@chromium.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
3 years agoBluetooth: use inclusive language in SMP
Archie Pusaka [Mon, 31 May 2021 08:37:25 +0000 (16:37 +0800)]
Bluetooth: use inclusive language in SMP

This patch replaces some non-inclusive terms based on the appropriate
language mapping table compiled by the Bluetooth SIG:
https://specificationrefs.bluetooth.com/language-mapping/Appropriate_Language_Mapping_Table.pdf

Specifically, these terms are replaced:
master -> initiator
slave  -> responder

Signed-off-by: Archie Pusaka <apusaka@chromium.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
3 years agoBluetooth: use inclusive language in HCI LE features
Archie Pusaka [Mon, 31 May 2021 08:37:23 +0000 (16:37 +0800)]
Bluetooth: use inclusive language in HCI LE features

This patch replaces some non-inclusive terms based on the appropriate
language mapping table compiled by the Bluetooth SIG:
https://specificationrefs.bluetooth.com/language-mapping/Appropriate_Language_Mapping_Table.pdf

Specifically, these terms are replaced:
master -> central
slave  -> peripheral

Signed-off-by: Archie Pusaka <apusaka@chromium.org>
Reviewed-by: Miao-chen Chou <mcchou@chromium.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
3 years agoBluetooth: use inclusive language to describe CPB
Archie Pusaka [Mon, 31 May 2021 08:37:22 +0000 (16:37 +0800)]
Bluetooth: use inclusive language to describe CPB

This patch replaces some non-inclusive terms based on the appropriate
language mapping table compiled by the Bluetooth SIG:
https://specificationrefs.bluetooth.com/language-mapping/Appropriate_Language_Mapping_Table.pdf

Specifically, these terms are replaced when describing the
connectionless peripheral broadcast feature:
master -> central
slave  -> peripheral

Signed-off-by: Archie Pusaka <apusaka@chromium.org>
Reviewed-by: Miao-chen Chou <mcchou@chromium.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
3 years agoBluetooth: use inclusive language in hci_core.h
Archie Pusaka [Mon, 31 May 2021 08:37:21 +0000 (16:37 +0800)]
Bluetooth: use inclusive language in hci_core.h

This patch replaces some non-inclusive terms based on the appropriate
language mapping table compiled by the Bluetooth SIG:
https://specificationrefs.bluetooth.com/language-mapping/Appropriate_Language_Mapping_Table.pdf

Specifically, these terms are replaced:
master -> central
slave  -> peripheral

These attributes are not used elsewhere in the code.

Signed-off-by: Archie Pusaka <apusaka@chromium.org>
Reviewed-by: Miao-chen Chou <mcchou@chromium.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
3 years agoBluetooth: mgmt: Fix slab-out-of-bounds in tlv_data_is_valid
Luiz Augusto von Dentz [Fri, 28 May 2021 18:45:02 +0000 (11:45 -0700)]
Bluetooth: mgmt: Fix slab-out-of-bounds in tlv_data_is_valid

This fixes parsing of LTV entries when the length is 0.

Found with:

tools/mgmt-tester -s "Add Advertising - Success (ScRsp only)"

Add Advertising - Success (ScRsp only) - run
  Sending Add Advertising (0x003e)
  Test condition added, total 1
[   11.004577] ==================================================================
[   11.005292] BUG: KASAN: slab-out-of-bounds in tlv_data_is_valid+0x87/0xe0
[   11.005984] Read of size 1 at addr ffff888002c695b0 by task mgmt-tester/87
[   11.006711]
[   11.007176]
[   11.007429] Allocated by task 87:
[   11.008151]
[   11.008438] The buggy address belongs to the object at ffff888002c69580
[   11.008438]  which belongs to the cache kmalloc-64 of size 64
[   11.010526] The buggy address is located 48 bytes inside of
[   11.010526]  64-byte region [ffff888002c69580ffff888002c695c0)
[   11.012423] The buggy address belongs to the page:
[   11.013291]
[   11.013544] Memory state around the buggy address:
[   11.014359]  ffff888002c69480: fa fb fb fb fb fb fb fb fc fc fc fc fc fc fc fc
[   11.015453]  ffff888002c69500: fb fb fb fb fb fb fb fb fc fc fc fc fc fc fc fc
[   11.016232] >ffff888002c69580: 00 00 00 00 00 00 fc fc fc fc fc fc fc fc fc fc
[   11.017010]                                      ^
[   11.017547]  ffff888002c69600: 00 00 00 00 00 00 fc fc fc fc fc fc fc fc fc fc
[   11.018296]  ffff888002c69680: fb fb fb fb fb fb fb fb fc fc fc fc fc fc fc fc
[   11.019116] ==================================================================

Fixes: 2bb36870e8cb2 ("Bluetooth: Unify advertising instance flags check")
Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
3 years agoBluetooth: btrtl: rename USB fw for RTL8761
Joakim Tjernlund [Fri, 28 May 2021 15:26:44 +0000 (17:26 +0200)]
Bluetooth: btrtl: rename USB fw for RTL8761

According Realteks own BT drivers firmware RTL8761B is for UART
and RTL8761BU is for USB.

Change existing 8761B to UART and add an 8761BU entry for USB

Signed-off-by: Joakim Tjernlund <Joakim.Tjernlund@infinera.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
3 years agoBluetooth: btusb: Add 0x0b05:0x190e Realtek 8761BU (ASUS BT500) device.
Joakim Tjernlund [Fri, 28 May 2021 15:26:45 +0000 (17:26 +0200)]
Bluetooth: btusb: Add 0x0b05:0x190e Realtek 8761BU (ASUS BT500) device.

T:  Bus=01 Lev=01 Prnt=01 Port=08 Cnt=04 Dev#= 18 Spd=12   MxCh= 0
D:  Ver= 1.10 Cls=e0(wlcon) Sub=01 Prot=01 MxPS=64 #Cfgs=  1
P:  Vendor=0b05 ProdID=190e Rev= 2.00
S:  Manufacturer=Realtek
S:  Product=ASUS USB-BT500
S:  SerialNumber=xxxxxxxx
C:* #Ifs= 2 Cfg#= 1 Atr=e0 MxPwr=500mA
I:* If#= 0 Alt= 0 #EPs= 3 Cls=e0(wlcon) Sub=01 Prot=01 Driver=btusb
E:  Ad=81(I) Atr=03(Int.) MxPS=  16 Ivl=1ms
E:  Ad=02(O) Atr=02(Bulk) MxPS=  64 Ivl=0ms
E:  Ad=82(I) Atr=02(Bulk) MxPS=  64 Ivl=0ms
I:* If#= 1 Alt= 0 #EPs= 2 Cls=e0(wlcon) Sub=01 Prot=01 Driver=btusb
E:  Ad=03(O) Atr=01(Isoc) MxPS=   0 Ivl=1ms
E:  Ad=83(I) Atr=01(Isoc) MxPS=   0 Ivl=1ms
I:  If#= 1 Alt= 1 #EPs= 2 Cls=e0(wlcon) Sub=01 Prot=01 Driver=btusb
E:  Ad=03(O) Atr=01(Isoc) MxPS=   9 Ivl=1ms
E:  Ad=83(I) Atr=01(Isoc) MxPS=   9 Ivl=1ms
I:  If#= 1 Alt= 2 #EPs= 2 Cls=e0(wlcon) Sub=01 Prot=01 Driver=btusb
E:  Ad=03(O) Atr=01(Isoc) MxPS=  17 Ivl=1ms
E:  Ad=83(I) Atr=01(Isoc) MxPS=  17 Ivl=1ms
I:  If#= 1 Alt= 3 #EPs= 2 Cls=e0(wlcon) Sub=01 Prot=01 Driver=btusb
E:  Ad=03(O) Atr=01(Isoc) MxPS=  25 Ivl=1ms
E:  Ad=83(I) Atr=01(Isoc) MxPS=  25 Ivl=1ms
I:  If#= 1 Alt= 4 #EPs= 2 Cls=e0(wlcon) Sub=01 Prot=01 Driver=btusb
E:  Ad=03(O) Atr=01(Isoc) MxPS=  33 Ivl=1ms
E:  Ad=83(I) Atr=01(Isoc) MxPS=  33 Ivl=1ms
I:  If#= 1 Alt= 5 #EPs= 2 Cls=e0(wlcon) Sub=01 Prot=01 Driver=btusb
E:  Ad=03(O) Atr=01(Isoc) MxPS=  49 Ivl=1ms
E:  Ad=83(I) Atr=01(Isoc) MxPS=  49 Ivl=1ms
Signed-off-by: Joakim Tjernlund <Joakim.Tjernlund@infinera.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
3 years agoBluetooth: hci_h5: Add RTL8822CS capabilities
Archie Pusaka [Thu, 27 May 2021 07:19:54 +0000 (15:19 +0800)]
Bluetooth: hci_h5: Add RTL8822CS capabilities

RTL8822 chipset supports WBS, and this information is conveyed in
btusb.c. However, the UART driver doesn't have this information just
yet.

Signed-off-by: Archie Pusaka <apusaka@chromium.org>
Reviewed-by: Abhishek Pandit-Subedi <abhishekpandit@chromium.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
3 years agoBluetooth: mgmt: Fix the command returns garbage parameter value
Tedd Ho-Jeong An [Wed, 26 May 2021 17:36:22 +0000 (10:36 -0700)]
Bluetooth: mgmt: Fix the command returns garbage parameter value

When the Get Device Flags command fails, it returns the error status
with the parameters filled with the garbage values. Although the
parameters are not used, it is better to fill with zero than the random
values.

Signed-off-by: Tedd Ho-Jeong An <tedd.an@intel.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
3 years agoBluetooth: disable filter dup when scan for adv monitor
Yun-Hao Chung [Thu, 20 May 2021 05:12:09 +0000 (13:12 +0800)]
Bluetooth: disable filter dup when scan for adv monitor

Disable duplicates filter when scanning for advertisement monitor for
the following reasons. The scanning includes active scan and passive
scan.

For HW pattern filtering (ex. MSFT), Realtek and Qualcomm controllers
ignore RSSI_Sampling_Period when the duplicates filter is enabled.

For SW pattern filtering, when we're not doing interleaved scanning, it
is necessary to disable duplicates filter, otherwise hosts can only
receive one advertisement and it's impossible to know if a peer is still
in range.

Signed-off-by: Yun-Hao Chung <howardchung@chromium.org>
Reviewed-by: Archie Pusaka <apusaka@chromium.org>
Reviewed-by: Manish Mandlik <mmandlik@chromium.org>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
3 years agoBluetooth: Translate additional address type during le_conn_comp
Sathish Narasimman [Thu, 20 May 2021 11:42:01 +0000 (17:12 +0530)]
Bluetooth: Translate additional address type during le_conn_comp

When using controller based address resolution, then the destination
address type during le_conn_complete uses 0x02 & 0x03 if controller
resolves the destination address(RPA).
These address types need to be converted back into either 0x00 0r 0x01

Signed-off-by: Sathish Narasimman <sathish.narasimman@intel.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
3 years agoBluetooth: RFCOMM: Use DEVICE_ATTR_RO macro
YueHaibing [Thu, 20 May 2021 13:32:35 +0000 (21:32 +0800)]
Bluetooth: RFCOMM: Use DEVICE_ATTR_RO macro

Use DEVICE_ATTR_RO helper instead of plain DEVICE_ATTR,
which makes the code a bit shorter and easier to read.

Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
3 years agoBluetooth: btusb: Add support USB ALT 3 for WBS
Hilda Wu [Fri, 14 May 2021 03:19:01 +0000 (11:19 +0800)]
Bluetooth: btusb: Add support USB ALT 3 for WBS

Because mSBC frames do not need to be aligned to the SCO packet
boundary. Using USB ALT 3 let HCI payload >= 60 bytes, let mSBC
data satisfy 60 Bytes avoid payload unaligned situation and fixed
some headset no voise issue.

USB Alt 3 supported also need HFP support transparent MTU in 72 Bytes.

Signed-off-by: Hilda Wu <hildawu@realtek.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
3 years agoBluetooth: L2CAP: Fix invalid access on ECRED Connection response
Luiz Augusto von Dentz [Wed, 19 May 2021 20:41:51 +0000 (13:41 -0700)]
Bluetooth: L2CAP: Fix invalid access on ECRED Connection response

The use of l2cap_chan_del is not safe under a loop using
list_for_each_entry.

Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>
3 years agoBluetooth: L2CAP: Fix invalid access if ECRED Reconfigure fails
Luiz Augusto von Dentz [Wed, 19 May 2021 20:41:50 +0000 (13:41 -0700)]
Bluetooth: L2CAP: Fix invalid access if ECRED Reconfigure fails

The use of l2cap_chan_del is not safe under a loop using
list_for_each_entry.

Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Signed-off-by: Marcel Holtmann <marcel@holtmann.org>