ravb: Initialize GbEthernet E-MAC

Initialize GbEthernet E-MAC found on RZ/G2L SoC.
This patch also renames ravb_set_rate to ravb_set_rate_rcar and
ravb_rcar_emac_init to ravb_emac_init_rcar to be consistent with
the naming convention used in sh_eth driver.

Signed-off-by: Biju Das <[email protected]>
Reviewed-by: Lad Prabhakar <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

ravb: Add half_duplex to struct ravb_hw_info

RZ/G2L supports half duplex mode.
Add a half_duplex hw feature bit to struct ravb_hw_info for
supporting half duplex mode for RZ/G2L.

Signed-off-by: Biju Das <[email protected]>
Reviewed-by: Sergey Shtylyov <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

ravb: Add magic_pkt to struct ravb_hw_info

E-MAC on R-Car supports magic packet detection, whereas RZ/G2L
does not support this feature. Add magic_pkt to struct ravb_hw_info
and enable this feature only for R-Car.

Signed-off-by: Biju Das <[email protected]>
Reviewed-by: Lad Prabhakar <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

ravb: Add tsrq to struct ravb_hw_info

R-Car AVB-DMAC has 4 Transmit start request queues, whereas
RZ/G2L has only 1 Transmit start request queue.

Add a tsrq variable to struct ravb_hw_info to handle this
difference.

Signed-off-by: Biju Das <[email protected]>
Reviewed-by: Lad Prabhakar <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

ravb: Exclude gPTP feature support for RZ/G2L

R-Car supports gPTP feature whereas RZ/G2L does not support it.
This patch excludes gtp feature support for RZ/G2L.

Signed-off-by: Biju Das <[email protected]>
Reviewed-by: Lad Prabhakar <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

ravb: Initialize GbEthernet DMAC

Initialize GbEthernet DMAC found on RZ/G2L SoC.

Signed-off-by: Biju Das <[email protected]>
Reviewed-by: Lad Prabhakar <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

ravb: Add support for RZ/G2L SoC

RZ/G2L SoC has Gigabit Ethernet IP consisting of Ethernet controller
(E-MAC), Internal TCP/IP Offload Engine (TOE) and Dedicated Direct
memory access controller (DMAC).

This patch adds compatible string for RZ/G2L and fills up the
ravb_hw_info struct. Function stubs are added which will be used by
gbeth_hw_info and will be filled incrementally.

Signed-off-by: Biju Das <[email protected]>
Reviewed-by: Lad Prabhakar <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

ravb: Add nc_queue to struct ravb_hw_info

R-Car supports network control queue whereas RZ/G2L does not support
it. Add nc_queue to struct ravb_hw_info, so that NC queue is handled
only by R-Car.

This patch also renames ravb_rcar_dmac_init to ravb_dmac_init_rcar
to be consistent with the naming convention used in sh_eth driver.

Signed-off-by: Biju Das <[email protected]>
Reviewed-by: Lad Prabhakar <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

ravb: Rename "no_ptp_cfg_active" and "ptp_cfg_active" variables

Rename the variable "no_ptp_cfg_active" with "gptp" and
"ptp_cfg_active" with "ccc_gac" to match the HW features.

There is no functional change.

Signed-off-by: Biju Das <[email protected]>
Suggested-by: Sergey Shtylyov <[email protected]>
Reviewed-by: Lad Prabhakar <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

ravb: Rename "ravb_set_features_rx_csum" function to "ravb_set_features_rcar"

Rename "ravb_set_features_rx_csum" function to "ravb_set_features_rcar" and
replace the function pointer "set_rx_csum_feature" with "set_feature".

Signed-off-by: Biju Das <[email protected]>
Suggested-by: Sergey Shtylyov <[email protected]>
Reviewed-by: Sergey Shtylyov <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

Merge https://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf-next

Daniel Borkmann says:

====================
bpf-next 2021-10-02

We've added 85 non-merge commits during the last 15 day(s) which contain
a total of 132 files changed, 13779 insertions(+), 6724 deletions(-).

The main changes are:

1) Massive update on test_bpf.ko coverage for JITs as preparatory work for
   an upcoming MIPS eBPF JIT, from Johan Almbladh.

2) Add a batched interface for RX buffer allocation in AF_XDP buffer pool,
   with driver support for i40e and ice from Magnus Karlsson.

3) Add legacy uprobe support to libbpf to complement recently merged legacy
   kprobe support, from Andrii Nakryiko.

4) Add bpf_trace_vprintk() as variadic printk helper, from Dave Marchevsky.

5) Support saving the register state in verifier when spilling <8byte bounded
   scalar to the stack, from Martin Lau.

6) Add libbpf opt-in for stricter BPF program section name handling as part
   of libbpf 1.0 effort, from Andrii Nakryiko.

7) Add a document to help clarifying BPF licensing, from Alexei Starovoitov.

8) Fix skel_internal.h to propagate errno if the loader indicates an internal
   error, from Kumar Kartikeya Dwivedi.

9) Fix build warnings with -Wcast-function-type so that the option can later
   be enabled by default for the kernel, from Kees Cook.

10) Fix libbpf to ignore STT_SECTION symbols in legacy map definitions as it
    otherwise errors out when encountering them, from Toke Høiland-Jørgensen.

11) Teach libbpf to recognize specialized maps (such as for perf RB) and
    internally remove BTF type IDs when creating them, from Hengqi Chen.

12) Various fixes and improvements to BPF selftests.
====================

Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Jakub Kicinski <[email protected]>

Bluetooth: Rename driver .prevent_wake to .wakeup

prevent_wake logic is backward since what it is really checking is
if the device may wakeup the system or not, not that it will prevent
the to be awaken.

Also looking on how other subsystems have the entry as power/wakeup
this also renames the force_prevent_wake to force_wakeup in vhci driver.

Signed-off-by: Luiz Augusto von Dentz <[email protected]>
Signed-off-by: Marcel Holtmann <[email protected]>

net/core: disable NET_RX_BUSY_POLL on PREEMPT_RT

napi_busy_loop() disables preemption and performs a NAPI poll. We can't acquire
sleeping locks with disabled preemption which would be required while
__napi_poll() invokes the callback of the driver.

A threaded interrupt performing the NAPI-poll can be preempted on PREEMPT_RT.
A RT thread on another CPU may observe NAPIF_STATE_SCHED bit set and busy-spin
until it is cleared or its spin time runs out. Given it is the task with the
highest priority it will never observe the NEED_RESCHED bit set.
In this case the time is better spent by simply sleeping.

The NET_RX_BUSY_POLL is disabled by default (the system wide sysctls for
poll/read are set to zero). Disabling NET_RX_BUSY_POLL on PREEMPT_RT to avoid
wrong locking context in case it is used.

Signed-off-by: Sebastian Andrzej Siewior <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Jakub Kicinski <[email protected]>

Merge branch 'libbpf: Support uniform BTF-defined key/value specification across all BPF maps'

Hengqi Chen says:

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

Currently a bunch of (usually pretty specialized) BPF maps do not support
specifying BTF types for they key and value. For such maps, specifying
their definition like this:

  struct {
      __uint(type, BPF_MAP_TYPE_PERF_EVENT_ARRAY);
      __type(key, int);
      __type(value, int);
  } my_perf_buf SEC(".maps");

Would actually produce warnings about retrying BPF map creation without BTF.
Users are forced to know such nuances and use __uint(key_size, 4) instead.
This is non-uniform, annoying, and inconvenient.

This patch set teaches libbpf to recognize those specialized maps and removes
BTF type IDs when creating BPF map. Also, update existing BPF selftests to
exericse this change.
====================

Signed-off-by: Andrii Nakryiko <[email protected]>

selftests/bpf: Use BTF-defined key/value for map definitions

Change map definitions in BPF selftests to use BTF-defined
key/value types. This unifies the map definitions and ensures
libbpf won't emit warning about retrying map creation.

Signed-off-by: Hengqi Chen <[email protected]>
Signed-off-by: Andrii Nakryiko <[email protected]>
Link: https://lore.kernel.org/bpf/[email protected]

libbpf: Support uniform BTF-defined key/value specification across all BPF maps

A bunch of BPF maps do not support specifying BTF types for key and value.
This is non-uniform and inconvenient[0]. Currently, libbpf uses a retry
logic which removes BTF type IDs when BPF map creation failed. Instead
of retrying, this commit recognizes those specialized maps and removes
BTF type IDs when creating BPF map.

  [0] Closes: https://github.com/libbpf/libbpf/issues/355

Signed-off-by: Hengqi Chen <[email protected]>
Signed-off-by: Andrii Nakryiko <[email protected]>
Link: https://lore.kernel.org/bpf/[email protected]

Merge tag 'mlx5-updates-2021-09-30' of git://git.kernel.org/pub/scm/linux/kernel/git/saeed/linux

Saeed Mahameed says:

====================
mlx5-updates-2021-09-30

1) From Yevgeny Kliteynik:

This patch series deals with vport handling in SW steering.

For every vport, SW steering queries FW for this vport's properties,
such as RX/TX ICM addresses to be able to add this vport as dest action.
The following patches rework vport capabilities managements and add support
for Scalable Functions (SFs).

 - Patch 1 fixes the vport number data type all over the DR code to 16 bits
   in accordance with HW spec.
 - Patch 2 replaces local SW steering WIRE_PORT macro with the existing
   mlx5 define.
 - Patch 3 adds missing query for vport 0 and and handles eswitch manager
   capabilities for ECPF (BlueField in embedded CPU mode).
 - Patch 4 fixes error messages for failure to obtain vport caps from
   different locations in the code to have the same verbosity level and
   similar wording.
 - Patch 5 adds support for csum recalculation flow tables on SFs: it
   implements these FTs management in XArray instead of the fixed size array,
   thus adding support for csum recalculation table for any valid vport.
 - Patch 6 is the main patch of this whole series: it refactors vports
   capabilities handling and adds SFs support.

2) Minor and trivial updates and cleanups

* tag 'mlx5-updates-2021-09-30' of git://git.kernel.org/pub/scm/linux/kernel/git/saeed/linux:
  net/mlx5e: Use array_size() helper
  net/mlx5: Use struct_size() helper in kvzalloc()
  net/mlx5: Use kvcalloc() instead of kvzalloc()
  net/mlx5: Tolerate failures in debug features while driver load
  net/mlx5: Warn for devlink reload when there are VFs alive
  net/mlx5: DR, Add missing string for action type SAMPLER
  net/mlx5: DR, init_next_match only if needed
  net/mlx5: DR, Fix typo 'offeset' to 'offset'
  net/mlx5: DR, Increase supported num of actions to 32
  net/mlx5: DR, Add support for SF vports
  net/mlx5: DR, Support csum recalculation flow table on SFs
  net/mlx5: DR, Align error messages for failure to obtain vport caps
  net/mlx5: DR, Add missing query for vport 0
  net/mlx5: DR, Replace local WIRE_PORT macro with the existing MLX5_VPORT_UPLINK
  net/mlx5: DR, Fix vport number data type to u16
====================

Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Jakub Kicinski <[email protected]>

bpf/tests: Add test of LDX_MEM with operand aliasing

This patch adds a set of tests of BPF_LDX_MEM where both operand registers
are the same register. Mainly testing 32-bit JITs that may load a 64-bit
value in two 32-bit loads, and must not overwrite the address register.

Signed-off-by: Johan Almbladh <[email protected]>
Signed-off-by: Daniel Borkmann <[email protected]>
Link: https://lore.kernel.org/bpf/[email protected]

bpf/tests: Add test of ALU shifts with operand register aliasing

This patch adds a tests of ALU32 and ALU64 LSH/RSH/ARSH operations for the
case when the two operands are the same register. Mainly intended to test
JITs that implement ALU64 shifts with 32-bit CPU instructions.

Also renamed related helper functions for consistency with the new tests.

Signed-off-by: Johan Almbladh <[email protected]>
Signed-off-by: Daniel Borkmann <[email protected]>
Link: https://lore.kernel.org/bpf/[email protected]

bpf/tests: Add exhaustive tests of BPF_ATOMIC register combinations

This patch adds tests of all register combinations for BPF_ATOMIC
operations on both BPF_W and BPF_DW sizes.

Signed-off-by: Johan Almbladh <[email protected]>
Signed-off-by: Daniel Borkmann <[email protected]>
Link: https://lore.kernel.org/bpf/[email protected]

bpf/tests: Add exhaustive tests of ALU register combinations

This patch replaces the current register combination test with new
exhaustive tests. Before, only a subset of register combinations was
tested for ALU64 DIV. Now, all combinatons of operand registers are
tested, including the case when they are the same, and for all ALU32
and ALU64 operations.

Signed-off-by: Johan Almbladh <[email protected]>
Signed-off-by: Daniel Borkmann <[email protected]>
Link: https://lore.kernel.org/bpf/[email protected]

bpf/tests: Minor restructuring of ALU tests

This patch moves the ALU LSH/RSH/ARSH reference computations into the
common reference value function. Also fix typo in constants so they
now have the intended values.

Signed-off-by: Johan Almbladh <[email protected]>
Signed-off-by: Daniel Borkmann <[email protected]>
Link: https://lore.kernel.org/bpf/[email protected]

bpf/tests: Add more tests for ALU and ATOMIC register clobbering

This patch expands the register-clobbering-during-function-call tests
to cover more all ALU32/64 MUL, DIV and MOD operations and all ATOMIC
operations. In short, if a JIT implements a complex operation with
a call to an external function, it must make sure to save and restore
all its caller-saved registers that may be clobbered by the call.

Signed-off-by: Johan Almbladh <[email protected]>
Signed-off-by: Daniel Borkmann <[email protected]>
Link: https://lore.kernel.org/bpf/[email protected]

bpf/tests: Add tests to check source register zero-extension

This patch adds tests to check that the source register is preserved when
zero-extending a 32-bit value. In particular, it checks that the source
operand is not zero-extended in-place.

Signed-off-by: Johan Almbladh <[email protected]>
Signed-off-by: Daniel Borkmann <[email protected]>
Link: https://lore.kernel.org/bpf/[email protected]

bpf/tests: Add exhaustive tests of BPF_ATOMIC magnitudes

This patch adds a series of test to verify the operation of BPF_ATOMIC
with BPF_DW and BPF_W sizes, for all power-of-two magnitudes of the
register value operand.

Also fixes a confusing typo in the comment for a related test.

Signed-off-by: Johan Almbladh <[email protected]>
Signed-off-by: Daniel Borkmann <[email protected]>
Link: https://lore.kernel.org/bpf/[email protected]

bpf/tests: Add zero-extension checks in BPF_ATOMIC tests

This patch updates the existing tests of BPF_ATOMIC operations to verify
that a 32-bit register operand is properly zero-extended. In particular,
it checks the operation on archs that require 32-bit operands to be
properly zero-/sign-extended or the result is undefined, e.g. MIPS64.

Signed-off-by: Johan Almbladh <[email protected]>
Signed-off-by: Daniel Borkmann <[email protected]>
Link: https://lore.kernel.org/bpf/[email protected]

bpf/tests: Add tests of BPF_LDX and BPF_STX with small sizes

This patch adds a series of tests to verify the behavior of BPF_LDX and
BPF_STX with BPF_B//W sizes in isolation. In particular, it checks that
BPF_LDX zero-extendeds the result, and that BPF_STX does not overwrite
adjacent bytes in memory.

BPF_ST and operations on BPF_DW size are deemed to be sufficiently
tested by existing tests.

Signed-off-by: Johan Almbladh <[email protected]>
Signed-off-by: Daniel Borkmann <[email protected]>
Link: https://lore.kernel.org/bpf/[email protected]

fix up for "net: add new socket option SO_RESERVE_MEM"

Some architectures do not include uapi/asm/socket.h

Fixes: 2bb2f5fb21b0 ("net: add new socket option SO_RESERVE_MEM")
Signed-off-by: Stephen Rothwell <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

Revert "Merge branch 'mctp-kunit-tests'"

This reverts commit 4f42ad2011d2fcbd89f5cdf56121271a8cd5ee5d, reversing
changes made to ea2dd331bfaaeba74ba31facf437c29044f7d4cb.

These chanfges break the build when mctp is modular.

Signed-off-by: David S. Miller <[email protected]>

sparc: add SO_RESERVE_MEM definition.

Reported-by: Stephen Rothwell <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

devlink: report maximum number of snapshots with regions

Each region has an independently configurable number of maximum
snapshots. This information is not reported to userspace, making it not
very discoverable. Fix this by adding a new
DEVLINK_ATTR_REGION_MAX_SNAPSHOST attribute which is used to report this
maximum.

Ex:

  $devlink region
  pci/0000:af:00.0/nvm-flash: size 10485760 snapshot [] max 1
  pci/0000:af:00.0/device-caps: size 4096 snapshot [] max 10
  pci/0000:af:00.1/nvm-flash: size 10485760 snapshot [] max 1
  pci/0000:af:00.1/device-caps: size 4096 snapshot [] max 10

This information enables users to understand why a new region command
may fail due to having too many existing snapshots.

Reported-by: Gurucharan G <[email protected]> (A Contingent worker at Intel)
Signed-off-by: Jacob Keller <[email protected]>
Acked-by: Jakub Kicinski <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

Merge branch 'mctp-kunit-tests'

Jeremy Kerr says:

====================
MCTP kunit tests

This change adds some initial kunit tests for the MCTP core. We'll
expand the coverage in a future series, and augment with a few
selftests, but this establishes a baseline set of tests for now.

Thanks to the kunit folks for the framework!
====================

Signed-off-by: David S. Miller <[email protected]>

mctp: Add input reassembly tests

Add multi-packet route input tests, for message reassembly. These will
feed packets to be received by a bound socket, or dropped.

Signed-off-by: Jeremy Kerr <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

mctp: Add route input to socket tests

Add a few tests for single-packet route inputs, testing the
mctp_route_input function.

Signed-off-by: Jeremy Kerr <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

mctp: Add packet rx tests

Add a few tests for the initial packet ingress through
mctp_pkttype_receive function; mainly packet header sanity checks. Full
input routing checks will be added as a separate change.

Signed-off-by: Jeremy Kerr <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

mctp: Add test utils

Add a new object for shared test utilities

Signed-off-by: Jeremy Kerr <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

mctp: Add initial test structure and fragmentation test

This change adds the first kunit test for the mctp subsystem, and an
initial test for the fragmentation path.

We're adding tests under a new net/mctp/test/ directory.

Signed-off-by: Jeremy Kerr <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

Merge tag 'mlx5-fixes-2021-09-30' of git://git.kernel.org/pub/scm/linux/kernel/git/saeed/linux

Saeed Mahameed says:

====================
mlx5 fixes 2021-09-30

This series introduces some fixes to mlx5 driver.
Please pull and let me know if there is any problem.
====================

Signed-off-by: David S. Miller <[email protected]>

Bluetooth: btrtl: Ask ic_info to drop firmware

Some un-support wakeup platforms keep USB power and suspend signal
is coming late, this makes Realtek some chip keep its firmware,
and make it never load new firmware.

So use vendor specific HCI command to ask them drop its firmware after
system shutdown or resume.

Signed-off-by: Hilda Wu <[email protected]>
Signed-off-by: Marcel Holtmann <[email protected]>

Bluetooth: btusb: Add support for TP-Link UB500 Adapter

Add support for TP-Link UB500 Adapter (RTL8761B)

* /sys/kernel/debug/usb/devices
T:  Bus=01 Lev=02 Prnt=05 Port=01 Cnt=01 Dev#= 78 Spd=12   MxCh= 0
D:  Ver= 1.10 Cls=e0(wlcon) Sub=01 Prot=01 MxPS=64 #Cfgs=  1
P:  Vendor=2357 ProdID=0604 Rev= 2.00
S:  Manufacturer=
S:  Product=TP-Link UB500 Adapter
S:  SerialNumber=E848B8C82000
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: Nicholas Flintham <[email protected]>
Signed-off-by: Marcel Holtmann <[email protected]>

Bluetooth: hci_sock: Replace use of memcpy_from_msg with bt_skb_sendmsg

This makes use of bt_skb_sendmsg instead of allocating a different
buffer to be used with memcpy_from_msg which cause one extra copy.

Tested-by: Tedd Ho-Jeong An <[email protected]>
Signed-off-by: Luiz Augusto von Dentz <[email protected]>
Signed-off-by: Marcel Holtmann <[email protected]>

net: sched: Use struct_size() helper in kvmalloc()

Make use of the struct_size() helper instead of an open-coded version,
in order to avoid any potential type mistakes or integer overflows
that, in the worst scenario, could lead to heap overflows.

Link: https://github.com/KSPP/linux/issues/160
Signed-off-by: Gustavo A. R. Silva <[email protected]>
Link: https://lore.kernel.org/r/20210929201718.GA342296@embeddedor
Signed-off-by: Jakub Kicinski <[email protected]>

net/mlx5e: Use array_size() helper

Use array_size() helper to aid in 2-factor allocation instances.

Link: https://github.com/KSPP/linux/issues/160
Signed-off-by: Gustavo A. R. Silva <[email protected]>
Signed-off-by: Saeed Mahameed <[email protected]>

net/mlx5: Use struct_size() helper in kvzalloc()

Make use of the struct_size() helper instead of an open-coded version,
in order to avoid any potential type mistakes or integer overflows that,
in the worse scenario, could lead to heap overflows.

Link: https://github.com/KSPP/linux/issues/160
Signed-off-by: Gustavo A. R. Silva <[email protected]>
Signed-off-by: Saeed Mahameed <[email protected]>

net/mlx5: Use kvcalloc() instead of kvzalloc()

Use 2-factor argument form kvcalloc() instead of kvzalloc().

Link: https://github.com/KSPP/linux/issues/162
Signed-off-by: Gustavo A. R. Silva <[email protected]>
Signed-off-by: Saeed Mahameed <[email protected]>

net/mlx5: Tolerate failures in debug features while driver load

FW tracer and resource dump are debug features. Although failing to
initialize them may indicate an error, don't let this stop device
loading.

Signed-off-by: Aya Levin <[email protected]>
Reviewed-by: Tariq Toukan <[email protected]>
Signed-off-by: Saeed Mahameed <[email protected]>

net/mlx5: Warn for devlink reload when there are VFs alive

When performing PF reload, VF can't communicate with FW until
it recovers and reloads as well.

Add a warning message when performing devlink reload while
VFs are still present. Thus, giving a notice of an unfavorable
behavior that might occur as a result of a consequential reloads
and cause interruption of VF recovery.

Signed-off-by: Lama Kayal <[email protected]>
Reviewed-by: Moshe Shemesh <[email protected]>
Signed-off-by: Saeed Mahameed <[email protected]>

net/mlx5: DR, Add missing string for action type SAMPLER

Add missing string value for DR_ACTION_TYP_SAMPLER action type

Signed-off-by: Yevgeny Kliteynik <[email protected]>
Signed-off-by: Saeed Mahameed <[email protected]>

net/mlx5: DR, init_next_match only if needed

Allocate next steering table entry only if the remaining space requires to.

Signed-off-by: Yevgeny Kliteynik <[email protected]>
Signed-off-by: Saeed Mahameed <[email protected]>

net/mlx5: DR, Fix typo 'offeset' to 'offset'

Signed-off-by: Yevgeny Kliteynik <[email protected]>
Signed-off-by: Saeed Mahameed <[email protected]>

net/mlx5: DR, Increase supported num of actions to 32

Increase max supported number of actions in the same rule.

Signed-off-by: Hamdan Igbaria <[email protected]>
Signed-off-by: Yevgeny Kliteynik <[email protected]>
Signed-off-by: Saeed Mahameed <[email protected]>

net/mlx5: DR, Add support for SF vports

Move all the vport capabilities to a separate struct and store vport caps
in XArray: SFs vport numbers will not come in the same range as VF vports,
so the existing implementation of vport capabilities as a fixed size array
is not suitable here.

XArray is a perfect fit: it is efficient when the indices used are densely
clustered. In addition to being a perfect fit as a dynamic data structure,
XArray also provides locking - it uses RCU and an internal spinlock to
synchronise access, so no additional protection needed.

Now except for the eswitch manager vport, all other vports (including the
uplink vport) are handled in the same way: when a new go-to-vport action
is added, this vport's caps are loaded from the xarray. If it is the first
time for this particular vport number, then its capabilities are queried
from FW and filled in into the appropriate entry.

Signed-off-by: Yevgeny Kliteynik <[email protected]>
Reviewed-by: Muhammad Sammar <[email protected]>
Signed-off-by: Saeed Mahameed <[email protected]>

net/mlx5: DR, Support csum recalculation flow table on SFs

Implement csum recalculation flow tables in XAarray instead of a fixed
array, thus adding support for csum recalc table on any valid vport
number, which enables this support for SFs.

Signed-off-by: Yevgeny Kliteynik <[email protected]>
Reviewed-by: Muhammad Sammar <[email protected]>
Signed-off-by: Saeed Mahameed <[email protected]>

net/mlx5: DR, Align error messages for failure to obtain vport caps

Print similar error messages when an invalid vport number is
provided during action creation and during STEv0/1 creation.

Signed-off-by: Yevgeny Kliteynik <[email protected]>
Reviewed-by: Muhammad Sammar <[email protected]>
Signed-off-by: Saeed Mahameed <[email protected]>

net/mlx5: DR, Add missing query for vport 0

Currently, vport 0 capabilities are not set.
To fix this, we now querying both eswitch manager and vport 0.
Eswitch manager has an access to all the vports - for eswitch manager PF, all
vports can be referred as other vports. The exception is embedded CPU mode,
where there is vport 0 of ECPF and the PF vport 0.

Here is how vport are queried:

For Connect-X5/6:
    PF vport (0) and vports 1..n: vport number, other = true
    esw_manager is vport 0 (PF)
For BlueField (in embedded CPU mode):
    ECPF vport: vport = 0, other = false
    PF vport (0) and 1..n: vport number, other = true
    esw_manager = vport 0 (ECPF)

Also, note that there's no need for other_vport function parameter
in dr_domain_query_vport - this value is now deduced locally in the
function.

Signed-off-by: Yuval Avnery <[email protected]>
Signed-off-by: Yevgeny Kliteynik <[email protected]>
Reviewed-by: Muhammad Sammar <[email protected]>
Signed-off-by: Saeed Mahameed <[email protected]>

net/mlx5: DR, Replace local WIRE_PORT macro with the existing MLX5_VPORT_UPLINK

SW steering defines its own macro for uplink vport number.
Replace this macro with an already existing mlx5 macro.

Signed-off-by: Yevgeny Kliteynik <[email protected]>
Signed-off-by: Saeed Mahameed <[email protected]>

net/mlx5: DR, Fix vport number data type to u16

According to the HW spec, vport number is a 16-bit value.
Fix vport usage all over the code to u16 data type.

Signed-off-by: Yevgeny Kliteynik <[email protected]>
Reviewed-by: Muhammad Sammar <[email protected]>
Signed-off-by: Saeed Mahameed <[email protected]>

Merge git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net

drivers/net/phy/bcm7xxx.c
  d88fd1b546ff ("net: phy: bcm7xxx: Fixed indirect MMD operations")
  f68d08c437f9 ("net: phy: bcm7xxx: Add EPHY entry for 72165")

net/sched/sch_api.c
  b193e15ac69d ("net: prevent user from passing illegal stab size")
  69508d43334e ("net_sched: Use struct_size() and flex_array_size() helpers")

Both cases trivial - adjacent code additions.

Signed-off-by: Jakub Kicinski <[email protected]>

Merge tag 'net-5.15-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net

Pull networking fixes from Jakub Kicinski:
 "Networking fixes, including fixes from mac80211, netfilter and bpf.

  Current release - regressions:

   - bpf, cgroup: assign cgroup in cgroup_sk_alloc when called from
     interrupt

   - mdio: revert mechanical patches which broke handling of optional
     resources

   - dev_addr_list: prevent address duplication

  Previous releases - regressions:

   - sctp: break out if skb_header_pointer returns NULL in sctp_rcv_ootb
     (NULL deref)

   - Revert "mac80211: do not use low data rates for data frames with no
     ack flag", fixing broadcast transmissions

   - mac80211: fix use-after-free in CCMP/GCMP RX

   - netfilter: include zone id in tuple hash again, minimize collisions

   - netfilter: nf_tables: unlink table before deleting it (race -> UAF)

   - netfilter: log: work around missing softdep backend module

   - mptcp: don't return sockets in foreign netns

   - sched: flower: protect fl_walk() with rcu (race -> UAF)

   - ixgbe: fix NULL pointer dereference in ixgbe_xdp_setup

   - smsc95xx: fix stalled rx after link change

   - enetc: fix the incorrect clearing of IF_MODE bits

   - ipv4: fix rtnexthop len when RTA_FLOW is present

   - dsa: mv88e6xxx: 6161: use correct MAX MTU config method for this
     SKU

   - e100: fix length calculation & buffer overrun in ethtool::get_regs

  Previous releases - always broken:

   - mac80211: fix using stale frag_tail skb pointer in A-MSDU tx

   - mac80211: drop frames from invalid MAC address in ad-hoc mode

   - af_unix: fix races in sk_peer_pid and sk_peer_cred accesses (race
     -> UAF)

   - bpf, x86: Fix bpf mapping of atomic fetch implementation

   - bpf: handle return value of BPF_PROG_TYPE_STRUCT_OPS prog

   - netfilter: ip6_tables: zero-initialize fragment offset

   - mhi: fix error path in mhi_net_newlink

   - af_unix: return errno instead of NULL in unix_create1() when over
     the fs.file-max limit

  Misc:

   - bpf: exempt CAP_BPF from checks against bpf_jit_limit

   - netfilter: conntrack: make max chain length random, prevent
     guessing buckets by attackers

   - netfilter: nf_nat_masquerade: make async masq_inet6_event handling
     generic, defer conntrack walk to work queue (prevent hogging RTNL
     lock)"

* tag 'net-5.15-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net: (77 commits)
  af_unix: fix races in sk_peer_pid and sk_peer_cred accesses
  net: stmmac: fix EEE init issue when paired with EEE capable PHYs
  net: dev_addr_list: handle first address in __hw_addr_add_ex
  net: sched: flower: protect fl_walk() with rcu
  net: introduce and use lock_sock_fast_nested()
  net: phy: bcm7xxx: Fixed indirect MMD operations
  net: hns3: disable firmware compatible features when uninstall PF
  net: hns3: fix always enable rx vlan filter problem after selftest
  net: hns3: PF enable promisc for VF when mac table is overflow
  net: hns3: fix show wrong state when add existing uc mac address
  net: hns3: fix mixed flag HCLGE_FLAG_MQPRIO_ENABLE and HCLGE_FLAG_DCB_ENABLE
  net: hns3: don't rollback when destroy mqprio fail
  net: hns3: remove tc enable checking
  net: hns3: do not allow call hns3_nic_net_open repeatedly
  ixgbe: Fix NULL pointer dereference in ixgbe_xdp_setup
  net: bridge: mcast: Associate the seqcount with its protecting lock.
  net: mdio-ipq4019: Fix the error for an optional regs resource
  net: hns3: fix hclge_dbg_dump_tm_pg() stack usage
  net: mdio: mscc-miim: Fix the mdio controller
  af_unix: Return errno instead of NULL in unix_create1().
  ...

bpf, xdp, docs: Correct some English grammar and spelling

Header DOC on include/net/xdp.h contained a few English grammer and
spelling errors.

Signed-off-by: Kev Jackson <[email protected]>
Signed-off-by: Daniel Borkmann <[email protected]>
Acked-by: Jesper Dangaard Brouer <[email protected]>
Link: https://lore.kernel.org/bpf/YVVaWmKqA8l9Tm4J@kev-VirtualBox

selftests/bpf: Use kselftest skip code for skipped tests

There are several test cases in the bpf directory are still using
exit 0 when they need to be skipped. Use kselftest framework skip
code instead so it can help us to distinguish the return status.

Criterion to filter out what should be fixed in bpf directory:
  grep -r "exit 0" -B1 | grep -i skip

This change might cause some false-positives if people are running
these test scripts directly and only checking their return codes,
which will change from 0 to 4. However I think the impact should be
small as most of our scripts here are already using this skip code.
And there will be no such issue if running them with the kselftest
framework.

Signed-off-by: Po-Hsu Lin <[email protected]>
Signed-off-by: Daniel Borkmann <[email protected]>
Link: https://lore.kernel.org/bpf/[email protected]

net/mlx5e: Mutually exclude setting of TX-port-TS and MQPRIO in channel mode

TX-port-TS hijacks the PTP traffic to a specific HW TX-queue. This
conflicts with MQPRIO in channel mode, which specifies explicitly which
TC accepts the packet. This patch mutually excludes the above
configuration.

Fixes: ec60c4581bd9 ("net/mlx5e: Support MQPRIO channel mode")
Signed-off-by: Aya Levin <[email protected]>
Reviewed-by: Tariq Toukan <[email protected]>
Signed-off-by: Saeed Mahameed <[email protected]>

net/mlx5e: Fix the presented RQ index in PTP stats

PTP-RQ counters title format contains PTP-RQ identifier, which is
mistakenly not passed to sprinft().
This leads to unexpected garbage values instead.
This patch fixes it.

Before applying the patch:
ethtool -S eth3 | grep ptp_rq
     ptp_rq15_packets: 0
     ptp_rq8_bytes: 0
     ptp_rq6_csum_complete: 0
     ptp_rq14_csum_complete_tail: 0
     ptp_rq3_csum_complete_tail_slow : 0
     ptp_rq9_csum_unnecessary: 0
     ptp_rq1_csum_unnecessary_inner: 0
     ptp_rq7_csum_none: 0
     ptp_rq10_xdp_drop: 0
     ptp_rq9_xdp_redirect: 0
     ptp_rq13_lro_packets: 0
     ptp_rq12_lro_bytes: 0
     ptp_rq10_ecn_mark: 0
     ptp_rq9_removed_vlan_packets: 0
     ptp_rq5_wqe_err: 0
     ptp_rq8_mpwqe_filler_cqes: 0
     ptp_rq2_mpwqe_filler_strides: 0
     ptp_rq5_oversize_pkts_sw_drop: 0
     ptp_rq6_buff_alloc_err: 0
     ptp_rq15_cqe_compress_blks: 0
     ptp_rq2_cqe_compress_pkts: 0
     ptp_rq2_cache_reuse: 0
     ptp_rq12_cache_full: 0
     ptp_rq11_cache_empty: 256
     ptp_rq12_cache_busy: 0
     ptp_rq11_cache_waive: 0
     ptp_rq12_congst_umr: 0
     ptp_rq11_arfs_err: 0
     ptp_rq9_recover: 0

After applying the patch:
ethtool -S eth3 | grep ptp_rq
     ptp_rq0_packets: 0
     ptp_rq0_bytes: 0
     ptp_rq0_csum_complete: 0
     ptp_rq0_csum_complete_tail: 0
     ptp_rq0_csum_complete_tail_slow : 0
     ptp_rq0_csum_unnecessary: 0
     ptp_rq0_csum_unnecessary_inner: 0
     ptp_rq0_csum_none: 0
     ptp_rq0_xdp_drop: 0
     ptp_rq0_xdp_redirect: 0
     ptp_rq0_lro_packets: 0
     ptp_rq0_lro_bytes: 0
     ptp_rq0_ecn_mark: 0
     ptp_rq0_removed_vlan_packets: 0
     ptp_rq0_wqe_err: 0
     ptp_rq0_mpwqe_filler_cqes: 0
     ptp_rq0_mpwqe_filler_strides: 0
     ptp_rq0_oversize_pkts_sw_drop: 0
     ptp_rq0_buff_alloc_err: 0
     ptp_rq0_cqe_compress_blks: 0
     ptp_rq0_cqe_compress_pkts: 0
     ptp_rq0_cache_reuse: 0
     ptp_rq0_cache_full: 0
     ptp_rq0_cache_empty: 256
     ptp_rq0_cache_busy: 0
     ptp_rq0_cache_waive: 0
     ptp_rq0_congst_umr: 0
     ptp_rq0_arfs_err: 0
     ptp_rq0_recover: 0

Fixes: a28359e922c6 ("net/mlx5e: Add PTP-RX statistics")
Signed-off-by: Lama Kayal <[email protected]>
Reviewed-by: Tariq Toukan <[email protected]>
Signed-off-by: Saeed Mahameed <[email protected]>

net/mlx5: Fix setting number of EQs of SFs

When setting number of completion EQs of the SF, consider number of
online CPUs.
Without this consideration, when number of online cpus are less than 8,
unnecessary 8 completion EQs are allocated.

Fixes: c36326d38d93 ("net/mlx5: Round-Robin EQs over IRQs")
Signed-off-by: Shay Drory <[email protected]>
Reviewed-by: Parav Pandit <[email protected]>
Signed-off-by: Saeed Mahameed <[email protected]>

net/mlx5: Fix length of irq_index in chars

The maximum irq_index can be 2047, This means irq_name should have 4
characters reserve for the irq_index. Hence, increase it to 4.

Fixes: 3af26495a247 ("net/mlx5: Enlarge interrupt field in CREATE_EQ")
Signed-off-by: Shay Drory <[email protected]>
Reviewed-by: Parav Pandit <[email protected]>
Signed-off-by: Saeed Mahameed <[email protected]>

net/mlx5: Avoid generating event after PPS out in Real time mode

When in Real-time mode, HW clock is synced with the PTP daemon. Hence
driver should not re-calibrate the next pulse (via MTPPSE repetitive
events mechanism).

This patch arms repetitive events only in free-running mode.

Fixes: 432119de33d9 ("net/mlx5: Add cyc2time HW translation mode support")
Signed-off-by: Aya Levin <[email protected]>
Reviewed-by: Eran Ben Elisha <[email protected]>
Signed-off-by: Saeed Mahameed <[email protected]>

net/mlx5: Force round second at 1PPS out start time

Allow configuration of 1PPS start time only with time-stamp representing
a round second. Prior to this patch driver allowed setting of a
non-round-second which is not supported by the device. Avoid unexpected
behavior by restricting start-time configuration to a round-second.

Fixes: 4272f9b88db9 ("net/mlx5e: Change 1PPS out scheme")
Signed-off-by: Aya Levin <[email protected]>
Reviewed-by: Eran Ben Elisha <[email protected]>
Signed-off-by: Saeed Mahameed <[email protected]>

net/mlx5: E-Switch, Fix double allocation of acl flow counter

Flow counter is allocated in eswitch legacy acl setting functions
without checking if already allocated by previous setting. Add a check
to avoid such double allocation.

Fixes: 07bab9502641 ("net/mlx5: E-Switch, Refactor eswitch ingress acl codes")
Fixes: ea651a86d468 ("net/mlx5: E-Switch, Refactor eswitch egress acl codes")
Signed-off-by: Moshe Shemesh <[email protected]>
Reviewed-by: Tariq Toukan <[email protected]>
Signed-off-by: Saeed Mahameed <[email protected]>

net/mlx5e: Improve MQPRIO resiliency

* Add netdev->tc_to_txq rollback in case of failure in
  mlx5e_update_netdev_queues().
* Fix broken transition between the two modes:
  MQPRIO DCB mode with tc==8, and MQPRIO channel mode.
* Disable MQPRIO channel mode if re-attaching with a different number
  of channels.
* Improve code sharing.

Fixes: ec60c4581bd9 ("net/mlx5e: Support MQPRIO channel mode")
Signed-off-by: Tariq Toukan <[email protected]>
Reviewed-by: Maxim Mikityanskiy <[email protected]>
Signed-off-by: Saeed Mahameed <[email protected]>

net/mlx5e: Keep the value for maximum number of channels in-sync

The value for maximum number of channels is first calculated based
on the netdev's profile and current function resources (specifically,
number of MSIX vectors, which depends among other things on the number
of online cores in the system).
This value is then used to calculate the netdev's number of rxqs/txqs.
Once created (by alloc_etherdev_mqs), the number of netdev's rxqs/txqs
is constant and we must not exceed it.

To achieve this, keep the maximum number of channels in sync upon any
netdevice re-attach.

Use mlx5e_get_max_num_channels() for calculating the number of netdev's
rxqs/txqs. After netdev is created, use mlx5e_calc_max_nch() (which
coinsiders core device resources, profile, and netdev) to init or
update priv->max_nch.

Before this patch, the value of priv->max_nch might get out of sync,
mistakenly allowing accesses to out-of-bounds objects, which would
crash the system.

Track the number of channels stats structures used in a separate
field, as they are persistent to suspend/resume operations. All the
collected stats of every channel index that ever existed should be
preserved. They are reset only when struct mlx5e_priv is,
in mlx5e_priv_cleanup(), which is part of the profile changing flow.

There is no point anymore in blocking a profile change due to max_nch
mismatch in mlx5e_netdev_change_profile(). Remove the limitation.

Fixes: a1f240f18017 ("net/mlx5e: Adjust to max number of channles when re-attaching")
Signed-off-by: Tariq Toukan <[email protected]>
Reviewed-by: Aya Levin <[email protected]>
Reviewed-by: Maxim Mikityanskiy <[email protected]>
Signed-off-by: Saeed Mahameed <[email protected]>

net/mlx5e: IPSEC RX, enable checksum complete

Currently in Rx data path IPsec crypto offloaded packets uses
csum_none flag, so checksum is handled by the stack, this naturally
have some performance/cpu utilization impact on such flows. As Nvidia
NIC starting from ConnectX6DX provides checksum complete value out of
the box also for such flows there is no sense in taking csum_none path,
furthermore the stack (xfrm) have the method to handle checksum complete
corrections for such flows i.e. IPsec trailer removal and consequently
checksum value adjustment.

Because of the above and in addition the ConnectX6DX is the first HW
which supports IPsec crypto offload then it is safe to report csum
complete for IPsec offloaded traffic.

Fixes: b2ac7541e377 ("net/mlx5e: IPsec: Add Connect-X IPsec Rx data path offload")
Signed-off-by: Raed Salem <[email protected]>
Signed-off-by: Saeed Mahameed <[email protected]>

Bluetooth: btrtl: Add support for MSFT extension to rtl8821c devices

The rtl8821c class of Realtek devices does support the Microsoft
extensions and thus enable them.

Signed-off-by: Marcel Holtmann <[email protected]>
Signed-off-by: Luiz Augusto von Dentz <[email protected]>

Merge tag 'gpio-fixes-for-v5.15-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/brgl/linux

Pull gpio fixes from Bartosz Golaszewski:
 "A single fix for the gpio-pca953x driver and two commits updating the
  MAINTAINERS entries for Mun Yew Tham (GPIO specific) and myself
  (treewide after a change in professional situation).

  Summary:

   - don't ignore I2C errors in gpio-pca953x

   - update MAINTAINERS entries for Mun Yew Tham and myself"

* tag 'gpio-fixes-for-v5.15-rc4' of git://git.kernel.org/pub/scm/linux/kernel/git/brgl/linux:
  MAINTAINERS: Update Mun Yew Tham as Altera Pio Driver maintainer
  MAINTAINERS: update my email address
  gpio: pca953x: do not ignore i2c errors

Merge tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rdma/rdma

Pull rdma fixes from Jason Gunthorpe:
 "Not much too exciting here, although two syzkaller bugs that seem to
  have 9 lives may have finally been squashed.

  Several core bugs and a batch of driver bug fixes:

   - Fix compilation problems in qib and hfi1

   - Do not corrupt the joined multicast group state when using
     SEND_ONLY

   - Several CMA bugs, a reference leak for listening and two syzkaller
     crashers

   - Various bug fixes for irdma

   - Fix a Sleeping while atomic bug in usnic

   - Properly sanitize kernel pointers in dmesg

   - Two bugs in the 64b CQE support for hns"

* tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rdma/rdma:
  RDMA/hns: Add the check of the CQE size of the user space
  RDMA/hns: Fix the size setting error when copying CQE in clean_cq()
  RDMA/hfi1: Fix kernel pointer leak
  RDMA/usnic: Lock VF with mutex instead of spinlock
  RDMA/hns: Work around broken constant propagation in gcc 8
  RDMA/cma: Ensure rdma_addr_cancel() happens before issuing more requests
  RDMA/cma: Do not change route.addr.src_addr.ss_family
  RDMA/irdma: Report correct WC error when there are MW bind errors
  RDMA/irdma: Report correct WC error when transport retry counter is exceeded
  RDMA/irdma: Validate number of CQ entries on create CQ
  RDMA/irdma: Skip CQP ring during a reset
  MAINTAINERS: Update Broadcom RDMA maintainers
  RDMA/cma: Fix listener leak in rdma_cma_listen_on_all() failure
  IB/cma: Do not send IGMP leaves for sendonly Multicast groups
  IB/qib: Fix clang confusion of NULL pointer comparison

af_unix: fix races in sk_peer_pid and sk_peer_cred accesses

Jann Horn reported that SO_PEERCRED and SO_PEERGROUPS implementations
are racy, as af_unix can concurrently change sk_peer_pid and sk_peer_cred.

In order to fix this issue, this patch adds a new spinlock that needs
to be used whenever these fields are read or written.

Jann also pointed out that l2cap_sock_get_peer_pid_cb() is currently
reading sk->sk_peer_pid which makes no sense, as this field
is only possibly set by AF_UNIX sockets.
We will have to clean this in a separate patch.
This could be done by reverting b48596d1dc25 "Bluetooth: L2CAP: Add get_peer_pid callback"
or implementing what was truly expected.

Fixes: 109f6e39fa07 ("af_unix: Allow SO_PEERCRED to work across namespaces.")
Signed-off-by: Eric Dumazet <[email protected]>
Reported-by: Jann Horn <[email protected]>
Cc: Eric W. Biederman <[email protected]>
Cc: Luiz Augusto von Dentz <[email protected]>
Cc: Marcel Holtmann <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

Merge branch 'snmp-optimizations'

Eric Dumazet says:

====================
net: snmp: minor optimizations

Fetching many SNMP counters on hosts with large number of cpus
takes a lot of time. mptcp still uses the old non-batched
fashion which is not cache friendly.
====================

Signed-off-by: David S. Miller <[email protected]>

mptcp: use batch snmp operations in mptcp_seq_show()

Using snmp_get_cpu_field_batch() allows for better cpu cache
utilization, especially on hosts with large number of cpus.

Also remove special handling when mptcp mibs where not yet
allocated.

I chose to use temporary storage on the stack to keep this patch simple.
We might in the future use the storage allocated in netstat_seq_show().

Combined with prior patch (inlining snmp_get_cpu_field)
time to fetch and output mptcp counters on a 256 cpu host [1]
goes from 75 usec to 16 usec.

[1] L1 cache size is 32KB, it is not big enough to hold all dataset.

Signed-off-by: Eric Dumazet <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

net: snmp: inline snmp_get_cpu_field()

This trivial function is called ~90,000 times on 256 cpus hosts,
when reading /proc/net/netstat. And this number keeps inflating.

Inlining it saves many cycles.

Signed-off-by: Eric Dumazet <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

net/mlx4_en: Add XDP_REDIRECT statistics

Add counters for XDP REDIRECT success and failure. This brings the
redirect path in line with metrics gathered via the other XDP paths.

Signed-off-by: Joshua Roys <[email protected]>
Reviewed-by: Tariq Toukan <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

net: stmmac: fix EEE init issue when paired with EEE capable PHYs

When STMMAC is paired with Energy-Efficient Ethernet(EEE) capable PHY,
and the PHY is advertising EEE by default, we need to enable EEE on the
xPCS side too, instead of having user to manually trigger the enabling
config via ethtool.

Fixed this by adding xpcs_config_eee() call in stmmac_eee_init().

Fixes: 7617af3d1a5e ("net: pcs: Introducing support for DWC xpcs Energy Efficient Ethernet")
Cc: Michael Sit Wei Hong <[email protected]>
Signed-off-by: Wong Vee Khee <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

ixgbe: let the xdpdrv work with more than 64 cpus

Originally, ixgbe driver doesn't allow the mounting of xdpdrv if the
server is equipped with more than 64 cpus online. So it turns out that
the loading of xdpdrv causes the "NOMEM" failure.

Actually, we can adjust the algorithm and then make it work through
mapping the current cpu to some xdp ring with the protect of @tx_lock.

Here are some numbers before/after applying this patch with xdp-example
loaded on the eth0X:

As client (tx path):
                     Before    After
TCP_STREAM send-64   734.14    714.20
TCP_STREAM send-128  1401.91   1395.05
TCP_STREAM send-512  5311.67   5292.84
TCP_STREAM send-1k   9277.40   9356.22 (not stable)
TCP_RR     send-1    22559.75  21844.22
TCP_RR     send-128  23169.54  22725.13
TCP_RR     send-512  21670.91  21412.56

As server (rx path):
                     Before    After
TCP_STREAM send-64   1416.49   1383.12
TCP_STREAM send-128  3141.49   3055.50
TCP_STREAM send-512  9488.73   9487.44
TCP_STREAM send-1k   9491.17   9356.22 (not stable)
TCP_RR     send-1    23617.74  23601.60
...

Notice: the TCP_RR mode is unstable as the official document explains.

I tested many times with different parameters combined through netperf.
Though the result is not that accurate, I cannot see much influence on
this patch. The static key is places on the hot path, but it actually
shouldn't cause a huge regression theoretically.

Co-developed-by: Shujin Li <[email protected]>
Signed-off-by: Shujin Li <[email protected]>
Signed-off-by: Jason Xing <[email protected]>
Tested-by: Sandeep Penigalapati <[email protected]>
Signed-off-by: Tony Nguyen <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

Merge branch 'SO_RESEVED_MEM'

Wei Wang says:

====================
net: add new socket option SO_RESERVE_MEM

This patch series introduces a new socket option SO_RESERVE_MEM.
This socket option provides a mechanism for users to reserve a certain
amount of memory for the socket to use. When this option is set, kernel
charges the user specified amount of memory to memcg, as well as
sk_forward_alloc. This amount of memory is not reclaimable and is
available in sk_forward_alloc for this socket.
With this socket option set, the networking stack spends less cycles
doing forward alloc and reclaim, which should lead to better system
performance, with the cost of an amount of pre-allocated and
unreclaimable memory, even under memory pressure.
With a tcp_stream test with 10 flows running on a simulated 100ms RTT
link, I can see the cycles spent in __sk_mem_raise_allocated() dropping
by ~0.02%. Not a whole lot, since we already have logic in
sk_mem_uncharge() to only reclaim 1MB when sk_forward_alloc has more
than 2MB free space. But on a system suffering memory pressure
constently, the savings should be more.

The first patch is the implementation of this socket option. The
following 2 patches change the tcp stack to make use of this reserved
memory when under memory pressure. This makes the tcp stack behavior
more flexible when under memory pressure, and provides a way for user to
control the distribution of the memory among its sockets.
With a TCP connection on a simulated 100ms RTT link, the default
throughput under memory pressure is ~500Kbps. With SO_RESERVE_MEM set to
100KB, the throughput under memory pressure goes up to ~3.5Mbps.

Change since v2:
- Added description for new field added in struct sock in patch 1
Change since v1:
- Added performance stats in cover letter and rebased
====================

Signed-off-by: David S. Miller <[email protected]>

tcp: adjust rcv_ssthresh according to sk_reserved_mem

When user sets SO_RESERVE_MEM socket option, in order to utilize the
reserved memory when in memory pressure state, we adjust rcv_ssthresh
according to the available reserved memory for the socket, instead of
using 4 * advmss always.

Signed-off-by: Wei Wang <[email protected]>
Signed-off-by: Eric Dumazet <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

tcp: adjust sndbuf according to sk_reserved_mem

If user sets SO_RESERVE_MEM socket option, in order to fully utilize the
reserved memory in memory pressure state on the tx path, we modify the
logic in sk_stream_moderate_sndbuf() to set sk_sndbuf according to
available reserved memory, instead of MIN_SOCK_SNDBUF, and adjust it
when new data is acked.

Signed-off-by: Wei Wang <[email protected]>
Signed-off-by: Eric Dumazet <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

net: add new socket option SO_RESERVE_MEM

This socket option provides a mechanism for users to reserve a certain
amount of memory for the socket to use. When this option is set, kernel
charges the user specified amount of memory to memcg, as well as
sk_forward_alloc. This amount of memory is not reclaimable and is
available in sk_forward_alloc for this socket.
With this socket option set, the networking stack spends less cycles
doing forward alloc and reclaim, which should lead to better system
performance, with the cost of an amount of pre-allocated and
unreclaimable memory, even under memory pressure.

Note:
This socket option is only available when memory cgroup is enabled and we
require this reserved memory to be charged to the user's memcg. We hope
this could avoid mis-behaving users to abused this feature to reserve a
large amount on certain sockets and cause unfairness for others.

Signed-off-by: Wei Wang <[email protected]>
Signed-off-by: Eric Dumazet <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

net: dev_addr_list: handle first address in __hw_addr_add_ex

struct dev_addr_list is used for device addresses, unicast addresses
and multicast addresses. The first of those needs special handling
of the main address - netdev->dev_addr points directly the data
of the entry and drivers write to it freely, so we can't maintain
it in the rbtree (for now, at least, to be fixed in net-next).

Current work around sprinkles special handling of the first
address on the list throughout the code but it missed the case
where address is being added. First address will not be visible
during subsequent adds.

Syzbot found a warning where unicast addresses are modified
without holding the rtnl lock, tl;dr is that team generates
the same modification multiple times, not necessarily when
right locks are held.

In the repro we have:

  macvlan -> team -> veth

macvlan adds a unicast address to the team. Team then pushes
that address down to its memebers (veths). Next something unrelated
makes team sync member addrs again, and because of the bug
the addr entries get duplicated in the veths. macvlan gets
removed, removes its addr from team which removes only one
of the duplicated addresses from veths. This removal is done
under rtnl. Next syzbot uses iptables to add a multicast addr
to team (which does not hold rtnl lock). Team syncs veth addrs,
but because veths' unicast list still has the duplicate it will
also get sync, even though this update is intended for mc addresses.
Again, uc address updates need rtnl lock, boom.

Reported-by: [email protected]
Fixes: 406f42fa0d3c ("net-next: When a bond have a massive amount of VLANs with IPv6 addresses, performance of changing link state, attaching a VRF, changing an IPv6 address, etc. go down dramtically.")
Signed-off-by: Jakub Kicinski <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

net: phy: marvell10g: add downshift tunable support

Add support for the downshift tunable for the Marvell 88x3310 PHY.
Downshift is only usable with firmware 0.3.5.0 and later.

Signed-off-by: Russell King <[email protected]>
Reviewed-by: Andrew Lunn <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

net: sched: flower: protect fl_walk() with rcu

Patch that refactored fl_walk() to use idr_for_each_entry_continue_ul()
also removed rcu protection of individual filters which causes following
use-after-free when filter is deleted concurrently. Fix fl_walk() to obtain
rcu read lock while iterating and taking the filter reference and temporary
release the lock while calling arg->fn() callback that can sleep.

KASAN trace:

[  352.773640] ==================================================================
[  352.775041] BUG: KASAN: use-after-free in fl_walk+0x159/0x240 [cls_flower]
[  352.776304] Read of size 4 at addr ffff8881c8251480 by task tc/2987

[  352.777862] CPU: 3 PID: 2987 Comm: tc Not tainted 5.15.0-rc2+ #2
[  352.778980] Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS rel-1.13.0-0-gf21b5a4aeb02-prebuilt.qemu.org 04/01/2014
[  352.781022] Call Trace:
[  352.781573]  dump_stack_lvl+0x46/0x5a
[  352.782332]  print_address_description.constprop.0+0x1f/0x140
[  352.783400]  ? fl_walk+0x159/0x240 [cls_flower]
[  352.784292]  ? fl_walk+0x159/0x240 [cls_flower]
[  352.785138]  kasan_report.cold+0x83/0xdf
[  352.785851]  ? fl_walk+0x159/0x240 [cls_flower]
[  352.786587]  kasan_check_range+0x145/0x1a0
[  352.787337]  fl_walk+0x159/0x240 [cls_flower]
[  352.788163]  ? fl_put+0x10/0x10 [cls_flower]
[  352.789007]  ? __mutex_unlock_slowpath.constprop.0+0x220/0x220
[  352.790102]  tcf_chain_dump+0x231/0x450
[  352.790878]  ? tcf_chain_tp_delete_empty+0x170/0x170
[  352.791833]  ? __might_sleep+0x2e/0xc0
[  352.792594]  ? tfilter_notify+0x170/0x170
[  352.793400]  ? __mutex_unlock_slowpath.constprop.0+0x220/0x220
[  352.794477]  tc_dump_tfilter+0x385/0x4b0
[  352.795262]  ? tc_new_tfilter+0x1180/0x1180
[  352.796103]  ? __mod_node_page_state+0x1f/0xc0
[  352.796974]  ? __build_skb_around+0x10e/0x130
[  352.797826]  netlink_dump+0x2c0/0x560
[  352.798563]  ? netlink_getsockopt+0x430/0x430
[  352.799433]  ? __mutex_unlock_slowpath.constprop.0+0x220/0x220
[  352.800542]  __netlink_dump_start+0x356/0x440
[  352.801397]  rtnetlink_rcv_msg+0x3ff/0x550
[  352.802190]  ? tc_new_tfilter+0x1180/0x1180
[  352.802872]  ? rtnl_calcit.isra.0+0x1f0/0x1f0
[  352.803668]  ? tc_new_tfilter+0x1180/0x1180
[  352.804344]  ? _copy_from_iter_nocache+0x800/0x800
[  352.805202]  ? kasan_set_track+0x1c/0x30
[  352.805900]  netlink_rcv_skb+0xc6/0x1f0
[  352.806587]  ? rht_deferred_worker+0x6b0/0x6b0
[  352.807455]  ? rtnl_calcit.isra.0+0x1f0/0x1f0
[  352.808324]  ? netlink_ack+0x4d0/0x4d0
[  352.809086]  ? netlink_deliver_tap+0x62/0x3d0
[  352.809951]  netlink_unicast+0x353/0x480
[  352.810744]  ? netlink_attachskb+0x430/0x430
[  352.811586]  ? __alloc_skb+0xd7/0x200
[  352.812349]  netlink_sendmsg+0x396/0x680
[  352.813132]  ? netlink_unicast+0x480/0x480
[  352.813952]  ? __import_iovec+0x192/0x210
[  352.814759]  ? netlink_unicast+0x480/0x480
[  352.815580]  sock_sendmsg+0x6c/0x80
[  352.816299]  ____sys_sendmsg+0x3a5/0x3c0
[  352.817096]  ? kernel_sendmsg+0x30/0x30
[  352.817873]  ? __ia32_sys_recvmmsg+0x150/0x150
[  352.818753]  ___sys_sendmsg+0xd8/0x140
[  352.819518]  ? sendmsg_copy_msghdr+0x110/0x110
[  352.820402]  ? ___sys_recvmsg+0xf4/0x1a0
[  352.821110]  ? __copy_msghdr_from_user+0x260/0x260
[  352.821934]  ? _raw_spin_lock+0x81/0xd0
[  352.822680]  ? __handle_mm_fault+0xef3/0x1b20
[  352.823549]  ? rb_insert_color+0x2a/0x270
[  352.824373]  ? copy_page_range+0x16b0/0x16b0
[  352.825209]  ? perf_event_update_userpage+0x2d0/0x2d0
[  352.826190]  ? __fget_light+0xd9/0xf0
[  352.826941]  __sys_sendmsg+0xb3/0x130
[  352.827613]  ? __sys_sendmsg_sock+0x20/0x20
[  352.828377]  ? do_user_addr_fault+0x2c5/0x8a0
[  352.829184]  ? fpregs_assert_state_consistent+0x52/0x60
[  352.830001]  ? exit_to_user_mode_prepare+0x32/0x160
[  352.830845]  do_syscall_64+0x35/0x80
[  352.831445]  entry_SYSCALL_64_after_hwframe+0x44/0xae
[  352.832331] RIP: 0033:0x7f7bee973c17
[  352.833078] Code: 0c 00 f7 d8 64 89 02 48 c7 c0 ff ff ff ff eb b7 0f 1f 00 f3 0f 1e fa 64 8b 04 25 18 00 00 00 85 c0 75 10 b8 2e 00 00 00 0f 05 <48> 3d 00 f0 ff ff 77 51 c3 48 83 ec 28 89 54 24 1c 48 89 74 24 10
[  352.836202] RSP: 002b:00007ffcbb368e28 EFLAGS: 00000246 ORIG_RAX: 000000000000002e
[  352.837524] RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00007f7bee973c17
[  352.838715] RDX: 0000000000000000 RSI: 00007ffcbb368e50 RDI: 0000000000000003
[  352.839838] RBP: 00007ffcbb36d090 R08: 00000000cea96d79 R09: 00007f7beea34a40
[  352.841021] R10: 00000000004059bb R11: 0000000000000246 R12: 000000000046563f
[  352.842208] R13: 0000000000000000 R14: 0000000000000000 R15: 00007ffcbb36d088

[  352.843784] Allocated by task 2960:
[  352.844451]  kasan_save_stack+0x1b/0x40
[  352.845173]  __kasan_kmalloc+0x7c/0x90
[  352.845873]  fl_change+0x282/0x22db [cls_flower]
[  352.846696]  tc_new_tfilter+0x6cf/0x1180
[  352.847493]  rtnetlink_rcv_msg+0x471/0x550
[  352.848323]  netlink_rcv_skb+0xc6/0x1f0
[  352.849097]  netlink_unicast+0x353/0x480
[  352.849886]  netlink_sendmsg+0x396/0x680
[  352.850678]  sock_sendmsg+0x6c/0x80
[  352.851398]  ____sys_sendmsg+0x3a5/0x3c0
[  352.852202]  ___sys_sendmsg+0xd8/0x140
[  352.852967]  __sys_sendmsg+0xb3/0x130
[  352.853718]  do_syscall_64+0x35/0x80
[  352.854457]  entry_SYSCALL_64_after_hwframe+0x44/0xae

[  352.855830] Freed by task 7:
[  352.856421]  kasan_save_stack+0x1b/0x40
[  352.857139]  kasan_set_track+0x1c/0x30
[  352.857854]  kasan_set_free_info+0x20/0x30
[  352.858609]  __kasan_slab_free+0xed/0x130
[  352.859348]  kfree+0xa7/0x3c0
[  352.859951]  process_one_work+0x44d/0x780
[  352.860685]  worker_thread+0x2e2/0x7e0
[  352.861390]  kthread+0x1f4/0x220
[  352.862022]  ret_from_fork+0x1f/0x30

[  352.862955] Last potentially related work creation:
[  352.863758]  kasan_save_stack+0x1b/0x40
[  352.864378]  kasan_record_aux_stack+0xab/0xc0
[  352.865028]  insert_work+0x30/0x160
[  352.865617]  __queue_work+0x351/0x670
[  352.866261]  rcu_work_rcufn+0x30/0x40
[  352.866917]  rcu_core+0x3b2/0xdb0
[  352.867561]  __do_softirq+0xf6/0x386

[  352.868708] Second to last potentially related work creation:
[  352.869779]  kasan_save_stack+0x1b/0x40
[  352.870560]  kasan_record_aux_stack+0xab/0xc0
[  352.871426]  call_rcu+0x5f/0x5c0
[  352.872108]  queue_rcu_work+0x44/0x50
[  352.872855]  __fl_put+0x17c/0x240 [cls_flower]
[  352.873733]  fl_delete+0xc7/0x100 [cls_flower]
[  352.874607]  tc_del_tfilter+0x510/0xb30
[  352.886085]  rtnetlink_rcv_msg+0x471/0x550
[  352.886875]  netlink_rcv_skb+0xc6/0x1f0
[  352.887636]  netlink_unicast+0x353/0x480
[  352.888285]  netlink_sendmsg+0x396/0x680
[  352.888942]  sock_sendmsg+0x6c/0x80
[  352.889583]  ____sys_sendmsg+0x3a5/0x3c0
[  352.890311]  ___sys_sendmsg+0xd8/0x140
[  352.891019]  __sys_sendmsg+0xb3/0x130
[  352.891716]  do_syscall_64+0x35/0x80
[  352.892395]  entry_SYSCALL_64_after_hwframe+0x44/0xae

[  352.893666] The buggy address belongs to the object at ffff8881c8251000
                which belongs to the cache kmalloc-2k of size 2048
[  352.895696] The buggy address is located 1152 bytes inside of
                2048-byte region [ffff8881c8251000, ffff8881c8251800)
[  352.897640] The buggy address belongs to the page:
[  352.898492] page:00000000213bac35 refcount:1 mapcount:0 mapping:0000000000000000 index:0x0 pfn:0x1c8250
[  352.900110] head:00000000213bac35 order:3 compound_mapcount:0 compound_pincount:0
[  352.901541] flags: 0x2ffff800010200(slab|head|node=0|zone=2|lastcpupid=0x1ffff)
[  352.902908] raw: 002ffff800010200 0000000000000000 dead000000000122 ffff888100042f00
[  352.904391] raw: 0000000000000000 0000000000080008 00000001ffffffff 0000000000000000
[  352.905861] page dumped because: kasan: bad access detected

[  352.907323] Memory state around the buggy address:
[  352.908218]  ffff8881c8251380: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
[  352.909471]  ffff8881c8251400: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
[  352.910735] >ffff8881c8251480: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
[  352.912012]                    ^
[  352.912642]  ffff8881c8251500: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
[  352.913919]  ffff8881c8251580: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
[  352.915185] ==================================================================

Fixes: d39d714969cd ("idr: introduce idr_for_each_entry_continue_ul()")
Signed-off-by: Vlad Buslov <[email protected]>
Acked-by: Cong Wang <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

octeontx2-af: Remove redundant initialization of variable pin

The variable pin is being initialized with a value that is never
read, it is being updated later on in only one case of a switch
statement.  The assignment is redundant and can be removed.

Addresses-Coverity: ("Unused value")
Signed-off-by: Colin Ian King <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

net: macb: ptp: Switch to gettimex64() interface

The macb PTP support currently implements the `gettime64` callback to allow
to retrieve the hardware clock time. Update the implementation to provide
the `gettimex64` callback instead.

The difference between the two is that with `gettime64` a snapshot of the
system clock is taken before and after invoking the callback. Whereas
`gettimex64` expects the callback itself to take the snapshots.

To get the time from the macb Ethernet core multiple register accesses have
to be done. Only one of which will happen at the time reported by the
function. This leads to a non-symmetric delay and adds a slight offset
between the hardware and system clock time when using the `gettime64`
method. This offset can be a few 100 nanoseconds. Switching to the
`gettimex64` method allows for a more precise correlation of the hardware
and system clocks and results in a lower offset between the two.

On a Xilinx ZynqMP system `phc2sys` reports a delay of 1120 ns before and
300 ns after the patch. With the latter being mostly symmetric.

Signed-off-by: Lars-Peter Clausen <[email protected]>
Acked-by: Richard Cochran <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

dissector: do not set invalid PPP protocol

The following flower filter fails to match non-PPP_IP{V6} packets
wrapped in PPP_SES protocol:

tc filter add dev eth0 ingress protocol ppp_ses flower \
        action simple sdata hi64

The reason is that proto local variable is being set even when
FLOW_DISSECT_RET_OUT_BAD status is returned.

The fix is to avoid setting proto variable if the PPP protocol is unknown.

Signed-off-by: Boris Sukholitko <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

net: dsa: rtl8366rb: Use core filtering tracking

We added a state variable to track whether a certain port
was VLAN filtering or not, but we can just inquire the DSA
core about this.

Cc: Vladimir Oltean <[email protected]>
Cc: Mauri Sandberg <[email protected]>
Cc: DENG Qingfang <[email protected]>
Cc: Alvin Šipraga <[email protected]>
Cc: Reviewed-by: Florian Fainelli <[email protected]>
Signed-off-by: Linus Walleij <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

net: introduce and use lock_sock_fast_nested()

Syzkaller reported a false positive deadlock involving
the nl socket lock and the subflow socket lock:

MPTCP: kernel_bind error, err=-98
============================================
WARNING: possible recursive locking detected
5.15.0-rc1-syzkaller #0 Not tainted
--------------------------------------------
syz-executor998/6520 is trying to acquire lock:
ffff8880795718a0 (k-sk_lock-AF_INET){+.+.}-{0:0}, at: mptcp_close+0x267/0x7b0 net/mptcp/protocol.c:2738

but task is already holding lock:
ffff8880787c8c60 (k-sk_lock-AF_INET){+.+.}-{0:0}, at: lock_sock include/net/sock.h:1612 [inline]
ffff8880787c8c60 (k-sk_lock-AF_INET){+.+.}-{0:0}, at: mptcp_close+0x23/0x7b0 net/mptcp/protocol.c:2720

other info that might help us debug this:
 Possible unsafe locking scenario:

       CPU0
       ----
  lock(k-sk_lock-AF_INET);
  lock(k-sk_lock-AF_INET);

 *** DEADLOCK ***

 May be due to missing lock nesting notation

3 locks held by syz-executor998/6520:
 #0: ffffffff8d176c50 (cb_lock){++++}-{3:3}, at: genl_rcv+0x15/0x40 net/netlink/genetlink.c:802
 #1: ffffffff8d176d08 (genl_mutex){+.+.}-{3:3}, at: genl_lock net/netlink/genetlink.c:33 [inline]
 #1: ffffffff8d176d08 (genl_mutex){+.+.}-{3:3}, at: genl_rcv_msg+0x3e0/0x580 net/netlink/genetlink.c:790
 #2: ffff8880787c8c60 (k-sk_lock-AF_INET){+.+.}-{0:0}, at: lock_sock include/net/sock.h:1612 [inline]
 #2: ffff8880787c8c60 (k-sk_lock-AF_INET){+.+.}-{0:0}, at: mptcp_close+0x23/0x7b0 net/mptcp/protocol.c:2720

stack backtrace:
CPU: 1 PID: 6520 Comm: syz-executor998 Not tainted 5.15.0-rc1-syzkaller #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
Call Trace:
 __dump_stack lib/dump_stack.c:88 [inline]
 dump_stack_lvl+0xcd/0x134 lib/dump_stack.c:106
 print_deadlock_bug kernel/locking/lockdep.c:2944 [inline]
 check_deadlock kernel/locking/lockdep.c:2987 [inline]
 validate_chain kernel/locking/lockdep.c:3776 [inline]
 __lock_acquire.cold+0x149/0x3ab kernel/locking/lockdep.c:5015
 lock_acquire kernel/locking/lockdep.c:5625 [inline]
 lock_acquire+0x1ab/0x510 kernel/locking/lockdep.c:5590
 lock_sock_fast+0x36/0x100 net/core/sock.c:3229
 mptcp_close+0x267/0x7b0 net/mptcp/protocol.c:2738
 inet_release+0x12e/0x280 net/ipv4/af_inet.c:431
 __sock_release net/socket.c:649 [inline]
 sock_release+0x87/0x1b0 net/socket.c:677
 mptcp_pm_nl_create_listen_socket+0x238/0x2c0 net/mptcp/pm_netlink.c:900
 mptcp_nl_cmd_add_addr+0x359/0x930 net/mptcp/pm_netlink.c:1170
 genl_family_rcv_msg_doit+0x228/0x320 net/netlink/genetlink.c:731
 genl_family_rcv_msg net/netlink/genetlink.c:775 [inline]
 genl_rcv_msg+0x328/0x580 net/netlink/genetlink.c:792
 netlink_rcv_skb+0x153/0x420 net/netlink/af_netlink.c:2504
 genl_rcv+0x24/0x40 net/netlink/genetlink.c:803
 netlink_unicast_kernel net/netlink/af_netlink.c:1314 [inline]
 netlink_unicast+0x533/0x7d0 net/netlink/af_netlink.c:1340
 netlink_sendmsg+0x86d/0xdb0 net/netlink/af_netlink.c:1929
 sock_sendmsg_nosec net/socket.c:704 [inline]
 sock_sendmsg+0xcf/0x120 net/socket.c:724
 sock_no_sendpage+0x101/0x150 net/core/sock.c:2980
 kernel_sendpage.part.0+0x1a0/0x340 net/socket.c:3504
 kernel_sendpage net/socket.c:3501 [inline]
 sock_sendpage+0xe5/0x140 net/socket.c:1003
 pipe_to_sendpage+0x2ad/0x380 fs/splice.c:364
 splice_from_pipe_feed fs/splice.c:418 [inline]
 __splice_from_pipe+0x43e/0x8a0 fs/splice.c:562
 splice_from_pipe fs/splice.c:597 [inline]
 generic_splice_sendpage+0xd4/0x140 fs/splice.c:746
 do_splice_from fs/splice.c:767 [inline]
 direct_splice_actor+0x110/0x180 fs/splice.c:936
 splice_direct_to_actor+0x34b/0x8c0 fs/splice.c:891
 do_splice_direct+0x1b3/0x280 fs/splice.c:979
 do_sendfile+0xae9/0x1240 fs/read_write.c:1249
 __do_sys_sendfile64 fs/read_write.c:1314 [inline]
 __se_sys_sendfile64 fs/read_write.c:1300 [inline]
 __x64_sys_sendfile64+0x1cc/0x210 fs/read_write.c:1300
 do_syscall_x64 arch/x86/entry/common.c:50 [inline]
 do_syscall_64+0x35/0xb0 arch/x86/entry/common.c:80
 entry_SYSCALL_64_after_hwframe+0x44/0xae
RIP: 0033:0x7f215cb69969
Code: 28 00 00 00 75 05 48 83 c4 28 c3 e8 e1 14 00 00 90 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 c7 c1 c0 ff ff ff f7 d8 64 89 01 48
RSP: 002b:00007ffc96bb3868 EFLAGS: 00000246 ORIG_RAX: 0000000000000028
RAX: ffffffffffffffda RBX: 00007f215cbad072 RCX: 00007f215cb69969
RDX: 0000000000000000 RSI: 0000000000000004 RDI: 0000000000000005
RBP: 0000000000000000 R08: 00007ffc96bb3a08 R09: 00007ffc96bb3a08
R10: 0000000100000002 R11: 0000000000000246 R12: 00007ffc96bb387c
R13: 431bde82d7b634db R14: 0000000000000000 R15: 0000000000000000

the problem originates from uncorrect lock annotation in the mptcp
code and is only visible since commit 2dcb96bacce3 ("net: core: Correct
the sock::sk_lock.owned lockdep annotations"), but is present since
the port-based endpoint support initial implementation.

This patch addresses the issue introducing a nested variant of
lock_sock_fast() and using it in the relevant code path.

Fixes: 1729cf186d8a ("mptcp: create the listening socket for new port")
Fixes: 2dcb96bacce3 ("net: core: Correct the sock::sk_lock.owned lockdep annotations")
Suggested-by: Thomas Gleixner <[email protected]>
Reported-and-tested-by: [email protected]
Signed-off-by: Paolo Abeni <[email protected]>
Reviewed-by: Thomas Gleixner <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

octeontx2-pf: Add XDP support to netdev PF

Adds XDP_PASS, XDP_TX, XDP_DROP and XDP_REDIRECT support
for netdev PF.

Signed-off-by: Geetha sowjanya <[email protected]>
Signed-off-by: Sunil Goutham <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

octeontx2-af: Adjust LA pointer for cpt parse header

In case of ltype NPC_LT_LA_CPT_HDR, LA pointer is pointing to the
start of cpt parse header. Since cpt parse header has veriable
length padding, this will be a problem for DMAC extraction. Adding
KPU profile changes to adjust the LA pointer to start at ether header
in case of cpt parse header by
   - Adding ptr advance in pkind 58 to a fixed value 40
   - Adding variable length offset 7 and mask 7 (pad len in
     CPT_PARSE_HDR).
Also added the missing static declaration for npc_set_var_len_offset_pkind
function.

Signed-off-by: Kiran Kumar K <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

libbpf: Fix skel_internal.h to set errno on loader retval < 0

When the loader indicates an internal error (result of a checked bpf
system call), it returns the result in attr.test.retval. However, tests
that rely on ASSERT_OK_PTR on NULL (returned from light skeleton) may
miss that NULL denotes an error if errno is set to 0. This would result
in skel pointer being NULL, while ASSERT_OK_PTR returning 1, leading to
a SEGV on dereference of skel, because libbpf_get_error relies on the
assumption that errno is always set in case of error for ptr == NULL.

In particular, this was observed for the ksyms_module test. When
executed using `./test_progs -t ksyms`, prior tests manipulated errno
and the test didn't crash when it failed at ksyms_module load, while
using `./test_progs -t ksyms_module` crashed due to errno being
untouched.

Fixes: 67234743736a (libbpf: Generate loader program out of BPF ELF file.)
Signed-off-by: Kumar Kartikeya Dwivedi <[email protected]>
Signed-off-by: Alexei Starovoitov <[email protected]>
Link: https://lore.kernel.org/bpf/[email protected]

net_sched: Use struct_size() and flex_array_size() helpers

Make use of the struct_size() and flex_array_size() helpers instead of
an open-coded version, in order to avoid any potential type mistakes
or integer overflows that, in the worse scenario, could lead to heap
overflows.

Link: https://github.com/KSPP/linux/issues/160
Signed-off-by: Gustavo A. R. Silva <[email protected]>
Link: https://lore.kernel.org/r/20210928193107.GA262595@embeddedor
Signed-off-by: Jakub Kicinski <[email protected]>

libbpf: Properly ignore STT_SECTION symbols in legacy map definitions

The previous patch to ignore STT_SECTION symbols only added the ignore
condition in one of them. This fails if there's more than one map
definition in the 'maps' section, because the subsequent modulus check will
fail, resulting in error messages like:

libbpf: elf: unable to determine legacy map definition size in ./xdpdump_xdp.o

Fix this by also ignoring STT_SECTION in the first loop.

Fixes: c3e8c44a9063 ("libbpf: Ignore STT_SECTION symbols in 'maps' section")
Signed-off-by: Toke Høiland-Jørgensen <[email protected]>
Signed-off-by: Andrii Nakryiko <[email protected]>
Link: https://lore.kernel.org/bpf/[email protected]

bpf: Do not invoke the XDP dispatcher for PROG_RUN with single repeat

We have a unit test that invokes an XDP program with 1m different
inputs, aka 1m BPF_PROG_RUN syscalls. We run this test concurrently
with slight variations in how we generated the input.

Since commit f23c4b3924d2 ("bpf: Start using the BPF dispatcher in BPF_TEST_RUN")
the unit test has slowed down significantly. Digging deeper reveals that
the concurrent tests are serialised in the kernel on the XDP dispatcher.
This is a global resource that is protected by a mutex, on which we contend.

Fix this by not calling into the XDP dispatcher if we only want to perform
a single run of the BPF program.

See: https://lore.kernel.org/bpf/CACAyw9_y4QumOW35qpgTbLsJ532uGq-kVW-VESJzGyiZkypnvw@mail.gmail.com/

Signed-off-by: Lorenz Bauer <[email protected]>
Signed-off-by: Alexei Starovoitov <[email protected]>
Link: https://lore.kernel.org/bpf/[email protected]

Bluetooth: hci_vhci: Add force_prevent_wake entry

This adds force_prevent_wake which can be used to force a certain state
while interacting with force_suspend.

Signed-off-by: Luiz Augusto von Dentz <[email protected]>
Signed-off-by: Marcel Holtmann <[email protected]>