platform/kernel/linux-rpi.git
2 years agoRevert "e1000e: Fix possible HW unit hang after an s0ix exit"
Sasha Neftin [Mon, 9 May 2022 08:52:54 +0000 (11:52 +0300)]
Revert "e1000e: Fix possible HW unit hang after an s0ix exit"

This reverts commit 1866aa0d0d6492bc2f8d22d0df49abaccf50cddd.

Commit 1866aa0d0d64 ("e1000e: Fix possible HW unit hang after an s0ix
exit") was a workaround for CSME problem to handle messages comes via H2ME
mailbox. This problem has been fixed by patch "e1000e: Enable the GPT
clock before sending message to the CSME".

Fixes: 3e55d231716e ("e1000e: Add handshake with the CSME to support S0ix")
Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=214821
Signed-off-by: Sasha Neftin <sasha.neftin@intel.com>
Tested-by: Naama Meir <naamax.meir@linux.intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
2 years agoe1000e: Enable GPT clock before sending message to CSME
Sasha Neftin [Sun, 8 May 2022 07:09:05 +0000 (10:09 +0300)]
e1000e: Enable GPT clock before sending message to CSME

On corporate (CSME) ADL systems, the Ethernet Controller may stop working
("HW unit hang") after exiting from the s0ix state. The reason is that
CSME misses the message sent by the host. Enabling the dynamic GPT clock
solves this problem. This clock is cleared upon HW initialization.

Fixes: 3e55d231716e ("e1000e: Add handshake with the CSME to support S0ix")
Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=214821
Reviewed-by: Dima Ruinskiy <dima.ruinskiy@intel.com>
Signed-off-by: Sasha Neftin <sasha.neftin@intel.com>
Tested-by: Chia-Lin Kao (AceLan) <acelan.kao@canonical.com>
Tested-by: Naama Meir <naamax.meir@linux.intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
2 years agoselftests/net: test nexthop without gw
Nicolas Dichtel [Wed, 13 Jul 2022 11:48:53 +0000 (13:48 +0200)]
selftests/net: test nexthop without gw

This test implement the scenario described in the commit
"ip: fix dflt addr selection for connected nexthop".
The test configures a nexthop object with an output device only (no gateway
address) and a route that uses this nexthop. The goal is to check if the
kernel selects a valid source address.

Link: https://lore.kernel.org/netdev/20220712095545.10947-1-nicolas.dichtel@6wind.com/
Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Link: https://lore.kernel.org/r/20220713114853.29406-2-nicolas.dichtel@6wind.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2 years agoip: fix dflt addr selection for connected nexthop
Nicolas Dichtel [Wed, 13 Jul 2022 11:48:52 +0000 (13:48 +0200)]
ip: fix dflt addr selection for connected nexthop

When a nexthop is added, without a gw address, the default scope was set
to 'host'. Thus, when a source address is selected, 127.0.0.1 may be chosen
but rejected when the route is used.

When using a route without a nexthop id, the scope can be configured in the
route, thus the problem doesn't exist.

To explain more deeply: when a user creates a nexthop, it cannot specify
the scope. To create it, the function nh_create_ipv4() calls fib_check_nh()
with scope set to 0. fib_check_nh() calls fib_check_nh_nongw() wich was
setting scope to 'host'. Then, nh_create_ipv4() calls
fib_info_update_nhc_saddr() with scope set to 'host'. The src addr is
chosen before the route is inserted.

When a 'standard' route (ie without a reference to a nexthop) is added,
fib_create_info() calls fib_info_update_nhc_saddr() with the scope set by
the user. iproute2 set the scope to 'link' by default.

Here is a way to reproduce the problem:
ip netns add foo
ip -n foo link set lo up
ip netns add bar
ip -n bar link set lo up
sleep 1

ip -n foo link add name eth0 type dummy
ip -n foo link set eth0 up
ip -n foo address add 192.168.0.1/24 dev eth0

ip -n foo link add name veth0 type veth peer name veth1 netns bar
ip -n foo link set veth0 up
ip -n bar link set veth1 up

ip -n bar address add 192.168.1.1/32 dev veth1
ip -n bar route add default dev veth1

ip -n foo nexthop add id 1 dev veth0
ip -n foo route add 192.168.1.1 nhid 1

Try to get/use the route:
> $ ip -n foo route get 192.168.1.1
> RTNETLINK answers: Invalid argument
> $ ip netns exec foo ping -c1 192.168.1.1
> ping: connect: Invalid argument

Try without nexthop group (iproute2 sets scope to 'link' by dflt):
ip -n foo route del 192.168.1.1
ip -n foo route add 192.168.1.1 dev veth0

Try to get/use the route:
> $ ip -n foo route get 192.168.1.1
> 192.168.1.1 dev veth0 src 192.168.0.1 uid 0
>     cache
> $ ip netns exec foo ping -c1 192.168.1.1
> PING 192.168.1.1 (192.168.1.1) 56(84) bytes of data.
> 64 bytes from 192.168.1.1: icmp_seq=1 ttl=64 time=0.039 ms
>
> --- 192.168.1.1 ping statistics ---
> 1 packets transmitted, 1 received, 0% packet loss, time 0ms
> rtt min/avg/max/mdev = 0.039/0.039/0.039/0.000 ms

CC: stable@vger.kernel.org
Fixes: 597cfe4fc339 ("nexthop: Add support for IPv4 nexthops")
Reported-by: Edwin Brossette <edwin.brossette@6wind.com>
Signed-off-by: Nicolas Dichtel <nicolas.dichtel@6wind.com>
Link: https://lore.kernel.org/r/20220713114853.29406-1-nicolas.dichtel@6wind.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2 years agonet: atlantic: remove aq_nic_deinit() when resume
Chia-Lin Kao (AceLan) [Wed, 13 Jul 2022 11:12:24 +0000 (19:12 +0800)]
net: atlantic: remove aq_nic_deinit() when resume

aq_nic_deinit() has been called while suspending, so we don't have to call
it again on resume.
Actually, call it again leads to another hang issue when resuming from
S3.

Jul 8 03:09:44 u-Precision-7865-Tower kernel: [ 5910.992345] Call Trace:
Jul 8 03:09:44 u-Precision-7865-Tower kernel: [ 5910.992346] <TASK>
Jul 8 03:09:44 u-Precision-7865-Tower kernel: [ 5910.992348] aq_nic_deinit+0xb4/0xd0 [atlantic]
Jul 8 03:09:44 u-Precision-7865-Tower kernel: [ 5910.992356] aq_pm_thaw+0x7f/0x100 [atlantic]
Jul 8 03:09:44 u-Precision-7865-Tower kernel: [ 5910.992362] pci_pm_resume+0x5c/0x90
Jul 8 03:09:44 u-Precision-7865-Tower kernel: [ 5910.992366] ? pci_pm_thaw+0x80/0x80
Jul 8 03:09:44 u-Precision-7865-Tower kernel: [ 5910.992368] dpm_run_callback+0x4e/0x120
Jul 8 03:09:44 u-Precision-7865-Tower kernel: [ 5910.992371] device_resume+0xad/0x200
Jul 8 03:09:44 u-Precision-7865-Tower kernel: [ 5910.992373] async_resume+0x1e/0x40
Jul 8 03:09:44 u-Precision-7865-Tower kernel: [ 5910.992374] async_run_entry_fn+0x33/0x120
Jul 8 03:09:44 u-Precision-7865-Tower kernel: [ 5910.992377] process_one_work+0x220/0x3c0
Jul 8 03:09:44 u-Precision-7865-Tower kernel: [ 5910.992380] worker_thread+0x4d/0x3f0
Jul 8 03:09:44 u-Precision-7865-Tower kernel: [ 5910.992382] ? process_one_work+0x3c0/0x3c0
Jul 8 03:09:44 u-Precision-7865-Tower kernel: [ 5910.992384] kthread+0x12a/0x150
Jul 8 03:09:44 u-Precision-7865-Tower kernel: [ 5910.992386] ? set_kthread_struct+0x40/0x40
Jul 8 03:09:44 u-Precision-7865-Tower kernel: [ 5910.992387] ret_from_fork+0x22/0x30
Jul 8 03:09:44 u-Precision-7865-Tower kernel: [ 5910.992391] </TASK>
Jul 8 03:09:44 u-Precision-7865-Tower kernel: [ 5910.992392] ---[ end trace 1ec8c79604ed5e0d ]---
Jul 8 03:09:44 u-Precision-7865-Tower kernel: [ 5910.992394] PM: dpm_run_callback(): pci_pm_resume+0x0/0x90 returns -110
Jul 8 03:09:44 u-Precision-7865-Tower kernel: [ 5910.992397] atlantic 0000:02:00.0: PM: failed to resume async: error -110

Fixes: 1809c30b6e5a ("net: atlantic: always deep reset on pm op, fixing up my null deref regression")
Signed-off-by: Chia-Lin Kao (AceLan) <acelan.kao@canonical.com>
Link: https://lore.kernel.org/r/20220713111224.1535938-2-acelan.kao@canonical.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2 years agonet: atlantic: remove deep parameter on suspend/resume functions
Chia-Lin Kao (AceLan) [Wed, 13 Jul 2022 11:12:23 +0000 (19:12 +0800)]
net: atlantic: remove deep parameter on suspend/resume functions

Below commit claims that atlantic NIC requires to reset the device on pm
op, and had set the deep to true for all suspend/resume functions.
commit 1809c30b6e5a ("net: atlantic: always deep reset on pm op, fixing up my null deref regression")
So, we could remove deep parameter on suspend/resume functions without
any functional change.

Fixes: 1809c30b6e5a ("net: atlantic: always deep reset on pm op, fixing up my null deref regression")
Signed-off-by: Chia-Lin Kao (AceLan) <acelan.kao@canonical.com>
Link: https://lore.kernel.org/r/20220713111224.1535938-1-acelan.kao@canonical.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2 years agosfc: fix kernel panic when creating VF
Íñigo Huguet [Wed, 13 Jul 2022 09:21:16 +0000 (11:21 +0200)]
sfc: fix kernel panic when creating VF

When creating VFs a kernel panic can happen when calling to
efx_ef10_try_update_nic_stats_vf.

When releasing a DMA coherent buffer, sometimes, I don't know in what
specific circumstances, it has to unmap memory with vunmap. It is
disallowed to do that in IRQ context or with BH disabled. Otherwise, we
hit this line in vunmap, causing the crash:
  BUG_ON(in_interrupt());

This patch reenables BH to release the buffer.

Log messages when the bug is hit:
 kernel BUG at mm/vmalloc.c:2727!
 invalid opcode: 0000 [#1] PREEMPT SMP NOPTI
 CPU: 6 PID: 1462 Comm: NetworkManager Kdump: loaded Tainted: G          I      --------- ---  5.14.0-119.el9.x86_64 #1
 Hardware name: Dell Inc. PowerEdge R740/06WXJT, BIOS 2.8.2 08/27/2020
 RIP: 0010:vunmap+0x2e/0x30
 ...skip...
 Call Trace:
  __iommu_dma_free+0x96/0x100
  efx_nic_free_buffer+0x2b/0x40 [sfc]
  efx_ef10_try_update_nic_stats_vf+0x14a/0x1c0 [sfc]
  efx_ef10_update_stats_vf+0x18/0x40 [sfc]
  efx_start_all+0x15e/0x1d0 [sfc]
  efx_net_open+0x5a/0xe0 [sfc]
  __dev_open+0xe7/0x1a0
  __dev_change_flags+0x1d7/0x240
  dev_change_flags+0x21/0x60
  ...skip...

Fixes: d778819609a2 ("sfc: DMA the VF stats only when requested")
Reported-by: Ma Yuying <yuma@redhat.com>
Signed-off-by: Íñigo Huguet <ihuguet@redhat.com>
Acked-by: Edward Cree <ecree.xilinx@gmail.com>
Link: https://lore.kernel.org/r/20220713092116.21238-1-ihuguet@redhat.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2 years agoMerge branch 'seg6-fix-skb-checksum-for-srh-encapsulation-insertion'
Paolo Abeni [Thu, 14 Jul 2022 08:15:15 +0000 (10:15 +0200)]
Merge branch 'seg6-fix-skb-checksum-for-srh-encapsulation-insertion'

Andrea Mayer says:

====================
seg6: fix skb checksum for SRH encapsulation/insertion

The Linux kernel supports Segment Routing Header (SRH)
encapsulation/insertion operations by providing the capability to: i)
encapsulate a packet in an outer IPv6 header with a specified SRH; ii)
insert a specified SRH directly after the IPv6 header of the packet.
Note that the insertion operation is also referred to as 'injection'.

The two operations are respectively supported by seg6_do_srh_encap() and
seg6_do_srh_inline(), which operate on the skb associated to the packet as
needed (e.g. adding the necessary headers and initializing them, while
taking care to recalculate the skb checksum).

seg6_do_srh_encap() and seg6_do_srh_inline() do not initialize the payload
length of the IPv6 header, which is carried out by the caller functions.
However, this approach causes the corruption of the skb checksum which
needs to be updated only after initialization of headers is completed
(thanks to Paolo Abeni for detecting this issue).

The patchset fixes the skb checksum corruption by moving the IPv6 header
payload length initialization from the callers of seg6_do_srh_encap() and
seg6_do_srh_inline() directly into these functions.

This patchset is organized as follows:
 - patch 1/3, seg6: fix skb checksum evaluation in SRH
   encapsulation/insertion;
    (* SRH encapsulation/insertion available since v4.10)

 - patch 2/3, seg6: fix skb checksum in SRv6 End.B6 and End.B6.Encaps
   behaviors;
    (* SRv6 End.B6 and End.B6.Encaps behaviors available since v4.14)

 - patch 3/3, seg6: bpf: fix skb checksum in bpf_push_seg6_encap();
    (* bpf IPv6 Segment Routing helpers available since v4.18)

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

Link: https://lore.kernel.org/r/20220712175837.16267-1-andrea.mayer@uniroma2.it
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2 years agoseg6: bpf: fix skb checksum in bpf_push_seg6_encap()
Andrea Mayer [Tue, 12 Jul 2022 17:58:37 +0000 (19:58 +0200)]
seg6: bpf: fix skb checksum in bpf_push_seg6_encap()

Both helper functions bpf_lwt_seg6_action() and bpf_lwt_push_encap() use
the bpf_push_seg6_encap() to encapsulate the packet in an IPv6 with Segment
Routing Header (SRH) or insert an SRH between the IPv6 header and the
payload.
To achieve this result, such helper functions rely on bpf_push_seg6_encap()
which, in turn, leverages seg6_do_srh_{encap,inline}() to perform the
required operation (i.e. encap/inline).

This patch removes the initialization of the IPv6 header payload length
from bpf_push_seg6_encap(), as it is now handled properly by
seg6_do_srh_{encap,inline}() to prevent corruption of the skb checksum.

Fixes: fe94cc290f53 ("bpf: Add IPv6 Segment Routing helpers")
Signed-off-by: Andrea Mayer <andrea.mayer@uniroma2.it>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2 years agoseg6: fix skb checksum in SRv6 End.B6 and End.B6.Encaps behaviors
Andrea Mayer [Tue, 12 Jul 2022 17:58:36 +0000 (19:58 +0200)]
seg6: fix skb checksum in SRv6 End.B6 and End.B6.Encaps behaviors

The SRv6 End.B6 and End.B6.Encaps behaviors rely on functions
seg6_do_srh_{encap,inline}() to, respectively: i) encapsulate the
packet within an outer IPv6 header with the specified Segment Routing
Header (SRH); ii) insert the specified SRH directly after the IPv6
header of the packet.

This patch removes the initialization of the IPv6 header payload length
from the input_action_end_b6{_encap}() functions, as it is now handled
properly by seg6_do_srh_{encap,inline}() to avoid corruption of the skb
checksum.

Fixes: 140f04c33bbc ("ipv6: sr: implement several seg6local actions")
Signed-off-by: Andrea Mayer <andrea.mayer@uniroma2.it>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2 years agoseg6: fix skb checksum evaluation in SRH encapsulation/insertion
Andrea Mayer [Tue, 12 Jul 2022 17:58:35 +0000 (19:58 +0200)]
seg6: fix skb checksum evaluation in SRH encapsulation/insertion

Support for SRH encapsulation and insertion was introduced with
commit 6c8702c60b88 ("ipv6: sr: add support for SRH encapsulation and
injection with lwtunnels"), through the seg6_do_srh_encap() and
seg6_do_srh_inline() functions, respectively.
The former encapsulates the packet in an outer IPv6 header along with
the SRH, while the latter inserts the SRH between the IPv6 header and
the payload. Then, the headers are initialized/updated according to the
operating mode (i.e., encap/inline).
Finally, the skb checksum is calculated to reflect the changes applied
to the headers.

The IPv6 payload length ('payload_len') is not initialized
within seg6_do_srh_{inline,encap}() but is deferred in seg6_do_srh(), i.e.
the caller of seg6_do_srh_{inline,encap}().
However, this operation invalidates the skb checksum, since the
'payload_len' is updated only after the checksum is evaluated.

To solve this issue, the initialization of the IPv6 payload length is
moved from seg6_do_srh() directly into the seg6_do_srh_{inline,encap}()
functions and before the skb checksum update takes place.

Fixes: 6c8702c60b88 ("ipv6: sr: add support for SRH encapsulation and injection with lwtunnels")
Reported-by: Paolo Abeni <pabeni@redhat.com>
Link: https://lore.kernel.org/all/20220705190727.69d532417be7438b15404ee1@uniroma2.it
Signed-off-by: Andrea Mayer <andrea.mayer@uniroma2.it>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2 years agoMerge branch '100GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/net...
Jakub Kicinski [Thu, 14 Jul 2022 03:16:03 +0000 (20:16 -0700)]
Merge branch '100GbE' of git://git./linux/kernel/git/tnguy/net-queue

Tony Nguyen says:

====================
Intel Wired LAN Driver Updates 2022-07-12

This series contains updates to ice driver only.

Paul fixes detection of E822 devices for firmware update and changes NVM
read for snapshot creation to be done in chunks as some systems cannot
read the entire NVM in the allotted time.
====================

Link: https://lore.kernel.org/r/20220712164829.7275-1-anthony.l.nguyen@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agosfc: fix use after free when disabling sriov
Íñigo Huguet [Tue, 12 Jul 2022 06:26:42 +0000 (08:26 +0200)]
sfc: fix use after free when disabling sriov

Use after free is detected by kfence when disabling sriov. What was read
after being freed was vf->pci_dev: it was freed from pci_disable_sriov
and later read in efx_ef10_sriov_free_vf_vports, called from
efx_ef10_sriov_free_vf_vswitching.

Set the pointer to NULL at release time to not trying to read it later.

Reproducer and dmesg log (note that kfence doesn't detect it every time):
$ echo 1 > /sys/class/net/enp65s0f0np0/device/sriov_numvfs
$ echo 0 > /sys/class/net/enp65s0f0np0/device/sriov_numvfs

 BUG: KFENCE: use-after-free read in efx_ef10_sriov_free_vf_vswitching+0x82/0x170 [sfc]

 Use-after-free read at 0x00000000ff3c1ba5 (in kfence-#224):
  efx_ef10_sriov_free_vf_vswitching+0x82/0x170 [sfc]
  efx_ef10_pci_sriov_disable+0x38/0x70 [sfc]
  efx_pci_sriov_configure+0x24/0x40 [sfc]
  sriov_numvfs_store+0xfe/0x140
  kernfs_fop_write_iter+0x11c/0x1b0
  new_sync_write+0x11f/0x1b0
  vfs_write+0x1eb/0x280
  ksys_write+0x5f/0xe0
  do_syscall_64+0x5c/0x80
  entry_SYSCALL_64_after_hwframe+0x44/0xae

 kfence-#224: 0x00000000edb8ef95-0x00000000671f5ce1, size=2792, cache=kmalloc-4k

 allocated by task 6771 on cpu 10 at 3137.860196s:
  pci_alloc_dev+0x21/0x60
  pci_iov_add_virtfn+0x2a2/0x320
  sriov_enable+0x212/0x3e0
  efx_ef10_sriov_configure+0x67/0x80 [sfc]
  efx_pci_sriov_configure+0x24/0x40 [sfc]
  sriov_numvfs_store+0xba/0x140
  kernfs_fop_write_iter+0x11c/0x1b0
  new_sync_write+0x11f/0x1b0
  vfs_write+0x1eb/0x280
  ksys_write+0x5f/0xe0
  do_syscall_64+0x5c/0x80
  entry_SYSCALL_64_after_hwframe+0x44/0xae

 freed by task 6771 on cpu 12 at 3170.991309s:
  device_release+0x34/0x90
  kobject_cleanup+0x3a/0x130
  pci_iov_remove_virtfn+0xd9/0x120
  sriov_disable+0x30/0xe0
  efx_ef10_pci_sriov_disable+0x57/0x70 [sfc]
  efx_pci_sriov_configure+0x24/0x40 [sfc]
  sriov_numvfs_store+0xfe/0x140
  kernfs_fop_write_iter+0x11c/0x1b0
  new_sync_write+0x11f/0x1b0
  vfs_write+0x1eb/0x280
  ksys_write+0x5f/0xe0
  do_syscall_64+0x5c/0x80
  entry_SYSCALL_64_after_hwframe+0x44/0xae

Fixes: 3c5eb87605e85 ("sfc: create vports for VFs and assign random MAC addresses")
Reported-by: Yanghang Liu <yanghliu@redhat.com>
Signed-off-by: Íñigo Huguet <ihuguet@redhat.com>
Acked-by: Martin Habets <habetsm.xilinx@gmail.com>
Link: https://lore.kernel.org/r/20220712062642.6915-1-ihuguet@redhat.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agonet: sunhme: output link status with a single print.
Nick Bowler [Wed, 13 Jul 2022 01:58:35 +0000 (21:58 -0400)]
net: sunhme: output link status with a single print.

This driver currently prints the link status using four separate
printk calls, which these days gets presented to the user as four
distinct messages, not exactly ideal:

  [   32.582778] eth0: Link is up using
  [   32.582828] internal
  [   32.582837] transceiver at
  [   32.582888] 100Mb/s, Full Duplex.

Restructure the display_link_mode function to use a single netdev_info
call to present all this information as a single message, which is much
nicer:

  [   33.640143] hme 0000:00:01.1 eth0: Link is up using internal transceiver at 100Mb/s, Full Duplex.

The display_forced_link_mode function has a similar structure, so adjust
it in a similar fashion.

Signed-off-by: Nick Bowler <nbowler@draconx.ca>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agor8152: fix accessing unset transport header
Hayes Wang [Wed, 13 Jul 2022 03:31:11 +0000 (11:31 +0800)]
r8152: fix accessing unset transport header

A warning is triggered by commit 66e4c8d95008 ("net: warn if transport
header was not set"). The warning is harmless, because the value from
skb_transport_offset() is only used for skb_is_gso() is true or the
skb->ip_summed is equal to CHECKSUM_PARTIAL.

Fixes: 66e4c8d95008 ("net: warn if transport header was not set")
Signed-off-by: Hayes Wang <hayeswang@realtek.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoMerge tag 'wireless-2022-07-13' of git://git.kernel.org/pub/scm/linux/kernel/git...
David S. Miller [Wed, 13 Jul 2022 13:27:38 +0000 (14:27 +0100)]
Merge tag 'wireless-2022-07-13' of git://git./linux/kernel/git/wireless/wireless

Johannes Berg says:

====================
A small set of fixes for
 * queue selection in mesh/ocb
 * queue handling on interface stop
 * hwsim virtio device vs. some other virtio changes
 * dt-bindings email addresses
 * color collision memory allocation
 * a const variable in rtw88
 * shared SKB transmit in the ethernet format path
 * P2P client port authorization
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: stmmac: fix leaks in probe
Dan Carpenter [Tue, 12 Jul 2022 14:42:25 +0000 (17:42 +0300)]
net: stmmac: fix leaks in probe

These two error paths should clean up before returning.

Fixes: 2bb4b98b60d7 ("net: stmmac: Add Ingenic SoCs MAC support.")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: ftgmac100: Hold reference returned by of_get_child_by_name()
Liang He [Tue, 12 Jul 2022 06:14:17 +0000 (14:14 +0800)]
net: ftgmac100: Hold reference returned by of_get_child_by_name()

In ftgmac100_probe(), we should hold the refernece returned by
of_get_child_by_name() and use it to call of_node_put() for
reference balance.

Fixes: 39bfab8844a0 ("net: ftgmac100: Add support for DT phy-handle property")
Signed-off-by: Liang He <windhl@126.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoMerge branch 'net-sysctl-races'
David S. Miller [Wed, 13 Jul 2022 11:56:50 +0000 (12:56 +0100)]
Merge branch 'net-sysctl-races'

Kuniyuki Iwashima says:

====================
sysctl: Fix data-races around ipv4_net_table (Roun).

This series fixes data-races around the first 13 knobs and
nexthop_compat_mode in ipv4_net_table.

I will post another patch for three early_demux knobs later,
so the next round will start from ip_default_ttl.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonexthop: Fix data-races around nexthop_compat_mode.
Kuniyuki Iwashima [Tue, 12 Jul 2022 00:15:33 +0000 (17:15 -0700)]
nexthop: Fix data-races around nexthop_compat_mode.

While reading nexthop_compat_mode, it can be changed concurrently.
Thus, we need to add READ_ONCE() to its readers.

Fixes: 4f80116d3df3 ("net: ipv4: add sysctl for nexthop api compatibility mode")
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoipv4: Fix data-races around sysctl_ip_dynaddr.
Kuniyuki Iwashima [Tue, 12 Jul 2022 00:15:32 +0000 (17:15 -0700)]
ipv4: Fix data-races around sysctl_ip_dynaddr.

While reading sysctl_ip_dynaddr, it can be changed concurrently.
Thus, we need to add READ_ONCE() to its readers.

Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agotcp: Fix a data-race around sysctl_tcp_ecn_fallback.
Kuniyuki Iwashima [Tue, 12 Jul 2022 00:15:31 +0000 (17:15 -0700)]
tcp: Fix a data-race around sysctl_tcp_ecn_fallback.

While reading sysctl_tcp_ecn_fallback, it can be changed concurrently.
Thus, we need to add READ_ONCE() to its reader.

Fixes: 492135557dc0 ("tcp: add rfc3168, section 6.1.1.1. fallback")
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agotcp: Fix data-races around sysctl_tcp_ecn.
Kuniyuki Iwashima [Tue, 12 Jul 2022 00:15:30 +0000 (17:15 -0700)]
tcp: Fix data-races around sysctl_tcp_ecn.

While reading sysctl_tcp_ecn, it can be changed concurrently.
Thus, we need to add READ_ONCE() to its readers.

Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoraw: Fix a data-race around sysctl_raw_l3mdev_accept.
Kuniyuki Iwashima [Tue, 12 Jul 2022 00:15:29 +0000 (17:15 -0700)]
raw: Fix a data-race around sysctl_raw_l3mdev_accept.

While reading sysctl_raw_l3mdev_accept, it can be changed concurrently.
Thus, we need to add READ_ONCE() to its reader.

Fixes: 6897445fb194 ("net: provide a sysctl raw_l3mdev_accept for raw socket lookup with VRFs")
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoicmp: Fix a data-race around sysctl_icmp_ratemask.
Kuniyuki Iwashima [Tue, 12 Jul 2022 00:15:28 +0000 (17:15 -0700)]
icmp: Fix a data-race around sysctl_icmp_ratemask.

While reading sysctl_icmp_ratemask, it can be changed concurrently.
Thus, we need to add READ_ONCE() to its reader.

Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoicmp: Fix a data-race around sysctl_icmp_ratelimit.
Kuniyuki Iwashima [Tue, 12 Jul 2022 00:15:27 +0000 (17:15 -0700)]
icmp: Fix a data-race around sysctl_icmp_ratelimit.

While reading sysctl_icmp_ratelimit, it can be changed concurrently.
Thus, we need to add READ_ONCE() to its reader.

Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoicmp: Fix a data-race around sysctl_icmp_errors_use_inbound_ifaddr.
Kuniyuki Iwashima [Tue, 12 Jul 2022 00:15:26 +0000 (17:15 -0700)]
icmp: Fix a data-race around sysctl_icmp_errors_use_inbound_ifaddr.

While reading sysctl_icmp_errors_use_inbound_ifaddr, it can be changed
concurrently.  Thus, we need to add READ_ONCE() to its reader.

Fixes: 1c2fb7f93cb2 ("[IPV4]: Sysctl configurable icmp error source address.")
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoicmp: Fix a data-race around sysctl_icmp_ignore_bogus_error_responses.
Kuniyuki Iwashima [Tue, 12 Jul 2022 00:15:25 +0000 (17:15 -0700)]
icmp: Fix a data-race around sysctl_icmp_ignore_bogus_error_responses.

While reading sysctl_icmp_ignore_bogus_error_responses, it can be changed
concurrently.  Thus, we need to add READ_ONCE() to its reader.

Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoicmp: Fix a data-race around sysctl_icmp_echo_ignore_broadcasts.
Kuniyuki Iwashima [Tue, 12 Jul 2022 00:15:24 +0000 (17:15 -0700)]
icmp: Fix a data-race around sysctl_icmp_echo_ignore_broadcasts.

While reading sysctl_icmp_echo_ignore_broadcasts, it can be changed
concurrently.  Thus, we need to add READ_ONCE() to its reader.

Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoicmp: Fix data-races around sysctl_icmp_echo_enable_probe.
Kuniyuki Iwashima [Tue, 12 Jul 2022 00:15:23 +0000 (17:15 -0700)]
icmp: Fix data-races around sysctl_icmp_echo_enable_probe.

While reading sysctl_icmp_echo_enable_probe, it can be changed
concurrently.  Thus, we need to add READ_ONCE() to its readers.

Fixes: d329ea5bd884 ("icmp: add response to RFC 8335 PROBE messages")
Fixes: 1fd07f33c3ea ("ipv6: ICMPV6: add response to ICMPV6 RFC 8335 PROBE messages")
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoicmp: Fix a data-race around sysctl_icmp_echo_ignore_all.
Kuniyuki Iwashima [Tue, 12 Jul 2022 00:15:22 +0000 (17:15 -0700)]
icmp: Fix a data-race around sysctl_icmp_echo_ignore_all.

While reading sysctl_icmp_echo_ignore_all, it can be changed concurrently.
Thus, we need to add READ_ONCE() to its reader.

Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agotcp: Fix a data-race around sysctl_max_tw_buckets.
Kuniyuki Iwashima [Tue, 12 Jul 2022 00:15:21 +0000 (17:15 -0700)]
tcp: Fix a data-race around sysctl_max_tw_buckets.

While reading sysctl_max_tw_buckets, it can be changed concurrently.
Thus, we need to add READ_ONCE() to its reader.

Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agosysctl: Fix data-races in proc_dointvec_ms_jiffies().
Kuniyuki Iwashima [Tue, 12 Jul 2022 00:15:20 +0000 (17:15 -0700)]
sysctl: Fix data-races in proc_dointvec_ms_jiffies().

A sysctl variable is accessed concurrently, and there is always a chance
of data-race.  So, all readers and writers need some basic protection to
avoid load/store-tearing.

This patch changes proc_dointvec_ms_jiffies() to use READ_ONCE() and
WRITE_ONCE() internally to fix data-races on the sysctl side.  For now,
proc_dointvec_ms_jiffies() itself is tolerant to a data-race, but we still
need to add annotations on the other subsystem's side.

Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agosysctl: Fix data-races in proc_dou8vec_minmax().
Kuniyuki Iwashima [Tue, 12 Jul 2022 00:15:19 +0000 (17:15 -0700)]
sysctl: Fix data-races in proc_dou8vec_minmax().

A sysctl variable is accessed concurrently, and there is always a chance
of data-race.  So, all readers and writers need some basic protection to
avoid load/store-tearing.

This patch changes proc_dou8vec_minmax() to use READ_ONCE() and
WRITE_ONCE() internally to fix data-races on the sysctl side.  For now,
proc_dou8vec_minmax() itself is tolerant to a data-race, but we still
need to add annotations on the other subsystem's side.

Fixes: cb9444130662 ("sysctl: add proc_dou8vec_minmax()")
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoMerge branch 'bnxt_en-5-bug-fixes'
Jakub Kicinski [Wed, 13 Jul 2022 03:36:00 +0000 (20:36 -0700)]
Merge branch 'bnxt_en-5-bug-fixes'

Michael Chan says:

====================
bnxt_en: 5 Bug fixes

This patchset fixes various issues, including SRIOV error unwinding,
one error recovery path, live patch reporting, XDP transmit path,
and PHC clock reading.
====================

Link: https://lore.kernel.org/r/1657592778-12730-1-git-send-email-michael.chan@broadcom.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agobnxt_en: Fix bnxt_refclk_read()
Pavan Chebbi [Tue, 12 Jul 2022 02:26:18 +0000 (22:26 -0400)]
bnxt_en: Fix bnxt_refclk_read()

The upper 32-bit PHC register is not latched when reading the lower
32-bit PHC register.  Current code leaves a small window where we may
not read correct higher order bits if the lower order bits are just about
to wrap around.

This patch fixes this by reading higher order bits twice and makes
sure that final value is correctly paired with its lower 32 bits.

Fixes: 30e96f487f64 ("bnxt_en: Do not read the PTP PHC during chip reset")
Cc: Richard Cochran <richardcochran@gmail.com>
Signed-off-by: Pavan Chebbi <pavan.chebbi@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agobnxt_en: Fix and simplify XDP transmit path
Michael Chan [Tue, 12 Jul 2022 02:26:17 +0000 (22:26 -0400)]
bnxt_en: Fix and simplify XDP transmit path

Fix the missing length hint in the TX BD for the XDP transmit path.  The
length hint is required on legacy chips.

Also, simplify the code by eliminating the first_buf local variable.
tx_buf contains the same value.  The opaque value only needs to be set
on the first BD.  Fix this also for correctness.

Fixes: a7559bc8c17c ("bnxt: support transmit and free of aggregation buffers")
Reviewed-by: Andy Gospodarek <andrew.gospodarek@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agobnxt_en: fix livepatch query
Vikas Gupta [Tue, 12 Jul 2022 02:26:16 +0000 (22:26 -0400)]
bnxt_en: fix livepatch query

In the livepatch query fw_target BNXT_FW_SRT_PATCH is
applicable for P5 chips only.

Fixes: 3c4153394e2c ("bnxt_en: implement firmware live patching")
Reviewed-by: Saravanan Vajravel <saravanan.vajravel@broadcom.com>
Reviewed-by: Somnath Kotur <somnath.kotur@broadcom.com>
Signed-off-by: Vikas Gupta <vikas.gupta@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agobnxt_en: Fix bnxt_reinit_after_abort() code path
Michael Chan [Tue, 12 Jul 2022 02:26:15 +0000 (22:26 -0400)]
bnxt_en: Fix bnxt_reinit_after_abort() code path

bnxt_reinit_after_abort() is called during ifup when a previous
FW reset sequence has aborted or a previous ifup has failed after
detecting FW reset.  In all cases, it is safe to assume that a
previous FW reset has completed and the driver may not have fully
reinitialized.

Prior to this patch, it is assumed that the
FUNC_DRV_IF_CHANGE_RESP_FLAGS_HOT_FW_RESET_DONE flag will always be
set by the firmware in bnxt_hwrm_if_change().  This may not be true if
the driver has already attempted to register with the firmware.  The
firmware may not set the RESET_DONE flag again after the driver has
registered, assuming that the driver has seen the flag already.

Fix it to always go through the FW reset initialization path if
the BNXT_STATE_FW_RESET_DET flag is set.  This flag is always set
by the driver after successfully going through bnxt_reinit_after_abort().

Fixes: 6882c36cf82e ("bnxt_en: attempt to reinitialize after aborted reset")
Reviewed-by: Pavan Chebbi <pavan.chebbi@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agobnxt_en: reclaim max resources if sriov enable fails
Kashyap Desai [Tue, 12 Jul 2022 02:26:14 +0000 (22:26 -0400)]
bnxt_en: reclaim max resources if sriov enable fails

If bnxt_sriov_enable() fails after some resources have been reserved
for the VFs, the current code is not unwinding properly and the
reserved resources become unavailable afterwards.  Fix it by
properly unwinding with a call to bnxt_hwrm_func_qcaps() to
reset all maximum resources.

Also, add the missing bnxt_ulp_sriov_cfg() call to let the RDMA
driver know to abort.

Fixes: c0c050c58d84 ("bnxt_en: New Broadcom ethernet driver.")
Signed-off-by: Kashyap Desai <kashyap.desai@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoice: change devlink code to read NVM in blocks
Paul M Stillwell Jr [Wed, 8 Jun 2022 21:48:32 +0000 (14:48 -0700)]
ice: change devlink code to read NVM in blocks

When creating a snapshot of the NVM the driver needs to read the entire
contents from the NVM and store it. The NVM reads are protected by a lock
that is shared between the driver and the firmware.

If the driver takes too long to read the entire NVM (which can happen on
some systems) then the firmware could reclaim the lock and cause subsequent
reads from the driver to fail.

We could fix this by increasing the timeout that we pass to the firmware,
but we could end up in the same situation again if the system is slow.
Instead have the driver break the reading of the NVM into blocks that are
small enough that we have confidence that the read will complete within the
timeout time, but large enough not to cause significant AQ overhead.

Fixes: dce730f17825 ("ice: add a devlink region for dumping NVM contents")
Signed-off-by: Paul M Stillwell Jr <paul.m.stillwell.jr@intel.com>
Tested-by: Gurucharan <gurucharanx.g@intel.com> (A Contingent worker at Intel)
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
2 years agoice: handle E822 generic device ID in PLDM header
Paul M Stillwell Jr [Wed, 8 Jun 2022 21:09:52 +0000 (14:09 -0700)]
ice: handle E822 generic device ID in PLDM header

The driver currently presumes that the record data in the PLDM header
of the firmware image will match the device ID of the running device.
This is true for E810 devices. It appears that for E822 devices that
this is not guaranteed to be true.

Fix this by adding a check for the generic E822 device.

Fixes: d69ea414c9b4 ("ice: implement device flash update via devlink")
Signed-off-by: Paul M Stillwell Jr <paul.m.stillwell.jr@intel.com>
Tested-by: Gurucharan <gurucharanx.g@intel.com> (A Contingent worker at Intel)
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
2 years agonet: marvell: prestera: fix missed deinit sequence
Yevhen Orlov [Sun, 10 Jul 2022 12:20:21 +0000 (15:20 +0300)]
net: marvell: prestera: fix missed deinit sequence

Add unregister_fib_notifier as rollback of register_fib_notifier.

Fixes: 4394fbcb78cf ("net: marvell: prestera: handle fib notifications")
Signed-off-by: Yevhen Orlov <yevhen.orlov@plvision.eu>
Link: https://lore.kernel.org/r/20220710122021.7642-1-yevhen.orlov@plvision.eu
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
2 years agoselftest: net: add tun to .gitignore
Jakub Kicinski [Sat, 9 Jul 2022 02:41:41 +0000 (19:41 -0700)]
selftest: net: add tun to .gitignore

Add missing .gitignore entry.

Fixes: 839b92fede7b ("selftest: tun: add test for NAPI dismantle")
Link: https://lore.kernel.org/r/20220709024141.321683-1-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/netfilter/nf
David S. Miller [Mon, 11 Jul 2022 10:58:38 +0000 (11:58 +0100)]
Merge git://git./linux/kernel/git/netfilter/nf

Pablo Neira Ayuso says:

====================
Netfilter fixes for net

The following patchset contains Netfilter fixes for net:

1) refcount_inc_not_zero() is not semantically equivalent to
   atomic_int_not_zero(), from Florian Westphal. My understanding was
   that refcount_*() API provides a wrapper to easier debugging of
   reference count leaks, however, there are semantic differences
   between these two APIs, where refcount_inc_not_zero() needs a barrier.
   Reason for this subtle difference to me is unknown.

2) packet logging is not correct for ARP and IP packets, from the
   ARP family and netdev/egress respectively. Use skb_network_offset()
   to reach the headers accordingly.

3) set element extension length have been growing over time, replace
   a BUG_ON by EINVAL which might be triggerable from userspace.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoMerge branch 'mptcp-fixes'
David S. Miller [Mon, 11 Jul 2022 10:31:38 +0000 (11:31 +0100)]
Merge branch 'mptcp-fixes'

Mat Martineau says:

====================
mptcp: Disconnect and selftest fixes

Patch 1 switches to a safe list iterator in the MPTCP disconnect code.

Patch 2 adds the userspace_pm.sh selftest script to the MPTCP selftest
Makefile, resolving the netdev/check_selftest CI failure.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoselftests: mptcp: validate userspace PM tests by default
Matthieu Baerts [Fri, 8 Jul 2022 23:36:10 +0000 (16:36 -0700)]
selftests: mptcp: validate userspace PM tests by default

The new script was not listed in the programs to test.

By consequence, some CIs running MPTCP selftests were not validating
these new tests. Note that MPTCP CI was validating it as it executes all
.sh scripts from 'tools/testing/selftests/net/mptcp' directory.

Fixes: 259a834fadda ("selftests: mptcp: functional tests for the userspace PM type")
Reported-by: Jakub Kicinski <kuba@kernel.org>
Reviewed-by: Mat Martineau <mathew.j.martineau@linux.intel.com>
Signed-off-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Signed-off-by: Mat Martineau <mathew.j.martineau@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agomptcp: fix subflow traversal at disconnect time
Paolo Abeni [Fri, 8 Jul 2022 23:36:09 +0000 (16:36 -0700)]
mptcp: fix subflow traversal at disconnect time

At disconnect time the MPTCP protocol traverse the subflows
list closing each of them. In some circumstances - MPJ subflow,
passive MPTCP socket, the latter operation can remove the
subflow from the list, invalidating the current iterator.

Address the issue using the safe list traversing helper
variant.

Reported-by: van fantasy <g1042620637@gmail.com>
Fixes: b29fcfb54cd7 ("mptcp: full disconnect implementation")
Tested-by: Matthieu Baerts <matthieu.baerts@tessares.net>
Reviewed-by: Mat Martineau <mathew.j.martineau@linux.intel.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Mat Martineau <mathew.j.martineau@linux.intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agowifi: mac80211: fix queue selection for mesh/OCB interfaces
Felix Fietkau [Sat, 2 Jul 2022 14:52:27 +0000 (16:52 +0200)]
wifi: mac80211: fix queue selection for mesh/OCB interfaces

When using iTXQ, the code assumes that there is only one vif queue for
broadcast packets, using the BE queue. Allowing non-BE queue marking
violates that assumption and txq->ac == skb_queue_mapping is no longer
guaranteed. This can cause issues with queue handling in the driver and
also causes issues with the recent ATF change, resulting in an AQL
underflow warning.

Cc: stable@vger.kernel.org
Signed-off-by: Felix Fietkau <nbd@nbd.name>
Link: https://lore.kernel.org/r/20220702145227.39356-1-nbd@nbd.name
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
2 years agonetfilter: nf_tables: replace BUG_ON by element length check
Pablo Neira Ayuso [Tue, 5 Jul 2022 09:41:59 +0000 (11:41 +0200)]
netfilter: nf_tables: replace BUG_ON by element length check

BUG_ON can be triggered from userspace with an element with a large
userdata area. Replace it by length check and return EINVAL instead.
Over time extensions have been growing in size.

Pick a sufficiently old Fixes: tag to propagate this fix.

Fixes: 7d7402642eaf ("netfilter: nf_tables: variable sized set element keys / data")
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
2 years agovlan: fix memory leak in vlan_newlink()
Eric Dumazet [Fri, 8 Jul 2022 15:11:53 +0000 (15:11 +0000)]
vlan: fix memory leak in vlan_newlink()

Blamed commit added back a bug I fixed in commit 9bbd917e0bec
("vlan: fix memory leak in vlan_dev_set_egress_priority")

If a memory allocation fails in vlan_changelink() after other allocations
succeeded, we need to call vlan_dev_free_egress_priority()
to free all allocated memory because after a failed ->newlink()
we do not call any methods like ndo_uninit() or dev->priv_destructor().

In following example, if the allocation for last element 2000:2001 fails,
we need to free eight prior allocations:

ip link add link dummy0 dummy0.100 type vlan id 100 \
egress-qos-map 1:2 2:3 3:4 4:5 5:6 6:7 7:8 8:9 2000:2001

syzbot report was:

BUG: memory leak
unreferenced object 0xffff888117bd1060 (size 32):
comm "syz-executor408", pid 3759, jiffies 4294956555 (age 34.090s)
hex dump (first 32 bytes):
09 00 00 00 00 a0 00 00 00 00 00 00 00 00 00 00 ................
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................
backtrace:
[<ffffffff83fc60ad>] kmalloc include/linux/slab.h:600 [inline]
[<ffffffff83fc60ad>] vlan_dev_set_egress_priority+0xed/0x170 net/8021q/vlan_dev.c:193
[<ffffffff83fc6628>] vlan_changelink+0x178/0x1d0 net/8021q/vlan_netlink.c:128
[<ffffffff83fc67c8>] vlan_newlink+0x148/0x260 net/8021q/vlan_netlink.c:185
[<ffffffff838b1278>] rtnl_newlink_create net/core/rtnetlink.c:3363 [inline]
[<ffffffff838b1278>] __rtnl_newlink+0xa58/0xdc0 net/core/rtnetlink.c:3580
[<ffffffff838b1629>] rtnl_newlink+0x49/0x70 net/core/rtnetlink.c:3593
[<ffffffff838ac66c>] rtnetlink_rcv_msg+0x21c/0x5c0 net/core/rtnetlink.c:6089
[<ffffffff839f9c37>] netlink_rcv_skb+0x87/0x1d0 net/netlink/af_netlink.c:2501
[<ffffffff839f8da7>] netlink_unicast_kernel net/netlink/af_netlink.c:1319 [inline]
[<ffffffff839f8da7>] netlink_unicast+0x397/0x4c0 net/netlink/af_netlink.c:1345
[<ffffffff839f9266>] netlink_sendmsg+0x396/0x710 net/netlink/af_netlink.c:1921
[<ffffffff8384dbf6>] sock_sendmsg_nosec net/socket.c:714 [inline]
[<ffffffff8384dbf6>] sock_sendmsg+0x56/0x80 net/socket.c:734
[<ffffffff8384e15c>] ____sys_sendmsg+0x36c/0x390 net/socket.c:2488
[<ffffffff838523cb>] ___sys_sendmsg+0x8b/0xd0 net/socket.c:2542
[<ffffffff838525b8>] __sys_sendmsg net/socket.c:2571 [inline]
[<ffffffff838525b8>] __do_sys_sendmsg net/socket.c:2580 [inline]
[<ffffffff838525b8>] __se_sys_sendmsg net/socket.c:2578 [inline]
[<ffffffff838525b8>] __x64_sys_sendmsg+0x78/0xf0 net/socket.c:2578
[<ffffffff845ad8d5>] do_syscall_x64 arch/x86/entry/common.c:50 [inline]
[<ffffffff845ad8d5>] do_syscall_64+0x35/0xb0 arch/x86/entry/common.c:80
[<ffffffff8460006a>] entry_SYSCALL_64_after_hwframe+0x46/0xb0

Fixes: 37aa50c539bc ("vlan: introduce vlan_dev_free_egress_priority")
Reported-by: syzbot <syzkaller@googlegroups.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Xin Long <lucien.xin@gmail.com>
Reviewed-by: Xin Long <lucien.xin@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonfp: fix issue of skb segments exceeds descriptor limitation
Baowen Zheng [Fri, 8 Jul 2022 10:07:18 +0000 (11:07 +0100)]
nfp: fix issue of skb segments exceeds descriptor limitation

TCP packets will be dropped if the segments number in the tx skb
exceeds limitation when sending iperf3 traffic with --zerocopy option.

we make the following changes:

Get nr_frags in nfp_nfdk_tx_maybe_close_block instead of passing from
outside because it will be changed after skb_linearize operation.

Fill maximum dma_len in first tx descriptor to make sure the whole
head is included in the first descriptor.

Fixes: c10d12e3dce8 ("nfp: add support for NFDK data path")
Signed-off-by: Baowen Zheng <baowen.zheng@corigine.com>
Reviewed-by: Louis Peens <louis.peens@corigine.com>
Signed-off-by: Simon Horman <simon.horman@corigine.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonetfilter: nf_log: incorrect offset to network header
Pablo Neira Ayuso [Tue, 5 Jul 2022 08:26:15 +0000 (10:26 +0200)]
netfilter: nf_log: incorrect offset to network header

NFPROTO_ARP is expecting to find the ARP header at the network offset.

In the particular case of ARP, HTYPE= field shows the initial bytes of
the ethernet header destination MAC address.

 netdev out: IN= OUT=bridge0 MACSRC=c2:76:e5:71:e1:de MACDST=36:b0:4a:e2:72:ea MACPROTO=0806 ARP HTYPE=14000 PTYPE=0x4ae2 OPCODE=49782

NFPROTO_NETDEV egress hook is also expecting to find the IP headers at
the network offset.

Fixes: 35b9395104d5 ("netfilter: add generic ARP packet logger")
Reported-by: Tom Yan <tom.ty89@gmail.com>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
2 years agoMerge branch 'selftests-forwarding-install-two-missing-tests'
Jakub Kicinski [Sat, 9 Jul 2022 03:31:02 +0000 (20:31 -0700)]
Merge branch 'selftests-forwarding-install-two-missing-tests'

Martin Blumenstingl says:

====================
selftests: forwarding: Install two missing tests

For some distributions (e.g. OpenWrt) we don't want to rely on rsync
to copy the tests to the target as some extra dependencies need to be
installed. The Makefile in tools/testing/selftests/net/forwarding
already installs most of the tests.

This series adds the two missing tests to the list of installed tests.
That way a downstream distribution can build a package using this
Makefile (and add dependencies there as needed).
====================

Link: https://lore.kernel.org/r/20220707135532.1783925-1-martin.blumenstingl@googlemail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoselftests: forwarding: Install no_forwarding.sh
Martin Blumenstingl [Thu, 7 Jul 2022 13:55:32 +0000 (15:55 +0200)]
selftests: forwarding: Install no_forwarding.sh

When using the Makefile from tools/testing/selftests/net/forwarding/
all tests should be installed. Add no_forwarding.sh to the list of
"to be installed tests" where it has been missing so far.

Fixes: 476a4f05d9b83f ("selftests: forwarding: add a no_forwarding.sh test")
Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoselftests: forwarding: Install local_termination.sh
Martin Blumenstingl [Thu, 7 Jul 2022 13:55:31 +0000 (15:55 +0200)]
selftests: forwarding: Install local_termination.sh

When using the Makefile from tools/testing/selftests/net/forwarding/
all tests should be installed. Add local_termination.sh to the list of
"to be installed tests" where it has been missing so far.

Fixes: 90b9566aa5cd3f ("selftests: forwarding: add a test for local_termination.sh")
Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoMerge https://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf
Jakub Kicinski [Fri, 8 Jul 2022 22:24:16 +0000 (15:24 -0700)]
Merge https://git./linux/kernel/git/bpf/bpf

Daniel Borkmann says:

====================
bpf 2022-07-08

We've added 3 non-merge commits during the last 2 day(s) which contain
a total of 7 files changed, 40 insertions(+), 24 deletions(-).

The main changes are:

1) Fix cBPF splat triggered by skb not having a mac header, from Eric Dumazet.

2) Fix spurious packet loss in generic XDP when pushing packets out (note
   that native XDP is not affected by the issue), from Johan Almbladh.

3) Fix bpf_dynptr_{read,write}() helper signatures with flag argument before
   its set in stone as UAPI, from Joanne Koong.

* https://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf:
  bpf: Add flags arg to bpf_dynptr_read and bpf_dynptr_write APIs
  bpf: Make sure mac_header was set before using it
  xdp: Fix spurious packet loss in generic XDP TX path
====================

Link: https://lore.kernel.org/r/20220708213418.19626-1-daniel@iogearbox.net
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoMerge branch 'sysctl-data-races'
David S. Miller [Fri, 8 Jul 2022 11:10:34 +0000 (12:10 +0100)]
Merge branch 'sysctl-data-races'

Kuniyuki Iwashima says:

====================
sysctl: Fix data-races around ipv4_table.

A sysctl variable is accessed concurrently, and there is always a chance
of data-race.  So, all readers and writers need some basic protection to
avoid load/store-tearing.

The first half of this series changes some proc handlers used in ipv4_table
to use READ_ONCE() and WRITE_ONCE() internally to fix data-races on the
sysctl side.  Then, the second half adds READ_ONCE() to the other readers
of ipv4_table.

Changes:
  v2:
    * Drop some changes that makes backporting difficult
      * First cleanup patch
      * Lockless helpers and .proc_handler changes
    * Drop the tracing part for .sysctl_mem
      * Steve already posted a fix
    * Drop int-to-bool change for cipso
      * Should be posted to net-next later
    * Drop proc_dobool() change
      * Can be included in another series

  v1: https://lore.kernel.org/netdev/20220706052130.16368-1-kuniyu@amazon.com/
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoipv4: Fix a data-race around sysctl_fib_sync_mem.
Kuniyuki Iwashima [Wed, 6 Jul 2022 23:40:03 +0000 (16:40 -0700)]
ipv4: Fix a data-race around sysctl_fib_sync_mem.

While reading sysctl_fib_sync_mem, it can be changed concurrently.
So, we need to add READ_ONCE() to avoid a data-race.

Fixes: 9ab948a91b2c ("ipv4: Allow amount of dirty memory from fib resizing to be controllable")
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoicmp: Fix data-races around sysctl.
Kuniyuki Iwashima [Wed, 6 Jul 2022 23:40:02 +0000 (16:40 -0700)]
icmp: Fix data-races around sysctl.

While reading icmp sysctl variables, they can be changed concurrently.
So, we need to add READ_ONCE() to avoid data-races.

Fixes: 4cdf507d5452 ("icmp: add a global rate limitation")
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agocipso: Fix data-races around sysctl.
Kuniyuki Iwashima [Wed, 6 Jul 2022 23:40:01 +0000 (16:40 -0700)]
cipso: Fix data-races around sysctl.

While reading cipso sysctl variables, they can be changed concurrently.
So, we need to add READ_ONCE() to avoid data-races.

Fixes: 446fda4f2682 ("[NetLabel]: CIPSOv4 engine")
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Acked-by: Paul Moore <paul@paul-moore.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: Fix data-races around sysctl_mem.
Kuniyuki Iwashima [Wed, 6 Jul 2022 23:40:00 +0000 (16:40 -0700)]
net: Fix data-races around sysctl_mem.

While reading .sysctl_mem, it can be changed concurrently.
So, we need to add READ_ONCE() to avoid data-races.

Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoinetpeer: Fix data-races around sysctl.
Kuniyuki Iwashima [Wed, 6 Jul 2022 23:39:59 +0000 (16:39 -0700)]
inetpeer: Fix data-races around sysctl.

While reading inetpeer sysctl variables, they can be changed
concurrently.  So, we need to add READ_ONCE() to avoid data-races.

Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agotcp: Fix a data-race around sysctl_tcp_max_orphans.
Kuniyuki Iwashima [Wed, 6 Jul 2022 23:39:58 +0000 (16:39 -0700)]
tcp: Fix a data-race around sysctl_tcp_max_orphans.

While reading sysctl_tcp_max_orphans, it can be changed concurrently.
So, we need to add READ_ONCE() to avoid a data-race.

Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agosysctl: Fix data races in proc_dointvec_jiffies().
Kuniyuki Iwashima [Wed, 6 Jul 2022 23:39:57 +0000 (16:39 -0700)]
sysctl: Fix data races in proc_dointvec_jiffies().

A sysctl variable is accessed concurrently, and there is always a chance
of data-race.  So, all readers and writers need some basic protection to
avoid load/store-tearing.

This patch changes proc_dointvec_jiffies() to use READ_ONCE() and
WRITE_ONCE() internally to fix data-races on the sysctl side.  For now,
proc_dointvec_jiffies() itself is tolerant to a data-race, but we still
need to add annotations on the other subsystem's side.

Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agosysctl: Fix data races in proc_doulongvec_minmax().
Kuniyuki Iwashima [Wed, 6 Jul 2022 23:39:56 +0000 (16:39 -0700)]
sysctl: Fix data races in proc_doulongvec_minmax().

A sysctl variable is accessed concurrently, and there is always a chance
of data-race.  So, all readers and writers need some basic protection to
avoid load/store-tearing.

This patch changes proc_doulongvec_minmax() to use READ_ONCE() and
WRITE_ONCE() internally to fix data-races on the sysctl side.  For now,
proc_doulongvec_minmax() itself is tolerant to a data-race, but we still
need to add annotations on the other subsystem's side.

Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agosysctl: Fix data races in proc_douintvec_minmax().
Kuniyuki Iwashima [Wed, 6 Jul 2022 23:39:55 +0000 (16:39 -0700)]
sysctl: Fix data races in proc_douintvec_minmax().

A sysctl variable is accessed concurrently, and there is always a chance
of data-race.  So, all readers and writers need some basic protection to
avoid load/store-tearing.

This patch changes proc_douintvec_minmax() to use READ_ONCE() and
WRITE_ONCE() internally to fix data-races on the sysctl side.  For now,
proc_douintvec_minmax() itself is tolerant to a data-race, but we still
need to add annotations on the other subsystem's side.

Fixes: 61d9b56a8920 ("sysctl: add unsigned int range support")
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agosysctl: Fix data races in proc_dointvec_minmax().
Kuniyuki Iwashima [Wed, 6 Jul 2022 23:39:54 +0000 (16:39 -0700)]
sysctl: Fix data races in proc_dointvec_minmax().

A sysctl variable is accessed concurrently, and there is always a chance
of data-race.  So, all readers and writers need some basic protection to
avoid load/store-tearing.

This patch changes proc_dointvec_minmax() to use READ_ONCE() and
WRITE_ONCE() internally to fix data-races on the sysctl side.  For now,
proc_dointvec_minmax() itself is tolerant to a data-race, but we still
need to add annotations on the other subsystem's side.

Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agosysctl: Fix data races in proc_douintvec().
Kuniyuki Iwashima [Wed, 6 Jul 2022 23:39:53 +0000 (16:39 -0700)]
sysctl: Fix data races in proc_douintvec().

A sysctl variable is accessed concurrently, and there is always a chance
of data-race.  So, all readers and writers need some basic protection to
avoid load/store-tearing.

This patch changes proc_douintvec() to use READ_ONCE() and WRITE_ONCE()
internally to fix data-races on the sysctl side.  For now, proc_douintvec()
itself is tolerant to a data-race, but we still need to add annotations on
the other subsystem's side.

Fixes: e7d316a02f68 ("sysctl: handle error writing UINT_MAX to u32 fields")
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agosysctl: Fix data races in proc_dointvec().
Kuniyuki Iwashima [Wed, 6 Jul 2022 23:39:52 +0000 (16:39 -0700)]
sysctl: Fix data races in proc_dointvec().

A sysctl variable is accessed concurrently, and there is always a chance
of data-race.  So, all readers and writers need some basic protection to
avoid load/store-tearing.

This patch changes proc_dointvec() to use READ_ONCE() and WRITE_ONCE()
internally to fix data-races on the sysctl side.  For now, proc_dointvec()
itself is tolerant to a data-race, but we still need to add annotations on
the other subsystem's side.

Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Signed-off-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: sock: tracing: Fix sock_exceed_buf_limit not to dereference stale pointer
Steven Rostedt (Google) [Wed, 6 Jul 2022 14:50:40 +0000 (10:50 -0400)]
net: sock: tracing: Fix sock_exceed_buf_limit not to dereference stale pointer

The trace event sock_exceed_buf_limit saves the prot->sysctl_mem pointer
and then dereferences it in the TP_printk() portion. This is unsafe as the
TP_printk() portion is executed at the time the buffer is read. That is,
it can be seconds, minutes, days, months, even years later. If the proto
is freed, then this dereference will can also lead to a kernel crash.

Instead, save the sysctl_mem array into the ring buffer and have the
TP_printk() reference that instead. This is the proper and safe way to
read pointers in trace events.

Link: https://lore.kernel.org/all/20220706052130.16368-12-kuniyu@amazon.com/
Cc: stable@vger.kernel.org
Fixes: 3847ce32aea9f ("core: add tracepoints for queueing skb to rcvbuf")
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Acked-by: Kuniyuki Iwashima <kuniyu@amazon.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agobpf: Add flags arg to bpf_dynptr_read and bpf_dynptr_write APIs
Joanne Koong [Wed, 6 Jul 2022 23:25:47 +0000 (16:25 -0700)]
bpf: Add flags arg to bpf_dynptr_read and bpf_dynptr_write APIs

Commit 13bbbfbea759 ("bpf: Add bpf_dynptr_read and bpf_dynptr_write")
added the bpf_dynptr_write() and bpf_dynptr_read() APIs.

However, it will be needed for some dynptr types to pass in flags as
well (e.g. when writing to a skb, the user may like to invalidate the
hash or recompute the checksum).

This patch adds a "u64 flags" arg to the bpf_dynptr_read() and
bpf_dynptr_write() APIs before their UAPI signature freezes where
we then cannot change them anymore with a 5.19.x released kernel.

Fixes: 13bbbfbea759 ("bpf: Add bpf_dynptr_read and bpf_dynptr_write")
Signed-off-by: Joanne Koong <joannelkoong@gmail.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/r/20220706232547.4016651-1-joannelkoong@gmail.com
2 years agonet: ocelot: fix wrong time_after usage
Pavel Skripkin [Wed, 6 Jul 2022 13:28:45 +0000 (16:28 +0300)]
net: ocelot: fix wrong time_after usage

Accidentally noticed, that this driver is the only user of
while (time_after(jiffies...)).

It looks like typo, because likely this while loop will finish after 1st
iteration, because time_after() returns true when 1st argument _is after_
2nd one.

There is one possible problem with this poll loop: the scheduler could put
the thread to sleep, and it does not get woken up for
OCELOT_FDMA_CH_SAFE_TIMEOUT_US. During that time, the hardware has done
its thing, but you exit the while loop and return -ETIMEDOUT.

Fix it by using sane poll API that avoids all problems described above

Fixes: 753a026cfec1 ("net: ocelot: add FDMA support")
Suggested-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Pavel Skripkin <paskripkin@gmail.com>
Reviewed-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Link: https://lore.kernel.org/r/20220706132845.27968-1-paskripkin@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoMerge tag 'mlx5-fixes-2022-07-06' of git://git.kernel.org/pub/scm/linux/kernel/git...
Jakub Kicinski [Fri, 8 Jul 2022 00:44:45 +0000 (17:44 -0700)]
Merge tag 'mlx5-fixes-2022-07-06' of git://git./linux/kernel/git/saeed/linux

Saeed Mahameed says:

====================
mlx5 fixes 2022-07-06

This series provides bug fixes to mlx5 driver.

* tag 'mlx5-fixes-2022-07-06' of git://git.kernel.org/pub/scm/linux/kernel/git/saeed/linux:
  net/mlx5e: Ring the TX doorbell on DMA errors
  net/mlx5e: Fix capability check for updating vnic env counters
  net/mlx5e: CT: Use own workqueue instead of mlx5e priv
  net/mlx5: Lag, correct get the port select mode str
  net/mlx5e: Fix enabling sriov while tc nic rules are offloaded
  net/mlx5e: kTLS, Fix build time constant test in RX
  net/mlx5e: kTLS, Fix build time constant test in TX
  net/mlx5: Lag, decouple FDB selection and shared FDB
  net/mlx5: TC, allow offload from uplink to other PF's VF
====================

Link: https://lore.kernel.org/r/20220706231309.38579-1-saeed@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agonet: ethernet: ti: am65-cpsw: Fix devlink port register sequence
Siddharth Vadapalli [Wed, 6 Jul 2022 07:02:08 +0000 (12:32 +0530)]
net: ethernet: ti: am65-cpsw: Fix devlink port register sequence

Renaming interfaces using udevd depends on the interface being registered
before its netdev is registered. Otherwise, udevd reads an empty
phys_port_name value, resulting in the interface not being renamed.

Fix this by registering the interface before registering its netdev
by invoking am65_cpsw_nuss_register_devlink() before invoking
register_netdev() for the interface.

Move the function call to devlink_port_type_eth_set(), invoking it after
register_netdev() is invoked, to ensure that netlink notification for the
port state change is generated after the netdev is completely initialized.

Fixes: 58356eb31d60 ("net: ti: am65-cpsw-nuss: Add devlink support")
Signed-off-by: Siddharth Vadapalli <s-vadapalli@ti.com>
Link: https://lore.kernel.org/r/20220706070208.12207-1-s-vadapalli@ti.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agonet: stmmac: dwc-qos: Disable split header for Tegra194
Jon Hunter [Wed, 6 Jul 2022 08:39:13 +0000 (09:39 +0100)]
net: stmmac: dwc-qos: Disable split header for Tegra194

There is a long-standing issue with the Synopsys DWC Ethernet driver
for Tegra194 where random system crashes have been observed [0]. The
problem occurs when the split header feature is enabled in the stmmac
driver. In the bad case, a larger than expected buffer length is
received and causes the calculation of the total buffer length to
overflow. This results in a very large buffer length that causes the
kernel to crash. Why this larger buffer length is received is not clear,
however, the feedback from the NVIDIA design team is that the split
header feature is not supported for Tegra194. Therefore, disable split
header support for Tegra194 to prevent these random crashes from
occurring.

[0] https://lore.kernel.org/linux-tegra/b0b17697-f23e-8fa5-3757-604a86f3a095@nvidia.com/

Fixes: 67afd6d1cfdf ("net: stmmac: Add Split Header support and enable it in XGMAC cores")
Signed-off-by: Jon Hunter <jonathanh@nvidia.com>
Link: https://lore.kernel.org/r/20220706083913.13750-1-jonathanh@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agonetfilter: conntrack: fix crash due to confirmed bit load reordering
Florian Westphal [Wed, 6 Jul 2022 14:50:04 +0000 (16:50 +0200)]
netfilter: conntrack: fix crash due to confirmed bit load reordering

Kajetan Puchalski reports crash on ARM, with backtrace of:

__nf_ct_delete_from_lists
nf_ct_delete
early_drop
__nf_conntrack_alloc

Unlike atomic_inc_not_zero, refcount_inc_not_zero is not a full barrier.
conntrack uses SLAB_TYPESAFE_BY_RCU, i.e. it is possible that a 'newly'
allocated object is still in use on another CPU:

CPU1 CPU2
encounter 'ct' during hlist walk
 delete_from_lists
 refcount drops to 0
 kmem_cache_free(ct);
 __nf_conntrack_alloc() // returns same object
refcount_inc_not_zero(ct); /* might fail */

/* If set, ct is public/in the hash table */
test_bit(IPS_CONFIRMED_BIT, &ct->status);

In case CPU1 already set refcount back to 1, refcount_inc_not_zero()
will succeed.

The expected possibilities for a CPU that obtained the object 'ct'
(but no reference so far) are:

1. refcount_inc_not_zero() fails.  CPU2 ignores the object and moves to
   the next entry in the list.  This happens for objects that are about
   to be free'd, that have been free'd, or that have been reallocated
   by __nf_conntrack_alloc(), but where the refcount has not been
   increased back to 1 yet.

2. refcount_inc_not_zero() succeeds. CPU2 checks the CONFIRMED bit
   in ct->status.  If set, the object is public/in the table.

   If not, the object must be skipped; CPU2 calls nf_ct_put() to
   un-do the refcount increment and moves to the next object.

Parallel deletion from the hlists is prevented by a
'test_and_set_bit(IPS_DYING_BIT, &ct->status);' check, i.e. only one
cpu will do the unlink, the other one will only drop its reference count.

Because refcount_inc_not_zero is not a full barrier, CPU2 may try to
delete an object that is not on any list:

1. refcount_inc_not_zero() successful (refcount inited to 1 on other CPU)
2. CONFIRMED test also successful (load was reordered or zeroing
   of ct->status not yet visible)
3. delete_from_lists unlinks entry not on the hlist, because
   IPS_DYING_BIT is 0 (already cleared).

2) is already wrong: CPU2 will handle a partially initited object
that is supposed to be private to CPU1.

Add needed barriers when refcount_inc_not_zero() is successful.

It also inserts a smp_wmb() before the refcount is set to 1 during
allocation.

Because other CPU might still see the object, refcount_set(1)
"resurrects" it, so we need to make sure that other CPUs will also observe
the right content.  In particular, the CONFIRMED bit test must only pass
once the object is fully initialised and either in the hash or about to be
inserted (with locks held to delay possible unlink from early_drop or
gc worker).

I did not change flow_offload_alloc(), as far as I can see it should call
refcount_inc(), not refcount_inc_not_zero(): the ct object is attached to
the skb so its refcount should be >= 1 in all cases.

v2: prefer smp_acquire__after_ctrl_dep to smp_rmb (Will Deacon).
v3: keep smp_acquire__after_ctrl_dep close to refcount_inc_not_zero call
    add comment in nf_conntrack_netlink, no control dependency there
    due to locks.

Cc: Peter Zijlstra <peterz@infradead.org>
Link: https://lore.kernel.org/all/Yr7WTfd6AVTQkLjI@e126311.manchester.arm.com/
Reported-by: Kajetan Puchalski <kajetan.puchalski@arm.com>
Diagnosed-by: Will Deacon <will@kernel.org>
Fixes: 719774377622 ("netfilter: conntrack: convert to refcount_t api")
Signed-off-by: Florian Westphal <fw@strlen.de>
Acked-by: Will Deacon <will@kernel.org>
2 years agobpf: Make sure mac_header was set before using it
Eric Dumazet [Thu, 7 Jul 2022 12:39:00 +0000 (12:39 +0000)]
bpf: Make sure mac_header was set before using it

Classic BPF has a way to load bytes starting from the mac header.

Some skbs do not have a mac header, and skb_mac_header()
in this case is returning a pointer that 65535 bytes after
skb->head.

Existing range check in bpf_internal_load_pointer_neg_helper()
was properly kicking and no illegal access was happening.

New sanity check in skb_mac_header() is firing, so we need
to avoid it.

WARNING: CPU: 1 PID: 28990 at include/linux/skbuff.h:2785 skb_mac_header include/linux/skbuff.h:2785 [inline]
WARNING: CPU: 1 PID: 28990 at include/linux/skbuff.h:2785 bpf_internal_load_pointer_neg_helper+0x1b1/0x1c0 kernel/bpf/core.c:74
Modules linked in:
CPU: 1 PID: 28990 Comm: syz-executor.0 Not tainted 5.19.0-rc4-syzkaller-00865-g4874fb9484be #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 06/29/2022
RIP: 0010:skb_mac_header include/linux/skbuff.h:2785 [inline]
RIP: 0010:bpf_internal_load_pointer_neg_helper+0x1b1/0x1c0 kernel/bpf/core.c:74
Code: ff ff 45 31 f6 e9 5a ff ff ff e8 aa 27 40 00 e9 3b ff ff ff e8 90 27 40 00 e9 df fe ff ff e8 86 27 40 00 eb 9e e8 2f 2c f3 ff <0f> 0b eb b1 e8 96 27 40 00 e9 79 fe ff ff 90 41 57 41 56 41 55 41
RSP: 0018:ffffc9000309f668 EFLAGS: 00010216
RAX: 0000000000000118 RBX: ffffffffffeff00c RCX: ffffc9000e417000
RDX: 0000000000040000 RSI: ffffffff81873f21 RDI: 0000000000000003
RBP: ffff8880842878c0 R08: 0000000000000003 R09: 000000000000ffff
R10: 000000000000ffff R11: 0000000000000001 R12: 0000000000000004
R13: ffff88803ac56c00 R14: 000000000000ffff R15: dffffc0000000000
FS: 00007f5c88a16700(0000) GS:ffff8880b9b00000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00007fdaa9f6c058 CR3: 000000003a82c000 CR4: 00000000003506e0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
Call Trace:
<TASK>
____bpf_skb_load_helper_32 net/core/filter.c:276 [inline]
bpf_skb_load_helper_32+0x191/0x220 net/core/filter.c:264

Fixes: f9aefd6b2aa3 ("net: warn if mac header was not set")
Reported-by: syzbot <syzkaller@googlegroups.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/bpf/20220707123900.945305-1-edumazet@google.com
2 years agoMerge tag 'net-5.19-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net
Linus Torvalds [Thu, 7 Jul 2022 17:08:20 +0000 (10:08 -0700)]
Merge tag 'net-5.19-rc6' of git://git./linux/kernel/git/netdev/net

Pull networking fixes from Paolo Abeni:
 "Including fixes from bpf, netfilter, can, and bluetooth.

  Current release - regressions:

   - bluetooth: fix deadlock on hci_power_on_sync

  Previous releases - regressions:

   - sched: act_police: allow 'continue' action offload

   - eth: usbnet: fix memory leak in error case

   - eth: ibmvnic: properly dispose of all skbs during a failover

  Previous releases - always broken:

   - bpf:
       - fix insufficient bounds propagation from
         adjust_scalar_min_max_vals
       - clear page contiguity bit when unmapping pool

   - netfilter: nft_set_pipapo: release elements in clone from
     abort path

   - mptcp: netlink: issue MP_PRIO signals from userspace PMs

   - can:
       - rcar_canfd: fix data transmission failed on R-Car V3U
       - gs_usb: gs_usb_open/close(): fix memory leak

  Misc:

   - add Wenjia as SMC maintainer"

* tag 'net-5.19-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net: (57 commits)
  wireguard: Kconfig: select CRYPTO_CHACHA_S390
  crypto: s390 - do not depend on CRYPTO_HW for SIMD implementations
  wireguard: selftests: use microvm on x86
  wireguard: selftests: always call kernel makefile
  wireguard: selftests: use virt machine on m68k
  wireguard: selftests: set fake real time in init
  r8169: fix accessing unset transport header
  net: rose: fix UAF bug caused by rose_t0timer_expiry
  usbnet: fix memory leak in error case
  Revert "tls: rx: move counting TlsDecryptErrors for sync"
  mptcp: update MIB_RMSUBFLOW in cmd_sf_destroy
  mptcp: fix local endpoint accounting
  selftests: mptcp: userspace PM support for MP_PRIO signals
  mptcp: netlink: issue MP_PRIO signals from userspace PMs
  mptcp: Acquire the subflow socket lock before modifying MP_PRIO flags
  mptcp: Avoid acquiring PM lock for subflow priority changes
  mptcp: fix locking in mptcp_nl_cmd_sf_destroy()
  net/mlx5e: Fix matchall police parameters validation
  net/sched: act_police: allow 'continue' action offload
  net: lan966x: hardcode the number of external ports
  ...

2 years agoMerge tag 'pinctrl-v5.19-2' of git://git.kernel.org/pub/scm/linux/kernel/git/linusw...
Linus Torvalds [Thu, 7 Jul 2022 17:02:38 +0000 (10:02 -0700)]
Merge tag 'pinctrl-v5.19-2' of git://git./linux/kernel/git/linusw/linux-pinctrl

Pull pin control fixes from Linus Walleij:

 - Tag Intel pin control as supported in MAINTAINERS

 - Fix a NULL pointer exception in the Aspeed driver

 - Correct some NAND functions in the Sunxi A83T driver

 - Use the right offset for some Sunxi pins

 - Fix a zero base offset in the Freescale (NXP) i.MX93

 - Fix the IRQ support in the STM32 driver

* tag 'pinctrl-v5.19-2' of git://git.kernel.org/pub/scm/linux/kernel/git/linusw/linux-pinctrl:
  pinctrl: stm32: fix optional IRQ support to gpios
  pinctrl: imx: Add the zero base flag for imx93
  pinctrl: sunxi: sunxi_pconf_set: use correct offset
  pinctrl: sunxi: a83t: Fix NAND function name for some pins
  pinctrl: aspeed: Fix potential NULL dereference in aspeed_pinmux_set_mux()
  MAINTAINERS: Update Intel pin control to Supported

2 years agosignal handling: don't use BUG_ON() for debugging
Linus Torvalds [Wed, 6 Jul 2022 19:20:59 +0000 (12:20 -0700)]
signal handling: don't use BUG_ON() for debugging

These are indeed "should not happen" situations, but it turns out recent
changes made the 'task_is_stopped_or_trace()' case trigger (fix for that
exists, is pending more testing), and the BUG_ON() makes it
unnecessarily hard to actually debug for no good reason.

It's been that way for a long time, but let's make it clear: BUG_ON() is
not good for debugging, and should never be used in situations where you
could just say "this shouldn't happen, but we can continue".

Use WARN_ON_ONCE() instead to make sure it gets logged, and then just
continue running.  Instead of making the system basically unusuable
because you crashed the machine while potentially holding some very core
locks (eg this function is commonly called while holding 'tasklist_lock'
for writing).

Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2 years agoMerge branch 'wireguard-patches-for-5-19-rc6'
Jakub Kicinski [Thu, 7 Jul 2022 03:04:09 +0000 (20:04 -0700)]
Merge branch 'wireguard-patches-for-5-19-rc6'

Jason A. Donenfeld says:

====================
wireguard patches for 5.19-rc6

1) A few small fixups to the selftests, per usual. Of particular note is
   a fix for a test flake that occurred on especially fast systems that
   boot in less than a second.

2) An addition during this cycle of some s390 crypto interacted with the
   way wireguard selects dependencies, resulting in linker errors
   reported by the kernel test robot. So Vladis sent in a patch for
   that, which also required a small preparatory fix moving some Kconfig
   symbols around.
====================

Link: https://lore.kernel.org/r/20220707003157.526645-1-Jason@zx2c4.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agowireguard: Kconfig: select CRYPTO_CHACHA_S390
Vladis Dronov [Thu, 7 Jul 2022 00:31:57 +0000 (02:31 +0200)]
wireguard: Kconfig: select CRYPTO_CHACHA_S390

Select the new implementation of CHACHA20 for S390 when available.
It is faster than the generic software implementation, but also prevents
some linker errors in certain situations.

Reported-by: kernel test robot <lkp@intel.com>
Link: https://lore.kernel.org/linux-kernel/202207030630.6SZVkrWf-lkp@intel.com/
Signed-off-by: Vladis Dronov <vdronov@redhat.com>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agocrypto: s390 - do not depend on CRYPTO_HW for SIMD implementations
Jason A. Donenfeld [Thu, 7 Jul 2022 00:31:56 +0000 (02:31 +0200)]
crypto: s390 - do not depend on CRYPTO_HW for SIMD implementations

Various accelerated software implementation Kconfig values for S390 were
mistakenly placed into drivers/crypto/Kconfig, even though they're
mainly just SIMD code and live in arch/s390/crypto/ like usual. This
gives them the very unusual dependency on CRYPTO_HW, which leads to
problems elsewhere.

This patch fixes the issue by moving the Kconfig values for non-hardware
drivers into the usual place in crypto/Kconfig.

Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agowireguard: selftests: use microvm on x86
Jason A. Donenfeld [Thu, 7 Jul 2022 00:31:55 +0000 (02:31 +0200)]
wireguard: selftests: use microvm on x86

This makes for faster tests, faster compile time, and allows us to ditch
ACPI finally.

Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agowireguard: selftests: always call kernel makefile
Jason A. Donenfeld [Thu, 7 Jul 2022 00:31:54 +0000 (02:31 +0200)]
wireguard: selftests: always call kernel makefile

These selftests are used for much more extensive changes than just the
wireguard source files. So always call the kernel's build file, which
will do something or nothing after checking the whole tree, per usual.

Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agowireguard: selftests: use virt machine on m68k
Jason A. Donenfeld [Thu, 7 Jul 2022 00:31:53 +0000 (02:31 +0200)]
wireguard: selftests: use virt machine on m68k

This should be a bit more stable hopefully.

Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agowireguard: selftests: set fake real time in init
Jason A. Donenfeld [Thu, 7 Jul 2022 00:31:52 +0000 (02:31 +0200)]
wireguard: selftests: set fake real time in init

Not all platforms have an RTC, and rather than trying to force one into
each, it's much easier to just set a fixed time. This is necessary
because WireGuard's latest handshakes parameter is returned in wallclock
time, and if the system time isn't set, and the system is really fast,
then this returns 0, which trips the test.

Turning this on requires setting CONFIG_COMPAT_32BIT_TIME=y, as musl
doesn't support settimeofday without it.

Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agor8169: fix accessing unset transport header
Heiner Kallweit [Tue, 5 Jul 2022 19:15:22 +0000 (21:15 +0200)]
r8169: fix accessing unset transport header

66e4c8d95008 ("net: warn if transport header was not set") added
a check that triggers a warning in r8169, see [0].

The commit referenced in the Fixes tag refers to the change from
which the patch applies cleanly, there's nothing wrong with this
commit. It seems the actual issue (not bug, because the warning
is harmless here) was introduced with bdfa4ed68187
("r8169: use Giant Send").

[0] https://bugzilla.kernel.org/show_bug.cgi?id=216157

Fixes: 8d520b4de3ed ("r8169: work around RTL8125 UDP hw bug")
Reported-by: Erhard F. <erhard_f@mailbox.org>
Tested-by: Erhard F. <erhard_f@mailbox.org>
Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com>
Link: https://lore.kernel.org/r/1b2c2b29-3dc0-f7b6-5694-97ec526d51a0@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agonet: rose: fix UAF bug caused by rose_t0timer_expiry
Duoming Zhou [Tue, 5 Jul 2022 12:56:10 +0000 (20:56 +0800)]
net: rose: fix UAF bug caused by rose_t0timer_expiry

There are UAF bugs caused by rose_t0timer_expiry(). The
root cause is that del_timer() could not stop the timer
handler that is running and there is no synchronization.
One of the race conditions is shown below:

    (thread 1)             |        (thread 2)
                           | rose_device_event
                           |   rose_rt_device_down
                           |     rose_remove_neigh
rose_t0timer_expiry        |       rose_stop_t0timer(rose_neigh)
  ...                      |         del_timer(&neigh->t0timer)
                           |         kfree(rose_neigh) //[1]FREE
  neigh->dce_mode //[2]USE |

The rose_neigh is deallocated in position [1] and use in
position [2].

The crash trace triggered by POC is like below:

BUG: KASAN: use-after-free in expire_timers+0x144/0x320
Write of size 8 at addr ffff888009b19658 by task swapper/0/0
...
Call Trace:
 <IRQ>
 dump_stack_lvl+0xbf/0xee
 print_address_description+0x7b/0x440
 print_report+0x101/0x230
 ? expire_timers+0x144/0x320
 kasan_report+0xed/0x120
 ? expire_timers+0x144/0x320
 expire_timers+0x144/0x320
 __run_timers+0x3ff/0x4d0
 run_timer_softirq+0x41/0x80
 __do_softirq+0x233/0x544
 ...

This patch changes rose_stop_ftimer() and rose_stop_t0timer()
in rose_remove_neigh() to del_timer_sync() in order that the
timer handler could be finished before the resources such as
rose_neigh and so on are deallocated. As a result, the UAF
bugs could be mitigated.

Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Signed-off-by: Duoming Zhou <duoming@zju.edu.cn>
Link: https://lore.kernel.org/r/20220705125610.77971-1-duoming@zju.edu.cn
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agousbnet: fix memory leak in error case
Oliver Neukum [Tue, 5 Jul 2022 12:53:51 +0000 (14:53 +0200)]
usbnet: fix memory leak in error case

usbnet_write_cmd_async() mixed up which buffers
need to be freed in which error case.

v2: add Fixes tag
v3: fix uninitialized buf pointer

Fixes: 877bd862f32b8 ("usbnet: introduce usbnet 3 command helpers")
Signed-off-by: Oliver Neukum <oneukum@suse.com>
Link: https://lore.kernel.org/r/20220705125351.17309-1-oneukum@suse.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agonet/mlx5e: Ring the TX doorbell on DMA errors
Maxim Mikityanskiy [Mon, 30 May 2022 11:01:37 +0000 (14:01 +0300)]
net/mlx5e: Ring the TX doorbell on DMA errors

TX doorbells may be postponed, because sometimes the driver knows that
another packet follows (for example, when xmit_more is true, or when a
MPWQE session is closed before transmitting a packet).

However, the DMA mapping may fail for the next packet, in which case a
new WQE is not posted, the doorbell isn't updated either, and the
transmission of the previous packet will be delayed indefinitely.

This commit fixes the described rare error flow by posting a NOP and
ringing the doorbell on errors to flush all the previous packets. The
MPWQE session is closed before that. DMA mapping in the MPWQE flow is
moved to the beginning of mlx5e_sq_xmit_mpwqe, because empty sessions
are not allowed. Stop room always has enough space for a NOP, because
the actual TX WQE is not posted.

Fixes: e586b3b0baee ("net/mlx5: Ethernet Datapath files")
Signed-off-by: Maxim Mikityanskiy <maximmi@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5e: Fix capability check for updating vnic env counters
Gal Pressman [Mon, 27 Jun 2022 12:05:53 +0000 (15:05 +0300)]
net/mlx5e: Fix capability check for updating vnic env counters

The existing capability check for vnic env counters only checks for
receive steering discards, although we need the counters update for the
exposed internal queue oob counter as well. This could result in the
latter counter not being updated correctly when the receive steering
discards counter is not supported.
Fix that by checking whether any counter is supported instead of only
the steering counter capability.

Fixes: 0cfafd4b4ddf ("net/mlx5e: Add device out of buffer counter")
Signed-off-by: Gal Pressman <gal@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5e: CT: Use own workqueue instead of mlx5e priv
Roi Dayan [Tue, 21 Jun 2022 07:43:55 +0000 (10:43 +0300)]
net/mlx5e: CT: Use own workqueue instead of mlx5e priv

Allocate a ct priv workqueue instead of using mlx5e priv one
so flushing will only be of related CT entries.
Also move flushing of the workqueue before rhashtable destroy
otherwise entries won't be valid.

Fixes: b069e14fff46 ("net/mlx5e: CT: Fix queued up restore put() executing after relevant ft release")
Signed-off-by: Roi Dayan <roid@nvidia.com>
Reviewed-by: Oz Shlomo <ozsh@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5: Lag, correct get the port select mode str
Liu, Changcheng [Fri, 27 May 2022 08:49:14 +0000 (16:49 +0800)]
net/mlx5: Lag, correct get the port select mode str

mode & mode_flags is updated at the end of mlx5_activate_lag which
may not reflect the actual mode as shown in below logic:
    mlx5_activate_lag(struct mlx5_lag *ldev,
    |-- unsigned long flags = 0;
    |-- err = mlx5_lag_set_flags(ldev, mode, tracker, shared_fdb, &flags);
    |-- err = mlx5_create_lag(ldev, tracker, mode, flags);
              |-- mlx5_get_str_port_sel_mode(ldev);
    |-- ldev->mode = mode;
    |-- ldev->mode_flags = flags;
Use mode & flag as parameters to get port select mode info.

Fixes: 94db33177819 ("net/mlx5: Support multiport eswitch mode")
Signed-off-by: Liu, Changcheng <jerrliu@nvidia.com>
Reviewed-by: Eli Cohen <elic@nvidia.com>
Reviewed-by: Mark Bloch <mbloch@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5e: Fix enabling sriov while tc nic rules are offloaded
Paul Blakey [Wed, 22 Jun 2022 10:11:18 +0000 (13:11 +0300)]
net/mlx5e: Fix enabling sriov while tc nic rules are offloaded

There is a total of four 4M entries flow tables. In sriov disabled
mode, ct, ct_nat and post_act take three of them. When adding the
first tc nic rule in this mode, it will take another 4M table
for the tc <chain,prio> table. If user then enables sriov, the legacy
flow table tries to take another 4M and fails, and so enablement fails.

To fix that, have legacy fdb take the next available maximum
size from the fs ft pool.

Fixes: 4a98544d1827 ("net/mlx5: Move chains ft pool to be used by all firmware steering")
Signed-off-by: Paul Blakey <paulb@nvidia.com>
Reviewed-by: Roi Dayan <roid@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5e: kTLS, Fix build time constant test in RX
Tariq Toukan [Mon, 6 Jun 2022 18:21:10 +0000 (21:21 +0300)]
net/mlx5e: kTLS, Fix build time constant test in RX

Use the correct constant (TLS_DRIVER_STATE_SIZE_RX) in the comparison
against the size of the private RX TLS driver context.

Fixes: 1182f3659357 ("net/mlx5e: kTLS, Add kTLS RX HW offload support")
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
Reviewed-by: Maxim Mikityanskiy <maximmi@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5e: kTLS, Fix build time constant test in TX
Tariq Toukan [Mon, 6 Jun 2022 18:20:29 +0000 (21:20 +0300)]
net/mlx5e: kTLS, Fix build time constant test in TX

Use the correct constant (TLS_DRIVER_STATE_SIZE_TX) in the comparison
against the size of the private TX TLS driver context.

Fixes: df8d866770f9 ("net/mlx5e: kTLS, Use kernel API to extract private offload context")
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
Reviewed-by: Maxim Mikityanskiy <maximmi@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5: Lag, decouple FDB selection and shared FDB
Mark Bloch [Tue, 24 May 2022 09:08:10 +0000 (12:08 +0300)]
net/mlx5: Lag, decouple FDB selection and shared FDB

Multiport eswitch is required to use native FDB selection instead of
affinity, This was achieved by passing the shared_fdb flag down
the HW lag creation path. While it did accomplish the goal of setting
FDB selection mode to native, it had the side effect of also
creating a shared FDB configuration.

This created a few issues:
- TC rules are inserted into a non active FDB, which means traffic isn't
  offloaded as all traffic will reach only a single FDB.
- All wire traffic is treated as if a single physical port received it; while
  this is true for a bond configuration, this shouldn't be the case for
  multiport eswitch.

Create a new flag MLX5_LAG_MODE_FLAG_FDB_SEL_MODE_NATIVE
to indicate what FDB selection mode should be used.

Fixes: 94db33177819 ("net/mlx5: Support multiport eswitch mode")
Signed-off-by: Mark Bloch <mbloch@nvidia.com>
Reviewed-by: Eli Cohen <elic@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5: TC, allow offload from uplink to other PF's VF
Eli Cohen [Wed, 8 Jun 2022 14:34:16 +0000 (17:34 +0300)]
net/mlx5: TC, allow offload from uplink to other PF's VF

Redirecting traffic from uplink to a VF is a legal operation of
mulitport eswitch mode. Remove the limitation.

Fixes: 94db33177819 ("net/mlx5: Support multiport eswitch mode")
Signed-off-by: Eli Cohen <elic@nvidia.com>
Reviewed-by: Maor Dickman <maord@nvidia.com>
Reviewed-by: Mark Bloch <mbloch@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>