linux-2.6-microblaze.git
2 years agonet: caif: remove redundant assignment to variable expectlen
Colin Ian King [Tue, 28 Dec 2021 00:45:42 +0000 (00:45 +0000)]
net: caif: remove redundant assignment to variable expectlen

Variable expectlen is being assigned a value that is never read, the
assignment occurs before a return statement. The assignment is
redundant and can be removed.

Signed-off-by: Colin Ian King <colin.i.king@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet:Remove initialization of static variables to 0
Wen Zhiwei [Mon, 27 Dec 2021 08:22:01 +0000 (16:22 +0800)]
net:Remove initialization of static variables to 0

Delete the initialization of three static variables
because it is meaningless.

Signed-off-by: Wen Zhiwei <wenzhiwei@kylinos.cn>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: bridge: Get SIOCGIFBR/SIOCSIFBR ioctl working in compat mode
Remi Pommarel [Fri, 24 Dec 2021 11:46:40 +0000 (12:46 +0100)]
net: bridge: Get SIOCGIFBR/SIOCSIFBR ioctl working in compat mode

In compat mode SIOC{G,S}IFBR ioctls were only supporting
BRCTL_GET_VERSION returning an artificially version to spur userland
tool to use SIOCDEVPRIVATE instead. But some userland tools ignore that
and use SIOC{G,S}IFBR unconditionally as seen with busybox's brctl.

Example of non working 32-bit brctl with CONFIG_COMPAT=y:
$ brctl show
brctl: SIOCGIFBR: Invalid argument

Example of fixed 32-bit brctl with CONFIG_COMPAT=y:
$ brctl show
bridge name     bridge id               STP enabled     interfaces
br0

Signed-off-by: Remi Pommarel <repk@triplefau.lt>
Co-developed-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Acked-by: Nikolay Aleksandrov <nikolay@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: ethernet: ti: davinci_emac: Use platform_get_irq() to get the interrupt
Lad Prabhakar [Fri, 24 Dec 2021 19:26:26 +0000 (19:26 +0000)]
net: ethernet: ti: davinci_emac: Use platform_get_irq() to get the interrupt

platform_get_resource(pdev, IORESOURCE_IRQ, ..) relies on static
allocation of IRQ resources in DT core code, this causes an issue
when using hierarchical interrupt domains using "interrupts" property
in the node as this bypasses the hierarchical setup and messes up the
irq chaining.

In preparation for removal of static setup of IRQ resource from DT core
code use platform_get_irq() for DT users only.

While at it propagate error code in case request_irq() fails instead of
returning -EBUSY.

Signed-off-by: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: xilinx: emaclite: Use platform_get_irq() to get the interrupt
Lad Prabhakar [Fri, 24 Dec 2021 19:26:23 +0000 (19:26 +0000)]
net: xilinx: emaclite: Use platform_get_irq() to get the interrupt

platform_get_resource(pdev, IORESOURCE_IRQ, ..) relies on static
allocation of IRQ resources in DT core code, this causes an issue
when using hierarchical interrupt domains using "interrupts" property
in the node as this bypasses the hierarchical setup and messes up the
irq chaining.

In preparation for removal of static setup of IRQ resource from DT core
code use platform_get_irq().

Signed-off-by: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: ethoc: Use platform_get_irq() to get the interrupt
Lad Prabhakar [Fri, 24 Dec 2021 19:26:22 +0000 (19:26 +0000)]
net: ethoc: Use platform_get_irq() to get the interrupt

platform_get_resource(pdev, IORESOURCE_IRQ, ..) relies on static
allocation of IRQ resources in DT core code, this causes an issue
when using hierarchical interrupt domains using "interrupts" property
in the node as this bypasses the hierarchical setup and messes up the
irq chaining.

In preparation for removal of static setup of IRQ resource from DT core
code use platform_get_irq().

Signed-off-by: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agofsl/fman: Use platform_get_irq() to get the interrupt
Lad Prabhakar [Fri, 24 Dec 2021 19:26:21 +0000 (19:26 +0000)]
fsl/fman: Use platform_get_irq() to get the interrupt

platform_get_resource(pdev, IORESOURCE_IRQ, ..) relies on static
allocation of IRQ resources in DT core code, this causes an issue
when using hierarchical interrupt domains using "interrupts" property
in the node as this bypasses the hierarchical setup and messes up the
irq chaining.

In preparation for removal of static setup of IRQ resource from DT core
code use platform_get_irq(). While doing so return error pointer
from read_dts_node() as platform_get_irq() may return -EPROBE_DEFER.

Signed-off-by: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: pxa168_eth: Use platform_get_irq() to get the interrupt
Lad Prabhakar [Fri, 24 Dec 2021 19:26:20 +0000 (19:26 +0000)]
net: pxa168_eth: Use platform_get_irq() to get the interrupt

platform_get_resource(pdev, IORESOURCE_IRQ, ..) relies on static
allocation of IRQ resources in DT core code, this causes an issue
when using hierarchical interrupt domains using "interrupts" property
in the node as this bypasses the hierarchical setup and messes up the
irq chaining.

In preparation for removal of static setup of IRQ resource from DT core
code use platform_get_irq().

Signed-off-by: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoethernet: netsec: Use platform_get_irq() to get the interrupt
Lad Prabhakar [Fri, 24 Dec 2021 19:26:19 +0000 (19:26 +0000)]
ethernet: netsec: Use platform_get_irq() to get the interrupt

platform_get_resource(pdev, IORESOURCE_IRQ, ..) relies on static
allocation of IRQ resources in DT core code, this causes an issue
when using hierarchical interrupt domains using "interrupts" property
in the node as this bypasses the hierarchical setup and messes up the
irq chaining.

In preparation for removal of static setup of IRQ resource from DT core
code use platform_get_irq().

Signed-off-by: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: wwan: iosm: Keep device at D0 for s2idle case
Kai-Heng Feng [Fri, 24 Dec 2021 08:19:14 +0000 (16:19 +0800)]
net: wwan: iosm: Keep device at D0 for s2idle case

We are seeing spurious wakeup caused by Intel 7560 WWAN on AMD laptops.
This prevent those laptops to stay in s2idle state.

>From what I can understand, the intention of ipc_pcie_suspend() is to
put the device to D3cold, and ipc_pcie_suspend_s2idle() is to keep the
device at D0. However, the device can still be put to D3hot/D3cold by
PCI core.

So explicitly let PCI core know this device should stay at D0, to solve
the spurious wakeup.

Signed-off-by: Kai-Heng Feng <kai.heng.feng@canonical.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: wwan: iosm: Let PCI core handle PCI power transition
Kai-Heng Feng [Fri, 24 Dec 2021 08:19:13 +0000 (16:19 +0800)]
net: wwan: iosm: Let PCI core handle PCI power transition

pci_pm_suspend_noirq() and pci_pm_resume_noirq() already handle power
transition for system-wide suspend and resume, so it's not necessary to
do it in the driver.

Signed-off-by: Kai-Heng Feng <kai.heng.feng@canonical.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: lan966x: Fix the vlan used by host ports
Horatiu Vultur [Thu, 23 Dec 2021 14:01:13 +0000 (15:01 +0100)]
net: lan966x: Fix the vlan used by host ports

The blamed commit changed the vlan used by the host ports to be 4095
instead of 0.
Because of this change the following issues are seen:
- when the port is probed first it was adding an entry in the MAC table
  with the wrong vlan (port->pvid which is default 0) and not HOST_PVID
- when the port is removed from a bridge, it was using the wrong vlan to
  add entries in the MAC table. It was using the old PVID and not the
  HOST_PVID

This patch fixes this two issues by using the HOST_PVID instead of
port->pvid.

Fixes: 6d2c186afa5d5d ("net: lan966x: Add vlan support.")
Signed-off-by: Horatiu Vultur <horatiu.vultur@microchip.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoMerge branch 'bnxt_en-next'
David S. Miller [Mon, 27 Dec 2021 12:00:29 +0000 (12:00 +0000)]
Merge branch 'bnxt_en-next'

Michael Chan says:

====================
bnxt_en: Update for net-next

This series includes some added error logging for firmware error reports,
DIM interrupt sampling for the latest 5750X chips, CQE interrupt
coalescing mode support, and to use RX page frag buffers for better
software GRO performance.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agobnxt_en: Use page frag RX buffers for better software GRO performance
Jakub Kicinski [Mon, 27 Dec 2021 08:00:32 +0000 (03:00 -0500)]
bnxt_en: Use page frag RX buffers for better software GRO performance

If NETIF_F_GRO_HW is disabled, the existing driver code uses kmalloc'ed
data for RX buffers.  This causes inefficient SW GRO performance
because the GRO data is merged using the less efficient frag_list.
Use netdev_alloc_frag() and friends instead so that GRO data can be
merged into skb_shinfo(skb)->frags for better performance.

[Use skb_free_frag() - Vikas Gupta]

Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Reviewed-by: Pavan Chebbi <pavan.chebbi@broadcom.com>
Reviewed-by: Andy Gospodarek <gospo@broadcom.com>
Signed-off-by: Vikas Gupta <vikas.gupta@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agobnxt_en: convert to xdp_do_flush
Edwin Peer [Mon, 27 Dec 2021 08:00:31 +0000 (03:00 -0500)]
bnxt_en: convert to xdp_do_flush

The xdp_do_flush_map function has been replaced with the more general
xdp_do_flush().

Signed-off-by: Edwin Peer <edwin.peer@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agobnxt_en: Support CQE coalescing mode in ethtool
Michael Chan [Mon, 27 Dec 2021 08:00:30 +0000 (03:00 -0500)]
bnxt_en: Support CQE coalescing mode in ethtool

Support showing and setting the CQE mode in ethtool.

Reviewed-by: Edwin Peer <edwin.peer@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agobnxt_en: Support configurable CQE coalescing mode
Michael Chan [Mon, 27 Dec 2021 08:00:29 +0000 (03:00 -0500)]
bnxt_en: Support configurable CQE coalescing mode

CQE coalescing mode is the same as the timer reset coalescing mode
on Broadcom devices.  Currently this mode is always enabled if it
is supported by the device.  Restructure the code slightly to support
dynamically changing this mode.

Add a flags field to struct bnxt_coal.  Initially, the CQE flag will
be set for the RX and TX side if the device supports it.  We need to
move bnxt_init_dflt_coal() to set up default coalescing until the
capability is determined.

Reviewed-by: Andy Gospodarek <gospo@broadcom.com>
Reviewed-by: Edwin Peer <edwin.peer@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agobnxt_en: enable interrupt sampling on 5750X for DIM
Andy Gospodarek [Mon, 27 Dec 2021 08:00:28 +0000 (03:00 -0500)]
bnxt_en: enable interrupt sampling on 5750X for DIM

5750X (P5) chips handle receiving packets on the NQ rather than the main
completion queue so we need to get and set stats from the correct spots
for dynamic interrupt moderation.

Signed-off-by: Andy Gospodarek <gospo@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agobnxt_en: Log error report for dropped doorbell
Michael Chan [Mon, 27 Dec 2021 08:00:27 +0000 (03:00 -0500)]
bnxt_en: Log error report for dropped doorbell

Log the unrecognized error report type value as well.

Reviewed-by: Andy Gospodarek <gospo@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agobnxt_en: Add event handler for PAUSE Storm event
Somnath Kotur [Mon, 27 Dec 2021 08:00:26 +0000 (03:00 -0500)]
bnxt_en: Add event handler for PAUSE Storm event

FW has been modified to send a new async event when it detects
a pause storm. Register for this new event and log it upon receipt.

Reviewed-by: Andy Gospodarek <gospo@broadcom.com>
Reviewed-by: Edwin Peer <edwin.peer@broadcom.com>
Signed-off-by: Somnath Kotur <somnath.kotur@broadcom.com>
Signed-off-by: Michael Chan <michael.chan@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agolib: objagg: Use the bitmap API when applicable
Christophe JAILLET [Thu, 23 Dec 2021 21:33:42 +0000 (22:33 +0100)]
lib: objagg: Use the bitmap API when applicable

Use 'bitmap_zalloc()' to simplify code, improve the semantic and reduce
some open-coded arithmetic in allocator arguments.

Also change the corresponding 'kfree()' into 'bitmap_free()' to keep
consistency.

Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Link: https://lore.kernel.org/r/f9541b085ec68e573004e1be200c11c9c901181a.1640295165.git.christophe.jaillet@wanadoo.fr
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agonet: phy: micrel: Add config_init for LAN8814
Horatiu Vultur [Thu, 23 Dec 2021 08:28:26 +0000 (09:28 +0100)]
net: phy: micrel: Add config_init for LAN8814

Add config_init for LAN8814. This function is required for the following
reasons:
- we need to make sure that the PHY is reset,
- disable ANEG with QSGMII PCS Host side
- swap the MDI-X A,B transmit so that there will not be any link flip-flaps
  when the PHY gets a link.

Signed-off-by: Horatiu Vultur <horatiu.vultur@microchip.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: wan/lmc: fix spelling of "its"
Randy Dunlap [Thu, 23 Dec 2021 06:26:11 +0000 (22:26 -0800)]
net: wan/lmc: fix spelling of "its"

Use the possessive "its" instead of the contraction of "it is" ("it's")
in user messages.

Signed-off-by: Randy Dunlap <rdunlap@infradead.org>
Cc: Krzysztof Halasa <khc@pm.waw.pl>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net
Jakub Kicinski [Fri, 24 Dec 2021 00:09:58 +0000 (16:09 -0800)]
Merge git://git./linux/kernel/git/netdev/net

include/net/sock.h
  commit 8f905c0e7354 ("inet: fully convert sk->sk_rx_dst to RCU rules")
  commit 43f51df41729 ("net: move early demux fields close to sk_refcnt")
  https://lore.kernel.org/all/20211222141641.0caa0ab3@canb.auug.org.au/

Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agonet: stmmac: dwmac-visconti: Fix value of ETHER_CLK_SEL_FREQ_SEL_2P5M
Nobuhiro Iwamatsu [Thu, 23 Dec 2021 07:36:33 +0000 (16:36 +0900)]
net: stmmac: dwmac-visconti: Fix value of ETHER_CLK_SEL_FREQ_SEL_2P5M

ETHER_CLK_SEL_FREQ_SEL_2P5M is not 0 bit of the register. This is a
value, which is 0. Fix from BIT(0) to 0.

Reported-by: Yuji Ishikawa <yuji2.ishikawa@toshiba.co.jp>
Fixes: b38dd98ff8d0 ("net: stmmac: Add Toshiba Visconti SoCs glue driver")
Signed-off-by: Nobuhiro Iwamatsu <nobuhiro1.iwamatsu@toshiba.co.jp>
Link: https://lore.kernel.org/r/20211223073633.101306-1-nobuhiro1.iwamatsu@toshiba.co.jp
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoMerge branch 'r8152-fix-bugs'
Jakub Kicinski [Thu, 23 Dec 2021 17:56:08 +0000 (09:56 -0800)]
Merge branch 'r8152-fix-bugs'

Hayes Wang says:

====================
r8152: fix bugs

Patch #1 fix the issue of force speed mode for RTL8156.
Patch #2 fix the issue of unexpected ocp_base.
====================

Link: https://lore.kernel.org/r/20211223092702.23841-386-nic_swsd@realtek.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agor8152: sync ocp base
Hayes Wang [Thu, 23 Dec 2021 09:27:02 +0000 (17:27 +0800)]
r8152: sync ocp base

There are some chances that the actual base of hardware is different
from the value recorded by driver, so we have to reset the variable
of ocp_base to sync it.

Set ocp_base to -1. Then, it would be updated and the new base would be
set to the hardware next time.

Signed-off-by: Hayes Wang <hayeswang@realtek.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agor8152: fix the force speed doesn't work for RTL8156
Hayes Wang [Thu, 23 Dec 2021 09:27:01 +0000 (17:27 +0800)]
r8152: fix the force speed doesn't work for RTL8156

It needs to set mdio force mode. Otherwise, link off always occurs when
setting force speed.

Fixes: 195aae321c82 ("r8152: support new chips")
Signed-off-by: Hayes Wang <hayeswang@realtek.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agonet: bridge: fix ioctl old_deviceless bridge argument
Remi Pommarel [Thu, 23 Dec 2021 15:31:38 +0000 (16:31 +0100)]
net: bridge: fix ioctl old_deviceless bridge argument

Commit 561d8352818f ("bridge: use ndo_siocdevprivate") changed the
source and destination arguments of copy_{to,from}_user in bridge's
old_deviceless() from args[1] to uarg breaking SIOC{G,S}IFBR ioctls.

Commit cbd7ad29a507 ("net: bridge: fix ioctl old_deviceless bridge
argument") fixed only BRCTL_{ADD,DEL}_BRIDGES commands leaving
BRCTL_GET_BRIDGES one untouched.

The fixes BRCTL_GET_BRIDGES as well and has been tested with busybox's
brctl.

Example of broken brctl:
$ brctl show
bridge name     bridge id               STP enabled     interfaces
brctl: can't get bridge name for index 0: No such device or address

Example of fixed brctl:
$ brctl show
bridge name     bridge id               STP enabled     interfaces
br0             8000.000000000000       no

Fixes: 561d8352818f ("bridge: use ndo_siocdevprivate")
Signed-off-by: Remi Pommarel <repk@triplefau.lt>
Reviewed-by: Arnd Bergmann <arnd@arndb.de>
Acked-by: Nikolay Aleksandrov <nikolay@nvidia.com>
Link: https://lore.kernel.org/all/20211223153139.7661-2-repk@triplefau.lt/
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agonet: stmmac: ptp: fix potentially overflowing expression
Xiaoliang Yang [Thu, 23 Dec 2021 07:39:28 +0000 (15:39 +0800)]
net: stmmac: ptp: fix potentially overflowing expression

Convert the u32 variable to type u64 in a context where expression of
type u64 is required to avoid potential overflow.

Fixes: e9e3720002f6 ("net: stmmac: ptp: update tas basetime after ptp adjust")
Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com>
Link: https://lore.kernel.org/r/20211223073928.37371-1-xiaoliang.yang_1@nxp.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agonet: dsa: tag_ocelot: use traffic class to map priority on injected header
Xiaoliang Yang [Thu, 23 Dec 2021 07:22:11 +0000 (15:22 +0800)]
net: dsa: tag_ocelot: use traffic class to map priority on injected header

For Ocelot switches, the CPU injected frames have an injection header
where it can specify the QoS class of the packet and the DSA tag, now it
uses the SKB priority to set that. If a traffic class to priority
mapping is configured on the netdevice (with mqprio for example ...), it
won't be considered for CPU injected headers. This patch make the QoS
class aligned to the priority to traffic class mapping if it exists.

Fixes: 8dce89aa5f32 ("net: dsa: ocelot: add tagger for Ocelot/Felix switches")
Signed-off-by: Xiaoliang Yang <xiaoliang.yang_1@nxp.com>
Signed-off-by: Marouen Ghodhbane <marouen.ghodhbane@nxp.com>
Link: https://lore.kernel.org/r/20211223072211.33130-1-xiaoliang.yang_1@nxp.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoveth: ensure skb entering GRO are not cloned.
Paolo Abeni [Wed, 22 Dec 2021 18:39:52 +0000 (19:39 +0100)]
veth: ensure skb entering GRO are not cloned.

After commit d3256efd8e8b ("veth: allow enabling NAPI even without XDP"),
if GRO is enabled on a veth device and TSO is disabled on the peer
device, TCP skbs will go through the NAPI callback. If there is no XDP
program attached, the veth code does not perform any share check, and
shared/cloned skbs could enter the GRO engine.

Ignat reported a BUG triggered later-on due to the above condition:

[   53.970529][    C1] kernel BUG at net/core/skbuff.c:3574!
[   53.981755][    C1] invalid opcode: 0000 [#1] PREEMPT SMP KASAN PTI
[   53.982634][    C1] CPU: 1 PID: 19 Comm: ksoftirqd/1 Not tainted 5.16.0-rc5+ #25
[   53.982634][    C1] Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 0.0.0 02/06/2015
[   53.982634][    C1] RIP: 0010:skb_shift+0x13ef/0x23b0
[   53.982634][    C1] Code: ea 03 0f b6 04 02 48 89 fa 83 e2 07 38 d0
7f 08 84 c0 0f 85 41 0c 00 00 41 80 7f 02 00 4d 8d b5 d0 00 00 00 0f
85 74 f5 ff ff <0f> 0b 4d 8d 77 20 be 04 00 00 00 4c 89 44 24 78 4c 89
f7 4c 89 8c
[   53.982634][    C1] RSP: 0018:ffff8881008f7008 EFLAGS: 00010246
[   53.982634][    C1] RAX: 0000000000000000 RBX: ffff8881180b4c80 RCX: 0000000000000000
[   53.982634][    C1] RDX: 0000000000000002 RSI: ffff8881180b4d3c RDI: ffff88810bc9cac2
[   53.982634][    C1] RBP: ffff8881008f70b8 R08: ffff8881180b4cf4 R09: ffff8881180b4cf0
[   53.982634][    C1] R10: ffffed1022999e5c R11: 0000000000000002 R12: 0000000000000590
[   53.982634][    C1] R13: ffff88810f940c80 R14: ffff88810f940d50 R15: ffff88810bc9cac0
[   53.982634][    C1] FS:  0000000000000000(0000) GS:ffff888235880000(0000) knlGS:0000000000000000
[   53.982634][    C1] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[   53.982634][    C1] CR2: 00007ff5f9b86680 CR3: 0000000108ce8004 CR4: 0000000000170ee0
[   53.982634][    C1] Call Trace:
[   53.982634][    C1]  <TASK>
[   53.982634][    C1]  tcp_sacktag_walk+0xaba/0x18e0
[   53.982634][    C1]  tcp_sacktag_write_queue+0xe7b/0x3460
[   53.982634][    C1]  tcp_ack+0x2666/0x54b0
[   53.982634][    C1]  tcp_rcv_established+0x4d9/0x20f0
[   53.982634][    C1]  tcp_v4_do_rcv+0x551/0x810
[   53.982634][    C1]  tcp_v4_rcv+0x22ed/0x2ed0
[   53.982634][    C1]  ip_protocol_deliver_rcu+0x96/0xaf0
[   53.982634][    C1]  ip_local_deliver_finish+0x1e0/0x2f0
[   53.982634][    C1]  ip_sublist_rcv_finish+0x211/0x440
[   53.982634][    C1]  ip_list_rcv_finish.constprop.0+0x424/0x660
[   53.982634][    C1]  ip_list_rcv+0x2c8/0x410
[   53.982634][    C1]  __netif_receive_skb_list_core+0x65c/0x910
[   53.982634][    C1]  netif_receive_skb_list_internal+0x5f9/0xcb0
[   53.982634][    C1]  napi_complete_done+0x188/0x6e0
[   53.982634][    C1]  gro_cell_poll+0x10c/0x1d0
[   53.982634][    C1]  __napi_poll+0xa1/0x530
[   53.982634][    C1]  net_rx_action+0x567/0x1270
[   53.982634][    C1]  __do_softirq+0x28a/0x9ba
[   53.982634][    C1]  run_ksoftirqd+0x32/0x60
[   53.982634][    C1]  smpboot_thread_fn+0x559/0x8c0
[   53.982634][    C1]  kthread+0x3b9/0x490
[   53.982634][    C1]  ret_from_fork+0x22/0x30
[   53.982634][    C1]  </TASK>

Address the issue by skipping the GRO stage for shared or cloned skbs.
To reduce the chance of OoO, try to unclone the skbs before giving up.

v1 -> v2:
 - use avoid skb_copy and fallback to netif_receive_skb  - Eric

Reported-by: Ignat Korchagin <ignat@cloudflare.com>
Fixes: d3256efd8e8b ("veth: allow enabling NAPI even without XDP")
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Tested-by: Ignat Korchagin <ignat@cloudflare.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Link: https://lore.kernel.org/r/b5f61c5602aab01bac8d711d8d1bfab0a4817db7.1640197544.git.pabeni@redhat.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoMerge tag 'wireless-drivers-next-2021-12-23' of git://git.kernel.org/pub/scm/linux...
Jakub Kicinski [Thu, 23 Dec 2021 17:12:36 +0000 (09:12 -0800)]
Merge tag 'wireless-drivers-next-2021-12-23' of git://git./linux/kernel/git/kvalo/wireless-drivers-next

Kalle Valo says:

====================
wireless-drivers-next patches for v5.17

Third set of patches for v5.17, and the final one if all goes well. We
have Specific Absorption Rate (SAR) support for both mt76 and rtw88.
Also iwlwifi should be now W=1 warning free. But otherwise nothing
really special this time, business as usual.

Major changes:

mt76
 * Specific Absorption Rate (SAR) support
 * mt7921: new PCI ids
 * mt7921: 160 MHz channel support

iwlwifi
 * fix W=1 and sparse warnings
 * BNJ hardware support
 * add new killer device ids
 * support for Time-Aware-SAR (TAS) from the BIOS
 * Optimized Connectivity Experience (OCE) scan support

rtw88
 * hardware scan
 * Specific Absorption Rate (SAR) support

ath11k
 * qca6390/wcn6855: report signal and tx bitrate
 * qca6390: rfkill support
 * qca6390/wcn6855: regdb.bin support

ath5k
 * switch to rate table based lookup

wilc1000
 * spi: reset/enable GPIO support

* tag 'wireless-drivers-next-2021-12-23' of git://git.kernel.org/pub/scm/linux/kernel/git/kvalo/wireless-drivers-next: (148 commits)
  mt76: mt7921: fix a possible race enabling/disabling runtime-pm
  wilc1000: Document enable-gpios and reset-gpios properties
  wilc1000: Add reset/enable GPIO support to SPI driver
  wilc1000: Convert static "chipid" variable to device-local variable
  rtw89: 8852a: correct bit definition of dfs_en
  rtw88: don't consider deep PS mode when transmitting packet
  ath11k: Fix unexpected return buffer manager error for QCA6390
  ath11k: add support of firmware logging for WCN6855
  ath11k: Fix napi related hang
  ath10k: replace strlcpy with strscpy
  rtw88: support SAR via kernel common API
  rtw88: 8822c: add ieee80211_ops::hw_scan
  iwlwifi: mei: wait before mapping the shared area
  iwlwifi: mei: clear the ownership when the driver goes down
  iwlwifi: yoyo: fix issue with new DBGI_SRAM region read.
  iwlwifi: fw: fix some scan kernel-doc
  iwlwifi: pcie: make sure prph_info is set when treating wakeup IRQ
  iwlwifi: mvm: remove card state notification code
  iwlwifi: mvm: drop too short packets silently
  iwlwifi: mvm: fix AUX ROC removal
  ...
====================

Link: https://lore.kernel.org/r/20211223141108.78808C36AE9@smtp.kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoMerge branch 'mlxsw-tests'
David S. Miller [Thu, 23 Dec 2021 11:24:01 +0000 (11:24 +0000)]
Merge branch 'mlxsw-tests'

Amit Cohen says:

====================
mlxsw: Add tests for VxLAN with IPv6 underlay

mlxsw driver lately added support for VxLAN with IPv6 underlay.
This set adds tests for IPv6, which are dedicated for mlxsw.

Patch set overview:
Patches #1-#2 make vxlan.sh test more flexible and extend it for IPv6
Patches #3-#4 make vxlan_fdb_veto.sh test more flexible and extend it
for IPv6
Patches #5-#6 add tests for VxLAN flooding for different ASICs
Patches #7-#8 add test for VxLAN related traps and align the existing
test
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoselftests: mlxsw: devlink_trap_tunnel_vxlan: Fix 'decap_error' case
Amit Cohen [Thu, 23 Dec 2021 07:30:02 +0000 (09:30 +0200)]
selftests: mlxsw: devlink_trap_tunnel_vxlan: Fix 'decap_error' case

Change the case that sends packets with "too short inner packet" to
include part of ethernet header, to make the trap to be triggered due to
the correct reason.

According to ASIC arch, the trap is triggered if overlay packet length is
less than 18B, and the minimum inner packet should include source MAC and
destination MAC.

Till now the case passed because one of the reserved bits in VxLAN
header was used. This issue was found while adding an equivalent test
for IPv6.

Signed-off-by: Amit Cohen <amcohen@nvidia.com>
Reviewed-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoselftests: mlxsw: Add test for VxLAN related traps for IPv6
Amit Cohen [Thu, 23 Dec 2021 07:30:01 +0000 (09:30 +0200)]
selftests: mlxsw: Add test for VxLAN related traps for IPv6

The test configures VxLAN with IPv6 underlay and verifies that the
expected traps are triggered under the right conditions.

The test is similar to the existing IPv4 test.

Signed-off-by: Amit Cohen <amcohen@nvidia.com>
Reviewed-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoselftests: mlxsw: spectrum-2: Add a test for VxLAN flooding with IPv6
Amit Cohen [Thu, 23 Dec 2021 07:30:00 +0000 (09:30 +0200)]
selftests: mlxsw: spectrum-2: Add a test for VxLAN flooding with IPv6

The device stores flood records in a singly linked list where each
record stores up to X IP addresses of remote VTEPs.

The number of records is changed according to ASIC type and address
family.

Add a test which is similar to the existing IPv4 test to check IPv6.
The test is dedicated for Spectrum-2 and above, which support up to four
IPv6 addresses in one record.

The test verifies that packets are correctly flooded in various cases such
as deletion of a record in the middle of the list.

Signed-off-by: Amit Cohen <amcohen@nvidia.com>
Reviewed-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoselftests: mlxsw: spectrum: Add a test for VxLAN flooding with IPv6
Amit Cohen [Thu, 23 Dec 2021 07:29:59 +0000 (09:29 +0200)]
selftests: mlxsw: spectrum: Add a test for VxLAN flooding with IPv6

The device stores flood records in a singly linked list where each
record stores up to X IP addresses of remote VTEPs.

The number of records is changed according to ASIC type and address
family.

Add a test which is similar to the existing IPv4 test to check IPv6.
The test is dedicated for Spectrum-1 switches, which support up to five
IPv6 addresses in one record.

The test verifies that packets are correctly flooded in various cases such
as deletion of a record in the middle of the list.

Signed-off-by: Amit Cohen <amcohen@nvidia.com>
Reviewed-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoselftests: mlxsw: Add VxLAN FDB veto test for IPv6
Amit Cohen [Thu, 23 Dec 2021 07:29:58 +0000 (09:29 +0200)]
selftests: mlxsw: Add VxLAN FDB veto test for IPv6

Add test to verify FDB vetos of VxLAN with IPv6 underlay.
Use the existing test which checks IPv4.

Signed-off-by: Amit Cohen <amcohen@nvidia.com>
Reviewed-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoselftests: mlxsw: vxlan_fdb_veto: Make the test more flexible for future use
Amit Cohen [Thu, 23 Dec 2021 07:29:57 +0000 (09:29 +0200)]
selftests: mlxsw: vxlan_fdb_veto: Make the test more flexible for future use

vxlan_fdb_veto.sh cases are dedicated to test VxLAN with IPv4 underlay.
The main changes to test IPv6 underlay are IP addresses and some flags.

Add variables to define all the values which supposed to be different
for IPv6 testing, set them to use the existing values by default.

The next patch will define the new added variables in a separated file,
so the same tests can be used for IPv6 also.

Signed-off-by: Amit Cohen <amcohen@nvidia.com>
Reviewed-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoselftests: mlxsw: Add VxLAN configuration test for IPv6
Amit Cohen [Thu, 23 Dec 2021 07:29:56 +0000 (09:29 +0200)]
selftests: mlxsw: Add VxLAN configuration test for IPv6

Add test to verify configuration of VxLAN with IPv6 underlay.
Use the existing test which checks IPv4.

Add separated test cases for learning which is not supported for IPv6
and for UDP checksum flags which are different from IPv4 flags.

Signed-off-by: Amit Cohen <amcohen@nvidia.com>
Reviewed-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoselftests: mlxsw: vxlan: Make the test more flexible for future use
Amit Cohen [Thu, 23 Dec 2021 07:29:55 +0000 (09:29 +0200)]
selftests: mlxsw: vxlan: Make the test more flexible for future use

vxlan.sh cases are dedicated to test VxLAN with IPv4 underlay.
The main changes to test IPv6 underlay are IP addresses and some flags.

Add variables to define all the values which supposed to be different
for IPv6 testing, set them to use the existing values by default.

The next patch will define the new added variables in a separated file,
so the same tests can be used for IPv6 also.

Rename some functions to include "ipv4", so the next patch will add
equivalent functions for IPv6.

Signed-off-by: Amit Cohen <amcohen@nvidia.com>
Reviewed-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agowwan: Replace kernel.h with the necessary inclusions
Andy Shevchenko [Wed, 22 Dec 2021 16:32:56 +0000 (18:32 +0200)]
wwan: Replace kernel.h with the necessary inclusions

When kernel.h is used in the headers it adds a lot into dependency hell,
especially when there are circular dependencies are involved.

Replace kernel.h inclusion with the list of what is really being used.

Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Reviewed-by: Sergey Ryazanov <ryazanov.s.a@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: stmmac: add tc flower filter for EtherType matching
Ong Boon Leong [Wed, 22 Dec 2021 14:43:10 +0000 (22:43 +0800)]
net: stmmac: add tc flower filter for EtherType matching

This patch adds basic support for EtherType RX frame steering for
LLDP and PTP using the hardware offload capabilities.

Example steps for setting up RX frame steering for LLDP and PTP:
$ IFDEVNAME=eth0
$ tc qdisc add dev $IFDEVNAME ingress
$ tc qdisc add dev $IFDEVNAME root mqprio num_tc 8 \
     map 0 1 2 3 4 5 6 7 0 0 0 0 0 0 0 0 \
     queues 1@0 1@1 1@2 1@3 1@4 1@5 1@6 1@7 hw 0

For LLDP
$ tc filter add dev $IFDEVNAME parent ffff: protocol 0x88cc \
     flower hw_tc 5
OR
$ tc filter add dev $IFDEVNAME parent ffff: protocol LLDP \
     flower hw_tc 5

For PTP
$ tc filter add dev $IFDEVNAME parent ffff: protocol 0x88f7 \
     flower hw_tc 6

Show tc ingress filter
$ tc filter show dev $IFDEVNAME ingress

v1->v2:
 Thanks to Kurt's and Sebastian's suggestion.
 - change from __be16 to u16 etype
 - change ETHER_TYPE_FULL_MASK to use cpu_to_be16() macro

Signed-off-by: Ong Boon Leong <boon.leong.ong@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agonet: lan966x: Add support for multiple bridge flags
Horatiu Vultur [Wed, 22 Dec 2021 11:07:59 +0000 (12:07 +0100)]
net: lan966x: Add support for multiple bridge flags

This patch series extends the current supported bridge flags with the
following flags: BR_FLOOD, BR_BCAST_FLOOD and BR_LEARNING.

Signed-off-by: Horatiu Vultur <horatiu.vultur@microchip.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoflow_offload: fix suspicious RCU usage when offloading tc action
Baowen Zheng [Wed, 22 Dec 2021 04:25:46 +0000 (12:25 +0800)]
flow_offload: fix suspicious RCU usage when offloading tc action

Fix suspicious rcu_dereference_protected() usage when offloading tc action.

We should hold tcfa_lock to offload tc action in action initiation.

Without these changes, the following warning will be observed:

WARNING: suspicious RCU usage
5.16.0-rc5-net-next-01504-g7d1f236dcffa-dirty #50 Tainted: G          I
-----------------------------
include/net/tc_act/tc_tunnel_key.h:33 suspicious rcu_dereference_protected() usage!
1 lock held by tc/12108:
CPU: 4 PID: 12108 Comm: tc Tainted: G
Hardware name: Dell Inc. PowerEdge R740/07WCGN, BIOS 1.6.11 11/20/2018
Call Trace:
<TASK>
dump_stack_lvl+0x49/0x5e
dump_stack+0x10/0x12
lockdep_rcu_suspicious+0xed/0xf8
tcf_tunnel_key_offload_act_setup+0x1de/0x300 [act_tunnel_key]
tcf_action_offload_add_ex+0xc0/0x1f0
tcf_action_init+0x26a/0x2f0
tcf_action_add+0xa9/0x1f0
tc_ctl_action+0xfb/0x170
rtnetlink_rcv_msg+0x169/0x510
? sched_clock+0x9/0x10
? rtnl_newlink+0x70/0x70
netlink_rcv_skb+0x55/0x100
rtnetlink_rcv+0x15/0x20
netlink_unicast+0x1a8/0x270
netlink_sendmsg+0x245/0x490
sock_sendmsg+0x65/0x70
____sys_sendmsg+0x219/0x260
? __import_iovec+0x2c/0x150
___sys_sendmsg+0xb7/0x100
? __lock_acquire+0x3d5/0x1f40
? __this_cpu_preempt_check+0x13/0x20
? lock_is_held_type+0xe4/0x140
? sched_clock+0x9/0x10
? ktime_get_coarse_real_ts64+0xbe/0xd0
? __this_cpu_preempt_check+0x13/0x20
? lockdep_hardirqs_on+0x7e/0x100
? ktime_get_coarse_real_ts64+0xbe/0xd0
? trace_hardirqs_on+0x2a/0xf0
__sys_sendmsg+0x5a/0xa0
? syscall_trace_enter.constprop.0+0x1dd/0x220
__x64_sys_sendmsg+0x1f/0x30
do_syscall_64+0x3b/0x90
entry_SYSCALL_64_after_hwframe+0x44/0xae
RIP: 0033:0x7f4db7bb7a60

Fixes: 8cbfe939abe9 ("flow_offload: allow user to offload tc action to net device")
Reported-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: Baowen Zheng <baowen.zheng@corigine.com>
Signed-off-by: Louis Peens <louis.peens@corigine.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agosctp: move hlist_node and hashent out of sctp_ep_common
Xin Long [Tue, 21 Dec 2021 21:40:30 +0000 (16:40 -0500)]
sctp: move hlist_node and hashent out of sctp_ep_common

Struct sctp_ep_common is included in both asoc and ep, but hlist_node
and hashent are only needed by ep after asoc_hashtable was dropped by
Commit b5eff7128366 ("sctp: drop the old assoc hashtable of sctp").

So it is better to move hlist_node and hashent from sctp_ep_common to
sctp_endpoint, and it saves some space for each asoc.

Signed-off-by: Xin Long <lucien.xin@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoMerge branch 'add-tests-for-vxlan-with-ipv6-underlay'
Jakub Kicinski [Thu, 23 Dec 2021 01:14:44 +0000 (17:14 -0800)]
Merge branch 'add-tests-for-vxlan-with-ipv6-underlay'

Amit Cohen says:

====================
Add tests for VxLAN with IPv6 underlay

mlxsw driver lately added support for VxLAN with IPv6 underlay.
This set adds the relevant tests for IPv6, most of them are same to
IPv4 tests with the required changes.

Patch set overview:
Patch #1 relaxes requirements for offloading TC filters that
match on 802.1q fields. The following selftests make use of these
newly-relaxed filters.
Patch #2 adds preparation as part of selftests API, which will be used
later.
Patches #3-#4 add tests for VxLAN with bridge aware and unaware.
Patche #5 cleans unused function.
Patches #6-#7 add tests for VxLAN symmetric and asymmetric.
Patch #8 adds test for Q-in-VNI.
====================

Link: https://lore.kernel.org/r/20211221144949.2527545-1-amcohen@nvidia.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoselftests: forwarding: Add Q-in-VNI test for IPv6
Amit Cohen [Tue, 21 Dec 2021 14:49:49 +0000 (16:49 +0200)]
selftests: forwarding: Add Q-in-VNI test for IPv6

Add test to check Q-in-VNI traffic with IPv6 underlay and overlay.
The test is similar to the existing IPv4 test.

Signed-off-by: Amit Cohen <amcohen@nvidia.com>
Reviewed-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoselftests: forwarding: Add a test for VxLAN symmetric routing with IPv6
Amit Cohen [Tue, 21 Dec 2021 14:49:48 +0000 (16:49 +0200)]
selftests: forwarding: Add a test for VxLAN symmetric routing with IPv6

In a similar fashion to the asymmetric test, add a test for symmetric
routing. In symmetric routing both the ingress and egress VTEPs perform
routing in the overlay network into / from the VxLAN tunnel. Packets in
different directions use the same VNI - the L3 VNI.
Different tenants (VRFs) use different L3 VNIs.

Add a test which is similar to the existing IPv4 test to check IPv6.

Signed-off-by: Amit Cohen <amcohen@nvidia.com>
Reviewed-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoselftests: forwarding: Add a test for VxLAN asymmetric routing with IPv6
Amit Cohen [Tue, 21 Dec 2021 14:49:47 +0000 (16:49 +0200)]
selftests: forwarding: Add a test for VxLAN asymmetric routing with IPv6

In asymmetric routing the ingress VTEP routes the packet into the
correct VxLAN tunnel, whereas the egress VTEP only bridges the packet to
the correct host. Therefore, packets in different directions use
different VNIs - the target VNI.

Add a test which is similar to the existing IPv4 test to check IPv6.

The test uses a simple topology with two VTEPs and two VNIs and verifies
that ping passes between hosts (local / remote) in the same VLAN (VNI)
and in different VLANs belonging to the same tenant (VRF).

While the test does not check VM mobility, it does configure an anycast
gateway using a macvlan device on both VTEPs.

Signed-off-by: Amit Cohen <amcohen@nvidia.com>
Reviewed-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoselftests: forwarding: vxlan_bridge_1q: Remove unused function
Amit Cohen [Tue, 21 Dec 2021 14:49:46 +0000 (16:49 +0200)]
selftests: forwarding: vxlan_bridge_1q: Remove unused function

Remove `vxlan_ping_test()` which is not used and probably was copied
mistakenly from vxlan_bridge_1d.sh.

This was found while adding an equivalent test for IPv6.

Signed-off-by: Amit Cohen <amcohen@nvidia.com>
Reviewed-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoselftests: forwarding: Add VxLAN tests with a VLAN-aware bridge for IPv6
Amit Cohen [Tue, 21 Dec 2021 14:49:45 +0000 (16:49 +0200)]
selftests: forwarding: Add VxLAN tests with a VLAN-aware bridge for IPv6

The tests are very similar to their VLAN-unaware counterpart
(vxlan_bridge_1d_ipv6.sh and vxlan_bridge_1d_port_8472_ipv6.sh),
but instead of using multiple VLAN-unaware bridges, a single VLAN-aware
bridge is used with multiple VLANs.

Signed-off-by: Amit Cohen <amcohen@nvidia.com>
Reviewed-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoselftests: forwarding: Add VxLAN tests with a VLAN-unaware bridge for IPv6
Amit Cohen [Tue, 21 Dec 2021 14:49:44 +0000 (16:49 +0200)]
selftests: forwarding: Add VxLAN tests with a VLAN-unaware bridge for IPv6

Add tests similar to vxlan_bridge_1d.sh and vxlan_bridge_1d_port_8472.sh.

The tests set up a topology with three VxLAN endpoints: one
"local", possibly offloaded, and two "remote", formed using veth pairs
and likely purely software bridges. The "local" endpoint is connected to
host systems by a VLAN-unaware bridge.

Since VxLAN tunnels must be unique per namespace, each of the "remote"
endpoints is in its own namespace. H3 forms the bridge between the three
domains.

Send IPv4 packets and IPv6 packets with IPv6 underlay.
Use `TC_FLAG`, which is defined in `forwarding.config` file, for TC
checks. `TC_FLAG` allows testing that on HW datapath, the traffic
actually goes through HW.

Signed-off-by: Amit Cohen <amcohen@nvidia.com>
Reviewed-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoselftests: lib.sh: Add PING_COUNT to allow sending configurable amount of packets
Amit Cohen [Tue, 21 Dec 2021 14:49:43 +0000 (16:49 +0200)]
selftests: lib.sh: Add PING_COUNT to allow sending configurable amount of packets

Currently `ping_do()` and `ping6_do()` send 10 packets.

There are cases that it is not possible to catch only the interesting
packets using tc rule, so then, it is possible to send many packets and
verify that at least this amount of packets hit the rule.

Add `PING_COUNT` variable, which is set to 10 by default, to allow tests
sending more than 10 packets using the existing ping API.

Signed-off-by: Amit Cohen <amcohen@nvidia.com>
Reviewed-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agomlxsw: spectrum_flower: Make vlan_id limitation more specific
Amit Cohen [Tue, 21 Dec 2021 14:49:42 +0000 (16:49 +0200)]
mlxsw: spectrum_flower: Make vlan_id limitation more specific

Spectrum ASICs do not support matching of VLAN ID at egress.
Currently, mlxsw driver forbids matching of all VLAN related fields at
egress, which is too strict check.

For example, the following filter is not supported by the driver:
$ tc filter add dev swpX egress protocol 802.1q pref 1 handle 101 flower
vlan_ethtype ipv4 src_ip .. dst_ip .. skip_sw action pass
Error: mlxsw_spectrum: vlan_id key is not supported on egress.
We have an error talking to the kernel

The filter above does not match on VLAN ID, but is bounced anyway.

Make the check more specific, forbid only matching of 'vlan_id' at egress.

Signed-off-by: Amit Cohen <amcohen@nvidia.com>
Reviewed-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoMerge tag 'mlx5-updates-2021-12-21' of git://git.kernel.org/pub/scm/linux/kernel...
Jakub Kicinski [Thu, 23 Dec 2021 01:13:00 +0000 (17:13 -0800)]
Merge tag 'mlx5-updates-2021-12-21' of git://git./linux/kernel/git/saeed/linux

Saeed Mahameed says:

====================
mlx5-updates-2021-12-21

1) From Shay Drory: Devlink user knobs to control device's EQ size

This series provides knobs which will enable users to
minimize memory consumption of mlx5 Functions (PF/VF/SF).
mlx5 exposes two new generic devlink params for EQ size
configuration and uses devlink generic param max_macs.

LINK: https://lore.kernel.org/netdev/20211208141722.13646-1-shayd@nvidia.com/
2) From Tariq and Lama, allocate software channel objects and statistics
  of a mlx5 netdevice private data dynamically upon first demand to save on
  memory.

* tag 'mlx5-updates-2021-12-21' of git://git.kernel.org/pub/scm/linux/kernel/git/saeed/linux:
  net/mlx5e: Take packet_merge params directly from the RX res struct
  net/mlx5e: Allocate per-channel stats dynamically at first usage
  net/mlx5e: Use dynamic per-channel allocations in stats
  net/mlx5e: Allow profile-specific limitation on max num of channels
  net/mlx5e: Save memory by using dynamic allocation in netdev priv
  net/mlx5e: Add profile indications for PTP and QOS HTB features
  net/mlx5e: Use bitmap field for profile features
  net/mlx5: Remove the repeated declaration
  net/mlx5: Let user configure max_macs generic param
  devlink: Clarifies max_macs generic devlink param
  net/mlx5: Let user configure event_eq_size param
  devlink: Add new "event_eq_size" generic device param
  net/mlx5: Let user configure io_eq_size param
  devlink: Add new "io_eq_size" generic device param
====================

Link: https://lore.kernel.org/r/20211222031604.14540-1-saeed@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agocodel: remove unnecessary pkt_sched.h include
Jakub Kicinski [Tue, 21 Dec 2021 19:39:41 +0000 (11:39 -0800)]
codel: remove unnecessary pkt_sched.h include

Commit d068ca2ae2e6 ("codel: split into multiple files") moved all
Qdisc-related code to codel_qdisc.h, move the include of pkt_sched.h
as well.

This is similar to the previous commit, although we don't care as
much about incremental builds after pkt_sched.h was touched itself
it is included by net/sch_generic.h which is modified ~20 times
a year.

This decreases the incremental build size after touching pkt_sched.h
from 1592 to 617 objects.

Fix unmasked missing includes in WiFi drivers.

Acked-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20211221193941.3805147-2-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agocodel: remove unnecessary sock.h include
Jakub Kicinski [Tue, 21 Dec 2021 19:39:40 +0000 (11:39 -0800)]
codel: remove unnecessary sock.h include

Since sock.h is modified relatively often (60 times in the last
12 months) it seems worthwhile to decrease the incremental build
work.

CoDel's header includes net/inet_ecn.h which in turn includes net/sock.h.
codel.h is itself included by mac80211 which is included by much of
the WiFi stack and drivers. Removing the net/inet_ecn.h include from
CoDel breaks the dependecy between WiFi and sock.h.

Commit d068ca2ae2e6 ("codel: split into multiple files") moved all
the code which actually needs ECN helpers out to net/codel_impl.h,
the include can be moved there as well.

This decreases the incremental build size after touching sock.h
from 4999 objects to 4051 objects.

Fix unmasked missing includes in WiFi drivers.

Acked-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20211221193941.3805147-1-kuba@kernel.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agonet: broadcom: bcm4908enet: remove redundant variable bytes
Colin Ian King [Wed, 22 Dec 2021 00:39:37 +0000 (00:39 +0000)]
net: broadcom: bcm4908enet: remove redundant variable bytes

The variable bytes is being used to summate slot lengths,
however the value is never used afterwards. The summation
is redundant so remove variable bytes.

Signed-off-by: Colin Ian King <colin.i.king@gmail.com>
Link: https://lore.kernel.org/r/20211222003937.727325-1-colin.i.king@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoice: trivial: fix odd indenting
Jesse Brandeburg [Tue, 21 Dec 2021 23:05:38 +0000 (15:05 -0800)]
ice: trivial: fix odd indenting

Fix an odd indent where some code was left indented, and causes smatch
to warn:
ice_log_pkg_init() warn: inconsistent indenting

While here, for consistency, add a break after the default case.

This commit has a Fixes: but we caught this while it was only in net-next.

Fixes: 247dd97d713c ("ice: Refactor status flow for DDP load")
Signed-off-by: Jesse Brandeburg <jesse.brandeburg@intel.com>
Link: https://lore.kernel.org/r/20211221230538.2546315-1-jesse.brandeburg@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoasix: fix wrong return value in asix_check_host_enable()
Pavel Skripkin [Tue, 21 Dec 2021 20:10:43 +0000 (23:10 +0300)]
asix: fix wrong return value in asix_check_host_enable()

If asix_read_cmd() returns 0 on 30th interation, 0 will be returned from
asix_check_host_enable(), which is logically wrong. Fix it by returning
-ETIMEDOUT explicitly if we have exceeded 30 iterations

Also, replaced 30 with #define as suggested by Andrew

Fixes: a786e3195d6a ("net: asix: fix uninit value bugs")
Reported-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Pavel Skripkin <paskripkin@gmail.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Link: https://lore.kernel.org/r/ecd3470ce6c2d5697ac635d0d3b14a47defb4acb.1640117288.git.paskripkin@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoasix: fix uninit-value in asix_mdio_read()
Pavel Skripkin [Tue, 21 Dec 2021 20:10:36 +0000 (23:10 +0300)]
asix: fix uninit-value in asix_mdio_read()

asix_read_cmd() may read less than sizeof(smsr) bytes and in this case
smsr will be uninitialized.

Fail log:
BUG: KMSAN: uninit-value in asix_check_host_enable drivers/net/usb/asix_common.c:82 [inline]
BUG: KMSAN: uninit-value in asix_check_host_enable drivers/net/usb/asix_common.c:82 [inline] drivers/net/usb/asix_common.c:497
BUG: KMSAN: uninit-value in asix_mdio_read+0x3c1/0xb00 drivers/net/usb/asix_common.c:497 drivers/net/usb/asix_common.c:497
 asix_check_host_enable drivers/net/usb/asix_common.c:82 [inline]
 asix_check_host_enable drivers/net/usb/asix_common.c:82 [inline] drivers/net/usb/asix_common.c:497
 asix_mdio_read+0x3c1/0xb00 drivers/net/usb/asix_common.c:497 drivers/net/usb/asix_common.c:497

Fixes: d9fe64e51114 ("net: asix: Add in_pm parameter")
Reported-and-tested-by: syzbot+f44badb06036334e867a@syzkaller.appspotmail.com
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Pavel Skripkin <paskripkin@gmail.com>
Link: https://lore.kernel.org/r/8966e3b514edf39857dd93603fc79ec02e000a75.1640117288.git.paskripkin@gmail.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agoMerge branch '100GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/next...
Jakub Kicinski [Wed, 22 Dec 2021 22:25:41 +0000 (14:25 -0800)]
Merge branch '100GbE' of git://git./linux/kernel/git/tnguy/next-queue

Tony Nguyen says:

====================
100GbE Intel Wired LAN Driver Updates 2021-12-21

This series contains updates to ice driver only.

Karol modifies the reset flow to correct issues with PTP reset.

Jake extends PTP support for E822 based devices. This includes a few
cleanup patches, that fix some minor issues. In addition, there are some
slight refactors to ease the addition of E822 support, followed by adding
the new hardware implementation ice_ptp_hw.c.

There are a few major differences with E822 support compared to E810
support:

*) The E822 device has a Clock Generation Unit which must be initialized in
order to generate proper clock frequencies on the output that drives the PTP
hardware clock registers

*) The E822 PHY is a bit different and requires a more complex
initialization procedure which must be rerun any time the link configuration
changes.

*) The E822 devices support enhanced timestamp calibration by making use of
a process called Vernier offset measurement. This allows the hardware to
measure phase offset related to the PHY clocks for Serdes and FEC, reducing
the inaccuracy of the timestamp relative to the actual packet transmission
and receipt. Making use of this requires data gathered from the first
transmitted and received packets, and waiting for the PHY to complete the
calibration measurements. This is done as part of a new kthread, ov_work.
Note that to avoid delay in enabling timestamps, we start the PHY in
'bypass' mode which allows timestamps to be captured without the Vernier
calibration measurement. Once the first packets have been sent and received,
we then complete the calibration setup and exit bypass mode and begin using
the more precise timestamps. According to the datasheet, timestamps without
calibration data can be incorrect relative to actual receipt or transmission
by up to 1 clock cycle (~1.25 nanoseconds), while calibrated timestamps
should be correct to within 1/8th of a clock cycle (~0.15 nanoseconds).

*) E822 devices support crosstimestamping via PCIe PTM, which we enable when
available on the platform.

There is a fair amount of logic required to perform PHY and CGU
initialization, which is the vast majority of the new code, but it is fairly
self contained within ice_ptp_hw.c, with the exception of monitoring for
offset validity being handled by a kthread.

* '100GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/next-queue:
  ice: support crosstimestamping on E822 devices if supported
  ice: exit bypass mode once hardware finishes timestamp calibration
  ice: ensure the hardware Clock Generation Unit is configured
  ice: implement basic E822 PTP support
  ice: convert clk_freq capability into time_ref
  ice: introduce ice_ptp_init_phc function
  ice: use 'int err' instead of 'int status' in ice_ptp_hw.c
  ice: PTP: move setting of tstamp_config
  ice: introduce ice_base_incval function
  ice: Fix E810 PTP reset flow
====================

Link: https://lore.kernel.org/r/20211221174845.3063640-1-anthony.l.nguyen@intel.com
Acked-by: Richard Cochran <richardcochran@gmail.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agosfc: falcon: Check null pointer of rx_queue->page_ring
Jiasheng Jiang [Mon, 20 Dec 2021 14:03:44 +0000 (22:03 +0800)]
sfc: falcon: Check null pointer of rx_queue->page_ring

Because of the possible failure of the kcalloc, it should be better to
set rx_queue->page_ptr_mask to 0 when it happens in order to maintain
the consistency.

Fixes: 5a6681e22c14 ("sfc: separate out SFC4000 ("Falcon") support into new sfc-falcon driver")
Signed-off-by: Jiasheng Jiang <jiasheng@iscas.ac.cn>
Acked-by: Martin Habets <habetsm.xilinx@gmail.com>
Link: https://lore.kernel.org/r/20211220140344.978408-1-jiasheng@iscas.ac.cn
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agosfc: Check null pointer of rx_queue->page_ring
Jiasheng Jiang [Mon, 20 Dec 2021 13:56:03 +0000 (21:56 +0800)]
sfc: Check null pointer of rx_queue->page_ring

Because of the possible failure of the kcalloc, it should be better to
set rx_queue->page_ptr_mask to 0 when it happens in order to maintain
the consistency.

Fixes: 5a6681e22c14 ("sfc: separate out SFC4000 ("Falcon") support into new sfc-falcon driver")
Signed-off-by: Jiasheng Jiang <jiasheng@iscas.ac.cn>
Acked-by: Martin Habets <habetsm.xilinx@gmail.com>
Link: https://lore.kernel.org/r/20211220135603.954944-1-jiasheng@iscas.ac.cn
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agomt76: mt7921: fix a possible race enabling/disabling runtime-pm
Lorenzo Bianconi [Tue, 21 Dec 2021 13:57:09 +0000 (14:57 +0100)]
mt76: mt7921: fix a possible race enabling/disabling runtime-pm

Fix a possible race enabling/disabling runtime-pm between
mt7921_pm_set() and mt7921_poll_rx() since mt7921_pm_wake_work()
always schedules rx-napi callback and it will trigger
mt7921_pm_power_save_work routine putting chip to in low-power state
during mt7921_pm_set processing.

Suggested-by: Deren Wu <deren.wu@mediatek.com>
Tested-by: Deren Wu <deren.wu@mediatek.com>
Fixes: 1d8efc741df8 ("mt76: mt7921: introduce Runtime PM support")
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/0f3e075a2033dc05f09dab4059e5be8cbdccc239.1640094847.git.lorenzo@kernel.org
2 years agowilc1000: Document enable-gpios and reset-gpios properties
David Mosberger-Tang [Tue, 21 Dec 2021 21:25:34 +0000 (21:25 +0000)]
wilc1000: Document enable-gpios and reset-gpios properties

Add documentation for the ENABLE and RESET GPIOs that may be needed by
wilc1000-spi.

Signed-off-by: David Mosberger-Tang <davidm@egauge.net>
Reviewed-by: Rob Herring <robh@kernel.org>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20211221212531.4011609-3-davidm@egauge.net
2 years agowilc1000: Add reset/enable GPIO support to SPI driver
David Mosberger-Tang [Tue, 21 Dec 2021 21:25:34 +0000 (21:25 +0000)]
wilc1000: Add reset/enable GPIO support to SPI driver

For the SDIO driver, the RESET/ENABLE pins of WILC1000 are controlled
through the SDIO power sequence driver.  This commit adds analogous
support for the SPI driver.  Specifically, during initialization, the
chip will be ENABLEd and taken out of RESET and during
deinitialization, the chip will be placed back into RESET and disabled
(both to reduce power consumption and to ensure the WiFi radio is
off).

Both RESET and ENABLE GPIOs are optional.  However, if the ENABLE GPIO
is specified, then the RESET GPIO should normally also be specified as
otherwise there is no way to ensure proper timing of the ENABLE/RESET
sequence.

Signed-off-by: David Mosberger-Tang <davidm@egauge.net>
Reviewed-by: Claudiu Beznea <claudiu.beznea@microchip.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20211221212531.4011609-2-davidm@egauge.net
2 years agowilc1000: Convert static "chipid" variable to device-local variable
David Mosberger-Tang [Tue, 21 Dec 2021 21:05:41 +0000 (21:05 +0000)]
wilc1000: Convert static "chipid" variable to device-local variable

Move "chipid" variable into the per-driver structure so the code
doesn't break if more than one wilc1000 module is present.

Signed-off-by: David Mosberger-Tang <davidm@egauge.net>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20211221210538.4011227-1-davidm@egauge.net
2 years agortw89: 8852a: correct bit definition of dfs_en
Chung-Hsuan Hung [Tue, 21 Dec 2021 02:48:00 +0000 (10:48 +0800)]
rtw89: 8852a: correct bit definition of dfs_en

Since there are other protections in the set channel flow, fortunately old
wrong setting won't affect the performance.

Signed-off-by: Chung-Hsuan Hung <hsuan8331@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20211221024800.23814-1-pkshih@realtek.com
2 years agortw88: don't consider deep PS mode when transmitting packet
Chin-Yen Lee [Tue, 21 Dec 2021 02:02:30 +0000 (10:02 +0800)]
rtw88: don't consider deep PS mode when transmitting packet

In original flow, driver needs to ensure chip leave deep ps mode
before transmitting any packet, and don't enter deep ps mode beofre
PCIE DMA is finished. Now with the support of 8822ce's hardware
setting and firmware after v9.9.11, these limits are removed.

Signed-off-by: Chin-Yen Lee <timlee@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20211221020230.20764-1-pkshih@realtek.com
2 years agoMerge ath-next from git://git.kernel.org/pub/scm/linux/kernel/git/kvalo/ath.git
Kalle Valo [Wed, 22 Dec 2021 17:46:02 +0000 (19:46 +0200)]
Merge ath-next from git://git./linux/kernel/git/kvalo/ath.git

ath.git patches for v5.17. Major changes:

ath11k

* qca6390/wcn6855: report signal and tx bitrate

* qca6390: rfkill support

* qca6390/wcn6855: regdb.bin support

ath5k

* switch to rate table based lookup

2 years agoath11k: Fix unexpected return buffer manager error for QCA6390
Baochen Qiang [Wed, 22 Dec 2021 01:35:35 +0000 (09:35 +0800)]
ath11k: Fix unexpected return buffer manager error for QCA6390

We are seeing below error on QCA6390:
...
[70211.671189] ath11k_pci 0000:72:00.0: failed to parse rx error in wbm_rel ring desc -22
[70212.696154] ath11k_pci 0000:72:00.0: failed to parse rx error in wbm_rel ring desc -22
[70213.092941] ath11k_pci 0000:72:00.0: failed to parse rx error in wbm_rel ring desc -22
...

The reason is that, with commit 734223d78428 ("ath11k: change return
buffer manager for QCA6390"), ath11k expects the return buffer manager
(RBM) field of descriptor configured as HAL_RX_BUF_RBM_SW1_BM when
parsing error frames from WBM2SW3_RELEASE ring. This is a wrong change
cause the RBM field is set as HAL_RX_BUF_RBM_SW3_BM.

The same issue also applies to REO2TCL ring though we have not got any
error reported.

Fix it by changing RBM from HAL_RX_BUF_RBM_SW1_BM to HAL_RX_BUF_RBM_SW3_BM
for these two rings.

Tested-on: QCA6390 hw2.0 PCI WLAN.HST.1.0.1-01740-QCAHSTSWPLZ_V2_TO_X86-1

Fixes: 734223d78428 ("ath11k: change return buffer manager for QCA6390")
Signed-off-by: Baochen Qiang <quic_bqiang@quicinc.com>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://lore.kernel.org/r/20211222013536.582527-1-quic_bqiang@quicinc.com
2 years agonet: ks8851: Check for error irq
Jiasheng Jiang [Wed, 22 Dec 2021 07:59:44 +0000 (15:59 +0800)]
net: ks8851: Check for error irq

Because platform_get_irq() could fail and return error irq.
Therefore, it might be better to check it if order to avoid the use of
error irq.

Fixes: 797047f875b5 ("net: ks8851: Implement Parallel bus operations")
Signed-off-by: Jiasheng Jiang <jiasheng@iscas.ac.cn>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agodrivers: net: smc911x: Check for error irq
Jiasheng Jiang [Wed, 22 Dec 2021 07:41:12 +0000 (15:41 +0800)]
drivers: net: smc911x: Check for error irq

Because platform_get_irq() could fail and return error irq.
Therefore, it might be better to check it if order to avoid the use of
error irq.

Fixes: ae150435b59e ("smsc: Move the SMC (SMSC) drivers")
Signed-off-by: Jiasheng Jiang <jiasheng@iscas.ac.cn>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agofjes: Check for error irq
Jiasheng Jiang [Wed, 22 Dec 2021 07:12:07 +0000 (15:12 +0800)]
fjes: Check for error irq

I find that platform_get_irq() will not always succeed.
It will return error irq in case of the failure.
Therefore, it might be better to check it if order to avoid the use of
error irq.

Fixes: 658d439b2292 ("fjes: Introduce FUJITSU Extended Socket Network Device driver")
Signed-off-by: Jiasheng Jiang <jiasheng@iscas.ac.cn>
Signed-off-by: David S. Miller <davem@davemloft.net>
2 years agoath11k: add support of firmware logging for WCN6855
Cheng Wang [Mon, 20 Dec 2021 12:10:53 +0000 (20:10 +0800)]
ath11k: add support of firmware logging for WCN6855

Host enables WMI firmware logging feature via QMI message.
Host receives firmware logging messages on WMI_DIAG_EVENTID, then
sends logging messages to user space via event tracing infrastructure.

Tested-on: WCN6855 hw2.0 PCI WLAN.HSP.1.1-01720.1-QCAHSPSWPL_V1_V2_SILICONZ_LITE-1

Signed-off-by: Cheng Wang <quic_chengwan@quicinc.com>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://lore.kernel.org/r/20211220121053.357087-1-quic_chengwan@quicinc.com
2 years agoath11k: Fix napi related hang
Ben Greear [Thu, 3 Sep 2020 19:52:54 +0000 (12:52 -0700)]
ath11k: Fix napi related hang

Similar to the same bug in ath10k, a napi disable w/out it being enabled
will hang forever.  I believe I saw this while trying rmmod after driver
had some failure on startup.  Fix it by keeping state on whether napi is
enabled or not.

And, remove un-used napi pointer in ath11k driver base struct.

Signed-off-by: Ben Greear <greearb@candelatech.com>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://lore.kernel.org/r/20200903195254.29379-1-greearb@candelatech.com
2 years agoath10k: replace strlcpy with strscpy
Jason Wang [Tue, 21 Dec 2021 10:04:20 +0000 (12:04 +0200)]
ath10k: replace strlcpy with strscpy

The strlcpy should not be used because it doesn't limit the source
length. So that it will lead some potential bugs.

But the strscpy doesn't require reading memory from the src string
beyond the specified "count" bytes, and since the return value is
easier to error-check than strlcpy()'s. In addition, the implementation
is robust to the string changing out from underneath it, unlike the
current strlcpy() implementation.

Thus, replace strlcpy with strscpy.

Signed-off-by: Jason Wang <wangborong@cdjrlc.com>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://lore.kernel.org/r/20211221070931.725720-1-wangborong@cdjrlc.com
2 years agonet/mlx5e: Take packet_merge params directly from the RX res struct
Tariq Toukan [Tue, 14 Sep 2021 11:46:05 +0000 (14:46 +0300)]
net/mlx5e: Take packet_merge params directly from the RX res struct

As packet_merge params structure is saved on the RX resources structure, there
is no need to pass it separately.

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: Allocate per-channel stats dynamically at first usage
Lama Kayal [Wed, 22 Sep 2021 06:32:33 +0000 (09:32 +0300)]
net/mlx5e: Allocate per-channel stats dynamically at first usage

Make stats allocation per-channel dynamic on demand, at channel open
operation.

Previously the stats array was pre-allocated for the maximum possible
number of channels. Here we defer the per-channel stats instance allocation
upon its first usage, so that it's allocated only if really needed.

Allocating stats on demand helps maintain a more memory-efficient code,
as we're saving memory when the used number of channels is smaller than
the maximum.

The stats memory instances are still freed in mlx5e_priv_arrays_free(),
so that they are persistent to channels' closure.

Memory size allocated for struct mlx5e_channel_stats is 3648 bytes.
If maximum number of channel stands for 64, the total memory space
allocated for stats is 3648x64 = 228K bytes. In scenarios where the
number of channels in use is significantly smaller than maximum number,
the memory saved can be remarkable.

Signed-off-by: Lama Kayal <lkayal@nvidia.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5e: Use dynamic per-channel allocations in stats
Tariq Toukan [Sun, 5 Dec 2021 14:04:14 +0000 (16:04 +0200)]
net/mlx5e: Use dynamic per-channel allocations in stats

Make stats array an array of pointer. This patch comes in to prepare for
the next patch where allocations of the stats are to be performed
dynamically on first usage.

Signed-off-by: Lama Kayal <lkayal@nvidia.com>
Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
Reviewed-by: Moshe Shemesh <moshe@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5e: Allow profile-specific limitation on max num of channels
Tariq Toukan [Wed, 29 Sep 2021 14:29:24 +0000 (17:29 +0300)]
net/mlx5e: Allow profile-specific limitation on max num of channels

Let SF/VF representor's netdev use profile-specific limitation on
max_nch to reduce its memory and HW resources consumption.

This is particularly important for environments with limited memory
and high number of SFs.

Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Vu Pham <vuhuong@nvidia.com>
Reviewed-by: Moshe Shemesh <moshe@nvidia.com>
Reviewed-by: Parav Pandit <parav@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5e: Save memory by using dynamic allocation in netdev priv
Tariq Toukan [Thu, 2 Sep 2021 07:24:06 +0000 (10:24 +0300)]
net/mlx5e: Save memory by using dynamic allocation in netdev priv

Many arrays in priv are statically allocated with a pre-defined maximum
(for num channels, num TCs, etc...), that is in some cases significantly
larger than the actual maximum. Examples:
- The more VFs are supported, the less MSIX vectors each of them could
  have. This limits the max_nch for each.
- Systems with limited number of cores or MSIX (< 64).
- Netdev profiles that do not support: QoS (DCB / HTB), PTP TX port
  timestamping.

Here we save some amount of memory by moving several structures
and arrays to follow the actual maximum instead.
This patch also prepares the code for even more savings to follow.

For example, on a system where the maximum num of channel is 8,
the channels stats structs alone go down from 3648*64 = 228 KB to
3648*8 = 28.5 KB per interface.

This is important for environments with high number of VFs/SFs or
limited memory.

Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
Reviewed-by: Moshe Shemesh <moshe@nvidia.com>
Reviewed-by: Parav Pandit <parav@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5e: Add profile indications for PTP and QOS HTB features
Tariq Toukan [Tue, 7 Dec 2021 06:38:42 +0000 (08:38 +0200)]
net/mlx5e: Add profile indications for PTP and QOS HTB features

Let the profile indicate support of the PTP and HTB (QOS) features.
This unifies the logic that calculates the number of netdev queues needed
for the features, and allows simplification of mlx5e_create_netdev(),
which no longer requires number of rx/tx queues as parameters.

Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
Reviewed-by: Moshe Shemesh <moshe@nvidia.com>
Reviewed-by: Aya Levin <ayal@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5e: Use bitmap field for profile features
Tariq Toukan [Sun, 5 Dec 2021 10:31:47 +0000 (12:31 +0200)]
net/mlx5e: Use bitmap field for profile features

Use a features bitmap field in mlx5e_profile to declare profile support
state of the different features.  Let it replace the existing
rx_ptp_support boolean. It will be extended to cover more features in a
downstream patch.

Signed-off-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5: Remove the repeated declaration
Shaokun Zhang [Tue, 7 Dec 2021 12:35:15 +0000 (20:35 +0800)]
net/mlx5: Remove the repeated declaration

Function 'mlx5_esw_vport_match_metadata_supported' and
'mlx5_esw_offloads_vport_metadata_set' are declared twice, so remove
the repeated declaration and blank line.

Cc: Saeed Mahameed <saeedm@nvidia.com>
Cc: Leon Romanovsky <leon@kernel.org>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Shaokun Zhang <zhangshaokun@hisilicon.com>
Reviewed-by: Leon Romanovsky <leonro@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5: Let user configure max_macs generic param
Shay Drory [Thu, 9 Dec 2021 10:09:29 +0000 (12:09 +0200)]
net/mlx5: Let user configure max_macs generic param

Currently, max_macs is taking 70Kbytes of memory per function. This
size is not needed in all use cases, and is critical with large scale.
Hence, allow user to configure the number of max_macs.

For example, to reduce the number of max_macs to 1, execute::
$ devlink dev param set pci/0000:00:0b.0 name max_macs value 1 \
              cmode driverinit
$ devlink dev reload pci/0000:00:0b.0

Signed-off-by: Shay Drory <shayd@nvidia.com>
Reviewed-by: Moshe Shemesh <moshe@nvidia.com>
Reviewed-by: Parav Pandit <parav@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agodevlink: Clarifies max_macs generic devlink param
Shay Drory [Thu, 9 Dec 2021 10:09:28 +0000 (12:09 +0200)]
devlink: Clarifies max_macs generic devlink param

The generic param max_macs documentation isn't clear.
Replace it with a more descriptive documentation

Signed-off-by: Shay Drory <shayd@nvidia.com>
Reviewed-by: Jiri Pirko <jiri@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5: Let user configure event_eq_size param
Shay Drory [Thu, 9 Dec 2021 10:09:27 +0000 (12:09 +0200)]
net/mlx5: Let user configure event_eq_size param

Event EQ is an EQ which received the notification of almost all the
events generated by the NIC.
Currently, each event EQ is taking 512KB of memory. This size is not
needed in most use cases, and is critical with large scale. Hence,
allow user to configure the size of the event EQ.

For example to reduce event EQ size to 64, execute::
$ devlink dev param set pci/0000:00:0b.0 name event_eq_size value 64 \
              cmode driverinit
$ devlink dev reload pci/0000:00:0b.0

Signed-off-by: Shay Drory <shayd@nvidia.com>
Reviewed-by: Moshe Shemesh <moshe@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agodevlink: Add new "event_eq_size" generic device param
Shay Drory [Thu, 9 Dec 2021 10:09:26 +0000 (12:09 +0200)]
devlink: Add new "event_eq_size" generic device param

Add new device generic parameter to determine the size of the
asynchronous control events EQ.

For example, to reduce event EQ size to 64, execute:
$ devlink dev param set pci/0000:06:00.0 \
              name event_eq_size value 64 cmode driverinit
$ devlink dev reload pci/0000:06:00.0

Signed-off-by: Shay Drory <shayd@nvidia.com>
Reviewed-by: Moshe Shemesh <moshe@nvidia.com>
Reviewed-by: Jiri Pirko <jiri@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agonet/mlx5: Let user configure io_eq_size param
Shay Drory [Thu, 9 Dec 2021 10:09:25 +0000 (12:09 +0200)]
net/mlx5: Let user configure io_eq_size param

Currently, each I/O EQ is taking 128KB of memory. This size
is not needed in all use cases, and is critical with large scale.
Hence, allow user to configure the size of I/O EQs.

For example, to reduce I/O EQ size to 64, execute:
$ devlink dev param set pci/0000:00:0b.0 name io_eq_size value 64 \
              cmode driverinit
$ devlink dev reload pci/0000:00:0b.0

Signed-off-by: Shay Drory <shayd@nvidia.com>
Reviewed-by: Moshe Shemesh <moshe@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agodevlink: Add new "io_eq_size" generic device param
Shay Drory [Thu, 9 Dec 2021 10:09:24 +0000 (12:09 +0200)]
devlink: Add new "io_eq_size" generic device param

Add new device generic parameter to determine the size of the
I/O completion EQs.

For example, to reduce I/O EQ size to 64, execute:
$ devlink dev param set pci/0000:06:00.0 \
              name io_eq_size value 64 cmode driverinit
$ devlink dev reload pci/0000:06:00.0

Signed-off-by: Shay Drory <shayd@nvidia.com>
Reviewed-by: Moshe Shemesh <moshe@nvidia.com>
Reviewed-by: Jiri Pirko <jiri@nvidia.com>
Signed-off-by: Saeed Mahameed <saeedm@nvidia.com>
2 years agoMerge branch '1GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/next-queue
Jakub Kicinski [Wed, 22 Dec 2021 01:20:31 +0000 (17:20 -0800)]
Merge branch '1GbE' of git://git./linux/kernel/git/tnguy/next-queue

Tony Nguyen says:

====================
1GbE Intel Wired LAN Driver Updates 2021-12-21

This series contains updates to igc, igb, igbvf, and fm10k drivers.

Sasha removes unused defines and enum values from igc driver.

Jason Wang removes a variable whose value never changes and, instead,
returns the value directly for igb.

Karen adjusts a reset message from warning to info for igbvf.

Xiang wangx removes a repeated word for fm10k.

* '1GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/next-queue:
  fm10k: Fix syntax errors in comments
  igbvf: Refactor trace
  igb: remove never changed variable `ret_val'
  igc: Remove obsolete define
  igc: Remove obsolete mask
  igc: Remove obsolete nvm type
  igc: Remove unused phy type
  igc: Remove unused _I_PHY_ID define
====================

Link: https://lore.kernel.org/r/20211221180200.3176851-1-anthony.l.nguyen@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agonet: phy: micrel: Adding interrupt support for Link up/Link down in LAN8814 Quad phy
Divya Koppera [Tue, 21 Dec 2021 11:22:17 +0000 (16:52 +0530)]
net: phy: micrel: Adding interrupt support for Link up/Link down in LAN8814 Quad phy

This patch add support for Link up or Link down
interrupt support in LAN8814 Quad phy

Signed-off-by: Divya Koppera <Divya.Koppera@microchip.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Link: https://lore.kernel.org/r/20211221112217.9502-1-Divya.Koppera@microchip.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agobonding: fix ad_actor_system option setting to default
Fernando Fernandez Mancera [Tue, 21 Dec 2021 11:13:45 +0000 (12:13 +0100)]
bonding: fix ad_actor_system option setting to default

When 802.3ad bond mode is configured the ad_actor_system option is set to
"00:00:00:00:00:00". But when trying to set the all-zeroes MAC as actors'
system address it was failing with EINVAL.

An all-zeroes ethernet address is valid, only multicast addresses are not
valid values.

Fixes: 171a42c38c6e ("bonding: add netlink support for sys prio, actor sys mac, and port key")
Signed-off-by: Fernando Fernandez Mancera <ffmancera@riseup.net>
Acked-by: Jay Vosburgh <jay.vosburgh@canonical.com>
Link: https://lore.kernel.org/r/20211221111345.2462-1-ffmancera@riseup.net
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2 years agortw88: support SAR via kernel common API
Zong-Zhe Yang [Mon, 20 Dec 2021 09:36:56 +0000 (17:36 +0800)]
rtw88: support SAR via kernel common API

Register cfg80211_sar_capa with type NL80211_SAR_TYPE_POWER and four
frequency ranges for configurations in unit of 0.25 dBm. And handle
callback set_sar_specs.

Originally, TX power has three main parameters, i.e. power base,
power by rate, and power limit. The formula can be simply considered
as TX power = power base + min(power by rate, power limit). With the
support of SAR which can be treated as another power limit, there is
one more parameter for TX power. And the formula will evolve into
TX power = power base + min(power by rate, power limit, power sar).

Besides, debugfs tx_pwr_tbl is also refined to show SAR information.
The following is an example for the difference.
Before supporting SAR,
-----------------------------------
...
path rate       pwr       base      (byr  lmt ) rem
   A  CCK_1M     66(0x42)   78  -12 (  12  -12)    0
   A  CCK_2M     66(0x42)   78  -12 (   8  -12)    0
...
-----------------------------------
After supporting SAR and making some configurations,
-----------------------------------
...
path rate       pwr       base      (byr  lmt  sar ) rem
   A  CCK_1M     62(0x3e)   78  -16 (  12  -12  -16)    0
   A  CCK_2M     62(0x3e)   78  -16 (   8  -12  -16)    0
...
-----------------------------------

Signed-off-by: Zong-Zhe Yang <kevin_yang@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20211220093656.65312-1-pkshih@realtek.com
2 years agortw88: 8822c: add ieee80211_ops::hw_scan
Po-Hao Huang [Tue, 21 Dec 2021 08:50:10 +0000 (16:50 +0800)]
rtw88: 8822c: add ieee80211_ops::hw_scan

Declare this function allows us to use customized scanning policy.
By doing so we can be more time efficient on each scan. In order to
make existing coex mechanism work as usual, firmware notifies driver
on each channel switch event, then decide antenna ownership based on
the current channel/band. Do note that this new mechanism affects
throughput more than the sw_scan we used to have, but the overall
average throughput is not affected since each scan take less time.
Since the firmware size is limited, we only support probe requests
with custom IEs length under 128 bytes for now, if any user space
tools requires more than that, we'll introduce related changes
afterwards. For backward compatibility, we fallback to sw_scan when
using older firmware that does not support this feature.

Signed-off-by: Po-Hao Huang <phhuang@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20211221085010.39421-1-pkshih@realtek.com
2 years agoMerge tag 'iwlwifi-next-for-kalle-2021-12-21-v2' of git://git.kernel.org/pub/scm...
Kalle Valo [Tue, 21 Dec 2021 18:07:09 +0000 (20:07 +0200)]
Merge tag 'iwlwifi-next-for-kalle-2021-12-21-v2' of git://git./linux/kernel/git/iwlwifi/iwlwifi-next

wlwifi patches for v5.17 v2

* Support for Time-Aware-SAR (TAS) as read from the BIOS;
* Fix scan timeout issue when 6GHz is enabled;
* Work continues for new HW family Bz;
* Support for Optimized Connectivity Experience (OCE) scan;
* A bunch of FW debugging improvements and fixes;
* Fix one 32-bit compilation issue;
* Some RX changes for new HW family
* Some fixes for 6 GHz scan;
* Fix SAR table fixes with newer platforms;
* Fix early restart crash;
* Small fix in the debugging code;
* Add new Killer device IDs;
* Datapath updates for Bz family continues;
* A couple of important fixes in iwlmei;
* Some other small fixes, clean-ups and improvements.