Until now the virtual address of the received buffer were stored in the
cookie field of the rx descriptor. However, this field is 32-bits only
which prevents to use the driver on a 64-bits architecture.
With this patch the virtual address is stored in an array not shared with
the hardware (no more need to use the DMA API). Thanks to this, it is
possible to use cache contrary to the access of the rx descriptor member.
The change is done in the swbm path only because the hwbm uses the cookie
field, this also means that currently the hwbm is not usable in 64-bits.
Signed-off-by: Gregory CLEMENT <gregory.clement@free-electrons.com>
Reviewed-by: Jisheng Zhang <jszhang@marvell.com>
Tested-by: Marcin Wojtas <mw@semihalf.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
For HWBM all buffers are allocated in mvneta_bm_construct() and in runtime
they are put into descriptors by hardware. There is no need to fill them
at this point.
Suggested-by: Marcin Wojtas <mw@semihalf.com>
Signed-off-by: Gregory CLEMENT <gregory.clement@free-electrons.com>
Tested-by: Marcin Wojtas <mw@semihalf.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
For small frame reuse the phys_addr variable instead of accessing the
uncacheable value in the rx descriptor.
Signed-off-by: Gregory CLEMENT <gregory.clement@free-electrons.com>
Tested-by: Marcin Wojtas <mw@semihalf.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Make sure to deregister and free any fixed-link PHY registered using
of_phy_register_fixed_link() on probe errors and on driver unbind.
Fixes: 83895bedee ("net: mvneta: add support for fixed links")
Signed-off-by: Johan Hovold <johan@kernel.org>
Reviewed-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
udplite conflict is resolved by taking what 'net-next' did
which removed the backlog receive method assignment, since
it is no longer necessary.
Two entries were added to the non-priv ethtool operations
switch statement, one in 'net' and one in 'net-next, so
simple overlapping changes.
Signed-off-by: David S. Miller <davem@davemloft.net>
The mvneta driver advertises it supports IFF_UNICAST_FLT. However, it
actually does not. The hardware probably does support it, but there is
no code to configure the filter. As a quick and simple fix, remove the
flag. This will cause the core to fall back to promiscuous mode.
Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Fixes: b50b72de2f ("net: mvneta: enable features before registering the driver")
Signed-off-by: David S. Miller <davem@davemloft.net>
Implement ethtool::nway_reset using phy_ethtool_nway_reset. We are
already using dev->phydev all over the place so this comes for free.
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
mvneta: min_mtu 68, max_mtu 9676
- mtu validation routine mostly did range check, merge back into
mvneta_change_mtu for simplicity
mvpp2: min_mtu 68, max_mtu 9676
- mtu validation routine mostly did range check, merge back into
mvpp2_change_mtu for simplicity
pxa168_eth: min_mtu 68, max_mtu 9500
skge: min_mtu 60, max_mtu 9000
sky2: min_mtu 68, max_mtu 1500 or 9000, depending on hw
CC: netdev@vger.kernel.org
CC: Mirko Lindner <mlindner@marvell.com>
CC: Stephen Hemminger <stephen@networkplumber.org>
CC: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Jarod Wilson <jarod@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Pull networking updates from David Miller:
1) BBR TCP congestion control, from Neal Cardwell, Yuchung Cheng and
co. at Google. https://lwn.net/Articles/701165/
2) Do TCP Small Queues for retransmits, from Eric Dumazet.
3) Support collect_md mode for all IPV4 and IPV6 tunnels, from Alexei
Starovoitov.
4) Allow cls_flower to classify packets in ip tunnels, from Amir Vadai.
5) Support DSA tagging in older mv88e6xxx switches, from Andrew Lunn.
6) Support GMAC protocol in iwlwifi mwm, from Ayala Beker.
7) Support ndo_poll_controller in mlx5, from Calvin Owens.
8) Move VRF processing to an output hook and allow l3mdev to be
loopback, from David Ahern.
9) Support SOCK_DESTROY for UDP sockets. Also from David Ahern.
10) Congestion control in RXRPC, from David Howells.
11) Support geneve RX offload in ixgbe, from Emil Tantilov.
12) When hitting pressure for new incoming TCP data SKBs, perform a
partial rathern than a full purge of the OFO queue (which could be
huge). From Eric Dumazet.
13) Convert XFRM state and policy lookups to RCU, from Florian Westphal.
14) Support RX network flow classification to igb, from Gangfeng Huang.
15) Hardware offloading of eBPF in nfp driver, from Jakub Kicinski.
16) New skbmod packet action, from Jamal Hadi Salim.
17) Remove some inefficiencies in snmp proc output, from Jia He.
18) Add FIB notifications to properly propagate route changes to
hardware which is doing forwarding offloading. From Jiri Pirko.
19) New dsa driver for qca8xxx chips, from John Crispin.
20) Implement RFC7559 ipv6 router solicitation backoff, from Maciej
Żenczykowski.
21) Add L3 mode to ipvlan, from Mahesh Bandewar.
22) Support 802.1ad in mlx4, from Moshe Shemesh.
23) Support hardware LRO in mediatek driver, from Nelson Chang.
24) Add TC offloading to mlx5, from Or Gerlitz.
25) Convert various drivers to ethtool ksettings interfaces, from
Philippe Reynes.
26) TX max rate limiting for cxgb4, from Rahul Lakkireddy.
27) NAPI support for ath10k, from Rajkumar Manoharan.
28) Support XDP in mlx5, from Rana Shahout and Saeed Mahameed.
29) UDP replicast support in TIPC, from Richard Alpe.
30) Per-queue statistics for qed driver, from Sudarsana Reddy Kalluru.
31) Support BQL in thunderx driver, from Sunil Goutham.
32) TSO support in alx driver, from Tobias Regnery.
33) Add stream parser engine and use it in kcm.
34) Support async DHCP replies in ipconfig module, from Uwe
Kleine-König.
35) DSA port fast aging for mv88e6xxx driver, from Vivien Didelot.
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next: (1715 commits)
mlxsw: switchx2: Fix misuse of hard_header_len
mlxsw: spectrum: Fix misuse of hard_header_len
net/faraday: Stop NCSI device on shutdown
net/ncsi: Introduce ncsi_stop_dev()
net/ncsi: Rework the channel monitoring
net/ncsi: Allow to extend NCSI request properties
net/ncsi: Rework request index allocation
net/ncsi: Don't probe on the reserved channel ID (0x1f)
net/ncsi: Introduce NCSI_RESERVED_CHANNEL
net/ncsi: Avoid unused-value build warning from ia64-linux-gcc
net: Add netdev all_adj_list refcnt propagation to fix panic
net: phy: Add Edge-rate driver for Microsemi PHYs.
vmxnet3: Wake queue from reset work
i40e: avoid NULL pointer dereference and recursive errors on early PCI error
qed: Add RoCE ll2 & GSI support
qed: Add support for memory registeration verbs
qed: Add support for QP verbs
qed: PD,PKEY and CQ verb support
qed: Add support for RoCE hw init
qede: Add qedr framework
...
We get 2 warnings when building kernel with W=1:
drivers/net/ethernet/marvell/mvneta.c:639:27: warning: no previous prototype for 'mvneta_get_stats64' [-Wmissing-prototypes]
drivers/net/ethernet/marvell/mvneta.c:3529:5: warning: no previous prototype for 'mvneta_ethtool_set_link_ksettings' [-Wmissing-prototypes]
In fact, these two functions are only used in the file in which they are
declared and don't need a declaration, but can be made static.
so this patch marks these functions with 'static'.
Signed-off-by: Baoyou Xie <baoyou.xie@linaro.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Install the callbacks via the state machine and let the core invoke
the callbacks on the already online CPUs.
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: netdev@vger.kernel.org
Cc: rt@linutronix.de
Link: http://lkml.kernel.org/r/20160818125731.27256-9-bigeasy@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
The ethtool api {get|set}_settings is deprecated.
We move the mvneta driver to new api {get|set}_link_ksettings.
We use the generic function phy_ethtool_get_link_ksettings,
and update old mvneta_ethtool_set_settings to the new api.
Signed-off-by: Philippe Reynes <tremyfr@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The private structure contain a pointer to phydev, but the structure
net_device already contain such pointer. So we can remove the pointer
phy_dev in the private structure, and update the driver to use the
one contained in struct net_device.
Signed-off-by: Philippe Reynes <tremyfr@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
of_node_put needs to be called when the device node which is got
from of_parse_phandle has finished using.
Signed-off-by: Peter Chen <peter.chen@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Commit aebea2ba0f ("net: mvneta: fix Tx interrupt delay") intended to
set coalescing threshold to a value guaranteeing interrupt generation
per each sent packet, so that buffers can be released with no delay.
In fact setting threshold to '1' was wrong, because it causes interrupt
every two packets. According to the documentation a reason behind it is
following - interrupt occurs once sent buffers counter reaches a value,
which is higher than one specified in MVNETA_TXQ_SIZE_REG(q). This
behavior was confirmed during tests. Also when testing the SoC working
as a NAS device, better performance was observed with int-per-packet,
as it strongly depends on the fact that all transmitted packets are
released immediately.
This commit enables NETA controller work in interrupt per sent packet mode
by setting coalescing threshold to 0.
Signed-off-by: Dmitri Epshtein <dima@marvell.com>
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Cc: <stable@vger.kernel.org> # v3.10+
Fixes aebea2ba0f ("net: mvneta: fix Tx interrupt delay")
Acked-by: Willy Tarreau <w@1wt.eu>
Signed-off-by: David S. Miller <davem@davemloft.net>
Since commit 3b9d6da67e ("cpu/hotplug: Fix rollback during error-out
in __cpu_disable()") it is ensured that callbacks of CPU_ONLINE and
CPU_DOWN_PREPARE are processed on the hotplugged CPU. Due to this SMP
function calls are no longer required.
Replace smp_call_function_single() with a direct call to
mvneta_percpu_enable() or mvneta_percpu_disable(). The functions do
not require to be called with interrupts disabled, therefore the
smp_call_function_single() calling convention is not preserved.
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: netdev@vger.kernel.org
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
After enabling per-cpu processing it appeared that under heavy load
changing MTU can result in blocking all port's interrupts and
transmitting data is not possible after the change.
This commit fixes above issue by disabling percpu interrupts for the
time, when TXQs and RXQs are reconfigured.
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
L1_CACHE_BYTES may not be the real cacheline size, use cache_line_size
to determine the cacheline size in runtime.
Signed-off-by: Jisheng Zhang <jszhang@marvell.com>
Suggested-by: Marcin Wojtas <mw@semihalf.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The mvneta is also used in some Marvell berlin family SoCs which may
have 64bytes cacheline size. Replace the MVNETA_CPU_D_CACHE_LINE_SIZE
usage with L1_CACHE_BYTES.
And since dma_alloc_coherent() is always cacheline size aligned, so
remove the align checks.
Signed-off-by: Jisheng Zhang <jszhang@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Some literal values are actually already defined by macros, so let's use
them.
[gregory.clement@free-electrons.com: split intial commit in two
individual changes]
Signed-off-by: Dmitri Epshtein <dima@marvell.com>
Signed-off-by: Gregory CLEMENT <gregory.clement@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This commit corrects error printing when shutting down the port.
[gregory.clement@free-electrons.com: split initial commit in two
individual changes]
Signed-off-by: Dmitri Epshtein <dima@marvell.com>
Signed-off-by: Gregory CLEMENT <gregory.clement@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Function eth_prepare_mac_addr_change() is called as part of MAC
address change. This function check if interface is running.
To enable change MAC address when interface is running:
IFF_LIVE_ADDR_CHANGE flag must be set to dev->priv_flags field
Fixes: c5aff18204 ("net: mvneta: driver for Marvell Armada 370/XP
network unit")
Cc: stable@vger.kernel.org
Signed-off-by: Dmitri Epshtein <dima@marvell.com>
Signed-off-by: Gregory CLEMENT <gregory.clement@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In the previous patch, the spinlock was not initialized. While it didn't
cause any trouble yet it could be a problem to use it uninitialized.
The most annoying part was the critical section protected by the spinlock
in mvneta_stop(). Some of the functions could sleep as pointed when
activated CONFIG_DEBUG_ATOMIC_SLEEP. Actually, in mvneta_stop() we only
need to protect the is_stopped flagged, indeed the code of the notifier
for CPU online is protected by the same spinlock, so when we get the
lock, the notifer work is done.
Reported-by: Patrick Uiterwijk <patrick@puiterwijk.org>
Signed-off-by: Gregory CLEMENT <gregory.clement@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The mvneta_percpu_notifier() hotplug callback lacks handling of the
CPU_DOWN_FAILED case. That means, if CPU_DOWN_PREPARE failes, the
driver is not well configured on the CPU.
Add handling for CPU_DOWN_FAILED[_FROZEN] hotplug notifier transition
to setup the driver.
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: netdev@vger.kernel.org
Signed-off-by: Anna-Maria Gleixner <anna-maria@linutronix.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
Now that the hardware buffer management framework had been introduced,
let's use it.
Tested-by: Sebastian Careba <nitroshift@yahoo.com>
Signed-off-by: Gregory CLEMENT <gregory.clement@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Buffer manager (BM) is a dedicated hardware unit that can be used by all
ethernet ports of Armada XP and 38x SoC's. It allows to offload CPU on RX
path by sparing DRAM access on refilling buffer pool, hardware-based
filling of descriptor ring data and better memory utilization due to HW
arbitration for using 'short' pools for small packets.
Tests performed with A388 SoC working as a network bridge between two
packet generators showed increase of maximum processed 64B packets by
~20k (~555k packets with BM enabled vs ~535 packets without BM). Also
when pushing 1500B-packets with a line rate achieved, CPU load decreased
from around 25% without BM to 20% with BM.
BM comprise up to 4 buffer pointers' (BP) rings kept in DRAM, which
are called external BP pools - BPPE. Allocating and releasing buffer
pointers (BP) to/from BPPE is performed indirectly by write/read access
to a dedicated internal SRAM, where internal BP pools (BPPI) are placed.
BM hardware controls status of BPPE automatically, as well as assigning
proper buffers to RX descriptors. For more details please refer to
Functional Specification of Armada XP or 38x SoC.
In order to enable support for a separate hardware block, common for all
ports, a new driver has to be implemented ('mvneta_bm'). It provides
initialization sequence of address space, clocks, registers, SRAM,
empty pools' structures and also obtaining optional configuration
from DT (please refer to device tree binding documentation). mvneta_bm
exposes also a necessary API to mvneta driver, as well as a dedicated
structure with BM information (bm_priv), whose presence is used as a
flag notifying of BM usage by port. It has to be ensured that mvneta_bm
probe is executed prior to the ones in ports' driver. In case BM is not
used or its probe fails, mvneta falls back to use software buffer
management.
A sequence executed in mvneta_probe function is modified in order to have
an access to needed resources before possible port's BM initialization is
done. According to port-pools mapping provided by DT appropriate registers
are configured and the buffer pools are filled. RX path is modified
accordingly. Becaues the hardware allows a wide variety of configuration
options, following assumptions are made:
* using BM mechanisms can be selectively disabled/enabled basing
on DT configuration among the ports
* 'long' pool's single buffer size is tied to port's MTU
* using 'long' pool by port is obligatory and it cannot be shared
* using 'short' pool for smaller packets is optional
* one 'short' pool can be shared among all ports
This commit enables hardware buffer management operation cooperating with
existing mvneta driver. New device tree binding documentation is added and
the one of mvneta is updated accordingly.
[gregory.clement@free-electrons.com: removed the suspend/resume part]
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Signed-off-by: Gregory CLEMENT <gregory.clement@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When stopping the port, the CPU notifier are still there whereas the
mvneta_stop_dev function calls mvneta_percpu_disable() on each CPUs.
It was possible to have a new CPU coming at this point which could be
racy.
This patch adds a flag preventing executing the code notifier for a new
CPU when the port is stopping. It also uses the spinlock introduces
previously. To avoid the deadlock, the lock has been moved outside the
mvneta_percpu_elect function.
Signed-off-by: Gregory CLEMENT <gregory.clement@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Electing a CPU must be done in an atomic way: it should be done after or
before the removal/insertion of a CPU and this function is not reentrant.
During the loop of mvneta_percpu_elect we associates the queues to the
CPUs, if there is a topology change during this loop, then the mapping
between the CPUs and the queues could be wrong. During this loop the
interrupt mask is also updating for each CPUs, It should not be changed
in the same time by other part of the driver.
This patch adds spinlock to create the needed critical sections.
Signed-off-by: Gregory CLEMENT <gregory.clement@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In the MVNETA_INTR_* registers, the queues related fields are per cpu,
according to the datasheet (comment in [] are added by me):
"In a multi-CPU system, bits of RX[or TX] queues for which the access by
the reading[or writing] CPU is disabled are read as 0, and cannot be
cleared[or written]."
That means that each time we want to manipulate these bits we had to do
it on each cpu and not only on the current cpu.
Signed-off-by: Gregory CLEMENT <gregory.clement@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Since the commit 2dcf75e279 ("net: mvneta: Associate RX queues with
each CPU") all the percpu irq are used and disabled at initialization, so
there is no point to disable them first.
Signed-off-by: Gregory CLEMENT <gregory.clement@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Instead of using a for_each_* loop in which we just call the
smp_call_function_single macro, it is more simple to directly use the
on_each_cpu macro. Moreover, this macro ensures that the calls will be
done all at once.
Suggested-by: Russell King <rmk+kernel@arm.linux.org.uk>
Signed-off-by: Gregory CLEMENT <gregory.clement@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When passing to the management of multiple RX queue, the
mvneta_percpu_elect function was broken. The use of the modulo can lead
to elect the wrong cpu. For example with rxq_def=2, if the CPU 2 goes
offline and then online, we ended with the third RX queue activated in
the same time on CPU 0 and CPU2, which lead to a kernel crash.
With this fix, we don't try to get "the closer" CPU if the default CPU is
gone, now we just use CPU 0 which always be there. Thanks to this, the
code becomes more readable, easier to maintain and more predicable.
Cc: stable@vger.kernel.org
Fixes: 2dcf75e279 ("net: mvneta: Associate RX queues with each CPU")
Signed-off-by: Gregory CLEMENT <gregory.clement@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch convert the for_each_present in on_each_cpu, instead of
applying on the present cpus it will be applied only on the online cpus.
This fix a bug reported on
http://thread.gmane.org/gmane.linux.ports.arm.kernel/468173.
Using the macro on_each_cpu (instead of a for_each_* loop) also ensures
that all the calls will be done all at once.
Fixes: f864288544 ("net: mvneta: Statically assign queues to CPUs")
Reported-by: Stefan Roese <stefan.roese@gmail.com>
Suggested-by: Jisheng Zhang <jszhang@marvell.com>
Suggested-by: Russell King <rmk+kernel@arm.linux.org.uk>
Signed-off-by: Gregory CLEMENT <gregory.clement@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Some platforms may provide more than one clk for the mvneta IP, for
example Marvell BG4CT provides one clk for the mac core, and one
clk for the AXI bus logic. Obviously this bus clk also need to
be enabled. This patch adds this optional "bus" clk support.
Signed-off-by: Jisheng Zhang <jszhang@marvell.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Some platforms may provide more than one clk for the mvneta IP, for
example Marvell BG4CT provides one clk for the mac core, and one
clk for the AXI bus logic.
To support for more than one clock, we'll need to distinguish between
the clock by name. Change clock probing to first try to get "core"
clock before falling back to unnamed clock.
Signed-off-by: Jisheng Zhang <jszhang@marvell.com>
Acked-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Sorting the headers in alphabetic order will help to reduce the conflict
when adding new headers in the future.
Signed-off-by: Jisheng Zhang <jszhang@marvell.com>
Acked-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When s->type is T_REG_64, the high 32bits are lost in val. This patch
fixes this trivial issue.
Signed-off-by: Jisheng Zhang <jszhang@marvell.com>
Fixes: 9b0cdefa4c ("net: mvneta: add ethtool statistics")
Signed-off-by: David S. Miller <davem@davemloft.net>
Not all devices attached to an MDIO bus are phys. So add an
mdio_device structure to represent the generic parts of an mdio
device, and place this structure into the phy_device.
Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: David S. Miller <davem@davemloft.net>
With this patch each CPU is associated with its own set of TX queues.
It also setup the XPS with an initial configuration which set the
affinity matching the hardware configuration.
Suggested-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Gregory CLEMENT <gregory.clement@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch adds the support for the RSS related ethtool
function. Currently it only uses one entry in the indirection table which
allows associating an mvneta interface to a given CPU.
Signed-off-by: Gregory CLEMENT <gregory.clement@free-electrons.com>
Tested-by: Marcin Wojtas <mw@semihalf.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
We enable the percpu interrupt for all the CPU and we just associate a
CPU to a few queue at the neta level. The mapping between the CPUs and
the queues is static. The queues are associated to the CPU module the
number of CPUs. However currently we only use on RX queue for a given
Ethernet port.
Signed-off-by: Gregory CLEMENT <gregory.clement@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Instead of using the same default queue for all the port. Move it in the
port struct. It will allow have a different default queue for each port.
Signed-off-by: Gregory CLEMENT <gregory.clement@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Conflicts:
drivers/net/ethernet/renesas/ravb_main.c
kernel/bpf/syscall.c
net/ipv4/ipmr.c
All three conflicts were cases of overlapping changes.
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch allows to do
ethtool -s eth0 autoneg off
ethtool -s eth0 autoneg on
to disable or enable autonegotiation at run-time.
Without that functionality, the only way to control the autonegotiation
is to modify the device tree.
This is needed if you plan to use the same kernel with
different ethernet switches, the ones that support the in-band
status and the ones that not.
CC: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
CC: netdev@vger.kernel.org
CC: linux-kernel@vger.kernel.org
Signed-off-by: Stas Sergeev <stsp@users.sourceforge.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
This moves autoneg-related bit manipulations to the single place.
CC: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
CC: netdev@vger.kernel.org
CC: linux-kernel@vger.kernel.org
Signed-off-by: Stas Sergeev <stsp@users.sourceforge.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
Since Armada 38x SoC can support IP checksum for jumbo frames only on
a single port, it means that this feature should be enabled per-port,
rather than for the whole SoC.
This patch enables setting custom TX IP checksum limit by adding new
optional property to the mvneta device tree node. If not used, by
default 1600B is set for "marvell,armada-370-neta" and 9800B for other
strings, which ensures backward compatibility. Binding documentation
is updated accordingly.
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In the actual RX processing, there is same error path for both descriptor
ring refilling and building skb fails. This is not correct, because after
successful refill, the ring is already updated with newly allocated
buffer. Then, in case of build_skb() fail, hitherto code left the original
buffer unmapped.
This patch fixes above situation by swapping error check of skb build with
DMA-unmap of original buffer.
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Acked-by: Simon Guinot <simon.guinot@sequanux.org>
Cc: <stable@vger.kernel.org> # v4.2+
Fixes a84e328941 ("net: mvneta: fix refilling for Rx DMA buffers")
Signed-off-by: David S. Miller <davem@davemloft.net>
A value originally defined in the driver was inappropriate. Even though
the ingress was somehow working, writing MVNETA_RXQ_INTR_ENABLE_ALL_MASK
to MVNETA_INTR_ENABLE didn't make any effect, because the bits [31:16]
are reserved and read-only.
This commit updates MVNETA_RXQ_INTR_ENABLE_ALL_MASK to be compliant with
the controller's documentation.
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Fixes: c5aff18204 ("net: mvneta: driver for Marvell Armada 370/XP network
unit")
Signed-off-by: David S. Miller <davem@davemloft.net>
MVNETA_RXQ_HW_BUF_ALLOC bit which controls enabling hardware buffer
allocation was mistakenly set as BIT(1). This commit fixes the assignment.
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Reviewed-by: Gregory CLEMENT <gregory.clement@free-electrons.com>
Fixes: c5aff18204 ("net: mvneta: driver for Marvell Armada 370/XP network
unit")
Signed-off-by: David S. Miller <davem@davemloft.net>
This commit adds missing configuration of MBUS windows access protection
in mvneta_conf_mbus_windows function - a dedicated variable for that
purpose remained there unused since v3.8 initial mvneta support. Because
of that the register contents were inherited from the bootloader.
Signed-off-by: Marcin Wojtas <mw@semihalf.com>
Reviewed-by: Gregory CLEMENT <gregory.clement@free-electrons.com>
Fixes: c5aff18204 ("net: mvneta: driver for Marvell Armada 370/XP network
unit")
Signed-off-by: David S. Miller <davem@davemloft.net>
After changing an interface's MTU, then bringing the interface down and
back up again, I immediately saw tons of kernel messages like below.
The reason for this bad behavior is mvneta_rxq_drop_pkts(), which calls
dma_unmap_single() on already-freed memory. So we need to switch the
order of those two operations.
[ 152.388518] BUG: Bad page state in process ifconfig pfn:1b518
[ 152.388526] page:dff3dbc0 count:0 mapcount:0 mapping: (null) index:0x0
[ 152.395178] flags: 0x200(arch_1)
[ 152.398441] page dumped because: PAGE_FLAGS_CHECK_AT_PREP flag set
[ 152.398446] bad because of flags:
[ 152.398450] flags: 0x200(arch_1)
[ 152.401716] Modules linked in:
[ 152.401728] CPU: 0 PID: 1453 Comm: ifconfig Tainted: P B O 4.1.12.armada.1 #1
[ 152.401733] Hardware name: Marvell Armada 370/XP (Device Tree)
[ 152.401749] [<c0015b1c>] (unwind_backtrace) from [<c0011d8c>] (show_stack+0x10/0x14)
[ 152.401762] [<c0011d8c>] (show_stack) from [<c06aa68c>] (dump_stack+0x74/0x90)
[ 152.401772] [<c06aa68c>] (dump_stack) from [<c0096c08>] (bad_page+0xc4/0x124)
[ 152.401783] [<c0096c08>] (bad_page) from [<c0099378>] (get_page_from_freelist+0x4e4/0x644)
[ 152.401794] [<c0099378>] (get_page_from_freelist) from [<c0099620>] (__alloc_pages_nodemask+0x148/0x784)
[ 152.401805] [<c0099620>] (__alloc_pages_nodemask) from [<c00ac658>] (kmalloc_order+0x10/0x20)
[ 152.401818] [<c00ac658>] (kmalloc_order) from [<c04c6f44>] (mvneta_rx_refill+0xc4/0xe8)
[ 152.401830] [<c04c6f44>] (mvneta_rx_refill) from [<c04c96c0>] (mvneta_setup_rxqs+0x298/0x39c)
[ 152.401842] [<c04c96c0>] (mvneta_setup_rxqs) from [<c04c9904>] (mvneta_open+0x3c/0x150)
[ 152.401853] [<c04c9904>] (mvneta_open) from [<c0597764>] (__dev_open+0xac/0x124)
[ 152.401864] [<c0597764>] (__dev_open) from [<c05979e4>] (__dev_change_flags+0x8c/0x148)
[ 152.401875] [<c05979e4>] (__dev_change_flags) from [<c0597ac0>] (dev_change_flags+0x18/0x48)
[ 152.401886] [<c0597ac0>] (dev_change_flags) from [<c060d308>] (devinet_ioctl+0x620/0x6d0)
[ 152.401897] [<c060d308>] (devinet_ioctl) from [<c057d810>] (sock_ioctl+0x64/0x288)
[ 152.401908] [<c057d810>] (sock_ioctl) from [<c00dcb7c>] (do_vfs_ioctl+0x78/0x608)
[ 152.401918] [<c00dcb7c>] (do_vfs_ioctl) from [<c00dd170>] (SyS_ioctl+0x64/0x74)
[ 152.401930] [<c00dd170>] (SyS_ioctl) from [<c000f3a0>] (ret_fast_syscall+0x0/0x3c)
Signed-off-by: Justin Maggard <jmaggard@netgear.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The existing function to clear the MIB statatistics was using the
wrong address for the registers. Also, the counters would of been
cleared when the interface was brought up, not during the
probe. Fix both of these.
Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
Add support for the ethtool statistic interface, returning the full set
of statistics which both Armada 370, 38x and Armada XP can support.
Tested-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
Since the switch to per-CPU interrupts, we lost the ability to set which
CPU was going to receive our RX interrupt, which was now only the CPU on
which the mvneta_open function was run.
We can now assign our queues to their respective CPUs, and make sure only
this CPU is going to handle our traffic.
This also paves the road to be able to change that at runtime, and later on
to support RSS.
[gregory.clement@free-electrons.com]: hardened the CPU hotplug support.
Signed-off-by: Maxime Ripard <maxime.ripard@free-electrons.com>
Signed-off-by: Gregory CLEMENT <gregory.clement@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The mvneta driver allows to change the default RX queue trough the rxq_def
kernel parameter.
However, the current code doesn't allow to have any value but 0. It is
actively checked for in the driver's probe because the drivers makes a
number of assumption and takes a number of shortcuts in order to just use
that RX queue.
Remove these limitations in order to be able to specify any available
queue.
Signed-off-by: Maxime Ripard <maxime.ripard@free-electrons.com>
Signed-off-by: Gregory CLEMENT <gregory.clement@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Now that our interrupt controller is allowing us to use per-CPU interrupts,
actually use it in the mvneta driver.
This involves obviously reworking the driver to have a CPU-local NAPI
structure, and report for incoming packet using that structure.
Signed-off-by: Maxime Ripard <maxime.ripard@free-electrons.com>
Signed-off-by: Gregory CLEMENT <gregory.clement@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The CPU_MAP register is duplicated for each CPUs at different addresses,
each instance being at a different address.
However, the code so far was using CONFIG_NR_CPUS to initialise the CPU_MAP
registers for each registers, while the SoCs embed at most 4 CPUs.
This is especially an issue with multi_v7_defconfig, where CONFIG_NR_CPUS
is currently set to 16, resulting in writes to registers that are not
CPU_MAP.
Fixes: c5aff18204 ("net: mvneta: driver for Marvell Armada 370/XP network unit")
Signed-off-by: Maxime Ripard <maxime.ripard@free-electrons.com>
Cc: <stable@vger.kernel.org> # v3.8+
Signed-off-by: Gregory CLEMENT <gregory.clement@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
of_phy_find_device() increments the phy struct device refcount, which
we need to properly balance. Add code to network drivers using this
function to ensure that the struct device refcount is correctly
balanced.
For xgene, looking back in the history, we should be able to use
of_phy_connect() with a zero flags argument for the DT case as this is
how the driver used to operate prior to de7b5b3d79 ("net: eth: xgene:
change APM X-Gene SoC platform ethernet to support ACPI").
This leaves the Cavium Thunder BGX unfixed; fixing this driver is a
complicated task, one which the maintainers need to be involved with.
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch fixes a regression introduced by the commit a84e328941
("net: mvneta: fix refilling for Rx DMA buffers"). Due to this commit
the newly allocated Rx buffers are DMA-unmapped in place of those passed
to the networking stack. Obviously, this causes data corruptions.
This patch fixes the issue by ensuring that the right Rx buffers are
DMA-unmapped.
Reported-by: Oren Laskin <oren@igneous.io>
Signed-off-by: Simon Guinot <simon.guinot@sequanux.org>
Fixes: a84e328941 ("net: mvneta: fix refilling for Rx DMA buffers")
Cc: <stable@vger.kernel.org> # v3.8+
Tested-by: Oren Laskin <oren@igneous.io>
Signed-off-by: David S. Miller <davem@davemloft.net>
Conflicts:
net/bridge/br_mdb.c
br_mdb.c conflict was a function call being removed to fix a bug in
'net' but whose signature was changed in 'net-next'.
Signed-off-by: David S. Miller <davem@davemloft.net>
The commit 898b2970e2 ("mvneta: implement SGMII-based in-band link state
signaling") implemented the link parameters auto-negotiation unconditionally.
Unfortunately it appears that some HW that implements SGMII protocol,
doesn't generate the inband status, so it is not possible to auto-negotiate
anything with such HW.
This patch enables the auto-negotiation only if explicitly requested with
the 'managed' DT property.
This patch fixes the following regression:
https://lkml.org/lkml/2015/7/8/865
Signed-off-by: Stas Sergeev <stsp@users.sourceforge.net>
CC: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
CC: netdev@vger.kernel.org
CC: linux-kernel@vger.kernel.org
Signed-off-by: David S. Miller <davem@davemloft.net>
With the actual code, if a memory allocation error happens while
refilling a Rx descriptor, then the original Rx buffer is both passed
to the networking stack (in a SKB) and let in the Rx ring. This leads
to various kernel oops and crashes.
As a fix, this patch moves Rx descriptor refilling ahead of building
SKB with the associated Rx buffer. In case of a memory allocation
failure, data is dropped and the original DMA buffer is put back into
the Rx ring.
Signed-off-by: Simon Guinot <simon.guinot@sequanux.org>
Fixes: c5aff18204 ("net: mvneta: driver for Marvell Armada 370/XP network unit")
Cc: <stable@vger.kernel.org> # v3.8+
Tested-by: Yoann Sculo <yoann@sculo.fr>
Signed-off-by: David S. Miller <davem@davemloft.net>
The Ethernet controller found in the Armada 370, 380 and 385 SoCs don't
support TCP/IP checksumming with frame sizes larger than 1600 bytes.
This patch fixes the issue by disabling the features NETIF_F_IP_CSUM and
NETIF_F_TSO for the Armada 370 and compatibles SoCs when the MTU is set
to a value greater than 1600 bytes.
Signed-off-by: Simon Guinot <simon.guinot@sequanux.org>
Fixes: c5aff18204 ("net: mvneta: driver for Marvell Armada 370/XP network unit")
Cc: <stable@vger.kernel.org> # v3.8+
Acked-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The mvneta driver supports the Ethernet IP found in the Armada 370, XP,
380 and 385 SoCs. Since at least one more hardware feature is available
for the Armada XP SoCs then a way to identify them is needed.
This patch introduces a new compatible string "marvell,armada-xp-neta".
Signed-off-by: Simon Guinot <simon.guinot@sequanux.org>
Fixes: c5aff18204 ("net: mvneta: driver for Marvell Armada 370/XP network unit")
Cc: <stable@vger.kernel.org> # v3.8+
Acked-by: Gregory CLEMENT <gregory.clement@free-electrons.com>
Acked-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Conflicts:
drivers/net/ethernet/mellanox/mlx4/main.c
net/packet/af_packet.c
Both conflicts were cases of simple overlapping changes.
Signed-off-by: David S. Miller <davem@davemloft.net>
The commit 898b2970e2 ("mvneta: implement SGMII-based in-band link state
signaling")
changed mvneta_adjust_link() so that it does not clear the auto-negotiation
bits in MVNETA_GMAC_AUTONEG_CONFIG register. This was necessary for
auto-negotiation mode to work.
Unfortunately I haven't checked if these bits are ever initialized.
It appears they are not.
This patch adds the missing initialization of the auto-negotiation bits
in the MVNETA_GMAC_AUTONEG_CONFIG register.
It fixes the following regression:
https://www.mail-archive.com/netdev@vger.kernel.org/msg67928.html
Since the patch was tested to fix a regression, it should be applied to
stable tree.
Tested-by: Arnaud Ebalard <arno@natisbad.org>
CC: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
CC: Florian Fainelli <f.fainelli@gmail.com>
CC: netdev@vger.kernel.org
CC: linux-kernel@vger.kernel.org
CC: stable@vger.kernel.org
Signed-off-by: Stas Sergeev <stsp@users.sourceforge.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
Conflicts:
drivers/net/ethernet/mellanox/mlx4/cmd.c
net/core/fib_rules.c
net/ipv4/fib_frontend.c
The fib_rules.c and fib_frontend.c conflicts were locking adjustments
in 'net' overlapping addition and removal of code in 'net-next'.
The mlx4 conflict was a bug fix in 'net' happening in the same
place a constant was being replaced with a more suitable macro.
Signed-off-by: David S. Miller <davem@davemloft.net>
mvneta_adjust_link() is a callback for of_phy_connect() and should
not be called directly. The result of calling it directly is as below:
Signed-off-by: David S. Miller <davem@davemloft.net>
When MDIO bus is unavailable (common setup for SGMII), the in-band
signaling must be used to correctly track link state.
This patch enables the in-band status delivery for link state changes, namely:
- link up/down
- link speed
- duplex full/half
fixed_phy_update_state() is used to update phy status.
CC: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
CC: Florian Fainelli <f.fainelli@gmail.com>
CC: netdev@vger.kernel.org
CC: linux-kernel@vger.kernel.org
Signed-off-by: Stas Sergeev <stsp@users.sourceforge.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
Conflicts:
drivers/net/ethernet/amd/xgbe/xgbe-desc.c
drivers/net/ethernet/renesas/sh_eth.c
Overlapping changes in both conflict cases.
Signed-off-by: David S. Miller <davem@davemloft.net>
mvneta_tx() dereferences skb to get skb->len too late,
as hardware might have completed the transmit and TX completion
could have freed the skb from another cpu.
Fixes: 71f6d1b31f ("net: mvneta: replace Tx timer with a real interrupt")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The mvneta driver sets the amount of Tx coalesce packets to 16 by
default. Normally that does not cause any trouble since the driver
uses a much larger Tx ring size (532 packets). But some sockets
might run with very small buffers, much smaller than the equivalent
of 16 packets. This is what ping is doing for example, by setting
SNDBUF to 324 bytes rounded up to 2kB by the kernel.
The problem is that there is no documented method to force a specific
packet to emit an interrupt (eg: the last of the ring) nor is it
possible to make the NIC emit an interrupt after a given delay.
In this case, it causes trouble, because when ping sends packets over
its raw socket, the few first packets leave the system, and the first
15 packets will be emitted without an IRQ being generated, so without
the skbs being freed. And since the socket's buffer is small, there's
no way to reach that amount of packets, and the ping ends up with
"send: no buffer available" after sending 6 packets. Running with 3
instances of ping in parallel is enough to hide the problem, because
with 6 packets per instance, that's 18 packets total, which is enough
to grant a Tx interrupt before all are sent.
The original driver in the LSP kernel worked around this design flaw
by using a software timer to clean up the Tx descriptors. This timer
was slow and caused terrible network performance on some Tx-bound
workloads (such as routing) but was enough to make tools like ping
work correctly.
Instead here, we simply set the packet counts before interrupt to 1.
This ensures that each packet sent will produce an interrupt. NAPI
takes care of coalescing interrupts since the interrupt is disabled
once generated.
No measurable performance impact nor CPU usage were observed on small
nor large packets, including when saturating the link on Tx, and this
fixes tools like ping which rely on too small a send buffer. If one
wants to increase this value for certain workloads where it is safe
to do so, "ethtool -C $dev tx-frames" will override this default
setting.
This fix needs to be applied to stable kernels starting with 3.10.
Tested-By: Maggie Mae Roxas <maggie.mae.roxas@gmail.com>
Signed-off-by: Willy Tarreau <w@1wt.eu>
Signed-off-by: David S. Miller <davem@davemloft.net>
Use phy_print_status() to report a change in the PHY status.
The current message is not verbose enough, so this commit improves
it by using the generic status message.
After this change, the kernel reports PHY status down and up events as:
mvneta f1070000.ethernet eth0: Link is Down
mvneta f1070000.ethernet eth0: Link is Up - 1Gbps/Full - flow control rx/tx
Signed-off-by: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
drivers/net/ethernet/marvell/mvneta.c: In function 'mvneta_skb_tx_csum':
drivers/net/ethernet/marvell/mvneta.c:1374:3: error: implicit declaration of function 'vlan_get_protocol' [-Werror=implicit-function-declaration]
__be16 l3_proto = vlan_get_protocol(skb);
^
Reporeted-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: David S. Miller <davem@davemloft.net>
This driver doesn't appear to support vlan acceleration at
all. However, it does claim to support TSO and IP checksums
for vlan devices. Thus any configured vlan device would
end up passing down partial checksums or TSO frames.
The driver also uses the value from skb->protocol to
determine TSO and checksum offload information, but assumes
that skb->protocol holds the l3 protocol information.
As a result, vlan traffic with partial checksums or TSO
will fail those checks and TSO will not happen.
Fix this by using vlan_get_protocol() helper.
CC: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: Vladislav Yasevich <vyasevic@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
If there is a "phy" handle the probe function returns with holding a
reference to that node. Make sure that in the fixed phy case there is
also held a reference to yield a consistant state.
Also add the corresponding of_node_put in the error path and the remove
function.
Fixes: 83895bedee ("net: mvneta: add support for fixed links")
Fixes: c5aff18204 ("net: mvneta: driver for Marvell Armada 370/XP network unit")
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
This commit fixes the command value generated for CSUM calculation
when running in big endian mode. The Ethernet protocol ID for IP was
being unconditionally byte-swapped in the layer 3 protocol check (with
swab16), which caused the mvneta driver to not function correctly in
big endian mode. This patch byte-swaps the ID conditionally with
htons.
Cc: <stable@vger.kernel.org> # v3.13+
Signed-off-by: Thomas Fitzsimmons <fitzsim@fitzsim.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
As reported by Maggie Mae Roxas, the mvneta driver doesn't behave
properly in 10 Mbit/s mode. This is due to a misconfiguration of the
MVNETA_GMAC_AUTONEG_CONFIG register: bit MVNETA_GMAC_CONFIG_MII_SPEED
must be set for a 100 Mbit/s speed, but cleared for a 10 Mbit/s speed,
which the driver was not properly doing. This commit adjusts that by
setting the MVNETA_GMAC_CONFIG_MII_SPEED bit only in 100 Mbit/s mode,
and relying on the fact that all the speed related bits of this
register are cleared at the beginning of the mvneta_adjust_link()
function.
This problem exists since c5aff18204 ("net: mvneta: driver for
Marvell Armada 370/XP network unit") which is the commit that
introduced the mvneta driver in the kernel.
Cc: <stable@vger.kernel.org> # v3.8+
Fixes: c5aff18204 ("net: mvneta: driver for Marvell Armada 370/XP network unit")
Reported-by: Maggie Mae Roxas <maggie.mae.roxas@gmail.com>
Cc: Maggie Mae Roxas <maggie.mae.roxas@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The buffers for the TSO headers belong to a DMA coherent region which is
allocated at ndo_open() time, and released at ndo_stop() time.
Therefore, and contrary to the TSO payload descriptor buffers, the TSO header
buffers don't need to be unmapped. This commit adds a check to detect a
TSO header buffer and explicitly prevent the unmap.
Signed-off-by: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The Tx descriptor release code currently calls dma_unmap_single() and
dev_kfree_skb_any() if the descriptor is associated with a non-NULL skb.
This is true only for the last fragment of the packet.
This is wrong, however, since every descriptor buffer is DMA mapped and needs
to be unmapped.
Signed-off-by: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Currently small MSS values may require too many TSO descriptors for
the default queue size. This commit prevents this situation by fixing
the maximum supported TSO number of segments to 100 and by setting a
minimum Tx queue size. The minimum Tx queue size is set so that at
least 2 worst-case skb can be accommodated.
In addition, the queue stop and wake thresholds values are adjusted
accordingly. The queue is stopped when there's room for only 1 worst-case
skb and waked when the number of descriptors is half that value.
Signed-off-by: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This driver has no need for a custom NAPI weigth. Use the default
one, which has the same value.
Signed-off-by: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The 'weight' field is only used to pass the weigth to napi initialization
function. This commit removes the field, and instead uses a fixed value to
initialize the napi context.
Signed-off-by: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The driver does not support multiple rx queues, and so it's a waste
of resources to have a default number larger than one (1).
Signed-off-by: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Use eth_prepare_mac_addr_change and eth_commit_mac_addr_change, instead
of manually checking and storing the MAC address, which makes the
code slightly more robust. This fixes the lack of valid MAC address check
in the driver's .ndo_set_mac_address hook.
Signed-off-by: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This commit cleans-up mvneta_init(), which initializes the hardware
and allocates the rx/qx queues. The queue allocation is simplified
by using devm_kcalloc instead of kzalloc. The unused phy_addr parameter
is removed. While here, the 'hal' references in the comments are removed.
This commit makes no functionality change.
Signed-off-by: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This commit checks the return code of mvneta_setup_txq() call
in mvneta_change_mtu(). Also, use the netdevice pointer directly
instead of dereferencing the port structure. While here, let's
fix a tiny comment typo.
Signed-off-by: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
A tiny clean-up to improve readability. This commit makes no functionality
change.
Signed-off-by: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Now that the TSO helper API has been introduced, this commit makes use
of it to implement the TSO in this driver.
Using iperf to test and vmstat to check the CPU usage, shows a substantial
CPU usage drop when TSO is on (~15% vs. ~25%). HTTP-based tests performed
by Willy Tarreau have shown performance improvements.
Signed-off-by: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Rework mvneta_tx() so that the code that performs the final handling
before a sk_buff is transmitted is done only if the numbers of fragments
processed if positive.
This is preparation work to add the support for software TSO.
Signed-off-by: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
In order to ease the addition of new features, let's factorize the
feature list.
Signed-off-by: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Following the introduction of of_phy_register_fixed_link(), this patch
introduces fixed link support in the mvneta driver, for Marvell Armada
370/XP SOCs.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
net: get rid of SET_ETHTOOL_OPS
Dave Miller mentioned he'd like to see SET_ETHTOOL_OPS gone.
This does that.
Mostly done via coccinelle script:
@@
struct ethtool_ops *ops;
struct net_device *dev;
@@
- SET_ETHTOOL_OPS(dev, ops);
+ dev->ethtool_ops = ops;
Compile tested only, but I'd seriously wonder if this broke anything.
Suggested-by: Dave Miller <davem@davemloft.net>
Signed-off-by: Wilfried Klaebe <w-lkml@lebenslange-mailadresse.de>
Acked-by: Felipe Balbi <balbi@ti.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Commit 5445eaf309 ('mvneta: Try to fix mvneta when compiled as
module') fixed the mvneta driver to make it work properly when loaded
as a module in SGMII configuration, which was tested successful by the
author on the Armada XP OpenBlocks AX3, which uses SGMII.
However, some other platforms, namely the Armada XP GP don't use
SGMII, but a QSGMII connection between the MAC and the PHY, and this
case was not supported by the mvneta driver, which was relying on
configuration put in place by the bootloader. While this works when
the mvneta driver is built-in (because clocks are not gated), it
breaks when mvneta is built as a module, because the clock is gated
(all configuration is lost) and then re-enabled when the mvneta driver
is loaded.
In order to support all of RGMII, SGMII and QSGMII, this commit
reworks how the PHY interface configuration is done, and simplifies
it: it removes the mvneta_port_sgmii_config() and
mvneta_gmac_rgmii_set() functions, which were strange because
mvneta_gmac_rgmii_set() was called in all cases, even for SGMII
configurations. Also, the mvneta_gmac_rgmii_set() function was taking
a boolean as argument, which was always true.
Instead, all the PHY interface configuration logic is moved into the
mvneta_port_power_up() function, in a much simpler 'switch' construct,
with four cases:
- QSGMII: the RGMIIEn bit, the PCSEn bit in GMAC_CTRL_2 are set, and
the SERDES is configured in QSGMII. Technically speaking,
configuring the SERDES of the first port would be sufficient, but
it is simpler to do it on all ports.
- SGMII: the RGMIIEn bit, the PCSEn bit in GMAC_CTRL_2 are set, and
the SERDES is configured as SGMII.
- RGMII: the RGMIIEn bit in GMAC_CTRL_2 is set. The PCSEn bit is kept
cleared, and no SERDES configuration is done, because RGMII is not
using SERDES lanes.
- other: an error is returned. For this reason, the
mvneta_port_power_up() now returns an int instead of nothing, and
the return value is checked by mvneta_probe().
This has been successfully tested on:
* Armada XP DB, which has two RGMII and two SGMII connections
* Armada XP GP, which uses QSGMII for its four interfaces
* Armada 370 Mirabox, which has two RGMII connections
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This reverts commit e3a8786c10. While
this commit allows to use the mvneta driver as a module on some
configurations, it breaks other configurations even if mvneta is used
built-in.
This breakage is due to the fact that on some RGMII platforms, the PCS
bit has to be set, and on some other platforms, it has to be
cleared. At the moment, we lack informations to know exactly the
significance of this bit (the datasheet only says "enables PCS"), and
so we can't produce a patch that will work on all platforms at this
point. And since this change is breaking the network completely for
many users, it's much better to revert it for now. We'll come back
later with a proper fix that takes into account all platforms.
Basically:
* Armada XP GP is configured as RGMII-ID, and needs the PCS bit to be
set.
* Armada 370 Mirabox is configured as RGMII-ID, and needs the PCS bit
to be cleared.
And at the moment, we don't know how to make the distinction between
those two cases. One hint is that the Armada XP GP appears in fact to
be using a QSGMII connection with the PHY (Quad-SGMII), but
configuring it as SGMII doesn't work, while RGMII-ID works. This needs
more investigation, but in the mean time, let's unbreak the network
for all those users.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Reported-by: Arnaud Ebalard <arno@natisbad.org>
Reported-by: Alexander Reuter <Alexander.Reuter@gmx.net>
Fixes: https://bugzilla.kernel.org/show_bug.cgi?id=73401
Cc: stable@vger.kernel.org
Signed-off-by: David S. Miller <davem@davemloft.net>
Conflicts:
drivers/net/ethernet/marvell/mvneta.c
The mvneta.c conflict is a case of overlapping changes,
a conversion to devm_ioremap_resource() vs. a conversion
to netdev_alloc_pcpu_stats.
Signed-off-by: David S. Miller <davem@davemloft.net>
The mvneta driver currently uses of_iomap(), which has two drawbacks:
it doesn't request the resource, and it isn't devm-style so some error
handling is needed.
This commit switches to use devm_ioremap_resource() instead, which
automatically requests the resource (so the I/O registers region shows
up properly in /proc/iomem), and also is devm-style, which allows to
get rid of some error handling to unmap the I/O registers region.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The mvneta driver currently uses of_iomap(), which has two drawbacks:
it doesn't request the resource, and it isn't devm-style so some error
handling is needed.
This commit switches to use devm_ioremap_resource() instead, which
automatically requests the resource (so the I/O registers region shows
up properly in /proc/iomem), and also is devm-style, which allows to
get rid of some error handling to unmap the I/O registers region.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Commit 5445eaf309 ('mvneta: Try to fix mvneta when compiled as
module') fixed the mvneta driver to make it work properly when loaded
as a module in SGMII configuration, which was tested successful by the
author on the Armada XP OpenBlocks AX3, which uses SGMII.
However, it turns out that the Armada XP GP, which uses RGMII, is
affected by a similar problem: its SERDES configuration is lost when
mvneta is loaded as a module, because this configuration is set by the
bootloader, and then lost because the clock is gated by the clock
framework until the mvneta driver is loaded again and the clock is
re-enabled.
However, it turns out that for the RGMII case, setting the SERDES
configuration is not sufficient: the PCS enable bit in the
MVNETA_GMAC_CTRL_2 register must also be set, like in the SGMII
configuration.
Therefore, this commit reworks the SGMII/RGMII initialization: the
only difference between the two now is a different SERDES
configuration, all the rest is identical.
In detail, to achieve this, the commit:
* Renames MVNETA_SGMII_SERDES_CFG to MVNETA_SERDES_CFG because it is
not specific to SGMII, but also used on RGMII configurations.
* Adds a MVNETA_RGMII_SERDES_PROTO definition, that must be used as
the MVNETA_SERDES_CFG value in RGMII configurations.
* Removes the mvneta_gmac_rgmii_set() and mvneta_port_sgmii_config()
functions, and instead directly do the SGMII/RGMII configuration in
mvneta_port_up(), from where those functions where called. It is
worth mentioning that mvneta_gmac_rgmii_set() had an 'enable'
parameter that was always passed as '1', so it was pretty useless.
* Reworks the mvneta_port_up() function to set the MVNETA_SERDES_CFG
register to the appropriate value depending on the RGMII vs. SGMII
configuration. It also unconditionally set the PCS_ENABLE bit (was
already done for SGMII, but is now also needed for RGMII), and sets
the PORT_RGMII bit (which was already done for both SGMII and
RGMII).
This commit was successfully tested with mvneta compiled as a module,
on both the OpenBlocks AX3 (SGMII configuration) and the Armada XP GP
(RGMII configuration).
Reported-by: Steve McIntyre <steve@einval.com>
Cc: stable@vger.kernel.org # 3.11.x: 5445eaf309 mvneta: Try to fix mvneta when compiled as module
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Bit 3 of the MVNETA_GMAC_CTRL_2 is actually used to enable the PCS,
not the PSC: there was a typo in the name of the define, which this
commit fixes.
Cc: stable@vger.kernel.org
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Replace the bh safe variant with the hard irq safe variant.
We need a hard irq safe variant to deal with netpoll transmitting
packets from hard irq context, and we need it in most if not all of
the places using the bh safe variant.
Except on 32bit uni-processor the code is exactly the same so don't
bother with a bh variant, just have a hard irq safe variant that
everyone can use.
Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
There are many drivers calling alloc_percpu() to allocate pcpu stats
and then initializing ->syncp. So just introduce a helper function for them.
Cc: David S. Miller <davem@davemloft.net>
Signed-off-by: Cong Wang <xiyou.wangcong@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The function return parameter is not used in mvneta_tx_done_gbe(),
where the function is called. This patch makes the function return
void.
Reviewed-by: Willy Tarreau <w@1wt.eu>
Signed-off-by: Arnaud Ebalard <arno@natisbad.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
mvneta_tx_done_gbe() return value and third parameter are no more
used. This patch changes the function prototype and removes a useless
variable where the function is called.
Reviewed-by: Willy Tarreau <w@1wt.eu>
Signed-off-by: Arnaud Ebalard <arno@natisbad.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
calling dma_map_single()/dma_unmap_single() is quite expensive compared
to copying a small packet. So let's copy short frames and keep the buffers
mapped. We set the limit to 256 bytes which seems to give good results both
on the XP-GP board and on the AX3/4.
The Rx small packet rate increased by 16.4% doing this, from 486kpps to
573kpps. It is worth noting that even the call to the function
dma_sync_single_range_for_cpu() is expensive (300 ns) although less
than dma_unmap_single(). Without it, the packet rate raises to 711kpps
(+24% more). Thus on systems where coherency from device to CPU is
guaranteed by a snoop control unit, this patch should provide even more
gains, and probably rx_copybreak could be increased.
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Gregory CLEMENT <gregory.clement@free-electrons.com>
Tested-by: Arnaud Ebalard <arno@natisbad.org>
Signed-off-by: Willy Tarreau <w@1wt.eu>
Signed-off-by: David S. Miller <davem@davemloft.net>
Make use of build_skb() to allocate frags on the RX path. When frag size
is lower than a page size, we can use netdev_alloc_frag(), and we fall back
to kmalloc() for larger sizes. The frag size is stored into the mvneta_port
struct. The alloc/free functions check the frag size to decide what alloc/
free method to use. MTU changes are safe because the MTU change function
stops the device and clears the queues before applying the change.
With this patch, I observed a reproducible 2% performance improvement on
HTTP-based benchmarks, and 5% on small packet RX rate.
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Gregory CLEMENT <gregory.clement@free-electrons.com>
Tested-by: Arnaud Ebalard <arno@natisbad.org>
Signed-off-by: Willy Tarreau <w@1wt.eu>
Signed-off-by: David S. Miller <davem@davemloft.net>
Currently, the mvneta driver tries to prefetch the current Rx
descriptor during read. Tests have shown that prefetching the
next one instead increases general performance by about 1% on
HTTP traffic.
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Gregory CLEMENT <gregory.clement@free-electrons.com>
Tested-by: Arnaud Ebalard <arno@natisbad.org>
Signed-off-by: Willy Tarreau <w@1wt.eu>
Signed-off-by: David S. Miller <davem@davemloft.net>
At several places, we already know the value of the rx status but
we call functions which dereference the pointer again to get it
and don't need the descriptor for anything else. Simplify this
task by replacing the rx desc pointer by the status word itself.
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Gregory CLEMENT <gregory.clement@free-electrons.com>
Tested-by: Arnaud Ebalard <arno@natisbad.org>
Signed-off-by: Willy Tarreau <w@1wt.eu>
Signed-off-by: David S. Miller <davem@davemloft.net>
Make mvneta_rxq_fill() use mvneta_rx_refill() instead of using
duplicate code.
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Gregory CLEMENT <gregory.clement@free-electrons.com>
Tested-by: Arnaud Ebalard <arno@natisbad.org>
Signed-off-by: Willy Tarreau <w@1wt.eu>
Signed-off-by: David S. Miller <davem@davemloft.net>
Currently, mvneta_txq_bufs_free() calls mvneta_tx_done_policy() with
a non-null cause to retrieve the pointer to the next queue to process.
There are useless tests on the return queue number and on the pointer,
all of which are well defined within a known limited set. This code
path is fast, although not critical. Removing 3 tests here that the
compiler could not optimize (verified) is always desirable.
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Gregory CLEMENT <gregory.clement@free-electrons.com>
Tested-by: Arnaud Ebalard <arno@natisbad.org>
Signed-off-by: Willy Tarreau <w@1wt.eu>
Signed-off-by: David S. Miller <davem@davemloft.net>
Right now the mvneta driver doesn't handle Tx IRQ, and relies on two
mechanisms to flush Tx descriptors : a flush at the end of mvneta_tx()
and a timer. If a burst of packets is emitted faster than the device
can send them, then the queue is stopped until next wake-up of the
timer 10ms later. This causes jerky output traffic with bursts and
pauses, making it difficult to reach line rate with very few streams.
A test on UDP traffic shows that it's not possible to go beyond 134
Mbps / 12 kpps of outgoing traffic with 1500-bytes IP packets. Routed
traffic tends to observe pauses as well if the traffic is bursty,
making it even burstier after the wake-up.
It seems that this feature was inherited from the original driver but
nothing there mentions any reason for not using the interrupt instead,
which the chip supports.
Thus, this patch enables Tx interrupts and removes the timer. It does
the two at once because it's not really possible to make the two
mechanisms coexist, so a split patch doesn't make sense.
First tests performed on a Mirabox (Armada 370) show that less CPU
seems to be used when sending traffic. One reason might be that we now
call the mvneta_tx_done_gbe() with a mask indicating which queues have
been done instead of looping over all of them.
The same UDP test above now happily reaches 987 Mbps / 87.7 kpps.
Single-stream TCP traffic can now more easily reach line rate. HTTP
transfers of 1 MB objects over a single connection went from 730 to
840 Mbps. It is even possible to go significantly higher (>900 Mbps)
by tweaking tcp_tso_win_divisor.
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Gregory CLEMENT <gregory.clement@free-electrons.com>
Cc: Arnaud Ebalard <arno@natisbad.org>
Cc: Eric Dumazet <eric.dumazet@gmail.com>
Tested-by: Arnaud Ebalard <arno@natisbad.org>
Signed-off-by: Willy Tarreau <w@1wt.eu>
Signed-off-by: David S. Miller <davem@davemloft.net>
Marvell has not published the chip's datasheet yet, so it's very hard
to find the relevant bits to manipulate to change the IRQ behaviour.
Fortunately, these bits are described in the proprietary LSP patch set
which is publicly available here :
http://www.plugcomputer.org/downloads/mirabox/
So let's put them back in the driver in order to reduce the burden of
current and future maintenance.
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Gregory CLEMENT <gregory.clement@free-electrons.com>
Tested-by: Arnaud Ebalard <arno@natisbad.org>
Signed-off-by: Willy Tarreau <w@1wt.eu>
Signed-off-by: David S. Miller <davem@davemloft.net>
If a queue timeout is reported, we can oops because of some
schedules while the caller is atomic, as shown below :
mvneta d0070000.ethernet eth0: tx timeout
BUG: scheduling while atomic: bash/1528/0x00000100
Modules linked in: slhttp_ethdiv(C) [last unloaded: slhttp_ethdiv]
CPU: 2 PID: 1528 Comm: bash Tainted: G WC 3.13.0-rc4-mvebu-nf #180
[<c0011bd9>] (unwind_backtrace+0x1/0x98) from [<c000f1ab>] (show_stack+0xb/0xc)
[<c000f1ab>] (show_stack+0xb/0xc) from [<c02ad323>] (dump_stack+0x4f/0x64)
[<c02ad323>] (dump_stack+0x4f/0x64) from [<c02abe67>] (__schedule_bug+0x37/0x4c)
[<c02abe67>] (__schedule_bug+0x37/0x4c) from [<c02ae261>] (__schedule+0x325/0x3ec)
[<c02ae261>] (__schedule+0x325/0x3ec) from [<c02adb97>] (schedule_timeout+0xb7/0x118)
[<c02adb97>] (schedule_timeout+0xb7/0x118) from [<c0020a67>] (msleep+0xf/0x14)
[<c0020a67>] (msleep+0xf/0x14) from [<c01dcbe5>] (mvneta_stop_dev+0x21/0x194)
[<c01dcbe5>] (mvneta_stop_dev+0x21/0x194) from [<c01dcfe9>] (mvneta_tx_timeout+0x19/0x24)
[<c01dcfe9>] (mvneta_tx_timeout+0x19/0x24) from [<c024afc7>] (dev_watchdog+0x18b/0x1c4)
[<c024afc7>] (dev_watchdog+0x18b/0x1c4) from [<c0020b53>] (call_timer_fn.isra.27+0x17/0x5c)
[<c0020b53>] (call_timer_fn.isra.27+0x17/0x5c) from [<c0020cad>] (run_timer_softirq+0x115/0x170)
[<c0020cad>] (run_timer_softirq+0x115/0x170) from [<c001ccb9>] (__do_softirq+0xbd/0x1a8)
[<c001ccb9>] (__do_softirq+0xbd/0x1a8) from [<c001cfad>] (irq_exit+0x61/0x98)
[<c001cfad>] (irq_exit+0x61/0x98) from [<c000d4bf>] (handle_IRQ+0x27/0x60)
[<c000d4bf>] (handle_IRQ+0x27/0x60) from [<c000843b>] (armada_370_xp_handle_irq+0x33/0xc8)
[<c000843b>] (armada_370_xp_handle_irq+0x33/0xc8) from [<c000fba9>] (__irq_usr+0x49/0x60)
Ben Hutchings attempted to propose a better fix consisting in using a
scheduled work for this, but while it fixed this panic, it caused other
random freezes and panics proving that the reset sequence in the driver
is unreliable and that additional fixes should be investigated.
When sending multiple streams over a link limited to 100 Mbps, Tx timeouts
happen from time to time, and the driver correctly recovers only when the
function is disabled.
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Gregory CLEMENT <gregory.clement@free-electrons.com>
Cc: Ben Hutchings <ben@decadent.org.uk>
Tested-by: Arnaud Ebalard <arno@natisbad.org>
Signed-off-by: Willy Tarreau <w@1wt.eu>
Signed-off-by: David S. Miller <davem@davemloft.net>
Stats writers are mvneta_rx() and mvneta_tx(). They don't lock anything
when they update the stats, and as a result, it randomly happens that
the stats freeze on SMP if two updates happen during stats retrieval.
This is very easily reproducible by starting two HTTP servers and binding
each of them to a different CPU, then consulting /proc/net/dev in loops
during transfers, the interface should immediately lock up. This issue
also randomly happens upon link state changes during transfers, because
the stats are collected in this situation, but it takes more attempts to
reproduce it.
The comments in netdevice.h suggest using per_cpu stats instead to get
rid of this issue.
This patch implements this. It merges both rx_stats and tx_stats into
a single "stats" member with a single syncp. Both mvneta_rx() and
mvneta_rx() now only update the a single CPU's counters.
In turn, mvneta_get_stats64() does the summing by iterating over all CPUs
to get their respective stats.
With this change, stats are still correct and no more lockup is encountered.
Note that this bug was present since the first import of the mvneta
driver. It might make sense to backport it to some stable trees. If
so, it depends on "d33dc73 net: mvneta: increase the 64-bit rx/tx stats
out of the hot path".
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Gregory CLEMENT <gregory.clement@free-electrons.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Tested-by: Arnaud Ebalard <arno@natisbad.org>
Signed-off-by: Willy Tarreau <w@1wt.eu>
Signed-off-by: David S. Miller <davem@davemloft.net>
Better count packets and bytes in the stack and on 32 bit then
accumulate them at the end for once. This saves two memory writes
and two memory barriers per packet. The incoming packet rate was
increased by 4.7% on the Openblocks AX3 thanks to this.
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Gregory CLEMENT <gregory.clement@free-electrons.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Tested-by: Arnaud Ebalard <arno@natisbad.org>
Signed-off-by: Willy Tarreau <w@1wt.eu>
Signed-off-by: David S. Miller <davem@davemloft.net>
The current code unmaps the DMA mapping created for rx skb_buff's by
using the data_size as the the mapping size. This is wrong since the
correct size to specify should match the size used to create the mapping.
This commit removes the following DMA_API_DEBUG warning:
------------[ cut here ]------------
WARNING: at lib/dma-debug.c:887 check_unmap+0x3a8/0x860()
mvneta d0070000.ethernet: DMA-API: device driver frees DMA memory with different size [device address=0x000000002eb80000] [map size=1600 bytes] [unmap size=66 bytes]
CPU: 0 PID: 0 Comm: swapper/0 Not tainted 3.10.21-01444-ga88ae13-dirty #92
[<c0013600>] (unwind_backtrace+0x0/0xf8) from [<c0010fb8>] (show_stack+0x10/0x14)
[<c0010fb8>] (show_stack+0x10/0x14) from [<c001afa0>] (warn_slowpath_common+0x48/0x68)
[<c001afa0>] (warn_slowpath_common+0x48/0x68) from [<c001b01c>] (warn_slowpath_fmt+0x30/0x40)
[<c001b01c>] (warn_slowpath_fmt+0x30/0x40) from [<c018d0fc>] (check_unmap+0x3a8/0x860)
[<c018d0fc>] (check_unmap+0x3a8/0x860) from [<c018d734>] (debug_dma_unmap_page+0x64/0x70)
[<c018d734>] (debug_dma_unmap_page+0x64/0x70) from [<c0233f78>] (mvneta_rx+0xec/0x468)
[<c0233f78>] (mvneta_rx+0xec/0x468) from [<c023436c>] (mvneta_poll+0x78/0x16c)
[<c023436c>] (mvneta_poll+0x78/0x16c) from [<c02db468>] (net_rx_action+0x94/0x160)
[<c02db468>] (net_rx_action+0x94/0x160) from [<c0021e68>] (__do_softirq+0xe8/0x1d0)
[<c0021e68>] (__do_softirq+0xe8/0x1d0) from [<c0021ff8>] (do_softirq+0x4c/0x58)
[<c0021ff8>] (do_softirq+0x4c/0x58) from [<c0022228>] (irq_exit+0x58/0x90)
[<c0022228>] (irq_exit+0x58/0x90) from [<c000e7c8>] (handle_IRQ+0x3c/0x94)
[<c000e7c8>] (handle_IRQ+0x3c/0x94) from [<c0008548>] (armada_370_xp_handle_irq+0x4c/0xb4)
[<c0008548>] (armada_370_xp_handle_irq+0x4c/0xb4) from [<c000dc20>] (__irq_svc+0x40/0x50)
Exception stack(0xc04f1f70 to 0xc04f1fb8)
1f60: c1fe46f8 00000000 00001d92 00001d92
1f80: c04f0000 c04f0000 c04f84a4 c03e081c c05220e7 00000001 c05220e7 c04f0000
1fa0: 00000000 c04f1fb8 c000eaf8 c004c048 60000113 ffffffff
[<c000dc20>] (__irq_svc+0x40/0x50) from [<c004c048>] (cpu_startup_entry+0x54/0x128)
[<c004c048>] (cpu_startup_entry+0x54/0x128) from [<c04c1a14>] (start_kernel+0x29c/0x2f0)
[<c04c1a14>] (start_kernel+0x29c/0x2f0) from [<00008074>] (0x8074)
---[ end trace d4955f6acd178110 ]---
Mapped at:
[<c018d600>] debug_dma_map_page+0x4c/0x11c
[<c0235d6c>] mvneta_setup_rxqs+0x398/0x598
[<c0236084>] mvneta_open+0x40/0x17c
[<c02dbbd4>] __dev_open+0x9c/0x100
[<c02dbe58>] __dev_change_flags+0x7c/0x134
Signed-off-by: Ezequiel Garcia <ezequiel.garcia@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Pull core locking changes from Ingo Molnar:
"The biggest changes:
- add lockdep support for seqcount/seqlocks structures, this
unearthed both bugs and required extra annotation.
- move the various kernel locking primitives to the new
kernel/locking/ directory"
* 'core-locking-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (21 commits)
block: Use u64_stats_init() to initialize seqcounts
locking/lockdep: Mark __lockdep_count_forward_deps() as static
lockdep/proc: Fix lock-time avg computation
locking/doc: Update references to kernel/mutex.c
ipv6: Fix possible ipv6 seqlock deadlock
cpuset: Fix potential deadlock w/ set_mems_allowed
seqcount: Add lockdep functionality to seqcount/seqlock structures
net: Explicitly initialize u64_stats_sync structures for lockdep
locking: Move the percpu-rwsem code to kernel/locking/
locking: Move the lglocks code to kernel/locking/
locking: Move the rwsem code to kernel/locking/
locking: Move the rtmutex code to kernel/locking/
locking: Move the semaphore core to kernel/locking/
locking: Move the spinlock code to kernel/locking/
locking: Move the lockdep code to kernel/locking/
locking: Move the mutex code to kernel/locking/
hung_task debugging: Add tracepoint to report the hang
x86/locking/kconfig: Update paravirt spinlock Kconfig description
lockstat: Report avg wait and hold times
lockdep, x86/alternatives: Drop ancient lockdep fixup message
...
In order to enable lockdep on seqcount/seqlock structures, we
must explicitly initialize any locks.
The u64_stats_sync structure, uses a seqcount, and thus we need
to introduce a u64_stats_init() function and use it to initialize
the structure.
This unfortunately adds a lot of fairly trivial initialization code
to a number of drivers. But the benefit of ensuring correctness makes
this worth while.
Because these changes are required for lockdep to be enabled, and the
changes are quite trivial, I've not yet split this patch out into 30-some
separate patches, as I figured it would be better to get the various
maintainers thoughts on how to best merge this change along with
the seqcount lockdep enablement.
Feedback would be appreciated!
Signed-off-by: John Stultz <john.stultz@linaro.org>
Acked-by: Julian Anastasov <ja@ssi.bg>
Signed-off-by: Peter Zijlstra <peterz@infradead.org>
Cc: Alexey Kuznetsov <kuznet@ms2.inr.ac.ru>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Eric Dumazet <eric.dumazet@gmail.com>
Cc: Hideaki YOSHIFUJI <yoshfuji@linux-ipv6.org>
Cc: James Morris <jmorris@namei.org>
Cc: Jesse Gross <jesse@nicira.com>
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Mirko Lindner <mlindner@marvell.com>
Cc: Patrick McHardy <kaber@trash.net>
Cc: Roger Luethi <rl@hellgate.ch>
Cc: Rusty Russell <rusty@rustcorp.com.au>
Cc: Simon Horman <horms@verge.net.au>
Cc: Stephen Hemminger <stephen@networkplumber.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Wensong Zhang <wensong@linux-vs.org>
Cc: netdev@vger.kernel.org
Link: http://lkml.kernel.org/r/1381186321-4906-2-git-send-email-john.stultz@linaro.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Checking if MAC address is valid using is_valid_ether_addr() is already done in
of_get_mac_address().
Signed-off-by: Luka Perkov <luka@openwrt.org>
Acked-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
CC: David Miller <davem@davemloft.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
Conflicts:
drivers/net/ethernet/stmicro/stmmac/stmmac_platform.c
net/bridge/br_multicast.c
net/ipv6/sit.c
The conflicts were minor:
1) sit.c changes overlap with change to ip_tunnel_xmit() signature.
2) br_multicast.c had an overlap between computing max_delay using
msecs_to_jiffies and turning MLDV2_MRC() into an inline function
with a name using lowercase instead of uppercase letters.
3) stmmac had two overlapping changes, one which conditionally allocated
and hooked up a dma_cfg based upon the presence of the pbl OF property,
and another one handling store-and-forward DMA made. The latter of
which should not go into the new of_find_property() basic block.
Signed-off-by: David S. Miller <davem@davemloft.net>
This commit implements the ->ndo_do_ioctl() operation so that the
PHY-related ioctl() calls can work from userspace, which allows
applications like mii-tool or mii-diag to do their job.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Tested-by: Gregory CLEMENT <gregory.clement@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This commit fixes a long-standing bug that has been reported by many
users: on some Armada 370 platforms, only the network interface that
has been used in U-Boot to tftp the kernel works properly in
Linux. The other network interfaces can see a 'link up', but are
unable to transmit data. The reports were generally made on the Armada
370-based Mirabox, but have also been given on the Armada 370-RD
board.
The network MAC in the Armada 370/XP (supported by the mvneta driver
in Linux) has a functionality that allows it to continuously poll the
PHY and directly update the MAC configuration accordingly (speed,
duplex, etc.). The very first versions of the driver submitted for
review were using this hardware mechanism, but due to this, the driver
was not integrated with the kernel phylib. Following reviews, the
driver was changed to use the phylib, and therefore a software based
polling. In software based polling, Linux regularly talks to the PHY
over the MDIO bus, and sees if the link status has changed. If it's
the case then the adjust_link() callback of the driver is called to
update the MAC configuration accordingly.
However, it turns out that the adjust_link() callback was not
configuring the hardware in a completely correct way: while it was
setting the speed and duplex bits correctly, it wasn't telling the
hardware to actually take into account those bits rather than what the
hardware-based PHY polling mechanism has concluded. So, in fact the
adjust_link() callback was basically a no-op.
However, the network happened to be working because on the network
interfaces used by U-Boot for tftp on Armada 370 platforms because the
hardware PHY polling was enabled by the bootloader, and left enabled
by Linux. However, the second network interface not used for tftp (or
both network interfaces if the kernel is loaded from USB, NAND or SD
card) didn't had the hardware PHY polling enabled.
This patch fixes this situation by:
(1) Making sure that the hardware PHY polling is disabled by clearing
the MVNETA_PHY_POLLING_ENABLE bit in the MVNETA_UNIT_CONTROL
register in the driver ->probe() function.
(2) Making sure that the duplex and speed selections made by the
adjust_link() callback are taken into account by clearing the
MVNETA_GMAC_AN_SPEED_EN and MVNETA_GMAC_AN_DUPLEX_EN bits in the
MVNETA_GMAC_AUTONEG_CONFIG register.
This patch has been tested on Armada 370 Mirabox, and now both network
interfaces are usable after boot.
[ Problem introduced by commit c5aff18 ("net: mvneta: driver for
Marvell Armada 370/XP network unit") ]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Willy Tarreau <w@1wt.eu>
Cc: Jochen De Smet <jochen.armkernel@leahnim.org>
Cc: Peter Sanford <psanford@nearbuy.io>
Cc: Ethan Tuttle <ethan@ethantuttle.com>
Cc: Chény Yves-Gael <yves@cheny.fr>
Cc: Ryan Press <ryan@presslab.us>
Cc: Simon Guinot <simon.guinot@sequanux.org>
Cc: vdonnefort@lacie.com
Cc: stable@vger.kernel.org
Acked-by: Jason Cooper <jason@lakedaemon.net>
Tested-by: Vincent Donnefort <vdonnefort@gmail.com>
Tested-by: Yves-Gael Cheny <yves@cheny.fr>
Tested-by: Gregory CLEMENT <gregory.clement@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
When the mvneta driver is compiled as module, the clock is disabled before
it's loading. This will reset the registers values and all configuration
made by the bootloader.
This patch sets the "sgmii serdes configuration" register to a magical value
found in:
https://github.com/yellowback/ubuntu-precise-armadaxp/blob/master/arch/arm/mach-armadaxp/armada_xp_family/ctrlEnv/mvCtrlEnvLib.c
With this change, the interrupts are working/generated and ethernet is
working.
Signed-off-by: Arnaud Patard <arnaud.patard@rtp-net.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
When the mvneta driver is compiled, it'll be loaded with clocks disabled.
This implies that the clocks should be enabled again before any register
access or it'll hang.
To fix it:
- enable clock earlier
- move timer callback after setting timer.data
Signed-off-by: Arnaud Patard <arnaud.patard@rtp-net.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
Use the "swap descriptor" feature of the hardware to properly swap the
descriptors when running in big endian mode. Since the swapping occurs
on 64 bits words, we also need to provide a separate structure layout
for the DMA descriptors between little endian and big endian mode,
like is done in the mv643xx_eth driver.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The macros used for the various fields of the RX and TX descriptions
are currently declared next to those fields within the structure
definitions of the RX and TX descriptors.
However, in order to support big endian, we'll have to use the "swap
descriptors" features of the hardware, which swaps every byte within
each 64 bits word of the descriptors. This requires a separate
definition of the RX and TX descriptor structures for little and big
endian, as is done in the mv643xx_eth. Those macros can therefore no
longer be defined inside those structures.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This patch improves the logic used by the mvneta driver to find a MAC
address for a particular interface. Until now, it was only looking at
the Device Tree, and if no address was found, was falling back to
generating a random MAC address.
This patch adds the intermediate solution of reading the MAC address
from the hardware registers, in case it has been set by the
bootloader. So the order is now:
1) MAC address from the Device Tree
2) MAC address from the hardware registers
3) Random MAC address
This requires moving the MAC address initialization a little bit later
in the ->probe() code, because it now requires the hardware registers
to be remapped.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Sergei Shtylyov <sergei.shtylyov@cogentembedded.com>
Cc: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The driver core clears the driver data to NULL after device_release
or on probe failure, since commit 0998d06310
(device-core: Ensure drvdata = NULL when no driver is bound).
Thus, it is not needed to manually clear the device driver data to NULL.
Signed-off-by: Jingoo Han <jg1.han@samsung.com>
Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com>
Acked-by: Rob Herring <rob.herring@calxeda.com>
Acked-by: Roland Stigge <stigge@antcom.de>
Acked-by: Mugunthan V N <mugunthanvnm@ti.com>
Reviewed-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Tested-by: Roland Stigge <stigge@antcom.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
Conflicts:
drivers/net/ethernet/emulex/benet/be_main.c
drivers/net/ethernet/intel/igb/igb_main.c
drivers/net/wireless/brcm80211/brcmsmac/mac80211_if.c
include/net/scm.h
net/batman-adv/routing.c
net/ipv4/tcp_input.c
The e{uid,gid} --> {uid,gid} credentials fix conflicted with the
cleanup in net-next to now pass cred structs around.
The be2net driver had a bug fix in 'net' that overlapped with the VLAN
interface changes by Patrick McHardy in net-next.
An IGB conflict existed because in 'net' the build_skb() support was
reverted, and in 'net-next' there was a comment style fix within that
code.
Several batman-adv conflicts were resolved by making sure that all
calls to batadv_is_my_mac() are changed to have a new bat_priv first
argument.
Eric Dumazet's TS ECR fix in TCP in 'net' conflicted with the F-RTO
rewrite in 'net-next', mostly overlapping changes.
Thanks to Stephen Rothwell and Antonio Quartulli for help with several
of these merge resolutions.
Signed-off-by: David S. Miller <davem@davemloft.net>
mvneta_tx() was using a static tx queue number causing crashes as
soon as a little bit of traffic was sent via the interface, because
it is normally expected that the same queue should be used as in
dev_queue_xmit().
As suggested by Ben Hutchings, let's use skb_get_queue_mapping() to
get the proper Tx queue number, and use alloc_etherdev_mqs() instead
of alloc_etherdev_mq() to create the queues.
Both my Mirabox and my OpenBlocks AX3 used to crash without this patch
and don't anymore with it. The issue appeared in 3.8 but became more
visible after the fix allowing GSO to be enabled.
Original work was done by Dmitri Epshtein and Thomas Petazzoni. I
just adapted it to take care of Ben's comments.
Signed-off-by: Willy Tarreau <w@1wt.eu>
Cc: Dmitri Epshtein <dima@marvell.com>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Gregory CLEMENT <gregory.clement@free-electrons.com>
Cc: Ben Hutchings <bhutchings@solarflare.com>
Tested-by: Gregory CLEMENT <gregory.clement@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
It seems that the reason why the dev features were ignored was because
they were enabled after registeration.
Signed-off-by: Willy Tarreau <w@1wt.eu>
Acked-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
I believe these error messages are already logged
on allocation failure by warn_alloc_failed and so
get a dump_stack on OOM.
Remove the unnecessary additional error logging.
Around these deletions:
o Alignment neatening.
o Remove unnecessary casts of dma_alloc_coherent.
o Hoist assigns from ifs.
Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
The file uses nothing from the version.h header, so there is no reason
to include it.
Signed-off-by: Jesper Juhl <jj@chaosbits.net>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
Remove some __dev* markings that snuck in the 3.8-rc1 merge window in
the drivers/net/* directory.
Cc: Bill Pemberton <wfp5p@virginia.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
mvneta_deinit() can be called from the ->probe() hook in the error
path, so it shouldn't be marked as __devexit. It fixes the following
section mismatch warning:
WARNING: vmlinux.o(.devinit.text+0x239c): Section mismatch in reference
from the function mvneta_probe() to the function .devexit.text:mvneta_deinit()
The function __devinit mvneta_probe() references
a function __devexit mvneta_deinit().
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Now that the Armada 370/XP platform has gained proper integration with
the clock framework, we add clk support in the Marvell Armada 370/XP
Ethernet driver.
Since the existing Device Tree binding that exposes a
'clock-frequency' property has never been exposed in any stable kernel
release, we take the freedom of removing this property to replace it
with the standard 'clocks' clock pointer property.
The Device Tree binding documentation is updated accordingly.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
As reported by checkpatch, the multiline comments for net/ and
drivers/net/ have a slightly different format than the one used in the
rest of the kernel, so we adjust our multiline comments accordingly.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
This patch contains a new network driver for the network unit of the
ARM Marvell Armada 370 and the Armada XP. Both SoCs use the PJ4B
processor, a Marvell-developed ARM core that implements the ARMv7
instruction set.
Compared to previous ARM Marvell SoCs (Kirkwood, Orion, Discovery),
the network unit in Armada 370 and Armada XP is highly different. This
is the reason why this new 'mvneta' driver is needed, while the older
ARM Marvell SoCs use the 'mv643xx_eth' driver.
Here is an overview of the most important hardware changes that
require a new, specific, driver for the network unit of Armada 370/XP:
- The new network unit has a completely different design and layout
for the RX and TX descriptors. They are now organized as a simple
array (each RX and TX queue has base address and size of this
array) rather than a linked list as in the old SoCs.
- The new network unit has a different RXQ and TXQ management: this
management is done using special read/write counter registers,
while in the Old SocS, it was done using the Ownership bit in RX
and TX descriptors.
- The new network unit has different interrupt registers
- The new network unit way of cleaning of interrupts is not done by
writing to the cause register, but by updating per-queue counters
- The new network unit has different GMAC registers (link, speed,
duplex configuration) and different WRR registers.
- The new network unit has lots of new units like PnC (Parser and
Classifier), PMT, BM (Memory Buffer Management), xPON, and more.
The driver proposed in the current patch only handles the basic
features. Additional hardware features will progressively be supported
as needed.
This code has originally been written by Rami Rosen
<rosenr@marvell.com>, and then reviewed and cleaned up by Thomas
Petazzoni <thomas.petazzoni@free-electrons.com>.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Acked-by: David S. Miller <davem@davemloft.net>
Gregory CLEMENT