The Atmel DFP headers define two "component typedef styles": RFO and
NTO; where the latter makes use of bit field structs to access hardware
registers.
The default component typedef style assumed by the DFP headers (i.e.
when `COMPONENT_TYPEDEF_STYLE` is not explicitly defined) is "RFO" and
this is indeed the component typedef style used throughout the Zephyr
Atmel SAM drivers, except in the particular instance which this commit
addresses.
The use of `GMAC_TA_Type` bit field struct, which is an "NTO" style
construct, is no longer possible with the latest DFPs because
conditional compilation checks for the bit field struct definitions
were added to restrict the use of such constructs to only when the
global component typedef style is set to "NTO".
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
Convert older DT_INST_ macro use in silab drivers to the new
include/devicetree.h DT_INST macro APIs.
Signed-off-by: Kumar Gala <kumar.gala@linaro.org>
Convert older DT_INST_ macro use in stellaris drivers to the new
include/devicetree.h DT_INST macro APIs.
Signed-off-by: Kumar Gala <kumar.gala@linaro.org>
Starting a SMI write operation without waiting for completion of the
preceding SMI read operation cause the write operation to fail if
the time between the 2 operations it too short. This leads to the
PHY being in an unusable state on the MIMRT1060-EVK eval board.
In addition the value of the register was not used, as consequence
some bits were not preserved.
The solution is to do a read/modify/write to set only the
ref clock bit, which sets the PHY into 50MHz clock mode,
and keep the value of the other bits.
Signed-off-by: Armand Ciejak <armand@riedonetworks.com>
Force PHY normal operation in eth_mcux_phy_setup in case strap-in
pins configure the PHY in factory test mode.
Signed-off-by: Armand Ciejak <armand@riedonetworks.com>
This commit adds the "monitor task" to detect and report any changes
in the PHY link status to the operating system.
The monitor task is perodically executed to poll the link status from
the PHY and call `net_eth_carrier_{off,on}` based on the detected
link status change.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
The `link_configure` function currently configures the `GMAC_UR`
register and forces the RMII interface.
This is not necessary because the `GMAC_UR` register is already
configured with an appropriate value based on `CONFIG_ETH_SAM_GMAC_MII`
in `gmac_init`.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
This commit adds the PHY link status query function that can be used
for detecting the Ethernet connection.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
Add interface name to link up/down information message
to help tracking which port is changing status.
Signed-off-by: Armand Ciejak <armandciejak@users.noreply.github.com>
Copy paste error, base address of second controller was set to the
same as the first one, corrected to ENET2.
Signed-off-by: Armand Ciejak <armandciejak@users.noreply.github.com>
With Ethernet controller base address is not straight forward
to figure out if it is ENET or ENET2l, using name is much better.
Signed-off-by: Armand Ciejak <armandciejak@users.noreply.github.com>
This will allow having a different PHY address for each controller.
It also avoids for the PHY address to be hard-coded in several
functions.
Signed-off-by: Armand Ciejak <armandciejak@users.noreply.github.com>
Add new config allowing to enable the second Ethernet port
on i.MX RT1060 SoC family.
Signed-off-by: Armand Ciejak <armandciejak@users.noreply.github.com>
This reverts commit 8739517107.
Pull Request #23437 was merged by mistake with an invalid manifest.
Signed-off-by: Carles Cufi <carles.cufi@nordicsemi.no>
Replace all occurences of BUILD_ASSERT_MSG() with BUILD_ASSERT()
as a result of merging BUILD_ASSERT() and BUILD_ASSERT_MSG().
Signed-off-by: Oleg Zhurakivskyy <oleg.zhurakivskyy@intel.com>
Add Virtual LAN support to stm32 Ethernet driver. Refactor the
eth_iface_init() and move device configuration settings to
eth_initialize() as the eth_iface_init() is called multiple
times (once / configured VLAN).
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
The RX fragment debug print does not work (compile error)
if memory allocation debugging is enabled, so disable it
for time being.
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
The current design of the network-specific stack dumping APIs
is fundamentally unsafe. You cannot properly dump stack data
without information which is only available in the thread object.
In addition, this infrastructure is unnecessary. There is already
a core shell command which dumps stack information for all
active threads.
Signed-off-by: Andrew Boie <andrew.p.boie@intel.com>
Instead of hardcoding the "zeth" network interface name, use the
name defined in Kconfig so that user can change it if needed.
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
Move auto-negotiate sequence from driver initialization to link up event
Previously when booting without ethernet cable connected the
initialization would fail and never recover.
Now we can connect the ethernet cable any time and multiple times.
This also drastically reduces boot time to main.
Logging Link up and Link down events.
Logging speed and duplex from eth_gecko logger instead of eth_gecko_phy.
Signed-off-by: Luuk Bosma <l.bosma@interay.com>
Do not use the content of the status register to write into the
override register because it may have unpredictable effect,
instead to a read/modify/write on the override register.
Signed-off-by: Armand Ciejak <armandciejak@users.noreply.github.com>
It is necessary to poll the ENET_EIR_MII bit before reading
the data register as explained in the i.MX RT1060 reference
manual in chapter 41.7.17.4.
Use PHY_* functions from NXP HAL to correctly access the PHY
registers.
Signed-off-by: Armand Ciejak <armandciejak@users.noreply.github.com>
Improve priority queue conditional build. Now priority queue code is
enabled only if device have support to it. This enables GMAC driver
for devices with only one queue for RX/TX.
Signed-off-by: Gerson Fernando Budke <nandojve@gmail.com>
The Atmel SAM SoC with ethernet port uses same GMAC driver. However,
there are differences between SoC GMAC implementation. Some SoCs have
priority queue and system can configure 0 up to 5, depending of SoC
version. This update current GMAC driver adding missing definitions.
Co-authored-by: Stephanos Ioannidis <root@stephanos.io>
Signed-off-by: Gerson Fernando Budke <nandojve@gmail.com>
Add missing queue entries for sam gmac. This update the queue selection
to proper handle all supported SAM SoC that uses GMAC driver.
Signed-off-by: Gerson Fernando Budke <nandojve@gmail.com>
The first revision of the SAM E70 soc had three queues. The current
revision B has six queues. If we don't initialize all queues, the DMA
engine gets stuck when trying to read a descriptor from NULL. To enable
the initialization of the additional queues, the correct soc has to be
selected in the config options, f.ex. CONFIG_SOC_PART_NUMBER_SAME70Q21B
instead of CONFIG_SOC_PART_NUMBER_SAME70Q21.
Also rename GMAC_QUEUE_NO to GMAC_QUEUE_NUM as requested during review.
Signed-off-by: Daniel Glöckner <dg@emlix.com>
SMI initialization is required to enable PHY communication.
PHY setups needs to run after SMI initialization.
Signed-off-by: Andrei Gansari <andrei.gansari@nxp.com>
Same deal as in commit eddd98f811 ("kconfig: Replace some single-symbol
'if's with 'depends on'"), for the remaining cases outside defconfig
files. See that commit for an explanation.
Will do the defconfigs separately in case there are any complaints
there.
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
If the Ethernet driver has VLAN enabled (only native_posix, mcux
or gmac has VLAN supported), then the iface pointer in ethernet
context should contain the main network interface. This is needed
so that the interface will get link address set to it properly.
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
The build infrastructure should not be adding the drivers subdirectory
to the include path. Fix the legacy uses that depended on that
addition.
Signed-off-by: Peter Bigot <peter.bigot@nordicsemi.no>
Ethernet MAC present in Silicon Labs EFM32GG11B4xx and
EFM32GG11B8xx SoCs.
DMA based driver with support for link up/down detection.
Signed-off-by: Oane Kingma <o.kingma@interay.com>
TCP segment with several EOL TCP options causes echo server to block and
use 100% of CPU. This patch fixes that issue by using k_yield to let
also other threads to run.
Fixes#21949
Signed-off-by: Ruslan Mstoi <ruslan.mstoi@intel.com>
echo_server app doesn't compile (it uses minimal libc which lacks
unistd.h), let's switch to a more fine-grained include here.
Signed-off-by: Oleg Zhurakivskyy <oleg.zhurakivskyy@intel.com>
Do not enable hardware accelerated checksum calculation by
default. It does not work for frdm-k64f and is causing
confusion among users.
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
Don't sleep 50ms after each received packet, sleep only when there
wasn't anything to receive. Otherwise data could get stuck for a long
time if there was more than 20 packets coming in per second. The
read() call on a TUN/TAP device returns only a single packet per call.
Also remove the call to eth_stats_update_errors_rx() because this else
clause isn't actually a receive error, we're just waiting for more
packets.
Signed-off-by: Jonas Norling <jonas.norling@greeneggs.se>
How prompts work is better documented nowadays, and these comments might
not be that helpful if you don't know.
There are lots promptless symbols that don't have a comment.
Also fix up some comments in arch/Kconfig that seem misplaced/redundant,
and clean up some whitespace (no blank line after a comment makes it
look like it only applies to the symbol directly after it to me).
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
Enable ETHERNET_HW_TX_CHKSUM_OFFLOAD, ETHERNET_HW_RX_CHKSUM_OFFLOAD,
ETHERNET_AUTO_NEGOTIATION_SET and the equivalent driver configuration
in eth_mcux driver.
Autonegoitiation was done at driver initialization.
Signed-off-by: Andrei Gansari <andrei.gansari@nxp.com>
It is possible that the network interface is not yet initialized
when status of the PHY changes. In this case we must not call
net_eth_carrier_on() as that will cause a crash.
This was noticed with mimxrt1050_evk board.
Fixes: #21257
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
Bool symbols implicitly default to 'n'.
A 'default n' can make sense e.g. in a Kconfig.defconfig file, if you
want to override a 'default y' on the base definition of the symbol. It
isn't used like that on any of these symbols though.
Also replace some
config
prompt "foo"
bool/int
with the more common shorthand
config
bool/int "foo"
See the 'Style recommendations and shorthands' section in
https://docs.zephyrproject.org/latest/guides/kconfig/index.html.
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
Set the network interface up / down according to link status.
This means that we call Ethernet carrier on/off function in
proper places.
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
This commit refactors kernel and arch headers to establish a boundary
between private and public interface headers.
The refactoring strategy used in this commit is detailed in the issue
This commit introduces the following major changes:
1. Establish a clear boundary between private and public headers by
removing "kernel/include" and "arch/*/include" from the global
include paths. Ideally, only kernel/ and arch/*/ source files should
reference the headers in these directories. If these headers must be
used by a component, these include paths shall be manually added to
the CMakeLists.txt file of the component. This is intended to
discourage applications from including private kernel and arch
headers either knowingly and unknowingly.
- kernel/include/ (PRIVATE)
This directory contains the private headers that provide private
kernel definitions which should not be visible outside the kernel
and arch source code. All public kernel definitions must be added
to an appropriate header located under include/.
- arch/*/include/ (PRIVATE)
This directory contains the private headers that provide private
architecture-specific definitions which should not be visible
outside the arch and kernel source code. All public architecture-
specific definitions must be added to an appropriate header located
under include/arch/*/.
- include/ AND include/sys/ (PUBLIC)
This directory contains the public headers that provide public
kernel definitions which can be referenced by both kernel and
application code.
- include/arch/*/ (PUBLIC)
This directory contains the public headers that provide public
architecture-specific definitions which can be referenced by both
kernel and application code.
2. Split arch_interface.h into "kernel-to-arch interface" and "public
arch interface" divisions.
- kernel/include/kernel_arch_interface.h
* provides private "kernel-to-arch interface" definition.
* includes arch/*/include/kernel_arch_func.h to ensure that the
interface function implementations are always available.
* includes sys/arch_interface.h so that public arch interface
definitions are automatically included when including this file.
- arch/*/include/kernel_arch_func.h
* provides architecture-specific "kernel-to-arch interface"
implementation.
* only the functions that will be used in kernel and arch source
files are defined here.
- include/sys/arch_interface.h
* provides "public arch interface" definition.
* includes include/arch/arch_inlines.h to ensure that the
architecture-specific public inline interface function
implementations are always available.
- include/arch/arch_inlines.h
* includes architecture-specific arch_inlines.h in
include/arch/*/arch_inline.h.
- include/arch/*/arch_inline.h
* provides architecture-specific "public arch interface" inline
function implementation.
* supersedes include/sys/arch_inline.h.
3. Refactor kernel and the existing architecture implementations.
- Remove circular dependency of kernel and arch headers. The
following general rules should be observed:
* Never include any private headers from public headers
* Never include kernel_internal.h in kernel_arch_data.h
* Always include kernel_arch_data.h from kernel_arch_func.h
* Never include kernel.h from kernel_struct.h either directly or
indirectly. Only add the kernel structures that must be referenced
from public arch headers in this file.
- Relocate syscall_handler.h to include/ so it can be used in the
public code. This is necessary because many user-mode public codes
reference the functions defined in this header.
- Relocate kernel_arch_thread.h to include/arch/*/thread.h. This is
necessary to provide architecture-specific thread definition for
'struct k_thread' in kernel.h.
- Remove any private header dependencies from public headers using
the following methods:
* If dependency is not required, simply omit
* If dependency is required,
- Relocate a portion of the required dependencies from the
private header to an appropriate public header OR
- Relocate the required private header to make it public.
This commit supersedes #20047, addresses #19666, and fixes#3056.
Signed-off-by: Stephanos Ioannidis <root@stephanos.io>
Use this short header style in all Kconfig files:
# <description>
# <copyright>
# <license>
...
Also change all <description>s from
# Kconfig[.extension] - Foo-related options
to just
# Foo-related options
It's clear enough that it's about Kconfig.
The <description> cleanup was done with this command, along with some
manual cleanup (big letter at the start, etc.)
git ls-files '*Kconfig*' | \
xargs sed -i -E '1 s/#\s*Kconfig[\w.-]*\s*-\s*/# /'
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
Clean up space errors and use a consistent style throughout the Kconfig
files. This makes reading the Kconfig files more distraction-free, helps
with grepping, and encourages the same style getting copied around
everywhere (meaning another pass hopefully won't be needed).
Go for the most common style:
- Indent properties with a single tab, including for choices.
Properties on choices work exactly the same syntactically as
properties on symbols, so not sure how the no-indentation thing
happened.
- Indent help texts with a tab followed by two spaces
- Put a space between 'config' and the symbol name, not a tab. This
also helps when grepping for definitions.
- Do '# A comment' instead of '#A comment'
I tweaked Kconfiglib a bit to find most of the stuff.
Some help texts were reflowed to 79 columns with 'gq' in Vim as well,
though not all, because I was afraid I'd accidentally mess up
formatting.
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
Re-run with updated script to convert integer literal delay arguments to
k_sleep to use the standard timeout macros.
Signed-off-by: Peter Bigot <peter.bigot@nordicsemi.no>
The zephyr/subsys/net/l2 include directory has been added through the
'zephyr_library_' API to modify the 'zephyr' library, when the
'zephyr_' API should have been used.
This patch fixes this problem. Using 'zephyr_library_' in this context
works by accident when 'zephyr' is the current library but has no
guarantees of working in the future.
Signed-off-by: Sebastian Bøe <sebastian.boe@nordicsemi.no>
This patch adds support for configuring the MAC address through the
Network Management API to the STM32 Ethernet driver.
Signed-off-by: Markus Fuchs <markus.fuchs@de.sauter-bc.com>
When randomly generating MAC addresses they will always be
locally administrated addresses, so the LAA bit should be set.
The LAA bit is the 2nd bit of the 1st byte of the MAC address
not the 2nd bit of the 4th byte.
Fixes: #16452
Signed-off-by: Erwin Rol <erwin@erwinrol.com>
This was only used on the Quark SoCs. It is no longer used, can no
longer be tested, and it's reliant upon the deprecated legacy PCI
subsystem. Remove it to prevent bitrot.
Signed-off-by: Charles E. Youse <charles.youse@intel.com>
This commit adds support for a 6LoCAN Ethernet border translator.
CAN frames with the translator CAN address are translated and forwarded
to Ethernet. Ethernet frames with the first 34 bits matching the MAC
address of the translator are translated and forwarded to 6LoCAN.
Signed-off-by: Alexander Wachter <alexander.wachter@student.tugraz.at>
ETH_LITEETH is defined in drivers/ethernet/Kconfig.liteeth, which is
source'd within a menu that has 'depends on ETH_LITEETH', in
drivers/ethernet/Kconfig.
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
Liteeth exposes two memory regions:
* set of rx/tx buffers (aka slots) to exchange packets,
* control and status registers.
Signed-off-by: Mateusz Holenko <mholenko@antmicro.com>
For STM32F7 MCU the actual implementation doesn't work when the
DMA buffers are placed in the SRAM.
This might be a problem with caches.
To overcome this problem, the buffer is moved to the DTCM.
Signed-off-by: Alexander Wachter <alexander.wachter@student.tugraz.at>
uint8_array values are now generated as structure initializers. Update
the code accordingly. The implementation assumes that existing
devicetree source does not provide the correct OUI so preserves the
in-driver override of the value provided by devicetree and its setting
for random/unique addresses.
Signed-off-by: Peter A. Bigot <pab@pabigot.com>
uint8_array values are now generated as structure initializers. Update
the code accordingly. The implementation assumes that existing
devicetree source does not provide the correct OUI so preserves the
in-driver override of the value provided by devicetree.
Signed-off-by: Peter A. Bigot <pab@pabigot.com>
move misc/util.h to sys/util.h and
create a shim for backward-compatibility.
No functional changes to the headers.
A warning in the shim can be controlled with CONFIG_COMPAT_INCLUDES.
Related to #16539
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
move misc/printk.h to sys/printk.h and
create a shim for backward-compatibility.
No functional changes to the headers.
A warning in the shim can be controlled with CONFIG_COMPAT_INCLUDES.
Related to #16539
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
move misc/__assert.h to sys/__assert.h and
create a shim for backward-compatibility.
No functional changes to the headers.
A warning in the shim can be controlled with CONFIG_COMPAT_INCLUDES.
Related to #16539
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
move spi.h to drivers/spi.h and
create a shim for backward-compatibility.
No functional changes to the headers.
A warning in the shim can be controlled with CONFIG_COMPAT_INCLUDES.
Related to #16539
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
move i2c.h to drivers/i2c.h and
create a shim for backward-compatibility.
No functional changes to the headers.
A warning in the shim can be controlled with CONFIG_COMPAT_INCLUDES.
Related to #16539
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
move gpio.h to drivers/gpio.h and
create a shim for backward-compatibility.
No functional changes to the headers.
A warning in the shim can be controlled with CONFIG_COMPAT_INCLUDES.
Related to #16539
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
move clock_control.h to drivers/clock_control.h and
create a shim for backward-compatibility.
No functional changes to the headers.
A warning in the shim can be controlled with CONFIG_COMPAT_INCLUDES.
Related to #16539
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
move sys_io.h to sys/sys_io.h and
create a shim for backward-compatibility.
No functional changes to the headers.
A warning in the shim can be controlled with CONFIG_COMPAT_INCLUDES.
Related to #16539
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
Convert DT_.*_GPIO_{CONTROLLER,PIN,FLAGS} ->
DT_.*_GPIOS_{CONTROLLER,PIN,FLAGS)
Used the following commands to make these conversions:
git grep -l DT_.*_GPIO_CONTROLLER | xargs sed -i 's/DT_\(.*\)_GPIO_CONTROLLER/DT_\1_GPIOS_CONTROLLER/g'
git grep -l DT_.*_GPIO_PIN | xargs sed -i 's/DT_\(.*\)_GPIO_PIN/DT_\1_GPIOS_PIN/g'
git grep -l DT_.*_GPIO_FLAGS | xargs sed -i 's/DT_\(.*\)_GPIO_FLAGS/DT_\1_GPIOS_FLAGS/g'
Signed-off-by: Kumar Gala <kumar.gala@linaro.org>
Found a few annoying typos and figured I better run script and
fix anything it can find, here are the results...
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
Change code from using now deprecated DT_<COMPAT>_<INSTANCE>_<PROP>
defines to using DT_INST_<INSTANCE>_<COMPAT>_<PROP>.
Signed-off-by: Kumar Gala <kumar.gala@linaro.org>
Change code from using now deprecated DT_<COMPAT>_<INSTANCE>_<PROP>
defines to using DT_INST_<INSTANCE>_<COMPAT>_<PROP>.
Signed-off-by: Kumar Gala <kumar.gala@linaro.org>
This is mainly testing purpose from native_posize ethernet
driver. Enable CONFIG_ETH_NATIVE_POSIX_VLAN_TAG_STRIP to have
VLAN tag strip feature on ethernet Rx frames.
Signed-off-by: Ravi kumar Veeramally <ravikumar.veeramally@linux.intel.com>
Migrate from "legacy" PCI support (drivers/pci) to new PCI(e) support.
The e1000 driver is merely for testing with QEMU and so should not be
a model for the use of PCI(e) functions. Consult instead "real-world"
PCI(e) drivers like the NS16550 UART (drivers/serial/uart_ns16550.c).
Signed-off-by: Charles Youse <charles.youse@intel.com>
As part of the ll_reserve refactoring effort, the packet length now
includes header size as well. Before the refactor, when the packet
length was written to the device, it did not include the header size,
which is the required value as per the LM3S6965 datasheet. After the
refactor the packet length includes the header size as well. The
header size has to subtracted from the packet length before writing to
the device. Fixes#13943.
Signed-off-by: Vijay Kumar B <vijaykumar@zilogic.com>
Update the files which contain no license information with the
'Apache-2.0' SPDX license identifier. Many source files in the tree are
missing licensing information, which makes it harder for compliance
tools to determine the correct license.
By default all files without license information are under the default
license of Zephyr, which is Apache version 2.
Signed-off-by: Anas Nashif <anas.nashif@intel.com>
Rename reserved function names in drivers/ subdirectory. Update
function macros concatenatenating function names with '##'. As
there is a conflict between the existing gpio_sch_manage_callback()
and _gpio_sch_manage_callback() names, leave the latter unmodified.
Signed-off-by: Patrik Flykt <patrik.flykt@intel.com>
When building with -fno-inline, the compiler complains about
undefined reference to this function. This happens when
building for code coverage. Since this function is only called
within the file, mark it static also.
Signed-off-by: Daniel Leung <daniel.leung@intel.com>
In the conversion of net_pkt_read_new to net_pkt_read, we missed
changing the function in the eth_smsc911x and eswifi_offload.
Signed-off-by: Kumar Gala <kumar.gala@linaro.org>
Some places were still using the old allocator. Using the new one does
not change any behavior. This will help to remove the useless data_len
attribute in net_pkt which legacy allocator was still setting.
Signed-off-by: Tomasz Bursztyka <tomasz.bursztyka@linux.intel.com>
Remove magic numbers from Ethernet drivers and tests by defining
NET_ETH_MAX_DATAGRAM_SIZE and NET_ETH_MAX_FRAME_SIZE.
Signed-off-by: Andrei Emeltchenko <andrei.emeltchenko@intel.com>
Removing the TX timeout handling in the GMAC driver (commit 18b07e09e0)
revealed some issues with the way hardware priority queues work.
For cases with both hardware priority queues enabled, with the default
recommended delta bandwidths (0% - 75%), the lower priority queue (0%)
is hardly able to send any packets. This became visible, because without
the timeout mechanism, we quickly ran out of available TX buffers if
there were multiple packets being queued to the queue.
Here is an excerpt from 802.1Q, chapter 34.3.1 which describes how Qav
bandwidth sharing SHOULD work:
The deltaBandwidth(N) for a given N, plus the deltaBandwidth(N) values
for any higher priority queues (larger values of N) defines the total
percentage of the Port’s bandwidth that can be reserved for that queue
and all higher priority queues. For the highest priority queue, this
means that the maximum value of operIdleSlope(N) is deltaBandwidth(N)%
of portTransmitRate. However, if operIdleSlope(N) is actually less
than this maximum value, any lower priority queue that supports the
credit-based shaper algorithm can make use of the reservable bandwidth
that is unused by the higher priority queue. So, for queue N-1, the
maximum value of (operIdleSlope(N) + operIdleSlope(N-1)) is
(deltaBandwidth(N) + deltaBandwidth(N1))% of portTransmitRate.
However in reality, the lower priority queues (N-1) on the SAM GMAC
hardware DO NOT use the bandwidth available from the higher priority
queues (N).
This commits fixes the issue by changing the defaults. These are still
set to the recommended 75% (total), but this percentage is split between
the priority queues manually.
Signed-off-by: Tomasz Gorochowik <tgorochowik@antmicro.com>
There is a log message printed for the non-priority queue (Queue 0). Add
the same message for the priority queues too when they are enabled, and
a corresponding message when the queue is not used (set to idle).
Signed-off-by: Tomasz Gorochowik <tgorochowik@antmicro.com>
Throughout the manual, the queues GMAC is equipped with are identified
by a number - Queue 0, Queue 1 and Queue 2.
However in the context of Qav, the queues are identified with a
character (note that there are only two queues as Qav is not used for
the non-priority queue) - Queue A and Queue B.
Queue B and Queue 2 are also called "the highest priority queues".
Based on that, the previous implementation was using the following
mapping:
Queue 1 - Queue A
Queue 2 - Queue B
However when running some specific tests, that is for example forcing
all the traffic to Queue 1, it showed that this queue is actually
affected by the Queue B registers. Similarly, Queue 2 seems to be
affected by the Queue A registers.
Based on that observation, this commit changes the registers used to
work with the following mapping:
Queue 1 - Queue B
Queue 2 - Queue A
Note that this is based solely on observations, there is nothing in the
datasheet that confirms this, and the "highest priority" label suggests
it is otherwise.
Signed-off-by: Tomasz Gorochowik <tgorochowik@antmicro.com>
This option is meant to be used only for debugging. Use it to force all
traffic to be routed through a specific hardware queue. With this option
enabled it is easier to verify whether the chosen hardware queue
actually works.
This works only if there are four or fewer RX traffic classes enabled as
the SAM GMAC hardware supports screening up to four traffic classes.
Signed-off-by: Tomasz Gorochowik <tgorochowik@antmicro.com>
The interface name is IFNAMSIZ long so we must not copy it
full which would overwrite the terminating null byte.
Coverity-CID: 195770
Fixes#14419
Signed-off-by: Jukka Rissanen <jukka.rissanen@linux.intel.com>
Some of these are from 'source'ing a file within a menu that has a
'depends on NET_L2_ETHERNET' (in drivers/ethernet/Kconfig) and then
adding another 'depends on NET_L2_ETHERNET' within it.
Similarly, subsys/net/l2/ethernet/Kconfig sources files within an
'if NET_L2_ETHERNET'.
'if FOO' is just shorthand for adding 'depends on FOO' to each item
within the 'if'. Dependencies on menus work similarly. There are no
"conditional includes" in Kconfig, so 'if FOO' has no special meaning
around a source. Conditional includes wouldn't be possible, because an
if condition could include (directly or indirectly) forward references
to symbols not defined yet.
Tip: When adding a symbol, check its dependencies in the menuconfig
('ninja menuconfig', then / to jump to the symbol). The menuconfig also
shows how the file with the symbol got included, so if you see
duplicated dependencies, it's easy to hunt down where they come from.
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
Fix calculation of frame length in eth_enc28j60_rx().
The calculation was incorrect because the CRC size was
subtracted only from the lower byte.
Signed-off-by: Johann Fischer <j.fischer@phytec.de>
The j iterator is used for priorities, and the i iterator is used for
choosing the actual screening register. Therefore the screening register
availability needs to be checked with i.
This commit also improves the readability by doing two things:
* by moving the screening register index check to a place where the
index is going to be actually used
* by using the ARRAY_SIZE macro instead of a hard-coded number as the
upper limit
Signed-off-by: Tomasz Gorochowik <tgorochowik@antmicro.com>
The feature that used this symbol was removed in 18b07e09e0.
This is just a cleanup commit that removes the unused symbol.
Signed-off-by: Tomasz Gorochowik <tgorochowik@antmicro.com>
These options were removed from kconfig in previous patch. So rename
the leftovers as shared_irq driver options are defined in DTS now.
Signed-off-by: Daniel Leung <daniel.leung@intel.com>
Same change as in commit 8cf8db3a73 ("Kconfig: Use a short, consistent
style for prompts"), fixing stuff that got introduced since then.
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
In was generating these warnings:
eth_stellaris.c:66:8: warning: unused variable ‘eth_hdr’
eth_stellaris.c:65:8: warning: unused variable ‘head_len_left’
Signed-off-by: Tomasz Bursztyka <tomasz.bursztyka@linux.intel.com>
Wait in the send callback for the packet to be actually sent.
After this change, only one TX packet will be handled at once.
This is needed because of the way the TX packets are currently handled
in L2 after this PR: #12563
This is similar to what #13167 did for the SAM GMAC on SAM E-70.
Without this, packet time-stamping does not work with the current stack.
This commit is minimalistic on purpose to make it easily revertible when
the network stack is able to properly handle DMA drivers for TX packets
again.
Signed-off-by: Tomasz Gorochowik <tgorochowik@antmicro.com>
The ring_buf implementation is basically used as an array of net_buf
pointers in the RX path. The tail and head indexes are taken from the
RX descriptors and not from the ring_buf. That's why for example the
fact that head is never initialized doesn't cause problem. Only len is
used in free_rx_bufs(), but anyway this function is plainly broken as
it always free the first net_buf and doesn't set it back to NULL.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Scale down the TX path of the GMAC driver by waiting for a packet to be
fully sent before returning from the send function. This has a small
performance impact, but has a few advantages:
- It allows the Ethernet code to modify the packet afterward, fixing PTP
support on this board (see PR #12563).
- It returns an error to the IP stack in case of a transmit failure.
- It doesn't require net_buf to be thread safe.
This change can be reverted by changing GMAC_MULTIPLE_TX_PACKETS from 0
to 1.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
The GMAC driver tries to ensure that it can always send a complete
Ethernet frame. However in the TX path this is the problem of the IP
stack: if the buffers can't be allocated, they won't be sent. Therefore
just drop the check.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Now that the RX packets are limited to the RX workqueue only, we can
reduce ETH_SAM_GMAC_BUF_RX_COUNT to 12 and still be able to receive a
full Ethernet frame. This reduces the minimum NET_BUF_RX_COUNT required
by this driver to 24.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
This reverts commit afbee4c96a. The IP
stack has been fixed and does not use the RX packets or buffers to
transmit data.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Commit 040d6b6e99 (eth: eth_mcux: Convert to use DT_ prefixed defines)
changed all the defines, but to incorrect ones.
This commit changes them to what actually gets generated.
Signed-off-by: Tomasz Gorochowik <tgorochowik@antmicro.com>
The eth mcux driver has been using non DT_ prefixed defines for DT
generated defines. Switch to use DT_ prefixed ones as we want to
deprecated the non DT_ prefixed defines.
Signed-off-by: Kumar Gala <kumar.gala@linaro.org>
The SAM GMAC Ethernet driver currently keeps a reference to the packet
being sent in addition than keeping of references of the fragments. In
practice this is only needed when PTP is enabled, otherwise the driver
only need to prevent the fragment (or even their content) to be changed.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
The SAM GMAC Ethernet driver uses scatter gather DMA to transmit data.
Each fragment of a network packet is mapped from a set of descriptors
that is used by the controller to do the DMA transfer. This means that
the packet is not necessary sent when the send() function returns. For
that reason the driver calls net_pkt_ref() on the packet to prevent it
from being freed. It is then unreferenced with net_pkt_unref() in the
TX ISR when the packet has effectively been sent.
However this doesn't work if the packet is modified in the meantime,
like it will be done in PR #12563 to remove the Ethernet header
contained in the first fragment. To avoid that, call net_pkt_frag_ref()
on each fragment of the packet, and unreferenced them with
net_pkt_frag_unref() in the TX ISR when the packet has effectively been
sent.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Replace the sole use of i2c_burst_read_addr with a more generic API
function to allow the former to be deprecated along with its unreliable
sibling write and update functions.
Signed-off-by: Peter A. Bigot <pab@pabigot.com>
This allows the shared_irq driver to be configured by device tree.
With previous implementation, only the board configuration can
override the IRQ trigger, as the trigger config is a "choice" rather
than "config". With this patch, the driver can be fully configued at
the SoC level.
Signed-off-by: Daniel Leung <daniel.leung@intel.com>
Use the new API where relevant. Only sam_gmac is left aside for now.
This simplifies a lot the code as the caller should only care about
allocating net_pkt and its buffer once, and thus will not need to mess
with "frags" etc...
Signed-off-by: Tomasz Bursztyka <tomasz.bursztyka@linux.intel.com>
Force Operation Mode Strap Override register to disable NANDTree. This
is due to some users reporting PHY entering NANDTree.
Signed-off-by: Andrei Gansari <andrei.gansari@nxp.com>
Boot PHY initialization timeout, caching mechanism fixes and networking
buffer descriptors moved to no cache section. Enabled cache management
in networking driver and manual barriers.
Signed-off-by: Andrei Gansari <andrei.gansari@nxp.com>
Previous SAM E70 HAL version was patched by commit 4dcfc8706 ("Add
missing interrupt number definitions") to add missing interrupt
definitions for SAM GMAC Priority Queues. This has been fixed in the
latest HAL by using a slightly different name. This patch updates the
driver accordingly.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Now that the descriptor lists are in non-cached memory, the helpers are
just accessing the w0 or w1 members of the structure. Just drop them and
access the members directly.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
Atmel SAM Ethernet module is using a scatter-gather technique to
exchange data with the Ethernet driver. To communicate the location
of the data buffers the driver sets-up a so called descriptor list. This
is effectively a place in RAM containing a sequence of 32-bit words
representing buffer location and its status.
Currently the cache coherency is handled using cache clean or
invalidate. Unfortunately this strategy only works correctly when the
corresponding data size is a multiple of the cache line. This is not
the case here and might lead to data loss or corruption.
Instead of using cache operations, this patch moves the descriptor
listed to the recently added non-cached memory region, as recommended by
ATMEL. A data synchronisation barrier is still required for writes, as
the non-cached memory is defined with TEX=1, i.e. it is not strongly
ordered. The descriptor lists alignment can be decreased to 4 bytes, as
required by the SAM Ethernet module.
The RX/TX buffer are left unchanged, still managed by cache operations.
Fixes#9812
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
The RX buffer accounting list is not a list shared with Ethernet device,
therefore there is no need to run a cache clean operation on it.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
The SAM E70 GMAC ethernet driver reference a packet with net_pkt_ref()
when queueing a packet, and unreference it with net_pkt_unref() in the
ISR when it has been fully sent.
The call to net_pkt_ref() is done just after re-enabling the
interruptions, so there is however a small race condition that might
cause the packet to be unreference before being referenced. This is
only theoretical and has not been seen in practice.
Fix that by moving the call to net_pkt_ref() just before re-enabling
the interruptions.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
The previous commit replaced the net_pkt element ref with an element
atomic_ref. CI tests turned up more places where ref was used directly.
This commit converts them to use the new element.
Signed-off-by: Daniel Glöckner <dg@emlix.com>
For some reason, there was sequence like:
1. Get size of RX packet.
2. Allocate pkt buffer.
3. Check if the size of RX packet is too large, then deallocate pkt
buffer and error out.
Instead, reorder operations to check size before allocating buf.
Signed-off-by: Paul Sokolovsky <paul.sokolovsky@linaro.org>
Adds a new config HAS_MCUX_ENET to constrain which socs can enable the
mcux ethernet driver. This will prevent users from enabling the driver
on socs like kl25z or kw41z which do not have ethernet mac hardware.
Signed-off-by: Maureen Helm <maureen.helm@nxp.com>
We now generate CS GPIO defines from the DTS that we can utilize. We
needed to slightly update the #defines in the driver from:
DT_MICROCHIP_ENC28J60_0_CS_GPIOS_{PIN,CONTROLLER} to
DT_MICROCHIP_ENC28J60_0_CS_GPIO_{PIN,CONTROLLER}
Fixes#12640
Signed-off-by: Kumar Gala <kumar.gala@linaro.org>
The SAM E70 Ethernet driver uses scatter gather DMA to transmit data.
Each fragment of a network packet is mapped from a set of descriptors
that is used by the controller to do the DMA transfer. Each descriptor
contain an address and a length/status. The important status bits are
GMAC_TXW1_LASTBUFFER to indicate the last fragment of a packet and
GMAC_TXW1_USED to indicate that a descriptor has been processed by the
controller.
When starting a transmission, the controller start at the descriptor
after the last one that has been processed. If the descriptor is NOT
flagged by GMAC_TXW1_USED, it sends a first packet by sending all the
fragments up to a descriptor flagged with GMAC_TXW1_LASTBUFFER. The
first descriptor of a packet *and only the first descriptor of a packet*
is then modified to flag it with GMAC_TXW1_USED and to provide a status
(mostly related to errors and checksum offloading). It then continues
with the next packet and so on and only stops if the next descriptor
after GMAC_TXW1_LASTBUFFER is flagged with GMAC_TXW1_USED.
Therefore in order for the controller to stop processing descriptors,
the strategy is to flag the next descriptor after the last fragment to
be sent with GMAC_TXW1_USED. When the next packet has to be queued, the
flag can be removed before starting a transmission.
This is what is currently done in the current driver. However there is a
small race condition in the implementation: if packets are queued fast
enough, the controller is still sending the fragment of the previous
packet when the descriptor are written. When writing the first
descriptor, the GMAC_TXW1_USED flag is removed. This is done after
writing the address (with a memory barrier) so that looks safe. However
given that the GMAC_TXW1_USED flag is only added by the controller to
the first descriptor of a packet it means the next descriptor might
have it cleared. In that case the descriptor is processed, and a junk
packet is sent. That also desynchronize eth_tx and tx_complete as one
or more packets than expected are transmitted.
In order to fix that the strategy is slightly changed to initially write
the first descriptor with the GMAC_TXW1_USED flag set. Once all the
descriptors from the packet are written the bit is cleared (after a
memory barrier). Then the transmission can be started safely.
The patch also does a small optimization writing the next descriptor
with only the GMAC_TXW1_USED bit set instead of setting this bit. As
this will be a non-cached area, it's better avoiding a read followed
by a write if not necessary.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
If a pkt has more frags than the number of TX descriptors, we end up in
a deadlock situation, as the whole packet and thus all the frags have to
be mapped in the descriptors at once. That is why the number of
descriptors is defined as CONFIG_NET_BUF_TX_COUNT + 1.
This wrongly assumes that only TX buffers can be used to send data,
however the packets might also come from the RX buffers, like for
example with ICMPv4.
Therefore define the number of descriptors as the maximum of
CONFIG_NET_BUF_RX_COUNT + 1 and CONFIG_NET_BUF_TX_COUNT + 1. This fixes
a deadlock when CONFIG_NET_BUF_TX_COUNT is much smaller than
CONFIG_NET_BUF_RX_COUNT.
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
