| Age | Commit message (Collapse) | Author |
|---|
|
As no platforms enable the ethernet driver, remove it.
Signed-off-by: Tom Rini <[email protected]>
|
|
A large number of network drivers cannot build without access to some
platform specific header files. Express those requirements in Kconfig as
well. This covers the QUICC engine drivers as that is networking
driver infrastructure.
Signed-off-by: Tom Rini <[email protected]>
|
|
The common portable header for CPU related functions such as cache
flushing and invalidation is <cpu_func.h> so add that to these drivers.
Signed-off-by: Tom Rini <[email protected]>
|
|
In the debug print in ftgmac100_send we want to say where the packet is
in memory and what the length is, so use %p to print that.
Signed-off-by: Tom Rini <[email protected]>
|
|
Instead of strcpy which is unbounded use strlcpy to ensure that the
receiving buffer cannot be overflowed.
This issue found by Smatch.
Signed-off-by: Andrew Goodbody <[email protected]>
|
|
The kernel-doc comment for the axi_mrmac_recv function was missing
the colon (':') after the '@packetp' parameter tag.
The kernel-doc comment for the axi_mrmac_free_pkt function was missing
the colon (':') after the 'length' parameter tag.
This caused a Sparse warnings regarding the 'packetp' and 'length'
parameters not being described. Fix the formatting to align
with kernel-doc standards and resolve the warning.
drivers/net/xilinx_axi_mrmac.c:357:
warning: Function parameter or member 'packetp' not described
in 'axi_mrmac_recv'
drivers/net/xilinx_axi_mrmac.c:411:
warning: Function parameter or member 'length' not described
in 'axi_mrmac_free_pkt'
Signed-off-by: Venkatesh Yadav Abbarapu <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Michal Simek <[email protected]>
|
|
The Designware IP integrated in TH1520 SoC requires extra clock
configuration to operate correctly. The Linux kernel's T-Head DWMAC glue
driver is ported and adapted to U-Boot's API.
Signed-off-by: Yao Zi <[email protected]>
Acked-by: Leo Yu-Chi Liang <[email protected]>
|
|
Mikhail Kshevetskiy <[email protected]> says:
Several fixes for the airoha ethernet driver.
Link: https://lore.kernel.org/r/[email protected]
|
|
Transmission of short packets does not work good for XFI (GDM2) and
HSGMII (GDM3) interfaces. The issue can be solved with:
- padding of short packets to 60 bytes
- setting of PAD_EN bit in the corresponding REG_GDM_FWD_CFG(n)
register.
The issue should present for the lan switch (GDM1) as well, but it does
does not appear due to unknown reason.
This patch set PAD_EN bit for the used GDM.
Signed-off-by: Mikhail Kshevetskiy <[email protected]>
|
|
ARCH_DMA_MINALIGN is 64 for ARMv7a/ARMv8a architectures, but RX/TX
descriptors are 32 bytes long. So they may not be aligned on an
ARCH_DMA_MINALIGN boundary. In case of RX path, this may cause the
following problem
1) Assume that a packet has arrived and the EVEN rx descriptor has been
updated with the incoming data. The driver will invalidate and check
the corresponding rx descriptor.
2) Now suppose the next descriptor (ODD) has not yet completed.
Please note that all even descriptors starts on 64-byte boundary,
and the odd ones are NOT aligned on 64-byte boundary.
Inspecting even descriptor, we will read the entire CPU cache line
(64 bytes). So we read and sore in CPU cache also the next (odd)
descriptor.
3) Now suppose the next packet (for the odd rx descriptor) arrived
while the first packet was being processed. So we have new data
in memory but old data in cache.
4) After packet processing (in arht_eth_free_pkt() function) we will
cleanup the descriptor and put it back to rx queue.
This will call flush_dcache_range() function for the even descriptor,
so the odd one will be flushed as well (it is in the same cache line).
So the old data will be written to the next rx descriptor.
5) We get a freeze. The next descriptor is empty (so the driver is
waiting for packets), but the hardware will continue to receive
packets on other available descriptors. This will continue until
the last available rx descriptor is full. Then the hardware will
also freeze.
The problem will be solved if the previous descriptor will be put back
to the queue instead of the current one.
If the current descriptor is even (starts on a 64-byte boundary),
then putting the previous descriptor to the rx queue will affect
the previous cache line. To be 100% ok, we must make sure that the
previous and the one before the previous descriptor cannot be used
for receiving at this moment.
If the current descriptor is odd, then the previous descriptor is on
the same cache line. Both (current and previous) descriptors are not
currently in use, so issue will not arrise.
WARNING: The following restrictions on PKTBUFSRX must be held:
* PKTBUFSRX is even,
* PKTBUFSRX >= 4.
The bug appears on 32-bit airoha platform, but should be present on
64-bit as well.
The code was tested both on 32-bit and 64-bit airoha boards.
Signed-off-by: Mikhail Kshevetskiy <[email protected]>
|
|
The dma_map_single() function calls one of the functions
* invalidate_dcache_range(),
* flush_dcache_range().
Both of them expect that 'vaddr' is aligned to the ARCH_DMA_MINALIGN
boundary. Unfortunately, RX/TX descriptors are 32-byte long. Thus they
might not be aligned to the ARCH_DMA_MINALIGN boundary. Data flushing
(or invalidating) might do nothing in this case.
The same applies to dma_unmap_single() function.
In the TX path case the issue might prevent package transmission (filled
TX descriptor was not flushed).
To fix an issue a special wrappers for
* dma_map_single(),
* dma_unmap_single()
functions were created. The patch fix flushing/invalidatiog for the
RX path as well.
The bug appears on 32-bit airoha platform, but should be present on
64-bit as well.
The code was tested both on 32-bit and 64-bit airoha boards.
Signed-off-by: Mikhail Kshevetskiy <[email protected]>
|
|
Compatible device list must have a terminator. If terminator is missed
the u-boot driver subsystem will access random data placed after the
list in the memory.
The issue can be observed with the "dm compat" command.
Signed-off-by: Mikhail Kshevetskiy <[email protected]>
|
|
The TPID value should be 0x9100 instead of 0x8100 according to the
datasheet.
Fixes: cedafee9ff3 (net: mediatek: add support for Airoha AN8855 ethernet switch)
Signed-off-by: Weijie Gao <[email protected]>
|
|
- SH Ether clean ups, RZ/A1 clean ups, RZ/A1 Genmai support
- Gen3 EEPROM DT node clean up
- V4H SA0 BootROM compatible binman etype, SCIF compatible SREC
generation for Gen4
|
|
Drop CFG_SH_ETHER_PHY_ADDR from README and configuration files, this
value is never used, PHY address is extracted from control DT instead.
No functional change intended.
Signed-off-by: Marek Vasut <[email protected]>
|
|
Drop unused struct sh_eth_info *port_info .phy_addr member assignment.
PHY address is extracted from control DT. No functional change intended.
Signed-off-by: Marek Vasut <[email protected]>
|
|
Turn the current cache operation macros into static functions to improve
compiler coverage checking. This does change the driver behavior slightly,
the driver now expects those cache operation functions to be available on
all architectures on which it is used. This should pose no problem, as the
driver is only used on 32bit and 64bit ARM, which both have those operations.
The CFG_SH_ETHER_ALIGNE_SIZE is converted to SH_ETHER_ALIGN_SIZE and defined
as either 64 on ARM or 16 on SH.
Signed-off-by: Marek Vasut <[email protected]>
|
|
The CFG_SH_ETHER_USE_PORT configuration option is a remnant from
before U-Boot DM existed and SH Ethernet made full use of it, and
is no longer used, remove it.
Signed-off-by: Marek Vasut <[email protected]>
|
|
The struct sh_eth_dev .port member is always set to 0, therefore only
single-ported SH Ethernet is ever used. Support for multiple SH Ethernet
ports implemented on driver level is a remnant from before U-Boot DM
existed.
Pass struct sh_eth_info port_info around directly and remove the
struct sh_eth_dev entirely. Handling of multiple ports should be done
by U-Boot DM and multiple per-driver-instance private data.
No functional change intended.
Signed-off-by: Marek Vasut <[email protected]>
|
|
Update the ICSSG PRU Ethernet driver to read PRU/RTU/TXPRU firmware names
from the Device Tree using the "firmware-name" property, instead of relying
on the hard-coded firmware names. The firmware names are parsed during
prueth_probe() and stored in the prueth->firmwares for each slice. The
driver now uses these dynamically loaded names when starting the PRU cores.
This change improves flexibility and allows firmware selection to be
controlled via the device tree, making the driver more adaptable to
different platforms and firmware configurations.
Signed-off-by: MD Danish Anwar <[email protected]>
|
|
If phy node isn't found in ethernet-phy subnode, try to get it from
phy-handle. And when this fails, only then use Ethernet node. This
fix results in getting correct phy handle properties for the
phy node if defined.
Signed-off-by: Naresh Kumar Ravulapalli <[email protected]>
|
|
Green Eco"
Romain Gantois <[email protected]> says:
This is version one of my series which enables Ethernet support on the BBGE
board. This requires three main changes:
- Describing the MAC<->PHY link and DP83867 PHY accurately in the device
tree
- Enabling the RGMII1 pinmux configuration
- Enabling the DP83867 driver
These changes are all applied in patch 2.
Patch 1 enables excluding the DP83867 driver from SPL. This is done to
avoid size issues when adding the DP83867 driver to the am335x-evm
defconfig.
Link: https://lore.kernel.org/r/[email protected]
|
|
The DP83867 PHY driver is used by the BeagleBoneGreen Eco board, but adding
it to the am335x-evm defconfig causes SPL to overflow its size limits.
Add a separate option to enable this driver in SPL, so that it can be
enabled in U-Boot without adding unnecessary volume to SPL.
Signed-off-by: Romain Gantois <[email protected]>
Tested-by: Judith Mendez <[email protected]>
|
|
Prepare v2025.07-rc5
With this merge, tighten up the LTO_FLAGS removal we added to not
trigger on ARMv7 (which is Thumb-2 and should be fine).
|
|
Device bus address might not be valid for direct access when the bus
address and CPU address are not the same. Use dev_bus_to_phys() to
translate bus address back to CPU address.
Fixes: 3d98b8c504e15 ("net: designware: Invalidate RX buffer cache before freeing the DMA descriptor")
Signed-off-by: Baruch Siach <[email protected]>
Reviewed-by: Ramon Fried <[email protected]>
|
|
Marek Vasut <[email protected]> says:
Rename the environment related variables and add ENV_ prefix, so that
all configuration options which are related to environment would have
an CONFIG_ENV_ prefix. No functional change.
Link: https://lore.kernel.org/r/[email protected]
|
|
Rename the variable and add ENV_ prefix, so that all configuration
options which are related to environment would have an CONFIG_ENV_
prefix. No functional change.
Use ENV_MMC_DEVICE_INDEX to clarify this is the SD/MMC device
index, a number, as enumerated by U-Boot. Update the help text
accordingly.
Signed-off-by: Marek Vasut <[email protected]>
|
|
Rasmus Villemoes <[email protected]> says:
While looking through list.h, I saw that the regular list_* helpers
(and one of the hlist_* ones) still contain the prefetch() that was
removed in linux 14 years ago. It doesn't do anything, but makes the
macros harder to read, so get rid of it, and the fallback, no-op
definition that they relied on. That requires removing a few uses
outside list.h as well.
checkpatch warns about some whitespace issues in list.h, but as I've
copied whole kerneldoc+#define blocks directly from the linux kernel,
I think it's better to just accept that so that we don't introduce
needless diffs. The "macro argument reuse" arguments should also be
ignored, as e.g. the "member" arguments are obviously always just bare
identifiers, and the "pos" arguments must be assigned to multiple
times.
Link: https://lore.kernel.org/r/[email protected]
|
|
These all end up using the no-op prefetch() defined in linux/list.h,
because the only possible real implementation is in
arch/mips/include/asm/processor.h, which is behing
CONFIG_CPU_HAS_PREFETCH which is nowhere defined.
In order to be able to drop that fallback definition from list.h,
first remove all uses.
Signed-off-by: Rasmus Villemoes <[email protected]>
|
|
E Shattow <[email protected]> says:
Make consistent use of lowercase hexadecimal prefix '0x' throughout U-Boot.
There are a few remaining uses of uppercase 'X' to denote hexadecimal prefix
or placeholder in documentation and error messages.
External devicetree-rebasing dts/upstream and the generated code of
xilinx/zynq are ignored for the series.
Link: https://lore.kernel.org/r/[email protected]
|
|
Use consistent lowercase hex prefix style in drivers/*
Does not change hex prefix case in allcaps uppercase style error messages
Signed-off-by: E Shattow <[email protected]>
|
|
Prepare v2025.07-rc4
|
|
Tom Rini <[email protected]> says:
Hey all,
Related to my other series I've posted recently on cleaning up some
headers, this series here is the result of at least lightly auditing the
#includes used in include/[a-m]*.h. This ignores subdirectories, as at
least in part I think the top-level includes we've constructed are the
most likely places to have some extra transitive include paths. I'm sure
there's exceptions and I'll likely audit deeper once this first pass is
done. This only gets as far as "include/m*.h" because I didn't want this
to get too big. This also sets aside <miiphy.h> and <phy.h>. While
miiphy.h does not directly need <phy.h> there are *so* many users and I
think I had half of the tree just about not building when I first tried.
It might be worth further investigation, but it might just be OK as-is.
Link: https://lore.kernel.org/r/[email protected]
|
|
The value of "ETH_ALEN" is defined to 6 in <linux/if_ether.h>. This file
is included in <net.h>. In the places where we had ETH_ALEN but no
direct include of <net.h>, add <linux/if_ether.h>. In the places where
we had a custom name used, make use of ETH_ALEN instead.
Signed-off-by: Tom Rini <[email protected]>
|
|
If the MII interface is used, the PHY is the clock master, thus don't
set the clock rate. On Zynq-7000, this will prevent the following
error:
zynq_gem ethernet@e000b000: failed to set tx clock rate 25000000
Signed-off-by: Martin Kaistra <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Michal Simek <[email protected]>
|
|
Tom Rini <[email protected]> says:
Hey all,
This is a v3 of Simon's series[1] and depends on the series[2] I posted
the other day that removes <env.h> from <command.h>. With this series
done, I believe we've tackled all of the current cases of headers which
include <env.h> without directly needing it. Much of this series is in
fact Simon's v2 with the main differneces being:
- Removing <env.h> from <net.h> at the end
- Removing env_to_ip() given how little it's used rather than shuffling
around where it's declared and un-inline'ing it. For a rarely used
helper, this ends up being cleaner I think. Especially looking at some
of the users (which called env_get repeatedly). If there's strong
opinion here about using the other method[3] we can do that instead.
- Setting aside for now how to handle CMD_ELF=y and NO_NET=y because
today it's actually fine as we unconditionally build lib/net_utils.c
where string_to_ip() is defined. I'm unsure if a further series is
warranted here or not. We rely on link-time optimization to keep code
readable too.
[1]: https://lore.kernel.org/all/[email protected]
[2]: https://lore.kernel.org/all/[email protected]
[3]: https://lore.kernel.org/all/[email protected]
Link: https://lore.kernel.org/r/[email protected]
|
|
Now that env_get_ip() has been removed, the include file <net.h> does
not need anything from <env.h>. Furthermore, include/env.h itself
includes other headers which can lead to longer indirect inclusion
paths. To prepare to remove <env.h> from <net.h> fix all of the
remaining places which had relied on this indirect inclusion to instead
include <env.h> directly.
Reviewed-by: Jerome Forissier <[email protected]> # net/lwip
Acked-by: Ilias Apalodimas <[email protected]>
Reviewed-by: Wolfgang Wallner <[email protected]>
Reviewed-by: Martyn Welch <[email protected]>
Signed-off-by: Tom Rini <[email protected]>
|
|
This file uses the environment but does not include the header file.
Update it.
Signed-off-by: Simon Glass <[email protected]>
|
|
Currently, we have the function env_get_ip which takes an IP address
in string form and returns a struct in_addr representation of that
address. It is however little used and means that a number of places
indirectly (and unclearly) get <env.h> via <net.h>. To clean this up
start by replacing env_get_ip() calls with string_to_ip() calls. This is
generally a no-op as env_get_ip(str) is an inline of
string_to_ip(env_get(str)) but in a few cases we can or already have
stored the result of env_get(str) and can save the additional call.
Signed-off-by: Tom Rini <[email protected]>
|
|
Prepare v2025.07-rc3
|
|
Remove unnecessary CL22 phy reset before ethernet phy driver
probe. Lots of ethernet phys requires driver to load firmware.
Before that, CL22 phy reset may lead to malfunction.
Signed-off-by: SkyLake.Huang <[email protected]>
|
|
With commit cc4e8af2c552, fsl_enetc register accessors have been split to
handle different register offsets on different SoCs. However, for
internal MDIO register calculation, only ENETC_PM_IMDIO_BASE was fixed
without adding the SoC specific MAC register offset.
As a result, the network support for the Kontron SMARC-sAL28 and
probably other boards based on the LS1028A CPU is broken.
Add the SoC specific MAC register offset to calculation of imdio.priv to
fix this.
Fixes: cc4e8af2c552 ("net: fsl_enetc: Split register accessors")
Signed-off-by: Thomas Schaefer <[email protected]>
Signed-off-by: Heiko Thiery <[email protected]>
Reviewed-by: Michael Walle <[email protected]>
Reviewed-by: Vladimir Oltean <[email protected]>
Tested-by: Vladimir Oltean <[email protected]> # LS1028A
Tested-by: Tim Harvey <[email protected]> # imx95_19x19_evk
Reviewed-by: Marek Vasut <[email protected]>
Reviewed-by: Alice Guo <[email protected]>
|
|
Currently, Synopsis xgmac driver only works if DMA region is under 4GB.
This change enables the DMA buffers allocations above 4GB memory
regions.
Signed-off-by: Nikunj Kela <[email protected]>
|
|
Add RMII-specific handling to rk3288_gmac_fix_mac_speed() so that it
properly sets the RMII clock (2.5 MHz vs. 25 MHz) and speed bits
(10 Mbps vs. 100 Mbps). Also define a new rk3288_gmac_set_to_rmii()
function to set the PHY interface field and RMII_MODE bit.
Signed-off-by: Christoph Fritz <[email protected]>
Reviewed-by: Kever Yang <[email protected]>
|
|
1. Add pcie device id 0x15f3
2. Add IIC phy id 0x67C9DC00
Signed-off-by: ZhiJie.Zhang <[email protected]>
|
|
Add compatible "st,stm32mp25-dwmac" to manage STM32MP2 boards
Signed-off-by: Christophe Roullier <[email protected]>
Signed-off-by: Patrice Chotard <[email protected]>
Reviewed-by: Marek Vasut <[email protected]>
Reviewed-by: Patrick Delaunay <[email protected]>
|
|
The Renesas R9A07G044L (RZ/G2L) SoC includes two Gigabit Ethernet
interfaces which can be supported using the ravb driver. Some RZ/G2L
specific steps need to be taken during initialization due to differences
between this SoC and previously supported SoCs. We also need to ensure
that the module reset is de-asserted after the module clock is enabled
but before any Ethernet register reads/writes take place.
Signed-off-by: Paul Barker <[email protected]>
|
|
In order to add support for the Renesas RZ/G2L Ethernet IP in a
subsequent patch, we introduce optional de-assertion and re-assertion of
a reset signal in ravb_probe() and ravb_remove().
Signed-off-by: Paul Barker <[email protected]>
|
|
In order to add support for the Renesas RZ/G2L Ethernet IP in a
subsequent patch, we move all R-Car specific code into new functions and
introduce a device_ops function pointer table.
Signed-off-by: Paul Barker <[email protected]>
|
|
Correctly handle RX errors in ravb_recv() by returning 0 instead
of -EAGAIN on RX error.
In case the RAVB driver detects an RX error in ravb_recv(), it must
not return the -EAGAIN, but instead must return 0. Both error codes
are handled in eth-uclass.c eth_rx() and -EAGAIN is rewritten to 0
at the end of eth_rx(), but negative return code from the .recv()
callback does not trigger .free_pkt() callback, which would clean
up and re-enqueue the descriptor which holds the currently received
corrupted packet. The .free_pkt() must be called for this descriptor,
otherwise the follow up received data become corrupted too, even if
those packets are correctly received. Returning 0 from the .recv()
callback assures the corrupted packet is not processed by the network
stack, but is skipped instead.
For TFTP loading, an RX error produces the timeout "T" output and
resumes the TFTP loading operation shortly afterward, without any
data corruption.
Signed-off-by: Marek Vasut <[email protected]>
Reviewed-by: Paul Barker <[email protected]>
|
