Due to introducing the new peripheral clock handle functions,
use these functions to reduce the duplicated code.
Signed-off-by: Wenyou Yang <wenyou.yang@atmel.com>
Tested-by: Heiko Schocher <hs@denx.de>
[fixup for missing clk.h in at91_emac.c]
Signed-off-by: Andreas Bießmann <andreas.devel@googlemail.com>
With commit a058052c [net: phy: do not read configuration register on
reset], phy_reset() will clear the BMCR register. Resulting in bit 12
being cleared (A/N enable). This leads to autonegotiation link problems,
at least on the Marvell Armada ClearFog board. I suspect that other
boards using this driver will be affected as well.
At the of m88e1111s_config(), phy_reset() is called. This is not needed
for the PHY to load the changed configuration, as phy_reset() is called
a few lines before already. So lets call genphy_restart_aneg() here
instead to start the AN correctly.
Tested on clearfog.
Signed-off-by: Stefan Roese <sr@denx.de>
Cc: Stefan Agner <stefan@agner.ch>
Cc: Hao Zhang <hzhang@ti.com>
Cc: Michal Simek <monstr@monstr.eu>
Cc: Andy Fleming <afleming@gmail.com>
Cc: Joe Hershberger <joe.hershberger@ni.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
Instead of coding the soft PHY reset function multiple times in marvell.c,
lets call the common phy_reset() function from phy.c.
Signed-off-by: Stefan Roese <sr@denx.de>
Cc: Stefan Agner <stefan@agner.ch>
Cc: Hao Zhang <hzhang@ti.com>
Cc: Michal Simek <monstr@monstr.eu>
Cc: Andy Fleming <afleming@gmail.com>
Cc: Joe Hershberger <joe.hershberger@ni.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
When Micrel phy is selected without CONFIG_PHY_MICREL_KSZ9031 or
CONFIG_PHY_MICREL_KSZ9021 there is a build error. Fixing this
by adding proper ifdefs
drivers/net/phy/micrel.c:370:39: error: array type has incomplete element type
static const struct ksz90x1_reg_field ksz9031_ctl_grp[] =
^
drivers/net/phy/micrel.c:372:39: error: array type has incomplete element type
static const struct ksz90x1_reg_field ksz9031_clk_grp[] =
^
drivers/net/phy/micrel.c: In function ‘ksz9031_of_config’:
drivers/net/phy/micrel.c:377:23: error: array type has incomplete element type
struct ksz90x1_ofcfg ofcfg[] = {
^
drivers/net/phy/micrel.c:379:13: error: ‘ksz90x1_rxd_grp’ undeclared (first use in this function)
{ MII_KSZ9031_EXT_RGMII_RX_DATA_SKEW, 2, ksz90x1_rxd_grp, 4 },
^
drivers/net/phy/micrel.c:379:13: note: each undeclared identifier is reported only once for each function it appears in
drivers/net/phy/micrel.c:380:13: error: ‘ksz90x1_txd_grp’ undeclared (first use in this function)
{ MII_KSZ9031_EXT_RGMII_TX_DATA_SKEW, 2, ksz90x1_txd_grp, 4 },
^
drivers/net/phy/micrel.c:386:3: warning: implicit declaration of function ‘ksz90x1_of_config_group’ [-Wimplicit-function-declaration]
ret = ksz90x1_of_config_group(phydev, &(ofcfg[i]));
^
drivers/net/phy/micrel.c:377:23: warning: unused variable ‘ofcfg’ [-Wunused-variable]
struct ksz90x1_ofcfg ofcfg[] = {
^
drivers/net/phy/micrel.c: At top level:
drivers/net/phy/micrel.c:370:39: warning: ‘ksz9031_ctl_grp’ defined but not used [-Wunused-variable]
static const struct ksz90x1_reg_field ksz9031_ctl_grp[] =
^
drivers/net/phy/micrel.c:372:39: warning: ‘ksz9031_clk_grp’ defined but not used [-Wunused-variable]
static const struct ksz90x1_reg_field ksz9031_clk_grp[] =
^
scripts/Makefile.build:277: recipe for target 'drivers/net/phy/micrel.o' failed
make[1]: *** [drivers/net/phy/micrel.o] Error 1
Makefile:1201: recipe for target 'drivers/net/phy' failed
make: *** [drivers/net/phy] Error 2
make: *** Waiting for unfinished jobs....
Signed-off-by: Mugunthan V N <mugunthanvnm@ti.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
NULL check is made after the pointer dereference. This patch
fixes this issue.
Signed-off-by: Vishwas Srivastava <vishu.kernel@gmail.com>
CC: Joe Hershberger <joe.hershberger@ni.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
Signed-off-by: Anatolij Gustschin <agust@denx.de>
emac module of the davinci platform supports only 8 tx and 8
rx channels (total 16). emac driver for davinci platform,
however, while doing initialization of the dma descriptor
head pointers, wrongly initializes the 16 head pointers
(instead of 8) for tx dma and 16 head pointers (insted of 8)
for rx dma, which is wrong. The result is, that this register
initilization spills over the other registers which was not
intended and is undesirable. This patch fixes this problem.
Signed-off-by: Vishwas Srivastava <vishu.kernel@gmail.com>
CC: Joe Hershberger <joe.hershberger@ni.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
Signed-off-by: Anatolij Gustschin <agust@denx.de>
Correct spelling of "U-Boot" shall be used in all written text
(documentation, comments in source files etc.).
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Heiko Schocher <hs@denx.de>
Reviewed-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Minkyu Kang <mk7.kang@samsung.com>
This driver implements MAC and MII layer of the ethernet controller.
Network data transfer is handled by controller internal DMA engine.
Ethernet controller is configurable through device-tree file.
Signed-off-by: Purna Chandra Mandal <purna.mandal@microchip.com>
Update printf with dpbp_exit to match with previous function call.
Signed-off-by: Itai Katz <itai.katz@nxp.com>
Signed-off-by: Prabhakar Kushwaha <prabhakar.kushwaha@nxp.com>
Reviewed-by: York Sun <york.sun@nxp.com>
All fields of struct dprc_cfg are not being configured while creating
child container. "Not" configured fields are assumed to be 0.
So memset dprc_cfg before configuring the fields.
Signed-off-by: Itai Katz <itai.katz@nxp.com>
Signed-off-by: Prabhakar Kushwaha <prabhakar.kushwaha@nxp.com>
Reviewed-by: York Sun <york.sun@nxp.com>
Cortina phy cannot support soft reset, this commit implements probe
for Cortina PHY to tell phylib to skip phy soft reset by setting
PHY_FLAG_BROKEN_RESET in flags.
Signed-off-by: Shaohui Xie <Shaohui.Xie@nxp.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
Current driver always performs a phy soft reset when connecting the phy
device, but soft reset is not always supported by a phy device, so
introduce a quirk PHY_FLAG_BROKEN_RESET to let such a phy device to skip
soft reset. This commit uses 'flags' of phy device structure to store the
quirk.
Signed-off-by: Shaohui Xie <Shaohui.Xie@nxp.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
The picoseconds to register value divisor(ps_to_regval) should be 60 and not
200. Linux has KSZ9031_PS_TO_REG defined to be 60 as well. 60 is the correct
divisor because the 4-bit skew values are defined from 0x0000(-420ps) to
0xffff(480ps), increments of 60.
For example, a DTS skew value of 420, represents 0ps delay, which should be 0x7.
With the previous divisor of 200, it would result in 0x2, which represents a
-300ps delay.
With this patch, ethernet on the SoCFPGA DE0 Atlas is now able to work with
1Gb ethernet.
References:
http://www.micrel.com/_PDF/Ethernet/datasheets/KSZ9031RNX.pdf -> page 26
Signed-off-by: Dinh Nguyen <dinguyen@opensource.altera.com>
Acked-by: Marek Vasut <marex@denx.de>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
Use the 'autoneg' flag available in phydev when checking if
autoneg is in use.
The previous implementation was checking directly in the PHY
if autoneg was supported. Some PHYs will report that autoneg
is supported, even when it is disabled. Thus it is not possible
to use that bit to determine if autoneg is currently in use or
not.
Signed-off-by: Alexandre Messier <amessier@tycoint.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
When configuring a PHY in fixed (forced) link mode, in order for
the changes to be applied, either one of these conditions must
be triggered:
1- PHY is reset
2- Autoneg is restarted
3- PHY transitions from power-down to power-up
Neither of these is currently done, so effectively the fixed link
configuration is not applied in the PHY.
Fix this by setting the Autoneg restart bit.
Signed-off-by: Alexandre Messier <amessier@tycoint.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
Micrel PHYs KSZ8021/31 and KSZ8081 have a feature where MDIO address 0
is considered as a broadcast address; the PHY will respond even if it
is not its configured (pinstrapped) address. This feature is enabled
by default.
The Linux kernel disables that feature at initialisation, but not
before it probes the MDIO bus. This causes an issue, because a PHY
at address 3 will be discovered at addresses 0 and 3, but will then
only respond at address 3. Because Linux attaches the first PHY it
discovers on 'eth0', it will attach the PHY from address 0, which
will never answer again.
Fix the issue by disabling the broadcast feature in U-Boot, before
Linux is started.
Signed-off-by: Alexandre Messier <amessier@tycoint.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
This property allows to specify fastest connection mode supported by
the MAC (as opposed to features of the phy).
There are situations when phy may handle faster modes than the
MAC (or even it's particular implementation or even due to CPU being too
slow).
This property is a standard one in Linux kernel these days and some
boards do already use it in their device tree descriptions.
Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
Cc: Bin Meng <bmeng.cn@gmail.com>
Cc: Joe Hershberger <joe.hershberger@ni.com>
Cc: Sonic Zhang <sonic.zhang@analog.com>
cc: Simon Glass <sjg@chromium.org>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
Current implementation only sets "port select" bit for non-1Gb mode.
That works fine if GMAC has just exited reset state but we may as well
change connection mode in runtime. Then we'll need to reprogram GMAC for
that new mode of operation and if previous mode was 10 or 100 Mb and new
one is 1 Gb we'll need to reset port mode bit.
Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
Cc: Bin Meng <bmeng.cn@gmail.com>
Cc: Joe Hershberger <joe.hershberger@ni.com>
Cc: Sonic Zhang <sonic.zhang@analog.com>
cc: Simon Glass <sjg@chromium.org>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
This new function will allow MAC drivers to override supported
capabilities of the phy. It is required when MAC cannot handle all
speeds supported by phy.
For example phy supports up-to 1Gb connections while MAC may only work
in modes up to 100 or even 10 Mbit/sec.
Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
Cc: Joe Hershberger <joe.hershberger@ni.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
of_set_phy_supported allows overwiting hardware capabilities of
a phy with values from the devicetree. This does not work with
the genphy driver though because the genphys config_init function
will overwrite all values adjusted by of_set_phy_supported. Fix
this by initialising the genphy features in the phy_driver struct
and in config_init just limit the features to the ones the hardware
can actually support. The resulting features are a subset of the
devicetree specified features and the hardware features.
This is a copy of the patch from Linux kernel, see
http://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=c242a47238fa2a6a54af8a16e62b54e6e031d4bc
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
Cc: Joe Hershberger <joe.hershberger@ni.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
When a Gigabit PHY device is connected to a 10/100Mbits capable Ethernet
MAC, the driver will restrict the phydev->supported modes to mask off
Gigabit. If the Gigabit PHY comes out of reset with the Gigabit features
set by default in MII_CTRL1000, it will keep advertising these feature,
so by the time we call genphy_config_advert(), the condition on
phydev->supported having the Gigabit features on is false, and we do not
update MII_CTRL1000 with updated values, and we keep advertising Gigabit
features, eventually configuring the PHY for Gigabit whilst the Ethernet
MAC does not support that.
This patches fixes the problem by ensuring that the Gigabit feature bits
are always cleared in MII_CTRL1000, if the PHY happens to be a Gigabit
PHY, and then, if Gigabit features are supported, setting those and
updating MII_CTRL1000 accordingly.
This is a copy of patch from Linux kernel, see
http://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=5273e3a5ca94fbeb8e07d31203069220d5e682aa
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
Cc: Joe Hershberger <joe.hershberger@ni.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
Add a new member 'tbiaddr' to tsec_private struct. For non-DM driver,
it is initialized as CONFIG_SYS_TBIPA_VALUE, but for DM driver, we
can get this from device tree. Update the bindings doc as well.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
This adds driver model support to Freescale TSEC ethernet driver.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
For internal routines like redundant_init(), startup_tsec() and
init_phy(), change to use tsec_private pointer as the parameter.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
Adjust static functions in a proper order so that forward declaration
of tsec_send() can be avoided.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
rxbd and txbd are declared static with 8 byte alignment requirement,
but they can be put into struct tsec_private as well and are natually
aligned to 8 byte.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
At present rx_idx and tx_idx are declared as static variables
in the driver codes. To support multiple interfaces, move it to
struct tsec_private.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
Clean up the tsec and fsl_mdio driver codes a little bit, by:
- Fix misuse of tab and space here and there
- Use correct multi-line comment format
- Replace license identifier to GPL-2.0+
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
You can now configure LAG on VSC9953's ports using the command:
ethsw [port <port_no>] aggr {[help] | show | <lag_group_no>}
A port must belong to a single LAG. By default, a port
belongs to a LAG equal to the port's number.
For each frame, a hash will be calculated based on
Source/Destination MAC addresses, Source/Destination IP(v4/v6)
addresses, Source/Destination ports. This hash will be used to
select a single egress port from LAG. This also assures
that frames from the same flow will always have the
same egress port.
Signed-off-by: Codrin Ciubotariu <codrin.ciubotariu@freescale.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
By default, the aging period is set to 0, so the dynamic
FDB entries are never removed. This patch sets the aging
time to 300 seconds.
Signed-off-by: Codrin Ciubotariu <codrin.ciubotariu@freescale.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
When doing a software reset, the reset flag should be written without
other bits set. Writing the current state will lead to restoring the
state of the PHY (e.g. Powerdown), which is not what is expected from
a software reset.
Signed-off-by: Stefan Agner <stefan@agner.ch>
Acked-by: Michael Welling <mwelling@ieee.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
Management Complex firmware 9.0 has fixed the issue of
dprc_destroy_container i.e. the used portal is not return to the
free pool. Which was resulting in error ethernet driver want to
use this portal via either DPL or dynamically in Linux.
Hard-coding of portal id is removed.
Signed-off-by: Prabhakar Kushwaha <prabhakar@freescale.com>
Reviewed-by: York Sun <york.sun@nxp.com>
Add debug information prints to provide DPMAC statistics
- Number of bytes received
- Number of received and discard frames
- Number of bytes transferred
- Number of frames transferred
etc.
Signed-off-by: Prabhakar Kushwaha <prabhakar@freescale.com>
Reviewed-by: York Sun <york.sun@nxp.com>
Management Complex FW 9.0 set the hardware depletion to be 20
buffers in order to support multiple pools in DPNI. This requires
driver to fill the pool with at least 21 to be able to receive
frames. So, Increase number of buffers for a pool.
Signed-off-by: Prabhakar Kushwaha <prabhakar@freescale.com>
Reviewed-by: York Sun <york.sun@nxp.com>
Management Complex FW 9.0 puts a new requirement to provide Tx
confirmation and error queue configuration by calling
dpni_set_tx_conf API.
Configure report of only error frames for a tx frame.
Signed-off-by: Prabhakar Kushwaha <prabhakar@freescale.com>
Reviewed-by: York Sun <york.sun@nxp.com>
Management Complex FW 9.0 puts a new requirement to prepare extended
parameters which should be provided as input in dpni_create. extended
parameters includes traffic class and IP reassembly configurations.
So prepare extended parameters with default "0" as input for
dpni_create.
Signed-off-by: Prabhakar Kushwaha <prabhakar@freescale.com>
Reviewed-by: York Sun <york.sun@nxp.com>
MC firmware version 9.0.0 contains
- Support of new APIs
- Update in existing APIs
- Change in Major and minor version of DPAA2 objects
This patch contains modifications in FLIB files to support new
MC firmware version.
Signed-off-by: Prabhakar Kushwaha <prabhakar@freescale.com>
Reviewed-by: York Sun <york.sun@nxp.com>
Check and compare version of management complex's object with
the version supported by Freescale ldpaa2 ethernet driver.
Signed-off-by: Prabhakar Kushwaha <prabhakar@freescale.com>
Reviewed-by: York Sun <york.sun@nxp.com>
Size of this snprintf "lltemac.%lx" is bigger than 16 characters.
Replacing it with "ll_tem.%lx"
Signed-off-by: Ricardo Ribalda Delgado <ricardo.ribalda@gmail.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Reviewed-by: Michal Simek <michal.simek@xilinx.com>
MDC setting depends on pclk input clocks which varies across SoC. This
driver is used by xilinx zynq and zynqmp SOC.
Input clock frequence on silicon is 125MHz where divider 64 put
frequency below 2.5MHz requires by spec (125/64=1.95).
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Enable automatic crossover cable detection.
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Reviewed-by: Edgar E. Iglesias <edgar.iglesias@xilinx.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
Add PHYLIB and MII dependencies and enable it by default for Microblaze.
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
When IP is configured with pong buffers, IP is receiving packets to ping
and then to pong buffer and than ping again.
The original logic in the driver remains there that when ping buffer is
free, pong buffer is checked too and return if both are free.
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
The original logic in the driver was exchanging buffers which are used for
sending packet and tx_ping and tx_pong buffers were exchanged all the
time to ensure that IP has enough time to send the packet out.
Based on this "feature" send function was using nextbuffertouse variable
to save which buffer should be used.
Before this algorithm was called driver checked that there is free
buffer available.
This checking remains in the driver but driver tries to use tx_ping
first if available. If not, tx_pong buffer is used instead.
To reach this code the original condition is met that at least one of the
buffer should be available.
Testing doesn't show any performance drop when this patch is applied.
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
This flag is not documented anywhere in the latest documentation that's
why this patch removes it.
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
Simplify logic how to find out if there is free TX buffer.
Both buffers are checked all the time that's why logic around order
can be removed.
Also add check when only one buffer is available.
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
Move to use indirect register access when timeout expires for resetting
TX buffers.
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
Do initialization via indirect register access.
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
Use u-boot coding style how to setup and access MDIO bus.
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
Also add dependency on PHYLIB and MII which is required.
Clean PHYLIB dependency from the driver too.
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
Rename few functions to fit to the new name convention used by DM.
Suggested-by: Joe Hershberger <joe.hershberger@ni.com>
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Saving iobase directly to private structure helps with moving to DM.
There is an option to load iobase from pdata but it is additional load.
Pointer to private structure is available all the time.
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
Simplify mdio_wait function by passing regs directly.
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
Fix these compilation warning by proper grouping:
In function 'axi_dma_init': drivers/net/xilinx_axi_emac.c:391:7:
warning: suggest parentheses around operand of '!' or change '&' to '&&'
or '!' to '~' [-Wparentheses]
if (!(in_be32(&priv->dmatx->control) |
^
In function 'axiemac_send': drivers/net/xilinx_axi_emac.c:501:21:
warning: suggest parentheses around operand of '!' or change '&' to '&&'
or '!' to '~' [-Wparentheses]
while (timeout && (!in_be32(&priv->dmatx->status) &
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
During the initialization of PHY the gigabit bit capable is set if the
controller is a GEM. However, for sama5d2 and sama5d4, the GEM is
configured to support only 10/100.
Improperly setting the GBE capability leads to an unresponsive MAC
controller. This patch fixes this behavior allowing using the gmac with
these SoCs.
Suggested-by: Nicolas Ferre <nicolas.ferre@atmel.com>
Signed-off-by: Gregory CLEMENT <gregory.clement@free-electrons.com>
Acked-by: Nicolas Ferre <nicolas.ferre@atmel.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
Reviewed-by: Andreas Bießmann <andreas.devel@googlemail.com>
[fixed minor checkpatch warning]
Signed-off-by: Andreas Bießmann <andreas.devel@googlemail.com>
Current code compares the return pointer of function
qbman_cena_write_start with NULL. Instead the value of the return
pointer should be compared.
Signed-off-by: Pratiyush Mohan Srivastava <pratiyush.srivastava@freescale.com>
Acked-by: Prabhakar Kushwaha <prabhakar.kushwaha@nxp.com>
Reviewed-by: York Sun <york.sun@nxp.com>
When building katmai, it reports quite a lot
warning: cast to pointer from integer of different size [-Wint-to-pointer-cast]
Fix this by casting the dev->iobase with u_long.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
BUILD_BUG_* macros have been defined in several headers. It would
be nice to collect them in include/linux/bug.h like Linux.
This commit is cherry-picking useful macros from include/linux/bug.h
of Linux 4.4.
I did not import BUILD_BUG_ON_MSG() because it would not work if it
is used with include/common.h in U-Boot. I'd like to postpone it
until the root cause (the "error()" macro in include/common.h causes
the name conflict with "__attribute__((error()))") is fixed.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
pci_virt_to_mem() uses virt_to_phys() to get the physical address.
But pci_virt_to_mem() is also called with uncached addresses which
is wrong according to the documentation of virt_to_phys().
Refactor the PCI_TO_MEM() macro to optionally map an uncached address
back to a cached one before calling pci_virt_to_mem().
Currently pcnet works because virt_to_phys() is incorrectly implemented
on MIPS. With the upcoming asm header file update for MIPS, the
virt_to_phys() implementation will be fixed. Thus this patch is needed
to keep pcnet working on MIPS Malta board.
Signed-off-by: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
With format-security errors turned on, GCC picks up the use of sprintf with
a format parameter not being a string literal.
Simple uses of sprintf are also converted to use strcpy.
Signed-off-by: Ben Whitten <ben.whitten@gmail.com>
Acked-by: Wolfgang Denk <wd@denx.de>
Reviewed-by: Tom Rini <trini@konsulko.com>
Update this driver to support driver model. As all MVEBU boards using
this driver are converted with this patch, the non-driver-model code
can be removed completely. This is also the reason why this patch
is quite big and includes a) the driver change and b) the
platform change. As its not git-bisect save otherwise.
With this conversion, some parameters are now extracted from the
DT instread of using the config header defines. The supported
properties right now are:
PHY-mode ("phy-mode") and PHY-address ("reg").
The base addresses for the ethernet controllers can be removed from
the header files as well.
Please note that this patch also removes the E1000 network driver
from some MVEBU config headers. This is necessary, as with DM_ETH
configured and the e1000 driver enabled, the PCI driver also needs
to support DM. But the MVEBU PCI(e) driver still needs to get
ported to DM. When this is done, the E1000 driver can be enabled
again.
Signed-off-by: Stefan Roese <sr@denx.de>
Cc: Luka Perkov <luka.perkov@sartura.hr>
Cc: Dirk Eibach <dirk.eibach@gdsys.cc>
Cc: Joe Hershberger <joe.hershberger@ni.com>
Cc: Simon Glass <sjg@chromium.org>
The change fixes PHY write operation, which incorrectly waits for
released busy state before issuing a write operation, this breaks
sequential write/read operation logic, because read operation
starts immediately on request and it completes, when busy state is
gone.
Instead of adding the second preceding busy state check to read
function, do busy state release check after issuing a write operation,
this method of operation is also recommended by the LPC32xx User's
Manual, see MII Mgmt Indicators Register notes:
For PHY Write if scan is not used:
1. Write 0 to MCMD
2. Write PHY address and register address to MADR
3. Write data to MWTD
4. Wait for busy bit to be cleared in MIND
Reported-by: Alexandre Messier <amessier@tycoint.com>
Signed-off-by: Vladimir Zapolskiy <vz@mleia.com>
Tested-by: Alexandre Messier <amessier@tycoint.com>
This will allow the implementation to make use of data in the block_dev
structure beyond the base device number. This will be useful so that eMMC
block devices can encompass the HW partition ID rather than treating this
out-of-band. Equally, the existence of the priv field is crying out for
this patch to exist.
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
Update this driver to use the proper driver-model PCI API functions.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
Most driver model PCI functions have a dm_ prefix. At some point, when the
old code is converted to driver model and the old functions are removed, we
will drop that prefix.
For consistency, we should use the dm_ prefix for all driver model
functions. Update pci_get_bdf() accordingly.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Introduced in 45b4773 (net/arp: account for ARP delay, avoid duplicate packets on timeout)
Check the arp timeout and adjust the timeout start time before the call
to eth_recv() so that the sandbox driver has the opportunity to adjust
the sandbox timer after the new start time has been recorded.
Also, change the adjustment amount by 11 seconds instead of exactly the
10 seconds that the ping timout is expecting since the timeout check is
looking for the time elapsed to be greater than but not equal to the
specified delay.
Signed-off-by: Joe Hershberger <joe.hershberger@ni.com>
Reviewed-by: Stefan Brüns <stefan.bruens@rwth-aachen.de>
Acked-by: Simon Glass <sjg@chromium.org>
Commit 08ad9b068a (" ar8031: modify the config func of ar8031 to
ar8021_config") selected 'ar8021_config' as the configuration function
for AR8031.
The correct would be to use 'ar8035_config' instead as AR8031/AR8035
have the same programming model and even share the same phy driver
in the linux kernel: drivers/net/phy/at803x.c.
Tested on a mx6qsabresd and wandboard, which now can work without
any PHY setup code in the board files.
Signed-off-by: Fabio Estevam <fabio.estevam@nxp.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
In struct e1000_rx_desc, field 'length' is declared as
uint16_t, so use le16_to_cpu() to do endianness conversion.
Also drop conversion on 'status' which is declared as
uint8_t.
Signed-off-by: Miao Yan <yanmiaobest@gmail.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
If fecmxc_initialize_multi() fails, it frees but does not unregister
the mdio bus, causing subsequent uses of the "mii" command to crash.
Fix this by adding mdio_unregister() calls where needed.
Signed-off-by: Mans Rullgard <mans@mansr.com>
Reviewed-by: Eric Nelson <eric@nelint.com>
If the host clock frequency is higher than 100 MHz, the MDIO hold
time needs to be increased from its current setting of one cycle in
order to meet the specified minium of 10 ns. Writing an appropriate
value to the HOLDTIME field of the MII_SPEED register achieves this.
Comment copied from Linux kernel.
Signed-off-by: Mans Rullgard <mans@mansr.com>
Reviewed-by: Eric Nelson <eric@nelint.com>
Trailing backslashes are necessary only in macros, not in the actual
code, so remove them.
Signed-off-by: Marek Vasut <marex@denx.de>
Cc: Dinh Nguyen <dinguyen@opensource.altera.com>
Cc: Joe Hershberger <joe.hershberger@ni.com>
Select PHYLIB in drivers/net/Kconfig. And remove CONFIG_PHYLIB
from legacy board header files.
This fixed the warnings when both ALTERA_TSE and ETH_DESIGNWARE
are selected.
Signed-off-by: Thomas Chou <thomas@wytron.com.tw>
Reported-by: Pavel Machek <pavel@denx.de>
Acked-by: Chin Liang See <clsee@altera.com>
Acked-by: Pavel Machek <pavel@denx.de>
Tested-by: Pavel Machek <pavel@denx.de>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Per the comments, e1000_spi_eeprom_disable_wr() and
e1000_spi_eeprom_write_status() have been tested.
Remove the #if 0, #endif and mark them as __maybe_unused.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
CONFIG_MVBC_1G is not referenced anywhere, hence remove it.
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
Add code to process the KSZ9021/KSZ9031 OF props if they are present
and configure skew registers based on the information from the OF.
This code is only enabled if the DM support for ethernet is also
enabled.
Signed-off-by: Marek Vasut <marex@denx.de>
Cc: Joe Hershberger <joe.hershberger@ni.com>
Cc: Chin Liang See <clsee@altera.com>
Cc: Dinh Nguyen <dinguyen@opensource.altera.com>
V2: - Implement struct ksz90x1_reg_field to describe the skew register
fields more accurately.
- Fix RXDV/TXEN skew register default value and offset.
Use core to call net_process_received_packet() instead of call inside
the driver.
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
recv function should return 0 instead of frame_len not to
proceed the same packet again in core.
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
We don't disable unused FM1-DTSEC1 MAC node in FMAN v2 since it is
used by MDIO. For FMAN v3, MDIO uses dedicated controller, so we
can disable unused FM1-DTSEC1 MAC node to avoid being probed in
Linux.
Signed-off-by: Shaohui Xie <Shaohui.Xie@freescale.com>
[York Sun: revised commit message]
Reviewed-by: York Sun <yorksun@freescale.com>
Remove verbose message for FMan port.
Signed-off-by: Mingkai Hu <Mingkai.Hu@freescale.com>
[York Sun: Added commit message]
Reviewed-by: York Sun <yorksun@freescale.com>
The MC version numbers provide no meaningful information
about binary interface compatibility, so remove the
check which refuses to start the MC unless a specific
version is found.
Version checking is supposed to be done at the individual
object level, and individual drivers are responsible
for their own version checking.
Signed-off-by: Stuart Yoder <stuart.yoder@freescale.com>
Acked-by: Prabhakar Kushwaha <prabhakar@freescale.com>
Reviewed-by: York Sun <yorksun@freescale.com>
With gcc-5.x we get a warning about the ambiguity of BUG_ON(!a != b) and
becomes BUG_ON((!a) != b). In this case reading of the function leads to
us wanting to rewrite this as BUG_ON(a != b).
Cc: Prabhakar Kushwaha <prabhakar@freescale.com>
Cc: Geoff Thorpe <Geoff.Thorpe@freescale.com>
Cc: Haiying Wang <Haiying.Wang@freescale.com>
Cc: Roy Pledge <Roy.Pledge@freescale.com>
Cc: York Sun <yorksun@freescale.com>
Signed-off-by: Tom Rini <trini@konsulko.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: York Sun <yorksun@freescale.com>
Do not set interface via configs. Read information from DT.
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Reviewed-by: Jagan Teki <jteki@openedev.com>
- Enable DM_ETH by default for Zynq and ZynqMP
- Remove board_eth_init code
- Change miiphy_read function to return value instead of error code
based on DM requirement
- Do not enable EMIO DT support by default
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Reviewed-by: Jagan Teki <jteki@openedev.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
This function was used for OF init before DM.
Remove this function as the part of move to DM.
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Jagan Teki <jteki@openedev.com>
Enable access to MDIO before zynq_gem_init is called.
It enables read information about phy earlier.
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Move phy init code out of zynq_gem_init. DM drivers are normally calling
this code from probe function.
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Next step to move driver to driver model. Do not use eth_device
structure. Use private structure instead.
Add iobase to private structure to store gem iobase.
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Add return value for phy detection algorithm to stop init function when
phy is not found.
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Reviewed-by: Jagan Teki <jteki@openedev.com>
We now use the generic delay method which specifies the timeout as
microseconds instead of ticks.
Signed-off-by: Francois Retief <fgretief@spaceteq.co.za>
The phy can share driver with other aquantia PHYs, so we only
add PHY ID.
Signed-off-by: Shaohui Xie <Shaohui.Xie@freescale.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
Reviewed-by: York Sun <yorksun@freescale.com>
Freescale's LS2085A is a another personality of LS2080A SoC with
support of AIOP and DP-DDR.
This Patch adds support of LS2085A Personality.
Signed-off-by: Pratiyush Mohan Srivastava <pratiyush.srivastava@freescale.com>
Signed-off-by: Prabhakar Kushwaha <prabhakar@freescale.com>
[York Sun: Updated MAINTAINERS files
Dropped #ifdef in cpu.h
Add CONFIG_SYS_NS16550=y in defconfig]
Reviewed-by: York Sun <yorksun@freescale.com>
LS2080A is a prime personality of Freescale’s LS2085A. It is a non-AIOP
personality without support of DP-DDR, L2 switch, 1588, PCIe endpoint etc.
So renaming existing LS2085A code base to reflect LS2080A (Prime personality)
Signed-off-by: Pratiyush Mohan Srivastava <pratiyush.srivastava@freescale.com>
Signed-off-by: Prabhakar Kushwaha <prabhakar@freescale.com>
[York Sun: Dropped #ifdef in cpu.c for cpu_type_list]
Reviewed-by: York Sun <yorksun@freescale.com>
MC 0.7.1.2 enforces limitation i.e.: "Packets may be corrupted
in several combinations of buffer size and frame offsets.
Workaround: Use buffers that are of size that is a multiple of 256, and
frame offset that is a multiple of 256"
Updating the DPNI Eth driver to comply with the restriction.
Signed-off-by: Bogdan Hamciuc <bogdan.hamciuc@freescale.com>
Signed-off-by: Prabhakar Kushwaha <prabhakar@freescale.com>
Reviewed-by: York Sun <yorksun@freescale.com>
Add following debug information in the driver
- Get various DPNI counter values
- Get link status of DPNI objects
- Get information of both ends of connection (DPMAC - DPNI)
Signed-off-by: Prabhakar Kushwaha <prabhakar@freescale.com>
Reviewed-by: York Sun <yorksun@freescale.com>
As per current implementation of DPAA2 ethernet driver DPNI is used as
net device. DPNI is tangible objects can be multiple connected to same physical lane.
Use DPMAC as net device where it represents physical lane.
Below modification done in driver
- Use global DPNI object
- Connect DPMAC to DPNI
- Create and destroy DPMAC
Signed-off-by: Prabhakar Kushwaha <prabhakar@freescale.com>
Reviewed-by: York Sun <yorksun@freescale.com>
Freescale's DPAA2 ethernet driver depends upon the static DPL for the
DPRC, DPNI, DPBP, DPIO objects.
Instead of static objects, Create DPNI, DPBP, DPIO objects at run-time.
Signed-off-by: Prabhakar Kushwaha <prabhakar@freescale.com>
Reviewed-by: York Sun <yorksun@freescale.com>
Management complex Firmware, DPL and DPC are depolyed during u-boot boot
sequence.
Add new DPAA2 commands to manage Management Complex (MC) i.e. start mc, aiop
and apply DPL from u-boot command prompt.
Signed-off-by: Prabhakar Kushwaha <prabhakar@freescale.com>
Reviewed-by: York Sun <yorksun@freescale.com>
dpni_create API take takes more time as comapred to existing supported
APIs of MC Flib.
So increase MC command timeout.
Signed-off-by: Prabhakar Kushwaha <prabhakar@freescale.com>
Reviewed-by: York Sun <yorksun@freescale.com>
DPMAC represents physical line on the board. This physical
line eventually asscociate with on-board PHY.
So Add an api to return linked PHY ID of DPMAC object.
Signed-off-by: Prabhakar Kushwaha <prabhakar@freescale.com>
Reviewed-by: York Sun <yorksun@freescale.com>
DPMAC object of Management complex controls Physical MAC and MDIO controller.
It provides APIs for MDIO and link state updates. It also provides APIs for
PHY/link configuration.
Signed-off-by: Prabhakar Kushwaha <prabhakar@freescale.com>
Reviewed-by: York Sun <yorksun@freescale.com>
Current Management Complex Flibs does not support APIs for adding and
destroying the objects.
Add APIs to create and destroy objects for DPBP, DPIO, DPNI and DPRC.
Signed-off-by: Prabhakar Kushwaha <prabhakar@freescale.com>
Reviewed-by: York Sun <yorksun@freescale.com>
Current implementation only consider SGMIIs for dpmac initialization.
XFI serdes protocols also uses dpmac.
Also, fix lane protocol parsing logic to consider both XFIs and SGMIIs.
Signed-off-by: Prabhakar Kushwaha <prabhakar@freescale.com>
Reviewed-by: York Sun <yorksun@freescale.com>
The console includes a global variable and several functions that are only
used by a small subset of U-Boot files. Before adding more functions, move
the definitions into their own header file.
Signed-off-by: Simon Glass <sjg@chromium.org>
Based on spec:
"MDC must not exceed 2.5 MHz (MDC is only active during MDIO read and
write operations)"
Zynq is running on 111MHz. Current setting is 32 which is 111/32=3.47
which is above of 2.5MHz.
Using 48 divider will give us correct setting according spec
(111/48=2.31).
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
Using set and clear macro is incorrect because it is not overwritting
origin mdc clock division setup.
For example origin setup is 8(0b001) and new setup is 64(0b100) which
means 0b101 is setup which is 96 divider.
Using writel to rewrite all setting like for 1000Mbit/s case.
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
Zynq has no priority queues.
ZynqMP has one priority queue and this change is required
to get ethernet working.
This patch was not needed on ep108 for uknown reason even
it should be used.
Tested on Zynq and ZynqMP.
Signed-off-by: Edgar E. Iglesias <edgar.iglesias@xilinx.com>
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Target is duplicating packets. IP prefetches another BD and process it
when the first one is sent. Adding one dummy BD to the chain fix the
problem with packet duplication.
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
BD_SEPRN_SPACE should not have hard coded value and it will be
calculated based on the number of buffer descriptors that we
would like to use.
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
Previous loop was completely bogus. Iterration should go just over
statistic counters.
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
