Testing in two circumstances:
1. back to back optical SFP+ connection between two LS1028A-QDS ports
with the SCH-26908 riser card
2. T1042 with on-board AQR115 PHY using "OCSGMII", as per
https://lore.kernel.org/lkml/aIuEvaSCIQdJWcZx@FUE-ALEWI-WINX/
strongly suggests that enabling in-band auto-negotiation is actually
possible when the lane baud rate is 3.125 Gbps.
It was previously thought that this would not be the case, because it
was only tested on 2500base-x links with on-board Aquantia PHYs, where
it was noticed that MII_LPA is always reported as zero, and it was
thought that this is because of the PCS.
Test case #1 above shows it is not, and the configured MII_ADVERTISE on
system A ends up in the MII_LPA on system B, when in 2500base-x mode
(IF_MODE=0).
Test case #2, which uses "SGMII" auto-negotiation (IF_MODE=3) for the
3.125 Gbps lane, is actually a misconfiguration, but it is what led to
the discovery.
There is actually an old bug in the Lynx PCS driver - it expects all
register values to contain their default out-of-reset values, as if the
PCS were initialized by the Reset Configuration Word (RCW) settings.
There are 2 cases in which this is problematic:
- if the bootloader (or previous kexec-enabled Linux) wrote a different
IF_MODE value
- if dynamically changing the SerDes protocol from 1000base-x to
2500base-x, e.g. by replacing the optical SFP module.
Specifically in test case #2, an accidental alignment between the
bootloader configuring the PCS to expect SGMII in-band code words, and
the AQR115 PHY actually transmitting SGMII in-band code words when
operating in the "OCSGMII" system interface protocol, led to the PCS
transmitting replicated symbols at 3.125 Gbps baud rate. This could only
have happened if the PCS saw and reacted to the SGMII code words in the
first place.
Since test #2 is invalid from a protocol perspective (there seems to be
no standard way of negotiating the data rate of 2500 Mbps with SGMII,
and the lower data rates should remain 10/100/1000), in-band auto-negotiation
for 2500base-x effectively means Clause 37 (i.e. IF_MODE=0).
Make 2500base-x be treated like 1000base-x in this regard, by removing
all prior limitations and calling lynx_pcs_config_giga().
This adds a new feature: LINK_INBAND_ENABLE and at the same time fixes
the Lynx PCS's long standing problem that the registers (specifically
IF_MODE, but others could be misconfigured as well) are not written by
the driver to the known valid values for 2500base-x.
Co-developed-by: Alexander Wilhelm <alexander.wilhelm@westermo.com>
Signed-off-by: Alexander Wilhelm <alexander.wilhelm@westermo.com>
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Link: https://patch.msgid.link/20251125103507.749654-1-vladimir.oltean@nxp.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The fbnic driver is planning to make use of the XPCS driver to enable
support for PCS and better integration with phylink. To do this though we
will need to enable several workarounds since the PMD interface for fbnic
is likely to be unique since it is a mix of two different vendor products
with a unique wrapper around the IP.
I have generated a PHY identifier based on IEEE 802.3-2022 22.2.4.3.1 using
an OUI belonging to Meta Platforms and used with our NICs. Using this we
will provide it as the PMD ID via the SW based MDIO interface so that
the fbnic device can be identified and necessary workarounds enabled in the
XPCS driver.
As an initial workaround this change adds an exception so that soft_reset
is not set when the driver is initially bound to the PCS.
In addition I have added logic to integrate the PMD Rx signal detect state
into the link state for the PCS. With this we can avoid the link coming up
too soon on the FBNIC PMD and as a result of it being in the training state
so we can avoid link flaps.
Signed-off-by: Alexander Duyck <alexanderduyck@fb.com>
Link: https://patch.msgid.link/176374321695.959489.6648161125012056619.stgit@ahduyck-xeon-server.home.arpa
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
The XPCS driver was mangling the PMA identifier as the original code
appears to have been focused on just capturing the OUI. Rather than store a
mangled ID it is better to work with the actual PMA ID and instead just
mask out the values that don't apply rather than shifting them and
reordering them as you still don't get the original OUI for the NIC without
having to bitswap the values as per the definition of the layout in IEEE
802.3-2022 22.2.4.3.1.
By laying it out as it was in the hardware it is also less likely for us to
have an unintentional collision as the enum values will occupy the revision
number area while the OUI occupies the upper 22 bits.
Signed-off-by: Alexander Duyck <alexanderduyck@fb.com>
Link: https://patch.msgid.link/176374320920.959489.17267159479370601070.stgit@ahduyck-xeon-server.home.arpa
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
With this change we are adding support for 25G, 50G, and 100G interface
types to the XPCS driver. This had supposedly been enabled with the
addition of XLGMII but I don't see any capability for configuration there
so I suspect it may need to be refactored in the future.
With this change we can enable the XPCS driver with the selected interface
and it should be able to detect link, speed, and report the link status to
the phylink interface.
Signed-off-by: Alexander Duyck <alexanderduyck@fb.com>
Link: https://patch.msgid.link/176374320248.959489.11649590675011158859.stgit@ahduyck-xeon-server.home.arpa
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Make MIIC accessory register unlock/lock behaviour selectable via SoC/OF
data. Add init_unlock_lock_regs and miic_write to struct miic_of_data so
the driver can either perform the traditional global unlock sequence (as
used on RZ/N1) or use a different policy for other SoCs (for example
RZ/T2H, which does not require leaving registers unlocked).
miic_reg_writel() now calls the per-SoC miic_write callback to perform
register writes. Provide miic_reg_writel_unlocked() as the default writer
and set it for the RZ/N1 OF data so existing platforms keep the same
behaviour. Add a miic_unlock_regs() helper that implements the accessory
register unlock sequence so the unlock/lock sequence can be reused where
needed (for example when a SoC requires explicit unlock/lock around
individual accesses).
This change is preparatory work for supporting RZ/T2H.
Signed-off-by: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>
Tested-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Link: https://patch.msgid.link/20250910204132.319975-9-prabhakar.mahadev-lad.rj@bp.renesas.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Add reset-line handling to the RZN1 MIIC driver and move reset
configuration into the SoC/OF data. Introduce MIIC_MAX_NUM_RSTS (= 2),
add storage for reset_control_bulk_data in struct miic and add
reset_ids and reset_count fields to miic_of_data.
When reset_ids are present in the OF data, the driver obtains the reset
lines with devm_reset_control_bulk_get_exclusive(), deasserts them during
probe and registers a devres action to assert them on remove or on error.
This change is preparatory work to support the RZ/T2H SoC, which exposes
two reset lines for the ETHSS IP. The driver remains backward compatible
for platforms that do not provide reset lines.
Signed-off-by: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Tested-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Link: https://patch.msgid.link/20250910204132.319975-8-prabhakar.mahadev-lad.rj@bp.renesas.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Define per-SoC miic_port_start and miic_port_max fields in struct
miic_of_data and use them to validate the device-tree "reg" port number
and to compute the driver's internal zero-based port index as
(port - miic_port_start). Replace uses of the hard-coded MIIC_MAX_NR_PORTS
with the SoC-provided miic_port_max when iterating over ports.
On RZ/N1 the MIIC ports are numbered 1..5, whereas RZ/T2H numbers its MIIC
ports 0..3. By making the port base and range part of the OF data the
driver no longer assumes a fixed numbering scheme and can support SoCs that
enumerate ports from either zero or one and that expose different numbers
of ports.
This change is preparatory work for adding RZ/T2H support.
Signed-off-by: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Tested-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Link: https://patch.msgid.link/20250910204132.319975-6-prabhakar.mahadev-lad.rj@bp.renesas.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Move configuration data such as the modctrl matching table, converter
count, and string lookup tables into the SoC-specific miic_of_data
structure. Update the helper functions to use the per-SoC configuration
instead of relying on fixed-size arrays or global tables, and allocate
DT configuration memory dynamically.
This refactoring keeps the existing RZ/N1 support intact while preparing
the driver to handle the different configuration requirements of the
RZ/T2H SoC.
Signed-off-by: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>
Tested-by: Wolfram Sang <wsa+renesas@sang-engineering.com>
Link: https://patch.msgid.link/20250910204132.319975-5-prabhakar.mahadev-lad.rj@bp.renesas.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
The xpcs_switch_interface_mode function was introduced to handle
interface switching.
According to the XPCS datasheet, a soft reset is required to initiate
Clause 37 auto-negotiation when the XPCS switches interface modes.
When the interface mode switches from 2500BASE-X to SGMII,
re-initiating Clause 37 auto-negotiation is required for the SGMII
interface mode to function properly.
Signed-off-by: Choong Yong Liang <yong.liang.choong@linux.intel.com>
Link: https://patch.msgid.link/20250227121522.1802832-3-yong.liang.choong@linux.intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
As all PCS are using the neg_mode parameter rather than the legacy
an_mode, remove the ability to use the legacy an_mode. We remove the
tests in the phylink code, unconditionally passing the PCS neg_mode
parameter to PCS methods, and remove setting the flag from drivers.
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Link: https://patch.msgid.link/E1tidPn-0040hd-2R@rmk-PC.armlinux.org.uk
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
There is now no need to pass the mult_fact into xpcs_config_eee(), so
let's remove that argument and use xpcs->eee_mult_fact directly. While
changing the function signature, as we pass true/false for enable, use
"bool" instead of "int" for this.
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Link: https://patch.msgid.link/E1thRQY-003w7U-IG@rmk-PC.armlinux.org.uk
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Add a function to separate out the EEE clock multiplying factor. This
will be called by the stmmac driver to configure this value.
It would have been better had the driver used the CLK API to retrieve
this clock, get its rate and calculate the appropriate multiplier, but
that door has closed.
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Link: https://patch.msgid.link/E1thRQ8-003w70-VT@rmk-PC.armlinux.org.uk
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
On a port with SGMII fixed-link at SPEED_1000, DW_VR_MII_DIG_CTRL1 gets
set to 0x2404. This is incorrect, because bit 2 (DW_VR_MII_DIG_CTRL1_2G5_EN)
is set.
It comes from the previous write to DW_VR_MII_AN_CTRL, because the "val"
variable is reused and is dirty. Actually, its value is 0x4, aka
FIELD_PREP(DW_VR_MII_PCS_MODE_MASK, DW_VR_MII_PCS_MODE_C37_SGMII).
Resolve the issue by clearing "val" to 0 when writing to a new register.
After the fix, the register value is 0x2400.
Prior to the blamed commit, when the read-modify-write was open-coded,
the code saved the content of the DW_VR_MII_DIG_CTRL1 register in the
"ret" variable.
Fixes: ce8d6081fc ("net: pcs: xpcs: add _modify() accessors")
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Maxime Chevallier <maxime.chevallier@bootlin.com>
Reviewed-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Link: https://patch.msgid.link/20250114164721.2879380-1-vladimir.oltean@nxp.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
While using "return" when calling a void returning function inside a
function that returns void doesn't cause a compiler warning, it looks
weird. Convert the bunch of if() statements to a switch() and remove
these return statements.
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Tested-by: Serge Semin <fancer.lancer@gmail.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
xpcs_config_usxgmii() is only called from the xpcs_link_up() method, so
let's name it similarly to the SGMII and 1000BASEX functions.
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Tested-by: Serge Semin <fancer.lancer@gmail.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
xpcs_link_up_sgmii() and xpcs_link_up_1000basex() are almost identical
with the exception of checking the speed and duplex for 1000BASE-X.
Combine the two functions.
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Tested-by: Serge Semin <fancer.lancer@gmail.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
We can now see that we have an open-coded version of
mii_bmcr_encode_fixed() when this is called with SPEED_1000:
val = BMCR_SPEED1000;
if (duplex == DUPLEX_FULL)
val |= BMCR_FULLDPLX;
Replace this with a call to mii_bmcr_encode_fixed().
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Tested-by: Serge Semin <fancer.lancer@gmail.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Remove an unnecessary switch() statement in xpcs_link_up_1000basex().
The only value this switch statement is interested in is SPEED_1000,
all other values lead to an error. Replace this with a simple if()
statement.
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Tested-by: Serge Semin <fancer.lancer@gmail.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
As a general policy, we refer our generic register definitions over
vendor specific definitions. In XPCS, it appears that the register
layout follows a BMCR, BMSR and ADVERTISE register definition. We
already refer to this BMCR register using several different macros
which is confusing.
Convert the following register definitions to generic versions:
DW_VR_MII_MMD_CTRL => MII_BMCR
MDIO_CTRL1 => MII_BMCR
AN_CL37_EN => BMCR_ANENABLE
SGMII_SPEED_SS6 => BMCR_SPEED1000
SGMII_SPEED_SS13 => BMCR_SPEED100
MDIO_CTRL1_RESET => BMCR_RESET
DW_VR_MII_MMD_STS => MII_BMSR
DW_VR_MII_MMD_STS_LINK_STS => BMSR_LSTATUS
DW_FULL_DUPLEX => ADVERTISE_1000XFULL
iDW_HALF_DUPLEX => ADVERTISE_1000XHALF
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Tested-by: Serge Semin <fancer.lancer@gmail.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
According to commits 2a22b7ae2f ("net: pcs: xpcs: adapt Wangxun NICs
for SGMII mode") and 2deea43f38 ("net: pcs: xpcs: add 1000BASE-X AN
interrupt support"), Wangxun devices need special VR_XS_PCS_DIG_CTRL1
settings for SGMII and 1000BASE-X. Both SGMII and 1000BASE-X use the
same settings.
Rather than placing these in the individual xpcs_config_*() functions,
move it to where we already test for the Wangxun devices in
xpcs_do_config().
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
