phy: qcom-qmp-combo: restructure PHY creation

In preparation for supporting devicetree bindings which do not use child
nodes, move the PHY creation to probe() proper and parse the serdes,
dp_com and dp_serdes resources in a dedicated legacy devicetree helper.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221121085058.31213-5-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-combo: drop v4 reference-clock source

The source clock for the reference clock should not be described by the
devicetree and instead this relationship should be modelled in the clock
driver.

Drop the management of the source clock from the driver for SC8180X and
SC8280XP. Note that support for the former is not yet in mainline.

Also note that the binding has never been updated to describe the v4
clocks for SC8180X.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221121085058.31213-4-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

dt-bindings: phy: qcom,qmp-usb3-dp: fix sc8280xp binding

The current QMP USB3-DP PHY bindings are based on the original MSM8996
binding which provided multiple PHYs per IP block and these in turn were
described by child nodes.

The QMP USB3-DP PHY block provides a single multi-protocol PHY and even
if some resources are only used by either the USB or DP part of the
device there is no real benefit in describing these resources in child
nodes.

The original MSM8996 binding also ended up describing the individual
register blocks as belonging to either the wrapper node or the PHY child
nodes.

This is an unnecessary level of detail which has lead to problems when
later IP blocks using different register layouts have been forced to fit
the original mould rather than updating the binding. The bindings are
arguable also incomplete as they only the describe register blocks used
by the current Linux drivers (e.g. does not include the PCS LANE
registers).

This is specifically true for later USB4-USB3-DP QMP PHYs where the TX
registers are used by both the USB3 and DP parts of the PHY (and where
the USB4 part of the PHY was not covered by the binding at all). Notably
there are also no DP "RX" (sic) registers as described by the current
bindings and the DP "PCS" region is really a set of DP_PHY registers.

Add a new binding for the USB4-USB3-DP QMP PHYs found on SC8280XP which
further bindings can be based on.

Note that the binding uses a PHY index to access either the USB3 or DP
part of the PHY and that this can later be used also for the USB4 part
if needed.

Similarly, the clock inputs and outputs can later be extended to support
USB4.

Also note that the current binding is simply removed instead of being
deprecated as it was only recently merged and would not allow for
supporting DP mode.

Reviewed-by: Rob Herring <robh@kernel.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221121085058.31213-3-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

dt-bindings: phy: qcom,qmp-usb3-dp: rename current bindings

The current QMP USB3-DP PHY bindings are based on the original MSM8996
binding which provided multiple PHYs per IP block and these in turn were
described by child nodes.

The QMP USB3-DP PHY block provides a single multi-protocol PHY and
even if some resources are only used by either the USB or DP part of the
device there is no real benefit in describing these resources in child
nodes.

The original MSM8996 binding also ended up describing the individual
register blocks as belonging to either the wrapper node or the PHY child
nodes.

This is an unnecessary level of detail which has lead to problems when
later IP blocks using different register layouts have been forced to fit
the original mould rather than updating the binding. The bindings are
arguable also incomplete as they only the describe register blocks used
by the current Linux drivers (e.g. does not include the PCS_LANE
registers).

In preparation for adding new bindings for SC8280XP which further
bindings can be based on, rename the current schema file after SC7180,
which was the first supported platform, and add a reference to the
SC8280XP bindings.

Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221121085058.31213-2-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-combo: clean up DP callback names

Clean up and unify the DP callbacks by dropping the redundant "qcom" and
"phy" prefix and infix and by using a common naming scheme ("qmp" +
version + callback name).

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221114110621.4639-23-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-combo: clean up probe initialisation

Stop abusing the driver data pointer and instead pass the driver state
structure directly to the initialisation helpers during probe.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221114110621.4639-22-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-combo: clean up device-tree parsing

Since the QMP driver split there will be precisely two child nodes so
drop the obsolete iteration construct.

While at it, drop the verbose error logging that would have been printed
also on probe deferrals.

Note that there is no need to check if there are additional child nodes
(the kernel is not a devicetree validator), but let's return an error if
either child node is missing.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221114110621.4639-21-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-combo: merge driver data

The QMP combo driver manages a single PHY (even if it provides two
interfaces for USB and DP, respectively) so merge the old qcom_qmp and
qmp_phy structures and drop the PHY array.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221114110621.4639-20-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-combo: merge USB and DP configurations

It does not really make any sense to keep separate configuration
structures for the USB and DP parts of the same PHY so merge them.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221114110621.4639-19-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-combo: drop lanes config parameter

Since the QMP driver split there is really no need for the 'lanes'
configuration parameter as all of these USB-C PHYs support dual-lane
SuperSpeed USB and quad-lane (uni-directional) DP (even if the driver
still only supports CC1 orientation using lanes 2 and 3).

Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Link: https://lore.kernel.org/r/20221114110621.4639-18-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-combo: add DP configuration tables

In preparation for merging the USB and DP configurations, add dedicated
pointers for the DP serdes and tx tables to the configurations.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221114110621.4639-17-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-combo: rename sc8280xp config

In preparation for merging the USB and DP configurations, drop the
"combo" infix from the SC8280XP combined configuration for consistency
with the other platforms.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221114110621.4639-16-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-combo: clean up DP configurations

In preparation for merging the USB and DP configurations, align the
initialisations of the DP function pointers.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221114110621.4639-15-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-combo: add dedicated DP iomem pointers

In preparation for merging the driver data, add separate iomem pointers
for the DP part of the PHY.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221114110621.4639-14-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-combo: separate USB and DP devicetree parsing

Separate the devicetree parsing of the USB and DP child nodes in two
dedicated helpers in preparation for merging the driver data.

Note that only the USB part of the PHY has a pipe clock and that the DP
implementation only uses the tx/tx2 and pcs register regions.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221114110621.4639-13-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-combo: clean up serdes initialisation

Clean up serdes initialisation somewhat by making the current helper a
dedicated helper for the DP part of the PHY.

Note that no error is currently returned for non-supported link rates.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221114110621.4639-12-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-combo: separate USB and DP power-on ops

Separate the USB and DP power-on and power-off operations in two
dedicated implementations.

Note that the pipe clock is only used by the USB part of the PHY and
that no DP configuration has a pcs (or rx) table or has has_pwrdn_delay
set.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221114110621.4639-11-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-combo: rename DP PHY ops

Rename the configure and calibrate DP PHY ops using the common prefix
for consistency.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221114110621.4639-10-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-combo: separate USB and DP init ops

Separate the USB and DP init and exit operations by calling the common
initialisation code directly from the USB operation and adding a "dp"
infix to the DP callbacks.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221114110621.4639-9-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-combo: drop unnecessary debug message

Drop the unnecessary (verbose) debug message from the init PHY op.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221114110621.4639-8-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-combo: rename USB PHY ops

Add a "usb" infix to the USB PHY operation functions and name them
after the corresponding operations (e.g. "init" rather than "enable").

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221114110621.4639-7-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-combo: drop unused DP PHY mode op

The set-mode operation is currently only used by the USB part of the
PHY so drop the corresponding callback from the DP PHY ops.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221114110621.4639-6-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-combo: rename PHY ops structures

Rename the PHY operation structures so that they have a "phy_ops" suffix
and move them next to the implementation.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221114110621.4639-5-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-combo: move pm ops

Move the PM ops structure next to the implementation to keep the driver
callbacks grouped.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221114110621.4639-4-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-combo: move device-id table

Move the device-id table after probe() and next to the driver structure
to keep the driver callback functions grouped together.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221114110621.4639-3-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-combo: sort device-id table

Sort the device-id table by compatible string to make it easier to find
and add new entries.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221114110621.4639-2-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-combo: clean up common initialisation

Commit 52e013d0bffa ("phy: qcom-qmp: Add support for DP in USB3+DP combo
phy") added support for the DisplayPort part of QMP PHYs but
unfortunately did so by duplicating parts of the shared configuration,
something which has lead to subtle bugs depending on probe order.

As the resources have always been requested based on the USB
configuration, make sure to not rely on fields from the DP configuration
when using them (e.g. in case they get out of sync) and remove the now
unused fields from the DP configurations.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221114081346.5116-7-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-combo: fix runtime suspend

Drop the confused runtime-suspend type check which effectively broke
runtime PM if the DP child node happens to be parsed before the USB
child node during probe (e.g. due to order of child nodes in the
devicetree).

Instead use the new driver data USB PHY pointer to access the USB
configuration and resources.

Fixes: 52e013d0bffa ("phy: qcom-qmp: Add support for DP in USB3+DP combo phy")
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221114081346.5116-6-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-combo: fix broken power on

The PHY is powered on during phy-init by setting the SW_PWRDN bit in the
COM_POWER_DOWN_CTRL register and then setting the same bit in the in the
PCS_POWER_DOWN_CONTROL register that belongs to the USB part of the
PHY.

Currently, whether power on succeeds depends on probe order and having
the USB part of the PHY be initialised first. In case the DP part of the
PHY is instead initialised first, the intended power on of the USB block
results in a corrupted DP_PHY register (e.g. DP_PHY_AUX_CFG8).

Add a pointer to the USB part of the PHY to the driver data and use that
to power on the PHY also if the DP part of the PHY is initialised first.

Fixes: 52e013d0bffa ("phy: qcom-qmp: Add support for DP in USB3+DP combo phy")
Cc: stable@vger.kernel.org # 5.10
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221114081346.5116-5-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-combo: fix sc8180x reset

The SC8180X has two resets but the DP configuration erroneously
described only one.

In case the DP part of the PHY is initialised before the USB part (e.g.
depending on probe order), then only the first reset would be asserted.

Fixes: 1633802cd4ac ("phy: qcom: qmp: Add SC8180x USB/DP combo")
Cc: stable@vger.kernel.org # 5.15
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221114081346.5116-4-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-combo: fix sdm845 reset

The SDM845 has two resets but the DP configuration erroneously described
only one.

In case the DP part of the PHY is initialised before the USB part (e.g.
depending on probe order), then only the first reset would be asserted.

Add a dedicated configuration for SDM845 rather than reuse the
incompatible SC7180 configuration.

Fixes: d88497fb6bbd ("phy: qualcomm: phy-qcom-qmp: add support for combo USB3+DP phy on SDM845")
Cc: stable@vger.kernel.org # 6.1
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221114081346.5116-3-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-combo: fix out-of-bounds clock access

The SM8250 only uses three clocks but the DP configuration erroneously
described four clocks.

In case the DP part of the PHY is initialised before the USB part, this
would lead to uninitialised memory beyond the bulk-clocks array to be
treated as a clock pointer as the clocks are requested based on the USB
configuration.

Fixes: aff188feb5e1 ("phy: qcom-qmp: add support for sm8250-usb3-dp phy")
Cc: stable@vger.kernel.org # 5.13
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221114081346.5116-2-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: renesas: Add Renesas Ethernet SERDES driver for R-Car S4-8

Add Renesas Ethernet SERDES driver for R-Car S4-8 (r8a779f0).
The datasheet describes initialization procedure without any information
about registers' name/bits. So, this is all black magic to initialize
the hardware. Especially, all channels should be initialized at once.

Signed-off-by: Yoshihiro Shimoda <yoshihiro.shimoda.uh@renesas.com>
Link: https://lore.kernel.org/r/20221108005500.3011449-3-yoshihiro.shimoda.uh@renesas.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>

dt-bindings: phy: renesas: Document Renesas Ethernet SERDES

Document Renesas Etherent SERDES for R-Car S4-8 (r8a779f0).

Signed-off-by: Yoshihiro Shimoda <yoshihiro.shimoda.uh@renesas.com>
Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Link: https://lore.kernel.org/r/20221108005500.3011449-2-yoshihiro.shimoda.uh@renesas.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-pcie: Fix sm8450_qmp_gen4x2_pcie_pcs_tbl[] register names

sm8450_qmp_gen4x2_pcie_pcs_tbl[] contains the init sequence for PCS
registers of QMP PHY v5.20. So use the v5.20 specific register names.
Only major change is the rename of PCS_EQ_CONFIG{2/3} registers to
PCS_EQ_CONFIG{4/5}.

Fixes: 2c91bf6bf290 ("phy: qcom-qmp: Add SM8450 PCIe1 PHY support")
Signed-off-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
Link: https://lore.kernel.org/r/20221102081835.41892-2-manivannan.sadhasivam@linaro.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-pcie: Fix high latency with 4x2 PHY when ASPM is enabled

The PCIe QMP 4x2 RC PHY generates high latency when ASPM is enabled. This
seem to be fixed by clearing the QPHY_V5_20_PCS_PCIE_PRESET_P10_POST
register of the pcs_misc register space.

Fixes: 2c91bf6bf290 ("phy: qcom-qmp: Add SM8450 PCIe1 PHY support")
Signed-off-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
Link: https://lore.kernel.org/r/20221102081835.41892-1-manivannan.sadhasivam@linaro.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-pcie: split pcs_misc init cfg for ipq8074 pcs table

Commit af6643242d3a ("phy: qcom-qmp-pcie: split pcs_misc region for ipq6018
pcie gen3") reworked the pcs regs values and removed the 0x400 offset
for each pcs_misc regs.

This change caused the malfunction of ipq8074 downstream since it still
has the legacy pcs table where pcs_misc are not placed on a different
table and instead put together assuming the offset of 0x400 for the
related pcs_misc regs.

Split pcs_misc init cfg from the ipq8074 pcs init table to be handled
correctly to prepare for actual support for gen3 pcie for ipq8074.

Fixes: af6643242d3a ("phy: qcom-qmp-pcie: split pcs_misc region for ipq6018 pcie gen3")
Reported-by: Robert Marko <robimarko@gmail.com>
Tested-by: Robert Marko <robimarko@gmail.com>
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
Link: https://lore.kernel.org/r/20221103212125.17156-1-ansuelsmth@gmail.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-pcie: add support for sc8280xp 4-lane PHYs

The PCIe2 and PCIe3 controllers and PHYs on SC8280XP can be used in
4-lane mode or as separate controllers and PHYs in 2-lane mode (e.g. as
PCIe2A and PCIe2B).

Add support for fetching the 4-lane configuration from the TCSR and
programming the lane registers of the second port when in 4-lane mode.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221105145939.20318-17-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-pcie: add support for sc8280xp

Add support for the single and dual-lane PHYs found on SC8280XP.

Note that the SC8280XP binding does not try to describe every register
subregion and instead the driver holds the corresponding offsets.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221105145939.20318-16-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-pcie: add support for pipediv2 clock

Some QMP PHYs have a second fixed-divider pipe clock that needs to be
enabled along with the pipe clock.

Add support for an optional "pipediv2" clock.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221105145939.20318-15-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-pcie: fix initialisation reset

Add the missing delay after asserting reset. This is specifically needed
for the reset to have any effect on SC8280XP.

The vendor driver uses a 1 ms delay, but that seems a bit excessive.
Instead use a 200 us delay which appears to be more than enough and also
matches the UFS reset delay added by commit 870b1279c7a0 ("scsi:
ufs-qcom: Add reset control support for host controller").

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221105145939.20318-14-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-pcie: restructure PHY creation

In preparation for supporting devicetree bindings which do not use a
child node, move the PHY creation to probe() proper and parse the serdes
resource in what is now the legacy devicetree helper.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221105145939.20318-13-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

dt-bindings: phy: qcom,qmp-pcie: add sc8280xp bindings

Add bindings for the PCIe QMP PHYs found on SC8280XP.

The PCIe2 and PCIe3 controllers and PHYs on SC8280XP can be used in
4-lane mode or as separate controllers and PHYs in 2-lane mode (e.g. as
PCIe2A and PCIe2B).

The configuration for a specific system can be read from a TCSR register.

Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221105145939.20318-12-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

dt-bindings: phy: qcom,qmp-pcie: rename current bindings

The current QMP PCIe PHY bindings are based on the original MSM8996
binding which provided multiple PHYs per IP block and these in turn were
described by child nodes.

Later QMP PCIe PHY blocks only provide a single PHY and the remnant
child node does not really reflect the hardware.

The original MSM8996 binding also ended up describing the individual
register blocks as belonging to either the wrapper node or the PHY child
nodes.

This is an unnecessary level of detail which has lead to problems when
later IP blocks using different register layouts have been forced to fit
the original mould rather than updating the binding. The bindings are
arguable also incomplete as they only the describe register blocks used
by the current Linux drivers (e.g. does not include the per lane PCS
registers).

In preparation for adding new bindings for SC8280XP which further
bindings can be based on, rename the current schema file after IPQ8074,
which was the first SoC added to the bindings after MSM8996 (which has
already been split out), and add a reference to the SC8280XP bindings.

Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221105145939.20318-11-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-pcie: add register init helper

Generalise the serdes initialisation helper so that it can be used to
initialise all the PHY registers (e.g. serdes, tx, rx, pcs).

Note that this defers the ungating of the PIPE clock somewhat, which is
fine as it isn't needed until starting the PHY.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221105145939.20318-10-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-pcie: use shorter tables identifiers

The QMP drivers all use 'tbl' to refer to their register initialisation
tables.

For consistency use 'tbls' rather than 'tables' to refer to the new
aggregate table structures.

Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Link: https://lore.kernel.org/r/20221105145939.20318-9-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-pcie: clean up PHY lane init

Clean up the PHY lane initialisation somewhat by adding further
temporary variables and programming both tx and rx for the second lane
after the first lane.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221105145939.20318-8-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-pcie: rename PHY ops structure

Rename the PHY operation structure so that it has a "phy_ops" suffix and
move it next to the implementation.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221105145939.20318-7-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-pcie: clean up probe initialisation

Stop abusing the driver data pointer and instead pass the driver state
structure directly to the initialisation helpers during probe.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221105145939.20318-6-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-pcie: clean up device-tree parsing

Since the QMP driver split there will be at most a single child node so
drop the obsolete iteration construct.

While at it, drop the verbose error logging that would have been
printed also on probe deferrals.

Note that there's no need to check if there are additional child nodes
(the kernel is not a devicetree validator), but let's return an error if
there are no child nodes at all for now.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221105145939.20318-5-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-pcie: merge driver data

The PCIe QMP PHY driver only manages a single PHY so merge the old
qcom_qmp and qmp_phy structures and drop the PHY array.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221105145939.20318-4-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-pcie: move device-id table

Move the device-id table below probe() and next to the driver structure
to keep the driver callback functions grouped together.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221105145939.20318-3-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-pcie: sort device-id table

Sort the device-id table by compatible string to make it easier to find
and add new entries.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221105145939.20318-2-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

doc: phy: Document typical order of API calls

Document the typical order of API calls to used by new drivers and
controllers. Many existing controllers follow this order, but some do
not. This is especially true for controllers designed to work with one
particular PHY driver, which may not need a call to (for example)
phy_init.

Signed-off-by: Sean Anderson <sean.anderson@seco.com>
Link: https://lore.kernel.org/r/20221018175841.1906611-1-sean.anderson@seco.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: tegra: xusb: Remove usb3 supply

Remove redundant codes for getting the vbus supply of usb3 ports because
we get and control the vbus supply by the companion usb2 ports

Signed-off-by: Wayne Chang <waynec@nvidia.com>
Signed-off-by: Haotien Hsu <haotienh@nvidia.com>
Reviewed-by: Jon Hunter <jonathanh@nvidia.com>
Link: https://lore.kernel.org/r/20221005084031.2154251-1-haotienh@nvidia.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: phy-brcm-usb: Utilize platform_get_irq_byname_optional()

The wake-up interrupt lines are entirely optional, avoid printing
messages that interrupts were not found by switching to the _optional
variant.

Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Acked-by: Justin Chen <justinpopo6@gmail.com>
Link: https://lore.kernel.org/r/20221026224450.2958762-1-f.fainelli@gmail.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: ti: gmii-sel: Add support for CPSW9G GMII SEL in J721e

Each of the CPSW9G ports in J721e support additional modes like QSGMII.
Add a new compatible for J721e to support the additional modes.

In TI's J721e, each of the CPSW9G ethernet interfaces can act as a
QSGMII main or QSGMII-SUB port. The QSGMII main interface is responsible
for performing auto-negotiation between the MAC and the PHY while the rest
of the interfaces are designated as QSGMII-SUB interfaces, indicating that
they will not be taking part in the auto-negotiation process.

Signed-off-by: Siddharth Vadapalli <s-vadapalli@ti.com>
Reviewed-by: Roger Quadros <rogerq@kernel.org>
Link: https://lore.kernel.org/r/20221026074532.109220-4-s-vadapalli@ti.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: ti: gmii-sel: Update methods for fetching and using qsgmii main port

The number of QSGMII main ports are specific to the device. TI's J7200 for
which the QSGMII main port property is fetched from the device-tree has
only one QSGMII main port. However, devices like TI's J721e support up to
two QSGMII main ports. Thus, the existing methods for fetching and using
the QSGMII main port are not scalable.

Update the existing methods for handling the QSGMII main ports and its
associated requirements to make it scalable for future devices.

Signed-off-by: Siddharth Vadapalli <s-vadapalli@ti.com>
Reviewed-by: Roger Quadros <rogerq@kernel.org>
Link: https://lore.kernel.org/r/20221026074532.109220-3-s-vadapalli@ti.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>

dt-bindings: phy: ti: phy-gmii-sel: Add bindings for J721e

TI's J721e SoC supports additional PHY modes like QSGMII and SGMII
that are not supported on earlier SoCs. Add a compatible for it.

Extend ti,qsgmii-main-ports property to support selection of upto
two main ports at once across the two QSGMII interfaces.

Signed-off-by: Siddharth Vadapalli <s-vadapalli@ti.com>
Reviewed-by: Rob Herring <robh@kernel.org>
Link: https://lore.kernel.org/r/20221026074532.109220-2-s-vadapalli@ti.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: sun4i-usb: Add support for the H616 USB PHY

The USB PHY used in the Allwinner H616 SoC inherits some traits from its
various predecessors: it has four full PHYs like the H3, needs some
extra bits to be set like the H6, and puts SIDDQ on a different bit like
the A100. Plus it needs this weird PHY2 quirk.

Name all those properties in a new config struct and assign a new
compatible name to it.

Signed-off-by: Andre Przywara <andre.przywara@arm.com>
Reviewed-by: Samuel Holland <samuel@sholland.org>
Link: https://lore.kernel.org/r/20221031111358.3387297-5-andre.przywara@arm.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: sun4i-usb: Introduce port2 SIDDQ quirk

At least the Allwinner H616 SoC requires a weird quirk to make most
USB PHYs work: Only port2 works out of the box, but all other ports
need some help from this port2 to work correctly: The CLK_BUS_PHY2 and
RST_USB_PHY2 clock and reset need to be enabled, and the SIDDQ bit in
the PMU PHY control register needs to be cleared. For this register to
be accessible, CLK_BUS_ECHI2 needs to be ungated. Don't ask ....

Instead of disguising this as some generic feature, treat it more like
a quirk (what it really is):
If the quirk bit is set, and we initialise a PHY other than PHY2, ungate
this one special clock, and clear the SIDDQ bit. We also pick the clock
and reset from PHY2 and enable them as well.

Signed-off-by: Andre Przywara <andre.przywara@arm.com>
Link: https://lore.kernel.org/r/20221031111358.3387297-4-andre.przywara@arm.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>

dt-bindings: phy: Add special clock for Allwinner H616 PHY

The USB PHY IP in the Allwinner H616 SoC requires a quirk that involves
some resources from port 2's PHY and HCI IP. In particular the PMU clock
for port 2 must be surely ungated before accessing the REG_HCI_PHY_CTL
register of port 2. To allow each USB port to be controlled
independently of port 2, we need a handle to that particular PMU clock
in the *PHY* node, as the HCI and PHY part might be handled by separate
drivers.

Add that clock to the requirements of the H616 PHY binding, so that a
PHY driver can apply the quirk in isolation, without requiring help from
port 2's HCI driver.

Signed-off-by: Andre Przywara <andre.przywara@arm.com>
Reviewed-by: Rob Herring <robh@kernel.org>
Link: https://lore.kernel.org/r/20221031111358.3387297-3-andre.przywara@arm.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: usb: Fix clock imbalance for suspend/resume

We should be disabling clocks when wake from USB is not needed. Since
this wasn't done, we had a clock imbalance since clocks were always
being enabled on resume.

Fixes: ae532b2b7aa5 ("phy: usb: Add "wake on" functionality for newer Synopsis XHCI controllers")
Fixes: b0c0b66c0b43 ("phy: usb: Add support for wake and USB low power mode for 7211 S2/S5")
Signed-off-by: Justin Chen <justinpopo6@gmail.com>
Acked-by: Florian Fainelli <f.fainelli@gmail.com>
Link: https://lore.kernel.org/r/1665005418-15807-7-git-send-email-justinpopo6@gmail.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: usb: Use slow clock for wake enabled suspend

The logic was incorrect when switching to slow clock. We want the slow
clock if wake_enabled is set.

Fixes: ae532b2b7aa5 ("phy: usb: Add "wake on" functionality for newer Synopsis XHCI controllers")
Signed-off-by: Justin Chen <justinpopo6@gmail.com>
Acked-by: Florian Fainelli <f.fainelli@gmail.com>
Link: https://lore.kernel.org/r/1665005418-15807-6-git-send-email-justinpopo6@gmail.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: usb: Disable phy auto-suspend

The BDC block requires the PLL lock in order to grab the PLL clock.
The phy auto-suspend feature turns off the phy when nothing is attached
leading to the PLL to not lock. This leads the BDC block to grab the AUX
clock instead of the PLL clock. This is not ideal, so lets turn this
feature off.

Signed-off-by: Justin Chen <justinpopo6@gmail.com>
Acked-by: Florian Fainelli <f.fainelli@gmail.com>
Link: https://lore.kernel.org/r/1665005418-15807-5-git-send-email-justinpopo6@gmail.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: usb: Migrate to BIT and BITMASK macros

Using BIT and BITMASK macros makes it much easier to read and make
modifications. Also reordered some constants to be in numerical order.

Signed-off-by: Justin Chen <justinpopo6@gmail.com>
Acked-by: Florian Fainelli <f.fainelli@gmail.com>
Link: https://lore.kernel.org/r/1665005418-15807-4-git-send-email-justinpopo6@gmail.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: usb: s2 WoL wakeup_count not incremented for USB->Eth devices

The PHY's "wakeup_count" is not incrementing when waking from
WoL. The wakeup count can be found in sysfs at:
/sys/bus/platform/devices/rdb/*.usb-phy/power/wakeup_count.
The problem is that the system wakup event handler was being passed
the wrong "device" by the PHY driver.

Fixes: f1c0db40a3ad ("phy: usb: Add "wake on" functionality")
Signed-off-by: Al Cooper <alcooperx@gmail.com>
Signed-off-by: Justin Chen <justinpopo6@gmail.com>
Acked-by: Florian Fainelli <f.fainelli@gmail.com>
Link: https://lore.kernel.org/r/1665005418-15807-3-git-send-email-justinpopo6@gmail.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: usb: Improve port mode selection

Split port modes into two different variables. Supported port modes
is what the hardware supports. While port mode is how the hardware
is currently configured and can be dynamically changed through the
sysfs. We initialize all supported port modes on init even though
the port mode may not be selected because we cannot guarantee the
downstream interface from the phy will be active or not.

This also fixes an issue where port modes selected via sysfs were
not being saved through suspend/resume.

Signed-off-by: Justin Chen <justinpopo6@gmail.com>
Acked-by: Florian Fainelli <f.fainelli@gmail.com>
Link: https://lore.kernel.org/r/1665005418-15807-2-git-send-email-justinpopo6@gmail.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-usb: add support for updated sc8280xp binding

Add support for the new SC8280XP binding.

Note that the binding does not try to describe every register subregion
and instead the driver holds the corresponding offsets. This includes
the PCS_USB region which was initially overlooked.

Note that the driver will no longer accept the old binding due to the
fixed "phy_phy" reset name.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221028160435.26948-14-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-usb: restructure PHY creation

In preparation for supporting devicetree bindings which do not use a
child node, move the PHY creation to probe() proper and parse the serdes
and dp_com resources in what is now the legacy devicetree helper.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221028160435.26948-13-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

dt-bindings: phy: qcom,qmp-usb: fix sc8280xp binding

The current QMP USB PHY bindings are based on the original MSM8996 PCIe
PHY binding which provided multiple PHYs per IP block and these in turn
were described by child nodes.

The QMP USB PHY block only provide a single PHY and the remnant child
node does not really reflect the hardware.

The original MSM8996 binding also ended up describing the individual
register blocks as belonging to either the wrapper node or the PHY child
nodes.

This is an unnecessary level of detail which has lead to problems when
later IP blocks using different register layouts have been forced to fit
the original mould rather than updating the binding. The bindings are
arguable also incomplete as they only the describe register blocks used
by the current Linux drivers (e.g. does not include the per lane PCS
registers).

Note that PCS_USB region is also not described by the current bindings
despite being used by the driver and this has led to people increasing
the size of the PCS region in the devicetree so that it includes PCS_USB
registers even though other regions like TX and RX may lie in between.

Add a new binding for the QMP USB PHYs found on SC8280XP which further
bindings can be based on.

Note that this also fixes the SC8280XP "phy_phy" reset name.

Also note that the current binding is simply removed instead of being
deprecated as it was only recently merged and support for SC8280XP is
still under development. And, specifically, there is no support in
mainline for the multiport controller that uses these PHYs.

Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Reviewed-by: Rob Herring <robh@kernel.org>
Link: https://lore.kernel.org/r/20221028160435.26948-12-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

dt-bindings: phy: qcom,qmp-usb: rename current bindings

The current QMP USB PHY bindings are based on the original MSM8996
binding which provided multiple PHYs per IP block and these in turn were
described by child nodes.

Later QMP USB PHY blocks only provide a single PHY and the remnant child
node does not really reflect the hardware.

The original MSM8996 binding also ended up describing the individual
register blocks as belonging to either the wrapper node or the PHY child
nodes.

This is an unnecessary level of detail which has lead to problems when
later IP blocks using different register layouts have been forced to fit
the original mould rather than updating the binding. The bindings are
arguable also incomplete as they only the describe register blocks used
by the current Linux drivers (e.g. does not include the per lane PCS
registers).

In preparation for adding new bindings for SC8280XP which further
bindings can be based on, rename the current bindings after MSM8996 and
add a reference to the SC8280XP bindings.

Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221028160435.26948-11-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-usb: clean up PHY init

Clean up the PHY initialisation somewhat by programming both tx and rx
for the second lane after the first lane.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221028160435.26948-10-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-usb: rename PHY ops structure

Rename the PHY operation structure so that it has a "phy_ops" suffix and
move it next to the implementation.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221028160435.26948-9-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-usb: clean up probe initialisation

Stop abusing the driver data pointer and instead pass the driver state
structure directly to the initialisation helpers during probe.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221028160435.26948-8-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-usb: clean up device-tree parsing

Since the QMP driver split there will be at most a single child node so
drop the obsolete iteration construct.

While at it, drop the verbose error logging that would have been
printed also on probe deferrals.

Note that there's no need to check if there are additional child nodes
(the kernel is not a devicetree validator), but let's return an error if
there are no child nodes at all for now.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221028160435.26948-7-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-usb: merge driver data

The USB QMP PHY driver only manages a single PHY so merge the old
qcom_qmp and qmp_phy structures and drop the PHY array.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221028160435.26948-6-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-usb: move pm ops

Move the PM ops structure next to the implementation to keep the driver
callbacks grouped.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221028160435.26948-5-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-usb: move device-id table

Move the device-id table below probe() and next to the driver structure
to keep the driver callback functions grouped together.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221028160435.26948-4-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-usb: sort device-id table

Sort the device-id table by compatible string to make it easier to find
and add new entries.

Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221028160435.26948-3-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-usb: fix sc8280xp PCS_USB offset

The PCS_USB register block lives at an offset of 0x1000 from the PCS
region on SC8280XP so add the missing offset to avoid corrupting
unrelated registers on runtime suspend.

Note that the current binding is broken as it does not describe the
PCS_USB region and the PCS register size does not cover PCS_USB and the
regions in between. As Linux currently maps full pages, simply adding
the offset to driver works until the binding has been fixed.

Fixes: c0c7769cdae2 ("phy: qcom-qmp: Add SC8280XP USB3 UNI phy")
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Link: https://lore.kernel.org/r/20221028160435.26948-2-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

Merge branch 'fixes' into next

Merge the fixes to next as both carry qmp-phy patches

phy: qcom-qmp-ufs: add support for updated sc8280xp binding

Add support for the new SC8280XP binding.

Note that the binding does not try to describe every register subregion
and instead the driver holds the corresponding offsets.

Also note that the driver will continue to accept the old binding, at
least for the time being.

Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Link: https://lore.kernel.org/r/20221024090041.19574-11-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-ufs: restructure PHY creation

In preparation for supporting devicetree bindings which do not use a
child node, move the PHY creation to probe() proper and parse the serdes
resource in what is now the legacy devicetree helper.

Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Link: https://lore.kernel.org/r/20221024090041.19574-10-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

dt-bindings: phy: qcom,qmp-ufs: fix sc8280xp binding

The current QMP UFS PHY bindings are based on the original MSM8996 PCIe
PHY binding which provided multiple PHYs per IP block and these in turn
were described by child nodes.

The QMP UFS PHY block only provide a single PHY and the remnant child
node does not really reflect the hardware.

The original MSM8996 binding also ended up describing the individual
register blocks as belonging to either the wrapper node or the PHY child
nodes.

This is an unnecessary level of detail which has lead to problems when
later IP blocks using different register layouts have been forced to fit
the original mould rather than updating the binding. The bindings are
arguable also incomplete as they only the describe register blocks used
by the current Linux drivers.

Add a new binding for the UFS QMP PHYs found on SC8280XP which further
bindings can be based on.

Note that the current binding is simply removed instead of being
deprecated as it was only recently merged and support for SC8280XP is
still under development.

Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Link: https://lore.kernel.org/r/20221024090041.19574-9-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

dt-bindings: phy: qcom,qmp-ufs: rename current bindings

The current QMP UFS PHY bindings are based on the original MSM8996 PCIe
PHY binding which provided multiple PHYs per IP block and these in turn
were described by child nodes.

The QMP UFS PHY block only provide a single PHY and the remnant child
node does not really reflect the hardware.

The original MSM8996 binding also ended up describing the individual
register blocks as belonging to either the wrapper node or the PHY child
nodes.

This is an unnecessary level of detail which has lead to problems when
later IP blocks using different register layouts have been forced to fit
the original mould rather than updating the binding. The bindings are
arguable also incomplete as they only the describe register blocks used
by the current Linux drivers.

In preparation for adding new bindings for SC8280XP which further
bindings can be based on, rename the current bindings after MSM8996 and
add a reference to the SC8280XP bindings.

Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Link: https://lore.kernel.org/r/20221024090041.19574-8-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-ufs: clean up PHY init

Clean up the PHY initialisation somewhat programming both tx and rx for
the second lane after the first lane.

Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Link: https://lore.kernel.org/r/20221024090041.19574-7-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-ufs: rename PHY ops structure

Rename the PHY operation structure so that it has a "phy_ops" suffix and
move it next to the implementation.

Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Link: https://lore.kernel.org/r/20221024090041.19574-6-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-ufs: clean up probe initialisation

Stop abusing the driver data pointer and instead pass the driver state
structure directly to the initialisation helpers during probe.

Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Link: https://lore.kernel.org/r/20221024090041.19574-5-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-ufs: clean up device-tree parsing

Since the QMP driver split there will be at most a single child node so
drop the obsolete iteration construct.

While at it, drop the verbose error logging that would have been
printed also on probe deferrals.

Note that there's no need to check if there are additional child nodes
(the kernel is not a devicetree validator), but let's return an error if
there are no child nodes at all for now.

Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Link: https://lore.kernel.org/r/20221024090041.19574-4-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-ufs: merge driver data

The UFS QMP PHY driver only manages a single PHY so merge the old
qcom_qmp and qmp_phy structures and drop the PHY array.

Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Link: https://lore.kernel.org/r/20221024090041.19574-3-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-ufs: move device-id table

Move the device-id table below probe() and next to the driver structure
to keep the driver callback functions grouped together.

Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Link: https://lore.kernel.org/r/20221024090041.19574-2-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-combo: fix NULL-deref on runtime resume

Commit fc64623637da ("phy: qcom-qmp-combo,usb: add support for separate
PCS_USB region") started treating the PCS_USB registers as potentially
separate from the PCS registers but used the wrong base when no PCS_USB
offset has been provided.

Fix the PCS_USB base used at runtime resume to prevent dereferencing a
NULL pointer on platforms that do not provide a PCS_USB offset (e.g.
SC7180).

Fixes: fc64623637da ("phy: qcom-qmp-combo,usb: add support for separate PCS_USB region")
Cc: stable@vger.kernel.org # 5.20
Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Reviewed-by: Andrew Halaney <ahalaney@redhat.com>
Link: https://lore.kernel.org/r/20221026162116.26462-1-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-usb: correct registers layout for IPQ8074 USB3 PHY

According to the kernel 4.4 sources from NHSS.QSDK.9.0.2 and according
to hardware docs, the PHY registers layout used for IPQ8074 USB3 PHY is
incorrect. This platform uses offset 0x174 for the PCS_STATUS register,
0xd8 for PCS_AUTONOMOUS_MODE_CTRL, etc.

Correct the PHY registers layout.

Fixes: 94a407cc17a4 ("phy: qcom-qmp: create copies of QMP PHY driver")
Fixes: 507156f5a99f ("phy: qcom-qmp: Add USB QMP PHY support for IPQ8074")
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Reviewed-by: Kathiravan T<quic_kathirav@quicinc.com>
Link: https://lore.kernel.org/r/20220929190017.529207-1-dmitry.baryshkov@linaro.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-usb: drop start and pwrdn-ctrl abstraction

Drop the start and pwrdn-ctrl abstractions which are no longer needed
since the QMP driver split.

Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Link: https://lore.kernel.org/r/20221012085002.24099-20-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-ufs: drop start and pwrdn-ctrl abstraction

Drop the start and pwrdn-ctrl abstractions which are no longer needed
since the QMP driver split.

Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Link: https://lore.kernel.org/r/20221012085002.24099-19-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-pcie-msm8996: drop start and pwrdn-ctrl abstraction

Drop the start and pwrdn-ctrl abstractions which are no longer needed
since the QMP driver split.

Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Link: https://lore.kernel.org/r/20221012085002.24099-18-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-pcie: add config sanity checks

The driver expects every configuration to set the pwrdn_ctrl and
phy_status masks. Add some probe WARN_ON_ONCE() to probe to catch any
new driver support that fails to provide them.

Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Link: https://lore.kernel.org/r/20221012085002.24099-17-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-pcie: drop start-ctrl abstraction

All PCIe PHYs need to start and stop the SerDes and PCS so drop the
start-ctrl abstraction which is no longer needed since the QMP driver
split.

Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Link: https://lore.kernel.org/r/20221012085002.24099-16-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>

phy: qcom-qmp-combo: drop start and pwrdn-ctrl abstraction

All USB PHYs need to start and stop the SerDes and PCS so drop the
start-ctrl abstraction which is no longer needed since the QMP driver
split.

Similarly, drop the pwrdn-ctrl abstraction which also is not needed
since the split.

Signed-off-by: Johan Hovold <johan+linaro@kernel.org>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Link: https://lore.kernel.org/r/20221012085002.24099-15-johan+linaro@kernel.org
Signed-off-by: Vinod Koul <vkoul@kernel.org>