ram: rk3399: Add LPPDR4 mr detection

Like data training in other sdram types, mr detection need
to taken care for lpddr4 with looped rank and associated
channel to make sure the proper configuration held.

Once the mr detection successful for active and configured
rank with channel number, the same can later reused during
actual LPDDR4 initialization.

So, add code to support for it.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: YouMin Chen <cym@rock-chips.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

ram: rk3399: Handle data training via ops

data training can be even required for lpddr4 and we
need to keep the lpddr4 code to compile only for relevant
boards which do support lpddr4.

For this requirement, and for code readability handle
data training via sdram_rk3399_ops and same will update
in future while supporting lpddr4 code.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

ram: rk3399: Simplify data training first argument

data training is using chan_info as first argument with
channel number as second argument instead of that use
dram_info as first argument so-that we can get the
chan_info at data training definition.

This was the argument handling is meaningful, readable
and it would help to add similar data training for
lpddr4 in future.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

ram: rk3399: Update lpddr4 vref_mode_ac

Update vref_mode_ac for lpddr4 based on VDDQ/3/2=16.8%

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: YouMin Chen <cym@rock-chips.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

ram: rk3399: Update lpddr4 mode_sel based on io settings

The mode_sel on lpddr4 value is depending on IO settings
of rd_vref.

Add support for it.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: YouMin Chen <cym@rock-chips.com>

ram: rk3399: Update lpddr4 vref based on io settings

The vref_mode_dq, vref_value_dq on lpddr4 value is depending
on IO settings of rd_vref.

Add support for it.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: YouMin Chen <cym@rock-chips.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

ram: rk3399: Get lpddr4 tsel_rd_en from io settings

For base.odt 1 the lpddr4 tsel_rd_en value is depending
on IO settings of rd_odt_en.

Add support for it.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: YouMin Chen <cym@rock-chips.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

ram: rk3399: Configure soc odt support

CTL 145, 146, 159, 160 registers are used to configure
soc odt on rk3399.

These soc odt values are updated from CS0_MR22_VAL and
CS1_MR22_VAL and for lpddr4 these values ORed with
tsel_rd_select_n.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: YouMin Chen <cym@rock-chips.com>

ram: rk3399: Add tsel control clock drive

tsel contrl clock drives are required to configure PHY
929, 939 controls drive settings.

Add support for these control clock for all dramtype
sdrams.

Thse control clock drives are configure via tsel_ckcs_select_p
and tsel_ckcs_select_n variables.

tsel_ckcs_select_n is PHY_DRV_ODT_34_3 value where as
tsel_ckcs_select_p is retrived from IO settings for lpddr4
and rest uses PHY_DRV_ODT_34_3.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: YouMin Chen <cym@rock-chips.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

ram: sdram: Configure lpddr4 tsel rd, wr based on IO settings

Now we have IO settings available for all supported sdram
frequencies, so retrieve these IO settings and make used
for LPDDR4 ds odt configuration.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: YouMin Chen <cym@rock-chips.com>

ram: rk3399: Add IO settings

Add IO settings for dram ctl and phy.

IO settings are useful for configuring ctl, phy odt, vref,
mr5, mode select and other needed input output operations
for lpddr4 or any other dramtype sdram.

Right now, this patch added IO setting for all supported
sdram frequencies.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: YouMin Chen <cym@rock-chips.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

ram: rk3399: Don't disable dfi dram clk for lpddr4, rank 1

The hardware for LPDDR4 with
- CLK0P/N connect to lower 16-bits
- CLK1P/N connect to higher 16-bits

and usually dfi dram clk is configured via CLK1P/N, so
disabling dfi dram clk will disable the CLK1P/N as well.

So, add patch to not to disable dfi dram clk for lpddr4,
with rank 1.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: YouMin Chen <cym@rock-chips.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

ram: rk3399: Configure tsel write ca for lpddr4

tsel write ca_p and ca_n values need to write on PHY 544, 672
and 800 to configure ds odt.

Configure the same PHY register for lpddr4 would require a mask
value of (300 << 8).

Add support for it.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: YouMin Chen <cym@rock-chips.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

ram: rk3399: Map chipselect for lpddr4

Assign desired cs_map values for lpddr4 during set memory map.

Initial cs_map values is based on the sdram parameters, so
the same will adjusted based dramtype as LPDDR4.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: YouMin Chen <cym@rock-chips.com>

ram: rk3399: Configure PHY RX_CM_INPUT for lpddr4

Configure PHY RX_CM_INPUT for lpddr4 during phy IO config.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: YouMin Chen <cym@rock-chips.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

ram: rk3399: Configure SLEWP_EN, SLEWN_EN for lpddr4

Configure SLEWP_EN, SLEWN_EN for lpddr4 during phy IO config.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: YouMin Chen <cym@rock-chips.com>

ram: rk3399: Configure BOOSTP_EN, BOOSTN_EN for lpddr4

Configure BOOSTP_EN, BOOSTN_EN for lpddr4 during phy IO config.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: YouMin Chen <cym@rock-chips.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

ram: rk3399: Configure PHY_898, PHY_919 for lpddr4

PHY_898, PHY_919 would require to configure PHY LP4 boot
pll control and ca for lpddr4.

So, configure the same in pctl_cfg for LPDDR4.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: YouMin Chen <cym@rock-chips.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

ram: rk3399: Avoid two channel ZQ Cal Start at the same time

It is possible in lpddr4 dram, where both the channels would
start at same time with ZQ Cal Start. If it uses ZQ Call start
then it will use RZQ.

For example LPDDR4 366 Dual-Die, Quad-Channel Package, RZQ maybe
connect to both channel. If ZQ Cal Start at the same time,
it will use the same RZQ.

It is not a problem of using RZQ in both the channels, but can not
use at the same time.

So, to avoid this, we have an option of dram tINIT3 value for
increasing the frequency for channel 1.

This patch increase the available tINIT3 with existing running
dram frequency.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: YouMin Chen <cym@rock-chips.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

ram: rk3399: Don't wait for PLL lock in lpddr4

lpddr4 has PLL bypass mode during phy initialization phase,
which does all pll configurations.

So no need to wait explicitly during pctl config.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: YouMin Chen <cym@rock-chips.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

ram: rk3399: Move mode_sel assignment

mode_sel assignment is based on dram type.

In phy_io_config, already have vref setting based
on the dram type, so move this mode_sel assignment
on vref setting area.

No functionality change.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

ram: rk3399: Add lpddr4 rank mask for wdql training

Add rank_mask based on the rank number for lpddr4.

This would keep the wdql data training loop based on the
desired rank mask value instead of looping for all values.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: YouMin Chen <cym@rock-chips.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

ram: rk3399: Add lpddr4 rank mask for ca training

Add rank_mask based on the rank number for lpddr4.

This would keep the ca data training loop based on the
desired rank mask value instead of looping for all values.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: YouMin Chen <cym@rock-chips.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

ram: rockchip: Kconfig: Add RK3399 LPDDR4 entry

Supporting LPDDR4 code support in RK3399 would increases
the size of SPL/TPL.

So add kconfig entry for RK3399 LPDDR4 code so-that
the boards have LPDDR4 can enable them via defconfig.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

ram: rk3399: Configure phy IO in ds odt

Some dramtypes like lpddr4 initialization would required to
configure phy IO even after pctl_cfg and after set_ds_odt.

For those cases the set_ds_odt would be an initial call to
setup the phy.

To satisfy all the cases, trigger phy IO from set_ds_odt.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

ram: rk3399: Add DdrMode

Add DdrMode structure with associated bit fields.

These would help to reconfigure sdram capabilities during
lpddr4 setup related configs.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

ram: rk3399: Add ddrtimingC0

Add DdrTimingC0 structure with associated bit fields.

These would help to reconfigure sdram capabilities during
lpddr4 setup related configs.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

ram: rk3399: Add ddr version enc macro

Add dram config macro for handling ddr version number.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: YouMin Chen <cym@rock-chips.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

ram: rk3399: Introduce sys_reg3 for more capacity info

cs0_row, cs1_row and cs1_col needs more bits to show its
correct value, update to make use of both sys_reg2,
sys_reg3.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: YouMin Chen <cym@rock-chips.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>
(Squash similar patches into one patch)
Signed-off-by: Kever Yang <Kever.yang@rock-chips.com>

ram: rk3399: Rename sys_reg with sys_reg2

Use dram config variable name as sys_reg2 instead of sys_reg
since the final variable value is to written into a pmugrf
register named as sys_reg2.

This reflect the both variable and associated register
names are same and also help to add next sys_reg's to
add it in future.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: YouMin Chen <cym@rock-chips.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

ram: rk3399: Simply existing dram enc macro

Add simplified and meaningful macro for all setting.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: YouMin Chen <cym@rock-chips.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>
(Squash the similar patches into 1 patch)
Signed-off-by: Kever Yang <Kever.yang@rock-chips.com>

ram: rk3399: Enable sdram debug functions

This would help to debug the sdram base parameters while
debugging existing chip or while supporting new sdram type.

It require explicit enablement of CONFIG_RAM_ROCKCHIP_DEBUG
for showing the debug prints.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

ram: rk3399: Add rank detection support

Right now the rk3399 sdram driver assume that the board
has configured with 2 channels, so any possibility to
enable single channel on the same driver will encounter
channel #1 data training failure.

Log:
U-Boot TPL board init
sdram_init: data training failed
rk3399_dmc_init DRAM init failed -5

So, add an algorithm that can capable to compute the active
or configured rank with associated channel like
a) do rank loop to compute the active rank, with associated
   channel numbers
b) then, succeed the data training only for configured channel
c) preserve the rank for given channel
d) do channel loop for setting the active channel
e) if given rank is zero or inactive on the specific channel,
   clear the timings for the associated channel
f) finally, return error if number of channels is zero

Tested in NanoPI-NEO4 since it support single channel sdram
configuration.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: YouMin Chen <cym@rock-chips.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>
(add PI_READ_GATE_TRAINING for LPDDR3 to support rk3399-evb case)
Signed-off-by: Kever Yang <Kever.yang@rock-chips.com>

ram: rk3399: Compute stride for 1 channel a

Add stride computation for the sdram which support
single channel a

This configuration available in NanoPi NEO4 and the
same can work with existing rk3399-sdram-ddr3-1866.dtsi

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: YouMin Chen <cym@rock-chips.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

ram: rk3399: Compute stride for 2 channels

stride value from sdram timings can be computed dynamically
based on the determined capacity for the given channel.

Right now these stride values are taken as part of sdram timings
via dtsi, but it possible to use same timings dtsi for given
frequency even though the configured board sdram do support
single channel with different size by dynamically detect the
stride value.

Example, NanoPi NEO4 do have DDR3-1866, but with single channel
and 1GB size with dynamic stride detection it is possible to
use existing rk3399-sdram-ddr3-1866.dtsi whose stride,
number of channels and capacity it support is d efferent.

So, add initial support to calculate the stride value for
2 channels sdram, which is available by default on existing
boards.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: YouMin Chen <cym@rock-chips.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

ram: rk3399: debug: Add sdram_print_stride

Add code to print the channel stride, this would help to
print the stride of associated channel.

Here is sample print on LPDDR4, 50MHz.
256B stride

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: YouMin Chen <cym@rock-chips.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

ram: rockchip: debug: Get the cs capacity

Add code to get the channel capacity, this would help to
print the capacity of specific channel.

Here is sample print on LPDDR4, 50MHz channel 0
BW=32 Col=10 Bk=8 CS0 Row=15 CS1 Row=15 CS=2 Die BW=16 Size=2048MB

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: YouMin Chen <cym@rock-chips.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

ram: rockchip: debug: Add sdram_print_ddr_info

Add sdram ddr info print support, this would help to
observe the sdram base parameters.

Here is sample print on LPDDR4, 50MHz channel 0
BW=32 Col=10 Bk=8 CS0 Row=15 CS1 Row=15 CS=2 Die BW=16

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: YouMin Chen <cym@rock-chips.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

ram: rockchip: Add debug sdram driver

Add sdram driver to handle debug across rockchip SoCs.

This would help to improve code debugging feature for
sdram drivers in rockchip family, whoever wants to
debug the driver should call these core debug code on
their respective platform sdram drivers.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: YouMin Chen <cym@rock-chips.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

debug_uart: Add printdec

Add printdec, this would help to print an
output a decimalism value.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: YouMin Chen <cym@rock-chips.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

ram: rockchip: Add initial Kconfig

Right now sdram drivers in rockchip SoC are built based
on the SoC configs which may not be an adequate solutions
while adding common or debug driver.

So, add meaningful Kconfig options start with rk3399.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

ram: rk3399: Add pctl start support

Add support for pctl start for both channel 0, 1 control
and phy registers.

This would also handle pwrup_srefresh_exit init based
on the channel number.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: YouMin Chen <cym@rock-chips.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

ram: rk3399: Move pwrup_srefresh_exit to dram_info

Add pwrup_srefresh_exit to be part of dram_info so-that
the it can help to support pwrup_srefresh_exit in individual
channels while starting pctl in future.

No functionality change.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

ram: rk3399: Add phy pctrl reset support

Add support for phy pctrl reset support for both channel 0, 1.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: YouMin Chen <cym@rock-chips.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

ram: rk3399: Use rank mask in wdql data training

Add rank_mask based on the rank number, this would keep
the wdql data training loop based on the desired rank mask
value instead of looping for all values.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: YouMin Chen <cym@rock-chips.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

ram: rk3399: Use rank mask in ca data training

Add rank_mask based on the rank number, this would keep
the ca data training loop based on the desired rank mask
value instead of looping for all values.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: YouMin Chen <cym@rock-chips.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

ram: rk3399: Clear PI_175 interrupts in data training

Clear the PI_175 interrupts before processing actual
data training in all relevant calls.

This would help to clear interrupt from previous training.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: YouMin Chen <cym@rock-chips.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

ram: rk3399: Handle data training return types

data trainings calls like ca, wl, rg, rl, wdql have proper
return types with -EIO and the return type missed to handle
in data_training function.

This patch, add proper return type checks along with useful
debug statement on each data training calls.

Incidentally this would help to prevent the sdram initialization
hang for single channel dram and when the code is trying to
initialize second channel with proper return type of relevant
data training call might failed.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

clk: rockchip: rk3399: Fix check patch warnings and checks

- CHECK: spaces preferred around that '*'
- CHECK: spaces preferred around that '/'
- CHECK: space preferred before that '|'
- WARNING: macros should not use a trailing semicolon
- CHECK: Unnecessary parentheses around 'fbdiv <= min_fbdiv'
- CHECK: Unnecessary parentheses around 'parent->id == SCLK_MAC'
- CHECK: Unnecessary parentheses around 'parent->dev == clk->dev'
- WARNING: line over 80 characters
- CHECK: Prefer kernel type 'u8' over 'uint8_t'
- Add proper macro definitions arrangements

Note: there are still line over 80 characters and other warnings but
fixing those making code look unreadable, so I kept it as it is.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

arm: include: rockchip: Add DDR4 enum

Add DDR4 enum number in common header.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: YouMin Chen <cym@rock-chips.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

arm: include: rockchip: Move dramtypes to common header

dramtype enum numbers as common across all dram controllers
in rockchip, so move the eneum values in common header.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

ram: rk3399: Move common sdram structures in common header

Move common sdram structures like sdram_cap_info, sdram_base_params
into sdram_common header, this would help to reuse the same
from another controllers like px30.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

ram: rk3399: s/rk3399_base_params/sdram_base_params

Most of the ddr parameters are common in rk3399_base_params
structure and which would reuse it in another controller like
px30 in future.

So, rename the structure from rk3399_base_params into
sdram_base_params.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: YouMin Chen <cym@rock-chips.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

ram: rockchip: rk3399: Add cap_info structure

Group common ddr attributes like
- rank
- col
- bk
- bw
- dbw
- row_3_4
- cs0_row
- cs1_row
- ddrconfig

into a common cap_info structure for more code readability and extend
if possible based on the new features.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

ram: rk3399: Order tsel variables

Order tsel* variable declarations and assignment in proper
and meaningful way.

No functionality change.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: YouMin Chen <cym@rock-chips.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

ram: rk3399: s/ca_tsel_wr_select_p/tsel_wr_select_ca_p

Rename ca_tsel_wr_select_p to tsel_wr_select_ca_p based
on the bsp code.

No functionality change.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: YouMin Chen <cym@rock-chips.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

ram: rk3399: s/ca_tsel_wr_select_n/tsel_wr_select_ca_n

Rename ca_tsel_wr_select_n to tsel_wr_select_ca_n based
on the bsp code.

No functionality change.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: YouMin Chen <cym@rock-chips.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

ram: rk3399: s/tsel_wr_select_p/tsel_wr_select_dq_p

Rename tsel_wr_select_p to tsel_wr_select_dq_p based
on the bsp code.

No functionality change.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: YouMin Chen <cym@rock-chips.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

ram: rk3399: s/tsel_wr_select_n/tsel_wr_select_dq_n

Rename tsel_wr_select_n to tsel_wr_select_dq_n based
on the bsp code.

No functionality change.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: YouMin Chen <cym@rock-chips.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

ram: rk3399: Handle pctl_cfg return type

Add proper return type handling of pctl_cfg with
meaningful print statement.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

ram: rk3399: s/sdram_params/params

Rename variable name of struct rk3399_sdram_params
from sdram_params with params for more code readability.

No functionality change.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

ram: rk3399: Some trivial code fixes

- Add proper spaces in data training, rk3399_dmc_init, pctl_cfg
- Order include files
- Move macro after include files

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

ram: rk3399: Fix code warnings

Fix checkpatch warninigs on sdram_rk3399.c like
- Avoid CamelCase
- Unnecessary parentheses
- Alignment should match open parenthesis
- multiple blank lines
- misspelled
- spaces preferred around that '>>'

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

rockchip: rock960-rk3399: fix mail format in MAINTAINER file

The mail format should have '<>', or else the patman won't
recognize it correctley.

Signed-off-by: Kever Yang <kever.yang@rock-chips.com>

rockchip: dts: rk3399: Add 'same-as-spl' for Rock PI 4

Let the board continue boot from the storage device where
it bootup.

Signed-off-by: Andy Yan <andy.yan@rock-chips.com>
Reviewed-by: Kever Yang <Kever.yang@rock-chips.com>

rockchip: dts: rk3399: Add spl-boot-order for Rock PI 4

RK3399 use sdhci for eMMC and DW MMC for SD Card, and
spl will only try to boot from SDMMC if we don't specify
other boot device for spl-boot-order. So add sdhci and sdmmc
for spl-boot-order here.

Signed-off-by: Andy Yan <andy.yan@rock-chips.com>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>

configs: rockchip: rock960: enable USB3 support

Enable USB3 support via the dwc3 XHCI driver.

Signed-off-by: Peter Robinson <pbrobinson@gmail.com>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
Reviewed-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>

configs: rockchip: rock960: Add support for USB ethernet adapters

As the Rock960 doesn't have an onboard wired ethernet interface
it's useful to have some common USB wired ethernet devices added
to enable testing.

Signed-off-by: Peter Robinson <pbrobinson@gmail.com>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
Reviewed-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>

configs: rockchip: rock960: enable DMA for SDHCI controller

Enable the SDMA controller so the eMMC connected to the SDHCI
controller (sdhci@fe330000) can make use of it.

Signed-off-by: Peter Robinson <pbrobinson@gmail.com>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
Reviewed-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>

configs: rockchip: rock960: enable pmic and regulator commands

We have both PMIC and Regulator functionality so it's useful to
be able to see output and debug with the commands enabled.

Signed-off-by: Peter Robinson <pbrobinson@gmail.com>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
Reviewed-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>

configs: rockchip: rock960: drop options for non-existent HW

The Rock960 doesn't contain SPI flash so drop related config options.

Signed-off-by: Peter Robinson <pbrobinson@gmail.com>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
Reviewed-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>

arm64: rockchip: rock960: sync dts files from Linux 5.2-rc6

Sync the dts files for the Rock960 boards from Linux to get the
latest changes and fixes for the devices.

Signed-off-by: Peter Robinson <pbrobinson@gmail.com>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
Reviewed-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>

rockchip: xhci: Remove RK3399 support

Remove RK3399 compatible strings as this driver is no longer
used on that SoC.

Signed-off-by: Mark Kettenis <kettenis@openbsd.org>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>

usb: xhci-dwc3: Add USB2 PHY configuration

Configure USB2 PHY register based on "phy_type" property and
handle all the quirks that are relevant for Rockchip RK3399 SoCs.

Signed-off-by: Mark Kettenis <kettenis@openbsd.org>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>

usb: dwc3-of-simple: Add support for RK3399

Add compatible string for RK3399 and enable it by default on
Rockchip platforms with USB3 support.

Signed-off-by: Mark Kettenis <kettenis@openbsd.org>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>

rockchip: clk: rk3399: handle clk_enable requests for USB3

The "simple" OF glue layer for the Designware USB3 core enables
all refernced clocks.  These need to be need to be implemented
otherwise the driver fails to probe.  A dummy implementation
that simply returns success is sufficient since the RK3399 comes
out of reset with all clock gates open.

Signed-off-by: Mark Kettenis <kettenis@openbsd.org>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>

Merge branch '2019-07-17-master-imports'

- Various FS/disk related fixes with security implications.
- Proper fix for the pci_ep test.
- Assorted bugfixes
- Some MediaTek updates.
- 'env erase' support.

Revert "test: Disable pci_ep test for now"

We now have a proper fix for this test, stop disabling it in CI.

This reverts commit ae8d23a668755d804748a1cf848426b28338b3d5.

Signed-off-by: Tom Rini <trini@konsulko.com>

pci_ep: fix wrong addressing to barno

barno was mistakely readed from the target structure,
resulting in undefined behavious depending on the previous memory
content. fix that.

Fixes: bb413337826e ("pci_ep: add pci endpoint sandbox driver")
Signed-off-by: Ramon Fried <rfried.dev@gmail.com>
[trini: Drop unused bar_idx]
Signed-off-by: Tom Rini <trini@konsulko.com>

board: Arcturus: ucp1020: Removing obsoleted stuff

Removed one of the defconfig(obsoleted) file
and unused CONFIG_MMC_SPI definition to avoid confusion
about if this board using non-DM stuff or not.
uCP1020 is completely DM free board, tested and runs well.

Signed-off-by: Oleksandr Zhadan <oleks@arcturusnetworks.com>
Signed-off-by: Michael Durrant <mdurrant@arcturusnetworks.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>

blk: Invalidate block cache when switching hwpart

Some storage devices have multiple hw partitions and both address from
zero, for example eMMC.
However currently block cache invalidation only applies to block
write/erase.
This can cause a problem that data of current hw partition is cached
before switching to another hw partition. And the following read
operation of the latter hw partition will get wrong data when reading
from the addresses that have been cached previously.

To solve this problem, invalidate block cache after a successful
select_hwpart operation.

Signed-off-by: Weijie Gao <weijie.gao@mediatek.com>

arm: dts: MediaTek: remove tick-timer from mt7629.dtsi

This patch removes tick-timer as all mt7629 boards should use arch timer.

Signed-off-by: Weijie Gao <weijie.gao@mediatek.com>

configs: mt7629_rfb: use arm arch timer instead of mtk timer

This patch changes mt7629_rfb to use ARM's generic arch timer instead of
MediaTek's soc timer.

Signed-off-by: Weijie Gao <weijie.gao@mediatek.com>

arm: dts: MediaTek: fix clock order for timer0 node of mt7629.dtsi

The timer0 node has its two clocks written in reversed order. The timer0
is used as the tick timer which causes a problem that the time a delay
function used is 4 times longer.

This patch reverses these two clocks to solve this issue.

Signed-off-by: Weijie Gao <weijie.gao@mediatek.com>

chromium: Update docs to clone vboot_reference directly

We don't need a full checkout of Chrome OS to build U-Boot with
Chromium OS verified boot. Update the instructions accordingly and fix a
typo which joins the output directory and defconfig.

Signed-off-by: Simon Glass <sjg@chromium.org>

arm: mediatek: add missing arch timer configuration for MT7629

This patch sets CNTVOFF of ARM CP15 timer to zero to make sure the virtual
counter is fully usable for linux kernel.

Cc: Albert Aribaud <albert.u.boot@aribaud.net>
Signed-off-by: Weijie Gao <weijie.gao@mediatek.com>

power-domain.h: Fix typo

%s/ot/to/

Signed-off-by: Anatolij Gustschin <agust@denx.de>

CVE-2019-13106: ext4: fix out-of-bounds memset

In ext4fs_read_file in ext4fs.c, a memset can overwrite the bounds of
the destination memory region. This patch adds a check to disallow
this.

Signed-off-by: Paul Emge <paulemge@forallsecure.com>

ext4: gracefully fail on divide-by-0

This patch checks for 0 in several ext4 headers and gracefully
fails instead of raising a divide-by-0 exception.

Signed-off-by: Paul Emge <paulemge@forallsecure.com>

CVE-2019-13104: ext4: check for underflow in ext4fs_read_file

in ext4fs_read_file, it is possible for a broken/malicious file
system to cause a memcpy of a negative number of bytes, which
overflows all memory. This patch fixes the issue by checking for
a negative length.

Signed-off-by: Paul Emge <paulemge@forallsecure.com>

CVE-2019-13105: ext4: fix double-free in ext4_cache_read

ext_cache_read doesn't null cache->buf, after freeing, which results
in a later function double-freeing it. This patch fixes
ext_cache_read to call ext_cache_fini instead of free.

Signed-off-by: Paul Emge <paulemge@forallsecure.com>

CVE-2019-13103: disk: stop infinite recursion in DOS Partitions

part_get_info_extended and print_partition_extended can recurse infinitely
while parsing a self-referential filesystem or one with a silly number of
extended partitions. This patch adds a limit to the number of recursive
partitions.

Signed-off-by: Paul Emge <paulemge@forallsecure.com>

qemu-riscv: enable VIRTIO_PCI

libvirt v.5.3.0 with QEMU 4.0.0 or above uses PCI automatically and
thus devices (network, storage, etc) are connected via PCI.

Signed-off-by: David Abdurachmanov <david.abdurachmanov@sifive.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>

arm: qemu: fix failure in flash initialization if booting from TF-A

If U-Boot is loaded and started from TF-A (you need to change
SYS_TEXT_BASE to 0x60000000), it will hang up at flash initialization.

If secure mode is off (default, or -machine virt,secure=off) at qemu,
it will provide dtb with two flash memory banks:
flash@0 {
bank-width = <0x4>;
reg = <0x0 0x0 0x0 0x4000000 0x0 0x4000000 0x0 0x4000000>;
compatible = "cfi-flash";
};
If secure mode is on, on the other hand, qemu provides dtb with 1 bank:
flash@0 {
bank-width = <0x4>;
reg = <0x0 0x4000000 0x0 0x4000000>;
compatible = "cfi-flash";
};

As a result, flash_init()/flash_get_size() will eventually fail.
With this patch applied, relevant CONFIG values are modified.

Signed-off-by: AKASHI Takahiro <takahiro.akashi@linaro.org>
Tested-by: Heinrich Schuchardt <xypron.glpk@gmx.de>

arm: move CONFIG_TFABOOT to generic Kconfig

Currently, CONFIG_TFABOOT is located in armv8/fsl-layerscape Kconfig,
but it will be also useful for other targets if some additional
configuration are necessary.
So move it to arch/arm/Kconfig.

Please note that CONFIG_TFABOOT still depends on
CONFIG_ARCH_SUPPORT_TFABOOT and so the menu won't come up
if any target doesn't need its own customization for TF-A boot.
This will maintain the compatibility.

Signed-off-by: AKASHI Takahiro <takahiro.akashi@linaro.org>
Cc: Rajesh Bhagat <rajesh.bhagat@nxp.com>
Cc: York Sun <york.sun@nxp.com>
Cc: Prabhakar Kushwaha <prabhakar.kushwaha@nxp.com>
Cc: Priyanka Jain <priyanka.jain@nxp.com>
Cc: Sriram Dash <sriram.dash@nxp.com>
Cc: Hou Zhiqiang <Zhiqiang.Hou@nxp.com>
Cc: Peng Ma <peng.ma@nxp.com>
Cc: Yuantian Tang <andy.tang@nxp.com>
Cc: Pankit Garg <pankit.garg@nxp.com>

doc: Move fastboot protocol doc to android dir

Signed-off-by: Sam Protsenko <semen.protsenko@linaro.org>

cmd: mem: Add a command to fill the memory with random data

This command fills the memory with data produced by rand().

Signed-off-by: Jean-Jacques Hiblot <jjhiblot@ti.com>

tools: mkenvimage: Always consider non-regular files

At the moment mkenvimage has two separate read paths: One to read from
a potential pipe, while dynamically increasing the buffer size, and a
second one using mmap(2), using the input file's size. This is
problematic for two reasons:
- The "pipe" path will be chosen if the input filename is missing or
  "-".  Any named, but non-regular file will use the other path, which
  typically will cause mmap() to fail:
  $ mkenvimage -s 256 -o out <(echo "foo=bar")
- There is no reason to have *two* ways of reading a file, since the
  "pipe way" will always work, even for regular files.

Fix this (and simplify the code on the way) by always using the method
of dynamically resizing the buffer. The existing distinction between
the two cases will merely be used to use the open() syscall or not.

Signed-off-by: Andre Przywara <andre.przywara@arm.com>

tools: mkenvimage: Fix reading from slow pipe

It is perfectly fine for the read(2) syscall to return with less than
the requested number of bytes read (short read, see the "RETURN VALUE"
section of the man page). This typically happens with slow input
(keyboard, network) or with complex pipes.

So far mkenvimage expects the exact number of requested bytes to be
read, assuming an end-of-file condition otherwise. This wrong behaviour
can be easily shown with:
$ (echo "foo=bar"; sleep 1; echo "bar=baz") | mkenvimage -s 256 -o out -
The second line will be missing from the output.

Correct this by checking for any positive, non-zero return value.

This fixes a problem with a complex pipe in one of my scripts, where
the environment consist of two parts.

Signed-off-by: Andre Przywara <andre.przywara@arm.com>
Acked-by: Alexander Dahl <ada@thorsis.com>

test/py: gpt: Use long options for sgdisk

sgdisk 0.8.10.2 from AOSP doesn't support short options, failing with
errors like this:

    sgdisk: invalid option -- 'U'

Test fails due to that error. Let's use long options to make the test
work with any sgdisk version.

Signed-off-by: Sam Protsenko <semen.protsenko@linaro.org>
Acked-by: Stephen Warren <swarren@nvidia.com>