pci: layerscape: Modify the ls_pcie_dump_atu function

Modify the ls_pcie_dump_atu function, make it can print the INBOUND
windows registers.

Signed-off-by: Xiaowei Bao <xiaowei.bao@nxp.com>
Signed-off-by: Hou Zhiqiang <Zhiqiang.Hou@nxp.com>
Signed-off-by: Priyanka Jain <priyanka.jain@nxp.com>

pci_ep: layerscape: Add the SRIOV VFs of PF support

Add the INBOUND configuration for VFs of PF.

Signed-off-by: Xiaowei Bao <xiaowei.bao@nxp.com>
Signed-off-by: Hou Zhiqiang <Zhiqiang.Hou@nxp.com>
Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>

pci_ep: layerscape: Add Support for ls2085a and ls2080a EP mode

Due to the ls2085a and ls2080a use different way to set the BAR size,
so add the BAR size init code here.

Signed-off-by: Xiaowei Bao <xiaowei.bao@nxp.com>
Signed-off-by: Hou Zhiqiang <Zhiqiang.Hou@nxp.com>
Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>

pci_ep: layerscape: Add the workaround for errata A-009460

The VF_BARn_REG register's Prefetchable and Type bit fields
are overwritten by a write to VF's BAR Mask register.
workaround: Before writing to the VF_BARn_MASK_REG register,
write 0b to the PCIE_MISC_CONTROL_1_OFF register.

Signed-off-by: Xiaowei Bao <xiaowei.bao@nxp.com>
Signed-off-by: Hou Zhiqiang <Zhiqiang.Hou@nxp.com>
Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>

pcie_ep: layerscape: Add the multiple function support

Add the multiple function support for Layerscape platform, some PEXs
of Layerscaple platform have more than one PF.

Signed-off-by: Xiaowei Bao <xiaowei.bao@nxp.com>
Signed-off-by: Hou Zhiqiang <Zhiqiang.Hou@nxp.com>
Signed-off-by: Priyanka Jain <priyanka.jain@nxp.com>

armv8: dts: ls1046a: Add the PCIe EP node

Add the PCIe EP node for ls1046a.

Signed-off-by: Xiaowei Bao <xiaowei.bao@nxp.com>
Signed-off-by: Hou Zhiqiang <Zhiqiang.Hou@nxp.com>
Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>

pci_ep: Add the init function

Some EP deivces need to initialize before RC scan it, e.g. NXP
layerscape platform, so add the init function in pci_ep uclass.

Signed-off-by: Xiaowei Bao <xiaowei.bao@nxp.com>
Signed-off-by: Hou Zhiqiang <Zhiqiang.Hou@nxp.com>
Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>

pci: layerscape: Split the EP and RC driver

Split the RC and EP driver, and reimplement the EP driver base on
the EP framework.

Signed-off-by: Xiaowei Bao <xiaowei.bao@nxp.com>
Signed-off-by: Hou Zhiqiang <Zhiqiang.Hou@nxp.com>
[Rebased]
Signed-off-by: Priyanka Jain <priyanka.jain@nxp.com>

arm: dts: lx2160a: Increase configuration window size

lx2160a rev2 requires 4KB space for type0 and 4KB
space for type1 iATU window. Increase configuration
size to 8KB to have sufficient space for type0
and type1 window.

Signed-off-by: Wasim Khan <wasim.khan@nxp.com>
Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>

configs:ls1046afrwy: Add tfa secure boot defonfig

Add TFA secure boot defconfig and Enables secure boot related
configs in it.

Signed-off-by: Manish Tomar <manish.tomar@nxp.com>
Signed-off-by: Priyanka Jain <priyanka.jain@nxp.com>

freescale: ls1043aqds: drop ifdef CONFIG_SYS_I2C

- Drop ifdef CONFIG_SYS_I2C to initialize
  baudrate of i2c

- Drop warning of i2c_early_init_f as follows,
  warning: implicit declaration of function 'i2c_early_init_f'; did you
  mean 'arch_early_init_r'? [-Wimplicit-function-declaration]

Signed-off-by: Biwen Li <biwen.li@nxp.com>
Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>

freescale: ls1043aqds: enable secure system counter

Enable secure system counter in board_early_init_f for udelay()
to fix a bug that always return 0 by timer_read_counter()
when boot from qspi(No TFA)

Signed-off-by: Biwen Li <biwen.li@nxp.com>
Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>

freescale: ls1046aqds: drop ifdef CONFIG_SYS_I2C

- Drop ifdef CONFIG_SYS_I2C to initialize
  baudrate of i2c

- Drop warning of i2c_early_init_f as follows,
  warning: implicit declaration of function 'i2c_early_init_f'; did you
  mean 'arch_early_init_r'? [-Wimplicit-function-declaration]

Signed-off-by: Biwen Li <biwen.li@nxp.com>
Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>

freescale: ls1046aqds: enable secure system counter

Enable secure system counter in board_early_init_f for udelay()
to fix a bug that always return 0 by timer_read_counter()
when boot from qspi(No TFA)

Signed-off-by: Biwen Li <biwen.li@nxp.com>
Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>

i2c: mxc: move i2c_early_init_f to common function

Move i2c_early_init_f to common function
to initialize baudrate of i2c

Signed-off-by: Biwen Li <biwen.li@nxp.com>
Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>

crypto/fsl: add RNG support

Register the random number generator with the rng subsystem in u-boot.
This way it can be used by EFI as well as for the 'rng' command.

Signed-off-by: Michael Walle <michael@walle.cc>
Tested-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>

crypto/fsl: instantiate the RNG with prediciton resistance

If it is already instantiated tear it down first and then reinstanciate
it again with prediction resistance.

Signed-off-by: Michael Walle <michael@walle.cc>
Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>

crypto/fsl: don't regenerate secure keys

The secure keys (TDKEK, JDKEK, TDSK) can only be generated once after a
POR. Otherwise the RNG4 will throw an error.

Signed-off-by: Michael Walle <michael@walle.cc>
Reviewed-by: Horia Geantă <horia.geanta@nxp.com>
Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>

crypto/fsl: support newer SEC modules

Since Era 10, the version registers changed. Add the version registers
and use them on newer modules.

Signed-off-by: Michael Walle <michael@walle.cc>
Reviewed-by: Horia Geantă <horia.geanta@nxp.com>
Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>

crypto/fsl: export caam_get_era()

We need the era in other modules, too. For example, to get the RNG
version.

Signed-off-by: Michael Walle <michael@walle.cc>
Reviewed-by: Horia Geantă <horia.geanta@nxp.com>
Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>

crypto/fsl: make SEC%u status line consistent

Align the status line with all the other output in U-Boot.

Before the change:
DDR    3.9 GiB (DDR3, 32-bit, CL=11, ECC on)
SEC0: RNG instantiated
WDT:   Started with servicing (60s timeout)

After the change:
DDR    3.9 GiB (DDR3, 32-bit, CL=11, ECC on)
SEC0:  RNG instantiated
WDT:   Started with servicing (60s timeout)

Signed-off-by: Michael Walle <michael@walle.cc>
Reviewed-by: Horia Geantă <horia.geanta@nxp.com>
Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>

crypto/fsl: unused value in caam_hash_update()

The value 0 assigned to final is overwritten before ever being used.

Remove the assignment.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Reviewed-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>

crypto/fsl: correct printf() statement.

The sequence of arguments should match the format string.
For printing unsigned numbers we should use %u.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>

arm64: ls1043a: Remove the workaround of erratum A-009929

The workaround has been implemented in PBI phase, so remove
the duplicated implementation from U-Boot.

Signed-off-by: Hou Zhiqiang <Zhiqiang.Hou@nxp.com>
Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>

configs: ls1012a: Increase CONFIG_SYS_MALLOC_LEN value

Previous attempt to increase CONFIG_SYS_MALLOC_LEN was done in commit
c084a8edf4e2 ("configs: ls1012a: Increase CONFIG_SYS_MALLOC_LEN size")
which increased malloc memory to ~1M.

PFE firmware alone requires 3M of dynamic memory allocation and
therefore, increase the config value to a larger value i.e 5M. This size
should be enough as of now to accommodate further memory requirements.

Signed-off-by: Kuldeep Singh <kuldeep.singh@nxp.com>
Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>

configs: lx2160aqds: enable CONFIG_BOARD_EARLY_INIT_R

Enable CONFIG_BOARD_EARLY_INIT_R for SDHC adapter card
identification and configuration.

Signed-off-by: Yangbo Lu <yangbo.lu@nxp.com>
Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>

board: fsl: lx2160aqds: identify SDHC adapter during board init

Add support for SDHC adapter identification and configuration
during board init.

Signed-off-by: Yangbo Lu <yangbo.lu@nxp.com>
Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>

Move eSDHC adapter card identification to board files

The eSDHC adapter card identification and multiplexing configuration
through FPGA had been implemented in both common mmc driver and
fsl_esdhc driver. However it is proper to move these code to board
files and do it during board initialization. The FPGA registers are
also board specific.

This patch is to move eSDHC adapter card identification and
multiplexing configuration from mmc driver to specific board files.
And the option CONFIG_FSL_ESDHC_ADAPTER_IDENT is no longer needed.

Signed-off-by: Yangbo Lu <yangbo.lu@nxp.com>
[Rebased, Removed T1040QDS change as board does not exist]
Signed-off-by: Priyanka Jain <priyanka.jain@nxp.com>

Drop global data sdhc_adapter for powerpc

The sdhc_adapter of global data has not been used, and we
do not have to use it as global data even we may need it
in the future.

Signed-off-by: Yangbo Lu <yangbo.lu@nxp.com>
Reviewed-by: Peng Fan <peng.fan@nxp.com>
Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>

armv8: ls1028ardb: add xspi parameter to qixis command

Add xspi boot source to qixis command to let the soc boot from
flex-nor flash chip.

Signed-off-by: Yuantian Tang <andy.tang@nxp.com>
Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>

armv8: ls1012a: RGMII ports require internal delay

The correct setting for the RGMII ports on LS1012ARDB is to
enable delay on both Rx and Tx so the interface mode used should
be PHY_INTERFACE_MODE_RGMII_ID

Signed-off-by: Chaitanya Sakinam <chaitanya.sakinam@nxp.com>
Signed-off-by: Anji J <anji.jagarlmudi@nxp.com>
Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>

include/configs: ls1012a: Remove fdt_high env variable

Remove "fdt_high" environment variable to use the bootm_size
to safely contain a kernel, device tree and initrd for
relocation.

Signed-off-by: Udit Agarwal <udit.agarwal@nxp.com>
Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>

config: lx2160/2a: enable dspi

Enable dspi in lx2160aqds tfa defconfig
Enable CONFIG_SPI_FLASH_SST/EON in config file.

Signed-off-by: Zhao Qiang <qiang.zhao@nxp.com>
Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>

armv8: dts: fsl-lx2160a: add flash node under dspi to qds dts

Add flash node under dspi into fsl-lx2160a-qds.dtsi

Signed-off-by: Zhao Qiang <qiang.zhao@nxp.com>
Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>

crypto/fsl: fix unaligned access

On aarch64 running with dcache off, will result in an unaligned access
exception:

   => dcache off
   => hash sha1 $kernel_addr_r 100
   "Synchronous Abort" handler, esr 0x96000061
   elr: 00000000960317d8 lr : 00000000960316a4 (reloc)
   elr: 00000000fbd787d8 lr : 00000000fbd786a4
   [..]

The compiler emits a "stur x1, [x0, #12]". x1 is might just be 32 bit
aligned pointer. Remove the unused u64 element from the union to drop
the minimal alignment to 32 bit. Also remove the union, because it is
no more needed.

Signed-off-by: Michael Walle <michael@walle.cc>
Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>

I2C: ls1043a, ls1046a: enable SYS_I2C_MXC

This enables SYS_I2C_MXC to fix a bug that
failed to boot from sd card with
image u-boot-with-spl-pbl.bin

Signed-off-by: Biwen Li <biwen.li@nxp.com>
Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>

armv8: layerscape: rework spin table

There are two issues:

 (1) The spin table doesn't convert the endianness of the jump address.
     Although there is code for it, the result isn't used at all (x0).
 (2) If something goes wrong, the function returns. But that doesn't
     make sense at all.

Use the actual converted jump address as destination to fix. If
there is an error, jump to a trap loop. And rearrange the code exception
level switching code to make it smaller and clearer.

This reduces the size of the spin table code section from 696 bytes to
424 bytes. If CONFIG_ARMV8_SWITCH_TO_EL1 the code size reduced from 696
bytes to 632 bytes.

Signed-off-by: Michael Walle <michael@walle.cc>
Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>

armv8: layerscape: relocate spin table if EFI_LOADER is enabled

On ARM64, a 64kb region is reserved for the runtime services code.
Unfortunately, this code overlaps with the spin table code, which also
needs to be reserved. Thus now that the code is relocatable, allocate a
new page from EFI, copy the spin table code into it, update any pointers
to the old region and the start the secondary CPUs.

Signed-off-by: Michael Walle <michael@walle.cc>
Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>

armv8: layerscape: clean exported symbols in spintable.S

Add a new variable secondary_boot_code_start, which holds a pointer to
the start of the spin table code. This will help to relocate the code
section. While at it, move the size variable from the end to the
beginning so there is a common section for the variables. Remove any
other symbols.

Signed-off-by: Michael Walle <michael@walle.cc>
Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>

armv8: layerscape: drop first .ltorg directive in spintable.S

Now that the spin table is in a separate module, this is no longer
necessary. Drop it.

Signed-off-by: Michael Walle <michael@walle.cc>
Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>

armv8: layerscape: make wake_secondary_core_n() static

This function is not used outside the module. Make it static.

Signed-off-by: Michael Walle <michael@walle.cc>
Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>

armv8: layerscape: simplify get_spin_tbl_addr() calls

There is no need to cast around. Assign the address to the local
variable and use it.

Signed-off-by: Michael Walle <michael@walle.cc>
Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>

armv8: layerscape: remove determine_mp_bootpg()

Only the PowerPC architecture needs this function. Remove it.

Signed-off-by: Michael Walle <michael@walle.cc>
Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>

armv8: layerscape: fix alignment for spin table

Fix the alignment so it will match the comments. The spin table has to
be 8 byte aligned, so ".align 3" is enough.

Signed-off-by: Michael Walle <michael@walle.cc>
Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>

armv8: layerscape: load function pointer using ADR

Don't use LDR to load a pointer to a function. This will generate a
literal which cannot be relocated. Use ADR which is PC-relative and
therefore can easily be relocated.

Signed-off-by: Michael Walle <michael@walle.cc>
Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>

armv8: layerscape: move spin table into own module

Move it out of lowlevel.S into spintable.S. On layerscape, the secondary
CPUs are brought up in main u-boot. This will make it possible to only
compile the spin table code for the main u-boot and omit it in SPL.

This saves about 720 bytes in the SPL.

Signed-off-by: Michael Walle <michael@walle.cc>
Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>

armv8: layerscape: properly use CPU_RELEASE_ADDR

The generic armv8 code already has support to bring up the secondary
cores. Thus, don't hardcode the jump in the layerscape lowlevel_init to
the spin table code; instead just return early and let the common armv8
code handle the jump. This way we can actually use the CPU_RELEASE_ADDR
feature.

Signed-off-by: Michael Walle <michael@walle.cc>
[Rebased, Removed kontron_sl28.h change as file does not exist]
Signed-off-by: Priyanka Jain <priyanka.jain@nxp.com>

armv8: layerscape: pretty print info about SMP cores

Make the print of the starting address a debug output and pretty print
the info about online cores.

Signed-off-by: Michael Walle <michael@walle.cc>
Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>

armv8: layerscape: fix spin-table support

Spin tables are broken with bootefi. This is because - in contrast to
the booti call chain - there is no call to smp_kick_all_cpus(). Due to
this missing call the secondary CPUs are never released from their "wait
for interrupt state", see secondary_boot_func() in lowlevel.S.

Originally, this "wait for interrupt" is there to make sure, the spin
table is cleared before the secondary cores read it for the first time.
But the boot flow for the layerscape architecture is different from
that. The CPUs are release from their BootROM _after_ U-Boot's
spin-table is cleared, see fsl_layerscape_wake_seconday_cores() in mp.c.
Thus, there is no need to wait for this interrupt and no need for
kicking all cores on cpu_release. An atomic 64bit write to the
spin-table and a "sev" is sufficient.

Signed-off-by: Michael Walle <michael@walle.cc>
Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>

net: pfe_eth: Use spi_flash_read API to access flash memory

Current PFE firmware access spi-nor memory directly. New spi-mem
framework does not support direct memory access. So, let's use
spi_flash_read API to access memory instead of directly using it.

Signed-off-by: Kuldeep Singh <kuldeep.singh@nxp.com>
Reviewed-by: Frieder Schrempf <frieder.schrempf@kontron.de>
Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>

dm: armv8: gpio: include <asm/arch/gpio.h> for fsl-layerscape

Enable the gpio feature on fsl-layerscape platform.

Signed-off-by: hui.song <hui.song_1@nxp.com>
Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>

armv8: gpio: add gpio feature

add one struct mpc8xxx_gpio_plat to enable gpio feature.

Signed-off-by: hui.song <hui.song_1@nxp.com>
Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>

configs: ls1088ardb: Add support for usb boot target

LS1088A-RDB has MMC, SCSI, DHCP as boot targets,
but the USB support was missing.
Add support for USB as Boot_targets_devices.

Signed-off-by: Era Tiwari <era.tiwari@nxp.com>
Signed-off-by: Pramod Kumar <pramod.kumar_1@nxp.com>
[Rebased]
Signed-off-by: Priyanka Jain <priyanka.jain@nxp.com>

armv8: ls1028a: move FSL_LAYERSCAPE to kconfig

CONFIG_FSL_LAYERSCAPE is available in kconfig. There is no need to
define it per board; the ls1028a_common.h is really board dependent and
only fits to the NXP eval boards. Instead select CONFIG_FSL_LAYERSCAPE
when ARCH_LS1028A is selected.

Signed-off-by: Michael Walle <michael@walle.cc>
Reviewed-by: Priyanka Jain <priyanka.jain@nxp.com>

fsl_dspi: Introduce DT bindings for CS-SCK and SCK-CS delays

Communication with some SPI slaves just won't cut it if these delays
(before the beginning, and after the end of a transfer) are not added to
the Chip Select signal.

These are a straight copy from Linux:
Documentation/devicetree/bindings/spi/spi-fsl-dspi.txt
drivers/spi/spi-fsl-dspi.c

Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
[Rebased]
Signed-off-by: Priyanka Jain <priyanka.jain@nxp.com>

Merge https://gitlab.denx.de/u-boot/custodians/u-boot-riscv

- Fix SiFive HiFive Unleashed board booting failure problem.
- Enable SiFive fu540 PWM driver.
- Support SiFive fu540: SPI boot.
- Update OpenSBI used for RISC-V CI testing.
- Revert "riscv: Allow use of reset drivers".
- Revert "Revert "riscv: sifive: fu540: Add gpio-restart support"".
- sysreset: syscon:
        - Don't assume default value for offset and mask property.
        - Support value property.
- qemu: Add syscon reboot and poweroff support.
- Fix SIFIVE debug serial dependency.
- Fix linking error when building u-boot-spl with no SMP support.
- AE350 use fdtdec_get_addr_size_auto_noparent to parse smc reg.
- Make memory node available to SPL in hifive-unleashed-a00-u-boot.dtsi
- SiFive fu540 avoid using hardcoded ram base and size.

Revert "Merge tag 'dm-pull-20jul20' of git://git.denx.de/u-boot-dm"

This reverts commit 5d3a21df6694ebd66d5c34c9d62a26edc7456fc7, reversing
changes made to 56d37f1c564107e27d873181d838571b7d7860e7.

Unfortunately this is causing CI failures:
https://travis-ci.org/github/trini/u-boot/jobs/711313649

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

ram: sifive: Avoid using hardcoded ram base and size

At present the SiFive FU540 RAM driver uses hard-coded memory base
address and size to initialize the DDR controller. This may not be
true when this driver is used on another board based on FU540.

Update the driver to read the memory information from DT and use
that during the initialization.

Signed-off-by: Bin Meng <bin.meng@windriver.com>
Reviewed-by: Leo Liang <ycliang@andestech.com>
Reviewed-by: Pragnesh Patel <pragnesh.patel@sifive.com>

riscv: dts: hifive-unleashed-a00: Make memory node available to SPL

Make memory node available to SPL in prepration to updates to SiFive
DDR RAM driver to read memory information from DT.

Signed-off-by: Bin Meng <bin.meng@windriver.com>
Reviewed-by: Pragnesh Patel <pragnesh.patel@sifive.com>

riscv: ae350: Use fdtdec_get_addr_size_auto_noparent to parse smc reg

Use fdtdec_get_addr_size_auto_noparent to read the "reg" property
instead of fdtdec_get_addr. This will increase the compatibility
of dtb parsing.

Signed-off-by: Rick Chen <rick@andestech.com>
Acked-by: Leo Liang <ycliang@andestech.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Tested-by: Leo Liang <ycliang@andestech.com>

riscv: Fix linking error when building u-boot-spl with no SMP support

Switch off SMP support when building u-boot-spl would cause linking error as follow:
undefined reference to 'secondary hart relocate' and 'smp_call_function'.
Add macro to wrap up proper code region that needs SMP configuration on.

Signed-off by: Leo Liang <ycliang@andestech.com>
Cc: rick@andestech.com
Reviewed-by: Bin Meng <bin.meng@windriver.com>

serial: Fix SIFIVE debug serial dependency

The commit 4cc24aeaf420 ("serial: Add missing Kconfig dependencies for
debug consoles") has added incorrect dependency for SIFIVE debug uart which
should depend on SIFIVE driver instead of PL01x.

Fixes: 4cc24aeaf420 ("serial: Add missing Kconfig dependencies for debug consoles")
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bin.meng@windriver.com>
Reviewed-by: Leo Liang <ycliang@andestech.com>
Reviewed-by: Sean Anderson <seanga2@gmail.com>

riscv: qemu: Add syscon reboot and poweroff support

This adds syscon reboot and poweroff support to QEMU RISC-V.

Signed-off-by: Bin Meng <bin.meng@windriver.com>
Reviewed-by: Pragnesh Patel <pragnesh.patel@sifive.com>

sysreset: syscon: Support value property

Per the DT binding, <mask> and <value> property can have either one
or both, and if <value> is missing, <mask> should be used, which is
what current U-Boot sysreset_syscon driver supports.

This adds support to the <value> property to the driver, and <mask>
semantics is updated to really be a mask to the value if both exist.

Signed-off-by: Bin Meng <bin.meng@windriver.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Pragnesh Patel <pragnesh.patel@sifive.com>

sysreset: syscon: Don't assume default value for offset and mask property

Per the DT binding, <offset> is a required property. Let's abort
the probe if it is missing. For the <mask> property, current codes
assume a default value of zero, which is not correct either.

Signed-off-by: Bin Meng <bin.meng@windriver.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Pragnesh Patel <pragnesh.patel@sifive.com>

Revert "Revert "riscv: sifive: fu540: Add gpio-restart support""

This reverts commit 23da3c682a84a2ad67a67287979dd4f5259ff607.

Now the build failure of sifive_fu540_defconfig board has been fixed,
revert this "revert patch".

Signed-off-by: Bin Meng <bin.meng@windriver.com>

Revert "riscv: Allow use of reset drivers"

This reverts commit 958a3f464c7f8ef7e10db9feb663e9e80445ce2f.

A more appropriate change below is already in mainline.
Commit fd31e4fd184f ("riscv: Do not build reset.c if SYSRESET is on")

Revert this patch, so that U-Boot can be built successfully for
SiFive Fu540 board.

Signed-off-by: Bin Meng <bin.meng@windriver.com>
Reviewed-by: Sean Anderson <seanga2@gmail.com>
Reviewed-by: Leo Liang <ycliang@andestech.com>

azure: gitlab: travis: Update OpenSBI used for RISC-V testing

Change to use OpenSBI release v0.8 generic platform images for QEMU
RISC-V CI testing for azure, gitlab and travis-ci.

Signed-off-by: Bin Meng <bin.meng@windriver.com>
Reviewed-by: Tom Rini <trini@konsulko.com>

sifive: fu540: Enable SF distro bootcmd

Enable SPI flash(SF) distro boot command in Sifive FU540.

This distro boot will read the boot script at specific
location at the flash and start sourcing the same.

Included the SF device at the last of the target devices
list since all the rest of the devices on the list have
more possibility to boot the distribution due to the
size of the SPI flash is concern.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Reviewed-by: Bin Meng <bin.meng@windriver.com>

sifive: fu540: Add boot flash script offset, size

HiFive-Unleashed-A00 has SPI flash with 32MiB size.
So, let's use the script offset at the end of 4K.
This way it cannot overlap any offsets being used
by software components in flash layout.

So, SF distrocmd will pick the script at desired
script address and run.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Reviewed-by: Bin Meng <bin.meng@windriver.com>

sifive: fu540: Mark the default env as SPI flash

Mark the default U-Boot environment as SPI flash since
this is an on board flash device.

Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Reviewed-by: Bin Meng <bin.meng@windriver.com>
Tested-by: Bin Meng <bin.meng@windriver.com>

env: Enable SPI flash env for SiFive FU540

SPI flash device on HiFive Unleashed has 32MiB Size.

This patch adds SPI flash environment after U-Boot proper
partition with a size of 128KiB.

SPI flash partition layout(32MiB):
    0 - 34 : reserved for GPT header
   35 - 39 : unused
   40 - 2087 : loader1 (SPL, FSBL)
 2088 - 10279 : loader2 (U-Boot proper, U-Boot)
10280 - 10535 : environment
10536 - 65494 : rootfs
65528 - 65536 : distro script

Note: the loader1 must start from 40th sector even though
there are 6 free sectors prior since 40th sector is nearest
flash sector boundary. 

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Reviewed-by: Bin Meng <bin.meng@windriver.com>
Tested-by: Bin Meng <bin.meng@windriver.com>

sifive: fu540: Add Booting from SPI

Add booting from SPI for SiFive Unleashed board.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Reviewed-by: Bin Meng <bin.meng@windriver.com>
Tested-by: Bin Meng <bin.meng@windriver.com>

sifive: fu540: Add runtime boot mode detection

Add support to detect boot mode at runtime for
SiFive FU540 boards.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Reviewed-by: Bin Meng <bin.meng@windriver.com>
Tested-by: Bin Meng <bin.meng@windriver.com>

riscv: sifive: fu540: Enable SiFive PWM driver

This patch enables SiFive PWM driver for the SiFive
Unleashed board.

Signed-off-by: Pragnesh Patel <pragnesh.patel@sifive.com>
Reviewed-by: Bin Meng <bin.meng@windriver.com>

riscv: Make SiFive HiFive Unleashed board boot again

Commit 40686c394e53 ("riscv: Clean up IPI initialization code")
caused U-Boot failed to boot on SiFive HiFive Unleashed board.

The codes inside arch_cpu_init_dm() may call U-Boot timer APIs
before the call to riscv_init_ipi(). At that time the timer register
base (e.g.: the SiFive CLINT device in this case) is unknown yet.

It might be the name riscv_init_ipi() that misleads people to only
consider it is related to IPI, but in fact the timer capability is
provided by the same SiFive CLINT device that provides the IPI.
Timer capability is needed for both UP and SMP.

Considering that the original refactor does have benefits, that it
makes the IPI code more similar to U-Boot initialization idioms.
It also removes some quite ugly macros. Let's do the minimal revert
instead of a complete revert, plus a fixes to arch_cpu_init_dm() to
consider the SPL case.

Fixes: 40686c394e53 ("riscv: Clean up IPI initialization code")
Signed-off-by: Bin Meng <bin.meng@windriver.com>
Reviewed-by: Sean Anderson <seanga2@gmail.com>
Tested-by: Leo Liang <ycliang@andestech.com>

Merge tag 'dm-pull-20jul20' of git://git.denx.de/u-boot-dm

binman support for FIT
new UCLASS_SOC
patman switch 'test' command
minor fdt fixes
patman usability improvements

Merge tag 'efi-2020-10-rc1-5' of https://gitlab.denx.de/u-boot/custodians/u-boot-efi

Pull request for UEFI sub-system for efi-2020-10-rc1 (5)

The series provides bug fixes for:

* crash in OS when accessing UEFI variables
* returning from UEFI fit images to U-Boot
* error handling for variable services provided by OP-TEE
* error handling in EFI_FILE_PROTOCOL.Read()
* missing function documentation

The first patches needed to use intermediate certificates for
secure boot are added. (The rest of the series requires
updating sbsigntool in our CI systems.)

Logging is enabled in the bootefi command.

Merge tag 'u-boot-rockchip-20200722' of https://gitlab.denx.de/u-boot/custodians/u-boot-rockchip

- New rk3326 board: Hardkernel Odroid Go2;
- Update board config and dts for RockPI 4/N8/N10;
- Update led boot on support for roc-rk3399-pc;
- Enable SPI Flash suppor for rk3328 rock64 board;
- Update rockchip pcie phy to use generic framework;

rockchip: rock-p-n8/n10: Disable RAM_ROCKCHIP_DEBUG

Disable ram rockchip debug driver for ROCKPi N8/N10 boards
since we have verified ram in many instances with respective
U-Boot versions.

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

arm64: dts: rockchip: Add u-boot, spl-boot-order for ROCKPi N10

Add u-boot,spl-boot-order for ROCKPi N10, so-that it can able
to boot from eMMC and SDMMC in order.

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

rockchip: rock-pi-n10: Enable PCI/NVME

Enable PCI/NVME for M.2 Slot on RockPI-4 boards.

=> nvme info
Device 0: Vendor: 0x144d Rev: 4L1QCXB7 Prod: S35FNX0J623292
            Type: Hard Disk
            Capacity: 122104.3 MB = 119.2 GB (250069680 x 512)

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

rockchip: rock-pi-n8: Enable on board devices

Enable common on board devices for ROCKPi N8.

- USB 2.0 Host
- USB 2.0 OTG/Gadget
- HDMI Out

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

rockchip: evb_rk3288: Add console settings

Add console settings like stdin, stdout and stderr as
usbkbd and vidconsole respectively for evb-rk3288 targets.

This would certainly help to detect the attached video
devices (like HDMI) and print the console messages on display.

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

rockchip: rock-pi-n10: Enable on board devices

Enable common on board devices for ROCKPi N10.

- USB 2.0 Host
- USB 3.0 Host
- USB 3.0 Gadget
- HDMI Out

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

arm64: dts: rockchip: Add PCIe for RockPI N10

This patch adds support to enable PCIe for RockPI N10.

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

arm: dts: rockchip: Add HDMI out for RockPI N8/N10

This patch adds support to enable HDMI out for
N10 and N8 combinations SBCs.

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

ARM: dts: rockchip: Add USB for RockPI N8/N10

Radxa dalang carrier board has 2x USB 2.0 and 1x USB 3.0
ports.

This patch adds support to enable all these USB ports for
N10 and N8 combinations SBCs.

Note that the USB 3.0 port on RockPI N8 combination works
as USB 2.0 OTG since it is driven from RK3288.

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

ARM: dts: rockchip: Add usb host0 ohci node for rk3288

rk3288 and rk3288w have a usb host0 ohci controller.

Although rk3288 ohci doesn't actually work on hardware, but
rk3288w ohci can work well.

So add usb host0 ohci node in rk3288 dtsi and the quirk in
ohci platform driver will disable ohci on rk3288.

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

arm: dts: rockchip: Sync rk3288 dtsi from linux

This sync has changes required to use HDMI CEC pin in U-Boot.

Sync dts from linux v5.8-rc5 commit:
"ARM: dts: rockchip: define the two possible rk3288 CEC pins"
(sha1: 838980dd04e994bf81cf104fa01ae60802146b39)

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

ARM: dts: rockchip: Sync ROCKPi N8/N10 dts(i) from Linux

Sync ROCKPi N8/N10 dts(i) changes from Linux.

commit <afd9eb880414> ("ARM: dts: rockchip: Add Radxa Rock Pi N8 initial
support")

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

rockchip: roc-rk3399-pc: Set LED only during POR and pwr_key=y

ROC-RK3399-PC has specific set of configurations for
on-board led setup.

Due to easiness for user to know the state of the board
roc-rk339-pc board code will setup the low power led
on/off, and waiting for user to press power key and then
glow full power led.

All this needs to happen only during power-on-reset not
for soft reset or WDT.

Also, it is not a proper usage to ask the user to press
the Power key if the board connected remotely, so add
the environment variable 'pwr_key' to check as well.

So, user need to press Power key only
- during POR
- pwr_key=y

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

rockchip: Separate the reset cause from display cpuinfo

reset cause is a generic functionality based on the soc
cru registers in rockchip. This can be used for printing
the cause of reset in cpuinfo or some other place where
reset cause is needed. 

Other than cpuinfo, reset cause can also be using during
bootcount for checking the specific reset cause and glow
the led based on the reset cause.

So, let's separate the reset cause code from cpuinfo, and
add a check to build it for rk3399, rk3288 since these two
soc are supporting reset cause as of now.

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

rockchip: Don't clear the reset status reg

reset reason can be used several stages of U-Boot bootloader
like SPL, U-Boot proper based on the requirements.

Clearing the status register end of get_reset_cause will end
up showing the wrong reset cause when it read the second time.
For example, if board resets, SPL reads the reset status as
RST whereas U-Boot proper reads the status as POR.

However, based on the latest testing clearing reset status
won't be required for determine the last reset cause or
following resets.

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

rockchip: tpl: Remove board_early_init_f()

There is no need for board_early_init_f() in TPL, anything like this
should goes to SPL.

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

rockchip: roc-rk3399-pc: Add custom led_setup()

roc-rk3399-pc has some specific requirements to support LEDS,
environment. board detection and etc prior to U-Boot proper.

So as of now SPL would be a better stage for these custom board
requirements to support unlike TPL. Adding few of these custom
requirements like LEDS in TPL would require extra code pulling
and also the size of TPL can grow.

So, this patch moves the leds code from TPL into SPL Board init
led_setup code.

Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
(split tpl.c change as separate patch)
Signed-off-by: Kever Yang <kever.yang@rock-chips.com>

rockchip: rk3399: Add weak led_setup()

Add weak led_setup() so that board which has an uncommon
led setup code that can make use of custom implementation.

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

rockchip: rk3288: Add OF board setup

The new rk3288 revision rk3288w has some changes with respect
to legacy rk3288 like hclk_vio in cru and usb host0 ohci.

Linux clock driver already handle this via rockchip,rk3288w-cru
compatible.

USB ohci host can enable via dts for rk3288w based boards.

So, add fdt board setup code to update cru compatible with
rk3288w-cru compatible if the SOC revision is RK3288W.

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

rockchip: Add rk3288 SoC detection helper

Rockchip SoC's has a new revision chip for rk3288 SoCs.

RK3288 has a new revision chip called RK3288W which is similar
but different hclk_vio clock and fixed OHCI host.

Add common Rockchip SoC detection helper to support this rk3288w
detection.

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

arm64: dts: rockchip: Add Radxa ROCK Pi 4C support

Rock PI 4C has AP6256 Wifi/BT, PoE, miniDP, USB Host enabled
GPIO pin change compared to 4B, 4C.

So, add or enable difference nodes/properties in 4C dts
by including common dtsi.

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