sf: Remove unneeded SST_BP and SST_WP

SST parts added on sf_params.c supports both SST_WR which consits
of both BP and WP and there is a spi controller ich which supports
only BP so the relevent _write hook set based on "slave->op_mode_tx"
hence there is no respective change required from flash side hance
removed these.

Tested-by: Bin Meng <[email protected]>
Reviewed-by: Bin Meng <[email protected]>
Signed-off-by: Jagan Teki <[email protected]>

sf: Remove unneeded header includes

Removed unneeded header includes in sf_ops and sf_probe

Reviewed-by: Bin Meng <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
Signed-off-by: Jagan Teki <[email protected]>

sf: Flash power up read-only based on idcode0

Using macro's for flash power up read-only access code
leads wrong behaviour hence use idcode0 for runtime
detection, hence the flash which require this functionality
gets detected at runtime.

Tested-by: Jagan Teki <[email protected]>
Reviewed-by: Bin Meng <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
Signed-off-by: Jagan Teki <[email protected]>

sf: Use simple name for register access functions

Most of the register access function are static,
so used simple name to represent each.

Tested-by: Jagan Teki <[email protected]>
Reviewed-by: Bin Meng <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
Signed-off-by: Jagan Teki <[email protected]>

sf: Fix Makefile

This patch removes unneeded ifdef and fixed accordingly.

Reviewed-by: Bin Meng <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
Signed-off-by: Jagan Teki <[email protected]>

sf: Use static for file-scope functions

Use static for file-scope functions and removed
them from header files.

Tested-by: Jagan Teki <[email protected]>
Reviewed-by: Bin Meng <[email protected]>
Signed-off-by: Jagan Teki <[email protected]>

sf: probe: Code cleanup

- Move bar read code below the bar write hance both
  at once place, hence it easy for #ifdef macro only
  once and readable.
- Move read_cmd_array at top

Tested-by: Jagan Teki <[email protected]>
Reviewed-by: Bin Meng <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
Signed-off-by: Jagan Teki <[email protected]>

sf: Move read_id code to sf_ops

read_id code is related to spi_flash stuff
hence moved to sf_ops.

Tested-by: Jagan Teki <[email protected]>
Tested-by: Bin Meng <[email protected]>
Reviewed-by: Bin Meng <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
Signed-off-by: Jagan Teki <[email protected]>

sf: Move spi_flash_scan code to sf_ops

Intension is that sf_ops should deals all spi_flash
related stuff and sf_probe (which should renamed future)
should be an interface layer for spi_flash versus spi drivers.

sf_ops => spi_flash interface
sf_probe => interface layer vs spi_flash(sf_probe) to spi drivers

Tested-by: Jagan Teki <[email protected]>
Tested-by: Bin Meng <[email protected]>
Reviewed-by: Bin Meng <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
Signed-off-by: Jagan Teki <[email protected]>

sf: spi_flash_validate_params => spi_flash_scan

Rename spi_flash_validate_params to spi_flash_scan
as this code not only deals with params setup but
also configure all spi_flash attributes.

And also moved all flash related code into
spi_flash_scan for future functionality addition.

Tested-by: Jagan Teki <[email protected]>
Tested-by: Bin Meng <[email protected]>
Reviewed-by: Bin Meng <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
Signed-off-by: Jagan Teki <[email protected]>

spi: Kconfig: Fix correct target name for ZynqMP

ZynqMP is using different symbol.
Use correct one.

Reviewed-by: Jagan Teki <[email protected]>
Signed-off-by: Michal Simek <[email protected]>

spi: zynq_spi: Add cadence compatible string

Extend compatible list table for cdns,spi-r1p6 compatible string.

Reviewed-by: Jagan Teki <[email protected]>
Signed-off-by: Michal Simek <[email protected]>

imx: mx7dsabresd: Add QSPI support

Support qspi flashes for mx7dsabresd
1. introduce pin mux settings
2. enable qspi clock
3. introduce related macro definitions

Default QSPI is not enabled, since we need hardware rework to use QSPI,
see SPF-28590, page 9:
"
QSPI signals are muxed with EPDC_D[7:0]
When using QSPI: de-populate R388-R391, R396-R399
populate R392-R395, R299, R300
"

After hardware rework, define CONFIG_FSL_QSPI in mx7dsabresd.h. qspi
flashes can be deteced and read/erase/write. Log info:
"
=> sf probe
SF: Detected MX25L51235F with page size 256 Bytes, erase size 64 KiB, total 64 MiB
=> sf read 0x80000000 0 0x4000000
device 0 whole chip
SF: 67108864 bytes @ 0x0 Read: OK
=> sf erase 0 0x4000000
SF: 67108864 bytes @ 0x0 Erased: OK
=> sf write 0x80000000 0 0x4000000
device 0 whole chip
SF: 67108864 bytes @ 0x0 Written: OK
"

Cc: Fabio Estevam <[email protected]>
Cc: Adrian Alonso <[email protected]>
Reviewed-by: Stefano Babic <[email protected]>
Reviewed-by: Jagan Teki <[email protected]>
Signed-off-by: Peng Fan <[email protected]>

spi: ti_qspi: Fix SPI_3WIRE checking using mode

SPI_3WIRE is spi mode not spi flags, so this patch fixed
the spi-3wire checking throgh mode instead of flags.

Cc: Mugunthan V N <[email protected]>
Acked-by: Simon Glass <[email protected]>
Signed-off-by: Jagan Teki <[email protected]>

spi: Get spi-3wire from dts

spi-3wire is used when SI/SO signals shared so get
the same from dts node and assign to mode on slave
plat->mode.

Acked-by: Simon Glass <[email protected]>
Signed-off-by: Jagan Teki <[email protected]>

x86: Remove HAVE_ACPI_RESUME

These are currently dead codes. Until we have complete ACPI support,
we don't know if it works or not. Remove to avoid confusion.

Signed-off-by: Bin Meng <[email protected]>
Acked-by: Simon Glass <[email protected]>
Tested-by: Simon Glass <[email protected]>

x86: Remove CPU_INTEL_SOCKET_RPGA989

This Kconfig option name indicates it has something to do with cpu
socket, however it is actually not the case. Remove it and move
options inside it to NORTHBRIDGE_INTEL_IVYBRIDGE.

Signed-off-by: Bin Meng <[email protected]>
Acked-by: Simon Glass <[email protected]>
Tested-by: Simon Glass <[email protected]>

x86: Clean up ivybridge/chrome Kconfig options

There are some options which are never used, and also some options
which are selected by others but have never been a Kconfg option.
Clean these up.

Signed-off-by: Bin Meng <[email protected]>
Acked-by: Simon Glass <[email protected]>
Tested-by: Simon Glass <[email protected]>

x86: ivybridge: Remove NORTHBRIDGE_INTEL_SANDYBRIDGE

NORTHBRIDGE_INTEL_SANDYBRIDGE is for sandybridge, not ivybridge.

Signed-off-by: Bin Meng <[email protected]>
Acked-by: Simon Glass <[email protected]>
Tested-by: Simon Glass <[email protected]>

common: Remove timer_init() call for x86

With driver model timer support, there should not be an explict
call to timer_init(). Remove this call for x86.

Signed-off-by: Bin Meng <[email protected]>
Acked-by: Simon Glass <[email protected]>

x86: Move i8254_init() to x86_cpu_init_f()

Right now i8254_init() is called from timer_init() in the tsc timer
driver. But actually i8254 and tsc are completely different things.
Since tsc timer has been converted to driver model, we should find
a new place that is appropriate for U-Boot to call i8254_init(),
which is now x86_cpu_init_f().

Signed-off-by: Bin Meng <[email protected]>
Acked-by: Simon Glass <[email protected]>

x86: Fix PCI UART compatible string for crownbay and galileo

With recent ns16550 driver changes, we only changed the legacy UART
(at I/O port 0x3f8) compatible string, but forgot to change the PCI
UART compatible string. Now fix it.

Signed-off-by: Bin Meng <[email protected]>
Acked-by: Simon Glass <[email protected]>

fdt: Change OF_BAD_ADDR to FDT_ADDR_T_NONE

Currently OF_BAD_ADDR is always -1ULL. When using OF_BAD_ADDR as the
return value of dev_get_addr(), it creates potential size mismatch
as dev_get_addr() uses FDT_ADDR_T_NONE as the return value which can
be either -1U or -1ULL depending on CONFIG_PHYS_64BIT. Now we change
OF_BAD_ADDR to FDT_ADDR_T_NONE to avoid such case.

Signed-off-by: Bin Meng <[email protected]>
Acked-by: Simon Glass <[email protected]>
Reviewed-by: Stefan Roese <[email protected]>

axs10x: add support of generic EHCI USB 2.0 controller

This commit adds support of USB 2.0 storage media on AXS10x boards.

Signed-off-by: Alexey Brodkin <[email protected]>
Cc: Marek Vasut <[email protected]>
Cc: Simon Glass <[email protected]>

Prepare v2016.01-rc2

Signed-off-by: Tom Rini <[email protected]>

mkimage: Fix warning from fix for generating multi and script images again

Seems 6ae6e160 broke creating images in certain cases, there
are two problems with that patch.

First is that the expression "!x == 4 || !x == 6" is ambiguous.  The
intention here was "!(x == 4) || !(x == 6)" based on reading further in
the file, where this was borrowed from.  This however is interpreted by
gcc as "(!x) == 4 || (!x) == 6" and always false.  gcc-5.x will warn
about this case.

The second problem is that we do not want to test for the case of "(NOT x
is 4) OR (NOT x is 6)" but instead "(x is not equal to 4) AND (x is not
equal to 6)".  This is because in those two cases we already execute the
code question in another part of the file.  Rewrite the expression and
add parenthesis for clarity.

Signed-off-by: Marek Vasut <[email protected]>
Cc: Tom Rini <[email protected]>
Cc: Philippe De Swert <[email protected]>
Cc: Simon Glass <[email protected]>
[trini: Re-word Marek's explanation]

CONFIG_NEEDS_MANUAL_RELOC: Fix warnings when not set

Now that we may compile (but not link) code calling fixup_cmdtable when
this is not set, we need to always have the declaration available.  We
should also make sure that anyone calling the function includes
<command.h> as that's where the function declaration is.

Signed-off-by: Tom Rini <[email protected]>

Merge branch 'zynq' of git://denx.de/git/u-boot-microblaze

dm: core: Enable SPL_SIMPLE_BUS by default

This option is needed for all SoCs which have nodes on bus. Without
enabling this drivers are not found and probed.
Issue was found on Zynq MMC probe.
Enable this option by default.

Signed-off-by: Michal Simek <[email protected]>
Reviewed-by: Stefan Roese <[email protected]>
Reviewed-by: Simon Glass <[email protected]>

dm: cmd: Relocate subcommands when MANUAL_RELOC

Subcommands contain pointers to functions which are not updated when
MANUAL_RELOC is enabled. This patch fix it.

Signed-off-by: Michal Simek <[email protected]>
Reviewed-by: Simon Glass <[email protected]>

i2c: cmd: Relocate subcommands when MANUAL_RELOC

Subcommands contain pointers to functions which are not updated when
MANUAL_RELOC is enabled. This patch fix it.

Signed-off-by: Michal Simek <[email protected]>
Acked-by: Heiko Schocher <[email protected]>
Reviewed-by: Simon Glass <[email protected]>

serial: zynq: Remove duplicated header

debug_uart.h is included twice.

Signed-off-by: Michal Simek <[email protected]>
Reviewed-by: Bin Meng <[email protected]>
Reviewed-by: Simon Glass <[email protected]>

serial: zynq: Fix incorrect reference to s5p driver

Remove this c&p error from s5p driver.

Signed-off-by: Michal Simek <[email protected]>
Reviewed-by: Bin Meng <[email protected]>
Reviewed-by: Simon Glass <[email protected]>

zynq: sdhci: Move driver to DM

Move driver to DM

Signed-off-by: Michal Simek <[email protected]>
Reviewed-by: Simon Glass <[email protected]>

zynq: sdhci: Remove zynq_sdhci_of_init()

Prepare for using DM.

Signed-off-by: Michal Simek <[email protected]>

zynq: sdhci: Calculate minimum frequency based on max frequency

Calculate the minimum sd clock based on max clock. This will
be done by add_sdhci() if we pass minimum clock as zero.
It also does based on SD host contoller version.

Signed-off-by: Siva Durga Prasad Paladugu <[email protected]>
Signed-off-by: Michal Simek <[email protected]>

ARM: zynq: Remove unused SERIAL macros for serial_zynq

Remove unused macros when driver was moved to DM.

Signed-off-by: Michal Simek <[email protected]>

ARM: zynq: Extend malloc size

SPL DM MMC FAT requires more malloc space(3k fat buffers + dm)
that it is available now. Extend SPL malloc space.

Current OCM layout:
0xffff0000 - 0xfff2000 - Full malloc space
0xffff2000 - 0xffff300 - Stack location
0xfffff300 - CONFIG_SYS_MALLOC_F_LEN - Early malloc space
0xfffffd00 - sizeof(GD) - GD
0xfffffe00 - 0xffffffff - SoC specific boot code

Signed-off-by: Michal Simek <[email protected]>

ARM: zynq: Fix location of stack and malloc areas

The patch
"board_init: Change the logic to setup malloc_base"
(sha1: 9ac4fc82071ce346e3885118242ff45d22f69b82)
breaks SPL for Zynq because it puts early alloc area on the stack which
caused that stack was decreased by CONFIG_SYS_MALLOC_F_LEN (0x400)
and there was not enough space for regular stack.

This patch changes memory layout to better utilize the last 64k OCM
block.
0xffff0000 - 0xfff1000 - Full malloc space
0xffff1000 - 0xffff300 - Stack location
0xfffff300 - CONFIG_SYS_MALLOC_F_LEN - Early malloc space
0xfffffd00 - sizeof(GD) - GD
0xfffffe00 - 0xffffffff - SoC specific boot code

Signed-off-by: Michal Simek <[email protected]>
Tested-by: Moritz Fischer <[email protected]>

ARM: zynq: Enable debug console for zc702

Signed-off-by: Michal Simek <[email protected]>

ARM64: zynqmp: Enable FIT config option via Kconfig

Remove configuration options from board file.

Signed-off-by: Michal Simek <[email protected]>

net: gem: Enable CTRL+C in wait_for_bit

Enable to break waiting loop at any time.

Signed-off-by: Michal Simek <[email protected]>
Acked-by: Joe Hershberger <[email protected]>

net: gem: Move gem to Kconfig

Signed-off-by: Michal Simek <[email protected]>
Reviewed-by: Jagan Teki <[email protected]>

net: gem: Read information about interface from DT

Do not set interface via configs. Read information from DT.

Signed-off-by: Michal Simek <[email protected]>
Reviewed-by: Jagan Teki <[email protected]>

net: gem: Move driver to DM

- Enable DM_ETH by default for Zynq and ZynqMP
- Remove board_eth_init code
- Change miiphy_read function to return value instead of error code
  based on DM requirement
- Do not enable EMIO DT support by default

Signed-off-by: Michal Simek <[email protected]>
Reviewed-by: Jagan Teki <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
Reviewed-by: Bin Meng <[email protected]>

net: gem: Fix miiphy_read name

Sync it with write function.

Signed-off-by: Michal Simek <[email protected]>
Reviewed-by: Jagan Teki <[email protected]>

net: gem: Remove zynq_gem_of_init()

This function was used for OF init before DM.
Remove this function as the part of move to DM.

Signed-off-by: Michal Simek <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
Reviewed-by: Jagan Teki <[email protected]>

net: gem: Enable MDIO bus earlier

Enable access to MDIO before zynq_gem_init is called.
It enables read information about phy earlier.

Signed-off-by: Michal Simek <[email protected]>

net: gem: Check if priv->phydev is valid

Check return value.

Signed-off-by: Michal Simek <[email protected]>
Reviewed-by: Jagan Teki <[email protected]>

net: gem: Extract phy init code

Move phy init code out of zynq_gem_init. DM drivers are normally calling
this code from probe function.

Signed-off-by: Michal Simek <[email protected]>

net: gem: Remove phydev variable

Resort code to use priv->phydev variable directly.
It will simplify move to DM.

Signed-off-by: Michal Simek <[email protected]>

net: gem: Change mii function not to use eth_device structure

Next step to move driver to driver model. Do not use eth_device
structure. Use private structure instead.
Add iobase to private structure to store gem iobase.

Signed-off-by: Michal Simek <[email protected]>

net: gem: Change mdio_wait prototype to pass regs

Pass regs instead of dev because this will be chagned by
driver model.

Signed-off-by: Michal Simek <[email protected]>

net: gem: Do not continue if phy is not found

Add return value for phy detection algorithm to stop init function when
phy is not found.

Signed-off-by: Michal Simek <[email protected]>
Reviewed-by: Jagan Teki <[email protected]>

ARM: zynq: Remove CONFIG_API

CONFIG_API is causing compilation error when DM_ETH is enabled because
eth_get_dev() is not available.

Signed-off-by: Michal Simek <[email protected]>
Reviewed-by: Jagan Teki <[email protected]>

ARM: zynq: Remove PHYLIB from config to defconfig

Move PHYLIB from board config to defconfig

Signed-off-by: Michal Simek <[email protected]>
Reviewed-by: Jagan Teki <[email protected]>

Merge git://git.denx.de/u-boot-nios

Merge git://git.denx.de/u-boot-usb

arm: socfpga: Enable CONFIG_DM_MMC

Enable driver model MMC support on SoCFPGA.

Signed-off-by: Marek Vasut <[email protected]>
Cc: Chin Liang See <[email protected]>
Cc: Dinh Nguyen <[email protected]>
Cc: Pantelis Antoniou <[email protected]>
Cc: Simon Glass <[email protected]>
Cc: Tom Rini <[email protected]>

mmc: dwmmc: socfpga: Convert to DM

Convert the SoCFPGA shim for registering the DWMMC driver to DM.

Signed-off-by: Marek Vasut <[email protected]>
Cc: Chin Liang See <[email protected]>
Cc: Dinh Nguyen <[email protected]>
Cc: Pantelis Antoniou <[email protected]>
Cc: Simon Glass <[email protected]>
Cc: Tom Rini <[email protected]>

arm: socfpga: Remove cpu_mmc_init()

This function triggers the registration of the dwmmc driver on SoCFPGA,
but this is not needed in case the driver is correctly probed from DT.

Signed-off-by: Marek Vasut <[email protected]>
Cc: Chin Liang See <[email protected]>
Cc: Dinh Nguyen <[email protected]>
Cc: Pantelis Antoniou <[email protected]>
Cc: Simon Glass <[email protected]>
Cc: Tom Rini <[email protected]>

arm: socfpga: Add SoCFPGA SR1500 board

The SR1500 board is a CycloneV based board, similar to the EBV
SoCrates, equipped with the following devices:

- SPI NOR
- eMMC
- Ethernet

Signed-off-by: Stefan Roese <[email protected]>
Reviewed-by: Marek Vasut <[email protected]>
Cc: Pavel Machek <[email protected]>
Cc: Dinh Nguyen <[email protected]>
Acked-by: Pavel Machek <[email protected]>

usb: xhci: zynqmp: Removing unused function usb_phy_reset

This patch removes unsued function usb_phy_reset, rather common function
dwc3_phy_reset is used.

Signed-off-by: Marek Vasut <[email protected]>

usb: xhci: fsl: Removing unused function usb_phy_reset

This patch removes unsued function usb_phy_reset, rather common function
dwc3_phy_reset is used.

Signed-off-by: Rajesh Bhagat <[email protected]>

usb: xhci: dwc3: Adding reset delay requirement as per dwc3 databook

As per dwc3 databook, delay is required before taking the core out of reset.
This delay is required so that the PHY are stable, and then we can take core
out of reset.

Reference is taken from linux dwc3 code, file: drivers/usb/dwc3/core.c.

Signed-off-by: Sriram Dash <[email protected]>
Signed-off-by: Rajesh Bhagat <[email protected]>

usb: add support for generic EHCI devices

This driver is meant to be used with any EHCI-compatible host
controller in case if there's no need for platform-specific
glue such as setup of controller or PHY's power mode via
GPIOs etc.

Signed-off-by: Alexey Brodkin <[email protected]>
Cc: Stephen Warren <[email protected]>
Cc: Simon Glass <[email protected]>
Cc: Marek Vasut <[email protected]>

sunxi: usb: Rename the sunxi usb driver DM entry

The DM entry should be unique, otherwise it will collide with other
drivers. Fix this by assigning the driver a more unique name than
usb_ehci.

Signed-off-by: Marek Vasut <[email protected]>
Cc: Hans de Goede <[email protected]>
Cc: Ian Campbell <[email protected]>
Cc: Simon Glass <[email protected]>

usb: zynqmp: Fix build warnings

The driver does "return 0" in function with void type.

Signed-off-by: Marek Vasut <[email protected]>
Cc: Michal Simek <[email protected]>
Cc: Siva Durga Prasad Paladugu <[email protected]>

drivers:usb:fsl: Add T4080 as affected soc for Erratum A007792 sw workaround

Apply Erratum A007792 sw workaround for T4080

Signed-off-by: Sriram Dash <[email protected]>
Signed-off-by: Rajesh Bhagat <[email protected]>
Signed-off-by: Ramneek Mehresh <[email protected]>

usb: dwc3: fix build warnings

fix the following build warnings:

drivers/usb/dwc3/core.c: In function ‘dwc3_uboot_init’:
drivers/usb/dwc3/core.c:625:6: warning: ‘dev’ is used uninitialized in this function [-Wuninitialized]
mem = devm_kzalloc(dev, sizeof(*dwc) + DWC3_ALIGN_MASK, GFP_KERNEL);
        ^
drivers/usb/dwc3/dwc3-omap.c: In function ‘dwc3_omap_uboot_init’:
drivers/usb/dwc3/dwc3-omap.c:380:7: warning: ‘dev’ is used uninitialized in this function [-Wuninitialized]
omap = devm_kzalloc(dev, sizeof(*omap), GFP_KERNEL);

Signed-off-by: Felipe Balbi <[email protected]>

usb: host: xhci-omap: fix build break

Fix the following build break:

drivers/usb/host/xhci-omap.c:35:5: error: ‘board_usb_init’ aliased to external symbol ‘__board_usb_init’
 int board_usb_init(int index, enum usb_init_type init)
      ^

Signed-off-by: Felipe Balbi <[email protected]>

usb: zynqmp: Enable USB XHCI support

Enable USB XHCI support for ZynqMP

Signed-off-by: Siva Durga Prasad Paladugu <[email protected]>

usb: zynqmp: Add XHCI driver support

Added USB XHCI driver support for zynqmp.

Signed-off-by: Siva Durga Prasad Paladugu <[email protected]>

am33xx: Update serial pdata

We now want to make sure that we have the platform data for NS16550 when
we do not have OF_CONTROL set.

Signed-off-by: Tom Rini <[email protected]>
Reviewed-by: Simon Glass <[email protected]>

altera_qspi: fix erase and write error code

Fix erase and write error code, which should be "protected".

From the "Embedded Peripherals IP User Guide" of Altera,

The "Illegal write" flag indicates that a write instruction is
targeting a protected sector on the flash memory. This bit is
set to indicate that the IP has cancelled a write instruction.

The "Illegal erase" flag indicates that an erase instruction has
been set to a protected sector on the flash memory. This bit is
set to indicate that the IP has cancelled the erase instruction.

Signed-off-by: Thomas Chou <[email protected]>
Acked-by: Chin Liang See <[email protected]>
Reviewed-by: Marek Vasut <[email protected]>

altera_qspi: add lock unlock ops

Add lock() and unlock() mtd ops to altera_qspi.

Signed-off-by: Thomas Chou <[email protected]>
Acked-by: Chin Liang See <[email protected]>
Reviewed-by: Marek Vasut <[email protected]>

dm: tegra: Provide serial platform data for SPL

At present an incorrect #if term is preventing this data from being compiled
in. All tegra boards use driver model for serial, so we can just drop this.

Fixes: fde7e18938d8 ("dm: tegra: pci: Move CONFIG_PCI_TEGRA to Kconfig")
Signed-off-by: Simon Glass <[email protected]>
Reported-by: Stephen Warren <[email protected]>
Acked-by: Thomas Chou <[email protected]>

dm: serial: ns16550: Add pre-reloc flag to the driver

For platforms that don't use device tree in SPL the only way to mark this
driver as 'required by relocation' is with the DM_FLAG_PRE_RELOC flag. Add
this to ensure that the driver is bound.

Fixes: fde7e18938d8 ("dm: tegra: pci: Move CONFIG_PCI_TEGRA to Kconfig")
Signed-off-by: Simon Glass <[email protected]>
Reported-by: Stephen Warren <[email protected]>
Acked-by: Thomas Chou <[email protected]>

tegra: Remove merge markers in some Kconfig files

These were added by mistake in commit fde7e189. They cause a warning when
configuring the boards. Remove them.

Fixes: fde7e18938d8 ("dm: tegra: pci: Move CONFIG_PCI_TEGRA to Kconfig")
Signed-off-by: Simon Glass <[email protected]>
Reported-by: Stephen Warren <[email protected]>
Tested-by: Stephen Warren <[email protected]>
Reported-by: Kevin Hilman <[email protected]>
Acked-by: Thomas Chou <[email protected]>

rpi: add support for Raspberry Pi Zero

For U-Boot's purposes, at present all we care about is ensuring there's
a model table entry.

Signed-off-by: Stephen Warren <[email protected]>

rpi: fix revision scheme parsing

The RPi has two different schemes for encoding board revision values.
When adding RPi 2 support, I thought that the conflicting type field
values were to be interpreted based on bcm2835-vs-bcm2836. In fact, the
scheme bit determines the encoding. The RPi Zero uses the bcm2835 yet
uses the new encoding scheme. Fix the code to cater for this correctly.

Signed-off-by: Stephen Warren <[email protected]>

rpi: get rid of BCM2835_BOARD_REV_* macros

There are two numbering schemes for the RPi revision values; old and new
scheme. The values within each scheme overlap. Hence, it doesn't make
sense to have absolute/global names for the revision IDs. Get rid of the
names and just use the raw revision/type values to set up the array of
per-revision data.

This change makes most sense when coupled with the next change. However,
it's split out so that the mechanical cut/paste is separate from the
logic changes for easier review and problem bisection.

Signed-off-by: Stephen Warren <[email protected]>

mkimage : Fix generating multi and script images

Seems 92a655c3 broke creating multi and script type images.
Since the file1:file2:file3 string does not get split up,
it fails on trying to open an non-existing file.

mkimage -A arm -O linux -T multi -C none -d zImage:splash.bmp:device.dtb uimage
tools/mkimage: Can't open zImage:splash.bmp:device.dtb: No such file or directory

Since the sizes of the different parts seem to get added in the actual
routine that handles multi and script type images, we can probably skip the
bit of the code that causes the failure for that type of images.

Signed-off-by: Philippe De Swert <[email protected]>

common: board: Dont relocate FDT incase of CONFIG_OF_EMBED

Don't relocate fdt in case of CONFIG_OF EMBED as the fdt is
already embedded with u-boot image.
Update fdt_blob after relocation as the fdt will be copied
during u-boot relocation.

Signed-off-by: Siva Durga Prasad Paladugu <[email protected]>
Signed-off-by: Michal Simek <[email protected]>
Reviewed-by: Alexey Brodkin <[email protected]>
Reviewed-by: Bin Meng <[email protected]>
Tested-by: Bin Meng <[email protected]> (QEMU x86)
Tested-by: Thomas Chou <[email protected]> (Nios2)
Acked-by: Thomas Chou <[email protected]>
Acked-by: Simon Glass <[email protected]>

spl: mmc: Unify non/driver model spl_mmc_find_device()

It is risky to have two different functions with much the same code. Future
authors may update one but not the other. It is hard to see which parts are
the same and which are different.

Unify the functions and drop the differences that are not really needed.

Note that one puts() becomes printf() as Tom mentioned that this does not
affect image size:

https://patchwork.ozlabs.org/patch/537276/

Note: It would be better to have an empty printf() and avoid the #ifdef for
CONFIG_SPL_LIBCOMMON_SUPPORT.

Signed-off-by: Simon Glass <[email protected]>
Tested-by: Michal Simek <[email protected]>

spl: mmc: Rename 'mmc' variable to 'mmcp'

The 'p' suffix makes it more obvious that we are dealing with a pointer
to a (pointer) value that will be returned to its caller.

Signed-off-by: Simon Glass <[email protected]>
Acked-by: Nikita Kiryanov <[email protected]>
Tested-by: Michal Simek <[email protected]>

spl: mmc: Fix compiler warning with CONFIG_DM_MMC

Since commit 4188ba3 we get the following warning on rockchip boards:

common/spl/spl_mmc.c:31:24: warning: ‘mmc’ may be used uninitialized in this function [-Wmaybe-uninitialized]
  count = mmc->block_dev.block_read(0, sector, 1, header);
                        ^
common/spl/spl_mmc.c:251:14: note: ‘mmc’ was declared here
  struct mmc *mmc;

Correct this by move the variable init earlier.

Signed-off-by: Simon Glass <[email protected]>
Tested-by: Michal Simek <[email protected]>
Acked-by: Nikita Kiryanov <[email protected]>

drivers: remoteproc: rproc-uclass: Fix check for NULL pointers

Neither uc_pdata->name nor check_name are supposed to be NULL in
_rproc_name_is_unique(). if uc_pdata->name is NULL, we are not
intialized yet, however if check_data is NULL, we do not have
proper data. Further, if either were NULL, strlen will crap out
while attempting to derefence NULL.

Instead, just check if either of these are NULL and bail out.

This should also fix the following coverity scan warnings:
*** CID 132281:  Null pointer dereferences  (FORWARD_NULL)
/drivers/remoteproc/rproc-uclass.c: 73 in _rproc_name_is_unique()

Reported-by: Tom Rini <[email protected]>
Signed-off-by: Nishanth Menon <[email protected]>

Change e-mail address of Kamil Lulko

Signed-off-by: Kamil Lulko <[email protected]>

ns16550: change reg-shift property default to zero

Change reg-shift property default to zero. When the integer property
is missing, it should be taken as zero. This is consistent to Linux
drivers/tty/serial/of_serial.c.

The x86 and most powerpc use reg-shift of 0. Most others use reg-shift
of 2. While reg-shift of 1 is rarely used.

Signed-off-by: Thomas Chou <[email protected]>
Reviewed-by: Bin Meng <[email protected]>
Reviewed-by: Mugunthan V N <[email protected]>
Acked-by: Simon Glass <[email protected]>

arch/arm/cpu/arm920t/ep93xx/led.c: Mark inline functions as static inline

With gcc-5.x we get warning about inline non-static functions referring to
static elements.

Signed-off-by: Tom Rini <[email protected]>
Reviewed-by: Bin Meng <[email protected]>

common: cli_hush: avoid memory leak

Need to free memory avoid memory leak, when error.

Signed-off-by: Peng Fan <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
Cc: Simon Glass <[email protected]>
Cc: Tom Rini <[email protected]>

common/Makefile: Compile fdt_support is enabled in SPL

When CONFIG_SPL_OF_TRANSLATE is enabled fdt_support.c needs to get
compiled. Otherwise fdt_translate_address() is missing which is needed
in dev_get_addr().

Signed-off-by: Stefan Roese <[email protected]>
Cc: Simon Glass <[email protected]>
Acked-by: Simon Glass <[email protected]>

defconfig: dra74_evm: enable serial driver model

Enable serial driver model for dra74_evm as ns16550 supports
driver model.

Signed-off-by: Mugunthan V N <[email protected]>
Reviewed-by: Simon Glass <[email protected]>

arm: dts: dra7-evm: add chosen node to specify serial console device

Introduce chosen node and specify uart0 to be used as serial console.

Signed-off-by: Mugunthan V N <[email protected]>
Reviewed-by: Simon Glass <[email protected]>

arm: dts: omap: add reg-shift to serial device tree node

With the commit 'c7b9686d5d48 ("ns16550: unify serial_omap")' all
TI platforms are broken with DM/DT boot as ns16550 driver expects
reg-shift from DT which is not populated for TI platforms.
Earlier it worked as it was hard coded to 2 in serial-omap
driver. So adding the reg-shift to serial nodes for dra7, am4372
and am33xx dtsi files. Tested this patch on am437x-sk-evm,
am437x-gp-evm, am335x-boneblack, dra74x-evm and dra72x-evm.

Signed-off-by: Mugunthan V N <[email protected]>
Reviewed-by: Simon Glass <[email protected]>
Acked-by: Thomas Chou <[email protected]>

common: miiphyutil: avoid memory leak

The following code will alloc memory for new_dev and ldev:
"
new_dev = mdio_alloc();
ldev = malloc(sizeof(*ldev));
"
Either new_dev or ldev is NULL, directly return, but this may leak memory.
So before return, using free(ldev) and mdio_free(new_dev) to avoid
leaking memory, also free can handle NULL pointer.

Signed-off-by: Peng Fan <[email protected]>
Cc: Joe Hershberger <[email protected]>
Cc: Simon Glass <[email protected]>
Cc: Bin Meng <[email protected]>
Reviewed-by: Bin Meng <[email protected]>
Acked-by: Joe Hershberger <[email protected]>

common: mmc: unsigned char compared against 0

"enable" is unsigned char type and its value will not be
negative, so discard "enable < 0".

Signed-off-by: Peng Fan <[email protected]>
Cc: Diego Santa Cruz <[email protected]>
Cc: Pantelis Antoniou <[email protected]>
Cc: Andrew Gabbasov <[email protected]>
Cc: Simon Glass <[email protected]>
Cc: Stefano Babic <[email protected]>
Cc: Tom Rini <[email protected]>
Reviewed-by: Simon Glass <[email protected]>

eeprom: Zap CONFIG_SYS_I2C_MULTI_EEPROMS

This symbol is no longer used anywhere, remove it.

Signed-off-by: Marek Vasut <[email protected]>
Cc: Heiko Schocher <[email protected]>
Cc: Tom Rini <[email protected]>
Acked-by: Heiko Schocher <[email protected]>

Fix typo: firstly -> first.

Signed-off-by: Vagrant Cascadian <[email protected]>
Acked-by: Marek Vasut <[email protected]>