]> git.dujemihanovic.xyz Git - u-boot.git/log
u-boot.git
8 years agosh: generate position independent code for all platforms
Vladimir Zapolskiy [Sun, 27 Nov 2016 22:15:35 +0000 (00:15 +0200)]
sh: generate position independent code for all platforms

Finally add fpic compilation option to produce relocatable code.
Note that this requires to define CONFIG_NEEDS_MANUAL_RELOC for all
board files, also relocation support still has some limitations
(e.g. a developer should care not to overwrite the executing code or
memset() with zeroes not yet relocated data on malloc init etc.),
which may be fixed while switching to PIE.

Due to short investigation the architecture code is not ready for PIE
linking, this will require some manipulations with .dyn* sections.

Signed-off-by: Vladimir Zapolskiy <vz@mleia.com>
8 years agosh: share the correct version of start.S among all cpus
Vladimir Zapolskiy [Sun, 27 Nov 2016 22:15:34 +0000 (00:15 +0200)]
sh: share the correct version of start.S among all cpus

It is easy to note that SH2/SH3/SH4 start.S code is practically
the same with a minor difference for SH2 where a short data header is
present. To avoid unwanted code duplication and to automatically
convert SH2 and SH3 platforms to generic board support move fixed SH4
start.S into arch/sh/lib/start.S and share it among all platforms.

Signed-off-by: Vladimir Zapolskiy <vz@mleia.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
8 years agosh4: fix start.S by calling board_init_f() after first code relocation
Vladimir Zapolskiy [Sun, 27 Nov 2016 22:15:33 +0000 (00:15 +0200)]
sh4: fix start.S by calling board_init_f() after first code relocation

Like on ARM platform keep the first code relocation from a U-boot
image storage to RAM at CONFIG_SYS_TEXT_BASE, then pass execution to a
generic board_init_f() with empty GD flags. If CONFIG_SYS_TEXT_BASE is
equal to a calculated by board_init_f() relocation address there will
be no more code and data copy, however it's worth to mention that the
first copy happens even if $pc on _start is the same as
CONFIG_SYS_TEXT_BASE, on practice this works without a problem.

Also note that _sh_start is renamed back to _start to correct
gd->mon_len calculation by setup_mon_len(), the opposite rename was
done in pre-generic board commit 2024b968ee9 ("sh: Fix build in start.S").

Signed-off-by: Vladimir Zapolskiy <vz@mleia.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
8 years agosh: add shared relocate_code() function and call board_init_r()
Vladimir Zapolskiy [Sun, 27 Nov 2016 22:15:32 +0000 (00:15 +0200)]
sh: add shared relocate_code() function and call board_init_r()

Commits b61e90e6fd83 ("sh: Drop the arch-specific board init") and
f41e6088eb1a ("sh: Fix build errors for generic board") left code and
data relocation done in start.S, however further actual U-boot
configuration is not started anymore. Practically SH boards with the
code relocated into the expected position by start.S still can be
booted, so the change adds this option and provides an option how to
relocate code for board_init_r() execution.

Signed-off-by: Vladimir Zapolskiy <vz@mleia.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
8 years agosh: add common dram_init() function for all boards
Vladimir Zapolskiy [Sun, 27 Nov 2016 22:15:31 +0000 (00:15 +0200)]
sh: add common dram_init() function for all boards

Generic board support assumes a different method of specifying
DRAM size on board, also it can be shared among all boards, notably
only sh7763rdp board has a custom legacy dram_init(), however
the difference is only in printing some additional information,
this feature can be removed.

Signed-off-by: Vladimir Zapolskiy <vz@mleia.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
8 years agosh: define CONFIG_DISPLAY_BOARDINFO to print board information
Vladimir Zapolskiy [Sun, 27 Nov 2016 22:15:30 +0000 (00:15 +0200)]
sh: define CONFIG_DISPLAY_BOARDINFO to print board information

All SH boards define a checkboard() function which outputs basic board
information on boot, however generic board support requires to define
CONFIG_DISPLAY_BOARDINFO to do that, so define it for the boards.

Signed-off-by: Vladimir Zapolskiy <vz@mleia.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
8 years agosh: remove undefined DEBUG preprocessor token from board config files
Vladimir Zapolskiy [Sun, 27 Nov 2016 22:15:29 +0000 (00:15 +0200)]
sh: remove undefined DEBUG preprocessor token from board config files

By default this undef is a noop, moreover at this point when the
platform support is broken is prevents debugging of U-boot by manual
insertion of #define DEBUG into common files, so it makes sense to
remove the option from all SH boards as a harmful one.

Signed-off-by: Vladimir Zapolskiy <vz@mleia.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
8 years agosh: add MEMORY command to a shared linker script
Vladimir Zapolskiy [Sun, 27 Nov 2016 22:15:28 +0000 (00:15 +0200)]
sh: add MEMORY command to a shared linker script

At the moment in runtime all defined sections are copied into or
created in RAM, specify this explicitly to assert potential out of RAM
placements of the sections.

Signed-off-by: Vladimir Zapolskiy <vz@mleia.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
8 years agosh: define entry point and reloc_dst inside a linker script
Vladimir Zapolskiy [Sun, 27 Nov 2016 22:15:27 +0000 (00:15 +0200)]
sh: define entry point and reloc_dst inside a linker script

No functional change, concentrate linker script commands in one
place for convenience. Entry point is set to CONFIG_SYS_TEXT_BASE by
default on build, so this option can be omitted from being added to
the linker script.

Signed-off-by: Vladimir Zapolskiy <vz@mleia.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
8 years agosh: place board lowlevel_init code in the beginning of .text
Vladimir Zapolskiy [Sun, 27 Nov 2016 22:15:26 +0000 (00:15 +0200)]
sh: place board lowlevel_init code in the beginning of .text

Reference lowlevel_init of all supported SH2A/SH3/SH4/SH4A boards
from a shared linker script, the lowlevel_init function will be called
by a relative address.

Signed-off-by: Vladimir Zapolskiy <vz@mleia.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
8 years agosh4: use single u-boot linker script for all boards
Vladimir Zapolskiy [Sun, 27 Nov 2016 22:15:25 +0000 (00:15 +0200)]
sh4: use single u-boot linker script for all boards

Three supported SH4/SH4A boards with the bootloader image stored on
SPI flash have own flavour of a linker script, in turn they are equal
among each other. The only difference is that the text from
lowlevel_init.o is placed right after start.o, which makes sense.

Note that .bss section is not marked as NOLOAD, because for about
10 years this is a default option of a GNU linker, either the
attribute is found or not the resulting image file is the same.

Signed-off-by: Vladimir Zapolskiy <vz@mleia.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
8 years agocommon: sh: add necessary define bits to board_f
Vladimir Zapolskiy [Sun, 27 Nov 2016 22:15:24 +0000 (00:15 +0200)]
common: sh: add necessary define bits to board_f

Since a platform conversion to generic board support has not been
accomplished some architecture specific bits are missing from board_f
init sequence, the change adds a number of basic expected callbacks
into early init sequence.

Signed-off-by: Vladimir Zapolskiy <vz@mleia.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
8 years agor2dplus: select rtl8139 driver in defconfig
Vladimir Zapolskiy [Sun, 27 Nov 2016 22:15:23 +0000 (00:15 +0200)]
r2dplus: select rtl8139 driver in defconfig

CONFIG_RTL8139 was moved to a board defconfig by a commit 86e9dc86b1a2
("net: Move CONFIG_RTL8139 to Kconfig"), however it was done
incorrectly due to a missing CONFIG_NETDEVICES selection, thus
virtually it was just a removal of the driver compilation.

As an unlucky consequence the option was completely removed by a purge
commit adad96e60d0e ("configs: Re-sync HUSH options"), restore the
driver inclusion back.

Signed-off-by: Vladimir Zapolskiy <vz@mleia.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
8 years agor2dplus: use P1 area space for text base and PCI system memory
Vladimir Zapolskiy [Sun, 27 Nov 2016 22:15:22 +0000 (00:15 +0200)]
r2dplus: use P1 area space for text base and PCI system memory

While both options are acceptable use P1 area physical addresses
instead of external memory space of text base and PCI system memory
for unification purposes, all other supported superh boards have the
same selection.

This allows to easily ensure that CONFIG_SYS_TEXT_BASE is located
within available DRAM.

Signed-off-by: Vladimir Zapolskiy <vz@mleia.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
8 years agosh4: remove __io config options from r2dplus and r7780mp boards
Vladimir Zapolskiy [Sun, 27 Nov 2016 22:15:21 +0000 (00:15 +0200)]
sh4: remove __io config options from r2dplus and r7780mp boards

Defined __io is no-op for the SH architecture and it can be removed
from board files without any functional change.

Signed-off-by: Vladimir Zapolskiy <vz@mleia.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
8 years agopci: sh7751: map PCI memory space into SDRAM
Vladimir Zapolskiy [Sun, 27 Nov 2016 22:15:20 +0000 (00:15 +0200)]
pci: sh7751: map PCI memory space into SDRAM

For ease of use and accounting a condition that on SH4
pci_phys_to_bus() and pci_bus_to_phys() are one in one mappings due to
unimplemented __iomem() conversion, this change fixes access to SDRAM
memory by PCI devices.

This change also generalizes PCI system memory configuration, which is
taken from board specific defines rather than hardcoded in the PCI
host driver.

Signed-off-by: Vladimir Zapolskiy <vz@mleia.com>
8 years agopci: sh7751: fix up PCI I/O space address
Vladimir Zapolskiy [Sun, 27 Nov 2016 22:15:19 +0000 (00:15 +0200)]
pci: sh7751: fix up PCI I/O space address

The change actually maps PCI I/O window to the same address on PCI bus
as it is stated by a comment, before the change transfers to the PCI I/O
space are failed due to misconfiguration of the most significant 14 bits
of the PCI address in PCIIOBR (note that it is set to 0x0).

Most probably the problem remained unnoticed, because communcation
to all tested PCI devices is done over PCI memory space only.

Signed-off-by: Vladimir Zapolskiy <vz@mleia.com>
8 years agosh4: cache: move exported cache manipulation functions into cache.c
Vladimir Zapolskiy [Sun, 27 Nov 2016 22:15:18 +0000 (00:15 +0200)]
sh4: cache: move exported cache manipulation functions into cache.c

No functional change, moving cache manipulation functions into cache.c
allows to collect all of them in a single location and as a pleasant
side effect cache_control() function can be unexported now.

Signed-off-by: Vladimir Zapolskiy <vz@mleia.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
8 years agosh: cache: don't modify CCR from P1 area
Vladimir Zapolskiy [Sun, 27 Nov 2016 22:15:17 +0000 (00:15 +0200)]
sh: cache: don't modify CCR from P1 area

cache_wback_all() is a local function and it is called from
cache_control() only, which is in turn jumps to P2 area.

The change fixes an issue when cache_wback_all() returns from P2 to
P1, however cache_control() continues to manipulate with CCR
register, according to the User's Manual this is restricted.

Signed-off-by: Vladimir Zapolskiy <vz@mleia.com>
8 years agosh: cache use jump_to_P2() and back_to_P1() from asm/system.h
Vladimir Zapolskiy [Sun, 27 Nov 2016 22:15:16 +0000 (00:15 +0200)]
sh: cache use jump_to_P2() and back_to_P1() from asm/system.h

Both jump_to_P2() and back_to_P1() functions are found in asm/system.h
header file and functionally they are the same, don't redefine them.

Signed-off-by: Vladimir Zapolskiy <vz@mleia.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
8 years agosh3: remove unused cache.c file from being built
Vladimir Zapolskiy [Sun, 27 Nov 2016 22:15:15 +0000 (00:15 +0200)]
sh3: remove unused cache.c file from being built

The change is similar to commit 994b56616bae ("sh: delete an unused
source file") for SH2, however here the removed cache.c file was
built and included into an image as a dead code.

If it is needed in future the contents can be reused from a similar
arch/sh/cpu/sh4/cache.c file, which is in turn will be moved to
a shared among all core flavours location at arch/sh/lib/cache.c.

Signed-off-by: Vladimir Zapolskiy <vz@mleia.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
8 years agosh4: cache: correct flush_cache() to writeback and invalidate
Vladimir Zapolskiy [Sun, 27 Nov 2016 22:15:14 +0000 (00:15 +0200)]
sh4: cache: correct flush_cache() to writeback and invalidate

In common usecases flush_cache() assumes both cache invalidation and
write-back to memory, instead of doing cache invalidation only with
the wrapped 'ocbi' instruction pin flush_cache() to cache invalidation
with memory write-back done by 'ocbp'.

Signed-off-by: Vladimir Zapolskiy <vz@mleia.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
8 years agosh4: cache: correct dcache flush to invalidate with write-back
Vladimir Zapolskiy [Sun, 27 Nov 2016 22:15:13 +0000 (00:15 +0200)]
sh4: cache: correct dcache flush to invalidate with write-back

In common usecases flush_cache() assumes both cache invalidation and
write-back to memory, thus in flush_dcache_range() implementation
change SH4 cache write-back only instruction 'ocbwb' with cache purge
instruction 'ocbp', according to the User's Manual there should be no
performance penalty for that.

Note that under circumstances only cache invalidation is expected from
flush_cache() call, in these occasional cases the current version of
flush_cache() works, which is a wrapper over invalidate_dcache_range()
at the moment, this will be fixed in the following change.

Signed-off-by: Vladimir Zapolskiy <vz@mleia.com>
8 years agocompiler.h: use u-boot endian macros on OpenBSD
Jonathan Gray [Sat, 26 Nov 2016 04:18:00 +0000 (15:18 +1100)]
compiler.h: use u-boot endian macros on OpenBSD

When building u-boot on sparc64 and powerpc hosts it became clear that
u-boot expects endian conversion defines to be macros:

lib/crc32.c:87: error: braced-group within expression allowed only inside a function

For OpenBSD switch from using system definitions equivalent to the u-boot ones
and define glibc __BYTE_ORDER __BIG_ENDIAN __LITTLE_ENDIAN names, as at least
some parts of the non-cross build assumes those names are present (ie crc32.c).

Signed-off-by: Jonathan Gray <jsg@jsg.id.au>
8 years agoARM: am57xx_evm: enable DFU support
Sekhar Nori [Fri, 25 Nov 2016 08:55:55 +0000 (14:25 +0530)]
ARM: am57xx_evm: enable DFU support

AM57xx GP EVM has USB2 port of the SoC exposed as
USB client port.

It is useful to be able to use this port for USB
DFU downloads.

Enable USB DFU support. Tested on AM57x GP EVM Rev
A3 using DFU to download to connected SD card.

configs for HS version of the AM57x EVM are
included in the patch but not really tested.

Signed-off-by: Sekhar Nori <nsekhar@ti.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
8 years agoARM: ti: consolidate dfu environment variables
Sekhar Nori [Fri, 25 Nov 2016 08:55:54 +0000 (14:25 +0530)]
ARM: ti: consolidate dfu environment variables

Introduce include/environment/ti/dfu.h that
consolidates environment variable definitions
for various TI boards that support DFU today.

Tested on AM335x EVM, AM437x SK EVM and DRA74x
EVM by using DFU to write to SD card.

Signed-off-by: Sekhar Nori <nsekhar@ti.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
8 years agocmd: scsi: Make private functions static
Michal Simek [Wed, 30 Nov 2016 11:18:36 +0000 (12:18 +0100)]
cmd: scsi: Make private functions static

Two functions should be static because they are not exported to any
other file.
Warnings were reported by sparse C=1.

Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
8 years agoscsi: Separate SCSI private block description initialization
Michal Simek [Wed, 30 Nov 2016 11:50:58 +0000 (12:50 +0100)]
scsi: Separate SCSI private block description initialization

When blk_create_device() is called some parameters in blk_desc are
automatically filled. Separate SCSI private initialization and SCSI full
block device initialization not to rewrite already prepared data.

Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
8 years agoscsi: Change scsi_scan() to be able to return value
Michal Simek [Wed, 30 Nov 2016 11:12:31 +0000 (12:12 +0100)]
scsi: Change scsi_scan() to be able to return value

With DM_SCSI this function will return more than one return value to
cover errors.

Suggested-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
8 years agoscsi: Make private functions static
Michal Simek [Wed, 30 Nov 2016 10:53:43 +0000 (11:53 +0100)]
scsi: Make private functions static

Several functions should be static because they are not exported to any
other file.
Warnings were reported by sparse C=1.

Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
8 years agoscsi: Remove completely unused functions
Michal Simek [Wed, 30 Nov 2016 10:54:52 +0000 (11:54 +0100)]
scsi: Remove completely unused functions

These functions are not called for any location.
This patch removes them scsi_trim_trail(), scsi_get_disk_count()
and scsi_setup_read6().

Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
8 years agoscsi: Simplify scsi_read/scsi_write()
Michal Simek [Fri, 18 Nov 2016 15:22:42 +0000 (16:22 +0100)]
scsi: Simplify scsi_read/scsi_write()

There is no reason to directly point to static allocated array
when we have proper block_dev pointer available via parameter
in !CONFIG_BLK. For CONFIG_BLK this is read directly from uclass
platdata.

Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
8 years agoscsi: Move pccb buffer initalization directly to scsi_detect_dev
Michal Simek [Fri, 18 Nov 2016 15:14:24 +0000 (16:14 +0100)]
scsi: Move pccb buffer initalization directly to scsi_detect_dev

pccb is pointer to temporary buffer which is used only for sending
command. Make it local as is done in scsi_read/scsi_write.

Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
8 years agoscsi: Take lun from device block description
Michal Simek [Fri, 18 Nov 2016 14:52:51 +0000 (15:52 +0100)]
scsi: Take lun from device block description

Prepare LUN(Logical unit number) directly in block description structure
and reuse it.

Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
8 years agoscsi: Extract device detection algorithm
Michal Simek [Fri, 18 Nov 2016 14:42:13 +0000 (15:42 +0100)]
scsi: Extract device detection algorithm

The patch enables running detection algorithm on block device
description structure.

Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
8 years agoscsi: Extract block device initialization
Michal Simek [Fri, 18 Nov 2016 14:27:00 +0000 (15:27 +0100)]
scsi: Extract block device initialization

Extract block device initialization to specific function.

Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
8 years agocommon: miiphyutil: Work and report phy address in hex in mdio cmd
Michal Simek [Wed, 16 Nov 2016 07:41:01 +0000 (08:41 +0100)]
common: miiphyutil: Work and report phy address in hex in mdio cmd

It is confusing that mdio commands work and report phy id as
decimal value when mii is working with hex values.

For example:
ZynqMP> mdio list
gem:
21 - TI DP83867 <--> ethernet@ff0e0000
ZynqMP> mdio read ethernet@ff0e0000 0
Reading from bus gem
PHY at address 21:
0 - 0x1140
ZynqMP> mii dump 21 0
Incorrect PHY address. Range should be 0-31
...
ZynqMP> mii dump 15
0.     (1140)                 -- PHY control register --
  (8000:0000) 0.15    =     0    reset

U-Boot normally takes hex values that's why this patch is changing mdio
command to handle hex instead of changing mii command to handle decimal
values.

Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
8 years agoARM: zynq: Enable SD1 and qspi for picozed board
Siva Durga Prasad Paladugu [Fri, 7 Oct 2016 05:41:52 +0000 (11:11 +0530)]
ARM: zynq: Enable SD1 and qspi for picozed board

Enable SD1 and qspi for picozed board.

Signed-off-by: Siva Durga Prasad Paladugu <sivadur@xilinx.com>
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
8 years agoARM: zynq(mp): remove unneeded CONFIG_USB_MAX_CONTROLLER_COUNT defines
Masahiro Yamada [Thu, 13 Oct 2016 15:40:00 +0000 (00:40 +0900)]
ARM: zynq(mp): remove unneeded CONFIG_USB_MAX_CONTROLLER_COUNT defines

ARCH_ZYNQ(MP) selects DM_USB, where CONFIG_USB_MAX_CONTROLLER_COUNT
is not used.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Acked-by: Moritz Fischer <moritz.fischer@ettus.com>
Acked-by: Michal Simek <michal.simek@xilinx.com>
Signed-off-by: Michal Simek <michal.simek@xilinx.com>
8 years agoARM64: zynqmp: List secondary software boot modes
Michal Simek [Wed, 26 Oct 2016 07:24:32 +0000 (09:24 +0200)]
ARM64: zynqmp: List secondary software boot modes

Using alternative bootmode field to support automatic secondary boot
modes. It is purely software setting where SW modes are using free
bootmode combinations.

Signed-off-by: Michal Simek <michal.simek@xilinx.com>
8 years agoARM64: zynqmp: Use DTS name for different psu_init_gpl* files in SPL
Michal Simek [Tue, 29 Nov 2016 11:03:51 +0000 (12:03 +0100)]
ARM64: zynqmp: Use DTS name for different psu_init_gpl* files in SPL

CONFIG_SYS_CONFIG_NAME is not proper config option for different low
level init files because different board revisions requires different
psu_init_gpl* files.

Also at the end of moving drivers to DM all board specific configuration
files should be removed.

The same changes was done for Zynq.
"ARM: zynq: Simplify zynq configuration"
(sha1: ad5b5801264e573bfbf17a20b04c546985c5bfc1)

Signed-off-by: Michal Simek <michal.simek@xilinx.com>
8 years agoARM64: zynqmp: Force certain bootmode for SPL
Michal Simek [Tue, 30 Aug 2016 14:17:27 +0000 (16:17 +0200)]
ARM64: zynqmp: Force certain bootmode for SPL

ZynqMP provides an option to overwrite bootmode setting which
can change SPL behavior.
For example: boot SPL via JTAG and then SPL loads images from SD.

Signed-off-by: Michal Simek <michal.simek@xilinx.com>
8 years agoMerge git://www.denx.de/git/u-boot-marvell
Tom Rini [Thu, 1 Dec 2016 14:24:02 +0000 (09:24 -0500)]
Merge git://www.denx.de/git/u-boot-marvell

8 years agotools/kwbimage: add DEBUG option
Chris Packham [Wed, 9 Nov 2016 09:21:45 +0000 (22:21 +1300)]
tools/kwbimage: add DEBUG option

Offset 0x1 in the generated kwb image file is a set of flags, bit 0
enables debug output from the BootROM firmware.  Allow a DEBUG option in
the kwb configuration to request debug output from the BootROM firmware.

Signed-off-by: Chris Packham <judge.packham@gmail.com>
Signed-off-by: Stefan Roese <sr@denx.de>
8 years agotools/kwbimage: add BAUDRATE option
Chris Packham [Wed, 9 Nov 2016 09:07:45 +0000 (22:07 +1300)]
tools/kwbimage: add BAUDRATE option

Offset 0x18 in some Marvell datasheets this field is redacted as
"reserved". This offset is actually a set of options and bits 2:0 allow
the selection of the UART baudrate.

Allow a BAUDRATE option to set the UART baudrate for any messages coming
from the BootROM firmware.

Signed-off-by: Chris Packham <judge.packham@gmail.com>
Signed-off-by: Stefan Roese <sr@denx.de>
8 years agoarm: mvebu: move SYS_MVEBU_PLL_CLOCK to Kconfig
Chris Packham [Wed, 26 Oct 2016 01:08:30 +0000 (14:08 +1300)]
arm: mvebu: move SYS_MVEBU_PLL_CLOCK to Kconfig

The main PLL frequency is 2GHz for Armada-XP and 1GHZ for Armada 375,
38x and 39x.

[ Linux commit ae142bd9976532aa5232ab0b00e621690d8bfe6a ]

Signed-off-by: Chris Packham <judge.packham@gmail.com>
Signed-off-by: Stefan Roese <sr@denx.de>
8 years agomvebu: db-88f6820-amc: Enable FIT support
Chris Packham [Mon, 24 Oct 2016 22:40:55 +0000 (11:40 +1300)]
mvebu: db-88f6820-amc: Enable FIT support

Signed-off-by: Chris Packham <judge.packham@gmail.com>
Signed-off-by: Stefan Roese <sr@denx.de>
8 years agoSPL: mmc: Make spl_mmc_load_image available
Marek Vasut [Thu, 1 Dec 2016 01:06:35 +0000 (02:06 +0100)]
SPL: mmc: Make spl_mmc_load_image available

Make the spl_mmc_load_image() available globally, so it can be
invoked directly by SPL on extremely space-constrained systems.

Signed-off-by: Marek Vasut <marex@denx.de>
Cc: Tom Rini <trini@konsulko.com>
Cc: Simon Glass <sjg@chromium.org>
Cc: Jaehoon Chung <jh80.chung@samsung.com>
8 years agommc: Tinification of the mmc code
Marek Vasut [Thu, 1 Dec 2016 01:06:33 +0000 (02:06 +0100)]
mmc: Tinification of the mmc code

Add new configuration option CONFIG_MMC_TINY which strips away all
memory allocation within the MMC code and code for handling multiple
cards. This allows extremely space-constrained SPL code use the MMC
framework.

Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Marek Vasut <marex@denx.de>
Cc: Tom Rini <trini@konsulko.com>
Cc: Jaehoon Chung <jh80.chung@samsung.com>
8 years agommc: Fix warning if debug() is not used
Marek Vasut [Thu, 1 Dec 2016 01:06:32 +0000 (02:06 +0100)]
mmc: Fix warning if debug() is not used

If debug() is not used, then the whole content of debug(...) will
be removed by the preprocessor, which will result in the following
warning. This patch adds __maybe_unused annotation to fix this.

drivers/mmc/mmc.c: In function â€˜mmc_init’:
drivers/mmc/mmc.c:1685:11: warning: variable â€˜start’ set but not used [-Wunused-but-set-variable]
  unsigned start;

Reviewed-by: Tom Rini <trini@konsulko.com>
Signed-off-by: Marek Vasut <marex@denx.de>
Cc: Pantelis Antoniou <panto@antoniou-consulting.com>
Cc: Jaehoon Chung <jh80.chung@samsung.com>
8 years agommc: add bkops-enable command
Tomas Melin [Fri, 25 Nov 2016 09:01:03 +0000 (11:01 +0200)]
mmc: add bkops-enable command

Add new command that provides possibility to enable the
background operations handshake functionality
(BKOPS_EN, EXT_CSD byte [163]) on eMMC devices.

This is an optional feature of eMMCs, the setting is write-once.
The command must be explicitly taken into use with
CONFIG_CMD_BKOPS_ENABLE.

Signed-off-by: Tomas Melin <tomas.melin@vaisala.com>
8 years agommc: s5p_sdhci: fix to check proper pinmux id
Seung-Woo Kim [Thu, 24 Nov 2016 06:05:51 +0000 (15:05 +0900)]
mmc: s5p_sdhci: fix to check proper pinmux id

At sdhci_get_config(), there was wrong condition to check pimux
id, so this patch fixes to check proper pinmux id.

Signed-off-by: Seung-Woo Kim <sw0312.kim@samsung.com>
8 years agots4600: Disable CONFIG_DISPLAY_CPUINFO
Tom Rini [Wed, 30 Nov 2016 19:14:47 +0000 (14:14 -0500)]
ts4600: Disable CONFIG_DISPLAY_CPUINFO

Without this change we see:
../arch/arm/cpu/arm926ejs/mxs/mxs.c: In function â€˜print_cpuinfo’:
../arch/arm/cpu/arm926ejs/mxs/mxs.c:181:23: warning: unused variable â€˜data’ [-Wunused-variable]
../arch/arm/cpu/arm926ejs/mxs/mxs.c:180:6: warning: variable â€˜cpurev’ set but not used [-Wunused-but-set-variable]

So the easy solution is to disable CONFIG_DISPLAY_CPUINFO

Reviewed-by: Sebastien Bourdelin <sebastien.bourdelin@savoirfairelinux.com>
Signed-off-by: Tom Rini <trini@konsulko.com>
8 years agoMerge branch 'master' of git://git.denx.de/u-boot-mips
Tom Rini [Wed, 30 Nov 2016 19:08:28 +0000 (14:08 -0500)]
Merge branch 'master' of git://git.denx.de/u-boot-mips

8 years agotravis.yml: Split Freescale ARM job up more
Tom Rini [Wed, 30 Nov 2016 19:05:08 +0000 (14:05 -0500)]
travis.yml: Split Freescale ARM job up more

In order to avoid running into the time limit, split the 32bit and 64bit
Freescale boards into separate jobs.  We could either pass
"freescale & armv8" to buildman or exclude all of the 32bit CPUs.  While
the former is shorter I fear the amount of possible escaping required
would make things less readable.

Signed-off-by: Tom Rini <trini@konsulko.com>
8 years agoMIPS: Fix map_physmem for cached mappings
Paul Burton [Mon, 26 Sep 2016 18:28:57 +0000 (19:28 +0100)]
MIPS: Fix map_physmem for cached mappings

map_physmem should return a pointer that can be used by the CPU to
access the given memory - on MIPS simply returning the physical address
as it does prior to this patch doesn't achieve that. Instead return a
pointer to the memory within (c)kseg0, which matches up consistently
with the (c)kseg1 pointer that uncached mappings return via ioremap.

Signed-off-by: Paul Burton <paul.burton@imgtec.com>
8 years agoMIPS: Use ram_top, not bi_memsize, in arch_lmb_reserve
Paul Burton [Mon, 26 Sep 2016 18:28:56 +0000 (19:28 +0100)]
MIPS: Use ram_top, not bi_memsize, in arch_lmb_reserve

When calculating the region to reserve for the stack in
arch_lmb_reserve, make use of ram_top instead of adding bi_memsize to
CONFIG_SYS_SDRAM_BASE. This avoids overflow if the system has enough
memory to reach the end of the address space.

Signed-off-by: Paul Burton <paul.burton@imgtec.com>
8 years agomips: Let cache.h be included from assembly source
Marek Vasut [Fri, 25 Nov 2016 22:32:22 +0000 (23:32 +0100)]
mips: Let cache.h be included from assembly source

Add ifdef __ASSEMBLY__ around the function prototype to let cache.h
be included from assembly code.

Signed-off-by: Marek Vasut <marex@denx.de>
Cc: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
Cc: Paul Burton <paul.burton@imgtec.com>
8 years agocommon/board_f: enable initr_trap for MIPS
Daniel Schwierzeck [Sat, 9 Jan 2016 17:34:15 +0000 (18:34 +0100)]
common/board_f: enable initr_trap for MIPS

Enable initr_trap hook also for MIPS to install and enable
U-Boot's specific MIPS exception handlers.

Signed-off-by: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
8 years agoMIPS: add handling for generic and EJTAG exceptions
Daniel Schwierzeck [Sat, 9 Jan 2016 17:34:14 +0000 (18:34 +0100)]
MIPS: add handling for generic and EJTAG exceptions

Add exception handlers for generic and EJTAG exceptions. Most of
the assembly code is imported from Linux kernel and adapted to U-Boot.
The exception vector table will be reserved above the stack before
U-Boot is relocated. The exception handlers will be installed and
activated after relocation in the initr_traps hook function.

Generic exceptions are handled by showing a CPU register dump similar
to Linux kernel. For example:

malta # md 1
00000001:
Ooops:
$ 0   : 00000000 00000000 00000009 00000004
$ 4   : 8ff7e108 00000000 0000003a 00000000
$ 8   : 00000008 00000001 8ff7cd18 00000004
$12   : 00000002 00000000 00000005 0000003a
$16   : 00000004 00000040 00000001 00000001
$20   : 00000000 8fff53c0 00000008 00000004
$24   : ffffffff 8ffdea44
$28   : 90001650 8ff7cd00 00000004 8ffe6818
Hi    : 00000000
Lo    : 00000004
epc   : 8ffe6848 (text bfc28848)
ra    : 8ffe6818 (text bfc28818)
Status: 00000006
Cause : 00000410 (ExcCode 04)
BadVA : 8ff9e928
PrId  : 00019300
 ### ERROR ### Please RESET the board ###

EJTAG exceptions are checked for SDBBP and delegated to the SDBBP handler
if necessary. Otherwise the debug mode will simply be exited. The SDBBP
handler currently prints the contents of registers c0_depc and c0_debug.
This could be extended in the future to handle semi-hosting according to
the MIPS UHI specification.

Signed-off-by: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
Reviewed-by: Paul Burton <paul.burton@imgtec.com>
Tested-by: Paul Burton <paul.burton@imgtec.com>
8 years agoMIPS: reserve space for exception vectors
Daniel Schwierzeck [Sat, 9 Jan 2016 17:34:14 +0000 (18:34 +0100)]
MIPS: reserve space for exception vectors

In order to set own exception handlers, a table with the exception
vectors must be built in DRAM and the CPU EBase register must be
set to the base address of this table.

Reserve the space above the stack and use gd->irq_sp as storage
for the exception base address.

Signed-off-by: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
8 years agoMIPS: add asm-offsets for struct pt_regs
Daniel Schwierzeck [Sat, 9 Jan 2016 17:34:14 +0000 (18:34 +0100)]
MIPS: add asm-offsets for struct pt_regs

Import asm-offsets.c from kernel to generate offset for struct pt_regs
needed by exception handlers.

Signed-off-by: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
8 years agoMIPS: add possibility to setup initial stack and global data in SRAM
Daniel Schwierzeck [Sat, 4 Jun 2016 14:13:21 +0000 (16:13 +0200)]
MIPS: add possibility to setup initial stack and global data in SRAM

This adds a new Kconfig option CONFIG_MIPS_INIT_STACK_IN_SRAM which
a SoC can select if it supports some kind of SRAM. Together with
CONFIG_SYS_INIT_SP_ADDR the initial stack and global data can be
set up in that SRAM. This can be used to provide a C environment
also for lowlevel_init().

Signed-off-by: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
8 years agoMIPS: factor out code for initial stack and global data
Daniel Schwierzeck [Sun, 25 Sep 2016 16:36:38 +0000 (18:36 +0200)]
MIPS: factor out code for initial stack and global data

Move the code for setting up the initial stack and global data
to a macro to be able to use it more than once.

Signed-off-by: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
8 years agoMIPS: fix iand optimize setup of CP0 registers
Daniel Schwierzeck [Sun, 7 Feb 2016 23:37:59 +0000 (00:37 +0100)]
MIPS: fix iand optimize setup of CP0 registers

Clear cp0 status while preserving implementation specific bits.
Set bits BEV and ERL as the arch specification requires after
a reset or soft-reset exception.

Extend and fix initialization of watch registers. Check if additional
watch register sets are implemented and initialize them too.

Initialize cp0 count as early as possible to get the most
accurate boot timing.

Signed-off-by: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
8 years agoMIPS: fix ROM exception vectors
Daniel Schwierzeck [Sun, 7 Feb 2016 18:39:58 +0000 (19:39 +0100)]
MIPS: fix ROM exception vectors

When booting from ROM, early exceptions can't be handled
properly. Instead of busy-looping give the developer the
possibilty to examine the situation. Invoke an UHI
exception operation which can be read as unhandled exception
by a hardware debugger if one is attached. If the debugger
doesn't support UHI, the exception is read as unexpected
breakpoint.

Signed-off-by: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
8 years agoMIPS: make inclusion of ROM exception vectors configurable
Daniel Schwierzeck [Sun, 14 Feb 2016 17:52:57 +0000 (18:52 +0100)]
MIPS: make inclusion of ROM exception vectors configurable

This adds a compile time option to include code for static
exception vectors. Static exception vectors are only needed,
when the U-Boot entry point is equal to the CPU reset exception
vector address. For instance this is the case when U-Boot is
used as ROM in Qemu or booted from parallel NOR flash. When
U-Boot is booted from RAM (e.g. loaded there by SPL), the
exception vectors need to be setup dynamically, which is done
in follow-up commits.

Signed-off-by: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
8 years agoMerge branch 'master' of git://www.denx.de/git/u-boot-imx
Tom Rini [Wed, 30 Nov 2016 14:57:52 +0000 (09:57 -0500)]
Merge branch 'master' of git://www.denx.de/git/u-boot-imx

8 years agoimx6: clock: Enable External Memory Interface [EIM] clock (eim_slow_clock)
Lukasz Majewski [Mon, 28 Nov 2016 06:18:14 +0000 (07:18 +0100)]
imx6: clock: Enable External Memory Interface [EIM] clock (eim_slow_clock)

This patch extends the imx6 clock code to enable or disable the EIM
slow clock, which in necessary when one wants to use EIM interface t
o read/write from external memory (e.g. NOR).

Signed-off-by: Lukasz Majewski <l.majewski@majess.pl>
8 years agomx6sx: Add initial support for Samtec VIN|ING 2000 board
Christoph Fritz [Tue, 29 Nov 2016 15:13:40 +0000 (16:13 +0100)]
mx6sx: Add initial support for Samtec VIN|ING 2000 board

This patch adds initial support for Samtec VIN|ING 2000 board.

Signed-off-by: Christoph Fritz <chf.fritz@googlemail.com>
Reviewed-by: Stefano Babic <sbabic@denx.de>
Acked-by: Marek Vasut <marex@denx.de>
8 years agoMerge git://git.denx.de/u-boot-mpc85xx
Tom Rini [Wed, 30 Nov 2016 00:42:48 +0000 (19:42 -0500)]
Merge git://git.denx.de/u-boot-mpc85xx

8 years agotravis.yml: Add samsung and rockchip builds
Tom Rini [Tue, 29 Nov 2016 17:41:19 +0000 (12:41 -0500)]
travis.yml: Add samsung and rockchip builds

The catch-all job is failing due to time limits depending on factors out
of our control, so move Samsung and Rockchip boards into their own jobs
and then exclude them from the general ARM and AArch64 jobs.

Signed-off-by: Tom Rini <trini@konsulko.com>
8 years agoimx7: SPI: add suport for SPI flash in mikroBUS slot
Angus Ainslie [Fri, 11 Nov 2016 18:31:39 +0000 (11:31 -0700)]
imx7: SPI: add suport for SPI flash in mikroBUS slot

Enable the escpi3 nets attached to the mikroBUS slot
on the i.MX7 Sabre evalution board. Also enble the SPI flash
commands to work with the "flash click" board.

This is V2 of this patch with changes recommended by the maintainer

CC: Jagan Teki <jteki@openedev.com>
8 years agoarm: mx6: specify SPL padding
Stefan Agner [Tue, 15 Nov 2016 18:38:23 +0000 (10:38 -0800)]
arm: mx6: specify SPL padding

Specify standard padding for payload to 68KB. This is derived from
the maximum header size plus maximum SPL size. It matches the
already defined offset for SD/eMMC devices (69KB) too. This allows
to use the u-boot-with-spl.imx build target to generate a directly
flashable image which can be flashed using:

  dd if=u-boot-with-spl.imx of=/dev/mmcblk0 bs=512 skip=2

While the patch has been created with SD/eMMC in mind, this also
works with other boot media. The board file needs to configure the
media specific (absolute) payload offset accordingly. Especially
the IVT offset is boot media specific and can be retrieved from the
reference manual (Table 8-25. Image Vector Table Offset and Initial
Load Region Size). For NAND boot a define like this should do the
job:

 #define CONFIG_SYS_NAND_U_BOOT_OFFS (SPL_PAD_TO + 0x400)

Signed-off-by: Stefan Agner <stefan.agner@toradex.com>
8 years agomx6sx: Add initial support for UDOO Neo Board
Breno Lima [Fri, 25 Nov 2016 18:56:57 +0000 (16:56 -0200)]
mx6sx: Add initial support for UDOO Neo Board

UDOO Neo Board is a development board from Seco that has three models:
 - UDOO Neo Basic
 - UDOO Neo Basic Kick Starter
 - UDOO Neo Extended
 - UDOO Neo Full

All versions are based on the i.MX6 SoloX processor.

For more details about the UDOO Neo board, please refer to:
http://www.udoo.org/udoo-neo/

This work is based on a previous commit of Francesco Montefoschi
<francesco.monte@gmail.com>:
https://github.com/fmntf/u-boot/commit/877b71184a5105e708024f232d36aed574961844

Only tested on the UDOO Neo Full board.

Signed-off-by: Breno Lima <breno.lima@nxp.com>
Reviewed-by: Fabio Estevam <fabio.estevam@nxp.com>
8 years agocolibri_vf: Read kernel and device tree from static UBI volumes
Sanchayan Maity [Fri, 25 Nov 2016 10:49:17 +0000 (16:19 +0530)]
colibri_vf: Read kernel and device tree from static UBI volumes

Our update scripts write the kernel and device tree in seperate
UBI volumes. This allows to use a lot less UBI/UBIFS support in
U-Boot, which should lower the risk of hitting bugs in this area.

Signed-off-by: Sanchayan Maity <maitysanchayan@gmail.com>
8 years agoARM: ts4600: add basic board support
Sebastien Bourdelin [Tue, 8 Nov 2016 17:18:07 +0000 (12:18 -0500)]
ARM: ts4600: add basic board support

This commit adds basic support including:
MMC, Serial console

Signed-off-by: Sebastien Bourdelin <sebastien.bourdelin@savoirfairelinux.com>
Reviewed-by: Fabio Estevam <fabio.estevam@nxp.com>
8 years agoboard: ge: bx50v3: add the PMIC configuration support
Ken Lin [Fri, 18 Nov 2016 17:20:54 +0000 (12:20 -0500)]
board: ge: bx50v3: add the PMIC configuration support

Change the PMIC bulk configuration from auto mode to sync mode to avoid
voltage dropout issue seen in auto mode.

Signed-off-by: Ken Lin <ken.lin@advantech.com.tw>
Signed-off-by: Akshay Bhat <akshay.bhat@timesys.com>
8 years agoARM: mx6: ddr: use Kconfig for inclusion of DDR calibration routines
Eric Nelson [Sun, 30 Oct 2016 23:33:50 +0000 (16:33 -0700)]
ARM: mx6: ddr: use Kconfig for inclusion of DDR calibration routines

The DDR calibration routines are gated by conditionals for the
i.MX6DQ SOCs, but with the use of the sysinfo parameter, these
are usable on at least i.MX6SDL and i.MX6SL variants with DDR3.

Also, since only the Novena board currently uses the dynamic
DDR calibration routines, these routines waste space on other
boards using SPL.

Add a KConfig entry to allow boards to selectively include the
DDR calibration routines.

Signed-off-by: Eric Nelson <eric@nelint.com>
8 years agomx6: ddr: add routine to return DDR calibration data
Eric Nelson [Sun, 30 Oct 2016 23:33:49 +0000 (16:33 -0700)]
mx6: ddr: add routine to return DDR calibration data

Add routine mmdc_read_calibration() to return the output of DDR
calibration. This can be used for debugging or to aid in construction
of static memory configuration.

This routine will be used in a subsequent patch set adding a virtual
"mx6memcal" board, but could also be useful when gathering statistics
during an initial production run.

Signed-off-by: Eric Nelson <eric@nelint.com>
8 years agomx6: ddr: pass mx6_ddr_sysinfo to calibration routines
Eric Nelson [Sun, 30 Oct 2016 23:33:48 +0000 (16:33 -0700)]
mx6: ddr: pass mx6_ddr_sysinfo to calibration routines

The DDR calibration routines have scattered support for bus
widths other than 64-bits:

-- The mmdc_do_write_level_calibration() routine assumes the
presence of PHY1, and
-- The mmdc_do_dqs_calibration() routine tries to determine
whether one or two DDR PHYs are active by reading MDCTL.

Since a caller of these routines must have a valid struct mx6_ddr_sysinfo
for use in calling mx6_dram_cfg(), and the bus width is available in the
"dsize" field, use this structure to inform the calibration routines which
PHYs are active.

This allows the use of the DDR calibration routines on CPU variants
like i.MX6SL that only have a single MMDC port.

Signed-off-by: Eric Nelson <eric@nelint.com>
Reviewed-by: Marek Vasut <marex@denx.de>
8 years agomx6: ddr: allow 32 cycles for DQS gating calibration
Eric Nelson [Sun, 30 Oct 2016 23:33:47 +0000 (16:33 -0700)]
mx6: ddr: allow 32 cycles for DQS gating calibration

The DDR calibration code is only setting flag DG_CMP_CYC (DQS gating sample
cycle) for the first PHY.

Set the 32-cycle flag for both PHYs and clear when done so the MPDGCTRL0
output value isn't polluted with calibration artifacts.

Signed-off-by: Eric Nelson <eric@nelint.com>
Reviewed-by: Marek Vasut <marex@denx.de>
8 years agoimx: mx6: ddr: add register MPZQLP2CTL for LPDDR2
Eric Nelson [Fri, 28 Oct 2016 17:13:57 +0000 (10:13 -0700)]
imx: mx6: ddr: add register MPZQLP2CTL for LPDDR2

Add constants for the MPZQLP2CTL DDR register for both
banks to allow setting the LPDDR2 timing values in
.cfg files using a named constant instead of hex addresses
as is currently done in mx6slevk and other board files.

Signed-off-by: Eric Nelson <eric@nelint.com>
8 years agotools: imximage: display DCD block offset, length
Eric Nelson [Thu, 17 Nov 2016 00:13:41 +0000 (17:13 -0700)]
tools: imximage: display DCD block offset, length

These values can be used to sign a U-Boot image for use when
loading an image through the Serial Download Protocol (SDP).

Note that the address of 0x910000 is usable with the stock
configuration of imx_usb_loader on i.MX6 and i.MX7 SOCs:

https://github.com/boundarydevices/imx_usb_loader/blob/master/mx6_usb_work.conf#L3

Refer to the section on imx_usb_loader in this post for more
details:

https://boundarydevices.com/high-assurance-boot-hab-dummies/

Signed-off-by: Eric Nelson <eric@nelint.com>
8 years agoarm: imx: wandboard: fix compile error if CONFIG_VIDEO is deactivated
Sven Ebenfeld [Fri, 25 Nov 2016 20:42:53 +0000 (21:42 +0100)]
arm: imx: wandboard: fix compile error if CONFIG_VIDEO is deactivated

When I tried to deactivate VIDEO support for the Wandboard, it still
tried to initialize the Framebuffer and so on. That is the reason for
the added ifdefs. CONFIG_VIDEO is enabled in the configuration as default
and therefore nothing changes for the default user.

The structs mx6dl_i2c2_pad_info and mx6q_i2c2_pad_info are only available
when CONFIG_IPUV3 are set and should not be tried to access, when that
define is not defined.

Signed-off-by: Sven Ebenfeld <sven.ebenfeld@gmail.com>
8 years agopwm: imx: increase support up to PWM8 for i.MX6SX
Christoph Fritz [Tue, 22 Nov 2016 11:01:28 +0000 (12:01 +0100)]
pwm: imx: increase support up to PWM8 for i.MX6SX

This patch increases supported PWMs from previously PWM4 now up to PWM8
if i.MX6SX is in use.

Signed-off-by: Christoph Fritz <chf.fritz@googlemail.com>
8 years agotbs2910: Make Ethernet functional again
Soeren Moch [Sun, 27 Nov 2016 15:02:19 +0000 (16:02 +0100)]
tbs2910: Make Ethernet functional again

Configure the PHY to output a 125MHz clk from CLK_25M and set tx clock delay.
This patch is similar to commit 4b6035da482cccda06aeb419634f99937c9fc783
("mx6sabresd: Make Ethernet functional again").

Signed-off-by: Soeren Moch <smoch@web.de>
8 years agoimx: make ipu's di configurable
Max Krummenacher [Tue, 1 Nov 2016 14:04:21 +0000 (15:04 +0100)]
imx: make ipu's di configurable

The ipu has two display interfaces. Make the used one a parameter
in struct display_info_t instead of using unconditionally DI0.
DI0 is the default setting.

Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
Reviewed-by: Eric Nelson <eric@nelint.com>
8 years agospl: mmc: fix switch statement
Max Krummenacher [Tue, 1 Nov 2016 14:04:20 +0000 (15:04 +0100)]
spl: mmc: fix switch statement

If CONFIG_SPL_LIBCOMMON_SUPPORT is not defined there is a lone case statement
at the end of the switch leading to a compile error.
Remove the offending case statement.

| common/spl/spl_mmc.c:339:7: error: label at end of compound statement

Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
Acked-by: Marek Vasut <marex@denx.de>
8 years agoARM: mx6: add MMC2 boot device detection support in SPL
Marcin Niestroj [Fri, 21 Oct 2016 11:53:54 +0000 (13:53 +0200)]
ARM: mx6: add MMC2 boot device detection support in SPL

Check BOOT_CFG2[3:4] to determine which SD/MMC port is selected to boot
from. If MMC2 is selected return BOOT_DEVICE_MMC2. In all other cases
return BOOT_DEVICE_MMC1, as we do not have corresponding macro for MMC3
and MMC4.

Signed-off-by: Marcin Niestroj <m.niestroj@grinn-global.com>
8 years agoMerge branch 'master' of git://git.denx.de/u-boot
Stefano Babic [Tue, 29 Nov 2016 15:28:28 +0000 (16:28 +0100)]
Merge branch 'master' of git://git.denx.de/u-boot

Signed-off-by: Stefano Babic <sbabic@denx.de>
8 years agoarmv7: psci: cpu_off: flush D-Cache before disable D-Cache
Peng Fan [Tue, 22 Nov 2016 11:41:09 +0000 (19:41 +0800)]
armv7: psci: cpu_off: flush D-Cache before disable D-Cache

Before disable cache, need to first flush cache.

There maybe dirty data in D-Cache before disable D-Cache.
After disable D-Cache, the first store instructions in
psci_v7_flush_dcache_all will directly store registers
{r4-r5, r7, r9-r11, lr} to memory.
If there is dirty data before disable D-Cache,
psci_v7_flush_dcache_all will flush data to memory,
and may overwrite the memory that hold the registers
{r4-r5, r7, r9-r11, lr}.

So before disable cache, first flush D-Cache.

Signed-off-by: Peng Fan <peng.fan@nxp.com>
Cc: Albert Aribaud <albert.u.boot@aribaud.net>
Cc: Chen-Yu Tsai <wens@csie.org>
Cc: Hans de Goede <hdegoede@redhat.com>
Cc: Hongbo Zhang <hongbo.zhang@nxp.com>
Cc: York Sun <york.sun@nxp.com>
Cc: Marc Zyngier <marc.zyngier@arm.com>
Cc: Tom Rini <trini@konsulko.com>
8 years agovexpress64: Juno: Change PCI buss addresses for IO to start from zero.
Liviu Dudau [Tue, 22 Nov 2016 11:19:18 +0000 (11:19 +0000)]
vexpress64: Juno: Change PCI buss addresses for IO to start from zero.

Juno uses a 1:1 mapping between CPU and PCI addresses for IO. First,
that will trip devices that cannot use more than 16 bits of addresses
for IO, second it is un-necessary as the system can handle zero-based
PCI addresses just fine.

Change the mapping to start IO bus addresses from zero.

Signed-off-by: Liviu Dudau <Liviu.Dudau@foss.arm.com>
8 years agobcm2835: Reserve the spin table in efi memory map
Alexander Graf [Wed, 2 Nov 2016 09:36:20 +0000 (10:36 +0100)]
bcm2835: Reserve the spin table in efi memory map

Firmware provides a spin table on the raspberry pi. This table shouldn't
get overwritten by payloads, so we need to mark it as reserved.

Signed-off-by: Alexander Graf <agraf@suse.de>
Acked-by: Stephen Warren <swarren@wwwdotorg.org>
8 years agobcm2835 video: Map frame buffer as 32bpp
Alexander Graf [Wed, 2 Nov 2016 09:36:19 +0000 (10:36 +0100)]
bcm2835 video: Map frame buffer as 32bpp

To enable working efifb support, let's map the frame buffer as 32bpp
instead of 16bpp.

Signed-off-by: Alexander Graf <agraf@suse.de>
Acked-by: Stephen Warren <swarren@wwwdotorg.org>
8 years agoARM: bcm283x: Implement EFI RTS reset_system
Alexander Graf [Wed, 2 Nov 2016 09:36:18 +0000 (10:36 +0100)]
ARM: bcm283x: Implement EFI RTS reset_system

The rpi has a pretty simple way of resetting the whole system. All it takes
is to poke a few registers at a well defined location in MMIO space.

This patch adds support for the EFI loader implementation to allow an OS to
reset and power off the system when we're outside of boot time.

Signed-off-by: Alexander Graf <agraf@suse.de>
8 years agodrivers: net: keystone_net: add rgmii link type support when parsing dt
Mugunthan V N [Thu, 11 Aug 2016 14:34:03 +0000 (20:04 +0530)]
drivers: net: keystone_net: add rgmii link type support when parsing dt

Add support to detect RGMII link interface from link-interface
device tree entry. Also rename the existing link type enums so
that it provides meaningful interface like SGMII.

Signed-off-by: Mugunthan V N <mugunthanvnm@ti.com>
Reported-by: Sekhar Nori <nsekhar@ti.com>
Tested-by: Sekhar Nori <nsekhar@ti.com>
Acked-by: Joe Hershberger <joe.hershberger@ni.com>
8 years agoARM: bcm283x: use OF_CONTROL for bcm283x
Fabian Vogt [Mon, 26 Sep 2016 12:26:51 +0000 (14:26 +0200)]
ARM: bcm283x: use OF_CONTROL for bcm283x

This patch removes use of U_BOOT_DEVICE in board/raspberrypi/rpi/rpi.c,
enables OF_CONTROL in the config and adjusts the rpi_*defconfig configs.

Signed-off-by: Fabian Vogt <fvogt@suse.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
8 years agoboard: rpi: move uart deactivation to board_init
Fabian Vogt [Mon, 26 Sep 2016 12:26:50 +0000 (14:26 +0200)]
board: rpi: move uart deactivation to board_init

When using OF_CONTROL, the disabled value of the mini UART platdata
gets reset after board_early_init_f. So move detection and disabling
to board_init and remove board_early_init_f.
This uses the first device using the mini uart driver, as this method
works reliably with different device trees or even no device tree at all.

Signed-off-by: Fabian Vogt <fvogt@suse.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
8 years agoserial: bcm283x_mu: support disabling after initialization
Fabian Vogt [Mon, 26 Sep 2016 12:26:49 +0000 (14:26 +0200)]
serial: bcm283x_mu: support disabling after initialization

For the Raspberry Pi 3 it needs to be possible to disable the serial
device after initialization happens, as only after the GPIO device is available
it is known whether the mini uart is usable.

Signed-off-by: Fabian Vogt <fvogt@suse.com>
Reviewed-by: Simon Glass <sjg@chromium.org>