cmd_zfs: normalize 'file not found' errors

Signed-off-by: Luka Perkov <[email protected]>

cmd_reiser: normalize 'file not found' errors

Signed-off-by: Luka Perkov <[email protected]>

cmd_ubifs: normalize 'file not found' errors

Signed-off-by: Tim Harvey <[email protected]>

include: delete include/linux/config.h

Linux Kernel abolished include/linux/config.h long time ago.
(around version v2.6.18..v2.6.19)

We don't need to provide Linux copatibility any more.

This commit deletes include/linux/config.h
and fixes source files not to include this.

Signed-off-by: Masahiro Yamada <[email protected]>

config/sandbox: Add EFI and GPT support

Signed-off-by: Egbert Eich <[email protected]>

config: Define HAVE_BLOCK_DEVICE when CONFIG_CMD_GPT is set

Signed-off-by: Egbert Eich <[email protected]>

cmd/gpt: Support gpt command for all devices

The gpt command was only implemented for mmc devices. There is no reason
why this command should not be generalized and be applied all other
storage device classes.
This change both simplifies the implementation and eliminates a
build failure for systems that don't support mmcs.

Signed-off-by: Egbert Eich <[email protected]>
Tested-by: Piotr Wilczek <[email protected]>
[trini: Change coding style slightly]
Signed-off-by: Tom Rini <[email protected]>

cosmetic: UDM-net: clean up the remainders of dead driver

This commit omits non-existing drivers/net/netarm_eth.c from the list.
This driver is deleted by commit b411eb30f.

Signed-off-by: Masahiro Yamada <[email protected]>

cosmetic: UDM-serial: clean up the remainders of dead driver

The following serial drivers do not exist any more.

 - ns9750_serial.c: deleted by commit 4cfc611b4
 - s3c4510b_uart.c: deleted by commit afad40299
 - serial_clps7111.c: deleted by commit f2e080156
 - serial_netarm.c: deleted by commit b411eb30f

This commit cleans up UDM-serial.txt.

Signed-off-by: Masahiro Yamada <[email protected]>

cosmetic: doc: driver-model: Do not number driver lists

Everytime a dead driver is removed from the list,
we must re-number. This is a painful task.

Try
  git show e53232250 -- doc/driver-model/UDM-serial.txt
  git show 6f62f4207 -- doc/driver-model/UDM-serial.txt
  git show b9f4bc34a -- doc/driver-model/UDM-serial.txt
to see what I mean.

Signed-off-by: Masahiro Yamada <[email protected]>

README: remove wrong config name

There is no CONFIG_PCA953X_INFO symbol.
U-Boot uses CONFIG_CMD_PCA953X_INFO instead, which is described in
"Monitor Functions" section and thus no need to be repeated in the
"GPIO Support" section.
Remove the whole line.

Signed-off-by: Igor Grinberg <[email protected]>

cosmetic: remove empty lines at the top of file

Signed-off-by: Masahiro Yamada <[email protected]>

common/cmd_bootm.c: fix subcommand processing in OS specific do_bootm_xxx() functions

In commit "5c427e4: use BOOTM_STATE_OS_CMDLINE flag for plain bootm"
and "3d187b3: Only pass BOOTM_STATE_OS_CMDLINE on PowerPC/MIPS",
BOOTM_STATE_OS_CMDLINE was added to do_bootm for PowerPC and MIPS. This
breaks other OSes (vxworks, netbsd, plan9,...) that don't support
subcommand processing, e.g. they all contain the following code in their
do_bootm_xxx():

    if (flag & BOOTM_STATE_OS_PREP)
            return 0;
    if ((flag != 0) && (flag != BOOTM_STATE_OS_GO))
            return 1;

which will result a "subcommand not supported" error.
This patch changes the above logic to:

    /* if not go command, pretend everything to be OK */
    if (flag != BOOTM_STATE_OS_GO)
         return 0;

Signed-off-by: Miao Yan <[email protected]>

Makefile: do not create a symbolic link to arch/${ARCH}/include/asm

In-tree build:
  - Do not create a symbolic link
      from include/asm to arch/${ARCH}/include/asm
  - Add ${SRCTREE}/arch/arm/include into the header search path

Out-of-tree build:
  - Do not create a directory ${OBJTREE}/include2
  - Do not create a symbolic link
      from ${OBJTREE}/include2/asm to ${SRCTREE}/arch/${ARCH}/include/asm
  - Add ${SRCTREE}/arch/arm/include into the header search path

Signed-off-by: Masahiro Yamada <[email protected]>

Add support for SX151x SPI GPIO Expanders

Signed-off-by: Viktar Palstsiuk <[email protected]>

boards.cfg: remove git leftovers

Remove 'HEAD' line which is most likely left there while rebasing. It was
introduced in commit 877bfe37dc00b0ae59f37742954a62bce3fdf3a0.

Signed-off-by: Luka Perkov <[email protected]>

boards.cfg: remove trailing whitespace

Commit 93e14596 "Coding Style cleanup: replace leading SPACEs by TABs"
added trailing whitespace to a single line of boards.cfg. I presume this
was unintentional, and it causes the file to change after running it
through the reformat.py script. Remove the offending character.

Signed-off-by: Paul Burton <[email protected]>

cmd_nvedit.c: Add env exists command

env exists is a way to test (in hush) if an environment variable
exists.  A workaround existed using printenv but this new command
doesn't require all the stdout/stderr redirection to prevent
printing information to the screen.

Example:
$ set testexists 1
$ env exists testexists && echo "yes"
yes
$ env exists testexists || echo "no"
$ set testexists
$ env exists testexists && echo "yes"
$ env exists testexists || echo "no"
no
$

Signed-off-by: Andrew Ruder <[email protected]>

MAKEALL: rename boards_by_* functions to targets_by_*

We expect boards_by_* function to return the 7th filed, 'Target',
not the 6th field, 'Board name'.

So the function names, boards_by_* are a little misleading,
and should be renamed to targets_by_*.

Signed-off-by: Masahiro Yamada <[email protected]>
Cc: Albert ARIBAUD <[email protected]>

MAKEALL: fix boards_by_field function

Commit 27af930e changed the boards.cfg format
and it changed boards_by_field() function incorrectly.
For tegra cpus it returned Board Name field,
not Target field.

This commit restores the behavior prior to 27af930e in the right way.

Signed-off-by: Masahiro Yamada <[email protected]>
Cc: Albert ARIBAUD <[email protected]>

MAKEALL: fix a bug to use CROSS_COMPILE_<ARCH>

Commit 27af930e changed the boards.cfg format but
missed to change get_target_arch() fuction.
This commit adjusts it for CROSS_COMPILE_<ARCH>
to work correctly.

Signed-off-by: Masahiro Yamada <[email protected]>
Cc: Albert ARIBAUD <[email protected]>

MAKEALL: fix awk warning message

If you do `./MAKEALL -M ` or `./MAKEALL -m`
GNU awk would display warnings like follows:

    awk: warning: escape sequence `\ ' treated as plain ` '

In the first place, we do not explicitly set the field separator.

Signed-off-by: Masahiro Yamada <[email protected]>
Cc: Albert ARIBAUD <[email protected]>

sparc: include config.h to start.S

arch/sparc/cpu/leon3/start.S requires CONFIG_SYS_SPARC_NWINDOES
to be defined:

  #ifndef CONFIG_SYS_SPARC_NWINDOWS
  #error Must define number of SPARC register windows, default is 8
  #endif

But it missed to include <config.h>, which always ended up in compile error.

This commit fixes this problem.

Signed-off-by: Masahiro Yamada <[email protected]>
Cc: Daniel Hellstrom <[email protected]>

EP88x: remove remainders of dead board

Commit 1b0757e deleted the EP88x entry from boards.cfg file.
But it missed to remove include/configs/EP88x.h and board/ep88x/.

This commit removes them and adds EP88x to README.scrapyard.

Signed-off-by: Masahiro Yamada <[email protected]>
Cc: Wolfgang Denk <[email protected]>

README.scrapyard: fix broken format

Some tabs have been replaced with spaces
because doc/README.scrapyard is consistently using spaces.

Signed-off-by: Masahiro Yamada <[email protected]>

kup: Delete an unused Makefile

Signed-off-by: Masahiro Yamada <[email protected]>
Cc: Klaus Heydeck <[email protected]>

Merge branch 'master' of git://denx.de/git/u-boot-coldfire

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

ColdFire: fix some typoes for CF platform

Signed-off-by: Jason Jin <[email protected]>

coldfire: cpu5282: increase malloc space to fix crash on start u-boot

The malloc space is to small to boot, the current uboot 2013.10-rcX,
This will fix the startup problems by increasing the mallog space to 4MiB.

Signed-off-by: Jens Scharsig (BuS Elektronik) <[email protected]>

Makfile: fix a rule to build u-boot.sb

Signed-off-by: Masahiro Yamada <[email protected]>

freescale: p1_p2_rdb_pc: rename COBJS-y to obj-y

Signed-off-by: Masahiro Yamada <[email protected]>

fpga: Add support for gzip images with bitstreams

Here is the set of command which has been performed
to proof this feature.

gzip < fpga.bin > fpga.bin.gz
mkimage -A arm -O u-boot -T firmware -C gzip \
-a 20000000 -n "zc702_fpga_bin" -d fpga.bin.gz fpga.bin.gz.ub

tftp 100000 fpga.bin.gz.ub
fpga loadmk 0 100000

This flow should speedup loading bitstream data
from external memory and save image footprint in non volatile
memory.

Signed-off-by: Michal Simek <[email protected]>
Acked-by: Jagannadha Sutradharudu Teki <[email protected]>

fpga: zynqpl: Do not place bitstream below 1MB

DMA doesn't work when src is placed below 1MB limit.

Signed-off-by: Michal Simek <[email protected]>
Acked-by: Jagannadha Sutradharudu Teki <[email protected]>

fpga: zynqpl: Add dcache flush support

Buffers must be cache and dma aligned.

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

pxe: fix handling of absolute paths

pxelinux and syslinux differ in their handling of absolute paths in menu
files. A pxelinux path is aways prepended with the bootfile path while
syslinux allows for absolute paths. u-boot was always treating a leading
/ as an absolute path breaking some pxelinux setups. Fix this by adding
a flag to distinguish pxelinux vs. syslinux behavior.

Reported-by: Ian Campbell <[email protected]>
Signed-off-by: Rob Herring <[email protected]>

ARM: versatile: convert to common timer code

Convert versatile to use the commmon timer code.

Signed-off-by: Rob Herring <[email protected]>

ARM: tegra: convert to common timer code

Convert tegra to use the commmon timer code.

Signed-off-by: Rob Herring <[email protected]>

ARM: socfpga: convert to common timer code

Convert socfpga to use the commmon timer code.

Signed-off-by: Rob Herring <[email protected]>

ARM: vexpress: convert to common timer code

Convert vexpress to use the commmon timer code.

Signed-off-by: Rob Herring <[email protected]>

ARM: mx25: convert to common timer code

Convert mx25 to use the commmon timer code.

Signed-off-by: Rob Herring <[email protected]>

ARM: highbank: convert to common timer code

Convert highbank to use the commmon timer code.

Signed-off-by: Rob Herring <[email protected]>

sh: convert to common timer code

Convert sh to use the commmon timer code. Remove reset_timer and
set_timer as they are unused on sh.

Signed-off-by: Rob Herring <[email protected]>

Introduce common timer functions

Many platforms duplicate pretty much the same timer code yet they all have
a 32-bit freerunning counter register. Create a common implementation that
minimally requires 2 or 3 defines to add timer support:

CONFIG_SYS_TIMER_RATE - Clock rate of the timer counter
CONFIG_SYS_TIMER_COUNTER - Address of 32-bit counter
CONFIG_SYS_TIMER_COUNTS_DOWN - Define if counter counts down

All functions are weak or ifdef'ed so they can still be overriden by any
platform.

Signed-off-by: Rob Herring <[email protected]>

examples: enable gc-sections option

This fixes building time.c when unreferenced functions are added.

Signed-off-by: Rob Herring <[email protected]>

config: remove platform CONFIG_SYS_HZ definition part 2/2

Remove platform CONFIG_SYS_HZ definition for configs a-z*.

Signed-off-by: Rob Herring <[email protected]>

config: remove platform CONFIG_SYS_HZ definition part 1/2

Remove platform CONFIG_SYS_HZ definition for configs A-Z*.

Signed-off-by: Rob Herring <[email protected]>

config: consolidate CONFIG_SYS_HZ definition

According to the README, CONFIG_SYS_HZ must be 1000 and most platforms
follow that. In preparation to remove CONFIG_SYS_HZ from all these
platforms, provide a common definition. The platforms which use a value
other than 1000 will get build warning now. These configs are:

include/configs/M5271EVB.h:#define CONFIG_SYS_HZ                        1000000
include/configs/balloon3.h:#define      CONFIG_SYS_HZ                   3250000         /* Timer @ 3250000 Hz */
include/configs/idmr.h:#define CONFIG_SYS_HZ                    (50000000 / 64)
include/configs/mini2440.h:#define CONFIG_SYS_HZ                        1562500
include/configs/mx1ads.h:#define CONFIG_SYS_HZ                  3686400
include/configs/omap3_zoom2.h:#define CONFIG_SYS_HZ                     ((V_SCLK) / (2 << CONFIG_SYS_PTV))
include/configs/omap730p2.h:#define CONFIG_SYS_HZ                       ((CONFIG_SYS_CLK_FREQ)/(2 << CONFIG_SYS_PTV))
include/configs/palmld.h:#define        CONFIG_SYS_HZ                   3250000         /* Timer @ 3250000 Hz */
include/configs/palmtc.h:#define        CONFIG_SYS_HZ                   3686400         /* Timer @ 3686400 Hz */
include/configs/rsk7203.h:#define CONFIG_SYS_HZ                 (CONFIG_SYS_CLK_FREQ / CMT_CLK_DIVIDER)
include/configs/rsk7264.h:#define CONFIG_SYS_HZ         (CONFIG_SYS_CLK_FREQ / CMT_CLK_DIVIDER)
include/configs/rsk7269.h:#define CONFIG_SYS_HZ         (CONFIG_SYS_CLK_FREQ / CMT_CLK_DIVIDER)
include/configs/scb9328.h:#define CONFIG_SYS_HZ                 3686400      /* incrementer freq: 3.6864 MHz */
include/configs/versatile.h:#define CONFIG_SYS_HZ                       (1000000 / 256)
include/configs/zipitz2.h:#define       CONFIG_SYS_HZ                   3250000         /* Timer @ 3250000 Hz */

Signed-off-by: Rob Herring <[email protected]>

config: Add a default CONFIG_SYS_PROMPT

The definitions for CONFIG_SYS_PROMPT are varied with little reason other
than to display the board name. Over half the definitions are "==> ", so
make this the default. The rest of the boards remain unchanged to avoid
breaking any external scripts expecting a certain prompt.

Signed-off-by: Rob Herring <[email protected]>
Reviewed-by: Fabio Estevam <[email protected]>

Merge branch 'master' of git://git.denx.de/u-boot-blackfin

Easy to resolve conflict on the GPIO change.

Conflicts:
arch/blackfin/cpu/Makefile

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

blackfin: Move machine specific gpio_port_t structure back to blackfin arch folder.

The gpio register mappings are different among blackfin processors.

Signed-off-by: Steven Miao <[email protected]>
Signed-off-by: Sonic Zhang <[email protected]>

blackfin: fix a warning in arch/blackfin/cpu/cpu.c

This commit fixes:
    cpu.c:107: warning: ‘noreturn’ function does return

Signed-off-by: Masahiro Yamada <[email protected]>
Signed-off-by: Sonic Zhang <[email protected]>

board/keymile/kmp204x/Makefile: Convert to SPDX tag

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

board: powerpc: convert more makefiles to Kbuild style

Cc: Masahiro Yamada <[email protected]>
Cc: Wolfgang Denk <[email protected]>
Cc: Kim Phillips <[email protected]>
Cc: York Sun <[email protected]>
Signed-off-by: Tom Rini <[email protected]>

Makefile: convert makefiles to Kbuild style and delete grep switch

We have converted all makefiles needed to build $(LIBS).

Until this commit we used to grep switch so that U-Boot style
and Kbuild style makefiles coexist.
But we do not need any more.

Goint forward, use always Kbuild style Makefile when adding
a new Makefile

Signed-off-by: Masahiro Yamada <[email protected]>

dts, api, test: convert makefiles to Kbuild style

Signed-off-by: Masahiro Yamada <[email protected]>

post: convert makefiles to Kbuild style

This commit also deletes post/rules.mk,
which in not necessary any more.

Signed-off-by: Masahiro Yamada <[email protected]>

board: powerpc: convert makefiles to Kbuild style

Signed-off-by: Masahiro Yamada <[email protected]>
Cc: Wolfgang Denk <[email protected]>
Cc: Kim Phillips <[email protected]>
Cc: York Sun <[email protected]>
Cc: Stefan Roese <[email protected]>

board: arm: convert makefiles to Kbuild style

Signed-off-by: Masahiro Yamada <[email protected]>
Cc: Albert ARIBAUD <[email protected]>
Cc: Andreas Bießmann <[email protected]>
Cc: Stefano Babic <[email protected]>
Cc: Prafulla Wadaskar <[email protected]>
Cc: Minkyu Kang <[email protected]>
Cc: Vipin Kumar <[email protected]>
Cc: Tom Warren <[email protected]>
Cc: Tom Rini <[email protected]>

blackfin: convert makefiles to Kbuild style

Signed-off-by: Masahiro Yamada <[email protected]>
Cc: Sonic Zhang <[email protected]>

m68k: convert makefiles to Kbuild style

Signed-off-by: Masahiro Yamada <[email protected]>
Cc: Jason Jin <[email protected]>

x86: convert makefiles to Kbuild style

Signed-off-by: Masahiro Yamada <[email protected]>
Cc: Simon Glass <[email protected]>

nios2: convert makefiles to Kbuild style

Signed-off-by: Masahiro Yamada <[email protected]>
Cc: Thomas Chou <[email protected]>

nds32: convert makefiles to Kbuild style

Signed-off-by: Masahiro Yamada <[email protected]>
Cc: Macpaul Lin <[email protected]>

mips: convert makefiles to Kbuild style

Signed-off-by: Masahiro Yamada <[email protected]>
Cc: Daniel Schwierzeck <[email protected]>

microblaze: convert makefiles to Kbuild style

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

openrisc: convert makefiles to Kbuild style

Signed-off-by: Masahiro Yamada <[email protected]>
Cc: Stefan Kristiansson <[email protected]>

avr32: convert makefiles to Kbuild style

Signed-off-by: Masahiro Yamada <[email protected]>
Cc: Andreas Bießmann <[email protected]>

sh: convert makefiles to Kbuild style

Signed-off-by: Masahiro Yamada <[email protected]>
Cc: Nobuhiro Iwamatsu <[email protected]>

sparc: convert makefiles to Kbuild style

Signed-off-by: Masahiro Yamada <[email protected]>
Cc: Daniel Hellstrom <[email protected]>

sh: Do not include start.o in lib$(CPU).o

Signed-off-by: Masahiro Yamada <[email protected]>
Cc: Nobuhiro Iwamatsu <[email protected]>

sparc: fix a link error

Before this commit, arch/sparc/lib/Makefile used
both COBJS and COBJS-y.
And it missed to add COBJS-y into OBJS.
This means bootm.o was never compiled even if
CONFIG_CMD_BOOTM=y

Signed-off-by: Masahiro Yamada <[email protected]>
Cc: Daniel Hellstrom <[email protected]>

ARM: s5pc, exynos: move Samsung ARM SoC specific code under arch/arm/

This patch moves S5PC, EXYNOS specific directory entries
from the toplevel Makefile to arch/arm/cpu/armv7/Makefile
using Kbuild descending feature.

Signed-off-by: Masahiro Yamada <[email protected]>
Cc: Minkyu Kang <[email protected]>

ARM: omap: move OMAP specific code under arch/arm/

This patch moves OMAP specific directory entries
from the toplevel Makefile and spl/Makefile
to arch/arm/cpu/armv7/Makefile using Kbuild descending feature.

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

ARM: tegra: move Tegra specific code under arch/arm/

This patch moves Tegra specific directory entries
from the toplevel Makefile and spl/Makefile
to arch/arm/cpu/*/Makefile using Kbuild descending feature.

Signed-off-by: Masahiro Yamada <[email protected]>
Cc: Tom Warren <[email protected]>

fs: move some file system to fs/Makefile

This commit moves some subdirectories of fs
from the toplevel Makefile to fs/Makefile
using Kbuild descending feature.

Signed-off-by: Masahiro Yamada <[email protected]>
Acked-by: Simon Glass <[email protected]>

drivers: move some drivers to drivers/Makefile

This commit moves some drivers subdirectory entry
from the toplevel Makefile to drivers/Makefile
using Kbuild descending feature.

Signed-off-by: Masahiro Yamada <[email protected]>

Makefile: support descending down to subdirectories

This patch tweaks scripts/Makefile.build to allow
the build system to descend into subdirectories like Kbuild.

To use this feature, use "obj-y += foo/" syntax.

Example:
    obj-$(CONFIG_FOO) += foo/

Signed-off-by: Masahiro Yamada <[email protected]>
Cc: Simon Glass <[email protected]>

board: ti: convert makefiles to Kbuild style

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

powerpc: convert makefiles to Kbuild style

Note:
arch/powerpc/cpu/mpc8260/Makefile is originally like follows:

    ---<snip>---
    START   = start.o kgdb.o
    COBJS   = traps.o serial_smc.o serial_scc.o cpu.o cpu_init.o speed.o \
    ---<snip>---
    COBJS-$(CONFIG_ETHER_ON_SCC) = ether_scc.o
    ---<snip>---
    $(LIB): $(OBJS)
            $(call cmd_link_o_target, $(OBJS) $(obj)kgdb.o)

The link rule `$(call cmd_link_o_target, $(OBJS) $(obj)kgdb.o)'
is weird.
kbdg.o is not included in $(OBJS) but linked into $(LIB)
and $(LIB) is not dependent on kgdb.o.
(Broken dependency tracking)

So,
    START   = start.o kgdb.o
shoud have been
    START   = start.o
    SOBJS   = kgdb.o

That is why this commit adds kgdb.o to obj-y, not to extra-y.

Signed-off-by: Masahiro Yamada <[email protected]>
Cc: Wolfgang Denk <[email protected]>
Cc: Stefan Roese <[email protected]>

sandbox: convert makefiles to Kbuild style

Signed-off-by: Masahiro Yamada <[email protected]>
Cc: Simon Glass <[email protected]>
Acked-by: Simon Glass <[email protected]>

disk: convert a makefile to Kbuild style

Signed-off-by: Masahiro Yamada <[email protected]>

lib: convert makefiles to Kbuild style

Signed-off-by: Masahiro Yamada <[email protected]>

net: convert a makefile to Kbuild style

Signed-off-by: Masahiro Yamada <[email protected]>

common: convert makefiles to Kbuild style

Signed-off-by: Masahiro Yamada <[email protected]>

fs: convert makefiles to Kbuild style

Signed-off-by: Masahiro Yamada <[email protected]>

drivers: convert makefiles to Kbuild style

Signed-off-by: Masahiro Yamada <[email protected]>

drivers: usb: convert makefiles to Kbuild style

Signed-off-by: Masahiro Yamada <[email protected]>

drivers: mtd: convert makefiles to Kbuild style

Signed-off-by: Masahiro Yamada <[email protected]>

drivers: net: convert makefiles to Kbuild style

Signed-off-by: Masahiro Yamada <[email protected]>

ARM: imx-common: convert makefiles to Kbuild style

Multiple targets are included in arch/arm/imx-common/Makefile
In order to refactor it,
we need to tweak Makefile and spl/Makefile.

Signed-off-by: Masahiro Yamada <[email protected]>

ARM: convert makefiles to Kbuild style

Signed-off-by: Masahiro Yamada <[email protected]>

arm720t: convert makefiles to Kbuild style

Signed-off-by: Masahiro Yamada <[email protected]>

arm920t: convert makefiles to Kbuild style

Signed-off-by: Masahiro Yamada <[email protected]>

arm926ejs: convert makefiles to Kbuild style

Note1:
In arch/arm/cpu/arm926ejs/spear/Makefile
START := start.o
was changed
extra-$(CONFIG_SPL_BUILD) := start.o
because spear/start.o is only used for SPL.

Note2:
START := start.o
was missing from arch/arm/cpu/arm926ejs/mxs/Makefile.
This commit simply adds
extra-$(CONFIG_SPL_BUILD) := start.o

Signed-off-by: Masahiro Yamada <[email protected]>

armv7: convert makefiles to Kbuild style

Signed-off-by: Masahiro Yamada <[email protected]>

Makefile: prepare for using Kbuild-style Makefile

Every makefile in sub directories has common lines
at the top and the bottom.
This commit pushes the common parts into script/Makefile.build.

Going forward sub-makefiles only need to describe this part:

    COBJS := ...
    COBJS += ...
    SOBJS := ...

But using obj-y is preferable to prepare for switching to Kbuild.

The conventional (non-Kbuild) Makefile style is still supported.
This is achieved by greping the Makefile before entering into it.
U-Boot conventional sub makefiles always include some other makefiles.
So the build system searches a line beginning with "include" keyword
in the makefile in order to distinguish which style it is.
If the Makefile include a "include" line, we assume it is a conventional
U-Boot style. Otherwise, it is treated as a Kbuild-style makefile.

With this tweak, we can switch sub-makefiles
from U-Boot style to Kbuild style little by little.

obj-y := foo/
syntax (descending into the sub directory) is not supportd yet.
It will be implemented in the upcomming commit.

Signed-off-by: Masahiro Yamada <[email protected]>
Cc: Simon Glass <[email protected]>
Cc: Tom Rini <[email protected]>

nand_util.c: Correct licensing

Prior to SPDX licensing this file was GPL-2.0 with Freescale granting
rights for "or later" for their contributed code.  We incorrectly moved
this file to GPL-2.0+, so correct it to GPL-2.0.  In addition we cannot
easily denote in the file where or what code is "or later", so just set
that aside for now and the file as a whole is GPL-2.0 regardless.

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

powerpc/esdhc: Add 3.3v voltage support in esdhc capacity register

T4240QDS eSDHC host capabilities reigster should have VS33 bit define.
Add quirk CONFIG_SYS_FSL_MMC_HAS_CAPBLT_VS33 to deal with capacity
missing

Signed-off-by: Roy Zang <[email protected]>
Signed-off-by: Haijun Zhang <[email protected]>
Acked-by: Pantelis Antoniou <[email protected]>

esdhc: memset mmc struct before putting into use

struct mmc should be clear to all '0' after malloc to avoid
unexpect variable value.

Like mmc->has_init = xxx.
In this case mmcinfo will believe the card had been initialized before
and skip the initialization.

Test on P5040 and T4240,
Error Log:

=> mmcinfo
Device: FSL_SDHC
Manufacturer ID: 0
OEM: 0
Name: Tran Speed: 0
Rd Block Len: 0
MMC version 0.0
High Capacity: No
Capacity: 0 Bytes
Bus Width: 0-bit
=>

Signed-off-by: Haijun Zhang <[email protected]>
Signed-off-by: Xie Shaohui-B21989 <[email protected]>
Tested-by: Ryan Barnett <[email protected]>
Acked-by: Pantelis Antoniou <[email protected]>