kbuild: switch over to thin archive ...

Starting v4.13, Linux always uses thin archive instead of relocatable
ELF to combine builtin objects.

The thin archive is basically a text file that only contains paths to
object files.

As Linux commit 98ced886dd79 mentioned, this has a lot of benefits:

 - save disk space for builds

 - speed-up building a little

 - fix some issues when liking a giant executable (for example,
   allyesconfig on ARM Linux)

 - work better with dead code elimination

The last one is important for barebox because pbl highly relies on the
dead code elimination (-f{function,data}-sections and --gc-sections)
dropping symbols that are unreachable from the linker's entry point.
The dead code elimination does not work if the same symbol names are
used in the pevious incremental link mechanism because the same name
symbols all go into the same section.

Commit a83c97f2a406 ("ARM: socfpga: generate smaller images when
multiple boards are selected") worked around it by giving a dedicate
section to each of the same name symbols. This workaround can go away.

built-in.o was renamed to built-in.a since it is now an archive.
built-in-pbl.o was renamed to built-in.pbl.a for consistency.

Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

kbuild: unify barebox and pbl build commands ...

The current Linux uses the unified build command for built-in objects
and module objects. This is possible because part-of-module is set to
y when the target is being built for a module. If so, quiet_modtag is
set to [M], and modkern_cflags is set to KBUILD_CFLAGS_MODULE instead
of KBUILD_CFLAGS_KERNEL.

Currently, barebox uses a separate build command for pbl, but the
difference between cmd_cc_o_c and cmd_pbl_cc_o_c is just
"-D__PBL__ $(PBL_CPPFLAGS)".

By using the same approach as Linux, they can be unified.

Set part-of-pbl flag to y when the target is being built for pbl.
Merge cmd_as_o_S and cmd_link_o_target as well.

This also changes the shortlog style as follows:

  PBLCC  ->  CC [P]
  PBLAS  ->  AS [P]
  PBLLD  ->  LD [P]

This is more consistent with the policy in the Linux build system.

In Linux, a capital letter enclosed square blackets indicates which
target the object is being built for.

  CC      object built for built-in
  CC [M]  object built for modules
  CC [U]  object built for userspace (proposed for Linux 5.8-rc1)

The idea behind this is we use the same compiler $(CC) for them.
Only the difference is the compiler flags. On the other hand, host
programs are compiled by a different tool, $(HOSTCC), hence the
shortlog is 'HOSTCC'.

Barebox supports module builds (it seems), but the [P] notation does
not conflict with [M] because pbl has a limited memory footprint.
We never expect pbl can support the module feature.

Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

kbuild: resync modkern_{c, a}flags and quiet_modtag with Linux 5.7-rc6 ...

The latest Linux calculates them more simply; part-of-module is set
when the target is being built for a module.

Also, rename real-objs-{y,m} to real-obj-{y,m}.

Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

kbuild: append $(bbenv-y) to obj-y earlier ...

You can save one line.

Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

kbuild: resync mkdir code with Linux 5.7-rc6 ...

Kbuild automatically creates the output directories for O= builds.
Previously it called mkdir too much. Linux optimized this a lot.
Let's import the outcome so it works faster.

Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

kbuild: remove obj-dtb-y, pbl-dtb-y, lwl-dtb-y syntax ...

The embedded DTBs are compiled by the chain of pattern rules as follows:

  %.dts -> %.dtb -> %.dtb.S -> %.dtb.o       for barebox proper
  %.dts -> %.dtb -> %.dtb.S -> %.dtb.pbl.o   for pbl

Barebox introduced {obj,pbl,lwl}-dtb-y syntax to put the intermediate
files into extra-y. The purposes of doing so were:

 [1] prevent GNU Make from deleting the intermediate files
 [2] include .*.cmd files

In contrast, Linux does not use a special syntax for embedding DTBs
into the kernel.

For example, as you see in arch/sh/boot/dts/Makefile of Linux 5.6,

   obj-y += <basename>.dtb.o

... just works.

This is because scripts/Kbuild.include specifies .SECONDARY to cater
to [1], and scripts/Makefile.build adds the intermediates to 'targets'
to deal with [2].

Barebox had already imported the same code from Linux, so you can use
obj-y instead of obj-dtb-y, like Linux.

pbl-dtb-y and lwl-dtb-y are barebox-specific cases, so I added the
%.dtb.pbl.o pattern to intermediate_targets.

Going forward, please use obj-y, pbl-y, or lwl-y.

Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

kbuild: rename pbl object pbl-*.o to *.pbl.o ...

Currently, pbl objects are output to:

  <directory-path>/pbl-<basename>.o

This commit changes as follows:

  <directory-path>/<basename>.pbl.o

The motivation is not only to get rid of the ugly code introduced by
commit 257abdaa36c8 ("Do not rm the path from pbl-y target"), but also
to make it easier to remove obj-dtb-y, pbl-dtb-y, lwl-dtb-y syntax in
the next commit.

Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

kbuild: build modules by 'make' ...

In Linux, 'make' and 'make all' work equivalently; it builds both
vmlinux and modules.

In barebox, strangely, 'make all' builds both barebox and modules,
but 'make' builds only barebox because KBUILD_MODULES is unset.

The check for 'ifeq ($(MAKECMDGOALS),)' is missing since the day-one.

Fixes: b02e0966412c ("add additional linking stage for modules")
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

kconfig: add 'option modules' to allow tristate options to become m ...

Since commit 18895514dc5e ("kconfig: update to v3.18-rc6"), tristate
symbols cannot become 'm'.

In old days of Linux, CONFIG_MODULES was the special option hard-coded
in Kconfig.

The current Kconfig does not know which option is it. You need to
specify 'option modules' to teach Kconfig that it is the module
support switch.

Add 'option modules' to revive the module builds.

Fixes: 18895514dc5e ("kconfig: update to v3.18-rc6")
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

kbuild: add -std=gnu89 to KBUILD_CFLAGS ...

Barebox imports lots of code from Linux, so should use the same
language standard as Linux.

Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

Merge branch 'for-next/stm32'

Merge branch 'for-next/powerpc'

Merge branch 'for-next/mxs'

Merge branch 'for-next/misc'

Merge branch 'for-next/kconfig'

Merge branch 'for-next/kbuild'

Merge branch 'for-next/dts'

Merge branch 'for-next/cpsw'

Merge branch 'for-next/compile-test'

Release v2020.05.0 ...

Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

phy: drop of_phy_simple_xlate() stub ...

The function of_phy_simple_xlate() is not used.

Signed-off-by: Antony Pavlov <antonynpavlov@gmail.com>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

net: designware: eqos: attach PHY earlier ...

If the PHY isn't driving the refclock, the software reset of the
controller will time out. Some PHYs need some board specific
configuration to properly drive the reflock. Attach the PHY before
attempting the software reset, so PHY fixups have a chance to run.

Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

regulator: stpmic1: move to coredevice level ...

Those regulators are needed by other devices. Probe them early.

Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

mfd: stpmic1: move to coredevice level ...

The PMIC provides regulators, which are needed before other
devices can probe.

Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

i2c: add coredevice_i2c_driver() macro ...

Some i2c device want to be coredevices. Add a macro to avoid
boilerplate code.

Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

mci: stm32_sdmmc2: move command timeout errors to debug level ...

Command timeouts are sometimes a normal operating condition
(for example when probing whether the connected card is SD or MMC),
so we don't want to print an error message for all of them. Move
those messages to the debug level, the MCI core will print some
more meaningful messages on most command errors anyways.

Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

mci: stm32_sdmmc2: don't retry commands ...

No other MCI does retry failing commands on its own (if at all
this should be done by the MCI core). Remove the retry, as we
don't want to paper over board level issues.

Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

image: add stm32 images to gitignore ...

Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

scripts: add stm32image to .gitignore ...

Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

video: backlight-pwm: set backlight dev.parent to hardware dev ...

backlight_register allocates a new backlight device. Populating the
device's parent field with the hardware device gives us more
informational devinfo output.

Signed-off-by: Ahmad Fatoum <ahmad@a3f.at>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>