i2c_probe command: Use kstrtoint ...

Use kstrtoint rather than simple_strtoul to catch erroneous input like
"i2c1" which previously was silently interpreted as "0".

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

commands: boot: create boot entries on demand ...

We currently create all boot entries before attempting boot. This is
less than optimal, because this may involve probing devices that won't
be used for actual boot.

Fix this by not creating boot entries till the previous boot argument
(command line argument or boot.default word) was found to be unbootable.

This means that "boot mmc1 mmc0" will now not touch mmc0 if mmc1 had a
bootable entry. This is only done when no menu or list was requested.

As the boot entries are in a linked list, the allocation done for each
boot argument could be omitted, but as the saving from skipping an
allocation is easily dwarfed by the boot medium access, we just
reallocate and enjoy the improved code clarity.

Signed-off-by: Ahmad Fatoum <a.fatoum@pengutronix.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

commands: boot: refactor boot entry creation to use iterators ...

We currently create all boot entries before attempting boot. This is
less than optimal, because this may involve probing devices that won't
be used for actual boot.

In preparation for changing this, refactor the code, so we only have one
loop we need to touch.

Signed-off-by: Ahmad Fatoum <a.fatoum@pengutronix.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

ARM: stm32mp: dk2: barebox_set_model to shorter name ...

The full "STMicroelectronics STM32MP157C-DK2 Discovery Board" is a tad
too long. Abbreviate it to STM32MP157C-DK2 instead.

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

i2c: tegra: correct error message ...

reset_control_get returns NULL when the controller reset is missing.
The error pointer is used for malformed reset controller specification.

Correct the message accordingly.

Signed-off-by: Ahmad Fatoum <a.fatoum@pengutronix.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

watchdog: dw_wdt: remove duplicated error message ...

The wd command already warns that "Watchdog cannot be disabled",
when the stop operation returns -ENOSYS. We do that now, so telling the
user that it will not stop is superfluous. Remove it.

Signed-off-by: Ahmad Fatoum <a.fatoum@pengutronix.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

watchdog: dw_wdt: inform user on missing reset control line ...

The driver has an error message on probe and one on each watchdog stop
attempt that warns the user if the reset line is missing.

Missing reset line (because the "reset"-property is missing) is
indicated by a NULL pointer though, so these warnings were only
triggered when the reset controller specification is malformed.

Fix this by propagating malformed reset pointer specification and
continuing only if it's either valid or completely missing.

Signed-off-by: Ahmad Fatoum <a.fatoum@pengutronix.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

net: designware: socfpga: fix possible invalid pointer deref ...

In cases where the reset controller specification in the device tree is
malformed, we get an error pointer back from reset_control_get. This
compares unequal to NULL and would cause an access violation when passed
to reset_control_(de)?assert.

Fix this by propagating the error. When the reset controller is missing,
reset_control_(de)?assert will be passed NULL pointers, rendering them
no-ops.

Signed-off-by: Ahmad Fatoum <a.fatoum@pengutronix.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

mci: stm32_sdmmc2: don't ignore reset_control_get errors ...

reset_control_get returns a NULL pointer when no resets were specified
via device tree properties. If there's a malformed "resets"-property, we
get an error pointer. This error should be propagated, only the NULL
returns are the ones we can safely ignore.

Signed-off-by: Ahmad Fatoum <a.fatoum@pengutronix.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

watchdog: dw_wdt: populate driver_d.name ...

Apparently, somewhere between v2017.01.0 and v2020.01.0, barebox became
dependent on drivers having a name as having this driver enabled in
v2020.01.0 has barebox crash due to a null pointer dereference in the
device_match's second strcmp.

A previous commit introduced a palatable warning for this case.
Now in order to be able to use the driver again, give it a name.

Cc: Steffen Trumtrar <s.trumtrar@pengutronix.de>
Signed-off-by: Ahmad Fatoum <a.fatoum@pengutronix.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

driver: bail out, don't crash, if drv->name is not set ...

Currently, we expect each driver to have a name. If a driver doesn't,
we run into a NULL pointer dereference. Make this error scenario more
pleasant by checking if a name is set and failing otherwise.

The only in-tree driver affected by this is drivers/watchdog/dw_wdt.c,
which will be fixed in the follow-up commit.

Affected drivers can be determined with following pipeline, which should
return all driver_d structure instantiations that don't contain a name:

  ack --A 4 'struct driver_d.*' | perl -pe 's/^--$/\n/' | \
	perl -000 -ne 'print if /=.*\{.*\}/s && !/name/s'

With this change, these drivers should now give a more pleasant message:

  ERROR: initcall dw_wdt_driver_register+0x1/0xc failed: Invalid argument

Signed-off-by: Ahmad Fatoum <a.fatoum@pengutronix.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

clocksource: rename driver for ARMv8 Timer to ARM architected timer ...

The driver matches against both "arm,armv7-timer" and "arm,armv8-timer"
compatibles, thus rename it to better reflect that it's not only ARMv8
specific.

Only functional change intended is name changes. The symbol is default
y, so the rename shouldn't cause it to get lost with olddefconfig.

Signed-off-by: Ahmad Fatoum <a.fatoum@pengutronix.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

scripts: imx: Remove unused variable ...

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

fs: jffs2: add initial support for reading jffs2 ...

Import the jffs2 filesystem code from Linux v5.5-rc1 and convert it
to work with barebox.

Writing is *not* supported.

Testing was done with a n25q256a-compatible QuadSPI chip on a
SoCFPGA-based Socrates board.

Testing was done with a combination of:
	user@somelinuxhost: mkfs.jffs2 --eraseblock=4 -d fs/jffs2/ -o jffs2.img
	root@target:~ flash_erase -j /dev/mtd5 0 0
	root@target:~ dd if=jffs2.img of=/dev/mtd5
	barebox@EBV SOCrates:/ mount /dev/mtd0.data
	(...)
	mounted /dev/mtd0.data on /mnt/mtd0.data

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

sandbox: hostfile: fix UB-inducing shift ...

On 32-bit hosts, UBSan reports:

====================================================================
UBSAN: Undefined behaviour in ./arch/sandbox/board/hostfile.c:135:12
shift exponent 32 is too large for 32-bit type 'long unsigned int'
====================================================================

Fix this by using a unsigned long long for base. The upper 32-bit
won't be set, but device tree fixups can now read them.

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

sandbox: fix signed integer overflow ...

On a 32-bit host system, UBSan reports:

===============================================================
UBSAN: Undefined behaviour in ./arch/sandbox/os/common.c:115:32
signed integer overflow:
83598 * 1000000000 cannot be represented in type 'long int'
===============================================================

Fix this.

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

commands/test: Implement -b and -c to test for character and block devices ...

These match the same options on coreutil's test(1).

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

commands/test: Improve option parsing to handle "]" less special ...

Testing for *ap[1] != ']' is bogus during option processing. Instead
test if there are options left to be processed.

This fixes the return value for e.g.

	test -z ']lala'

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

commands/test: Bail out on incomplete command line options ...

This makes test emit an error (and fail) on e.g.

	test -f

and also on unimplemented options like

	test -c /dev/null

.

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

regulator: fixed: take regulator-boot-on into account ...

There are regulators which are turned on by default (hw-design) and
shouldn't be disabled during the boot. Upstream uses the
regulator-boot-on dt-property for it:

8<--------------------------------------------------
  regulator-boot-on:
    description: bootloader/firmware enabled regulator
    type: boolean
8<--------------------------------------------------

The difference between boot-on and always-on is the linux suspend
handling. Regulators marked as boot-on can be disabled during suspend.
This case isn't interesting for barebox so we can mark those as
always-on too.

Signed-off-by: Marco Felsch <m.felsch@pengutronix.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

ubsan: use a big if block in Kconfig ...

A side effect (which was the motivation for this change) is that the
config item CONFIG_UBSAN_ALIGNMENT doesn't appear in configurations that
have CONFIG_UBSAN disabled.

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

Documentation/sandbox: Add hint how to end a sandboxed barebox ...

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

images: Use separate LDFLAGS for PBL ...

When setting CONFIG_PBL_RELOCATABLE=y (selected by PBL_MULTI_IMAGES) and
CONFIG_RELOCATABLE=n (trying to reduce image size), the PBL is also
linked with "-static" instead of "-pie". The result is a non-working
PBL.

As a side effect, also get rid of erroneous "-Map barebox.map" when
linking the PBL.

Only changed for ARM, are any other platforms affected?

Signed-off-by: Christian Eggers <ceggers@arri.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

gpiolib: introduce helper functions working on gpio_info structs ...

gpioinfo_*() assume their gpio_info pointer parameter to be valid and
don't ensure the gpio to be requested.

This drops several checks for being requested and allows further
extensions to work with unrequested gpios.

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

gpiolib: reorder functions ...

gpio_request_one() and gpio_request_array() make use of
gpio_direction_*, so it is natural to define the former them further low
in the file. It doesn't matter for the compiler here as all functions
are declared in a header, but the next commit adds a few more functions
and it benefits from this reordering by not needing forward
declarations.

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

common: make CONFIG_BAREBOX_UPDATE visible ...

The barebox update infrastructure is also useful without
CONFIG_CMD_BAREBOX_UPDATE (e.g. from a own barebox_main() with
CONFIG_SHELL_NONE).

Signed-off-by: Christian Eggers <ceggers@arri.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

gadget: f_fastboot: Fix memory leak ...

bootm_data::os_file is not required to be a copied string. It isn't
freed anywhere.

Signed-off-by: Christian Eggers <ceggers@arri.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

eeprom: at25: Allow page sizes greater than 16 bit ...

For FRAM devices the page size is usually equal to the device size. Some
available devices are greater than 64kB, so a 32 bit variable must be
used.

The same change has to be done for the Linux driver. I'll try to
upstream this as soon as possible.

Signed-off-by: Christian Eggers <ceggers@arri.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

watchdog: Fix error path ...

In case of an error (e.g. dev_add_param_tristate_ro() returns -ENOSYS),
device registration must be reverted.

Signed-off-by: Christian Eggers <ceggers@arri.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>

param: Make enum param_tristate always visible ...

When disabling CONFIG_SHELL_HUSH, also CONFIG_PARAMETER may be disabled.
In this case, the definition "enum param_tristate" is not reachable from
watchdog.h.

I also considered moving almost everything in watchdog.h inside the
CONFIG_WATCHDOG guard. But this doesn't work for me because
drivers/watchdog/imxwd.c can also be built without CONFIG_WATCHDOG.

Signed-off-by: Christian Eggers <ceggers@arri.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>