bootm: Drop #ifdef from do_bootm()

Drop the #ifdefs from this command by using a variable to hold the states
that should be executed.

Signed-off-by: Simon Glass <sjg@chromium.org>

bootm: Avoid returning error codes from command

Functions which implement commands must return a CMD_RET_... error code.
At present bootm can return a negative errno value in some cases, thus
causing strange behaviour such as trying to exit the shell and printing
usage information.

Fix this by returning the correct value.

Signed-off-by: Simon Glass <sjg@chromium.org>

bootm: Change incorrect 'unsupported' error

At present when bootm fails, it says:

    subcommand not supported

and then prints help for the bootm command. This is not very useful, since
generally the error is related to something else, such as fixups failing.
It is quite confusing to see this in a test run.

Change the error and show the error code.

We could update the OS functions to return -ENOSYS when they do not
support the bootm subcommand. But this involves some thought since this is
arch-specific code and proper errno error codes are not always returned.
Also, with the code as is, all required subcommands are of course
supported - a problem would only come if someone added a new one or
removed support for one from an existing OS. Therefore it seems better to
leave that sort of effort for when our bootm tests are improved.

Note: v1 of this patch generated a discussion[1] about printing error
strings automatically using printf(). That is outside the scope of this
patch but will be dealt with separately.

[1] https://patchwork.ozlabs.org/project/uboot/patch/20220909151801.336551-3-sjg@chromium.org/

Signed-off-by: Simon Glass <sjg@chromium.org>

dm: core: Switch uclass_*_device_err to use uclass_*_device_check

The _err variant iterators use the simple iterators without suffix as
basis.

However, there is no user that uclass_next_device_err for iteration,
many users of uclass_first_device_err use it to get the first and
(assumed) only device of an uclass, and a couple that use
uclass_next_device_err to get the device following a known device in the
uclass list.

While there are some truly singleton device classes in which more than
one device cannot exist these are quite rare, and most classes can have
multiple devices even if it is not the case on the SoC's EVB.

In a later patch the simple iterators will be updated to not stop on
error and return next device instead. With this in many cases the code
that expects the first device or an error if it fails to probe may get
the next device instead. Use the _check iterators as the basis of _err
iterators to preserve the old behavior.

Signed-off-by: Michal Suchanek <msuchanek@suse.de>
Reviewed-by: Simon Glass <sjg@chromium.org>

dm: core: Switch uclass_foreach_dev_probe to use simple iterator

The return value is not used for anythig, and in a later patch the
behavior of the _err iterator will change in an incompatible way.

Signed-off-by: Michal Suchanek <msuchanek@suse.de>
Reviewed-by: Simon Glass <sjg@chromium.org>
Update pvblock_probe() to avoid using internal var:
Signed-off-by: Simon Glass <sjg@chromium.org>

mpc83xx: gazerbeam: Update sysinfo_get error handling

In a later patch sysinfo_get will be changed to return the device in cae
of an error. Set sysinfo to NULL on error to preserve previous behavior.

Signed-off-by: Michal Suchanek <msuchanek@suse.de>
Reviewed-by: Simon Glass <sjg@chromium.org>

net: eth-uclass: Do not set device on error

eth_get_dev relies on the broken behavior that returns an error but not
the device on which the error happened which gives the caller no
reasonable way to report or handle the error.

In a later patch uclass_first_device_err will be changed to return the
device on error but eth_get_dev stores the returned device pointer
directly in a global state without checking the return value. Unset the
pointer again in the error case.

Signed-off-by: Michal Suchanek <msuchanek@suse.de>
Reviewed-by: Simon Glass <sjg@chromium.org>

dm: blk: Do not use uclass_next_device_err

blk_first_device_err/blk_next_device_err uses
uclass_first_device_err/uclass_next_device_err for device iteration.

Although the function names superficially match the return value from
uclass_first_device_err/uclass_next_device_err is never used
meaningfully, and uclass_first_device/uclass_next_device works equally
well for this purpose.

In the following patch the semantic of
uclass_first_device_err/uclass_next_device_err will be changed to be
based on uclass_first_device_check/uclass_next_device_check breaking
this sole user that uses uclass_next_device_err for iteration.

Signed-off-by: Michal Suchanek <msuchanek@suse.de>
Reviewed-by: Simon Glass <sjg@chromium.org>

dm: treewide: Use uclass_next_device_err when accessing second device

There are a couple users of uclass_next_device return value that get the
first device by other means and use uclass_next_device assuming the
following device in the uclass is related to the first one.

Use uclass_next_device_err because the return value from
uclass_next_device will be removed in a later patch.

Signed-off-by: Michal Suchanek <msuchanek@suse.de>
Reviewed-by: Simon Glass <sjg@chromium.org>

dm: treewide: Use uclass_first_device_err when accessing one device

There is a number of users that use uclass_first_device to access the
first and (assumed) only device in uclass.

Some check the return value of uclass_first_device and also that a
device was returned which is exactly what uclass_first_device_err does.

Some are not checking that a device was returned and can potentially
crash if no device exists in the uclass. Finally there is one that
returns NULL on error either way.

Convert all of these to use uclass_first_device_err instead, the return
value will be removed from uclass_first_device in a later patch.

Signed-off-by: Michal Suchanek <msuchanek@suse.de>
Reviewed-by: Simon Glass <sjg@chromium.org>

cmd: List all uclass devices regardless of probe error

There are a few commands that iterate uclass with
uclass_first_device/uclass_next_device or the _err variant.

Use the _check class iterator variant to get devices that fail to probe
as well, and print the status.

Signed-off-by: Michal Suchanek <msuchanek@suse.de>
Reviewed-by: Simon Glass <sjg@chromium.org>

w1: Fix bus counting in w1_get_bus

Use uclass_first_device_check/uclass_next_device_check to correctly
count buses that fail to probe.

Fixes: d3e19cf919 ("w1: Add 1-Wire uclass")
Signed-off-by: Michal Suchanek <msuchanek@suse.de>
Reviewed-by: Simon Glass <sjg@chromium.org>

video: ipuv3: Fix error handling when getting the display

The code checks that uclass_first_device returned a device but the
returned value that is assigned is never used. Use
uclass_first_device_err instead, and move the error return outside of
the if block.

Fixes: f4ec1ae08e ("mxc_ipuv3_fb.c: call display_enable")
Signed-off-by: Michal Suchanek <msuchanek@suse.de>
Reviewed-by: Simon Glass <sjg@chromium.org>

stdio: Fix class iteration in stdio_add_devices()

There is a complaint in the code that iterates keyboards that we don't
have the _check variant of class iterator but we in fact do, use it.

In the code that iterates video devices there is an attempt to print
errors but the simple iterator does not return a device when there is an
error. Use the _check variant of the iterator as well.

Also format error messages consistently.

Signed-off-by: Michal Suchanek <msuchanek@suse.de>
Reviewed-by: Simon Glass <sjg@chromium.org>

usb: ether: Fix error handling in usb_ether_init

The code checks the return value from uclass_first_device as well as
that the device exists but it passes on the return value which may be
zero if there are no gadget devices. Just check that a device was
returned and return -ENODEV otherwise.

Also remove the dev variable which is not really used for anything.

Signed-off-by: Michal Suchanek <msuchanek@suse.de>
Reviewed-by: Simon Glass <sjg@chromium.org>

bootstd: Fix listing boot devices

bootdev_list() uses uclass_*_device_err() to iterate devices.
However, the only value _err adds is returning an error when the device
pointer is null, and that's checked anyway.

Also there is some intent to report errors, and that's what
uclass_*_device_check() is for, use it.

Signed-off-by: Michal Suchanek <msuchanek@suse.de>
Reviewed-by: Simon Glass <sjg@chromium.org>

dm: pci: Fix device PCI iteration

When there is no PCI bus uclass_first_device will return no bus and no
error which will result in pci_find_first_device calling
skip_to_next_device with no bus, and the bus is only checked at the end
of the while cycle, not the beginning.

Fixes: 76c3fbcd3d ("dm: pci: Add a way to iterate through all PCI devices")
Signed-off-by: Michal Suchanek <msuchanek@suse.de>
Reviewed-by: Simon Glass <sjg@chromium.org>

dm: treewide: Do not opencode uclass_probe_all()

We already have a function for probing all devices of a specific class,
use it.

Signed-off-by: Michal Suchanek <msuchanek@suse.de>
Reviewed-by: Simon Glass <sjg@chromium.org>

dm: core: Fix uclass_probe_all to really probe all devices

uclass_probe_all uses uclass_first_device/uclass_next_device assigning
the return value.

The interface for getting meaningful error is
uclass_first_device_check/uclass_next_device_check, use it.

Also do not stop iteration when an error is encountered. Probing all
devices includes those that happen to be after a failing device in the
uclass order.

Fixes: a59153dfeb ("dm: core: add function uclass_probe_all() to probe all devices")
Signed-off-by: Michal Suchanek <msuchanek@suse.de>
Reviewed-by: Simon Glass <sjg@chromium.org>

dm: regmap: Disable range checks in SPL

A recent change to regmap breaks building of phycore-rk3288 for me. The
difference is only a few bytes. Somehow CI seems to pass, even though it
fails when I run docker locally. But it prevents me from sending any more
pull requests.

In any case this board is clearly near the limit. We could revert the
offending change, but it is needed for sandbox tests.

Instead, add a way to drop the range checks in SPL, since they end up
doing nothing if everything is working as expected.

This makes phycore-rk3288 build again for me and reduces the size of SPL
slightly for a number of boards.

Signed-off-by: Simon Glass <sjg@chromium.org>
Fixes: 947d4f132b4 ("regmap: fix range checks")

CI: Update to jammy-20221003-17Oct2022 tag

This includes python3-pyelftools so we can drop it from one of the tests
directly.

Signed-off-by: Tom Rini <trini@konsulko.com>

docker: Install pyelftools for builds

Binman needs this module to build sandbox_vpl and it is needed elsewhere
in CI.

Add it to the docker file.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Heinrich Schuchardt <xpyron.glpk@gmx.de>

Merge tag 'efi-2023-01-rc1-3' of https://source.denx.de/u-boot/custodians/u-boot-efi

Pull request for efi-2023-01-rc1-3

UEFI:
* replace EFI_CALL() by internal functions
* delete loadfile2 handle by uninstalling all protocols

Other:

* Provide spi_set_speed() needed for implementation of
  EFI SPI I/O protocol

efi_loader: remove efi_delete_handle on loadfile2

Loadfile2 code is installing two protocols on it's own handle
and uses efi_delete_handle() to clean it up on failure(s). However
commit 05c4c9e21ae6 ("efi_loader: define internal implementations of
install/uninstallmultiple") prepares the ground for us to clean up
efi_delete_handle() used in favor of Install/UninstallMultipleProtocol.

While at it clean up the non needed void casts to (void *) on the
protolcol installation.

Signed-off-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>
Reviewed-by: Heinrich Schuchardt <heinrich.schuchardt@canonical.com>

cmd: simplify efidebug

* don't use EFI_CALL() for variable services
* don't use runtime pointer to access exported function

Signed-off-by: Heinrich Schuchardt <heinrich.schuchardt@canonical.com>

efi_loader: avoid EFI_CALL() when draining console

Use internal function.

Signed-off-by: Heinrich Schuchardt <heinrich.schuchardt@canonical.com>

efi_loader: avoid EFI_CALL() for clearing screen

Carve out function efi_clear_screen.

Signed-off-by: Heinrich Schuchardt <heinrich.schuchardt@canonical.com>

spi: Implement spi_set_speed

This function is already defined in spi.h but no implementation of it
currently exists in the tree. The implementation is based on the static
function spi_set_speed_mode(). The function prototype is modified so
that an success or error condition can be returned to the caller.

Signed-off-by: Paul Barker <paul.barker@sancloud.com>
Reviewed-by: Heinrich Schuchardt <xypron.glpk@gmx.de>

Merge branch '2022-10-12-additional-fixes'

- Add "ubi list", Nokia RX51 fixes, other assorted fixes

Nokia RX-51: Fix double space key press

Space key is indicated by two different bits. Some HW models indicate press
of space key only by the first bit. Qemu indicates it by both bits at the
same time, which is currently interpreted by u-boot as double key press.

Fix this issue by setting first bit when only second is set (to support HW
models which indicate press only by second bit) and always clearing second
bit before processing to not report double space key press.

Signed-off-by: Pali Rohár <pali@kernel.org>

Nokia RX-51: Remove CONFIG_UBI_SIZE

CONFIG_UBI_SIZE option is not used, so remove it.

Signed-off-by: Pali Rohár <pali@kernel.org>

Fix typo in CONFIG_USBNET_DEVADDR

Fix typo that was caused by the same feature being split in to 2 different
configuration options. Replace CONFIG_USBNET_DEVADDR with
CONFIG_USBNET_DEV_ADDR

Signed-off-by: Ignacio Zamora <nachopitt@gmail.com>

tools: k3_gen_x509_cert: Do not print SWRV on build

This matches the others here.

Signed-off-by: Andrew Davis <afd@ti.com>

smem: msm: add missing <linux/sizes.h>

MSM SMEM driver is currently missing <linux/sizes.h> header and throws
the following compile error:

drivers/smem/msm_smem.c: In function ‘qcom_smem_get_ptable’:
drivers/smem/msm_smem.c:635:71: error: ‘SZ_4K’ undeclared (first use in this function)
  635 |         ptable = smem->regions[0].virt_base + smem->regions[0].size - SZ_4K;

Signed-off-by: Robert Marko <robert.marko@sartura.hr>
Cc: luka.perkov@sartura.hr

test: run setexpr test only on sandbox

The test relies on memory being available at 0x0. This in not valid for
many boards.

Signed-off-by: Heinrich Schuchardt <heinrich.schuchardt@canonical.com>
Reviewed-by: Simon Glass <sjg@chromium.org>

firmware: scmi: fix the multi channel support with CCF

When the CCF is activated, the dev->parent is not necessary
the reference to SCMI transport and the function devm_scmi_of_get_channel
failed for the registered SCMI clock, child for protocol@14,
the channel is null and the SCMI clock driver crash for any operations.

This patch changes the first parameter of the ops of_get_channel(),
aligned with other process_msg() to pass directly the good reference,
i.e. parent result of find_scmi_transport_device(dev)
which return the reference of the scmi transport device.

Fixes: 8e96801aa6a ("firmware: scmi: add multi-channel support")
Signed-off-by: Patrick Delaunay <patrick.delaunay@foss.st.com>

cmd: ubi: Add 'ubi list' command for printing list of all UBI volumes

To allow easily iterate over all UBI volumes, add a new command which
either print all user UBI volumes on output or set them into env variable.

As UBI volumes can have arbitrary name/label, in most cases it is useful to
iterate them by their numbers. This can be achieved by -numeric flag.

This functionality is similar to already existing 'part list' command which
prints partitions on formatted block device.

Signed-off-by: Pali Rohár <pali@kernel.org>

Merge branch '2022-10-11-assorted-fixes-and-updates'

- Assorted code cleanups and fixes, along with two new commands.

test: cmd: add test for temperature command

Add simple test for the temperature command.

Signed-off-by: Robert Marko <robert.marko@sartura.hr>
Reviewed-by: Simon Glass <sjg@chromium.org>

thermal: add sandbox driver

Provide a simple sandbox driver for the thermal uclass.
It simply registers and returns 100 degrees C if requested.

Signed-off-by: Robert Marko <robert.marko@sartura.hr>
Reviewed-by: Simon Glass <sjg@chromium.org>

doc: cmd: temperature: add documentation

Add documentation for the temperature command.

Signed-off-by: Robert Marko <robert.marko@sartura.hr>
Reviewed-by: Simon Glass <sjg@chromium.org>

cmd: add temperature command

Currently, there is no way for users to check the readings from thermal
sensors from U-boot console, only some boards print it during boot.

So, lets add a simple "temperature" command that allows listing thermal
uclass devices and getting their value.

Note that the thermal devices are intenionally probed if list is used as
almost always they will not get probed otherwise and there is no way for
users to manually call probe on a certain device from console.

Assumption is made that temperature is returned in degrees C and not
milidegrees like in Linux as this is what most drivers seem to return.

Signed-off-by: Robert Marko <robert.marko@sartura.hr>
Reviewed-by: Simon Glass <sjg@chromium.org>

cmd: xxd: add new command

Add xxd command to print file content as hexdump to standard out

Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Roger Knecht <rknecht@pm.me>

cmd: cat: add new command

Add cat command to print file content to standard out

Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Roger Knecht <rknecht@pm.me>

pinctrl: fix buffer size for pinctrl_generic_set_state_prefix()

This buffer has the concatenated prefix and name written into it, so it
must be large enough to cover both strings plus the terminating NUL.

Fixes: 92c4a95ec7 ("pinctrl: Add new function pinctrl_generic_set_state_prefix()")
Signed-off-by: John Keeping <john@metanate.com>
Reviewed-by: Pali Rohár <pali@kernel.org>
Reviewed-by: Simon Glass <sjg@chromium.org>

pinctrl: nuvoton: fix set persist error

CA9C is cortex A9 watchdog reset control bit.
if device set persist mode, it shouldn't set this bit.

Signed-off-by: Jim Liu <JJLIU0@nuvoton.com>

boot: image-pre-load: Check environment for location of signature info

Setting an alternative signature info node in "pre_load_sig_info_path"
allows verification of an image using the bootm pre-load mechanism with
a different key, e.g.: setenv pre_load_sig_info_path "/alt/sig" ; bootm
preload [addr]

Signed-off-by: Steven Lawrance <steven.lawrance@softathome.com>
Reviewed-by: Simon Glass <sjg@chromium.org>

image-pre-load: Move macros/definitions to image.h

Putting these definitions in a header will allow signatures to be
validated independently of bootm.

Signed-off-by: Steven Lawrance <steven.lawrance@softathome.com>
Reviewed-by: Simon Glass <sjg@chromium.org>

bootcount: pmic: Correct misleading comment

Fix a copy-paste error I did when inserting the comment.

Signed-off-by: Philip Oberfichtner <pro@denx.de>
Reviewed-by: Jaehoon Chung <jh80.chung@samsung.com>

lib: crypt: fix selecting a non-existent option

The option SHA256_ALGO does not exist. Remove selecting it.

Fixes: 26dd9936574 ("lib: add crypt subsystem")
Signed-off-by: Oleksandr Suvorov <oleksandr.suvorov@foundries.io>

cmd: pxe: add alias devicetree-overlay for fdtoverlays

This adds keyword devicetree-overlay as an alias for fdtoverlays in
extlinux (sysboot) and pxe to better follow the Boot Loader Specification
[1], improves documentation around them by adding an example for both
fdtoverlays and devicetree-overlay and the environment variable required
for this feature. The link for the spec is updated to the current one.

[1] https://systemd.io/BOOT_LOADER_SPECIFICATION/

Signed-off-by: Edoardo Tomelleri <e.tomell@gmail.com>
Reviewed-by: Neil Armstrong <neil.armstrong@linaro.org>
Reviewed-by: Ramon Fried <rfried.dev@gmail.com>

Merge tag 'xilinx-for-v2023.01-rc1-v3' of https://source.denx.de/u-boot/custodians/u-boot-microblaze

Xilinx changes for v2023.01-rc1 (round 3)

fpga:
- Create new uclass
- Get rid of FPGA_DEBUG and use logging infrastructure

zynq:
- Enable early EEPROM decoding
- Some DT updates

zynqmp:
- Use OCM_BANK_0 to check config loading permission
- Change config object loading in SPL
- Some DT updates

net:
- emaclite: Enable driver for RISC-V

xilinx:
- Fix static checker warnings
- Fix GCC12 warning

sdhci:
- Read PD id from DT

Merge branch '2022-10-10-LLVM-related-improvements'

- A few patches to make building with LLVM/LLD easier

spl: atf: Fix clang -Wasm-operand-widths warning

common/spl/spl_atf.c:187:51: warning: value size does not match register
  size specified by the constraint and modifier [-Wasm-operand-widths]
    __asm__ __volatile__("msr DAIF, %0\n\t" : : "r" (daif) : "memory");
                                                     ^
common/spl/spl_atf.c:187:34: note: use constraint modifier "w"
    __asm__ __volatile__("msr DAIF, %0\n\t" : : "r" (daif) : "memory");
                                    ^~
                                    %w0

Use %x0 to match what Linux does in <asm/sysreg.h> write_sysreg().

Signed-off-by: Alistair Delva <adelva@google.com>
Cc: Kever Yang <kever.yang@rock-chips.com>
Cc: Michael Walle <michael@walle.cc>
Cc: Simon Glass <sjg@chromium.org>
Cc: Tom Rini <trini@konsulko.com>
Cc: Nick Desaulniers <ndesaulniers@google.com>

Makefile: apply dynamic relocations for LLD

It seems that for aarch64, unless we apply dynamic relocations to the
location being relocated, we fail to boot.

As Fangrui notes:
  For dynamic relocations using the RELA format (readelf -Wr), GNU ld
  sets the initial content to r_addend; ld.lld doesn't do that by
  default (needs --apply-dynamic-relocs).

Otherwise .rodata appears to be full of NUL-bytes before relocation,
causing crashes when trying to invoke the function pointers in
init_sequence_f from initcall_run_list().

Link: https://reviews.llvm.org/D42797
Suggested-by: Fangrui Song <maskray@google.com>
Signed-off-by: Nick Desaulniers <ndesaulniers@google.com>
Signed-off-by: Alistair Delva <adelva@google.com>
Cc: Simon Glass <sjg@chromium.org>
Cc: Tom Rini <trini@konsulko.com>
Cc: Nick Desaulniers <ndesaulniers@google.com>

examples: standalone: Fix build with LLVM toolchain

When building the standalone example with llvm, the link step fails:

examples/standalone/libstubs.o: In function `dummy':
include/_exports.h:10: undefined reference to `jt'
include/_exports.h:11: undefined reference to `jt'
include/_exports.h:12: undefined reference to `jt'
include/_exports.h:13: undefined reference to `jt'
include/_exports.h:14: undefined reference to `jt'
examples/standalone/libstubs.o:include/_exports.h:15:
  more undefined references to `jt' follow

Indeed, the standalone libstubs.o does use the jt symbol, but it was
marked 'static' in stubs.c. It's strange how gcc builds are working.

Signed-off-by: Alistair Delva <adelva@google.com>
Cc: Rick Chen <rick@andestech.com>
Cc: Simon Glass <sjg@chromium.org>
Cc: Tom Rini <trini@konsulko.com>
Cc: Nick Desaulniers <ndesaulniers@google.com>

Merge https://source.denx.de/u-boot/custodians/u-boot-usb

Merge tag 'efi-2023-01-rc1-2' of https://source.denx.de/u-boot/custodians/u-boot-efi

Pull request for efi-2023-01-rc1-2

Documentation:

* man-page for cls command
* update build requirements

UEFI

* Fix bugs in the Stop() function of the EFi Driver Binding protocol
  for block devices
* Avoid EFI_CALL() when invoking CloseProtocol()

usb: gadget: fastboot: detach usb on reboot commands

When host issues "fastboot reboot fastboot", it's expected that the
board drops the USB connection before resetting.

On some boards, such as Khadas VIM3L and SEI610, this is not the case.

We observe the following error:
$ fastboot reboot fastboot
Rebooting into fastboot                            OKAY [  0.004s]
fastboot: error: Failed to boot into userspace fastboot; one or more components might be unbootable.

This does not happen when we use the RST button on the board.
It can be reproduced in linux with:
  # echo b > /proc/sysrq-trigger

In this case, we hit a undefined hardware behavior, where D+ and D-
are in an unknown state. Therefore the host can't detect usb
disconnection.

Make sure we always call usb_gadget_release() when a "fastboot reboot"
command is issued.

Note: usb_gadget_release() should be called before g_dnl_unregister()
because g_dnl_unregister() triggers a complete() call on each
endpoint (thus calling do_reset()).

Signed-off-by: Mattijs Korpershoek <mkorpershoek@baylibre.com>

usb: ehci: Remove unused ehci_{setup,shutdown}_phy() helpers

Remove unused ehci_{setup,shutdown}_phy() helpers now replaced by
generic_{setup,shutdown}_phy().

Signed-off-by: Patrice Chotard <patrice.chotard@foss.st.com>
Cc: Marek Vasut <marex@denx.de>
Cc: Simon Glass <sjg@chromium.org>

usb: ehci: Make usage of generic_{setup,shutdown}_phy() helpers

Replace ehci_setup_phy() and ehci_shutdown_phy () by respectively
generic_setup_phy() and generic_shutdown_phy().

Signed-off-by: Patrice Chotard <patrice.chotard@foss.st.com>
Cc: Marek Vasut <marex@denx.de>
Cc: Simon Glass <sjg@chromium.org>

usb: ohci: Make usage of generic_{setup,shutdown}_phy() helpers

Replace ohci_setup_phy() and ohci_shutdown_phy () by respectively
generic_setup_phy() and generic_shutdown_phy().

Signed-off-by: Patrice Chotard <patrice.chotard@foss.st.com>
Cc: Marek Vasut <marex@denx.de>
Cc: Simon Glass <sjg@chromium.org>

phy: Add generic_{setup,shutdown}_phy() helpers

In drivers usb/host/{ehci,ohci}-generic.c, {ehci,ohci}_setup_phy() and
{ehci,ohci}_shutdown_phy() shares 95% of common code.
Factorize this code in new generic_{setup,shudown}_phy() functions.

Signed-off-by: Patrice Chotard <patrice.chotard@foss.st.com>
Cc: Marek Vasut <marex@denx.de>
Cc: Simon Glass <sjg@chromium.org>

usb: gadget: Clean up Makefile ifdeffery

Take the USB_ETHER ifdef block apart and make use of obj-$(VAR) instead
to include the source files in build. The duplicate CI_UDC entry is now
removed, the USB_DEVICE ifdef is now reduced to core.o ep.o addition,
the ether.o can be conditionally compiled in using USB_ETHER.

No functional change.

Signed-off-by: Marek Vasut <marex@denx.de>
Reviewed-by: Patrice Chotard <patrice.chotard@foss.st.com>

usb: Add missing guard around env_get() in usb_hub

The env_get() might be undefined in case ENV_SUPPORT is disabled,
which may happen e.g. in SPL. Add missing ifdef guard around the
env_get() to prevent build failure.

Signed-off-by: Marek Vasut <marex@denx.de>
Tested-by: Fabio Estevam <festevam@denx.de>

efi_loader: reformat efi_disk_add_dev()

Make it clearer why InstallMultipleProtocolInterfaces is invoked with two
NULLs:

* rename guid to esp_guid
* put protocol GUIDs and the related interfaces on same lines
* add comment

Signed-off-by: Heinrich Schuchardt <heinrich.schuchardt@canonical.com>
Reviewed-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>

efi_loader: simplify efi_set_load_options()

* Replace the OpenProtocol() call by efi_search_protocol().
* Remove the CloseProtocol() call.

Signed-off-by: Heinrich Schuchardt <heinrich.schuchardt@canonical.com>
Reviewed-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>

efi_driver: use efi_close_protocol

Avoid EFI_CALL() by using efi_close_protocol().

Signed-off-by: Heinrich Schuchardt <heinrich.schuchardt@canonical.com>
Reviewed-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>

efi_loader: internal CloseProtocol

Allow avoiding using EFI_CALL() when closing a protocol by providing an
internal function.

Signed-off-by: Heinrich Schuchardt <heinrich.schuchardt@canonical.com>
Reviewed-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>

efi_loader: CloseProtocol in efi_fmp_find

The CloseProtocol() boot service requires a handle as first argument.
Passing the protocol interface is incorrect.

CloseProtocol() only has an effect if called with a non-zero value for
agent_handle. HandleProtocol() uses an opaque agent_handle when invoking
OpenProtocol() (currently NULL). Therefore HandleProtocol() should be
avoided.

* Replace the LocateHandle() call by efi_search_protocol().
* Remove the CloseProtocol() call.

Fixes: 8d99026f0697 ("efi_loader: capsule: support firmware update")
Signed-off-by: Heinrich Schuchardt <heinrich.schuchardt@canonical.com>

efi_loader: CloseProtocol in tcg2_measure_gpt_data

The CloseProtocol() boot service requires a handle as first argument.
Passing the protocol interface is incorrect.

CloseProtocol() only has an effect if called with a non-zero value for
agent_handle. HandleProtocol() uses an opaque agent_handle when invoking
OpenProtocol() (currently NULL). Therefore HandleProtocol() should be
avoided.

* Replace the LocateHandle() call by efi_search_protocol().
* Remove the CloseProtocol() call.
* Remove a superfluous goto.

Fixes: ce3dbc5d080d ("efi_loader: add UEFI GPT measurement")
Signed-off-by: Heinrich Schuchardt <heinrich.schuchardt@canonical.com>
Reviewed-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>

efi_driver: fix efi_uc_stop()

We must always call EFI_EXIT() when returning from an EFIAPI function.

Fixes: 05ef48a2484b ("efi_driver: EFI block driver")
Signed-off-by: Heinrich Schuchardt <heinrich.schuchardt@canonical.com>
Reviewed-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>

efi_loader: don't export efi_remove_all_protocols

This function is only used inside efi_boottime.c and is not safe to use
outside of this context.

Signed-off-by: Heinrich Schuchardt <heinrich.schuchardt@canonical.com>
Reviewed-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>

efi_loader: printf code in efi_disk_get_device_name()

part is unsigned. So it must be printed with %u.

Signed-off-by: Heinrich Schuchardt <heinrich.schuchardt@canonical.com>
Reviewed-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>

doc: describe usage of virt-make-fs for testing

We want tests to avoid the usage of sudo. Describe that virt-make-fs can
generate disk images without being root.

Signed-off-by: Heinrich Schuchardt <heinrich.schuchardt@canonical.com>

doc: add python3-filelock python3-pytest-xdist

Packages python3-filelock python3-pytest-xdist as required to run
'make tests'. Add them to the required packages list in the documentation.

Signed-off-by: Heinrich Schuchardt <heinrich.schuchardt@canonical.com>
Reviewed-by: Simon Glass <sjg@chromium.org>

doc: man-page for cls

Provide a man-page for the cls command.

Signed-off-by: Heinrich Schuchardt <heinrich.schuchardt@canonical.com>
Reviewed-by: Simon Glass <sjg@chromium.org>

cmd: CMD_CLS should not depend on video

The cls command works fine on the serial console. There is no reason to
let it depend on the availability of video.

Signed-off-by: Heinrich Schuchardt <heinrich.schuchardt@canonical.com>

Merge tag 'u-boot-nand-20221009' of https://source.denx.de/u-boot/custodians/u-boot-nand-flash

- mtd: Update the function name to 'rfree'
- Support NAND ONFI EDO mode for imx8mn architecture
- dm: clk: add missing stub when CONFIG_CLK is deactivated

fpga: virtex2: Use logging feature instead of FPGA_DEBUG

Instead of using DEBUG or LOG_DEBUG the driver still had its own
definition for debug output.

Signed-off-by: Alexander Dahl <ada@thorsis.com>
Signed-off-by: Michal Simek <michal.simek@amd.com>
Link: https://lore.kernel.org/r/20221007122003.11239-11-ada@thorsis.com

fpga: spartan3: Use logging feature instead of FPGA_DEBUG

Instead of using DEBUG or LOG_DEBUG the driver still had its own
definition for debug output.

Signed-off-by: Alexander Dahl <ada@thorsis.com>
Signed-off-by: Michal Simek <michal.simek@amd.com>
Link: https://lore.kernel.org/r/20221007122003.11239-10-ada@thorsis.com

fpga: spartan2: Use logging feature instead of FPGA_DEBUG

Instead of using DEBUG or LOG_DEBUG the driver still had its own
definition for debug output.

Signed-off-by: Alexander Dahl <ada@thorsis.com>
Signed-off-by: Michal Simek <michal.simek@amd.com>
Link: https://lore.kernel.org/r/20221007122003.11239-9-ada@thorsis.com

fpga: ACEX1K: Use logging feature instead of FPGA_DEBUG

Instead of using DEBUG or LOG_DEBUG the driver still had its own
definition for debug output.

Signed-off-by: Alexander Dahl <ada@thorsis.com>
Signed-off-by: Michal Simek <michal.simek@amd.com>
Link: https://lore.kernel.org/r/20221007122003.11239-8-ada@thorsis.com

fpga: cyclon2: Use logging feature instead of FPGA_DEBUG

Instead of using DEBUG or LOG_DEBUG the driver still had its own
definition for debug output.

Signed-off-by: Alexander Dahl <ada@thorsis.com>
Signed-off-by: Michal Simek <michal.simek@amd.com>
Link: https://lore.kernel.org/r/20221007122003.11239-7-ada@thorsis.com

fpga: altera: Use logging feature instead of FPGA_DEBUG

Instead of using DEBUG or LOG_DEBUG the driver still had its own
definition for debug output.

Signed-off-by: Alexander Dahl <ada@thorsis.com>
Signed-off-by: Michal Simek <michal.simek@amd.com>
Link: https://lore.kernel.org/r/20221007122003.11239-6-ada@thorsis.com

fpga: virtex2: Fix printf format string warnings

Warning appears if built with FPGA_DEBUG defined:

  CC      drivers/fpga/virtex2.o
/mnt/data/adahl/src/u-boot/drivers/fpga/virtex2.c: In function ‘virtex2_ssm_load’:
/mnt/data/adahl/src/u-boot/drivers/fpga/virtex2.c:333:11: warning: format ‘%d’ expects argument of type ‘int’, but argument 4 has type ‘size_t’ {aka ‘long unsigned int’} [-Wformat=]
    PRINTF("%s:%d:done went active early, bytecount = %d\n",
           ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
           __func__, __LINE__, bytecount);
                               ~~~~~~~~~
/mnt/data/adahl/src/u-boot/drivers/fpga/virtex2.c:25:37: note: in definition of macro ‘PRINTF’
 #define PRINTF(fmt, args...) printf(fmt, ##args)
                                     ^~~
/mnt/data/adahl/src/u-boot/drivers/fpga/virtex2.c: In function ‘virtex2_ss_load’:
/mnt/data/adahl/src/u-boot/drivers/fpga/virtex2.c:468:12: warning: format ‘%d’ expects argument of type ‘int’, but argument 4 has type ‘size_t’ {aka ‘long unsigned int’} [-Wformat=]
     PRINTF("%s:%d:done went active early, bytecount = %d\n",
            ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
            __func__, __LINE__, bytecount);
                                ~~~~~~~~~
/mnt/data/adahl/src/u-boot/drivers/fpga/virtex2.c:25:37: note: in definition of macro ‘PRINTF’
 #define PRINTF(fmt, args...) printf(fmt, ##args)
                                     ^~~

Signed-off-by: Alexander Dahl <ada@thorsis.com>
Signed-off-by: Michal Simek <michal.simek@amd.com>
Link: https://lore.kernel.org/r/20221007122003.11239-5-ada@thorsis.com

fpga: spartan3: Fix printf arguments warning

The additional comma messes up the arguments.
Warning appears if built with FPGA_DEBUG defined:

      CC      drivers/fpga/spartan3.o
    /mnt/data/adahl/src/u-boot/drivers/fpga/spartan3.c: In function ‘spartan3_sp_load’:
    /mnt/data/adahl/src/u-boot/drivers/fpga/spartan3.c:118:11: warning: too many arguments for format [-Wformat-extra-args]
       PRINTF ("%s: Function Table:\n"
               ^~~~~~~~~~~~~~~~~~~~~~~
    /mnt/data/adahl/src/u-boot/drivers/fpga/spartan3.c:18:37: note: in definition of macro ‘PRINTF’
     #define PRINTF(fmt,args...) printf (fmt ,##args)
                                         ^~~

Fixes: 875c78934ee2 ("Add Xilinx Spartan3 family FPGA support Patch by Kurt Stremerch, 14 February 2005")
Signed-off-by: Alexander Dahl <ada@thorsis.com>
Signed-off-by: Michal Simek <michal.simek@amd.com>
Link: https://lore.kernel.org/r/20221007122003.11239-4-ada@thorsis.com

fpga: spartan2: Fix printf arguments warning

That extra comma messes up format arguments.
Warning appears if built with FPGA_DEBUG defined:

      CC      drivers/fpga/spartan2.o
    /mnt/data/adahl/src/u-boot/drivers/fpga/spartan2.c: In function ‘spartan2_sp_load’:
    /mnt/data/adahl/src/u-boot/drivers/fpga/spartan2.c:112:11: warning: too many arguments for format [-Wformat-extra-args]
       PRINTF ("%s: Function Table:\n"
               ^~~~~~~~~~~~~~~~~~~~~~~
    /mnt/data/adahl/src/u-boot/drivers/fpga/spartan2.c:12:37: note: in definition of macro ‘PRINTF’
     #define PRINTF(fmt,args...) printf (fmt ,##args)
                                         ^~~
      CC      drivers/fpga/spartan3.o
    /mnt/data/adahl/src/u-boot/drivers/fpga/spartan3.c: In function ‘spartan3_sp_load’:
    /mnt/data/adahl/src/u-boot/drivers/fpga/spartan3.c:117:11: warning: too many arguments for format [-Wformat-extra-args]
       PRINTF ("%s: Function Table:\n"
               ^~~~~~~~~~~~~~~~~~~~~~~
    /mnt/data/adahl/src/u-boot/drivers/fpga/spartan3.c:17:37: note: in definition of macro ‘PRINTF’
     #define PRINTF(fmt,args...) printf (fmt ,##args)
                                         ^~~

Fixes: e221174377d7 ("Initial revision")
Signed-off-by: Alexander Dahl <ada@thorsis.com>
Signed-off-by: Michal Simek <michal.simek@amd.com>
Link: https://lore.kernel.org/r/20221007122003.11239-3-ada@thorsis.com

fpga: Add missing Kconfig symbols for old FPGA drivers

Those drivers could not be built anymore without those options present.

Signed-off-by: Alexander Dahl <ada@thorsis.com>
Signed-off-by: Michal Simek <michal.simek@amd.com>
Link: https://lore.kernel.org/r/20221007122003.11239-2-ada@thorsis.com

clk: versal: Mark versal_clock_setup() as static

Fix the following sparse and compile time warning
triggered with W=1:

drivers/clk/clk_versal.c:605:5:
warning: no previous prototype for 'versal_clock_setup'
[-Wmissing-prototypes]
605 | int versal_clock_setup(void)

Signed-off-by: Venkatesh Yadav Abbarapu <venkatesh.abbarapu@amd.com>
Link: https://lore.kernel.org/r/20221007105535.31902-1-venkatesh.abbarapu@amd.com
Signed-off-by: Michal Simek <michal.simek@amd.com>

xilinx: zynqmp: Load pmufw configuration before checking access

Before this patch you could see in the log:
U-Boot SPL 2022.10-rc5 (Sep 29 2022 - 15:29:27 +0200)
PMUFW: v1.1
Loading new PMUFW cfg obj (32 bytes)
PMUFW:  No permission to change config object
Loading new PMUFW cfg obj (2032 bytes)

where it is visible that permission is check before sending PMUFW
configuration (big size).

When this patch is applied it is visible that order is correct.
U-Boot SPL 2022.10-rc5 (Sep 29 2022 - 15:47:08 +0200)
Loading new PMUFW cfg obj (2032 bytes)
PMUFW: v1.1
Loading new PMUFW cfg obj (32 bytes)

Signed-off-by: Michal Simek <michal.simek@amd.com>
Link: https://lore.kernel.org/r/a0bf4f46d670023da4f848790eece6fff22090c2.1664962765.git.michal.simek@amd.com

xilinx: common: fix board_late_init_xilinx()

Compiling with GCC-12 leads to an error:

    +board/xilinx/common/board.c:479:37: error: the comparison will always
     evaluate as 'true' for the address of 'mac_addr' will never be NULL
     [-Werror=address]
    +  479 |                                 if (!desc->mac_addr[i])
    +      |                                     ^

Remove the redundant check.

Fixes: a03b594738f8 ("xilinx: board: Add support for additional card detection")
Signed-off-by: Heinrich Schuchardt <heinrich.schuchardt@canonical.com>
Link: https://lore.kernel.org/r/20221008091317.52838-1-heinrich.schuchardt@canonical.com
Signed-off-by: Michal Simek <michal.simek@amd.com>

xilinx: zynqmp: change the type of multiboot variable

In function ‘set_dfu_alt_info’ a comparison of a u8 value against
0 is done. Since it is always false, change the signature of this
function to use an `int` instead, which match the type used in caller:
`multi_boot()`.

Fix the following warning triggered with W=1:

board/xilinx/zynqmp/zynqmp.c:651:23:
warning: comparison is always false due to limited range of data type
[-Wtype-limits]
651 |         if (multiboot < 0)

Signed-off-by: Venkatesh Yadav Abbarapu <venkatesh.abbarapu@amd.com>
Link: https://lore.kernel.org/r/20221004055254.26246-1-venkatesh.abbarapu@amd.com
Signed-off-by: Michal Simek <michal.simek@amd.com>

soc: xilinx: zynqmp: Mark soc_xilinx_zynqmp_get_machine() as static

Fix the following sparse and compile time warning triggered with W=1:

drivers/soc/soc_xilinx_zynqmp.c:288:5:
warning: no previous prototype for 'soc_xilinx_zynqmp_get_machine'
[-Wmissing-prototypes]

Signed-off-by: Venkatesh Yadav Abbarapu <venkatesh.abbarapu@amd.com>
Link: https://lore.kernel.org/r/20221004055201.26146-1-venkatesh.abbarapu@amd.com
Signed-off-by: Michal Simek <michal.simek@amd.com>

xilinx: common: Add print_cpuinfo() declaration

cpu-info.c defines print_cpuinfo(), but neglected to
include its declaration, causing the following sparse and
compile time warnings:

board/xilinx/common/cpu-info.c:10:5:
warning: no previous prototype for 'print_cpuinfo'
[-Wmissing-prototypes]

Include init.h, which includes the missing declaration.

Signed-off-by: Venkatesh Yadav Abbarapu <venkatesh.abbarapu@amd.com>
Link: https://lore.kernel.org/r/20221004055053.26047-1-venkatesh.abbarapu@amd.com
Signed-off-by: Michal Simek <michal.simek@amd.com>

mtd: Update the function name to 'rfree'

Since commit 8d38a8459b0d ("mtd: Rename free() to rfree()")
the function has been renamed to rfree(), so update the description
inside the mtd_oob_region structure as well.

Fixes: 8d38a8459b0d ("mtd: Rename free() to rfree()")
Reported-by: Mikhail Kshevetskiy <mikhail.kshevetskiy@iopsys.eu>
Signed-off-by: Fabio Estevam <festevam@denx.de>
Acked-by: Michael Trimarchi <michael@amarulasolutions.com>
Reviewed-by: Dario Binacchi <dario.binacchi@amarulasolutions.com>
Signed-off-by: Dario Binacchi <dario.binacchi@amarulasolutions.com>

mtd: mxs_nand: Support EDO mode for imx8mn architecture

Add support for imx8mn architecture in order to run the NAND
in fast edo mode.

Signed-off-by: Michael Trimarchi <michael@amarulasolutions.com>
Reviewed-by: Dario Binacchi <dario.binacchi@amarulasolutions.com>
Signed-off-by: Dario Binacchi <dario.binacchi@amarulasolutions.com>

mtd: mxs_nand: get the clock with the right name

Rename the gpmi_apb_bch clock name to gpmi_bch_apb, as you can find in
the device tree.

Fixes: commit a59691280daca ("MXS_NAND: Add clock support for iMX8")
Signed-off-by: Dario Binacchi <dario.binacchi@amarulasolutions.com>
Reviewed-by: Michael Trimarchi <michael@amarulasolutions.com>

mtd: mxs_nand: don't get the gpmi_apbh_dma clock

This clock name is not present in any U-boot and Linux kernel device
tree.

Fixes: commit a59691280daca ("MXS_NAND: Add clock support for iMX8")
Signed-off-by: Dario Binacchi <dario.binacchi@amarulasolutions.com>
Reviewed-by: Michael Trimarchi <michael@amarulasolutions.com>

imx: gpmi: Add register needed to control nand bus timing

It is used as delay for gpmi write strobe.

Signed-off-by: Michael Trimarchi <michael@amarulasolutions.com>
Reviewed-by: Dario Binacchi <dario.binacchi@amarulasolutions.com>
Signed-off-by: Dario Binacchi <dario.binacchi@amarulasolutions.com>