oweals/u-boot.git
6 years agoefi_loader: helper function to add EFI object to list
Heinrich Schuchardt [Sun, 26 Nov 2017 13:05:23 +0000 (14:05 +0100)]
efi_loader: helper function to add EFI object to list

To avoid duplicate coding provide a helper function that
initializes an EFI object and adds it to the EFI object
list.

efi_exit() is the only place where we dereference a handle
to obtain a protocol interface. Add a comment to the function.

Suggested-by: Alexander Graf <agraf@suse.de>
Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_loader: pass handle of loaded image
Heinrich Schuchardt [Sun, 26 Nov 2017 13:05:22 +0000 (14:05 +0100)]
efi_loader: pass handle of loaded image

The handle of a loaded image is the value of the handle
member of the loaded image info object and not the
address of the loaded image info.

Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agotest/py: check return code of helloworld
Heinrich Schuchardt [Sun, 26 Nov 2017 13:05:21 +0000 (14:05 +0100)]
test/py: check return code of helloworld

Check that helloworld.efi returns EFI_SUCCESS.

Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_loader: output load options in helloworld
Heinrich Schuchardt [Sun, 26 Nov 2017 13:05:20 +0000 (14:05 +0100)]
efi_loader: output load options in helloworld

We need to test if we pass a valid image handle when loading
and EFI application. This cannot be done in efi_selftest as
it is not loaded as an image.

So let's enhance helloworld a bit.

Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_selftest: add missing line feed
Heinrich Schuchardt [Sun, 26 Nov 2017 13:05:19 +0000 (14:05 +0100)]
efi_selftest: add missing line feed

Add a missing line feed for an error message.

Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_selftest: compile without special compiler flags
Heinrich Schuchardt [Sun, 26 Nov 2017 13:05:18 +0000 (14:05 +0100)]
efi_selftest: compile without special compiler flags

As the selftest is not compiled as an EFI binary we do not
need special compiler flags.

This avoids the checkarmreloc error on vexpress_ca15_tc2.

Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_loader: manage protocols in a linked list
Heinrich Schuchardt [Sun, 26 Nov 2017 13:05:17 +0000 (14:05 +0100)]
efi_loader: manage protocols in a linked list

Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_loader: simplify find_obj
Heinrich Schuchardt [Sun, 26 Nov 2017 13:05:16 +0000 (14:05 +0100)]
efi_loader: simplify find_obj

Use function efi_search_protocol().

Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_loader: simplify efi_open_protocol
Heinrich Schuchardt [Sun, 26 Nov 2017 13:05:15 +0000 (14:05 +0100)]
efi_loader: simplify efi_open_protocol

Use function efi_search_protocol.

Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_loader: efi_gop: use efi_add_protocol
Heinrich Schuchardt [Sun, 26 Nov 2017 13:05:14 +0000 (14:05 +0100)]
efi_loader: efi_gop: use efi_add_protocol

Use efi_add_protocol to add protocol.

Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_loader: efi_net: use efi_add_protocol
Heinrich Schuchardt [Sun, 26 Nov 2017 13:05:13 +0000 (14:05 +0100)]
efi_loader: efi_net: use efi_add_protocol

Use efi_add_protocol to add protocols.

Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_loader: efi_disk: use efi_add_protocol
Heinrich Schuchardt [Sun, 26 Nov 2017 13:05:12 +0000 (14:05 +0100)]
efi_loader: efi_disk: use efi_add_protocol

Use efi_add_protocol to install protocols.

Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_selftest: test EFI_DEVICE_PATH_TO_TEXT_PROTOCOL
Heinrich Schuchardt [Sun, 26 Nov 2017 13:05:11 +0000 (14:05 +0100)]
efi_selftest: test EFI_DEVICE_PATH_TO_TEXT_PROTOCOL

Provide a test for the EFI_DEVICE_PATH_TO_TEXT_PROTOCOL protocol.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_loader: reimplement LocateDevicePath
Heinrich Schuchardt [Sun, 26 Nov 2017 13:05:10 +0000 (14:05 +0100)]
efi_loader: reimplement LocateDevicePath

The current implementation of efi_locate_device_path does not match
the UEFI specification. It completely ignores the protocol
parameters.

Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_loader: fix efi_convert_device_node_to_text
Heinrich Schuchardt [Sun, 26 Nov 2017 13:05:09 +0000 (14:05 +0100)]
efi_loader: fix efi_convert_device_node_to_text

We need to implement to different functions for the
EFI_DEVICE_PATH_TO_TEXT_PROTOCOL:
ConvertDeviceNodeToText
ConvertDevicePathToText

A recent patch screwed up efi_convert_device_node_to_text
to expect a device path and not a node.

The patch makes both service functions work again.

efi_convert_device_node_to_text is renamed to
efi_convert_single_device_node_to_text and
efi_convert_device_node_to_text_ext is renamed to
efi_convert_device_node_to_text to avoid future
confusion.

A test of ConvertDeviceNodeToText will be provided in
a follow-up patch.

Fixes: adae4313cdd efi_loader: flesh out device-path to text
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_loader: efi_dp_str should print path not node
Heinrich Schuchardt [Sun, 26 Nov 2017 13:05:08 +0000 (14:05 +0100)]
efi_loader: efi_dp_str should print path not node

efi_dp_str is meant to print a device path and not a device
node.

The old coding only worked because efi_convert_device_node_to_text
was screwed up to expect paths instead of nodes.

Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_loader: size of media device path node represenation
Heinrich Schuchardt [Sun, 26 Nov 2017 13:05:07 +0000 (14:05 +0100)]
efi_loader: size of media device path node represenation

In the format specifier we want to specify the maximum width
in case an ending \0 is missing.

So slen must be used as precision and not as field width.

Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_loader: Exit in efi_set_bootdev() upon invalid "desc"
Stefan Roese [Fri, 17 Nov 2017 07:47:09 +0000 (08:47 +0100)]
efi_loader: Exit in efi_set_bootdev() upon invalid "desc"

When trying to load an image from a non-existent USB key, U-Boot v2017.11
crashes on my x86 platform:

=> load usb 0:1 03000000 abc
General Protection
EIP: 0010:[<7b59030d>] EFLAGS: 00010286
Original EIP :[<fff4330d>]
...

This used to work in v2017.09. Testing has shown, that this bug was
introduced with patch 95c5553e [efi_loader: refactor boot device and
loaded_image handling].

This patch now checks if a valid "desc" is returned from blk_get_dev()
and only continues when "desc" is available. Resulting in this cmd
output (again):

=> load usb 0:1 03000000 abc
** Bad device usb 0 **

Signed-off-by: Stefan Roese <sr@denx.de>
Reviewed-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_loader: helloworld.c: remove superfluous include
Heinrich Schuchardt [Sun, 26 Nov 2017 13:05:06 +0000 (14:05 +0100)]
efi_loader: helloworld.c: remove superfluous include

Remove a superfluous include from helloworld.c

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_loader: initialise partition_signature memory
Jonathan Gray [Wed, 22 Nov 2017 03:18:59 +0000 (14:18 +1100)]
efi_loader: initialise partition_signature memory

Zero partition_signature in the efi_device_path_hard_drive_path
structure when signature_type is 0 (no signature) as required by the
UEFI specification.

This is required so that efi_dp_match() will work as expected
when doing memcmp() comparisons.  Previously uninitialised memory
would cause it not match nodes when it should have when the signature
type was not GUID.

Corrects a problem where the loaded image protocol would not return a
device path with MEDIA_DEVICE causing the OpenBSD bootloader to fail
on rpi_3 and other targets.

v2: Also handle signature_type 1 (MBR) as described in the specification

Signed-off-by: Jonathan Gray <jsg@jsg.id.au>
Tested-by: Artturi Alm <artturi.alm@gmail.com>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_loader: efi_console: use helper functions
Heinrich Schuchardt [Thu, 26 Oct 2017 17:25:59 +0000 (19:25 +0200)]
efi_loader: efi_console: use helper functions

Use helper functions efi_created_handle and efi_add_protocol
for creating the console handles and instaling the respective
protocols.

This change is needed if we want to move from an array of
protocols to a linked list of protocols.

Eliminate EFI_PROTOCOL_OBJECT which is not used anymore.

Currently we have not defined protocol interfaces to be const.
So efi_con_out and efi_console_control cannot be defined as const.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_loader: refactor efi_setup_loaded_image
Heinrich Schuchardt [Thu, 26 Oct 2017 17:25:58 +0000 (19:25 +0200)]
efi_loader: refactor efi_setup_loaded_image

Use helper functions to add protocols.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_loader: simplify efi_locate_protocol
Heinrich Schuchardt [Thu, 26 Oct 2017 17:25:57 +0000 (19:25 +0200)]
efi_loader: simplify efi_locate_protocol

Use helper function efi_search_protocol.

Do not print protocol guid twice in debug mode.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_loader: simplify efi_uninstall_protocol_interface
Heinrich Schuchardt [Thu, 26 Oct 2017 17:25:56 +0000 (19:25 +0200)]
efi_loader: simplify efi_uninstall_protocol_interface

Use function efi_search_obj, efi_search_protocol and
efi_remove_protocol to simplify the coding.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_loader: simplify efi_search
Heinrich Schuchardt [Thu, 26 Oct 2017 17:25:55 +0000 (19:25 +0200)]
efi_loader: simplify efi_search

Use helper function efi_search_protocol in efi_search.
Add missing comments.
Put default handling into default branch of switch statement.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_loader: simplify efi_install_protocol_interface
Heinrich Schuchardt [Thu, 26 Oct 2017 17:25:54 +0000 (19:25 +0200)]
efi_loader: simplify efi_install_protocol_interface

Use helper functio efi_add_protocol.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_loader: helper functions for protocol management
Heinrich Schuchardt [Thu, 26 Oct 2017 17:25:53 +0000 (19:25 +0200)]
efi_loader: helper functions for protocol management

This patch provides helper functions to manage protocols.
efi_search_protocol - find a protocol on a handle
efi_add_protocol - install a protocol on a handle
efi_remove_protocol - remove a protocol from a handle
efi_remove_all_protocols - remove all protocols from a handle

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_selftest: test for graphics output protocol
Heinrich Schuchardt [Thu, 26 Oct 2017 17:25:52 +0000 (19:25 +0200)]
efi_selftest: test for graphics output protocol

Supply a test for the graphics output protocol.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_loader: efi_gop: use correct types for parameters
Heinrich Schuchardt [Thu, 26 Oct 2017 17:25:51 +0000 (19:25 +0200)]
efi_loader: efi_gop: use correct types for parameters

Use efi_uintn_t instead of unsigned long.

EFI_GRAPHICS_OUTPUT_BLT_OPERATION is an enum. If we don't
define an enum we have to pass it as u32.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_loader: argument of efi_search_obj should be const
Heinrich Schuchardt [Thu, 26 Oct 2017 17:25:50 +0000 (19:25 +0200)]
efi_loader: argument of efi_search_obj should be const

The argument of efi_search_obj is not changed so it should
be marked as const.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_loader: make efi_create_handle non-static
Heinrich Schuchardt [Thu, 26 Oct 2017 17:25:49 +0000 (19:25 +0200)]
efi_loader: make efi_create_handle non-static

Export function efi_create_handle.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_loader: efi_dp_match should have const arguments
Heinrich Schuchardt [Thu, 26 Oct 2017 17:25:48 +0000 (19:25 +0200)]
efi_loader: efi_dp_match should have const arguments

efi_dp_match does not change its arguments.
So they should be marked as const.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_loader: efi_net: check return value of calloc
Heinrich Schuchardt [Thu, 26 Oct 2017 17:25:47 +0000 (19:25 +0200)]
efi_loader: efi_net: check return value of calloc

Calloc may return NULL. So we must check the return value.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_loader: efi_disk: check return value of calloc
Heinrich Schuchardt [Thu, 26 Oct 2017 17:25:46 +0000 (19:25 +0200)]
efi_loader: efi_disk: check return value of calloc

Calloc may return NULL. We should check the return value.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_loader: efi_gop: check calloc return value
Heinrich Schuchardt [Thu, 26 Oct 2017 17:25:45 +0000 (19:25 +0200)]
efi_loader: efi_gop: check calloc return value

Calloc may return NULL. We have to check the return value.

Fixes: be8d324191f efi_loader: Add GOP support
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_loader implement UninstallMultipleProtocolInterfaces
Heinrich Schuchardt [Thu, 26 Oct 2017 17:25:44 +0000 (19:25 +0200)]
efi_loader implement UninstallMultipleProtocolInterfaces

Implement UninstallMultipleProtocolInterfaces.
The efi_uninstall_multipled_protocol_interfaces tries to
uninstall protocols one by one. If an error occurs all
uninstalled protocols are reinstalled.

As the implementation efi_uninstall_protocol_interface is
still incomplete the function will fail.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_loader: debug output efi_install_protocol_interface
Heinrich Schuchardt [Thu, 26 Oct 2017 17:25:43 +0000 (19:25 +0200)]
efi_loader: debug output efi_install_protocol_interface

efi_install_protocol_interface should provide the created or
provided handle in the debug output.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_loader: fix typo efi_install_multiple_protocol_interfaces
Heinrich Schuchardt [Thu, 26 Oct 2017 17:25:42 +0000 (19:25 +0200)]
efi_loader: fix typo efi_install_multiple_protocol_interfaces

%s/occured/occurred/g

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_loader: rework efi_search_obj
Heinrich Schuchardt [Mon, 6 Nov 2017 20:17:50 +0000 (21:17 +0100)]
efi_loader: rework efi_search_obj

EFI_HANDLEs are used both in boottime and in runtime services.
efi_search_obj is a function that can be used to validate
handles. So let's make it accessible via efi_loader.h.

We can simplify the coding using list_for_each_entry.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_loader: rework efi_locate_handle
Heinrich Schuchardt [Mon, 6 Nov 2017 20:17:49 +0000 (21:17 +0100)]
efi_loader: rework efi_locate_handle

Check the parameters in efi_locate_handle.

Use list_for_each_entry instead of list_for_each.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_loader: consistently use efi_uintn_t in boot services
Heinrich Schuchardt [Mon, 6 Nov 2017 20:17:48 +0000 (21:17 +0100)]
efi_loader: consistently use efi_uintn_t in boot services

Consistenly use efi_uintn_t wherever the UEFI spec uses
UINTN in boot services interfaces.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_loader: replace UINTN by efi_uintn_t
Heinrich Schuchardt [Mon, 6 Nov 2017 20:17:47 +0000 (21:17 +0100)]
efi_loader: replace UINTN by efi_uintn_t

UINTN is used in the UEFI specification for unsigned integers
matching the bitness of the CPU.

Types in U-Boot should be lower case. The patch replaces it
by efi_uintn_t.

Suggested-by: Simon Glass <sjg@chromium.org>
Suggested-by: Rob Clark <robdclark@gmail.com>
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_loader: remove unused typedef for INTN
Heinrich Schuchardt [Mon, 6 Nov 2017 20:17:46 +0000 (21:17 +0100)]
efi_loader: remove unused typedef for INTN

INTN is not used in the coding.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_loader: eliminate efi_uninstall_protocol_interface_ext
Heinrich Schuchardt [Mon, 6 Nov 2017 20:17:45 +0000 (21:17 +0100)]
efi_loader: eliminate efi_uninstall_protocol_interface_ext

As we now have EFI_CALL there is no need for separate
functions efi_uninstall_protocol_interface_ext and
efi_uninstall_protocol_interface.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_loader: eliminate efi_install_protocol_interface_ext
Heinrich Schuchardt [Mon, 6 Nov 2017 20:17:44 +0000 (21:17 +0100)]
efi_loader: eliminate efi_install_protocol_interface_ext

As we now have EFI_CALL there is no need for separate
functions efi_install_protocol_interface_ext and
efi_install_protocol_interface.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_selftest: test protocol management
Heinrich Schuchardt [Mon, 6 Nov 2017 20:17:43 +0000 (21:17 +0100)]
efi_selftest: test protocol management

This unit test checks the following protocol services:
InstallProtocolInterface, UninstallProtocolInterface,
InstallMultipleProtocolsInterfaces,
UninstallMultipleProtocolsInterfaces,
HandleProtocol, ProtocolsPerHandle,
LocateHandle, LocateHandleBuffer.

As UninstallProtocolInterface and UninstallMultipleProtocolsInterfaces
are not completely implemented a TODO message will shown for
their failure.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_loader: capitalize EFI_LOCATE_SEARCH_TYPE values
Heinrich Schuchardt [Mon, 6 Nov 2017 20:17:42 +0000 (21:17 +0100)]
efi_loader: capitalize EFI_LOCATE_SEARCH_TYPE values

Constants should be capitalized.
So rename the values of enum efi_locate_search_type.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_loader: set parent handle in efi_load_image
Heinrich Schuchardt [Wed, 18 Oct 2017 16:13:20 +0000 (18:13 +0200)]
efi_loader: set parent handle in efi_load_image

The parent_handle of the loaded image must be set.
Set the system table.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_selftest: provide test for EFI_SIMPLE_TEXT_OUTPUT_PROTOCOL
Heinrich Schuchardt [Wed, 18 Oct 2017 16:13:19 +0000 (18:13 +0200)]
efi_selftest: provide test for EFI_SIMPLE_TEXT_OUTPUT_PROTOCOL

The following services are tested:
OutputString, TestString, SetAttribute.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agotest/py: fix typo in test_efi_loader.py
Heinrich Schuchardt [Wed, 18 Oct 2017 16:13:18 +0000 (18:13 +0200)]
test/py: fix typo in test_efi_loader.py

Make a comment line easier to read.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agotest/py: test reboot by EFI watchdog
Heinrich Schuchardt [Wed, 18 Oct 2017 16:13:17 +0000 (18:13 +0200)]
test/py: test reboot by EFI watchdog

Clear environment variable efi_selftest before executing the
default tests.

Provide a test verifying that the EFI watchdog
reboots the system upon timeout.

The test depends on CONFIG_CMD_EFI_SELFTEST=y.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_selftest: test reboot by watchdog
Heinrich Schuchardt [Wed, 18 Oct 2017 16:13:16 +0000 (18:13 +0200)]
efi_selftest: test reboot by watchdog

A test is added that verifies that the watchdog timer actually
causes a reboot upon timeout. The test is only executed on
request using

    setenv efi_selftest watchdog reboot
    bootefi selftest

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_loader: use bootargs as load options
Heinrich Schuchardt [Wed, 18 Oct 2017 16:13:15 +0000 (18:13 +0200)]
efi_loader: use bootargs as load options

Use environment variable bootargs used as load options
for bootefi payloads.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_selftest: correctly cleanup after selftest
Heinrich Schuchardt [Wed, 18 Oct 2017 16:13:14 +0000 (18:13 +0200)]
efi_selftest: correctly cleanup after selftest

After executing bootefi selftest
* restore GD
* unlink the load image handle
* return 0 or 1 and not a truncated efi_status_t.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_selftest: allow to select a single test for execution
Heinrich Schuchardt [Wed, 18 Oct 2017 16:13:13 +0000 (18:13 +0200)]
efi_selftest: allow to select a single test for execution

Environment variable efi_selftest is passed as load options
to the selftest application. It is used to select a single
test to be executed.

The load options are an UTF8 string. Yet I decided to keep
the name propertiy of the tests as char[] to reduce code
size.

Special value 'list' displays a list of all available tests.

Tests get an on_request property. If this property is set
the tests are only executed if explicitly requested.

The invocation of efi_selftest is changed to reflect that
bootefi selftest with efi_selftest = 'list' will call the
Exit bootservice.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_selftest: deduplicate code
Heinrich Schuchardt [Wed, 18 Oct 2017 16:13:12 +0000 (18:13 +0200)]
efi_selftest: deduplicate code

Move duplicate code to the new function efi_st_do_tests.

Suggested-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_selftest: efi_st_memcmp should return 0
Heinrich Schuchardt [Wed, 18 Oct 2017 16:13:11 +0000 (18:13 +0200)]
efi_selftest: efi_st_memcmp should return 0

If the compared memory areas match the return value should be 0.
We should not use the unrelated constant EFI_ST_SUCCESS.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_selftest: reformat code
Heinrich Schuchardt [Wed, 18 Oct 2017 16:13:10 +0000 (18:13 +0200)]
efi_selftest: reformat code

Remove superfluous spaces.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_selftest: provide a dummy device path
Heinrich Schuchardt [Wed, 18 Oct 2017 16:13:09 +0000 (18:13 +0200)]
efi_selftest: provide a dummy device path

Currently we pass bootefi_device_path and bootefi_image_path as
device and image path without initializing them. They may carry
values from previous calls to bootefi.

With the patch the variables are initialized valid dummy values.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_loader: consistently use efi_status_t in bootefi
Heinrich Schuchardt [Wed, 18 Oct 2017 16:13:08 +0000 (18:13 +0200)]
efi_loader: consistently use efi_status_t in bootefi

Where ulong or unsigned long are used to hold an EFI status
code we should consistenly use efi_status_t.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_loader: guard against double inclusion of efi_loader.h
Heinrich Schuchardt [Wed, 18 Oct 2017 16:13:07 +0000 (18:13 +0200)]
efi_loader: guard against double inclusion of efi_loader.h

Use a define to detect double inclusion of efi_loader.h.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_loader: new function utf8_to_utf16
Heinrich Schuchardt [Wed, 18 Oct 2017 16:13:06 +0000 (18:13 +0200)]
efi_loader: new function utf8_to_utf16

Provide a conversion function from utf8 to utf16.

Add missing #include <linux/types.h> in include/charset.h.
Remove superfluous #include <common.h> in lib/charset.c.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_selftest: provide test for watchdog timer
Heinrich Schuchardt [Wed, 18 Oct 2017 16:13:05 +0000 (18:13 +0200)]
efi_selftest: provide test for watchdog timer

The test verifies that resetting the watchdog timer ensures
that it is not called during the timeout period.

Testing that the watchdog timer actually executes a reset
would require a test outside the efi_selftest framework.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_loader: implement SetWatchdogTimer
Heinrich Schuchardt [Wed, 18 Oct 2017 16:13:04 +0000 (18:13 +0200)]
efi_loader: implement SetWatchdogTimer

The watchdog is initialized with a 5 minute timeout period.
It can be reset by SetWatchdogTimer.
It is stopped by ExitBoottimeServices.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoefi_loader: move efi_search_obj up in code
Heinrich Schuchardt [Wed, 18 Oct 2017 16:13:03 +0000 (18:13 +0200)]
efi_loader: move efi_search_obj up in code

To avoid a forward declaration move efi_search_obj before
all protocol services functions.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
6 years agoMerge branch 'rmobile-mx' of git://git.denx.de/u-boot-sh
Tom Rini [Thu, 30 Nov 2017 15:39:04 +0000 (10:39 -0500)]
Merge branch 'rmobile-mx' of git://git.denx.de/u-boot-sh

6 years agoMerge git://git.denx.de/u-boot-marvell
Tom Rini [Thu, 30 Nov 2017 15:38:40 +0000 (10:38 -0500)]
Merge git://git.denx.de/u-boot-marvell

6 years agoMerge git://git.denx.de/u-boot-x86
Tom Rini [Thu, 30 Nov 2017 15:37:43 +0000 (10:37 -0500)]
Merge git://git.denx.de/u-boot-x86

6 years agoARM: rmobile: Rework the ULCB CPLD driver
Marek Vasut [Sun, 26 Nov 2017 19:32:44 +0000 (20:32 +0100)]
ARM: rmobile: Rework the ULCB CPLD driver

Rework the ULCB CPLD driver and make it into a sysreset driver,
since that is what the ULCB CPLD driver is mostly for.

Signed-off-by: Marek Vasut <marek.vasut+renesas@gmail.com>
Cc: Nobuhiro Iwamatsu <iwamatsu@nigauri.org>
6 years agoarm: mvebu: correct comments around cas_wl/cas_l
Chris Packham [Tue, 28 Nov 2017 21:38:34 +0000 (10:38 +1300)]
arm: mvebu: correct comments around cas_wl/cas_l

The order of members in struct hws_topology_map is cas_wl, cas_l. The
comments in the original db-88f6820-gp.c had this wrong and have been
copied to other Armada-385 based boards. Practically this hasn't made a
difference since all these boards set both cas_wl and cas_l to 0
(autodetect) but if there were ever a board that did need to set these
explicitly they would run into unexpected issued.

Update the comments to reflect the correct order of structure members.

Reported-by: Tobi Wulff <tobi.wulff@alliedtelesis.co.nz>
Signed-off-by: Chris Packham <judge.packham@gmail.com>
Reviewed-by: Stefan Roese <sr@denx.de>
Signed-off-by: Stefan Roese <sr@denx.de>
6 years agoarm64: mvebu: armada-7k/8k: drop useless #ifdef
Baruch Siach [Sun, 26 Nov 2017 07:21:23 +0000 (09:21 +0200)]
arm64: mvebu: armada-7k/8k: drop useless #ifdef

CONFIG_ENV_IS_IN_NAND has been removed in commit 2be296538e2e (Convert
CONFIG_ENV_IS_IN_MMC/NAND/UBI and NOWHERE to Kconfig).
CONFIG_ENV_IS_IN_SPI_FLASH has been removed in commit 91c868fe7cd
(Convert CONFIG_ENV_IS_IN_SPI_FLASH to Kconfig). The environment #ifdef
is now empty. Remove it.

Signed-off-by: Baruch Siach <baruch@tkos.co.il>
Signed-off-by: Stefan Roese <sr@denx.de>
6 years agoarm: mvebu: enable boot from NAND
Sean Nyekjaer [Fri, 24 Nov 2017 13:01:47 +0000 (14:01 +0100)]
arm: mvebu: enable boot from NAND

Check if we are booting from NAND and let the bootrom
continue to load the rest of the bootloader

Signed-off-by: Sean Nyekjaer <sean.nyekjaer@prevas.dk>
Signed-off-by: Stefan Roese <sr@denx.de>
6 years agoarm: mvebu: fix boot from UART when in fallback mode
Sean Nyekjaer [Fri, 24 Nov 2017 13:01:28 +0000 (14:01 +0100)]
arm: mvebu: fix boot from UART when in fallback mode

It's the first 8 bits of the bootrom error register that
contain the boot error/fallback error code. Let's check that
and continue to boot from UART.

Signed-off-by: Sean Nyekjaer <sean.nyekjaer@prevas.dk>
Signed-off-by: Stefan Roese <sr@denx.de>
6 years agoarm: mvebu: add nand pins
Sean Nyekjaer [Fri, 24 Nov 2017 13:00:48 +0000 (14:00 +0100)]
arm: mvebu: add nand pins

Signed-off-by: Sean Nyekjaer <sean.nyekjaer@prevas.dk>
Signed-off-by: Stefan Roese <sr@denx.de>
6 years agofix: nand: pxa3xx: fix defined but not used warnings
Sean Nyekjaer [Wed, 22 Nov 2017 12:39:08 +0000 (13:39 +0100)]
fix: nand: pxa3xx: fix defined but not used warnings

bbt_mirror_descr and bbt_main_descr is defined but not used
when compiling without CONFIG_SYS_NAND_USE_FLASH_BBT set.

Signed-off-by: Sean Nyekjaer <sean.nyekjaer@prevas.dk>
Signed-off-by: Stefan Roese <sr@denx.de>
6 years agox86: lib: Implement standalone __udivdi3 etc instead of libgcc ones
Stefan Roese [Wed, 29 Nov 2017 15:23:31 +0000 (16:23 +0100)]
x86: lib: Implement standalone __udivdi3 etc instead of libgcc ones

This patch removes the inclusion of the libgcc math functions and
replaces them by functions coded in C, taken from the coreboot
project. This makes U-Boot building more independent from the toolchain
installed / available on the build system.

The code taken from coreboot is authored from Vadim Bendebury
<vbendeb@chromium.org> on 2014-11-28 and committed with commit
ID e63990ef [libpayload: provide basic 64bit division implementation]
(coreboot git repository located here [1]).

I modified the code so that its checkpatch clean without any
functional changes.

[1] git://github.com/coreboot/coreboot.git

Signed-off-by: Stefan Roese <sr@denx.de>
Cc: Simon Glass <sjg@chromium.org>
Cc: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Tested-by: Bin Meng <bmeng.cn@gmail.com>
6 years agodistro bootcmd: define bootloader name for x86
Heinrich Schuchardt [Fri, 24 Nov 2017 21:32:35 +0000 (22:32 +0100)]
distro bootcmd: define bootloader name for x86

Currently X86 does not properly support distro defaults.
This patch is only a partial fix.

It provides the name of the bootloader EFI application
for the X86 architecture.

The architecture dependent file names are defined in the UEFI
specification.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Reviewed-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
6 years agoRevert "x86: bootm: Fix FIT image booting on x86"
Anatolij Gustschin [Thu, 23 Nov 2017 17:59:45 +0000 (18:59 +0100)]
Revert "x86: bootm: Fix FIT image booting on x86"

This reverts commit 13c531e52a09b4e6ffa8b5a1457199b0a574cb27.

The error message with FIT style image mentioned in the above commit
only happens when booting using FIT image containing bzImage kernel
and without setup node (setup.bin). The current documentation for
x86 FIT support in doc/uImage.FIT/x86-fit-boot.txt mentions that
kernel's setup.bin file is required for building x86 FIT images.
The above commit breaks FIT images generated as described in the
documentation. Revert it to allow booting with images built in the
documented way.

Signed-off-by: Anatolij Gustschin <agust@denx.de>
Reviewed-by: Stefan Roese <sr@denx.de>
Acked-by: Bin Meng <bmeng.cn@gmail.com>
6 years agox86: don't compare pointers to 0
Heinrich Schuchardt [Mon, 20 Nov 2017 18:45:56 +0000 (19:45 +0100)]
x86: don't compare pointers to 0

x86_vendor_name is defined as

static const char *const x86_vendor_name[]

So its elements should not be compared to 0.

Remove superfluous paranthesis.

Problem identified with Coccinelle.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
6 years agox86: conga-qeval20-qa3-e3845: Adjust VGA rom address
Anatolij Gustschin [Fri, 17 Nov 2017 01:29:59 +0000 (02:29 +0100)]
x86: conga-qeval20-qa3-e3845: Adjust VGA rom address

Adjust VGA rom address to 0xfffb0000 so that u-boot.rom image
can be built again.

Signed-off-by: Anatolij Gustschin <agust@denx.de>
Reviewed-by: Stefan Roese <sr@denx.de>
Acked-by: Bin Meng <bmeng.cn@gmail.com>
6 years agox86: qemu: Move Cache-As-RAM memory from area mapped to ROM
Anton Gerasimov [Tue, 14 Nov 2017 17:47:16 +0000 (18:47 +0100)]
x86: qemu: Move Cache-As-RAM memory from area mapped to ROM

ROM has been made read-only in qemu recently (namely commit 208fa0e4:
"pc: make 'pc.rom' readonly when machine has PCI enabled"). So this
patch restores compatibility between U-Boot and qemu.

Signed-off-by: Anton Gerasimov <anton@advancedtelematic.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Tested-by: Bin Meng <bmeng.cn@gmail.com>
[bmeng: mention qemu commit title in the commit message]
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
6 years agofat: Use cache aligned buffers for fat_opendir
Neil Armstrong [Fri, 24 Nov 2017 08:54:41 +0000 (09:54 +0100)]
fat: Use cache aligned buffers for fat_opendir

Before this patch one could receive following errors when executing "fatls"
command on machine with cache enabled (ex i.MX6Q) :

=> fatls mmc 0:1
CACHE: Misaligned operation at range [4f59dfc84f59e7c8]
CACHE: Misaligned operation at range [4f59dfc84f59e7c8]
ERROR: v7_outer_cache_inval_range - start address is not aligned - 0x4f59dfc8
ERROR: v7_outer_cache_inval_range - stop address is not aligned - 0x4f59e7c8
CACHE: Misaligned operation at range [4f59dfc84f59e7c8]
CACHE: Misaligned operation at range [4f59dfc84f59e7c8]
ERROR: v7_outer_cache_inval_range - start address is not aligned - 0x4f59dfc8
ERROR: v7_outer_cache_inval_range - stop address is not aligned - 0x4f59e7c8

To alleviate this problem - the calloc()s have been replaced with
malloc_cache_aligned() and memset().

After those changes the buffers are properly aligned (with both start
address and size) to SoC cache line.

Fixes: 09fa964bba80 ("fs/fat: Fix 'CACHE: Misaligned operation at range' warnings")
Suggested-by: Lukasz Majewski <lukma@denx.de>
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Reviewed-by: Lukasz Majewski <lukma@denx.de>
Reviewed-by: Fabio Estevam <fabio.estevam@nxp.com>
6 years agoboard: laird: add WB50N CPU module
Ben Whitten [Thu, 23 Nov 2017 13:47:48 +0000 (13:47 +0000)]
board: laird: add WB50N CPU module

This board is based on the Atmel sama5d3 eval boards.
Supporting the following features:
 - Boot from NAND Flash
 - Ethernet
 - FIT
 - SPL

Signed-off-by: Ben Whitten <ben.whitten@lairdtech.com>
Signed-off-by: Dan Kephart <dan.kephart@lairdtech.com>
6 years agoboard: laird: add WB45N CPU module
Ben Whitten [Thu, 23 Nov 2017 13:47:47 +0000 (13:47 +0000)]
board: laird: add WB45N CPU module

This board is based on the Atmel 9x5 eval board.
Supporting the following features:
 - Boot from NAND Flash
 - Ethernet
 - FIT
 - SPL

Signed-off-by: Ben Whitten <ben.whitten@lairdtech.com>
Signed-off-by: Dan Kephart <dan.kephart@lairdtech.com>
6 years agoenv: suppress a spurious warning with GCC 7.1
Philipp Tomsich [Tue, 21 Nov 2017 22:29:40 +0000 (23:29 +0100)]
env: suppress a spurious warning with GCC 7.1

GCC 7.1 seems to be smart enough to track val through the various
static inline functions, but not smart enough to see that val will
always be initialised when no error is returned.  This triggers
the following warning:
  env/mmc.c: In function 'mmc_get_env_addr':
  env/mmc.c:121:12: warning: 'val' may be used uninitialized in this function [-Wmaybe-uninitialized]

To make it easier for compiler to understand what is going on, let's
initialise val.

Signed-off-by: Philipp Tomsich <philipp.tomsich@theobroma-systems.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
6 years agospl: make CONFIG_OF_EMBED pass dts through fdtgrep
Goldschmidt Simon [Tue, 21 Nov 2017 12:29:56 +0000 (12:29 +0000)]
spl: make CONFIG_OF_EMBED pass dts through fdtgrep

Building spl with CONFIG_OF_EMBED enabled results in an error message
on my board: "SPL image too big". This is because the fdtgrep build
step is only executed for CONFIG_OF_SEPARATE.

Fix this by moving the fdtgrep build step ('cmd_fdtgreo') from
scripts/Makefile.spl to dts/Makefile so that the reduced dtb is
available for all kinds of spl builds.

The resulting variable name for the embedded device tree blob changes,
too, which is why common.h and fdtdec.c have tiny changes.

Signed-off-by: Simon Goldschmidt <sgoldschmidt@de.pepperl-fuchs.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
6 years agoARM: dts: Rename logicpd-toredp-37xx-devkit in U-Boot
Adam Ford [Mon, 20 Nov 2017 15:56:03 +0000 (09:56 -0600)]
ARM: dts: Rename logicpd-toredp-37xx-devkit in U-Boot

In U-Boot, this device tree is compatible with both the Torpedo and
SOM-LV kits.  Let's rename it in the device tree since the U-Boot code and
show a more generic OMAP3 name.  The code auto detects between the two and
loads the proper DTB file for Linux.  This would eliminate the SOM-LV showing
the name Torpedo during boot and hopefully eliminate some confusion.

Signed-off-by: Adam Ford <aford173@gmail.com>
6 years agofs: avoid possible NULL dereference in fs_devread
Heinrich Schuchardt [Sun, 19 Nov 2017 22:49:21 +0000 (23:49 +0100)]
fs: avoid possible NULL dereference in fs_devread

It is unwise to first dereference a variable
and then to check if it was NULL.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Reviewed-by: Marek Behun <marek.behun@nic.cz>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
6 years agocmd: blk: remove unreachable code
Heinrich Schuchardt [Sun, 19 Nov 2017 22:25:32 +0000 (23:25 +0100)]
cmd: blk: remove unreachable code

Remove an unreachable return statement.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
6 years agocommon: command: tempory buffer should have size of command line buf
Heinrich Schuchardt [Sun, 19 Nov 2017 22:07:50 +0000 (23:07 +0100)]
common: command: tempory buffer should have size of command line buf

When copying the command line buffer the target array should
at least have the same size.

Cf. definition of console_buffer in common/cli_readline.c.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
6 years agodisk: efi: correct the allocation size for mbr header in stack
Patrick Delaunay [Fri, 17 Nov 2017 09:08:18 +0000 (10:08 +0100)]
disk: efi: correct the allocation size for mbr header in stack

use ALLOC_CACHE_ALIGN_BUFFER_PAD for mbr header allocation
in stack to fix alloc issue in is_gpt_valid()

this patch fix also issue for GPT partition handling
with blocksize != 512 in set_protective_mbr()

Signed-off-by: Patrick Delaunay <patrick.delaunay@st.com>
6 years agoboard: atmel: add sama5d2_ptc_ek board
Ludovic Desroches [Fri, 17 Nov 2017 06:57:12 +0000 (14:57 +0800)]
board: atmel: add sama5d2_ptc_ek board

Add the SAMA5D2 PTC EK board and remove the SAMA5D2 PTC ENGI board
which was a prototype.

Signed-off-by: Ludovic Desroches <ludovic.desroches@microchip.com>
Signed-off-by: Wenyou Yang <wenyou.yang@microchip.com>
6 years agoARM: at91: add sama5d2 smc header
Ludovic Desroches [Fri, 17 Nov 2017 06:57:11 +0000 (14:57 +0800)]
ARM: at91: add sama5d2 smc header

Add a header for SAMA5D2 SMC since it's not compatible with
SAMA5D3 one.

Signed-off-by: Ludovic Desroches <ludovic.desroches@microchip.com>
[wenyou: fix the wrong base address of the SMC register]
Signed-off-by: Wenyou Yang <wenyou.yang@microchip.com>
6 years agommc: atmel_sdhci: not on capabilities to set gck rate
Ludovic Desroches [Fri, 17 Nov 2017 06:51:27 +0000 (14:51 +0800)]
mmc: atmel_sdhci: not on capabilities to set gck rate

The capabilities have default values which doesn't reflect the reality
when it concerns the base clock and the mul value. Use a fixe rate
for the gck. 240 MHz is an arbitrary choice, it is a multiple of the
maximum SD clock frequency handle by the controller and it allows to
get a 400 kHz clock for the card initialisation.

Signed-off-by: Ludovic Desroches <ludovic.desroches@microchip.com>
Signed-off-by: Wenyou Yang <wenyou.yang@microchip.com>
6 years agoclk: at91: clk-generated: fix incorrect index of clk source
Wenyou Yang [Fri, 17 Nov 2017 06:50:22 +0000 (14:50 +0800)]
clk: at91: clk-generated: fix incorrect index of clk source

Differentiate the generic clock source selection value from the parent
clock index to fix the incorrect assignment of the generic clock
source selection.

Signed-off-by: Wenyou Yang <wenyou.yang@microchip.com>
6 years agoclk: at91: clk-generated: select absolute closest rate
Ludovic Desroches [Fri, 17 Nov 2017 06:50:21 +0000 (14:50 +0800)]
clk: at91: clk-generated: select absolute closest rate

To get the same behavior as the Linux driver, instead of selecting
the closest inferior rate, select the closest inferior or superior
rate

Signed-off-by: Ludovic Desroches <ludovic.desroches@microchip.com>
Signed-off-by: Wenyou Yang <wenyou.yang@microchip.com>
6 years agoclk: at91: Kconfig: fix the dependency of AT91_UTMI
Wenyou Yang [Fri, 17 Nov 2017 06:46:43 +0000 (14:46 +0800)]
clk: at91: Kconfig: fix the dependency of AT91_UTMI

What the AT91_UTMI depends on SPL_DM isn't right. AT91_UTMI is not
only used in SPL, also in other place, even if SPL_DM isn't enabled.

Signed-off-by: Wenyou Yang <wenyou.yang@microchip.com>
6 years agomach-stm32: Factorize MPU's region config for STM32 SoCs
Patrice Chotard [Thu, 16 Nov 2017 07:59:21 +0000 (08:59 +0100)]
mach-stm32: Factorize MPU's region config for STM32 SoCs

MPU's region setup can be factorized between STM32F4/F7/H7 SoCs family
and used a common MPU's region config.

Only one exception for STM32H7 which doesn't have device area
located at 0xA000 0000.

For STM32F4, configure_clocks() need to be moved from arch_cpu_init()
to board_early_init_f().

Signed-off-by: Patrice Chotard <patrice.chotard@st.com>
Reviewed-by: Vikas Manocha <vikas.manocha@st.com>
6 years agostm32: migrate clock structs in include/stm32_rcc.h
Patrice Chotard [Wed, 15 Nov 2017 12:14:53 +0000 (13:14 +0100)]
stm32: migrate clock structs in include/stm32_rcc.h

In order to factorize code between STM32F4 and STM32F7
migrate all structs related to RCC clocks in include/stm32_rcc.h

Signed-off-by: Patrice Chotard <patrice.chotard@st.com>
Reviewed-by: Vikas Manocha <vikas.manocha@st.com>
6 years agoclk: clk_stm32fx: add clock configuration for mmc usage
Patrice Chotard [Wed, 15 Nov 2017 12:14:52 +0000 (13:14 +0100)]
clk: clk_stm32fx: add clock configuration for mmc usage

MMC block needs 48Mhz source clock, for that we choose
to select the SAI PLL.
Update also stm32_clock_get_rate() to retrieve the MMC
clock source needed in MMC driver.

STM32F4 uses a different RCC variant than STM32F7. For STM32F4
sdmmc clocks bit are located into dckcfgr register whereas there
are located into dckcfgr2 registers on STM32F7.
In both registers, bits CK48MSEL and SDMMC1SEL are located at
the same position.

Signed-off-by: Christophe Priouzeau <christophe.priouzeau@st.com>
Signed-off-by: Patrice Chotard <patrice.chotard@st.com>
Reviewed-by: Vikas Manocha <vikas.manocha@st.com>