oweals/u-boot.git
5 years agoefi_loader: provide freestanding library
Heinrich Schuchardt [Sun, 20 Jan 2019 07:20:30 +0000 (08:20 +0100)]
efi_loader: provide freestanding library

GCC requires that freestanding programs provide memcpy(), memmove(),
memset(), and memcmp().

Provide the required library functions.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
5 years agoefi_selftest: fix variables test for GetNextVariableName()
AKASHI Takahiro [Mon, 21 Jan 2019 11:43:14 +0000 (12:43 +0100)]
efi_selftest: fix variables test for GetNextVariableName()

There is a bug in efi variables test.
Fix it with some cosmetic improvements.

Please note that efi variables test still fails at QueryVariableInfo()
and GetVariable(), but this is not due to a change in this patch.
  ==8<==
  Testing EFI API implementation

  Selected test: 'variables'

  Setting up 'variables'
  Setting up 'variables' succeeded

  Executing 'variables'
  .../u-boot/lib/efi_selftest/efi_selftest_variables.c(60):
  TODO: QueryVariableInfo failed
  .../u-boot/lib/efi_selftest/efi_selftest_variables.c(131):
  TODO: GetVariable returned wrong length 7
  .../u-boot/lib/efi_selftest/efi_selftest_variables.c(133):
  TODO: GetVariable returned wrong value
  Executing 'variables' succeeded

  Boot services terminated

  Summary: 0 failures
  ==>8==

Signed-off-by: AKASHI Takahiro <takahiro.akashi@linaro.org>
Reviewed-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
5 years agoefi_loader: implement GetNextVariableName()
AKASHI Takahiro [Mon, 21 Jan 2019 11:43:13 +0000 (12:43 +0100)]
efi_loader: implement GetNextVariableName()

The current GetNextVariableName() is a placeholder.
With this patch, it works well as expected.

Signed-off-by: AKASHI Takahiro <takahiro.akashi@linaro.org>
rebased on efi-next
Reviewed-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
5 years agoefi_loader: comments for variable services
Heinrich Schuchardt [Fri, 18 Jan 2019 18:52:05 +0000 (19:52 +0100)]
efi_loader: comments for variable services

Comment the functions implementing the runtime variable services.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
5 years agoefi_loader: remove duplicate function mem2hex()
Heinrich Schuchardt [Fri, 18 Jan 2019 17:54:26 +0000 (18:54 +0100)]
efi_loader: remove duplicate function mem2hex()

Replace duplicate function mem2hex() by inline function bin2hex().

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
5 years agoefi_loader: eliminate duplicate function hex2mem()
Heinrich Schuchardt [Fri, 18 Jan 2019 11:31:54 +0000 (12:31 +0100)]
efi_loader: eliminate duplicate function hex2mem()

Use existing inline function hex2bin() instead of defining a new one.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
5 years agoefi_loader: remove duplicate GUID definition
Heinrich Schuchardt [Tue, 15 Jan 2019 04:44:56 +0000 (05:44 +0100)]
efi_loader: remove duplicate GUID definition

Remove duplicate definition o EFI_SIMPLE_FILE_SYSTEM_PROTOCOL_GUID.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
5 years agoefi_loader: consistent build flags for EFI applications
Heinrich Schuchardt [Sat, 12 Jan 2019 14:32:06 +0000 (15:32 +0100)]
efi_loader: consistent build flags for EFI applications

At the same time adding and removing the -Os flag does not make any sense.
Actually it leads to -Os not being used.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
5 years agoefi_loader: use u16* for file name
Heinrich Schuchardt [Sat, 12 Jan 2019 11:02:33 +0000 (12:02 +0100)]
efi_loader: use u16* for file name

UTF-16 strings in our code should all be u16 *. Fix an inconsistency for
file names which may lead to a warning for printf("%ls", ).

Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
5 years agoefi_loader: fix CopyMem()
Heinrich Schuchardt [Wed, 9 Jan 2019 20:41:13 +0000 (21:41 +0100)]
efi_loader: fix CopyMem()

CopyMem() must support overlapping buffers. So replace memcpy() by
memmove().

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
5 years agoefi_loader: refactor switch to non-secure mode
Heinrich Schuchardt [Tue, 8 Jan 2019 17:13:06 +0000 (18:13 +0100)]
efi_loader: refactor switch to non-secure mode

Refactor the switch from supervisor to hypervisor to a new function called
at the beginning of do_bootefi().

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
5 years agoefi_driver: simplify error message
Heinrich Schuchardt [Sun, 6 Jan 2019 16:12:43 +0000 (17:12 +0100)]
efi_driver: simplify error message

Stating the function module is sufficient. We don't need file and line
number. Anyway the format code for the line number was incorrect (should
be %d).

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
5 years agoefi_selftest: tpl unit test, check return values
Heinrich Schuchardt [Sun, 6 Jan 2019 15:44:16 +0000 (16:44 +0100)]
efi_selftest: tpl unit test, check return values

For some API calls checks for the return values are missing.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
5 years agoefi_selftest: events unit test, check return values
Heinrich Schuchardt [Sun, 6 Jan 2019 15:38:57 +0000 (16:38 +0100)]
efi_selftest: events unit test, check return values

For some API calls checks for the return values are missing.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
5 years agoefi_selftest: SNP unit test on sandbox
Heinrich Schuchardt [Sat, 5 Jan 2019 22:50:41 +0000 (23:50 +0100)]
efi_selftest: SNP unit test on sandbox

Running the simple network protocol test on the sandbox requires setting
the environment variable ethact to a network interface connected to a DHCP
server and ethrotate to 'no'. So let's make it an on-request test on the
sandbox (selectable by setting environment variable efi_selftest).

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
5 years agoefi_loader: efi_set_variable use const void *
Heinrich Schuchardt [Sun, 30 Dec 2018 20:03:15 +0000 (21:03 +0100)]
efi_loader: efi_set_variable use const void *

The SetVariable() runtime service does not change the data passed to it.
So mark the parameter as constant.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
5 years agoefi_loader: use const efi_guid_t * for variable services
Heinrich Schuchardt [Sun, 30 Dec 2018 19:53:51 +0000 (20:53 +0100)]
efi_loader: use const efi_guid_t * for variable services

The runtime variable services never change GUIDs. So we should declare
the GUID parameters as constant.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
5 years agoefi_loader: move efi_init_obj_list() to a new efi_setup.c
AKASHI Takahiro [Sun, 30 Dec 2018 14:16:55 +0000 (15:16 +0100)]
efi_loader: move efi_init_obj_list() to a new efi_setup.c

The function, efi_init_obj_list(), can be shared in different pseudo efi
applications, like bootefi/bootmgr as well as my efishell. Moreover, it
will be utilized to extend efi initialization, for example, my "removable
disk support" patch and "capsule-on-disk support" patch in the future.

So with this patch, it will be moved to a new file, efi_setup.c, under
lib/efi_loader and exported, making no changes in functionality.

Signed-off-by: AKASHI Takahiro <takahiro.akashi@linaro.org>
Remove lines deactivated by #if 1 #else
Reviewed-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
5 years agodoc: README.uefi: fix typos
Heinrich Schuchardt [Sun, 30 Dec 2018 11:54:36 +0000 (12:54 +0100)]
doc: README.uefi: fix typos

%s/specfication/specification/
%s/selftest/self-test/
%s/little endian/little-endian/

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
5 years agoefi_selftest: allow building on ARMv7-M
Heinrich Schuchardt [Sun, 30 Dec 2018 09:52:41 +0000 (10:52 +0100)]
efi_selftest: allow building on ARMv7-M

ARMv7-M only supports the Thumb instruction set. Our current crt0 code does
not support it. With the patch we can build all unit tests of the EFI
subsystem that do not require crt0.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
5 years agoefi_loader: signature of StartImage and Exit
Heinrich Schuchardt [Fri, 28 Dec 2018 11:41:15 +0000 (12:41 +0100)]
efi_loader: signature of StartImage and Exit

We use u16* for Unicode strings and efi_uintn_t for UINTN. Correct the
signature of efi_exit() and efi_start_image().

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
5 years agoefi_loader: avoid unnecessary pointer to long conversion
Heinrich Schuchardt [Fri, 28 Dec 2018 11:41:14 +0000 (12:41 +0100)]
efi_loader: avoid unnecessary pointer to long conversion

debug() support supports %p to print pointers.

The debug message is unique. So there is not need to write a possibly
distracting line number.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
5 years agoefi_loader: fix memory allocation on sandbox
Heinrich Schuchardt [Sat, 5 Jan 2019 22:41:36 +0000 (23:41 +0100)]
efi_loader: fix memory allocation on sandbox

Commit 7b78d6438a2b ("efi_loader: Reserve unaccessible memory") introduced
a comparison between RAM top and RAM start that was not known at the time
when the patch of commit 49759743bf09 ("efi_loader: eliminate sandbox
addresses") was written.

The sandbox uses an address space that is only relevant in the sandbox
context. We have to map ram_top from the sandbox address space to the
physical address space before using it in the EFI subsystem.

Fixes: 49759743bf09 ("efi_loader: eliminate sandbox addresses")
Fixes: 7b78d6438a2b ("efi_loader: Reserve unaccessible memory")
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Alexander Graf <agraf@suse.de>
5 years agoefi_loader: CMD_BOOTEFI_HELLO_COMPILE in configs
Heinrich Schuchardt [Sun, 30 Dec 2018 09:11:14 +0000 (10:11 +0100)]
efi_loader: CMD_BOOTEFI_HELLO_COMPILE in configs

It should not be necessary to adjust CMD_BOOTEFI_HELLO_COMPILE in config
files.

arch/arm/lib/crt0_arm_efi.S cannot be compiled in thumbs mode. We can
disable CMD_BOOTEFI_HELLO_COMPILE for CONFIG_CPU_V7M. So there is no longer
a need to disable it in stm32 configs.

helloworld.efi can be built without problems on x86_64. So there is no need
to disable it in chromebook_link64_defconfig and qemu-x86_64_defconfig.

Same is true for ARM V7A. So do not disable CMD_BOOTEFI_HELLO_COMPILE in
kp_imx6q_tpc_defconfig.

Some architecture checks are already make for EFI_LOADER. There is no need
to repeat them for CMD_BOOTEFI_HELLO_COMPILE

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Reviewed-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Patrice.Chotard@st.com
Signed-off-by: Alexander Graf <agraf@suse.de>
5 years agoefi_loader: use named constant for efi_dp_from_mem()
Heinrich Schuchardt [Wed, 26 Dec 2018 21:23:00 +0000 (22:23 +0100)]
efi_loader: use named constant for efi_dp_from_mem()

When calling efi_dp_from_mem() use a named constant for the memory type.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
5 years agodistro: not taint environment variables if possible
AKASHI Takahiro [Tue, 4 Dec 2018 06:46:55 +0000 (15:46 +0900)]
distro: not taint environment variables if possible

The aim of this patch is not to have temporary variables used in
distro_bootcmd left as environment variables after run something.
See the discussion[1].
Without this patch, saveenv command also saves those variables, too.
While they are apparently safe, scsi_need_init can be harmful.

Please note that, in most cases, a variable should be converted to
hush's local variable, while "devplist" cannot because it is created
by "part" command as an environment variable.

[1] https://lists.denx.de/pipermail/u-boot/2018-December/350209.html

Signed-off-by: AKASHI Takahiro <takahiro.akashi@linaro.org>
Signed-off-by: Alexander Graf <agraf@suse.de>
5 years agoefi_loader: efi_connect_controller() use %pD
Heinrich Schuchardt [Sun, 9 Dec 2018 15:39:20 +0000 (16:39 +0100)]
efi_loader: efi_connect_controller() use %pD

EFI_ENTRY in efi_connect_controller() should use %pD to print the remaining
device path.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
5 years agoefi_loader: efi_add_runtime_mmio()
Heinrich Schuchardt [Sun, 23 Dec 2018 01:35:13 +0000 (02:35 +0100)]
efi_loader: efi_add_runtime_mmio()

The first parameter of efi_add_runtime_mmio() is a pointer to a
pointer. This should be reflected in the documentation.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
5 years agodistro_bootcmd: add NVME support
Heinrich Schuchardt [Fri, 21 Dec 2018 01:18:16 +0000 (02:18 +0100)]
distro_bootcmd: add NVME support

Some boards support NVME drives. We should be able to use them as boot
devices.

NVME access requires running 'nvme scan'.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Alexander Graf <agraf@suse.de>
5 years agotest: tests for u16_strdup() and u16_strcpy()
Heinrich Schuchardt [Fri, 14 Dec 2018 21:00:37 +0000 (22:00 +0100)]
test: tests for u16_strdup() and u16_strcpy()

Provide unit tests for u16_strdup() and u16_strcpy().

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
5 years agolib: add u16_strcpy/strdup functions
Akashi, Takahiro [Fri, 14 Dec 2018 10:10:38 +0000 (19:10 +0900)]
lib: add u16_strcpy/strdup functions

Add u16_strcpy() and u16_strdup(). The latter function will be
used later in implementing efi HII database protocol.

Signed-off-by: Akashi Takahiro <takahiro.akashi@linaro.org>
Reviewed-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
5 years agoefi_loader: struct efi_configuration_table
Heinrich Schuchardt [Mon, 17 Dec 2018 23:06:05 +0000 (00:06 +0100)]
efi_loader: struct efi_configuration_table

Commit 393fccdf6c73 ("efi_loader: efi_guid_t must be 64-bit aligned")
has changed the alignment of efi_guid_t. This changed the size of
struct efi_configuration_table on 32-bit systems form 20 to 24 bytes. As
an array of this type is pointed to by the system table this breaks
compatibility with existing versions of GRUB and Linux. Let's get back the
original size by using the attribute __packed.

Fixes: 393fccdf6c73 ("efi_loader: efi_guid_t must be 64-bit aligned")
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
5 years agoefi_loader: efi_guid_t must be 64-bit aligned
Heinrich Schuchardt [Sun, 16 Dec 2018 10:16:03 +0000 (11:16 +0100)]
efi_loader: efi_guid_t must be 64-bit aligned

The UEFI Specification Version 2.7 Errata A defines:

"EFI_GUID
128-bit buffer containing a unique identifier value.
Unless otherwise specified, aligned on a 64-bit boundary."

Before this patch efi_guid_t was 8-bit aligned.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Acked-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Alexander Graf <agraf@suse.de>
5 years agotest/py: use default load address for tftp
Heinrich Schuchardt [Sat, 26 Jan 2019 14:25:12 +0000 (15:25 +0100)]
test/py: use default load address for tftp

On x86_64 the size of the file u-boot loaded by the tftp test has grown in
size such that when loading the file to 0x200000 it overwrites a memory
area reserved for PCI.

If no load address is specified for tftp do not use the ram base address
(or if zero 0x200000) but the default address.

Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
5 years agoMerge git://git.denx.de/u-boot-x86
Tom Rini [Tue, 12 Feb 2019 12:16:46 +0000 (07:16 -0500)]
Merge git://git.denx.de/u-boot-x86

- Edison switch to CONFIG_OF_SEPARATE.
- Tangier initial ACPI support for PMIC device.
- TSC timer driver update to support native calibration.
- Fixes to 64-bit U-Boot proper.

5 years agox86: tangier: Add initial ACPI support for PMIC device
Andy Shevchenko [Tue, 5 Feb 2019 11:06:57 +0000 (13:06 +0200)]
x86: tangier: Add initial ACPI support for PMIC device

Basin Cove PMIC is connected to I2C0 bus which is hidden from the OS
and access is going via SCU device, enumerated via PCI.

For now, we add just a minimum support of PMIC device to allow enabling,
e.g. USB OTG, in the OS.

Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Acked-by: Bin Meng <bmeng.cn@gmail.com>
5 years agox86: Use the existing GDT in the ROM for 64-bit U-Boot proper
Bin Meng [Thu, 31 Jan 2019 16:22:14 +0000 (08:22 -0800)]
x86: Use the existing GDT in the ROM for 64-bit U-Boot proper

It is unnecessary to use a RAM version GDT for 64-bit U-Boot proper.
In fact we can just use the ROM version directly, which not only
eliminates the risk of being overwritten by application, but also
removes the complexity of patching the cpu_call64().

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
5 years agox86: Don't copy the cpu_call64() function to a hardcoded address
Bin Meng [Thu, 31 Jan 2019 16:22:13 +0000 (08:22 -0800)]
x86: Don't copy the cpu_call64() function to a hardcoded address

Before jumping to 64-bit U-Boot proper, SPL copies the cpu_call64()
function to a hardcoded address 0x3000000. This can have potential
conflicts with application usage. Switch the destination address
to be allocated from the heap to avoid such risk.

Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
5 years agox86: Change 4-level page table base address to low memory
Bin Meng [Thu, 31 Jan 2019 16:22:12 +0000 (08:22 -0800)]
x86: Change 4-level page table base address to low memory

At present the 4-level page table base address for 64-bit U-Boot
proper is assigned an address that conflicts with CONFIG_LOADADDR.
Change it to an address within the low memory range instead.

Fixes crashes seen when 'dhcp' on QEMU x86_64 with
"-net nic -net user,tftp=.,bootfile=u-boot".

Reported-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Bin Meng <bmeng.cn@gmail.com>
Tested-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Reviewed-by: Simon Glass <sjg@chromium.org>
5 years agox86: tsc: Add support for native calibration of TSC freq
Bernhard Messerklinger [Mon, 7 Jan 2019 11:14:40 +0000 (12:14 +0100)]
x86: tsc: Add support for native calibration of TSC freq

Add native tsc calibration function. Calibrate the tsc timer the same
way as linux does in arch/x86/kernel/tsc.c.

Fixes booting for Apollo Lake processors.

Signed-off-by: Bernhard Messerklinger <bernhard.messerklinger@br-automation.com>
Reviewed-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
5 years agodoc: Fix CONFIG_OF_SEPARATE description
Andy Shevchenko [Tue, 8 Jan 2019 14:17:42 +0000 (16:17 +0200)]
doc: Fix CONFIG_OF_SEPARATE description

CONFIG_OF_SEPARATE description is not in align with actual code in Makefile and
thus has misleading instructions and explanation.

Make it aligned with the actual code.

Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
5 years agox86: edison: Switch to CONFIG_OF_SEPARATE
Andy Shevchenko [Tue, 8 Jan 2019 14:17:43 +0000 (16:17 +0200)]
x86: edison: Switch to CONFIG_OF_SEPARATE

There is no need for Intel Edison to have CONFIG_OF_EMBED to be enabled.
Replace it with CONFIG_OF_SEPARATE.

There is no functional change since u-boot.bin always contains DTB
either embedded or attached.

Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Tested-by: Ferry Toth <ftoth@exalondelft.nl>
5 years agoMerge branch 'master' of git://git.denx.de/u-boot-i2c
Tom Rini [Mon, 11 Feb 2019 16:15:34 +0000 (11:15 -0500)]
Merge branch 'master' of git://git.denx.de/u-boot-i2c

- DM I2C improvements

5 years agoMerge git://git.denx.de/u-boot-marvell
Tom Rini [Mon, 11 Feb 2019 15:35:52 +0000 (10:35 -0500)]
Merge git://git.denx.de/u-boot-marvell

- Fix BUILD_TARGET for ARCH_MVEBU from Baruch
- Fix MVEBU PCIe reset issues from Baruch
- Increase DDR stability on x530 from Chris

5 years agoARM: mvebu: x530: use MV_DDR_FREQ_SAR
Chris Packham [Mon, 11 Feb 2019 01:19:56 +0000 (14:19 +1300)]
ARM: mvebu: x530: use MV_DDR_FREQ_SAR

MV_DDR_FREQ_SAR lets the DDR frequency be determined by hardware
strapping. This also has the side effect of running the DDR clock in
synchronous mode with the CPU core clock rather than from an independent
PLL. We've seen this improve reliability in operation across a number of
boards and temperature ranges.

Signed-off-by: Chris Packham <judge.packham@gmail.com>
Reviewed-by: Stefan Roese <sr@denx.de>
Signed-off-by: Stefan Roese <sr@denx.de>
5 years agoKconfig: fix BUILD_TARGET for ARCH_MVEBU
Baruch Siach [Thu, 7 Feb 2019 11:21:16 +0000 (13:21 +0200)]
Kconfig: fix BUILD_TARGET for ARCH_MVEBU

Commit dc146ca11187 ("Kconfig: Migrate CONFIG_BUILD_TARGET") made the
mvebu default build target depend on CONFIG_SPL_BUILD. Unfortunately,
there is no such Kconfig symbol. Use the CONFIG_SPL symbol instead to
fix that.

Cc: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: Baruch Siach <baruch@tkos.co.il>
Reviewed-by: Stefan Roese <sr@denx.de>
Reviewed-by: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: Stefan Roese <sr@denx.de>
5 years agoarm: mvebu: cf gt-8k: dts: add PCIe slot reset support
Baruch Siach [Sun, 3 Feb 2019 13:15:40 +0000 (15:15 +0200)]
arm: mvebu: cf gt-8k: dts: add PCIe slot reset support

Describe the mini-PCIe slot gpio reset signal. This enables PCIe devices
on Clearfog GT-8K.

Signed-off-by: Baruch Siach <baruch@tkos.co.il>
Reviewed-by: Stefan Roese <sr@denx.de>
Signed-off-by: Stefan Roese <sr@denx.de>
5 years agopcie: designware: mvebu: fix reset release polarity
Baruch Siach [Sun, 3 Feb 2019 13:15:39 +0000 (15:15 +0200)]
pcie: designware: mvebu: fix reset release polarity

The dm_gpio_set_value() routine sets signal logical level, with
GPIO_ACTIVE_LOW/HIGH value taken into account. Reset active value is 1
(asserted), while reset inactive value is 0 (de-asserted). Fix the reset
toggle code to set the correct reset logic value.

Reported-by: Sven Auhagen <sven.auhagen@voleatech.de>
Signed-off-by: Baruch Siach <baruch@tkos.co.il>
Reviewed-by: Stefan Roese <sr@denx.de>
Signed-off-by: Stefan Roese <sr@denx.de>
5 years agoarm: mvebu: mcbin: dts: fix PCIe reset polarity
Baruch Siach [Sun, 3 Feb 2019 13:15:38 +0000 (15:15 +0200)]
arm: mvebu: mcbin: dts: fix PCIe reset polarity

The PCIe slot PERST signal is active low. Fix the gpio signal
description in the dts.

This happened to work because the pcie_dw_mvebu driver sets the reset
gpio level to 1 (high) to release the reset. The following commit will
fix that.

Signed-off-by: Baruch Siach <baruch@tkos.co.il>
Reviewed-by: Stefan Roese <sr@denx.de>
Signed-off-by: Stefan Roese <sr@denx.de>
5 years agoi2c: mux: Generate longer i2c mux name
Michal Simek [Thu, 31 Jan 2019 15:31:03 +0000 (16:31 +0100)]
i2c: mux: Generate longer i2c mux name

For !DM case busses are listed as
ZynqMP> i2c bus
Bus 0: zynq_0
Bus 1: zynq_0->PCA9544A@0x75:0
Bus 2: zynq_0->PCA9544A@0x75:1
Bus 3: zynq_0->PCA9544A@0x75:2
Bus 4: zynq_1
Bus 5: zynq_1->PCA9548@0x74:0
Bus 6: zynq_1->PCA9548@0x74:1
Bus 7: zynq_1->PCA9548@0x74:2
Bus 8: zynq_1->PCA9548@0x74:3
Bus 9: zynq_1->PCA9548@0x74:4
Bus 10: zynq_1->PCA9548@0x75:0
Bus 11: zynq_1->PCA9548@0x75:1
Bus 12: zynq_1->PCA9548@0x75:2
Bus 13: zynq_1->PCA9548@0x75:3
Bus 14: zynq_1->PCA9548@0x75:4
Bus 15: zynq_1->PCA9548@0x75:5
Bus 16: zynq_1->PCA9548@0x75:6
Bus 17: zynq_1->PCA9548@0x75:7

where is exactly describing i2c bus topology.
By moving to DM case i2c mux buses are using names from DT and because
i2c-muxes describing sub busses with the same names like i2c@0, etc it
is hard to identify which bus is where.
Linux is adding topology information to i2c-mux busses to identify them
better.
This patch is doing the same and composing bus name with topology
information.

When patch is applied with topology information on zcu102-revA.
ZynqMP> i2c bus
Bus 0: i2c@ff020000
   20: gpio@20, offset len 1, flags 0
   21: gpio@21, offset len 1, flags 0
   75: i2c-mux@75, offset len 1, flags 0
Bus 2: i2c@ff020000->i2c-mux@75->i2c@0
Bus 3: i2c@ff020000->i2c-mux@75->i2c@1
Bus 4: i2c@ff020000->i2c-mux@75->i2c@2
Bus 1: i2c@ff030000  (active 1)
   74: i2c-mux@74, offset len 1, flags 0
   75: i2c-mux@75, offset len 1, flags 0
Bus 5: i2c@ff030000->i2c-mux@74->i2c@0  (active 5)
   54: eeprom@54, offset len 1, flags 0
Bus 6: i2c@ff030000->i2c-mux@74->i2c@1
Bus 7: i2c@ff030000->i2c-mux@74->i2c@2
Bus 8: i2c@ff030000->i2c-mux@74->i2c@3
Bus 9: i2c@ff030000->i2c-mux@74->i2c@4
Bus 10: i2c@ff030000->i2c-mux@75->i2c@0
Bus 11: i2c@ff030000->i2c-mux@75->i2c@1
Bus 12: i2c@ff030000->i2c-mux@75->i2c@2
Bus 13: i2c@ff030000->i2c-mux@75->i2c@3
Bus 14: i2c@ff030000->i2c-mux@75->i2c@4
Bus 15: i2c@ff030000->i2c-mux@75->i2c@5
Bus 16: i2c@ff030000->i2c-mux@75->i2c@6
Bus 17: i2c@ff030000->i2c-mux@75->i2c@7

Behavior before the patch is applied.
ZynqMP> i2c bus
Bus 0: i2c@ff020000
   20: gpio@20, offset len 1, flags 0
   21: gpio@21, offset len 1, flags 0
   75: i2c-mux@75, offset len 1, flags 0
Bus 2: i2c@0
Bus 3: i2c@1
Bus 4: i2c@2
Bus 1: i2c@ff030000  (active 1)
   74: i2c-mux@74, offset len 1, flags 0
   75: i2c-mux@75, offset len 1, flags 0
Bus 5: i2c@0  (active 5)
   54: eeprom@54, offset len 1, flags 0
Bus 6: i2c@1
Bus 7: i2c@2
Bus 8: i2c@3
Bus 9: i2c@4
Bus 10: i2c@0
Bus 11: i2c@1
Bus 12: i2c@2
Bus 13: i2c@3
Bus 14: i2c@4
Bus 15: i2c@5
Bus 16: i2c@6
Bus 17: i2c@7

Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Heiko Schocher <hs@denx.de>
5 years agoi2c: Fill req_seq in i2c_post_bind()
Michal Simek [Thu, 31 Jan 2019 15:31:02 +0000 (16:31 +0100)]
i2c: Fill req_seq in i2c_post_bind()

For i2c controllers which are missing alias in DT there is no req_seq
setup. This function is setting up proper ID based on highest found
alias ID.

On zcu102 this is the behavior when patch is applied.
ZynqMP> i2c bus
Bus 0: i2c@ff020000
   20: gpio@20, offset len 1, flags 0
   21: gpio@21, offset len 1, flags 0
   75: i2c-mux@75, offset len 1, flags 0
Bus 2: i2c@0
Bus 3: i2c@1
Bus 4: i2c@2
Bus 1: i2c@ff030000  (active 1)
   74: i2c-mux@74, offset len 1, flags 0
   75: i2c-mux@75, offset len 1, flags 0
Bus 5: i2c@0  (active 5)
   54: eeprom@54, offset len 1, flags 0
Bus 6: i2c@1
Bus 7: i2c@2
Bus 8: i2c@3
Bus 9: i2c@4
Bus 10: i2c@0
Bus 11: i2c@1
Bus 12: i2c@2
Bus 13: i2c@3
Bus 14: i2c@4
Bus 15: i2c@5
Bus 16: i2c@6
Bus 17: i2c@7

Before this patch applied (controllers have -1 ID)
ZynqMP> i2c bus
Bus 0: i2c@ff020000
   20: gpio@20, offset len 1, flags 0
   21: gpio@21, offset len 1, flags 0
   75: i2c-mux@75, offset len 1, flags 0
Bus -1: i2c@0
Bus -1: i2c@1
Bus -1: i2c@2
Bus 1: i2c@ff030000  (active 1)
   74: i2c-mux@74, offset len 1, flags 0
   75: i2c-mux@75, offset len 1, flags 0
Bus -1: i2c@0  (active 0)
   54: eeprom@54, offset len 1, flags 0
Bus -1: i2c@1
Bus -1: i2c@2
Bus -1: i2c@3
Bus -1: i2c@4
Bus -1: i2c@0
Bus -1: i2c@1
Bus -1: i2c@2
Bus -1: i2c@3
Bus -1: i2c@4
Bus -1: i2c@5
Bus -1: i2c@6
Bus -1: i2c@7

Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Reviewed-by: Heiko Schocher <hs@denx.de>
5 years agoi2c: dm: Record maximum id of devices before probing devices
Michal Simek [Thu, 31 Jan 2019 15:31:01 +0000 (16:31 +0100)]
i2c: dm: Record maximum id of devices before probing devices

There is a need to find out the first free i2c ID which can be used for
i2s buses (including i2c buses connected to i2c mux). Do it early in
init and share this variable with other i2c classes for uniq bus
identification.

add from hs:
fix build problem in i2c-uclass.c for omap devices

Signed-off-by: Michal Simek <michal.simek@xilinx.com>
Reviewed-by: Heiko Schocher <hs@denx.de>
5 years agoMerge branch 'master' of git://git.denx.de/u-boot-sh
Tom Rini [Sun, 10 Feb 2019 13:11:53 +0000 (08:11 -0500)]
Merge branch 'master' of git://git.denx.de/u-boot-sh

- SD/MMC fixes and ext4 memory leak fix

5 years agoMerge tag 'dm-pull-10feb19' of git://git.denx.de/u-boot-dm
Tom Rini [Sun, 10 Feb 2019 13:11:32 +0000 (08:11 -0500)]
Merge tag 'dm-pull-10feb19' of git://git.denx.de/u-boot-dm

Samsung sound patches (applied for Samsung maintainer)
Common sound support
buildman environment support
of-platdata documentation improvements

5 years agoMerge branch '2019-02-08-master-imports'
Tom Rini [Sun, 10 Feb 2019 13:04:53 +0000 (08:04 -0500)]
Merge branch '2019-02-08-master-imports'

- bcm6345 watchdog, bcm63158/bcm963158 initial support.
- Various TI platform resyncs and improvements.
- FDT support in Android-format images.
- stm32mp1 improvements.

5 years agosound: Allow audio codecs to be used by other SoCs
Simon Glass [Thu, 27 Dec 2018 20:24:39 +0000 (13:24 -0700)]
sound: Allow audio codecs to be used by other SoCs

At present there is still some samsung-specific code in the audio codecs.
Remove it so that these can be used by other SoCs.

Signed-off-by: Simon Glass <sjg@chromium.org>
5 years agosound: i2s: Tidy up a few comments
Simon Glass [Thu, 27 Dec 2018 20:24:38 +0000 (13:24 -0700)]
sound: i2s: Tidy up a few comments

Fix a struct typo and drop a comment (and function prototype) which is not
actually used.

Signed-off-by: Simon Glass <sjg@chromium.org>
5 years agosound: samsung: Fix 'regiter' typo
Simon Glass [Thu, 27 Dec 2018 20:24:37 +0000 (13:24 -0700)]
sound: samsung: Fix 'regiter' typo

Fix a typo that appears many times in this file.

Signed-off-by: Simon Glass <sjg@chromium.org>
5 years agomisc: Allow child devices
Simon Glass [Thu, 27 Dec 2018 20:24:36 +0000 (13:24 -0700)]
misc: Allow child devices

Allow misc devices to have children, so that we can use this uclass for
cases where a child device (e.g. I2S) needs to access a misc driver for
transferring data.

Signed-off-by: Simon Glass <sjg@chromium.org>
5 years agobuildman: fix typo
Chris Packham [Fri, 18 Jan 2019 07:40:29 +0000 (20:40 +1300)]
buildman: fix typo

Fix a typo in the error message from CheckOutputDir().

Signed-off-by: Chris Packham <judge.packham@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
5 years agoof-platdata: improve documentation
Simon Goldschmidt [Wed, 16 Jan 2019 19:40:18 +0000 (20:40 +0100)]
of-platdata: improve documentation

Improve some things in the documentation of OF_PLATDATA that I found
while porting socfgpa_gen5 to it.

Signed-off-by: Simon Goldschmidt <simon.k.r.goldschmidt@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
5 years agobuildman: Write the environment out to an 'env' file
Simon Glass [Mon, 7 Jan 2019 23:44:23 +0000 (16:44 -0700)]
buildman: Write the environment out to an 'env' file

Sometimes it is useful to see the environment that was used to build
U-Boot. Write this out to a file in the build directory.

Signed-off-by: Simon Glass <sjg@chromium.org>
5 years agommc: omap_hsmmc: Use regulator_set_enable_if_allowed for enabling regulator
Lokesh Vutla [Fri, 11 Jan 2019 09:45:52 +0000 (15:15 +0530)]
mmc: omap_hsmmc: Use regulator_set_enable_if_allowed for enabling regulator

Use regulator_set_enable_if_allowed() api instead of regulator_set_enable()
while enabling io regulators. This way the driver doesn't see an error
when disabling an always-on regulator and when enabling is not supported.

Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Lokesh Vutla <lokeshvutla@ti.com>
5 years agopower: regulator: Introduce regulator_set_enable_if_allowed api
Lokesh Vutla [Fri, 11 Jan 2019 09:45:51 +0000 (15:15 +0530)]
power: regulator: Introduce regulator_set_enable_if_allowed api

regulator_set_enable() api throws an error in the following three cases:
- when requested to disable an always-on regulator
- when set_enable() ops not provided by regulator driver
- when enabling is actually failed.(Error returned by the regulator driver)

Sometimes consumer drivers doesn't want to track the first two scenarios
and just need to worry about the case where enabling is actually failed.
But it is also a good practice to have an error value returned in the
first two cases.

So introduce an api regulator_set_enable_if_allowed() which ignores the
first two error cases and returns an error as given by regulator driver.
Consumer drivers can use this api need not worry about the first two
error conditions.

Signed-off-by: Lokesh Vutla <lokeshvutla@ti.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
5 years agoRevert "power: regulator: Return success on attempt to disable an always-on regulator"
Lokesh Vutla [Fri, 11 Jan 2019 09:45:50 +0000 (15:15 +0530)]
Revert "power: regulator: Return success on attempt to disable an always-on regulator"

This reverts commit e17e0ceb83538c015a50b965547f2f4d38f81c5d.

It is advised to return an error when trying to disable an always-on
regulator and let the consumer driver handle the error if needed.

Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Lokesh Vutla <lokeshvutla@ti.com>
5 years agofdt: tegra: Drop COMPAT_AMS_AS3722
Simon Glass [Sat, 12 Jan 2019 01:37:15 +0000 (18:37 -0700)]
fdt: tegra: Drop COMPAT_AMS_AS3722

This is no-longer used. Drop it.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Lukasz Majewski <lukma@denx.de>
5 years agofdt: samsung: Drop unused fdt_compat_id values
Simon Glass [Sat, 12 Jan 2019 01:37:14 +0000 (18:37 -0700)]
fdt: samsung: Drop unused fdt_compat_id values

This enum still exists but we can shrink it a little based on recent
driver-model conversions with samsung. Update it to remove unused items.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Lukasz Majewski <lukma@denx.de>
5 years agosamsung: mmc: Drop old MMC init code
Simon Glass [Sat, 12 Jan 2019 01:37:13 +0000 (18:37 -0700)]
samsung: mmc: Drop old MMC init code

Now that these boards use driver model we can drop the old code. At
present s5p_mmc_init() is still used by goni and smdkv310 so cannot be
removed unless we remove those boards.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Lukasz Majewski <lukma@denx.de>
Reviewed-by: Minkyu Kang <mk7.kang@samsung.com>
5 years agosamsung: Drop board_enable_audio_codec()
Simon Glass [Sat, 12 Jan 2019 01:37:12 +0000 (18:37 -0700)]
samsung: Drop board_enable_audio_codec()

This function is not needed now since the audio codecs have been converted
to proper drivers. The codec-enable GPIO is handled there.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Lukasz Majewski <lukma@denx.de>
Reviewed-by: Minkyu Kang <mk7.kang@samsung.com>
5 years agospring: Update flashmap details
Simon Glass [Sat, 12 Jan 2019 01:37:11 +0000 (18:37 -0700)]
spring: Update flashmap details

Update the flashmap so that this board can be started over USB A-A. It
is slightly different from snow.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Lukasz Majewski <lukma@denx.de>
5 years agospring: Update sound to use max98088 codec
Simon Glass [Sat, 12 Jan 2019 01:37:10 +0000 (18:37 -0700)]
spring: Update sound to use max98088 codec

Update the spring settings to use this codec, which is what it actually
shipped with.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Lukasz Majewski <lukma@denx.de>
5 years agosound: Add a driver for max98088
Simon Glass [Sat, 12 Jan 2019 01:37:09 +0000 (18:37 -0700)]
sound: Add a driver for max98088

This chip is used by spring. Add a driver for it and update the
samsung_sound driver to pick it up.

Signed-off-by: Simon Glass <sjg@chromium.org>
5 years agoexynos: Convert to use CONFIG_BLK
Simon Glass [Sat, 12 Jan 2019 01:37:08 +0000 (18:37 -0700)]
exynos: Convert to use CONFIG_BLK

Move all exynos boards over to use CONFIG_BLK.

This converts s5p_goni also, but adding dummy functions for pinmux and
peripheral ID. This will not function correctly, but gives the maintainer
more time to convert the board if desired.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Lukasz Majewski <lukma@denx.de>
Reviewed-by: Minkyu Kang <mk7.kang@samsung.com>
5 years agoexynos: Drop duplicate 'model' line
Simon Glass [Sat, 12 Jan 2019 01:37:07 +0000 (18:37 -0700)]
exynos: Drop duplicate 'model' line

At present the model is shown twice, once in the generic code and once
in the exynos code. Drop the latter.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Lukasz Majewski <lukma@denx.de>
Reviewed-by: Minkyu Kang <mk7.kang@samsung.com>
5 years agoConvert CONFIG_BOARD_TYPES to Kconfig
Simon Glass [Sat, 12 Jan 2019 01:37:06 +0000 (18:37 -0700)]
Convert CONFIG_BOARD_TYPES to Kconfig

This converts the following to Kconfig:
   CONFIG_BOARD_TYPES

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Lukasz Majewski <lukma@denx.de>
Acked-by: Alexey Brodkin <abrodkin@synopsys.com>
Reviewed-by: Minkyu Kang <mk7.kang@samsung.com>
5 years agoarm: mach-k3: common: Clean up ATF image startup function
Andreas Dannenberg [Mon, 4 Feb 2019 18:58:47 +0000 (12:58 -0600)]
arm: mach-k3: common: Clean up ATF image startup function

Perform some cosmetic cleanup of the ATF image startup function, namely
fixing a spelling mistake, capitalization of a few words, spacing, as
well aligning how errors are printed and as using panic() for cases that
were using a combination of printf() + hang().

Signed-off-by: Andreas Dannenberg <dannenberg@ti.com>
Reviewed-by: Lokesh Vutla <lokeshvutla@ti.com>
5 years agobcm968580xref: switch to CONFIG_OF_SEPARATE
Philippe Reynes [Mon, 4 Feb 2019 14:14:18 +0000 (15:14 +0100)]
bcm968580xref: switch to CONFIG_OF_SEPARATE

The option OF_EMBED is deprecated,
so we switch to CONFIG_OF_SEPARATE

Signed-off-by: Philippe Reynes <philippe.reynes@softathome.com>
5 years agobcm968580: rename to bcm968580xref
Philippe Reynes [Mon, 4 Feb 2019 14:14:17 +0000 (15:14 +0100)]
bcm968580: rename to bcm968580xref

The name of the board is bcm968580xref,
so rename the config to bcm968580xref too.

Signed-off-by: Philippe Reynes <philippe.reynes@softathome.com>
5 years agoarmv7R: K3: am654: Fix order of debug elements in x509 template
Andrew F. Davis [Fri, 1 Feb 2019 21:04:58 +0000 (15:04 -0600)]
armv7R: K3: am654: Fix order of debug elements in x509 template

The first element in the debug section is expected to be debugUID.
ROM will not parse this correctly when out of order, fix this here.

Signed-off-by: Andrew F. Davis <afd@ti.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
Reviewed-by: Lokesh Vutla <lokeshvutla@ti.com>
5 years agoam65x_evm: Allow bootm to load larger kernels
Andrew F. Davis [Fri, 1 Feb 2019 21:04:57 +0000 (15:04 -0600)]
am65x_evm: Allow bootm to load larger kernels

Bootm will fail to load kernels over 8MB, this is not enough
for our 64bit kernel images. Increase this to 64MB.

Signed-off-by: Andrew F. Davis <afd@ti.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
Reviewed-by: Lokesh Vutla <lokeshvutla@ti.com>
5 years agoconfigs: am65x_evm_r5: Enable GPT support
Andrew F. Davis [Fri, 1 Feb 2019 21:04:56 +0000 (15:04 -0600)]
configs: am65x_evm_r5: Enable GPT support

The second loader stages may be stored on GPT partitions,
enable support for this here.

Signed-off-by: Andrew F. Davis <afd@ti.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
Reviewed-by: Lokesh Vutla <lokeshvutla@ti.com>
5 years agocmd: sata: add null pointer check for dev
Marcel Ziswiler [Fri, 1 Feb 2019 15:01:08 +0000 (16:01 +0100)]
cmd: sata: add null pointer check for dev

Calling sata_scan() with a null pointer probably won't make much sense.

Signed-off-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
5 years agodm: device: fail uclass_find_first_device() if list_empty
Marcel Ziswiler [Fri, 1 Feb 2019 15:01:07 +0000 (16:01 +0100)]
dm: device: fail uclass_find_first_device() if list_empty

While uclass_find_device() fails with -ENODEV in case of list_empty
strangely uclass_find_first_device() returns 0.

Fix uclass_find_first_device() to also fail with -ENODEV instead.

Signed-off-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
5 years agobcm963158: enable watchdog and reboot with watchdog
Philippe Reynes [Thu, 31 Jan 2019 17:57:39 +0000 (18:57 +0100)]
bcm963158: enable watchdog and reboot with watchdog

Enable watchdog and reboot with watchdog in the configuration.

Signed-off-by: Philippe Reynes <philippe.reynes@softathome.com>
5 years agodt: bcm63158: add watchdog
Philippe Reynes [Thu, 31 Jan 2019 17:57:38 +0000 (18:57 +0100)]
dt: bcm63158: add watchdog

This commit add watchdog and sysreset watchdog
in the bcm63158 device tree.

Signed-off-by: Philippe Reynes <philippe.reynes@softathome.com>
5 years agowatchdog: bcm6345: allow to use this driver on arm bcm63158
Philippe Reynes [Thu, 31 Jan 2019 17:57:37 +0000 (18:57 +0100)]
watchdog: bcm6345: allow to use this driver on arm bcm63158

This IP is also used on some arm SoC, so we allow to
use it on arm bcm63158 too.

Signed-off-by: Philippe Reynes <philippe.reynes@softathome.com>
5 years agobcm963158: add initial support
Philippe Reynes [Thu, 31 Jan 2019 17:57:36 +0000 (18:57 +0100)]
bcm963158: add initial support

This add the initial support of the broadcom reference
board bcm963158 with a bcm63158 SoC.

This board has 1 GB of ram, 512 MB of flash (nand),
2 usb port, 1 uart, 4 ethernet ports (LAN), 1 ethernet port (WAN).

Signed-off-by: Philippe Reynes <philippe.reynes@softathome.com>
5 years agobcm63158: add initial support
Philippe Reynes [Thu, 31 Jan 2019 17:57:35 +0000 (18:57 +0100)]
bcm63158: add initial support

This add the initial support of the broadcom bcm63158 SoC family,
only the cpu, dram and uart are supported.

Signed-off-by: Philippe Reynes <philippe.reynes@softathome.com>
5 years agoinitcall: Move to inline function
Alexander Graf [Thu, 31 Jan 2019 15:06:23 +0000 (16:06 +0100)]
initcall: Move to inline function

The board_r init function was complaining that we are looping through
an array, calling all our tiny init stubs sequentially via indirect
function calls (which can't be speculated, so they are slow).

The solution to that is pretty easy though. All we need to do is inline
the function that loops through the functions and the compiler will
automatically convert almost all indirect calls into direct inlined code.

With this patch, the overall code size drops (by 40 bytes on riscv64)
and boot time should become measurably faster for every target.

Signed-off-by: Alexander Graf <agraf@suse.de>
5 years agoboard/BuR/brppt1: drop DM_I2C_COMPAT
Hannes Schmelzer [Thu, 31 Jan 2019 08:24:45 +0000 (09:24 +0100)]
board/BuR/brppt1: drop DM_I2C_COMPAT

The TPS62517 PMIC driver has been partially converted to DM, so the
legacy I2C access layer isn't needed anymore.

Signed-off-by: Hannes Schmelzer <hannes.schmelzer@br-automation.com>
5 years agommc: omap_hsmmc: Workaround errata regarding SDR104/HS200 tuning failures (i929)
Faiz Abbas [Wed, 30 Jan 2019 12:38:42 +0000 (18:08 +0530)]
mmc: omap_hsmmc: Workaround errata regarding SDR104/HS200 tuning failures (i929)

Errata i929 in certain OMAP5/DRA7XX/AM57XX silicon revisions
(SPRZ426D - November 2014 - Revised February 2018 [1]) mentions
unexpected tuning pattern errors. A small failure band may be present
in the tuning range which may be missed by the current algorithm.
Furthermore, the failure bands vary with temperature leading to
different optimum tuning values for different temperatures.

As suggested in the related Application Report (SPRACA9B - October 2017
- Revised July 2018 [2]), tuning should be done in two stages.
In stage 1, assign the optimum ratio in the maximum pass window for the
current temperature. In stage 2, if the chosen value is close to the
small failure band, move away from it in the appropriate direction.

References:
[1] http://www.ti.com/lit/pdf/sprz426
[2] http://www.ti.com/lit/pdf/SPRACA9

Signed-off-by: Faiz Abbas <faiz_abbas@ti.com>
5 years agoclk: stm32mp1: correctly handle Clock Spreading Generator
Patrick Delaunay [Wed, 30 Jan 2019 12:07:06 +0000 (13:07 +0100)]
clk: stm32mp1: correctly handle Clock Spreading Generator

To activate the csg option, the driver need to set the bit2
of PLLNCR register = SSCG_CTRL: Spread Spectrum Clock Generator
of PLLn enable.

Signed-off-by: Patrick Delaunay <patrick.delaunay@st.com>
5 years agodts: stm32mp1: clock tree update
Patrick Delaunay [Wed, 30 Jan 2019 12:07:05 +0000 (13:07 +0100)]
dts: stm32mp1: clock tree update

- Add st,digbypass on clk_hse node (needed for board rev.C)
- MLAHB/AHB max frequency increased from 200 to 209MHz, with:
  - PLL3P set to 208.8MHz for MCU sub-system
  - PLL3Q set to 24.57MHz for 48kHz SAI/SPI2S
  - PLL3R set to 11.29MHz for 44.1kHz SAI/SPI2S
  - PLL4P set to 99MHz for SDMMC and SPDIFRX
  - PLL4Q set to 74.25MHz for EVAL board

Signed-off-by: Patrick Delaunay <patrick.delaunay@st.com>
5 years agoclk: stm32mp1: add debug information
Patrick Delaunay [Wed, 30 Jan 2019 12:07:04 +0000 (13:07 +0100)]
clk: stm32mp1: add debug information

Add support of clk dump command and
display information during probe (under CONFIG_DISPLAY_CPUINFO).

Signed-off-by: Patrick Delaunay <patrick.delaunay@st.com>
5 years agoclk: stm32mp1: recalculate counter when switching freq
Patrick Delaunay [Wed, 30 Jan 2019 12:07:03 +0000 (13:07 +0100)]
clk: stm32mp1: recalculate counter when switching freq

Because stgen is initialized with HSI clock, we need to
recalculate the counter when changing frequency.

Signed-off-by: Lionel Debieve <lionel.debieve@st.com>
Signed-off-by: Patrick Delaunay <patrick.delaunay@st.com>
5 years agoclk: stm32mp1: correct access to RCC_OCENSETR/RCC_OCENCLRR
Patrick Delaunay [Wed, 30 Jan 2019 12:07:02 +0000 (13:07 +0100)]
clk: stm32mp1: correct access to RCC_OCENSETR/RCC_OCENCLRR

Remove unnecessary setbits on set/clear registers.
Avoid to deactivate HSI with HSE.

Signed-off-by: Patrick Delaunay <patrick.delaunay@st.com>
5 years agoclk: stm32mp1: add IPCC clock
Patrick Delaunay [Wed, 30 Jan 2019 12:07:01 +0000 (13:07 +0100)]
clk: stm32mp1: add IPCC clock

Add support for enable/disable of IPCC clock using AHB3 registers

Signed-off-by: Patrick Delaunay <patrick.delaunay@st.com>
5 years agoclk: stm32mp1: no more get ck_usbo_48m in device tree
Patrick Delaunay [Wed, 30 Jan 2019 12:07:00 +0000 (13:07 +0100)]
clk: stm32mp1: no more get ck_usbo_48m in device tree

Remove support of ck_usbo_48m clock node in device tree,
but force 48MHz frequency to prepare alignment
with kernel device tree.

Signed-off-by: Patrick Delaunay <patrick.delaunay@st.com>
5 years agodrivers: serial: dm: Enable DM_FLAG_PRE_RELOC in SBSA pl011 uart driver
Vabhav Sharma [Wed, 30 Jan 2019 11:24:00 +0000 (11:24 +0000)]
drivers: serial: dm: Enable DM_FLAG_PRE_RELOC in SBSA pl011 uart driver

The DM_FLAG_PRE_RELOC shall be enabled in SBSA PL011 uart driver
as this driver is used in NXP based SoCs

It is necessary to have Serial console running before relocation

The !CONFIG_IS_ENABLED(OF_CONTROL) [*] check is set as "workaround"
for DM problem : 4687919684e

This flag is set if board does not support device-tree and using
platform data, In DM Model either of device tree or platform data
can be used to fetch device configuration

It is possible to use SBSA UART with CONFIG_DM_SERIAL but witout
corresponding device tree description (OF_CONTROL)

Other board/SoCs have this flag set unconditionally

Signed-off-by: Vabhav Sharma <vabhav.sharma@nxp.com>
5 years agohashtable: fix environment variable corruption
Roman Kapl [Wed, 30 Jan 2019 10:39:54 +0000 (11:39 +0100)]
hashtable: fix environment variable corruption

Only first previously deleted entry was recognized, leading hsearch_r
to think that there was no previously deleted entry. It then conluded
that a free entry was found, even if there were no free entries and it
overwrote a random entry.

This patch makes sure all deleted or free entries are always found and
also introduces constants for the 0 and -1 numbers. Unit tests to excersise a
simple hash table usage and catch the corruption were added.

To trash your environment, simply run this loop:

setenv i 0
while true; do
    setenv v_$i $i
    setenv v_$i
    setexpr i $i + 1
done

Signed-off-by: Roman Kapl <rka@sysgo.com>