]> git.dujemihanovic.xyz Git - u-boot.git/log
u-boot.git
8 weeks agoarm: total_compute: Dynamically detect block device
Leo Yan [Fri, 25 Oct 2024 17:18:19 +0000 (18:18 +0100)]
arm: total_compute: Dynamically detect block device

Dynamically detect block device in the boot command, this allows to
support both MMC and virtio block devices.

Signed-off-by: Leo Yan <leo.yan@arm.com>
8 weeks agoarm: total_compute: Minor improvement for boot arguments
Boyan Karatotev [Fri, 25 Oct 2024 17:18:18 +0000 (18:18 +0100)]
arm: total_compute: Minor improvement for boot arguments

Tell the AVB command that is loading from MMC.

Signed-off-by: Boyan Karatotev <boyan.karatotev@arm.com>
Signed-off-by: Leo Yan <leo.yan@arm.com>
8 weeks agoarm: total_compute: move the boot command to an env file
Boyan Karatotev [Fri, 25 Oct 2024 17:18:17 +0000 (18:18 +0100)]
arm: total_compute: move the boot command to an env file

The boot command for Total Compute has many aspects and changes from time
to time. So move it to an .env file where it can be a proper script.

Signed-off-by: Boyan Karatotev <boyan.karatotev@arm.com>
Signed-off-by: Leo Yan <leo.yan@arm.com>
8 weeks agoarm: total_compute: Remove unused bootm_size
Boyan Karatotev [Fri, 25 Oct 2024 17:18:16 +0000 (18:18 +0100)]
arm: total_compute: Remove unused bootm_size

The whole DRAM bank is used for loading and U-boot can detect the
overlap between the kernel and initramfs. So it is safe to drop
bootm_size.

Signed-off-by: Boyan Karatotev <boyan.karatotev@arm.com>
Signed-off-by: Leo Yan <leo.yan@arm.com>
8 weeks agoarm: total_compute: Initialize environment variables
Boyan Karatotev [Fri, 25 Oct 2024 17:18:15 +0000 (18:18 +0100)]
arm: total_compute: Initialize environment variables

Initialize the environment variables 'fdt_addr_r' and 'kernel_addr_r'
during the misc init phase. The static configurations are not needed,
remove them.

Signed-off-by: Boyan Karatotev <boyan.karatotev@arm.com>
Signed-off-by: Leo Yan <leo.yan@arm.com>
8 weeks agoarm: total_compute: depend on TF-A for hardware description
Boyan Karatotev [Fri, 25 Oct 2024 17:18:14 +0000 (18:18 +0100)]
arm: total_compute: depend on TF-A for hardware description

On Total Compute, TF-A passes the info via DT binding for the hardware
description - includes the serial, memory, and arm_ffa nodes.

This commit initializes the fdt base address based on the passed the
register x1.

The similar implementation has already been done for the raspberry pi,
so borrow a lot of it.

Co-developed-by: Jackson Cooper-Driver <jackson.cooper-driver@arm.com>
Signed-off-by: Jackson Cooper-Driver <jackson.cooper-driver@arm.com>
Signed-off-by: Boyan Karatotev <boyan.karatotev@arm.com>
Signed-off-by: Leo Yan <leo.yan@arm.com>
2 months agotest: boot: fix bootflow_cmd_label for when DSA_SANDBOX is disabled
Jerome Forissier [Wed, 16 Oct 2024 09:56:26 +0000 (11:56 +0200)]
test: boot: fix bootflow_cmd_label for when DSA_SANDBOX is disabled

When DSA_SANDBOX is not set, the sandbox tests fail as follows:

 $ ./test/py/test.py --build-dir=$(pwd) -k bootdev_test_any
 [...]
 Scanning for bootflows with label '9'
 [...]
 Cannot find '9' (err=-19)

This is due to the device list containing two less entries than
expected. Therefore, look for label '7' when DSA_SANDBOX is disabled.

The actual use case is NET_LWIP=y (to be introduced in later patches)
which implies DSA_SANDBOX=n for the time being.

Signed-off-by: Jerome Forissier <jerome.forissier@linaro.org>
2 months agotest: boot: fix bootdev_test_any for when DSA_SANDBOX is disabled
Jerome Forissier [Wed, 16 Oct 2024 09:56:25 +0000 (11:56 +0200)]
test: boot: fix bootdev_test_any for when DSA_SANDBOX is disabled

When DSA_SANDBOX is not set, the sandbox tests fail as follows:

 $ ./test/py/test.py --build-dir=$(pwd) -k bootdev_test_any
 [...]
 Test: bootdev_test_any: bootdev.c
 test/boot/bootdev.c:156, bootdev_test_any(): "mmc2" = media->name: Expected "mmc2", got "mmc0"
 [...]

This is due to the device list containing two less entries than
expected. Therefore, adjust the expected index to be two less when
DSA_SANDBOX is disabled.

The actual use case is NET_LWIP=y (to be introduced in later patches)
which implies DSA_SANDBOX=n for the time being.

Signed-off-by: Jerome Forissier <jerome.forissier@linaro.org>
Reviewed-by: Simon Glass <sjg@chromium.org>
2 months agosandbox: add dummy driver ETH_SANDBOX_LWIP
Jerome Forissier [Wed, 16 Oct 2024 09:56:24 +0000 (11:56 +0200)]
sandbox: add dummy driver ETH_SANDBOX_LWIP

Introduce ETH_SANDBOX_LWIP which enables a mock driver similar to
ETH_SANDOX but without the dependencies on the legacy network stack
(NET) so that it may be enabled when the lwIP stack (NET_LWIP) is
introduced. The driver does nothing at this stage but its presence
will allow dm_test_iommu_noiommu [1] to pass.

[1] ./u-boot -T -c "ut dm dm_test_iommu_noiommu"

Signed-off-by: Jerome Forissier <jerome.forissier@linaro.org>
Acked-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>
Reviewed-by: Simon Glass <sjg@chromium.org>
2 months agoconfigs: use syntax CONFIG_FOO=n in tools-only_defconfig
Jerome Forissier [Wed, 16 Oct 2024 09:56:23 +0000 (11:56 +0200)]
configs: use syntax CONFIG_FOO=n in tools-only_defconfig

The tools-only defconfig causes troubles on MacOSX due to the default
C compiler being Clang (LLVM) rather than GCC and more specifically
due to [1]. Therefore replace "# CONFIG_FOO is not set" with the
equivalent "CONFIG_FOO=n" using the following command:

 $ sed -i -e 's/# \(CONFIG_[^ ]*\) is not set/\1=n/' \
       configs/tools-only_defconfig

This fixes the tools_only_macOS CI job on GitHub [2].

[1] https://github.com/llvm/llvm-project/issues/78778
[2] https://dev.azure.com/u-boot/u-boot/_build/results?buildId=9105&view=results

Suggested-by: Tom Rini <trini@konsulko.com>
Signed-off-by: Jerome Forissier <jerome.forissier@linaro.org>
Reviewed-by: Tom Rini <trini@konsulko.com>
2 months agoMerge commit 'f3f86fd1fe0fb288356bff78f8a6fa2edf89e3fc' as 'lib/lwip/lwip'
Tom Rini [Wed, 16 Oct 2024 14:10:14 +0000 (08:10 -0600)]
Merge commit 'f3f86fd1fe0fb288356bff78f8a6fa2edf89e3fc' as 'lib/lwip/lwip'

2 months agoSquashed 'lib/lwip/lwip/' content from commit 0a0452b2c39b
Tom Rini [Wed, 16 Oct 2024 14:10:14 +0000 (08:10 -0600)]
Squashed 'lib/lwip/lwip/' content from commit 0a0452b2c39b

git-subtree-dir: lib/lwip/lwip
git-subtree-split: 0a0452b2c39bdd91e252aef045c115f88f6ca773

2 months agoRevert "Makefile: Drop SPL_FIT_GENERATOR / SPL_FIT_SOURCE support" changes
Tom Rini [Tue, 15 Oct 2024 22:51:05 +0000 (16:51 -0600)]
Revert "Makefile: Drop SPL_FIT_GENERATOR / SPL_FIT_SOURCE support" changes

:hile we had hoped to be able to remove these options finally, it was
missed that zynq still requires these currently.

This reverts commit 5b9261fb0b1ed087387f2036d279fd3f4bb20a61 and
commit 099b6df556c95f5d06864612e9199eab7ba50ed3.

Reported-by: Jonas Karlman <jonas@kwiboo.se>
Signed-off-by: Tom Rini <trini@konsulko.com>
2 months agoMerge https://source.denx.de/u-boot/custodians/u-boot-usb
Tom Rini [Tue, 15 Oct 2024 22:40:23 +0000 (16:40 -0600)]
Merge https://source.denx.de/u-boot/custodians/u-boot-usb

2 months agoMerge patch series "Make EFI memory allocations synchronous with LMB"
Tom Rini [Tue, 15 Oct 2024 19:46:08 +0000 (13:46 -0600)]
Merge patch series "Make EFI memory allocations synchronous with LMB"

Sughosh Ganu <sughosh.ganu@linaro.org> says:

This is part two of the series to have the EFI and LMB modules have a
coherent view of memory. Part one of this goal was to change the LMB
module to have a global and persistent memory map. Those patches have
now been applied to the next branch.

These patches are changing the EFI memory allocation API's such that
they rely on the LMB module to allocate RAM memory. This fixes the
current scenario where the EFI memory module has no visibility of the
allocations/reservations made by the LMB module. One thing to note
here is that this is limited to the RAM memory region, i.e. the
EFI_CONVENTIONAL_MEMORY type. Any other memory type that is to be
added to the EFI memory map, still gets handled by the EFI memory
module.

Changes since V3:
* Add comments for the LMB_NOOVERWRITE and LMB_NONOTIFY flags
* Drop use of is_addr_in_ram() function
* Drop use of CONFIG_MEM_MAP_UPDATE_NOTIFY symbol to check if the
  notification needs to be sent.
* s/lmb_notify/lmb_should_notify
* Put a check for EFI_LOADER in the lmb_should_notify() function

Some test logs to highlight the issue that is being fixed by the series.

Without patch series
--------------------

lmb_dump_all:
 memory.count = 0x1
 memory[0] [0x40000000-0x820fffff], 0x42100000 bytes flags: none
 reserved.count = 0x3
 reserved[0] [0xe100000-0xeffffff], 0x00f00000 bytes flags: no-map
 reserved[1] [0x42000000-0x421fffff], 0x00200000 bytes flags: no-map
 reserved[2] [0x7f77da00-0x820fffff], 0x02982600 bytes flags: no-overwrite

=> efidebug memmap -- does not show regions allocated by lmb

Missing TPMv2 device for EFI_TCG_PROTOCOL
Type             Start            End              Attributes
================ ================ ================ ==========
CONVENTIONAL     0000000040000000-000000007f751000 WB
BOOT DATA        000000007f751000-000000007f756000 WB
RUNTIME DATA     000000007f756000-000000007f757000 WB|RT
BOOT DATA        000000007f757000-000000007f758000 WB
RUNTIME DATA     000000007f758000-000000007f77a000 WB|RT
BOOT DATA        000000007f77a000-000000007f781000 WB
BOOT CODE        000000007f781000-00000000807b5000 WB
RUNTIME DATA     00000000807b5000-00000000807b6000 WB|RT
BOOT CODE        00000000807b6000-00000000817c0000 WB
RUNTIME CODE     00000000817c0000-00000000817d0000 WB|RT
BOOT CODE        00000000817d0000-0000000082100000 WB
=>

Trying to allocate EFI memory with already allocated region succeeds(should fail)
---------------------------------------------------------------------------------

=> efi_mem alloc 2000 42000000
Address returned 0x42000000

=> efidebug memmap
Type             Start            End              Attributes
================ ================ ================ ==========
CONVENTIONAL     0000000040000000-0000000042000000 WB
BOOT DATA        0000000042000000-0000000042002000 WB
CONVENTIONAL     0000000042002000-000000007f751000 WB
BOOT DATA        000000007f751000-000000007f756000 WB
RUNTIME DATA     000000007f756000-000000007f757000 WB|RT
BOOT DATA        000000007f757000-000000007f758000 WB
RUNTIME DATA     000000007f758000-000000007f77a000 WB|RT
BOOT DATA        000000007f77a000-000000007f781000 WB
BOOT CODE        000000007f781000-00000000807b5000 WB
RUNTIME DATA     00000000807b5000-00000000807b6000 WB|RT
BOOT CODE        00000000807b6000-00000000817c0000 WB
RUNTIME CODE     00000000817c0000-00000000817d0000 WB|RT
BOOT CODE        00000000817d0000-0000000082100000 WB
=>

With patch series
-----------------

lmb_dump_all:
 memory.count = 0x1
 memory[0] [0x40000000-0x820fffff], 0x42100000 bytes flags: none
 reserved.count = 0x4
 reserved[0] [0xe100000-0xeffffff], 0x00f00000 bytes flags: no-map
 reserved[1] [0x42000000-0x421fffff], 0x00200000 bytes flags: no-map
 reserved[2] [0x7f74f000-0x7f77dfff], 0x0002f000 bytes flags: no-notify, no-overwrite
 reserved[3] [0x7f77ea00-0x820fffff], 0x02981600 bytes flags: no-overwrite

=> efidebug memmap
Type             Start            End              Attributes
================ ================ ================ ==========
BOOT DATA        000000000e100000-000000000f000000 WB
CONVENTIONAL     0000000040000000-0000000042000000 WB
BOOT DATA        0000000042000000-0000000042200000 WB
CONVENTIONAL     0000000042200000-000000007f74e000 WB
BOOT DATA        000000007f74e000-000000007f753000 WB
RUNTIME DATA     000000007f753000-000000007f754000 WB|RT
BOOT DATA        000000007f754000-000000007f755000 WB
RUNTIME DATA     000000007f755000-000000007f777000 WB|RT
BOOT DATA        000000007f777000-00000000807b6000 WB
RUNTIME DATA     00000000807b6000-00000000807b7000 WB|RT
BOOT DATA        00000000807b7000-00000000817c0000 WB
RUNTIME CODE     00000000817c0000-00000000817d0000 WB|RT
BOOT DATA        00000000817d0000-0000000082100000 WB

Trying to allocate EFI memory with already allocated region fails
-----------------------------------------------------------------

=> efi_mem alloc 2000 42000000
efi_allocate_pages failed 800000000000000e
=>

Trying to allocate EFI memory with non-allocated region succeeds
----------------------------------------------------------------

=> efi_mem alloc 2000 42200000
Address returned 0x42200000

=> efidebug memmap
Type             Start            End              Attributes
================ ================ ================ ==========
BOOT DATA        000000000e100000-000000000f000000 WB
CONVENTIONAL     0000000040000000-0000000042000000 WB
BOOT DATA        0000000042000000-0000000042202000 WB
CONVENTIONAL     0000000042202000-000000007f74d000 WB
BOOT DATA        000000007f74d000-000000007f752000 WB
RUNTIME DATA     000000007f752000-000000007f753000 WB|RT
BOOT DATA        000000007f753000-000000007f754000 WB
RUNTIME DATA     000000007f754000-000000007f776000 WB|RT
BOOT DATA        000000007f776000-00000000807b5000 WB
RUNTIME DATA     00000000807b5000-00000000807b6000 WB|RT
BOOT DATA        00000000807b6000-00000000817c0000 WB
RUNTIME CODE     00000000817c0000-00000000817d0000 WB|RT
BOOT DATA        00000000817d0000-0000000082100000 WB
=>

lmb_dump_all:
 memory.count = 0x1
 memory[0] [0x40000000-0x820fffff], 0x42100000 bytes flags: none
 reserved.count = 0x5
 reserved[0] [0xe100000-0xeffffff], 0x00f00000 bytes flags: no-map
 reserved[1] [0x42000000-0x421fffff], 0x00200000 bytes flags: no-map
 reserved[2] [0x42200000-0x42201fff], 0x00002000 bytes flags: no-notify, no-overwrite
 reserved[3] [0x7f74e000-0x7f77cfff], 0x0002f000 bytes flags: no-notify, no-overwrite
 reserved[4] [0x7f77da00-0x820fffff], 0x02982600 bytes flags: no-overwrite

Link: https://lore.kernel.org/r/20241015153717.401371-1-sughosh.ganu@linaro.org
2 months agolmb: replace the double-underscore with single-underscore for all functions
Sughosh Ganu [Tue, 15 Oct 2024 15:37:17 +0000 (21:07 +0530)]
lmb: replace the double-underscore with single-underscore for all functions

A bunch of static functions in the LMB module have used a
double-undersore for the function names. It was suggested to use a
single-underscore instead, as the double-underscore is usually used
by library functions. Replace the double-underscore with
single-underscore for all functions.

Signed-off-by: Sughosh Ganu <sughosh.ganu@linaro.org>
Suggested-by: Simon Glass <sjg@chromium.org>
2 months agoefi_memory: rename variable to highlight overlap with free memory
Sughosh Ganu [Tue, 15 Oct 2024 15:37:16 +0000 (21:07 +0530)]
efi_memory: rename variable to highlight overlap with free memory

The variable overlap_only_ram is used to specify that the new memory
region that is being created needs to come from the free memory pool
-- this is done by carving out the memory region from the free
memory. The name is a bit confusing though, as other allocated memory
regions, like boot-services code and data are also part of the RAM
memory. Rename the variable to overlap_conventional to highlight the
fact that it is the free/conventional memory that is being referred to
in this context.

Signed-off-by: Sughosh Ganu <sughosh.ganu@linaro.org>
Reviewed-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>
Reviewed-by: Simon Glass <sjg@chromium.org>
2 months agolmb: remove call to efi_lmb_reserve()
Sughosh Ganu [Tue, 15 Oct 2024 15:37:15 +0000 (21:07 +0530)]
lmb: remove call to efi_lmb_reserve()

The EFI memory allocations are now being done through the LMB
module. With this change, there is no need to get the EFI memory map
and set aside EFI allocated memory.

Signed-off-by: Sughosh Ganu <sughosh.ganu@linaro.org>
Reviewed-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>
Reviewed-by: Simon Glass <sjg@chromium.org>
2 months agoefi_memory: do not add RAM memory to the memory map
Sughosh Ganu [Tue, 15 Oct 2024 15:37:14 +0000 (21:07 +0530)]
efi_memory: do not add RAM memory to the memory map

The EFI_CONVENTIONAL_MEMORY type, which is the usable RAM memory is
now being managed by the LMB module. Remove the addition of this
memory type to the EFI memory map. This memory now gets added to the
EFI memory map as part of the LMB memory map update event handler.

Signed-off-by: Sughosh Ganu <sughosh.ganu@linaro.org>
Reviewed-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>
2 months agox86: e820: use the lmb API for adding RAM memory
Sughosh Ganu [Tue, 15 Oct 2024 15:37:13 +0000 (21:07 +0530)]
x86: e820: use the lmb API for adding RAM memory

The EFI_CONVENTIONAL_MEMORY type is now being managed through the LMB
module. Add a separate function, lmb_arch_add_memory() to add the RAM
memory to the LMB memory map. The efi_add_known_memory() function is
now used for adding any other memory type to the EFI memory map.

Signed-off-by: Sughosh Ganu <sughosh.ganu@linaro.org>
2 months agolayerscape: use the lmb API's to add RAM memory
Sughosh Ganu [Tue, 15 Oct 2024 15:37:12 +0000 (21:07 +0530)]
layerscape: use the lmb API's to add RAM memory

The EFI memory allocations are now being done through the LMB module,
and hence the memory map is maintained by the LMB module. Use the
lmb_arch_add_memory() API function to add the usable RAM memory to the
LMB's memory map.

Signed-off-by: Sughosh Ganu <sughosh.ganu@linaro.org>
2 months agolmb: allow for boards to specify memory map
Sughosh Ganu [Tue, 15 Oct 2024 15:37:11 +0000 (21:07 +0530)]
lmb: allow for boards to specify memory map

Some architectures have special or unique aspects which need
consideration when adding memory ranges to the list of available
memory map. Enable this config in such scenarios which allow
architectures and boards to define their own memory map.

Signed-off-by: Sughosh Ganu <sughosh.ganu@linaro.org>
2 months agostm32mp: remove efi_add_known_memory() function definition
Sughosh Ganu [Tue, 15 Oct 2024 15:37:10 +0000 (21:07 +0530)]
stm32mp: remove efi_add_known_memory() function definition

The efi_add_known_memory() function for the stm32mp platforms is adding
the EFI_CONVENTIONAL_MEMORY type. This memory is now being handled
through the LMB module -- the lmb_add_memory() adds this memory to the
memory map. Remove the definition of the now superfluous
efi_add_known_memory() function.

Signed-off-by: Sughosh Ganu <sughosh.ganu@linaro.org>
Reviewed-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>
Reviewed-by: Simon Glass <sjg@chromium.org>
2 months agoti: k3: remove efi_add_known_memory() function definition
Sughosh Ganu [Tue, 15 Oct 2024 15:37:09 +0000 (21:07 +0530)]
ti: k3: remove efi_add_known_memory() function definition

The efi_add_known_memory() function for the TI K3 platforms is adding
the EFI_CONVENTIONAL_MEMORY type. This memory is now being handled
through the LMB module -- the lmb_add_memory() adds this memory to the
memory map. Remove the definition of the now superfluous
efi_add_known_memory() function.

Signed-off-by: Sughosh Ganu <sughosh.ganu@linaro.org>
Reviewed-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>
Reviewed-by: Simon Glass <sjg@chromium.org>
2 months agoefi_memory: do not add U-Boot memory to the memory map
Sughosh Ganu [Tue, 15 Oct 2024 15:37:08 +0000 (21:07 +0530)]
efi_memory: do not add U-Boot memory to the memory map

The memory region occupied by U-Boot is reserved by LMB, and gets
added to the EFI memory map through a call from the LMB module. Remove
this superfluous addition to the EFI memory map.

Signed-off-by: Sughosh Ganu <sughosh.ganu@linaro.org>
Reviewed-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>
2 months agolmb: notify of any changes to the LMB memory map
Sughosh Ganu [Tue, 15 Oct 2024 15:37:07 +0000 (21:07 +0530)]
lmb: notify of any changes to the LMB memory map

In U-Boot, LMB and EFI are two primary modules who provide memory
allocation and reservation API's. Both these modules operate with the
same regions of memory for allocations. Use the LMB memory map update
event to notify other interested listeners about a change in it's
memory map. This can then be used by the other module to keep track of
available and used memory.

There is no need to send these notifications when the LMB module is
being unit-tested. Add a flag to the lmb structure to indicate if the
memory map is being used for tests, and suppress sending any
notifications when running these unit tests.

Signed-off-by: Sughosh Ganu <sughosh.ganu@linaro.org>
2 months agoefi: memory: use the lmb API's for allocating and freeing memory
Sughosh Ganu [Tue, 15 Oct 2024 15:37:06 +0000 (21:07 +0530)]
efi: memory: use the lmb API's for allocating and freeing memory

Use the LMB API's for allocating and freeing up memory. With this, the
LMB module becomes the common backend for managing non U-Boot image
memory that might be requested by other modules.

Signed-off-by: Sughosh Ganu <sughosh.ganu@linaro.org>
2 months agolmb: add and reserve memory above ram_top
Sughosh Ganu [Tue, 15 Oct 2024 15:37:05 +0000 (21:07 +0530)]
lmb: add and reserve memory above ram_top

U-Boot does not use memory above ram_top. However, this memory does
need to get registered as part of the memory map, so that subsystems
like EFI pass it on to the operating system as part of the EFI memory
map. Add memory above ram_top and reserve it with the LMB_NOOVERWRITE
flag so that it does not get allocated or re-used.

Signed-off-by: Sughosh Ganu <sughosh.ganu@linaro.org>
Suggested-by: Mark Kettenis <kettenis@openbsd.org>
2 months agolmb: add a flag to allow suppressing memory map change notification
Sughosh Ganu [Tue, 15 Oct 2024 15:37:04 +0000 (21:07 +0530)]
lmb: add a flag to allow suppressing memory map change notification

Add a flag LMB_NONOTIFY that can be passed to the LMB API's for
reserving memory. This will then result in no notification being sent
from the LMB module for the changes to the LMB's memory map.

While here, also add a description of the memory attributes that the
flags signify.

Signed-off-by: Sughosh Ganu <sughosh.ganu@linaro.org>
2 months agolmb: add versions of the lmb API with flags
Sughosh Ganu [Tue, 15 Oct 2024 15:37:03 +0000 (21:07 +0530)]
lmb: add versions of the lmb API with flags

The LMB module is to be used as a backend for allocating and freeing
up memory requested from other modules like EFI. These memory requests
are different from the typical LMB reservations in that memory
required by the EFI module cannot be overwritten, or re-requested. Add
versions of the LMB API functions with flags for allocating and
freeing up memory. The caller can then use these API's for specifying
the type of memory that is required. For now, these functions will be
used by the EFI memory module.

Signed-off-by: Sughosh Ganu <sughosh.ganu@linaro.org>
Reviewed-by: Simon Glass <sjg@chromium.org>
2 months agotest: Drop mention of old flags in a comment
Simon Glass [Mon, 14 Oct 2024 20:17:53 +0000 (14:17 -0600)]
test: Drop mention of old flags in a comment

A comment in test-main.c was not updated with the recent rename. Fix it.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Mattijs Korpershoek <mkorpershoek@baylibre.com>
2 months agoMakefile: Drop SPL_FIT_GENERATOR support
Marek Vasut [Fri, 4 Oct 2024 23:07:14 +0000 (01:07 +0200)]
Makefile: Drop SPL_FIT_GENERATOR support

The SPL_FIT_GENERATOR is long superseded by binman, drop SPL_FIT_GENERATOR
support as there are no more users.

Signed-off-by: Marek Vasut <marex@denx.de>
Reviewed-by: Peter Robinson <pbrobinson@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
2 months agoMakefile: Drop SPL_FIT_SOURCE support
Marek Vasut [Fri, 4 Oct 2024 23:07:13 +0000 (01:07 +0200)]
Makefile: Drop SPL_FIT_SOURCE support

The SPL_FIT_SOURCE is long superseded by SPL_FIT_GENERATOR which
is long superseded by binman, drop SPL_FIT_SOURCE support as there
are no more users.

Signed-off-by: Marek Vasut <marex@denx.de>
Reviewed-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Peter Robinson <pbrobinson@gmail.com>
2 months agotest: Fix skip check for sleep command test
Andrew Goodbody [Tue, 8 Oct 2024 12:08:13 +0000 (13:08 +0100)]
test: Fix skip check for sleep command test

When the config option CMD_MISC was renamed to CMD_SLEEP the check
in the test for the sleep command was not updated. Do that now.

Fixes: 16060854095 ("cmd: Rename CMD_MISC to CMD_SLEEP")
Signed-off-by: Andrew Goodbody <andrew.goodbody@linaro.org>
Reviewed-by: Simon Glass <sjg@chromium.org>
2 months agoCI: Do not test "sleep" in QEMU
Tom Rini [Tue, 15 Oct 2024 18:28:26 +0000 (12:28 -0600)]
CI: Do not test "sleep" in QEMU

When we have platforms being emulated by QEMU we cannot rely on the
"sleep" command running for the expected wall-clock amount of time. Even
with our current allowance for deviation from expected time, it will
still fail from time to time. Exclude the sleep test here.

Signed-off-by: Tom Rini <trini@konsulko.com>
2 months agonet: correct wget_connected debug messages
Heinrich Schuchardt [Tue, 8 Oct 2024 23:02:05 +0000 (01:02 +0200)]
net: correct wget_connected debug messages

* Remove duplicate debug message in wget_connected()
* Correct typo 'Connctd'

Signed-off-by: Heinrich Schuchardt <heinrich.schuchardt@canonical.com>
Reviewed-by: Peter Robinson <pbrobinson@gmail.com>
Reviewed-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>
2 months agonet/wget: set filesize
Heinrich Schuchardt [Wed, 9 Oct 2024 11:08:54 +0000 (13:08 +0200)]
net/wget: set filesize

After downloading a file with wget the file size may be needed in follow up
actions, e.g.

* write file to device
* calculate hash

Let wget set the environment variable filesize.

Reviewed-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>
Signed-off-by: Heinrich Schuchardt <heinrich.schuchardt@canonical.com>
2 months agoRevert "mmc: dw_mmc: Extract FIFO data transfer into a separate routine"
Jonas Karlman [Tue, 8 Oct 2024 19:18:31 +0000 (19:18 +0000)]
Revert "mmc: dw_mmc: Extract FIFO data transfer into a separate routine"

The commit 0252924ac6d4 ("mmc: dw_mmc: Extract FIFO data transfer into a
separate routine") unintentionally changed behavior of the FIFO data
transfer routine.

When data is read and size reaches 0 the original loop would wait on
DWMCI_INTMSK_DTO or timeout. The remaining size to read and buf position
is no longer tracked across dwmci_data_transfer_fifo() calls and because
of this an extra call to fifo() and dwmci_fifo_ready() may now trigger a
FIFO underflow timeout error and slows down FIFO reading.

  Buswidth = 4, clock: 50000000
  Sending CMD16
  Sending CMD17
  dwmci_fifo_ready: FIFO underflow timeout
  Sending CMD16
  Sending CMD18
  dwmci_fifo_ready: FIFO underflow timeout
  Sending CMD12
  ## Checking hash(es) for config config-1 ... OK

This reverts commit 0252924ac6d4af69061bb9589d16b30c5bdb7178 to restore
the old working behavior.

Fixes: 0252924ac6d4 ("mmc: dw_mmc: Extract FIFO data transfer into a separate routine")
Signed-off-by: Jonas Karlman <jonas@kwiboo.se>
Tested-by: Quentin Schulz <quentin.schulz@cherry.de> # RK3588 Tiger
Reviewed-by: Simon Glass <sjg@chromium.org>
2 months agoMerge patch series "test: Minor fixes to test.py"
Tom Rini [Tue, 15 Oct 2024 15:31:50 +0000 (09:31 -0600)]
Merge patch series "test: Minor fixes to test.py"

Simon Glass <sjg@chromium.org> says:

This series collects together the patches from the Labgrid series which
are not related to Labgrid, or at least can be applied independently of
using Labgrid to run the lab.

Link: https://lore.kernel.org/r/20241010002907.19383-1-sjg@chromium.org
2 months agoMerge patch series to add a "fallback" keyword to extlinux.conf parsing
Tom Rini [Tue, 15 Oct 2024 15:18:04 +0000 (09:18 -0600)]
Merge patch series to add a "fallback" keyword to extlinux.conf parsing

This series from Martyn Welch <martyn.welch@collabora.com> adds the
ability to have a "fallback" option in extlinux.conf parsing, which can
be in turn used in A/B style update mechanisms.

Link: https://lore.kernel.org/u-boot/20241009131548.929439-1-martyn.welch@collabora.com/
2 months agotest: Add tests for the bootmeth set command
Martyn Welch [Wed, 9 Oct 2024 13:15:41 +0000 (14:15 +0100)]
test: Add tests for the bootmeth set command

We have added a "set" sub command to bootmeth, add some tests to check
it's operation.

Signed-off-by: Martyn Welch <martyn.welch@collabora.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
2 months agotest: Fix mulptiplex_log typo
Simon Glass [Thu, 10 Oct 2024 00:29:05 +0000 (18:29 -0600)]
test: Fix mulptiplex_log typo

Fix a typo in a comment.

Signed-off-by: Simon Glass <sjg@chromium.org>
2 months agobootstd: Add command to enable setting of bootmeth specific properties
Martyn Welch [Wed, 9 Oct 2024 13:15:40 +0000 (14:15 +0100)]
bootstd: Add command to enable setting of bootmeth specific properties

We have previously added logic to allow a "fallback" option to be
specified in the extlinux configuration. Provide a command that allows
us to set this as the preferred default option when booting.

Combined with the bootcount functionality, this allows the "altbootcmd"
to provide a means of falling back to a previously known good state
after a failed update. For example, if "bootcmd" is set to:

    bootflow scan -lb

We would set "altbootcmd" to:

    bootmeth set extlinux fallback 1; bootflow scan -lb

Causing the boot process to boot from the fallback option.

Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Martyn Welch <martyn.welch@collabora.com>
2 months agotest: Tidy up remaining exceptions
Simon Glass [Thu, 10 Oct 2024 00:29:04 +0000 (18:29 -0600)]
test: Tidy up remaining exceptions

Use the new handle_exception() function from ConsoleBase also.

Signed-off-by: Simon Glass <sjg@chromium.org>
2 months agoboot: Add logic to enable booting from fallback option
Martyn Welch [Wed, 9 Oct 2024 13:15:39 +0000 (14:15 +0100)]
boot: Add logic to enable booting from fallback option

The "fallback" extlinux config option allows us to set an alternative
default boot option for when it has been detected that the default is
failing. Implement the logic required to boot from this option when
desired.

Signed-off-by: Martyn Welch <martyn.welch@collabora.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
2 months agotest: Detect dead connections
Simon Glass [Thu, 10 Oct 2024 00:29:03 +0000 (18:29 -0600)]
test: Detect dead connections

When the connection to a board dies, assume it is dead forever until
some user action is taken. Skip all remaining tests. This avoids CI
runs taking an hour, with hundreds of 30-second timeouts all to no
avail.

Signed-off-by: Simon Glass <sjg@chromium.org>
2 months agoboot: pxe_utils: Add fallback support
Martyn Welch [Wed, 9 Oct 2024 13:15:38 +0000 (14:15 +0100)]
boot: pxe_utils: Add fallback support

When configured correctly, we can detect when boot fails after the boot
process has been handed over to the kernel through the use of U-Boot's
bootcount support. In some instances, such as when we are performing
atomic updates via a system such as OSTree, it is desirable to provide a
fallback option so that we can return to a previous (hopefully working)
state.

Add a "fallback" option to the supported extlinux configuration options
that points to a label like "default" so that we can utilise this in
later commits.

Signed-off-by: Martyn Welch <martyn.welch@collabora.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
2 months agotest: Separate out the exception handling
Simon Glass [Thu, 10 Oct 2024 00:29:02 +0000 (18:29 -0600)]
test: Separate out the exception handling

The tests currently catch a very broad Exception in each case. This is
thrown even in the event of a coding error.

We want to handle exceptions differently depending on their severity,
so that we can avoid hour-long delays waiting for a board that is
clearly broken.

As a first step, create some new exception types, separating out those
which are simply an unexpected result from executed a command, from
those which indicate some kind of hardware failure.

Signed-off-by: Simon Glass <sjg@chromium.org>
2 months agotest: Move the receive code into a function
Simon Glass [Thu, 10 Oct 2024 00:29:01 +0000 (18:29 -0600)]
test: Move the receive code into a function

There is quite a bit of code to deal with receiving data from the target
so move it into its own receive() function.

Signed-off-by: Simon Glass <sjg@chromium.org>
2 months agotest: Create a common function to get the config
Simon Glass [Thu, 10 Oct 2024 00:29:00 +0000 (18:29 -0600)]
test: Create a common function to get the config

The settings are decoded in two places. Combine them into a new
function, before (in a future patch) expanding the number of items.

Signed-off-by: Simon Glass <sjg@chromium.org>
2 months agotest: Avoid failing skipped tests
Simon Glass [Thu, 10 Oct 2024 00:28:59 +0000 (18:28 -0600)]
test: Avoid failing skipped tests

When a test returns -EAGAIN this should not be considered a failure.
Fix what seems to be a problem case, where the pytests see a failure
when a test has merely been skipped.

We cannot squash the -EAGAIN error in ut_run_test() since the failure
count is incremented by its caller, ut_run_test_live_flat()

The specific example here is on snow, where a test is compiled into the
image but cannot run, so returns -EAGAIN to skip:

    test/py/tests/test_ut.py sssnow # ut bdinfo bdinfo_test_eth
    Test: bdinfo_test_eth: bdinfo.c
    Skipping: Console recording disabled
    test/test-main.c:486, ut_run_test_live_flat(): 0 == ut_run_test(uts,
    test, test->name): Expected 0x0 (0), got 0xfffffff5 (-11)
    Test bdinfo_test_eth failed 1 times
    Skipped: 1, Failures: 1
    snow # F+u-boot-test-reset snow snow

The fix is simply to respect the return code from ut_run_test(), so do
that.

Signed-off-by: Simon Glass <sjg@chromium.org>
2 months agotest: Use a constant for the test timeout
Simon Glass [Thu, 10 Oct 2024 00:28:58 +0000 (18:28 -0600)]
test: Use a constant for the test timeout

Declare a constant rather than open-coding the same value twice.

Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Tom Rini <trini@konsulko.com>
2 months agoMerge patch series "mtd: Make sure UBIFS does not do multi-pass page programming...
Tom Rini [Tue, 15 Oct 2024 15:03:17 +0000 (09:03 -0600)]
Merge patch series "mtd: Make sure UBIFS does not do multi-pass page programming on flashes that don't support it"

Takahiro Kuwano <Takahiro.Kuwano@infineon.com> says:

This series is equivalent to the one for Linux MTD submitted by
Pratyush Yadav.

https://patchwork.ozlabs.org/project/linux-mtd/list/?series=217759&state=*

Changes in v3:
  - Rebase

Changes in v2:
  - Fix an issue in setting macronix_octal_fixups
  - Rework fixup hooks

Takahiro Kuwano (6):
  mtd: ubi: Do not zero out EC and VID on ECC-ed NOR flashes
  mtd: spi-nor: Allow flashes to specify MTD writesize
  mtd: spi-nor: Check nor->info before setting macronix_octal_fixups
  mtd: spi-nor: Replace default_init() hook with late_init()
  mtd: spi-nor: Call spi_nor_post_sfdp_fixups() only after
    spi_nor_parse_sfdp()
  mtd: spi-nor: Set ECC unit size to MTD writesize in Infineon SEMPER
    flashes

drivers/mtd/spi/spi-nor-core.c | 88 +++++++++++++++++++++-------------
 drivers/mtd/ubi/build.c        |  4 +-
 drivers/mtd/ubi/io.c           |  9 +++-
 include/linux/mtd/spi-nor.h    |  1 +
 4 files changed, 65 insertions(+), 37 deletions(-)

Link: https://lore.kernel.org/r/cover.1728964655.git.Takahiro.Kuwano@infineon.com
2 months agomtd: spi-nor: Set ECC unit size to MTD writesize in Infineon SEMPER flashes
Takahiro Kuwano [Tue, 15 Oct 2024 04:08:36 +0000 (13:08 +0900)]
mtd: spi-nor: Set ECC unit size to MTD writesize in Infineon SEMPER flashes

The Infineon SEMPER NOR flash family uses 2-bit ECC by default with each
ECC block being 16 bytes. Under this scheme multi-pass programming to an
ECC block is not allowed. Set the writesize to make sure multi-pass
programming is not attempted on the flash.

Acked-by: Tudor Ambarus <tudor.ambarus@linaro.org>
Signed-off-by: Takahiro Kuwano <Takahiro.Kuwano@infineon.com>
2 months agomtd: spi-nor: Call spi_nor_post_sfdp_fixups() only after spi_nor_parse_sfdp()
Takahiro Kuwano [Tue, 15 Oct 2024 04:08:35 +0000 (13:08 +0900)]
mtd: spi-nor: Call spi_nor_post_sfdp_fixups() only after spi_nor_parse_sfdp()

This patch follows the upstream linux commit:
5273cc6df984("mtd: spi-nor: core: Call spi_nor_post_sfdp_fixups() only
when SFDP is defined")

spi_nor_post_sfdp_fixups() was called regardless of if
spi_nor_parse_sfdp() had been called or not. late_init() should be
instead used to initialize the parameters that are not defined in SFDP.

Ideally spi_nor_post_sfdp_fixups() is called only after successful parse
of SFDP. However, in case SFDP support is disabled by .config, that can
break current functionality. Therefore, we would call it after
spi_nor_parse_sfdp() regardless of its return value.

Acked-by: Tudor Ambarus <tudor.ambarus@linaro.org>
Signed-off-by: Takahiro Kuwano <Takahiro.Kuwano@infineon.com>
2 months agomtd: spi-nor: Replace default_init() hook with late_init()
Takahiro Kuwano [Tue, 15 Oct 2024 04:08:34 +0000 (13:08 +0900)]
mtd: spi-nor: Replace default_init() hook with late_init()

default_init() is wrong, it contributes to the maze of initializing
flash parameters. We'd like to get rid of it because the flash
parameters that it initializes are not really used at SFDP parsing time,
thus they can be initialized later.

Ideally we want SFDP to initialize all the flash parameters. If (when)
SFDP tables are wrong, we fix them with the post_sfdp/bfpt hooks, to
emphasize that SFDP is indeed wrong. When there are parameters that are
not covered by SFDP, we initialize them in late_init() - these
parameters have nothing to do with SFDP and they are not needed earlier.
With this we'll have a clearer view of who initializes what.

There are six default_init() hooks implemented just for initializing
octal_dtr_enable() and/or setup() hooks that called later on.
Just moving those to late_init() does not change functionality.

Suggested-by: Tudor Ambarus <tudor.ambarus@linaro.org>
Acked-by: Tudor Ambarus <tudor.ambarus@linaro.org>
Signed-off-by: Takahiro Kuwano <Takahiro.Kuwano@infineon.com>
2 months agomtd: spi-nor: Check nor->info before setting macronix_octal_fixups
Takahiro Kuwano [Tue, 15 Oct 2024 04:08:33 +0000 (13:08 +0900)]
mtd: spi-nor: Check nor->info before setting macronix_octal_fixups

The macronix_octal_fixups should be set only when mfr and flags match.

Fixes: df3d5f9e41 ("mtd: spi-nor: add support for Macronix Octal flash")
Acked-by: Tudor Ambarus <tudor.ambarus@linaro.org>
Signed-off-by: Takahiro Kuwano <Takahiro.Kuwano@infineon.com>
Cc: JaimeLiao <jaimeliao.tw@gmail.com>
2 months agomtd: spi-nor: Allow flashes to specify MTD writesize
Takahiro Kuwano [Tue, 15 Oct 2024 04:08:32 +0000 (13:08 +0900)]
mtd: spi-nor: Allow flashes to specify MTD writesize

Some flashes like the Infineon SEMPER NOR flash family use ECC. Under
this ECC scheme, multi-pass writes to an ECC block is not allowed.
In other words, once data is programmed to an ECC block, it can't be
programmed again without erasing it first.

Upper layers like file systems need to be given this information so they
do not cause error conditions on the flash by attempting multi-pass
programming. This can be done by setting 'writesize' in 'struct
mtd_info'.

Set the default to 1 but allow flashes to modify it in fixup hooks. If
more flashes show up with this constraint in the future it might be
worth it to add it to 'struct flash_info', but for now increasing its
size is not worth it.

This patch replicates the following upstream linux commit:
afd473e85827 ("mtd: spi-nor: core: Allow flashes to specify MTD writesize")

Acked-by: Tudor Ambarus <tudor.ambarus@linaro.org>
Signed-off-by: Takahiro Kuwano <Takahiro.Kuwano@infineon.com>
2 months agomtd: ubi: Do not zero out EC and VID on ECC-ed NOR flashes
Takahiro Kuwano [Tue, 15 Oct 2024 04:08:31 +0000 (13:08 +0900)]
mtd: ubi: Do not zero out EC and VID on ECC-ed NOR flashes

For NOR flashes EC and VID are zeroed out before an erase is issued to
make sure UBI does not mistakenly treat the PEB as used and associate it
with an LEB.

But on some flashes, like the Infineon Semper NOR flash family,
multi-pass page programming is not allowed on the default ECC scheme.
This means zeroing out these magic numbers will result in the flash
throwing a page programming error.

Do not zero out EC and VID for such flashes. A writesize > 1 is an
indication of an ECC-ed flash.

This patch replicates the following upstream linux commit:
f669e74be820 ("ubi: Do not zero out EC and VID on ECC-ed NOR flashes")

Acked-by: Tudor Ambarus <tudor.ambarus@linaro.org>
Acked-by: Pratyush Yadav <pratyush@kernel.org>
Signed-off-by: Takahiro Kuwano <Takahiro.Kuwano@infineon.com>
Reviewed-by: Tudor Ambarus <tudor.ambarus@microchip.com>
2 months agodts: Deduplicate dtbs target
Marek Vasut [Fri, 4 Oct 2024 00:10:42 +0000 (02:10 +0200)]
dts: Deduplicate dtbs target

The dtbs: target is almost identical in all architecture Makefiles.
All architecture Makefiles include scripts/Makefile.dts . Deduplicate
the dtbs: target into scripts/Makefile.dts . No functional change.

Reviewed-by: Sumit Garg <sumit.garg@linaro.org>
Signed-off-by: Marek Vasut <marex@denx.de>
Reviewed-by: Simon Glass <sjg@chromium.org>
Tested-by: Caleb Connolly <caleb.connolly@linaro.org> #qcom, OF_UPSTREAM
2 months agoMerge patch series "Integrate MbedTLS v3.6 LTS with U-Boot"
Tom Rini [Mon, 14 Oct 2024 19:34:06 +0000 (13:34 -0600)]
Merge patch series "Integrate MbedTLS v3.6 LTS with U-Boot"

Raymond Mao <raymond.mao@linaro.org> says:
Integrate MbedTLS v3.6 LTS (currently v3.6.0) with U-Boot.

Motivations:
------------

1. MbedTLS is well maintained with LTS versions.
2. LWIP is integrated with MbedTLS and easily to enable HTTPS.
3. MbedTLS recently switched license back to GPLv2.

Prerequisite:
-------------

This patch series requires mbedtls git repo to be added as a
subtree to the main U-Boot repo via:
    $ git subtree add --prefix lib/mbedtls/external/mbedtls \
          https://github.com/Mbed-TLS/mbedtls.git \
          v3.6.0 --squash
Moreover, due to the Windows-style files from mbedtls git repo,
we need to convert the CRLF endings to LF and do a commit manually:
    $ git add --renormalize .
    $ git commit

New Kconfig options:
--------------------

`MBEDTLS_LIB` is for MbedTLS general switch.
`MBEDTLS_LIB_CRYPTO` is for replacing original digest and crypto libs with
MbedTLS.
`MBEDTLS_LIB_CRYPTO_ALT` is for using original U-Boot crypto libs as
MbedTLS crypto alternatives.
`MBEDTLS_LIB_X509` is for replacing original X509, PKCS7, MSCode, ASN1,
and Pubkey parser with MbedTLS.
By default `MBEDTLS_LIB_CRYPTO_ALT` and `MBEDTLS_LIB_X509` are selected
when `MBEDTLS_LIB` is enabled.
`LEGACY_CRYPTO` is introduced as a main switch for legacy crypto library.
`LEGACY_CRYPTO_BASIC` is for the basic crypto functionalities and
`LEGACY_CRYPTO_CERT` is for the certificate related functionalities.
For each of the algorithm, a pair of `<alg>_LEGACY` and `<alg>_MBEDTLS`
Kconfig options are introduced. Meanwhile, `SPL_` Kconfig options are
introduced.

In this patch set, MBEDTLS_LIB, MBEDTLS_LIB_CRYPTO and MBEDTLS_LIB_X509
are by default enabled in qemu_arm64_defconfig and sandbox_defconfig
for testing purpose.

Patches for external MbedTLS project:
-------------------------------------

Since U-Boot uses Microsoft Authentication Code to verify PE/COFFs
executables which is not supported by MbedTLS at the moment,
addtional patches for MbedTLS are created to adapt with the EFI loader:
1. Decoding of Microsoft Authentication Code.
2. Decoding of PKCS#9 Authenticate Attributes.
3. Extending MbedTLS PKCS#7 lib to support multiple signer's certificates.
4. MbedTLS native test suites for PKCS#7 signer's info.

All above 4 patches (tagged with `mbedtls/external`) are submitted to
MbedTLS project and being reviewed, eventually they should be part of
MbedTLS LTS release.
But before that, please merge them into U-Boot, otherwise the building
will be broken when MBEDTLS_LIB_X509 is enabled.

See below PR link for the reference:
https://github.com/Mbed-TLS/mbedtls/pull/9001

Miscellaneous:
--------------

Optimized MbedTLS library size by tailoring the config file
and disabling all unnecessary features for EFI loader.
From v2, original libs (rsa, asn1_decoder, rsa_helper, md5, sha1, sha256,
sha512) are completely replaced when MbedTLS is enabled.
From v3, the size-growth is slightly reduced by refactoring Hash functions.
From v6, smaller implementations for SHA256 and SHA512 are enabled and
target size reduce significantly.
Target(QEMU arm64) size-growth when enabling MbedTLS:
v1: 6.03%
v2: 4.66%
v3 - v5: 4.55%
v6: 2.90%

Tests done:
-----------

EFI Secure Boot test (EFI variables loading and verifying, EFI signed image
verifying and booting) via U-Boot console.
EFI Secure Boot and Capsule sandbox test passed.

Known issues:
-------------

None.

Link: https://lore.kernel.org/u-boot/20241003215112.3103601-1-raymond.mao@linaro.org/
2 months agoconfigs: enable MbedTLS as default setting
Raymond Mao [Thu, 3 Oct 2024 21:50:40 +0000 (14:50 -0700)]
configs: enable MbedTLS as default setting

Enable MbedTLS as default setting for qemu arm64 and sandbox.

Signed-off-by: Raymond Mao <raymond.mao@linaro.org>
Reviewed-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>
2 months agotest: Remove ASN1 library test
Raymond Mao [Thu, 3 Oct 2024 21:50:39 +0000 (14:50 -0700)]
test: Remove ASN1 library test

With MBEDTLS_LIB_X509 enabled, we don't build the original ASN1 lib,
So remove it from test.

Signed-off-by: Raymond Mao <raymond.mao@linaro.org>
Reviewed-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>
2 months agoasn1_decoder: add build options for ASN1 decoder
Raymond Mao [Thu, 3 Oct 2024 21:50:38 +0000 (14:50 -0700)]
asn1_decoder: add build options for ASN1 decoder

When building with MbedTLS, we are using MbedTLS to decode ASN1 data
for x509, pkcs7 and mscode.
Introduce _LEGACY and _MBEDTLS kconfigs for ASN1 decoder legacy and
MbedTLS implementations respectively.

Signed-off-by: Raymond Mao <raymond.mao@linaro.org>
Reviewed-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>
2 months agolib/rypto: Adapt rsa_helper to MbedTLS
Raymond Mao [Thu, 3 Oct 2024 21:50:37 +0000 (14:50 -0700)]
lib/rypto: Adapt rsa_helper to MbedTLS

Previous patch has introduced MbedTLS porting layer for RSA helper,
here to adjust the makefile accordingly.

Signed-off-by: Raymond Mao <raymond.mao@linaro.org>
Reviewed-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>
2 months agombedtls: add RSA helper layer on MbedTLS
Raymond Mao [Thu, 3 Oct 2024 21:50:36 +0000 (14:50 -0700)]
mbedtls: add RSA helper layer on MbedTLS

Add RSA helper layer on top on MbedTLS PK and RSA library.
Introduce _LEGACY and _MBEDTLS kconfigs for RSA helper legacy and
MbedTLS implementations respectively.

Signed-off-by: Raymond Mao <raymond.mao@linaro.org>
Acked-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>
2 months agolib/crypto: Adapt mscode_parser to MbedTLS
Raymond Mao [Thu, 3 Oct 2024 21:50:35 +0000 (14:50 -0700)]
lib/crypto: Adapt mscode_parser to MbedTLS

Previous patch has introduced MbedTLS porting layer for mscode parser,
here to adjust the header and makefiles accordingly.
Adding _LEGACY Kconfig for legacy mscode implementation.

Signed-off-by: Raymond Mao <raymond.mao@linaro.org>
2 months agombedtls: add MSCode parser porting layer
Raymond Mao [Thu, 3 Oct 2024 21:50:34 +0000 (14:50 -0700)]
mbedtls: add MSCode parser porting layer

Add porting layer for MSCode on top of MbedTLS ASN1 library.
Introduce _MBEDTLS kconfigs for MSCode MbedTLS implementation.

Signed-off-by: Raymond Mao <raymond.mao@linaro.org>
Acked-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>
2 months agolib/crypto: Adapt PKCS7 parser to MbedTLS
Raymond Mao [Thu, 3 Oct 2024 21:50:33 +0000 (14:50 -0700)]
lib/crypto: Adapt PKCS7 parser to MbedTLS

Previous patch has introduced MbedTLS porting layer for PKCS7 parser,
here to adjust the header and makefiles accordingly.

Signed-off-by: Raymond Mao <raymond.mao@linaro.org>
2 months agombedtls: add PKCS7 parser porting layer
Raymond Mao [Thu, 3 Oct 2024 21:50:32 +0000 (14:50 -0700)]
mbedtls: add PKCS7 parser porting layer

Add porting layer for PKCS7 parser on top of MbedTLS PKCS7 library.
Introduce _LEGACY and _MBEDTLS kconfigs for PKCS7 parser legacy and
MbedTLS implementations respectively.

Signed-off-by: Raymond Mao <raymond.mao@linaro.org>
Reviewed-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>
2 months agolib/crypto: Adapt x509_cert_parser to MbedTLS
Raymond Mao [Thu, 3 Oct 2024 21:50:31 +0000 (14:50 -0700)]
lib/crypto: Adapt x509_cert_parser to MbedTLS

Previous patch has introduced MbedTLS porting layer for x509 cert parser,
here to adjust the header and makefiles accordingly.

Signed-off-by: Raymond Mao <raymond.mao@linaro.org>
2 months agombedtls: add X509 cert parser porting layer
Raymond Mao [Thu, 3 Oct 2024 21:50:30 +0000 (14:50 -0700)]
mbedtls: add X509 cert parser porting layer

Add porting layer for X509 cert parser on top of MbedTLS X509
library.
Introduce _LEGACY and _MBEDTLS kconfigs for X509 cert parser legacy
and MbedTLS implementations respectively.

Signed-off-by: Raymond Mao <raymond.mao@linaro.org>
2 months agolib/crypto: Adapt public_key header with MbedTLS
Raymond Mao [Thu, 3 Oct 2024 21:50:29 +0000 (14:50 -0700)]
lib/crypto: Adapt public_key header with MbedTLS

Previous patch has introduced MbedTLS porting layer for public key,
here to adjust the header and makefiles accordingly.

Signed-off-by: Raymond Mao <raymond.mao@linaro.org>
Reviewed-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>
2 months agombedtls: add public key porting layer
Raymond Mao [Thu, 3 Oct 2024 21:50:28 +0000 (14:50 -0700)]
mbedtls: add public key porting layer

Add porting layer for public key on top of MbedTLS X509 library.
Introduce _LEGACY and _MBEDTLS kconfigs for public key legacy and
MbedTLS implementations respectively.

Signed-off-by: Raymond Mao <raymond.mao@linaro.org>
Reviewed-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>
2 months agopkcs7: move common functions to PKCS7 helper
Raymond Mao [Thu, 3 Oct 2024 21:50:27 +0000 (14:50 -0700)]
pkcs7: move common functions to PKCS7 helper

Move pkcs7_get_content_data as a helper function that can be
shared by legacy crypto lib and MbedTLS implementation.

Signed-off-by: Raymond Mao <raymond.mao@linaro.org>
Reviewed-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>
2 months agox509: move common functions to x509 helper
Raymond Mao [Thu, 3 Oct 2024 21:50:26 +0000 (14:50 -0700)]
x509: move common functions to x509 helper

Move x509_check_for_self_signed as a common helper function
that can be shared by legacy crypto lib and MbedTLS implementation.

Signed-off-by: Raymond Mao <raymond.mao@linaro.org>
Reviewed-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>
2 months agopublic_key: move common functions to public key helper
Raymond Mao [Thu, 3 Oct 2024 21:50:25 +0000 (14:50 -0700)]
public_key: move common functions to public key helper

Move public_key_free and public_key_signature_free as helper
functions that can be shared by legacy crypto lib and MbedTLS
implementation.

Signed-off-by: Raymond Mao <raymond.mao@linaro.org>
Reviewed-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>
2 months agombedtls/external: update MbedTLS PKCS7 test suites
Raymond Mao [Thu, 3 Oct 2024 21:50:24 +0000 (14:50 -0700)]
mbedtls/external: update MbedTLS PKCS7 test suites

Update the PKCS7 test suites for multiple certs.

The PR for this patch is at:
https://github.com/Mbed-TLS/mbedtls/pull/9001

For enabling EFI loader PKCS7 features with MbedTLS build,
we need this patch on top of MbedTLS v3.6.0 before it is merged into
the next MbedTLS LTS release.

Signed-off-by: Raymond Mao <raymond.mao@linaro.org>
Acked-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>
2 months agombedtls/external: support decoding multiple signer's cert
Raymond Mao [Thu, 3 Oct 2024 21:50:23 +0000 (14:50 -0700)]
mbedtls/external: support decoding multiple signer's cert

Support decoding multiple signer's cert in the signed data within
a PKCS7 message.

The PR for this patch is at:
https://github.com/Mbed-TLS/mbedtls/pull/9001

For enabling EFI loader PKCS7 features with MbedTLS build,
we need this patch on top of MbedTLS v3.6.0 before it is merged into
the next MbedTLS LTS release.

Signed-off-by: Raymond Mao <raymond.mao@linaro.org>
2 months agombedtls/external: support PKCS9 Authenticate Attributes
Raymond Mao [Thu, 3 Oct 2024 21:50:22 +0000 (14:50 -0700)]
mbedtls/external: support PKCS9 Authenticate Attributes

Populate PKCS9 Authenticate Attributes from signer info if it exists
in a PKCS7 message.
Add OIDs for describing objects using for Authenticate Attributes.

The PR for this patch is at:
https://github.com/Mbed-TLS/mbedtls/pull/9001

For enabling EFI loader PKCS7 features with MbedTLS build,
we need this patch on top of MbedTLS v3.6.0 before it is merged into
the next MbedTLS LTS release.

Signed-off-by: Raymond Mao <raymond.mao@linaro.org>
Acked-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>
2 months agombedtls/external: support Microsoft Authentication Code
Raymond Mao [Thu, 3 Oct 2024 21:50:21 +0000 (14:50 -0700)]
mbedtls/external: support Microsoft Authentication Code

Populate Microsoft Authentication Code from the content data
into PKCS7 decoding context if it exists in a PKCS7 message.
Add OIDs for describing objects using for Microsoft Authentication
Code.

The PR for this patch is at:
https://github.com/Mbed-TLS/mbedtls/pull/9001

For enabling EFI loader PKCS7 features with MbedTLS build,
we need this patch on top of MbedTLS v3.6.0 before it is merged into
the next MbedTLS LTS release.

Signed-off-by: Raymond Mao <raymond.mao@linaro.org>
Acked-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>
2 months agombedtls: Enable smaller implementation for SHA256/512
Raymond Mao [Thu, 3 Oct 2024 21:50:20 +0000 (14:50 -0700)]
mbedtls: Enable smaller implementation for SHA256/512

Smaller implementation for SHA256 and SHA512 helps to reduce the
ROM footprint though it has a certain impact on performance.
As a trade-off, enable it as a default config when MbedTLS is
enabled can reduce the target size significantly with acceptable
performance loss.

Signed-off-by: Raymond Mao <raymond.mao@linaro.org>
Reviewed-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>
2 months agombedtls: add digest shim layer for MbedTLS
Raymond Mao [Thu, 3 Oct 2024 21:50:19 +0000 (14:50 -0700)]
mbedtls: add digest shim layer for MbedTLS

Implement digest shim layer on top of MbedTLS crypto library.
Introduce <alg>_MBEDTLS kconfig for MbedTLS crypto implementations.

Signed-off-by: Raymond Mao <raymond.mao@linaro.org>
2 months agosha1: Remove sha1 non-watchdog API
Raymond Mao [Thu, 3 Oct 2024 21:50:18 +0000 (14:50 -0700)]
sha1: Remove sha1 non-watchdog API

We don't need an API specially for non-watchdog since sha1_csum_wd
supports it by disabling CONFIG_HW_WATCHDOG and CONFIG_WATCHDOG.
Set 0x10000 as default chunk size for SHA1.

Signed-off-by: Raymond Mao <raymond.mao@linaro.org>
Reviewed-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>
2 months agomd5: Remove md5 non-watchdog API
Raymond Mao [Thu, 3 Oct 2024 21:50:17 +0000 (14:50 -0700)]
md5: Remove md5 non-watchdog API

We don't need an API specially for non-watchdog since md5_wd supports
it by disabling CONFIG_HW_WATCHDOG and CONFIG_WATCHDOG.
Set 0x10000 as default chunk size for MD5.

Signed-off-by: Raymond Mao <raymond.mao@linaro.org>
Reviewed-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>
Reviewed-by: Michal Simek <michal.simek@amd.com>
2 months agolib: Adapt digest header files to MbedTLS
Raymond Mao [Thu, 3 Oct 2024 21:50:16 +0000 (14:50 -0700)]
lib: Adapt digest header files to MbedTLS

Adapt digest header files to support both original libs and MbedTLS
by switching on/off MBEDTLS_LIB_CRYPTO.
Introduce <alg>_LEGACY kconfig for legacy hash implementations.
sha256.o should depend on SHA256 kconfig only but not SUPPORT_EMMC_RPMB,
SHA256 should be selected when SUPPORT_EMMC_RPMB is enabled instead.

`IS_ENABLED` or `CONFIG_IS_ENABLED` is not applicable here, since
including <linux/kconfig.h> causes undefined reference on schedule()
with sandbox build, as <linux/kconfig.h> includes <generated/autoconf.h>
which enables `CONFIG_HW_WATCHDOG` and `CONFIG_WATCHDOG` but no schedule()
are defined in sandbox build,
Thus we use `#if defined(CONFIG_MBEDTLS_LIB_CRYPTO)` instead.

Signed-off-by: Raymond Mao <raymond.mao@linaro.org>
Reviewed-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>
2 months agombedtls: add mbedtls into the build system
Raymond Mao [Thu, 3 Oct 2024 21:50:15 +0000 (14:50 -0700)]
mbedtls: add mbedtls into the build system

Port mbedtls with adapted libc header files.
Add mbedtls default config header file.
Optimize mbedtls default config by disabling unused features to
reduce the target size.
Add mbedtls kbuild makefile.
Add Kconfig skeleton and config submenu entry for selecting
crypto libraries between mbedtls and legacy ones.
Add the mbedtls include directories into the build system.
Port u-boot hash functions as MbedTLS crypto alternatives and set
it as default.

Subsequent patches will separate those Kconfigs into pairs of
_LEGACY and _MBEDTLS for controlling the implementations of legacy
crypto libraries and MbedTLS ones respectively.

The motivation of moving and adapting *INT* macros from kernel.h
to limits.h is to fulfill the MbedTLS building requirement.
The conditional compilation statements in MbedTLS expects the
*INT* macros as constant expressions, thus expressions like
`((int)(~0U >> 1))` will not work.

Prerequisite
------------

This patch series requires mbedtls git repo to be added as a
subtree to the main U-Boot repo via:

$ git subtree add --prefix lib/mbedtls/external/mbedtls \
      https://github.com/Mbed-TLS/mbedtls.git \
      v3.6.0 --squash

Moreover, due to the Windows-style files from mbedtls git repo,
we need to convert the CRLF endings to LF and do a commit manually:

$ git add --renormalize .
$ git commit

Signed-off-by: Raymond Mao <raymond.mao@linaro.org>
2 months agoCI: Exclude MbedTLS subtree for CONFIG checks
Raymond Mao [Thu, 3 Oct 2024 21:50:14 +0000 (14:50 -0700)]
CI: Exclude MbedTLS subtree for CONFIG checks

Since MbedTLS is an external repo with its own coding style,
exclude it from Azure and gitlab CI CONFIG checks.

Signed-off-by: Raymond Mao <raymond.mao@linaro.org>
Reviewed-by: Tom Rini <trini@konsulko.com>
Reviewed-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>
2 months agousb: xhci: Define 'XHCI_MAX_HALT_USEC' macro only once
Bhupesh Sharma [Wed, 23 Aug 2023 15:41:23 +0000 (21:11 +0530)]
usb: xhci: Define 'XHCI_MAX_HALT_USEC' macro only once

Right now xhci header file defines XHCI_MAX_HALT_USEC macro
twice. Fix the same.

Cc: Bin Meng <bmeng.cn@gmail.com>
Signed-off-by: Bhupesh Sharma <bhupesh.sharma@linaro.org>
Reviewed-by: Marek Vasut <marex@denx.de>
2 months agousb: xhci: pet watchdog during transfers
Godfrey Mwangi [Fri, 4 Aug 2023 19:00:39 +0000 (12:00 -0700)]
usb: xhci: pet watchdog during transfers

On some platforms with low USB throughput, tranfers
of huge files take a long time and watchdog timer can
expire resulting in hardware reset. Avoid this by
petting the watchdog as long as we have pending transfers.

Signed-off-by: Godfrey Mwangi <godmwan@microsoft.com>
Reviewed-by: Marek Vasut <marex@denx.de>
2 months agoMerge tag 'u-boot-ufs-next-20241014' of https://source.denx.de/u-boot/custodians...
Tom Rini [Mon, 14 Oct 2024 14:15:15 +0000 (08:15 -0600)]
Merge tag 'u-boot-ufs-next-20241014' of https://source.denx.de/u-boot/custodians/u-boot-ufs

- Set of fixes/updated on UFS core
 - Fix dcache/flush
 - Update & Sync Quirks
 - Backport fixes from Linux
 - Add missing memory barriers
 - Remove link_startup_again logic
 - Add Neil to UFS maintainers
- Add UFS DesignWare Controller for AMD vers2 platforms
- Add UFS Qualcomm controller driver

2 months agoMerge tag 'u-boot-amlogic-next-20241014' of https://source.denx.de/u-boot/custodians...
Tom Rini [Mon, 14 Oct 2024 14:14:46 +0000 (08:14 -0600)]
Merge tag 'u-boot-amlogic-next-20241014' of https://source.denx.de/u-boot/custodians/u-boot-amlogic

- Add Libre Computer boards into proper libre-computer board directory
- Add new Boards:
 - Libre Computer aml-s905d3-cc
 - Libre Computer aml-a311d-cc
- Add capsule update to libretech-ac and the new boards since they have an onboard SPI nor flash
- Fix HDMI support after sync to v6.11 and regulator enable from Marek
- Fix khadas-vim3 android config for android-mainline kernel
- Disable meson64 boot targets when configs are not eavailable

2 months agoMerge branch 'u-boot-nand-20241012' of https://source.denx.de/u-boot/custodians/u...
Tom Rini [Mon, 14 Oct 2024 14:13:23 +0000 (08:13 -0600)]
Merge branch 'u-boot-nand-20241012' of https://source.denx.de/u-boot/custodians/u-boot-nand-flash

This merge request add support for BCMBCA raw nand driver for bcm96846
board that switch using OF_UPSTREAM and allow use onfi ecc params when
they are available in the atmel nand controller

The patches pass the pipeline CI:
https://source.denx.de/u-boot/custodians/u-boot-nand-flash/-/pipelines/22638

2 months agoconfigs: meson64: Ensure commands for boot targets are available
Callum Parsey [Sat, 12 Oct 2024 08:03:51 +0000 (08:03 +0000)]
configs: meson64: Ensure commands for boot targets are available

When populating BOOT_TARGET_DEVICES on Amlogic Meson platforms, verify
that the shell commands for accessing the relevant media have been
enabled in the build configuration.

This was already being done for USB, NVME and SCSI devices, but not
for MMC devices or network boot. This meant that configuring U-Boot
without the networking or MMC subsystems/drivers would cause an error
to be thrown during preprocessing.

Signed-off-by: Callum Parsey <callum@neoninteger.au>
Reviewed-by: Neil Armstrong <neil.armstrong@linaro.org>
Link: https://lore.kernel.org/r/20241012080340.2596-1-callum@neoninteger.au
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
2 months agoconfigs: meson64_android: Remove dtbo support
Mattijs Korpershoek [Fri, 11 Oct 2024 15:51:03 +0000 (17:51 +0200)]
configs: meson64_android: Remove dtbo support

The device tree overlays are not part of the
Android Common Kernel code [1]

Historically, they existed to enable some Android specific
features but they have been removed.

Remove the dtbos logic since these files do no longer exist.

[1] https://android.googlesource.com/kernel/common/+/ac2ea0bb91a3b5f5e71eba66c441e58beca6767e/arch/arm64/boot/dts/amlogic/

Signed-off-by: Mattijs Korpershoek <mkorpershoek@baylibre.com>
Reviewed-by: Guillaume La Roque <glaroque@baylibre.com>
Reviewed-by: Neil Armstrong <neil.armstrong@linaro.org>
Link: https://lore.kernel.org/r/20241011-yukawa-aosp-parts-v1-2-7d74aa13ff8e@baylibre.com
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
2 months agoconfigs: khadas-vim3{l}: Increase boot/recovery partition size
Mattijs Korpershoek [Fri, 11 Oct 2024 15:51:02 +0000 (17:51 +0200)]
configs: khadas-vim3{l}: Increase boot/recovery partition size

The Android Common Kernel is over 34MiB now [1]. This does not
include any kernel modules nor the Android ramdisk.

To make sure we are future proof, increase both recovery
and boot partitions to 64 MiB.

[1] https://source.android.com/docs/core/architecture/kernel/gki-android15-6_6-release-builds

Signed-off-by: Mattijs Korpershoek <mkorpershoek@baylibre.com>
Reviewed-by: Guillaume La Roque <glaroque@baylibre.com>
Reviewed-by: Neil Armstrong <neil.armstrong@linaro.org>
Link: https://lore.kernel.org/r/20241011-yukawa-aosp-parts-v1-1-7d74aa13ff8e@baylibre.com
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
2 months agopower/domain: meson-ee-pwrc: make sure to not enable a domain twice
Neil Armstrong [Wed, 9 Oct 2024 09:15:21 +0000 (11:15 +0200)]
power/domain: meson-ee-pwrc: make sure to not enable a domain twice

The upstream Device Tree for GXBB/GXL/G12A was updated with VPU domain
shared between the VPU and HDMI node, causing a double enable.

Simply store the enable state and avoid enabling twice, fixing
HDMI output on all platforms.

Link: https://lore.kernel.org/r/20241009-u-boot-topic-fix-hdmi-v1-2-2479cd90c4ea@linaro.org
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
2 months agoclk: meson: gxbb: add HDMI clocks
Neil Armstrong [Wed, 9 Oct 2024 09:15:20 +0000 (11:15 +0200)]
clk: meson: gxbb: add HDMI clocks

Align with g12a driver to handle the CLKID_HDMI, CLKID_HDMI_SEL
and CLKID_HDMI_DIV clocks since they were added to the upstream
GXBB/GXL Devicetree on v6.11 with [1]

[1] https://lore.kernel.org/all/20240626152733.1350376-1-jbrunet@baylibre.com/

Link: https://lore.kernel.org/r/20241009-u-boot-topic-fix-hdmi-v1-1-2479cd90c4ea@linaro.org
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
2 months agovideo: meson: dw-hdmi: do not fail probe if HDMI regulator is already enabled
Neil Armstrong [Wed, 9 Oct 2024 08:01:56 +0000 (10:01 +0200)]
video: meson: dw-hdmi: do not fail probe if HDMI regulator is already enabled

If the regulator is already enabled, this happens if the regulator
is set in regulator-always-on, regulator_set_enable() return -EALREADY.

Ignore the -EALREADY return since it's not an error.

Suggested-by: Jonas Karlman <jonas@kwiboo.se>
Link: https://lore.kernel.org/r/20241009-u-boot-video-fix-hdmi-supply-already-on-v2-1-4478a28d1f49@linaro.org
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
2 months agodts: meson-g12-common-u-boot: do not disable canvas
Neil Armstrong [Tue, 8 Oct 2024 16:23:42 +0000 (18:23 +0200)]
dts: meson-g12-common-u-boot: do not disable canvas

We were disabling canvas, which causes meson vpu probe failure,
just stop and leave canvas alone.

Fixes: ce9fa7bffc5 ("ARM: dts: meson-g12a: add U-Boot specific DT for graphics")
Link: https://lore.kernel.org/r/20241008-u-boot-topic-g12-do-not-disable-canvas-v1-1-b89659370012@linaro.org
Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>