The UEFI spec mandates that unaligned memory access should be enabled if
supported by the CPU architecture.
This patch implements the function unaligned_access() to set the enable
unaligned data support flag and to clear the aligned flag in the system
control register (SCTLR). It is called when UEFI related commands like
bootefi are invoked.
efi_loader: restrict EFI_LOADER to armv7 and armv8 on ARM
fatload USB was reported to fail on the Sheevaplug. Debugging showed that
this was caused by an incorrect unaligned write to memory in
path_to_uefi().
UEFI on ARM requires that unaligned memory access is enabled.
* ARMv5 does not support unaligned access at all.
* ARMv6 supports unaligned access when we clear the A flag and set the
U flag.
* On ARMv7 unaligned access is possible when clearing the aligned flag,
which we do in function allow_unaligned() (arch/arm/cpu/armv7/sctlr.S).
For none of the other cpus in arch/arm/cpu/ we have implemented a
similar function.
* ARMv8 allows unaligned access.
Let EFI_LOADER on ARM depend on SYS_CPU=armv7 or SYS_CPU=armv8.
Once we have implemented allow_unaligned() for other ARM CPUs we can add
these to Kconfig.
Reported-by: Gray Remlin <gryrmln@gmail.com> Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Tom Rini [Mon, 18 Nov 2019 15:48:29 +0000 (10:48 -0500)]
Merge tag 'video-for-v2020.01-rc2' of https://gitlab.denx.de/u-boot/custodians/u-boot-video
- fix i.MX6ULL evk black screen observed while reboot stress tests
- remove "synopsys,dw-mipi-dsi" compatible to reduce the device tree
differences between Linux and U-Boot for stm32mp1 platform
Tom Rini [Mon, 18 Nov 2019 02:15:57 +0000 (21:15 -0500)]
Merge tag 'u-boot-rockchip-20191118' of https://gitlab.denx.de/u-boot/custodians/u-boot-rockchip
- Add support for rockchip SoC: PX30, RK3308
- Add and migrate to use common dram driver: PX30, RK3328, RK3399
- Add rk3399 board Tinker-s support
- Board config update for Rock960, Rockpro64
Thomas Hebb [Fri, 15 Nov 2019 16:48:57 +0000 (08:48 -0800)]
rockchip: imply instead of selecting SPL_SYS_MALLOC_SIMPLE
We shouldn't force which allocator the SPL uses, since there's no
platform requirement for one over the other: in fact, we currently allow
selection of the TPL allocator but not the SPL one!
Signed-off-by: Thomas Hebb <tommyhebb@gmail.com> Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
Thomas Hebb [Fri, 15 Nov 2019 16:48:56 +0000 (08:48 -0800)]
rockchip: allow DRAM init in SPL
The common SPL removed SoC-specific code for RK3399's SPL and in the
process caused the previously-unconditional DRAM initialization in
board_init_f() to only happen when compiling a configuration that does not
support TPL, meaning DRAM never gets initialized if TPL is supported but
disabled.
Fix this by omitting the DRAM init in SPL only when we are configured to
also build a TPL. This fixes custom configurations that have disabled
TPL, and it should also unbreak the "ficus-rk3399", "rock960-rk3399",
and "chromebook_bob" defconfigs, although since I don't have any of
those devices I can't confirm they're broken now.
Fixes: b7abef2ecbcc ("rockchip: rk3399: Migrate to use common spl board file") Signed-off-by: Thomas Hebb <tommyhebb@gmail.com> Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
Thomas Hebb [Fri, 15 Nov 2019 16:48:55 +0000 (08:48 -0800)]
rockchip: SPL: fix ordering of DRAM init
The common SPL code reordered the DRAM initialization before
rockchip_stimer_init(), which as far as I can tell causes the RK3399 to
lock up completely.
Fix this issue in the common code by putting the DRAM init back after
timer init. I have only tested this on the RK3399, but it wouldn't make
any sense for the timer init to require DRAM be set up on any system.
Fixes: b7abef2ecbcc ("rockchip: rk3399: Migrate to use common spl board file") Signed-off-by: Thomas Hebb <tommyhebb@gmail.com> Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
Andy Yan [Thu, 14 Nov 2019 03:21:12 +0000 (11:21 +0800)]
arm: rockchip: Add RK3308 SOC support
RK3308 is a quad Cortex A35 based SOC with rich audio
interfaces(I2S/PCM/TDM/PDM/SPDIF/VAD/HDMI ARC), which
designed for intelligent voice interaction and audio
input/output processing.
Signed-off-by: Andy Yan <andy.yan@rock-chips.com> Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
Peter Robinson [Sat, 9 Nov 2019 20:30:05 +0000 (20:30 +0000)]
rockchip: dts: rk3399: move the u-boot, dm-pre-reloc to the u-boot.dtsi
The u-boot specific pieces in the dts files should be in u-boot.dtsi
not the main files, this allows easier sync with upstream. The
rk3399.dtsi has a mix of both so move them all for consistency.
Signed-off-by: Peter Robinson <pbrobinson@gmail.com> Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
(Fix with missing pmugrf) Signed-off-by: Kever Yang <kever.yang@rock-chips.com>
rockchip: rk3399: split rockpro64 out of evb_rk3399
rockpro64 needs to setup I/O domains in order for USB to work in u-boot.
Since we currently don't have a driver to do that, split it into its own
board file and initialize I/O domains here.
Signed-off-by: Vasily Khoruzhick <anarsoul@gmail.com> Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
CONFIG_IS_ENABLED() needs the config name like used in Kconfig, so
without the leading CONFIG_. The clock drivers all wrongly check for
CONFIG_RESET_ROCKCHIP, fix that
Signed-off-by: Heiko Stuebner <heiko.stuebner@theobroma-systems.com> Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
Soeren Moch [Thu, 7 Nov 2019 11:11:22 +0000 (12:11 +0100)]
arm: dts: rk3399-rockpro64: slightly increase center voltage
The rk3399 VD_CENTER voltage domain is not subject to dynamic voltage
scaling. So the regulator reset voltage of 0.9V is used on this board.
Let u-boot initialize the center voltage to 0.95V as it is done for the
VD_LOGIC domain. This avoids instability and occasional linux kernel
Opses on this board.
Signed-off-by: Soeren Moch <smoch@web.de> Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
Cc: Kever Yang <kever.yang@rock-chips.com> Cc: Levin Du <djw@t-chip.com.cn> Signed-off-by: Jagan Teki <jagan@amarulasolutions.com> Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
Add core architecture code to support the px30 soc.
This includes a separate tpl board file due to very limited
sram size as well as a non-dm sdram driver, as this also has
to fit into the tiny sram.
Signed-off-by: Heiko Stuebner <heiko.stuebner@theobroma-systems.com> Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
rockchip: misc: read cpuid either from efuse or otp
Newer Rockchip socs use a different ip block to handle one-time-
programmable memory, so depending on what got enabled get the cpuid
from either source.
Signed-off-by: Heiko Stuebner <heiko.stuebner@theobroma-systems.com> Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
spl: separate SPL_FRAMEWORK config for spl and tpl
Right now enabling SPL_FRAMEWORK will also enable it for the TPL in all
cases, making the TPL bigger. There may be cases where the TPL is really
size constrained due to its underlying ram size.
Therefore introduce a new TPL_FRAMEWORK option and make the relevant
conditionals check for both. The default is set to "y if SPL_FRAMEWORK"
to mimic the previous behaviour where the TPL would always get the
SPL framework if it was enabled in SPL.
Signed-off-by: Heiko Stuebner <heiko.stuebner@theobroma-systems.com> Reviewed-by: Kever Yang <kever.yang@rock-chips.com>
YouMin Chen [Fri, 15 Nov 2019 03:04:53 +0000 (11:04 +0800)]
ram: rk3399: Fix dram setting to make dram more stable
There are some code different with rockchip vendor code which may lead
to different bugs, including:
1) Fix setting error about LPDDR3 dram size ODT.
2) Set phy io speed to 0x2.
3) Fix setting error about phy_pad_fdbk_drive.
4) Fix setting error about PI_WDQLVL_VREF_EN
Signed-off-by: YouMin Chen <cym@rock-chips.com> Signed-off-by: Kever Yang <kever.yang@rock-chips.com>
YouMin Chen [Fri, 15 Nov 2019 03:04:43 +0000 (11:04 +0800)]
ram: px30: add sdram driver
Add the sdram driver for PX30 to support ddr3, ddr4, lpddr2 and lpddr3.
For TPL_BUILD, the driver implement full dram init and without DM
support due to the limit of internal SRAM size.
For SPL and U-Boot proper, it's a simple driver with dm for get
dram_info like other SoCs.
Signed-off-by: YouMin Chen <cym@rock-chips.com> Signed-off-by: Kever Yang <kever.yang@rock-chips.com>
Kever Yang [Fri, 15 Nov 2019 03:04:42 +0000 (11:04 +0800)]
ram: rockchip: add common msch reg definition
The noc register bit definition may be the same for different SoC while
the offset of the register may be different, add the struction
definition as common code.
Signed-off-by: YouMin Chen <cym@rock-chips.com> Signed-off-by: Kever Yang <kever.yang@rock-chips.com>
Kever Yang [Fri, 15 Nov 2019 03:04:37 +0000 (11:04 +0800)]
ram: rockchip: add common code for sdram driver
There are some function like os_reg setting, capacity detect functions,
can be used as common code for different Rockchip SoCs, add a
sdram_common.c for all these functions.
Signed-off-by: YouMin Chen <cym@rock-chips.com> Signed-off-by: Kever Yang <kever.yang@rock-chips.com>
Kever Yang [Fri, 15 Nov 2019 03:04:35 +0000 (11:04 +0800)]
rockchip: sdram: extend to use sys_reg3 for capacity info
Since we have new DRAM type and to support different DRAM size in different
CS, we need more bits, so introduce sys_reg3 to record the info.
Note that the info in sys_reg3 is extension to sys_reg2 and the info in
sys_reg2 is the same as before. We define the DRAM_INFO with sys_reg3 as
VERSION2.
All the ENC macro are moved to sdram_common.h since the sdram.c only
need to do the info decode.
Signed-off-by: YouMin Chen <cym@rock-chips.com> Signed-off-by: Kever Yang <kever.yang@rock-chips.com>
Kever Yang [Fri, 15 Nov 2019 03:04:34 +0000 (11:04 +0800)]
rockchip: sdram: move cap structure and debug function to sdram_common.h
The sdram.h suppose to be helper file for sdram.c which including dram
size decode and some u-boot related dram init interface, and all
structure and function for dram driver move to sdram_common.h
Signed-off-by: YouMin Chen <cym@rock-chips.com> Signed-off-by: Kever Yang <kever.yang@rock-chips.com>
Kever Yang [Fri, 15 Nov 2019 03:04:33 +0000 (11:04 +0800)]
ram: rockchip: rename sdram_common.c/h to sdram.c
rename sdram_common.c in arch/arm/mach-rockchip to sdram.c;
so that we can use the file name sdram_common.c in dram driver for
better understand the code;
clean the related file who has use the header file at the same time.
Signed-off-by: YouMin Chen <cym@rock-chips.com> Signed-off-by: Kever Yang <kever.yang@rock-chips.com>
Tom Rini [Thu, 14 Nov 2019 14:48:26 +0000 (09:48 -0500)]
Merge tag 'efi-2020-01-rc3' of https://gitlab.denx.de/u-boot/custodians/u-boot-efi
Pull request for UEFI sub-system for efi-2020-01-rc3
The following changes for the UEFI subsystem are provided:
* allow building UEFI binaries on the sandbox
* enable access to file systems without partition tables
* correctly check the return value of efi_dp_from_file()
When compiling list of cover letter cc addresses, using null as a
separater, then encoding to utf-8 results in lots of "\x00" as
separators. patman then doesnt understand that when it comes to
repoting the list to send-email.
Fix this by not encoding to utf-8, as done for the other patch files.
Signed-off-by: Robert Beckett <bob.beckett@collabora.com> Reviewed-by: Simon Glass <sjg@chromium.org>
Compiling arch/sandbox/cpu/os.c results in an error
../arch/sandbox/cpu/os.c: In function ‘os_find_text_base’:
../arch/sandbox/cpu/os.c:823:12: error: cast to pointer from
integer of different size [-Werror=int-to-pointer-cast]
823 | base = (void *)addr;
| ^
cc1: all warnings being treated as errors
The size of void* differs from that of unsigned long long on 32bit
systems.
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de> Reviewed-by: Simon Glass <sjg@chromium.org>
Heiko Stuebner [Wed, 23 Oct 2019 14:46:41 +0000 (16:46 +0200)]
tests: add OP-TEE test suite
OP-TEE can get supplied with a devicetree and will then insert
its firmware node and reserved-memory sections into it.
As this devicetree often is not the one supplied to a later
loaded kernel, a previous commit added functionality to transfer
these nodes onto that new devicetree.
To make sure this functionality stays intact, also add a test
for the transfer functionality.
Signed-off-by: Heiko Stuebner <heiko.stuebner@theobroma-systems.com> Reviewed-by: Simon Glass <sjg@chromium.org>
Heiko Stuebner [Wed, 23 Oct 2019 14:46:40 +0000 (16:46 +0200)]
image: fdt: copy possible optee nodes to a loaded devicetree
The loading convention for optee or any other tee on arm64 is as bl32
parameter to the trusted-firmware. So TF-A gets invoked with the TEE as
bl32 and main u-boot as bl33. Once it has done its startup TF-A jumps
into the bl32 for the TEE startup, returns to TF-A and then jumps to bl33.
All of them get passed a devicetree as parameter and all components often
get loaded from a FIT image.
OP-TEE will create additional nodes in that devicetree namely a firmware
node and possibly multiple reserved-memory nodes.
While this devicetree is used in main u-boot, in most cases it won't be
the one passed to the actual kernel. Instead most boot commands will load
a new devicetree from somewhere like mass storage of the network, so if
that happens u-boot should transfer the optee nodes to that new devicetree.
To make that happen introduce optee_copy_fdt_nodes() called from the dt
setup function in image-fdt which after checking for the optee presence
in the u-boot dt will make sure a optee node is present in the kernel dt
and transfer any reserved-memory regions it can find.
Heiko Stuebner [Wed, 23 Oct 2019 14:46:39 +0000 (16:46 +0200)]
fdtdec: only create phandle if caller wants it in fdtdec_add_reserved_memory()
The phandlep pointer returning the phandle to the caller is optional
and if it is not set when calling fdtdec_add_reserved_memory() it is
highly likely that the caller is not interested in a phandle to the
created reserved-memory area and really just wants that area added.
So just don't create a phandle in that case.
Signed-off-by: Heiko Stuebner <heiko.stuebner@theobroma-systems.com> Reviewed-by: Simon Glass <sjg@chromium.org>
Heiko Stuebner [Wed, 23 Oct 2019 14:46:38 +0000 (16:46 +0200)]
fdtdec: protect against another NULL phandlep in fdtdec_add_reserved_memory()
The change adding fdtdec_add_reserved_memory() already protected the added
phandle against the phandlep being NULL - making the phandlep var optional.
But in the early code checking for an already existing carveout this check
was not done and thus the phandle assignment could run into trouble,
so add a check there as well, which makes the function still return
successfully if a matching region is found, even though no-one wants to
work with the phandle.
Patrick Delaunay [Wed, 23 Oct 2019 13:44:36 +0000 (15:44 +0200)]
pinctrol: dm: remove the function pinctrl_decode_pin_config
Remove the pinctrl_decode_pin_config() API, because this
function is unused and not compatible with livetree
(it uses fdtdec_get_bool instead of ofnode API).
Signed-off-by: Patrick Delaunay <patrick.delaunay@st.com> Reviewed-by: Simon Glass <sjg@chromium.org>
AKASHI Takahiro [Mon, 7 Oct 2019 05:59:39 +0000 (14:59 +0900)]
efi_loader: disk: install file system protocol to a whole disk
Currently, a whole disk without any partitions is not associated
with EFI_SIMPLE_FILE_SYSTEM_PROTOCOL. So even if it houses some
file system, there is a chance that we may not be able to access
it, particularly, when accesses are to be attempted after searching
that protocol against a device handle.
With this patch, EFI_SIMPLE_FILE_SYSTEM_PROTOCOL is installed
to such a disk if part_get_info() shows there is no partition
table installed on it.
Signed-off-by: AKASHI Takahiro <takahiro.akashi@linaro.org>
Only if no partition table exists, check for a file system on disk level. Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Correctly check the return value of efi_dp_from_file().
If we can determine device path for the file, should not depend on the
device path for the device being requested.
Provide a function description for efi_dp_from_name().
Reported-by: Coverity CID 273159, CID 273158 Fixes: 08c51fff30cc ("efi_loader: device_path: check against file path length") Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
- Migrate the symbol CONFIG_SYS_REDUNDAND_ENVIRONMENT to Kconfig. This
is size neutral outside of two platforms with latent bugs being fixed
now and they no longer have "ENV_IS_NOWHERE" set along with their
intended location.
Patrick Delaunay [Tue, 12 Nov 2019 14:39:58 +0000 (15:39 +0100)]
video: stm32: remove the compatible "synopsys, dw-mipi-dsi" support
Remove the compatible "synopsys,dw-mipi-dsi" added in U-Boot
(it don't exist in Linux kernel binding); it is only used
to bind the generic synopsys UCLASS_DSI_HOST "dw_mipi_dsi" to
the driver "stm32-display-dsi" UCLASS_VIDEO_BRIDGE
This binding is done in Linux kernel drivers without compatible
(dw_mipi_dsi_bind() is called in bind of driver, for example in
drivers/gpu/drm/rockchip/dw-mipi-dsi-rockchip.c).
This patch does the same in U-Boot driver, the STM32 driver
calls during its bind the function device_bind_driver_to_node
to bind the generic driver "dw_mipi_dsi" at the same address.
This patch reduces the device tree differences
between Linux kernel and U-Boot for stm32mp1 platform.
Tested with v2020.01-rc1 on STM32MP157C-EV1 and STM32MP157C-DK2.
The dependency of driver is clearer and the probe order is guaranteed.
Fabio Estevam [Fri, 1 Nov 2019 12:25:21 +0000 (09:25 -0300)]
video: mxsfb: Configure the clock after eLCDIF reset
Running stress reboot test on a i.MX6ULL evk board with a
custom LCD can lead to splash screen failure (black screen).
After comparing the eLCDIF initialization with the Linux kernel
driver, it was noticed that the eLCDIF reset is the first operation
that needs to be done.
So do like the eLCDIF Linux driver and move the eLCDIF clock
frequency configuration after the eLCDIF reset and just prior to
setting the RUN bit.
With this change applied, no more black screen issues has been
found during overnight reboot stress tests.
With recent update in u-boot gitattributes all files are treated as regular
text files. This creates problems with special files and repo always
shows uncommitted files like below.
Your branch is up-to-date with 'origin/master'.
Changes not staged for commit:
(use "git add <file>..." to update what will be committed)
(use "git checkout -- <file>..." to discard changes in working directory)
projects / u-boot.git / log