Prepare for booting Android images which lack any DTB in the second
area by using 'fdtaddr' environment variable as source/address of FDT.
No functional/behavioral change expected in this patch.
Signed-off-by: Eugeniu Rosca <erosca@de.adit-jv.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Increase the readability of boot_get_fdt().
No change in behavior is expected.
Signed-off-by: Eugeniu Rosca <erosca@de.adit-jv.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
The 'no_fdt' goto label was introduced by v2015.01 commit [0] and it
had two review stages [1-2]. The *documented* purpose behind commit [0]
is (excerpt from commit description):
> allows both FDT and non-FDT kernels to boot by making the
> third parameter to the bootm/bootz optional
While [1] and [2] share the same goal, they have very different
implementations:
- [1] was based on a very simple 'argc' check at function error out
with returning success to the caller if the third parameter was NOT
passed to bootm/bootz command. This approach had the downside of
returning success to the caller even in case of legitimate internal
errors, which should halt booting.
- [2] added the "no_fdt" label and several "goto no_fdt" statements.
This allowed to report the legitimate internal errors to the caller.
IOW the major difference between [1] and [2] is:
- [1] boot w/o FDT if FDT address is not passed to boot{m,z,*}
- [2] give *freedom* to the developer to boot w/o FDT from any
(more or less) arbitrary point in the function flow (and here
comes the peculiar aspect, which looks to be a leftover from [1])
with the precondition that the 3rd argument (FDT address) is NOT
provided to boot{m,z,*}. In practice, this means that only a subset
of "goto no_fdt" end up booting w/o FDT while the other subset is
returning an error to the caller.
This patch removes the peculiar behavior described above, such that
"goto no_fdt" performs really what it tells to the developer.
The motivation of this patch is to decrease the unneeded complexity
and increase the readability of boot_get_fdt().
[0] 48aead71c1 ("fdt: Allow non-FDT kernels to boot when CONFIG_OF_LIBFDT is defined")
[1] https://patchwork.ozlabs.org/patch/412923/
("[U-Boot,v1] fdt: Allow non-FDT kernels to boot when CONFIG_OF_LIBFDT is defined")
[2] https://patchwork.ozlabs.org/patch/415635/
("[U-Boot,v2] fdt: Allow non-FDT kernels to boot when CONFIG_OF_LIBFDT is defined")
Signed-off-by: Eugeniu Rosca <erosca@de.adit-jv.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Paranoid programming [1] lies at the foundation of proper software
development, but the repetitive zeroing-out of output arguments in the
context of the same function rather clutters the code and inhibits
further refactoring/optimization than is doing any good.
In boot_get_fdt(), we already perform zero/NULL-initialization of
*of_flat_tree and *of_size at the beginning of the function, so doing
the same at function error-out is redundant/superfluous.
Moreover, keeping the code unchanged might encourage the developers to
update *of_flat_tree and *of_size during some interim computations,
which is against the current design of boot_get_fdt(). Currently,
writing useful data into these arguments happens just before
successfully returning from boot_get_fdt() and it should better stay so.
[1] https://blog.regehr.org/archives/1106
Signed-off-by: Eugeniu Rosca <erosca@de.adit-jv.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
This fixes a disconnect issue detected with fastboot command, when using
dwc2 driver.
- On u-boot side:
uboot>$ fastboot 0
- On USB host PC side, few seconds after
PC>$ fastboot reboot # Get stuck, uboot target never reboots
By enabling DEBUG_ISR logs, the bus suspend interrupt is seen before the
PC command has been issued. When the USB bus suspend occurs, there's a HACK
that disables the fastboot (composite driver). Here is the call stack
upon USB bus suspend:
- dwc2_handle_usb_suspend_intr()
- dev->driver->disconnect()
- composite_disconnect()
- reset_config()
- f->disable()
- fastboot_disable()
- usb_ep_disable(f_fb->out_ep);
- usb_ep_disable(f_fb->in_ep);
.. other disable calls.
When the resume interrupt happens, everything has been disabled, then
nothing happens. fastboot command gets stuck on HOST side.
Remove original HACK, that disconnects the composite driver upon
USB bus suspend. Implement disconnect detection instead:
- check GINTSTS OTG interrupt
- read GOTGINT register
- check GOTGINT, SesEndDet bit (e.g. session end)
This is inspired by what is implemented currently in Linux dwc2 driver.
Signed-off-by: Fabrice Gasnier <fabrice.gasnier@st.com>
Reviewed-by: Marek Vasut <marex@denx.de>
Reload the watchdog in the mass storage command ums
to avoid reboot during the usb waiting loop
when the host doesn't send any request.
Signed-off-by: Patrick Delaunay <patrick.delaunay@st.com>
Our R-Car3 Android userspace relies on the output of 'fastboot
getvar platform' and U-Boot currently is not able to handle it:
host $> fastboot getvar platform
getvar:platform FAILED (remote: Variable not implemented)
finished. total time: 0.001s
We either have the option of adding 'fastboot.platform' variable
to the default/saved environment as a workaround or add proper
'fastboot getvar platform' support in U-Boot via this patch.
In the latter case, other platforms can benefit from it too.
Note that R-Car3 already exports 'platform' environment variable via
v2019.01 commit 00e4b57e9e ("ARM: rmobile: Set environment variable
containing CPU type").
Signed-off-by: Eugeniu Rosca <erosca@de.adit-jv.com>
Where we have to compute partition names, rather than using a hardcoded
32 for the partition name length, replace with PART_NAME_LEN.
Signed-off-by: Alex Kiernan <alex.kiernan@gmail.com>
Reviewed-by: Eugeniu Rosca <rosca.eugeniu@gmail.com>
The DK1 and DK2 boards use the USB Type-C controller STUSB1600.
This patch updates:
- the device tree to add the I2C node in the DT
- the board stm32mp1 to probe this I2C device and use this controller
to check cable detection.
- the DWC2 driver to support a new dt property
"u-boot,force-b-session-valid" which forces B session and
device mode; it is a workaround because the VBUS sensing and
ID detection isn't available with stusb1600.
Signed-off-by: Patrick Delaunay <patrick.delaunay@st.com>
Reviewed-by: Lukasz Majewski <lukma@denx.de>
Use the DWC2 device driver with DM_USB_GADGET support and
cleanup the USB support in STM32MP1 board.
Signed-off-by: Patrick Delaunay <patrick.delaunay@st.com>
Remove the HOST support for STM32MP1 USBO device = OTG DWC2.
The current DWC2 driver have no dynamic detection of device,
So it is dangerous to have start 3V3 when PC is
connected to the micro USB connector.
=> it is preferable to have only DEVICE support
CONFIG_USB_GADGET_DWC2_OTG for OTG port
See DWC3 driver for clean dual role support...
Signed-off-by: Patrick Delaunay <patrick.delaunay@st.com>
Add compatible "st,stm32mp1-hsotg" and associated driver data to manage
the usb33d-supply and the ST specific register for VBus sensing.
Signed-off-by: Patrick Delaunay <patrick.delaunay@st.com>
# Conflicts:
# drivers/usb/gadget/dwc2_udc_otg.c
Reviewed-by: Lukasz Majewski <lukma@denx.de>
All TX fifo size can be different, add tx_fifo_sz_array[]
into dwc2_plat_otg_data to be able to set them.
tx_fifo_sz_array[] is 17 Bytes long and can contains max 16
tx fifo size (synopsys IP supports max 16 IN endpoints).
First entry of tx_fifo_sz_array[] is the number of valid
fifo size the array contains.
In case of tx_fifo_sz_array[] doesn't contains the same
number of element than max hardware endpoint, display
a warning message.
Compatibility with board which doesn't use tx_fifo_sz_array[]
(Rockchip rk322x/rk3128/rv1108/rk3288/rk3036) is kept.
Signed-off-by: Patrice Chotard <patrice.chotard@st.com>
Signed-off-by: Patrick Delaunay <patrick.delaunay@st.com>
Reviewed-by: Lukasz Majewski <lukma@denx.de>
Some DWC2 ip variant doesn't use 16 hardware endpoint as hardcoded
in the driver. Bits INEps [29:26] of HWCFG4 register allows to get
this information.
Signed-off-by: Patrice Chotard <patrice.chotard@st.com>
Signed-off-by: Patrick Delaunay <patrick.delaunay@st.com>
Reviewed-by: Lukasz Majewski <lukma@denx.de>
Add a new function to check the session B validity, to be use to check
cable connection.
Signed-off-by: Patrick Delaunay <patrick.delaunay@st.com>
Reviewed-by: Lukasz Majewski <lukma@denx.de>
Reset the hardware to be sure of the device state.
Signed-off-by: Patrick Delaunay <patrick.delaunay@st.com>
Reviewed-by: Lukasz Majewski <lukma@denx.de>
Minimal conversion to driver model by using the uclass
UCLASS_USB_GADGET_GENERIC based on:
- reset uclass
- clock uclass
- generic uclass.
Signed-off-by: Patrick Delaunay <patrick.delaunay@st.com>
Reviewed-by: Lukasz Majewski <lukma@denx.de>
Since its inception in upstream fastboot android-n-preview-1 [1],
"has-slot" option has never taken the form of "has_slot". Amongst the
users of "getvar has-slot:" is the upstream bootloadertest.py [2].
Current U-Boot "has_slot" version must be a typo. Fix it.
[1] https://android.googlesource.com/platform/system/core/+/a797479bd51c
("Fix fastboot variable name")
[2] https://android.googlesource.com/platform/system/extras/+/72de393e118e3
("Bootloader verification for AndroidThings.")
Fixes: f73a7df984 ("net: fastboot: Merge AOSP UDP fastboot")
Signed-off-by: Eugeniu Rosca <erosca@de.adit-jv.com>
Acked-by: Alex Kiernan <alex.kiernan@gmail.com>
The compiler is not happy when neither USB nor TFTP transport for DFU defined:
cmd/dfu.c: In function ‘do_dfu’:
cmd/dfu.c:31:8: warning: unused variable ‘devstring’ [-Wunused-variable]
char *devstring = argv[3];
^~~~~~~~~
cmd/dfu.c:30:8: warning: unused variable ‘interface’ [-Wunused-variable]
char *interface = argv[2];
^~~~~~~~~
Surround those variables by #ifdef expression.
More serious, that comes under same circumstances, is a compilation error due
to absence of macro parameter:
In file included from include/image.h:45,
from include/common.h:35,
from cmd/dfu.c:13:
include/command.h:207:24: error: expected expression before ‘,’ token
# define _CMD_HELP(x) x,
^
include/command.h:286:18: note: in expansion of macro ‘_CMD_HELP’
_cmd, _usage, _CMD_HELP(_help) _CMD_COMPLETE(_comp) }
^~~~~~~~~
include/command.h:290:3: note: in expansion of macro ‘U_BOOT_CMD_MKENT_COMPLETE’
U_BOOT_CMD_MKENT_COMPLETE(_name, _maxargs, _rep, _cmd, \
^~~~~~~~~~~~~~~~~~~~~~~~~
include/command.h:332:2: note: in expansion of macro ‘U_BOOT_CMD_COMPLETE’
U_BOOT_CMD_COMPLETE(_name, _maxargs, _rep, _cmd, _usage, _help, NULL)
^~~~~~~~~~~~~~~~~~~
cmd/dfu.c:70:1: note: in expansion of macro ‘U_BOOT_CMD’
U_BOOT_CMD(dfu, CONFIG_SYS_MAXARGS, 1, do_dfu,
^~~~~~~~~~
make[1]: *** [scripts/Makefile.build:279: cmd/dfu.o] Error 1
make: *** [Makefile:1518: cmd] Error 2
Put empty string unconditionally to have macro parameter present.
Fixes: 0f44d33536 ("dfu: Fix up the Kconfig mess")
Cc: Marek Vasut <marek.vasut@gmail.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Acked-by: Lukasz Majewski <lukma@denx.de>
USB_MUSB_TI requires USB_MUSB_DSPS, failing at link time if it's not
selected:
drivers/usb/musb-new/built-in.o: In function `ti_musb_host_ofdata_to_platdata':
drivers/usb/musb-new/ti-musb.c:193: undefined reference to `musb_dsps_ops'
or if OF_CONTROL is not selected:
arch/arm/mach-omap2/built-in.o:(.data.usb0+0x24): undefined reference to `musb_dsps_ops'
Reviewed-by: Hannes Schmelzer <oe5hpm@oevsv.at>
Tested-by: Hannes Schmelzer <oe5hpm@oevsv.at>
Signed-off-by: Alex Kiernan <alex.kiernan@gmail.com>
Add the DM_MMC-compatible DesignWare MMC driver support for Synopsys
ARC devboards. It is created to switch ARC devboards to use DM_MMC.
It required information such as clocks (Bus Interface Unit clock,
Card Interface Unit clock) and SDIO bus width.
Signed-off-by: Eugeniy Paltsev <Eugeniy.Paltsev@synopsys.com>
Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
The generic timer count is an incrementing 64bit value and a timer driver
must return an incrementing 64bit value. The DW APB timer only provides a
32bit timer counting down, thus the result must be inverted and converted
to a 64bit value. The current implementation is however missing the 64bit
up-conversion and this results in random timer roll-overs, which in turn
triggers random timeouts throughout the codebase.
This patch adds the missing 64bit up-conversion to fix the issue.
Signed-off-by: Marek Vasut <marex@denx.de>
Cc: Chin Liang See <chin.liang.see@intel.com>
Cc: Dinh Nguyen <dinguyen@kernel.org>
Cc: Ley Foon Tan <ley.foon.tan@intel.com>
Cc: Simon Goldschmidt <simon.k.r.goldschmidt@gmail.com>
Cc: Tien Fong Chee <tien.fong.chee@intel.com>
Merge qspi dts node from Linux.
Commit 0cb140d07fc75fb (arm64: dts: stratix10: Add QSPI support for Stratix10)
Add -u-boot.dtsi files for non Linux dts properties and
update properties for Uboot.
- add u-boot,dm-pre-reloc
- add alias for spi0
- change compatible for flash
- support quad read and quad write
- change maximum frequency to 100MHz
Tested on Stratix 10 SoC devkit.
SOCFPGA_STRATIX10 # sf probe 0:0
SF: Detected mt25qu02g with page size 256 Bytes, erase size 64 KiB, total 256 MiB
Signed-off-by: Ley Foon Tan <ley.foon.tan@intel.com>
Scrub memory content if ECC is enabled and it is not
from warm reset boot.
Enable icache and dcache before scrub memory
and use "DC ZVA" instruction to clear memory
to zeros. This instruction writes a cache line
at a time and it can prevent false ECC error
trigger if write cache line partially.
Signed-off-by: Ley Foon Tan <ley.foon.tan@intel.com>
The stratix10 socdk ships with 4GB of memory. Modify the
device tree to represent this. Note that to access 4GB of
memory in Stratix 10, due to the IO space from 2GB to 4GB,
we use the fact that the DDR controller ignores upper address
bits outside of the configured DRAM's size. This means that
, the 4GB DRAM is mapped to memory every 4GB.
For an 8GB memory, you can either live with the 2GB IO space,
and loose access to that memory from the processor, or use
the same trick:
Loose 2GB of memory:
memory {
device_type = "memory";
/* 8GB */
/* first 2GB */
reg = <0 0x00000000 0 0x80000000>,
/* last 4GB */
<1 0x00000000 1 0x00000000>;
u-boot,dm-pre-reloc;
};
or to map it all:
memory {
device_type = "memory";
/* 8GB */
/* first 2GB */
reg = <0 0x00000000 0 0x80000000>,
/* next 6GB */
<2 0x80000000 1 0x80000000>;
u-boot,dm-pre-reloc;
};
Signed-off-by: Dalon Westergreen <dalon.westergreen@intel.com>
Signed-off-by: Ley Foon Tan <ley.foon.tan@intel.com>
Stratix10 maps dram in 2 address spans, from 0-2GB and from
2GB up to 128GB.
Signed-off-by: Dalon Westergreen <dalon.westergreen@intel.com>
Signed-off-by: Ley Foon Tan <ley.foon.tan@intel.com>
Stratix 10 maps dram from 0 to 128GB. There is a 2GB hole
in the memory for peripherals and other IO from 2GB to 4GB.
However the dram controller ignores upper address bits for
smaller dram configurations. Example: a 4GB dram
maps to multiple locations, every 4GB on the address.
Signed-off-by: Dalon Westergreen <dalon.westergreen@intel.com>
Signed-off-by: Ley Foon Tan <ley.foon.tan@intel.com>
Move SDRAM size check to SDRAM driver. sdram_calculate_size()
is called in SDRAM initialization already, avoid calling
twice in size check function.
Signed-off-by: Ley Foon Tan <ley.foon.tan@intel.com>
This commit removes ad-hoc reset handling for peripheral resets from SPL
for socfpga gen5.
This is done because as U-Boot drivers support reset handling by now.
Signed-off-by: Simon Goldschmidt <simon.k.r.goldschmidt@gmail.com>
This adds reset handling to the cadence qspi driver.
For backwards compatibility, only a warning is printed when failing to
get reset handles.
Signed-off-by: Simon Goldschmidt <simon.k.r.goldschmidt@gmail.com>
This adds reset handling to the devicetree-enabled Denali NAND driver.
For backwards compatibility, only a warning is printed when failing to
get reset handles.
Signed-off-by: Simon Goldschmidt <simon.k.r.goldschmidt@gmail.com>
To clean up reset handling for socfpga gen5, port the DDR driver to DM
using UCLASS_RAM and implement proper reset handling.
This gets us rid of one ad-hoc call to socfpga_per_reset().
The gen5 driver is implemented in 2 distinct files. One of it (containing
the calibration training) is not touched much and is kept at using
hard coded addresses since the code grows even more otherwise.
SPL is changed from calling hard into the DDR driver code to just
probing UCLASS_RESET and UCLASS_RAM. It is happy after finding a RAM
driver after that.
Signed-off-by: Simon Goldschmidt <simon.k.r.goldschmidt@gmail.com>
This adds code to take peripherals out of reset based on an environment
variable. This is in preparation for removing the code that does this from
SPL.
However, some drivers even in current Linux cannot handle peripheral reset,
so until this works, we need a compatibility workaround.
This workaround is implemented in the 'assert' and 'remove' callbacks of
this reset driver: the 'assert' callback does not disable peripherals that
were already taken out of reset, while the 'remove' callback, which is
called on OS_PREPARE, deasserts all peripheral resets if the environment
variable "socfpga_legacy_reset_compat" is set to 1, which is what the gen5
SPL did up to now.
This is in preparation to clean up the SPL and implementing proper reset
handling for U-Boot.
Signed-off-by: Simon Goldschmidt <simon.k.r.goldschmidt@gmail.com>
To keep the current behaviour of taking all peripherals out of reset
before booting the OS before removing that code from socfpga gen5 SPL,
this enables the new behaviour by default for all gen5 boards by adding
the environment variable "socfpga_legacy_reset_compat=1" to the default
environment.
This can be overridden in board config files or by saving an environment
without this variable enabled.
Signed-off-by: Simon Goldschmidt <simon.k.r.goldschmidt@gmail.com>
The only member of this driver's priv struct is a pointer, which is
called 'membase'. However, since this driver handles multiple sub-
architectures, this is not the base address from dts but the base
address of some common registers of those sub-arches.
Reflect this better in sourcecode by renaming 'membase' to 'modrst_base'.
Signed-off-by: Simon Goldschmidt <simon.k.r.goldschmidt@gmail.com>
