Instead of listing individual registers for controls to each processor
core, list them as an array of registers. This makes accessing controls
by core index easier.
Also rename "cpucfg_sun6i.h" (which was unused anyway) to the more generic
"cpucfg.h", and add packed attribute to struct sunxi_cpucfg.
Signed-off-by: Chen-Yu Tsai <wens@csie.org>
Acked-by: Marc Zyngier <marc.zyngier@arm.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
cpucfg_sun6i.h includes a register definition for the CPUCFG register
block. The types used are u32 and u8, which are defined in linux/types.h.
Signed-off-by: Chen-Yu Tsai <wens@csie.org>
Acked-by: Marc Zyngier <marc.zyngier@arm.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
struct sunxi_prcm_reg is a representation of the PRCM registers. Add
the packed attribute to prevent the compiler from doing funny things.
Signed-off-by: Chen-Yu Tsai <wens@csie.org>
Acked-by: Marc Zyngier <marc.zyngier@arm.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Use SUNXI_CPUCFG_BASE across all families. This makes writing common
PSCI code easier.
Signed-off-by: Chen-Yu Tsai <wens@csie.org>
Acked-by: Marc Zyngier <marc.zyngier@arm.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
The PSCI implementation expects at most 2 pages worth of space reserved
at the end of the secure section for its stacks. If PSCI is relocated to
secure SRAM, then everything is fine. If no secure SRAM is available,
and PSCI remains in main memory, the reserved memory space doesn't cover
the space used by the stack.
If one accesses PSCI after Linux has fully booted, the memory that should
have been reserved for the PSCI stacks may have been used by the kernel
or userspace, and would be corrupted. Observed after effects include the
system hanging or telinit core dumping when trying to reboot. It seems
the init process gets hit the most on my test bed.
This fix allocates the space used by the PSCI stacks in the secure
section by skipping pages in the linker script, but only when there is
no secure SRAM, to avoid bloating the binary.
This fix is only a stop gap. It would be better to rework the stack
allocation mechanism, maybe with proper usage of CONFIG_ macros and an
explicit symbol.
Signed-off-by: Chen-Yu Tsai <wens@csie.org>
Acked-by: Marc Zyngier <marc.zyngier@arm.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Some common PSCI functions are written in assembly, but it should be
possible to use them from C code.
Add function declarations for C code to consume.
Signed-off-by: Chen-Yu Tsai <wens@csie.org>
Acked-by: Marc Zyngier <marc.zyngier@arm.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
For psci_get_cpu_stack_top() to be usable in C code, it must adhere to
the ARM calling conventions. Since it could be called when the stack
is still unavailable, and the entry code to linux also expects r1 and
r2 to remain unchanged, stick to r0 and r3.
Signed-off-by: Chen-Yu Tsai <wens@csie.org>
Acked-by: Marc Zyngier <marc.zyngier@arm.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
This reverts commit 1a5f0de08e86("sunxi: make SoC variant choice
mandatory").
With the optional marking in the Kconfig "make savedefconfig"
will drop CONFIG_MACH_SUN4I=y from all the A10 boards, making it
hard to see at a glance which family of sunxi chips the defconfig
is for.
This commit therefore restores the optional, and restores
CONFIG_MACH_SUN4I=y to all defconfig's which had it dropped
because of this.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Ian Campbell <ijc@hellion.org.uk>
The inet86dz board is a board used in 7" tablets from various oems.
These tablets are a23 based 7" tablets featuring a 1024x600 LCD,
512MB RAM, 4G NAND, rtl8188etv usb wifi, gsl1680 touchschreen,
micro-sd slot, 3.5mm headphone jack and a micro-usb otg connector
which doubles as charging port.
The dts file this commit adds is identical to the one submitted to
the upstream kernel.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Ian Campbell <ijc@hellion.org.uk>
The Polaroid MID2407PXE03 is an a23 based 7" tablet based on a M86_MB V2.0
PCB, featuring a 800x480 LCD, 512MB RAM, 4G NAND, esp8089 wifi, gsl1680
touchschreen, micro-sd slot, 3.5mm headphone jack and a micro-usb otg
connector which doubles as charging port.
The dts file is identical to the one submitted to the upstream kernel.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Ian Campbell <ijc@hellion.org.uk>
This gives us a bit more breathing room wrt our SPL size.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Ian Campbell <ijc@hellion.org.uk>
Reviewed-by: Simon Glass <sjg@chromium.org>
vprintf is used by panic() which is used in various SPL paths on some
boards.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Ian Campbell <ijc@hellion.org.uk>
Reviewed-by: Simon Glass <sjg@chromium.org>
Since commit bb597c0eeb ("common: bootdelay: move CONFIG_BOOTDELAY
into a Kconfig option"), CONFIG_BOOTDELAY is defined for all boards.
Prior to that commit, it was allowed to unset CONFIG_BOOTDELAY to
not compile common/autoboot.c, as described in common/Makefile:
# This option is not just y/n - it can have a numeric value
ifdef CONFIG_BOOTDELAY
obj-y += autoboot.o
endif
It was a bit odd to enable/disable code with an integer type option,
but it was how this option worked before that commit, and several
boards actually unset it to opt out of the autoboot feature.
This commit adds a new bool option, CONFIG_AUTOBOOT, and makes
CONFIG_BOOTDELAY depend on it.
I chose "default y" for this option because most boards use the
autoboot. I added "# CONFIG_AUTOBOOT is not set" for the boards that
had not set CONFIG_BOOTDELAY prior to the bad commit.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Due to changes in distro environment, ENV_SIZE limit was reached on Dragonboard.
This patch increases environment size to 8KiB.
Signed-off-by: Mateusz Kulikowski <mateusz.kulikowski@gmail.com>
Some drivers are still directly accessing the chip->mtd field. Patch
them to use nand_to_mtd() instead.
Signed-off-by: Boris Brezillon <boris.brezillon@free-electrons.com>
nand torture currently works on exactly one nand block which is specified
by giving the byteoffset to the beginning of the block.
Extend this by allowing for a second parameter specifying the byte size
to be tested.
e.g.
==> nand torture 1000000
NAND torture: device 0 offset 0x1000000 size 0x20000 (block size 0x20000)
Passed: 1, failed: 0
==> nand torture 1000000 40000
NAND torture: device 0 offset 0x1000000 size 0x40000 (block size 0x20000)
Passed: 2, failed: 0
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
Reviewed-by: Benoît Thébaudeau <benoit.thebaudeau.dev@gmail.com>
[scottwood: fix usage to show size as optional, and add misssing braces]
Signed-off-by: Scott Wood <oss@buserror.net>
Somehow this got overlooked when getting rid of nand_info.
Small patch, won't affect anything else, no reason to wait for the
next cycle.
Signed-off-by: Sergey Kubushyn <ksi@koi8.net>
NAND chips are supposed to expose their capabilities through advanced
mechanisms like READID, ONFI or JEDEC parameter tables. While those
methods are appropriate for the bootloader itself, it's way to
complicated and takes too much space to fit in the SPL.
Replace those mechanisms by a dumb 'trial and error' mechanism.
With this new approach we can get rid of the fixed config list that was
used in the sunxi NAND SPL driver.
Signed-off-by: Boris Brezillon <boris.brezillon@free-electrons.com>
Acked-by: Hans de Goede <hdegoede@redhat.com>
Split the 'load page' and 'read page' logic in 2 different functions so
we can later load the page and test different ECC configs without the
penalty of reloading the same page in the NAND cache.
We also move common setup to a dedicated function (nand_apply_config()) to
avoid rewriting the same values in NFC registers each time we read a page.
These new functions are passed a pointer to an nfc_config struct to limit
the number of parameters.
Signed-off-by: Boris Brezillon <boris.brezillon@free-electrons.com>
Acked-by: Hans de Goede <hdegoede@redhat.com>
check_value_xxx() helpers are using a 1ms delay between each test, which
can be quite long for some operations (like a page read on an SLC NAND).
Since we don't have anything to do but to poll this register, reduce the
delay between each test to 1us.
While we're at it, rename the max_number_of_retries parameters and the
MAX_RETRIES macro into timeout_us and DEFAULT_TIMEOUT_US to reflect that
we're actually waiting a given amount of time and not only a number of
retries.
Signed-off-by: Boris Brezillon <boris.brezillon@free-electrons.com>
Acked-by: Hans de Goede <hdegoede@redhat.com>
Use CONFIG_SYS_NAND_U_BOOT_OFFS_REDUND value instead of trying to guess
where the redundant u-boot image is based on simple (and most of the time
erroneous) heuristics.
Signed-off-by: Boris Brezillon <boris.brezillon@free-electrons.com>
Acked-by: Hans de Goede <hdegoede@redhat.com>
# Conflicts:
# drivers/mtd/nand/sunxi_nand_spl.c
On modern NAND it's more than recommended to have a backup copy of the
u-boot binary to recover from corruption: bitflips are quite common on
MLC NANDs, and the read-disturbance will corrupt your u-boot partitition
more quickly than what you would see on an SLC NAND.
Add an extra Kconfig option to specify the offset of the redundant u-boot
image.
Signed-off-by: Boris Brezillon <boris.brezillon@free-electrons.com>
Acked-by: Hans de Goede <hdegoede@redhat.com>
[scottwood: added ifdef to fix build break]
Signed-off-by: Scott Wood <oss@buserror.net>
The SYS_NAND_U_BOOT_OFFS is quite generic, but the Kconfig entry is forced
to explicitly depend on platforms that are not already defining it in their
include/configs/<board>.h header.
Add the SYS_NAND_U_BOOT_LOCATIONS option, make the SYS_NAND_U_BOOT_OFFS
depends on it, remove the dependency on NAND_SUNXI and make it dependent
on SPL selection.
Signed-off-by: Boris Brezillon <boris.brezillon@free-electrons.com>
Acked-by: Hans de Goede <hdegoede@redhat.com>
The sunxi SPL NAND controller driver supports use 'BootROM'-like configs,
that is, configs where the ECC bytes and real data are interleaved in the
page instead of putting ECC bytes in the OOB area.
Doing that has several drawbacks:
- since you're interleaving data and ECC bytes you can't use the whole page
otherwise you might override the bad block marker with non-FF bytes.
- to solve the bad block marker problem, the ROM code supports partially
using the page, but this introduces a huge penalty both in term of read
speed and NAND memory usage. While this is fine for rather small
binaries(like the SPL one which is at maximum 24KB large), it becomes
non-negligible for the bootloader image (several hundred of KB).
- auto-detection of the page size is not reliable (this is in my opinion
the biggest problem). If you get the page size wrong, you'll end up
reading data at a different offset than what was specified by the caller
and the reading may succeed (if valid data were written at this address).
For all those reasons I think it's wiser to completely remove support for
'syndrome' configs. If we ever need to support it again, then I'd recommend
specifying all the config parameters through Kconfig options.
Signed-off-by: Boris Brezillon <boris.brezillon@free-electrons.com>
Acked-by: Hans de Goede <hdegoede@redhat.com>
The following changes are made to the clock API:
* The concept of "clocks" and "peripheral clocks" are unified; each clock
provider now implements a single set of clocks. This provides a simpler
conceptual interface to clients, and better aligns with device tree
clock bindings.
* Clocks are now identified with a single "struct clk", rather than
requiring clients to store the clock provider device and clock identity
values separately. For simple clock consumers, this isolates clients
from internal details of the clock API.
* clk.h is split so it only contains the client/consumer API, whereas
clk-uclass.h contains the provider API. This aligns with the recently
added reset and mailbox APIs.
* clk_ops .of_xlate(), .request(), and .free() are added so providers
can customize these operations if needed. This also aligns with the
recently added reset and mailbox APIs.
* clk_disable() is added.
* All users of the current clock APIs are updated.
* Sandbox clock tests are updated to exercise clock lookup via DT, and
clock enable/disable.
* rkclk_get_clk() is removed and replaced with standard APIs.
Buildman shows no clock-related errors for any board for which buildman
can download a toolchain.
test/py passes for sandbox (which invokes the dm clk test amongst
others).
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Acked-by: Simon Glass <sjg@chromium.org>
This adds a sandbox reset implementation (provider), a test client
device, instantiates them both from Sandbox's DT, and adds a DM test
that excercises everything.
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Acked-by: Simon Glass <sjg@chromium.org>
A reset controller is a hardware module that controls reset signals that
affect other hardware modules or chips.
This patch defines a standard API that connects reset clients (i.e. the
drivers for devices affected by reset signals) to drivers for reset
controllers/providers. Initially, DT is the only supported method for
connecting the two.
The DT binding specification (reset.txt) was taken from Linux kernel
v4.5's Documentation/devicetree/bindings/reset/reset.txt.
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Acked-by: Simon Glass <sjg@chromium.org>
Tegra186's HSP module implements doorbells, mailboxes, semaphores, and
shared interrupts. This patch provides a driver for HSP, and hooks it
into the mailbox API. Currently, only doorbells are supported.
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Acked-by: Simon Glass <sjg@chromium.org>
Rename mailbox*.h to match the naming convention requested during review
of the new reset subsystem.
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Acked-by: Simon Glass <sjg@chromium.org>
The documentation of parameters in arch/sandbox/include/asm/gpio.h is
either missing or faulty.
This patch corrects the documentation.
Signed-off-by: Mario Six <mario.six@gdsys.cc>
Acked-by: Simon Glass <sjg@chromium.org>
A number of style fixes across the files in this directory, including:
* Correct invalid kernel-doc content.
* Tidy up massive comment in fdt_region.c.
* Use correct spelling of "U-Boot".
* Replace tests of "! <var>" with "!<var>".
* Replace "libfdt_env.h" with <libfdt_env.h>.
Signed-off-by: Robert P. J. Day <rpjday@crashcourse.ca>
Acked-by: Simon Glass <sjg@chromium.org>
A bug in the pca953x driver prevents correct reading of GPIO input
values beyond the 8th GPIO; all values are reported as zero. Setting of
GPIO output values is not affected.
This patch fixes the reading behavior.
Signed-off-by: Mario Six <mario.six@gdsys.cc>
Reviewed-by: Peng Fan <van.freenix@gmail.com>
Acked-by: Simon Glass <sjg@chromium.org>
Some architectures as m68k still need to use CONFIG_NEEDS_MANUAL_RELOC,
and are not still using the device tree.
Signed-off-by: Angelo Dureghello <angelo@sysam.it>
Acked-by: Simon Glass <sjg@chromium.org>
I do not insist on CONFIG_BOOTDELAY=3. The default value in Kconfig,
CONFIG_BOOTDELAY=2, is just fine for these boards.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
The DRAM PHY layer on PH1-LD20 is able to calibrate PHY parameters
periodically. This compensates for the voltage and temperature
deviation and improves the PHY parameter adjustment. Instead, it
requires 64 byte scratch memory in each DRAM channel for the dynamic
training. The memory regions must be reserved in DT before jumping
to the kernel.
The scratch area can be anywhere in each DRAM channel, but the DRAM
init code in SPL currently assigns it at the end of each channel.
So, it makes sense to reserve the regions on run-time by U-Boot
instead of statically embedding it in the DT in Linux. Anyway,
a boot-loader should know much more about memory initialization
than the kernel.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
This will make it easier to select config options specific to
particular ARM processor generation.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
At first, 256 byte of the head of DRAM space was reserved for some
reasons. However, as the progress of development, it turned out
unnecessary, and it was never used in the end. Move the CPU release
address to leave no space.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Correct the error message in spl_load_image_ext() when image parsing
fails. Instead of "ext4fs_read failed" print "failed to parse image
header".
Signed-off-by: Petr Kulhavy <brain@jikos.cz>
CC: Guillaume GARDET <guillaume.gardet@free.fr>
CC: Tom Rini <trini@konsulko.com>
Fix the following build errors when building sandbox on x86 32-bit:
In file included from fs/cbfs/cbfs.c:8:0:
include/malloc.h:364:7: error: conflicting types for 'memset'
void* memset(void*, int, size_t);
^
In file included from include/compiler.h:123:0,
from include/cbfs.h:10,
from fs/cbfs/cbfs.c:7:
include/linux/string.h:78:15: note: previous declaration of 'memset' was here
extern void * memset(void *,int,__kernel_size_t);
^
In file included from fs/cbfs/cbfs.c:8:0:
include/malloc.h:365:7: error: conflicting types for 'memcpy'
void* memcpy(void*, const void*, size_t);
^
In file included from include/compiler.h:123:0,
from include/cbfs.h:10,
from fs/cbfs/cbfs.c:7:
include/linux/string.h:81:15: note: previous declaration of 'memcpy' was here
extern void * memcpy(void *,const void *,__kernel_size_t);
^
scripts/Makefile.build:280: recipe for target 'fs/cbfs/cbfs.o' failed
Signed-off-by: Guillaume GARDET <guillaume.gardet@free.fr>
Cc: Tom Rini <trini@konsulko.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
CONFIG_VERSION_VARIABLE isn't always defined, so we can't simply look up
its value directly, or an exception will occur if it isn't defined.
Instead, we must use .get() to supply a default value if the variable
isn't defined.
Fixes: da37f006e7 ("tests: py: disable main_signon check for printenv cmd")
Acked-by: Heiko Schocher <hs@denx.de>
Signed-off-by: Stephen Warren <swarren@nvidia.com>
