This patch adds very basic minimal support for the Marvell Octeon 3
CN73xx based EBB7304 EVK. Please note that the basic Octeon port does
not support DDR3/4 initialization yet. To still use U-Boot on with this
port, the L2 cache (4MiB) is used as RAM. This way, U-Boot can boot
to the prompt on this board.
Supported devices:
- UART
- reset
- CFI parallel NOR flash
Signed-off-by: Stefan Roese <sr@denx.de>
This patch adds the code to copy itself from bootrom location to a
different location (TEXT_BASE) to the Octeon platform. Its used in
this case to copy the complete U-Boot image into L2 cache, which
greatly improves the bootup time - especially in regard to the
very long and complex DDR4 init code.
The Kconfig symbol CONFIG_MIPS_MACH_EARLY_INIT is enabled with this
patch for Octeon.
Signed-off-by: Stefan Roese <sr@denx.de>
This patch adds very basic support for the Octeon III SoCs. Only
CFI parallel NOR flash and UART is supported for now.
Please note that the basic Octeon port does not include the DDR3/4
initialization yet. This will be added in some follow-up patches
later. To still use U-Boot on with this port, the L2 cache (4MiB on
Octeon III CN73xx) is used as RAM. This way, U-Boot can boot to the
prompt on such boards.
Signed-off-by: Aaron Williams <awilliams@marvell.com>
Signed-off-by: Stefan Roese <sr@denx.de>
This patch adds the optional call to mips_mach_early_init() to start.S
at a very early stage. Its disabled per default. It can be used for
very early machine / platform specific init code. Its called very
early and at this stage the PC is allowed to differ from the linking
address (CONFIG_TEXT_BASE) as no absolute jump has been performed until
this call.
It will be used by thje Octeon platform.
Signed-off-by: Stefan Roese <sr@denx.de>
Reviewed-by: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
Sync asm/mipsregs.h with Linux 5.7. Also replace the custom
symbols EBASE_CPUNUM and EBASE_WG with the according symbols
from Linux.
Signed-off-by: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
Reviewed-by: Stefan Roese <sr@denx.de>
Tested-by: Stefan Roese <sr@denx.de>
Sync asm/addrspace.h with Linux 5.7
Signed-off-by: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
Reviewed-by: Stefan Roese <sr@denx.de>
Tested-by: Stefan Roese <sr@denx.de>
Sync asm/asm.h with Linux 5.7.
Signed-off-by: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
Reviewed-by: Stefan Roese <sr@denx.de>
Tested-by: Stefan Roese <sr@denx.de>
This macro only served as a wrapper for CKSEG1ADDR() with an
exception for CONFIG_TB0229. CONFIG_TB0229 doesn't exist, thus
use CKSEG1ADDR() directly.
This also prepares for an upcoming asm header sync with Linux.
Signed-off-by: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
Reviewed-by: Stefan Roese <sr@denx.de>
Tested-by: Stefan Roese <sr@denx.de>
Add an own Kconfig symbol for the initial disabling of caches
invoked from generic start code.
Also add an own Kconfig symbols for the initialization of caches
invoked from generic start code.
Until now both code paths could only be disabled with
CONFIG_SKIP_LOWLEVEL_INIT. But this is not flexible enough for
RAM boot scenarios like EJTAG or SPL payload or for machines
which don't require cache initialization or which want to
provide their own cache implementation.
Signed-off-by: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
Reviewed-by: Stefan Roese <sr@denx.de>
Tested-by: Stefan Roese <sr@denx.de>
Logically this code belongs to cache_init.S.
If a complex SoC needs to replace the generic cache init,
mips_cache_disable() can now be called from custom start.S files.
Signed-off-by: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
Reviewed-by: Stefan Roese <sr@denx.de>
Tested-by: Stefan Roese <sr@denx.de>
change_k0_cca() is called multiple times. Move the code for
changing to KSEG1 to a macro to avoid code duplication.
Also fix missing change to KSEG1 when changing to CONF_CM_CACHABLE_COW.
Signed-off-by: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
Reviewed-by: Stefan Roese <sr@denx.de>
Tested-by: Stefan Roese <sr@denx.de>
Since commit 703ec9ddf9 ("MIPS: Stop building position independent code")
the relocation code was completely reworked and removed from start.S.
Remove some left-overs of the old code.
Signed-off-by: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
Reviewed-by: Stefan Roese <sr@denx.de>
Tested-by: Stefan Roese <sr@denx.de>
This patch adds __weak to invalidate_dcache_range() in lib/cache.c. This
makes it possible to overwrite this function by a platforms specific
version, which will be done for Octeon.
Signed-off-by: Stefan Roese <sr@denx.de>
This patch adds __weak to flush_cache() in lib/cache.c. This makes it
possible to overwrite this function by a platforms specific version,
like done with the Octeon base port.
Signed-off-by: Stefan Roese <sr@denx.de>
This patch enables the usage of CONFIG_MIPS_L2_CACHE without
CONFIG_MIPS_CM, which is what is needed for the newly added Octeon
platform.
Signed-off-by: Stefan Roese <sr@denx.de>
This patch changes the R_MIPS_NONE define from 0 to a magic value. This
makes it possible to better detect any forbidden pre-relocation usage
of BSS variables, as they are often zero'ed and then relocation is
stopped too early.
Additionally the error message is improved to also print the faulting
address. This helps finding the root-cause for this breakage by
comparing this address with the values in System.map.
This patch helps a lot when working on pre-relocation code, like the
Octeon DDR init code, where such variables have hit me multiple times
now.
Signed-off-by: Stefan Roese <sr@denx.de>
Cc: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
Cc: Aaron Williams <awilliams@marvell.com>
Cc: Chandrakala Chavva <cchavva@marvell.com>
This patch opts-out the compilation of get_tbclk() if
CONFIG_SYS_MIPS_TIMER_FREQ is not defined. This is used on the Octeon
platform, where the weak get_tbclk() function is overwritten by its
platform specific one.
Signed-off-by: Stefan Roese <sr@denx.de>
We should not use typedefs in U-Boot. They cannot be used as forward
declarations which means that header files must include the full header to
access them.
Drop the typedef and rename the struct to remove the _s suffix which is
now not useful.
This requires quite a few header-file additions.
Signed-off-by: Simon Glass <sjg@chromium.org>
Move this header out of the common header. Network support is used in
quite a few places but it still does not warrant blanket inclusion.
Note that this net.h header itself has quite a lot in it. It could be
split into the driver-mode support, functions, structures, checksumming,
etc.
Signed-off-by: Simon Glass <sjg@chromium.org>
Small patch to add support for VoCore/VoCore2 board.
VoCore is open hardware and runs OpenWrt/LEDE.
It has WIFI, USB, UART, 20+ GPIOs but is only one inch square.
It will help you to make a smart house, study embedded system
or even make the tiniest router in the world.
Details about this SoM can be found at "https://vocore.io/v2.html".
Signed-off-by: Mauro Condarelli <mc5686@mclink.it>
Reviewed-by: Stefan Roese <sr@denx.de>
The current implementations have some issues detecting the correct
values:
dqs_find_max() will return "last passing fieldval + 1" instead of
"last passing fieldval". Also it will return "maxval + 1" in the
case that all fieldvals are tested valid (without error).
dqs_find_min() will not test the "lowest" value because of using ">"
instead of ">=".
This patch now rewrites these functions to fix those issues. Also,
this patch uses the same approach of a for loop in both functions making
it easier to read and maintain.
Since the variables are integers now, we can use min()/max(), which
handles the wrap around case for fieldval=0: return (0 - 1).
Signed-off-by: Stefan Roese <sr@denx.de>
Cc: Weijie Gao <weijie.gao@mediatek.com>
Cc: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
Reviewed-by: Weijie Gao <weijie.gao@mediatek.com>
This change is made to enable comparison of integer variables, which
might be negative in the next patch. No functional change is intended
in this patch.
Signed-off-by: Stefan Roese <sr@denx.de>
Cc: Weijie Gao <weijie.gao@mediatek.com>
Cc: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
Reviewed-by: Weijie Gao <weijie.gao@mediatek.com>
This function returns "-1" (true) upon error. So the function name does
not match its implementation which is confusing. This patch renames the
function to dqs_test_error() which makes the code easier to read.
Also change the return type to bool and return "true" or "false".
Signed-off-by: Stefan Roese <sr@denx.de>
Cc: Weijie Gao <weijie.gao@mediatek.com>
Cc: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
Reviewed-by: Weijie Gao <weijie.gao@mediatek.com>
This patch adds support for mt7628 reference board. SPL_DM and DT are not
enabled for SPL to save about 17KiB for u-boot-spl.bin.
Reviewed-by: Stefan Roese <sr@denx.de>
Signed-off-by: Weijie Gao <weijie.gao@mediatek.com>
This patch enables SPL for all mtmips boards. And also remove defconfig
files which are intend to build ram bootable u-boot files.
SPL_DM and OF_CONTROL are enabled for both boards.
Reviewed-by: Stefan Roese <sr@denx.de>
Signed-off-by: Weijie Gao <weijie.gao@mediatek.com>
This patch adds SPL support for mtmips platform. The lowlevel architecture
is split into SPL and the rest parts are built into a memory loadable
u-boot image. Optional SPL_DM and OF_CONTROL are also supported.
The increment of size is very small (< 10 KiB) if SPL_DM and OF_CONTROL are
not enabled and the memory bootable u-boot (u-boot.img) is generated
automatically so there is not need to add a separate config for it.
A lzma compressed payload (u-boot-lzma.img) is also generated and it will
be combined with u-boot-spl.bin to form the unified ROM bootable binary
u-boot-mtmips.bin.
A spl loader is added to support uncompress the payload.
Reviewed-by: Stefan Roese <sr@denx.de>
Signed-off-by: Weijie Gao <weijie.gao@mediatek.com>
This patch rewrites the mtmips architecture with the following changes:
1. Move MT7628 soc parts into a subfolder.
2. Lock parts of D-Cache as temporary stack.
3. Reimplement DDR initialization in C language.
4. Reimplement DDR calibration in a clear logic.
5. Add full support for auto size detection for DDR1 and DDR2.
6. Use accurate CPU clock depending on the input xtal frequency for timer
and delay functions.
Note:
print_cpuinfo() has incompatible parts with MT7620 so it's moved into
mt7628 subfolder.
Reviewed-by: Stefan Roese <sr@denx.de>
Reviewed-by: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
Signed-off-by: Weijie Gao <weijie.gao@mediatek.com>
The initial stack on some platforms is too small to hold a large malloc
space. This patch adds a option to allow these platforms not reserving the
malloc space on initial stack. These platforms should set the malloc base
after DRAM is usable.
Reviewed-by: Stefan Roese <sr@denx.de>
Signed-off-by: Weijie Gao <weijie.gao@mediatek.com>
Currently CONFIG_MIPS_INIT_STACK_IN_SRAM assumes the memory space for the
initial stack can be used directly. However on some platform the SRAM needs
initialization, e.g. lock cache.
This patch adds an option to allow a new function mips_sram_init() being
called before setup_stack_gd.
Reviewed-by: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
Reviewed-by: Stefan Roese <sr@denx.de>
Signed-off-by: Weijie Gao <weijie.gao@mediatek.com>
This patch adds a new pinmux for UART2, which shares the pins with SPIS.
Reviewed-by: Stefan Roese <sr@denx.de>
Signed-off-by: Weijie Gao <weijie.gao@mediatek.com>
SoCs of mtmips can use different CPU frequencies depending on the HW/SW
configurations. For example mt7628 uses 580MHz clock if the input xtal
frequency is 40MHz, and 575MHz clock if the xtal is 25MHz. Upon cold boot
the CPU uses the xtal frequency directly.
So hardcoding the timer frequency (half of the CPU frequency) in
CONFIG_SYS_MIPS_TIMER_FREQ is not a good idea for this case.
This patch adds a mtmips-specific field timer_freq to arch_global_data.
This field will be used later in mtmips-specific get_tbclk() to provide
accurate timer frequency in different boot stage.
Reviewed-by: Stefan Roese <sr@denx.de>
Reviewed-by: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
Signed-off-by: Weijie Gao <weijie.gao@mediatek.com>
This patch replaces sysreset-syscon with sysreset-resetctrl for mt7628 soc.
Reviewed-by: Stefan Roese <sr@denx.de>
Reviewed-by: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
Signed-off-by: Weijie Gao <weijie.gao@mediatek.com>
This patch adds a MIPS specific jump_to_image_no_args() implementation,
which flushes the U-Boot proper image loaded from the boot device in
SPL before jumping to it.
It has been noticed on MT76x8, that this cache flush is needed. Other
MIPS platforms might need it as well.
Signed-off-by: Stefan Roese <sr@denx.de>
Cc: Weijie Gao <weijie.gao@mediatek.com>
Cc: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
Cc: Simon Goldschmidt <simon.k.r.goldschmidt@gmail.com>
Reviewed-by: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
u_boot_list is not only used by DM, but also by some SPL image load methods
such as spl_nor.c.
This patch adds an option CONFIG_SPL_LOADER_SUPPORT in conjunction with
CONFIG_SPL_DM surrounding the u_boot_list section to make sure SPL image
loaders can be correctly built into u-boot SPL without DM enabled.
Reviewed-by: Stefan Roese <sr@denx.de>
Reviewed-by: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
Signed-off-by: Weijie Gao <weijie.gao@mediatek.com>
If CONFIG_SPL_OF_CONTROL is enabled for SPL and CONFIG_OF_SEPARATE is also
enabled, the dtb will be appended to the u-boot-spl.bin.
When calling dm_init_and_scan() in SPL, fdtdec_setup() will try to locate
dtb at the end of u-boot-spl.bin, by referencing to _image_binary_end.
However _image_binary_end is currently missing in u-boot-spl.lds.
This patch adds _image_binary_end to u-boot-spl.lds to make sure linking
u-boot-spl will not fail.
Reviewed-by: Stefan Roese <sr@denx.de>
Reviewed-by: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
Signed-off-by: Weijie Gao <weijie.gao@mediatek.com>
When setting up initial stack, global data will also be put in the stack,
and being cleared.
The assembler instructions for clearing gd is as follows:
move t0, k0
1:
PTR_S zero, 0(t0)
blt t0, t1, 1b
PTR_ADDIU t0, PTRSIZE
t0 is the start address of gd, t1 is the end address of gd (t0 + GD_SIZE).
[PTR_ADDIU t0, PTRSIZE] is in the delay slot of [blt t0, t1, 1b], so it
will be executed before the branch operation.
However the comparison for the BLT instruction is done before executing the
delay slot. This means when the last word just before k1 is cleared, the
loop will continue to run once. This will clear an extra word at k1, which
is outside the global data.
Global data is placed at the top of the stack. If the initial stack is a
SRAM or locked cache, the area outside them may be inaccessible. A write
operation performed in this area may cause an exception.
To solve this, [PTR_ADDIU t0, PTRSIZE] should be placed before the BLT
instruction.
Reviewed-by: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
Reviewed-by: Stefan Roese <sr@denx.de>
Signed-off-by: Weijie Gao <weijie.gao@mediatek.com>
Both mt7620 and mt7628 has the same cache configuration. There is no need
to use CONFIG_SYS_CACHE_SIZE_AUTO to probe it at runtime.
Add them into Kconfig to reduce some code size.
Reviewed-by: Stefan Roese <sr@denx.de>
Signed-off-by: Weijie Gao <weijie.gao@mediatek.com>
In U-Boot the exception vector base will be moved to top of memory, to be
used to display register dump when exception occurs.
But some old linux kernel does not honor the base set in CP0_EBASE. A
modified exception vector base will cause kernel crash.
This patch adds an option to enable reset exception vector base to its
previous value, or a user configured value before booting linux kernel.
Reviewed-by: Daniel Schwierzeck <daniel.schwierzeck@gmail.com>
Signed-off-by: Weijie Gao <weijie.gao@mediatek.com>
