The L4T kernel complains about a CSITE clock rate above 144MHz, presumably
because the HW is only characterized for a clock less than that. Adjust the
rate to 136MHz to avoid the warning and stay in spec.
Signed-off-by: Bryan Wu <pengw@nvidia.com>
(swarren, re-wrote commit description)
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Signed-off-by: Tom Warren <twarren@nvidia.com>
Trimslice currently stores its environment at 512KiB into the SPI flash
chip. The U-Boot binary has grown such that the size of the boot image
(which includes the Tegra BCT, padding, and the U-Boot binary) is slightly
larger than 512K now. Consequently, writing the boot image to flash
corrupts the saved environment, and equally, writing to or erasing the
environment will corrupt the bootloader, which in turn will cause the
Tegra boot ROM to enter recovery mode during boot, making it look as if
the system is non-operational. Note that tegra-uboot-flasher writes to
the environment during the flashing process.
Solve this by moving the environment as high as possible in flash. This
will allow the U-Boot binary to roughly double in size before this problem
is hit again, at which point there's nothing we can do anyway since the
binary won't fit into flash.
99% of other Tegra boards store the environment in eMMC and use a negative
value for CONFIG_ENV_OFFSET, which already automatically places the
environment as near the end of boot flash as possible. The 1 remaining
board hard-codes CONFIG_ENV_OFFSET to 2MiB, which allows for plenty more
bloat.
Reported-by: Stephen L Arnold <nerdboy@gentoo.org>
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Tom Warren <twarren@nvidia.com>
Currently, ft_system_setup() is implemented by board*.c, which are a bit
of a dumping ground for a bunch of unrelated functionality, and separate
versions exist for pre-Tegra186 and Tegra186. Move the implementation into
a separate file to separate functionality, and allow sharing.
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Tom Warren <twarren@nvidia.com>
p2771-0000 has a couple of PCIe ports; one physically x4 desktop PCI
connector (which may run at x2 electrically, depending on the board
version and configuration) and a x1 connection to the M.2 slot (which may
not be active, depending on the board version and configuration). This
change enables those.
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Signed-off-by: Tom Warren <twarren@nvidia.com>
Now that clock and reset drivers exist for Tegra186, we can enable the SD
card controller. Now that a BPMP I2C driver exists for Tegra186, we can
communicate with the PMIC to enable power to the SD card. Hook up the DT
content and board code required to make the SD card work.
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Signed-off-by: Tom Warren <twarren@nvidia.com>
clk/reset API was tested on T186 platform and previous chip like
T210/T124 will still use the old APIs.
Signed-off-by: Bryan Wu <pengw@nvidia.com>
(swarren, simplified some ifdefs, removed indent level inside an ifdef)
(swarren, added comment about the ifdefs)
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Heiko Schocher <hs@denx.de>
Signed-off-by: Tom Warren <twarren@nvidia.com>
Tegra186 supports the new standard clock, reset, and power domain APIs.
Older Tegra SoCs still use custom APIs. Enhance the Tegra PCIe driver so
that it can operate with either set of APIs.
On Tegra186, the BPMP handles all aspects of PCIe PHY (UPHY) programming.
Consequently, this logic is disabled too.
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Tom Warren <twarren@nvidia.com>
Tegra186 supports the new standard clock and reset APIs. Older Tegra SoCs
still use custom APIs. Enhance the Tegra MMC driver so that it can operate
with either set of APIs.
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Tom Warren <twarren@nvidia.com>
On Tegra186, some I2C controllers are directly controlled by the main CPU,
whereas others are controlled by the BPMP, and can only be accessed by the
main CPU via IPC requests to the BPMP. This driver covers the latter case.
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Heiko Schocher <hs@denx.de>
Signed-off-by: Tom Warren <twarren@nvidia.com>
In Tegra186, SoC power domains are manipulated using IPC requests to
the BPMP (Boot and Power Management Processor). This change implements a
driver that does that.
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Tom Warren <twarren@nvidia.com>
In Tegra186, on-SoC reset signals are manipulated using IPC requests to
the BPMP (Boot and Power Management Processor). This change implements a
driver that does that. It is unconditionally selected by CONFIG_TEGRA186
since virtually any Tegra186 build of U-Boot will need the feature.
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Tom Warren <twarren@nvidia.com>
In Tegra186, on-SoC clocks are manipulated using IPC requests to the BPMP
(Boot and Power Management Processor). This change implements a driver
that does that. A tegra/ sub-directory is created to follow the existing
pattern. It is unconditionally selected by CONFIG_TEGRA186 since virtually
any Tegra186 build of U-Boot will need the feature.
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Tom Warren <twarren@nvidia.com>
The Tegra BPMP (Boot and Power Management Processor) is a separate
auxiliary CPU embedded into Tegra to perform power management work, and
controls related features such as clocks, resets, power domains, PMIC I2C
bus, etc. This driver provides the core low-level communication path by
which feature-specific drivers (such as clock) can make requests to the
BPMP. This driver is similar to an MFD driver in the Linux kernel. It is
unconditionally selected by CONFIG_TEGRA186 since virtually any Tegra186
build of U-Boot will need the feature.
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: Tom Warren <twarren@nvidia.com>
The call op requests that the callee pass a message to the underlying HW
or device, wait for a response, and then pass back the response error code
and message to the callee. It is useful for drivers that represent some
kind of messaging or IPC channel to a remote device.
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Acked-by: Simon Glass <sjg@chromium.org>
The voltage and control registers need to be looked up from the value in
driver_data. Adjust the get_value and get_enable functions to match the
corresponding set_* functions.
Signed-off-by: John Keeping <john@metanate.com>
Acked-by: Simon Glass <sjg@chromium.org>
Some code may want to read reg values from DT, but from nodes that aren't
associated with DM devices, so using dev_get_addr_index() isn't
appropriate. In this case, fdtdec_get_addr_size_*() are the functions to
use. However, "translation" (via the chain of ranges properties in parent
nodes) may still be desirable. Add a function parameter to request that,
and implement it. Update all call sites to default to the original
behaviour.
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Squashed in build fix from Stephen:
Signed-off-by: Simon Glass <sjg@chromium.org>
The next patch will call fdt_translate_address() from somewhere with a
"const void *blob" rather than a "void *blob", so fdt_translate_address()
must accept a const pointer too. Constify the minimum number of function
parameters to achieve this.
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Acked-by: Simon Glass <sjg@chromium.org>
Squashed in build fix from Stephen:
Signed-off-by: Simon Glass <sjg@chromium.org>
Linux stopped the use of keyword 'boolean' in Kconfig.
Refer to commit 6341e62b212a2541efb0160c470e90bd226d5496 ("kconfig:
use bool instead of boolean for type definition attributes")
in Linux Kernel.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Enable eth driver model for am43xx_evm as cpsw supports
driver model.
This was already added with the commit bc705ea1cf but with
commit 4c4e3b3775 to add fit support CONFIG_DM_ETH was missed.
Signed-off-by: Mugunthan V N <mugunthanvnm@ti.com>
Cc: Lokesh Vutla <lokeshvutla@ti.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
cpsw tries to flush dcache which is not in the range of PKTALIGN.
Because of this the following warning comes while flushing:
CACHE: Misaligned operation at range [dffecec0, dffed016]
Fix it by flushing cache of size aligned to PKTALIGN.
Signed-off-by: Lokesh Vutla <lokeshvutla@ti.com>
Enable eth driver model for dra7xx_evm as cpsw supports
driver model.
This was already added with the commit 641b936fa5 but with
commit bd7245849f to add fit support CONFIG_DM_ETH was missed.
Signed-off-by: Mugunthan V N <mugunthanvnm@ti.com>
Cc: Lokesh Vutla <lokeshvutla@ti.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
Acked-by: Lokesh Vutla <lokeshvutla@ti.com>
Enable DM based regulator framework and also fixed regulator support as
some IPs like mmc use regulators for there functioning.
Signed-off-by: Vignesh R <vigneshr@ti.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
Add a node for evm_3v3_sd using onboard PCF GPIO expander which feeds
on to mmc vdd.
Update mapping for vmmc-supply and vmmc_aux-supply.
evm_3v3_sd supplies to SD card vdd, and ldo1 to sdcard i/o lines.
Signed-off-by: Vignesh R <vigneshr@ti.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
This commit allows injecting a board/platform/device-specific post-
processing function into the FIT image data loading process, which can
include modifying the size and altering the starting source address of
an image data artifact. This might be desired to do things like strip
headers or footers attached to the images before they were packaged into
the FIT, or to perform operations such as decryption or authentication.
Introduce new configuration option CONFIG_FIT_IMAGE_POST_PROCESS to
allow controlling this feature. If enabled, a platform-specific post-
process function must be provided.
Signed-off-by: Andreas Dannenberg <dannenberg@ti.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Create drivers/sysreset and move sysreset-uclass and all sysreset
drivers there.
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
Acked-by: Simon Glass <sjg@chromium.org>
Flush operations need to be cacheline aligned to take effect, make
sure to flush always complete cachelines. This avoids messages such
as:
CACHE: Misaligned operation at range [00900000, 009004d9]
Signed-off-by: Stefan Agner <stefan.agner@toradex.com>
Tested-by: Fabio Estevam <fabio.estevam@nxp.com>
The i2c uclass has a default setting for per_child_platdata_auto_alloc_size
so drivers do not need to set it. Remove this from drivers to avoid
confusion.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
This outer cache allows to control active ways independently for
each CPU, so this function will be useful to set up active ways
for a specific CPU.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
The UniPhier outer cache (L2 cache on ARMv7 SoCs) has 128 byte line
length and its tags are also managed per 128 byte line.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Move this option to Kconfig, renaming it into CONFIG_CACHE_UNIPHIER.
The new option name makes sense enough, and the same as Linux has.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Now, all of these macros are only used in cache-uniphier.c, so
there is no need to export them in a header file.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
The DRAM is available at this point, so setup the temporary stack
and call the C function to reduce the code duplication a bit.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
The System Cache (outer cache) is used not only as L2 cache,
but also as locked SRAM. The functions for turning on/off it
is necessary whether the L2 cache is enabled or not.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
As the sLD3 Boot ROM has a complex page table, it is difficult to
set up the debug UART with enabling it. It will be much easier to
initialize the UART port after switching over to the straight-mapped
page table.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Commit 4b50369fb5 ("ARM: uniphier: create early page table at
run-time") broke the ROM boot mode for PH1-sLD3 SoC, because the
run-time page table creation requires the outer cache register
access but the page table in the sLD3 Boot ROM does not straight-map
virtual/physical addresses.
The idea here is to check the current page table to determine if
it is a straight map table. If not, adjust the outer cache register
base.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Here, the ldr pseudo-instruction falls into the ldr + data set.
The register access by [r1, #offset] produces shorter code.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
If CONFIG_UNIPHIER_L2CACHE_ON is undefined, the L2 cache is never
enabled, so there is no need for v7_outer_cache_disable(). The weak
stub avoids the compile error anyway.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
The UniPhier outer cache (L2 cache on ARMv7 SoCs) can be used as
SRAM by locking ways.
These functions will be used to transfer the trampoline code for SMP
into the locked SRAM.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Unify the range/all operation routines into the common function,
uniphier_cache_maint_common(), and sync code with Linux a bit more.
This reduces the code duplication.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Commit "ecc3066 Fix board init code to respect the C runtime environment"
broke platform support for ppc4xx.
start.S prepares a stackframe that is later rendered unusable by appending
the reserved space for global data.
Instead the reserved space has to be put first. Then the stackframe can
be pushed.
I can only test the 405EP OCM case. At least all other ppc4xx boards still
build.
Signed-off-by: Dirk Eibach <dirk.eibach@gdsys.cc>
Signed-off-by: Stefan Roese <sr@denx.de>
Bryan Wu
Stephen Warren
Tom Rini
John Keeping
Masahiro Yamada
Mugunthan V N
Lokesh Vutla
Vignesh R
Andreas Dannenberg
Max Filippov
Stefan Agner
Simon Glass
Alban Bedel
Dirk Eibach