This adds a message to lists_bind_fdt when it skips initializing a device
pre-relocation. I've had a couple errors where a device didn't initialize
properly because one of its dependencies was missing.
Signed-off-by: Sean Anderson <seanga2@gmail.com>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
The suffix should be ".bin" instead of ".dtb" .
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
Reviewed-by: Bin Meng <bmeng.cn@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
At present we only support two phases where devres can be used:
bind and probe. This is handled with a boolean. We want to add a new
phase (platdata), so change this to an enum.
Signed-off-by: Simon Glass <sjg@chromium.org>
The devres functionality has very few users in U-Boot, but it still should
have tests. Add a few basic tests of the main functions.
Signed-off-by: Simon Glass <sjg@chromium.org>
At present when CONFIG_DEBUG_DEVRES is enabled, U-Boot prints log messages
to the console with every devres allocation/free event. This causes most
tests to fail since the console output is not as expected.
In particular this prevents us from adding a device to sandbox which uses
devres in its bind method.
Move devres over to use U-Boot's logging feature instead, and add a new
category for devres.
Signed-off-by: Simon Glass <sjg@chromium.org>
The malloc() implementations provides a way of finding out the approximate
amount of memory that is allocated. Add helper functions to make it easier
to access this and see changes over time. This is useful for tests that
want to check if memory has been allocated or freed.
Signed-off-by: Simon Glass <sjg@chromium.org>
At present these functions are lumped in with the core device functions.
They have their own #ifdef to control their availability, so it seems
better to split them out.
Move them into their own header file.
Signed-off-by: Simon Glass <sjg@chromium.org>
We want to avoid allocating platform data twice. This could happen if
device_probe() is called after device_ofdata_to_platdata() for the same
device.
Add a flag to track whether device_ofdata_to_platdata() has been called on
a device. Check the flag to make sure it doesn't happen twice, and clear
the flag when the data is freed.
Signed-off-by: Simon Glass <sjg@chromium.org>
Add a new internal function, device_ofdata_to_platdata() to handle
allocating private space associated with each device and reading the
platform data from the device tree.
Call this new function from device_probe().
Signed-off-by: Simon Glass <sjg@chromium.org>
At present the parent is probed before the child's ofdata_to_platdata()
method is called. Adjust the logic slightly so that probing parents is
not done until afterwards.
Signed-off-by: Simon Glass <sjg@chromium.org>
This method is supposed to extract platform data from the device tree. It
should be done before the device itself is probed. Move it earlier in the
device_probe() function.
Signed-off-by: Simon Glass <sjg@chromium.org>
At present the clock driver reads its ofdata in the probe() method. This
is not correct although it is often harmless.
However in this case it causes a problem, something like this:
- ast_get_scu() is called (from somewhere) to get the SCI address
- this probes the clock
- first sets up ofdata (which does nothing at present)
- DM marks clock device as active
- DM calls pinctrl
- pinctrl probes and calls ast_get_scu() in ast2500_pinctrl_probe()
- ast_get_scu() probes the clock, but sees it already marked as
probed
- ast_get_scu() accesses the clock's private data, with scu as NULL
- DM calls clock probe function ast2500_clk_probe() which reads scu
By putting the read of scu into the correct method, scu is read as part of
ofdata setup, and everything is OK.
Note: This problem did not matter until now since DM always probed all
parents before reading a child's ofdata. The fact that pinctrl is a child
of clock seems to trigger this strange bug.
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
It is not possible to access a device on a PCI bus that has not yet been
probed, since the bus number is not known. Add a warning to catch this
error.
Signed-off-by: Simon Glass <sjg@chromium.org>
The PCI bus is not actually probed by the time the ofdata_to_platdata()
method is called since that happens in the uclass's post_probe() method.
Update the PMC and P2SB drivers to access the bus in its probe() method.
Signed-off-by: Simon Glass <sjg@chromium.org>
These macros use __FILE__ which inserts the full path of the object file
into U-Boot, thus increasing file size. Drop these usages.
An older version of this patch was submitted here:
http://patchwork.ozlabs.org/patch/1205784/
Signed-off-by: Simon Glass <sjg@chromium.org>
Some U-Boot code uses BUG_ON() and WARN_ON() macros. These use __FILE__
which can include quite a large path, depending on how U-Boot is built.
The existing assert() is only checked if DEBUG is enabled. Add a new one
which is always checked, and prints a (smaller) error in that case.
Signed-off-by: Simon Glass <sjg@chromium.org>
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1
iQEcBAABAgAGBQJeFDfVAAoJEB6zHgIOrC/IKfsH/2cUfXuc9IgA7L90L3g7htlS
zLAQSC8Xk0WbgbmBJh/lq6zFWj2Mu2yfkZqRzwN8bHkmOPq40F+pEAszVtjH77HM
FpSjMSDVWfNXuhmoiS571dehA0TGi1b2kknrAXqpOwVuMRncMCwaPi8yZYZSGlSV
VzMtguQxD0sZg4oMZVde3tCd41smPTPNDjzLZnXdLa0b09+EtW6enGuk3vzXlS/8
QUT68wUzkeoptJhf3pQ3BXNvkFwdNdYJ74uevDmkAYtvxyl10WNykcwYmH50Q9qq
9plF11EizmTHlj16rWpZLx66RP0s4pziKaeGMr6S8uln33Cr9N496wPqtuDBnvA=
=FYPj
-----END PGP SIGNATURE-----
Merge tag 'u-boot-atmel-2020.04-a' of https://gitlab.denx.de/u-boot/custodians/u-boot-atmel
First set of u-boot-atmel features for 2020.04 cycle
This feature set is a patch series from Tudor Ambarus which includes
parsing of the spi flash SFDP parser for SST flashes, and using those
tables to retrieve unique saved per device MAC address. This is then
used as base mac address on the SAMA5D2 Wireless SOM EK board.
It's true that booting normally doesn't take long enough for the
register to roll (which actually happens in a little over an hour, not
just a few seconds). However, the counter starts at power-on, and if
the board is held in reset to be booted over USB, one actually risks
hitting wrap-around during boot, which can both result in too short
delays (if the "st += delay" calculation makes st small) and
theoretically also unbound delays (if st ends up being UINT_MAX and
one just misses sampling digctl_microseconds at that point).
It doesn't take more code to DTRT, and once bitten, twice shy.
Signed-off-by: Rasmus Villemoes <rasmus.villemoes@prevas.dk>
Engicam i.CoreM6 1.5 Quad/Dual MIPI dtsi is reusing fec node
from Engicam i.CoreM6 dtsi but have sampe copy of phy-reset-gpio
and phy-mode properties.
So, drop this phy reset methods from imx6qdl-icore-1.5 dsti file.
Cc: Jacopo Mondi <jacopo@jmondi.org>
Signed-off-by: Michael Trimarchi <michael@amarulasolutions.com>
Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
The EDIMM STARTER KIT i.Core 1.5 MIPI Evaluation is based on
the 1.5 version of the i.Core MX6 cpu module. The 1.5 version
differs from the original one for a few details, including the
ethernet PHY interface clock provider.
With this commit, the ethernet interface works properly:
SMSC LAN8710/LAN8720 2188000.ethernet-1:00: attached PHY driver
While before using the 1.5 version, ethernet failed to startup
do to un-clocked PHY interface:
fec 2188000.ethernet eth0: could not attach to PHY
Similar fix has merged for i.Core MX6Q but missed to update for DL.
Signed-off-by: Michael Trimarchi <michael@amarulasolutions.com>
Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Sync Engicam device tree file from v5.4 linux-next.
Signed-off-by: Michael Trimarchi <michael@amarulasolutions.com>
Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Now all the fec related phy properties are now accessible
via dts. So drop the explicit config items from common
config file, imx6-engicam.h
Signed-off-by: Michael Trimarchi <michael@amarulasolutions.com>
Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
According to the SOM and reference board the clock
can be taken from the external pin or provided from
ENET_REF_CLK. Add a new function that make the proper
set according the board type.
Signed-off-by: Michael Trimarchi <michael@amarulasolutions.com>
Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
Make easy to map fdt file to board in order to use
this information later to apply specific change to
specific board combination.
Signed-off-by: Michael Trimarchi <michael@amarulasolutions.com>
Signed-off-by: Jagan Teki <jagan@amarulasolutions.com>
The data sheet says that the DIV field cannot change while the CLKGATE
bit is set or modified. So do it a little more carefully, by first
clearing the bit, waiting for that to appear, then setting the DIV
field.
Signed-off-by: Rasmus Villemoes <rasmus.villemoes@prevas.dk>
The comment says to clear the bypass bit, but in fact it sets it, thus
selecting ref_xtal. And the next line of code does not set the divider
to 12, but to (the reset value of) 1.
Signed-off-by: Rasmus Villemoes <rasmus.villemoes@prevas.dk>
I tried clearing a bit by writing to hw_clkctrl_gpmi_clr, then
busy-waiting for it to actually clear. My board hung. The data sheet
agrees, these registers do not have _set, _clr, _tog, so fix up the
definitions. git grep -E 'clkctrl_(gpmi|ssp[0-9])_' says that nobody
uses those non-existing ops registers.
Signed-off-by: Rasmus Villemoes <rasmus.villemoes@prevas.dk>
Adjust the environment for booting a mainline kernel by default.
Signed-off-by: Fabio Estevam <festevam@gmail.com>
Reviewed-by: Peng Fan <peng.fan@nxp.com>
Add info about supported i.MX7, improve details the usage of
bcbonly subcommand.
Signed-off-by: Igor Opaniuk <igor.opaniuk@toradex.com>
Reviewed-by: Oleksandr Suvorov <oleksandr.suvorov@toradex.com>
When pmic_get() is used with DM the first parameter must be
the complete node name plus the unit address, so fix it
accordingly.
Reported-by: Igor Opaniuk <igor.opaniuk@toradex.com>
Signed-off-by: Fabio Estevam <festevam@gmail.com>
Reviewed-by: Peng Fan <peng.fan@nxp.com>
When pmic_get() is used with DM the first parameter must be
the complete node name plus the unit address, so fix it
accordingly.
Reported-by: Igor Opaniuk <igor.opaniuk@toradex.com>
Signed-off-by: Fabio Estevam <festevam@gmail.com>
Reviewed-by: Peng Fan <peng.fan@nxp.com>
When pmic_get() is used with DM the first parameter must be
the complete node name plus the unit address, so fix it
accordingly.
Reported-by: Igor Opaniuk <igor.opaniuk@toradex.com>
Signed-off-by: Fabio Estevam <festevam@gmail.com>
Reviewed-by: Peng Fan <peng.fan@nxp.com>
In cases when the same SPL should run on boards with i.MX8MM, that
differ in DDR configuration, it is necessary to try different
parameters and check if the training done by the firmware suceeds or
not.
Therefore we return the DDR training/initialization success to the
upper layer in order to be able to retry with different settings if
necessary.
Signed-off-by: Frieder Schrempf <frieder.schrempf@kontron.de>
The DDR firmware binaries are not part of the U-Boot source code, so
we should look for them in the build directory, where they need to be
copied to before building U-Boot.
The ATF binary is already fetched from the build directory, but the
README files for the i.MX8M EVKs claim that it needs to be copied to
the source directory (which is still true for in-tree builds, but not
in general). Therefore we also fix the READMEs to use the build
directory as the correct location for all additional binary files.
Sined-off-by: Frieder Schrempf <frieder.schrempf@kontron.de>
In the case of i.MX6 and i.MX7 family SoCs it is safe (from an errata
point of view) to use thumb2 by default to save space.
Cc: Stefano Babic <sbabic@denx.de>
Cc: Fabio Estevam <festevam@gmail.com>
Signed-off-by: Tom Rini <trini@konsulko.com>
Reviewed-by: Fabio Estevam <festevam@gmail.com>
Currently imx-specific bootaux command doesn't support ELF format
firmware for Cortex-M4 core.
This patches introduces a PoC implementation of handling elf firmware
(load_elf_image_phdr() was copy-pasted from elf.c just for PoC).
ELF64 binaries isn't supported yet.
This has the advantage that the user does not need to know to which
address the binary has been linked to. However, in order to handle
and load the elf sections to the right address, we need to translate the
Cortex-M4 core memory addresses to primary/host CPU memory
addresses (Cortex A7/A9 cores).
This allows to boot firmwares from any location with just using
bootaux, e.g.:
> tftp ${loadaddr} hello_world.elf && bootaux ${loadaddr}
Similar translation table can be found in the Linux remoteproc
driver [1].
[1] https://elixir.bootlin.com/linux/latest/source/drivers/remoteproc/imx_rproc.c
Signed-off-by: Igor Opaniuk <igor.opaniuk@toradex.com>
Signed-off-by: Stefan Agner <stefan.agner@toradex.com>
Reviewed-by: Oleksandr Suvorov <oleksandr.suvorov@toradex.com>
Sync DTS with the mainline Linux and introduce fec node and
regulator configuration for rn5t567 PMU.
Signed-off-by: Igor Opaniuk <igor.opaniuk@toradex.com>
Reviewed-by: Oleksandr Suvorov <oleksandr.suvorov@toradex.com>
Sync DTS with the mainline Linux and introduce fec node.
Signed-off-by: Igor Opaniuk <igor.opaniuk@toradex.com>
Reviewed-by: Oleksandr Suvorov <oleksandr.suvorov@toradex.com>
Sync DTS with the mainline Linux and introduce fec node.
Signed-off-by: Igor Opaniuk <igor.opaniuk@toradex.com>
Reviewed-by: Oleksandr Suvorov <oleksandr.suvorov@toradex.com>
Take over maintainership for colibri_imx6/imx6ull/t30/vf modules.
Signed-off-by: Igor Opaniuk <igor.opaniuk@toradex.com>
Acked-by: Max Krummenacher <max.krummenacher@toradex.com>
Acked-by: Stefan Agner <stefan.agner@toradex.com>
