The config option doesn't make any sense anymore
when building a multiimage barebox. With a proper
DT built into the image we don't need the ODMdata
mechanism to find the debug UART anymore.
Signed-off-by: Lucas Stach <dev@lynxeye.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Sometimes a power domain didn't properly power up,
reading back the command register seems to fix this
by flushing the write.
Signed-off-by: Lucas Stach <dev@lynxeye.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
In order to use some devices we first have to power
up their power domain. Add support to do this in a
generic way.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
- Remove OFDEVICE dependency from PINCTRL. It won't do
much then, so add a comment to Kconfig when PINCTRL is
selected without OFDEVICE
- Let Architectures only select PINCTRL instead of the
particular driver. Change the drivers to 'default y if $SOC'
to make sure the drivers are still compiled if the corresponding
SoC is selected
This fixes Kconfig warnings like:
warning: (PINCTRL_ARMADA_370 && PINCTRL_ARMADA_XP && PINCTRL_DOVE && PINCTRL_KIRKWOOD) selects PINCTRL which has unmet direct dependencies (OFDEVICE)
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Acked-by: Sebastian Hesselbarth <sebastian.hesselbarth@gmail.com>
We can reuse the Tegra30 pinctrl driver, as the bit
layout is the same. Just add the pin and drivegroups
and some compile-time magic to avoid bloat.
Signed-off-by: Lucas Stach <dev@lynxeye.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
If the bootloader doesn't init the architectural timer
on Cortex A15 Linux falls over when trying to boot.
Signed-off-by: Lucas Stach <dev@lynxeye.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
I don't know why get_runtime_offset fails on T124 yet,
but this is a safe workaround, with the nice side-effect
of fixing second stage barebox loading.
Signed-off-by: Lucas Stach <dev@lynxeye.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Don't disable clk to unrelated devices in the process.
Signed-off-by: Lucas Stach <dev@lynxeye.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Those 3 are needed to power CPU0 from the CPUG cluster.
Signed-off-by: Lucas Stach <dev@lynxeye.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Even the lowlevel functions are growing to a
size where having a stack seem beneficial.
Signed-off-by: Lucas Stach <dev@lynxeye.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
In RCM aka recovery mode the BootROM waits for a
usbloader to take over control. On most boards this
is triggered by holding a physical switch which may
be inconvinient at times. Add a command to switch
into RCM from software.
Signed-off-by: Lucas Stach <dev@lynxeye.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
This isn't much different from the default 1,16V
and I haven't seen this make a difference on any
board, but it seems to be required for some T30 SKUs.
Signed-off-by: Lucas Stach <dev@lynxeye.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
It's not safe to ramp up the CPU clock speed to
1,4 GHz on all T30 SKUs, as this may result in failure
to start the kernel properly. Start CPU at 600 MHz,
which is safe even for the slowest SKUs.
Signed-off-by: Lucas Stach <dev@lynxeye.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Mostly to make it clear that boarddata needs to be
something we can dereference.
As this is a pretty invasive change, use the opportunity
to make the signature 64bit safe.
Signed-off-by: Lucas Stach <l.stach@pengutronix.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
We need to reprogram PLL_P at a later time, so
we have to make sure MSELECT is able to operate
correctly when we stop PLL_P.
Signed-off-by: Lucas Stach <dev@lynxeye.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
fno-jump-tables isn't enough to guard against
gcc switch optimizations that are unsafe to use
in code that runs before relocation.
The switch-tree-conversion opt pass may generate
lookup tables that are placed in the data section
and accessed via absolute adressing, which fails
prior to relocation.
Signed-off-by: Lucas Stach <dev@lynxeye.de>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
The current implementation fakes a memory-mapped I/O device
at 0x3f8 and 0x2f8, then uses platform read/write functions
to do the actual reading and writing. These platform functions
only exist for the x86 platform; better to move the I/O
routines into the driver and have the driver request I/O ports
using request_ioport_region.
Signed-off-by: Michel Stam <michel@reverze.net>
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>