SPL on mvebu loads proper U-Boot from custom Marvell kwbimage format and
therefore support for other binary formats is not required to be present in
SPL. Boot source of proper U-Boot is defined by compile time options and
therefore it is not required to enable all possible and unused peripherals
in SPL by default.
This change decrease size of SPL binaries.
Signed-off-by: Pali Rohár <pali@kernel.org>
Reviewed-by: Stefan Roese <sr@denx.de>
This allows to compile U-Boot without some boot option for some A3720 board
which does not have that peripheral.
Signed-off-by: Pali Rohár <pali@kernel.org>
Reviewed-by: Stefan Roese <sr@denx.de>
I am currently maintaing the Methode uDPU and eDPU boards so add myself
as the maintainer for them.
Remove the old entry from board/Marvell/mvebu_armada-37xx/MAINTAINERS.
Signed-off-by: Robert Marko <robert.marko@sartura.hr>
Reviewed-by: Stefan Roese <sr@denx.de>
Methode eDPU is an Armada 3720 power board based on the Methode uDPU.
They feature the same CPU, RAM, and storage as well as the form factor.
However, eDPU only has one SFP slot plus a copper G.hn port which does not
work under U-boot.
In order to reduce duplication, split the uDPU DTS into a common one.
Signed-off-by: Robert Marko <robert.marko@sartura.hr>
Reviewed-by: Stefan Roese <sr@denx.de>
board_get_usable_ram_top() conflated the RAM size with the top address
of RAM. On systems where RAM starts at address 0 these numbers are the
same so it went unnoticed. Update board_get_usable_ram_top() to take
CONFIG_SYS_SDRAM_BASE into account when determining the top address.
Signed-off-by: Chris Packham <judge.packham@gmail.com>
Reviewed-by: Stefan Roese <sr@denx.de>
It does not matter what is DT node name of atsha device. So find it via
atsha driver and not by DT node name.
Signed-off-by: Pali Rohár <pali@kernel.org>
Reviewed-by: Stefan Roese <sr@denx.de>
U-Boot does not implement down_write_trylock() and its stub always returns
true that lock was acquired. Therefore ubifs_assert_cmt_locked() assert
currently always fails.
Fix this issue by redefining ubifs_assert_cmt_locked() to just empty stub
as there is nothing to assert.
Signed-off-by: Pali Rohár <pali@kernel.org>
i2c changes for 2022.10
- new driver nuvoton, NPCM7xx from Jim Liu
Fixes:
- ast_i2c: Remove SCL direct drive mode
from Eddie James
- avoid dynamic stack use in dm_i2c_write
bloat-o-meter drivers/i2c/i2c-uclass.o.{0,1}
add/remove: 0/0 grow/shrink: 0/1 up/down: 0/-144 (-144)
Function old new delta
dm_i2c_write 552 408 -144
Total: Before=3828, After=3684, chg -3.76%
patch from Rasmus Villemoes
To quote Andre:
One prominent feature is the restructering of the clock driver, which
allows to end up with one actual driver for all variants, although we
still only compile in support for one SoC.
Also contained are some initial SPI fixes, which should fix some
problems, and enable SPI flash support for the F1C100s SoC. Those
patches revealed more problems, I will queue fixes later on, but for
now it should at least still work.
Apart from some smaller fixes (for instance for NAND operation), there
is also preparation for the upcoming Allwinner D1 support, in form of
the USB PHY driver. There are more driver support patches to come.
The gitlab CI completed successfully, including the build test for all
160 sunxi boards. I also boot tested on a few boards, but didn't have
time for more elaborate tests this time.
The size of the dynamic stack allocation here is bounded by the if()
statement. However, just allocating the maximum size up-front and
doing malloc() if necessary avoids code duplication (the
i2c_setup_offset() until the invocation of ->xfer), and generates much
better (smaller) code:
bloat-o-meter drivers/i2c/i2c-uclass.o.{0,1}
add/remove: 0/0 grow/shrink: 0/1 up/down: 0/-144 (-144)
Function old new delta
dm_i2c_write 552 408 -144
Total: Before=3828, After=3684, chg -3.76%
It also makes static analysis of maximum stack usage (using the .su
files that are automatically generated during build) easier if there
are no lines saying "dynamic".
[This is not entirely equivalent to the existing code; this now uses
the stack for len <= 64 rather than len <= 63, but that seems like a
more natural limit.]
Signed-off-by: Rasmus Villemoes <rasmus.villemoes@prevas.dk>
Reviewed-by: Heiko Schocher <hs@denx.de>
SCL direct drive mode prevents communication with devices that
do clock stretching, so disable. The Linux driver doesn't use
this mode, and the engine can handle clock stretching.
Signed-off-by: Eddie James <eajames@linux.ibm.com>
Reviewed-by: Heiko Schocher <hs@denx.de>
Reviewed-by: Joel Stanley <joel@jms.id.au>
Reviewed-by: ryan_chen <ryan_chen@aspeedtech.com>
D1 has a register layout like A100 and H616, with the moved SIDDQ bit.
Unlike H616 it does not have any dependencies between PHY instances.
Signed-off-by: Samuel Holland <samuel@sholland.org>
Reviewed-by: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
As Icenowy pointed out, newer manuals (starting with H6) actually
document the register block at offset 0x800 as "HCI controller and PHY
interface", also describe the bits in our "PMU_UNK1" register.
Let's put proper names to those "unknown" variables and symbols.
While we are at it, generalise the existing code by allowing a bitmap
of bits to clear and set, to cover newer SoCs: The A100 and H616 use a
different bit for the SIDDQ control.
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
Signed-off-by: Samuel Holland <samuel@sholland.org>
Reviewed-by: Jagan Teki <jagan@amarulasolutions.com>
Since commit 089ffd0aed ("phy: sun4i-usb: Use CLK and RESET support")
neither of these headers is used. Dropping them allows the driver to be
architecture-independent.
Signed-off-by: Samuel Holland <samuel@sholland.org>
Reviewed-by: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
This option is used only by the phy-sun4i-usb driver, which does not
inherently depend on the ARM architecture.
Signed-off-by: Samuel Holland <samuel@sholland.org>
Reviewed-by: Jagan Teki <jagan@amarulasolutions.com>
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
Do not suggest successful operation if a flash area to be changed is
actually locked, thus will not execute the request. Rather report an
error and bail out. That's way more user-friendly than asking them to
manually check for this case.
Derived from original patch by Chao Zeng.
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Acked-by: Jagan Teki <jagan@amarulasolutions.com>
There was no user of this callback after 5b66fdb29d anymore, and its
semantic as now inconsistent between stm and sst26. What we need for the
upcoming new usecase is a "completely unlocked" semantic. So consolidate
over this.
Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Acked-by: Jagan Teki <jagan@amarulasolutions.com>
On probe, the SPI NOR core will put a flash in 8D mode if it
supports it. But Linux as of now expects to get the flash in
1S mode. Handing the flash to Linux in Octal DTR mode means
the kernel will fail to detect the flash.
This commit adds an option to soft reset the flash after
spl_spi_load_image() so that the flash can be reset to 1S mode
and subsequent spi-nor probe in Linux does not fail, since
spl_spi_load_image() performs spi_flash_probe() the remove is
added after completion loading images in spi_flash_probe() itself.
Tested on J721E EVM with 5.10 Linux kernel.
Linux spi-nor probe without the fix:
root@j7-evm:~# dmesg | grep spi-nor
[ 4.928023] spi-nor spi0.0: unrecognized JEDEC id bytes: ff ff ff ff ff ff
[ 4.934938] spi-nor: probe of spi0.0 failed with error -2
Linux spi-nor probe with the fix:
root@j7-evm:~# dmesg | grep spi-nor
[ 4.904484] spi-nor spi0.0: mt35xu512aba (65536 Kbytes)
Signed-off-by: Vaishnav Achath <vaishnav.a@ti.com>
Acked-by: Jagan Teki <jagan@amarulasolutions.com>
Power-on-Reset is a method to restore flash back to 1S-1S-1S mode from 8D-8D-8D
in the begging of probe.
Command extension type is not standardized across flash vendors in DTR mode.
For suiting different vendor flash devices, adding a flag to seperate types for
soft reset on boot.
Signed-off-by: JaimeLiao <jaimeliao.tw@gmail.com>
Reviewed-by: Jagan Teki <jagan@amarulasolutions.com>
Follow patch <f6adec1af4b2f5d3012480c6cdce7743b74a6156> (Allow using Micron mt35xu512aba
in Octal DTR mode).
Enable Octal DTR mode with 20 dummy cycles to allow running at the
maximum supported frequency for adding Macronix flash in Octal DTR mode.
-https://www.mxic.com.tw/Lists/Datasheet/Attachments/7841/MX25LM51245G,%203V,%20512Mb,%20v1.1.pdf
Signed-off-by: JaimeLiao <jaimeliao.tw@gmail.com>
Reviewed-by: Jagan Teki <jagan@amarulasolutions.com>
The sunxi nand SPL loader was broken at least for SUN4I,
SUN5I and SUN7I SOCs since the implementation change
from DMA to PIO usage - commit 6ddbb1e.
Root cause for this issue is the NFC control flag NFC_CTL_RAM_METHOD
being set by method nand_apply_config.
This flag controls the bus being used for the NFCs internal RAM access.
It must be set for the DMA use case only.
See A33_Nand_Flash_Controller_Specification.pdf page 12.
This fix is tested by myself on a Cubietruck A20 board.
Others should test it on new generation SOCs as well.
Signed-off-by: Markus Hoffrogge <mhoffrogge@gmail.com>
Reviewed-by: Andre Przywara <andre.przywara@arm.com>
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
The boards that come with a flash memory pre-soldered have a Macronix
flash chip.
Fixes: 280294c5df ("sunxi: boards: Enable SPI flash support in U-Boot proper")
Signed-off-by: Michal Suchanek <msuchanek@suse.de>
Reviewed-by: Andre Przywara <andre.przywara@arm.com>
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
KConfig has range option, use it instead of notice in the option
descrition.
Signed-off-by: Michal Suchanek <msuchanek@suse.de>
Reviewed-by: Andre Przywara <andre.przywara@arm.com>
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
Many LicheePi Nano boards come with SPI flash soldered, which already
works for booting the SPL and loading U-Boot proper.
With the updated DTB, we can now also use the SPI flash from U-Boot
proper, so enable the bits in the defconfig, to allow loading binaries
from SPI flash.
There seem to be board revisions with a Winbond SPI chip, but also
others with an XTX chip, so include support for both: the actual chip
used will be autodetected.
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
The SPI controllers in the Allwinner F1Cx00 series of SoCs are
compatible to the H3 IP. The only difference in the integration is
the missing mod clock in the F1C100, instead the SPI clock is directly
derived from the AHB clock.
We *should* be able to model this through the DT, but the addition of
get_rate() requires quite some refactoring, so it's not really worth in
this simple case: We programmed both the PLL_PERIPH to 600 MHz and the
PLL/AHB divider to 3 in the SPL, so we know the SPI base clock is 200
MHz. Since we used a hard coded fixed clock rate of 24 MHz for all the
other SoCs so far, we can as well do the same for the F1C100.
Define the SPI input clock and maximum frequency differently when
compiling for the F1C100 SoC.
Also adjust the power-of-2 divider programming, because that uses a
"minus one" encoding, compared to the other SoCs.
This allows to enable SPI flash support for the F1C100 boards.
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
The current SPI clock divider calculation has two problems:
- We use a normal round-down division, which results in a divider
typically being too small, resulting in a too high frequency on the bus.
- The calculaction for the power-of-two divider is very inaccurate, and
again rounds down, which might lead to wild bus frequencies.
This wasn't a real problem so far, since most chips can handle slightly
higher bus frequencies just fine. Also the actual speed was mostly lost
anyway, due to release_bus() reseting the device. And the power-of-2
calculation was probably never used, because it only applies to
frequencies below 47 KHz.
However this will become a problem for the F1C100s support, due to its
much higher base frequency.
Calculate a safe divider correctly (using round-up), and re-use that
value when calculating the power-of-2 value. We also separate the
maximum frequency and the input clock on the way, since they will be
different for the F1C100s.
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
As George rightfully pointed out [1], the spi-sunxi driver programs the
speed and mode settings only when the respective functions are called,
but this gets lost over a call to release_bus(). That asserts the
reset line, thus forces each SPI register back to its default value.
Adding to that, trying to program SPI_CCR and SPI_TCR might be pointless
in the first place, when the reset line is still asserted (before
claim_bus()), so those setting won't apply most of the time. In reality
I see two nested claim_bus() calls for the first use, so settings between
the two would work (for instance for the initial "sf probe"). However
later on the speed setting is not programmed into the hardware anymore.
So far we get away with that default frequency, because that is a rather
tame 24 MHz, which most SPI flash chips can handle just fine.
Move the actual register programming into a separate function, and use
.set_speed and .set_mode just to set the variables in our priv structure.
Then we only call this new function in claim_bus(), when we are sure
that register accesses actually work and are preserved.
[1] https://lore.kernel.org/u-boot/20210725231636.879913-17-me@yifangu.com/
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
Reported-by: George Hilliard <thirtythreeforty@gmail.com>
The current detection of RX FIFO depth seems to be not reliable, and
XCH will self-clear when a transfer is done.
Check XCH bit when polling for transfer finish.
Signed-off-by: Icenowy Zheng <uwu@icenowy.me>
Reviewed-by: Andre Przywara <andre.przywara@arm.com>
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
This header is not used since commit abdbefba2a ("net: sun8i_emac: Use
consistent clock bitfield definitions"). Dropping it allows the driver
to be architecture-independent.
Signed-off-by: Samuel Holland <samuel@sholland.org>
Reviewed-by: Andre Przywara <andre.przywara@arm.com>
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
This just prints the PHY mode taken from the devicetree. It does not
need to be printed during every boot, and also avoids an unwanted
line break for the "net: " reporting line.
Signed-off-by: Samuel Holland <samuel@sholland.org>
Reviewed-by: Andre Przywara <andre.przywara@arm.com>
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
For mostly historic reasons we had configuration headers for each
Allwinner CPU "family". These days they are mostly just including one
common header, with the rest being somewhat empty.
There were attempts to remove them, and to just use the one common header
to begin with, but this has implications to the build system, which me
might not be ready for, yet.
To document this behaviour, and to avoid something sneaking in over
time, make those files all the same (minus the CPU family name and
the copyrights), and add a comment explaining that.
This makes it easier to just remove those files later on, when needed
and possible.
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
The clock and reset drivers use the exact same platform data. Simplify
them by sharing the object. This is safe because the parent device
(the clock device) always gets its driver model callbacks run first.
Signed-off-by: Samuel Holland <samuel@sholland.org>
Acked-by: Andre Przywara <andre.przywara@arm.com>
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
The reason here is the same as the reason for changing the clock driver:
platform data can be provided when binding the driver.
Signed-off-by: Samuel Holland <samuel@sholland.org>
Reviewed-by: Andre Przywara <andre.przywara@arm.com>
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
All of the driver private data should really be platform data since it
is determined statically (selected by the compatible string or extracted
from the devicetree). Move everything to platform data, so it can be
provided when binding the driver. This is useful for SPL, or for
instantiating the driver as part of an MFD.
Signed-off-by: Samuel Holland <samuel@sholland.org>
Reviewed-by: Andre Przywara <andre.przywara@arm.com>
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
Now that all of the variants use the same bind/probe functions and ops,
there is no need to have a separate driver for each variant. Since most
SoCs contain two variants (the main CCU and PRCM CCU), this saves a bit
of firmware size and RAM.
Signed-off-by: Samuel Holland <samuel@sholland.org>
Reviewed-by: Andre Przywara <andre.przywara@arm.com>
[Andre: add F1C100s support]
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
This allows all of the clock drivers to use a common bind function.
Signed-off-by: Samuel Holland <samuel@sholland.org>
Reviewed-by: Andre Przywara <andre.przywara@arm.com>
[Andre: add F1C100s support]
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
Because the gate arrays are not given explicit sizes, the arrays are
only as large as the highest-numbered gate described in the driver.
However, only a subset of the CCU clocks are needed by U-Boot. So there
are valid clock specifiers with indexes greater than the size of the
arrays. Referencing any of these clocks causes out-of-bounds access.
Fix this by checking the identifier against the size of the array.
Fixes: 0d47bc7056 ("clk: Add Allwinner A64 CLK driver")
Signed-off-by: Samuel Holland <samuel@sholland.org>
Reviewed-by: Andre Przywara <andre.przywara@arm.com>
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
The reset array size is currently used for bounds checking in the reset
driver. The same bounds check should really be done in the clock driver.
Currently, the array size is provided to the reset driver separately
from the CCU descriptor, which is a bit strange. Let's do this the usual
way, with the array sizes next to the arrays themselves.
Signed-off-by: Samuel Holland <samuel@sholland.org>
Reviewed-by: Andre Przywara <andre.przywara@arm.com>
[Andre: add F1C100s support]
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
Now that the PHY driver will not try to drive VBUS if it is already
driven by an external supply, there is no need to check the VBUS voltage
before powering on the PHY.
Signed-off-by: Samuel Holland <samuel@sholland.org>
