The a3700_fdt_fix_pcie_regions() function still computes nonsense.
It computes the fixup offset from the PCI address taken from the first
row of the "ranges" array, which means that:
- PCI address must equal CPU address (otherwise the computed fix offset
will be wrong),
- the first row must contain the lowest address.
This is the case for the default device-tree, which is why we didn't
notice it.
It also adds the fixup offset to all PCI and CPU addresses, which is
wrong.
Instead:
1) The fixup offset must be computed from the CPU address, not PCI
address.
2) The fixup offset must be computed from the row containing the lowest
CPU address, which is not necessarily contained in the first row.
3) The PCI address - the address to which the PCIe controller remaps the
address space as seen from the point of view of the PCIe device -
must be fixed by the fix offset in the same way as the CPU address
only in the special case when the CPU adn PCI addresses are the same.
Same addresses means that remapping is disabled, and thus if we
change the CPU address, we need also to change the PCI address so
that the remapping is still disabled afterwards.
Consider an example:
The ranges entries contain:
PCI address CPU address
70000000 EA000000
E9000000 E9000000
EB000000 EB000000
By default CPU PCIe window is at: E8000000 - F0000000
Consider the case when TF-A moves it to: F2000000 - FA000000
Until now the function would take the PCI address of the first entry:
70000000, and the new base, F2000000, to compute the fix offset:
F2000000 - 70000000 = 82000000, and then add 8200000 to all addresses,
resulting in
PCI address CPU address
F2000000 6C000000
6B000000 6B000000
6D000000 6D000000
which is complete nonsense - none of the CPU addresses is in the
requested window.
Now it will take the lowest CPU address, which is in second row,
E9000000, and compute the fix offset F2000000 - E9000000 = 09000000,
and then add it to all CPU addresses and those PCI addresses which
equal to their corresponding CPU addresses, resulting in
PCI address CPU address
70000000 F3000000
F2000000 F2000000
F4000000 F4000000
where all of the CPU addresses are in the needed window.
Fixes: 4a82fca8e3 ("arm: a37xx: pci: Fix a3700_fdt_fix_pcie_regions() function")
Signed-off-by: Pali Rohár <pali@kernel.org>
Signed-off-by: Marek Behún <marek.behun@nic.cz>
Reviewed-by: Stefan Roese <sr@denx.de>
First set of u-boot-atmel fixes for the 2022.04 cycle:
This fixes set includes only a single fix for the Ethernet on sama7g5ek
board which is broken at the moment.
Support for Apple M1 Pro and Max will allow using a single binary for
all M1 SoCs. The M1 Pro/Max have a different memory layout. The RAM
start address is 0x100_0000_0000 instead of 0x8_0000_0000.
Replace the hardcoded memory layout with dynamic initialized
environment variables in board_late_init().
Tested on Mac Mini (2020) and Macbook Pro 14-inch (2021).
Signed-off-by: Janne Grunau <j@jannau.net>
Reviewed-by: Mark Kettenis <kettenis@openbsd.org>
Simplify the binman config and fdt nodes by using the "@..-SEQ"
substitutions and CONFIG_OF_LIST.
Signed-off-by: Michael Walle <michael@walle.cc>
[Rebased]
Signed-off-by: Priyanka Jain <priyanka.jain@nxp.com>
omap_ehci_hcd_stop appears to be dead code, and omap_ehci_hcd_init
is only called by the probe function, so it can be static to that
function. Remove both from the header along with some additional
checking for DM_USB.
Signed-off-by: Adam Ford <aford173@gmail.com>
The default U-Boot environment variables and design are all set up for
the MAIN R5FSS cluster to be in Split-mode. This is the setting used
when the dts nodes were originally added in v2021.01 U-Boot and the
dt nodes are synched with the kernel binding property names in
commit 468ec2f3ef ("remoteproc: k3_r5: Sync to upstreamed kernel DT
property names") merged in v2021.04-rc2.
The modes for the MAIN R5FSS cluster got switched back to LockStep mode
by mistake in commit fa09b12dc5 ("arm: ti: k3: Resync dts files and
bindings with Linux Kernel v5.14") in v2022.01-rc1. This throws the
following warning messages when early-booting the cores using default
env variables,
k3_r5f_rproc r5f@5d00000: Invalid op: Trying to start secondary core 7 in lockstep mode
Load Remote Processor 3 with data@addr=0x82000000 83148 bytes: Failed!
Fix this by switching back both the clusters to the expected Split-mode.
Make this mode change in the u-boot specific dtsi file to avoid such
sync overrides in the future until the kernel dts is also switched to
Split-mode by default.
Fixes: fa09b12dc5 ("arm: ti: k3: Resync dts files and bindings with Linux Kernel v5.14")
Signed-off-by: Suman Anna <s-anna@ti.com>
There are a few memory functions for both the emif4 (AM3517)
and sdrc (OMAP35/DM37) code that can be defined as static,
because those functions are not used externally. Make them
static and clean up some of the corresponding headers.
Signed-off-by: Adam Ford <aford173@gmail.com>
With LTO enabled, some functions appear to be optimized in a
way that causes hanging on some OMAP3 boards after some
unrelated patches were applied. The solution appears to make
several functions __used. There also appears be to be some
dead code, so remove it while cleaning this up.
This has been tested on a general purpose OMAP3530, DM3730,
and AM3517.
Signed-off-by: Adam Ford <aford173@gmail.com>
Correct the min/max voltages of VDD_CPU. As per data sheet the VDD_CPU
minimum voltage is .6V & maximum voltage is .9V.
Correct the same. While at it fix the comment to reflect VDD_CPU
instead of VDD_MPU.
Data Sheet Link: https://www.ti.com/lit/gpn/dra829v
Signed-off-by: Keerthy <j-keerthy@ti.com>
Choose the memory map based on the compatible property from the
device tree passed to us by m1n1. Since DRAM on the M1 Pro/Max
starts at a different address avoid hardcoding the top of usable
memory. Also make sure that the addresses entered into the memory
map are page aligned such that we don't crash in dcache_enable().
Signed-off-by: Mark Kettenis <kettenis@openbsd.org>
Tested on: Macbook M1 Max
Tested-by: Janne Grunau <j@jannau.net>
The GW7902 is based on the i.MX 8M Mini / Nano SoC featuring:
- LPDDR4 DRAM
- eMMC FLASH
- Gateworks System Controller
- LTE CAT M1 modem
- USB 2.0 HUB
- M.2 Socket with USB2.0, PCIe, and dual-SIM
- IMX8M FEC
- PCIe based GbE
- RS232/RS485/RS422 serial transceiver
- GPS
- CAN bus
- WiFi / Bluetooth
- MIPI header (DSI/CSI/GPIO/PWM/I2S)
- PMIC
To add support for the i.MX8M Nano GW7902:
- Add imx8mn-venice dts/defconfig/include
- Add imx8mn-gw7902 dts
- Add imx8mn-2gb lpddr4 dram configs
- Add misc support for IMX8M Nano SoC
- rename imx8mm-venice.c to venice.c as it is no longer imx8mm specific
- update README with differences for IMX8MN vs IMX8MM
Signed-off-by: Tim Harvey <tharvey@gateworks.com>
Reviewed-by: Fabio Estevam <festevam@gmail.com>
The interface for NOR/OneNAND is called "EIM" not "EMI". Fix this.
Signed-off-by: Harald Seiler <hws@denx.de>
Reviewed-by: Fabio Estevam <festevam@gmail.com>
- a37xx: pci: Cleanup and minor fix for root port check (Pali)
- pci: mvebu: Ensure that root port is always on root zero bus (Pali)
- kwbimage: Fix dumping DATA registers for v0 images (Pali)
- kwbimage: Support for parsing extended v0 format (Pali)
- a37xx: Fix code and update DTS files to upstream version (Pali)
- a37xx: Fix and extend building memory map (Pali)
- ddr: marvell: a38x: fix BYTE_HOMOGENEOUS_SPLIT_OUT decision (Marek)
- mvebu: Optionally reset board on DDR training failure (Marek)
Some boards may occacionally fail DDR training. Currently we hang() in
this case. Add an option that makes the board do an immediate reset in
such a case, so that a new training is tried as soon as possible,
instead of hanging and possibly waiting for watchdog to reset the board.
(If the DDR training fails while booting the image via UART, we will
still hang - it doesn't make sense to reset in such a case, because
after reset the board will try booting from another medium, and the
UART booting utility does not expect that.)
Signed-off-by: Marek Behún <marek.behun@nic.cz>
Reviewed-by: Pali Rohár <pali@kernel.org>
Reviewed-by: Stefan Roese <sr@denx.de>
Function is named build_mem_map, not a3700_build_mem_map.
Signed-off-by: Pali Rohár <pali@kernel.org>
Reviewed-by: Stefan Roese <sr@denx.de>
Reviewed-by: Marek Behún <marek.behun@nic.cz>
In function build_mem_map() prepare also mapping for CCI-400 and
BootROM windows.
BootROM window is 1 MB long and by default starts at address 0xfff00000.
A53 AP BootROM is 16 kB long and repeats in this BootROM window 64 times.
RVBAR_EL3 register is set to value 0xffff0000, so by default A53 AP BootROM
is accessed via range 0xffff0000-0xffff3fff.
CCI-400 window when new TF-A version is used, starts at address 0xfe000000
and when old TF-A version is used, starts at address 0xd8000000.
Physical addresses are read directly from mvebu registers, so if TF-A
remaps it in future (again) then it would not cause any issue for U-Boot.
Signed-off-by: Pali Rohár <pali@kernel.org>
Reviewed-by: Stefan Roese <sr@denx.de>
Function build_mem_map() modifies global variable mem_map. This variable is
used by the get_page_table_size() function which is called by function
arm_reserve_mmu() (as aliased macro PGTABLE_SIZE). Function
arm_reserve_mmu() is called earlier than enable_caches() which calls
build_mem_map(). So arm_reserve_mmu() does not calculate reserved memory
correctly.
Fix this issue by calling build_mem_map() from a3700_dram_init() which is
called before arm_reserve_mmu().
Signed-off-by: Pali Rohár <pali@kernel.org>
Reviewed-by: Stefan Roese <sr@denx.de>
Reviewed-by: Marek Behún <marek.behun@nic.cz>
This change updates all Armada 37xx DTS files to version which is used by
Linux kernel v5.18.
Signed-off-by: Pali Rohár <pali@kernel.org>
Reviewed-by: Stefan Roese <sr@denx.de>
Official DT bindings for Espressobin have disabled eMMC node.
As U-Boot requires to have this node enabled by default, do it in
armada-3720-espressobin-u-boot.dtsi DTS file.
Signed-off-by: Pali Rohár <pali@kernel.org>
Reviewed-by: Stefan Roese <sr@denx.de>
U-Boot specific changes should be in armada-3720-espressobin-u-boot.dtsi DTS file.
Signed-off-by: Pali Rohár <pali@kernel.org>
Reviewed-by: Stefan Roese <sr@denx.de>
In Linux kernel version of armada-37xx.dtsi file sdhci1 pointer refers to
sdhci@d0000 node and sdhci0 pointer to sdhci@d8000 node.
Signed-off-by: Pali Rohár <pali@kernel.org>
Reviewed-by: Stefan Roese <sr@denx.de>
Official DT bindings have only one reg property: watchdog address space.
Convert armada-37xx-wdt.c driver to offical DT bindings and access sel_reg
register via MVEBU_REGISTER() macro, as its value (required by U-Boot
driver) is not in DT yet. In later stage can be driver cleaned to not use
it.
This change would allow U-Boot to use A3720 watchdog DTS structure from
Linux kernel.
Signed-off-by: Pali Rohár <pali@kernel.org>
Reviewed-by: Stefan Roese <sr@denx.de>
Official DT bindings use compatible string marvell,armada-3700-ehci.
Update drivers and DTS files.
Signed-off-by: Pali Rohár <pali@kernel.org>
Reviewed-by: Stefan Roese <sr@denx.de>
In commit d368e10705 ("phy: marvell: a3700: Convert to official DT
bindings in COMPHY driver") was done update to official DT bindings but
compatible string of official DT bindings was not updated.
Fix it now.
Fixes: d368e10705 ("phy: marvell: a3700: Convert to official DT bindings in COMPHY driver")
Signed-off-by: Pali Rohár <pali@kernel.org>
Reviewed-by: Stefan Roese <sr@denx.de>
A new lpddr4 configuration is introduced for J7 SK with 4266 MTs data
rate. Therefore, update the R5 DTS file to point to the new DDR config
file.
Signed-off-by: Sinthu Raja <sinthu.raja@ti.com>
EMIF tool for J721E SK is now updated to 0.6.1 that includes
* Updated write DQ training pattern to enable user pattern and clock
pattern (from 0x7 to 0x6).
* Updated IO drive strength to 40-80-80 Ohms.
J721E SK uses the lpddr4 configuration of 4266 MTs data rate which is
the same as J721E EVM but facing random failures. As the tool update is
specific to the SK board, add a new lpddr4 config of 4266 MTs.
Signed-off-by: Kevin Scholz <k-scholz@ti.com>
Signed-off-by: Sinthu Raja <sinthu.raja@ti.com>
J721E Starter Kit (SK)[1] is a low cost, small form factor board designed
for TI’s J721E SoC. TI’s J721E SoC comprises of dual core A72, high
performance vision accelerators, video codec accelerators, latest C71x
and C66x DSP, high bandwidth real-time IPs for capture and display, GPU,
dedicated safety island and security accelerators. The SoC is power
optimized to provide best in class performance for industrial and
automotive applications.
J721E SK supports the following interfaces:
* 4 GB LPDDR4 RAM
* x1 Gigabit Ethernet interface
* x1 USB 3.0 Type-C port
* x3 USB 3.0 Type-A ports
* x1 PCIe M.2 E Key
* x1 PCIe M.2 M Key
* 512 Mbit OSPI flash
* x2 CSI2 Camera interface (RPi and TI Camera connector)
* 40-pin Raspberry Pi GPIO header
Add A72 specific dts for J721E-SK.
[1] https://www.ti.com/tool/SK-TDA4VM
Signed-off-by: Sinthu Raja <sinthu.raja@ti.com>
Board ID I2C EEPROM will be probed before SYSFW is available.
So drop the power-domains property for wakup_i2c0 on which
board ID EEPROM is connected.
Signed-off-by: Sinthu Raja <sinthu.raja@ti.com>
Enable support for selecting DTB from FIT within SPL based on the
board name read from EEPROM. This will help to use single defconfig
for both EVM and SK.
Signed-off-by: Sinthu Raja <sinthu.raja@ti.com>
The changes introduced with commit 6337d53fdf ("arm: dts: sync am33xx
with Linux 5.9-rc7") prevent the PDU001 from operating correctly.
This patch fixes the configuration of the pin multiplexer and uart3.
Signed-off-by: Felix Brack <fb@ltec.ch>
EMIF tool for AM64 SK is now updated to 0.8.0 that includes
* disabled Write DQ training
* improve CA ODT to 60 ohms
The lpddr4 enabled with periodic WDQ training is causing periodic 26us
stall. This makes the SoC stall without doing anything which leads to
R5 interrupt latency in TCM memory. Due to this periodic training there
are some outstanding CPU transactions waiting for the lpddr4 to complete.
Hence, disable the periodic write DQ training during the
non-initialization stage of lpddr4 which results in an approximate 1us
stall. Also, update the lpddr4 config to improve CA ODT by 60 ohms
The rationales are as follows:
- PI_WDQLVL_EN: 2 Bits register field to support write DQ leveling,
disable bit 1 that supports Write DQ during non-initialization to
avoid ~26us stall during code execution.
- MR11_DATA_F1/F2_x register fields value changed to 0x66 that changes
the CA ODT from 48ohm to 60ohm to improve the eye margin on CA bus by
increasing the signal swing.
Signed-off-by: James Doublesin <doublesin@ti.com>
Signed-off-by: Sinthu Raja <sinthu.raja@ti.com>
Add a driver for the SPI controller integrated on Apple SoCs.
This is necessary to support the keyboard on Apple Silicon laopts
since their keyboard uses an Apple-specific HID over SPI protocol.
Signed-off-by: Mark Kettenis <kettenis@openbsd.org>
Reviewed-by: Simon Glass <sjg@chromium.org>
Tested on: Macbook Air M1
Tested-by: Simon Glass <sjg@chromium.org>
The power management controller found on Apple SoCs als provides
a way to reset all devices within a power domain. This is needed
to cleanly shutdown the NVMe controller before we hand over
control to the OS.
Signed-off-by: Mark Kettenis <kettenis@openbsd.org>
Reviewed-by: Simon Glass <sjg@chromium.org>
Tested on: Macbook Air M1
Tested-by: Simon Glass <sjg@chromium.org>
Reviewed-by: Jaehoon Chung <jh80.chung@samsung.com>
Most Apple IOPs run a firmware that is based on what Apple calls
RTKit. RTKit implements a common mailbox protocol. This code
provides an implementation of the AP side of this protocol,
providing a function to initialize RTKit-based firmwares as well
as a function to do a clean shutdown of this firmware.
Signed-off-by: Mark Kettenis <kettenis@openbsd.org>
Reviewed-by: Simon Glass <sjg@chromium.org>
Tested on: Macbook Air M1
Tested-by: Simon Glass <sjg@chromium.org>
U-Boot is expected to support multiple generations of Apple SoCs
in a single binary with a single defconfig. Therefore it makes
more sense to set SYS_SOC to "apple".
Signed-off-by: Mark Kettenis <kettenis@openbsd.org>
Reviewed-by: Simon Glass <sjg@chromium.org>
This mailbox driver provides a communication channel with the
Apple IOP controllers found on Apple SoCs. These IOP controllers
are used to implement various functions such as the System
Manegement Controller (SMC) and NVMe storage. It allows sending
and receiving a 96-bit message over a single channel.
The header file with the struct used for mailbox messages is taken
straight from Linux.
Signed-off-by: Mark Kettenis <kettenis@openbsd.org>
Signed-off-by: Sven Peter <sven@svenpeter.dev>
Reviewed-by: Simon Glass <sjg@chromium.org>
Tested on: Macbook Air M1
Tested-by: Simon Glass <sjg@chromium.org>
