Add method to reserve video framebuffer information using blob,
received from previous stage.
Signed-off-by: Nikhil M Jain <n-jain1@ti.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Use spl_dcache_enable, in place of setup_dram, arch_reserve_mmu to set
up pagetable, initialise DRAM and enable Dcache to avoid multiple
function calls.
Check for CONFIG_SPL_VIDEO in place of CONFIG_SPL_VIDEO_TIDSS to prevent
any build failure in case video config is not defined and video related
functions are called.
Check for CONFIG_SPL_SPLASH_SCREEN and CONFIG_SPL_BMP before calling
splash_display to avoid compilation failure.
Signed-off-by: Nikhil M Jain <n-jain1@ti.com>
In spl_dcache_enable after setting up page table, set gd->relocaddr
pointer to tlb_addr, to get next location to reserve memory. Align
tlb_addr with 64KB address.
Signed-off-by: Nikhil M Jain <n-jain1@ti.com>
Reviewed-by: Devarsh Thakkar <devarsht@ti.com>
At SPL stage when stack is relocated, the stack pointer needs to be
updated, the stack pointer may point to stack in on chip memory even
though stack is relocated.
Signed-off-by: Nikhil M Jain <n-jain1@ti.com>
Reviewed-by: Tom Rini <trini@konsulko.com>
There is little need to print the devstat information or when we exit a
function during a typical boot. Remove them to reduce the noise during
typical operation
Signed-off-by: Bryan Brattlof <bb@ti.com>
During LPDDR initialization we will loop through a series of frequency
changes in order to train at the various operating frequencies. During
this training, accessing the DRAM_CLASS bitfield could happen during a
frequency change and cause the read to hang.
Store the DRAM type into the main structure to avoid multiple readings
while the independent phy is training.
Signed-off-by: Bryan Brattlof <bb@ti.com>
The function declaration for force_emif_self_refresh takes no parameters
but does not specify this, only the prototype in the headers do. As
clang will warn about this, correct it.
Signed-off-by: Tom Rini <trini@konsulko.com>
Non-HS boards can use FIT images so include the env var commands
for these unconditionally.
Signed-off-by: Andrew Davis <afd@ti.com>
Reviewed-by: Nishanth Menon <nm@ti.com>
AM62x SoC is available in multiple variant:
- CPU cores (Cortex-A) AM62x1 (1 core), AM62x2 (2 cores), AM62x4 (4 cores)
- GPU AM625x with GPU, AM623x without GPU
- PRU (Programmable RT unit) can be present or not on AM62x2/AM62x4
Remove the relevant FDT nodes by reading the actual configuration
from the SoC registers, with that change is possible to have a single
dts/dtb file handling the different variant at runtime.
While removing GPU node and CPU nodes also the watchdog node
in the same Module Domain is removed.
A similar approach is implemented for example on i.MX8 and STM32MP1 SoC.
Signed-off-by: Emanuele Ghidoli <emanuele.ghidoli@toradex.com>
Signed-off-by: Francesco Dolcini <francesco.dolcini@toradex.com>
Add register address and relevant bitmasks and shifts.
Allow reading these information:
- device identification
- number of cores (part of device identification)
- features (currently: PRU / no PRU)
- security
- functional safety
- speed grade
- temperature grade
- package
Signed-off-by: Emanuele Ghidoli <emanuele.ghidoli@toradex.com>
Signed-off-by: Francesco Dolcini <francesco.dolcini@toradex.com>
Reviewed-by: Andrew Davis <afd@ti.com>
ft_system_setup cannot be enabled on SoC without msmc sram otherwise
fdt_fixup_msmc_ram function fails causing system reset.
Fix by moving fdt_fixup_msmc_ram to common_fdt.c file and creating
SoC (AM654, J721E and J721S2) specific files for fdt fixups.
This change was verified to not change anything on any existing board
(all the J721S2, AM654 and J721E boards requires it,
none of the remaining k3 boards require it).
Fixes: 30e96a2401 ("arm: mach-k3: Move MSMC fixup to SoC level")
Signed-off-by: Emanuele Ghidoli <emanuele.ghidoli@toradex.com>
Signed-off-by: Francesco Dolcini <francesco.dolcini@toradex.com>
On security enforced (HS-SE) devices ROM firewalls OSPI data region3 that
is present in above 64bit region. Open this up in bootloader to allow
Linux to access OSPI flashes in mmap mode.
Without this kernel will crash when accessing this region due to
firewall violations on HS-SE devices.
Signed-off-by: Vignesh Raghavendra <vigneshr@ti.com>
During compilation splash_source puts out below warning for type
conversion in splash_load_fit for bmp_load_addr and fit_header.
Change their type to uintptr_t to fix the warnings.
common/splash_source.c: In function ‘splash_load_fit’:
common/splash_source.c:366:22: warning: cast to pointer from integer of different size [-Wint-to-pointer-cast]
366 | img_header = (struct legacy_img_hdr *)bmp_load_addr;
| ^
common/splash_source.c:376:49: warning: cast from pointer to integer of different size [-Wpointer-to-int-cast]
376 | res = splash_storage_read_raw(location, (u32)fit_header, fit_size);
| ^
common/splash_source.c:401:25: warning: cast to pointer from integer of different size [-Wint-to-pointer-cast]
401 | memmove((void *)bmp_load_addr, internal_splash_data, internal_splash_size);
The above warnings are generated if CONFIG_FIT is enabled.
Signed-off-by: Nikhil M Jain <n-jain1@ti.com>
At the time of compilation evm.c gives below warning for implicit
declaration of enable_caches, to mitigate this include cpu_func.h.
board/ti/am62x/evm.c: In function ‘spl_board_init’:
board/ti/am62x/evm.c:90:9: warning: implicit declaration of function ‘enable_caches’ [-Wimplicit-function-declaration]
90 | enable_caches();
Signed-off-by: Nikhil M Jain <n-jain1@ti.com>
For once this adds USB support for two SoCs: the H616 and the F1C100s
series. The rest is support for LPDDR3 DRAM chips on H616 boards.
Gitlab CI passed, and I booted that briefly on an H616 and an F1C200s
board. I don't have an H616 board with LPDDR3 DRAM, but reportedly that
works for Mikhail, and doesn't regress on my DDR3 boards.
Xilinx changes for v2023.10-rc1 v2
axi_emac:
- Change return value if RX packet is not ready
cadence_qspi:
- Enable flash reset for Versal NET
dt:
- Various DT syncups with Linux kernel
- SOM - reserved pmufw memory location
fpga:
- Add load event
mtd:
- Add missing dependency for FLASH_CFI_MTD
spi/nand:
- Minor cleanup in Xilinx drivers
versal-net:
- Prioritize boot device in boot_targets
- Wire mini ospi/qspi/emmc configurations
watchdog:
- Use new versal-wwdt property
xilinx:
- fix sparse warnings in various places ps7_init*
- add missing headers
- consolidate code around zynqmp_mmio_read/write
- switch to amd.com email
zynqmp_clk:
- Add handling for gem rx/tsu clocks
zynq_gem:
- Configure mdio clock at run time
zynq:
- Enable fdt overlay support
zynq_sdhci:
- Call dll reset only for ZynqMP SOCs
If there is no incoming package than axiemac_recv will return -1 which
in turn leads to printing `eth_rx: recv() returned error -1` error
message in eth_rx function. But missing a package is not an fatal error,
so return -EAGAIN in that case would be more suitable.
Signed-off-by: Maksim Kiselev <bigunclemax@gmail.com>
Link: https://lore.kernel.org/r/20230719065337.69280-1-bigunclemax@gmail.com
Signed-off-by: Michal Simek <michal.simek@amd.com>
The commit a4180c3696 ("arm64: zynqmp: Add mode-pin GPIO controller DT
node") added usb phy reset over bootmode pins by default on usb0 only.
zcu100 is using usb0 as peripheral and usb1 as host. Unfortunately reset
line is shared for both usb ulpi phys but usb_rst_b is connected to usb5744
hub which is used only in host mode. Especially this chip requires reset to
operate properly that's why better assign gpio reset to usb1 instead of
usb0.
Without this change usb start crashed when runs.
Signed-off-by: Michal Simek <michal.simek@amd.com>
Link: https://lore.kernel.org/r/1ca80ec5bf7a595c03822f3e4e3683298205067a.1688992653.git.michal.simek@amd.com
Increase reset assert time for TI SGMII PHY on KR260 CC starting
6.1 kernel. This PHY does not come out of reset with the existing
100us pulse width as per testing on multiple carrier cards. The reset
is driven via a PCA9570 I2C expander. The expander driver was updated
to an upstream version in 6.1 where gpio_chip _set was optimized.
Delays in earlier kernels may have masked this issue. This is a safe
workaround value for assert pulse width before the discussions are
resolved with TI.
Signed-off-by: Harini Katakam <harini.katakam@amd.com>
Reviewed-by: Radhey Shyam Pandey <radhey.shyam.pandey@amd.com>
Signed-off-by: Michal Simek <michal.simek@amd.com>
Link: https://lore.kernel.org/r/8fb9f17d43a43ef504c9f29006cd686cce8ac98b.1688992653.git.michal.simek@amd.com
Enable flash reset functionality for Versal NET platform.
In cadence_qspi.c there is weak function defined for reset, hence remove
the check for config, so that it will work for Versal and Versal NET
platforms.
Add register defines in Versal NET hardware.h for mini U-Boot flash
reset.
Add read_delay initialization for Versal NET also.
Signed-off-by: Ashok Reddy Soma <ashok.reddy.soma@amd.com>
Link: https://lore.kernel.org/r/20230614120452.21019-1-ashok.reddy.soma@amd.com
Signed-off-by: Michal Simek <michal.simek@amd.com>
Currently xspi0 is used for all spi boot modes, it means it will use "sf
probe 0 0 0" for all spi's irrespective of which node it is wired.
Get boot sequence from dev_seq() and update boot command for xspi
dynamically.
As a result bootcmd for spi is updated as below when two instances of spi
are present in DT node.
bootcmd_xspi0=devnum_xspi=0; run xspi_boot
bootcmd_xspi1=devnum_xspi=1; run xspi_boot
xspi_boot=sf probe $devnum_xspi:0 0 0 && sf read $scriptaddr
$script_offset_f $script_size_f && echo XSPI: Trying to boot script at
${scriptaddr} && source ${scriptaddr}; echo XSPI: SCRIPT FAILED:
continuing...;
Signed-off-by: Ashok Reddy Soma <ashok.reddy.soma@amd.com>
Link: https://lore.kernel.org/r/20230614093058.30438-1-ashok.reddy.soma@amd.com
Signed-off-by: Michal Simek <michal.simek@amd.com>
