Age | Commit message (Collapse) | Author |
|
Using linux,usable-memory the board code carves out 1MiB of memory
for the Cortex-M4 core at the end of the 256MiB block. However,
in case a board has 256MiB of memory (Colibri iMX7S), that is the
area where U-Boot gets relocated to... Use board_get_usable_ram_top
to avoid using that area as relocation target.
Signed-off-by: Stefan Agner <stefan.agner@toradex.com>
Acked-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
|
|
i.MX 7's Cortex-M4 core can run from DDR and uses DDR memory for
the rpmsg communication. Both use cases need a fixed location of
memory reserved. For the rpmsg use case the reserved area needs
to be in sync with the kernel's hardcoded vring descriptor location.
Use the linux,usable-memory property to carve out 1MB of memory
in case the M4 core is running. Also make sure that the i.MX 7
specific rpmsg driver does not get loaded in case we do not carve
out memory.
Signed-off-by: Stefan Agner <stefan.agner@toradex.com>
Acked-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
|
|
Add support for setting the linux,usable-memory in the memory
node of device tree.
Signed-off-by: Sanchayan Maity <maitysanchayan@gmail.com>
Signed-off-by: Stefan Agner <stefan.agner@toradex.com>
Acked-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
|
|
This reverts commit 34fcaeec51fa3f519a13eb8d127abea1853e77e5.
If you have this version of U-Boot running, the change we warned
about has already been applied. No need to warn anymore.
Signed-off-by: Stefan Agner <stefan.agner@toradex.com>
Acked-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
|
|
Use two separate partitions for the two firmware instances. Also
resize them to be of the same size which also makes the start of
the UBI partition nicely aligned to 0x400000.
In order to detect the new MTD layout and whether we run a U-Boot
with the new BCB format or not, introduce a variable called
"updlevel" which we can use in update/upgrade scripts.
Signed-off-by: Stefan Agner <stefan.agner@toradex.com>
Acked-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
|
|
The writeboot command has been used to write the bootloader in
a format required by early boot ROM's do to errata e9609 (write
3/4 of a page only). Since U-Boot now writes also BCBs which use
the full page, we can use standard U-Boot nand write commands to
write the bootloader.
Signed-off-by: Stefan Agner <stefan.agner@toradex.com>
Acked-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
|
|
Create a FCB without work-around for errata e9609. With that FCB
in place the firmware can be written using the standard nand write
command like:
nand write ${loadaddr} u-boot ${filesize}
Using sectors_in_firmware{1|2} seems not to make any difference and
is also not set on Vybrid, hence do not set it anymore.
Signed-off-by: Stefan Agner <stefan.agner@toradex.com>
Acked-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
|
|
Commit 35c204d8a9d0 ("nand: reinstate lazy bad block scanning")
introduced lazy BBT scanning. However, some controller do parts
of the initialization (mxs_nand.c) during scan_bbt, hence for
those controllers the BBT must be scanned at initialization time.
Signed-off-by: Stefan Agner <stefan.agner@toradex.com>
Acked-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
|
|
Add sd card detect SD1_CD# applicable for V1.1 modules using GPIO_PV2.
Signed-off-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
Acked-by: Dominik Sliwa <dominik.sliwa@toradex.com>
|
|
Configure GPIO_PV2 as SD1_CD# according to latest V1.1 HW.
Leave SDMMC3_CLK_LB_OUT muxed as SDMMC3 with output driver enabled aka
not tristated and input driver enabled as well as it features some
magic properties even though the external loopback is disabled and the
internal loopback used as per SDMMC_VENDOR_MISC_CNTRL_0 register's
SDMMC_SPARE1 bits being set to 0xfffd according to the TRM! This pin is
now a not-connect on V1.1 HW in order to avoid any interference.
Signed-off-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
Acked-by: Dominik Sliwa <dominik.sliwa@toradex.com>
|
|
Configure DP_HPD_PFF0 pin as optional DisplayPort hot-plug detect.
Signed-off-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
Acked-by: Dominik Sliwa <dominik.sliwa@toradex.com>
|
|
Pull-up GPIO_PI6 connected to TMP451's ALERT#/THERM2#.
Signed-off-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
Acked-by: Dominik Sliwa <dominik.sliwa@toradex.com>
|
|
Get rid of obsolete non-Jetson configuration options.
Signed-off-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
Acked-by: Dominik Sliwa <dominik.sliwa@toradex.com>
|
|
Fix device tree pre-fix in DFU_ALT_EMMC_INFO.
Signed-off-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
Acked-by: Dominik Sliwa <dominik.sliwa@toradex.com>
|
|
Few GPIOs are being used as special function pins
and should not be initialized as GPIOs.
Signed-off-by: Dominik Sliwa <dominik.sliwa@toradex.com>
Acked-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
|
|
Enable CONFIG_SUPPORT_EMMC_BOOT to provision for eMMC boot specific
U-Boot commands.
Signed-off-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
Acked-by: Max Krummenacher <max.krummenacher@toradex.com>
|
|
Turns out CONFIG_ARCH_MISC_INIT was not defined therefore our
arch_misc_init() function was never even called.
While at it also enable the previously missing CONFIG_OF_LIBFDT.
Signed-off-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
Acked-by: Max Krummenacher <max.krummenacher@toradex.com>
|
|
Add a note about the non-working apalis_tk1_gpio_inits table caused by
a reservation failure. If it wouldn't be run in SPL without anything
ever getting printed one would see the following:
A: dir_input: error: gpio A1 not reserved
Please note: The current workaround is to configure pull-up/downs in
the apalis_tk1_pingrps table further below.
Signed-off-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
Acked-by: Max Krummenacher <max.krummenacher@toradex.com>
|
|
Implement optional PCIe EvalBoard initialisation which properly reset the PLX
(now Avago) PEX 8605 PCIe switch plus PCIe devices on the Apalis Evaluation
carrier board.
Please note that you will have to enable the second PCIe port in the dts as well
e.g.:
diff --git a/arch/arm/dts/tegra124-apalis.dts b/arch/arm/dts/tegra124-apalis.dts
index 595aae8..841c676 100644
--- a/arch/arm/dts/tegra124-apalis.dts
+++ b/arch/arm/dts/tegra124-apalis.dts
@@ -42,6 +42,10 @@
vddio-pex-ctl-supply = <®_3v3>;
avdd-pll-erefe-supply = <&avdd_1v05>;
+ pci@1,0 {
+ status = "okay";
+ };
+
/* I210 Gigabit Ethernet Controller (on-module) */
pci@2,0 {
status = "okay";
While at it also get rid of the DECLARE_GLOBAL_DATA_PTR which is probably a
remnant of the Apalis T30 board file where we use it to get the memory size.
Signed-off-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
Acked-by: Max Krummenacher <max.krummenacher@toradex.com>
|
|
Floating pins towards the camera input level-shifter cause it to
consume at least 100mW of power even in idle. This patch fixes this by
configuring pull-downs on all multiplexed camera input pins.
Signed-off-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
Acked-by: Max Krummenacher <max.krummenacher@toradex.com>
|
|
Add common code for i.MX SoC's which avoids loading the environment.
This reimplements the functionaltiy lost with ("common: autoboot:
do not reset environment") by not loading the environment from the
beginning. The main difference will be that dynamically set
environment variables such as "soc" in the i.MX 7 case will still
be set.
Signed-off-by: Stefan Agner <stefan.agner@toradex.com>
Acked-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
|
|
Do not reset environment when using Serial Downloader. This is
problematic for some modules where we set environment variables
dynamically during boot (e.g. soc).
Signed-off-by: Stefan Agner <stefan.agner@toradex.com>
Acked-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
|
|
Make should_load_env a weak function. This allows a board to
decide whether to load the environment or not. This can be
useful when detecting recovery mode to avoid loading hostile
environments.
Signed-off-by: Stefan Agner <stefan.agner@toradex.com>
Acked-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
|
|
We plan to no longer support the work-around for errata e9609 since
it does not allow to detect factory bad blocks. Print a warning to
prepare customers of the upcoming change.
Signed-off-by: Stefan Agner <stefan.agner@toradex.com>
Acked-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
|
|
Skip over data which have the tag flag reporting invalid rather than
abort parsing.
Break parsing when the first tag id set to 0xffff is encountered or the
maximum size of the config block is reached.
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
Signed-off-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
|
|
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
Signed-off-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
|
|
Cleanup whitespace found when preparing patches for mainline.
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
Signed-off-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
|
|
Cleanup whitespace found when preparing patches for mainline.
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
Signed-off-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
|
|
Cleanup whitespace found when preparing patches for mainline.
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
Signed-off-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
|
|
Correctly skip bad blocks when writing boot loader. Despite this
fix U-Boot will generate excessive amount of messages when hitting
a bad block:
Size of write exceeds partition or device limit
However, the command continues to write page by page until writing
succeeds.
Note that despite the correct bad block skip schema in place, the
current FCB uses DISBBM = 1 which seems to disable the skip bad
block handling in Boot ROM.
Signed-off-by: Stefan Agner <stefan.agner@toradex.com>
Acked-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
|
|
Use the correct array element for the first firmware offset argument.
Due to that bug, so far only the second firmware has been used and
set as the offset of the first firmware in the FCB.
Signed-off-by: Stefan Agner <stefan.agner@toradex.com>
Acked-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
|
|
Disable DCDC2 discharge as this leads to higher power consumption when
turning off DCDC2 on Colibri iMX7S. This is likely due to backfeeding
the DCDC2 through the VFB2 pin.
Signed-off-by: Stefan Agner <stefan.agner@toradex.com>
Acked-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
|
|
Configure sleep slots for LDO1 (Ethernet) and ARM CPU rail. This
makes sure the rails get disabled when the SoC enters sleep mode.
Signed-off-by: Stefan Agner <stefan.agner@toradex.com>
Acked-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
|
|
u-boot allows modifying a device tree after it is loaded into
memory. Add fdt_fixup hook in u-boot environment which can
facilitate such modifications.
Signed-off-by: Sanchayan Maity <sanchayan.maity@toradex.com>
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
|
|
While trying to display a BMP image by running the command
'bmp display <address>' the Colibri T30 was getting stuck (while on
iMX6, iMX7 and Vybrid it worked). After investigation on what was going
wrong, I noticed that the function 'lcd_set_cmap' should get a pointer
address for the cmap variable, but the function returned 00000000.
After further investigation, I noticed that this value should be held
in the structure.member 'panel_info.cmap', which memory was being
dynamically allocated in the function 'lcd_ctrl_init' of the file
'drivers/video/sandbox_sdl.c' (which was not being called). After
copying the allocation to the same function in the file
'drivers/video/tegra.c', the command 'bmp display' didn't break
anymore. The Colibri T20 module was working without the patch, as the
module does have valid memory at the address 0, and of course it kept
working after the patch now properly allocating memory for it.
Signed-off-by: Leonardo Graboski Veiga <leonardo.veiga@toradex.com>
Signed-off-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
|
|
netdevwait was used in the 2.6.36 based Vibrante kernel for
Colibri T20.
None of our kernel trees evaluate that cmdline parameter.
So drop it from the nfsargs environment variable.
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
Acked-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
|
|
Add board_usb_phy_mode function for detecting whether a port is
being used as host or client using a GPIO. On Colibri Vybrid we
provide GPIO 102 for this very same purpose.
Signed-off-by: Sanchayan Maity <maitysanchayan@gmail.com>
Signed-off-by: Stefan Agner <stefan.agner@toradex.com>
|
|
Add board_usb_phy_mode weak function on similar lines to ehci-mx6.
However since Vybrid USB does not have a true OTG, make this weak
functon just return 0. The function is supposed to be implemented
by the individual boards using a GPIO for providing the OTG pin
functionality.
Signed-off-by: Sanchayan Maity <maitysanchayan@gmail.com>
Signed-off-by: Stefan Agner <stefan.agner@toradex.com>
|
|
The current ehci-vf USB driver for Vybrid hardcodes the USB host
and client functionality. Remove this.
Signed-off-by: Sanchayan Maity <maitysanchayan@gmail.com>
Signed-off-by: Stefan Agner <stefan.agner@toradex.com>
|
|
This makes sure that the common code passes the Toradex serial
number, product id and revision to the device tree.
Signed-off-by: Stefan Agner <stefan.agner@toradex.com>
Acked-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
|
|
nand torture currently works on exactly one nand block which is specified
by giving the byteoffset to the beginning of the block.
Extend this by allowing for a second parameter specifying the byte size
to be tested.
e.g.
==> nand torture 1000000
NAND torture: device 0 offset 0x1000000 size 0x20000 (block size 0x20000)
Passed: 1, failed: 0
==> nand torture 1000000 40000
NAND torture: device 0 offset 0x1000000 size 0x40000 (block size 0x20000)
Passed: 2, failed: 0
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
Reviewed-by: Benoît Thébaudeau <benoit.thebaudeau.dev@gmail.com>
[scottwood: fix usage to show size as optional, and add misssing braces]
Signed-off-by: Scott Wood <oss@buserror.net>
(cherry picked from commit 1866be7d28ce807397e4aedd93f70564ac8bebc0)
|
|
The mtd subsystem deprecated and renamed the direct use of the mtd_info
struct's functionpointers. Instead the corresponding mtd_xxx function
should be used.
See also:
https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/?id=3c3c10bba1e4ccb75b41442e45c1a072f6cded19
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
(cherry picked from commit 667067faa18334f1e28c01b47530b5cce1b6182f)
|
|
A negative value for the offset is treated as a backwards offset for
from the end of the device/partition for block devices. This aligns
the behavior of the config file with the syntax of CONFIG_ENV_OFFSET
where the functionality has been introduced with
commit 5c088ee841f9 ("env_mmc: allow negative CONFIG_ENV_OFFSET").
Signed-off-by: Stefan Agner <stefan.agner@toradex.com>
Acked-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
|
|
Currently flash_read completes a crucial part of the environment
device configuration, the device type (mtd_type). This is rather
confusing as flash_io calls flash_read conditionally, and one might
think flash_write, which also makes use of mtd_type, gets called
before flash_read. But since flash_io is always called with O_RDONLY
first, this is not actually the case in reality.
However, it is much cleaner to complete and verify the config early
in parse_config. This also prepares the code for further extension.
Signed-off-by: Stefan Agner <stefan.agner@toradex.com>
Acked-by: Marcel Ziswiler <marcel.ziswiler@toradex.com>
|
|
Signed-off-by: Andreas Fenkart <andreas.fenkart@digitalstrom.com>
(cherry picked from commit 10667e15f48613e264a4da78a36bbd7daf5c4ce8)
|
|
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
Acked-by: Stefan Agner <stefan.agner@toradex.com>
|
|
Copy these from Linux v4.5-rc6 tag.
This is needed so that we can keep up with newer gcc versions. Note
that we don't have the uapi/ hierarchy from the kernel so continue to
use <linux/types.h>
Signed-off-by: Tom Rini <trini@konsulko.com>
(cherry picked from commit 9b2c282b348dfe966bbba967dc7a45ce817cce50)
|
|
The i.MX 7 detection taken from upstream U-Boot adds a new macro
MXC_CPU_MX7S. The downstream U-Boot still has one runtime occurence
which currently checks for MXC_CPU_MX7D only. Fix this SoC detection
to detect MXC_CPU_MX7S too.
Note: While the GPT timer is available on i.MX 7, it is currently
not configured (CONFIG_GPT_TIMER). Instead, the CPU internal
syscounter timer is currently used (CONFIG_SYSCOUNTER_TIMER).
Signed-off-by: Stefan Agner <stefan.agner@toradex.com>
Acked-by: Max Krummenacher <max.krummenacher@toradex.com>
|
|
For mx7 the code needn't make provisions, we can use the simple
variant of fuse_bank_physical. Also U-Boot does not allow multi-
platform (MX6/MX7) configurations, hence we can use compile time
defines which select the simpler variant for MX7 always.
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
Signed-off-by: Stefan Agner <stefan.agner@toradex.com>
|
|
Remove code which does not change anything. This also synchronizes
this function with the upstream variant of it.
Signed-off-by: Max Krummenacher <max.krummenacher@toradex.com>
Signed-off-by: Stefan Agner <stefan.agner@toradex.com>
|