Age | Commit message (Collapse) | Author |
|
Fix Linux USB OTG1 behaviour by not enabling USBO1_EN during boot as
USB host functionality on OTG1 port is anyway broken under U-Boot.
Enabling USBO1_EN fed back via USBO1_VBUS sensing and caused the OTG
state machine to get a bogus peripheral state change rather than
staying in suspend which failed subsequent host activation.
This was especially bad on Ixora where by default the OTG1 port is
jumpered for shared host port functionality.
|
|
Fix possible uninitialised array member issue in serial number to
hexadecimal coded decimal conversion routine.
|
|
In order to be able to distinguish what hardware one is running on
change the U-Boot shell prompt to something more meaningful.
|
|
In preparation for the new Apalis T30 SKUs implement eMMC PT offset
auto detection.
The PT offset has been calculated from the .cfg eMMC partition
configuration file using virtual linearised addressing across all eMMC
regions as expected by nvflash.
Due to the lack of a region control mechanism in nvflash/.cfg flashing
utility in order to obtain the actual PT offset from the start of the
user region the size of the boot regions must be subtracted.
An eMMC device is specified to have 0 or 2 boot regions and they must
have the same size ranging from 0 to 2 MiB. The boot region size can be
queried from the device through the BOOT_SIZE_MULT field of the
response to the EXT_CSD command sequence.
Tested on initial samples of Apalis T30 1GB V1.0A,
Apalis T30 2GB V1.0B, Apalis T30 2GB V1.0C and Colibri T30 V1.1C.
|
|
In preparation for the new Apalis T30 SKUs implement memory size auto
detection based on reading the aperture register set by boot ROM from
BCT and a simple mirroring detection.
Tested on initial samples of Apalis T30 1GB V1.0A,
Apalis T30 2GB V1.0B, Apalis T30 2GB V1.0C and Colibri T30 V1.1C.
|
|
Fix possible overflow condition in serial number to hexadecimal coded
decimal conversion routine.
|
|
As we have lots of impatient managers lets decrease the default boot
delay. One can still enter the interactive U-Boot console by hitting a
key just in time of boot/reset.
Please note currently at Toradex we from the Linux team concentrate on
feature stability rather than fast booting.
|
|
Due to a unoptimised PLL X table we were only running at 400MHz during
boot.
Incorporated the PLL X table from NVIDIA's latest public U-Boot sources:
http://nv-tegra.nvidia.com/gitweb/?p=3rdparty/u-boot.git;a=blob;f=arch/arm/cpu/arm720t/tegra-common/cpu.c;h=119342e9577f6b42f93d118b81c0e931c9c9423a;hb=chromeos/v2013.01.01-tegra114#l67
And actually set up the T30 PLLs regardless of slow flag as this is
anyway exclusively used on T20.
Issue report courtesy of Mariusz Bulkowski from Draminski.
|
|
As Apalis T30 uses DVI-D aka HDMI as its default external display
interface make sure to activate the framebuffer console on there.
|
|
Forgot to actually adjust the amount of available memory in the
previous commit. Wouldn't it be nice to actually be able to use the
full 2GB?
|
|
Add initial Apalis T30 support based off our current Colibri T30
implementation:
- Updated machine ID.
- USB host USBH2 and USBH3 support. Note: USBO1 support is currently broken.
- Updated MMC and SD card support.
- Adjusted available amount of memory.
|
|
Rather than relying on hard-coded offsets actually make use of
partition table parsing implementation.
While at it bring the whole config more in-line with the Colibri T20.
|
|
Rather than relying on hard-coded offsets actually make use of
partition table parsing implementation.
|
|
As the partition table parsing offset handling relies on the global
data structure prepare the EFI GPT implementation for that.
|
|
Rather than relying on hard-coded offsets actually make use of
partition table parsing implementation.
|
|
NVIDIA's MMC/SD layout includes a partition table that can be used to
generically determine U-Boot environment, kernel, configuration block
as well as GPT offsets. As an added benefit this is completely
independent of the underlying MMC/SD card used which might differ with
various future module versions or particularly cards used for T20 SD
boot. Also handles the case of T20 SD boot where the configuration
block is still taken from NAND flash while the rest resides on the SD
card.
Initial partition table parsing courtesy of Mitja Špes from LXNAV.
|
|
There was no space for the null byte. In our case it led to the lowest
byte of the pointer mtd used by nand_init_chip being cleared, which
made the system crash when it tried to call mtd->priv->select_chip.
Signed-off-by: Daniel Glöckner <dg@emlix.com>
|
|
The Colibri T30 V1.1 modules actually use a different card detect GPIO.
|
|
After having registered the following proper machine type migrate to
actually using it.
http://www.arm.linux.org.uk/developer/machines/list.php?id=4493
While at it clean-out some obsolete Cardhu specific device-tree
nodes resp. properties and clean-up the mach-types header file as well.
|
|
|
|
|
|
This is a purely cosmetic clean-up of the common Toradex board file.
|
|
|
|
|
|
Add optional aka commented out 800x480@60 timing suitable for EDT
ET070080DH6.
|
|
Turns out our simplistic approach of just blindly enabling LAN_V_BUS
and releasing LAN_RESET_N does not prove very reliable.
Properly resetting the chip for 5 microseconds after VBUS is stable
just like we do on the Colibri T20 seems to fix the issue.
|
|
Include "colibri_" prefix in our board compatibility tables in
preparation to properly distinguish future e.g. Apalis modules.
While at it bring the device tree matching more in par with Linux
kernel 3.1.10 from NVIDIA's L4T R16-R2.
|
|
- board.c: changes required by different PMIC variant
|
|
|
|
CLK_RST_CONTROLLER_RST_CPU_CMPLX_SET/CLR_0
|
|
description in Android code. no bug visible as function is not called with run = 1
|
|
here this use did not show any adverse effects
|
|
Turns out we completely missed properly resetting the ASIX USB to FastEthernet chip which from time-to-time on certain modules caused severe Ethernet detection faults only a complete hardware reset or power-cycle could eliminate. Properly resetting the chip for 5 microseconds after VBUS is stable seems to fix the issue.
|
|
|
|
- kernel can now be up to 8MB
- cleanup, e.g. remove any NAND config
- unify with T20 file layout
|
|
If the USR partition we usually mount as root file system could not be
found which is e.g. the case for Android make sure mtdparts does not
start with a spurious coma separator the kernel would interpret as an
empty partition entry.
|
|
|
|
Conflicts:
arch/arm/cpu/armv7/tegra3/warmboot_avp.c
arch/arm/include/asm/arch-tegra/clk_rst.h
|
|
|
|
tegra: add enterrcm command
Tegra's boot ROM supports a mode whereby code may be downloaded and flash
programmed over a USB connection. On dev boards, this is typically entered
by holding down a "force recovery" button and resetting the CPU. However,
not all boards have such a button (one example is the Compulab Trimslice),
so a method to enter RCM from software is useful.
This change implements the command "enterrcm" to do this, and enables it
for all Tegra boards by default. Even on boards other than Trimslice,
controlling this over a UART may be useful, e.g. to allow simple remote
control without the need for mechanical button actuators, or hooking up
relays/... to the button.
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Signed-off-by: Tom Warren <twarren@nvidia.com>
|
|
|
|
As the kernel recently passed the 4 MB size limit simply copying 4 MB
won't quite cut it. Increase to 8 MB for now. In the future properly
parsing the SD card's partition table would be the way to go.
|
|
Fix build issues introduced with NVIDIA partition table parsing
integration.
|
|
|
|
If booting from SD card BCT contains information specific to SD card
partition layout which is bogus if used for NAND partition parsing.
Simply fall back to default offset just like in recovery BCT case.
Note: in the future we could parse SD partition table as well to more
generically support SD booting from various card densities.
|
|
Rather than relying on hard-coded offsets actually make use of
partition table parsing implementation.
|
|
NVIDIA's NAND layout includes a partition table that can be used to
generically construct the mtdparts kernel boot argument. As an added
benefit this is completely independent of the underlying NAND part
used which differs with various module versions. It further allows our
customer easy adoption to their own custom partition layout.
Initial partition table parsing courtesy of Mitja Špes from LXNAV.
|
|
If booting from SD card BCT contains information specific to SD card
partition layout which is bogus if used for NAND partition parsing.
Simply fall back to default offset just like in recovery BCT case.
Note: in the future we could parse SD partition table as well to more
generically support SD booting from various card densities.
|
|
Rather than relying on hard-coded offsets actually make use of
partition table parsing implementation.
|
|
NVIDIA's NAND layout includes a partition table that can be used to
generically construct the mtdparts kernel boot argument. As an added
benefit this is completely independent of the underlying NAND part
used which differs with various module versions. It further allows our
customer easy adoption to their own custom partition layout.
Initial partition table parsing courtesy of Mitja Špes from LXNAV.
|