NVIDIA Tegra
NVIDIA Jetson Linux Developer Guide
32.4.3 Release

 

Jetson TX1 Platform Adaptation and Bring-Up

 
Board Naming
Placeholders in the Porting Instructions
Pinmux Changes
Updating the Bootloader Pinmux
Porting U-Boot
Porting the Linux Kernel
Pad Power Detection
USB Lane Mapping
R24 Required Device Tree Changes
R24 Example 1
R24 Example 2
R28 Required Device Tree Changes
R28 Example 1
R28 Example 2
Other Considerations When Porting
Flashing the Build Image
This document is for software developers whose target is the NVIDIA® Jetson™ TX1 module. It describes how to port NVIDIA® Jetson™ Linux Driver Package (L4T) and the U-Boot boot loader from NVIDIA® Jetson™ TX1 Developer Kit to another hardware platform.
For all of the procedures in this document, L4T includes code for the Jetson TX1 Developer Kit (P2371-2180) that can serve as an example.

Board Naming

To support a Jetson TX1 module with your carrier board, you must assign the module/carrier board a lower case alphanumeric name. The name may include dashes (-) and underscores (_), but not spaces:
p2371-2180
devboard
The name you choose will appear in file names and path names in U-Boot and Linux kernel source code, user-visible device tree file names, and be exposed to the user via the U-Boot command prompt and various Linux kernel /proc files.
In this document, <board> represents your board name.
You must also choose a similarly-constructed vendor name. The same character set rules apply, such as the following example:
nvidia
In this document, <vendor> represents your vendor name.
Note:
Do not simply re-use and modify the existing NVIDIA Jetson TX1 Developer Kit code without choosing and using your own board name. If you do not use your own board name it will not be obvious to Jetson TX1 users whether modified source code supports the original Jetson TX1 Developer Kit carrier board or your board.

Placeholders in the Porting Instructions

The following sections refer to filenames and pathnames that contain these placeholders. Substitute an appropriate value for each placeholder when you enter the commands.
<function> is a functional module name, which may be power-tree, pinmux, sdmmc-drv, keys, comm (WIFI/BT), camera, etc.
<board> is a name you have chosen to represent your platform. For example, p2597 is the name of the Jetson TX1 Developer Kit carrier board. Note that NVIDIA <board> names use lower case letters only.
<som> is a System on a Module (SOM) board name, such as 2180.
<version> is a board version number, such as a00. Files for NVIDIA reference boards include a version number. Files for customer platforms need not include one.
<vendor> is your organization’s name, or the name of your board’s vendor.
<root> is the device that holds the platform’s root file system. At present the only supported value is emmc.

Pinmux Changes

If your board schematic differs from that for Jetson TX1 Developer Kit carrier board, you must change the pinmux configuration applied by the software.
To define your board’s pinmux configuration, you must download the Jetson TX1 pinmux table from the Jetson Download Center and customize it for the configuration of your board using the following procedures.
To customize the pinmux spreadsheet
1. Create a copy of the file with a name based on your board name, e.g. <board>_pinmux.xlsm. Note the directory in which you stored the copy; you will need it in the procedure Updating the Bootloader Pinmux.
2. Ensure that the new file is writable.
3. On a Windows PC, open the new file in Microsoft Excel.
4. If Microsoft Excel displays any warnings such as “PROTECTED VIEW” or “SECURITY WARNING,” click Enable Editing or Enable Content, so that you can save your changes to the new file.
5. Rename the Jetson TX1 Configuration tab based on the name of your board:
Right-click on the Jetson TX1 Configuration tab at the bottom of the window.
Click the Rename menu option.
Type your board name into the tab, then press ENTER.
6. Modify the spreadsheet columns grouped under the heading “Filled in by Customers” as required by the pinmux configuration for your board, based on the schematic.
7. Once the spreadsheet reflects the configuration you want, click the Generate DT File button to run a macro that exports the configuration data to a pair of .dtsi files in the directory where the spreadsheet is stored. The macro prompts you to enter a prefix for the files’ names; enter jetson-tx1-default (the default).

Updating the Bootloader Pinmux

The CBoot bootloader uses device tree files generated from the pinmux spreadsheet to configure the pinmux.
The pinmux settings in device tree files are only applied by CBoot, and not reapplied by the Linux kernel.
To use the updated device tree files, you must rebuild the device tree image for Jetson Nano.
To update the device tree image
1. Download the Linux kernel sources. For instructions see the section Obtaining the Kernel Sources with Git or Manually Downloading and Expanding Kernel Sources’ in the topic Kernel Customization.
2. Set up the build environment by installing the toolchain and setting the environment variable CROSS_COMPILE as described in the section Jetson Linux Driver Package Toolchain in the topic Kernel Customization.
3. Enter these commands to copy the newly generated device tree files from your Windows host to the proper locations on your Linux host:
$ cd <src_path>/hardware/nvidia/platform/t210/jetson/kernel-dts/jetson-platforms/
$ cp <dt_path>/jetson-tx1-default-pinmux.dtsi tegra210-jetson-cv-pinmux-p2597-2180-a00.dtsi
$ cp <dt_path>/jetson-tx1-default-gpio.dtsi tegra210-jetson-cv-gpio-p2597-2180-a00.dtsi
Where:
<src_path> is the pathname of the kernel sources that you downloaded in step 1.
jetson-tx1-default is the filename prefix that you entered in step 7 of To customize the pinmux spreadsheet. If you did not use the recommended default, change the filenames accordingly.
<dt_path> is the pathname of the directory that contains the pinmux spreadsheet that you modified in To customize the pinmux spreadsheet. Remember that this pathname is on the Windows host, which must be accessible to Linux through the network.
4. Rebuild the device tree image:
$ cd <src_path>/kernel/kernel-4.9/
$ make ARCH=arm64 tegra_defconfig
$ make ARCH=arm64 dtbs
5. Copy the updated device tree image to the L4T release tree:
$ cp arch/arm64/boot/dts/tegra210-jetson-cv-*.dtb <path-to-L4T-release>/kernel/dtb/
6. To reflash the target device for the new pinmux configuration to be applied, follow the instructions in the section Basic Flashing Procedures of the topic Flashing and Booting the Target Device.

Porting U-Boot

Perform the following actions in the U-Boot source code to add support for your board.
1. Copy include/configs/jetson-p2371-2180.h to include/configs/<board>.h.
2. Edit the set of enabled devices and features in <board>.h as appropriate for your board. For example, you must change the following:
#define CONFIG_TEGRA_BOARD_STRING "NVIDIA P2371-2180"
3. Copy arch/arm/dts/tegra210-p2371-2180.dts to arch/arm/dts/tegra210-<board>.dts.
4. Edit the set of enabled devices and their parameters (e.g. GPIO and IRQ IDs) in tegra210-<board>.dts as appropriate for your board.
Nodes and properties might need to be added, removed, or edited.
Note:
U-Boot and the Linux kernel do not always use the exact same device tree schema (bindings) to represent the same data. Follow examples from U-Boot rather than from the Linux kernel.
5. Add tegra210-<board>.dtb to arch/arm/dts/Makefile.
6. Copy configs/p2371-2180_defconfig to configs/<board>_defconfig.
7. Edit <board>_defconfig to refer to your board name.
8. Edit arch/arm/mach-tegra/tegra210/Kconfig to add target config and Kconfig.
9. Copy the board/nvidia/p2371-2180/ directory to board/<vendor>/<board>/.
10. Edit all of the files in board/<vendor>/<board>/ to refer to your board name rather than the P2371-2180. The files in this directory contain many instances of the P2371-2180 board name.
11. Edit board/<vendor>/<board>/MAINTAINERS to provide the correct maintainer contact information for your board.
12. Edit board/<vendor>/<board>/<board>.c to provide the correct PMIC programming for your board, if required.

Porting the Linux Kernel

To port the kernel configuration code (the device tree) to your platform, modify one of the distributed configuration files to describe your platform’s design.
The configuration files are in arch/arm64/boot/dts/. Their names have the form tegra210-jetson-cv-base-<board>-<som>.dts, where <board> refers to one of the NVIDIA reference boards, such as p2597.
NVIDIA recommends that you use this file and the files included, which describes the device tree of reference board P2597:
arch/arm64/boot/dts/tegra210-jetson-cv-base-p2597-2180-a00.dts
This device tree file includes many .dtsi files for various types of hardware. To configure the kernel to work on your platform, make copies of the .dts file and the .dtsi files it references, and modify the copies to correspond to your platform’s design.
The following procedure will guide you through this process.
1. Copy the .dts file you have chosen to this location:
arch/arm64/boot/dts/tegra210-<board>.dts
2. From the arch/arm64/boot/dts/tegra210-platforms/ directory, copy each file whose name has the form:
tegra210-jetson-cv-<function>-<board>-<som>-<version>.dtsi
Copy each file to:
arch/arm64/boot/dts/tegra210-platforms/tegra210-<board>-<function>.dtsi
You may rename the copies if that makes your work easier.
3. Edit your copy of the .dts file to refer to your copies of the .dtsi files.
4. Edit the set of enabled devices and their parameters (e.g. GPIO and IRQ IDs) in each copied file as appropriate for your board.
You may need to add, remove, or edit .dtsi file nodes and properties.
Note:
U-Boot and the Linux kernel do not always use the exact same device tree schema (bindings) to represent the same data. Follow examples from the Linux kernel rather than from U-Boot.
5. Replace the content of tegra210-<board>-gpio.dtsi and tegra210-<board>-pinmux.dtsi with the content you generated from the kernel pinmux files earlier.
6. Edit arch/arm64/boot/dts/Makefile to add an entry for your board, modeled after the existing Jetson TX1 entry.
7. Copy this file:
Linux_for_Tegra/bootloader/t210ref/p2371-2180/extlinux.conf.emmc
To this location:
Linux_for_Tegra/bootloader/t210ref/<board>-<som>/extlinux.conf.emmc
8. Copy the generated DTB to the following directory for flashing:
Linux_for_Tegra/kernel/dtb/
For rel-28 releases, to provide plugin manager support, the kernel DTB is not included in the file system along with the kernel image in the boot directory. Instead, the kernel DTB is selected from the DTB partition and modified by Cboot. Cboot passes it on to Uboot, which in turn passes it on to the kernel without diluting any of the data.
To flash only the DTB, execute the command:
sudo ./flash.sh -k DTB <platform> mmcblk0p1
To flash only Uboot, execute the command:
sudo ./flash.sh -k LNX <platform> mmcblk0p1
To flash only the kernel:
1. copy the kernel image to the following directory.
Linux_for_Tegra/kernel/
2. Execute the command:
sudo ./flash.sh <platform> mmcblk0p1
Optionally, you can perform a secure copy to copy the kernel image to the boot partition of the target system and reboot.
9. Copy Linux_for_Tegra/jetson-tx1.conf to Linux_for_Tegra/<board>.conf.
10. Edit SYSBOOTFILE and DTB_FILE in <board>.conf to refer to your board.
Following are some examples of modifications that you may have to make.
Regulator
VDDIO of SDMMC1 comes from PMU LDO2. That section looks like this:
regulators {
ldo2 {
regulator-name = "vddio-sdmmc1";
};
};
If there is any change for this power tree, this item should be changed.

Pad Power Detection

Pad power on the Jetson TX1 module’s T210 defaults to 3.3 V. Unlike some other Jetson processors, the T210 does not have auto power detect cells. I/O pads that are powered at 1.8 V must be set manually in DTS to 1.8 V.
gpio {
nvidia,io-pad-init-voltage = <IO_PAD_VOLTAGE_1_8V>;
};
GPIO
There are many GPIO configurations in different hardware modules. To change the GPIO setting, check the related device tree file.
For example, VDD of SDMMC is controlled by a GPIO pin (GPIO_PZ3). That is a power tree module, so the definition of this part is in:
tegra210-platforms/tegra210-jetson-cv-power-tree-p2597-2180-a00.dtsi
And looks like this:
en_vdd_sd: regulator@4 {
gpio = <&gpio TEGRA_GPIO(Z, 3) 0>;
};
You can change this setting according to platform circuit.
Interrupt
For modules that make interrupt requests, the interrupt requests can also be declared in the .dts file.
For example, this file:
tegra210-platforms/tegra210-comms-p2530-0930.dtsi
Describes a WIFI interrupt like this:
bcmdhd_wlan {
compatible = "android,bcmdhd_wlan";
interrupt-parent = <&gpio>;
interrupts = <TEGRA_GPIO(H, 2) 0x14>;
wlan-pwr-gpio = <&gpio TEGRA_GPIO(H, 0) 0>;
status = "okay";
};
This specifies GPIO_PH02 (same pin as WIFI_WAKE_AP) as the interrupt request pin from the WIFI module.
Key
The key is defined in:
tegra210-platforms/tegra210-keys-p2530-0930.dtsi
The power key is defined as:
power {
label = "Power";
gpios = <&gpio TEGRA_GPIO(X, 5) GPIO_ACTIVE_LOW>;
linux,code = <KEY_POWER>;
gpio-key,wakeup;
};
Note:
This example is meant only to show how to define a key in a .dts file. 'Power key' is a special key, and cannot be changed.
For the detailed information about .dts files, refer to the documentation at Documentation/devicetree/bindings in the NVIDIA released Linux kernel source package.

USB Lane Mapping

USB lane mapping and the required device tree changes are due to kernel version changes.
Release 24.X (R24) kernel version is 3.10
Release 28.X (R28) kernel version is 4.4
USB 3.0 has 4 superspeed ports. Not all can be used in the same implementation because of lane sharing between PCIE, SATA, and XUSB. Possible combinations for USB 3.0 are shown in the tables below.
TX1 Pin Names
PEX1
PEX_RFU
PEX2
USB_SS1
PEX0
USB_SS0
NA
SATA
TX1 Lanes
Lane 0
Lane 1
Lane 2
Lane 3
Lane 4
Lane 5
Lane 6
SATA
Use Case
1
(Carrier)
PCIe#1_0
PCIe#0_3
PCIe#0_2
PCIe#0_1
PCIe#0_0
USB_SS#1
USB_SS#0 on Jetson TX1 for LAN
SATA
2
PCIe#1_0
PCIe#0_3
PCIe#0_2
PCIe#0_1
PCIe#0_0
USB_SS#1
USB_SS#3
3
USB_SS#2
PCIe#0_3
PCIe#0_2
PCIe#0_1
PCIe#0_0
USB_SS#1
SATA
4
PCIe#1_0
-
-
USB_SS#2
PCIe#0_0
USB_SS#1
SATA
5
PCIe#1_0
-
-
USB_SS#2
PCIe#0_0
USB_SS#1
USB_SS#3
 
Use Case
Available Outputs from Jetson TX1
USB 3.0
PCIe
SATA
1 (Carrier)
1
1x1 + 1x4
1
2
2
1x1 + 1x4
0
3
2
1x4
1
4
2
2x1
1
5
3
2x1
0
The customer pinmux spreadsheet contains all Jetson TX1 pin names and ball names for determining which ball names are used for the super-speed connector, and the pinmux configuration of those pins.
An example configuration is given in section 5.1 of the Jetson TX1 OEM Product Design Guide. Each external super-speed connector has both USB 2.0 (DP, DN) and USB 3.0 lines (TX+-, RX+-) linked to the connector. A possible exception is where a fixed, on-board device is connected to super-speed lines and does not require USB 2.0 compatibility.

R24 Required Device Tree Changes

The following R24 device tree properties must change when USB configuration changes.
Property
Description
XHCI
nvidia,portmap
Lists all ports, 2.0 and 3.0, available for the XUSB controller.
Each bit represents a port and the bit is set for ports controlled by XUSB.
Bits 0-7 represent USB 3.0 ports with Port 0 on the LSB.
Bits 8-15 represent USB 2.0 ports with Port 0 on Bit 8.
Bits 16-23 represent HSIC ports with Port 0 on Bit 16.
For example, <0x0e02> represents USB 3.0 Port 1 and USB 2.0 Port 1, 2, and 3 are enabled.
PADCTL
nvidia,ss_portmap
Mapping between USB 2.0 USB 2.0 port lines and USB 3.0 por lines available on a single connector.
Each nibble represents a USB 3.0 port starting from LSB. The matching USB 2.0 port must be entered.
Enter 7 if the superspeed port is NOT in use or is not available.
For example, <0x7730> represents the following mapping:
ssport0-usb2port0
ssport1-usb2port3
ssport2-disabled
ssport3-disabled
USB 3.0 standalone ports without a USB 2.0 port require mapping to a USB 2.0 port that is a valid host port. For standalone ports, two USB 3.0 ports can be mapped to a single USB 2.0 port with the same role and may not represent the end-connector mapping.
PEX/HSIO PEX/HSIO lanes used by USB3.0 ports
nvidia,lane_owner
Each nibble represents a USB 3.0 port starting from the LSB.
A matching lane number must be entered.
Enter 0xF if the port is not in use or the lane is not mapped.
For example, <0xFF56> represents the following mapping:
ssport0-lane6
ssport1-lane5
ssport2-not in use
ssport3-not in use

R24 Example 1

A case where USB 2.0 Port 3 and USB 3.0 Port 2 are linked to an external connector on a carrier board matches use case 3 in possible mappings of superspeed ports. In that case the USB_SS#2 Port is using Lane 0 (USB_SS1 pins, ball names E41, E42, H41, H42). USB 2.0 Port 3, ball names B42 and B43, is mapped to an external connector; the default configuration uses these lines for M2.Key.
PCIe x1 must be disabled
USB_SS#2 port must be enabled
Lane mapping should indicate that lane 0 is in use by USB
In this example, the following device tree properties must be changed:
nvidia,ss_portmap: Super-speed port 2 must be enabled. This port map should match USB 3.0 and USB 2.0 lines on the same connector. For a standalone USB 3.0 port, any valid USB 2.0 port is sufficient. In this case USB 2.0 port 3 is matched with USB 3.0 port 2, so the value changes from <0x7721> to <0x7321>.
nvidia,portmap: Control super-speed port 2 with XHCI by replacing <0x0e03> with <0x0e07>.
nvidia,lane_owner: The lane owner for lane 0 is XUSB and is used by super-speed port 2. The property value of <0xff56> becomes <0xf056>.
nvidia,lane-map: This property indicates PEX lanes are available in the PCIe module and how they are configured. Because lane 0 (PEX1) is in use by USB, the lane map is modified from <0x14> to <0x4>.
Overall device tree properties patch in this case are as follows:
pcie-controller {
[...]
- nvidia,lane-map = <0x14>;
+ nvidia,lane-map = <0x4>;
[...]
};
xusb_pad_ctl: xusb_padctl { /* Put common control config here */
[...]
- nvidia,ss_portmap = <0x21>;
- nvidia,lane_owner = <0xff56>; /* Use 0xF to disable lane assign */
- nvidia,lane-map = <0x14>;
+ nvidia,ss_portmap = <0x321>;
+ nvidia,lane_owner = <0xf056>; /* Use 0xF to disable lane assign */
+ nvidia,lane-map = <0x4>;
[...]
};
xusb@70090000 {
[...]
- nvidia,portmap = <0x0e03>;
+ nvidia,portmap = <0x0e07>;
[...]
};

R24 Example 2

USB 3.0 Port 2 is used as a standalone port using lane 3. This matches use case 4/5 in possible mappings of superspeed ports. USB_SS#2 port is using lane 3 (PEX1 pins, ball names G42, G43, D42, D43). PCIe configuration changes from 1x1+1x4 to 1x1+1x1. USB_SS#2 is enabled, and lane mapping indicates that lane 3 is in use by USB.
The following device tree properties must change:
nvidia,ss_portmap: Superspeed port 2 is enabled. This port map must match USB 3.0 and USB 2.0 lines on the same connector. For a standalone USB 3.0 port, specify any valid USB 2.0 Port. In this case, a USB 2.0 standalone port matches a valid USB 2.0 port. This value changes from <0x7721> to <0x7221>.
nvidia,portmap: Control superspeed Port 2 with XHCI by replacing <0x0e03> with <0x0e07>.
nvidia,lane_owner: The lane owner for lane 3 is XUSB and is used by superspeed Port 2. The property value of <0xff56> becomes <0xf356>.
nvidia,lane-map: This property indicates that PEX lanes are available in the PCIe module and describes how they are configured. Because lane 3 USB_SS1 is in use by USB, the lane map is modified from <0x14> to <0x11>.
Overall device tree properties patch for this use case are as follows:
pcie-controller {
 
- nvidia,lane-map = <0x14>;
+ nvidia,lane-map = <0x12>;
[...]
};
xusb_pad_ctl: xusb_padctl { /* Put common control config here */
[...]
- nvidia,ss_portmap = <0x7721>;
- nvidia,lane_owner = <0xff56>; /* Use 0xF to disable lane assign */
- nvidia,lane-map = <0x14>;
+ nvidia,ss_portmap = <0x7221>;
+ nvidia,lane_owner = <0xf356>; /* Use 0xF to disable lane assign */
+ nvidia,lane-map = <0x11>;
[...]
};
xusb@70090000 {
[...]
- nvidia,portmap = <0x0e03>;
+ nvidia,portmap = <0x0e07>;
nvidia,common_padctl = <&xusb_pad_ctl>;
[...]
};

R28 Required Device Tree Changes

The following R28 device tree properties must change when USB configuration changes.
Property
Description
pinctrl@7009f000/pinmux
nvidia,lanes
Identifies the lane used on this port.
For USB 2.0 port, set this property to otg-<N>.
Where <N> is the number of USB 2.0 port.
For USB 3.0/PCIe/SATA ports, set this property to uphy-lane-<N>.
Where <N> is the lane number for this port.
nvidia,function
Specifies the function running on this port.
Possible values include:
xusb: indicates this port is used by USB 2.0 controller.
usb3: indicates this port is used by USB 3.0 controller.
pcie: indicates this port is used by PCIe controller.
nvidia,port-cap
If this USB 2.0/3.0 port supports host mode, set this property to <TEGRA_PADCTL_PORT_HOST_ONLY>.
nvidia,usb3-port
Set this property on USB 3.0 port only because it specifies which USB 3.0 port is bound to the lane set by the nvidia,lanes property.
nvidia,usb2-map
Set this property on USB 3.0 port because it specifies which USB 2.0 port is used with this USB 3.0 on the same connector.
nvidia,pcie-lane-select
Specifies that this PCIe port supports x1 or x4.
xusb@70090000
phys
Identifies the PHY handles to the USB 2.0 and 3.0 ports used by XUSB controller.
phy-names
Identifies the PHY names of the USB 2.0 and 3.0 phys used by XUSB controller.
The sequence in phys and phy-names must match each other.
pcie-controller@1003000
nvidia,num-lanes
This property must be present in each root port sub-node of the PCIe controller parent node.
For example:
pci@1,0, pci@2,0 and pci@3,0 nodes.
It indicates the maximum link width of that root port.
For example:
For controller-0, use 0x4for all cases.
For controller-1, only 0x1 is supported.

R28 Example 1

A case where USB 2.0 port 3 and USB 3.0 port 2 are linked to an external connector on a carrier board matches use case 3 in possible mappings of super-speed ports. In that case the USB_SS#2 port is using Lane 0 (USB_SS1 pins, ball names E41, E42, H41, H42). USB 2.0 port 3 (ball names B42, B43) is mapped to an external connector (the default configuration uses these lines for M2.Key). PCIe x1 should be disabled, USB_SS#2 port should be enabled, and lane mapping should now indicate that lane 0 is in use by USB.
In this example, the following device tree properties must be changed:
1. Disable pcie-m2 sub node.
Disable PCIe x1 on M2.Key by adding status = "disabled" to the pinctrl@7009f000/pinmux/pcie-m2 sub node.
2. Create two new sub-nodes under pinctrl@7009f000/pinmux.
Create two sub-nodes, such as usb2-m2 to represent USB 2.0 port 3 and usb3-m2 as an instance of USB3.0 port 2, then set the following properties of them.
nvidia,lanes: In this case, USB 3.0 port 2 is using Lane 0, so set nvidia,lanes = "uphy-lane-0" under the sub-node usb3-m2. Since USB 2.0 port 3 is used on M2.Key, set nvidia,lanes = "otg-3" under the usb2-m2 sub-node.
nvidia,function: Set nvidia,function = "xusb" in usb2-m2 and nvidia,function = "usb3" for usb3-m2.
nvidia,port-cap: Set nvidia,port-cap = <TEGRA_PADCTL_PORT_HOST_ONLY> in both usb2-m2 and usb3-m2 sub-nodes because USB connector on M2.Key supports only host mode.
nvidia,usb3-port: This property must be set in usb3-m2 to indicate which USB 3.0 port is bound to the lane set in nvidia,lanes. In this case, it is USB 3.0 port 2, so the value is:
nvidia,usb3-port = <2> in usb3-m2.nvidia,ss_portmap
nvidia,usb2-map: This nvidia,usb2-mapportmap must match USB 3.0 and USB 2.0 lines on the same connector and only be set in usb3-m2 sub-node. For a standalone USB 3.0 port, any valid USB 2.0 port is sufficient. In this case, USB 2.0 port 3 is matched with USB 3.0 port 2, so the value must be set to nvidia,usb2-map = <3>.
phys: Add <&tegra_padctl_uphy TEGRA_PADCTL_UPHY_UTMI_P(3)> for USB 2.0 port 3 and <&tegra_padctl_uphy TEGRA_PADCTL_UPHY_USB3_P(2)> for USB3.0 port2.
phy-names: Add "utmi-3" for USB 2.0 port 3 and "usb3-2" for USB 3.0 port 2.
The overall device tree properties patch in this case are as follows:
pinctrl@7009f000 {
[...]
pinmux {
[...]
pcie-m2 {
nvidia,lanes = "uphy-lane-0";
nvidia,function = "pcie";
nvidia,pcie-controller = <1>;
nvidia,pcie-lane-select = <TEGRA_PADCTL_PCIE_LANE_X1>;
+ status = “disabled”;
};
+ usb2-m2 {
+ nvidia,lanes = "otg-3";
+ nvidia,function = "xusb";
+ nvidia,port-cap = <TEGRA_PADCTL_PORT_HOST_ONLY>;
+ };
+ usb3-m2 {
+ nvidia,lanes = "uphy-lane-0";
+ nvidia,function = "usb3";
+ nvidia,usb3-port = <2>;
+ nvidia,usb2-map = <3>;
+ nvidia,port-cap = <TEGRA_PADCTL_PORT_HOST_ONLY>;
+ };
};
};
[...]
xusb@70090000 {
[...]
phys = <&tegra_padctl_uphy TEGRA_PADCTL_UPHY_UTMI_P(2)>,
<&tegra_padctl_uphy TEGRA_PADCTL_UPHY_USB3_P(1)>,
<&tegra_padctl_uphy TEGRA_PADCTL_UPHY_UTMI_P(1)>,
<&tegra_padctl_uphy TEGRA_PADCTL_UPHY_USB3_P(0)>,
<&tegra_padctl_uphy TEGRA_PADCTL_UPHY_UTMI_P(0)>,
+ <&tegra_padctl_uphy TEGRA_PADCTL_UPHY_UTMI_P(3)>,
+ <&tegra_padctl_uphy TEGRA_PADCTL_UPHY_USB3_P(2)>;
phy-names = "utmi-2", "usb3-1", "utmi-1", "usb3-0", "utmi-0",
+ "utmi-3", "usb3-2";
[...]
};

R28 Example 2

USB 3.0 port 2 is used as a standalone port using lane 3. This matches use case 4/5 in possible mappings of super speed ports. USB_SS#2 port is using lane 3 (PEX1 pins, ball names G42, G43, D42, D43). PCIe configuration changes from 1x1+1x4 to 1x1+1x1. USB_SS#2 is enabled, and lane mapping indicates that lane 3 is in use by USB.
The following device tree properties must change:
1. Create two new sub-nodes under pinctrl@7009f000/pinmux.
Create two sub-nodes, such as usb2-standalone to represent USB 2.0 port 3 and usb3-standalone as an instance of USB3.0 port 2, then set the following properties of them.
nvidia,lanes: In this case, USB 3.0 port 2 is using Lane 3, so set nvidia,lanes = "uphy-lane-3" under usb3-standalone sub-node and set nvidia,lanes = "otg-3" of usb2- standalone sub node.
Since PCIe configuration changes from 1x1+1x4 to 1x1+1x1, change nvidia,lanes of sub-node pcie from 4 lanes("uphy-lane-1", "uphy-lane-2", "uphy-lane-3", "uphy-lane-4") to "uphy-lane-4".
nvidia,function: Set nvidia,function = "xusb" in usb2-standalone and nvidia,function = "usb3" for usb3-standalone.
nvidia,port-cap: Set nvidia,port-cap = <TEGRA_PADCTL_PORT_HOST_ONLY> in both usb2-standalone and usb3-standalone sub-nodes because USB connector supports only host mode.
nvidia,usb3-port: This property must be set in usb3-standalone to indicate which USB 3.0 port is bound to the lane set in nvidia,lanes. In this case, it is USB 3.0 port 2, so the value is nvidia,usb3-port = <2> in usb3-standalone.
nvidia,usb2-map: This nvidia,usb2-map must match USB 3.0 and USB 2.0 lines on the same connector and only be set in usb3-m2standalone sub-node. In this case, USB 2.0 port 3 is matched with USB 3.0 port 2, so the value must be set to nvidia,usb2-map = <3>.
nvidia,pcie-lane-select: PCIe configuration changes from 1x1+1x4 to 1x1+1x1, it must always be set to <TEGRA_PADCTL_PCIE_LANE_X4> for controller 0.
phys: Add <&tegra_padctl_uphy TEGRA_PADCTL_UPHY_UTMI_P(3)> for USB 2.0 port 3 and <&tegra_padctl_uphy TEGRA_PADCTL_UPHY_USB3_P(2)> for USB3.0 port2.
phy-names: Add "utmi-3" for USB 2.0 port 3 and "usb3-2" for USB 3.0 port 2.
The overall device tree properties patch for above case would look like:
pinctrl@7009f000 {
[...]
pinmux {
[...]
pcie {
- nvidia,lanes = "uphy-lane-1", "uphy-lane-2",
- "uphy-lane-3", "uphy-lane-4";
+ nvidia,lanes = "uphy-lane-4";
nvidia,function = "pcie";
nvidia,pcie-controller = <0>;
nvidia,pcie-lane-select =
<TEGRA_PADCTL_PCIE_LANE_X4>
};
pcie-m2 {
nvidia,lanes = "uphy-lane-0";
nvidia,function = "pcie";
nvidia,pcie-controller = <1>;
nvidia,pcie-lane-select = <TEGRA_PADCTL_PCIE_LANE_X1>;
+ status = “disabled”;
};
+ usb2-standalone {
+ nvidia,lanes = "otg-3";
+ nvidia,function = "xusb";
+ nvidia,port-cap = <TEGRA_PADCTL_PORT_HOST_ONLY>;
+ };
+ usb3-standalone {
+ nvidia,lanes = "uphy-lane-3";
+ nvidia,function = "usb3";
+ nvidia,usb3-port = <2>;
+ nvidia,usb2-map = <3>;
+ nvidia,port-cap = <TEGRA_PADCTL_PORT_HOST_ONLY>;
+ };
};
};
[...]
xusb@70090000 {
[...]
phys = <&tegra_padctl_uphy TEGRA_PADCTL_UPHY_UTMI_P(2)>,
<&tegra_padctl_uphy TEGRA_PADCTL_UPHY_USB3_P(1)>,
<&tegra_padctl_uphy TEGRA_PADCTL_UPHY_UTMI_P(1)>,
<&tegra_padctl_uphy TEGRA_PADCTL_UPHY_USB3_P(0)>,
<&tegra_padctl_uphy TEGRA_PADCTL_UPHY_UTMI_P(0)>,
+ <&tegra_padctl_uphy TEGRA_PADCTL_UPHY_UTMI_P(3)>,
+ <&tegra_padctl_uphy TEGRA_PADCTL_UPHY_USB3_P(2)>;
phy-names = "utmi-2", "usb3-1", "utmi-1", "usb3-0", "utmi-0",
+ "utmi-3", "usb3-2";
[...]
};

Other Considerations When Porting

Other considerations and recommendations to consider when porting are as follows.
To flash the build image
When flashing the build image, use your specific board name:
$ sudo ./flash.sh <board>-<som> mmcblk0p1
To flash with BOARDID if the design does not use EEPROM
BOARDID is either passed using an XML file during flashing or is read from EEPROM. The flashing software uses the BOARDID from the XML file if provided; otherwise it uses the EEPROM value. The file board_config_p2597-devkit.xml, shown below, illustrates the XML file format.
<?xml version="1.0"?>
<!-- Nvidia Tegra board info configuration file -->
<board_configs>
<board type="proc" id="2180" sku="1000" fab="0" />
<board type="display" id="0000" sku="0000"/>
<board type="pmu" id="2180" sku="0000" />
</board_configs>
This flashing config file p2371-2180-devkit.conf passes the name of the XML file as an option:
BCFFILE="bootloader/${target_board}/cfg/board_config_p2597-devkit.xml";
The file contains the processor module ID (type="proc"), display board ID (type="display"), and power management unit ID (type="pmu"). Since the processor and PMU are on the same module in the development kit, they have the same ID.
To change the UART port
1. In Linux_for_Tegra/<board>.conf, modify the ODMDATA assignment:
ODMDATA=0x60084000;
2. In the U-Boot boot loader, locate the following lines in /include/configs/p2371-2180.h :
#define CONFIG_TEGRA_ENABLE_UARTA
3. Modify that line to specify the UART you want to use. For example, to change UARTA to UARTD:
#define CONFIG_TEGRA_ENABLE_UARTD
4. In the kernel, modify the debug_uartport assignment:
debug_uartport=lsport,0

Flashing the Build Image

When flashing the build image, use your specific board name. The flashing script uses the configuration in the <board>.conf file during the flashing process.
Setting Optional Environment Variables
The flash.sh script updates the following environment variables based on board EEPROM values and other parameters. If you want to override these variables’ values, set them in the board-specific file <board>.conf.
# Optional Environment Variables:
# BCTFILE ---------------- Boot control table configuration file to be used.
# BOARDID ---------------- Pass boardid to override EEPROM value
# BOARDREV --------------- Pass board_revision to override EEPROM value
# BOARDSKU --------------- Pass board_sku to override EEPROM value
# BOOTLOADER ------------- Bootloader binary to be flashed
# BOOTPARTLIMIT ---------- GPT data limit. (== Max BCT size + PPT size)
# BOOTPARTSIZE ----------- Total eMMC HW boot partition size.
# CFGFILE ---------------- Partition table configuration file to be used.
# CMDLINE ---------------- Target cmdline. See help for more information.
# DEVSECTSIZE ------------ Device Sector size. (default = 512Byte).
# DTBFILE ---------------- Device Tree file to be used.
# EMMCSIZE --------------- Size of target device eMMC (boot0+boot1+user).
# FLASHAPP --------------- Flash application running in host machine.
# FLASHER ---------------- Flash server running in target machine.
# INITRD ----------------- Initrd image file to be flashed.
# KERNEL_IMAGE ----------- Linux kernel zImage file to be flashed.
# MTS -------------------- MTS file name such as mts_si.
# MTSPREBOOT ------------- MTS preboot file name such as mts_preboot_si.
# NFSARGS ---------------- Static Network assignments.
# <C-ipa>:<S-ipa>:<G-ipa>:<netmask>
# NFSROOT ---------------- NFSROOT i.e. <my IP addr>:/exported/rootfs_dir.
# ODMDATA ---------------- Odmdata to be used.
# PKCKEY ----------------- RSA key file to used to sign bootloader images.
# ROOTFSSIZE ------------- Linux RootFS size (internal emmc/nand only).
# ROOTFS_DIR ------------- Linux RootFS directory name.
# SBKKEY ----------------- SBK key file to used to encrypt bootloader images.
# FAB -------------------- Target board's FAB ID.
# TEGRABOOT -------------- lowerlayer bootloader such as nvtboot.bin.
# WB0BOOT ---------------- Warmboot code such as nvtbootwb0.bin
Here is an example of environment variable settings:
source "${LDK_DIR}/p2371-2180-devkit.conf";
DTB_FILE=tegra210-jetson-tx1-p2597-2180-a02-devkit-24x7.dtb;
ODMDATA=0x90000;
To flash the build image
Execute the following command:
$ sudo ./flash.sh <board> mmcblk0p1