When attaching the TX1 module to a custom carrier board, the module never lots to GUI. Instead, it opens up “root” access terminal. I have attached an image of the screen and some messages. Of note, the module operates normally when attached to the dev board. I’ll reply with the image.
If you boot to both carrier boards, and run this command, does file “edid” exist on both? Are both finding this file at “/sys/kernel/debug/tegradc.1/”? If found, what is the content of file “edid” on both…do they match, or do they differ? Are both found in the same directory?
find /sys -name 'edid'
Normally I would not expect GPIO changes to be a problem, but changes could if the GPIO had previously been used for some other purpose and then suddenly the other purpose had been broken For example, suppose changing GPIO of something which detects video suddenly went missing…that’s just a contrived example, but it illustrates the idea. From both the developer carrier board and the custom carrier board, what is the content (use “ls -l /sys/class/gpio/*”, you’re just listing files, and for the case of symbolic links, where they point) of “/sys/class/gpio/”.
Note that monitor detection of EDID uses the i2c protocol, so if a change to GPIO broke i2c it might break the creation of the “edid” file (which is a result of monitor EDID query over i2c).
EDIT: On both carrier boards, what is the content of:
GPIOs 0-255, platform/6000d000.gpio, tegra-gpio:
gpio-2 (pcie_wake ) in hi
gpio-28 (usb-vbus3 ) out lo
gpio-56 (wlan_pwr ) out hi
gpio-57 (rtl-5v0 ) out hi
gpio-59 (bt_ext_wake ) out hi
gpio-60 (reset_gpio ) out lo
gpio-61 (bt_host_wake ) in lo
gpio-67 (vdd-disp-3v0 ) out lo
gpio-87 (pwm-fan-tach ) in hi
gpio-148 (cam_reset_gpio ) in lo
gpio-151 (cam_pwdn_gpio ) in lo
gpio-169 (lcd-bl-en ) out lo
gpio-188 (temp_alert ) in hi
gpio-189 (Power ) in hi
gpio-190 (Volume Up ) in hi
gpio-192 (Volume Down ) in hi
gpio-200 (1.extcon ) in lo
gpio-201 (sdhci_cd ) in lo
gpio-203 (en-vdd-sd ) out hi
gpio-204 (sdhci_wp ) in lo
gpio-225 (hdmi2.0_hpd ) in lo
gpio-228 (usb-vbus1 ) in hi
“gpio” [readonly] 46L, 1786C
GPIOs 984-991, platform/max77620-gpio.0, max77620-gpio, can sleep:
gpio-984 (1.extcon ) in lo
gpio-985 (vdd-sys-boost ) out lo
gpio-987 (vdd-3v3 ) out lo
gpio-990 (regulator-pwm ) out lo
gpio-991 (max77620-gpio7 ) out lo
GPIOs 992-1007, i2c/1-0077, tca9539, can sleep:
gpio-995 (en-vdd-cam-1v2 ) out lo
gpio-1001 (en-vdd-cam ) out lo
gpio-1002 (en-vdd-cam-1v2-alt ) out lo
GPIOs 1008-1023, i2c/1-0074, tca9539, can sleep:
gpio-1009 (en-vdd-ts-1v8 ) out lo
gpio-1010 (en-vdd-ts-hv-3v3 ) out lo
gpio-1011 (en-vdd-disp-3v3 ) out lo
gpio-1012 (vdd-fan ) out hi
gpio-1015 (en-mdm-pwr-3v7 ) out lo
gpio-1017 (en-vdd-disp-1v8 ) out lo
gpio-1020 (vdd-hdmi ) out lo
gpio-1021 (en-vdd-cam-hv-2v8 ) out lo
Here’s an important part of the question…I see only one listing of the edid file…is this file an exact match regardless of which carrier board is used?
The files are present in both. After troubleshooting with the connectors the electronics seem fine. I do get to command line, though it never gets to the GUI on the carrier board/tx1, while it continues to work on the dev board. Could this have resulted in part from the update (ubuntu 14.04 to ubuntu 16.06)? I have another tx1 that is running ubuntu 14.04 and the GUI works fine.
How might I fix this? Is there a fix on the 16.06 ubuntu version, or must I try to get ubuntu 14.04 back up and running (I’d prefer not to have to…)?
Just some information: L4T prior to version R24.1 was 64-bit kernel space, 32-bit user space; L4T R24.1 offered two sample rootfs variations…one to maintain 32-bit user space, the other was a new 64-bit user space (most people used 64-bit user space when R24.1 became available). Anything newer than R24.1 will always be 64-bit in both user space and kernel space.
To see version:
head -n 1 /etc/nv_tegra_release
NOTE: It’s important to know if that edid file is the same on both carrier boards or not.
Thanks for the advice. The carrier board does work with a tx module that has the beta version of R24.1. The module that does not bring up GUI but instead, brings me to ‘root command line’ has the latest Jetpack version with ubuntu 16.04
Is serial console available? If so, then it should be possible to log the entire boot process on the carrier board in question.
Having the same edid file present on both systems using the same monitor and cables with the same release of L4T should produce the same results on either carrier board. In this case I’d wonder if something is wired differently on the non-working carrier board, but it seems any possible wiring failure is not catastrophic (or else you wouldn’t even get the root command line).
Note that having a console login show up is normal on a directly connected keyboard/monitor setup if X does not start. However, if this is literally a “root command line” (running as root and not user ubuntu) then boot has stopped at single user stage. Normal user login stage where a password prompt is given is multi-user stage and occurs after single user stage; if in single user stage then the issue is not video at all…no graphics ever run in single user mode…this would indicate your video is completely functional but never reached.
On the command line you see did it autologin? If so, what user is shown from the “whoami” command?