Editing a file might be the problem. Many configuration items have dependencies, and simply editing a file would leave those out with a result of failing to build the module (or perhaps building the module, but failing to load). Even if you did not have a dependency issue, then the module might still fail to load without a matching CONFIG_LOCALVERSION.
You might find this useful:
https://forums.developer.nvidia.com/t/topic/262647/12
I will suggest to revert the change to tegra_defconfig. Set up compile to an empty outside directory with the “O=/some/where” option. Then do the following (you’d still need to use other steps for cross compile if cross compiling, I’m just shortening the commands):
make O=/some/where tegra_defconfigmake O=/some/where nconfig
(you could use “menuconfig”, but I likenconfigsince it has a symbol search)- Save a backup of the original
tegra_defconfig. - Copy the “
/some/where/.config” to thetegra_defconfig.
Once that is in place you’ll know that any edits via the editor properly dealt with dependencies.
If you are only building modules, then you might set up “CONFIG_LOCALVERSION=-tegra” (which changes where modules are searched for).
Then build. If the module is still not present, then there might be an actual problem.