Missing Modules

[pilotnbr1]

In an effort to document and stay organized it pains me to open this 😔

[pilotnbr1]

Just started looking into the 88xx problem. Anyone aware of some of the problems reported by 5.3.4? Sorry if this is already mentioned. I am just getting up to speed..
t6x/reaver-wps-fork-t6x#265 (comment)

[pilotnbr1]

To further resolve this issue I am wondering if journald (or perhaps some other log) should be enabled to get more log output.

[Yes21]

The issue is : with 5.3.4 on the airPi, no video transmission to the ground and error message "injection failed".

[RespawnDespair]

Its definitely a better decision to move back to 5.2.20, there appear to be more people with issues (in regards to monitor mode etc.):
t6x/reaver-wps-fork-t6x#265
I will try to go back while maintaining all the current improvements.

[Yes21]

I agree. Thanks.
I will test and confirm if it works again.

[pilotnbr1]

To go back to 5.2.20 is it as simple as changing the git request? Or is it more involved than it appears?

[RespawnDespair]

A little more involved, since the rest of the modifications still needs to work

[pilotnbr1]

@JelleTigchelaar just want to point out that the makefile for 5.2.20 now (as of a week ago) has option specific to pi3b+

CONFIG_PLATFORM_ARM_RPI = n
CONFIG_PLATFORM_ARM64_RPI = n

[pilotnbr1]

something else interesting- it has been some time since I have gone thru the terminal output while an image is being built. So here is some output with warnings of duplicate module aliases when depmod is called

/usr/bin/make -C /lib/modules/4.14.71+/build M=/home/pi/flir/v4l2loopback modules_install
make[1]: Entering directory '/usr/src/linux-headers-4.14.71+'
INSTALL /home/pi/flir/v4l2loopback/v4l2loopback.ko
DEPMOD 4.14.71+
make[1]: Leaving directory '/usr/src/linux-headers-4.14.71+'
Unknown host QEMU_IFLA type: 47
Unknown host QEMU_IFLA type: 48
Unknown host QEMU_IFLA type: 43
Unknown host QEMU_IFLA type: 47
Unknown host QEMU_IFLA type: 48
Unknown host QEMU_IFLA type: 43
Unknown host QEMU_IFLA type: 47
Unknown host QEMU_IFLA type: 48
Unknown host QEMU_IFLA type: 43
depmod: WARNING: duplicate module alias:
usb:v043Ep7A32ddcdscdpiciscipin rt2800usb
depmod: WARNING: duplicate module alias:
usb:v0BDAp8176ddcdscdpiciscipin 8192cu
depmod: WARNING: duplicate module alias:
usb:v0BDAp8191ddcdscdpiciscipin 8192cu
depmod: WARNING: duplicate module alias:
usb:v0BDAp8813ddcdscdpiciscipin 88XXau
depmod: WARNING: duplicate module alias:
usb:v0BDApA811ddcdscdpiciscipin 88XXau
depmod: WARNING: duplicate module alias:
usb:v2001p3313ddcdscdpiciscipin 88XXau
depmod: WARNING: duplicate module alias:
usb:v0BDAp881Addcdscdpiciscipin 88XXau
depmod: WARNING: duplicate module alias:
usb:v0BDAp8812ddcdscdpiciscipin 88XXau
depmod: WARNING: duplicate module alias:
usb:v0421p00E9ddcdscdpic02isc02ipFFin* cdc_acm

[pilotnbr1]

So to get to the bottom of our missing modules I have attempted some fairly aggressive rollbacks on the code.

I have gone so far as to completely comment out all purges, auto-remove, completely deleted stage 4 04-run.sh.

I am about to comment out
DEBIAN_FRONTEND=noninteractive sudo apt-get -yq install raspberrypi-kernel-headers

Comparing @Yes21 11-05 image there are quite a few more drivers present within /rootfs/lib/modules/4.14.71-v7+/kernel/drivers than any new image I create. lsmod on a live image also confirms missing modules in the newer images

Ideas? not sure what changed to screw this up

[pilotnbr1]

Comparing the stage 2 kernel config file in @JelleTigchelaar repo and my older cloned repo there is a difference at line 1170.

my repo-
"# CONFIG_RTL8XXXU is not set"
"CONFIG_RTL8812AU=y"

this repo-
"# CONFIG_RTL8XXXU is not set"
"# CONFIG_RTL8812AU=y"
"CONFIG_RTL8821AU=m"

rtl8812au=y Its commented out. dont know if the patch comes later or some other process changes it down the road but it is a difference in the 2 repo files.

[Yes21]

I can confirm that the latest commit (ab0206d) does this #CONFIG_RTL8812AU=y.

But @RespawnDespair did explain me that's not an error :

Yves Chaufour
Jelle I've found 3 errors were you wrote 8821 instead of 8812. But now the driver is not compiling anymore !!
Jelle Tigchelaar
Those are not errors. The entire set if modules us 8821au to the kernel config. Within the makefile for the 88xx you determine the actual support. So i dont think that is a problem. The gnd loads the driver for you 8812 doesnt it?

[pilotnbr1]

@Yes21 thanks for the reply. I am sure @JelleTigchelaar is right then... I am running out of ideas and just going earlier in the stages to figure it out.

I’ve already been building an image with older config files (that appear correct) to see what that does. Sounds like it will not fix the problem

[Yes21]

But I don't understand why @RespawnDespair did change to 8821 for the v5.3.4 branch of the 8812 driver ...