Is there a way I can execute this script on a pre-built image?
P.S. Is there a switch layout for this router so I know what each internal port corresponds to the physical port? If so would it be worth adding to the device Table of Hardware page found here?
I've only ever used the image builder several times often placing pre-configured files in the files directory so that they get baked into the firmware. Is what you're mentioning easy to do? I have modified the MWLWIFI code to use the latest commit before. Is it similar?
It would be easier to just use the prebuilt with the patch... Or if you want, to configure the switch from the command line.
Anyway, patching and building a kernel it's not the same as using the image builder. Why? Because the patched is applied to the kernel source code, not to the kernel executable. You'll need to build a custom image from source.
I'm willing to try both. Does the patch fix LuCI? Failing that with some guidance on the VLAN setup I'm happy to configure it through the backend.
I'm just trying to reverse engineer eginnc's/etc/config/network file to see what port they used for the trunk line.
UPDATE
From looking over the config port 1 is tagged option ports '1t' across all of the VLANs. Port 0 I assume is the CPU for the software VLANs and if WAN port was listed we would have 6 options rather 5?
Yes, that's more or less how it works in the OpenWrt default. In fact, you'll need to double tag your ports.
You'll need to tag port 5 (the wan port) always. This is because, in reality, eth0 is not LAN but CPU and eth1 doesn't exist. However, the driver will "treat the wan port as eth1", then you'll need to tag both and select the corresponding interface (eth0.x and/or eth1.x according to your needs).
And yes, in my ROM vlan works as expected in LuCI.
I misunderstood. The configuration used by my testers and by myself is not compatible with OpenWrt Vanilla and will render the device unusable until a factory reset (or a network reset) is performed from Fail Safe mode. Therefore, I would not recommend trying it.
What I said: if you just want to add extra packages to or remove from your image, or include some extra scripts you made yourself, the image builder suffices.
git branch release
git checkout openwrt-19.07
./scripts/feeds update -a && ./scripts/feeds install -a
make menuconfig
make defconfig
make download
git appy --stat 715-net-essedma-disable-default-vlan.patch
git appy --check 715-net-essedma-disable-default-vlan.patch
the last command reports
git apply '715-net-essedma-disable-default-vlan.patch'
error: arch/arm/boot/dts/qcom-ipq4019.dtsi: No such file or directory
error: drivers/net/ethernet/qualcomm/essedma/edma.c: No such file or directory
After searching the root build directory I did accidentally come across this directory with other patches in ~/openwrt/target/linux/ipq40xx/patches-4.14
I was wondering if I could move the physical file into there and when it compiles it gets applied?
This is the board.d\02_network file correct? If so do I need to append a .patch file extension to the file and place it in the `patches-4.14' directory?
Nope. It's not a patch to the kernel. You'll need to replace the file inside the "base-files" package for the device. Not as a patch, a replacement of the whole file.
That's an option too. I would prefer to replace the original file to keep track of the changes as the files folder is ignored by git. But it's totally doable and the effect is the same.ñ
Okay I have just compiled the firmware and copied the file to the correct folder but I now have no switch page on LuCI.
If I type the address in manually it's just a blank page. I only thing I can think of is I need to change the permissions on the file.
Advanced Reboot also seems to be doing weird things. OpenWrt was on partition 1 so I clicked the button on the other partition Reboot to alternative partition.... A error page displayed and when back to the IP address to gain access to LuCI, went back to Adcanced Reboot and now both partitions are OpenWrt. I re-load the page again and the firmware switches back; one saying OpenWrt and the other Linksys.
I will re-compile now to see if it makes any difference and update with the verdict.