OpenWrt Support for Armor G5 (NBG7815)

Today router rebooted (kernel crash?) itself when changing SQM settings

Feb 11 23:48:59 router-sinan SQM Stopping SQM on wan
Feb 11 23:48:59 router-sinan SQM sqm_stop: Stopping SQM on interface wan (ifb@wan)
Feb 11 23:48:59 router-sinan SQM sqm_stop: Removing kernel module: act_nssmirred
Feb 11 23:49:00 router-sinan SQM sqm_stop: Removing kernel module: qca_nss_qdisc
Feb 11 23:49:00 router-sinan SQM sqm_stop: Removing kernel module: ifb
Feb 11 23:49:00 router-sinan SQM Starting SQM script: nss-zk.qos on wan, in: 83000 Kbps, out: 20000 Kbps
Feb 11 23:49:00 router-sinan SQM sqm_prepare_script: loading required kernel module: qca_nss_qdisc
Feb 11 23:49:00 router-sinan SQM sqm_prepare_script: loading required kernel module: act_nssmirred
Feb 11 23:49:00 router-sinan SQM sqm_start: egress shaping activated
Feb 11 23:49:00 router-sinan root [ethtool] Disabling feature: rx-gro-list: disabled on (ifb@wan)
Feb 11 23:49:00 router-sinan SQM sqm_start: ingress shaping activated
Feb 11 23:49:00 router-sinan SQM nss-zk.qos was started on wan successfully
Feb 11 23:55:51 router-sinan SQM Stopping SQM on wan
Feb 11 23:55:51 router-sinan SQM sqm_stop: Stopping SQM on interface wan (ifb@wan)
Feb 11 23:55:51 router-sinan SQM ERROR: cmd_wrapper: tc: FAILURE (2): /sbin/tc qdisc del dev ifb@wan parent 200:
Feb 11 23:55:51 router-sinan SQM ERROR: cmd_wrapper: tc: LAST ERROR: RTNETLINK answers: No such file or directory
Feb 11 23:55:51 router-sinan kernel [517469.407641] nss_mirred_release[47]:IFB device (null) not found in the linked list
Feb 11 23:55:51 router-sinan SQM sqm_stop: Removing kernel module: act_nssmirred
Feb 11 23:55:51 router-sinan kernel [517469.657468] Unable to handle kernel read from unreadable memory at virtual address 00000002001ffff8
Feb 11 23:55:51 router-sinan kernel [517469.657514] Mem abort info:
Feb 11 23:55:51 router-sinan kernel [517469.665675]   ESR = 0x0000000096000005
Feb 11 23:55:51 router-sinan kernel [517469.668556]   EC = 0x25: DABT (current EL), IL = 32 bits
Feb 11 23:55:51 router-sinan kernel [517469.672094]   SET = 0, FnV = 0
Feb 11 23:55:51 router-sinan kernel [517469.677659]   EA = 0, S1PTW = 0
Feb 11 23:55:51 router-sinan kernel [517469.680597]   FSC = 0x05: level 1 translation fault
Feb 11 23:55:51 router-sinan kernel [517469.684072] Data abort info:
Feb 11 23:55:51 router-sinan kernel [517469.689042]   ISV = 0, ISS = 0x00000005, ISS2 = 0x00000000
Feb 11 23:55:51 router-sinan kernel [517469.691886]   CM = 0, WnR = 0, TnD = 0, TagAccess = 0
Feb 11 23:55:51 router-sinan kernel [517469.697267]   GCS = 0, Overlay = 0, DirtyBit = 0, Xs = 0
Feb 11 23:55:51 router-sinan kernel [517469.702499] user pgtable: 4k pages, 39-bit VAs, pgdp=0000000058ad0000

Use debug on log verbosity to prevent the crash.

1 Like

@asvio I'm compiling from your sources. Why are there two 24.10 branches? nbg7815-24.10-nss and nbg7815-24.10.0-nss, what's the difference?

24.10.0-nss will not have any update. It is a fixed branch base on on openwrt 24.10.0 release tag.

24.10-nss branch will be update following oficial 24.10 branch so this branch will have patches to improve stability and to correct problems that appear with use, but it will not incorporate new functions that may affect stability.

2 Likes

@RaylynnKnight Could you add NBG7815's device page https://openwrt.org/inbox/toh/zyxel/nbg7815_armor_g5 to https://openwrt.org/toh/hwdata/zyxel/start ?

2 Likes

https://openwrt.org/toh/hwdata/zyxel/zyxel_nbg7815_armor_g5 has existed since 2024/06/19. So I'm not sure what your are asking me to do.

You've forgotten to add the device page link


The link you posted is incomplete and contains only basic info. Compare it with the device page link I posted

OK I thought you were talking about the Data Entry page, not the Device Details page. I've now created the Device Details page using the template and copied data over from your inbox page into the appropriate sections. I left your inbox page so that you can check to see if I missed anything. Once you've confirmed that the Device Details page contains all appropriate data then we can delete the inbox page.

It's not my inbox page, it's OpenWRT's original device page for NBG7815.

The new device template doesn't look good enough to me. I copy/pasted the original device page. If this is against OpenWRT's plans for the device pages, please revert it.

There is a reason that the template exists. It provides consistency in the pages and allows for data to be pulled from the Data Entry page. With your latest changes the page will need to be modified at the next patch release. This would not be necessary if you used the page template as requested.

1 Like

Previous page was pulling the firmware release links and hardware details.
Hardware details in the data entry page is not good.
Added the release links section back, so it won't need to be updated with each release.

If the hardware details in the data entry page are wrong then the correct way to handle this is to update the data entry page, not the device details page.

It's not wrong, it's insufficient. For NBG7815 instance
Where can I type chip info?
It has USB3.1 Gen2, not USB3.0
It has RGB led, where can I put it? Is it 1 or more than 20?

@asvio @AgustinLorenzo Some kernel modules are missing in Qualcommax packages repo. Is it okay to add them in advance to the firmware?

CONFIG_PACKAGE_kmod-fuse=y # required for some filesystems like ntfs-3g

I can include it in my next release, but do you really need it?
For the ntfs file system, I'm including the ntfs3 package, which is much faster than ntfs-3g.

If anyone needs any file systems other than those included in my releases, they should consider building their own firmware to suit their needs.

not needed, I'm building my own firmware

Where? I can't find ntfs3

CONFIG_PACKAGE_kmod-fs-ntfs3=y

Ah, it's missing too. Okay let's see how fast it is :slight_smile:

ntfs3 is unbelievably fast :open_mouth: Thank you a lot!

2 Likes

@asvio have you ever encountered the ntfs dirty bit activated? On the internet, they say ntfsfix only clears the dirty bit, rather than fixing the problem. So when the ntfs disk faces a problem, Windows is needed to fix it. Is this true?