[Preview] Asuswrt-Merlin 384.3 pre-release test builds

[HeadBanger]

Exactly the same occurs with the official ASUS firmware.

As to what I’d say - ‘IPTV and AiProtection engaged cause the router CPUs to max out and all traffic grinds to a holt. Can you help?’

HB


Sent from my iPhone using Tapatalk

 

[king006]

I have changed the WiFi channel of the 5Ghz band but the change is not implemented. This will be the same problem as with guest network.
After the command service restart_wireless on the command line, the channel has changed.

Can you figure this out @RMerlin?

I also see channel 36, 36, 40, 40, 44, 44 and then 48, 52, 56.... (Why twice the same number and after 44 no more?) This is when changing the WiFi channel of 5GhZ.

 

[Sinner]

Tried your suggestion, but to no avail. Toggled QoS after about 10 minutes, waited 5 minutes, no charts. Tried @brummygit 's suggestion (which I tried before, but wasn't sure whether that was on 382 or on 384), but that didn't work either. I'm letting QoS for what it is now, let's see how this progresses. I might do a clean install to make sure there are no things from the past causing issues.



One remark which I have to get off my chest: please keep in mind that a large amount of forum members are not native english speakers. I didn't read any actual 'demands' in the post that triggered you, so it might just have come out wrong. I do understand why it could be easily misinterpreted. This comment is only regarding the post which triggered you, I know there are many people who actually do behave like a bunch of morons. I'm a long time forum member myself who had his account removed at some point because, as a non-native english speaker, somehow things repeatedly and unintentedly came out wrong, which caused several frustrating miscommunications and apparently I annoyed some other developers, which was definitely not my intention. I made a new account after several months cause I missed it and now try even more to carefully choose my words, hoping I won't accidently offend anyone. I have nothing but good intentions and come here to learn and, wherever possible, to contribute. Please forgive me for going off-topic, but I just had to let you know.

I'm having similar issues.. my qos statistics page is blank on my ac3100 on 384 a3. tried removing the sig file as merlin suggested and then performed an update from admin.. it did successfully update so I toggled qos off and on again to restart the freshjr 382.3 script I found in another thread but it says no modification ness. the router feels good but I cant tell if the script is working. anyone have any idea if freshjr is compatible with 384 a3?

correction.. running 384.3 a1 apparently... guess flashed wrong version. will load the latest alpha2 and see how it goes.

 

[feranick]

I have changed the WiFi channel of the 5Ghz band but the change is not implemented. This will be the same problem as with guest network.
After the command service restart_wireless on the command line, the channel has changed.

Can you figure this out @RMerlin?

I also see channel 36, 36, 40, 40, 44, 44 and then 48, 52, 56.... (Why twice the same number and after 44 no more?) This is when changing the WiFi channel of 5GhZ.

The same happens to me on the 2.4GHz (384 a2, on rt-68u)

 

[brummygit]

Exactly the same occurs with the official ASUS firmware.

As to what I’d say - ‘IPTV and AiProtection engaged cause the router CPUs to max out and all traffic grinds to a holt. Can you help?’

HB


Sent from my iPhone using Tapatalk

Am I right in remembering you are running BT TV when you have problems? I also run these and would be interested to break out to a new thread for a discussion as I see other problem behaviour with it on my installation.


Sent from my iPhone using Tapatalk

 

[deddc23efb]

Exactly the same occurs with the official ASUS firmware.

As to what I’d say - ‘IPTV and AiProtection engaged cause the router CPUs to max out and all traffic grinds to a holt. Can you help?’

HB


Sent from my iPhone using Tapatalk

There's no harm in asking. The worst they can say is no.

 

[HeadBanger]

Am I right in remembering you are running BT TV when you have problems? I also run these and would be interested to break out to a new thread for a discussion as I see other problem behaviour with it on my installation.

Yes, that’s correct.
Please do and let me know? I’ll happily contribute to the thread.

Thanks,

HB

 

[LostFreq]

@RMerlin In 382.2_beta3 / 384.3_alpha I spotted a few pages with HTML errors. I have listed each fragment with the correction below it:

Advanced_TOR_Content.asp:

<a class="hintstyle" href="javascript:void(0);" onClick="openHint(5,10);">MAC
<a class="hintstyle" href="javascript:void(0);" onClick="openHint(5,10);">MAC</a>

Advanced_OpenVPNClient_Content.asp:

<textarea rows="8" class="textarea_ssh_table"id="edit_vpn_crt_client_extra" name="edit_vpn_crt_client_extra" cols="65" maxlength="3499"></textarea>
<textarea rows="8" class="textarea_ssh_table" id="edit_vpn_crt_client_extra" name="edit_vpn_crt_client_extra" cols="65" maxlength="3499"></textarea>

Advanced_VPN_OpenVPN.asp:

<input type="button" onclick="edit_Keys();" value="Edit..."</td>
<input type="button" onclick="edit_Keys();" value="Edit...">

Advanced_VPNStatus.asp:

<script language-:JavaScript" type="text/javascript" src="/js/jquery.js"></script>
<script language="JavaScript" type="text/javascript" src="/js/jquery.js"></script>

Tools_Sysinfo.asp: (there are two of these lines with a missing space between attributes)

<table width="100%" border="1" align="center" cellpadding="4" cellspacing="0"bordercolor="#6b8fa3" class="FormTable">
<table width="100%" border="1" align="center" cellpadding="4" cellspacing="0" bordercolor="#6b8fa3" class="FormTable">

And a few more. I hope to have contributed something useful to your great project this way.

Tools_Sysinfo.asp:

<th><#General_x_SystemUpTime_itemname#></a></th>
<th><#General_x_SystemUpTime_itemname#></th>

Advanced_VPN_OpenVPN.asp:

<input type="text" maxlength="2" class="input_6_table"name="vpn_server_verb" onKeyPress="return validator.isNumber(this,event);" value="<% nvram_get("vpn_server_verb"); %>">
<input type="text" maxlength="2" class="input_6_table" name="vpn_server_verb" onKeyPress="return validator.isNumber(this,event);" value="<% nvram_get("vpn_server_verb"); %>">

WiFi_Insight.asp:

<div id="tabMenu" class="submenuBlock"></div
<div id="tabMenu" class="submenuBlock"></div>

Advanced_PerformanceTuning_Content.asp:

<meta http-equiv="X-UA-Compatible" content="IE=EmulateIE7, IE=EmulateIE10" />
<meta http-equiv="X-UA-Compatible" content="IE=Edge" />

That last one is not an error correction, but to bring the visual rendering (mainly the vertical spacing) in MSIE in line with the rest of the pages.

 

[RMerlin]

And a few more. I hope to have contributed something useful to your great project this way.

Thanks for the corrections, I'll take a look at them in the coming days.

 

[RMerlin]

In your opinion would it be better to contact Trend Micro directly on this?

No, they do not provide direct customer suppport over their engine, Asus are the ones that need to be contacted.

 

[Adamm]

So I was finally able to get my hands on an AC86U and track down the weird issues introduced with this new device and my script.

One of the major ones seems to be non ascii characters randomly being inserted to files I do various modifications on with grep/sed/echo. The real problem is the null character when being processed by IPSet crashes the router with the following message.

Feb 14 10:00:17 kernel: _ Reboot message ... _______________________________________________________
Feb 14 10:00:18 kernel: Unable to handle kernel NULL pointer dereference at virtual address 00000000pgd = ffffffc006627000[00000000] *pgd=00000000051ec003, *pud=00000000051ec003, *pmd=000000000540b003, *pte=0000000000000000Internal error: Oops: 96000007 [#1] PREEMPT SMPModules linked in: ip_set_list_set init_addr(          (null) -           (null)), core_addr(ffffffbffc312000 - ffffffbffc313a40) ip_set_hash_ip init_addr(          (null) -           (null)), core_addr(ffffffbffc309000 - ffffffbffc30c95
Feb 14 10:00:19 kernel:  x28: ffffffc0058938d4 x27: ffffffc00abb2308 x26: ffffffc00bb3c400 x25: ffffffc0055367d8 x24: 0000000000000001 x23: 0000000000000140 x22: ffffffc0058938d8 x21: ffffffc00a7b00c0 x20: 000000001f910000 x19: ffffffc0055367e0 x18: 0000000000000000 x17: 0000000000000000 x16: ffffffc00039ff18 x15: 4009000800000000 x14: 00000000f7706660 x13: 0000000000000001 x12: 0000000000000028 x11: 0101010101010101 x10: 7f7f7f7f7f7f7f7f x9 : 0000000000000000 x8 : 0000000000000000 x7 : 00000000fffffff
Feb 14 10:00:19 kernel: ____________________________________________________________________________

Now with that being said, I can't reproduce this every time, it happens maybe %60 of the time a new "list" is generated by Skynets banmalware function. This issue is not apparent on any other devices (random characters being inserted that is, I haven't tested yet if IPSet crashes on other devices when trying to list/restore entries with this character in it).

In the example here, 934 "null" characters were added to the end of the file for no apparent reason, sometimes the random characters replace other text too so the location/point of input isn't consistent.

So the short version is, do you have any idea what would be causing an issue of this nature on only this device, and secondly can any mitigation be put in place so at-least IPSet doesn't hard reboot the router when dealing with such characters in both the restore/save commands.

IPSet v6.35 is also out, I don't see any changes directly related to an issue of this nature but potentially it has be addressed.

I will try my best to track it down further, but the hard reboots make it quite a slow process.

To add to my last post, it seems IPSet restore command is the most common point of failure. The random text isn't the only thing causing an issue, some other times on a perfectly clean file I will get reboots with the following message.

Feb 14 10:00:17 kernel: _ Reboot message ... _______________________________________________________
Feb 14 10:00:18 kernel: Unable to handle kernel paging request at virtual address fffffffepgd = ffffffc006669000[fffffffe] *pgd=000000000b061003, *pud=000000000b061003, *pmd=0000000000000000Internal error: Oops: 96000046 [#1] PREEMPT SMPModules linked in: xt_set init_addr(          (null) -           (null)), core_addr(ffffffbffc30f000 - ffffffbffc310050) ip_set_list_set init_addr(          (null) -           (null)), core_addr(ffffffbffc161000 - ffffffbffc162a40) ip_set_hash_ip init_addr(          (nu
Feb 14 10:00:18 kernel: 01140 x25: ffffffc01ffeee98 x24: 000000000000003c x23: 000000000000003c x22: ffffffc01e800100 x21: ffffffc0064e39e0 x20: 0000000000000001 x19: 0000000000000005 x18: ffffffbe00467c28 x17: 00000000fffffffe x16: ffffffc01e801168 x15: 0000004000000000 x14: dead000000000100 x13: dead000000000200 x12: ffffffc0006e3000 x11: ffffffc01e801148 x10: ffffffbe00000000 x9 : ffffffc01e801140 x8 : 0000000000000003 x7 : 0000000014237500 x6 : 0000000000508dd4 x5 : 000000000a11ba80 x4 : ffffffc006
Feb 14 10:00:18 kernel: ____________________________________________________________________________

Some good news though is the fork() issues don't seem to be apparent anymore that users were running into on earlier firmware's, or at-least I can't reproduce them.

@RMerlin if there's anything I can do to help track down what's exactly wrong with IPSet on this device (again its not apparent on other models), let me know.

 

[RMerlin]

@RMerlin if there's anything I can do to help track down what's exactly wrong with IPSet on this device (again its not apparent on other models), let me know.

I currently have my hands full with other things so I won't have time to look into this for a while. One place to start if you want to dig deeper would be the Kernel repo. Look for any commits to the ipset code following 4.1.21, might be something that was fixed upstream in a newer 4.1.xx update.

 

[john9527]

@RMerlin if there's anything I can do to help track down what's exactly wrong with IPSet on this device (again its not apparent on other models), let me know.

First thing to try would be to back out the commits I made to get ipset to work on the earlier kernel (worked around some missing rpc function that should now be present in the 4.1 kernel).

 

[Gregory Phillips]

Could be a mistake on Asus's update servers. I run my own update server (for obvious reasons).

You are probably right. I've tried looking at it several times and I keep getting different content when I visit that page on the router.

 

[Vexira]

OK thanks; I feared that this would be the case.

I don’t really want to switch it off as I have two young children and their friends using the WiFi and like the first line of defence that it offers. In your opinion would it be better to contact Trend Micro directly on this?

Many thanks,

HB

contact asus, if you look in the stock firmware section of the forum there are some asus staff you could ask.

 

[Gingernut]

Last night I had severe Netflix buffering, which made it almost unwatchable, for the first time since installing an 86U.

Logged into the router, which was even terribly slow to even change tabs, and saw the router logs flooded with these messages:

WARNING: Bad encapsulated packet length from peer (5635), which must be > 0 and <= 1626 -- please ensure that --tun-mtu or --link-mtu is equal on both peers -- this condition could also indicate a possible active attack on the TCP link -- [Attempting restart...]

I do have an OpenVPN server running on the router to connect from work, over TCP & port 443, but I was at home at the time.

TCP attack?

 

[DonnyJohnny]

443 tcp is a common port used for port sniffing and Attack. The unauthorised entry may have make your router work slightly harder, trying to authenticate the packet coming in. But of coz it will be reject.

Suggest either change port to non standard port or you can try install Skynet to block common port scanner / attacker ip even before they reach your door. They will be stop at gate.

Currently massive ddos activity going on around the world, it could also be another factor why internet is slower.

 

[Gingernut]

On the OpenVPN subject,

the exported OpenVPN configuration file gets your ddns info automatically appended to it, but if you use dnsomatic, I use it for OpenDNS and dns filtering setup on the router, it gets used but isn't your ddns address but more a ddns update server service.

The other day I spent quite a long time debugging my OpenVPN connection error until I realized what had happened.

Would it be possible to add a ddns address option section to the OpenVPN setup page?

 

[djrm]

OK thanks; I feared that this would be the case.

I don’t really want to switch it off as I have two young children and their friends using the WiFi and like the first line of defence that it offers. In your opinion would it be better to contact Trend Micro directly on this?

Many thanks,

HB

If you are able to reproduce the issue with the latest asus fw, install it, reproduce the issue and then save a copy of your router configuration as asus will ask for it later. Open a case with them and be bit patient... I opened a case for the same issue on AC88u for the 382 and they sent me after few emails a modified version of 382 base which seems that fixed the related issue and also the new 384 base fw was released.

 

[Adamm]

First thing to try would be to back out the commits I made to get ipset to work on the earlier kernel (worked around some missing rpc function that should now be present in the 4.1 kernel).

Good call, would make sense as IPSet was directly ported over from the 380 branch. Unfortunately I'm no good with C so I'd end up doing more harm then good, so guess I'll have to wait this one out.