OpenWrt Forum Archive

Topic: davidc502 1900ac 3200acm builds

The content of this topic has been archived between 26 Feb 2018 and 7 May 2018. Unfortunately there are posts – most likely complete pages – missing.

Notsofast wrote:

Hi, newbie thinking of trying a wrt1900acv2 build.  Did not see a search for specific build issues for the ACV2 but it seems the ACV1 is or has been a problem child. Since this thread is fairly huge, i figured just ask 1st, as i have a good idea on how to install and remove if needed.

I also have another question... is it me or just an oddity with ACSV1 build being about the same size as the ACV2, so is there minor partion differences between them since hardware is the same? I noticed a bunch of factory refurbished ACV2  being unloaded below $90 and the ACSV2 routers below $110.

Yes, the acs version 1 is different than the ac version 2.  You can read about some of the differences here. ->
https://wiki.openwrt.org/toh/linksys/wr … d_hardware

If possible I recommend staying away from the 1900ac Version 1 as it is having issues with the latest kernel.  AND I would highly recommend a 1900acs (V1 or V2) over the ac V2 if you can get it.

Hello,
I'm WRT3200ACM user with latest firmware (27-12-17)
I'm trying to setup a guest WiFi at 2.4GHz, and another hidden WiFi at 2.4 in the some network of my LAN and 5GHz WiFi.
The problem is the mutiple SSID at 2.4GHz, only one work... The first WiFi to receive a client.
Is a know bug this?

Thanks

@ makedir

Greetings makedir-

Are you still having difficulty with your WRT1200AC? I am running an older build r4901 without any problems at all. I don't have a VPN or anything complicated so that might be the reason. Your latest comment about solving your reboot problem is good news and interesting because it involves the sleep probably using the kernel timer or software clock.

Have you considered building your own version of LEDE? The are a few people here who do that and have said the build isn't that difficult. I have considered trying that except am busy with other projects.

I think the advantage is that you could tailor your build to omit the things you don't use or don't want. Surely there is a lot of bloat because davidc502 tries to build for everybody wanting anything. The bloat can cause problems because the unused capability can conflict with the wanted capability. Configuring all of those many features can also be a problem because the combinational complexity is exponential making me want to simplify as much as possible.

An additional benefit is being able to examine the source code to find the cause of the problems. That code can point to the developers working on the code and make reporting problems to the developers more helpful. The code can show comments and explanations or modifications they have tried and be very informative for solving problems. Reporting problems and solutions to the developers can also encourage patches leading to a desired feature or result showing up in the final build.

There are instructions for building the LEDE and davidc502 has posted the configs he uses so that should make beginning less difficult. Be sure to start with a powerful machine to avoid waiting for builds. Just building for the WRT1200AC without the other builds would also save time.

Any ideas?

makedir wrote:
meffovic wrote:

EDIT: I can confirm that the reboot issue isn't solved with this build - yet.
I'm pretty sure I triggered a reboot when accessing the luci from my android unit - this with only 5ghz (2.4ghz disabled) band enabled.. sad

This seems to be a different issue or a memory corruption problem, which no one cares on here to follow. Because I had the same issue with Luci so far. Two times, when I opened Luci, the same second, the router rebooted. I am certain, that there is something wrong with the memory on these routers (this rom). Maybe the memory is wrongly clocked and overclocked with kernel 4.9, leading to rare memory corruptions. For example, my router hasnt rebooted anymore, since I deactivated a script of mine. It is a self written watchdog for OpenVPN, and it uses "sleep 20" in it. I had two times now this in the logs:

[318506.951336] BUG: Bad page map in process sleep  pte:00060000 pmd:1bd74831

Since I deactivated the watchdog script, the router has not rebooted anymore. The sleep is triggered every 20 seconds in the script, which might cause a memory corruption way sooner than not using the script.


Right, I can believe that. *But*, this happen once for me.
All the other reboots have been triggered when client(s) connected to the wifi, almost always the 2.4ghz radio.
Let's say I've had close to 100 unwanted reboots, and once of those have been from accessing luci, and it was from a 4.9k as you mentioned. But I've tried other 4.9k builds too, and I didn't get any reboots from accessing luci with those.
So yeah, there might be an additional issue, causing memory corruption. But the main issue is still from the wifi, from what I can understand. I never had a reboot with the radios disabled, so..

Anyhow, I wanna add something. I'm getting the sense, that the country region and/or the dbi have something to do with the reboots (at least with the reboots caused by the wifi). When I'm running the radio set to the default country region, and the transmission power set to auto - I haven't been able to reproduce any reboots. If this is a coincidence, or if it actually has something to do with the problem, I do not know. Maybe worth looking in to?

Best regards - meffe

EDIT: may I ask where that openvpn script is located, which included 'sleep 20'?
cause I've seen that 'sleep 20' running on my router too, maybe I could stop the script myself - and have some success avoiding some reboots.. thanks

(Last edited by meffovic on 30 Dec 2017, 11:30)

davidc502 wrote:

Build r5621 has been uploaded to the server.

I've patched the 1900ac Version 1 that helps to resist the random reboot issue. However, keep in mind, that V1 owners use this build at their own risk. I don't own a V1 and can not verify any part of this patch including if it has any adverse affects or not, so please keep this in mind. Other V1 owners have expressed a slight temperature increase, so I recommend if you're gong to use it, keeping an eye on CPU temperature.

This build does include a few kernel bumps and various fixes.

*EDIT*

For V1 owners please verify the package repository is working correctly.

@davidc502
Thx for this new build, I can confirm the added ipv6 packages are working great.

Router: WRT1900ac v1
Firmware: @davidc502 build Lede SNAPSHOT, r5621-1064e76e4e
Status: STABLE, I'm also experience the "signature check failed" messages when using the package manager.

davidc502 wrote:
Notsofast wrote:

Hi, newbie thinking of trying a wrt1900acv2 build.  Did not see a search for specific build issues for the ACV2 but it seems the ACV1 is or has been a problem child. Since this thread is fairly huge, i figured just ask 1st, as i have a good idea on how to install and remove if needed.

I also have another question... is it me or just an oddity with ACSV1 build being about the same size as the ACV2, so is there minor partion differences between them since hardware is the same? I noticed a bunch of factory refurbished ACV2  being unloaded below $90 and the ACSV2 routers below $110.

Yes, the acs version 1 is different than the ac version 2.  You can read about some of the differences here. ->

If possible I recommend staying away from the 1900ac Version 1 as it is having issues with the latest kernel.  AND I would highly recommend a 1900acs (V1 or V2) over the ac V2 if you can get it.

Ah, so the ACS V1 and V2 are better than the ACV2?  Everything points to them 3 revisions being the same internals, just minor changes with firmware and how gain tables are stored.

Been in contact with Untangled, and it seems some have installed ACS firmware on ACV2 routers. Though cannot tell if they used the wrong usb stick, but i have seen corruption just by messing with wireless settings on the newer factory firmwares. Thus another reason to go with openwrt.

Actually had my new unit reboot and reset after a simple channel and frequency change. Plus 5ghz would drop when changing any setting in the 5ghz section.

So any way after a fresh reload of firmware and restoration of settings, i am plugging along, while waiting for break in time to pass. If anything fails within the first 30 days, it will be noticed, before dealing with open firmwares.

Notsofast wrote:
davidc502 wrote:
Notsofast wrote:

Hi, newbie thinking of trying a wrt1900acv2 build.  Did not see a search for specific build issues for the ACV2 but it seems the ACV1 is or has been a problem child. Since this thread is fairly huge, i figured just ask 1st, as i have a good idea on how to install and remove if needed.

I also have another question... is it me or just an oddity with ACSV1 build being about the same size as the ACV2, so is there minor partion differences between them since hardware is the same? I noticed a bunch of factory refurbished ACV2  being unloaded below $90 and the ACSV2 routers below $110.

Yes, the acs version 1 is different than the ac version 2.  You can read about some of the differences here. ->

If possible I recommend staying away from the 1900ac Version 1 as it is having issues with the latest kernel.  AND I would highly recommend a 1900acs (V1 or V2) over the ac V2 if you can get it.

Ah, so the ACS V1 and V2 are better than the ACV2?  Everything points to them 3 revisions being the same internals, just minor changes with firmware and how gain tables are stored.

Been in contact with Untangled, and it seems some have installed ACS firmware on ACV2 routers. Though cannot tell if they used the wrong usb stick, but i have seen corruption just by messing with wireless settings on the newer factory firmwares. Thus another reason to go with openwrt.

Actually had my new unit reboot and reset after a simple channel and frequency change. Plus 5ghz would drop when changing any setting in the 5ghz section.

So any way after a fresh reload of firmware and restoration of settings, i am plugging along, while waiting for break in time to pass. If anything fails within the first 30 days, it will be noticed, before dealing with open firmwares.

"better" is subjective.  It's been well over a year but seem to remember a few people pointing out small differences between the two. I haven't seen those type of arguments in a while, but seems at times people are drawn towards arguing minutia. What I know is that I've been very pleased with the acs V1, but your experiences may differ. If you can get away with running the acs firmware on the ac V2 then I suppose that is okay, though I'm not sure what is gained unless something isn't working right with the other firmware.

davidc502 wrote:
Notsofast wrote:
davidc502 wrote:

Yes, the acs version 1 is different than the ac version 2.  You can read about some of the differences here. ->

If possible I recommend staying away from the 1900ac Version 1 as it is having issues with the latest kernel.  AND I would highly recommend a 1900acs (V1 or V2) over the ac V2 if you can get it.

Ah, so the ACS V1 and V2 are better than the ACV2?  Everything points to them 3 revisions being the same internals, just minor changes with firmware and how gain tables are stored.

Been in contact with Untangled, and it seems some have installed ACS firmware on ACV2 routers. Though cannot tell if they used the wrong usb stick, but i have seen corruption just by messing with wireless settings on the newer factory firmwares. Thus another reason to go with openwrt.

Actually had my new unit reboot and reset after a simple channel and frequency change. Plus 5ghz would drop when changing any setting in the 5ghz section.

So any way after a fresh reload of firmware and restoration of settings, i am plugging along, while waiting for break in time to pass. If anything fails within the first 30 days, it will be noticed, before dealing with open firmwares.

"better" is subjective.  It's been well over a year but seem to remember a few people pointing out small differences between the two. I haven't seen those type of arguments in a while, but seems at times people are drawn towards arguing minutia. What I know is that I've been very pleased with the acs V1, but your experiences may differ. If you can get away with running the acs firmware on the ac V2 then I suppose that is okay, though I'm not sure what is gained unless something isn't working right with the other firmware.

Well, it is primarily the option for firmwares that do not cover ACV2 due to support from the developer. As for Untangled only supports the newer ACS V1 and V2 with just one firmware for all. But, that is due to restructuring the partitions to be read through USB 2 port or ESATA drive connection. I have not bothered to dive in and compare build for build between ACV2 to the ACSV1, but it seems possible to load one for the other... just no guarantees.  Now if there was a strict build for ACSV2, then that would kill any thoughts for even trying.

I figure, if It does brick, i should be able to recover via console and load a factory firmware, or have a boot fail reset to secondary partition.

Since you do have firmwares for each model, the issue is just a matter of curiosity into how close the two devices are similar.

Hi all,

Thanks David for this great firmware! It's been quite solid on my 3200acm, especially compared to the current state of the stock firmware :)

I'm having an issue with a connected device which refuses to connect to the 2.4ghz wifi network (the device does not support 5ghz anyway), while it works perfectly on the stock firmware (it doesn't work with dd-wrt either, though).
Being a connected device thing, the only feedback I get from it is "could not connect", which is not very helpful.
It seems the device is using an Espressif ESP8266 chip and the software is based on the esp2866/Arduino library on github (can't post links for now due to new account).

I've tried to use tcpdump to see if DHCP packets were going through or not using tcpdump + wireshark, but I have seen none
[img]i.imgur.com/o1xRZwf.png[/img]

Is there anything I can do to further debug this issue? Unfortunately I don't think I can run tcpdump on the stock firmware (no ssh access, right?).

Thanks!

Quick edit: I just tried WPA2 vs WPA vs no encryption, and the device can connect successfully when there is no encryption but not whenever there is encryption.

(Last edited by iXce on 30 Dec 2017, 19:26)

Hi Folks,

I am new to the forum, as I am trying LEDE for the holidays, and I really like it.  David, you have put something together really nice here for us WRT users.  I admit that the router is running better and smoother than ever and I have been using dd-wrt pretty faithfully for about 4 years. 

My question to the group is pretty simple and I have been looking around for the answer to no avail yet.  I am looking about how to extend a guest network on my wired access point.  Right now I have two WRT3200s set up - one is the gateway, the other is bridged on a wired lan.  The first one is set up fine and I found a link to a page here to set up the guest network.  I put the same code into the bridge and cannot get to the internet.  The DHCP settings got complicated.  Can anyone send me a link or some information about how to extend the guest network to the second router.  I am using the same SSIDs on each unit.

Thanks for the help.  I know I have a lot of studying to do, and I am looking forward to it.

UPDATE:

I worked on this for two+ days and finally figured it out with a little help from searching this forum.  If anyone else is having this problem, look at this link: forum.openwrt.org/viewtopic.php?id=72537 - valuable information there.

Now I have two guest networks on each router.  Good to go.

(Last edited by dex06 on 1 Jan 2018, 00:20)

iXce wrote:

Hi all,

It seems the device is using an Espressif ESP8266 chip and the software is based on the esp2866/Arduino library on github (can't post links for now due to new account).

Quick edit: I just tried WPA2 vs WPA vs no encryption, and the device can connect successfully when there is no encryption but not whenever there is encryption.

https://github.com/kaloz/mwlwifi/issues/170

I see the following error in System Log:

Sat Dec 30 17:43:57 2017 daemon.info dnscrypt-proxy[2809]: dnscrypt-proxy Refetching server certificates
Sat Dec 30 17:44:12 2017 daemon.err dnscrypt-proxy[2809]: dnscrypt-proxy Unable to retrieve server certificates

Endryu wrote:

I see the following error in System Log:

Sat Dec 30 17:43:57 2017 daemon.info dnscrypt-proxy[2809]: dnscrypt-proxy Refetching server certificates
Sat Dec 30 17:44:12 2017 daemon.err dnscrypt-proxy[2809]: dnscrypt-proxy Unable to retrieve server certificates

Be sure to refresh the resolver list in the Services/DNScrypt-Proxy tab and then try some more servers. That fixed my same error report.

Well.... I revert my 1900ACv1 back to 4.4.
It's not the stability issues, stability is ok, but other:

1. wifi ping but returning back - so ping time is unstable, and connection is 3 times slower according to iperf.
2. FCC-inspired changes in driver make wifi coverage half decreased (BTW, how they can rule this for other countries? Seems like "open source" is so "open" by the fact).
3. Errors in SATA driver:

Sun Dec 31 07:52:30 2017 kern.err kernel: [52991.049907] ata1: exception Emask 0x10 SAct 0x0 SErr 0x100000 action 0x6 frozen
Sun Dec 31 07:52:30 2017 kern.err kernel: [52991.057260] ata1: edma_err_cause=00000020 pp_flags=00000003, SError=00100000
Sun Dec 31 07:52:30 2017 kern.info kernel: [52991.064357] ata1: hard resetting link
Sun Dec 31 07:52:30 2017 kern.info kernel: [52991.564169] ata1: SATA link up 3.0 Gbps (SStatus 123 SControl F300)
Sun Dec 31 07:52:30 2017 kern.info kernel: [52991.644385] ata1.00: configured for UDMA/133
Sun Dec 31 07:52:30 2017 kern.info kernel: [52991.644395] ata1: EH complete

So, i decide to stay at 4.4 until new router or other firmware without this limitations.

T-Troll wrote:

Well.... I revert my 1900ACv1 back to 4.4.
It's not the stability issues, stability is ok, but other:
2. FCC-inspired changes in driver make wifi coverage half decreased (BTW, how they can rule this for other countries? Seems like "open source" is so "open" by the fact).

Are you sure you have the WRT1900ACv1?

There should be no FCC limitations either for the 4.4 or the 4.9 kernel builds for the WRT1900ACv1. My WRT1200ACv1 is a newer design than the WRT1900ACv1 and still does not have the FCC restrictions while the WRT1200ACv2 does have them. I have the transmit power below the maximum with my v1 where there is no way to make those changes with the more recent v2. I prefer to run cool when the higher power is unneeded.

(Last edited by beginner67890 on 31 Dec 2017, 03:13)

beginner67890 wrote:
T-Troll wrote:

Well.... I revert my 1900ACv1 back to 4.4.
It's not the stability issues, stability is ok, but other:
2. FCC-inspired changes in driver make wifi coverage half decreased (BTW, how they can rule this for other countries? Seems like "open source" is so "open" by the fact).

Are you sure you have the WRT1900ACv1?

There should be no FCC limitations either for the 4.4 or the 4.9 kernel builds for the WRT1900ACv1. My WRT1200ACv1 is a newer design than the WRT1900ACv1 and still does not have the FCC restrictions while the WRT1200ACv2 does have them. I have the transmit power below the maximum with my v1 where there is no way to make those changes with the more recent v2. I prefer to run cool when the higher power is unneeded.

That's a good question.

The 1200ACv2, 1900ACSv2 and the 3200acm and newer routers will have the power settings on chip. This limits the power output to FCC designated power levels on all frequencies. This change didn't affect me, but it could affect other people used to using higher power levels. All newer routers firmware will already play by the rules, but this change forces 3rd party software to abide by the rules if they want to or not smile lol.

sommlov wrote:
iXce wrote:

Hi all,

It seems the device is using an Espressif ESP8266 chip and the software is based on the esp2866/Arduino library on github (can't post links for now due to new account).

Quick edit: I just tried WPA2 vs WPA vs no encryption, and the device can connect successfully when there is no encryption but not whenever there is encryption.

github.com/kaloz/mwlwifi/issues/170

Thanks, that's extremely relevant smile
I have disabled WMM, forced CCMP/AES and my device seems to connect and work alright in this configuration. However I'm not sure this is the proper fix/workaround, since the mwlwifi contributor yuhhaurlin advises against disabling WMM. I've setup a specific AP (wlan1-1) for this device, so it will be the only one using these settings. Are there any known consequences of disabling WMM that may affect security or other devices using the other APs?

Is there any better workaround/fix that I missed in the thread? (using radio2 does not seem to be an option as it used to cause me a lot of trouble)

wrt1200ac v1: wifi coverage and/or signal strength also decreased in latest build from Dec 27.

edit:
any news on the RTC issue? Still having a system time of 2037 after bootup until NTP clients kicks in....

(Last edited by tcsoft on 31 Dec 2017, 10:11)

For those, whose router periodic reboots. I noticed, that in recent builds, there is new service, that checks internet connection, and reboots it in case of failure.

Services -> Watchcat

Althought default settings is pretty fail safe, you can try to disable it.

tcsoft wrote:

wrt1200ac v1: wifi coverage and/or signal strength also decreased in latest build from Dec 27.

edit:
any news on the RTC issue? Still having a system time of 2037 after bootup until NTP clients kicks in....

Is the coverage signal strength related to transmit power? Did that change for the same power output? If you installed the newer build using backup and restore settings then maybe the settings changed by themself so could you check the Network/Wireless/Radio transmit power. Did that change causing the difference? These issues affect my plans to upgrade my WRT1200ACv1. I use 5Ghz 80211n at 50Mw now with a fairly strong signal strength at about 10 meters.

The RTC was disabled a month ago or more so rebooting the router will cause the kernel to set the system time to a dummy time until the NTP resets the system time to a real value. So what you see now is the new norm since those changes to the RTC.

vbatichko wrote:

For those, whose router periodic reboots. I noticed, that in recent builds, there is new service, that checks internet connection, and reboots it in case of failure.

Services -> Watchcat

Althought default settings is pretty fail safe, you can try to disable it.

In LUCI?  I have nothing called "Services"

(Last edited by kirkgbr on 31 Dec 2017, 17:07)

@beginner67890

according to your signature you're on r4901. I guess you should stay on this build for now.
I'm experiencing troubles with the last few builds (from late november on).
biggest issue is, after a few hours of uptime DHCP stops working properly.
the router is sending out DHCPOFFER, but not receiving back DHCPACK.
so my clients will not get IP addresses anymore.
Nothing helps (firewall restart, dnsmasq restart, ...) - just reboot fixes the issue.
Really strange...

beginner67890 wrote:

The RTC was disabled a month ago or more so rebooting the router will cause the kernel to set the system time to a dummy time until the NTP resets the system time to a real value. So what you see now is the new norm since those changes to the RTC.

at least /dev/rtc0 is still available, but /sbin/hwclock has been removed.
So I guess - yes.

@tcsoft

Thanks for the advice. That seems like a good idea until things settle down especially if the older builds aren't available. I am looking for an archive now of the older releases.

kirkgbr wrote:

In LUCI?  I have nothing called "Services"

Check "Startup" under "System".