OpenWrt Forum Archive

Topic: Update on Linksys WRT1900AC support

The content of this topic has been archived between 16 Sep 2014 and 7 May 2018. Unfortunately there are posts – most likely complete pages – missing.

@sera, Same build but backed off one release of mwlwifi(10.3.0.18-20160804), wifi only, have not experienced a reboot, with uptime of:

 11:51:57 up 1 day,  2:19,  load average: 0.09, 0.07, 0.05

The newest firmware/blob occurred in the mwlwifi release included in this image. The current mwlwifi (10.3.0.18-20160823-1) (the one supposedly failing on 4.7.x & 4.8-rcx) release only has code changes, as can be seen on the github. Now for a strong cuppa, I had Russell's teapot around here somewhere.

@Villeneuve

So mwlwifi is almost certainly the cause. Time for git-bisect
  bad: HEAD
  good (last release): d48847cc83954930fdcd98818bbf33e75707b8e1

$ git clone https://github.com/kaloz/mwlwifi.git
$ git bisect start HEAD d48847cc83954930fdcd98818bbf33e75707b8e1
Bisecting: 3 revisions left to test after this (roughly 2 steps)
[58fe09164f187f5a28e4dceea5628414dd398a4a] Fixed rate adaptation problem of station mode.

Now you know the next revision to test. Mark as appropriate and git will give you the next revision to test depending on the your findings. Then in the end you know which commit broke your device.

This technique is obviously more useful if you have to find a rotten apple in 10 thousands of commits as between kernel releases.

For more info man git-bisect or just ask if you get stuck.

Hows trunk working now-a-days?

I see it uses a good wifi driver, dd-wrt uses the same ver.
kmod-mwlwifi_4.4.14+10.3.0.18-20160804-1_mvebu.ipk

(Last edited by gufus on 11 Sep 2016, 22:15)

Villeneuve wrote:

@sera, Same build but backed off one release of mwlwifi(10.3.0.18-20160804), wifi only, have not experienced a reboot, with uptime of:

 11:51:57 up 1 day,  2:19,  load average: 0.09, 0.07, 0.05

IMHO

It's the best we have (4 now)

root@DD-WRT 5d:e1:f6:~# iw ath1 station dump
Station 00:25:9c:13:34:6b (on ath1)
        inactive time:  3350 ms
        rx bytes:       44542750
        rx packets:     449567
        tx bytes:       69148
        tx packets:     768
        tx retries:     0
        tx failed:      0
        beacon loss:    0
        beacon rx:      220524
        rx drop misc:   4
        signal:         -50 dBm
        signal avg:     -49 dBm
        beacon signal avg:      207 dBm
        tx bitrate:     1300.0 MBit/s VHT-MCS 9 80MHz short GI VHT-NSS 3
        rx bitrate:     1053.0 MBit/s VHT-MCS 8 80MHz VHT-NSS 3
        authorized:     yes
        authenticated:  yes
        associated:     yes
        preamble:       long
        WMM/WME:        yes
        MFP:            no
        TDLS peer:      no
        DTIM period:    2
        beacon interval:100
        short slot time:yes
        connected time: 22595 seconds
root@DD-WRT 5d:e1:f6:~#

(Last edited by gufus on 11 Sep 2016, 22:44)

gufus wrote:

How's trunk working now-a-days?

I see it uses a good wifi driver, dd-wrt uses the same ver.
kmod-mwlwifi_4.4.14+10.3.0.18-20160804-1_mvebu.ipk

Quite stable, with certain variances across kernel versions (if building own image). 

  • While I utilize both OpenWrt and DD-WRT [Netgear R6300 v1, while V2 has OpenWrt support, v1 does not], I prefer OpenWrt due to the ability for customization, a more robust community, and in depth wikis.

I've been running trunk across varying 4.4.x kernel versions for almost a year now, and 4.4.14 for over a month with only a few minor issues here and there

System Host name OpenWrt
Model Linksys WRT1900AC
Firmware Version LEDE Reboot r1535 / LuCI Master (git-16.251.61906-44bf3f0)
Kernel Version 4.4.20
Local Time Sun Sep 11 21:58:19 2016 Uptime 3d 2h 44m 58s
Trunk has been stable I usually only flash a new image when I see something that pertains to mvebu.
I have seen decent uptime's for quite awhile now.

gufus wrote:

Hows trunk working now-a-days?

I see it uses a good wifi driver, dd-wrt uses the same ver.
kmod-mwlwifi_4.4.14+10.3.0.18-20160804-1_mvebu.ipk

the "dark side" uses 4.4.19 and wifi driver version to 10.3.0.18-20160823-1, and i have no issue with this version...


Villeneuve wrote:

There was no console output to capture.


hi, is there a way to capture kernel crash logs to usb? what to do in this scenario? surround the enemy lowering the wifi version commit by commit?

sera wrote:

    .......making packages musl compatible. So there is something to actually test wink

@sera i wonder, why do you need to make what packages musl compatible? didn't you just change kernel version on openwrt trunk?  please explaine.



@all
is there is some benefit to upgrafe gcc from 4.5 to 6.2 and then compile latest trunk? i suppose that the squashfs will be the same, i mean, no big performance on our devices, or maybe whole compiling will be faster on laptop?

thnx.

(Last edited by gsustek on 12 Sep 2016, 18:52)

@gufus, Have found both of the latest two mwlwifi releases to be very good. Most of the time I am running with the latest (10.3.0.18-20160823-1) on a lede build. The post you quoted was related to some of us playing with getting a owrt build running with kernels 4.7.x and 4.8-rcx, based on @sera patch-set. Currently experiencing a kernel panic reboot, at least on a mamba device.

@gsustek, Have not found any guaranteed method of forcing log output. Ya, as @sera posted, doing a git-bisect would be the way to narrow this down, I have not got around to doing anymore owrt builds yet. But there are not that many changes between working, and not working.

gsustek wrote:
sera wrote:

    .......making packages musl compatible. So there is something to actually test wink

@sera i wonder, why do you need to make what packages musl compatible? didn't you just change kernel version on openwrt trunk?  please explaine.

Remove patches 270, 271, 272 in generic. Packages that still work are musl compatible (vanilla or local patches). Anyway, those uapi patches are now gone as I "fixed" all packages that needed fixing.

4.7, 4.8 support is simply cheap once you fix some conceptional brokenness. Read the git log of my series for more info.

So 4.8-rc6 was released and I have some other goodies pending, this time with an updated series for packages.

swrt-2016-09-12
---------------

base

* linux-4.8: bump to 4.8-rc6
* etables: musl compat, drop dependency on linux-headers-musl-hacks
* conntrack-tools: musl compat, drop dependency on linux-headers-musl-hacks
* netifd: musl compat, drop dependency on linux-headers-musl-hacks
* firewall3: musl compat, drop dependency on linux-headers-musl-hacks
* linux-headers-musl-hacks: drop package
* xfsprogs: bump to 4.8-rc1, better upstream musl compatibility
* ppp-mod-pppoe: fix for 4.8-rc5
* ppp-mod-pppol2tp: fix for 4.8-rc5
* ppp-mod-pptp: fix for 4.8-rc5
* update linux-next config
* include/*: kernel version handling bug fixes for linux-next
* xtables-addons mark as broken on linux-next
* The change for PRINTK_TIME originally in 4.7 was now dropped from next.
  Restore old handling

packages

* squid: version bump, musl compat, drop dependency on linux-headers-musl-hacks
* exfat-nofuse: mark as broken with next, already broken in 4.8


Downloads
---------

swrt-2016-09-12.tar.xz: https://gpldr.in/v/sr4fXRRtmK/rD5pwIGZ948dYZec
sha256sum: 5c6f11e8f0e2d545b26e10fea85a59b96f281d3a224391b99fb185d11a69a869

swrt-packages-2016-09-12.tar.xz: https://gpldr.in/v/VuLXhyTWPR/fVy6rTlWm1XNw4gB
sha256sum: 2e762aeb093698133f0ad1b73156562172233e16aeea520be6416f8235a99639

@gsustek @Villeneuve  Couldn't a syslog server be configured to save the logs to a USB drive or network share? 

  • I've never configured one before, and just started looking into setting up a syslog & snmp server in a jail on my FreeNAS server for all network devices

Running a syslog server now, but that is no guarantee of getting logs during a kernel panic.

JW0914 wrote:

@gsustek @Villeneuve  Couldn't a syslog server be configured to save the logs to a USB drive or network share? 

  • I've never configured one before, and just started looking into setting up a syslog & snmp server in a jail on my FreeNAS server for all network devices

A couple of month back I tried to get OpenWRT syslogging to Nas4Free/FreeNas as a syslog server, but couldn't get it going for some reason, gave up and redirect syslog to a USB device; so if you get it working with FreeNas, details instructions would be appreciated.

JW0914 wrote:

@gsustek @Villeneuve  Couldn't a syslog server be configured to save the logs to a USB drive or network share? 

  • I've never configured one before, and just started looking into setting up a syslog & snmp server in a jail on my FreeNAS server for all network devices

Putty log from serial terminal:-)

gsustek wrote:
JW0914 wrote:

@gsustek @Villeneuve  Couldn't a syslog server be configured to save the logs to a USB drive or network share? 

  • I've never configured one before, and just started looking into setting up a syslog & snmp server in a jail on my FreeNAS server for all network devices

Putty log from serial terminal:-)

I thought you were referring to kernel panic after boot (I should have asked) =]

(Last edited by JW0914 on 13 Sep 2016, 19:05)

I am new at this and my English is not very good .
I tried to find an answer to my question and I have not found .
I need to handle that davice connect to my router. Able to name and know is this connected and power block as is done with the original firmware.
I have a linksys 1900 with luci .
Thansk...!!!!

@MatiasG I believe you're asking how to access a device connected to the router?  As well as how to set it's hostname (possibly a static IP?)?

  • If those are correct, you would access a device connected to the router via it's IP address or Hostname (if within the LAN), however, if you'd like to access it from WAN [the internet], you'll need a Dynamic DNS [DDNS] address from a provider such as DynDns (there's others, they're just one)

    • You can set the IP address or hostname by logging into LuCI [router Web UI], then navigating to Network > DHCP & DNS

    • To configure DDNS, you'd first get the service from a provider, configure a hostname for your router's WAN IP through the provider, then log in to LuCI and navigate to Services -> Dynamic DNS

However, I'm not sure what you mean by "power block"


@davidc502 I saw on the wiki page you listed your builds as having both SQM qos, as well as QOS... I thought one is only supposed to have one or the other installed?

(Last edited by JW0914 on 14 Sep 2016, 14:13)

JW0914 wrote:

@davidc502 I saw on the wiki page you listed your builds as having both SQM qos, as well as QOS... I thought one is only supposed to have one or the other installed?

@JW0914

I think you can have both installed, but only 1 can be enabled.

adri wrote:
JW0914 wrote:

@davidc502 I saw on the wiki page you listed your builds as having both SQM qos, as well as QOS... I thought one is only supposed to have one or the other installed?

@JW0914
I think you can have both installed, but only 1 can be enabled.

Yep. If both are installed to the compiled firmware image, you should have at least one package's "enabled" setting toggle set to 0 in the shipped default config of the package. (I think that both actually have the enabled 0 as default.)

(Last edited by hnyman on 14 Sep 2016, 16:33)

@sera,Feedback on your last patch-set. I did a 4.8-rc6 build on a new tree and flashed UBI image to mamba device; no change to build, I simply copied across my .config. My intent was to let that run on the off chance I might catch some useful output on the console when it failed. In the meantime I planned on creating a modified build to start looking for the mwlwifi commit causing the grief on previous builds. More than 24 hours later the build is still running and I have not experienced a reboot. Did a quick browse @ kernel.org for any commits as likely candidate for change, but nothing leaps out.

JW0914 wrote:

@davidc502 I saw on the wiki page you listed your builds as having both SQM qos, as well as QOS... I thought one is only supposed to have one or the other installed?

Thank you -- Checking into it.

So far 2 builds with both packages witch no issues, but that doesn't mean it isn't a problem wink

Best Regards,

JW0914 wrote:

@MatiasG I believe you're asking how to access a device connected to the router?  As well as how to set it's hostname (possibly a static IP?)?

My friend... thanks for your answer.... i need change the setting...
I need to change the setting as changes in the original firmware. If you can see the link.

"linksys.com/us/support-article?articleNum=138954#two"

Thanks

(Last edited by MatiasG on 14 Sep 2016, 23:48)

MatiasG wrote:
JW0914 wrote:

@MatiasG I believe you're asking how to access a device connected to the router?  As well as how to set it's hostname (possibly a static IP?)?

My friend... thanks for your answer.... i need change the setting...
I need to change the setting as changes in the original firmware. If you can see the link: Network Map

OpenWrt doesn't have a feature like that [at least not that I'm aware of]. 

The closest thing would either be the Overview > Status page that you're brought to upon logging into LuCI

  • DHCP Leases
    DHCPv6 Leases
    Associated Stations

the Network > DHCP and DNS page

  • Active DHCP Leases
    Active DHCPv6 Leases

and Network > Wireless page

  • Associated Stations

(Last edited by JW0914 on 15 Sep 2016, 01:27)

Just a note that I personally have no more major issues with the WRT1900AC(S) family -- the last issue I had (bi-directional speed on 5G) seems to have been mostly addressed with recent commits in LEDE source.

iperf3 -c <target ip> -P6 -t20 yields (ACS to AC, both on LEDE):

[SUM]   0.00-20.00  sec   934 MBytes   392 Mbits/sec  826             sender
[SUM]   0.00-20.00  sec   930 MBytes   390 Mbits/sec                  receiver

iperf3 -c <target ip> -R -P6 -t20 yields:

[SUM]   0.00-20.00  sec   287 MBytes   121 Mbits/sec   23             sender
[SUM]   0.00-20.00  sec   287 MBytes   120 Mbits/sec                  receiver

The result here is the same if I run the server on the ACS and don't use -R -- that direction is always slower but I may see rates around 200 Mbits/sec.

File transfer from AC to ACS:  ~25MB/s
File transfer from ACS to AC:  28-32 MB/s, averaging ~30MB/s

uptime
 20:39:11 up 2 days,  1:55,  0 users,  load average: 0.13, 0.08, 0.03

For some reason there's still an imbalance in iperf but significantly less imbalance with file transfers and the typical slowdown after a day or so has not (yet) occurred.

I remember lockups that happened after 4-5 days in the past and those are behind us -- hopefully the slowdowns are as well.  Unless a major reason to rebuild comes up I'm going to once again attempt the long-term stability test.

Maybe now is the time to complain that with stock firmware in the AC as a bridge I'd get better than 40 MB/sec transfer rates... gotta keep the dream alive.

JW0914 wrote:
MatiasG wrote:
JW0914 wrote:

@MatiasG I believe you're asking how to access a device connected to the router?  As well as how to set it's hostname (possibly a static IP?)?

My friend... thanks for your answer.... i need change the setting...
I need to change the setting as changes in the original firmware. If you can see the link: Network Map

OpenWrt doesn't have a feature like that [at least not that I'm aware of]. 

The closest thing would either be the Overview > Status page that you're brought to upon logging into LuCI

  • DHCP Leases
    DHCPv6 Leases
    Associated Stations

the Network > DHCP and DNS page

  • Active DHCP Leases
    Active DHCPv6 Leases

and Network > Wireless page

  • Associated Stations


Ok, My friend, thanks for your answer and help!!!!

You know if it is possible to disconnect a wireless connection? That is, if a phone connect to my wifi network is possible disconnect from the router? Thanks again!!!!