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The content of this topic has been archived between 26 Feb 2018 and 7 May 2018.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.

cybrnook2002 wrote:

Interesting, so you're suggesting to actually disable the HW acceleration and test without it? I was curious about that myself. If that is the case, I will make a new .config from scratch to eliminate any inter-dependencies that have crept in to try a clean build without HW accel. Yes, my goal is purely speed, not offloading work.

WRT1900ACS without CESA, aes-256-cbc

type              16 bytes     64 bytes    256 bytes   1024 bytes   8192 bytes
aes-256 cbc      32470.09k    34938.86k    35737.09k    35967.66k    36088.64k

WRT1900ACS with CESA, aes-256-cbc, command line 'openssl speed -elapsed -evp aes-256-cbc'

type              16 bytes     64 bytes    256 bytes   1024 bytes   8192 bytes
aes-256-cbc        896.45k     3902.31k    14125.06k    41816.41k    79828.31k

Summary:  software significantly faster than hardware unless encoding large chunks of data at a time.  Speed is VERY use case dependent.

drfranco wrote:

Far as upgrading to r2507 goes, is it okay to use the upgrade and keep the config in place?  Or do you recommend clearing and then re-installing the config from a tarball backup?

IMO, the best way to upgrade so that you always have the stock Linksys firmware on one partition for emergencies is to do the following in SSH:

First make a tar backup of your current settings from Luci.

Then, determine what partition you are currently on by:

fw_printenv boot_part

It will say 1 or 2. To boot to the other partition (which if you haven't upgraded from inside LEDE is still stock Linksys) issue the command:

fw_setenv boot_part X

Where X is the number 1 or 2; the opposite of what the first command gave you.

Then simply:

reboot

To reboot to the stock Linksys partition, then flash @david's latest build of LEDE, and lastly restore your backup.

First, I would like to thank you for your work! I'm on 2607 now which runs flawlessly on my ACS.
But there are two packages which I cannot find in your repository: ddns-script_cloudflare and ddns-script_cloudflare.com-v4. Could you integrate those, too?

Thanks and keep up the good work!

denk wrote:

First, I would like to thank you for your work! I'm on 2607 now which runs flawlessly on my ACS.
But there are two packages which I cannot find in your repository: ddns-script_cloudflare and ddns-script_cloudflare.com-v4. Could you integrate those, too?

Thanks and keep up the good work!

cloudflare was failing sad  that's why it wasn't in the build.  It will probably be fine on the next one.

*EDIT* checked again, and still failing.  Curl is being compiled, yet it says it can not satisfy dependencies.

Collected errors:
* satisfy_dependencies_for: Cannot satisfy the following dependencies for ddns-scripts_cloudflare.com-v4:
*     curl *
* opkg_install_cmd: Cannot install package ddns-scripts_cloudflare.com-v4.
package/Makefile:60: recipe for target 'package/install' failed


Currently the LEDE snapshot repository has cloudflare, and with the same kernel version would probably work --  https://downloads.lede-project.org/snapshots/packages/

(Last edited by davidc502 on 22 Dec 2016, 15:23)

Thanks for the new build!

Is anyone else out there running strongswan as VPN server? On my WRT1900ACv2 it seems like clients fail to connect to VPN (strongswan IKEv2) if kmod-cryptodev is loaded. When I forced removed it then the clients could connect. (Actually the same problem happens with r2317/2321, but it works fine back in r2221 because the cryptodev engine was not properly loaded.)

However for ocserv it works either with or without cryptodev.

installed r2607, updated from r2221 without issues. Saved all the settings so it was pretty seamless upgrade. WiFi radios both up, and can login, uptime 13 hrs, 20 min so far. System log contained the following: 

kern.err kernel: [47672.364007] mvneta f1074000.ethernet eth1: bad rx status 0f812000 (crc error), size=1520

daemon.err [2681]: iptables v1.4.21: can't initialize iptables table `raw': Table does not exist (do you need to insmod?)

daemon.err [2681]: Perhaps iptables or your kernel needs to be upgraded.


Cheers

davidc502 wrote:

Just uploaded a new build with the latest Wifi Drivers and small Kernel bump.  I have yet to copy the images into their respective directories because I have not tested them. However, anyone who wants to get a jump can find the images here >>  http://davidc502sis.dynamic-dns.net/mvebu/

r2607

@david
Just when you thought things might settle down with the driver I see there is a new wireless driver available,10.3.2.0-20161222. wink
https://github.com/kaloz/mwlwifi?files=1
Your r2607 build as been running without issue for me on my wrt1900acs.. Did streaming last night both Netflix and from NAS. The r2664 build had pausing issues on 5GHz wireless when streaming from the NAS for some reason. Been up for 16 hours.
--bill

(Last edited by bill1228 on 22 Dec 2016, 18:11)

hancor wrote:

installed r2607, updated from r2221 without issues. Saved all the settings so it was pretty seamless upgrade. WiFi radios both up, and can login, uptime 13 hrs, 20 min so far. System log contained the following: 

kern.err kernel: [47672.364007] mvneta f1074000.ethernet eth1: bad rx status 0f812000 (crc error), size=1520

daemon.err [2681]: iptables v1.4.21: can't initialize iptables table `raw': Table does not exist (do you need to insmod?)

daemon.err [2681]: Perhaps iptables or your kernel needs to be upgraded.


Cheers

As for the first error, never seen it.  The second two I see on every boot.

@david, what do you think? 

Also, as has been reported, wifi driver 1222 is out, but only has 1 minor commit.

starcms wrote:
hancor wrote:

daemon.err [2681]: iptables v1.4.21: can't initialize iptables table `raw': Table does not exist (do you need to insmod?)

daemon.err [2681]: Perhaps iptables or your kernel needs to be upgraded.

... The second two I see on every boot.

That iptables error was due to LuCI iptables parser (that was called during the boot process by LuCI statistics config generation script) and has been fixed yesterday with https://github.com/openwrt/luci/commit/ … 21a2c20fca (roughly at LEDE r2610)

(Last edited by hnyman on 22 Dec 2016, 19:01)

Thanks for the info, looking forward to the r26xx updates!  smile

Merry Christmas / Frohe Weihnachten

Cheers

davidc502 wrote:

Okay.... The image seems to be performing well, so they have been placed in their normal spots.

Appreciate the feedback thus far.

@david
Might want to update the site with the new build info. Still has old stuff and dated 12/10/2016. actual download files appear to be correct by date.

Feedback from me is it is working great on my wrt1900acs v1. No 5GHz wireless issues so far.

Thanks again for your efforts. wink

--bill

(Last edited by bill1228 on 22 Dec 2016, 23:58)

Flashed the new build to my 1900ac v1, working fine so far for 4 hours.

Thanks for your hard work, David and have a Merry Christmas and a Happy New Year!

(Last edited by apvm on 24 Dec 2016, 10:36)

The website has been updated to include some dynamic content, so be sure to update bookmarks. It just shows how many images have been downloaded through the day.

http://davidc502sis.dynamic-dns.net/index.shtml


Merry Christmas everyone!

@david, what is the difference between cobra (1900ac v2) and Shelby (1900 acs) builds?  They both share the same FCC-ID and hardware. So what's different?

Thanks as always and Merry Christmas to you and your family!

Edit:. Also, do you know if all versions (including 1200ac, caimen) have 2 cesa cores except for 1900ac v1 (mamba)?

(Last edited by starcms on 24 Dec 2016, 22:44)

I created a port forward and enabled NAT loopback, however, I'm not able to access the machine from inside the network. Outside the network it appears to work fine.

I have a 3200acm running the build from 12/21. Any thoughts?

(Last edited by cjohnston on 25 Dec 2016, 03:15)

starcms wrote:

@david, what is the difference between cobra (1900ac v2) and Shelby (1900 acs) builds?  They both share the same FCC-ID and hardware. So what's different?

Thanks as always and Merry Christmas to you and your family!

Edit:. Also, do you know if all versions (including 1200ac, caimen) have 2 cesa cores except for 1900ac v1 (mamba)?

If you are not aware of this site, here is a link to a Openwrt wiki that has about everything you ever wanted to know but were afraid to ask.

https://wiki.openwrt.org/toh/linksys/wr … wrt1900acs

It does appear that the hardware is the same for the two. I remember when the wrt1900a v2 came out there was a lot of confusion because the stated cpu speed was slower then the wrt1900acs, howerver folks were buying them and finding that the cpu speed was the same as the wrt1900acs. I don't think I have ever seen a real discussion about the differences if there are any.

--bill

Merry Christmas.

A question: Do I still use "sysupgrade -F ...." to upgrande .bin file or "-F" switch is not needed anymore ?

ambrosa wrote:

Merry Christmas.

A question: Do I still use "sysupgrade -F ...." to upgrande .bin file or "-F" switch is not needed anymore ?

Unknown, as it has not been reported either way.


Merry Christmas smile

Merry Christmas everyone!

@david, latest build still working fantastic.  Got one wierd entry in the kernel log not too long ago though:

[62127.966723] conntrack: generic helper won't handle protocol 47. Please consider loading the specific helper module.

Any ideas?  Much thanks!

davidc502 wrote:
ambrosa wrote:

Merry Christmas.

A question: Do I still use "sysupgrade -F ...." to upgrande .bin file or "-F" switch is not needed anymore ?

Unknown, as it has not been reported either way.


Merry Christmas smile

From r2446 ro r2607

Just made "sysupgrade /tmp/...."  without "-F" : it works.
So I can confirm that "-F" is not needed anymore.

ambrosa wrote:
davidc502 wrote:

Testing r2446

DDNS is working again. -- Wasn't working after r2221

I'm testing r2446 now.
DDNS NOT working. With r2371 DDNS worked very fine.

If I check log I see:

182415       : #> /bin/uclient-fetch -q -O /var/run/ddns/***_ORG.dat -Y off 'http://checkip.dyndns.com' 2>/var/run/ddns/***_ORG.err
 182431       : Local IP '31.****' detected on web at 'http://checkip.dyndns.com'
 182431       : Forced Update - L: '31.****' == R: '31.****'
 182431       : #> /bin/uclient-fetch -q -O /var/run/ddns/***_ORG.dat -Y off 'http://***:******@nic.changeip.com/nic/update?u=******&p=***&cmd=update&hostname=***.org&ip=31.*** 2>/var/run/ddns/***_ORG.err
 182433       : DDNS Provider answered:
200 Successful Update
 182433 ERROR : IP update not accepted by DDNS Provider

After a correct answer "200 Successful Update", DDNS script don't recognize it correctly and report an ERROR

Some posts ago I notify problem with r2446 DDNS (ChangeIP and NoIP).
With r2607 the Lede DDNS subsystem works fine without any problem.
Thanks.

(Last edited by ambrosa on 26 Dec 2016, 09:14)

starcms wrote:

Merry Christmas everyone!

@david, latest build still working fantastic.  Got one wierd entry in the kernel log not too long ago though:

[62127.966723] conntrack: generic helper won't handle protocol 47. Please consider loading the specific helper module.

Any ideas?  Much thanks!

@davidc502,

I have the same message in syslog.
As a result any PPTP vpn won't connect anymore. sad
PPTP was working in previous releases, but I don't know exactly which release it stopped working.

Adri.

Hope everyone had a great Christmas!

Peering over the lede commits, it appears many of these issues have been addressed, and there's also a updated wifi driver.

It seems to be a good time to put a new build together, so I'll start working on it this morning.

@david
Kong on DD-WRT released a build with the latest drivers, 10.3.2.0-20161222, and I am not seeing anyone reporting wireless issues so should be a good driver to go to. With the fixes as well the one you are getting ready should be a great build. Running your latest without issue.

--bill

@davidc502,

Merry Christmas to you.

I'm looking forward to the new build.
Would it be possible to include kmod-sched-cake in the build, instead of as a package?
SQM with Cake performs really well and I think it makes sense to include this small module as standard.