OpenWrt Forum Archive

Topic: Support for TP-Link Archer C2600

The content of this topic has been archived between 29 Mar 2018 and 6 May 2018. Unfortunately there are posts – most likely complete pages – missing.

Hi all,
  I'm using the Heinz build     LEDE Reboot r1348 / LuCI Master (git-16.229.34804-46755e2) that has LuCI installed and now I am trying to get my HE.NET ipv6 tunnel working.  Seems 6in4 is not installed by default so I tried to install it.  I get this error though.


Collected errors:
* satisfy_dependencies_for: Cannot satisfy the following dependencies for 6in4:
*     kmod-sit *
* opkg_install_cmd: Cannot install package 6in4.

When I try to install kmod-sit though, I can't find it.  Any ideas people?

thanks!

vildkatt wrote:

Hi all,
  I'm using the Heinz build     LEDE Reboot r1348 / LuCI Master (git-16.229.34804-46755e2) that has LuCI installed and now I am trying to get my HE.NET ipv6 tunnel working.  Seems 6in4 is not installed by default so I tried to install it.  I get this error though.


Collected errors:
* satisfy_dependencies_for: Cannot satisfy the following dependencies for 6in4:
*     kmod-sit *
* opkg_install_cmd: Cannot install package 6in4.

When I try to install kmod-sit though, I can't find it.  Any ideas people?

thanks!


Install packages from my ftp not work from luci, but works from ssh.

Connect to router using ssh ( putty etc) and enter command:
opkg update
opkg install 6in4

jmomo wrote:

TeutonJon here had posted before about the ath10k CT version of firmware. To ask again, has anyone tried that, and if so, what were the results? I've seen Ben Greear posting his patches.

I finally got my TRENDnet version of this device booting and the firmware is refusing to load for the qca9980 radios. I'll figure I will try out both, once I figure out what I've messed up.

Heinz had graciously tried them out for us (since he as AFAIK the only working serial connection). They didn't solve any of the problems we had been having at the time. It "seems" like they should be better since they are actually being worked on rather than not.

My Samsung Galaxy Note 5 does not see the wifi signal on radio 0 (the one on AC), but my laptop does. The wifi signal on radio 1 is detected by both devices. Is there any tweaking I can do in LEDE so that the Samsung phone can see radio 0?

It's the OS, Stupid.

Max Hopper wrote:

It's the OS, Stupid.

Please read both this thread and the One you linked to again. You will realize they Cover two different issues. Thanks.

Beste Regards
       M.

P.S.: You Can safely ignore Mr. Hopper in this thread, he has no clue about your issues.

Please not another Max Hopper thread.
OH YEA. Did they finally ban him? I see the "Read-only access" mark on his profile now.

TeutonJon, thanks for your info on the CT firmware. As soon as I get my platform up, I'll be doing some testing with the two.

(Last edited by jmomo on 1 Sep 2016, 01:53)

I cannot find the logrotate and syslog-ng3 packages in LEDE build r1476. Can someone please build these and share?

There is an rsyslog package. Is this a replacement for syslog-ng3?

Thanks.

Khurram wrote:

I cannot find the logrotate and syslog-ng3 packages in LEDE build r1476. Can someone please build these and share?

There is an rsyslog package. Is this a replacement for syslog-ng3?

Thanks.

I don't see these packages in my feed either. They must have been removed.

There was some glibc changes that went into source a few days ago. They might have been removed because of it. In such case, they will get re-added as soon as the bugs get worked out.

jmomo wrote:
Khurram wrote:

I cannot find the logrotate and syslog-ng3 packages in LEDE build r1476. Can someone please build these and share?

There is an rsyslog package. Is this a replacement for syslog-ng3?

Thanks.

I don't see these packages in my feed either. They must have been removed.

There was some glibc changes that went into source a few days ago. They might have been removed because of it. In such case, they will get re-added as soon as the bugs get worked out.

Hi jmomo,

This is the second time you mention possible issues because of "glibc changes", lede uses musl, you explicitly need to select glibc. So if you are doing a default build (one that includes musl) then none of those glibc changes affects any thing.
Just wanted you to not worry smile

james04 wrote:

.... lede uses musl, you explicitly need to select glibc. So if you are doing a default build (one that includes musl) then none of those glibc changes affects any thing.

Any chance of getting logrotate and syslog-ng3 packages?

Could someone please also comment about using rsyslog instead of syslog-ng3?

Heinz wrote:
vildkatt wrote:

Hi all,
  I'm using the Heinz build     LEDE Reboot r1348 / LuCI Master (git-16.229.34804-46755e2) that has LuCI installed and now I am trying to get my HE.NET ipv6 tunnel working.  Seems 6in4 is not installed by default so I tried to install it.  I get this error though.


Collected errors:
* satisfy_dependencies_for: Cannot satisfy the following dependencies for 6in4:
*     kmod-sit *
* opkg_install_cmd: Cannot install package 6in4.

When I try to install kmod-sit though, I can't find it.  Any ideas people?

thanks!


Install packages from my ftp not work from luci, but works from ssh.

Connect to router using ssh ( putty etc) and enter command:
opkg update
opkg install 6in4

That did the job!  I just needed to add a new interface called HENET and set for 6in4 and add all the details.  It took a while to take effect for some reason, but then the ipv6 address showed up for this interface and the LAN automagically.

I do notice that the WIFI interfaces are quite picky when changing the settings.  They don't want to come up again after a change, but a hard reboot does the job.  Fortunately AC works

I am coming from shibby on an R7000, and I like the change.  Lots more to play with.

thanks Heinz!

Khurram wrote:

My Samsung Galaxy Note 5 does not see the wifi signal on radio 0 (the one on AC), but my laptop does. The wifi signal on radio 1 is detected by both devices. Is there any tweaking I can do in LEDE so that the Samsung phone can see radio 0?

Which channel and country code on LEDE radio0 ?

What country of SIM card?

(Last edited by matheos on 1 Sep 2016, 12:21)

Openssl assembly is up in the latest trunk.

Heinz bendavid
Could you please check if ondemand cpu frequency scaling works ok on c2600?

matheos wrote:
Khurram wrote:

My Samsung Galaxy Note 5 does not see the wifi signal on radio 0 (the one on AC), but my laptop does. The wifi signal on radio 1 is detected by both devices. Is there any tweaking I can do in LEDE so that the Samsung phone can see radio 0?

Which channel and country code on LEDE radio0 ?

Channel 36, country code US

matheos wrote:

What country of SIM card?

SIM cards country Pakistan (its a dual sim phone)

in my short testing ondemand governer doesn't scale correctly on second core when router under load..
first core does though.

Its best to keep default as is (performance governer).

Not like these arm cores consume a whole lot of electricity to begin with.

Khurram wrote:
matheos wrote:
Khurram wrote:

My Samsung Galaxy Note 5 does not see the wifi signal on radio 0 (the one on AC), but my laptop does. The wifi signal on radio 1 is detected by both devices. Is there any tweaking I can do in LEDE so that the Samsung phone can see radio 0?

Which channel and country code on LEDE radio0 ?

Channel 36, country code US

matheos wrote:

What country of SIM card?

SIM cards country Pakistan (its a dual sim phone)

Pakistan Reg DB

country PK: DFS-JP
    (2402 - 2482 @ 40), (20)
    (5735 - 5835 @ 80), (30)

Channel 36 - 5180 not allowed in PK

You should use US channels 149-165 on radio0 or ... change country code on your smarphone (need root)

@dissent
Tested ondemand on C2600. Work OK on both cores.

When i run openssl speed sha256 in two terminals and see:

 cat /sys/devices/system/cpu/cpufreq/policy0/stats/time_in_state /sys/devices/system/cpu/cpufreq/policy1/stats/time_in_state

Looks like both cpu use 1400000kHz, when openssl stop working frequency instantly go to 384000kHz

On syslog i have : [ 2430.871000] failed to scale voltage down: -5


ps. openssl use only one cpu in one thread

(Last edited by Heinz on 1 Sep 2016, 13:48)

The freqscaling code was probably ported over from the mobile team, these krait cores are the same ones used in many mobile phones.
In phones every little bit helps in saving battery life.
Thats not very relevant to a router with its power coming from the wall.

If you guys feel the urge to work further on polishing performance, please try to port the huge hardware offload engine available in the ipq806x
https://www.youtube.com/watch?v=gci-pB9Fshk

Its all opensource available in codeaurora

There are advantages to keeping the network stack in open source code rather than proprietary hardware implementation.
(What is open source is the interface code needed to use the hardware.)

As I said before there are already further improvements to wired ethernet performance in the pipe in the mainline stmmac ethernet driver in newer kernels, and probably also from https://git.lede-project.org/?p=lede/nb … 5af414a273 if someone already wants to test that.

Khurram wrote:
james04 wrote:

.... lede uses musl, you explicitly need to select glibc. So if you are doing a default build (one that includes musl) then none of those glibc changes affects any thing.

Any chance of getting logrotate and syslog-ng3 packages?

Could someone please also comment about using rsyslog instead of syslog-ng3?

Never had to use those, but what are you trying to do? Log to file? You can do that without any additional packages..

bendavid wrote:

There are advantages to keeping the network stack in open source code rather than proprietary hardware implementation.
(What is open source is the interface code needed to use the hardware.)

As I said before there are already further improvements to wired ethernet performance in the pipe in the mainline stmmac ethernet driver in newer kernels, and probably also from https://git.lede-project.org/?p=lede/nb … 5af414a273 if someone already wants to test that.

Agreed!

(Last edited by james04 on 1 Sep 2016, 14:39)

matheos wrote:

Pakistan Reg DB

country PK: DFS-JP
    (2402 - 2482 @ 40), (20)
    (5735 - 5835 @ 80), (30)

Channel 36 - 5180 not allowed in PK

You should use US channels 149-165 on radio0 or ... change country code on your smarphone (need root)

I changed the channel to 153 and it worked smile  Thanks

james04 wrote:
Khurram wrote:

Any chance of getting logrotate and syslog-ng3 packages?

Could someone please also comment about using rsyslog instead of syslog-ng3?

Never had to use those, but what are you trying to do? Log to file? You can do that without any additional packages..

Well, I plan to install squid on the router and use logrotate to rotate its access logfile.

Same thing with syslog-ng3 (get the system log to a file) and have logrotate rotate it. If I can log to file (on the flash drive) without any packages, then I dont need syslog-ng3.

(Last edited by Khurram on 1 Sep 2016, 15:28)

Ok, so there is definitely a regression somewhere in wifi at/before r1382 (2016-08-19), which was when we got that new IRQ fix code merged.

I went back to r1261 (2016-08-05) and wifi lost all of the stuttering and slow loading of things from my mobile devices. Oddly, the actual throughput wasn't really bad, it was just the actual functionality wasn't working well. The symptom is when doing anything on a wifi device (all android) that it would start to load something and then just sit and spin for awhile before loading it. Even things like small image files would take numerous seconds, when on the old build would be instantaneous.

(Just as a side note, if you try to reinstall a bunch of packages at the same time, it tends to not go well -- mine was having all sorts of problems till I broke them down into pieces).

Edit: just tested today's build, r1480, and it seemed much better than 1392, but still a little slower than r1261. Hmmm..... Also, thinks are working strange on trunk -- packages keep installing incorrectly. If I install the material theme, half of the time it doesn't work and then I have to reinstall it.

(Last edited by TeutonJon78 on 2 Sep 2016, 03:37)

Hey everyone!

Today I'm looking at the wireless/802.11 MAC addresses with my TEW-827DRU.

Did this ever get fixed properly for the C2600? I think the last informative post on the issue was here, and ianchi mentioned that he wrote a tool, as mentioned here, to modify the calibration data after it was extracted from flash storage.

Is this right? Are there actually MAC addresses in the ART partition? I dumped mine and compared it to what my system is setting and didn't see such strings, but maybe it's encoded, or the driver is setting random junk because that section is all zeros. I have no idea.

My device has the layer 2 addresses in the u-boot env partition, so it's very easy to get. However, the act of setting it seem not be working. The Ath10k driver sets the hard addresses during driver load and then they can't be changed at that level.

Is my interpretation of the situation up-to-date, or did this get figured out some other way?

Thanks in advance.

--

EDIT: I see it now, in the calibration data file.

dd if=/lib/firmware/ath10k/cal-pci-0000:01:00.0.bin bs=2 conv=swab 2> /dev/null | hexdump | head

dd if=/lib/firmware/ath10k/cal-pci-0001:01:00.0.bin bs=2 conv=swab 2> /dev/null | hexdump | head

root@lede:/lib/firmware/ath10k# iw dev
phy#1
        Interface wlan1
                ifindex 6
                wdev 0x100000001
                addr 00:03:7f:12:0b:0b
                type managed
                txpower 0.00 dBm
phy#0
        Interface wlan0
                ifindex 5
                wdev 0x1
                addr 00:34:56:78:81:81
                type managed
                txpower 0.00 dBm
root@lede:/lib/firmware/ath10k# 
bin bs=2 conv=swab 2> /dev/null | hexdump | headare/ath10k/cal-pci-0001:01:00.0. 
0000000 202f 66c2 0101 0003 7f12 3456 0000 0000
0000010 2a60 0000 0060 0000 0a07 0000 0000 0000
root@lede:/lib/firmware/ath10k# 
bin bs=2 conv=swab 2> /dev/null | hexdump | headare/ath10k/cal-pci-0000:01:00.0. 
0000000 202f 2834 0101 1234 5678 9012 0000 0000
0000010 1560 0000 0060 0000 1507 0000 0000 0000

(Last edited by jmomo on 2 Sep 2016, 06:47)