CHAOS CALMER (Bleeding Edge, r43974) 3.18.2

Check CPU Frequency
cat /sys/devices/system/cpu/cpu0/cpufreq/cpuinfo_cur_freq
cat /sys/devices/system/cpu/cpu1/cpufreq/cpuinfo_cur_freq

Stress Test
stress --cpu 1 --timeout 60
stress --cpu 2 --timeout 60

Fan status
cat /sys/devices/pwm_fan/hwmon/hwmon0/pwm1

Turn Fan on/off
echo 0 > /sys/devices/pwm_fan/hwmon/hwmon0/pwm1
echo 255 > /sys/devices/pwm_fan/hwmon/hwmon0/pwm1

Temperature
sensors

(Last edited by gufus on 19 Jan 2015, 05:41)

Running Kaloz image. Fan seems to be constantly running at full speed. How can I set it back to automatic?

@qkpham,
Check out the previous post from gufus:
echo 0 > /sys/devices/pwm_fan/hwmon/hwmon0/pwm1

@DavidMcWRT:

thanks for the info. i was quite confused, that i could setup 5ghz on both radios. now i know, why it didn't work

i read that the tx and rx values are not accurate, but my wifi was very slow anyway. i will try the fix for iphones today.

dd-wrt on my r7000 is plagued with the same problem. i only use 5ghz wifi on this unit, because with 2.4ghz, iphones will cause random reboots. also i am running 10.10.1 on my mac, where there is a lot of wifi trouble... don't know what's their problem, but apple really sucks at wifi at the moment

EDIT:
you were right, 6mbits is not correct. i tried your fixes. for testing i only used radio 1 with 5ghz. with my mac i was able to get some decent wifi speeds on ac or n, but not great. with the iphone it remained terribly slow. is there anything i can provide to the devs? some logs maybe?

(Last edited by lechatnoir on 18 Jan 2015, 12:30)

Thanks for your feedback lechatnoir.

Out of interest, how is the 2.4 GHz performance? Obviously it's not AC and you'd be on the more congested 2.4GHz band and likely be restricted to 20MHz channels too, but it might be worth comparing 2.4GHz to 5GHz (N) performance.

We're collecting wireless reports here:
https://github.com/kaloz/mwlwifi/issues/

If you'd like to leave a report please do.  Make sure you state you've applied the disassoc_low_ack=0 fix, and also give your client hardware, OS versions and wireless chips and drivers (if you can find that out).

It's interesting to know that dd-wrt on the R7000 also has issues.  The TP-Link Archer C7 V2 also has problems with 5GHz on OpenWRT.

If slow wifi really is a problem for you, you could try the older OpenWRT AA-based McWRT image:
https://github.com/Chadster766/McWRT/re … 2-128k.img

Note this is much older code (and a different wireless driver).

For safety you might like to flash back to Linksys stock firmware first
(download from: http://support.linksys.com/en-us/suppor … /WRT1900AC ) and then flash to McWRT.

Do not use "keep settings" (make sure it's UNSELECTED) or try to import any saved settings between different images, as this seems to be one sure-fire way of bricking the router.

EDIT: NB, your router IP will be 192.168.200.1 with McWRT.

(Last edited by DavidMcWRT on 18 Jan 2015, 16:32)

@lifehacksback @Kaloz:

Linux 3.18.3 now out.

I've gone through the Changelog and didn't spot anything relevant to the WRT1900AC, but there were some iwlwifi
and ath9k_hw fixes that might be applicable to the wider OpenWRT community.

@DavidMcWRT:
Thanks again for your advice.

The performance on the 2.4Ghz band was as bad as on the 5Ghz on the iPhone. On my mac both were alright. I just did some simple online speedtest, which is not a great idea on sundays. So I'll probably repeat them and test some wlan to lan speeds. I guess I'll install kaloz' latest mid next week and try again.

On the r7000 it's the other way round. The broadcom driver is unstable for the 2.4Ghz band. As there is no other 5Ghz network around, 5Ghz is absolutely fine for me.
With the r7000 as my current main unit I am in no hurry, I just want to help iron out some bugs.

Kaloz's latest will likely have the same issue as it's the same driver.  It's Chadster's (older) McWRT image that has the older driver code that appears to be faster/more stable.

So I've finally moved from lifehacksback's 3.18.1 stable ("known good") build to try out some of the later releases.

I went to lifehacksback's testing build first, checked out the fan/temperature stats, then tried Kaloz's latest build

Confirmed that Kaloz's does have the fan on full-whack initially.  Turned it off then repeated temperature investigations.  Also used stress to load the system (and confirmed via cpu readings that cpu speed does go up), saw the temps rise, turned the fan on manually, watched them fall, turned the fan off manually.

@Kaloz: did bsoft's auto-fan control patch not get merged in?
https://forum.openwrt.org/viewtopic.php … 32#p260232
https://forum.openwrt.org/viewtopic.php … 43#p260243

Output below.

lifehacksback's 3.18.2 testing build (2015/01/14)

CHAOS CALMER (Bleeding Edge, r43961)

# cat /sys/devices/system/cpu/cpu0/cpufreq/cpuinfo_cur_freq
600000
# cat /sys/devices/system/cpu/cpu1/cpufreq/cpuinfo_cur_freq
600000

# cat /sys/devices/gpio_fan/hwmon/hwmon0/pwm1
0

# sensors
gpio_fan-isa-0000
Adapter: ISA adapter
fan1:           0 RPM  (min =    0 RPM, max = 4500 RPM)

tmp421-i2c-0-4c
Adapter: mv64xxx_i2c adapter
temp1:        +38.6 C
temp2:        +41.8 C

armada_thermal-virtual-0
Adapter: Virtual device
temp1:        +48.7 C

Kaloz's latest 3.18.2 snapshot (2015/01/15)

CHAOS CALMER (Bleeding Edge, r43974)

# cat /sys/devices/system/cpu/cpu0/cpufreq/cpuinfo_cur_freq
600000
# cat /sys/devices/system/cpu/cpu1/cpufreq/cpuinfo_cur_freq
600000

# cat /sys/devices/pwm_fan/hwmon/hwmon0/pwm1
255

# echo 0 > /sys/devices/pwm_fan/hwmon/hwmon0/pwm1

# sensors
tmp421-i2c-0-4c
Adapter: mv64xxx_i2c adapter
ddr:          +33.7 C
wifi:         +38.1 C

armada_thermal-virtual-0
Adapter: Virtual device
cpu:          +41.4 C

# stress --cpu 1 --timeout 60

# sensors
tmp421-i2c-0-4c         
Adapter: mv64xxx_i2c adapter
ddr:          +35.1 C       
wifi:         +38.0 C       

armada_thermal-virtual-0
Adapter: Virtual device
cpu:          +53.0 C   

# cat /sys/devices/system/cpu/cpu0/cpufreq/cpuinfo_cur_freq
1200000

# echo 255 > /sys/devices/pwm_fan/hwmon/hwmon0/pwm1

# cat /sys/devices/system/cpu/cpu0/cpufreq/cpuinfo_cur_freq
600000

# sensors
tmp421-i2c-0-4c
Adapter: mv64xxx_i2c adapter
ddr:          +34.6 C
wifi:         +38.0 C

armada_thermal-virtual-0
Adapter: Virtual device
cpu:          +41.4 C

# echo 0 > /sys/devices/pwm_fan/hwmon/hwmon0/pwm1

# cat /sys/devices/pwm_fan/hwmon/hwmon0/pwm1
0

(Last edited by DavidMcWRT on 18 Jan 2015, 18:08)

Does this work?

FYI
https://forum.openwrt.org/viewtopic.php … 82#p261882

yes i know. but that's not the driver that will end up in chaos calmer, right? i will try newer builds when they arrive and report back.

thanks

I'm having major issues trying to get the router to work using Kalos' image.

The first issue I'm having is that I cannot get the wireless interfaces to work with 802.11ac, and the 5GHz radio doesn't seem to want to work at all. The 2.4GHz radio only works if I configure it to use 802.11n mode. Any suggestions on how I can fix this, or any information that I can provide to help troubleshoot?

The second issue is that even with the radio working, I get very poor speeds via both wired and wireless interfaces. I ran speedtests (I have a gigabit WLAN connection) across multiple configurations:

Config                                          D/L Mbps    U/L Mbps
Direct (no router)                          505.58    926.84
Stock Wired                                  573.61    924.73
Kalos Wired                                  490.75    121.89
Stock Wireless (802.11ac, 5GHz)       367.75    173.69
Kalos Wirelsss (802.11n, 2.4GHz)      27.71            27.01

I haven't seen anything in the thread to help resolve this since the new driver release, but apologies if I missed anything (I'll go back and look).

255 was the default value on stock Kaloz, fan on full, so without thinking I used that (and 0) to turn on and off.

Now you've made me question it, I just tried 128 and fan is on but slower (by ear) than full-whack.

So it appears this is now a range, 0=off, 255=full speed; in-between varies it.

Thanks for querying!

Correct.

FYI I've been running Kaloz's latest snapshot for the last couple of hours.  Have (manually!) tweaked it back to how I had lifehacksback's stable version, and all appears well.

Make sure you manually set a channel for AC; using auto doesn't work.

Thanks for providing detailed testing speeds.

There are some issues with the wireless driver, although your speeds do seem very low.

Your wired results are new information however.  I'm not too concerned about the D/L difference between stock and Kaloz, but the upload difference (only ~122Mb compared to ~925Mb stock) is a bit more of a concern.

Most people won't have a Gb upload connection so you might therefore be the first to come across this.

(Last edited by DavidMcWRT on 18 Jan 2015, 20:30)

I thought I had picked a channel, but no worries - I'll try again.

Let me know if I can provide any information, captures or logs to help troubleshoot. It seems to be quite painless to switch between different firmwares, so it's easy enough for me to flash back and try something.

Also, any idea why both firmwares manage to keep their settings through reflashing? Where is each storing their settings and why do they persist through firmware wipes?

Yes, please double-check.  (And make sure radio0 is setup for 2.4GHz and radio1 for 5GHz.)

Hopefully others can provide you with directions on this.

I'd urge a bit of caution when switching between firmwares, as some scenarios have left people with "bricked" routers that can only be revived by buying a cable, opening up the router, and performing command-line actions to restore it.

Looks like you've been "lucky" so far (thankfully) - and I don't want to scare you off! - but for future flashing:

1) always flash back to stock before going from one OpenWRT release to another
2) DON'T "keep settings" OR "load previous settings"

There are two separate firmware areas so it may well be each are keeping their settings in each area.  So if you've been consistently going from stock to OpenWRT and back, each will be ending up back in its own parition and thereby possibly finding its old configuration.

(Last edited by DavidMcWRT on 18 Jan 2015, 20:50)

Are you sure?

I can't test right now, I'm not in my office. I'm working remotely right now.

Yes!

Just double checked (this is Kaloz 15 Jan build):

# echo 255 > /sys/devices/pwm_fan/hwmon/hwmon0/pwm1
# echo 192 > /sys/devices/pwm_fan/hwmon/hwmon0/pwm1
# echo 128 > /sys/devices/pwm_fan/hwmon/hwmon0/pwm1
# echo 80 > /sys/devices/pwm_fan/hwmon/hwmon0/pwm1
# echo 0 > /sys/devices/pwm_fan/hwmon/hwmon0/pwm1

255, 192 and 128 all sounded different.

Couldn't hear anything at 80.

(Last edited by DavidMcWRT on 18 Jan 2015, 21:26)

@kaloz

FYI
https://forum.openwrt.org/viewtopic.php … 76#p261976

Just a request but could you guys activate the wifi by default on your firmwares (kaloz & lifehacksback) ?  Thank you

Just had a look at his patch.

+fan_set() {
+    if $1; then
+        pwm=255
+    else
+        pwm=0

on=255
off=0

I've only switching between stock and OpenWRT, and I have a TTL cable ready (somewhere)

Is there any documentation yet on how the partitions are laid out?

@kop48

cat /proc/mtd

nitroshift

New Testing image 01/19/2015

Wifi enabled by default for the person asking.

Kernel 3.18.2, I'll see if it was updated sometime after the build.