OpenWrt Forum Archive

Topic: openwrt on linksys WRT3200ACM

The content of this topic has been archived between 29 Mar 2018 and 24 Apr 2018. There are no obvious gaps in this topic, but there may still be some posts missing at the end.

Villeneuve wrote:

Understood, I only meant that you could go back to OEM in order to move forward with a gargoyle flash, rather than trying to force a sysupgrade from ssh.

It's all good mate. I flashed back to Gargoyle thanks for your help. I did not know about the -F flag. I really like this router it is very fast I just wish that the 5 ghz was more stable.

@Chadster

What is the correct pin out for serial? I keep getting nothing while connecting my wrt1900acv1 works just fine (bought the plug that fits into the serial port) I'm using

GND  RX   N/A   TX  N/A  N/A
   x       x       x      x      x      x

Ok this is extremely odd,

Everything connected = unit wont boot

Disconnect serial = unit boots

reconnecting serial = serial prompt

me right now = crying in the fetal position

(Last edited by lifehacksback on 3 Feb 2017, 04:09)

@lifehacksback

Looking at the front of the unit (for perspective) that is the pinout I use on the ACM.  It is the same as the ACS.

So I have the Rango unit finally configured and running in the network but kind of isolated so any wifi issues don't disrupt anything.  Here are some early comparisons:

Shelby (WRT1900ACS) openssl benchmark (see https://wiki.openwrt.org/doc/howto/benchmark.openssl)

The 'numbers' are in 1000s of bytes per second processed.
type             16 bytes     64 bytes    256 bytes   1024 bytes   8192 bytes
md5              14288.89k    48218.41k   128161.50k   216256.17k   272110.93k
sha1             14597.11k    47043.48k   117922.33k   188190.72k   228089.86k
des cbc          30081.03k    31334.34k    31566.93k    31679.49k    31738.54k
des ede3         11096.16k    11333.48k    11368.87k    11387.56k    11389.61k
aes-128 cbc      59034.62k    65393.30k    67903.40k    68580.35k    68793.69k
aes-192 cbc      51448.90k    56057.11k    57933.99k    58432.17k    58763.22k
aes-256 cbc      45630.54k    49052.07k    50482.01k    50842.28k    50943.32k
sha256           20028.50k    47077.06k    86431.66k   109312.68k   117926.57k
sha512            8144.43k    32368.41k    47596.80k    65779.71k    74025.64k
                  sign    verify    sign/s verify/s
rsa 2048 bits 0.019704s 0.000618s     50.8   1617.0
                  sign    verify    sign/s verify/s
dsa 2048 bits 0.005880s 0.006456s    170.1    154.9
| | ARMv7 Processor rev 1 (v7l) | 50.00 | ARMv7 Processor rev 1 (v7l) | 50.00 | Marvell Armada 380/385 (Device Tree) | 1.0.2k | 216256170 | 188190720 | 109312680 | 65779710 | 31679490 | 11387560 | 68580350 | 58432170 | 50842280 | 50.8 | 1617.0 170.1 | 154.9 |

These numbers are down a bit from prior ACS measurements, I'm wondering if it's not the addition of stack smashing protection and the like on speed (those are my modifications to the build, not stock).  In any case, Rango is built the same way for this test:

Rango (WRT3200ACM) openssl benchmark

The 'numbers' are in 1000s of bytes per second processed.
type             16 bytes     64 bytes    256 bytes   1024 bytes   8192 bytes
md5              12398.37k    41353.18k   108480.52k   183312.78k   228420.27k
sha1             12705.97k    40465.12k   101400.41k   160964.92k   195555.92k
des cbc          25415.92k    26781.74k    26969.00k    27148.33k    27129.49k
des ede3          9441.41k     9617.05k     9766.36k     9688.75k     9734.19k
aes-128 cbc      46526.70k    54564.24k    57619.48k    58611.16k    58854.06k
aes-192 cbc      41011.60k    46989.49k    49061.73k    49909.55k    50129.58k
aes-256 cbc      36514.18k    41501.74k    43027.90k    43538.09k    43592.95k
sha256           16547.52k    40265.82k    74289.49k    92900.93k   100468.44k
sha512            6821.12k    27728.58k    40888.27k    56390.66k    63105.71k
                  sign    verify    sign/s verify/s
rsa 2048 bits 0.023270s 0.000725s     43.0   1378.4
                  sign    verify    sign/s verify/s
dsa 2048 bits 0.006922s 0.007498s    144.5    133.4
| | ARMv7 Processor rev 1 (v7l) | 50.00 | ARMv7 Processor rev 1 (v7l) | 50.00 | Marvell Armada 380/385 (Device Tree) | 1.0.2k | 183312780 | 160964920 | 92900930 | 56390660 | 27148330 | 9688750 | 58611160 | 49909550 | 43538090 | 43.0 | 1378.4 144.5 | 133.4 |

Shelby to Rango iperf3 test (iperf3 -c <rango_ip> -P6 -t20)

[ ID] Interval           Transfer     Bandwidth       Retr
[  4]   0.00-20.01  sec   353 MBytes   148 Mbits/sec    0             sender
[  4]   0.00-20.01  sec   350 MBytes   147 Mbits/sec                  receiver
[  6]   0.00-20.01  sec   288 MBytes   121 Mbits/sec    2             sender
[  6]   0.00-20.01  sec   286 MBytes   120 Mbits/sec                  receiver
[  8]   0.00-20.01  sec   101 MBytes  42.5 Mbits/sec    2             sender
[  8]   0.00-20.01  sec  98.2 MBytes  41.2 Mbits/sec                  receiver
[ 10]   0.00-20.01  sec   356 MBytes   149 Mbits/sec    2             sender
[ 10]   0.00-20.01  sec   354 MBytes   148 Mbits/sec                  receiver
[ 12]   0.00-20.01  sec   114 MBytes  47.7 Mbits/sec    2             sender
[ 12]   0.00-20.01  sec   111 MBytes  46.4 Mbits/sec                  receiver
[ 14]   0.00-20.01  sec   343 MBytes   144 Mbits/sec    1             sender
[ 14]   0.00-20.01  sec   341 MBytes   143 Mbits/sec                  receiver
[SUM]   0.00-20.01  sec  1.52 GBytes   652 Mbits/sec    9             sender
[SUM]   0.00-20.01  sec  1.50 GBytes   645 Mbits/sec                  receiver

Shelby to Mamba iperf3 tests (similar distances to units)

[ ID] Interval           Transfer     Bandwidth       Retr
[  4]   0.00-20.00  sec   125 MBytes  52.3 Mbits/sec  143             sender
[  4]   0.00-20.00  sec   123 MBytes  51.8 Mbits/sec                  receiver
[  6]   0.00-20.00  sec   123 MBytes  51.5 Mbits/sec   91             sender
[  6]   0.00-20.00  sec   121 MBytes  50.9 Mbits/sec                  receiver
[  8]   0.00-20.00  sec  83.2 MBytes  34.9 Mbits/sec   81             sender
[  8]   0.00-20.00  sec  82.6 MBytes  34.6 Mbits/sec                  receiver
[ 10]   0.00-20.00  sec   128 MBytes  53.8 Mbits/sec  123             sender
[ 10]   0.00-20.00  sec   127 MBytes  53.3 Mbits/sec                  receiver
[ 12]   0.00-20.00  sec   129 MBytes  54.2 Mbits/sec  121             sender
[ 12]   0.00-20.00  sec   128 MBytes  53.7 Mbits/sec                  receiver
[ 14]   0.00-20.00  sec   231 MBytes  97.0 Mbits/sec  151             sender
[ 14]   0.00-20.00  sec   230 MBytes  96.7 Mbits/sec                  receiver
[SUM]   0.00-20.00  sec   819 MBytes   344 Mbits/sec  710             sender
[SUM]   0.00-20.00  sec   813 MBytes   341 Mbits/sec                  receiver

Rango appears to have maybe 15-20% better openssl speed and has excellent wireless speed capability (and the test speaks well for the Shelby unit as well).

Has anybody managed to open the case for this device? I could easy take the blue part away, but the black part refuses to open, and I do not want to force it too much.

eduperez wrote:

Has anybody managed to open the case for this device? I could easy take the blue part away, but the black part refuses to open, and I do not want to force it too much.

You don't need to separate the black cover to access the TTL Serial port on this model.

https://www.snbforums.com/threads/links … ion.35393/

Chadster766 wrote:
eduperez wrote:

Has anybody managed to open the case for this device? I could easy take the blue part away, but the black part refuses to open, and I do not want to force it too much.

You don't need to separate the black cover to access the TTL Serial port on this model.

https://www.snbforums.com/threads/links … ion.35393/

Yes, I know about that, thanks; but I need to uncover the whole board for... other nefarious purposes.

Then you will have to separate it the same as a WRT1900AC V1. Starting at the open edge of one side, pry apart with a pair thin flat head screw drivers at the seam.

For those building their own images: do NOT select kmod-dsa-mv88e6352, kmod-dsa-mv88e-6xxx is enough.

nitroshift

Hello. Can someone confirm that the LEDs on WRT3200AC (Rango) are much brighter than those from WRT1900AC V1 (Mamba) ?

(Last edited by sorinello on 18 Feb 2017, 20:02)

sorinello wrote:

Hello. Can someone confirm that the LEDs on WRT3200AC (Rango) are much brighter than those from WRT1900AC V1 (Mamba) ?

OMG Thank God you brought it up. I was thinking my V1 was breaking down hahaha but yes rango is much much brighter than V1

nitroshift wrote:

For those building their own images: do NOT select kmod-dsa-mv88e6352, kmod-dsa-mv88e-6xxx is enough.

nitroshift


thnx. why is that?  so no SWconfig for rango?

Folks, i think that we shouldn't separate information for rango here, it should be in 1900 thread, because i find some interesting information on rango on 1900 thread as well... Now i need to follow tho threads:-)

@Chadster766 what are the perks if i run your McDebian on my router?

(Last edited by gsustek on 19 Feb 2017, 13:57)

My Trunk built image from OpenWRT (Build 50104) has a cron to run the famous fan_ctrl.sh script. AFAIK Rango doesn't have a fan cooler. Am i wrong ? If not, I think the script can be safely removed from the Rango image.

@gsustek

Swconfig doesn't support the switch chip in Rango. Building with kmod-dsa-mv88e6352 makes the unit inaccessible by wire. Regarding separate threads: the wrt1900ac thread is already bloated with all kind of posts, some of which not being directly related to that unit, so having a dedicated thread for each device makes more sense and also makes searching a lot easier, not to mention that you have a wrt1900acs too, so you'll most likely check the other thread anyway wink

nitroshift

@sorinello

Only the original WRT1900AC (v1 - Mamba) has a fan, all other models don't. Feel free to submit a patch smile

nitroshift

sorinello wrote:

My Trunk built image from OpenWRT (Build 50104) has a cron to run the famous fan_ctrl.sh script. AFAIK Rango doesn't have a fan cooler. Am i wrong ? If not, I think the script can be safely removed from the Rango image.

nitroshift wrote:

Only the original WRT1900AC (v1 - Mamba) has a fan, all other models don't. Feel free to submit a patch smile

This could just be backported from LEDE:
https://github.com/lede-project/source/ … 7138e9643e

(Last edited by hnyman on 20 Feb 2017, 12:57)

nitroshift wrote:

@gsustek

Swconfig doesn't support the switch chip in Rango. Building with kmod-dsa-mv88e6352 makes the unit inaccessible by wire. Regarding separate threads: the wrt1900ac thread is already bloated with all kind of posts, some of which not being directly related to that unit, so having a dedicated thread for each device makes more sense and also makes searching a lot easier, not to mention that you have a wrt1900acs too, so you'll most likely check the other thread anyway wink

nitroshift


so how is it possible for LEDE build to have swconfig? Please explain.

thnx.

@gsustek

While not going to start a flamewar, Kaloz's commit that brings mv88e6352 support in swconfig was only pushed by Noltari in LEDE tree, pending the merge between the two projects.

nitroshift

@nitroshift
I just want to know how is it possible at technical level...

thnx

p.s. i am not taking any side

I unboxed my Rango yesterday and put linux-next (aka 4.11) on it. Using current mwlwifi with in-kernel wireless stack and hostapd-2.6. As the region is locked to US I can't operate the 2.4GHz band, iw says 30 db instead of the allowed 20 db. 5GHz band is fine, same rules it seems.

Running iperf for 24h without a drop in wireless performance. The laptop (agn client) more or less gets constant 40 Mbit, which is slightly better than the same device and Shelby.

@sera

I have seen no issues with running a single instance of iperf with either kernel 4.4 or 4.9 -- in fact, the performance is excellent and very stable.  Where I see issues is if I run multiple instances of iperf and drive those from multiple clients.  The existing mwlwifi driver craters quickly.  I'm guessing this is similar to the situation where an access point is serving multiple independent clients.

InkblotAdmirer,

might not be that easy to reproduce for me then, not many 5GHz devices. Are 2 or 3 enough already?

I think you only need two other wireless clients attached.  I run an iperf server on each and then two terminals on the Rango unit.  Then run a couple independent iperf commands, one to each client.

InkblotAdmirer,

New setup, the iperf server is connected via ethernet, now 3 iperf client devices each connected via wireless. After about 3h I still haven't seen any degradation. All involved devices are running Linux.