wildcat_paris Posted January 9, 2016 Posted January 9, 2016 @Tido I am not blaming the B53125, it is just nice, complete and afaik efficient (I liked your link/comment some time earlier! tkaiser is overdoing "lamobo-r1 is crap" so "every piece of lamobo-r1 is crap") 1/ Lamobo implemented 1 GMAC PHY with the A20 SoC for 2 (or more) VLAN. 2/ Lamobo may have added a separate PHY for the WAN on the B53125 (but only little better for BW) 3/ A20 SoC with 2 GMAC (one for WAN, one for LAN) with B53125 in the middle 4/ A20 SoC with 2 GMAC (one for B53125, one for WAN PHY) Your diagram (google docs), you are pretty right -- the only issue is => the WAN port number is arbitrary (you can use swconfig to set 0,1,2,3 or 4 linked to A20-GMAC port 8, pick port 3 for WAN=eth0.101) and the remaining "LAN" ports (0,1,2,4) for eth0.102 => all of the ports are linked to the A20-GMAC port 8 anyway More simply, when you setup eth0.101 and eth0.102 VLAN both are sharing the same A20-GMAC port 8 so linux routing/iptables (=CPU) handle every moves from eth0.101 to eth0.102 and vice-versa (VLAN are theorically separate LANs). in doing so, the BW is limited to (as a mean empiric value) A20-GMAC BW divided by 2 (not counting the terrible CPU/IRQ load) As A20-GMAC has hardware/software/etc issues RX/TX are not optimal, so you can divide the BW by 2 again (empiric value, worst case). So I am not that far stating the BW is 1/4 of GMAC BW when transfering data with Lamobo-r1 as a LAN-to-LAN or LAN-to-WAN "router" when eth0.101 and eth0.102 are in charge. EDIT: hand written schemas http://tinyurl.com/gw2m6z5 note : a completely separate WAN PHY GMAC would be far more secure! EDIT2: The only great thing with Lamobo-r1 you can spy data secretly on a LAN (flood the network to make switches working as hubs / or spy you ISP between their BOX and Internet, how do they handle SIP-like telephony)
Tido Posted January 10, 2016 Posted January 10, 2016 A first short comment to your drawing, I come back later with more. The WLAN Chip RTL8192 is as far as I know connected via USB2 to A20, means your Plan-A is wrong, Plan-B is partially right, I guess. 1
tkaiser Posted January 10, 2016 Posted January 10, 2016 I tought of tkaiser saying the B53125 has only on GMAC (it would need 2 GMAC one "in" one "out" = at least 2 Gbit/s) to handle traffic in/out Nope, it's the other way around: the BCM53125 has 5 GbE PHYs (external Ethernet ports) and 2 RGMII ports (to connect one or two SoCs) but A20 has only one RGMII interface. And ONLY if the BCM53125 would act differently when both RGMII ports are used (not as a dumb layer2 switch by default but in this mode interconnecting only 1 GbE PHY with RGMII port A and the 4 other GbE PHYs with RGMII port B ) then it could be used as a router. We neither have a SoC that features 2 RGMII interfaces nor the necessary knowledge whether the switch IC can boot up in this different default state. Regarding the Lamobo R1S: This is actually two boards in one combining the A20 with a MIPS SoC using another 256 MB of RAM and running OpenWRT. It would have to be confirmed in which state the MT7621A internal 5 port switch boots (also bridging all ports or differentiating between WAN and LAN) to judge. BTW: Before buying a crappy SinoVoip product again I would combine any other A20 device with something like the WiTi board (based on the very same MT7621A but using all the features of this SoC like USB3 or 2 x SATA and so on). Fortunately the R1S seems to be a prototype only. 1
tkaiser Posted January 10, 2016 Posted January 10, 2016 A20 SoC with 2 GMAC A20 is simply the wrong SoC since it lacks the necessary interconnects. What you're thinking about might work with the i.MX6 Quad for example. It features SATA II (90-100 MB/s in both directions), a somewhat limited GbE implementation (400 Mbits/sec -- you might want to use this with RGMII and an external GBit Ethernet PHY for WAN) and a single PCIe 2.x lane to be used with a GbE PCIe controller connected to the BCM53125 for example. Then you've a quad core Cortex-A9 solution utilising up to 4 GB RAM (twice as much as with any Allwinner SoC) and acting like a router should act. How the BCM53125 behaves in this mode is irrelevant, you need not even drivers for this IC since it should be a dumb layer 2 switch. But you could also use MDIO to define additional VLANs (and you might even rely on VLANs for routing purposes if you don't care about security). In case you really care about powerful Wi-Fi too you would've to rely on a SoC made for that (providing a few more PCIe lanes to be used with different Wi-Fi mini-PCIe cards or combine i.MX6 with PCIe switches maybe just to realise that you're running in bandwidth issues). A good example is Marvell Armada 38x as used on the Clearfog or Turris Omnia. Again regarding the i.MX6: The Dual core variants have also PCIe (so you can combine the internal GbE MAC with a 2nd real PCIe NIC) but lack SATA therefore the quad core variant would be the best solution to build something comparable to Lamobo R1 that does NOT suck (Utilite Standard/Pro both use an Intel I211 as 2nd NIC that shows way better performance than the i.MX6's internal GbE implementation). There exist many vendors selling i.MX6 based SoM solutions so this would be the route to go to get something comparable to the R1 for 30 bucks more (the SoC alone is more expensive than an A20 and an additional PCIe controller increases costs also) But for a router/NAS combination encryption performance might also be interesting and so I would have a look at hardware accelerated encryption (Freescale's CAAM or Marvell's CESA). I would assume Marvell wins and the Turris folk will try hard to get CESA drivers upstream. Then Armbian support for the Clearfog might automagically benefit from this too. 1
wildcat_paris Posted January 10, 2016 Posted January 10, 2016 (edited) Nope, it's the other way around: @tkaiser yep, I wrote "B53125 has only on GMAC" but I thought "A20 has only on GMAC", my bad. EDIT: Thomas, I have fixed my previous post "A20 SoC has only one GMAC link to the B53125" Edited January 10, 2016 by wildcat_paris
Tido Posted January 10, 2016 Posted January 10, 2016 WiTi board (based on the very same MT7621A but using all the features of this SoC like USB3 or 2 x SATA and so on). Currently you cannot purchase the WiTi board, in the forum I read about issues with WiFi, some say a heatsink would help, max speed 20m. Looks like they suffer similiar problems, like we do. Although I think they have access to the schematic FireWRT from Firefly I found while reading about WiTi, they have opened all documents, incl. BOM, CAD, schematics, housing. I think both projects suffer on something - to many components. i.e. TK wants fast SATA, Ethernet and maybe USB3.0, but no WiFi i.e. Tido wants SATA, Ethernet and WiFi So the PCB may fit all, but not all compontents have to be placed = different Versions of the board to keep the price and troubleshooting low(er). 1
tkaiser Posted January 10, 2016 Posted January 10, 2016 So the PCB may fit all, but not all compontents have to be placed = different Versions of the board to keep the price and troubleshooting low(er). Whatever that means... The Turris Omnia is based on a SoC that does not guarantee lowest throughput as with Allwinner's A20 and costs me 125,- bucks including shipping/VAT. It's a no-brainer to prefer this over the crappy Lamobo stuff. When my Wi-Fi AP dies sometimes in the future, I will replace it temporarely with another AP and then check prices/situations regarding available mini-PCIe Wi-Fi cards. That's what you get with a modular design providing SATA, 2 x USB3, 3 GbE PHYs and a few PCIe lanes and a SoC providing plenty of I/O and network bandwidth. 1
Igor Posted January 10, 2016 Posted January 10, 2016 FireWRT looks like a dead project. https://forum.openwrt.org/viewtopic.php?pid=283363#p283363
Tido Posted January 10, 2016 Posted January 10, 2016 @TK, Whatever that means... Maybe if you read it again you get my idea. FireWRT looks like a dead project. I think so too, but as I wrote: they have opened all documents, incl. BOM, CAD, schematics, housing. So whoever would take over, does not have to start from zero. That saves a lot of money & time. Awesome
Shawn Wood Posted January 12, 2016 Posted January 12, 2016 @wildcat. Just wanted to update on this. Using the untouched image for Armbian Debian Jessie - Vanilla, on my Lamoba-R1 (with only an apt-get update and an apt-get dist-upgrad) - I was able to get solid, stable WIFI. I was originally testing it in an environment with 20 neighbors stepping on the channels, but when I took it to where it will be used - there is no interference. It was streaming 2 HD TV's and running speed tests at once, and delivering data as fast as my ISP allows - which is as good as I could hope for. So I've got to report that the build is working well. My glitch now is that the board appears to lock up at night. LED is on in the morning, ethernet lights are flickering, but no response from a ping or SSH. Not sure if one of the cron tasks is doing it, or what but will be experimenting with that. 1
tkaiser Posted January 12, 2016 Posted January 12, 2016 My glitch now is that the board appears to lock up at night. LED is on in the morning, ethernet lights are flickering, but no response from a ping or SSH. Not sure if one of the cron tasks is doing it, or what but will be experimenting with that. Nothing in the logs? BTW: Monitoring almost always helps (with mainline kernel unfortunately a few data sources are missing, but at least you might get an idea when the lock-up occurs): http://forum.armbian.com/index.php/topic/155-testers-wanted-sunxi-adjustments-for-rpi-monitor/
wildcat_paris Posted January 15, 2016 Posted January 15, 2016 @wildcat. Just wanted to update on this. Using the untouched image for Armbian Debian Jessie - Vanilla, on my Lamoba-R1 (with only an apt-get update and an apt-get dist-upgrad) - I was able to get solid, stable WIFI. Shawn, that's a miracle... 1
Shawn Wood Posted January 15, 2016 Posted January 15, 2016 And yet another follow up on to my noobie attempts at making this work. I discovered that Hostapd was just eating memory when the service was running. It would go about 24 hours before hosing the board. So I got a fresh install from Igors downloads (Debian Jessie Vanilla) - and this time I didn't apply any dist-upgrade, update or upgrade to it. The thing is now running perfectly. I'm getting solid, acceptable performance on the WIFI (it's not sizzling by any means, but given my slow ISP, it is as good as it needs to be). Been running several days and memory use has stabilized (with 800 megs free memory), no lock-ups yet. So - many thanks Igor for your most excellent images. I'm going to go back and re-image my other Banana Pi's to use these same versions! 2
pschasch Posted January 18, 2016 Posted January 18, 2016 good morning, @shawn woodif you think you have a stable 8192cu-wifi, please test wifi (ex. after about 2-3 days running R1) if your client is not directly next to the R1, so thake a bit distance and then start ex. a youtube-HD-videostream or some big OS-Updates on your client(s).In all my tests with all kind of software-variations (OS,8192cu-drivers,hostapd) my 8192cu-Wifi on R1 was hanging after some days. After this hanging WIFI, restarting any hostapd-versions nor reloading any tested drivers did help! Only reboot the system did fix the problem. On Amazon people with 8192-devices also write those "distance/stream"-PROBLEM-comments, even with windows drivers for 8192. Because of this chip-problem I gave up too on R1-8192cu and did a ra5572-hardware mod. see : http://forum.armbian.com/index.php/topic/372-hardware-mod-bpi-r1/?p=4314 My R1-Wifi is stable for 8 days now. Further testing in progress... regards 1
Tido Posted January 18, 2016 Posted January 18, 2016 After this hanging WIFI, restarting any hostapd-versions nor reloading any tested drivers did help! Only reboot the system did fix the problem. On Amazon people with 8192-devices also write those "distance/stream"-PROBLEM-comments, even with windows drivers for 8192. I realized your reloading problem long time ago, in my 15.04 Bananian manual I wrote therefore: This service must start after the wlan0, after isc-dhcp !! I am looking after a solution, if you have a working one, for: with service networking restart hostapd must be the last to restart. Solution: shutdown -r now Well, I played around with it for quite a while and finally gave up - armbian worked much better and still does. I am playing right now with Bananian 15.08 and its cold start on R1 with a spindle drive is horrible to listen to Beside WiFi does not work. Therefore I still don't have a manual for Bananian 15.08 released People with Windows drivers for 8192 have problems, by any chance do you still have a link to this?
pschasch Posted January 18, 2016 Posted January 18, 2016 Hi Tido, thanks for your work and your manual! I realized your reloading problem long time ago, in my 15.04 Bananian manual I wrote therefore:...Well, I played around with it for quite a while and finally gave up - armbian worked much better and still does. I had this problem on bananian and on a past version of armbian too,. I was also writing with an other user here in a personal conversation who tested this for me too on his R1 using armbian. Till this point he also thought that he had a stable wifi...but he got the same behaviour on distance to 8192cu and big-data-streams. :-) ...crash...only reboot fixed it... People with Windows drivers for 8192 have problems, by any chance do you still have a link to this? There may be a lot of good feedbacks for usb-8192cu, but for me those bad feedbacks are no coincidence. That's what I found for example on german Amazon: --------------------------------------------------------------------------------------------------------------------------------------------- TP-Link TL-WN822N (8192cu-version)=============================== "Vorsicht: Neue Version mit schlechtem Chipsatz"http://www.amazon.de/review/R1OO1828LHK3BM/ref=cm_cr_dp_title?ie=UTF8&ASIN=B00416Q5KI&channel=detail-glance&nodeID=340843031&store=computers....."Die Realtek-Chipsätze (inkl. 8192CU) sind für deutlichen Leistungsabbruch mit zunehmender Distanz zum Router bekannt. Mein PC ist etwa 4 Meter vom Router entfernt. Das ist bereits genug für deutliche Leistungseinbrüche und Verbindungsverlust."..... " In V3 nicht zu empfehlen"http://www.amazon.de/review/R37FFPKZWSOJW5/ref=cm_cr_dp_title?ie=UTF8&ASIN=B00416Q5KI&channel=detail-glance&nodeID=340843031&store=computers.....Sofern ich keine größeren Downloads vornehme funktioniert das Gerät einwandfrei. Die Geschinwidkeit liegt bei ca. 144mbit/s. Sobald ich jedoch einen schnellen Download durchführe (z.B. Game-Download in Steam) verliert er ständig die Verbindung. Kabel ziehen und wieder einstecken hilft nur bedingt - nach 30 Sekunden reist die Verbindung meist direkt wieder ab. Die aktuellsten Treiber helfen da leider auch nicht...... Asus USB-N13 N300 Wi-Fi USB Stick==============================http://www.amazon.de/product-reviews/B0033C7ZAK/ref=cm_cr_pr_viewopt_kywd?ie=UTF8&showViewpoints=1&sortBy=helpful&pageNumber=1&filterByKeyword=8192cuhttp://www.amazon.de/gp/customer-reviews/R2RW8FKQBHCXK2/ref=cm_cr_pr_rvw_ttl?ie=UTF8&ASIN=B0033C7ZAK: "ASUS hat einen besseren Namen als dieses Produkt zeig"http://www.amazon.de/gp/customer-reviews/R39WN8RVII520N/ref=cm_cr_pr_rvw_ttl?ie=UTF8&ASIN=B0033C7ZAK: "Verliert ständig die Verbindung zum Router"http://www.amazon.de/gp/customer-reviews/RHNTQ3U82FH5U/ref=cm_cr_pr_rvw_ttl?ie=UTF8&ASIN=B0033C7ZAK: "UNBRAUCHBAR"http://www.amazon.de/gp/customer-reviews/R315NOA5H9NN21/ref=cm_cr_pr_rvw_ttl?ie=UTF8&ASIN=B0033C7ZAK: " Internetverbindung bricht zwischendurch kurz ab" http://www.amazon.de/gp/customer-reviews/R3H7J50F6GOASH/ref=cm_cr_pr_rvw_ttl?ie=UTF8&ASIN=B0033C7ZAK: "kommt dazu das ab und zu keine Verbindung aufgebaut wird"--------------------------------------------------------------------------------------------------------------------------------------------- After "ONLY" testing (no real/productive use) the R1 as "Stable"-AP for more then 1 year after I bought it, those amazon-comments made me stop testing on 8192cu-R1. I hope that someone has a stable 8192cu-R1 now or in the future, in that case he should test the distance/big-datastream-problem and on success he has to post all used software-versions (OS,hostapd,kernel,8192cu-driver) and configs for those in this forum. For me the R1-Lamobo-MT5572-hardware-mod is the best solution for now, With this "15+5 Euro"-mod I have a Lamobo-R1 with a stable WIFI-chip, a free USB-host-port and a WIFI-chip which supports more modes (ex. monitor-mode which is interesting to me). regards 1
Tido Posted January 18, 2016 Posted January 18, 2016 (edited) Somewhere I also read, that only one (1) Antenna is active - can't remember if there is a fix for that And if you want to use 40 MHz bandwidth you need to hack the hostapd file Thank you for your detailed information. I consider to embed this into the manual. Did you document on a website or such your modification with some more pictures /text ? I just wonder which WiFi chip is in my old: FON2201 (now in my closet for 2 years) this bloody device also lost connection all the time - it made me sooo angry Edit: I just found out, that I even have a second FON the 'simpl', but this has only 4MB flash. With the 2201 I will try to put OpenWrt on it Edited January 18, 2016 by Tido
pschasch Posted January 18, 2016 Posted January 18, 2016 Hi, >Somewhere I also read, that only one (1) Antenna is active - can't remember if there is a fix for that Yes, at my R1-8192cu also only 1 antenna gets better signal. is it because of MIMO? one antenna OUT, the other IN? But this Doesn't matter for me anymore... >And if you want to use 40 MHz bandwidth you need to hack the hostapd file I tested everything, also this patches. with/without-Debug-patched and with/without-40Mhz-patches ( http://forum.armbian.com/index.php/topic/136-lamobo-r1-wifi-unstable-in-ap-host-mode-better-buy-a-good-wifi-dongle-with-proper-linux-support/?p=2271 ) Nothing was stable with 8192cu. >Thank you for your detailed information. I consider to embed this into the manual. >Did you document on a website or such your modification with some more pictures /text ? No, Unfortunately not. But it's the same and you can test with every ra5572-USB-device for ex. : wildcat_paris his good "Dlink 160B1"-USB-Wifi. (thanks for config-help wildcat!!! :-) ) or the "CSL 300Mbit USB WLAN"-USB-ra5572 from Amazon which I used to test/prepare software before receiving the solder-solution from china. After I received I send the CSL back to Amazon! Test and Configure for those, after testing simply unsolder the 8192cu and solder the MT5572 like on my picture. After soldering replace wlan1 to wlan0 in your configs. I could help you with some more pictures from my mod, but you won't see more as in the published one from the mod-thread I guess. >I just wonder which WiFi chip is in my old: FON2201 (now in my closet for 2 years) >this bloody device also lost connection all the time - it made me sooo angry ...R1 Lamobo made me also very often very angry while testing 8192cu and also if it was in my closet and can't be used as AP. No hardware acceleration makes me still very angry... :-) regards 1
Tido Posted January 18, 2016 Posted January 18, 2016 while I was looking up my bloody FON routers, I read about the HW-mod to add a second 'real' antenna to the FON 2201. They wrote, this is only worth it if people move around with their devices. Reading this, I don't know how big the difference would be for the R1 if both Antenna's working nicely.
wildcat_paris Posted January 18, 2016 Posted January 18, 2016 And yet another follow up on to my noobie attempts at making this work. I discovered that Hostapd was just eating memory when the service was running. It would go about 24 hours before hosing the board. @Shawn hostapd didn't eat my memory ** but the quite not so useful irqbalance DID 51.7% (@tkaiser played with me about the swap memory so BIG, he is right!) gr@bpi:~$ top 1:Def - 00:26:03 up 11 days, 11:17, 2 users, load average: 0.06, 0.10, 0.13 Tasks: 120 total, 1 running, 119 sleeping, 0 stopped, 0 zombie %Cpu0 : 0.0 us, 2.6 sy, 0.0 ni, 97.4 id, 0.0 wa, 0.0 hi, 0.0 si, 0.0 st %Cpu1 : 0.0 us, 0.5 sy, 0.0 ni, 99.5 id, 0.0 wa, 0.0 hi, 0.0 si, 0.0 st KiB Mem: 1025632 total, 903728 used, 121904 free, 95248 buffers KiB Swap: 2158240 total, 681800 used, 1476440 free. 108696 cached Mem 1 PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 2844 gr 20 0 14060 3352 2292 S 0.5 0.3 0:00.49 sshd: gr@pts/2 6503 gr 20 0 4648 2092 1696 R 2.0 0.2 0:01.37 top 3330 root 20 0 S 0.5 0:05.59 [kworker/0:2] 2 PID PPID TIME+ %CPU %MEM PR NI S VIRT SWAP RES UID COMMAND 30322 2 20 0 S 0 kworker/0:1 29265 1 0:00.01 0.1 20 0 S 1640 108 1044 0 getty 25043 1 0:12.31 0.5 20 0 S 6164 5020 112 dhcpd 13451 2 0:05.49 20 0 S 0 kworker/1:2 8912 1 0:00.05 0.2 20 0 S 2540 1952 110 privoxy 6533 2882 0.0 20 0 S 1340 336 0 sleep 6503 2846 0:01.37 2.0 0.2 20 0 R 4648 2092 1000 top 6501 2 20 0 S 0 kworker/u4:1 6499 1 0:00.11 0.2 20 0 S 3872 2256 0 hostapd 6463 2 20 0 S 0 kworker/1:1 6430 2 20 0 S 0 kworker/0:0 6336 1 0:00.53 0.0 20 0 S 1336 60 0 rngd 3 PID %MEM VIRT SWAP RES CODE DATA SHR nMaj nDRT S PR NI %CPU COMMAND 2128 51.7 1185760 653100 530764 28 1183112 924 S 20 0 irqbalance 2262 2.6 83088 13840 26412 368 75344 2472 483 S 20 0 named 2828 2.4 32584 4556 24832 176 28288 1200 2348 S 20 0 dnsmasq 1858 2.1 57540 312 21260 336 55000 1336 79 S 20 0 rsyslogd 2846 0.7 8824 6784 632 4620 2216 S 20 0 bash 2718 0.6 13916 6280 564 784 5268 S 20 0 sshd 25043 0.5 6164 5020 1172 3684 1752 35 S 20 0 dhcpd 2690 0.5 5500 32 4936 128 4284 900 21 S 20 0 polipo 5211 0.3 5888 3524 564 588 3044 S 20 0 sshd 2844 0.3 14060 3352 564 928 2292 S 20 0 0.5 sshd 2855 0.2 5724 772 2492 132 1888 1316 149 S 20 0 lighttpd 3193 0.2 28752 800 2380 84 23908 2188 49 S 20 0 console-kit-dae 4 PID PPID UID USER RUSER TTY TIME+ %CPU %MEM S COMMAND 2855 1 33 www-data www-data ? 0:54.46 0.2 S lighttpd 1858 1 102 syslog syslog ? 6:40.41 2.1 S rsyslogd 1 0 0 root root ? 0:43.27 0.2 S init 2 0 0 root root ? 0:00.46 S kthreadd 3 2 0 root root ? 5:22.11 S ksoftirqd/0 5 2 0 root root ? S kworker/0:0H 7 2 0 root root ? 12:22.95 S rcu_sched 8 2 0 root root ? 0:00.03 S rcu_bh 9 2 0 root root ? 0:13.05 S migration/0 10 2 0 root root ? 0:10.87 S migration/1 11 2 0 root root ? 27:35.24 S ksoftirqd/1 13 2 0 root root ? S kworker/1:0H @Tido This service must start after the wlan0, after isc-dhcp !! I am looking after a solution, if you have a working one, for: with service networking restart hostapd must be the last to restart. Solution: shutdown -r now Well, I played around with it for quite a while and finally gave up - armbian worked much better and still does. depending on the use of upstart or systemd, Ubuntu 14.04 has /etc/rc.local and assuming you are using bridge br0 with eth/wlan, is it an option to do in bash/natural language : " service hostapd stop ;ifdown wlan0;kill isc-dhcpd; isc-dhcpd ( in daemon mode); ifup wlan0 ; service hostapd start" ? but is it possible? As there is a graph dependancy of network/network-related services an other option would be in /etc/network/interfaces playing with pre-up / post-up / etc. options (including /bin/sleep 10 ?) but it can get tricky and slow bootup very much (it is a very dirty fix) (joke) Bananian has talented people but they don't have the "full-time" day and night hobbist crew like Igor/Zador&Co @Sascha && @Tido Do you use the mini-board rt5572 antennas pins now? Because maybe the wiring/soldering of wifi pins on the Lamobo-r1 is faulty making the antennas "broken" (Or, more simply, as you write, everyone in the world *hates* the 8192cu chip)
tkaiser Posted January 19, 2016 Posted January 19, 2016 the quite not so useful irqbalance DID 51.7% Why is irqbalanced running? And just in case you want to play with it: https://github.com/dv1/irqbalanced
aaiyar Posted February 2, 2016 Posted February 2, 2016 Hello - First post to this forum - hope it isn't out of line. I have been running Armbian 4.5 on my Lamobo-R1 for a few months. WiFi has been stable, in that connections do not drop, as long as I disable 802.11n. However, there are two issues that may be related: 1) Throughput is crap - I max out at 8-10 Mbps - which fortunately is sufficient for my needs 2) Huge number of dropped RX packets. Anything I can do to improve this? I've cut & paste details below. Regards and thanks, Ashok P.S Thomas - if you read this post, perhaps you remember me from more than a decade ago on the netatalk list? hostapd -v hostapd v2.5-devel User space daemon for IEEE 802.11 AP management, IEEE 802.1X/WPA/WPA2/EAP/RADIUS Authenticator Copyright © 2002-2015, Jouni Malinen <j@w1.fi> and contributorsl ifconfig wlan0 wlan0 Link encap:Ethernet HWaddr 20:f4:1b:07:81:e2 UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1 RX packets:197826 errors:0 dropped:3382973 overruns:0 frame:0 TX packets:207831 errors:0 dropped:104 overruns:0 carrier:0 collisions:0 txqueuelen:1000 RX bytes:95269704 (90.8 MiB) TX bytes:93574764 (89.2 MiB) uname -a Linux bpi-r1 3.4.110-sun7i #10 SMP PREEMPT Mon Dec 28 11:48:43 CET 2015 armv7l GNU/Linux cat /etc/hostapd/hostapd.conf # interface interface=wlan0 bridge=br0 # CTRL-Interface ctrl_interface=/var/run/hostapd ctrl_interface_group=0 # driver driver=rtl871xdrv # SSID ssid=Ashok's Guest Network ignore_broadcast_ssid=0 # WLAN hw_mode=g ieee80211n=0 channel=1 max_num_sta=8 wmm_enabled=0 wme_enabled=0 # passphrase wpa_passphrase=###### # WPA2 auth_algs=1 wpa=2 wpa_key_mgmt=WPA-PSK wpa_pairwise=CCMP TKIP rsn_pairwise=CCMP
tkaiser Posted February 2, 2016 Posted February 2, 2016 P.S Thomas - if you read this post, perhaps you remember me from more than a decade ago on the netatalk list? Sure I do --I remember also the nice ebtables stuff to tunnel AppleTalk through IP Regarding dropped RX packets unfortunately no idea since I gave up on this device completely and used the Wi-Fi only once to have a short laugh and drop the idea to use it: The best you can do with the Wi-Fi module is to unsolder it: http://forum.armbian.com/index.php/topic/372-hardware-mod-bpi-r1/
aaiyar Posted February 3, 2016 Posted February 3, 2016 Sure I do --I remember also the nice ebtables stuff to tunnel AppleTalk through IP Regarding dropped RX packets unfortunately no idea since I gave up on this device completely and used the Wi-Fi only once to have a short laugh and drop the idea to use it: The best you can do with the Wi-Fi module is to unsolder it: http://forum.armbian.com/index.php/topic/372-hardware-mod-bpi-r1/ Thanks, Thomas for the suggestion! I know someone with a solder station - I think I will try the hardware mod. At worst I lose ~$30. Because of a traumatic event, I lost most of my memory of events from the early 90's to mid-2000's, including various Linux networking projects. Thanks to low-priced SBCs, I'm slowly getting back in the game .....
erpomata Posted May 18, 2016 Posted May 18, 2016 I did not understand how it was solved the problem.You must replace the soc wifi (link) or by the method published by Rui Ribeiro or that published by Chester? Thank you
Recommended Posts