raschid Posted June 26, 2017 Posted June 26, 2017 3 hours ago, martinayotte said: I will probably commit it soon, but leaving it disabled ... Thank you. That will certainly help people willing (and able) to compile their own kernels. I wish the rest could have access to the driver though ...
martinayotte Posted June 26, 2017 Posted June 26, 2017 3 hours ago, raschid said: Thank you. That will certainly help people willing (and able) to compile their own kernels. This commit should allow that : https://github.com/armbian/build/commit/cb48540d4a230e1ef0dfc507f0e4ac165d37ee18 (I've done quick try and it worked for me)
kutysam Posted June 27, 2017 Posted June 27, 2017 Just wondering, could this be enabled manually by the user on current kernels? In other words, those who wish to use this, is there an alternative way instead of building the kernel by 1) modprobe 2) load the dts in? It will greatly benefit us if we have either options.
zador.blood.stained Posted June 27, 2017 Posted June 27, 2017 5 hours ago, kutysam said: is there an alternative way instead of building the kernel Yes - building the module out of tree on the device and using a DT overlay (or DT patch). It's not documented, but for an experienced user it shouldn't be difficult.
ldiaz Posted June 27, 2017 Posted June 27, 2017 7 hours ago, kutysam said: Just wondering, could this be enabled manually by the user on current kernels? In other words, those who wish to use this, is there an alternative way instead of building the kernel by 1) modprobe 2) load the dts in? It will greatly benefit us if we have either options. This was my original proposal to use an overlay I posted one initial version that I wan't able to make work but I'm not a DTS expert maybe someone can simply aling it with the patch that @martinayotte created. 1
raschid Posted June 27, 2017 Posted June 27, 2017 Some test result using iperf. Distance to AP ca. 5m line of sight. Zero as client: [ ID] Interval Transfer Bandwidth [ 3] 0.0-10.0 sec 17.4 MBytes 14.6 Mbits/sec [ 3] 10.0-20.0 sec 17.6 MBytes 14.8 Mbits/sec [ 3] 20.0-30.0 sec 17.4 MBytes 14.6 Mbits/sec [ 3] 30.0-40.0 sec 17.9 MBytes 15.0 Mbits/sec [ 3] 40.0-50.0 sec 16.6 MBytes 13.9 Mbits/sec [ 3] 50.0-60.0 sec 17.2 MBytes 14.5 Mbits/sec [ 3] 0.0-60.1 sec 104 MBytes 14.5 Mbits/sec Zero as server: [ 4] 0.0-61.3 sec 96.9 MBytes 13.3 Mbits/sec dmesg reports no xradio related drops/errors.
Bubba Posted June 27, 2017 Posted June 27, 2017 OK so now that others see that zero wi-fi is not as crappy as others have stated. I have wireless streamed / transferred 67 gig across a zero this month and did not see any real issues. Now can we get it enabled for all the users?? NOT having it enabled really does not make much sense, if the powers to be do not wish to "support" then don't. They could move those post to Peer support. Some of us are only looking at IoT projects, moving txt / csv / xml files, and so on. I did a little project a few weeks back, call it a proof of concept, I was sent 10 zero's / sd cards and power supplies purchased as a package deal. I setup the zeros as asked and they went out to service, the Monkey tech installing the zero's had issues with a cisco catalyst ??, and another cisco switch??. The "tech" (firedog / geeksquad / Gurus to go) did not know what model switches.... He said one zero would not connect (wired), said he had to plug one into the switch with 12 " cable, going thru the punch down block did not work, another he had to use a cheap switch between the zero and cisco. I am told those 2 zeros have heavy packet loss over the wired connection and drop off the network once or twice a day. Now just to be clear I did not recommend the Zero, I said he should use the PC Plus. P.S. Because I got paid for this project I donated a small part (about 10%) to Igor..as I used Armbian build for the zeros. Hint... Hint... If everyone did this Igor and ??? could have a weekend long drunk on us. 1
zador.blood.stained Posted June 27, 2017 Posted June 27, 2017 3 minutes ago, Bubba said: OK so now that others see that zero wi-fi is not as crappy as others have stated. Can anybody test Creating an access point on the Zero Connecting the Zero (as client) to an open AP to confirm that it's not that crappy? 4 minutes ago, Bubba said: Now can we get it enabled for all the users?? For sun8i-dev (and sunxi-dev in the future) - possibly, as long as somebody will maintain it and update with future kernel updates. For sunxi-next (stable/supported releases) I will scream YOU SHALL NOT PASS each time anybody suggests adding it to the branch that is considered "stable".
Igor Posted June 28, 2017 Posted June 28, 2017 Updated dev / testing image for Opi Zero, wireless + upstream patches included: https://dl.armbian.com/orangepizero/Ubuntu_xenial_dev.7z 4
raschid Posted June 28, 2017 Posted June 28, 2017 quick check ... iperf - as client: [ ID] Interval Transfer Bandwidth [ 3] 0.0-10.0 sec 13.4 MBytes 11.2 Mbits/sec [ 3] 10.0-20.0 sec 15.4 MBytes 12.9 Mbits/sec [ 3] 20.0-30.0 sec 15.6 MBytes 13.1 Mbits/sec [ 3] 30.0-40.0 sec 16.0 MBytes 13.4 Mbits/sec [ 3] 40.0-50.0 sec 15.2 MBytes 12.8 Mbits/sec [ 3] 50.0-60.0 sec 15.9 MBytes 13.3 Mbits/sec [ 3] 0.0-60.0 sec 91.6 MBytes 12.8 Mbits/sec ... as server: [ ID] Interval Transfer Bandwidth [ 3] 0.0-10.0 sec 10.1 MBytes 8.49 Mbits/sec [ 3] 10.0-20.0 sec 10.0 MBytes 8.39 Mbits/sec [ 3] 20.0-30.0 sec 12.1 MBytes 10.2 Mbits/sec [ 3] 30.0-40.0 sec 9.12 MBytes 7.65 Mbits/sec [ 3] 40.0-50.0 sec 11.9 MBytes 9.96 Mbits/sec [ 3] 50.0-60.0 sec 11.1 MBytes 9.33 Mbits/sec [ 3] 0.0-60.3 sec 64.5 MBytes 8.98 Mbits/sec very few "missed interrupt" errors (1 per measurement).
tkaiser Posted June 28, 2017 Posted June 28, 2017 24 minutes ago, raschid said: quick check Using which Opi Zero PCB rev? 1.4 or lower?
Bubba Posted June 28, 2017 Posted June 28, 2017 On my zero ver 1.1 it was going good I was sucking down a 4 gig file, I pushed it and ran update / upgrade at the same time. On reboot wi-fi is gone, takes about 4 mins for the ethernet to come back up. Will reburn and test it once more.
raschid Posted June 28, 2017 Posted June 28, 2017 35 minutes ago, tkaiser said: Using which Opi Zero PCB rev? 1.4 or lower? rev 1.1
raschid Posted June 28, 2017 Posted June 28, 2017 31 minutes ago, Bubba said: On reboot wi-fi is gone, takes about 4 mins for the ethernet to come back up. @Bubbadid you update/upgrade before the reboot? That seems to kill wlan0 (by "upgrading" dtb and kernel) ...
tkaiser Posted June 28, 2017 Posted June 28, 2017 12 minutes ago, raschid said: rev 1.1 Thank you. Now we need also some 1.4 results since https://forum.armbian.com/index.php?/topic/4313-new-opi-zero-yet-another-high-temperature-issue/ @bozdenit's time for testing! Are you able to provide Wi-Fi throughput numbers made with identical conditions (antenna included) with one 1.4 OPi-Z and an earlier one?
Bubba Posted June 28, 2017 Posted June 28, 2017 @raschid I had rebooted before I ran the update. I am running some load test now, but I have replaced the "stock" antennas with these https://www.aliexpress.com/item/WIFI-2-4G-3dbi-PCB-Antenna-IPX-IPEX-WLAN-Laptop-Bluetooth-Zigbee-Wireless-Module-SIM900-SIM800L/32732546964.html?spm=2114.13010608.0.0.j0sOkS , https://www.aliexpress.com/item/Hot-Sale-Smart-Electronics-1set-ESP-8266-ESP8266-Serial-WIFI-Wireless-Transceiver-Module-MT7681-3DBI-Gain/32351052354.html?spm=2114.13010608.0.0.j0sOkS and with my bad eyesite and old hands it is not easy for to swap the stock ones. No luck with OPEN ap's, it talks to the AP but never connects. as of now very few errors are showing in client mode downloading from a local server on lan.
raschid Posted June 29, 2017 Posted June 29, 2017 21 hours ago, Bubba said: No luck with OPEN ap's, it talks to the AP but never connects. Same here. Also tested the fifteenhex-version of the driver. No luck either. This is no deal breaker for me - I fail to see significant use cases for open wifi with this device.
Bubba Posted June 29, 2017 Posted June 29, 2017 Yea I would NEVER connect to an AP. I was running load test on the zero, left it running over night, I guess it was to much for the zero, the ethernet port died. I have a 1.4 ver I will test this weekend.
Hamish Posted July 5, 2017 Posted July 5, 2017 I signed up to the forums just so I could add my voice to this ticket. I've been using the Orange pi zeros for a couple of months and was really happy with how well the wifi worked as an access point. So I am a little disturbed to discover that the wifi driver has been disabled and that I need to pin my kernel version (especially since I told thirty other people to get their own Orange pi zero based on my testing...) Is there a list of issues with this driver somewhere? 1
Bubba Posted July 5, 2017 Posted July 5, 2017 start here... https://forum.armbian.com/index.php?/topic/3243-orange-pi-zero-wireless-module-status-xradio-st-cw1200/ the zero works fine in IoT projects, thing is some folks think a cheap 7.00 board can replace 200.00 of x86 gear. ap limit of 5 connections, weak wi-fi signal, crappy stock antenna, weak psu = failure before they start Igor did build a 4.1x image with wi-fi it is linked here
martinayotte Posted July 5, 2017 Posted July 5, 2017 Although the WiFi patch is in disabled state, you can still build your own image with the patch re-enabled and KERNEL_CONFIGURE=yes.
kutysam Posted July 6, 2017 Posted July 6, 2017 On 6/28/2017 at 8:52 PM, Igor said: Updated dev / testing image for Opi Zero, wireless + upstream patches included: https://dl.armbian.com/orangepizero/Ubuntu_xenial_dev.7z Unfortunately, once apt-get upgrade kicks in, a reboot will make the wifi dissappear. Anyway to prevent this ?
Igor Posted July 6, 2017 Posted July 6, 2017 20 minutes ago, kutysam said: Unfortunately, once apt-get upgrade kicks in, a reboot will make the wifi dissappear. Anyway to prevent this ? Froze kernel upgrading in armbian-config and than apt-upgrade won't upgrade system packages ( kernel, uboot, bsp, ...) 1
Hamish Posted July 7, 2017 Posted July 7, 2017 On 05/07/2017 at 6:47 PM, Bubba said: start here... https://forum.armbian.com/index.php?/topic/3243-orange-pi-zero-wireless-module-status-xradio-st-cw1200/ Yeah, I read that topic some months ago - I was more hoping for a list of what would is needed to re-enable the driver in standard armbian builds. Yes, I can compile it myself, but that doesn't help me when I'm suggesting that other people get this board and use Armbian. Right now, I've ended up telling people to get Armbian installed and then manually install the older kernel version - but then I run the risk that the old kernel .deb will disappear.
Igor Posted July 7, 2017 Posted July 7, 2017 2 hours ago, Hamish said: what would is needed to re-enable the driver in standard armbian builds There are no standard Armbian builds for Orange pi Zero with kernel 4.x and you are obviously not a developer, whom this build is made for. Topic regarding this shitty wireless driver and related questions forced us to abandon support for it. We already throw into this problem insane amount of our time - it looks we hit the bottom long time ago - and all we get is negative reaction. Sorry, we only do our best to make this peace of shitty hardware work. We did not made it and you did not bought it from us.
Hamish Posted July 7, 2017 Posted July 7, 2017 2 hours ago, Igor said: There are no standard Armbian builds for Orange pi Zero with kernel 4.x and you are obviously not a developer, whom this build is made for. Topic regarding this shitty wireless driver and related questions forced us to abandon support for it. We already throw into this problem insane amount of our time - it looks we hit the bottom long time ago - and all we get is negative reaction. Sorry, we only do our best to make this peace of shitty hardware work. We did not made it and you did not bought it from us. If you would like my 'developer' credentials, perhaps I should point you at https://github.com/hamishcoleman That being said, I am not asking for support, nor am I complaining about this shitty device, I just am asking what specific development work - if done (by others - not by you) - would change the default status of this hardware driver back to "enabled"
Igor Posted July 7, 2017 Posted July 7, 2017 13 minutes ago, Hamish said: That being said, I am not asking for support, nor am I complaining about this shitty device, I just am asking what specific development work - if done (by others - not by you) - would change the default status of this hardware driver back to "enabled" Apologise for my misunderstanding and mistreatment. There is so much frustration surrounded this chip, that I don't even try to think. I can only wish you luck.
zador.blood.stained Posted July 7, 2017 Posted July 7, 2017 48 minutes ago, Hamish said: I just am asking what specific development work - if done (by others - not by you) - would change the default status of this hardware driver back to "enabled" This won't happen because people will still come to us for support. Feel free to build an image, share it with others and provide better level of support for this wireless module if you want, but at least this image won't be named "Armbian stable".
raschid Posted July 8, 2017 Posted July 8, 2017 On 7.7.2017 at 10:04 AM, zador.blood.stained said: This won't happen Thank you @zador.blood.stained for clearing this up. I had been genuinly under the impression that you thought about reactivating the driver when you were asking folks to test stuff : On 27.6.2017 at 8:08 PM, zador.blood.stained said: an anybody test Creating an access point on the Zero Connecting the Zero (as client) to an open AP Bit of a waste of time, right? One last request to @Igor: please update the armbian hompage regarding folks "becoming a part of our vibrant community, contribute ideas and have fun" - because it isn't, you can't and you won't. Just read this thread from top to bottom and you will see what I mean. Whoops, my stop - this is where I get of ... bye
zador.blood.stained Posted July 8, 2017 Posted July 8, 2017 8 minutes ago, raschid said: Thank you @zador.blood.stained for clearing this up. I had been genuinly under the impression that you thought about reactivating the driver when you were asking folks to test stuff For clarifications - this most likely won't happen unless there are significant improvements to the driver quality. These images (nightlies with the dev kernel) images are (or rather were) provided for developers and people who are willing to accept the current software situation, and not for people who will endlessly create threads like "I downloaded Orange Pi Zero mainline image and I can't connect to open AP please fix!!!!111". In my opinion compiling and installing an out-of-tree driver is an easy but powerful barrier to reduce the number of people who don't read the documentation and known issues.
Recommended Posts