The 2.6.6 wizard now allows the AP to have a wifi ap. I set it up and was able to connect wifi to my laptop to the ap’s ssid. Can you get to luci on the ap this way?
I can connect to the client’s wifi ap and access both the client and ap using the halow ip addresses from each
Another 2.6.6 wizard question: on the wifi access point page, when you select to enable, I don’t know what I’ve done to make the fields behave differently, but in one build, the SSID field will be a default “OpenWrt” with the passphrase field blank and in another instance, the SSID field will be a combo of project name and 4 other chars and the passphrase field will be 8 random characters. It’s just that as I was going from 2.5.3 to 2.6.6, slowly pulling unique changes in that were made in 2.5.3, the behavior was the former with OpenWrt as SSID and at some point whatever I did made it be the other behavior. I can’t recall what changes made that happen. What is it that controls whether you get one behavior or the other? I prefer the latter.
with the changes to luci for 2.6.6, how do you setup items like in the MM6108 - OpenWrt 2.5 Web GUI User Guide section 5 Setting up iPerf traffic testing in the table for iperf3 now since there’s no Morse/Halow Config menu item? What are the new steps?
Also, when I get the yellow warning banner that region is not set, if I back out of the wizard when it initially comes up, how is the region set if you don’t go through the initial page shown before the wizard? It used to be able to be set in 2.5.3 on the wizard Morse/Halow Config first page if it wasn’t set. Or should you just do that first page of the wizard where you pick region and hostname and then do the X to setup for iperf3 test?
building 2.6.6 is confusing. Sometimes when I’ve done a “make dirclean” and do a build, it takes about 4 hours (2.5.3 took about 2.5 hrs). But sometimes, it seems to take forever. For instance, I started a build about 8pm last night. At 8am this morning, it’s still building. Any ideas on this? I don’t know if I should just cancel it and start over again or what?
Attached is the log when the project was still building at 12 hours, normally a fresh build takes about 4 with 2.6.6. From the log, does that give you any clues as to what’s taking so long?
But one thing it didn’t do in the build that it’s done before was ask about some pinctrl debug something or other where the build stops until you enter Y or N. I enter N. Is there any way to have the build automatically take it as N without awaiting a user response?
(attachments)
log still building at 12 hours.zip (1.78 MB)
Sorry, forgot you’d be running a different config. Use the
luci-app-ttyd package (this is installed by default for our boards via utils_diffconfig).
If you check to see if it’s disabled as well, this would work regardless.
Yes; are you facing any difficulty with this?
Looks like on one build you have morse-bundle and one you don’t, perhaps? This is 98-morse-wireless-defaults file, which we use on ekh devices to provide ‘better’ defaults than normal OpenWrt (where as you see, the default behaviour is just to have OpenWrt as the SSID). Alternatively, there may be an issue with this script executing before the morse module insertion has caused updates to /etc/config/wireless. To fix this, you could add an extra delay in /etc/init.d/boot (see the sleep).
I never do a clean build on my local machine, so build times are < 2 minutes for no changes 
OpenWrt’s build system is usually pretty reasonable about not needing cleans, though occasionally you have to clean individual packages after certain changes.
Main variability I would think it whether you’re using a pre-compiled toolchain or building it yourself. It looks like the logs you attached were doing this. I’m still surprised it would take that long. If you want to never build a toolchain, you can use the ‘-E’ argument to morse_setup (which will set the appropriate config option and download one from the openwrt servers if it hasn’t already).
Sorry that we missed updating that part of the user guide for 2.6. Just set up your connection via the wizard and run iperf3 in the terminal (personally, I would just use ssh rather than using the luci terminal, but if you install luci-app-ttyd you can do it via the web interface). You can also run iperf3 via the diagnostics page.
Attached is a log for a build that wasn’t finished. After 14 hours I did ctrl-C. I did “make distclean”, then “make”. Yesterday I did the same thing and it was only just short of 4 hours or so. The log obviously isn’t verbose but does this make any sense?
As I finished typing this note, I then did “make -j” and the build finished in less than 10 minutes. Again, does that make any sense? Is it getting hung up in something the first time?
(attachments)
Re: check if wireless.radio1 is disabled won’t work. When you load a new image, before you do the very first luci page to set region, the entry “wireless.radio1.disabled=‘1’” is present in “uci show wireless”. However, as soon as you go to the next page which is the main wizard page where you select halow, mesh or easymesh, when you do uci show wireless again, “wireless.radio1.disabled=‘1’” is not in the list. Thus, that condition says the radio is enabled. Then if “wireless.default_radio1.mode=‘ap’” is there by default, that would mean radio is enabled and it’s an ap. Isn’t that all correct? I know I’m still grasping this whole architecture, but am I missing something?
what was it about building the 2.5.3 project after a “make dirclean” or new project where shortly after the build was started it paused and waited for the user to enter y or n for some pinctrl debug something or other? What is the pinctrl debug thing? I haven’t seen that in the 2.6.6 build, which is totally fine because I don’t think it’s a good idea to get user input in the middle of a time consuming build, if it can be helped.
I’m building the 2.6.6 sdk in ubuntu 22, usually I’ve been using only “make -j1”. I thought I’d build the sdk with “make” and “make -j” to see if it went any faster. However, doing this, the build bombs after a few minutes and ubuntu gives a black window with debug info for shutting down processes due to “out of memory”. Then ubuntu goes to the password screen and when the main screen comes back up, all previous windows are gone so it probably rebooted ubuntu. So far, using “make -j1” hasn’t been an issue, but I’m not ruling it out to not maybe happen sometime. Have y’all experienced anything like this? Btw, my ubuntu vm has 8gb memory. I tried boosting it to 32gb in virtualbox but that didn’t have any effect on the problem.