Wanted to give a thank you for the good documentation how to flash and the (till now) perfectly working RolandoMagic firmware I'm using. Bought the device some days ago at media markt germany for 18 euros. The device is just a "backhaul" for my Hackintosh to achieve wifi 6 speeds.
It is connected to my TP-Link EAP660HD where I'm using WPA2-EAP encryption. Would be nice to have to full wpad-wolfssl from the beginning since there is lots of space but was no problem to download wpad-wolfssl and wpa-cli into the /tmp folder, remove wpad-basic and set up the WPA2-EPA connection. Just a small hint from my side for people who want to connect to enterprise networks: The crt upload for the connection didn't work, i added it manually over ssh inside the /etc/luci-uploaded folder. Before uploading any certificate it makes sence to activate "use system certificates", log into ssh, check the log with "logread -f" and get the subject of the server certificate. Add subject and server cert to the connection settings and here we go.
Topology is like this:
Client (1000mbit) via Ethernet to DAP x1860 which is in Client mode to EAP660HD on 5ghz which is connected via 1000mbit ethernet to switch via 1000mbit to ESXI server with an debian instance and iperf3 server
This is what I got after let's say 1 1/2 walls (client is a hackintosh, connected via 1000mbit to the ethernet port):
Hi everyone,
I'm currently using the snapshot from Friday and see the following issue:
When scanning the wireless networks, independent which "Scan" button in Luci I use, I only see the 2.4GHz networks.
Manually joining a 5GHz network works as expected.
Can anybody confirm this issue.
Sorry, I am still on 2023-01-07 but was thinking of switching to a snapshot. There have been a lot of commits this weekend . There also have been some updates to the mt76 driver, which I assume will be merged to master snapshot soon, so I was waiting for that.
Btw. if I may ask Rolando, what packages did you select in the firmware selector to end up with LuCI and default config? I would not need lot's of fancy packages, just the usual stuff that would be comparable to the images you provide in your fork.
On a side note: To find the regression window, using the 50/50 rule of debugging is advised, as it's the fastest way:
Cut commits in half between last friday and 2023-01-16 and try the commit in the middle.
a) If bug is encountered, cut older half of that into another half and try middle commit....
b) If no bug is encoutnered, cut newer half of that into another half and try middle commit ...
Repeat until you can say for sure you found the commit that introduced the regression.
It's like cutting a cake into ever smaller pieces. 1/2 --> 1/4 --> 1/8 --> 1/16 --> 1/32 --> 1/64 --> ...
For the images I provided, I created a default configuration and selected LuCI as an additional package. It was not done by the firmware selector, I built them manually.
For the problem I reported above, I used the firmware selector with additional packages LuCI, relayd and the LuCI relayd plugin.
I’ll try to reduce the problem to a specific commit.
I understand, but my expectation was that there would be either a message indicating the problem or that scanning the 5Ghz network won’t show any results at all. Showing the 2,4 GHz networks when scanning 5GHz was confusing for me.
Somebody is having problems too. Not sure if actually related: Belkin RT1800 snapshot r21950 - mt7915e fail to load
I was neither able to find r21599, nor the r21950 commit. Would be nice, if people would actually provide links to the commit.
After some rounds of git bisect, builds and flashing:
f3d8de73988a79440d183050dc9711bb29d2a5e1 is the first bad commit
commit f3d8de73988a79440d183050dc9711bb29d2a5e1
Author: Christian Marangi <ansuelsmth@gmail.com>
Date: Sat Jan 21 01:23:22 2023 +0100
iwinfo: update to latest Git HEAD
git bisect start
# status: waiting for both good and bad commits
# bad: [d02d4556789f59f472c3134569a341a48f18c1ec] kernel: make it possible for packages to select page pool support
git bisect bad d02d4556789f59f472c3134569a341a48f18c1ec
# status: waiting for good commit(s), bad commit known
# good: [26c095cb4d27428acedf1b738507a585f1585fe4] ipq807x: add Dynalink DL-WRX36
git bisect good 26c095cb4d27428acedf1b738507a585f1585fe4
# bad: [183bad664167f73a6798980faab5d9b5bba06b5a] CI: tools: add gnu-getopt to macOS CI
git bisect bad 183bad664167f73a6798980faab5d9b5bba06b5a
# good: [00f1463df7e690862403208082f71fb4741baf02] mbedtls: move source modification to patch
git bisect good 00f1463df7e690862403208082f71fb4741baf02
# bad: [f183ce35b8ea2fd991ac489fb223b09a1ecb4db0] kernel: mtk-bmt: fix usage of _oob_read
git bisect bad f183ce35b8ea2fd991ac489fb223b09a1ecb4db0
# good: [84ee3436a44dac3809b05a1c900e34dbebf030c1] ipq806x: Point to externally compiled dtbs in recipes
git bisect good 84ee3436a44dac3809b05a1c900e34dbebf030c1
# good: [bd0f9d8ffcd1dc169fb66f9d7ddb335abbe49d37] fstools: bump to latest Git HEAD
git bisect good bd0f9d8ffcd1dc169fb66f9d7ddb335abbe49d37
# bad: [f3d8de73988a79440d183050dc9711bb29d2a5e1] iwinfo: update to latest Git HEAD
git bisect bad f3d8de73988a79440d183050dc9711bb29d2a5e1
# good: [ef649b0b14814a84a480e9b7805e4a401e9e06b6] ipq806x: Initial TP-Link and ASUS OnHub support
git bisect good ef649b0b14814a84a480e9b7805e4a401e9e06b6
# first bad commit: [f3d8de73988a79440d183050dc9711bb29d2a5e1] iwinfo: update to latest Git HEAD
While looking at older commits, two commits later there was:
commit b61404a6ad100b1e8305a84dbeacc2cae9d66ccf
Author: Christian Marangi <ansuelsmth@gmail.com>
Date: Sun Jan 22 19:51:39 2023 +0100
rssileds: bump PKG_RELEASE due to libiwinfo ABI change
Has anyone recently built an image and checked whether rssileds is now working?
// edit: ok this does probably not affect the naming of the interfaces, just resolves an issue introduced i the commit mentioned earlier (ABI change with new iinfo version).
Was thinking about upgrading to newest master snapshot, as the buildbot has recently finished its build for the ramips/mt7621 target. @RolandoMagico, do you still experience scanning issues? - iwinfo: bump to latest git HEAD (2023-02-06) has been merged, which was supposed to fix
Today I got two slightly used DAP-X1860 from amazon for 30€ a piece. Flashed the latest snapshot, used Win11 edge, everything went smoothly. Using 802.11s meshes with both radios without hickups. Thank you all. Nice small router.
This looks like the same result that other devices with MT7621 have achieved, not sure if iperf3 is indeed running single-threaded as some suggested, but it's a lot faster than qca9563 for example.
If we can still reproduce issues with newest snapshot, we probably should report them over there...