Repository navigation
v2.16-beta.40 — dashboard reboot loop fixed
Pre-releasev2.16-beta.40 — dashboard reboot loop fixed
Now released as stable: v2.16.0, the same firmware. The fix was confirmed on the Waveshare S3, the ATOM Lite and the T-2CAN (#223, #230).
If your board's Wi-Fi kept going off and on, or the dashboard sat on "Connection lost — retrying…" with every field on --, this is the fix. It affects beta.37 and beta.38 on every board; please update from beta.39 as well.
- ESP32: opening the dashboard no longer reboots the board (#219, #223). Since beta.37 the loop task ran out of stack the moment a browser connected its WebSocket. The board rebooted, the page reconnected two seconds later and rebooted it again. From the outside that looked like the Wi-Fi dropping in a loop, on every phone and PC. The Performance card in beta.37 had pushed the deepest dashboard path from about 7.6 KB to about 8.8 KB of an 8 KB stack. The loop task now gets 16 KB, and the CAN handlers no longer sit inside
loop()'s own stack frame. (PR #234) - ESP32 dashboard: Memory row. Device now shows the loop task's lowest free stack and the heap, and
/api/statuscarriesstack_free,heap_freeandheap_min. The boot banner prints the stack size. (PR #234)
This was worked out from the compiled frame sizes, nobody has posted a panic log, so I'd like to hear whether it holds on your board. If something still looks off, the Memory row under Device is the first thing to send.
Everything from beta.39 is included: the captive portal now completes for iPhones and Macs, the dashboard works inside the captive sheet, the footer names the real board and shows the release version, and library versions are pinned.
The Flipper app is unchanged.
Thanks to @olschmalz-cyber, whose test on two Windows PCs, an Android phone and an iPhone is what ruled out the captive portal, and to @nobless for the first report and the screenshots.
Full history in changelog.md.