RAK WisBlock 4631 firmware compile - big difference in file sizes from Web flasher vs. Platform IO

[rbreesems]

I have a WisMesh Pocket V2, whose internals are a RAK4630 WisDuo module + RAK19026 WisMesh Baseboard. When I use the Web Flasher, it selects a RAK WisBlock 4631 as the target device, with firmware 2.5.20.xxxxx that has 3015 blocks (flashing this firmware in the radio works). When I download the Meshtastic github repo, and use Platform IO to build, the closest match I get for a build is the RAK 4631 (there is also a rak_4631_eth but the final firmware build is not much different). When I build with Platform I/O, I get only 2521 blocks, which is a big difference, and it kills my radio when I flash it (I have to use a nanoDap with cloned firmware from a working radio to resurrect). Both firmware files start at address 0x26000, but the file from the Web Flasher has a lot more blocks in it. Why the differences in firmware file sizes? Why is the Platform IO build so different from the firmware build from the Web Flasher (which I though was built with Platform I/O). How is the firmware file from the Web Flasher build? I am confused. Thanks for any help!

[fifieldt]

There's a softdevice that lives in the earlier memory I believe. If you look under bin/build-nrf.sh you'll see where it's merged together with mergehex.

[rbreesems]

Thanks. I got brave and flashed my recovered radio with firmware that I compiled with VS Code/Platform I/O, and it worked this time. I somehow messed up the download the first time. The differences in file sizes is still puzzling, perhaps it is due to different compiler optimization flags, I don't know. Anyway, I am happy now that I can compile locally, flash, and it works.

I will checkout out bin/build-nrf.sh.