Note: I start out with a comparison to FreeBSD use with pftf/RPi4. But this is not a request for pftf/RPi4 changes.
FreeBSD crashes in early kernel activity for Graphical for pftf/RPi4 but is useful with Serial selected (log in, build FreeBSD, etc.). (I use disable-bt .)
Turns out FreeBSD crashes in the same way for pftf/RPi3 (on a RPi3B v1.2) as for Graphical with pftf/RPi4 --but there is no explicit/alternative selection present in pftf/RPi3: implicitly/always Graphical, or so it appears. I expect that, with Serial available to select and save, the result would work like it does for pftf/RPi4 .
Beyond the above:
The crash is such that USB input to the kernel prompt is non-functional and the serial console is not available (no input, output stopped earlier in the sequence when it switched to Graphical). This makes capturing the messages leading to the kernel-backtrace difficult to capture. (I do not claim to know how this fits with pftf/RPI3 vs. the FreeBSD kernel vs. both but it is a difficulty for evidence gathering.) With a sustained serial output stream I'd be able to just record it and then report it on a FreeBSD list.
Note: I start out with a comparison to FreeBSD use with pftf/RPi4. But this is not a request for pftf/RPi4 changes.
FreeBSD crashes in early kernel activity for Graphical for pftf/RPi4 but is useful with Serial selected (log in, build FreeBSD, etc.). (I use disable-bt .)
Turns out FreeBSD crashes in the same way for pftf/RPi3 (on a RPi3B v1.2) as for Graphical with pftf/RPi4 --but there is no explicit/alternative selection present in pftf/RPi3: implicitly/always Graphical, or so it appears. I expect that, with Serial available to select and save, the result would work like it does for pftf/RPi4 .
Beyond the above:
The crash is such that USB input to the kernel prompt is non-functional and the serial console is not available (no input, output stopped earlier in the sequence when it switched to Graphical). This makes capturing the messages leading to the kernel-backtrace difficult to capture. (I do not claim to know how this fits with pftf/RPI3 vs. the FreeBSD kernel vs. both but it is a difficulty for evidence gathering.) With a sustained serial output stream I'd be able to just record it and then report it on a FreeBSD list.