No need to stress-test the boot procedure because I'm already doing it. What I need from my beta testers (and thank you greatly for your help by the way!), is to test the regular features they use. I plan on releasing a newer beta soon with a new rtorrent and some fixes so you might want to wait for that instead.
Regarding my stability tests, I did a new one over the night. I restored my regular config (the one I used when merging and testing features) and I set WDTVEXT=ON and I rebooted and checked the box in a loop. Out of 186 reboots I had 186 times the OSD started correctly and 0 failures. So what the fuck?!?! It works correctly each time with my config, and only starts 8-15% of the time on the default config.
Now I need to devise a plan to see how to isolate the difference in my config (compared to the default config). The problem is I have about 450 settings in my config (a lot of UMSP stuff, also some older settings), so it would take about two days of continuous test to try each setting. And it wouldn't be conclusive, as we've seen in previous tests (there's a chance the box starts with wrong config).
I'll need to put my best people on it pronto, but I could use a different config where WDTVEXT is somewhat stable (@DjArty can you PM me your config as well?) to see what we have in common and eliminate the differences...
