On Fri 2018-08-24 23:59:49, Jan Kundrát wrote: > On pátek 24. srpna 2018 21:50:00 CEST, Pavel Machek wrote: > >I'd say this is "ti,our-hw-designers-forgot-to-connect-reset-line", > >and I see nothing OS specific about that... > > Actually, our board first resets the chip, and then U-Boot uses some LEDs as > status indicators to show progress of various stages of booting. That sounds > like a useful design to me, but it ends up with Linux claiming that all LEDs > are dim even though some channels are HW-blinking. > That's why I think that the chip should be always reset by the kernel. > Alternatively, all registers would have to be explicitly initialized so that > kernel's idea and reality match again. Explicitely initializing all registers sounds reasonable. Should work even on unusual i2c designs. Pavel -- (english) http://www.livejournal.com/~pavelmachek (cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
Attachment:
signature.asc
Description: Digital signature
