在2023年12月20日十二月 下午9:18,Thomas Bogendoerfer写道: > On Sun, Oct 29, 2023 at 02:53:01AM +0000, Jiaxun Yang wrote: >> The plan is to elimiate platform specific early_printk and >> cps-vec-ns16550 by debug_ll and earlycon. > > https://xkcd.com/927/ ? > > sorry I don't think that just another new function is good approach. > Doing this will end up with another method for early debugging and > all other will stay. To summarize how are we going to handle low-level debugging after this series: 1. You are lucky enough that the problem happens after console system initialized and you have stack working: go earlycon, or on very few old platforms with platform early_printk 2. You are debugging zboot: debug_ll 3. You are debugging SMP bootstrap code like cps-vec: debug_ll 4. Your kernel crashed at the middle of no where before trap_init: debug_ll Thanks > > Thomas. > > -- > Crap can work. Given enough thrust pigs will fly, but it's not necessarily a > good idea. [ RFC1925, 2.3 ] -- - Jiaxun