> On Apr 10, 2020, at 1:26 PM, Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx> wrote: > > On Fri, 10 Apr 2020 10:26:23 -0400 Qian Cai <cai@xxxxxx> wrote: > >> I don't set panic_on_warn. I'll deal with warnings afterwards. > > I'm not understanding why sysbot sets panic_on_warn. This decision > will needlessly turn many kernel errors into wont-boot situations and > will block further testing? I can feel that it is very reasonable to set panic_on_warn for the fully automatic systems because once those warnings happen, the rest of things can no longer to be trusted, so the goal to kill the first enemy on sight, and then deal the next one. It could be a good idea for some trees more stable like the mainline, but for linux-next, I could only dream of set it one day.
