On Fri, Jul 14, 2023 at 12:15:23PM -0700, Sandeep Dhavale wrote: > > > > Sandeep, thoughts? > > > I prefer to modify erofs check and contain the fix there as erofs > cares about it and it's > least invasive. For contexts where erofs cannot tell for sure, it will > always schedule kworker > (like CONFIG_PREEMPT_COUNT=n). > > I will also do measurements to see if erofs should continue to check > for context and > what are the benefits. At the end of the day, I guess it is Gao Xiang's and Chao Yu's decision. Thanx, Paul
