https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=289204
--- Comment #6 from Mark Johnston <[email protected]> --- (In reply to Konstantin Belousov from comment #5) Suppose the thread created some GEOM, and there is some asynchronous initialization path which creates /dev/foo etc. for the new GEOM. Then we want that initialization to finish before further system calls are executed, to avoid races. For kernel threads this is not really a problem, so perhaps we should only schedule the AST for user threads. Peter, are you able to test the patch? https://reviews.freebsd.org/D52421 -- You are receiving this mail because: You are the assignee for the bug.
