On Mon, 2018-05-14 at 16:58 -0700, Paul E. McKenney wrote: > OK, so if I define pr_fmt as follows, I get the old behavior? > > #define pr_fmt(fmt) fmt
yes. > I just queued > a commit to be squashed into my version of your patch 18/18 that adds > this to kernel/rcu/rcu_segcblist.c. This joins the ones that your > patch added to kernel/rcu/srcutree.c and kernel/rcu/tree.c. > > Should I also add "#define pr_fmt(fmt) "rcu: " fmt" to these files? > > kernel/rcu/srcutiny.c > kernel/rcu/sync.c > kernel/rcu/tiny.c Well, I don't actually remember why the define was added to rcu_segcblist.c. Most of this was done via a script which looked for pr_<level> uses without a pr_fmt in the same file. Dunno why rcu_segcblist was modified. Maybe it was a braino.
