On Tue, 15 Dec 2020 at 15:03, Mark Kettenis <mark.kette...@xs4all.nl> wrote: >
> > Thanks. This is committed now. However, there may be other case > where we use uvm_km_valloc() early on that will trip over the kernel > lock assertion that mpi@ added in uvm_km_pgremove(). Ideally we > should get rid of all the uvm_km_free() calls in the kernel. > Thank you and mpi@ for the fix. I wasn't able to test your diff, since my machine doesn't have the src checked out, so thank you to bluhm and Hrvoje for testing it out.