> The definition of v1 is very simple and absolutely precise. > The code is v1 when Roland thinks we can make the RFC patch and > send it for review.
Ok. My definition of success remains "whatever works to get it merged". > As I see it, we need the following changes. > > utrace: > > - utrace_control(SINGLESTEP) calls get_user_pages() under > spin_lock() Yes. > ptrace: I'm not actually tracking closely enough to verify which implementation holes are or aren't there. So I'd state it in the high-level way: it works without regressions and is not too much slower. That means verifying ptrace-tests (perhaps modulo job-control corners), gdb testsuite, and UML. > - simple cleanups, may be more comments. Right, I call this "whatever is needed for upstream review". Thanks, Roland
