Messages in this thread | | | From | Rusty Russell <> | Subject | Re: [PATCH v2 3/3] livepatch: Cleanup module page permission changes | Date | Mon, 09 Nov 2015 09:31:24 +1030 |
| |
Josh Poimboeuf <jpoimboe@redhat.com> writes: > On Fri, Nov 06, 2015 at 02:42:46PM +0100, Petr Mladek wrote: >> naming schemes. What about adding into the public API? >> >> set_module_ro() >> set_module_rw() >> >> It should modify everything: init, core, text, and data but only >> the ro/rw flags. > > Even that naming is not without its problems. For example, > set_module_ro() is false advertising -- it wouldn't change *all* module > memory to be read-only. (It wouldn't touch the r/w data areas.) > > But I don't really care what the interfaces are called. It's really > Rusty's call. I just stuck to the existing naming convention in the > module code with the set/unset ro_nx stuff.
I'm looking at the ro/nx stuff now, and it seems like a mess. For example, set_all_modules_text_rw() and set_all_modules_text_ro() use mod->core_text_size instead of mod->core_ro_size. Which is probably what they want (ftrace doesn't care about rodata) but pretty damn confusing.
So I'll extend your cleanup. Expect a patch for testing RSN...
Thanks, Rusty.
| |