From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: From: Rusty Russell To: Lucas De Marchi , Masami Hiramatsu Cc: Lucas De Marchi , Linux Kernel Mailing List , linux-modules , Josh Poimboeuf Subject: Re: [RFC PATCH 0/2] kmod: Support lockup option to make module un-removable In-Reply-To: References: <53FC52DA.9000303@hitachi.com> <20140826120442.8223.80618.stgit@kbuild-fedora.novalocal> Date: Mon, 13 Oct 2014 15:11:31 +1030 Message-ID: <87mw90eg3o.fsf@rustcorp.com.au> MIME-Version: 1.0 Content-Type: text/plain List-ID: Lucas De Marchi writes: > On Tue, Aug 26, 2014 at 9:04 AM, Masami Hiramatsu > wrote: >> Hi, >> >> Here is a pair of patches which adds --lockup option to >> modprobe and libkmod. >> >> As I sent a series of patches which removes stop_machine() >> from module removal: https://lkml.org/lkml/2014/8/25/142 >> it also adds lockup option which lock up the module in >> the kernel and makes it un-removable. >> >> These patches enables us to use that option when loading >> modules. Module lockup may be good for BIG SMP machines >> since the kernel skips module refcounting if the module >> is locked up :) >> >> Anyway, this is not needed if the lockup option is dropped >> from the series. I send this for testing. > > Ok. I'm not sure it's clear... I'm waiting for feedback on the kernel > patches in order to proceed with any review here. I'm not really > convinced we want this option when loading a module. > > Rusty, what do you think? I'm not convinced, I asked him to drop that patch. If we have significant performance issues, we'll have to do something smarter I think anyway. Cheers, Rusty.